PERLTIDY(1) User Contributed Perl Documentation PERLTIDY(1)

NAME

perltidy - a perl script indenter and reformatter

SYNOPSIS

perltidy [ options ] file1 file2 file3 ...
(output goes to file1.tdy, file2.tdy, file3.tdy, ...)
perltidy [ options ] file1 -o outfile
perltidy [ options ] file1 -st >outfile
perltidy [ options ] <infile >outfile

DESCRIPTION

Perltidy reads a perl script and writes an indented, reformatted
script. The formatting process involves converting the script into a
string of tokens, removing any non-essential whitespace, and then
rewriting the string of tokens with whitespace using whatever rules
are specified, or defaults. This happens in a series of operations
which can be controlled with the parameters described in this
document.

Perltidy is a commandline frontend to the module Perl::Tidy. For
documentation describing how to call the Perl::Tidy module from other
applications see the separate documentation for Perl::Tidy. It is
the file Perl::Tidy.pod in the source distribution.

Many users will find enough information in "EXAMPLES" to get started.
New users may benefit from the short tutorial which can be found at
http://perltidy.sourceforge.net/tutorial.html

A convenient aid to systematically defining a set of style parameters
can be found at http://perltidy.sourceforge.net/stylekey.html

Perltidy can produce output on either of two modes, depending on the
existence of an -html flag. Without this flag, the output is passed
through a formatter. The default formatting tries to follow the
recommendations in perlstyle(1), but it can be controlled in detail
with numerous input parameters, which are described in "FORMATTING
OPTIONS".

When the -html flag is given, the output is passed through an HTML
formatter which is described in "HTML OPTIONS".

EXAMPLES

Here are some example perltidy commands followed by their meanings:

perltidy somefile.pl

This will produce a file somefile.pl.tdy containing the script
reformatted using the default options, which approximate the style
suggested in perlstyle(1). The source file somefile.pl is unchanged.

perltidy *.pl

Execute perltidy on all .pl files in the current directory with the
default options. The output will be in files with an appended .tdy
extension. For any file with an error, there will be a file with
extension .ERR.

perltidy -b file1.pl file2.pl

Modify file1.pl and file2.pl in place, and backup the originals to
file1.pl.bak and file2.pl.bak. If file1.pl.bak and/or file2.pl.bak
already exist, they will be overwritten.

perltidy -b -bext='/' file1.pl file2.pl

Same as the previous example except that the backup files
file1.pl.bak and file2.pl.bak will be deleted if there are no errors.

perltidy -gnu somefile.pl

Execute perltidy on file somefile.pl with a style which approximates
the GNU Coding Standards for C programs. The output will be
somefile.pl.tdy.

perltidy -i=3 somefile.pl

Execute perltidy on file somefile.pl, with 3 columns for each level
of indentation (-i=3) instead of the default 4 columns. There will
not be any tabs in the reformatted script, except for any which
already exist in comments, pod documents, quotes, and here documents.
Output will be somefile.pl.tdy.

perltidy -i=3 -et=8 somefile.pl

Same as the previous example, except that leading whitespace will be
entabbed with one tab character per 8 spaces.

perltidy -ce -l=72 somefile.pl

Execute perltidy on file somefile.pl with all defaults except use
"cuddled elses" (-ce) and a maximum line length of 72 columns (-l=72)
instead of the default 80 columns.

perltidy -g somefile.pl

Execute perltidy on file somefile.pl and save a log file
somefile.pl.LOG which shows the nesting of braces, parentheses, and
square brackets at the start of every line.

perltidy -dbs -dbl=10 somefile.pl >blocks.csv

This will dump a table of comma-separated metrics for subroutines
longer than 10 lines to blocks.csv.

perltidy -duv somefile.pl >vars.txt

This will dump a list of unused and reused lexical variable names to
vars.txt.

perltidy -html somefile.pl

This will produce a file somefile.pl.html containing the script with
html markup. The output file will contain an embedded style sheet in
the "<HEAD"> section which may be edited to change the appearance.

perltidy -html -css=mystyle.css somefile.pl

This will produce a file somefile.pl.html containing the script with
html markup. This output file will contain a link to a separate
style sheet file mystyle.css. If the file mystyle.css does not
exist, it will be created. If it exists, it will not be overwritten.

perltidy -html -pre somefile.pl

Write an html snippet with only the "<PRE"> section to
somefile.pl.html. This is useful when code snippets are being
formatted for inclusion in a larger web page. No style sheet will be
written in this case.

perltidy -html -ss >mystyle.css

Write a style sheet to mystyle.css and exit.

perltidy -html -frm mymodule.pm

Write html with a frame holding a table of contents and the source
code. The output files will be mymodule.pm.html (the frame),
mymodule.pm.toc.html (the table of contents), and
mymodule.pm.src.html (the source code).

OPTIONS - OVERVIEW
The entire command line is scanned for options, and they are
processed before any files are processed. As a result, it does not
matter whether flags are before or after any filenames. However, the
relative order of parameters is important, with later parameters
overriding the values of earlier parameters.

For each parameter, there is a long name and a short name. The short
names are convenient for keyboard input, while the long names are
self-documenting and therefore useful in scripts. It is customary to
use two leading dashes for long names, but one may be used.

Most parameters which serve as on/off flags can be negated with a
leading "n" (for the short name) or a leading "no" or "no-" (for the
long name). For example, the flag to outdent long quotes is -olq or
--outdent-long-quotes. The flag to skip this is -nolq or
--nooutdent-long-quotes or --no-outdent-long-quotes.

Options may not be bundled together. In other words, options -q and
-g may NOT be entered as -qg.

Option names may be terminated early as long as they are uniquely
identified. For example, instead of --dump-token-types, it would be
sufficient to enter --dump-tok, or even --dump-t, to uniquely
identify this command.

I/O Control
The following parameters concern the files which are read and
written.

-h, --help
Show summary of usage and exit.

-v, --version
Show perltidy VERSION number and exit.

-o=filename, --outfile=filename
Name of the output file (only if a single input file is being
processed). If no output file is specified, and output is not
redirected to the standard output (see -st), the output will go
to filename.tdy. [Note: - does not redirect to standard output.
Use -st instead.]

-st, --standard-output
Perltidy must be able to operate on an arbitrarily large number
of files in a single run, with each output being directed to a
different output file. Obviously this would conflict with
outputting to the single standard output device, so a special
flag, -st, is required to request outputting to the standard
output. For example,

perltidy somefile.pl -st >somefile.new.pl

This option may only be used if there is just a single input
file. The default is -nst or --nostandard-output.

-se, --standard-error-output
If perltidy detects an error when processing file somefile.pl,
its default behavior is to write error messages to file
somefile.pl.ERR. Use -se to cause all error messages to be sent
to the standard error output stream instead. This directive may
be negated with -nse. Thus, you may place -se in a .perltidyrc
and override it when desired with -nse on the command line.

-oext=ext, --output-file-extension=ext
Change the extension of the output file to be ext instead of the
default tdy (or html in case the --html option is used). See
"Specifying File Extensions".

-opath=path, --output-path=path
When perltidy creates a filename for an output file, by default
it merely appends an extension to the path and basename of the
input file. This parameter causes the path to be changed to path
instead.

The path should end in a valid path separator character, but
perltidy will try to add one if it is missing.

For example

perltidy somefile.pl -opath=/tmp/

will produce /tmp/somefile.pl.tdy. Otherwise, somefile.pl.tdy
will appear in whatever directory contains somefile.pl.

If the path contains spaces, it should be placed in quotes.

This parameter will be ignored if output is being directed to
standard output, or if it is being specified explicitly with the
--outfile=s parameter.

-b, --backup-and-modify-in-place
Modify the input file or files in-place and save the original
with the extension .bak. Any existing .bak file will be deleted.
See next item for changing the default backup extension, and for
eliminating the backup file altogether.

Please Note: Writing back to the input file increases the risk of
data loss or corruption in the event of a software or hardware
malfunction. Before using the -b parameter please be sure to have
backups and verify that it works correctly in your environment
and operating system.

A -b flag will be ignored if input is from standard input or goes
to standard output, or if the -html flag is set.

In particular, if you want to use both the -b flag and the
--perl-best-practices (-pbp) flag, then you must put a
--nostandard-output (-nst) flag after the -pbp flag because it
contains a --standard-output flag as one of its components, which
means that output will go to the standard output stream.

-bext=ext, --backup-file-extension=ext
This parameter serves two purposes: (1) to change the extension
of the backup file to be something other than the default .bak,
and (2) to indicate that no backup file should be saved.

To change the default extension to something other than .bak see
"Specifying File Extensions".

A backup file of the source is always written, but you can
request that it be deleted at the end of processing if there were
no errors. This is risky unless the source code is being
maintained with a source code control system.

To indicate that the backup should be deleted include one forward
slash, /, in the extension. If any text remains after the slash
is removed it will be used to define the backup file extension
(which is always created and only deleted if there were no
errors).

Here are some examples:

Parameter Extension Backup File Treatment
-bext=bak .bak Keep (same as default behavior)
-bext='/' .bak Delete if no errors
-bext='/backup' .backup Delete if no errors
-bext='original/' .original Delete if no errors

-bm=s, --backup-method=s
This parameter should not normally be used but is available in
the event that problems arise as a transition is made from an
older implementation of the backup logic to a newer
implementation. The newer implementation is the default and is
specified with -bm='copy'. The older implementation is specified
with -bm='move'. The difference is that the older implementation
made the backup by moving the input file to the backup file, and
the newer implementation makes the backup by copying the input
file. The newer implementation preserves the file system inode
value. This may avoid problems with other software running
simultaneously. This change was made as part of issue git #103
at github.

-w, --warning-output
Setting -w causes any non-critical warning messages to be
reported as errors. These include messages about possible pod
problems, possibly bad starting indentation level, and cautions
about indirect object usage. The default, -nw or
--nowarning-output, is not to include these warnings.

-q, --quiet
Deactivate error messages (for running under an editor).

For example, if you use a vi-style editor, such as vim, you may
execute perltidy as a filter from within the editor using
something like

:n1,n2!perltidy -q

where "n1,n2" represents the selected text. Without the -q flag,
any error message may mess up your screen, so be prepared to use
your "undo" key.

-log, --logfile
Save the .LOG file, which has many useful diagnostics. Perltidy
always creates a .LOG file, but by default it is deleted unless a
program bug is suspected. Setting the -log flag forces the log
file to be saved.

-g=n, --logfile-gap=n
Set maximum interval between input code lines in the logfile.
This purpose of this flag is to assist in debugging nesting
errors. The value of "n" is optional. If you set the flag -g
without the value of "n", it will be taken to be 1, meaning that
every line will be written to the log file. This can be helpful
if you are looking for a brace, paren, or bracket nesting error.

Setting -g also causes the logfile to be saved, so it is not
necessary to also include -log.

If no -g flag is given, a value of 50 will be used, meaning that
at least every 50th line will be recorded in the logfile. This
helps prevent excessively long log files.

Setting a negative value of "n" is the same as not setting -g at
all.

-npro --noprofile
Ignore any .perltidyrc command file. Normally, perltidy looks
first in your current directory for a .perltidyrc file of
parameters. (The format is described below). If it finds one,
it applies those options to the initial default values, and then
it applies any that have been defined on the command line. If no
.perltidyrc file is found, it looks for one in your home
directory.

If you set the -npro flag, perltidy will not look for this file.

-pro=filename or --profile=filename
To simplify testing and switching .perltidyrc files, this command
may be used to specify a configuration file which will override
the default name of .perltidyrc. There must not be a space on
either side of the '=' sign. For example, the line

perltidy -pro=testcfg

would cause file testcfg to be used instead of the default
.perltidyrc.

A pathname begins with three dots, e.g. ".../.perltidyrc",
indicates that the file should be searched for starting in the
current directory and working upwards. This makes it easier to
have multiple projects each with their own .perltidyrc in their
root directories.

-opt, --show-options
Write a list of all options used to the .LOG file. Please see
--dump-options for a simpler way to do this.

-f, --force-read-binary
Force perltidy to process binary files. To avoid producing
excessive error messages, perltidy skips files identified by the
system as non-text. However, valid perl scripts containing
binary data may sometimes be identified as non-text, and this
flag forces perltidy to process them.

-ast, --assert-tidy
This flag asserts that the input and output code streams are
identical, or in other words that the input code is already
'tidy' according to the formatting parameters. If this is not
the case, an error message noting this is produced. This error
message will cause the process to return a non-zero exit code.
The test for this is made by comparing an MD5 hash value for the
input and output code streams. This flag has no other effect on
the functioning of perltidy. This might be useful for certain
code maintenance operations. Note: you will not see this message
if you have error messages turned off with the -quiet flag.

-asu, --assert-untidy
This flag asserts that the input and output code streams are
different, or in other words that the input code is 'untidy'
according to the formatting parameters. If this is not the case,
an error message noting this is produced. This flag has no other
effect on the functioning of perltidy.

-tos=n, --timeout-in-seconds=n
When the standard input supplies the input stream, and the input
has not been received within n seconds, perltidy will end with a
timeout message. The intention is to catch a situation where
perltidy is accidentally invoked without a file to process and
therefore waits for input from the system standard input (stdin),
which never arrives. The default is n=10 seconds. This check
can be turned off with n=0.

FORMATTING OPTIONS

Basic Options

-fmt=s, --format=s
If --format=tidy (the default) then perltidy will reformat the
input file, and if --format=html then perltidy will produce html
output.

For convenience, the abbreviation --tidy is equivalent to
--format=tidy, and -html is equivalent to --format=html.

--notidy
This flag disables all formatting and causes the input to be
copied unchanged to the output except for possible changes in
line ending characters and any pre- and post-filters. This can
be useful in conjunction with a hierarchical set of .perltidyrc
files to avoid unwanted code tidying. See also "Skipping
Selected Sections of Code" for a way to avoid tidying specific
sections of code.

-i=n, --indent-columns=n
Use n columns per indentation level (default n=4).

-l=n, --maximum-line-length=n
The default maximum line length is n=80 characters. Perltidy
will try to find line break points to keep lines below this
length. However, long quotes and side comments may cause lines to
exceed this length. And long lines may sometimes be broken at a
length less than n characters because some of the line break
decisions employ small tolerances to prevent formatting
instabilities.

The default length of 80 comes from the past when this was the
standard CRT screen width. Many programmers prefer to increase
this to something like 120.

Setting -l=0 is equivalent to setting -l=(a very large number).
But this is not recommended because, for example, a very long
list will be formatted in a single long line.

-vmll, --variable-maximum-line-length
A problem arises using a fixed maximum line length with very
deeply nested code and data structures because eventually the
amount of leading whitespace used for indicating indentation
takes up most or all of the available line width, leaving little
or no space for the actual code or data. One solution is to use
a very long line length. Another solution is to use the -vmll
flag, which basically tells perltidy to ignore leading whitespace
when measuring the line length.

To be precise, when the -vmll parameter is set, the maximum line
length of a line of code will be M+L*I, where

M is the value of --maximum-line-length=M (-l=M), default 80,
I is the value of --indent-columns=I (-i=I), default 4,
L is the indentation level of the line of code

When this flag is set, the choice of breakpoints for a block of
code should be essentially independent of its nesting depth.
However, the absolute line lengths, including leading whitespace,
can still be arbitrarily large. This problem can be avoided by
including the next parameter.

The default is not to do this (-nvmll).

-wc=n, --whitespace-cycle=n
This flag also addresses problems with very deeply nested code
and data structures. When the nesting depth exceeds the value n
the leading whitespace will be reduced and start at a depth of 1
again. The result is that blocks of code will shift back to the
left rather than moving arbitrarily far to the right. This
occurs cyclically to any depth.

For example if one level of indentation equals 4 spaces (-i=4,
the default), and one uses -wc=15, then if the leading whitespace
on a line exceeds about 4*15=60 spaces it will be reduced back to
4*1=4 spaces and continue increasing from there. If the
whitespace never exceeds this limit the formatting remains
unchanged.

The combination of --variable-maximum-line-length and
--whitespace-cycle=n provides a solution to the problem of
displaying arbitrarily deep data structures and code in a finite
window, although --whitespace-cycle=n may of course be used
without --variable-maximum-line-length.

The default is not to use this, which can also be indicated using
-wc=0.

Tabs
Using tab characters will almost certainly lead to future
portability and maintenance problems, so the default and
recommendation is not to use them. For those who prefer tabs,
however, there are two different options.

Except for possibly introducing tab indentation characters, as
outlined below, perltidy does not introduce any tab characters
into your file, and it removes any tabs from the code (unless
requested not to do so with --freeze-whitespace). If you have
any tabs in your comments, quotes, or here-documents, they will
remain.

-et=n, --entab-leading-whitespace
This flag causes each n leading space characters produced by
the formatting process to be replaced by one tab character.
The formatting process itself works with space characters.
The -et=n parameter is applied as a last step, after
formatting is complete, to convert leading spaces into tabs.
Before starting to use tabs, it is essential to first get the
indentation controls set as desired without tabs,
particularly the two parameters --indent-columns=n (or -i=n)
and --continuation-indentation=n (or -ci=n).

The value of the integer n can be any value but can be
coordinated with the number of spaces used for indentation.
For example, -et=4 -ci=4 -i=4 will produce one tab for each
indentation level and and one for each continuation
indentation level. You may want to coordinate the value of n
with what your display software assumes for the spacing of a
tab.

The default is not to use this, which can also be indicated
using -et=0.

-t, --tabs
This flag causes one leading tab character to be inserted for
each level of indentation. Certain other features are
incompatible with this option, and if these options are also
given, then a warning message will be issued and this flag
will be unset. One example is the --line-up-parentheses
option. This flag is retained for backwards compatibility,
but if you use tabs, the --entab-leading-whitespace=n flag is
recommended. If both --tabs and --entab-leading-whitespace=n
are set, then --entab-leading-whitespace=n is used.

-dt=n, --default-tabsize=n
If the first line of code passed to perltidy contains leading
tabs but no tab scheme is specified for the output stream
then perltidy must guess how many spaces correspond to each
leading tab. This number of spaces n corresponding to each
leading tab of the input stream may be specified with -dt=n.
The default is n=8.

This flag has no effect if a tab scheme is specified for the
output stream, because then the input stream is assumed to
use the same tab scheme and indentation spaces as for the
output stream (any other assumption would lead to unstable
editing).

-io, --indent-only
This flag is used to deactivate all whitespace and line break
changes within non-blank lines of code. When it is in effect,
the only change to the script will be to the indentation and to
the number of blank lines. And any flags controlling whitespace
and newlines will be ignored. You might want to use this if you
are perfectly happy with your whitespace and line breaks, and
merely want perltidy to handle the indentation. (This also
speeds up perltidy by well over a factor of two, so it might be
useful when perltidy is merely being used to help find a brace
error in a large script).

Setting this flag is equivalent to setting --freeze-newlines and
--freeze-whitespace.

If you also want to keep your existing blank lines exactly as
they are, you can add --freeze-blank-lines.

With this option perltidy is still free to modify the indenting
(and outdenting) of code and comments as it normally would. If
you also want to prevent long comment lines from being outdented,
you can add either --nooutdent-long-lines (-noll) or
--maximum-line-length=0 (l=0).

Setting this flag will prevent perltidy from doing any special
operations on closing side comments. You may still delete all
side comments however when this flag is in effect.

-enc=s, --character-encoding=s
This flag indicates if the input data stream uses a character
encoding. Perltidy does not look for the encoding directives in
the source stream, such as use utf8, and instead relies on this
flag to determine the encoding. (This is because perltidy often
works on snippets of code rather than complete files, so it
cannot rely on use utf8 directives). Consequently perltidy is
likely to encounter problems formatting a file which is only
partially encoded.

The possible values for s are:

-enc=none if no encoding is used, or
-enc=utf8 for encoding in utf8
-enc=guess if perltidy should guess between these two possibilities.

The value none causes the stream to be processed without special
encoding assumptions. This is appropriate for files which are
written in single-byte character encodings such as latin-1.

The value utf8 causes the stream to be read and written as UTF-8.
If the input stream cannot be decoded with this encoding then
processing is not done.

The value guess tells perltidy to guess between either utf8
encoding or no encoding (meaning one character per byte). The
guess option uses the Encode::Guess module which has been found
to be reliable at detecting if a file is encoded in utf8 or not.

The current default is guess.

The abbreviations -utf8 or -UTF8 are equivalent to -enc=utf8, and
the abbreviation -guess is equivalent to -enc=guess. So to
process a file named file.pl which is encoded in UTF-8 you can
use:

perltidy -utf8 file.pl

or

perltidy -guess file.pl

or simply

perltidy file.pl

since -guess is the default.

To process files with an encoding other than UTF-8, it would be
necessary to write a short program which calls the Perl::Tidy
module with some pre- and post-processing to handle decoding and
encoding.

-eos=s, --encode-output-strings=s
This flag was added to resolve an issue involving the interface
between Perl::Tidy and calling programs, and in particular
Code::TidyAll (tidyall).

If you only run the perltidy binary this flag has no effect. If
you run a program which calls the Perl::Tidy module and receives
a string in return, then the meaning of the flag is as follows:

+o The setting -eos means Perl::Tidy should encode any string
which it decodes. This is the default because it makes
perltidy behave well as a filter, and is the correct setting
for most programs.

+o The setting -neos means that a string should remain decoded
if it was decoded by Perl::Tidy. This is only appropriate if
the calling program will handle any needed encoding before
outputting the string.

The default was changed from -neos to -eos in versions after
20220217. If this change causes a program to start running
incorrectly on encoded files, an emergency fix might be to set
-neos. Additional information can be found in the man pages for
the Perl::Tidy module and also in
<https://github.com/perltidy/perltidy/blob/master/docs/eos_flag.md>.

-gcs, --use-unicode-gcstring
This flag controls whether or not perltidy may use module
Unicode::GCString to obtain accurate display widths of wide
characters. The default is --nouse-unicode-gcstring.

If this flag is set, and text is encoded, perltidy will look for
the module Unicode::GCString and, if found, will use it to obtain
character display widths. This can improve displayed vertical
alignment for files with wide characters. It is a nice feature
but it is off by default to avoid conflicting formatting when
there are multiple developers. Perltidy installation does not
require Unicode::GCString, so users wanting to use this feature
need set this flag and also to install Unicode::GCString
separately.

If this flag is set and perltidy does not find module
Unicode::GCString, a warning message will be produced and
processing will continue but without the potential benefit
provided by the module.

Also note that actual vertical alignment depends upon the fonts
used by the text display software, so vertical alignment may not
be optimal even when Unicode::GCString is used.

-ole=s, --output-line-ending=s
where s="win", "dos", "unix", or "mac". This flag tells perltidy
to output line endings for a specific system. Normally, perltidy
writes files with the line separator character of the host
system. The "win" and "dos" flags have an identical result.

-ple, --preserve-line-endings
This flag tells perltidy to write its output files with the same
line endings as the input file, if possible. It should work for
dos, unix, and mac line endings. It will only work if perltidy
input comes from a filename (rather than stdin, for example). If
perltidy has trouble determining the input file line ending, it
will revert to the default behavior of using the line ending of
the host system.

-atnl, --add-terminal-newline
This flag, which is enabled by default, allows perltidy to
terminate the last line of the output stream with a newline
character, regardless of whether or not the input stream was
terminated with a newline character. If this flag is negated,
with -natnl, then perltidy will add a terminal newline to the the
output stream only if the input stream is terminated with a
newline.

Negating this flag may be useful for manipulating one-line
scripts intended for use on a command line.

-it=n, --iterations=n
This flag causes perltidy to do n complete iterations. The
reason for this flag is that code formatting is an iterative
process and in some cases the output from perltidy can be
different if it is applied a second time. For most purposes the
default of n=1 should be satisfactory. However n=2 can be useful
when a major style change is being made, or when code is being
beautified on check-in to a source code control system. It has
been found to be extremely rare for the output to change after 2
iterations. If a value n is greater than 2 is input then a
convergence test will be used to stop the iterations as soon as
possible, almost always after 2 iterations. See the next item
for a simplified iteration control.

This flag has no effect when perltidy is used to generate html.

-conv, --converge
This flag is equivalent to -it=4 and is included to simplify
iteration control. Perltidy will check for convergence and stop
iterating as soon as possible. The default is -nconv (no
convergence check). Using -conv will approximately double run
time since typically one extra iteration is required to verify
convergence. No extra iterations are required if no new line
breaks are made, and two extra iterations are occasionally needed
when reformatting complex code structures, such as deeply nested
ternary statements.

Code Indentation Control

-ci=n, --continuation-indentation=n
Continuation indentation is extra indentation spaces applied when
a long line is broken. The default is n=2, illustrated here:

my $level = # -ci=2
( $max_index_to_go >= 0 ) ? $levels_to_go[0] : $last_output_level;

The same example, with n=0, is a little harder to read:

my $level = # -ci=0
( $max_index_to_go >= 0 ) ? $levels_to_go[0] : $last_output_level;

The value given to -ci is also used by some commands when a small
space is required. Examples are commands for outdenting labels,
--outdent-labels (-ola), and control keywords, --outdent-keywords
(-okw).

When default values are not used, it is recommended that either

(1) the value n given with -ci=n be no more than about one-half
of the number of spaces assigned to a full indentation level on
the -i=n command, or

(2) the flag --extended-continuation-indentation is used (see
next section).

-xci, --extended-continuation-indentation
This flag allows perltidy to use some improvements which have
been made to its indentation model. One of the things it does is
"extend" continuation indentation deeper into structures, hence
the name. The improved indentation is particularly noticeable
when the flags -ci=n and -i=n use the same value of n. There are
no significant disadvantages to using this flag, but to avoid
disturbing existing formatting the default is not to use it,
-nxci.

Please see the section "-pbp, --perl-best-practices" for an
example of how this flag can improve the formatting of ternary
statements. It can also improve indentation of some multiline qw
lists as shown below.

# perltidy
foreach $color (
qw(
AntiqueWhite3 Bisque1 Bisque2 Bisque3 Bisque4
SlateBlue3 RoyalBlue1 SteelBlue2 DeepSkyBlue3
),
qw(
LightBlue1 DarkSlateGray1 Aquamarine2 DarkSeaGreen2
SeaGreen1 Yellow1 IndianRed1 IndianRed2 Tan1 Tan4
)
)

# perltidy -xci
foreach $color (
qw(
AntiqueWhite3 Bisque1 Bisque2 Bisque3 Bisque4
SlateBlue3 RoyalBlue1 SteelBlue2 DeepSkyBlue3
),
qw(
LightBlue1 DarkSlateGray1 Aquamarine2 DarkSeaGreen2
SeaGreen1 Yellow1 IndianRed1 IndianRed2 Tan1 Tan4
)
)

-mci, --minimize-continuation-indentation
This flag allows perltidy to remove continuation indentation in
some special cases where it is not really unnecessary. For a
simple example, the default formatting for the following snippet
is:

# perltidy -nmci
$self->blurt( "Error: No INPUT for type '$type', typekind '"
. $type->xstype
. "' found" );

The second and third lines are one level deep in a container, and
are also statement continuations, so they get indented by the sum
of the full indentation -i value and the continuation indentation
-ci value. If this flag is set, the indentation is reduced by
-ci spaces, giving

# perltidy -mci
$self->blurt( "Error: No INPUT for type '$type', typekind '"
. $type->xstype
. "' found" );

This flag is off by default.

-sil=n --starting-indentation-level=n
By default, perltidy examines the input file and tries to
determine the starting indentation level. While it is often
zero, it may not be zero for a code snippet being sent from an
editing session.

To guess the starting indentation level perltidy simply assumes
that indentation scheme used to create the code snippet is the
same as is being used for the current perltidy process. This is
the only sensible guess that can be made. It should be correct
if this is true, but otherwise it probably won't. For example,
if the input script was written with -i=2 and the current
perltidy flags have -i=4, the wrong initial indentation will be
guessed for a code snippet which has non-zero initial
indentation. Likewise, if an entabbing scheme is used in the
input script and not in the current process then the guessed
indentation will be wrong.

If the default method does not work correctly, or you want to
change the starting level, use -sil=n, to force the starting
level to be n.

List indentation using --line-up-parentheses, -lp or
--extended--line-up-parentheses , -xlp
These flags provide an alternative indentation method for list
data. The original flag for this is -lp, but it has some
limitations (explained below) which are avoided with the newer
-xlp flag. So -xlp is probably the better choice for new work,
but the -lp flag is retained to minimize changes to existing
formatting. If you enter both -lp and -xlp, then -xlp will be
used.

