6 my $arg_string = undef;
8 # give Macs a chance to provide command line parameters
10 $arg_string = MacPerl::Ask(
11 'Please enter @ARGV (-h for help)',
12 defined $ARGV[0] ? "\"$ARGV[0]\"" : ""
16 # Exit codes returned by perltidy:
18 # 1 = perltidy could not run to completion due to errors
19 # 2 = perltidy ran to completion with error messages
20 exit Perl::Tidy::perltidy( argv => $arg_string );
26 perltidy - a perl script indenter and reformatter
30 perltidy [ options ] file1 file2 file3 ...
31 (output goes to file1.tdy, file2.tdy, file3.tdy, ...)
32 perltidy [ options ] file1 -o outfile
33 perltidy [ options ] file1 -st >outfile
34 perltidy [ options ] <infile >outfile
38 Perltidy reads a perl script and writes an indented, reformatted script.
40 Many users will find enough information in L<"EXAMPLES"> to get
41 started. New users may benefit from the short tutorial
43 http://perltidy.sourceforge.net/tutorial.html
45 A convenient aid to systematically defining a set of style parameters
47 http://perltidy.sourceforge.net/stylekey.html
49 Perltidy can produce output on either of two modes, depending on the
50 existence of an B<-html> flag. Without this flag, the output is passed
51 through a formatter. The default formatting tries to follow the
52 recommendations in perlstyle(1), but it can be controlled in detail with
53 numerous input parameters, which are described in L<"FORMATTING
56 When the B<-html> flag is given, the output is passed through an HTML
57 formatter which is described in L<"HTML OPTIONS">.
63 This will produce a file F<somefile.pl.tdy> containing the script reformatted
64 using the default options, which approximate the style suggested in
65 perlstyle(1). The source file F<somefile.pl> is unchanged.
69 Execute perltidy on all F<.pl> files in the current directory with the
70 default options. The output will be in files with an appended F<.tdy>
71 extension. For any file with an error, there will be a file with extension
74 perltidy -b file1.pl file2.pl
76 Modify F<file1.pl> and F<file2.pl> in place, and backup the originals to
77 F<file1.pl.bak> and F<file2.pl.bak>. If F<file1.pl.bak> and/or F<file2.pl.bak>
78 already exist, they will be overwritten.
80 perltidy -b -bext='/' file1.pl file2.pl
82 Same as the previous example except that the backup files F<file1.pl.bak> and F<file2.pl.bak> will be deleted if there are no errors.
84 perltidy -gnu somefile.pl
86 Execute perltidy on file F<somefile.pl> with a style which approximates the
87 GNU Coding Standards for C programs. The output will be F<somefile.pl.tdy>.
89 perltidy -i=3 somefile.pl
91 Execute perltidy on file F<somefile.pl>, with 3 columns for each level of
92 indentation (B<-i=3>) instead of the default 4 columns. There will not be any
93 tabs in the reformatted script, except for any which already exist in comments,
94 pod documents, quotes, and here documents. Output will be F<somefile.pl.tdy>.
96 perltidy -i=3 -et=8 somefile.pl
98 Same as the previous example, except that leading whitespace will
99 be entabbed with one tab character per 8 spaces.
101 perltidy -ce -l=72 somefile.pl
103 Execute perltidy on file F<somefile.pl> with all defaults except use "cuddled
104 elses" (B<-ce>) and a maximum line length of 72 columns (B<-l=72>) instead of
105 the default 80 columns.
107 perltidy -g somefile.pl
109 Execute perltidy on file F<somefile.pl> and save a log file F<somefile.pl.LOG>
110 which shows the nesting of braces, parentheses, and square brackets at
111 the start of every line.
113 perltidy -html somefile.pl
115 This will produce a file F<somefile.pl.html> containing the script with
116 html markup. The output file will contain an embedded style sheet in
117 the <HEAD> section which may be edited to change the appearance.
119 perltidy -html -css=mystyle.css somefile.pl
121 This will produce a file F<somefile.pl.html> containing the script with
122 html markup. This output file will contain a link to a separate style
123 sheet file F<mystyle.css>. If the file F<mystyle.css> does not exist,
124 it will be created. If it exists, it will not be overwritten.
126 perltidy -html -pre somefile.pl
128 Write an html snippet with only the PRE section to F<somefile.pl.html>.
129 This is useful when code snippets are being formatted for inclusion in a
130 larger web page. No style sheet will be written in this case.
132 perltidy -html -ss >mystyle.css
134 Write a style sheet to F<mystyle.css> and exit.
136 perltidy -html -frm mymodule.pm
138 Write html with a frame holding a table of contents and the source code. The
139 output files will be F<mymodule.pm.html> (the frame), F<mymodule.pm.toc.html>
140 (the table of contents), and F<mymodule.pm.src.html> (the source code).
142 =head1 OPTIONS - OVERVIEW
144 The entire command line is scanned for options, and they are processed
145 before any files are processed. As a result, it does not matter
146 whether flags are before or after any filenames. However, the relative
147 order of parameters is important, with later parameters overriding the
148 values of earlier parameters.
150 For each parameter, there is a long name and a short name. The short
151 names are convenient for keyboard input, while the long names are
152 self-documenting and therefore useful in scripts. It is customary to
153 use two leading dashes for long names, but one may be used.
155 Most parameters which serve as on/off flags can be negated with a
156 leading "n" (for the short name) or a leading "no" or "no-" (for the
157 long name). For example, the flag to outdent long quotes is B<-olq>
158 or B<--outdent-long-quotes>. The flag to skip this is B<-nolq>
159 or B<--nooutdent-long-quotes> or B<--no-outdent-long-quotes>.
161 Options may not be bundled together. In other words, options B<-q> and
162 B<-g> may NOT be entered as B<-qg>.
164 Option names may be terminated early as long as they are uniquely identified.
165 For example, instead of B<--dump-token-types>, it would be sufficient to enter
166 B<--dump-tok>, or even B<--dump-t>, to uniquely identify this command.
170 The following parameters concern the files which are read and written.
174 =item B<-h>, B<--help>
176 Show summary of usage and exit.
178 =item B<-o>=filename, B<--outfile>=filename
180 Name of the output file (only if a single input file is being
181 processed). If no output file is specified, and output is not
182 redirected to the standard output (see B<-st>), the output will go to
183 F<filename.tdy>. [Note: - does not redirect to standard output. Use
186 =item B<-st>, B<--standard-output>
188 Perltidy must be able to operate on an arbitrarily large number of files
189 in a single run, with each output being directed to a different output
190 file. Obviously this would conflict with outputting to the single
191 standard output device, so a special flag, B<-st>, is required to
192 request outputting to the standard output. For example,
194 perltidy somefile.pl -st >somefile.new.pl
196 This option may only be used if there is just a single input file.
197 The default is B<-nst> or B<--nostandard-output>.
199 =item B<-se>, B<--standard-error-output>
201 If perltidy detects an error when processing file F<somefile.pl>, its
202 default behavior is to write error messages to file F<somefile.pl.ERR>.
203 Use B<-se> to cause all error messages to be sent to the standard error
204 output stream instead. This directive may be negated with B<-nse>.
205 Thus, you may place B<-se> in a F<.perltidyrc> and override it when
206 desired with B<-nse> on the command line.
208 =item B<-oext>=ext, B<--output-file-extension>=ext
210 Change the extension of the output file to be F<ext> instead of the
211 default F<tdy> (or F<html> in case the -B<-html> option is used).
212 See L<Specifying File Extensions>.
214 =item B<-opath>=path, B<--output-path>=path
216 When perltidy creates a filename for an output file, by default it merely
217 appends an extension to the path and basename of the input file. This
218 parameter causes the path to be changed to F<path> instead.
220 The path should end in a valid path separator character, but perltidy will try
221 to add one if it is missing.
225 perltidy somefile.pl -opath=/tmp/
227 will produce F</tmp/somefile.pl.tdy>. Otherwise, F<somefile.pl.tdy> will
228 appear in whatever directory contains F<somefile.pl>.
230 If the path contains spaces, it should be placed in quotes.
232 This parameter will be ignored if output is being directed to standard output,
233 or if it is being specified explicitly with the B<-o=s> parameter.
235 =item B<-b>, B<--backup-and-modify-in-place>
237 Modify the input file or files in-place and save the original with the
238 extension F<.bak>. Any existing F<.bak> file will be deleted. See next
239 item for changing the default backup extension, and for eliminating the
240 backup file altogether.
242 A B<-b> flag will be ignored if input is from standard input or goes to
243 standard output, or if the B<-html> flag is set.
245 In particular, if you want to use both the B<-b> flag and the B<-pbp>
246 (--perl-best-practices) flag, then you must put a B<-nst> flag after the
247 B<-pbp> flag because it contains a B<-st> flag as one of its components,
248 which means that output will go to the standard output stream.
250 =item B<-bext>=ext, B<--backup-file-extension>=ext
252 This parameter serves two purposes: (1) to change the extension of the backup
253 file to be something other than the default F<.bak>, and (2) to indicate
254 that no backup file should be saved.
256 To change the default extension to something other than F<.bak> see
257 L<Specifying File Extensions>.
259 A backup file of the source is always written, but you can request that it
260 be deleted at the end of processing if there were no errors. This is risky
261 unless the source code is being maintained with a source code control
264 To indicate that the backup should be deleted include one forward slash,
265 B</>, in the extension. If any text remains after the slash is removed
266 it will be used to define the backup file extension (which is always
267 created and only deleted if there were no errors).
269 Here are some examples:
271 Parameter Extension Backup File Treatment
272 <-bext=bak> F<.bak> Keep (same as the default behavior)
273 <-bext='/'> F<.bak> Delete if no errors
274 <-bext='/backup'> F<.backup> Delete if no errors
275 <-bext='original/'> F<.original> Delete if no errors
277 =item B<-w>, B<--warning-output>
279 Setting B<-w> causes any non-critical warning
280 messages to be reported as errors. These include messages
281 about possible pod problems, possibly bad starting indentation level,
282 and cautions about indirect object usage. The default, B<-nw> or
283 B<--nowarning-output>, is not to include these warnings.
285 =item B<-q>, B<--quiet>
287 Deactivate error messages (for running under an editor).
289 For example, if you use a vi-style editor, such as vim, you may execute
290 perltidy as a filter from within the editor using something like
294 where C<n1,n2> represents the selected text. Without the B<-q> flag,
295 any error message may mess up your screen, so be prepared to use your
298 =item B<-log>, B<--logfile>
300 Save the F<.LOG> file, which has many useful diagnostics. Perltidy always
301 creates a F<.LOG> file, but by default it is deleted unless a program bug is
302 suspected. Setting the B<-log> flag forces the log file to be saved.
304 =item B<-g=n>, B<--logfile-gap=n>
306 Set maximum interval between input code lines in the logfile. This purpose of
307 this flag is to assist in debugging nesting errors. The value of C<n> is
308 optional. If you set the flag B<-g> without the value of C<n>, it will be
309 taken to be 1, meaning that every line will be written to the log file. This
310 can be helpful if you are looking for a brace, paren, or bracket nesting error.
312 Setting B<-g> also causes the logfile to be saved, so it is not necessary to
313 also include B<-log>.
315 If no B<-g> flag is given, a value of 50 will be used, meaning that at least
316 every 50th line will be recorded in the logfile. This helps prevent
317 excessively long log files.
319 Setting a negative value of C<n> is the same as not setting B<-g> at all.
321 =item B<-npro> B<--noprofile>
323 Ignore any F<.perltidyrc> command file. Normally, perltidy looks first in
324 your current directory for a F<.perltidyrc> file of parameters. (The format
325 is described below). If it finds one, it applies those options to the
326 initial default values, and then it applies any that have been defined
327 on the command line. If no F<.perltidyrc> file is found, it looks for one
328 in your home directory.
330 If you set the B<-npro> flag, perltidy will not look for this file.
332 =item B<-pro=filename> or B<--profile=filename>
334 To simplify testing and switching .perltidyrc files, this command may be
335 used to specify a configuration file which will override the default
336 name of .perltidyrc. There must not be a space on either side of the
337 '=' sign. For example, the line
339 perltidy -pro=testcfg
341 would cause file F<testcfg> to be used instead of the
342 default F<.perltidyrc>.
344 A pathname begins with three dots, e.g. ".../.perltidyrc", indicates that
345 the file should be searched for starting in the current directory and
346 working upwards. This makes it easier to have multiple projects each with
347 their own .perltidyrc in their root directories.
349 =item B<-opt>, B<--show-options>
351 Write a list of all options used to the F<.LOG> file.
352 Please see B<--dump-options> for a simpler way to do this.
354 =item B<-f>, B<--force-read-binary>
356 Force perltidy to process binary files. To avoid producing excessive
357 error messages, perltidy skips files identified by the system as non-text.
358 However, valid perl scripts containing binary data may sometimes be identified
359 as non-text, and this flag forces perltidy to process them.
361 =item B<-ast>, B<--assert-tidy>
363 This flag asserts that the input and output code streams are identical, or in
364 other words that the input code is already 'tidy' according to the formatting
365 parameters. If this is not the case, an error message noting this is produced.
366 This error message will cause the process to return a non-zero exit code.
367 The test for this is made by comparing an MD5 hash value for the input and
368 output code streams. This flag has no other effect on the functioning of
369 perltidy. This might be useful for certain code maintenance operations.
370 Note: you will not see this message if you have error messages turned off with the
373 =item B<-asu>, B<--assert-untidy>
375 This flag asserts that the input and output code streams are different, or in
376 other words that the input code is 'untidy' according to the formatting
377 parameters. If this is not the case, an error message noting this is produced.
378 This flag has no other effect on the functioning of perltidy.
380 =item B<-sal=s>, B<--sub-alias-list=s>
382 This flag causes one or more words to be treated the same as if they were the keyword 'sub'. The string B<s> contains one or more alias words, separated by spaces or commas.
386 perltidy -sal='method fun _sub M4'
388 will cause the perltidy to treate the words 'method', 'fun', '_sub' and 'M4' to be treated 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.
390 Note that several other parameters accept a list of keywords, including 'sub' (see L<Specifying Block Types>).
391 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.
395 =head1 FORMATTING OPTIONS
403 This flag disables all formatting and causes the input to be copied unchanged
404 to the output except for possible changes in line ending characters and any
405 pre- and post-filters. This can be useful in conjunction with a hierarchical
406 set of F<.perltidyrc> files to avoid unwanted code tidying. See also
407 L<Skipping Selected Sections of Code> for a way to avoid tidying specific
410 =item B<-i=n>, B<--indent-columns=n>
412 Use n columns per indentation level (default n=4).
414 =item B<-l=n>, B<--maximum-line-length=n>
416 The default maximum line length is n=80 characters. Perltidy will try
417 to find line break points to keep lines below this length. However, long
418 quotes and side comments may cause lines to exceed this length.
420 The default length of 80 comes from the past when this was the standard CRT
421 screen width. Many programmers prefer to increase this to something like 120.
423 Setting B<-l=0> is equivalent to setting B<-l=(a very large number)>. But this is
424 not recommended because, for example, a very long list will be formatted in a
427 =item B<-vmll>, B<--variable-maximum-line-length>
429 A problem arises using a fixed maximum line length with very deeply nested code
430 and data structures because eventually the amount of leading whitespace used
431 for indicating indentation takes up most or all of the available line width,
432 leaving little or no space for the actual code or data. One solution is to use
433 a vary long line length. Another solution is to use the B<-vmll> flag, which
434 basically tells perltidy to ignore leading whitespace when measuring the line
437 To be precise, when the B<-vmll> parameter is set, the maximum line length of a
438 line of code will be M+L*I, where
440 M is the value of --maximum-line-length=M (-l=M), default 80,
441 I is the value of --indent-columns=I (-i=I), default 4,
442 L is the indentation level of the line of code
444 When this flag is set, the choice of breakpoints for a block of code should be
445 essentially independent of its nesting depth. However, the absolute line
446 lengths, including leading whitespace, can still be arbitrarily large. This
447 problem can be avoided by including the next parameter.
449 The default is not to do this (B<-nvmll>).
451 =item B<-wc=n>, B<--whitespace-cycle=n>
453 This flag also addresses problems with very deeply nested code and data
454 structures. When the nesting depth exceeds the value B<n> the leading
455 whitespace will be reduced and start at a depth of 1 again. The result is that
456 blocks of code will shift back to the left rather than moving arbitrarily far
457 to the right. This occurs cyclically to any depth.
459 For example if one level of indentation equals 4 spaces (B<-i=4>, the default),
460 and one uses B<-wc=15>, then if the leading whitespace on a line exceeds about
461 4*15=60 spaces it will be reduced back to 4*1=4 spaces and continue increasing
462 from there. If the whitespace never exceeds this limit the formatting remains
465 The combination of B<-vmll> and B<-wc=n> provides a solution to the problem of
466 displaying arbitrarily deep data structures and code in a finite window,
467 although B<-wc=n> may of course be used without B<-vmll>.
469 The default is not to use this, which can also be indicated using B<-wc=0>.
473 Using tab characters will almost certainly lead to future portability
474 and maintenance problems, so the default and recommendation is not to
475 use them. For those who prefer tabs, however, there are two different
478 Except for possibly introducing tab indentation characters, as outlined
479 below, perltidy does not introduce any tab characters into your file,
480 and it removes any tabs from the code (unless requested not to do so
481 with B<-fws>). If you have any tabs in your comments, quotes, or
482 here-documents, they will remain.
486 =item B<-et=n>, B<--entab-leading-whitespace>
488 This flag causes each B<n> initial space characters to be replaced by
491 The value of the integer B<n> can be any value but can be coordinated with the
492 number of spaces used for intentation. For example, B<-et=4 -ci=4 -i=4> will
493 produce one tab for each indentation level and and one for each continuation
494 indentation level. You may want to coordinate the value of B<n> with what your
495 display software assumes for the spacing of a tab.
497 =item B<-t>, B<--tabs>
499 This flag causes one leading tab character to be inserted for each level
500 of indentation. Certain other features are incompatible with this
501 option, and if these options are also given, then a warning message will
502 be issued and this flag will be unset. One example is the B<-lp>
503 option. This flag is retained for backwards compatibility, but
504 if you use tabs, the B<-et=n> flag is recommended.
506 =item B<-dt=n>, B<--default-tabsize=n>
508 If the first line of code passed to perltidy contains leading tabs but no
509 tab scheme is specified for the output stream then perltidy must guess how many
510 spaces correspond to each leading tab. This number of spaces B<n>
511 corresponding to each leading tab of the input stream may be specified with
512 B<-dt=n>. The default is B<n=8>.
514 This flag has no effect if a tab scheme is specified for the output stream,
515 because then the input stream is assumed to use the same tab scheme and
516 indentation spaces as for the output stream (any other assumption would lead to
521 =item B<-xs>, B<--extended-syntax>
523 A problem with formatting Perl code is that some modules can introduce new
524 syntax. This flag allows perltidy to handle certain common extensions
525 to the standard syntax without complaint.
527 For example, without this flag a structure such as the following would generate
528 a syntax error and the braces would not be balanced:
530 method deposit( Num $amount) {
531 $self->balance( $self->balance + $amount );
534 For one of the extensions, module Switch::Plain, colons are marked as labels.
535 If you use this module, you may want to also use the B<--nooutdent-labels> flag
536 to prevent lines such as 'default:' from being outdented.
538 This flag is enabled by default but it can be deactivated with B<-nxs>.
539 Probably the only reason to deactivate this flag is to generate more diagnostic
540 messages when debugging a script.
542 For another method of handling extended syntax see the section L<Skipping Selected Sections of Code>.
544 =item B<-io>, B<--indent-only>
546 This flag is used to deactivate all whitespace and line break changes
547 within non-blank lines of code.
548 When it is in effect, the only change to the script will be
549 to the indentation and to the number of blank lines.
550 And any flags controlling whitespace and newlines will be ignored. You
551 might want to use this if you are perfectly happy with your whitespace
552 and line breaks, and merely want perltidy to handle the indentation.
553 (This also speeds up perltidy by well over a factor of two, so it might be
554 useful when perltidy is merely being used to help find a brace error in
557 Setting this flag is equivalent to setting B<--freeze-newlines> and
558 B<--freeze-whitespace>.
560 If you also want to keep your existing blank lines exactly
561 as they are, you can add B<--freeze-blank-lines>.
563 With this option perltidy is still free to modify the indenting (and
564 outdenting) of code and comments as it normally would. If you also want to
565 prevent long comment lines from being outdented, you can add either B<-noll> or
568 Setting this flag will prevent perltidy from doing any special operations on
569 closing side comments. You may still delete all side comments however when
570 this flag is in effect.