In the default indentation method perltidy indents lists with 4
spaces, or whatever value is specified with -i=n. Here is a
small list formatted in this way:

# perltidy (default)
@month_of_year = (
'Jan', 'Feb', 'Mar', 'Apr', 'May', 'Jun',
'Jul', 'Aug', 'Sep', 'Oct', 'Nov', 'Dec'
);

The -lp or -xlp flags add extra indentation to cause the data to
begin past the opening parentheses of a sub call or list, or
opening square bracket of an anonymous array, or opening curly
brace of an anonymous hash. With this option, the above list
would become:

# perltidy -lp or -xlp
@month_of_year = (
'Jan', 'Feb', 'Mar', 'Apr', 'May', 'Jun',
'Jul', 'Aug', 'Sep', 'Oct', 'Nov', 'Dec'
);

If the available line length (see -l=n ) does not permit this
much space, perltidy will use less. For alternate placement of
the closing paren, see the next section.

These flags have no effect on code BLOCKS, such as if/then/else
blocks, which always use whatever is specified with -i=n.

Some limitations on these flags are:

+o A limitation on -lp, but not -xlp, occurs in situations where
perltidy does not have complete freedom to choose line
breaks. Then it may temporarily revert to its default
indentation method. This can occur for example if there are
blank lines, block comments, multiline quotes, or side
comments between the opening and closing parens, braces, or
brackets. It will also occur if a multiline anonymous sub
occurs within a container since that will impose specific
line breaks (such as line breaks after statements).

+o For both the -lp and -xlp flags, any parameter which
significantly restricts the ability of perltidy to choose
newlines will conflict with these flags and will cause them
to be deactivated. These include --indent-only,
--freeze-newlines, --noadd-newlines, and
--nodelete-old-newlines.

+o The -lp and -xlp options may not be used together with the -t
tabs option. They may, however, be used with the -et=n tab
method

There are some potential disadvantages of this indentation method
compared to the default method that should be noted:

+o The available line length can quickly be used up if variable
names are long. This can cause deeply nested code to quickly
reach the line length limit, and become badly formatted, much
sooner than would occur with the default indentation method.

+o Since the indentation depends on the lengths of variable
names, small changes in variable names can cause changes in
indentation over many lines in a file. This means that minor
name changes can produce significant file differences. This
can be annoying and does not occur with the default
indentation method.

Some things that can be done to minimize these problems are:

+o Increase --maximum-line-length=n above the default n=80
characters if necessary.

+o If you use -xlp then long side comments can limit the
indentation over multiple lines. Consider adding the flag
--ignore-side-comment-lengths to prevent this, or minimizing
the use of side comments.

+o Apply this style in a limited way. By default, it applies to
all list containers (not just lists in parentheses). The
next section describes how to limit this style to, for
example, just function calls. The default indentation method
will be applied elsewhere.

-lpil=s, --line-up-parentheses-inclusion-list and -lpxl=s,
--line-up-parentheses-exclusion-list
The following discussion is written for -lp but applies equally
to the newer -xlp version. By default, the -lp flag applies to
as many containers as possible. The set of containers to which
the -lp style applies can be reduced by either one of these two
flags:

Use --line-up-parentheses-inclusion-list=s (-lpil=s) to specify
the containers to which -lp applies, or

use --line-up-parentheses-exclusion-list=s (-lpxl=s) to specify
the containers to which -lp does NOT apply.

Only one of these two flags may be used. Both flags can achieve
the same result, but the -lpil=s flag is much easier to describe
and use and is recommended. The -lpxl=s flag was the original
implementation and is only retained for backwards compatibility.

This list s for these parameters is a string with space-separated
items. Each item consists of up to three pieces of information
in this order: (1) an optional letter code (2) a required
container type, and (3) an optional numeric code.

The only required piece of information is a container type, which
is one of '(', '[', or '{'. For example the string

-lpil='('

means use -lp formatting only on lists within parentheses, not
lists in square-brackets or braces. The same thing could
alternatively be specified with

-lpxl = '[ {'

which says to exclude lists within square-brackets and braces.
So what remains is lists within parentheses.

A second optional item of information which can be given for
parentheses is an alphanumeric letter which is used to limit the
selection further depending on the type of token immediately
before the paren. The possible letters are currently 'k', 'K',
'f', 'F', 'w', and 'W', with these meanings for matching whatever
precedes an opening paren:

'k' matches if the previous nonblank token is a perl keyword
(such as 'if', 'while'),
'K' matches if 'k' does not: previous token is not a keyword
'f' matches if previous token is a function (not a keyword)
'F' matches if 'f' does not.
'w' matches if either 'k' or 'f' match.
'W' matches if 'w' does not.

For example:

-lpil = 'f('

means only apply -lp to function calls, and

-lpil = 'w('

means only apply -lp to parenthesized lists which follow a
function or a keyword.

This last example could alternatively be written using the
-lpxl=s flag as

-lpxl = '[ { W('

which says exclude -lp for lists within square-brackets, braces,
and parens NOT preceded by a keyword or function. Clearly, the
-lpil=s method is easier to understand.

An optional numeric code may follow any of the container types to
further refine the selection based on container contents. The
numeric codes are:

'0' or blank: no restriction is placed on container contents
'1' the container contents must be a simple list without sublists
'2' the container contents must be a simple list without sublists,
without code blocks, and without ternary operators

For example,

-lpil = 'f(2'

means only apply -lp to function calls with simple lists (not
containing any sublists, code blocks or ternary expressions).

-cti=n, --closing-token-indentation
The -cti=n flag controls the indentation of a line beginning with
a ")", "]", or a non-block "}". Such a line receives:

-cti = 0 no extra indentation (default)
-cti = 1 extra indentation such that the closing token
aligns with its opening token.
-cti = 2 one extra indentation level if the line looks like:
); or ]; or };
-cti = 3 one extra indentation level always

The flags -cti=1 and -cti=2 work well with the
--line-up-parentheses (-lp) flag (previous section).

# perltidy -lp -cti=1
@month_of_year = (
'Jan', 'Feb', 'Mar', 'Apr', 'May', 'Jun',
'Jul', 'Aug', 'Sep', 'Oct', 'Nov', 'Dec'
);

# perltidy -lp -cti=2
@month_of_year = (
'Jan', 'Feb', 'Mar', 'Apr', 'May', 'Jun',
'Jul', 'Aug', 'Sep', 'Oct', 'Nov', 'Dec'
);

These flags are merely hints to the formatter and they may not
always be followed. In particular, if -lp is not being used, the
indentation for cti=1 is constrained to be no more than one
indentation level.

If desired, this control can be applied independently to each of
the closing container token types. In fact, -cti=n is merely an
abbreviation for -cpi=n -csbi=n -cbi=n, where: -cpi or
--closing-paren-indentation controls )'s, -csbi or
--closing-square-bracket-indentation controls ]'s, -cbi or
--closing-brace-indentation controls non-block }'s.

-icp, --indent-closing-paren
The -icp flag is equivalent to -cti=2, described in the previous
section. The -nicp flag is equivalent -cti=0. They are included
for backwards compatibility.

-icb, --indent-closing-brace
The -icb option gives one extra level of indentation to a brace
which terminates a code block . For example,

if ($task) {
yyy();
} # -icb
else {
zzz();
}

The default is not to do this, indicated by -nicb.

-ils, --indent-leading-semicolon
A line which begins with a leading semicolon will, by default,
have the extra number of indentation spaces defined by
--continuation-indentation=n. This extra indentation can be
removed by setting -nils.

# default
$z = sqrt( $x**2 + $y**2 )

; # <-- indented by ci spaces

# -nils
$z = sqrt( $x**2 + $y**2 )

; # <-- not indented by ci spaces

Note that leading semicolons do not normally occur unless
requested with --break-at-old-semicolon-breakpoints or forced,
for example by a blank line as in this example.

-nib, --non-indenting-braces
Normally, lines of code contained within a pair of block braces
receive one additional level of indentation. This flag, which is
enabled by default, causes perltidy to look for opening block
braces which are followed by a special side comment. This special
side comment is #<<< by default. If found, the code between this
opening brace and its corresponding closing brace will not be
given the normal extra indentation level. For example:

{ #<<< a closure to contain lexical vars

my $var; # this line does not get one level of indentation
...

}

# this line does not 'see' $var;

This can be useful, for example, when combining code from
different files. Different sections of code can be placed within
braces to keep their lexical variables from being visible to the
end of the file. To keep the new braces from causing all of
their contained code to be indented if you run perltidy, and
possibly introducing new line breaks in long lines, you can mark
the opening braces with this special side comment.

Only the opening brace needs to be marked, since perltidy knows
where the closing brace is. Braces contained within marked
braces may also be marked as non-indenting.

If your code happens to have some opening braces followed by
'#<<<', and you don't want this behavior, you can use -nnib to
deactivate it. To make it easy to remember, the default string
is the same as the string for starting a format-skipping section.
There is no confusion because in that case it is for a block
comment rather than a side-comment.

The special side comment can be changed with the next parameter.

-nibp=s, --non-indenting-brace-prefix=s
The -nibp=string parameter may be used to change the marker for
non-indenting braces. The default is equivalent to -nibp='#<<<'.
The string that you enter must begin with a # and should be in
quotes as necessary to get past the command shell of your system.
This string is the leading text of a regex pattern that is
constructed by prepending a '^' and appending a'\s', so you must
also include backslashes for characters to be taken literally
rather than as patterns.

For example, to match the side comment '#++', the parameter would
be

-nibp='#\+\+'

-olq, --outdent-long-quotes
When -olq is set, lines which is a quoted string longer than the
value maximum-line-length will have their indentation removed to
make them more readable. This is the default. To prevent such
out-denting, use -nolq.

-oll, --outdent-long-lines
This command is equivalent to the combination
--outdent-long-quotes and --outdent-long-comments, and it is
included for compatibility with previous versions of perltidy.
The negation of this also works, -noll or --nooutdent-long-lines,
and is equivalent to setting -nolq and -nolc.

Outdenting Labels: -ola, --outdent-labels
This command will cause labels to be outdented by the number of
spaces defined by --continuation-indentation=n, if possible.
This is the default. For example:

my $i;
LOOP: while ( $i = <FOTOS> ) {
chomp($i);
next unless $i;
fixit($i);
}

Use -nola to prevent this. To control line breaks after labels
see "-bal=n, --break-after-labels=n".

Outdenting Keywords
-okw, --outdent-keywords
The command -okw will cause certain leading control keywords
to be outdented by the number of spaces defined by
--continuation-indentation=nspaces, if possible. By default,
these keywords are "redo", "next", "last", "goto", and
"return". The intention is to make these control keywords
easier to see. To change this list of keywords being
outdented, see the next section.

For example, using "perltidy -okw" on the previous example
gives:

my $i;
LOOP: while ( $i = <FOTOS> ) {
chomp($i);
next unless $i;
fixit($i);
}

Notice that the keyword next has been outdented. The default
is not to do this.

Specifying Outdented Keywords: -okwl=string,
--outdent-keyword-list=string
This command can be used to change the keywords which are
outdented with the -okw command. The parameter string is a
required list of perl keywords, which should be placed in
quotes if there are more than one. By itself, it does not
cause any outdenting to occur, so the -okw command is still
required.

For example, the commands "-okwl="next last redo goto" -okw"
will cause those four keywords to be outdented.

Whitespace Control

Whitespace refers to the blank space between variables, operators,
and other code tokens.

-fws, --freeze-whitespace
This flag causes your original whitespace to remain unchanged,
and causes the rest of the whitespace commands in this section,
the Code Indentation section, and the Comment Control section to
be ignored.

Tightness of curly braces, parentheses, and square brackets
Here the term "tightness" will mean the closeness with which
pairs of enclosing tokens, such as parentheses, contain the
quantities within. A numerical value of 0, 1, or 2 defines the
tightness, with 0 being least tight and 2 being most tight.
Spaces within containers are always symmetric, so if there is a
space after a "(" then there will be a space before the
corresponding ")".

The -pt=n or --paren-tightness=n parameter controls the space
within parens. The example below shows the effect of the three
possible values, 0, 1, and 2:

if ( ( my $len_tab = length( $tabstr ) ) > 0 ) { # -pt=0
if ( ( my $len_tab = length($tabstr) ) > 0 ) { # -pt=1 (default)
if ((my $len_tab = length($tabstr)) > 0) { # -pt=2

When n is 0, there is always a space to the right of a '(' and to
the left of a ')'. For n=2 there is never a space. For n=1, the
default, there is a space unless the quantity within the parens
is a single token, such as an identifier or quoted string.

Likewise, the parameter -sbt=n or --square-bracket-tightness=n
controls the space within square brackets, as illustrated below.

$width = $col[ $j + $k ] - $col[ $j ]; # -sbt=0
$width = $col[ $j + $k ] - $col[$j]; # -sbt=1 (default)
$width = $col[$j + $k] - $col[$j]; # -sbt=2

Curly braces which do not contain code blocks are controlled by
the parameter -bt=n or --brace-tightness=n.

$obj->{ $parsed_sql->{ 'table' }[0] }; # -bt=0
$obj->{ $parsed_sql->{'table'}[0] }; # -bt=1 (default)
$obj->{$parsed_sql->{'table'}[0]}; # -bt=2

And finally, curly braces which contain blocks of code are
controlled by the parameter -bbt=n or --block-brace-tightness=n
as illustrated in the example below (-bbt=0 is the default).

%bf = map { $_ => -M $_ } grep { /\.deb$/ } dirents '.'; # -bbt=0
%bf = map { $_ => -M $_ } grep {/\.deb$/} dirents '.'; # -bbt=1
%bf = map {$_ => -M $_} grep {/\.deb$/} dirents '.'; # -bbt=2

To simplify input in the case that all of the tightness flags
have the same value n, the parameter -act=n or
--all-containers-tightness=n is an abbreviation for the
combination -pt=n -sbt=n -bt=n -bbt=n.

-mutt=s, --multiple-token-tightness=s
To review, the tightness controls described in the previous
section have three possible integer values: 0, 1, and 2, where
n=0 always adds a space, and n=2 never adds a space.

The default value n=1 adds space only if the container holds
multiple tokens. Some perltidy tokens may be rather long, and it
can be preferable to space some of them as if they were multple
tokens. This can be done with this paramter.

For example, in the following expression the "qw" list is a
single token and therefore there the default formatting does not
put spaces within the square brackets:

my $rlist = [qw(alpha beta gamma)];

This can be changed with

# perltidy -mutt='qw'
my $rlist = [ qw(alpha beta gamma) ];

This tells perltidy to space a "qw"list as if it were multiple
tokens when the default tightness is used.

The parameter s may contain any of the following perl operators:

qw q qq qx qr s y tr m

Other possible specifications are

q* - all of the above operators
Q - all of the above plus any quoted text
h - a here-doc target, such as '<<EOT'
<<>> - the double-diamond operator

A symbol may be negated by preceding it with a carat ^ symbol.
The double-diamond operator is always included unless negated in
this way.

-xbt, --extended-block-tightness
There are two controls for spacing within curly braces, namely
--block-brace-tightness=n for code block braces and
--brace-tightness=n for all other braces.

There is a little fuzziness in this division of brace types
though because the curly braces considered by perltidy to contain
code blocks for formatting purposes, such as highlighting code
structure, exclude some of the small code blocks used by Perl
mainly for isolating terms. These include curly braces following
a keyword where an indirect object might occur, or curly braces
following a type symbol. For example, perltidy does not mark the
following braces as code block braces:

print {*STDERR} $message;
return ${$foo};

Consequently, the spacing within these small braced containers by
default follows the flag --brace-tightness=n rather than
--block-brace-tightness=n, as one might expect.

If desired, small blocks such as these can be made to instead
follow the spacing defined by the --block-brace-tightness=n flag
by setting --extended-block-tightness. The specific types of
small blocks to which this parameter applies is controlled by a
companion control parameter, described in the next section.

Note that if the two flags --block-brace-tightness=n and
--brace-tightness=n have the same value n then there would be no
reason to set this flag.

-xbtl=s, --extended-block-tightness-list=s
The previous parameter --extended-block-tightness (-xbt) can be
made to apply to curly braces preceded by any of the keywords

print printf exec system say

and/or the special symbols

$ @ % & * $#

The parameter string s may contain a selection of these keywords
and symbols to indicate the brace types to which
--extended-block-tightness applies. For convenience, all of the
keywords can be selected with 'k', and all of the special symbols
can be selected with 't'. The default is equivalent to
-xbtl='k', which selects all of the keywords.

Examples:

-xbtl='k' # selects just the keywords [DEFAULT]
-xbtl="t" # selects just the special type symbols
-xbtl="k t" # selects all keywords and symbols, or simply
-xbtl="kt" # selects all keywords and symbols
-xbtl="print say" # selects just keywords B<print> and B<say>:

Here are some formatting examples using the default values of
--brace-tightness=n and --block-brace-tightness=n. Note that in
these examples $ref is in block braces but $key is not.

# default formatting
print {*STDERR} $message;
my $val = ${$ref}{$key};

# perltidy -xbt or
# perltidy -xbt -xbtl=k
print { *STDERR } $message;
my $val = ${$ref}{$key};

# perltidy -xbt -xbtl=t
print {*STDERR} $message;
my $val = ${ $ref }{$key};

# perltidy -xbt -xbtl=kt
print { *STDERR } $message;
my $val = ${ $ref }{$key};

Finally, note that this parameter merely changes the way that the
parameter --extended-block-tightness works. It has no effect
unless --extended-block-tightness is actually set.

-tso, --tight-secret-operators
The flag -tso causes certain perl token sequences (secret
operators) which might be considered to be a single operator to
be formatted "tightly" (without spaces). The operators currently
modified by this flag are:

0+ +0 ()x!! ~~<> ,=> =( )=

For example the sequence 0 +, which converts a string to a
number, would be formatted without a space: 0+ when the -tso flag
is set. This flag is off by default.

-sts, --space-terminal-semicolon
Some programmers prefer a space before all terminal semicolons.
The default is for no such space, and is indicated with -nsts or
--nospace-terminal-semicolon.

$i = 1 ; # -sts
$i = 1; # -nsts (default)

-sfs, --space-for-semicolon
Semicolons within for loops may sometimes be hard to see,
particularly when commas are also present. This option places
spaces on both sides of these special semicolons, and is the
default. Use -nsfs or --nospace-for-semicolon to deactivate it.

for ( @a = @$ap, $u = shift @a ; @a ; $u = $v ) { # -sfs (default)
for ( @a = @$ap, $u = shift @a; @a; $u = $v ) { # -nsfs

-asc, --add-semicolons
Setting -asc allows perltidy to add any missing optional
semicolon at the end of a line which is followed by a closing
curly brace on the next line. This is the default, and may be
deactivated with -nasc or --noadd-semicolons.

-dsm, --delete-semicolons
Setting -dsm allows perltidy to delete extra semicolons which are
simply empty statements. This is the default, and may be
deactivated with -ndsm or --nodelete-semicolons. (Such
semicolons are not deleted, however, if they would promote a side
comment to a block comment).

-aws, --add-whitespace
Setting this option allows perltidy to add certain whitespace to
improve code readability. This is the default. If you do not
want any whitespace added, but are willing to have some
whitespace deleted, use -naws. (Use --freeze-whitespace (-fws)
to leave whitespace completely unchanged).

-dws, --delete-old-whitespace
Setting this option allows perltidy to remove optional whitespace
between characters in the input file. The default is to not to
do this (-nodelete-old-whitespace). This parameter has little
effect by itself. But in combination with --noadd-whitespace it
will cause most of the whitespace in a file to be removed.

Detailed whitespace controls around tokens
For those who want more detailed control over the whitespace
around tokens, there are four parameters which can directly
modify the default whitespace rules built into perltidy for any
token. They are:

-wls=s or --want-left-space=s,

-nwls=s or --nowant-left-space=s,

-wrs=s or --want-right-space=s,

-nwrs=s or --nowant-right-space=s.

These parameters are each followed by a quoted string, s,
containing a list of token types. No more than one of each of
these parameters should be specified, because repeating a
command-line parameter always overwrites the previous one before
perltidy ever sees it.

To illustrate how these are used, suppose it is desired that
there be no space on either side of the token types = + - / *.
The following two parameters would specify this desire:

-nwls="= + - / *" -nwrs="= + - / *"

(Note that the token types are in quotes, and that they are
separated by spaces). With these modified whitespace rules, the
following line of math:

$root = -$b + sqrt( $b * $b - 4. * $a * $c ) / ( 2. * $a );

becomes this:

$root=-$b+sqrt( $b*$b-4.*$a*$c )/( 2.*$a );

These parameters should be considered to be hints to perltidy
rather than fixed rules, because perltidy must try to resolve
conflicts that arise between them and all of the other rules that
it uses. One conflict that can arise is if, between two tokens,
the left token wants a space and the right one doesn't. In this
case, the token not wanting a space takes priority.

It is necessary to have a list of all token types in order to
create this type of input. Such a list can be obtained by the
command --dump-token-types. Also try the -D flag on a short
snippet of code and look at the .DEBUG file to see the
tokenization.

To illustrate, suppose we do not want a space after a colon which
introduces a sub attribute. We need to know its type. It is not a
colon because that is the type of a ternary operator. The output
of --dump-token-types states that it is token type 'A'. To
verify this, we can run "perltidy -D" on a short piece of code
containing such a colon, such as

sub foo : lvalue;

This produces a .DEBUG file which contains

1: sub foo : lvalue;
1: SSSSSSSbAbwwwwww;

The top line is the input, and the bottom line shows the token
types. The 'A' beneath the colon verifies that it is type A.

So to prevent a space on the right of this colon we can use

# perltidy -nwrs='A'
sub foo :lvalue;

WARNING Be sure to put these tokens in quotes to avoid having
them misinterpreted by your command shell.

Note1: Perltidy does always follow whitespace controls
The various parameters controlling whitespace within a program
are requests which perltidy follows as well as possible, but
there are a number of situations where changing whitespace could
change program behavior and is not done. Some of these are
obvious; for example, we should not remove the space between the
two plus symbols in '$x+ +$y' to avoid creating a '++' operator.
Some are more subtle and involve the whitespace around bareword
symbols and locations of possible filehandles. For example,
consider the problem of formatting the following subroutine:

sub print_div {
my ($x,$y)=@_;
print $x/$y;
}

Suppose the user requests that / signs have a space to the left
but not to the right. Perltidy will refuse to do this, but if
this were done the result would be

sub print_div {
my ($x,$y)=@_;
print $x /$y;
}

If formatted in this way, the program will not run (at least with
recent versions of perl) because the $x is taken to be a
filehandle and / is assumed to start a quote. In a complex
program, there might happen to be a / which terminates the
multiline quote without a syntax error, allowing the program to
run, but not as intended.

Related issues arise with other binary operator symbols, such as
+ and -, and in older versions of perl there could be problems
with ternary operators. So to avoid changing program behavior,
perltidy has the simple rule that whitespace around possible
filehandles is left unchanged. Likewise, whitespace around
unknown barewords is left unchanged. The reason is that if the
barewords are defined in other modules, or in code that has not
even been written yet, perltidy will not have seen their
prototypes and must treat them cautiously.

In perltidy this is implemented in the tokenizer by marking token
following a print keyword as a special type Z. When formatting
is being done, whitespace following this token type is generally
left unchanged as a precaution against changing program behavior.
This is excessively conservative but simple and easy to
implement. Keywords which are treated similarly to print include
printf, sort, exec, system. Changes in spacing around parameters
following these keywords may have to be made manually. For
example, the space, or lack of space, after the parameter $foo in
the following line will be unchanged in formatting.

system($foo );
system($foo);

To find if a token is of type Z you can use perltidy -DEBUG. For
the first line above the result is

1: system($foo );
1: kkkkkk{ZZZZb};

which shows that system is type k (keyword) and $foo is type Z.

Note2: Perltidy's whitespace rules are not perfect
Despite these precautions, it is still possible to introduce
syntax errors with some asymmetric whitespace rules, particularly
when call parameters are not placed in containing parens or
braces. For example, the following two lines will be parsed by
perl without a syntax error:

# original programming, syntax ok
my @newkeys = map $_-$nrecs+@data, @oldkeys;

# perltidy default, syntax ok
my @newkeys = map $_ - $nrecs + @data, @oldkeys;

But the following will give a syntax error:

# perltidy -nwrs='-'
my @newkeys = map $_ -$nrecs + @data, @oldkeys;

For another example, the following two lines will be parsed
without syntax error:

# original programming, syntax ok
for my $severity ( reverse $LOWEST+1 .. $HIGHEST ) { ... }

# perltidy default, syntax ok
for my $severity ( reverse $LOWEST + 1 .. $HIGHEST ) { ... }

But the following will give a syntax error:

# perltidy -nwrs='+', syntax error:
for my $severity ( reverse $LOWEST +1 .. $HIGHEST ) { ... }

To avoid subtle parsing problems like this, it is best to avoid
spacing a binary operator asymmetrically with a space on the left
but not on the right.

Space between specific keywords and opening paren
When an opening paren follows a Perl keyword, no space is
introduced after the keyword, unless it is (by default) one of
these:

my local our state and or xor err eq ne if else elsif until unless
while for foreach return switch case given when catch

These defaults can be modified with two commands:

-sak=s or --space-after-keyword=s adds keywords.

-nsak=s or --nospace-after-keyword=s removes keywords.

where s is a list of keywords (in quotes if necessary). For
example,

my ( $a, $b, $c ) = @_; # default
my( $a, $b, $c ) = @_; # -nsak="my local our"

The abbreviation -nsak='*' is equivalent to including all of the
keywords in the above list.

When both -nsak=s and -sak=s commands are included, the -nsak=s
command is executed first. For example, to have space after only
the keywords (my, local, our) you could use -nsak="*" -sak="my
local our".

To put a space after all keywords, see the next item.

Space between all keywords and opening parens
When an opening paren follows a function or keyword, no space is
introduced after the keyword except for the keywords noted in the
previous item. To always put a space between a function or
keyword and its opening paren, use the command:

-skp or --space-keyword-paren

You may also want to use the flag -sfp (next item) too.

Space between all function names and opening parens
When an opening paren follows a function the default and
recommended formatting is not to introduce a space. To cause a
space to be introduced use:

-sfp or --space-function-paren

myfunc( $a, $b, $c ); # default
myfunc ( $a, $b, $c ); # -sfp

You will probably also want to use the flag -skp (previous item)
too.

The parameter is not recommended because spacing a function paren
can make a program vulnerable to parsing problems by Perl. For
example, the following two-line program will run as written but
will have a syntax error if reformatted with -sfp:

if ( -e filename() ) { print "I'm here\n"; }
sub filename { return $0 }

In this particular case the syntax error can be removed if the
line order is reversed, so that Perl parses 'sub filename' first.

-fpva or --function-paren-vertical-alignment
A side-effect of using the -sfp flag is that the parens may
become vertically aligned. For example,

# perltidy -sfp
myfun ( $aaa, $b, $cc );
mylongfun ( $a, $b, $c );

This is the default behavior. To prevent this alignment use
-nfpva:

# perltidy -sfp -nfpva
myfun ( $aaa, $b, $cc );
mylongfun ( $a, $b, $c );

-spp=n or --space-prototype-paren=n
This flag can be used to control whether a function prototype is
preceded by a space. For example, the following prototype does
not have a space.

sub usage();

This integer n may have the value 0, 1, or 2 as follows:

-spp=0 means no space before the paren
-spp=1 means follow the example of the source code [DEFAULT]
-spp=2 means always put a space before the paren

The default is -spp=1, meaning that a space will be used if and
only if there is one in the source code. Given the above line of
code, the result of applying the different options would be:

sub usage(); # n=0 [no space]
sub usage(); # n=1 [default; follows input]
sub usage (); # n=2 [space]

-ssp=n or --space-signature-paren=n
This flag is analogous to the previous except that it applies to
the space before the opening paren of a sub signature rather than
a sub prototype.