573 =item B<-enc=s>, B<--character-encoding=s>
575 This flag indicates the character encoding, if any, of the input data stream.
576 Perltidy does not look for the encoding directives in the soure stream, such
577 as B<use utf8>, and instead relies on this flag to determine the encoding.
578 (Note that perltidy often works on snippets of code rather than complete files
579 so it cannot rely on B<use utf8> directives).
581 The possible values for B<s> are (1) the name of an encoding recognized by the
582 Encode.pm module, (2) B<none> if no encoding is used, or (3) <guess> if
583 perltidy should guess.
585 For example, the value B<utf8> causes the stream to be read and written as
586 UTF-8. If the input stream cannot be decoded with a specified encoding then
587 processing is not done.
589 The value B<none> causes the stream to be processed without special encoding
590 assumptions. This is appropriate for files which are written in single-byte
591 character encodings such as latin-1.
593 The value B<guess> tells perltidy to guess between either utf8 encoding or no
594 encoding (meaning one character per byte). The guess uses the Encode::Guess
595 module and this restricted range of guesses covers the most common cases.
596 Testing showed that considering any greater number of encodings as guess
597 suspects is too risky.
599 The current default is B<guess>.
601 The abbreviations B<-utf8> or B<-UTF8> are equivalent to B<-enc=utf8>, and the
602 abbreviation B<-guess> is equivalent to <-enc=guess>. So to process a file
603 named B<file.pl> which is encoded in UTF-8 you can use:
605 perltidy -utf8 file.pl
608 perltidy -guess file.pl
610 To process a file in B<euc-jp> you could use
612 perltidy -enc=euc-jp file.pl
614 A perltidy output file is unencoded if the input file is unencoded, and
615 otherwise it is encoded as B<utf8>, even if the input encoding was not
618 =item B<-gcs>, B<--use-unicode-gcstring>
620 This flag controls whether or not perltidy may use module Unicode::GCString to
621 obtain accurate display widths of wide characters. The default
622 is B<--nouse-unicode-gcstring>.
624 If this flag is set, and text is encoded, perltidy will look for the module
625 Unicode::GCString and, if found, will use it to obtain character display
626 widths. This can improve displayed vertical alignment for files with wide
627 characters. It is a nice feature but it is off by default to avoid conflicting
628 formatting when there are multiple developers. Perltidy installation does not
629 require Unicode::GCString, so users wanting to use this feature need set this
630 flag and also to install Unicode::GCString separately.
632 If this flag is set and perltidy does not find module Unicode::GCString,
633 a warning message will be produced and processing will continue but without
634 the potential benefit provided by the module.
636 Also note that actual vertical alignment depends upon the fonts used by the
637 text display software, so vertical alignment may not be optimal even when
638 Unicode::GCString is used.
640 =item B<-ole=s>, B<--output-line-ending=s>
642 where s=C<win>, C<dos>, C<unix>, or C<mac>. This flag tells perltidy
643 to output line endings for a specific system. Normally,
644 perltidy writes files with the line separator character of the host
645 system. The C<win> and C<dos> flags have an identical result.
647 =item B<-ple>, B<--preserve-line-endings>
649 This flag tells perltidy to write its output files with the same line
650 endings as the input file, if possible. It should work for
651 B<dos>, B<unix>, and B<mac> line endings. It will only work if perltidy
652 input comes from a filename (rather than stdin, for example). If
653 perltidy has trouble determining the input file line ending, it will
654 revert to the default behavior of using the line ending of the host system.
656 =item B<-atnl>, B<--add-terminal-newline>
658 This flag, which is enabled by default, allows perltidy to terminate the last
659 line of the output stream with a newline character, regardless of whether or
660 not the input stream was terminated with a newline character. If this flag is
661 negated, with B<-natnl>, then perltidy will add a terminal newline to the the
662 output stream only if the input stream is terminated with a newline.
664 Negating this flag may be useful for manipulating one-line scripts intended for
665 use on a command line.
667 =item B<-it=n>, B<--iterations=n>
669 This flag causes perltidy to do B<n> complete iterations. The reason for this
670 flag is that code beautification is an iterative process and in some
671 cases the output from perltidy can be different if it is applied a second time.
672 For most purposes the default of B<n=1> should be satisfactory. However B<n=2>
673 can be useful when a major style change is being made, or when code is being
674 beautified on check-in to a source code control system. It has been found to
675 be extremely rare for the output to change after 2 iterations. If a value
676 B<n> is greater than 2 is input then a convergence test will be used to stop
677 the iterations as soon as possible, almost always after 2 iterations. See
678 the next item for a simplified iteration control.
680 This flag has no effect when perltidy is used to generate html.
682 =item B<-conv>, B<--converge>
684 This flag is equivalent to B<-it=4> and is included to simplify iteration
685 control. For all practical purposes one either does or does not want to be
686 sure that the output is converged, and there is no penalty to using a large
687 iteration limit since perltidy will check for convergence and stop iterating as
688 soon as possible. The default is B<-nconv> (no convergence check). Using
689 B<-conv> will approximately double run time since typically one extra iteration
690 is required to verify convergence. No extra iterations are required if no new
691 line breaks are made, and two extra iterations are occasionally needed when
692 reformatting complex code structures, such as deeply nested ternary statements.
696 =head2 Code Indentation Control
700 =item B<-ci=n>, B<--continuation-indentation=n>
702 Continuation indentation is extra indentation spaces applied when
703 a long line is broken. The default is n=2, illustrated here:
706 ( $max_index_to_go >= 0 ) ? $levels_to_go[0] : $last_output_level;
708 The same example, with n=0, is a little harder to read:
711 ( $max_index_to_go >= 0 ) ? $levels_to_go[0] : $last_output_level;
713 The value given to B<-ci> is also used by some commands when a small
714 space is required. Examples are commands for outdenting labels,
715 B<-ola>, and control keywords, B<-okw>.
717 When default values are not used, it is recommended that either
719 (1) the value B<n> given with B<-ci=n> be no more than about one-half of the
720 number of spaces assigned to a full indentation level on the B<-i=n> command, or
722 (2) the flag B<-extended-continuation-indentation> is used (see next section).
724 =item B<-xci>, B<--extended-continuation-indentation>
726 This flag allows perltidy to use some improvements which have been made to its
727 indentation model. One of the things it does is "extend" continuation
728 indentation deeper into structures, hence the name. The improved indentation
729 is particularly noticeable when the flags B<-ci=n> and B<-i=n> use the same value of
730 B<n>. There are no significant disadvantages to using this flag, but to avoid
731 disturbing existing formatting the default is not to use it, B<-nxci>.
733 Please see the section L<"B<-pbp>, B<--perl-best-practices>"> for an example of
734 how this flag can improve the formatting of ternary statements. It can also
735 improve indentation of some multi-line qw lists as shown below.
740 AntiqueWhite3 Bisque1 Bisque2 Bisque3 Bisque4
741 SlateBlue3 RoyalBlue1 SteelBlue2 DeepSkyBlue3
744 LightBlue1 DarkSlateGray1 Aquamarine2 DarkSeaGreen2
745 SeaGreen1 Yellow1 IndianRed1 IndianRed2 Tan1 Tan4
752 AntiqueWhite3 Bisque1 Bisque2 Bisque3 Bisque4
753 SlateBlue3 RoyalBlue1 SteelBlue2 DeepSkyBlue3
756 LightBlue1 DarkSlateGray1 Aquamarine2 DarkSeaGreen2
757 SeaGreen1 Yellow1 IndianRed1 IndianRed2 Tan1 Tan4
761 =item B<-sil=n> B<--starting-indentation-level=n>
763 By default, perltidy examines the input file and tries to determine the
764 starting indentation level. While it is often zero, it may not be
765 zero for a code snippet being sent from an editing session.
767 To guess the starting indentation level perltidy simply assumes that
768 indentation scheme used to create the code snippet is the same as is being used
769 for the current perltidy process. This is the only sensible guess that can be
770 made. It should be correct if this is true, but otherwise it probably won't.
771 For example, if the input script was written with -i=2 and the current peltidy
772 flags have -i=4, the wrong initial indentation will be guessed for a code
773 snippet which has non-zero initial indentation. Likewise, if an entabbing
774 scheme is used in the input script and not in the current process then the
775 guessed indentation will be wrong.
777 If the default method does not work correctly, or you want to change the
778 starting level, use B<-sil=n>, to force the starting level to be n.
780 =item B<List indentation> using B<-lp>, B<--line-up-parentheses>
782 By default, perltidy indents lists with 4 spaces, or whatever value
783 is specified with B<-i=n>. Here is a small list formatted in this way:
787 'Jan', 'Feb', 'Mar', 'Apr', 'May', 'Jun',
788 'Jul', 'Aug', 'Sep', 'Oct', 'Nov', 'Dec'
791 Use the B<-lp> flag to add extra indentation to cause the data to begin
792 past the opening parentheses of a sub call or list, or opening square
793 bracket of an anonymous array, or opening curly brace of an anonymous
794 hash. With this option, the above list would become:
798 'Jan', 'Feb', 'Mar', 'Apr', 'May', 'Jun',
799 'Jul', 'Aug', 'Sep', 'Oct', 'Nov', 'Dec'
802 If the available line length (see B<-l=n> ) does not permit this much
803 space, perltidy will use less. For alternate placement of the
804 closing paren, see the next section.
806 This option has no effect on code BLOCKS, such as if/then/else blocks,
807 which always use whatever is specified with B<-i=n>.
809 In situations where perltidy does not have complete freedom to choose line
810 breaks it may temporarily revert to its default indentation method. This can
811 occur for example if there are blank lines, block comments, multi-line quotes,
812 or side comments between the opening and closing parens, braces, or brackets.
814 In addition, any parameter which significantly restricts the ability of
815 perltidy to choose newlines will conflict with B<-lp> and will cause
816 B<-lp> to be deactivated. These include B<-io>, B<-fnl>, B<-nanl>, and
817 B<-ndnl>. The reason is that the B<-lp> indentation style can require
818 the careful coordination of an arbitrary number of break points in
819 hierarchical lists, and these flags may prevent that.
821 The B<-lp> option may not be used together with the B<-t> tabs option.
822 It may, however, be used with the B<-et=n> tab method.
825 =item B<-lpxl=s>, B<--line-up-parentheses-exclusion-list>
827 This is an experimental parameter; the details might change as experience
830 The B<-lp> indentation style works well for some types of coding but can
831 produce very long lines when variables have long names and/or containers are
832 very deeply nested. The B<-lpxl=s> flag is intended to help mitigate this problem by
833 providing control over the containers to which the B<-lp> indentation style is
834 applied. The B<-lp> flag by default is "greedy" and applies to as many
835 containers as possible. This flag specifies a list of things which should
836 B<not> be use B<-lp> indentation.
838 This list is a string with space-separated items. Each item consists of up to
839 three pieces of information in this order: (1) an optional letter code (2) a
840 required container type, and (3) an optional numeric code.
842 The only required piece of information is a container type, which is one of
843 '(', '[', or '{'. For example the string
847 means do B<NOT> include use -lp formatting within square-bracets or braces. The only unspecified
848 container is '(', so this string means that only the contents within parens will use -lp indentation.
850 An optional numeric code may follow any of the container types to further refine the selection based
851 on container contents. The numeric codes are:
853 '0' or blank: no check on contents
854 '1' reject -lp unless the contents is a simple list without sublists
855 '2' reject -lp unless the contents is a simple list without sublists, without
856 code blocks, and without ternary operators
862 means only apply -lp to parenthesized lists which do not contain any sublists,
863 code blocks or ternary expressions.
865 A third optional item of information which can be given for parens is an alphanumeric
866 letter which is used to limit the selection further depending on the type of
867 token immediately before the paren. The possible letters are currently 'k',
868 'K', 'f', 'F', 'w', and 'W', with these meanings:
870 'k' matches if the previous nonblank token is a perl builtin keyword (such as 'if', 'while'),
871 'K' matches if 'k' does not, meaning that the previous token is not a keyword.
872 'f' matches if the previous token is a function other than a keyword.
873 'F' matches if 'f' does not.
874 'w' matches if either 'k' or 'f' match.
875 'W' matches if 'w' does not.
881 means only apply -lp to parenthesized lists which follow a function call and
882 which do not contain any sublists, code blocks or ternary expressions. The logic
883 of writing these codes is somewhat counter-intuitive because they describe what is not
884 getting the -lp indentation. So the 'F' indicates that non-function calls are
885 not getting -lp, or in other words that function calls are getting the -lp indentation.
887 =item B<-cti=n>, B<--closing-token-indentation>
889 The B<-cti=n> flag controls the indentation of a line beginning with
890 a C<)>, C<]>, or a non-block C<}>. Such a line receives:
892 -cti = 0 no extra indentation (default)
893 -cti = 1 extra indentation such that the closing token
894 aligns with its opening token.
895 -cti = 2 one extra indentation level if the line looks like:
897 -cti = 3 one extra indentation level always
899 The flags B<-cti=1> and B<-cti=2> work well with the B<-lp> flag (previous
902 # perltidy -lp -cti=1
904 'Jan', 'Feb', 'Mar', 'Apr', 'May', 'Jun',
905 'Jul', 'Aug', 'Sep', 'Oct', 'Nov', 'Dec'
908 # perltidy -lp -cti=2
910 'Jan', 'Feb', 'Mar', 'Apr', 'May', 'Jun',
911 'Jul', 'Aug', 'Sep', 'Oct', 'Nov', 'Dec'
914 These flags are merely hints to the formatter and they may not always be
915 followed. In particular, if -lp is not being used, the indentation for
916 B<cti=1> is constrained to be no more than one indentation level.
918 If desired, this control can be applied independently to each of the
919 closing container token types. In fact, B<-cti=n> is merely an
920 abbreviation for B<-cpi=n -csbi=n -cbi=n>, where:
921 B<-cpi> or B<--closing-paren-indentation> controls B<)>'s,
922 B<-csbi> or B<--closing-square-bracket-indentation> controls B<]>'s,
923 B<-cbi> or B<--closing-brace-indentation> controls non-block B<}>'s.
925 =item B<-icp>, B<--indent-closing-paren>
927 The B<-icp> flag is equivalent to
928 B<-cti=2>, described in the previous section. The B<-nicp> flag is
929 equivalent B<-cti=0>. They are included for backwards compatibility.
931 =item B<-icb>, B<--indent-closing-brace>
933 The B<-icb> option gives one extra level of indentation to a brace which
934 terminates a code block . For example,
943 The default is not to do this, indicated by B<-nicb>.
946 =item B<-nib>, B<--non-indenting-braces>
948 Normally, lines of code contained within a pair of block braces receive one
949 additional level of indentation. This flag, which is enabled by default,
950 causes perltidy to look for
951 opening block braces which are followed by a special side comment. This special
952 side comment is B<#<<<> by default. If found, the code between this opening brace and its
953 corresponding closing brace will not be given the normal extra indentation
956 { #<<< a closure to contain lexical vars
958 my $var; # this line does not get one level of indentation
963 # this line does not 'see' $var;
965 This can be useful, for example, when combining code from different files.
966 Different sections of code can be placed within braces to keep their lexical
967 variables from being visible to the end of the file. To keep the new braces
968 from causing all of their contained code to be indented if you run perltidy,
969 and possibly introducing new line breaks in long lines, you can mark the
970 opening braces with this special side comment.
972 Only the opening brace needs to be marked, since perltidy knows where the
973 closing brace is. Braces contained within marked braces may also be marked
976 If your code happens to have some opening braces followed by '#<<<', and you
977 don't want this behavior, you can use B<-nnib> to deactivate it. To make it
978 easy to remember, the default string is the same as the string for starting a
979 B<format-skipping> section. There is no confusion because in that case it is
980 for a block comment rather than a side-comment.
982 The special side comment can be changed with the next parameter.
985 =item B<-nibp=s>, B<--non-indenting-brace-prefix=s>
987 The B<-nibp=string> parameter may be used to change the marker for
988 non-indenting braces. The default is equivalent to -nibp='#<<<'. The string
989 that you enter must begin with a # and should be in quotes as necessary to get
990 past the command shell of your system. This string is the leading text of a
991 regex pattern that is constructed by appending pre-pending a '^' and appending
992 a'\s', so you must also include backslashes for characters to be taken
993 literally rather than as patterns.
995 For example, to match the side comment '#++', the parameter would be
1000 =item B<-olq>, B<--outdent-long-quotes>
1002 When B<-olq> is set, lines which is a quoted string longer than the
1003 value B<maximum-line-length> will have their indentation removed to make
1004 them more readable. This is the default. To prevent such out-denting,
1005 use B<-nolq> or B<--nooutdent-long-lines>.
1007 =item B<-oll>, B<--outdent-long-lines>
1009 This command is equivalent to B<--outdent-long-quotes> and
1010 B<--outdent-long-comments>, and it is included for compatibility with previous
1011 versions of perltidy. The negation of this also works, B<-noll> or
1012 B<--nooutdent-long-lines>, and is equivalent to setting B<-nolq> and B<-nolc>.
1014 =item B<Outdenting Labels:> B<-ola>, B<--outdent-labels>
1016 This command will cause labels to be outdented by 2 spaces (or whatever B<-ci>
1017 has been set to), if possible. This is the default. For example:
1020 LOOP: while ( $i = <FOTOS> ) {
1026 Use B<-nola> to not outdent labels.
1028 =item B<Outdenting Keywords>
1032 =item B<-okw>, B<--outdent-keywords>
1034 The command B<-okw> will cause certain leading control keywords to
1035 be outdented by 2 spaces (or whatever B<-ci> has been set to), if
1036 possible. By default, these keywords are C<redo>, C<next>, C<last>,
1037 C<goto>, and C<return>. The intention is to make these control keywords
1038 easier to see. To change this list of keywords being outdented, see
1041 For example, using C<perltidy -okw> on the previous example gives:
1044 LOOP: while ( $i = <FOTOS> ) {
1050 The default is not to do this.
1052 =item B<Specifying Outdented Keywords:> B<-okwl=string>, B<--outdent-keyword-list=string>
1054 This command can be used to change the keywords which are outdented with
1055 the B<-okw> command. The parameter B<string> is a required list of perl
1056 keywords, which should be placed in quotes if there are more than one.
1057 By itself, it does not cause any outdenting to occur, so the B<-okw>
1058 command is still required.
1060 For example, the commands C<-okwl="next last redo goto" -okw> will cause
1061 those four keywords to be outdented. It is probably simplest to place
1062 any B<-okwl> command in a F<.perltidyrc> file.
1068 =head2 Whitespace Control
1070 Whitespace refers to the blank space between variables, operators,
1071 and other code tokens.
1075 =item B<-fws>, B<--freeze-whitespace>
1077 This flag causes your original whitespace to remain unchanged, and
1078 causes the rest of the whitespace commands in this section, the
1079 Code Indentation section, and
1080 the Comment Control section to be ignored.
1082 =item B<Tightness of curly braces, parentheses, and square brackets>
1084 Here the term "tightness" will mean the closeness with which
1085 pairs of enclosing tokens, such as parentheses, contain the quantities
1086 within. A numerical value of 0, 1, or 2 defines the tightness, with
1087 0 being least tight and 2 being most tight. Spaces within containers
1088 are always symmetric, so if there is a space after a C<(> then there
1089 will be a space before the corresponding C<)>.
1091 The B<-pt=n> or B<--paren-tightness=n> parameter controls the space within
1092 parens. The example below shows the effect of the three possible
1093 values, 0, 1, and 2:
1095 if ( ( my $len_tab = length( $tabstr ) ) > 0 ) { # -pt=0
1096 if ( ( my $len_tab = length($tabstr) ) > 0 ) { # -pt=1 (default)
1097 if ((my $len_tab = length($tabstr)) > 0) { # -pt=2
1099 When n is 0, there is always a space to the right of a '(' and to the left
1100 of a ')'. For n=2 there is never a space. For n=1, the default, there
1101 is a space unless the quantity within the parens is a single token, such
1102 as an identifier or quoted string.
1104 Likewise, the parameter B<-sbt=n> or B<--square-bracket-tightness=n>
1105 controls the space within square brackets, as illustrated below.
1107 $width = $col[ $j + $k ] - $col[ $j ]; # -sbt=0
1108 $width = $col[ $j + $k ] - $col[$j]; # -sbt=1 (default)
1109 $width = $col[$j + $k] - $col[$j]; # -sbt=2
1111 Curly braces which do not contain code blocks are controlled by
1112 the parameter B<-bt=n> or B<--brace-tightness=n>.