For example, consider the following line:

sub circle( $xc, $yc, $rad )

This space before the opening paren can be controlled with
integer n which may have the value 0, 1, or 2 with these
meanings:

-ssp=0 means no space before the paren
-ssp=1 means follow the example of the source code [DEFAULT]
-ssp=2 means always put a space before the paren

The default is -ssp=1, meaning that will be a space in the output
if, and only if, there is one in the input. Given the above line
of code, the result of applying the different options would be:

sub circle( $xc, $yc, $rad ) # n=0 [no space]
sub circle( $xc, $yc, $rad ) # n=1 [default; same as input]
sub circle ( $xc, $yc, $rad ) # n=2 [space]

-kpit=n or --keyword-paren-inner-tightness=n
The space inside of an opening paren, which itself follows a
certain keyword, can be controlled by this parameter. The space
on the inside of the corresponding closing paren will be treated
in the same (balanced) manner. This parameter has precedence
over any other paren spacing rules. The values of n are as
follows:

-kpit=0 means always put a space (not tight)
-kpit=1 means ignore this parameter [default]
-kpit=2 means never put a space (tight)

To illustrate, the following snippet is shown formatted in three
ways:

if ( seek( DATA, 0, 0 ) ) { ... } # perltidy (default)
if (seek(DATA, 0, 0)) { ... } # perltidy -pt=2
if ( seek(DATA, 0, 0) ) { ... } # perltidy -pt=2 -kpit=0

In the second case the -pt=2 parameter makes all of the parens
tight. In the third case the -kpit=0 flag causes the space within
the 'if' parens to have a space, since 'if' is one of the
keywords to which the -kpit flag applies by default. The
remaining parens are still tight because of the -pt=2 parameter.

The set of keywords to which this parameter applies are by
default are:

if elsif unless while until for foreach

These can be changed with the parameter -kpitl=s described in the
next section.

-kpitl=string or --keyword-paren-inner-tightness-list=string
This command can be used to change the keywords to which the
previous parameter, -kpit=n, applies. The parameter string is a
required list either keywords or functions, which should be
placed in quotes if there are more than one. By itself, this
parameter does not cause any change in spacing, so the -kpit=n
command is still required.

For example, the commands "-kpitl="if else while" -kpit=2" will
cause the just the spaces inside parens following 'if', 'else',
and 'while' keywords to follow the tightness value indicated by
the -kpit=2 flag.

-lop or --logical-padding
In the following example some extra space has been inserted on
the second line between the two open parens. This extra space is
called "logical padding" and is intended to help align similar
things vertically in some logical or ternary expressions.

# perltidy [default formatting]
$same =
( ( $aP eq $bP )
&& ( $aS eq $bS )
&& ( $aT eq $bT )
&& ( $a->{'title'} eq $b->{'title'} )
&& ( $a->{'href'} eq $b->{'href'} ) );

Note that this is considered to be a different operation from
"vertical alignment" because space at just one line is being
adjusted, whereas in "vertical alignment" the spaces at all lines
are being adjusted. So it is sort of a local version of vertical
alignment.

Here is an example involving a ternary operator:

# perltidy [default formatting]
$bits =
$top > 0xffff ? 32
: $top > 0xff ? 16
: $top > 1 ? 8
: 1;

This behavior is controlled with the flag --logical-padding,
which is set 'on' by default. If it is not desired it can be
turned off using --nological-padding or -nlop. The above two
examples become, with -nlop:

# perltidy -nlop
$same =
( ( $aP eq $bP )
&& ( $aS eq $bS )
&& ( $aT eq $bT )
&& ( $a->{'title'} eq $b->{'title'} )
&& ( $a->{'href'} eq $b->{'href'} ) );

# perltidy -nlop
$bits =
$top > 0xffff ? 32
: $top > 0xff ? 16
: $top > 1 ? 8
: 1;

Trimming whitespace around "qw" quotes
-tqw or --trim-qw provide the default behavior of trimming spaces
around multiline "qw" quotes and indenting them appropriately.

-ntqw or --notrim-qw cause leading and trailing whitespace around
multiline "qw" quotes to be left unchanged. This option will not
normally be necessary, but was added for testing purposes,
because in some versions of perl, trimming "qw" quotes changes
the syntax tree.

-sbq=n or --space-backslash-quote=n
lines like

$str1=\"string1";
$str2=\'string2';

can confuse syntax highlighters unless a space is included
between the backslash and the single or double quotation mark.

this can be controlled with the value of n as follows:

-sbq=0 means no space between the backslash and quote
-sbq=1 means follow the example of the source code
-sbq=2 means always put a space between the backslash and quote

The default is -sbq=1, meaning that a space will be used if there
is one in the source code.

Trimming trailing whitespace from lines of POD
-trp or --trim-pod will remove trailing whitespace from lines of
POD. The default is not to do this.

Comment Controls

Perltidy has a number of ways to control the appearance of both block
comments and side comments. The term block comment here refers to a
full-line comment, whereas side comment will refer to a comment which
appears on a line to the right of some code.

Perltidy does not do any word wrapping of commented text to match a
selected maximum line length. This is because there is no way to
determine if this is appropriate for the given content. However, an
interactive program named perlcomment.pl is available in the examples
folder of the perltidy distribution which can assist in doing this.

-ibc, --indent-block-comments
Block comments normally look best when they are indented to the
same level as the code which follows them. This is the default
behavior, but you may use -nibc to keep block comments left-
justified. Here is an example:

# this comment is indented (-ibc, default)
if ($task) { yyy(); }

The alternative is -nibc:

# this comment is not indented (-nibc)
if ($task) { yyy(); }

See also the next item, -isbc, as well as -sbc, for other ways to
have some indented and some outdented block comments.

-isbc, --indent-spaced-block-comments
If there is no leading space on the line, then the comment will
not be indented, and otherwise it may be.

If both -ibc and -isbc are set, then -isbc takes priority.

-olc, --outdent-long-comments
When -olc is set, lines which are full-line (block) comments
longer than the value maximum-line-length will have their
indentation removed. This is the default; use -nolc to prevent
outdenting.

-msc=n, --minimum-space-to-comment=n
Side comments look best when lined up several spaces to the right
of code. Perltidy will try to keep comments at least n spaces to
the right. The default is n=4 spaces.

-fpsc=n, --fixed-position-side-comment=n
This parameter tells perltidy to line up side comments in column
number n whenever possible. The default, n=0, will not do this.

-iscl, --ignore-side-comment-lengths
This parameter causes perltidy to ignore the length of side
comments when setting line breaks. The default, -niscl, is to
include the length of side comments when breaking lines to stay
within the length prescribed by the -l=n maximum line length
parameter. For example, the following single line would remain
intact with -l=50 and -iscl:

# perltidy -l=50 -iscl
$vmsfile =~ s/;[\d\-]*$//; # Clip off version number

whereas without the -iscl flag the line will be broken:

# perltidy -l=50
$vmsfile =~
s/;[\d\-]*$//; # Clip off version number

-ipc, --ignore-perlcritic-comments
Perltidy, by default, will look for side comments beginning with
## no critic and ignore their lengths when making line break
decisions, even if the user has not set -iscl. The reason is
that an unwanted line break can make these special comments
ineffective in controlling perlcritic.

Setting --ignore-perlcritic-comments tells perltidy not to look
for these ## no critic comments.

-hsc, --hanging-side-comments
By default, perltidy tries to identify and align "hanging side
comments", which are something like this:

my $IGNORE = 0; # This is a side comment
# This is a hanging side comment
# And so is this

A comment is considered to be a hanging side comment if (1) it
immediately follows a line with a side comment, or another
hanging side comment, and (2) there is some leading whitespace on
the line. To deactivate this feature, use -nhsc or
--nohanging-side-comments. If block comments are preceded by a
blank line, or have no leading whitespace, they will not be
mistaken as hanging side comments.

Closing Side Comments
A closing side comment is a special comment which perltidy can
automatically create and place after the closing brace of a code
block. They can be useful for code maintenance and debugging.
The command -csc (or --closing-side-comments) adds or updates
closing side comments. For example, here is a small code snippet

sub message {
if ( !defined( $_[0] ) ) {
print("Hello, World\n");
}
else {
print( $_[0], "\n" );
}
}

And here is the result of processing with "perltidy -csc":

sub message {
if ( !defined( $_[0] ) ) {
print("Hello, World\n");
}
else {
print( $_[0], "\n" );
}
} ## end sub message

A closing side comment was added for "sub message" in this case,
but not for the "if" and "else" blocks, because they were below
the 6 line cutoff limit for adding closing side comments. This
limit may be changed with the -csci command, described below.

The command -dcsc (or --delete-closing-side-comments) reverses
this process and removes these comments.

Several commands are available to modify the behavior of these
two basic commands, -csc and -dcsc:

-csci=n, or --closing-side-comment-interval=n
where "n" is the minimum number of lines that a block must
have in order for a closing side comment to be added. The
default value is "n=6". To illustrate:

# perltidy -csci=2 -csc
sub message {
if ( !defined( $_[0] ) ) {
print("Hello, World\n");
} ## end if ( !defined( $_[0] ))
else {
print( $_[0], "\n" );
} ## end else [ if ( !defined( $_[0] ))
} ## end sub message

Now the "if" and "else" blocks are commented. However, now
this has become very cluttered.

-cscp=string, or --closing-side-comment-prefix=string
where string is the prefix used before the name of the block
type. The default prefix, shown above, is "## end". This
string will be added to closing side comments, and it will
also be used to recognize them in order to update, delete,
and format them. Any comment identified as a closing side
comment will be placed just a single space to the right of
its closing brace.

-cscl=string, or --closing-side-comment-list
where "string" is a list of block types to be tagged with
closing side comments. By default, all code block types
preceded by a keyword or label (such as "if", "sub", and so
on) will be tagged. The -cscl command changes the default
list to be any selected block types; see "Specifying Block
Types". For example, the following command requests that
only "sub"'s, labels, "BEGIN", and "END" blocks be affected
by operations which add (-csc) or delete (-dcsc) closing side
comments:

--closing-side-comment-list='sub : BEGIN END'

-cscxl=string, or --closing-side-comment-exclusion-list
where "string" is a list of block types which should NOT be
tagged with closing side comments. If a block type appears in
both -cscl and -cscxl, then -cscxl has priority and the block
will not be tagged.

For example, the following command requests that anonymous
subs should not be affected by any -csc or -dcsc operation:

--closing-side-comment-exclusion-list='asub'

By default, no block types are excluded.

-csct=n, or --closing-side-comment-maximum-text=n
The text appended to certain block types, such as an "if"
block, is whatever lies between the keyword introducing the
block, such as "if", and the opening brace. Since this might
be too much text for a side comment, there needs to be a
limit, and that is the purpose of this parameter. The
default value is "n=20", meaning that no additional tokens
will be appended to this text after its length reaches 20
characters. Omitted text is indicated with "...". (Tokens,
including sub names, are never truncated, however, so actual
lengths may exceed this). To illustrate, in the above
example, the appended text of the first block is " (
!defined( $_[0] )...". The existing limit of "n=20" caused
this text to be truncated, as indicated by the "...". See
the next flag for additional control of the abbreviated text.

-cscb, or --closing-side-comments-balanced
As discussed in the previous item, when the closing-side-
comment-maximum-text limit is exceeded the comment text must
be truncated. Older versions of perltidy terminated with
three dots, and this can still be achieved with -ncscb:

perltidy -csc -ncscb
} ## end foreach my $foo (sort { $b cmp $a ...

However this causes a problem with editors which cannot
recognize comments or are not configured to do so because
they cannot "bounce" around in the text correctly. The -cscb
flag has been added to help them by appending appropriate
balancing structure:

perltidy -csc -cscb
} ## end foreach my $foo (sort { $b cmp $a ... })

The default is -cscb.

-csce=n, or --closing-side-comment-else-flag=n
The default, n=0, places the text of the opening "if"
statement after any terminal "else".

If n=2 is used, then each "elsif" is also given the text of
the opening "if" statement. Also, an "else" will include the
text of a preceding "elsif" statement. Note that this may
result some long closing side comments.

If n=1 is used, the results will be the same as n=2 whenever
the resulting line length is less than the maximum allowed.

-cscb, or --closing-side-comments-balanced
When using closing-side-comments, and the closing-side-
comment-maximum-text limit is exceeded, then the comment text
must be abbreviated. It is terminated with three dots if the
-cscb flag is negated:

perltidy -csc -ncscb
} ## end foreach my $foo (sort { $b cmp $a ...

This causes a problem with older editors which do not
recognize comments because they cannot "bounce" around in the
text correctly. The -cscb flag tries to help them by
appending appropriate terminal balancing structures:

perltidy -csc -cscb
} ## end foreach my $foo (sort { $b cmp $a ... })

The default is -cscb.

-cscw, or --closing-side-comment-warnings
This parameter is intended to help make the initial
transition to the use of closing side comments. It causes
two things to happen if a closing side comment replaces an
existing, different closing side comment: first, an error
message will be issued, and second, the original side comment
will be placed alone on a new specially marked comment line
for later attention.

The intent is to avoid clobbering existing hand-written side
comments which happen to match the pattern of closing side
comments. This flag should only be needed on the first run
with -csc.

Important Notes on Closing Side Comments:

+o Closing side comments are only placed on lines terminated
with a closing brace. Certain closing styles, such as the
use of cuddled elses (-ce), preclude the generation of some
closing side comments.

+o Please note that adding or deleting of closing side comments
takes place only through the commands -csc or -dcsc. The
other commands, if used, merely modify the behavior of these
two commands.

+o It is recommended that the -cscw flag be used along with -csc
on the first use of perltidy on a given file. This will
prevent loss of any existing side comment data which happens
to have the csc prefix.

+o Once you use -csc, you should continue to use it so that any
closing side comments remain correct as code changes.
Otherwise, these comments will become incorrect as the code
is updated.

+o If you edit the closing side comments generated by perltidy,
you must also change the prefix to be different from the
closing side comment prefix. Otherwise, your edits will be
lost when you rerun perltidy with -csc. For example, you
could simply change "## end" to be "## End", since the test
is case sensitive. You may also want to use the -ssc flag to
keep these modified closing side comments spaced the same as
actual closing side comments.

+o Temporarily generating closing side comments is a useful
technique for exploring and/or debugging a perl script,
especially one written by someone else. You can always
remove them with -dcsc.

Static Block Comments
Static block comments are block comments with a special leading
pattern, "##" by default, which will be treated slightly
differently from other block comments. They effectively behave
as if they had glue along their left and top edges, because they
stick to the left edge and previous line when there is no blank
spaces in those places. This option is particularly useful for
controlling how commented code is displayed.

-sbc, --static-block-comments
When -sbc is used, a block comment with a special leading
pattern, "##" by default, will be treated specially.

Comments so identified are treated as follows:

+o If there is no leading space on the line, then the
comment will not be indented, and otherwise it may be,

+o no new blank line will be inserted before such a comment,
and

+o such a comment will never become a hanging side comment.

For example, assuming @month_of_year is left-adjusted, the
default formatting for the following snippet is:

# perltidy -sbc [default]
@month_of_year = (
'Jan', 'Feb', 'Mar', 'Apr', 'May', 'Jun',
## 'Jul', 'Aug', 'Sep', 'Oct', 'Dec', 'Nov'
'Jul', 'Aug', 'Sep', 'Oct', 'Nov', 'Dec'
);

This can be useful for displaying code which has been
commented out. Without this convention, the above code would
become

# perltidy -nsbc
@month_of_year = (
'Jan', 'Feb', 'Mar', 'Apr', 'May', 'Jun',

## 'Jul', 'Aug', 'Sep', 'Oct', 'Dec', 'Nov'
'Jul', 'Aug', 'Sep', 'Oct', 'Nov', 'Dec'
);

which is less clear. The default is to use -sbc. This may
be deactivated with -nsbc.

-sbcp=string, --static-block-comment-prefix=string
This parameter defines the prefix used to identify static
block comments when the -sbc parameter is set. The default
prefix is "##", corresponding to "-sbcp=##". The prefix is
actually part of a perl pattern used to match lines and it
must either begin with "#" or "^#". In the first case a
prefix ^\s* will be added to match any leading whitespace,
while in the second case the pattern will match only comments
with no leading whitespace. For example, to identify all
comments as static block comments, one would use "-sbcp=#".
To identify all left-adjusted comments as static block
comments, use "-sbcp='^#'".

Please note that -sbcp merely defines the pattern used to
identify static block comments; it will not be used unless
the switch -sbc is set. Also, please be aware that since
this string is used in a perl regular expression which
identifies these comments, it must enable a valid regular
expression to be formed.

A pattern which can be useful is:

-sbcp=^#{2,}[^\s#]

This pattern requires a static block comment to have at least
one character which is neither a # nor a space. It allows a
line containing only '#' characters to be rejected as a
static block comment. Such lines are often used at the start
and end of header information in subroutines and should not
be separated from the intervening comments, which typically
begin with just a single '#'.

-osbc, --outdent-static-block-comments
The command -osbc will cause static block comments to be
outdented by the number of --continuation-spaces=n, if
possible.

Static Side Comments
Static side comments are side comments with a special leading
pattern. This option can be useful for controlling how commented
code is displayed when it is a side comment.

-ssc, --static-side-comments
When -ssc is used, a side comment with a static leading
pattern, which is "##" by default, will be spaced only a
single space from previous character, and it will not be
vertically aligned with other side comments.

The default is -nssc.

-sscp=string, --static-side-comment-prefix=string
This parameter defines the prefix used to identify static
side comments when the -ssc parameter is set. The default
prefix is "##", corresponding to "-sscp=##".

Please note that -sscp merely defines the pattern used to
identify static side comments; it will not be used unless the
switch -ssc is set. Also, note that this string is used in a
perl regular expression which identifies these comments, so
it must enable a valid regular expression to be formed.

Skipping Selected Sections of Code

Selected lines of code may be passed verbatim to the output without
any formatting by marking the starting and ending lines with special
comments. There are two options for doing this. The first option is
called --format-skipping or -fs, and the second option is called
--code-skipping or -cs.

In both cases the lines of code will be output without any changes.
The difference is that in --format-skipping perltidy will still parse
the marked lines of code and check for errors, whereas in
--code-skipping perltidy will simply pass the lines to the output
without any checking.

Both of these features are enabled by default and are invoked with
special comment markers. --format-skipping uses starting and ending
markers '#<<<' and '#>>>', like this:

#<<< format skipping: do not let perltidy change my nice formatting
my @list = (1,
1, 1,
1, 2, 1,
1, 3, 3, 1,
1, 4, 6, 4, 1,);
#>>>

--code-skipping uses starting and ending markers '#<<V' and '#>>V',
like this:

#<<V code skipping: perltidy passes this verbatim, no error checking

token ident_digit {
[ [ <?word> | _ | <?digit> ] <?ident_digit>
| <''>
]
};

#>>V

(The last character V is like a < or > rotated 90 degrees).
Additional text may appear on the special comment lines provided that
it is separated from the marker by at least one space to highlight
the sign, as in the above examples.

Any number of code-skipping or format-skipping sections may appear in
a file. If an opening code-skipping or format-skipping comment is
not followed by a corresponding closing comment, then skipping
continues to the end of the file. If a closing code-skipping or
format-skipping comment appears in a file but does not follow a
corresponding opening comment, then it is treated as an ordinary
comment without any special meaning.

It is recommended to use --code-skipping only if you need to hide a
block of an extended syntax which would produce errors if parsed by
perltidy, and use --format-skipping otherwise. This is because the
--format-skipping option provides the benefits of error checking, and
there are essentially no limitations on which lines to which it can
be applied. The --code-skipping option, on the other hand, does not
do error checking and its use is more restrictive because the code
which remains, after skipping the marked lines, must be syntactically
correct code with balanced containers.

These features should be used sparingly to avoid littering code with
markers, but they can be helpful for working around occasional
problems.

Note that it may be possible to avoid the use of --format-skipping
for the specific case of a comma-separated list of values, as in the
above example, by simply inserting a blank or comment somewhere
between the opening and closing parens. See the section "Controlling
List Formatting".

The following sections describe the available controls for these
options. They should not normally be needed.

-fs, --format-skipping
As explained above, this flag, which is enabled by default,
causes any code between special beginning and ending comment
markers to be passed to the output without formatting. The code
between the comments is still checked for errors however. The
default beginning marker is #<<< and the default ending marker is
#>>>.

Format skipping begins when a format skipping beginning comment
is seen and continues until a format-skipping ending comment is
found.

This feature can be disabled with -nfs. This should not
normally be necessary.

-fsb=string, --format-skipping-begin=string
This and the next parameter allow the special beginning and
ending comments to be changed. However, it is recommended that
they only be changed if there is a conflict between the default
values and some other use. If they are used, it is recommended
that they only be entered in a .perltidyrc file, rather than on a
command line. This is because properly escaping these parameters
on a command line can be difficult.

If changed comment markers do not appear to be working, use the
-log flag and examine the .LOG file to see if and where they are
being detected.

The -fsb=string parameter may be used to change the beginning
marker for format skipping. The default is equivalent to
-fsb='#<<<'. The string that you enter must begin with a # and
should be in quotes as necessary to get past the command shell of
your system. It is actually the leading text of a pattern that
is constructed by appending a '\s', so you must also include
backslashes for characters to be taken literally rather than as
patterns.

Some examples show how example strings become patterns:

-fsb='#\{\{\{' becomes /^#\{\{\{\s/ which matches #{{{ but not #{{{{
-fsb='#\*\*' becomes /^#\*\*\s/ which matches #** but not #***
-fsb='#\*{2,}' becomes /^#\*{2,}\s/ which matches #** and #*****

-fse=string, --format-skipping-end=string
The -fse=string is the corresponding parameter used to change the
ending marker for format skipping. The default is equivalent to
-fse='#<<<'.

The beginning and ending strings may be the same, but it is
preferable to make them different for clarity.

-cs, --code-skipping
As explained above, this flag, which is enabled by default,
causes any code between special beginning and ending comment
markers to be directly passed to the output without any error
checking or formatting. Essentially, perltidy treats it as if it
were a block of arbitrary text. The default beginning marker is
#<<V and the default ending marker is #>>V.

This feature can be disabled with -ncs. This should not
normally be necessary.

-csb=string, --code-skipping-begin=string
This may be used to change the beginning comment for a
--code-skipping section, and its use is similar to the
-fsb=string. The default is equivalent to -csb='#<<V'.

-cse=string, --code-skipping-end=string
This may be used to change the ending comment for a
--code-skipping section, and its use is similar to the
-fse=string. The default is equivalent to -cse='#>>V'.

Formatting a Limited Range of Lines

A command --line-range-tidy=n1:n2 is available to process just a
selected range of lines of an input stream with perltidy. This
command is mainly of interest for programming interactive code
editors. When it is used, the entire input stream is read but just
the selected range of lines of the input file are processed by the
perltidy tokenizer and formatter, and then the stream is reassembled
for output. The selected lines need to contain a complete statement
or balanced container. Otherwise, a syntax error will occur and the
code will not be tidied. There are a couple of limitations on the
use of this command: (1) it may not be applied to multiple files, and
(2) it only applies to code tidying and not, for example, html
formatting.

-lrt=n1:n2, --line-range-tidy=n1:n2
The range of lines is specified by integers n1 and n2, where n1
is the first line number to be formatted (start counting with 1)
and n2 is the last line number to be formatted. If n2 is not
given, or exceeds the actual number of lines, then formatting
continues to the end of the file.

Examples:

--line-range-tidy=43:109 # tidy lines 43 through 109
--line-range-tidy=' 43 : 109' # tidy lines 43 through 109
--line-range-tidy=1: # tidy all lines
--line-range-tidy=0:90 # ERROR (n1 must be >= 1)

The second example shows that spaces are okay if placed in
quotes.

Line Break Control

The parameters in this and the next sections control breaks after
non-blank lines of code. Blank lines are controlled separately by
parameters in the section "Blank Line Control".

-dnl, --delete-old-newlines
By default, perltidy first deletes all old line break locations,
and then it looks for good break points to match the desired line
length. Use -ndnl or --nodelete-old-newlines to force perltidy
to retain all old line break points.

-anl, --add-newlines
By default, perltidy will add line breaks when necessary to
create continuations of long lines and to improve the script
appearance. Use -nanl or --noadd-newlines to prevent any new
line breaks.

This flag does not prevent perltidy from eliminating existing
line breaks; see --freeze-newlines to completely prevent changes
to line break points.

-fnl, --freeze-newlines
If you do not want any changes to the line breaks within lines of
code in your script, set -fnl, and they will remain fixed, and
the rest of the commands in this section and sections
"Controlling List Formatting", "Retaining or Ignoring Existing
Line Breaks". You may want to use --nooutdent-long-lines with
this.

Note: If you also want to keep your blank lines exactly as they
are, you can use the --freeze-blank-lines flag which is described
in the section "Blank Line Control".

Controlling Breaks at Braces, Parens, and Square Brackets
-ce, --cuddled-else
Enable the "cuddled else" style, in which "else" and "elsif" are
follow immediately after the curly brace closing the previous
block. The default is not to use cuddled elses, and is indicated
with the flag -nce or --nocuddled-else. Here is a comparison of
the alternatives:

# -ce
if ($task) {
yyy();
} else {
zzz();
}

# -nce (default)
if ($task) {
yyy();
}
else {
zzz();
}

In this example the keyword else is placed on the same line which
begins with the preceding closing block brace and is followed by
its own opening block brace on the same line. Other keywords and
function names which are formatted with this "cuddled" style are
elsif, continue, catch, finally.

Other block types can be formatted by specifying their names on a
separate parameter --cuddled-block-list, described in a later
section.

Cuddling between a pair of code blocks requires that the closing
brace of the first block start a new line. If this block is
entirely on one line in the input file, it is necessary to decide
if it should be broken to allow cuddling. This decision is
controlled by the flag --cuddled-break-options=n (-cbo=n)
discussed below. The default and recommended value of -cbo=1
bases this decision on the first block in the chain. If it spans
multiple lines then cuddling is made and continues along the
chain, regardless of the sizes of subsequent blocks. Otherwise,
short lines remain intact.

So for example, the --cuddled-else flag would not have any effect
if the above snippet is rewritten as

if ($task) { yyy() }
else { zzz() }

If the first block spans multiple lines, then cuddling can be
done and will continue for the subsequent blocks in the chain, as
illustrated in the previous snippet.

If there are blank lines between cuddled blocks they will be
eliminated. If there are comments after the closing brace where
cuddling would occur then cuddling will be prevented. If this
occurs, cuddling will restart later in the chain if possible.

-cb, --cuddled-blocks
This flag is equivalent to --cuddled-else (-ce).

-cbl, --cuddled-block-list
The built-in default cuddled block types are else, elsif,
continue, catch, finally.

Additional block types to which the --cuddled-blocks style
applies can be defined by this parameter. This parameter is a
character string, giving a list of block types separated by
commas or spaces. For example, to cuddle code blocks of type
sort, map and grep, in addition to the default types, the string
could be set to

-cbl="sort map grep"

or equivalently

-cbl=sort,map,grep

Note however that these particular block types are typically
short so there might not be much opportunity for the cuddled
format style.

Using commas avoids the need to protect spaces with quotes.

As a diagnostic check, the flag --dump-cuddled-block-list or
-dcbl can be used to view the hash of values that are generated
by this flag.

Finally, note that the --cuddled-block-list parameter by itself
merely specifies which blocks are formatted with the cuddled
format. It has no effect unless this formatting style is
activated with --cuddled-else.

-cblx, --cuddled-block-list-exclusive
When cuddled else formatting is selected with --cuddled-else,
setting this flag causes perltidy to ignore its built-in defaults
and rely exclusively on the block types specified on the
--cuddled-block-list flag described in the previous section. For
example, to avoid using cuddled catch and finally, which are
among the defaults, the following set of parameters could be
used:

perltidy -ce -cbl='else elsif continue' -cblx

-cbo=n, --cuddled-break-option=n
Cuddled formatting is only possible between a pair of code blocks
if the closing brace of the first block starts a new line. If a
block is encountered which is entirely on a single line, and
cuddled formatting is selected, it is necessary to make a
decision as to whether or not to "break" the block, meaning to
cause it to span multiple lines. This parameter controls that
decision. The options are:

cbo=0 Never force a short block to break.
cbo=1 If the first of a pair of blocks is broken in the input
file, then break the second [DEFAULT].
cbo=2 Break open all blocks for maximal cuddled formatting.

The default and recommended value is cbo=1. With this value, if
the starting block of a chain spans multiple lines, then a
cascade of breaks will occur for remaining blocks causing the
entire chain to be cuddled.

The option cbo=0 can produce erratic cuddling if there are
numerous one-line blocks.

The option cbo=2 produces maximal cuddling but will not allow any
short blocks.

-bl, --opening-brace-on-new-line, or --brace-left
Use the flag -bl to place an opening block brace on a new line:

if ( $input_file eq '-' )
{
...
}

By default it applies to all structural blocks except sort map
grep eval and anonymous subs.

The default is -nbl which places an opening brace on the same
line as the keyword introducing it if possible. For example,

# default
if ( $input_file eq '-' ) {
...
}

When -bl is set, the blocks to which this applies can be
controlled with the parameters --brace-left-list and
--brace-left-exclusion-list described in the next sections.