1114 $obj->{ $parsed_sql->{ 'table' }[0] }; # -bt=0
1115 $obj->{ $parsed_sql->{'table'}[0] }; # -bt=1 (default)
1116 $obj->{$parsed_sql->{'table'}[0]}; # -bt=2
1118 And finally, curly braces which contain blocks of code are controlled by the
1119 parameter B<-bbt=n> or B<--block-brace-tightness=n> as illustrated in the
1122 %bf = map { $_ => -M $_ } grep { /\.deb$/ } dirents '.'; # -bbt=0 (default)
1123 %bf = map { $_ => -M $_ } grep {/\.deb$/} dirents '.'; # -bbt=1
1124 %bf = map {$_ => -M $_} grep {/\.deb$/} dirents '.'; # -bbt=2
1126 To simplify input in the case that all of the tightness flags have the same
1127 value <n>, the parameter <-act=n> or B<--all-containers-tightness=n> is an
1128 abbreviation for the combination <-pt=n -sbt=n -bt=n -bbt=n>.
1131 =item B<-tso>, B<--tight-secret-operators>
1133 The flag B<-tso> causes certain perl token sequences (secret operators)
1134 which might be considered to be a single operator to be formatted "tightly"
1135 (without spaces). The operators currently modified by this flag are:
1137 0+ +0 ()x!! ~~<> ,=> =( )=
1139 For example the sequence B<0 +>, which converts a string to a number,
1140 would be formatted without a space: B<0+> when the B<-tso> flag is set. This
1141 flag is off by default.
1143 =item B<-sts>, B<--space-terminal-semicolon>
1145 Some programmers prefer a space before all terminal semicolons. The
1146 default is for no such space, and is indicated with B<-nsts> or
1147 B<--nospace-terminal-semicolon>.
1150 $i = 1; # -nsts (default)
1152 =item B<-sfs>, B<--space-for-semicolon>
1154 Semicolons within B<for> loops may sometimes be hard to see,
1155 particularly when commas are also present. This option places spaces on
1156 both sides of these special semicolons, and is the default. Use
1157 B<-nsfs> or B<--nospace-for-semicolon> to deactivate it.
1159 for ( @a = @$ap, $u = shift @a ; @a ; $u = $v ) { # -sfs (default)
1160 for ( @a = @$ap, $u = shift @a; @a; $u = $v ) { # -nsfs
1162 =item B<-asc>, B<--add-semicolons>
1164 Setting B<-asc> allows perltidy to add any missing optional semicolon at the end
1165 of a line which is followed by a closing curly brace on the next line. This
1166 is the default, and may be deactivated with B<-nasc> or B<--noadd-semicolons>.
1168 =item B<-dsm>, B<--delete-semicolons>
1170 Setting B<-dsm> allows perltidy to delete extra semicolons which are
1171 simply empty statements. This is the default, and may be deactivated
1172 with B<-ndsm> or B<--nodelete-semicolons>. (Such semicolons are not
1173 deleted, however, if they would promote a side comment to a block
1176 =item B<-aws>, B<--add-whitespace>
1178 Setting this option allows perltidy to add certain whitespace to improve
1179 code readability. This is the default. If you do not want any
1180 whitespace added, but are willing to have some whitespace deleted, use
1181 B<-naws>. (Use B<-fws> to leave whitespace completely unchanged).
1183 =item B<-dws>, B<--delete-old-whitespace>
1185 Setting this option allows perltidy to remove some old whitespace
1186 between characters, if necessary. This is the default. If you
1187 do not want any old whitespace removed, use B<-ndws> or
1188 B<--nodelete-old-whitespace>.
1190 =item B<Detailed whitespace controls around tokens>
1192 For those who want more detailed control over the whitespace around
1193 tokens, there are four parameters which can directly modify the default
1194 whitespace rules built into perltidy for any token. They are:
1196 B<-wls=s> or B<--want-left-space=s>,
1198 B<-nwls=s> or B<--nowant-left-space=s>,
1200 B<-wrs=s> or B<--want-right-space=s>,
1202 B<-nwrs=s> or B<--nowant-right-space=s>.
1204 These parameters are each followed by a quoted string, B<s>, containing a
1205 list of token types. No more than one of each of these parameters
1206 should be specified, because repeating a command-line parameter
1207 always overwrites the previous one before perltidy ever sees it.
1209 To illustrate how these are used, suppose it is desired that there be no
1210 space on either side of the token types B<= + - / *>. The following two
1211 parameters would specify this desire:
1213 -nwls="= + - / *" -nwrs="= + - / *"
1215 (Note that the token types are in quotes, and that they are separated by
1216 spaces). With these modified whitespace rules, the following line of math:
1218 $root = -$b + sqrt( $b * $b - 4. * $a * $c ) / ( 2. * $a );
1222 $root=-$b+sqrt( $b*$b-4.*$a*$c )/( 2.*$a );
1224 These parameters should be considered to be hints to perltidy rather
1225 than fixed rules, because perltidy must try to resolve conflicts that
1226 arise between them and all of the other rules that it uses. One
1227 conflict that can arise is if, between two tokens, the left token wants
1228 a space and the right one doesn't. In this case, the token not wanting
1229 a space takes priority.
1231 It is necessary to have a list of all token types in order to create
1232 this type of input. Such a list can be obtained by the command
1233 B<--dump-token-types>. Also try the B<-D> flag on a short snippet of code
1234 and look at the .DEBUG file to see the tokenization.
1236 B<WARNING> Be sure to put these tokens in quotes to avoid having them
1237 misinterpreted by your command shell.
1239 =item B<Note1: Perltidy does always follow whitespace controls>
1241 The various parameters controlling whitespace within a program are requests which perltidy follows as well as possible, but there are a number of situations where changing whitespace could change program behavior and is not done. Some of these are obvious; for example, we should not remove the space between the two plus symbols in '$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:
1248 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
1255 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.
1257 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 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.
1259 In perltidy this is implemented in the tokenizer by marking token following a
1260 B<print> keyword as a special type B<Z>. When formatting is being done,
1261 whitespace following this token type is generally left unchanged as a precaution
1262 against changing program behavior. This is excessively conservative but simple
1263 and easy to implement. Keywords which are treated similarly to B<print> include
1264 B<printf>, B<sort>, B<exec>, B<system>. Changes in spacing around parameters
1265 following these keywords may have to be made manually. For example, the space,
1266 or lack of space, after the parameter $foo in the following line will be
1267 unchanged in formatting.
1272 To find if a token is of type B<Z> you can use B<perltidy -DEBUG>. For the
1273 first line above the result is
1278 which shows that B<system> is type B<k> (keyword) and $foo is type B<Z>.
1280 =item B<Note2: Perltidy's whitespace rules are not perfect>
1282 Despite these precautions, it is still possible to introduce syntax errors with
1283 some asymmetric whitespace rules, particularly when call parameters are not
1284 placed in containing parens or braces. For example, the following two lines will
1285 be parsed by perl without a syntax error:
1287 # original programming, syntax ok
1288 my @newkeys = map $_-$nrecs+@data, @oldkeys;
1290 # perltidy default, syntax ok
1291 my @newkeys = map $_ - $nrecs + @data, @oldkeys;
1293 But the following will give a syntax error:
1295 # perltidy -nwrs='-'
1296 my @newkeys = map $_ -$nrecs + @data, @oldkeys;
1298 For another example, the following two lines will be parsed without syntax error:
1300 # original programming, syntax ok
1301 for my $severity ( reverse $SEVERITY_LOWEST+1 .. $SEVERITY_HIGHEST ) { ... }
1303 # perltidy default, syntax ok
1304 for my $severity ( reverse $SEVERITY_LOWEST + 1 .. $SEVERITY_HIGHEST ) { ... }
1306 But the following will give a syntax error:
1308 # perltidy -nwrs='+', syntax error:
1309 for my $severity ( reverse $SEVERITY_LOWEST +1 .. $SEVERITY_HIGHEST ) { ... }
1311 To avoid subtle parsing problems like this, it is best to avoid spacing a
1312 binary operator asymetrically with a space on the left but not on the right.
1314 =item B<Space between specific keywords and opening paren>
1316 When an opening paren follows a Perl keyword, no space is introduced after the
1317 keyword, unless it is (by default) one of these:
1319 my local our and or xor eq ne if else elsif until unless
1320 while for foreach return switch case given when
1322 These defaults can be modified with two commands:
1324 B<-sak=s> or B<--space-after-keyword=s> adds keywords.
1326 B<-nsak=s> or B<--nospace-after-keyword=s> removes keywords.
1328 where B<s> is a list of keywords (in quotes if necessary). For example,
1330 my ( $a, $b, $c ) = @_; # default
1331 my( $a, $b, $c ) = @_; # -nsak="my local our"
1333 The abbreviation B<-nsak='*'> is equivalent to including all of the
1334 keywords in the above list.
1336 When both B<-nsak=s> and B<-sak=s> commands are included, the B<-nsak=s>
1337 command is executed first. For example, to have space after only the
1338 keywords (my, local, our) you could use B<-nsak="*" -sak="my local our">.
1340 To put a space after all keywords, see the next item.
1342 =item B<Space between all keywords and opening parens>
1344 When an opening paren follows a function or keyword, no space is introduced
1345 after the keyword except for the keywords noted in the previous item. To
1346 always put a space between a function or keyword and its opening paren,
1349 B<-skp> or B<--space-keyword-paren>
1351 You may also want to use the flag B<-sfp> (next item) too.
1353 =item B<Space between all function names and opening parens>
1355 When an opening paren follows a function the default and recommended formatting
1356 is not to introduce a space. To cause a space to be introduced use:
1358 B<-sfp> or B<--space-function-paren>
1360 myfunc( $a, $b, $c ); # default
1361 myfunc ( $a, $b, $c ); # -sfp
1363 You will probably also want to use the flag B<-skp> (previous item) too.
1365 The reason this is not recommended is that spacing a function paren can make a
1366 program vulnerable to parsing problems by Perl. For example, the following
1367 two-line program will run as written but will have a syntax error if
1368 reformatted with -sfp:
1370 if ( -e filename() ) { print "I'm here\n"; }
1371 sub filename { return $0 }
1373 In this particular case the syntax error can be removed if the line order is
1374 reversed, so that Perl parses 'sub filename' first.
1376 =item B<-fpva> or B<--function-paren-vertical-alignment>
1378 A side-effect of using the B<-sfp> flag is that the parens may become vertically
1379 aligned. For example,
1382 myfun ( $aaa, $b, $cc );
1383 mylongfun ( $a, $b, $c );
1385 This is the default behavior. To prevent this alignment use B<-nfpva>:
1387 # perltidy -sfp -nfpva
1388 myfun ( $aaa, $b, $cc );
1389 mylongfun ( $a, $b, $c );
1391 =item B<-spp=n> or B<--space-prototype-paren=n>
1393 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.
1397 This integer B<n> may have the value 0, 1, or 2 as follows:
1399 -spp=0 means no space before the paren
1400 -spp=1 means follow the example of the source code [DEFAULT]
1401 -spp=2 means always put a space before the paren
1403 The default is B<-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
1404 applying the different options would be:
1406 sub usage(); # n=0 [no space]
1407 sub usage(); # n=1 [default; follows input]
1408 sub usage (); # n=2 [space]
1410 =item B<-kpit=n> or B<--keyword-paren-inner-tightness=n>
1412 The space inside of an opening paren, which itself follows a certain keyword,
1413 can be controlled by this parameter. The space on the inside of the
1414 corresponding closing paren will be treated in the same (balanced) manner.
1415 This parameter has precedence over any other paren spacing rules. The values
1416 of B<n> are as follows:
1418 -kpit=0 means always put a space (not tight)
1419 -kpit=1 means ignore this parameter [default]
1420 -kpit=2 means never put a space (tight)
1422 To illustrate, the following snippet is shown formatted in three ways:
1424 if ( seek( DATA, 0, 0 ) ) { ... } # perltidy (default)
1425 if (seek(DATA, 0, 0)) { ... } # perltidy -pt=2
1426 if ( seek(DATA, 0, 0) ) { ... } # perltidy -pt=2 -kpit=0
1428 In the second case the -pt=2 parameter makes all of the parens tight. In the
1429 third case the -kpit=0 flag causes the space within the 'if' parens to have a
1430 space, since 'if' is one of the keywords to which the -kpit flag applies by
1431 default. The remaining parens are still tight because of the -pt=2 parameter.
1433 The set of keywords to which this parameter applies are by default are:
1435 if elsif unless while until for foreach
1437 These can be changed with the parameter B<-kpitl=s> described in the next section.
1440 =item B<-kpitl=string> or B<--keyword-paren-inner-tightness=string>
1442 This command can be used to change the keywords to which the the B<-kpit=n>
1443 command applies. The parameter B<string> is a required list either keywords or
1444 functions, which should be placed in quotes if there are more than one. By
1445 itself, this parameter does not cause any change in spacing, so the B<-kpit=n>
1446 command is still required.
1448 For example, the commands C<-kpitl="if else while" -kpit=2> will cause the just
1449 the spaces inside parens following 'if', 'else', and 'while' keywords to
1450 follow the tightness value indicated by the B<-kpit=2> flag.
1452 =item B<-lop> or B<--logical-padding>
1454 In the following example some extra space has been inserted on the second
1455 line between the two open parens. This extra space is called "logical padding"
1456 and is intended to help align similar things vertically in some logical
1457 or ternary expressions.
1459 # perltidy [default formatting]
1464 && ( $a->{'title'} eq $b->{'title'} )
1465 && ( $a->{'href'} eq $b->{'href'} ) );
1467 Note that this is considered to be a different operation from "vertical
1468 alignment" because space at just one line is being adjusted, whereas in
1469 "vertical alignment" the spaces at all lines are being adjusted. So it sort of
1470 a local version of vertical alignment.
1472 Here is an example involving a ternary operator:
1474 # perltidy [default formatting]
1481 This behavior is controlled with the flag B<--logical-padding>, which is set
1482 'on' by default. If it is not desired it can be turned off using
1483 B<--nological-padding> or B<-nlop>. The above two examples become, with
1491 && ( $a->{'title'} eq $b->{'title'} )
1492 && ( $a->{'href'} eq $b->{'href'} ) );
1502 =item B<Trimming whitespace around C<qw> quotes>
1504 B<-tqw> or B<--trim-qw> provide the default behavior of trimming
1505 spaces around multi-line C<qw> quotes and indenting them appropriately.
1507 B<-ntqw> or B<--notrim-qw> cause leading and trailing whitespace around
1508 multi-line C<qw> quotes to be left unchanged. This option will not
1509 normally be necessary, but was added for testing purposes, because in
1510 some versions of perl, trimming C<qw> quotes changes the syntax tree.
1512 =item B<-sbq=n> or B<--space-backslash-quote=n>
1519 can confuse syntax highlighters unless a space is included between the backslash and the single or double quotation mark.
1521 this can be controlled with the value of B<n> as follows:
1523 -sbq=0 means no space between the backslash and quote
1524 -sbq=1 means follow the example of the source code
1525 -sbq=2 means always put a space between the backslash and quote
1527 The default is B<-sbq=1>, meaning that a space will be used if there is one in the source code.
1529 =item B<Trimming trailing whitespace from lines of POD>
1531 B<-trp> or B<--trim-pod> will remove trailing whitespace from lines of POD.
1532 The default is not to do this.
1536 =head2 Comment Controls
1538 Perltidy has a number of ways to control the appearance of both block comments
1539 and side comments. The term B<block comment> here refers to a full-line
1540 comment, whereas B<side comment> will refer to a comment which appears on a
1541 line to the right of some code.
1545 =item B<-ibc>, B<--indent-block-comments>
1547 Block comments normally look best when they are indented to the same
1548 level as the code which follows them. This is the default behavior, but
1549 you may use B<-nibc> to keep block comments left-justified. Here is an
1552 # this comment is indented (-ibc, default)
1553 if ($task) { yyy(); }
1555 The alternative is B<-nibc>:
1557 # this comment is not indented (-nibc)
1558 if ($task) { yyy(); }
1560 See also the next item, B<-isbc>, as well as B<-sbc>, for other ways to
1561 have some indented and some outdented block comments.
1563 =item B<-isbc>, B<--indent-spaced-block-comments>
1565 If there is no leading space on the line, then the comment will not be
1566 indented, and otherwise it may be.
1568 If both B<-ibc> and B<-isbc> are set, then B<-isbc> takes priority.
1570 =item B<-olc>, B<--outdent-long-comments>
1572 When B<-olc> is set, lines which are full-line (block) comments longer
1573 than the value B<maximum-line-length> will have their indentation
1574 removed. This is the default; use B<-nolc> to prevent outdenting.
1576 =item B<-msc=n>, B<--minimum-space-to-comment=n>
1578 Side comments look best when lined up several spaces to the right of
1579 code. Perltidy will try to keep comments at least n spaces to the
1580 right. The default is n=4 spaces.
1582 =item B<-fpsc=n>, B<--fixed-position-side-comment=n>
1584 This parameter tells perltidy to line up side comments in column number B<n>
1585 whenever possible. The default, n=0, will not do this.
1587 =item B<-iscl>, B<--ignore-side-comment-lengths>
1589 This parameter causes perltidy to ignore the length of side comments when
1590 setting line breaks. The default, B<-niscl>, is to include the length of
1591 side comments when breaking lines to stay within the length prescribed
1592 by the B<-l=n> maximum line length parameter. For example, the following
1593 long single line would remain intact with -l=80 and -iscl:
1595 perltidy -l=80 -iscl
1596 $vmsfile =~ s/;[\d\-]*$//; # Clip off version number; we can use a newer version as well
1598 whereas without the -iscl flag the line will be broken:
1601 $vmsfile =~ s/;[\d\-]*$//
1602 ; # Clip off version number; we can use a newer version as well
1605 =item B<-hsc>, B<--hanging-side-comments>
1607 By default, perltidy tries to identify and align "hanging side
1608 comments", which are something like this:
1610 my $IGNORE = 0; # This is a side comment
1611 # This is a hanging side comment
1614 A comment is considered to be a hanging side comment if (1) it immediately
1615 follows a line with a side comment, or another hanging side comment, and
1616 (2) there is some leading whitespace on the line.
1617 To deactivate this feature, use B<-nhsc> or B<--nohanging-side-comments>.
1618 If block comments are preceded by a blank line, or have no leading
1619 whitespace, they will not be mistaken as hanging side comments.
1621 =item B<Closing Side Comments>
1623 A closing side comment is a special comment which perltidy can
1624 automatically create and place after the closing brace of a code block.
1625 They can be useful for code maintenance and debugging. The command
1626 B<-csc> (or B<--closing-side-comments>) adds or updates closing side
1627 comments. For example, here is a small code snippet
1630 if ( !defined( $_[0] ) ) {
1631 print("Hello, World\n");
1634 print( $_[0], "\n" );
1638 And here is the result of processing with C<perltidy -csc>:
1641 if ( !defined( $_[0] ) ) {
1642 print("Hello, World\n");
1645 print( $_[0], "\n" );
1647 } ## end sub message
1649 A closing side comment was added for C<sub message> in this case, but not
1650 for the C<if> and C<else> blocks, because they were below the 6 line
1651 cutoff limit for adding closing side comments. This limit may be
1652 changed with the B<-csci> command, described below.
1654 The command B<-dcsc> (or B<--delete-closing-side-comments>) reverses this
1655 process and removes these comments.
1657 Several commands are available to modify the behavior of these two basic
1658 commands, B<-csc> and B<-dcsc>:
1662 =item B<-csci=n>, or B<--closing-side-comment-interval=n>
1664 where C<n> is the minimum number of lines that a block must have in
1665 order for a closing side comment to be added. The default value is
1666 C<n=6>. To illustrate:
1668 # perltidy -csci=2 -csc
1670 if ( !defined( $_[0] ) ) {
1671 print("Hello, World\n");
1672 } ## end if ( !defined( $_[0] ))
1674 print( $_[0], "\n" );
1675 } ## end else [ if ( !defined( $_[0] ))
1676 } ## end sub message
1678 Now the C<if> and C<else> blocks are commented. However, now this has
1679 become very cluttered.