-bll=s, --brace-left-list=s
Use this parameter to change the types of block braces for which
the -bl flag applies; see "Specifying Block Types". For example,
-bll='if elsif else sub' would apply it to only "if/elsif/else"
and named sub blocks. The default is all blocks, -bll='*'.

-blxl=s, --brace-left-exclusion-list=s
Use this parameter to exclude types of block braces for which the
-bl flag applies; see "Specifying Block Types". For example, the
default settings -bll='*' and -blxl='sort map grep eval asub'
mean all blocks except sort map grep eval and anonymous sub
blocks.

Note that the lists -bll=s and -blxl=s control the behavior of
the -bl flag but have no effect unless the -bl flag is set. These
two lists provide complete control for this flag, but two
shortcut flags are available and described in the next sections.

-sbl, --opening-sub-brace-on-new-line
The flag -sbl provides a shortcut way to turn on -bl just for
named subs. The same effect can be achieved by turning on -bl
with the block list set as -bll='sub'. To avoid conflicts, it is
recommended to either use the more general list method described
above to control -bl, or this shortcut method, but not both.

For example,

perltidy -sbl

produces this result:

sub message
{
if (!defined($_[0])) {
print("Hello, World\n");
}
else {
print($_[0], "\n");
}
}

This negative version of this flag, -nsbl, turns off -bl for
named subs. The same effect can be achieved with the exclusion
list method, -blxl=sub.

-asbl, --opening-anonymous-sub-brace-on-new-line
The flag -asbl is like the -sbl flag except that it applies to
anonymous sub's instead of named subs. The same effect can be
achieved by turning on -bl with the block list set to include
-bll='asub'.

For example

perltidy -asbl

produces this result:

$a = sub
{
if ( !defined( $_[0] ) ) {
print("Hello, World\n");
}
else {
print( $_[0], "\n" );
}
};

This negative version of this flag, -nasbl, turns off -bl for
anonymous subs.

-bli, --brace-left-and-indent
The flag -bli is similar to the -bl flag but in addition it
causes one unit of continuation indentation ( see
--continuation-indentation ) to be placed before an opening and
closing block braces.

For example

# perltidy -bli
if ( $input_file eq '-' )
{
important_function();
}

By default, this extra indentation occurs for block types: if,
elsif, else, unless, while, for, foreach, do, and also named subs
and blocks preceded by a label. The next item shows how to
change this.

Note: The -bli flag is similar to the -bl flag, with the
difference being that braces get indented. But these two flags
are implemented independently, and have different default
settings for historical reasons. If desired, a mixture of
effects can be achieved if desired by turning them both on with
different -list settings. In the event that both settings are
selected for a certain block type, the -bli style has priority.

-blil=s, --brace-left-and-indent-list=s
Use this parameter to change the types of block braces for which
the -bli flag applies; see "Specifying Block Types".

The default is -blil='if else elsif unless while for foreach do :
sub'.

-blixl=s, --brace-left-and-indent-exclusion-list=s
Use this parameter to exclude types of block braces for which the
-bli flag applies; see "Specifying Block Types".

This might be useful in conjunction with selecting all blocks
-blil='*'. The default setting is -blixl=' ', which does not
exclude any blocks.

Note that the two parameters -blil and -blixl control the
behavior of the -bli flag but have no effect unless the -bli flag
is set.

-bar, --opening-brace-always-on-right
The default style, -nbl places the opening code block brace on a
new line if it does not fit on the same line as the opening
keyword, like this:

if ( $bigwasteofspace1 && $bigwasteofspace2
|| $bigwasteofspace3 && $bigwasteofspace4 )
{
big_waste_of_time();
}

To force the opening brace to always be on the right, use the
-bar flag. In this case, the above example becomes

if ( $bigwasteofspace1 && $bigwasteofspace2
|| $bigwasteofspace3 && $bigwasteofspace4 ) {
big_waste_of_time();
}

A conflict occurs if both --opening-brace-on_new-line (-bl) and
-bar are specified.

-cpb, --cuddled-paren-brace
A related parameter, --cuddled-paren-brace, causes perltidy to
join two lines which otherwise would be

)
{

to be

) {

For example:

# default
foreach my $dir (
'05_lexer', '07_token', '08_regression', '11_util',
'13_data', '15_transform'
)
{
...
}

# perltidy -cpb
foreach my $dir (
'05_lexer', '07_token', '08_regression', '11_util',
'13_data', '15_transform'
) {
...;
}

-otr, --opening-token-right and related flags
The -otr flag is a hint that perltidy should not place a break
between a comma and an opening token. For example:

# default formatting
push @{ $self->{$module}{$key} },
{
accno => $ref->{accno},
description => $ref->{description}
};

# perltidy -otr
push @{ $self->{$module}{$key} }, {
accno => $ref->{accno},
description => $ref->{description}
};

The flag -otr is actually an abbreviation for three other flags
which can be used to control parens, hash braces, and square
brackets separately if desired:

-opr or --opening-paren-right
-ohbr or --opening-hash-brace-right
-osbr or --opening-square-bracket-right

-bbhb=n, --break-before-hash-brace=n and related flags
When a list of items spans multiple lines, the default formatting
is to place the opening brace (or other container token) at the
end of the starting line, like this:

$romanNumerals = {
one => 'I',
two => 'II',
three => 'III',
four => 'IV',
};

This flag can change the default behavior to cause a line break
to be placed before the opening brace according to the value
given to the integer n:

-bbhb=0 never break [default]
-bbhb=1 stable: break if the input script had a break
-bbhb=2 break if list is 'complex' (see note below)
-bbhb=3 always break

For example,

# perltidy -bbhb=3
$romanNumerals =
{
one => 'I',
two => 'II',
three => 'III',
four => 'IV',
};

There are several points to note about this flag:

+o This parameter only applies if the opening brace is preceded
by an '=' or '=>'.

+o This parameter only applies if the contents of the container
looks like a list. The contents need to contain some commas
or '=>'s at the next interior level to be considered a list.

+o For the n=2 option, a list is considered 'complex' if it is
part of a nested list structure which spans multiple lines in
the input file.

+o If multiple opening tokens have been 'welded' together with
the -wn parameter, then this parameter has no effect.

+o The indentation of the braces will normally be one level of
continuation indentation by default. This can be changed
with the parameter -bbhbi=n in the next section.

+o Similar flags for controlling parens and square brackets are
given in the subsequent section.

-bbhbi=n, --break-before-hash-brace-and-indent=n
This flag is a companion to -bbhb=n for controlling the
indentation of an opening hash brace which is placed on a new
line by that parameter. The indentation is as follows:

-bbhbi=0 one continuation level [default]
-bbhbi=1 outdent by one continuation level
-bbhbi=2 indent one full indentation level

For example:

# perltidy -bbhb=3 -bbhbi=1
$romanNumerals =
{
one => 'I',
two => 'II',
three => 'III',
four => 'IV',
};

# perltidy -bbhb=3 -bbhbi=2
$romanNumerals =
{
one => 'I',
two => 'II',
three => 'III',
four => 'IV',
};

Note that this parameter has no effect unless -bbhb=n is also
set.

-bbsb=n, --break-before-square-bracket=n
This flag is similar to the flag described above, except it
applies to lists contained within square brackets.

-bbsb=0 never break [default]
-bbsb=1 stable: break if the input script had a break
-bbsb=2 break if list is 'complex' (part of nested list structure)
-bbsb=3 always break

-bbsbi=n, --break-before-square-bracket-and-indent=n
This flag is a companion to -bbsb=n for controlling the
indentation of an opening square bracket which is placed on a new
line by that parameter. The indentation is as follows:

-bbsbi=0 one continuation level [default]
-bbsbi=1 outdent by one continuation level
-bbsbi=2 indent one full indentation level

-bbp=n, --break-before-paren=n
This flag is similar to -bbhb=n, described above, except it
applies to lists contained within parens.

-bbp=0 never break [default]
-bbp=1 stable: break if the input script had a break
-bpb=2 break if list is 'complex' (part of nested list structure)
-bbp=3 always break

-bbpi=n, --break-before-paren-and-indent=n
This flag is a companion to -bbp=n for controlling the
indentation of an opening paren which is placed on a new line by
that parameter. The indentation is as follows:

-bbpi=0 one continuation level [default]
-bbpi=1 outdent by one continuation level
-bbpi=2 indent one full indentation level

-bfvt=n, --brace-follower-vertical-tightness=n
Some types of closing block braces, such as eval, may be followed
by additional code. A line break may be inserted between such a
closing brace and the following code depending on the parameter n
and the length of the trailing code, as follows:

If the trailing code fits on a single line, then

-bfvt=0 Follow the input style regarding break/no-break
-bfvt=1 Follow the input style regarding break/no-break [Default]
-bfvt=2 Do not insert a line break

If the trailing code requires multiple lines, then

-bfvt=0 Insert a line break
-bfvt=1 Insert a line break except for a cuddled block chain
-bfvt=2 Do not insert a line break

The default is -bfvt=1. The most compact code is achieved with
-bfvt=2.

Example (non-cuddled, multiple lines ):

# -bfvt=0 or -bvft=1 [DEFAULT]
eval {
( $line, $cond ) = $self->_normalize_if_elif($line);
1;
}
or die sprintf "Error at line %d\nLine %d: %s\n%s",
( $line_info->start_line_num() ) x 2, $line, $@;

# -bfvt=2
eval {
( $line, $cond ) = $self->_normalize_if_elif($line);
1;
} or die sprintf "Error at line %d\nLine %d: %s\n%s",
( $line_info->start_line_num() ) x 2, $line, $@;

Example (cuddled, multiple lines):

# -bfvt=0
eval {
#STUFF;
1; # return true
}
or do {
##handle error
};

# -bfvt=1 [DEFAULT] or -bfvt=2
eval {
#STUFF;
1; # return true
} or do {
##handle error
};

Welding

-wn, --weld-nested-containers
The -wn flag causes closely nested pairs of opening and closing
container symbols (curly braces, brackets, or parens) to be
"welded" together, meaning that they are treated as if combined
into a single unit, with the indentation of the innermost code
reduced to be as if there were just a single container symbol.

For example:

# default formatting
do {
{
next if $x == $y;
}
} until $x++ > $z;

# perltidy -wn
do { {
next if $x == $y;
} } until $x++ > $z;

When this flag is set perltidy makes a preliminary pass through
the file and identifies all nested pairs of containers. To
qualify as a nested pair, the closing container symbols must be
immediately adjacent and the opening symbols must either (1) be
adjacent as in the above example, or (2) have an anonymous sub
declaration following an outer opening container symbol which is
not a code block brace, or (3) have an outer opening paren
separated from the inner opening symbol by any single non-
container symbol or something that looks like a function
evaluation, as illustrated in the next examples. An additional
option (4) which can be turned on with the flag --weld-fat-comma
is when the opening container symbols are separated by a hash key
and fat comma (=>).

Any container symbol may serve as both the inner container of one
pair and as the outer container of an adjacent pair.
Consequently, any number of adjacent opening or closing symbols
may join together in weld. For example, here are three levels of
wrapped function calls:

# default formatting
my (@date_time) = Localtime(
Date_to_Time(
Add_Delta_DHMS(
$year, $month, $day, $hour, $minute, $second,
'0', $offset, '0', '0'
)
)
);

# perltidy -wn
my (@date_time) = Localtime( Date_to_Time( Add_Delta_DHMS(
$year, $month, $day, $hour, $minute, $second,
'0', $offset, '0', '0'
) ) );

Notice how the indentation of the inner lines are reduced by two
levels in this case. This example also shows the typical result
of this formatting, namely it is a sandwich consisting of an
initial opening layer, a central section of any complexity
forming the "meat" of the sandwich, and a final closing layer.
This predictable structure helps keep the compacted structure
readable.

The inner sandwich layer is required to be at least one line
thick. If this cannot be achieved, welding does not occur. This
constraint can cause formatting to take a couple of iterations to
stabilize when it is first applied to a script. The --converge
flag can be used to insure that the final format is achieved in a
single run.

Here is an example illustrating a welded container within a
welded containers:

# default formatting
$x->badd(
bmul(
$class->new(
abs(
$sx * int( $xr->num() ) & $sy * int( $yr->num() )
)
),
$m
)
);

# perltidy -wn
$x->badd( bmul(
$class->new( abs(
$sx * int( $xr->num() ) & $sy * int( $yr->num() )
) ),
$m
) );

The welded closing tokens are by default on a separate line but
this can be modified with the --vertical-tightness-closing=n
(-vtc=n) flag (described in the next section). For example, the
same example adding -vtc=2 is

# perltidy -wn -vtc=2
$x->badd( bmul(
$class->new( abs(
$sx * int( $xr->num() ) & $sy * int( $yr->num() ) ) ),
$m ) );

This format option is quite general but there are some
limitations.

One limitation is that any line length limit still applies and
can cause long welded sections to be broken into multiple lines.

Also, the stacking of containers defined by this flag have
priority over any other container stacking flags. This is
because any welding is done first.

-wfc, --weld-fat-comma
When the -wfc flag is set, along with -wn
(--weld-nested-containers), perltidy is allowed to weld an
opening paren to an inner opening container when they are
separated by a hash key and fat comma (=>). for example

# perltidy -wn -wfc
elf->call_method( method_name_foo => {
some_arg1 => $foo,
some_other_arg3 => $bar->{'baz'},
} );

This option is off by default.

-wnxl=s, --weld-nested-exclusion-list
The -wnxl=s flag provides some control over the types of
containers which can be welded. The -wn flag by default is
"greedy" in welding adjacent containers. If it welds more types
of containers than desired, this flag provides a capability to
reduce the amount of welding by specifying a list of things which
should not be welded.

The logic in perltidy to apply this is straightforward. As each
container token is being considered for joining a weld, any
exclusion rules are consulted and used to reject the weld if
necessary.

This list is a string with space-separated items. Each item
consists of up to three pieces of information: (1) an optional
position, (2) an optional preceding type, and (3) a container
type.

The only required piece of information is a container type, which
is one of '(', '[', '{' or 'q'. The first three of these are
container tokens and the last represents a quoted list. For
example the string

-wnxl='[ { q'

means do NOT include square-brackets, braces, or quotes in any
welds. The only unspecified container is '(', so this string
means that only welds involving parens will be made.

To illustrate, following welded snippet consists of a chain of
three welded containers with types '(' '[' and 'q':

# perltidy -wn
skip_symbols( [ qw(
Perl_dump_fds
Perl_ErrorNo
Perl_GetVars
PL_sys_intern
) ] );

Even though the qw term uses parens as the quote delimiter, it
has a special type 'q' here. If it appears in a weld it always
appears at the end of the welded chain.

Any of the container types '[', '{', and '(' may be prefixed with
a position indicator which is either '^', to indicate the first
token of a welded sequence, or '.', to indicate an interior token
of a welded sequence. (Since a quoted string 'q' always ends a
chain it does need a position indicator).

For example, if we do not want a sequence of welded containers to
start with a square bracket we could use

-wnxl='^['

In the above snippet, there is a square bracket but it does not
start the chain, so the formatting would be unchanged if it were
formatted with this restriction.

A third optional item of information which can be given is an
alphanumeric letter which is used to limit the selection further
depending on the type of token immediately before the container.
If given, it goes just before the container symbol. The possible
letters are currently 'k', 'K', 'f', 'F', 'w', and 'W', with
these meanings:

'k' matches if the previous nonblank token is a perl keyword
(such as 'if', 'while'),
'K' matches if 'k' does not: previous token is not a keyword
'f' matches if previous token is a function (not a keyword)
'F' matches if 'f' does not
'w' matches if either 'k' or 'f' match
'W' matches if 'w' does not

For example, compare

# perltidy -wn
if ( defined( $_Cgi_Query{
$Config{'methods'}{'auth'}{'remote'}{'cgi'}{'username'}
} ) )

with

# perltidy -wn -wnxl='^K( {'
if ( defined(
$_Cgi_Query{ $Config{'methods'}{'auth'}{'remote'}{'cgi'}
{'username'} }
) )

The first case does maximum welding. In the second case the
leading paren is retained by the rule (it would have been
rejected if preceded by a non-keyword) but the curly brace is
rejected by the rule.

Here are some additional example strings and their meanings:

'^(' - the weld must not start with a paren
'.(' - second and later tokens may not be parens
'.w(' - second and later tokens may not be a keyword or call parens
'(' - no parens in a weld
'^K(' - exclude a leading paren preceded by a non-keyword
'.k(' - exclude a secondary paren preceded by a keyword
'[ {' - exclude all brackets and braces
'[ ( ^K{' - exclude all except nested structures like do {{ ... }}

Vertical tightness of non-block curly braces, parentheses, and square
brackets.
These parameters control what shall be called vertical tightness.
Here are the main points:

+o Opening tokens (except for block braces) are controlled by
-vt=n, or --vertical-tightness=n, where

-vt=0 always break a line after opening token (default).
-vt=1 do not break unless this would produce more than one
step in indentation in a line.
-vt=2 never break a line after opening token

+o You must also use the -lp (--line-up-parentheses) flag when
you use the -vt flag; the reason is explained below.

+o Closing tokens (except for block braces) are controlled by
-vtc=n, or --vertical-tightness-closing=n, where

-vtc=0 always break a line before a closing token (default),
-vtc=1 do not break before a closing token which is followed
by a semicolon or another closing token, and is not in
a list environment.
-vtc=2 never break before a closing token.
-vtc=3 Like -vtc=1 except always break before a closing token
if the corresponding opening token follows an = or =>.

The rules for -vtc=1 and -vtc=3 are designed to maintain a
reasonable balance between tightness and readability in
complex lists.

+o Different controls may be applied to different token types,
and it is also possible to control block braces; see below.

+o Finally, please note that these vertical tightness flags are
merely hints to the formatter, and it cannot always follow
them. Things which make it difficult or impossible include
comments, blank lines, blocks of code within a list, and
possibly the lack of the --line-up-parentheses parameter.
Also, these flags may be ignored for very small lists (2 or 3
lines in length).

Here are some examples:

# perltidy -lp -vt=0 -vtc=0
%romanNumerals = (
one => 'I',
two => 'II',
three => 'III',
four => 'IV',
);

# perltidy -lp -vt=1 -vtc=0
%romanNumerals = ( one => 'I',
two => 'II',
three => 'III',
four => 'IV',
);

# perltidy -lp -vt=1 -vtc=1
%romanNumerals = ( one => 'I',
two => 'II',
three => 'III',
four => 'IV', );

# perltidy -vtc=3
my_function(
one => 'I',
two => 'II',
three => 'III',
four => 'IV', );

# perltidy -vtc=3
%romanNumerals = (
one => 'I',
two => 'II',
three => 'III',
four => 'IV',
);

In the last example for -vtc=3, the opening paren is preceded by
an equals so the closing paren is placed on a new line.

The difference between -vt=1 and -vt=2 is shown here:

# perltidy -lp -vt=1
$init->add(
mysprintf( "(void)find_threadsv(%s);",
cstring( $threadsv_names[ $op->targ ] )
)
);

# perltidy -lp -vt=2
$init->add( mysprintf( "(void)find_threadsv(%s);",
cstring( $threadsv_names[ $op->targ ] )
)
);

With -vt=1, the line ending in "add(" does not combine with the
next line because the next line is not balanced. This can help
with readability, but -vt=2 can be used to ignore this rule.

The tightest, and least readable, code is produced with both
"-vt=2" and "-vtc=2":

# perltidy -lp -vt=2 -vtc=2
$init->add( mysprintf( "(void)find_threadsv(%s);",
cstring( $threadsv_names[ $op->targ ] ) ) );

Notice how the code in all of these examples collapses vertically
as -vt increases, but the indentation remains unchanged. This is
because perltidy implements the -vt parameter by first formatting
as if -vt=0, and then simply overwriting one output line on top
of the next, if possible, to achieve the desired vertical
tightness. The -lp (--line-up-parentheses) indentation style has
been designed to allow this vertical collapse to occur, which is
why it is required for the -vt parameter.

The -vt=n and -vtc=n parameters apply to each type of container
token. If desired, vertical tightness controls can be applied
independently to each of the closing container token types.

The parameters for controlling parentheses are -pvt=n or
--paren-vertical-tightness=n, and -pvtc=n or
--paren-vertical-tightness-closing=n.

Likewise, the parameters for square brackets are -sbvt=n or
--square-bracket-vertical-tightness=n, and -sbvtc=n or
--square-bracket-vertical-tightness-closing=n.

Finally, the parameters for controlling non-code block braces are
-bvt=n or --brace-vertical-tightness=n, and -bvtc=n or
--brace-vertical-tightness-closing=n.

In fact, the parameter -vt=n is actually just an abbreviation for
-pvt=n -bvt=n sbvt=n, and likewise -vtc=n is an abbreviation for
-pvtc=n -bvtc=n -sbvtc=n.

-bbvt=n or --block-brace-vertical-tightness=n
The -bbvt=n flag is just like the -vt=n flag but applies to
opening code block braces.

-bbvt=0 break after opening block brace (default).
-bbvt=1 do not break unless this would produce more than one
step in indentation in a line.
-bbvt=2 do not break after opening block brace.

It is necessary to also use either -bl or -bli for this to work,
because, as with other vertical tightness controls, it is
implemented by simply overwriting a line ending with an opening
block brace with the subsequent line. For example:

# perltidy -bli -bbvt=0
if ( open( FILE, "< $File" ) )
{
while ( $File = <FILE> )
{
$In .= $File;
$count++;
}
close(FILE);
}

# perltidy -bli -bbvt=1
if ( open( FILE, "< $File" ) )
{ while ( $File = <FILE> )
{ $In .= $File;
$count++;
}
close(FILE);
}

By default this applies to blocks associated with keywords if,
elsif, else, unless, for, foreach, sub, while, until, and also
with a preceding label. This can be changed with the parameter
-bbvtl=string, or --block-brace-vertical-tightness-list=string,
where string is a space-separated list of block types. For more
information on the possible values of this string, see
"Specifying Block Types"

For example, if we want to just apply this style to "if",
"elsif", and "else" blocks, we could use "perltidy -bli -bbvt=1
-bbvtl='if elsif else'".

There is no vertical tightness control for closing block braces;
with one exception they will be placed on separate lines. The
exception is that a cascade of closing block braces may be
stacked on a single line. See --stack-closing-block-brace.

-sot, --stack-opening-tokens and related flags
The -sot flag tells perltidy to "stack" opening tokens when
possible to avoid lines with isolated opening tokens.

For example:

# default
$opt_c = Text::CSV_XS->new(
{
binary => 1,
sep_char => $opt_c,
always_quote => 1,
}
);

# -sot
$opt_c = Text::CSV_XS->new( {
binary => 1,
sep_char => $opt_c,
always_quote => 1,
}
);

For detailed control of individual closing tokens the following
controls can be used:

-sop or --stack-opening-paren
-sohb or --stack-opening-hash-brace
-sosb or --stack-opening-square-bracket
-sobb or --stack-opening-block-brace

The flag -sot is an abbreviation for -sop -sohb -sosb.

The flag -sobb is an abbreviation for -bbvt=2 -bbvtl='*'. This
will case a cascade of opening block braces to appear on a single
line, although this an uncommon occurrence except in test
scripts.

-sct, --stack-closing-tokens and related flags
The -sct flag tells perltidy to "stack" closing tokens when
possible to avoid lines with isolated closing tokens.

For example:

# default
$opt_c = Text::CSV_XS->new(
{
binary => 1,
sep_char => $opt_c,
always_quote => 1,
}
);

# -sct
$opt_c = Text::CSV_XS->new(
{
binary => 1,
sep_char => $opt_c,
always_quote => 1,
} );

The -sct flag is somewhat similar to the -vtc
(--vertical-tightness-closing flags, and in some cases it can
give a similar result. The difference is that the -vtc flags try
to avoid lines with leading opening tokens by "hiding" them at
the end of a previous line, whereas the -sct flag merely tries to
reduce the number of lines with isolated closing tokens by
stacking them but does not try to hide them. For example:

# -vtc=2
$opt_c = Text::CSV_XS->new(
{
binary => 1,
sep_char => $opt_c,
always_quote => 1, } );

For detailed control of the stacking of individual closing tokens
the following controls can be used:

-scp or --stack-closing-paren
-schb or --stack-closing-hash-brace
-scsb or --stack-closing-square-bracket
-scbb or --stack-closing-block-brace

The flag -sct is an abbreviation for stacking the non-block
closing tokens, -scp -schb -scsb.

Stacking of closing block braces, -scbb, causes a cascade of
isolated closing block braces to be combined into a single line
as in the following example:

# -scbb:
for $w1 (@w1) {
for $w2 (@w2) {
for $w3 (@w3) {
for $w4 (@w4) {
push( @lines, "$w1 $w2 $w3 $w4\n" );
} } } }

To simplify input even further for the case in which both opening
and closing non-block containers are stacked, the flag -sac or
--stack-all-containers is an abbreviation for
--stack-opening-tokens --stack-closing-tokens.

Please note that if both opening and closing tokens are to be
stacked, then the newer flag --weld-nested-containers may be
preferable because it insures that stacking is always done
symmetrically. It also removes an extra level of unnecessary
indentation within welded containers. It is able to do this
because it works on formatting globally rather than locally, as
the --stack-opening-tokens and --stack-closing-tokens flags do.

Breaking Before or After Operators

Four command line parameters provide some control over whether a line
break should be before or after specific token types. Two parameters
give detailed control:

-wba=s or --want-break-after=s, and

-wbb=s or --want-break-before=s.

These parameters are each followed by a quoted string, s, containing
a list of token types (separated only by spaces). No more than one
of each of these parameters should be specified, because repeating a
command-line parameter always overwrites the previous one before
perltidy ever sees it.

By default, perltidy breaks after these token types:
% + - * / x != == >= <= =~ !~ < > | &
= **= += *= &= <<= &&= -= /= |= >>= ||= //= .= %= ^= x=

And perltidy breaks before these token types by default:
. << >> -> && || //

To illustrate, to cause a break after a concatenation operator, '.',
rather than before it, the command line would be

-wba="."

As another example, the following command would cause a break before
math operators '+', '-', '/', and '*':

-wbb="+ - / *"

These commands should work well for most of the token types that
perltidy uses (use --dump-token-types for a list). Also try the -D
flag on a short snippet of code and look at the .DEBUG file to see
the tokenization. However, for a few token types there may be
conflicts with hardwired logic which cause unexpected results. One
example is curly braces, which should be controlled with the
parameter --opening-brace-on-new-line provided for that purpose.

WARNING Be sure to put these tokens in quotes to avoid having them
misinterpreted by your command shell.

Two additional parameters are available which, though they provide no
further capability, can simplify input are:

-baao or --break-after-all-operators,

-bbao or --break-before-all-operators.

The -baao sets the default to be to break after all of the following
operators:

% + - * / x != == >= <= =~ !~ < > | &
= **= += *= &= <<= &&= -= /= |= >>= ||= //= .= %= ^= x=
. : ? && || and or err xor

and the -bbao flag sets the default to break before all of these
operators. These can be used to define an initial break preference
which can be fine-tuned with the --want-break-after and
--want-break-before flags. For example, to break before all
operators except an = one could use "-bbao -wba='='" rather than
listing every single perl operator except = on a -wbb flag.

-bal=n, --break-after-labels=n
This flag controls whether or not a line break occurs after a
label. There are three possible values for n:

-bal=0 break if there is a break in the input [DEFAULT]
-bal=1 always break after a label
-bal=2 never break after a label

For example,

# perltidy -bal=1
RETURN:
return;

# perltidy -bal=2
RETURN: return;

Controlling List Formatting

Perltidy attempts to format lists of comma-separated values in tables
which look good. Its default algorithms usually work well, but
sometimes they don't. In this case, there are several methods
available to control list formatting.

A very simple way to prevent perltidy from changing the line breaks
within a comma-separated list of values is to insert a blank line,
comment, or side-comment anywhere between the opening and closing
parens (or braces or brackets). This causes perltidy to skip over
its list formatting logic. (The reason is that any of these items
put a constraint on line breaks, and perltidy needs complete control
over line breaks within a container to adjust a list layout). For
example, let us consider

my @list = (1,
1, 1,
1, 2, 1,
1, 3, 3, 1,
1, 4, 6, 4, 1,);

The default formatting, which allows a maximum line length of 80,
will flatten this down to one line:

# perltidy (default)
my @list = ( 1, 1, 1, 1, 2, 1, 1, 3, 3, 1, 1, 4, 6, 4, 1, );

This formatting loses the nice structure. The original line breaks
can be retained by adding comment or a blank line somewhere between
the two parens. For example,

my @list = (
1, # a side comment forces the original breakpoints to be kept
1, 1,
1, 2, 1,
1, 3, 3, 1,
1, 4, 6, 4, 1,
);

We could achieve the same result with a blank line or full comment
anywhere between the opening and closing parens. Vertical alignment
of the list items will still occur if possible. The blank line method
is shown here:

my @list = (

1,
1, 1,
1, 2, 1,
1, 3, 3, 1,
1, 4, 6, 4, 1,
);

For another possibility see the -fs flag in "Skipping Selected
Sections of Code".