1681 =item B<-cscp=string>, or B<--closing-side-comment-prefix=string>
1683 where string is the prefix used before the name of the block type. The
1684 default prefix, shown above, is C<## end>. This string will be added to
1685 closing side comments, and it will also be used to recognize them in
1686 order to update, delete, and format them. Any comment identified as a
1687 closing side comment will be placed just a single space to the right of
1690 =item B<-cscl=string>, or B<--closing-side-comment-list>
1692 where C<string> is a list of block types to be tagged with closing side
1693 comments. By default, all code block types preceded by a keyword or
1694 label (such as C<if>, C<sub>, and so on) will be tagged. The B<-cscl>
1695 command changes the default list to be any selected block types; see
1696 L<Specifying Block Types>.
1697 For example, the following command
1698 requests that only C<sub>'s, labels, C<BEGIN>, and C<END> blocks be
1699 affected by any B<-csc> or B<-dcsc> operation:
1701 -cscl="sub : BEGIN END"
1703 =item B<-csct=n>, or B<--closing-side-comment-maximum-text=n>
1705 The text appended to certain block types, such as an C<if> block, is
1706 whatever lies between the keyword introducing the block, such as C<if>,
1707 and the opening brace. Since this might be too much text for a side
1708 comment, there needs to be a limit, and that is the purpose of this
1709 parameter. The default value is C<n=20>, meaning that no additional
1710 tokens will be appended to this text after its length reaches 20
1711 characters. Omitted text is indicated with C<...>. (Tokens, including
1712 sub names, are never truncated, however, so actual lengths may exceed
1713 this). To illustrate, in the above example, the appended text of the
1714 first block is C< ( !defined( $_[0] )...>. The existing limit of
1715 C<n=20> caused this text to be truncated, as indicated by the C<...>. See
1716 the next flag for additional control of the abbreviated text.
1718 =item B<-cscb>, or B<--closing-side-comments-balanced>
1720 As discussed in the previous item, when the
1721 closing-side-comment-maximum-text limit is exceeded the comment text must
1722 be truncated. Older versions of perltidy terminated with three dots, and this
1723 can still be achieved with -ncscb:
1725 perltidy -csc -ncscb
1726 } ## end foreach my $foo (sort { $b cmp $a ...
1728 However this causes a problem with editors which cannot recognize
1729 comments or are not configured to do so because they cannot "bounce" around in
1730 the text correctly. The B<-cscb> flag has been added to
1731 help them by appending appropriate balancing structure:
1734 } ## end foreach my $foo (sort { $b cmp $a ... })
1736 The default is B<-cscb>.
1738 =item B<-csce=n>, or B<--closing-side-comment-else-flag=n>
1740 The default, B<n=0>, places the text of the opening C<if> statement after any
1743 If B<n=2> is used, then each C<elsif> is also given the text of the opening
1744 C<if> statement. Also, an C<else> will include the text of a preceding
1745 C<elsif> statement. Note that this may result some long closing
1748 If B<n=1> is used, the results will be the same as B<n=2> whenever the
1749 resulting line length is less than the maximum allowed.
1751 =item B<-cscb>, or B<--closing-side-comments-balanced>
1753 When using closing-side-comments, and the closing-side-comment-maximum-text
1754 limit is exceeded, then the comment text must be abbreviated.
1755 It is terminated with three dots if the B<-cscb> flag is negated:
1757 perltidy -csc -ncscb
1758 } ## end foreach my $foo (sort { $b cmp $a ...
1760 This causes a problem with older editors which do not recognize comments
1761 because they cannot "bounce" around in the text correctly. The B<-cscb>
1762 flag tries to help them by appending appropriate terminal balancing structures:
1765 } ## end foreach my $foo (sort { $b cmp $a ... })
1767 The default is B<-cscb>.
1770 =item B<-cscw>, or B<--closing-side-comment-warnings>
1772 This parameter is intended to help make the initial transition to the use of
1773 closing side comments.
1775 things to happen if a closing side comment replaces an existing, different
1776 closing side comment: first, an error message will be issued, and second, the
1777 original side comment will be placed alone on a new specially marked comment
1778 line for later attention.
1780 The intent is to avoid clobbering existing hand-written side comments
1781 which happen to match the pattern of closing side comments. This flag
1782 should only be needed on the first run with B<-csc>.
1786 B<Important Notes on Closing Side Comments:>
1792 Closing side comments are only placed on lines terminated with a closing
1793 brace. Certain closing styles, such as the use of cuddled elses
1794 (B<-ce>), preclude the generation of some closing side comments.
1798 Please note that adding or deleting of closing side comments takes
1799 place only through the commands B<-csc> or B<-dcsc>. The other commands,
1800 if used, merely modify the behavior of these two commands.
1804 It is recommended that the B<-cscw> flag be used along with B<-csc> on
1805 the first use of perltidy on a given file. This will prevent loss of
1806 any existing side comment data which happens to have the csc prefix.
1810 Once you use B<-csc>, you should continue to use it so that any
1811 closing side comments remain correct as code changes. Otherwise, these
1812 comments will become incorrect as the code is updated.
1816 If you edit the closing side comments generated by perltidy, you must also
1817 change the prefix to be different from the closing side comment prefix.
1818 Otherwise, your edits will be lost when you rerun perltidy with B<-csc>. For
1819 example, you could simply change C<## end> to be C<## End>, since the test is
1820 case sensitive. You may also want to use the B<-ssc> flag to keep these
1821 modified closing side comments spaced the same as actual closing side comments.
1825 Temporarily generating closing side comments is a useful technique for
1826 exploring and/or debugging a perl script, especially one written by someone
1827 else. You can always remove them with B<-dcsc>.
1831 =item B<Static Block Comments>
1833 Static block comments are block comments with a special leading pattern,
1834 C<##> by default, which will be treated slightly differently from other
1835 block comments. They effectively behave as if they had glue along their
1836 left and top edges, because they stick to the left edge and previous line
1837 when there is no blank spaces in those places. This option is
1838 particularly useful for controlling how commented code is displayed.
1842 =item B<-sbc>, B<--static-block-comments>
1844 When B<-sbc> is used, a block comment with a special leading pattern, C<##> by
1845 default, will be treated specially.
1847 Comments so identified are treated as follows:
1853 If there is no leading space on the line, then the comment will not
1854 be indented, and otherwise it may be,
1858 no new blank line will be
1859 inserted before such a comment, and
1863 such a comment will never become
1864 a hanging side comment.
1868 For example, assuming C<@month_of_year> is
1871 @month_of_year = ( # -sbc (default)
1872 'Jan', 'Feb', 'Mar', 'Apr', 'May', 'Jun', 'Jul', 'Aug', 'Sep', 'Oct',
1876 Without this convention, the above code would become
1878 @month_of_year = ( # -nsbc
1879 'Jan', 'Feb', 'Mar', 'Apr', 'May', 'Jun', 'Jul', 'Aug', 'Sep', 'Oct',
1885 which is not as clear.
1886 The default is to use B<-sbc>. This may be deactivated with B<-nsbc>.
1888 =item B<-sbcp=string>, B<--static-block-comment-prefix=string>
1890 This parameter defines the prefix used to identify static block comments
1891 when the B<-sbc> parameter is set. The default prefix is C<##>,
1892 corresponding to C<-sbcp=##>. The prefix is actually part of a perl
1893 pattern used to match lines and it must either begin with C<#> or C<^#>.
1894 In the first case a prefix ^\s* will be added to match any leading
1895 whitespace, while in the second case the pattern will match only
1896 comments with no leading whitespace. For example, to
1897 identify all comments as static block comments, one would use C<-sbcp=#>.
1898 To identify all left-adjusted comments as static block comments, use C<-sbcp='^#'>.
1900 Please note that B<-sbcp> merely defines the pattern used to identify static
1901 block comments; it will not be used unless the switch B<-sbc> is set. Also,
1902 please be aware that since this string is used in a perl regular expression
1903 which identifies these comments, it must enable a valid regular expression to
1906 A pattern which can be useful is:
1910 This pattern requires a static block comment to have at least one character
1911 which is neither a # nor a space. It allows a line containing only '#'
1912 characters to be rejected as a static block comment. Such lines are often used
1913 at the start and end of header information in subroutines and should not be
1914 separated from the intervening comments, which typically begin with just a
1917 =item B<-osbc>, B<--outdent-static-block-comments>
1919 The command B<-osbc> will cause static block comments to be outdented by 2
1920 spaces (or whatever B<-ci=n> has been set to), if possible.
1924 =item B<Static Side Comments>
1926 Static side comments are side comments with a special leading pattern.
1927 This option can be useful for controlling how commented code is displayed
1928 when it is a side comment.
1932 =item B<-ssc>, B<--static-side-comments>
1934 When B<-ssc> is used, a side comment with a static leading pattern, which is
1935 C<##> by default, will be spaced only a single space from previous
1936 character, and it will not be vertically aligned with other side comments.
1938 The default is B<-nssc>.
1940 =item B<-sscp=string>, B<--static-side-comment-prefix=string>
1942 This parameter defines the prefix used to identify static side comments
1943 when the B<-ssc> parameter is set. The default prefix is C<##>,
1944 corresponding to C<-sscp=##>.
1946 Please note that B<-sscp> merely defines the pattern used to identify
1947 static side comments; it will not be used unless the switch B<-ssc> is
1948 set. Also, note that this string is used in a perl regular expression
1949 which identifies these comments, so it must enable a valid regular
1950 expression to be formed.
1956 =head2 Skipping Selected Sections of Code
1958 Selected lines of code may be passed verbatim to the output without any
1959 formatting by marking the starting and ending lines with special comments.
1960 There are two options for doing this. The first option is called
1961 B<--format-skipping> or B<-fs>, and the second option is called
1962 B<--code-skipping> or B<-cs>.
1964 In both cases the lines of code will be output without any changes.
1965 The difference is that in B<--format-skipping>
1966 perltidy will still parse the marked lines of code and check for errors,
1967 whereas in B<--code-skipping> perltidy will simply pass the lines to the output without any checking.
1969 Both of these features are enabled by default and are invoked with special
1970 comment markers. B<--format-skipping> uses starting and ending markers '#<<<'
1971 and '#>>>', like this:
1973 #<<< format skipping: do not let perltidy change my nice formatting
1981 B<--code-skipping> uses starting and ending markers '#<<V' and '#>>V', like
1984 #<<V code skipping: perltidy will pass this verbatim without error checking
1987 [ [ <?word> | _ | <?digit> ] <?ident_digit>
1994 Additional text may appear on the special comment lines provided that it
1995 is separated from the marker by at least one space, as in the above examples.
1997 It is recommended to use B<--code-skipping> only if you need to hide a block of
1998 an extended syntax which would produce errors if parsed by perltidy, and use
1999 B<--format-skipping> otherwise. This is because the B<--format-skipping>
2000 option provides the benefits of error checking, and there are essentially no
2001 limitations on which lines to which it can be applied. The B<--code-skipping>
2002 option, on the other hand, does not do error checking and its use is more
2003 restrictive because the code which remains, after skipping the marked lines,
2004 must be syntactically correct code with balanced containers.