-boc, --break-at-old-comma-breakpoints
The -boc flag is another way to prevent comma-separated lists
from being reformatted. Using -boc on the above example, plus
additional flags to retain the original style, yields

# perltidy -boc -lp -pt=2 -vt=1 -vtc=1
my @list = (1,
1, 1,
1, 2, 1,
1, 3, 3, 1,
1, 4, 6, 4, 1,);

A disadvantage of this flag compared to the methods discussed
above is that all tables in the file must already be nicely
formatted.

-btct=s, --break-at-trailing-comma-types=s
A trailing comma is an optional comma following the last item of
a list. The -btct=s tells perltidy to end a line at selected
trailing commas. The string s selects the trailing commas, as
follows:

s=1 or '*' : every trailing comma
s=m a trailing comma in a multiline list
s=b a bare trailing comma
s=0 none

For example, given the following input

$w->bind(
'<Page_Down>' => xx,
);

The default formatting would flatten this into a single line.
But the container can be kept open with

# perltidy -btct='b'
$w->bind(
'<Page_Down>' => xx,
);

This can be particularly useful for short function calls like
this, where the default perltidy formatting does not keep the
container open.

The options s=m and s=1 can be used to open up containers with
non-bare trailing commas. For example, given this input

$w->bind( '<Page_Down>' => xx, );

we can break it open with

# perltidy -btct=1
$w->bind(
'<Page_Down>' => xx,
);

Afterwards, we could switch to -btct='b' since the trailing comma
is now bare. But the -btct parameter must be retained in this
case because otherwise this small list will be flattened the next
time it is formatted.

This logic can be restricted to specific container types by
including an opening token ahead of the letter in the above
table. For example

-btct='(b'

means that this operation should only apply to bare trailing
commas within parentheses.

For parentheses, an additional item of information which can be
given is an alphanumeric letter which is used to limit the
selection further depending on the type of token immediately
before the opening paren. The possible letters are currently
'k', 'K', 'f', 'F', 'w', and 'W', with these meanings for
matching whatever precedes an opening paren:

'k' matches if the previous nonblank token is a perl keyword
(such as 'if', 'while'),
'K' matches if 'k' does not: previous token is not a keyword
'f' matches if previous token is a function (not a keyword)
'F' matches if 'f' does not.
'w' matches if either 'k' or 'f' match.
'W' matches if 'w' does not.

These are the same codes used for
--line-up-parentheses-inclusion-list. For example,

-btct='f(b'

means that bare trailing commas in function call lists in the
input stream should be followed by line breaks in the formatted
output stream.

The section "Adding and Deleting Commas" describes additional
controls for working with trailing commas. These can be combined
with the -break-trailing-comma-types parameter for additional
control of list formatting.

-mft=n, --maximum-fields-per-table=n
If n is a positive number, and the computed number of fields for
any table exceeds n, then it will be reduced to n. This
parameter might be used on a small section of code to force a
list to have a particular number of fields per line, and then
either the -boc flag could be used to retain this formatting, or
a blank line or comment could be introduced somewhere to freeze
the formatting in future applications of perltidy. For example

# perltidy -mft=2
@month_of_year = (
'Jan', 'Feb',
'Mar', 'Apr',
'May', 'Jun',
'Jul', 'Aug',
'Sep', 'Oct',
'Nov', 'Dec'
);

The default value is n=0, which does not place a limit on the
number of fields in a table.

-cab=n, --comma-arrow-breakpoints=n
A comma which follows a comma arrow, '=>', is given special
consideration. In a long list, it is common to break at all such
commas. This parameter can be used to control how perltidy
breaks at these commas. (However, it will have no effect if old
comma breaks are being forced because -boc is used). The
possible values of n are:

n=0 break at all commas after =>
n=1 stable: break at all commas after => if container is open,
EXCEPT FOR one-line containers
n=2 break at all commas after =>, BUT try to form the maximum
one-line container lengths
n=3 do not treat commas after => specially at all
n=4 break everything: like n=0 but ALSO break a short container with
a => not followed by a comma when -vt=0 is used
n=5 stable: like n=1 but ALSO break at open one-line containers when
-vt=0 is used (default)

For example, given the following single line, perltidy by default
will not add any line breaks because it would break the existing
one-line container:

bless { B => $B, Root => $Root } => $package;

Using -cab=0 will force a break after each comma-arrow item:

# perltidy -cab=0:
bless {
B => $B,
Root => $Root
} => $package;

If perltidy is subsequently run with this container broken, then
by default it will break after each '=>' because the container is
now broken. To reform a one-line container, the parameter -cab=2
could be used.

The flag -cab=3 can be used to prevent these commas from being
treated specially. In this case, an item such as "01" => 31 is
treated as a single item in a table. The number of fields in
this table will be determined by the same rules that are used for
any other table. Here is an example.

# perltidy -cab=3
my %last_day = (
"01" => 31, "02" => 29, "03" => 31, "04" => 30,
"05" => 31, "06" => 30, "07" => 31, "08" => 31,
"09" => 30, "10" => 31, "11" => 30, "12" => 31
);

-qwaf, --qw-as-function
This option tells perltidy to format a qw list which is delimited
with parentheses as if it were a function call whose call args
are a list of quoted items. Normally, a qw list is output
verbatim except for an adjustment of leading whitespace to
indicate the indentation level. For example, here is an example
of the default formatting of a poorly formatted qw list:

# perltidy
@fields = qw( $st_dev $st_ino $st_mode $st_nlink $st_uid
$st_gid $st_rdev $st_size $st_atime $st_mtime $st_ctime
$st_blksize $st_blocks);

If we format with -qwaf then the result will be:

# perltidy -qwaf
@fields = qw(
$st_dev $st_ino $st_mode $st_nlink
$st_uid $st_gid $st_rdev $st_size
$st_atime $st_mtime $st_ctime $st_blksize
$st_blocks
);

The way this works is that just before formatting begins, the
tokens of the qw text are replaced with the tokens of an
equivalent function call with a comma-separated list of quoted
items as call args. Then it is formatted like any other list.
Special comma tokens are employed which have no display text, so
when the code is eventually displayed it remains a valid qw
quote.

Some things to note are:

+o This only works for qw quotes which begin with qw(, with no
space before the paren.

+o If the option --space-function-paren is employed, it is
ignored for these special function calls because it would
deactivate them.

+o Otherwise the various formatting control flags operate on
these lists the same as for other comma-separated lists. In
particular, note that if --break-at-old-comma-breakpoints, or
-boc, is set, then the old line break locations will be
retained. And old line breaks will be retained if there are
any blank lines between the opening and closing parens.

+o Before using this option for the first time, it is a good
idea to scan the code and decide if any lists have a special
order which should be retained. This can be accomplished for
example by changing the quote delimiters to something other
than parens, or by inserting a blank line as discussed at the
start of this section.

Adding and Deleting Commas

-drc, --delete-repeated-commas
This option causes repeated commas to be removed. For example,
given this list with a repeated comma

ignoreSpec( $file, "file",, \%spec, \%Rspec );

we can remove it with -drc

# perltidy -drc:
ignoreSpec( $file, "file", \%spec, \%Rspec );

This parameter also deletes repeated fat commas, '=>'. The
complete list of actions taken when this flag is set are as
follows:

+o Repeated commas like ',,' are removed with a warning.

+o Repeated fat commas like '=> =>' are removed with a warning.

+o The combination '=>,' produces a warning but is not changed
(it is likely an error but only its author would know how to
fix it).

+o The remaining combination ',=>' (sometimes called a 'winking
fat comma') is ignored by this parameter.

+o These warnings are only output if the --warning-output, or
-w, flag is set.

This option is on by default. Use -ndrc to turn it off.

Adding and Deleting Trailing Commas
A trailing comma is a comma following the last item of a list.
Perl allows trailing commas but they are not required. Including
them can sometimes simplify the maintenance of large or complex
lists, and help display structure. But they may not be
appropriate in all lists, for example in a list which always has
just one term. By default, perltidy does not add or delete
trailing commas, but it is possible to manipulate them with the
following set of related parameters:

+o --want-trailing-commas=s, -wtc=s - defines where trailing
commas are wanted (the style)

+o --add-trailing-commas, -atc - gives permission to add
trailing commas to match the style wanted

+o --delete-trailing-commas, -dtc - gives permission to delete
trailing commas which do not match the style wanted

The parameter --want-trailing-commas=s, or -wtc=s, defines a
preferred style. The string s indicates which lists should get
trailing commas, as follows:

s=1 or '*' : every list
s=m a multiline list
s=b a multiline list, bare trailing comma
s=i a multiline list, bare trailing comma, about one comma per line
s=h a multiline list, bare trailing comma, about one key=>value
pair per line
s=0 : no list

s=' ' or not defined : leave trailing commas unchanged [DEFAULT]

where:

+o A list here is basically taken to be a container of items
(parens, square brackets, or braces), which is not a code
block, which contains one or more commas or fat commas.
These parameters only apply to something that fits this
definition of a list.

A paren-less list of parameters is not a list by this
definition, so these parameters do not apply to a paren-less
list.

+o A multiline list is a list for which the opening and closing
brackets on different lines.

+o A bare trailing comma is a comma which is at the end of a
line. That is, the closing container token follows on a
different line. So a list with a bare trailing comma is a
special case of a multiline list.

+o In fact the above options for trailing commas can be seen as
a hierarchy of nesting sets which can be expressed as

1 > m > b > i > h > 0

This indicates that multiline trailing commas m are a subset
of all trailing commas, and bare trailing commas b are a
subset of all multiline trailing commas, and so on.

This parameter by itself only indicates where trailing commas are
wanted. Perltidy only adds these trailing commas if permission
is granted by setting the flag --add-trailing-commas, or -atc.
And perltidy only removes unwanted trailing commas if the flag
--delete-trailing-commas, or -dtc is set.

Here are some example parameter combinations and their meanings

-wtc=0 -dtc : delete all trailing commas
-wtc=1 -atc : add trailing commas to all lists
-wtc=m -atc : add trailing commas to all multiline lists
(single line lists remain unchanged)
-wtc=b -atc : add commas so that all lists whose closing
bracket starts a new line have trailing commas
-wtc=b -dtc : all trailing commas which are not bare
(not followed by a newline) get deleted.
-wtc=b -atc -dtc : do both of the above operations so that
all trailing commas are bare

For example, given the following input

$wine_list = $top->Box(
"-length" => 5,
"-width" => 3
)->select( "red", "white", "gold", );

we have

# perltidy -wtc=b -atc -dtc
$wine_list = $top->Box(
"-length" => 5,
"-width" => 3,
)->select( "red", "white", "gold" );

A comma was added after the 3, since it is bare, and a comma was
removed after "gold", since it not bare.

It is possible to apply a different style to each type of
container token by including an opening token ahead of the style
character in the above table. For example

-wtc='(m [b'

means that lists within parens should have multiline trailing
commas, and that lists within square brackets have bare trailing
commas. Since there is no specification for curly braces in this
example, their trailing commas would remain unchanged.

For parentheses, an additional item of information which can be
given is an alphanumeric letter which is used to limit the
selection further depending on the type of token immediately
before the opening paren. The possible letters are currently
'k', 'K', 'f', 'F', 'w', and 'W', with these meanings for
matching whatever precedes an opening paren:

'k' matches if the previous nonblank token is a perl keyword
(such as 'if', 'while'),
'K' matches if 'k' does not: previous token is not a keyword
'f' matches if previous token is a function (not a keyword)
'F' matches if 'f' does not.
'w' matches if either 'k' or 'f' match.
'W' matches if 'w' does not.

These are the same codes used for
--line-up-parentheses-inclusion-list. For example,

-wtc='w(m'

means that trailing commas are wanted for multiline parenthesized
lists following a function call or keyword.

Finally, a leading + can be placed on any term to indicate that
it only applies when adding commas. A leading - indicates that it
only applies when deleting commas. For example,

-wtc='+h -b' -atc -dtc

means that missing trailing commas should be added to lists of
key => value pairs, and trailing commas which are not bare should
be removed. No other changes are made. When both plus and minus
terms are used like this, they must not be in conflict. There is
no conflict in this example because the trailing comma locations
of the key=>value pairs selected by the +h term are a subset of
all bare trailing commas, and thus will not be deleted by the -b
term. The general rule is that the letter of the plus term
should occur after the letter of the minus term in the
hierarchical nesting order, 1 > m > b > i > h > 0.

Some points to note regarding adding and deleting trailing
commas:

+o It is recommended to also use the --converge parameter when
adding and/or deleting trailing commas, especially if the
formatter may be making other line break changes at the same
time. The reason is that the decision regarding whether or
not a list is multiline or bare is made based on the input
stream if only one iteration is made, which is the default.

When iterations are requested with the --converge parameter,
any comma deletion operations are postponed until the start
of the second iteration, after most changes in line breaks
have been made.

To illustrate, if we start with

f(
a => 1,
b => 2, );

and attempt to delete non-bare commas,

# perltidy -wtc=b -dtc
f(
a => 1,
b => 2
);

we delete a comma which has become bare, which is not what is
wanted. This happened because the change was based on the
starting rather than the final line breaks. Running with
--converge gives the desired result:

# perltidy -wtc=b -dtc --converge
f(
a => 1,
b => 2,
);

because comma changes are based on the line breaks after the
first iteration.

A parameter --delay-trailing-comma-operations, or -dtco, is
available to control this behavior if desired. Negating this
parameter, with -ndtco, tells perltidy to always use the
starting state to make decisions regarding comma addition and
deletion, even when iterations are requested. This should not
normally be necessary.

+o Perltidy does not add a trailing comma in some edge cases
which appear to be near a stability limit. So if a comma is
unexpectedly not added, this is probably the reason.

+o If the parameter --break-at-trailing-comma-types, or -btct.
is also employed, it operates on the state after any adding
or deleting of commas. And it will allow trailing commas to
be added in most edge cases. For example, given the
following input text

plot(
'g', Canvas => $overview_canvas
);

formatting with "-wtc=f(b) -atc" will not add a trailing
comma because the list will be flattened and the comma will
not remain bare. But we can add a trailing comma, and keep
the container open, with

# perltidy -wtc='f(b' -atc -btct='f(b'
plot(
'g', Canvas => $overview_canvas,
);

As another example, given the same text on a single line
without a trailing comma

plot( 'g', Canvas => $overview_canvas );

we can add a trailing comma and break the container open with

# perltidy -wtc=1 -atc -btct=1
plot(
'g', Canvas => $overview_canvas,
);

After that, we could use "-btct='f(b'" to keep the container
open.

+o When using these parameters for the first time it is a good
idea to practice on some test scripts and verify that the
results are as expected.

Special Considerations for Lone Trailing Commas

Adding or deleting the only comma in a list can have some
implications which need to be explained and possibly controlled.
Two additional controls are available for these lone commas:

+o --add-lone-trailing-commas, -altc - gives permission to add
a comma if it will be the only comma. This is on by default
and explained below.

+o --delete-lone-trailing-commas, -dltc - gives permission to
delete the only comma in a list. This is on by default and
explained below.

One issue with deleting a lone comma is that if it is deleted,
then it might not be possible add it back automatically since
perltidy uses the existence of commas to help locate containers
where commas are appropriate. For example, given

my ( $self, ) = @_;

and if we remove the comma with

# perltidy -wtc=m -dtc
my ( $self ) = @_;

then we cannot use the trailing comma controls to add this comma
back. The parameter --delete-lone-trailing-commas allows such a
comma to be deleted, and is on by default, but can be turned off
to prevent this. This might be useful if one is experimenting
with formatting options and wants to restrict testing to
operations which are reversible. Note that this parameter is a
fine-tuning control for --delete-trailing-commas which must also
be set for it to have any effect.

However, if a single item in a list is itself is a list with
multiple lines, such as the item in braces here

$self->make_grammar(
{
iterator => $self->_iterator,
parser => $self,
}
);

then perltidy can add and/or delete a lone comma:

# perltidy -atc -wtc=b
$self->make_grammar(
{
iterator => $self->_iterator,
parser => $self,
},
);

But it turns out that these cases usually coincide with
situations where the --weld-nested-containers, or -wn, parameter
would apply, and adding such commas can block welding. For
example, the -wn parameter would succeed on the first of the
above snippets, but it would fail on the second because of the
added trailing comma.

The parameter --add-lone-trailing-commas, or -altc allows these
commas to be added, provide that --add-trailing-commas is also
set. It is on by default. Users of -wn may want to turn it off
with --noadd-lone-trailing-commas, -naltc to prevent such commas
from being added.

If such commas do get added, then can be removed to allow welding
with the control described in the next section.

-dwic, --delete-weld-interfering-commas
If the closing tokens of two nested containers are separated by a
comma, then welding requested with --weld-nested-containers
cannot occur. Any commas in this situation are optional trailing
commas and can be removed with -dwic. For example, a comma in
this script prevents welding:

# perltidy -wn
$self->make_grammar(
{
iterator => $self->_iterator,
parser => $self,
},
);

Adding -dwic removes the comma and allows welding:

# perltidy -wn -dwic
$self->make_grammar( {
iterator => $self->_iterator,
parser => $self,
} );

This feature is off by default. Here are some points to note
about the -dwic parameter

+o This operation is not always reversible, so please check
results of using this parameter carefully.

+o Removing this type of isolated trailing comma is necessary
for welding to be possible, but not sufficient. So welding
will not always occur where these commas are removed.

+o This operation is independent of --add-trailing-commas and
--delete-trailing-commas. If it conflicts with any of those
settings, it has priority.

Adding and Deleting Interbracket Arrows

In the following expression, the arrow operator '->' between the
closing and opening brackets of hash keys and array indexes are
optional:

return $self->{'commandline'}->{'args'}->[0]->[0]->{'hgroups'};

These will be called interbracket arrows here, for lack of a better
term. Perltidy will not change them by default, but they can be
added or removed with the following parameters.

-dia, --delete-interbracket-arrows
This parameter deletes interbracket arrows. Applied to the above
example we have

# perltidy -dia
return $self->{'commandline'}{'args'}[0][0]{'hgroups'};

By default this applies to all interbracket arrows, but selective
deletion is possible with controls described below.

-aia, --add-interbracket-arrows
This parameter adds interbracket arrows. Applied to the line of
code above, we get back the original line.

# perltidy -aia
return $self->{'commandline'}->{'args'}->[0]->[0]->{'hgroups'};

Selective changes can be made with controls described below.

-ias=s, --interbracket-arrow-style=s
By default the -add- and -delete- parameters apply to all
interbracket arrows.

An optional style can be specified with this parameter string s.
In that case the parameters --add-interbracket-arrows and
--delete-interbracket-arrows only apply where they would bring
the formatting into agreement with the specified style. They may
both be used in a single run if a mixed style is specified since
there is no conflict.

The style string s gives a graphical description of the desired
style. It lists up to four possible pairs of bracket types with
an optional arrow between them. For example:

-ias='][ }->{'

This means no arrows are wanted between '][' but arrows should be
between '}{'. And it means that the unlisted pairs ']{' and '}['
should remain unchanged, either with or without arrows.

In this particular example, if the parameter
--delete-interbracket-arrows is used, then only arrows like
']->[' will be deleted, since they are the only ones which
disagree with the style.

And likewise, if --add-interbracket-arrows is used, then arrows
will only be inserted between brackets like '}{' to bring the
formatting into conformity with the style in this example.

Spaces in the string s are optional. They are ignored when the
expression is parsed.

The style corresponding to all possible arrows is

-ias=']->[ ]->{ }->[ }->{'

For convenience, this may also be requested with -ias=1 or
-ias='*'.

The style corresponding to no interbracket arrows is

-ias='] [ ] { } [ } {'

which may also be requested with -ias=0.

-wia, --warn-interbracket-arrows
If this parameter is set, then a message is written to the error
file in the following cases:

+o If an arrow is added or deleted by an add or delete command.

+o If a style is defined and an arrow would have been added or
deleted if requested. So for example, the command

perltidy -wia -ias=']['

will show where a file has arrows like ]->[' since they do
not match the style, but no changes will be made because the
delete command -dia has not been given. And

perltidy -wia -ias=0

will warn if any arrows exist, since the flag -ias=0 means
that no arrows are wanted.

-iac=n, --interbracket-arrow-complexity=n
This parameter makes it possible to skip adding or deleting
arrows following a container which is complex in some sense.
Three levels of complexity can be specified with the integer n,
as follows:

n=0 the left container must contain be a single thing (token)
n=1 the left container must not contain other containers [DEFAULT]
n=2 the left container may contain anything

Some examples:

# Container complexity
{'commandline'} 0 single token OK by default
{ $type . $name } 1 multiple tokens OK by default
[ $plot{'x-axis'} - 1 ] 2 contains a container SKIPPED by default

So, with the default complexity level of 1, an arrow could be
added or deleted following the first two of these containers but
not the third.

Some points to consider when working with these parameters are:

+o There are no known bugs, but this is a relatively new feature. So
please carefully check file differences and run tests when
interbracket arrows are added or removed.

+o For some unusual spacing parameters, it could take an extra
iteration for the spaces between brackets to reach their final
state after arrows are added or deleted.

+o Any comments between brackets will prevent the adding and
deleting of arrows.

Missing Else Blocks

A defensive programming technique is to require that every if-elsif-
chain be terminated with an else block, even though it is not
strictly required. This helps insure that there are no holes in the
logic.

For example, consider the following snippet:

my $type = get_cards();
if ( $type = 1 ) { action("hold 'em") }
elsif ( $type = 2 ) { action("fold 'em") }
elsif ( $type = 3 ) { action("walk away") }

What if the variable $type is some other value? It might have been
obvious that this was okay when the code was first written, but it
might not be so clear when the code is reviewed a few years later. A
terminal else block with a comment would help clarify things.

The parameters in this section can help by either issuing a warning
if an else is missing, or even inserting an empty else block where
one is missing, or both.

-wme, --warn-missing-else
This flag tells perltidy to issue a warning if a program is
missing a terminal else block. The default is not to issue such
a warning.

-ame, --add-missing-else
This flag tells perltidy to output an empty else block wherever a
program is missing a terminal else block. To get a warning when
this is done you should also set -wme. The default is not to add
missing else blocks.

-amec=s, --add-missing-else-comment=s
This string is a side comment which will be written to highlight
a new empty else block. The default is:

-amec='##FIXME - added with perltidy -ame'

For example, on the above example we can add a missing else and also
get a warning notice with:

# perltidy -ame -wme
my $type = get_cards();
if ( $type == 1 ) { action("hold 'em") }
elsif ( $type == 2 ) { action("fold 'em") }
elsif ( $type == 3 ) { action("walk away") }
else {
##FIXME - added with perltidy -ame
}

Any ##FIXME comments created in this way should be reviewed and
changed appropriately. For example, one might decide that the code
fine as is, and just change the comment to indicate that nothing has
been overlooked:

my $type = get_cards();
if ( $type == 1 ) { action("hold 'em") }
elsif ( $type == 2 ) { action("fold 'em") }
elsif ( $type == 3 ) { action("walk away") }
else {
# ok - no worries
}

Or maybe a deeper analysis reveals that something was missed:

my $type = get_cards();
if ( $type == 1 ) { action("hold 'em") }
elsif ( $type == 2 ) { action("fold 'em") }
elsif ( $type == 3 ) { action("walk away") }
else { action("run") }

Sometimes it turns out that the else block should not reachable, in
which case an error exit might be appropriate. In any case, having
the else block can improve code maintainability.

Controlling Breaks at Certain Chain Operators

For a statment which consists of a chain of certain binary operators,
the default line break logic is: (1) if a statement fits on one line,
then it is formatted as one line, or (2) if the statement requires
multiple lines, then a break is made at all of the chain operators.
Additional rules apply, but this is the basic one.

The parameter --pack-operator-types=s, or -pot=s, can be used to
change this for certain operators which may be specified with the
string s. These are the concatenation operator s='.' and the method
call arrow s='->'. When this parameter is selected for an operator,
no special breaks are made, so there will be a minimal number of line
breaks. For example, here is the default formatting for a method
call chain:

# perltidy -l=60
Data::Dumper->new( [ \%errors ] )
->Indent(2)
->Terse(1)
->Sortkeys(1)
->Dump;

and this is the alternative formatting with this parameter:

# perltidy -l=60 -pot='->'
Data::Dumper->new( [ \%errors ] )->Indent(2)->Terse(1)
->Sortkeys(1)->Dump;

Note that this option only applies to arrows which follow a closing
paren.

Retaining or Ignoring Existing Line Breaks

Several additional parameters are available for controlling the
extent to which line breaks in the input script influence the output
script. In most cases, the default parameter values are set so that,
if a choice is possible, the output style follows the input style.
For example, if a short logical container is broken in the input
script, then the default behavior is for it to remain broken in the
output script.

Most of the parameters in this section would only be required for a
one-time conversion of a script from short container lengths to
longer container lengths. The opposite effect, of converting long
container lengths to shorter lengths, can be obtained by temporarily
using a short maximum line length.

-bol, --break-at-old-logical-breakpoints
By default, if a logical expression is broken at a "&&", "||",
"and", or "or", then the container will remain broken. Also,
breaks at internal keywords "if" and "unless" will normally be
retained. To prevent this, and thus form longer lines, use
-nbol.

Please note that this flag does not duplicate old logical
breakpoints. They are merely used as a hint with this flag that
a statement should remain broken. Without this flag, perltidy
will normally try to combine relatively short expressions into a
single line.

For example, given this snippet:

return unless $cmd = $cmd || ($dot
&& $Last) || &prompt('|');

# perltidy -bol [default]
return
unless $cmd = $cmd
|| ( $dot
&& $Last )
|| &prompt('|');

# perltidy -nbol
return unless $cmd = $cmd || ( $dot && $Last ) || &prompt('|');

-bom, --break-at-old-method-breakpoints
By default, a method call arrow "->" is considered a candidate
for a breakpoint, but method chains will fill to the line width
before a break is considered. With -bom, breaks before the arrow
are preserved, so if you have pre-formatted a method chain:

# perltidy -bom
$Document
->schild(0)
->schildren();

the flag -bom will keep these line breaks, rather than become
this:

# perltidy [DEFAULT]
$Document->schild(0)->schildren();

This flag will also look for and keep a 'cuddled' style of calls,
in which lines begin with a closing paren followed by a call
arrow, as in this example:

# perltidy -bom
my $q = $rs->related_resultset(
'CDs'
)->related_resultset(
'Tracks'
)->search(
{
'track.id' => { -ident => 'none_search.id' },
}
)->as_query;

-bos, --break-at-old-semicolon-breakpoints
Semicolons are normally placed at the end of a statement. This
means that formatted lines do not normally begin with semicolons.
If the input stream has some lines which begin with semicolons,
these can be retained by setting this flag. For example,
consider the following two-line input snippet:

$z = sqrt($x**2 + $y**2)
;

The default formatting will be:

$z = sqrt( $x**2 + $y**2 );

Using the -bos flag keeps the isolated semicolon:

# perltidy -bos
$z = sqrt( $x**2 + $y**2 )
;

The extra continuation indentation spaces on the semicolon can be
removed by also setting --noindent-leading-semicolon.

# perltidy -bos -nils
$z = sqrt( $x**2 + $y**2 )
;

The default is not to do this, -nbos.

-bok, --break-at-old-keyword-breakpoints
By default, perltidy will retain a breakpoint before keywords
which may return lists, such as "sort" and "map". This allows
chains of these operators to be displayed one per line. Use
-nbok to prevent retaining these breakpoints.

-bot, --break-at-old-ternary-breakpoints
By default, if a conditional (ternary) operator is broken at a
":", then it will remain broken. To prevent this, and thereby
form longer lines, use -nbot.

-boa, --break-at-old-attribute-breakpoints
By default, if an attribute list is broken at a ":" in the source
file, then it will remain broken. For example, given the
following code, the line breaks at the ':'s will be retained:

my @field
: field
: Default(1)
: Get('Name' => 'foo') : Set('Name');

If the attributes are on a single line in the source code then
they will remain on a single line if possible.

To prevent this, and thereby always form longer lines, use -nboa.

Keeping old breakpoints at specific token types
It is possible to override the choice of line breaks made by
perltidy, and force it to follow certain line breaks in the input
stream, with these two parameters:

-kbb=s or --keep-old-breakpoints-before=s, and

-kba=s or --keep-old-breakpoints-after=s

These parameters are each followed by a quoted string, s,
containing a list of token types (separated only by spaces). No
more than one of each of these parameters should be specified,
because repeating a command-line parameter always overwrites the
previous one before perltidy ever sees it.

For example, -kbb='=>' means that if an input line begins with a
'=>' then the output script should also have a line break before
that token.

For example, given the script:

method 'foo'
=> [ Int, Int ]
=> sub {
my ( $self, $x, $y ) = ( shift, @_ );
...;
};

# perltidy [default]
method 'foo' => [ Int, Int ] => sub {
my ( $self, $x, $y ) = ( shift, @_ );
...;
};

# perltidy -kbb='=>'
method 'foo'
=> [ Int, Int ]
=> sub {
my ( $self, $x, $y ) = ( shift, @_ );
...;
};

For the container tokens '{', '[' and '(' and, their closing
counterparts, use the token symbol. Thus, the command to keep a
break after all opening parens is:

perltidy -kba='('

It is possible to be more specific in matching parentheses by
preceding them with a letter. The possible letters are 'k', 'K',
'f', 'F', 'w', and 'W', with these meanings (these are the same
as used in the --weld-nested-exclusion-list and
--line-up-parentheses-exclusion-list parameters):

'k' matches if the previous nonblank token is a perl keyword
(such as 'if', 'while'),
'K' matches if 'k' does not: previous token is not a keyword
'f' matches if previous token is a function (not a keyword)
'F' matches if 'f' does not.
'w' matches if either 'k' or 'f' match.
'W' matches if 'w' does not.

So for example the the following parameter will keep breaks after
opening function call parens:

perltidy -kba='f('

NOTE: A request to break before an opening container, such as
-kbb='(', will be silently ignored because it can lead to
formatting instability. Likewise, a request to break after a
closing container, such as -kba=')', will also be silently
ignored.

-iob, --ignore-old-breakpoints
Use this flag to tell perltidy to ignore existing line breaks to
the maximum extent possible. This will tend to produce the
longest possible containers, regardless of type, which do not
exceed the line length limit. But please note that this
parameter has priority over all other parameters requesting that
certain old breakpoints be kept.

To illustrate, consider the following input text:

has subcmds => (
is => 'ro',
default => sub { [] },
);

The default formatting will keep the container broken, giving

# perltidy [default]
has subcmds => (
is => 'ro',
default => sub { [] },
);

If old breakpoints are ignored, the list will be flattened:

# perltidy -iob
has subcmds => ( is => 'ro', default => sub { [] }, );

Besides flattening lists, this parameter also applies to lines
broken at certain logical breakpoints such as 'if' and 'or'.

Even if this is parameter is not used globally, it provides a
convenient way to flatten selected lists from within an editor.

-kis, --keep-interior-semicolons
Use the -kis flag to prevent breaking at a semicolon if there was
no break there in the input file. Normally perltidy places a
newline after each semicolon which terminates a statement unless
several statements are contained within a one-line brace block.
To illustrate, consider the following input lines:

dbmclose(%verb_delim); undef %verb_delim;
dbmclose(%expanded); undef %expanded;

The default is to break after each statement, giving

dbmclose(%verb_delim);
undef %verb_delim;
dbmclose(%expanded);
undef %expanded;

With perltidy -kis the multiple statements are retained:

dbmclose(%verb_delim); undef %verb_delim;
dbmclose(%expanded); undef %expanded;

The statements are still subject to the specified value of
maximum-line-length and will be broken if this maximum is
exceeded.

Blank Line Control

Blank lines can improve the readability of a script if they are
carefully placed. Perltidy has several commands for controlling the
insertion, retention, and removal of blank lines.

-fbl, --freeze-blank-lines
Set -fbl if you want to the blank lines in your script to remain
exactly as they are. The rest of the parameters in this section
may then be ignored. (Note: setting the -fbl flag is equivalent
to setting -mbl=0 and -kbl=2).

-bbc, --blanks-before-comments
A blank line will be introduced before a full-line comment. This
is the default. Use -nbbc or --noblanks-before-comments to
prevent such blank lines from being introduced.

-blbs=n, --blank-lines-before-subs=n
The parameter -blbs=n requests that least n blank lines precede a
sub definition which does not follow a comment and which is more
than one-line long. The default is -blbs=1. BEGIN and END
blocks are included.

The requested number of blanks statement will be inserted
regardless of the value of --maximum-consecutive-blank-lines=n
(-mbl=n) with the exception that if -mbl=0 then no blanks will be
output.

This parameter interacts with the value k of the parameter
--maximum-consecutive-blank-lines=k (-mbl=k) as follows:

1. If -mbl=0 then no blanks will be output. This allows all
blanks to be suppressed with a single parameter. Otherwise,

2. If the number of old blank lines in the script is less than n
then additional blanks will be inserted to make the total n
regardless of the value of -mbl=k.

3. If the number of old blank lines in the script equals or
exceeds n then this parameter has no effect, however the total
will not exceed value specified on the -mbl=k flag.

-blbp=n, --blank-lines-before-packages=n
The parameter -blbp=n requests that least n blank lines precede a
package which does not follow a comment. The default is -blbp=1.

This parameter interacts with the value k of the parameter
--maximum-consecutive-blank-lines=k (-mbl=k) in the same way as
described for the previous item -blbs=n.

-bbs, --blanks-before-subs
For compatibility with previous versions, -bbs or
--blanks-before-subs is equivalent to -blbp=1 and -blbs=1.

Likewise, -nbbs or --noblanks-before-subs is equivalent to
-blbp=0 and -blbs=0.

-bbb, --blanks-before-blocks
A blank line will be introduced before blocks of coding delimited
by for, foreach, while, until, and if, unless, in the following
circumstances:

+o The block is not preceded by a comment.

+o The block is not a one-line block.

+o The number of consecutive non-blank lines at the current
indentation depth is at least -lbl (see next section).

This is the default. The intention of this option is to
introduce some space within dense coding. This is negated with
-nbbb or --noblanks-before-blocks.

-lbl=n --long-block-line-count=n
This controls how often perltidy is allowed to add blank lines
before certain block types (see previous section). The default
is 8. Entering a value of 0 is equivalent to entering a very
large number.

-blao=i or --blank-lines-after-opening-block=i
This control places a minimum of i blank lines after a line which
ends with an opening block brace of a specified type. By
default, this only applies to the block of a named sub, but this
can be changed (see -blaol below). The default is not to do this
(i=0).

Please see the note below on using the -blao and -blbc options.

-blbc=i or --blank-lines-before-closing-block=i
This control places a minimum of i blank lines before a line
which begins with a closing block brace of a specified type. By
default, this only applies to the block of a named sub, but this
can be changed (see -blbcl below). The default is not to do this
(i=0).

-blaol=s or --blank-lines-after-opening-block-list=s
The parameter s is a list of block type keywords to which the
flag -blao should apply. The section "Specifying Block Types"
explains how to list block types.

-blbcl=s or --blank-lines-before-closing-block-list=s
This parameter is a list of block type keywords to which the flag
-blbc should apply. The section "Specifying Block Types"
explains how to list block types.

Note on using the -blao and -blbc options.
These blank line controls introduce a certain minimum number of
blank lines in the text, but the final number of blank lines may
be greater, depending on values of the other blank line controls
and the number of old blank lines. A consequence is that
introducing blank lines with these and other controls cannot be
exactly undone, so some experimentation with these controls is
recommended before using them.

If necessary, it is possible to approximately undo these and
other blank line controls with the parameter
--keep-old-blank-lines-exceptions=s described below.

-mbl=n --maximum-consecutive-blank-lines=n
This parameter specifies the maximum number of consecutive blank
lines which will be output within code sections of a script. The
default is n=1. If the input file has more than n consecutive
blank lines, the number will be reduced to n except as noted
above for the --blank-lines-before-subs and
--blank-lines-before-subs parameters. If n=0 then no blank lines
will be output (unless all old blank lines are retained with the
--keep-old-blank-lines=2 flag of the next section).

This flag obviously does not apply to pod sections, here-
documents, and quotes.

-kbl=n, --keep-old-blank-lines=n
The -kbl=n flag gives you control over how your existing blank
lines are treated.

The possible values of n are:

n=0 ignore all old blank lines
n=1 stable: keep old blanks, but limited by the B<-mbl=n> flag, but
remove existing blanks matching the B<--kblx=s> flag (next section)
n=2 keep all old blank lines, regardless of the B<-mbl=n> flag

The default and recommended setting is n=1. With this option, you
may selectively filter out selected blank lines from the input
stream before formatting begins using the parameter described in
the next section.

The setting n=0 should be used with caution because it will
remove all blank lines, including lines which might be essential
for controlling vertical alignment, and for preventing unwanted
hanging side comments.

-kblx=s, --keep-old-blank-lines-exceptions=s
Most of the blank line controls described above specify a minimum
number of blank lines, rather than a maximum, so they cannot be
used remove lines. This parameter allows blank lines to be
removed by selectively ignoring them in the input stream. It
operates when --keep-old-blank-lines=1, which is the default as
described above. The selection of blank lines to exclude is made
with a string with one or more of the following special values:

'b{' 'b}' '{b' '}b' 'bp' 'bs' 'bc' 'cb'

In these symbols, the character b indicates the location of blank
lines with respect to the next or previous token type. The token
types { and } are opening and closing code blocks, the character
p is a package statement, s is a sub definition, and c is a block
comment.

So for example bs means one or more blank lines before a sub will
be ignored in the input stream. Note that this is approximately
the opposite of the parameter -blanks-before-subs described
above.

In fact, these parameters provide approximate inverses to all of
the previously described blank line options. To simplify things,
the block types to which the "{" and "}" operators apply are the
same as the corresponding operator which adds blank lines,
described above. The following table shows these relationships:

kblx= Inverse Applies to block types
b{ -bbb if unless while until for foreach
b} -blbc=i -blbcl=s [default = sub]
{b -blao=i -blaol=s [default = sub]
}b [none] -blaol=s
bp -blbp=n
bs -blbs=n
bc -bbc
cb [none - comments before blanks]

So, for example, the table indicates that to approximately remove
the blank lines introduced by a -blao operation we could use
-kblx='{b'.

Some notes on using this parameter:

+o The parameter string may contain multiple operators provided
that they are space-separated.

+o The comment deletion controls will keep at least one blank
line between groups of comments, and between a side comment
and following line comment.

+o These operations occur on the input stream, before other
blank line operations, so they may be used in the same run as
other blank line controls. So it can happen that these
controls will remove blanks from the input stream while other
controls add blank lines back.

+o These operations will normally only be needed occasionally to
correct some problems, or undo an unwanted operation.

-sob, --swallow-optional-blank-lines
This is equivalent to --keep-old-blank-lines=0 and is included
for compatibility with previous versions.

-nsob, --noswallow-optional-blank-lines
This is equivalent to --keep-old-blank-lines=1 and is included
for compatibility with previous versions.

Controls for blank lines around lines of consecutive keywords

The parameters in this section provide some control over the
placement of blank lines within and around groups of statements
beginning with selected keywords. These blank lines are called here
keyword group blanks, and all of the parameters begin with
--keyword-group-blanks*, or -kgb* for short. The default settings do
not employ these controls but they can be enabled with the following
parameters:

-kgbl=s or --keyword-group-blanks-list=s; s is a quoted string of
keywords

-kgbs=s or --keyword-group-blanks-size=s; s gives the number of
keywords required to form a group.

-kgbb=n or --keyword-group-blanks-before=n; n = (0, 1, or 2) controls
a leading blank

-kgba=n or --keyword-group-blanks-after=n; n = (0, 1, or 2) controls
a trailing blank

-kgbi or --keyword-group-blanks-inside is a switch for adding blanks
between subgroups

-kgbd or --keyword-group-blanks-delete is a switch for removing
initial blank lines between keywords

-kgbr=n or --keyword-group-blanks-repeat-count=n can limit the number
of times this logic is applied

In addition, the following abbreviations are available to for
simplified usage:

-kgb or --keyword-group-blanks is short for -kgbb=2 -kgba=2 kgbi

-nkgb or --nokeyword-group-blanks, is short for -kgbb=1 -kgba=1 nkgbi

Before describing the meaning of the parameters in detail let us look
at an example which is formatted with default parameter settings.

print "Entering test 2\n";
use Test;
use Encode qw(from_to encode decode
encode_utf8 decode_utf8
find_encoding is_utf8);
use charnames qw(greek);
my @encodings = grep( /iso-?8859/, Encode::encodings() );
my @character_set = ( '0' .. '9', 'A' .. 'Z', 'a' .. 'z' );
my @source = qw(ascii iso8859-1 cp1250);
my @destiny = qw(cp1047 cp37 posix-bc);
my @ebcdic_sets = qw(cp1047 cp37 posix-bc);
my $str = join( '', map( chr($_), 0x20 .. 0x7E ) );
return unless ($str);

using perltidy -kgb gives:

print "Entering test 2\n";
<----------this blank controlled by -kgbb
use Test;
use Encode qw(from_to encode decode
encode_utf8 decode_utf8
find_encoding is_utf8);
use charnames qw(greek);
<---------this blank controlled by -kgbi
my @encodings = grep( /iso-?8859/, Encode::encodings() );
my @character_set = ( '0' .. '9', 'A' .. 'Z', 'a' .. 'z' );
my @source = qw(ascii iso8859-1 cp1250);
my @destiny = qw(cp1047 cp37 posix-bc);
my @ebcdic_sets = qw(cp1047 cp37 posix-bc);
my $str = join( '', map( chr($_), 0x20 .. 0x7E ) );
<----------this blank controlled by -kgba
return unless ($str);

Blank lines have been introduced around the my and use sequences.
What happened is that the default keyword list includes my and use
but not print and return. So a continuous sequence of nine my and
use statements was located. This number exceeds the default
threshold of five, so blanks were placed before and after the entire
group. Then, since there was also a subsequence of six my lines, a
blank line was introduced to separate them.

Finer control over blank placement can be achieved by using the
individual parameters rather than the -kgb flag. The individual
controls are as follows.

-kgbl=s or --keyword-group-blanks-list=s, where s is a quoted string,
defines the set of keywords which will be formed into groups. The
string is a space separated list of keywords. The default set is
s="use require local our my", but any list of keywords may be used.
Comment lines may also be included in a keyword group, even though
they are not keywords. To include ordinary block comments, include
the symbol BC. To include static block comments (which normally begin
with '##'), include the symbol SBC.

-kgbs=s or --keyword-group-blanks-size=s, where s is a string
describing the number of consecutive keyword statements forming a
group (Note: statements separated by blank lines in the input file
are considered consecutive for purposes of this count). If s is an
integer then it is the minimum number required for a group. A
maximum value may also be given with the format s=min.max, where min
is the minimum number and max is the maximum number, and the min and
max values are separated by one or more dots. No groups will be
found if the maximum is less than the minimum. The maximum is
unlimited if not given. The default is s=5. Some examples:

s min max number for group
3 3 unlimited 3 or more
1.1 1 1 1
1..3 1 3 1 to 3
1.0 1 0 (no match)

There is no really good default value for this parameter. If it is
set too small, then an excessive number of blank lines may be
generated. However, some users may prefer reducing the value
somewhat below the default, perhaps to s=3.

-kgbb=n or --keyword-group-blanks-before=n specifies whether a blank
should appear before the first line of the group, as follows:

n=0 => (delete) an existing blank line will be removed
n=1 => (stable) no change to the input file is made [DEFAULT]
n=2 => (insert) a blank line is introduced if possible

-kgba=n or --keyword-group-blanks-after=n likewise specifies whether
a blank should appear after the last line of the group, using the
same scheme (0=delete, 1=stable, 2=insert).

-kgbi or --keyword-group-blanks-inside controls the insertion of
blank lines between the first and last statement of the entire group.
If there is a continuous run of a single statement type with more
than the minimum threshold number (as specified with -kgbs=s) then
this switch causes a blank line be inserted between this subgroup and
the others. In the example above this happened between the use and my
statements.

-kgbd or --keyword-group-blanks-delete controls the deletion of any
blank lines that exist in the the group when it is first scanned.
When statements are initially scanned, any existing blank lines are
included in the collection. Any such original blank lines will be
deleted before any other insertions are made when the parameter -kgbd
is set. The default is not to do this, -nkgbd.

-kgbr=n or --keyword-group-blanks-repeat-count=n specifies n, the
maximum number of times this logic will be applied to any file. The
special value n=0 is the same as n=infinity which means it will be
applied to an entire script [Default]. A value n=1 could be used to
make it apply just one time for example. This might be useful for
adjusting just the use statements in the top part of a module for
example.

-kgb or --keyword-group-blanks is an abbreviation equivalent to
setting -kgbb=1 -kgba=1 -kgbi. This turns on keyword group
formatting with a set of default values.

-nkgb or --nokeyword-group-blanks is equivalent to -kgbb=0 -kgba
nkgbi. This flag turns off keyword group blank lines and is the
default setting.

Here are a few notes about the functioning of this technique.

+o These parameters are probably more useful as part of a major code
reformatting operation rather than as a routine formatting
operation.

In particular, note that deleting old blank lines with -kgbd is
an irreversible operation so it should be applied with care.
Existing blank lines may be serving an important role in
controlling vertical alignment.

+o Conflicts which arise among these kgb* parameters and other blank
line controls are generally resolved by producing the maximum
number of blank lines implied by any parameter.

For example, if the flags --freeze-blank-lines, or
--keep-old-blank-lines=2, are set, then they have priority over
any blank line deletion implied by the -kgb flags of this
section, so no blank lines will be deleted.

For another example, if a keyword group ends at a sub and the
flag kgba=0 requests no blank line there, but we also have
--blank-lines-before-subs=2, then two blank lines will still be
introduced before the sub.

+o The introduction of blank lines does not occur if it would
conflict with other input controls or code validity. For example,
a blank line will not be placed within a here-doc or within a
section of code marked with format skipping comments. And in
general, a blank line will only be introduced at the end of a
group if the next statement is a line of code.

+o The count which is used to determine the group size is not the
number of lines but rather the total number of keywords which are
found. Individual statements with a certain leading keyword may
continue on multiple lines, but if any of these lines is nested
more than one level deep then that group will be ended.

+o The search for groups of lines with similar leading keywords is
based on the input source, not the final formatted source.
Consequently, if the source code is badly formatted, it would be
best to make a first formatting pass without these options.

Styles

A style refers to a convenient collection of existing parameters.

-gnu, --gnu-style
-gnu gives an approximation to the GNU Coding Standards (which do
not apply to perl) as they are sometimes implemented. At
present, this style overrides the default style with the
following parameters:

-lp -bl -noll -pt=2 -bt=2 -sbt=2 -icp

To use this style with -xlp (--extended-line-up-parentheses)
instead of -lp (--line-up-parentheses) use -gnu -xlp.

-pbp, --perl-best-practices
-pbp is an abbreviation for the parameters in the book Perl Best
Practices by Damian Conway:

-l=78 -i=4 -ci=4 -st -se -vt=2 -cti=0 -pt=1 -bt=1 -sbt=1 -bbt=1
-nsfs -nolq
-wbb="% + - * / x != == >= <= =~ !~ < > | & =
**= += *= &= <<= &&= -= /= |= >>= ||= //= .= %= ^= x="

Please note that this parameter set includes -st
(--standard-output) and -se (--standard-error-output) flags,
which make perltidy act as a filter on one file only. These can
be overridden by placing -nst and/or -nse after the -pbp
parameter.

Also note that the value of continuation indentation, -ci=4, is
equal to the value of the full indentation, -i=4. It is
recommended that the either (1) the parameter -ci=2 be used
instead, or (2) the flag -xci be set. This will help show
structure, particularly when there are ternary statements. The
following snippet illustrates these options.

# perltidy -pbp
$self->{_text} = (
!$section ? ''
: $type eq 'item' ? "the $section entry"
: "the section on $section"
)
. (
$page
? ( $section ? ' in ' : '' ) . "the $page_ext manpage"
: ' elsewhere in this document'
);

# perltidy -pbp -ci=2
$self->{_text} = (
!$section ? ''
: $type eq 'item' ? "the $section entry"
: "the section on $section"
)
. (
$page
? ( $section ? ' in ' : '' ) . "the $page_ext manpage"
: ' elsewhere in this document'
);

# perltidy -pbp -xci
$self->{_text} = (
!$section ? ''
: $type eq 'item' ? "the $section entry"
: "the section on $section"
)
. ( $page
? ( $section ? ' in ' : '' ) . "the $page_ext manpage"
: ' elsewhere in this document'
);

The -xci flag was developed after the -pbp parameters were
published so you need to include it separately.

Making a file unreadable
The goal of perltidy is to improve the readability of files, but
there are two commands which have the opposite effect, --mangle
and --extrude. They are actually merely aliases for combinations
of other parameters. Both of these strip all possible
whitespace, but leave comments and pod documents, so that they
are essentially reversible. The difference between these is that
--mangle puts the fewest possible line breaks in a script while
--extrude puts the maximum possible. Note that these options do
not provided any meaningful obfuscation, because perltidy can be
used to reformat the files. They were originally developed to
help test the tokenization logic of perltidy, but they have other
uses. One use for --mangle is the following:

perltidy --mangle myfile.pl -st | perltidy -o myfile.pl.new

This will form the maximum possible number of one-line blocks
(see next section), and can sometimes help clean up a badly
formatted script.

A similar technique can be used with --extrude instead of
--mangle to make the minimum number of one-line blocks.

Another use for --mangle is to combine it with
--delete-all-comments (-dac) to reduce the file size of a perl
script.

One-Line Blocks
A one-line block is a block of code where the contents within the
curly braces is short enough to fit on a single line. For example,

if ( -e $file ) { print "'$file' exists\n" }

The alternative, a block which spans multiple lines, is said to be a
broken block. With few exceptions, perltidy retains existing one-
line blocks, if it is possible within the line-length constraint, but
it does not attempt to form new ones. In other words, perltidy will
try to follow the input file regarding broken and unbroken blocks.

The main exception to this rule is that perltidy will attempt to form
new one-line blocks following the keywords "map", "eval", and "sort",
"eval", because these code blocks are often small and most clearly
displayed in a single line. This behavior can be controlled with the
flag --one-line-block-exclusion-list described below.

When the cuddled-else style is used, the default treatment of one-
line blocks may interfere with the cuddled style. In this case, the
default behavior may be changed with the flag
--cuddled-break-option=n described elsewhere.

When an existing one-line block is longer than the maximum line
length, and must therefore be broken into multiple lines, perltidy
checks for and adds any optional terminating semicolon (unless the
-nasc option is used) if the block is a code block.

-olbxl=s, --one-line-block-exclusion-list=s
As noted above, perltidy will, by default, attempt to create new
one-line blocks for certain block types. This flag allows the
user to prevent this behavior for the block types listed in the
string s. The list s may include any of the words "sort", "map",
"grep", "eval", or it may be "*" to indicate all of these.

So for example to prevent multiline eval blocks from becoming
one-line blocks, the command would be -olbxl='eval'. In this
case, existing one-line eval blocks will remain on one-line if
possible, and existing multiline eval blocks will remain
multiline blocks.

-olbn=n, --one-line-block-nesting=n
Nested one-line blocks are lines with code blocks which
themselves contain code blocks. For example, the following line
is a nested one-line block.

foreach (@list) { if ($_ eq $asked_for) { last } ++$found }

The default behavior is to break such lines into multiple lines,
but this behavior can be controlled with this flag. The values
of n are:

n=0 break nested one-line blocks into multiple lines [DEFAULT]
n=1 stable: keep existing nested-one line blocks intact

For the above example, the default formatting (-olbn=0) is

foreach (@list) {
if ( $_ eq $asked_for ) { last }
++$found;
}

If the parameter -olbn=1 is given, then the line will be left
intact if it is a single line in the source, or it will be broken
into multiple lines if it is broken in multiple lines in the
source.

-olbs=n, --one-line-block-semicolons=n
This flag controls the placement of semicolons at the end of one-
line blocks. Semicolons are optional before a closing block
brace, and frequently they are omitted at the end of a one-line
block containing just a single statement. By default, perltidy
follows the input file regarding these semicolons, but this
behavior can be controlled by this flag. The values of n are:

n=0 remove terminal semicolons in single-statement one-line blocks
n=1 stable; keep input terminal semicolons [DEFAULT ]
n=2 add terminal semicolons in all one-line blocks

Note that the n=2 option has no effect if adding semicolons is
prohibited with the -nasc flag. Also not that while n=2 adds
missing semicolons to all one-line blocks, regardless of
complexity, the n=0 option only removes ending semicolons which
terminate one-line blocks containing just one semicolon. So
these two options are not exact inverses.

Forming new one-line blocks
Sometimes it might be desirable to convert a script to have one-
line blocks whenever possible. Although there is currently no
flag for this, a simple workaround is to execute perltidy twice,
once with the flag --noadd-newlines and then once again with
normal parameters, like this:

cat infile | perltidy -nanl | perltidy >outfile

When executed on this snippet

if ( $? == -1 ) {
die "failed to execute: $!\n";
}
if ( $? == -1 ) {
print "Had enough.\n";
die "failed to execute: $!\n";
}

the result is

if ( $? == -1 ) { die "failed to execute: $!\n"; }
if ( $? == -1 ) {
print "Had enough.\n";
die "failed to execute: $!\n";
}

This shows that blocks with a single statement become one-line
blocks.

Breaking existing one-line blocks
There is no automatic way to break existing long one-line blocks
into multiple lines, but this can be accomplished by processing a
script, or section of a script, with a short value of the
parameter maximum-line-length=n. Then, when the script is
reformatted again with the normal parameters, the blocks which
were broken will remain broken (with the exceptions noted above).

Another trick for doing this for certain block types is to format
one time with the --cuddled-else flag and
--cuddled-break-option=2. Then format again with the normal
parameters. This will break any one-line blocks which are
involved in a cuddled-else style.

Controlling Vertical Alignment

Vertical alignment refers to lining up certain symbols in a list of
consecutive similar lines to improve readability. For example, the
"fat commas" are aligned in the following statement:

$data = $pkg->new(
PeerAddr => join( ".", @port[ 0 .. 3 ] ),
PeerPort => $port[4] * 256 + $port[5],
Proto => 'tcp'
);

Vertical alignment can be completely turned off using the -novalign
flag mentioned below. However, vertical alignment can be forced to
stop and restart by selectively introducing blank lines. For
example, a blank has been inserted in the following code to keep
somewhat similar things aligned.

%option_range = (
'format' => [ 'tidy', 'html', 'user' ],
'output-line-ending' => [ 'dos', 'win', 'mac', 'unix' ],
'character-encoding' => [ 'none', 'utf8' ],

'block-brace-tightness' => [ 0, 2 ],
'brace-tightness' => [ 0, 2 ],
'paren-tightness' => [ 0, 2 ],
'square-bracket-tightness' => [ 0, 2 ],
);

Vertical alignment is implemented by locally increasing an existing
blank space to produce alignment with an adjacent line. It cannot
occur if there is no blank space to increase. So if a particular
space is removed by one of the existing controls then vertical
alignment cannot occur. Likewise, if a space is added with one of the
controls, then vertical alignment might occur.

For example,

# perltidy -nwls='=>'
$data = $pkg->new(
PeerAddr=> join( ".", @port[ 0 .. 3 ] ),
PeerPort=> $port[4] * 256 + $port[5],
Proto=> 'tcp'
);

Completely turning off vertical alignment with -novalign
The default is to use vertical alignment, but vertical alignment
can be completely turned of with the -novalign flag.

A lower level of control of vertical alignment is possible with
three parameters -vc, -vsc, and -vbc. These independently control
alignment of code, side comments and block comments. They are
described in the next section.

The parameter -valign is in fact an alias for -vc -vsc -vbc, and
its negative -novalign is an alias for -nvc -nvsc -nvbc.

Controlling code alignment with --valign-code or -vc
The -vc flag enables alignment of code symbols such as =. The
default is -vc. For detailed control of which symbols to align,
see the -valign-exclude-list parameter below.

Controlling side comment alignment with --valign-side-comments or
-vsc
The -vsc flag enables alignment of side comments and is enabled
by default. If side comment alignment is disabled with -nvsc
they will appear at a fixed space from the preceding code token.
The default is -vsc

Controlling block comment alignment with --valign-block-comments or
-vbc
When -vbc is enabled, block comments can become aligned for
example if one comment of a consecutive sequence of comments
becomes outdented due a length in excess of the maximum line
length. If this occurs, the entire group of comments will remain
aligned and be outdented by the same amount. This coordinated
alignment will not occur if -nvbc is set. The default is -vbc.

Finer alignment control with --valign-exclusion-list=s or -vxl=s and
--valign-inclusion-list=s or -vil=s
More detailed control of alignment types is available with these
two parameters. Most of the vertical alignments in typical
programs occur at one of the tokens ',', '=', and '=>', but many
other alignments are possible and are given in the following
list:

= **= += *= &= <<= &&= -= /= |= >>= ||= //= .= %= ^= x=
{ ( ? : , ; => && || ~~ !~~ =~ !~ // <=> -> q
if unless and or err for foreach while until

These alignment types correspond to perl symbols, operators and
keywords except for 'q', which refers to the special case of
alignment in a 'use' statement of qw quotes and empty parens.

They are all enabled by default, but they can be selectively
disabled by including one or more of these tokens in the space-
separated list valign-exclusion-list=s. For example, the
following would prevent alignment at = and if:

--valign-exclusion-list='= if'

If it is simpler to specify only the token types which are to be
aligned, then include the types which are to be aligned in the
list of --valign-inclusion-list. In that case you may leave the
valign-exclusion-list undefined, or use the special symbol * for
the exclusion list. For example, the following parameters enable
alignment only at commas and 'fat commas':

--valign-inclusion-list=', =>'
--valign-exclusion-list='*' ( this is optional and may be omitted )

These parameter lists should consist of space-separated tokens
from the above list of possible alignment tokens, or a '*'. If
an unrecognized token appears, it is simply ignored. And if a
specific token is entered in both lists by mistake then the
exclusion list has priority.

The default values of these parameters enable all alignments and
are equivalent to

--valign-exclusion-list=' '
--valign-inclusion-list='*'

To illustrate, consider the following snippet with default
formatting

# perltidy
$co_description = ($color) ? 'bold cyan' : ''; # descr
$co_prompt = ($color) ? 'bold green' : ''; # prompt
$co_unused = ($color) ? 'on_green' : 'reverse'; # unused

To exclude all alignments except the equals (i.e., include only
equals) we could use:

# perltidy -vil='='
$co_description = ($color) ? 'bold cyan' : ''; # descr
$co_prompt = ($color) ? 'bold green' : ''; # prompt
$co_unused = ($color) ? 'on_green' : 'reverse'; # unused

To exclude only the equals we could use:

# perltidy -vxl='='
$co_description = ($color) ? 'bold cyan' : ''; # descr
$co_prompt = ($color) ? 'bold green' : ''; # prompt
$co_unused = ($color) ? 'on_green' : 'reverse'; # unused

Notice in this last example that although only the equals
alignment was excluded, the ternary alignments were also lost.
This happens because the vertical aligner sweeps from left-to-
right and usually stops if an important alignment cannot be made
for some reason.

But also notice that side comments remain aligned because their
alignment is controlled separately with the parameter
--valign-side-comments described above.

Aligning postfix unless and if with --valign-if-unless or -viu
By default, postfix if terms align and postfix unless terms
align, but separately. For example,

# perltidy [DEFAULT]
print "Tried to add: @Resolve\n" if ( @Resolve and !$Quiet );
print "Would need: @DepList\n" if ( @DepList and !$Quiet );
print "Output:\n" unless $Quiet;
print join( "\n", @Output ) . "\n" unless $Quiet;

The -viu flag causes a postfix unless to be treated as if it were
a postfix if for purposes of alignment, and thus they align:

# perltidy -viu
print "Tried to add: @Resolve\n" if ( @Resolve and !$Quiet );
print "Would need: @DepList\n" if ( @DepList and !$Quiet );
print "Output:\n" unless $Quiet;
print join( "\n", @Output ) . "\n" unless $Quiet;

Aligning signed numbers with --valign-signed-numbers or -vsn
Setting -vsn causes columns of numbers containing both signed and
unsigned values to have leading signs placed in their own column.
For example:

# perltidy -vsn
my @correct = (
[ 123456.79, 86753090000.868, 11 ],
[ -123456.79, -86753090000.868, -11 ],
[ 123456.001, 80.080, 10 ],
[ -123456.001, -80.080, 0 ],
[ 10.9, 10.9, 11 ],
);

The default is -vsn. This can be turned off to get is strict
left justification:

# perltidy -nvsn
my @correct = (
[ 123456.79, 86753090000.868, 11 ],
[ -123456.79, -86753090000.868, -11 ],
[ 123456.001, 80.080, 10 ],
[ -123456.001, -80.080, 0 ],
[ 10.9, 10.9, 11 ],
);

Some points regarding -vsn are:

+o This option works by inserting a single space ahead of
unsigned numbers when possible. This is not done if it would
require increasing the maximum width of a column.

+o This option is mainly limited to lists of comma-separated
numbers. For multiline lists of numbers, having trailing
commas can sometimes improve the results. If missing,
perltidy can add them for example with parameters -wtc=b
-atc. See "Adding and Deleting Commas".

+o This option has a control parameter
--valign-signed-number-limit=N, or -vsnl=N. This value
controls formatting of very long columns of numbers and
should not normally need to be changed. To see its purpose,
consider a very long column of just unsigned numbers, say
1000 lines. If we add a single negative number, it is
undesirable to move all of the other numbers over by one
space. This could create many lines of file differences but
not really improve the appearance when a local section of the
table was viewed. The number N avoids this problem by not
adding extra indentation to a run of more than N lines of
unsigned numbers. The default value, N=20, is set to be a
number of lines for which the ends of a long column of
unsigned numbers are not normally both in view.

Aligning assignment operators with --valign-wide-equals or -vwe
The following assignment operators are aligned independently by
default:

= **= += *= &= <<= &&= -= /= |= >>= ||= //= .= %= ^= x=

Setting --valign-wide-equals, or -vwe, causes them to be
vertically aligned together with the trailing = all aligned. For
example, here is the default formatting for some code with
several of these operators:

$str .= SPACE x $total_pad_count;
$str_len += $total_pad_count;
$total_pad_count = 0;
$str .= $rfields->[$j];
$str_len += $rfield_lengths->[$j];

And here is the same code formatted with -vme:

# perltidy -vme
$str .= SPACE x $total_pad_count;
$str_len += $total_pad_count;
$total_pad_count = 0;
$str .= $rfields->[$j];
$str_len += $rfield_lengths->[$j];

This option was added for issue git #135 and can improve
readability, but it is off by default to avoid changing existing
formatting.

Extended Syntax

This section describes some parameters for dealing with extended
syntax.

For another method of handling extended syntax see the section
"Skipping Selected Sections of Code".

Also note that the module Perl::Tidy supplies a pre-filter and post-
filter capability. This requires calling the module from a separate
program rather than through the binary perltidy.

-xs, --extended-syntax
This flag allows perltidy to handle certain common extensions to
the standard syntax without complaint.

For example, without this flag a structure such as the following
would generate a syntax error:

Method deposit( Num $amount) {
$self->balance( $self->balance + $amount );
}

This flag is enabled by default but it can be deactivated with
-nxs. Probably the only reason to deactivate this flag is to
generate more diagnostic messages when debugging a script.

-sal=s, --sub-alias-list=s
This flag causes one or more words to be treated the same as if
they were the keyword sub. The string s contains one or more
alias words, separated by spaces or commas.

For example,

perltidy -sal='method fun _sub M4'

will cause the perltidy to treat the words 'method', 'fun',
'_sub' and 'M4' the same as if they were 'sub'. Note that if the
alias words are separated by spaces then the string of words
should be placed in quotes.

Note that several other parameters accept a list of keywords,
including 'sub' (see "Specifying Block Types"). You do not need
to include any sub aliases in these lists. Just include keyword
'sub' if you wish, and all aliases are automatically included.

-gal=s, --grep-alias-list=s
This flag allows a code block following an external 'list
operator' function to be formatted as if it followed one of the
built-in keywords grep, map or sort. The string s contains the
names of one or more such list operators, separated by spaces or
commas.

By 'list operator' is meant a function which is invoked in the
form

word {BLOCK} @list

Perltidy tries to keep code blocks for these functions intact,
since they are usually short, and does not automatically break
after the closing brace since a list may follow. It also does
some special handling of continuation indentation.

For example, the code block arguments to functions 'My_grep' and
'My_map' can be given formatting like 'grep' with

perltidy -gal='My_grep My_map'

By default, the following list operators in List::Util are
automatically included:

all any first none notall reduce reductions

Any operators specified with --grep-alias-list are added to this
list. The next parameter can be used to remove words from this
default list.

-gaxl=s, --grep-alias-exclusion-list=s
The -gaxl=s flag provides a method for removing any of the
default list operators given above by listing them in the string
s. To remove all of the default operators use -gaxl='*'.

-uf=s, --use-feature=s
This flag tells perltidy to allow or disallow the syntax
associated a pragma in string s. The current possible settings
are:

+o --use-feature='class'. This tells perltidy to recognized the
special words class, method, field, and ADJUST as defined for
this feature.

+o --use-feature='noclass'. This tells perltidy not to treat
words class, method, field, ADJUST specially.

+o Neither of these (--use-feature not defined). This is the
DEFAULT and recommended setting. In this case perltidy will
try to automatically handle both the newer --use-feature
'class' syntax as well as some conflicting uses of some of
these special words by existing modules.

Note that this parameter is independent of any use feature
control lines within a script. Perltidy does not look for or read
such control lines. This is because perltidy must be able to
work on small chunks of code sent from an editor, so it cannot
assume that such lines will be within the lines being formatted.

Working around problems with older version of Perl
Perltidy contains a number of rules which help avoid known
subtleties and problems with older versions of perl, and these
rules always take priority over whatever formatting flags have
been set. For example, perltidy will usually avoid starting a
new line with a bareword, because this might cause problems if
"use strict" is active.

There is no way to override these rules.

Deleting and Extracting Pod or Comments

Deleting selected text
Perltidy can selectively delete comments and/or pod
documentation. The command -dac or --delete-all-comments will
delete all comments and all pod documentation, leaving just code
and any leading system control lines.

The command -dp or --delete-pod will remove all pod documentation
(but not comments).

Two commands which remove comments (but not pod) are: -dbc or
--delete-block-comments and -dsc or --delete-side-comments.
(Hanging side comments will be deleted with side comments here.)

When side comments are deleted, any special control side comments
for non-indenting braces will be retained unless they are
deactivated with a -nnib flag.

The negatives of these commands also work, and are the defaults.
When block comments are deleted, any leading 'hash-bang' will be
retained. Also, if the -x flag is used, any system commands
before a leading hash-bang will be retained (even if they are in
the form of comments).

Writing selected text to a file
When perltidy writes a formatted text file, it has the ability to
also send selected text to a file with a .TEE extension. This
text can include comments and pod documentation.

The command -tac or --tee-all-comments will write all comments
and all pod documentation.

The command -tp or --tee-pod will write all pod documentation
(but not comments).

The commands which write comments (but not pod) are: -tbc or
--tee-block-comments and -tsc or --tee-side-comments. (Hanging
side comments will be written with side comments here.)

The negatives of these commands also work, and are the defaults.

The perltidyrc file

Using a .perltidyrc command file
If you use perltidy frequently, you probably won't be happy until
you create a .perltidyrc file to avoid typing commonly-used
parameters. Perltidy will first look in your current directory
for a command file named .perltidyrc. If it does not find one,
it will continue looking for one in other standard locations.

These other locations are system-dependent, and may be displayed
with the command "perltidy -dpro". Under Unix systems, it will
first look for an environment variable PERLTIDY. Then it will
look for a .perltidyrc file in the home directory, and then for a
system-wide file /usr/local/etc/perltidyrc, and then it will look
for /etc/perltidyrc. Note that these last two system-wide files
do not have a leading dot. Further system-dependent information
will be found in the INSTALL file distributed with perltidy.

Under Windows, perltidy will also search for a configuration file
named perltidy.ini since Windows does not allow files with a
leading period (.). Use "perltidy -dpro" to see the possible
locations for your system. An example might be C:\Documents and
Settings\All Users\perltidy.ini.

Another option is the use of the PERLTIDY environment variable.
The method for setting environment variables depends upon the
version of Windows that you are using.

Under Windows NT / 2000 / XP the PERLTIDY environment variable
can be placed in either the user section or the system section.
The later makes the configuration file common to all users on the
machine. Be sure to enter the full path of the configuration
file in the value of the environment variable. Ex.
PERLTIDY=C:\Documents and Settings\perltidy.ini

The configuration file is free format, and simply a list of
parameters, just as they would be entered on a command line. Any
number of lines may be used, with any number of parameters per
line, although it may be easiest to read with one parameter per
line. Comment text begins with a #, and there must also be a
space before the # for side comments. It is a good idea to put
complex parameters in either single or double quotes.

Here is an example of a .perltidyrc file:

# This is a simple of a .perltidyrc configuration file
# This implements a highly spaced style
-se # errors to standard error output
-w # show all warnings
-bl # braces on new lines
-pt=0 # parens not tight at all
-bt=0 # braces not tight
-sbt=0 # square brackets not tight

The parameters in the .perltidyrc file are installed first, so
any parameters given on the command line will have priority over
them.

To avoid confusion, perltidy ignores any command in the
.perltidyrc file which would cause some kind of dump and an exit.
These include:

-h -v -ddf -dln -dop -dsn -dtt -dwls -dwrs -ss

There are several options may be helpful in debugging a
.perltidyrc file:

+o A very helpful command is --dump-profile or -dpro. It writes
a list of all configuration filenames tested to standard
output, and if a file is found, it dumps the content to
standard output before exiting. So, to find out where
perltidy looks for its configuration files, and which one if
any it selects, just enter

perltidy -dpro

+o It may be simplest to develop and test configuration files
with alternative names, and invoke them with -pro=filename on
the command line. Then rename the desired file to
.perltidyrc when finished.

+o The parameters in the .perltidyrc file can be switched off
with the -npro option on the command line.

+o Any parameter in the .perltidyrc file can be overridden with
a replacement value on the command line. This is because the
command line is processed after the .perltidyrc file.

+o The commands --dump-options, --dump-defaults,
--dump-long-names, and --dump-short-names, all described
below, may all be helpful.

Skipping a line with an unknown parameter
Versions of perltidy greater than 20240511 have an option to
filter unrecognized parameters from a perltidy command file. If
a line in the file begins with three dashes followed by a
parameter name (rather than one or two), then the line will be
removed if the parameter is unknown. Otherwise, a dash will be
removed to make the line valid. This option was added to allow a
single command file to be used during the transition to a new
version of perltidy.

Creating a new abbreviation
A special notation is available for use in a .perltidyrc file for
creating an abbreviation for a group of options. This can be
used to create a shorthand for one or more styles which are
frequently, but not always, used. The notation is to group the
options within curly braces which are preceded by the name of the
alias (without leading dashes), like this:

newword {
-opt1
-opt2
}

where newword is the abbreviation, and opt1, etc, are existing
parameters or other abbreviations. The main syntax requirement
is that the new abbreviation along with its opening curly brace
must begin on a new line. Space before and after the curly
braces is optional.

For a specific example, the following definition

oneliner {
--maximum-line-length=0
--noadd-newlines
--noadd-terminal-newline
}

or equivalently, with abbreviations,

oneliner { -l=0 -nanl -natnl }

could be placed in a .perltidyrc file to temporarily override the
maximum line length with a large value, to temporarily prevent
new line breaks from being added, and to prevent an extra newline
character from being added the file. All other settings in the
.perltidyrc file still apply. Thus it provides a way to format a
long 'one liner' when perltidy is invoked with

perltidy --oneliner ...

(Either "-oneliner" or "--oneliner" may be used).

Skipping leading non-perl commands with -x or --look-for-hash-bang
If your script has leading lines of system commands or other text
which are not valid perl code, and which are separated from the
start of the perl code by a "hash-bang" line, ( a line of the
form "#!...perl" ), you must use the -x flag to tell perltidy not
to parse and format any lines before the "hash-bang" line. This
option also invokes perl with a -x flag when checking the syntax.
This option was originally added to allow perltidy to parse
interactive VMS scripts, but it should be used for any script
which is normally invoked with "perl -x".

Please note: do not use this flag unless you are sure your script
needs it. Parsing errors can occur if it does not have a hash-
bang, or, for example, if the actual first hash-bang is in a
here-doc. In that case a parsing error will occur because the
tokenization will begin in the middle of the here-doc.

Debugging perltidy input

The --dump-... parameters
The following flags are available for debugging. Note that all
commands named --dump-... will simply write some requested
information to standard output and then immediately exit.

--dump-cuddled-block-list or -dcbl will dump to standard output
the internal hash of cuddled block types created by a
--cuddled-block-list input string.

--dump-defaults or -ddf will write the default option set to
standard output and quit

--dump-integer-option-range or -dior will write a list of comma-
separated values. Each line contains the name of an integer
option along with its minimum, maximum, and default values.

--dump-profile or -dpro will write the name of the current
configuration file and its contents to standard output and quit.

--dump-options or -dop will write current option set to standard
output and quit.

--dump-long-names or -dln will write all command line long names
(passed to Get_options) to standard output and quit.

--dump-short-names or -dsn will write all command line short
names to standard output and quit.

--dump-token-types or -dtt will write a list of all token types
to standard output and quit.

--dump-want-left-space or -dwls will write the hash
%want_left_space to standard output and quit. See the section on
controlling whitespace around tokens.

--dump-want-right-space or -dwrs will write the hash
%want_right_space to standard output and quit. See the section
on controlling whitespace around tokens.

See "Analyzing Code" for additional --dump- parameters.

Other parameters related to processing
--no-memoize or -nmem will turn of memoizing. Memoization can
reduce run time when running perltidy repeatedly in a single
process. It is on by default but can be deactivated for testing
with -nmem.

--no-timestamp or -nts will eliminate any time stamps in output
files to prevent differences in dates from causing test
installation scripts to fail. There are just a couple of places
where timestamps normally occur. One is in the headers of html
files, and another is when the -cscw option is selected. The
default is to allow timestamps (--timestamp or -ts).

--file-size-order or -fso will cause files to be processed in
order of increasing size, when multiple files are being
processed. This is particularly useful during program
development, when large numbers of files with varying sizes are
processed, because it can reduce virtual memory usage. This is
the default and can be deactivated with -nfso.

--maximum-file-size-mb=n or -maxfs=n specifies the maximum file
size in megabytes that perltidy will attempt to format. This
parameter is provided to avoid causing system problems by
accidentally attempting to format an extremely large data file.
Most perl scripts are less than about 2 MB in size. The integer n
has a default value of 10, so perltidy will skip formatting files
which have a size greater than 10 MB. The command to increase
the limit to 20 MB for example would be

perltidy -maxfs=20

This length test is applied to named files before they are read
into memory. It is applied to files arriving from standard input
after they are read into memory. It is not applied to character
strings arriving by a call to the Perl::Tidy module.

Controls for when to stop processing
--maximum-level-errors=n or -maxle=n specifies the maximum number
of indentation level errors are allowed before perltidy skips
formatting and just outputs a file verbatim. The default is n=1.
This means that if the final indentation of a script differs from
the starting indentation by more than 1 levels, the file will be
output verbatim. To avoid formatting if there are any
indentation level errors use -maxle=0. To skip this check you can
either set n equal to a large number, such as n=100, or set n=-1.

For example, the following script has level error of 3 and will
be output verbatim

Input and default output:
{{{


perltidy -maxle=100
{
{
{

--maximum-unexpected-errors=n or -maxue=n specifies the maximum
number of unexpected tokenization errors are allowed before
formatting is skipped and a script is output verbatim. The
intention is to avoid accidentally formatting a non-perl script,
such as an html file for example. This check can be turned off
by setting n=0.

A recommended value is n=3. However, the default is n=0 (skip
this check) to avoid causing problems with scripts which have
extended syntaxes.

Handling errors in options which take integer values
Many of the input parameters take integer values. Before
processing begins, a check is made to see if any of these integer
parameters exceed their valid ranges. The default behavior when
a range is exceeded is to write a warning message and reset the
value to its default setting. This default behavior can be
changed with the parameter --integer-range-check=n, or -irc=n, as
follows:

n=0 skip check completely (for stress-testing perltidy only)
n=1 reset bad values to defaults but do not issue a warning
n=2 reset bad values to defaults and issue warning [DEFAULT]
n=3 stop if any values are out of bounds

The values n=0 and n=1 are mainly useful for testing purposes.

Debugging perltidy tokenization
-DEBUG, -D will write a file with extension .DEBUG for each input
file showing the tokenization of all lines of code. This can
produce a lot of output and is mainly useful for debugging
tokenization issues during perltidy development.

Analyzing Code

Perltidy reports any obvious issues that are found during formatting,
such as unbalanced braces. But several parameters are available for
making certain additional checks for issues which might be of
interest to a programmer. These parameters fall into two categories
as indicated by their prefix, --dump- or --warn-:

+o The --dump- parameters read a file, write information to the
standard output, and then exit without doing any formatting.

+o The --warn- parameters, on the other hand, cause perltidy to
function normally but issue warnings to the error output when
certain conditions are encountered.

Some of these have associated control parameters.

Use --dump-block-summary to make a table of information on code
blocks
A table listing information about the blocks of code in a file
can be made with --dump-block-summary, or -dbs. This causes
perltidy to read and parse the file, write a table of comma-
separated values for selected code blocks to the standard output,
and then exit. This parameter must be on the command line, not
in a .perlticyrc file, and it requires a single file name on the
command line. For example

perltidy -dbs somefile.pl >blocks.csv

produces an output file blocks.csv whose lines hold these
parameters:

filename - name of the file
line - line number of the opening brace of this block
line_count - number of lines between opening and closing braces
code_lines - number of lines excluding blanks, comments, and pod
type - block type (sub, for, foreach, ...)
name - block name if applicable (sub or asub name, label..)
depth - nesting depth of the opening block brace
max_change - change in depth to the most deeply nested code block
block_count - total number of code blocks nested in this block
mccabe_count - McCabe complexity measure of this code block

This feature was developed to help identify complex sections of
code as an aid in refactoring. The McCabe complexity measure
follows the definition used by Perl::Critic. By default the
table contains these values for subroutines, but the user may
request them for any or all blocks of code or packages. For
blocks which are loops nested within loops, a postfix '+' to the
"type" is added to indicate possible code complexity. Although
the table does not otherwise indicate which blocks are nested in
other blocks, this can be determined by computing and comparing
the block ending line numbers.

For subroutines, the number of call arguments (args) is listed in
parentheses in the "type" column. For example, sub(9) indicates a
sub with 9 args. Subroutines whose arg count cannot easily be
determined are indicated as sub(*). The count includes any
leading object passed in a method call (such as $self).

By default the table lists subroutines with more than 20
"code_lines", but this can be changed with the following two
parameters:

--dump-block-minimum-lines=n, or -dbl=n, where n is the minimum
number of "code_lines" to be included. The default is -n=20.
Note that "code_lines" is the number of lines excluding comments,
blanks and pod.

--dump-block-types=s, or -dbt=s, where string s is a list of
block types to be included. The type of a block is either the
name of the perl builtin keyword for that block (such as sub if
elsif else for foreach ..) or the word immediately before the
opening brace. In addition, there are a few symbols for special
block types, as follows:

if elsif else for foreach ... any keyword introducing a block
sub - any sub or anonymous sub
asub - any anonymous sub
* - any block except nameless blocks
+ - any nested inner block loop
package - any package or class
closure - any nameless block
elsif3 - an if-elsif- chain with 3 or more elsif's (see below)

A chain of if-elsif-... blocks may be reported as a single line
item by entering the word elsif with an appended integer, as
indicated by the last item in this list. The integer indicates
the number of elsif blocks required for a chain to be reported.
If you use this, you may want to also use -dbl=n, with a smaller
number of lines n than the default.

In addition, specific block loop types which are nested in other
loops can be selected by adding a + after the block name. (Nested
loops are sometimes good candidates for restructuring).

The default is -dbt='sub'.

In the following examples a table "block.csv" is created for a
file "somefile.pl":

+o This selects both "subs" and "packages" which have 20 or more
lines of code. This can be useful in code which contains
multiple packages.

perltidy -dbs -dbt='sub package' somefile.pl >blocks.csv

+o This selects blocks with 2 or more code lines which are type
"sub" or which are inner loops.

perltidy -dbs -dbl=2 -dbt='sub +' somefile.pl >blocks.csv

+o This selects every block and package.

perltidy -dbs -dbl=1 -dbt='* closure' somefile.pl >blocks.csv

+o This selects every if-chain which contains 2 or more "elsif"
blocks:

perltidy -dbs -dbl=1 -dbt='elsif2' somefile.pl >blocks.csv

+o This selects every "while" block with 6 or more code lines

perltidy -dbs -dbt=while -dbl=6 somfile.pl >while.csv

Use --dump-unusual-variables to find unused, reused, and certain
other variables of interest
Variables with certain properties of interest to a programmer can
be listed with --dump-unusual-variables or -duv. This parameter
must be on the command line, along with a single file name. It
causes perltidy to scan the file for certain variable types,
write any found to the standard output, and then immediately exit
without doing any formatting. For example

perltidy -duv somefile.pl >vars.txt

produces a file with lines which look something like

1778:u: my $input_file
6089:r: my $j: reused - see line 6076

The values on the line are separated by colons and have the
following meaning:

line number - the number of the line of the input file
issue - a single letter indicating the issue, see below
variable name - the name of the variable, preceded by a keyword
note - an optional note referring to another line

If there are a large number of issues it can be convenient to
read the file into a spreadsheet.

The checks are made for variables introduced by the keywords my,
state, and our, along with variables defined with use vars and
use constant. It is a good idea to also set use strict in a
script so that Perl itself can find issues with variables which
appear in a script without one of these methods.

The types of checks which are made are identified in the output
with one of the letters r, s, p, u, and c as follows:

r: reused variable name
These are variables which are re-declared in the scope of a
variable with the identical name. This can be confusing,
perhaps not when the code is first written, but possibly
later during maintenance work. For example, this can make it
difficult to locate the correct variable with an editor when
changes are being made. This issue can be avoided by renaming
one of the conflicting variables. Note that this is similar
to the Perl::Critic policy Variables::ProhibitReusedNames.

s: sigil change but reused bareword
These are variables which have the same bareword name but a
different sigil ($, @, or %) as another variable in the same
scope. For example, this occurs if variables $data and %data
share the same scope. This can also be confusing for the
reasons mentioned above and can be avoided by renaming one of
the variables.

p: package-crossing variables
These are lexical variables which are declared in one package
and still visible in subroutines of a different package in
the same file. This can be confusing, and it might cause the
program to run differently, or fail, if the the packages were
ever split into separate files. This issue can usually be
avoided by placing code in block braces of some type. For
example, this issue is often found in test code and can
sometimes be fixed by using the structure

main();

sub main { #<<<
## old main code goes here
}

The non-indenting-braces side comment "#<<<" is not required
but will keep the indentation of the old code unchanged.

This check is only applied to package statements which are
not enclosed in block braces in order avoid warnings at
temporary package changes.

u: unused variables
These are lexical variables declared with "my" or "state"
(but not "our") and not referenced again within their scope.
Calling them unused is convenient but not really accurate;
this is a "gray area" for a program. There are some good
reasons for having such variables. For example, they might
occur in a list of values provided by another routine or data
structure, and therefore must be listed, even though they
might not be referenced again. Having such variables can
make them immediately available for future development and
debugging, and can be beneficial for program clarity.

But sometimes they can occur due to being orphaned by a
coding change, due to a misspelling, or by having an
unintentional preceding "my". So it is worth reviewing them,
especially for new code. Here is an example of an unused
variable in a script located with this method:

BEGIN { my $string = "" }
...
$string .= "ok";

This looks nice at first glance, but the scope of the "my"
declaration is limited to the surrounding braces, so it is
not the same variable as the other $string and must therefore
be reported as unused. This particular problem would have
also been caught by perl if the author had used "strict".

c: unused constants
These are names which are declared with a "use constant" and
a reference was not seen again within their package. They
might be needed by an external package, or a set of standard
definitions, or available for future development. And in
some unusual cases a reference may have been missed by
perltidy. But they might also be unused remnants from code
development, or due to a misspelling, so it can be worthwhile
reviewing them.

Exception: The following our variables are exempt from warnings:
$VERSION, @EXPORT, @EXPORT_OK, %EXPORT_TAGS, @ISA, $AUTOLOAD.

Use --warn-variable-types to warn about certain variable types
The flag --warn-variable-types=string, or -wvt=string, is the
--warn counterpart to --dump-unusual-variables, and can be used
to produce a warning message if certain of the above variable
types are encountered during formatting. All possible variable
warnings may be requested with -wvt='*' or -wvt=1.

For example,

perltidy -wvt='*' somefile.pl

The default is not to do any of these checks, and it can also be
indicated with -wvt=0.

To restrict the check to a specific set warnings, set the input
string to be a space-separated or comma-separated list of the
letters associated with the types of variables to be checked. For
example:

perltidy -wvt='s r' somefile.pl

will process somefile.pl normally but issue a warning if either
of the issues s or r, described above, are encountered.

A companion flag, --warn-variable-exclusion-list=string, or
-wvxl=string, can be used to skip warning checks for a list of
variable names. A leading and/or trailing '*' may be placed on
any of these variable names to allow a partial match.

For example,

perltidy -wvt=1 -wvxl='$self $class *_unused' somefile.pl

will do all possible checks but not report any warnings for
variables $self, $class, and for example $value_unused.

This partial match option provides a way to trigger a warning
message when a new unused variable is detected in a script. This
can be accomplished by adding a unique suffix to the names of
existing unused variables, such as "_unused". This suffix is then
added to the exclusion list.

As a specific example, consider the following line which is part
of some debug code which only references the latter three
variables (but might someday need to reference the package
variable too).

my ( $package_uu, $filename, $line, $subroutine ) = caller();

The unused variable, $package_uu, has been specially marked with
suffix "_uu". No type u (unused variable) warning will be
produced provided that this wildcard suffix is in the exclusion
list:

-wvxl='*_uu'

Use --dump-unique-keys to help locate misspelled hash keys
The parameter --dump-unique-keys, or -duk, dumps a list of hash
keys which appear to be used just once, and do not appear among
the quoted strings in a file. For example:

perltidy -duk File.pm >output.txt

The lines in the output file list unique keys and line numbers.

Many programs contain sets of hash keys which are largely unique
to that program but which are used for communication with other
programs. A filter is used to avoid listing such keys. This
filter is described in the next section.

A program "dump_unique_keys.pl" at
<https://github.com/perltidy/perltidy/tree/master/examples> can
run perltidy with -duk on multiple files, and then remove any
common keys from the list.

Use --warn-unique-keys-cutoff=N, or -wukc=N, to adjust the unique
keys filter.
This parameter controls the filter used by both
--dump-unique-keys and its counterpart --warn-unique-keys,
described below. The filter works as follows: if a related set
of keys has more than N unique keys, then none of them are
listed. The default value is N=1, which causes the minimum
number of keys to be reported. Increasing N above 1 will allow
more unique keys to be reported.

Thus, a value N=1 will catch up to 1 error per set of related
keys. A value N=2 will catch up to two errors per set, but at the
same time may produce more unwanted warnings.

For the special case N=1 some extra filtering is done to minimize
unwanted warnings. For examples, unique keys in ALL CAPS are not
reported for N=1.

Use --warn-unique-keys to warn of possible misspelled hash keys
The parameter --warn-unique-keys, or -wuk, can be used to produce
a warning of a possible misspelled hash key during normal
processing. It works like the previous -dump version but writes
to the standard error output if unique keys are found during
normal formatting.

For some scripts, this will produce unwanted warnings for a few
keys, even though the filter described above attempts to avoid
this. A way to avoid warnings for specific keys is to list them
somewhere in the program in a qw list. This will be seen during
the key scan, and the then keys will not be considered unique.

Use --dump-mixed-call-parens to find functions called both with and
without parens
The parameter --dump-mixed-call-parens, or -dmcp, provides
information on the use of call parens within a program. It
produces a list of keywords and sub names which occur both both
with and without parens. In other words, with a mixed style.
This might be useful if one is working to standardize the call
style for some particular keyword or function. For example,

perltidy -dmcp somefile.pl >output.txt

will analyze the text of somefile.pl, write the results to
output.txt, and then immediately exit (like all dump-
parameters).

The output shows a list of operators and the number of times they
were used with parens and the number of times without parens.
For example, here is a small section of the output from one file
in a past Perl distribution:

k:length:17:9
k:open:30:9
k:pop:3:4

The first line shows that the "length" function occurs 17 times
with parens and 9 times without parens. The 'k' indicates that
"length" is a Perl builtin keyword ('U' would mean user-defined
sub, and 'w' would mean unknown bareword). So from this partial
output we see that the author had a preference for parens around
the args of "length" and "open", whereas "pop" was about equally
likely to have parens as not.

More detailed information can be obtained with the parameters
described in the next section.

Use --want-call-parens=s and --nowant-call-parens=s to warn about
specific missing or extra call parens
The parameter --want-call-parens=s, or -wcp=s, can be used to to
produce a warning message if call parens are missing from
selected functions. Likewise, --nowant-call-parens=s, or
-nwcp=s, can warn if call parens exist for selected functions.
When either of these parameters are set, perltidy will report any
discrepancies from the requested style in its error output.

Before using either of these parameters, it may be helpful to
first use --dump-mixed-call-parens=s, described in the previous
section, to get an overview of the existing paren usage in a
file.

The string arguments s are space-separated lists of the names of
the functions to be checked. The function names may be builtin
keywords or user-defined subs. They may not include a package
prefix or sigil. To illustrate,

perltidy -wcp='length open' -nwcp='pop' somefile.pl

means that the builtin functions "length" and "open" should have
parens around their call args but "pop" should not. The error
output might contain lines such as:

2314:open FD_TO_CLOSE: no call parens
3652:pop (: has call parens
3783:length $DB: no call parens
...

For builtin keywords which have both a block form and a trailing
modifier form, such as "if", only the trailing modifier form will
be checked since parens are mandatory for the block form.

The symbol & may entered instead of a function name to mean all
user-defined subs not explicitly listed. So the compact
expression

perltidy -wcp='&' somefile.pl

means that calls to all user-defined subs in the file being
processed should have their call arguments enclosed in parens.

Perltidy does not have the ability to add or delete call parens
because it is difficult to automate, so changes must be made
manually. When adding or removing parentheses, it is essential
to pay attention to operator precedence issues. For example, if
the parens in the following statement are removed, then "||" must
be changed to "or":

open( IN, "<", $infile ) || die("cannot open $infile:$!\n");

Otherwise, the "||" will operate on $infile rather than the
return value of "open".

Use --dump-mismatched-args to find function calls with args not
matching sub declarations
The parameter --dump-mismatched-args, or -dma, causes perltidy to
examine the definitions of subroutines in a file, and calls to
those subs, and report certain differences. Like all --dump
commands, it writes its report to standard output and exits
immediately. For example

perltidy -dma somefile.pl >results.txt

Four types of issues are reported, types a, o, u, and i:

a: calls made to a sub both with and without the arrow operator
For example the following two lines would be reported as a
mismatch:

Fault();

and

$self->Fault();

This may or may not be an error, but it is worth checking. It
might become an error in the future if sub "Fault" starts to
access $self.

o: (overcount): the number of call args exceeds the expected
number.
u: (undercount): the number of call args is less than the
expected number.
For example

sub gnab_gib {
my $self=shift;
my ($v1,$v2)=@_;
...
}

$self->gnab_gib(42);

In this case, the sub is expecting a total of three args
($self, $v1, and $v2) but only receives two ($self and 42),
so an undercount is reported. This is not necessarily an
error because the sub may allow for this possibility, but it
is worth checking.

Although it is not possible to automatically determine which
sub args are optional, if optional sub args are enclosed in
an extra set of parentheses, perltidy will take this a signal
that they are optional and not issue a warning. So if the
above example is written as

sub gnab_gib {
my $self = shift;
my ( $v1, ($v2) ) = @_; # <-- $v2 is optional
...;
}

then perltidy will consider that the second arg is optional
and not issue a warning for:

$self->gnab_gib(42);

For multiple default call args, place one set of parens
around them all. Some examples:

my ( ($v1) ) = @_; # <-- $v1 is optional
my ( $v1, ( $v2, $v3 ) ) = @_; # <-- $v2, $v3 are optional

i: indeterminate: a specific number of expected args for a sub
could not be determined, but it is called with a specific number.
This issue is reported for the --dump- option but not the --warn-
option.

Notes and Limitations:

+o This option works best for subs which unpack call args in an
orderly manner near the beginning of the sub from @_ and/or
with "shift" operations. It will also work for direct access
to specific elements of the @_ array. However if the coding
for arg extraction is complex then the number of sub args
will be considered indeterminate and a count cannot be
checked.

+o Sub calls made without parentheses around the args are not
checked.

+o Anonymous subs and lexical subs (introduced with "my") are
not checked.

+o Only calls which appear to be to subs defined within the file
being processed are checked. But note that a file may contain
multiple packages.

Use --warn-mismatched-args to produce a warning for function calls
with args not matching sub declarations.
This is similar to the -dump parameter described above except
that any mismatches are reported in the error file and otherwise
formatting continues normally. The basic usage is

perltidy -wma somefile.pl

Several companion controls are available to avoid unwanted error
messages:

+o --warn-mismatched-arg-types=s, or -wmat=s, can be used to
select specific tests, type a (arrow test) or o (overcounts)
or u (undercounts). All checks may be requested with
-wmat='*' or -wmat=1. This is the default.

To restrict the check to a specific warning type, set the
string equal to the letter of that warning, any a, o, or u.
For example

perltidy -wmat='a o' somefile.pl

will format somefile.pl and report any arrow-type mismatches
and overcount mismatches, but will skip undercount
mismatches.

+o --warn-mismatched-arg-exclusion-list, or -wmaxl=string, can
be given to skip the warning checks for a list of subroutine
names, entered as a quoted string of space- or comma-
separated names, without a package prefix. All subs with
those names will be skipped, regardless of package. A leading
and/or trailing * on a name may be used to indicate a partial
string match.

+o --warn-mismatched-arg-undercount-cutoff=n, or -wmauc=n, can
be used to avoid undercount warnings when the expected number
of args is less than n. Please note that this number n is
the number of args from the point of view of the sub
definition, so an object like $self passed with an arrow
operator counts as one arg.

The default value is n=4. This has been found to allow most
programs to pass without warnings, but it should be reduced
if possible for better error checking. The minimum possible
value of n needed to avoid triggering an error for a program
can be determined by running with -wma -wmauc=0. If there
are undercount errors, a note at the bottom of the error
output indicates the value of n required to avoid reporting
them.

As noted above for the parameter --dump-mismatched-args, if
optional call args are enclosed in separate parentheses, then
perltidy will recognize them as optional args and avoid
needless warnings. If this method is used, then -wmauc=0
should be used for maximal checking.

+o --warn-mismatched-arg-overcount-cutoff=n, or -wmaoc=n, can be
used to avoid overcount warnings when the expected number of
args is less than n. The default value is n=1. This avoids
warning messages for subroutines which are dummy placeholders
for possible expansion.

To illustrate these controls,

perltidy -wma -wmat='o u' -wmaxl='new old' -wmauc=2 somefile.pl

means format somefile.pl as usual and check for mismatched
overcounts and undercounts but not arrows. Skip checking for any
sub named "new" or "old", and only warn of undercounts for subs
expecting more than 2 args.

Use --dump-mismatched-returns to find function calls where the number
of requested values may disagree with sub return statements
The parameter --dump-mismatched-returns, or -dmr, examines the
return side of sub call statements. Like all --dump commands, it
writes its report to standard output and exits immediately. For
example

perltidy -dmr somefile.pl >results.txt

The following types of issues are reported:

x: calls requesting an array from a sub with no return
statements.
y: calls requesting a scalar from a sub with no return
statements.
o: (overwant): calls requesting an array with a count which
exceeds the maximum number returned by the sub.
u: (underwant): calls requesting an array with a count which is
below the maximum and which does not match a number returned by
the sub.
s: calls requesting a scalar from a sub which only returns two or
more items.

These issue types are illustrated with the following code

sub macho {
...
( $name, $flags ); # 2 values but no 'return' statement
}

( $name, $flags ) = macho(); # 'x' (want array, but no return)
$name = macho(); # 'y' (want scalar but no return)

sub wimp {
...;
return ( $name, $flags ); # 2 values with 'return' statement
}

( $name, $flags, $access) = wimp(); # 'o' (want array 3 > 2)
($name) = wimp(); # 'u' (want array 1 < 2)
$name = wimp(); # 's' (want scalar but 2 values returned)

This analysis works by scanning all call statements and all sub
return statements, and comparing the the number of items wanted
with the possible number of items returned. If a specific value
for either of these numbers cannot be determined for a call then
it cannot be checked.

Since only return statements are scanned for return values, this
analysis will not be useful for programming which relies on the
default return mechanism, as in the first sub above. Note that
the Perl::Critic policy RequireFinalReturn can be used to check
for code in this situation.

Reported issues are not necessarily errors, but they might be, or
they might indicate potentially confusing code.

Use --warn-mismatched-returns to issue warnings when the number of
requested values may disagree with sub return statements
This is similar to the -dump parameter described above except
that any mismatches are reported in the error file and otherwise
formatting continues normally. The basic usage is

perltidy -wmr somefile.pl

The following companion controls are available to avoid unwanted
error messages:

+o --warn-mismatched-return-types=string, or -wmrt=string, can
be used to limit checks.

To restrict the checking, set the string equal to the
letter(s) of that warning, any x, y, o, u, or s. For example

perltidy -wmr -wmrt='x o s' somefile.pl

will format somefile.pl and report issue types x, o, and s
but not types u and y. All checks may be requested with
-wmrt='*' or -wmrt=1. This is the default if this parameter
is not specified.

+o --warn-mismatched-return-exclusion-list, or -wmrxl=string,
can be given to skip the warning checks for a list of
subroutine names, entered as a quoted string of space- or
comma-separated names, without a package prefix. All subs
with those names will be skipped, regardless of package. A
leading and/or trailing * on a name may be used to indicate a
partial string match.

Working with MakeMaker, AutoLoader and SelfLoader
The first $VERSION line of a file which might be eval'd by MakeMaker
is passed through unchanged except for indentation. The default
--pass-version-line, or -pvl, is to do this. Use
--nopass-version-line, or -npvl, to deactivate this feature.

If the AutoLoader module is used, perltidy will continue formatting
code after seeing an __END__ line. The default
--look-for-autoloader, or -lal, is to do this. Use
--nolook-for-autoloader, or -nlal, to deactivate this feature.

Likewise, if the SelfLoader module is used, perltidy will continue
formatting code after seeing a __DATA__ line. The default
--look-for-selfloader, or -lsl, is to do this. Use
--nolook-for-selfloader, or -nlsl, to deactivate this feature.

HTML OPTIONS

The -html master switch
The flag -html causes perltidy to write an html file with
extension .html. So, for example, the following command

perltidy -html somefile.pl

will produce a syntax-colored html file named somefile.pl.html
which may be viewed with a browser.

Please Note: In this case, perltidy does not do any formatting to
the input file, and it does not write a formatted file with
extension .tdy. This means that two perltidy runs are required
to create a fully reformatted, html copy of a script.

The -pre flag for code snippets
When the -pre flag is given, only the pre-formatted section,
within the "<PRE"> and "</PRE"> tags, will be output. This
simplifies inclusion of the output in other files. The default
is to output a complete web page.

The -nnn flag for line numbering
When the -nnn flag is given, the output lines will be numbered.

The -toc, or --html-table-of-contents flag
By default, a table of contents to packages and subroutines will
be written at the start of html output. Use -ntoc to prevent
this. This might be useful, for example, for a pod document
which contains a number of unrelated code snippets. This flag
only influences the code table of contents; it has no effect on
any table of contents produced by pod2html (see next item).

The -pod, or --pod2html flag
There are two options for formatting pod documentation. The
default is to pass the pod through the Pod::Html module (which
forms the basis of the pod2html utility). Any code sections are
formatted by perltidy, and the results then merged. Note:
perltidy creates a temporary file when Pod::Html is used; see
"FILES". Also, Pod::Html creates temporary files for its cache.

NOTE: Perltidy counts the number of "=cut" lines, and either
moves the pod text to the top of the html file if there is one
"=cut", or leaves the pod text in its original order (interleaved
with code) otherwise.

Most of the flags accepted by pod2html may be included in the
perltidy command line, and they will be passed to pod2html. In
some cases, the flags have a prefix "pod" to emphasize that they
are for the pod2html, and this prefix will be removed before they
are passed to pod2html. The flags which have the additional
"pod" prefix are:

--[no]podheader --[no]podindex --[no]podrecurse --[no]podquiet
--[no]podverbose --podflush

The flags which are unchanged from their use in pod2html are:

--backlink=s --cachedir=s --htmlroot=s --libpods=s --title=s
--podpath=s --podroot=s

where 's' is an appropriate character string. Not all of these
flags are available in older versions of Pod::Html. See your
Pod::Html documentation for more information.

The alternative, indicated with -npod, is not to use Pod::Html,
but rather to format pod text in italics (or whatever the
stylesheet indicates), without special html markup. This is
useful, for example, if pod is being used as an alternative way
to write comments.

The -frm, or --frames flag
By default, a single html output file is produced. This can be
changed with the -frm option, which creates a frame holding a
table of contents in the left panel and the source code in the
right side. This simplifies code browsing. Assume, for example,
that the input file is MyModule.pm. Then, for default file
extension choices, these three files will be created:

MyModule.pm.html - the frame
MyModule.pm.toc.html - the table of contents
MyModule.pm.src.html - the formatted source code

Obviously this file naming scheme requires that output be
directed to a real file (as opposed to, say, standard output).
If this is not the case, or if the file extension is unknown, the
-frm option will be ignored.

The -text=s, or --html-toc-extension flag
Use this flag to specify the extra file extension of the table of
contents file when html frames are used. The default is "toc".
See "Specifying File Extensions".

The -sext=s, or --html-src-extension flag
Use this flag to specify the extra file extension of the content
file when html frames are used. The default is "src". See
"Specifying File Extensions".

The -hent, or --html-entities flag
This flag controls the use of Html::Entities for html formatting.
By default, the module Html::Entities is used to encode special
symbols. This may not be the right thing for some
browser/language combinations. Use --nohtml-entities or -nhent
to prevent this.

Style Sheets
Style sheets make it very convenient to control and adjust the
appearance of html pages. The default behavior is to write a
page of html with an embedded style sheet.

An alternative to an embedded style sheet is to create a page
with a link to an external style sheet. This is indicated with
the -css=filename, where the external style sheet is filename.
The external style sheet filename will be created if and only if
it does not exist. This option is useful for controlling
multiple pages from a single style sheet.

To cause perltidy to write a style sheet to standard output and
exit, use the -ss, or --stylesheet, flag. This is useful if the
style sheet could not be written for some reason, such as if the
-pre flag was used. Thus, for example,

perltidy -html -ss >mystyle.css

will write a style sheet with the default properties to file
mystyle.css.

The use of style sheets is encouraged, but a web page without a
style sheets can be created with the flag -nss. Use this option
if you must to be sure that older browsers (roughly speaking,
versions prior to 4.0 of Netscape Navigator and Internet
Explorer) can display the syntax-coloring of the html files.

Controlling HTML properties
Note: It is usually more convenient to accept the default
properties and then edit the stylesheet which is produced.
However, this section shows how to control the properties with
flags to perltidy.

Syntax colors may be changed from their default values by flags
of the either the long form, -html-color-xxxxxx=n, or more
conveniently the short form, -hcx=n, where xxxxxx is one of the
following words, and x is the corresponding abbreviation:

Token Type xxxxxx x
---------- -------- --
comment comment c
number numeric n
identifier identifier i
bareword, function bareword w
keyword keyword k
quote, pattern quote q
here doc text here-doc-text h
here doc target here-doc-target hh
punctuation punctuation pu
parentheses paren p
structural braces structure s
semicolon semicolon sc
colon colon co
comma comma cm
label label j
sub definition name subroutine m
pod text pod-text pd

A default set of colors has been defined, but they may be changed
by providing values to any of the following parameters, where n
is either a 6 digit hex RGB color value or an ascii name for a
color, such as 'red'.

To illustrate, the following command will produce an html file
somefile.pl.html with "aqua" keywords:

perltidy -html -hck=00ffff somefile.pl

and this should be equivalent for most browsers:

perltidy -html -hck=aqua somefile.pl

Perltidy merely writes any non-hex names that it sees in the html
file. The following 16 color names are defined in the HTML 3.2
standard:

black => 000000,
silver => c0c0c0,
gray => 808080,
white => ffffff,
maroon => 800000,
red => ff0000,
purple => 800080,
fuchsia => ff00ff,
green => 008000,
lime => 00ff00,
olive => 808000,
yellow => ffff00
navy => 000080,
blue => 0000ff,
teal => 008080,
aqua => 00ffff,

Many more names are supported in specific browsers, but it is
safest to use the hex codes for other colors. Helpful color
tables can be located with an internet search for "HTML color
tables".

Besides color, two other character attributes may be set: bold,
and italics. To set a token type to use bold, use the flag
--html-bold-xxxxxx or -hbx, where xxxxxx or x are the long or
short names from the above table. Conversely, to set a token
type to NOT use bold, use --nohtml-bold-xxxxxx or -nhbx.

Likewise, to set a token type to use an italic font, use the flag
--html-italic-xxxxxx or -hix, where again xxxxxx or x are the
long or short names from the above table. And to set a token
type to NOT use italics, use --nohtml-italic-xxxxxx or -nhix.

For example, to use bold braces and lime color, non-bold, italics
keywords the following command would be used:

perltidy -html -hbs -hck=00FF00 -nhbk -hik somefile.pl

The background color can be specified with
--html-color-background=n, or -hcbg=n for short, where n is a 6
character hex RGB value. The default color of text is the value
given to punctuation, which is black as a default.

Here are some notes and hints:

1. If you find a preferred set of these parameters, you may want
to create a .perltidyrc file containing them. See the perltidy
man page for an explanation.

2. Rather than specifying values for these parameters, it is
probably easier to accept the defaults and then edit a style
sheet. The style sheet contains comments which should make this
easy.

3. The syntax-colored html files can be very large, so it may be
best to split large files into smaller pieces to improve download
times.

SOME COMMON INPUT CONVENTIONS

Specifying Block Types

Several parameters which refer to code block types may be customized
by also specifying an associated list of block types. The type of a
block is the name of the keyword which introduces that block, such as
if, else, or sub. An exception is a labeled block, which has no
keyword, and should be specified with just a colon. To specify all
blocks use '*'.

The keyword sub indicates a named sub. For anonymous subs, use the
special keyword asub.

For example, the following parameter specifies "sub", labels,
"BEGIN", and "END" blocks:

-cscl="sub : BEGIN END"

(the meaning of the -cscl parameter is described above.) Note that
quotes are required around the list of block types because of the
spaces. For another example, the following list specifies all block
types for vertical tightness:

-bbvtl='*'

Specifying File Extensions

Several parameters allow default file extensions to be overridden.
For example, a backup file extension may be specified with -bext=ext,
where ext is some new extension. In order to provides the user some
flexibility, the following convention is used in all cases to decide
if a leading '.' should be used. If the extension "ext" begins with
"A-Z", "a-z", or "0-9", then it will be appended to the filename with
an intermediate '.' (or perhaps a '_' on VMS systems). Otherwise, it
will be appended directly.

For example, suppose the file is somefile.pl. For "-bext=old", a '.'
is added to give somefile.pl.old. For "-bext=.old", no additional
'.' is added, so again the backup file is somefile.pl.old. For
"-bext=~", then no dot is added, and the backup file will be
somefile.pl~ .

SWITCHES WHICH MAY BE NEGATED

The following list shows all short parameter names which allow a
prefix 'n' to produce the negated form:

D aia altc ame anl asbl asc ast asu atc
atnl aws b baa baao bar bbao bbb bbc bbs
bl bli boa boc bok bol bom bos bot cblx
ce conv cpb cs csc cscb cscw dac dbc dbs
dcbl dcsc ddf dia dior dln dltc dma dmcp dmr
dnl dop dp dpro drc dsc dsm dsn dtc dtco
dtt duk duv dwic dwls dwrs dws eos f fpva
frm fs fso gcs hbc hbcm hbco hbh hbhh hbi
hbj hbk hbm hbn hbp hbpd hbpu hbq hbs hbsc
hbv hbw hent hic hicm hico hih hihh hii hij
hik him hin hip hipd hipu hiq his hisc hiv
hiw hsc html ibc icb icp ils iob ipc isbc
iscl kgb kgbd kgbi kis lal log lop lp lsl
mci mem nib ohbr okw ola olc oll olq opr
opt osbc osbr otr ple pod pvl q qwaf sac
sbc sbl scbb schb scp scsb sct se sfp sfs
skp sob sobb sohb sop sosb sot ssc st sts
t tac tbc toc tp tqw trp ts tsc tso
vbc vc viu vmll vsc vsn vwe w wfc wia
wma wme wmr wn wuk x xbt xci xlp xs

Equivalently, the prefix 'no' or 'no-' on the corresponding long
names may be used.

LIMITATIONS

Parsing Limitations
Perltidy should work properly on most perl scripts. It does a
lot of self-checking, but still, it is possible that an error
could be introduced and go undetected. Therefore, it is
essential to make careful backups and to test reformatted
scripts.

The main current limitation is that perltidy does not scan
modules included with 'use' statements. This makes it necessary
to guess the context of any bare words introduced by such
modules. Perltidy has good guessing algorithms, but they are not
infallible. When it must guess, it leaves a message in the log
file.

If you encounter a bug, please report it.

What perltidy does not parse and format
Perltidy indents but does not reformat comments and "qw" quotes.
Perltidy does not in any way modify the contents of here
documents or quoted text, even if they contain source code. (You
could, however, reformat them separately). Perltidy does not
format 'format' sections in any way. And, of course, it does not
modify pod documents.

FILES

Temporary files
Under the -html option with the default --pod2html flag, a
temporary file is required to pass text to Pod::Html. Unix
systems will try to use the POSIX tmpnam() function. Otherwise
the file perltidy.TMP will be temporarily created in the current
working directory.

Special files when standard input is used
When standard input is used, the log file, if saved, is
perltidy.LOG, and any errors are written to perltidy.ERR unless
the -se flag is set. These are saved in the current working
directory.

Files overwritten
The following file extensions are used by perltidy, and files
with these extensions may be overwritten or deleted: .ERR, .LOG,
.TEE, and/or .tdy, .html, and .bak, depending on the run type and
settings.

Files extensions limitations
Perltidy does not operate on files for which the run could
produce a file with a duplicated file extension. These
extensions include .LOG, .ERR, .TEE, and perhaps .tdy and .bak,
depending on the run type. The purpose of this rule is to
prevent generating confusing filenames such as
somefile.tdy.tdy.tdy.

ERROR HANDLING

An exit value of 0, 1, or 2 is returned by perltidy to indicate the
status of the result.

A exit value of 0 indicates that perltidy ran to completion with no
error messages.

A non-zero exit value indicates some kind of problem was detected.

An exit value of 1 indicates that perltidy terminated prematurely,
usually due to some kind of errors in the input parameters. This can
happen for example if a parameter is misspelled or given an invalid
value. Error messages in the standard error output will indicate the
cause of any problem. If perltidy terminates prematurely then no
output files will be produced.

An exit value of 2 indicates that perltidy was able to run to
completion but there there are (1) warning messages in the standard
error output related to parameter errors or problems and/or (2)
warning messages in the perltidy error file(s) relating to possible
syntax errors in one or more of the source script(s) being tidied.
When multiple files are being processed, an error detected in any
single file will produce this type of exit condition.

SEE ALSO

perlstyle(1), Perl::Tidy(3)

INSTALLATION

The perltidy binary uses the Perl::Tidy module and is installed when
that module is installed. The module name is case-sensitive. For
example, the basic command for installing with cpanm is 'cpanm
Perl::Tidy'.

VERSION

This man page documents perltidy version 20250214

BUG REPORTS

The source code repository is at
<https://github.com/perltidy/perltidy>.

To report a new bug or problem, use the "issues" link on this page.

COPYRIGHT

Copyright (c) 2000-2025 by Steve Hancock

LICENSE

This package is free software; you can redistribute it and/or modify
it under the terms of the "GNU General Public License".

Please refer to the file "COPYING" for details.

DISCLAIMER

This package is distributed in the hope that it will be useful, but
WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

See the "GNU General Public License" for more details.

perl v5.34.3 2025-02-12 PERLTIDY(1)