2006 These features should be used sparingly to avoid littering code with markers,
2007 but they can be helpful for working around occasional problems.
2009 Note that it may be possible to avoid the use of B<--format-skipping> for the
2010 specific case of a comma-separated list of values, as in the above example, by
2011 simply inserting a blank or comment somewhere between the opening and closing
2012 parens. See the section L<Controlling List Formatting>.
2014 The following sections describe the available controls for these options. They
2015 should not normally be needed.
2019 =item B<-fs>, B<--format-skipping>
2021 As explained above, this flag, which is enabled by default, causes any code
2022 between special beginning and ending comment markers to be passed to the output
2023 without formatting. The code between the comments is still checked for errors
2024 however. The default beginning marker is #<<< and the default ending marker is
2027 Format skipping begins when a format skipping beginning comment is seen and
2028 continues until a format-skipping ending comment is found.
2030 This feature can be disabled with B<-nfs>. This should not normally be necessary.
2032 =item B<-fsb=string>, B<--format-skipping-begin=string>
2034 This and the next parameter allow the special beginning and ending comments to
2035 be changed. However, it is recommended that they only be changed if there is a
2036 conflict between the default values and some other use. If they are used, it
2037 is recommended that they only be entered in a B<.perltidyrc> file, rather than
2038 on a command line. This is because properly escaping these parameters on a
2039 command line can be difficult.
2041 If changed comment markers do not appear to be working, use the B<-log> flag and
2042 examine the F<.LOG> file to see if and where they are being detected.
2044 The B<-fsb=string> parameter may be used to change the beginning marker for
2045 format skipping. The default is equivalent to -fsb='#<<<'. The string that
2046 you enter must begin with a # and should be in quotes as necessary to get past
2047 the command shell of your system. It is actually the leading text of a pattern
2048 that is constructed by appending a '\s', so you must also include backslashes
2049 for characters to be taken literally rather than as patterns.
2051 Some examples show how example strings become patterns:
2053 -fsb='#\{\{\{' becomes /^#\{\{\{\s/ which matches #{{{ but not #{{{{
2054 -fsb='#\*\*' becomes /^#\*\*\s/ which matches #** but not #***
2055 -fsb='#\*{2,}' becomes /^#\*{2,}\s/ which matches #** and #*****
2057 =item B<-fse=string>, B<--format-skipping-end=string>
2059 The B<-fse=string> is the corresponding parameter used to change the
2060 ending marker for format skipping. The default is equivalent to
2063 The beginning and ending strings may be the same, but it is preferable
2064 to make them different for clarity.
2066 =item B<-cs>, B<--code-skipping>
2068 As explained above, this flag, which is enabled by default, causes any code
2069 between special beginning and ending comment markers to be directly passed to
2070 the output without any error checking or formatting. Essentially, perltidy
2071 treats it as if it were a block of arbitrary text. The default beginning
2072 marker is #<<V and the default ending marker is #>>V.
2074 This feature can be disabled with B<-ncs>. This should not normally be
2077 =item B<-csb=string>, B<--code-skipping-begin=string>
2079 This may be used to change the beginning comment for a B<--code-skipping> section, and its use is similar to the B<-fsb=string>.
2080 The default is equivalent to -csb='#<<V'.
2082 =item B<-cse=string>, B<--code-skipping-end=string>
2084 This may be used to change the ending comment for a B<--code-skipping> section, and its use is similar to the B<-fse=string>.
2085 The default is equivalent to -cse='#>>V'.
2089 =head2 Line Break Control
2091 The parameters in this section control breaks after
2092 non-blank lines of code. Blank lines are controlled
2093 separately by parameters in the section L<Blank Line
2098 =item B<-fnl>, B<--freeze-newlines>
2100 If you do not want any changes to the line breaks within
2101 lines of code in your script, set
2102 B<-fnl>, and they will remain fixed, and the rest of the commands in
2103 this section and sections
2104 L<Controlling List Formatting>,
2105 L<Retaining or Ignoring Existing Line Breaks>.
2106 You may want to use B<-noll> with this.
2108 Note: If you also want to keep your blank lines exactly
2109 as they are, you can use the B<-fbl> flag which is described
2110 in the section L<Blank Line Control>.
2112 =item B<-ce>, B<--cuddled-else>
2114 Enable the "cuddled else" style, in which C<else> and C<elsif> are
2115 follow immediately after the curly brace closing the previous block.
2116 The default is not to use cuddled elses, and is indicated with the flag
2117 B<-nce> or B<--nocuddled-else>. Here is a comparison of the
2135 In this example the keyword B<else> is placed on the same line which begins with
2136 the preceding closing block brace and is followed by its own opening block brace
2137 on the same line. Other keywords and function names which are formatted with
2138 this "cuddled" style are B<elsif>, B<continue>, B<catch>, B<finally>.
2140 Other block types can be formatted by specifying their names on a
2141 separate parameter B<-cbl>, described in a later section.
2143 Cuddling between a pair of code blocks requires that the closing brace of the
2144 first block start a new line. If this block is entirely on one line in the
2145 input file, it is necessary to decide if it should be broken to allow cuddling.
2146 This decision is controlled by the flag B<-cbo=n> discussed below. The default
2147 and recommended value of B<-cbo=1> bases this decision on the first block in
2148 the chain. If it spans multiple lines then cuddling is made and continues
2149 along the chain, regardless of the sizes of subsequent blocks. Otherwise, short
2150 lines remain intact.
2152 So for example, the B<-ce> flag would not have any effect if the above snippet
2155 if ($task) { yyy() }
2158 If the first block spans multiple lines, then cuddling can be done and will
2159 continue for the subsequent blocks in the chain, as illustrated in the previous
2162 If there are blank lines between cuddled blocks they will be eliminated. If
2163 there are comments after the closing brace where cuddling would occur then
2164 cuddling will be prevented. If this occurs, cuddling will restart later in the
2167 =item B<-cb>, B<--cuddled-blocks>
2169 This flag is equivalent to B<-ce>.
2172 =item B<-cbl>, B<--cuddled-block-list>
2174 The built-in default cuddled block types are B<else, elsif, continue, catch, finally>.
2176 Additional block types to which the B<-cuddled-blocks> style applies can be defined by
2177 this parameter. This parameter is a character string, giving a list of
2178 block types separated by commas or spaces. For example, to cuddle code blocks
2179 of type sort, map and grep, in addition to the default types, the string could
2182 -cbl="sort map grep"
2188 Note however that these particular block types are typically short so there might not be much
2189 opportunity for the cuddled format style.
2191 Using commas avoids the need to protect spaces with quotes.
2193 As a diagnostic check, the flag B<--dump-cuddled-block-list> or B<-dcbl> can be
2194 used to view the hash of values that are generated by this flag.
2196 Finally, note that the B<-cbl> flag by itself merely specifies which blocks are formatted
2197 with the cuddled format. It has no effect unless this formatting style is activated with
2200 =item B<-cblx>, B<--cuddled-block-list-exclusive>
2202 When cuddled else formatting is selected with B<-ce>, setting this flag causes
2203 perltidy to ignore its built-in defaults and rely exclusively on the block types
2204 specified on the B<-cbl> flag described in the previous section. For example,
2205 to avoid using cuddled B<catch> and B<finally>, which among in the defaults, the
2206 following set of parameters could be used:
2208 perltidy -ce -cbl='else elsif continue' -cblx
2211 =item B<-cbo=n>, B<--cuddled-break-option=n>
2213 Cuddled formatting is only possible between a pair of code blocks if the
2214 closing brace of the first block starts a new line. If a block is encountered
2215 which is entirely on a single line, and cuddled formatting is selected, it is
2216 necessary to make a decision as to whether or not to "break" the block, meaning
2217 to cause it to span multiple lines. This parameter controls that decision. The
2220 cbo=0 Never force a short block to break.
2221 cbo=1 If the first of a pair of blocks is broken in the input file,
2222 then break the second [DEFAULT].
2223 cbo=2 Break open all blocks for maximal cuddled formatting.
2225 The default and recommended value is B<cbo=1>. With this value, if the starting
2226 block of a chain spans multiple lines, then a cascade of breaks will occur for
2227 remaining blocks causing the entire chain to be cuddled.
2229 The option B<cbo=0> can produce erratic cuddling if there are numerous one-line
2232 The option B<cbo=2> produces maximal cuddling but will not allow any short blocks.
2235 =item B<-bl>, B<--opening-brace-on-new-line>
2237 Use the flag B<-bl> to place the opening brace on a new line:
2239 if ( $input_file eq '-' ) # -bl
2241 important_function();
2244 This flag applies to all structural blocks, including named sub's (unless
2245 the B<-sbl> flag is set -- see next item).
2247 The default style, B<-nbl>, places an opening brace on the same line as
2248 the keyword introducing it. For example,
2250 if ( $input_file eq '-' ) { # -nbl (default)
2252 =item B<-sbl>, B<--opening-sub-brace-on-new-line>
2254 The flag B<-sbl> can be used to override the value of B<-bl> for
2255 the opening braces of named sub's. For example,
2259 produces this result:
2263 if (!defined($_[0])) {
2264 print("Hello, World\n");
2271 This flag is negated with B<-nsbl>. If B<-sbl> is not specified,
2272 the value of B<-bl> is used.
2274 =item B<-asbl>, B<--opening-anonymous-sub-brace-on-new-line>
2276 The flag B<-asbl> is like the B<-sbl> flag except that it applies
2277 to anonymous sub's instead of named subs. For example
2281 produces this result:
2285 if ( !defined( $_[0] ) ) {
2286 print("Hello, World\n");
2289 print( $_[0], "\n" );
2293 This flag is negated with B<-nasbl>, and the default is B<-nasbl>.
2295 =item B<-bli>, B<--brace-left-and-indent>
2297 The flag B<-bli> is the same as B<-bl> but in addition it causes one
2298 unit of continuation indentation ( see B<-ci> ) to be placed before
2299 an opening and closing block braces.
2303 if ( $input_file eq '-' ) # -bli
2305 important_function();
2308 By default, this extra indentation occurs for blocks of type:
2309 B<if>, B<elsif>, B<else>, B<unless>, B<for>, B<foreach>, B<sub>,
2310 B<while>, B<until>, and also with a preceding label. The next item
2311 shows how to change this.
2313 =item B<-blil=s>, B<--brace-left-and-indent-list=s>
2315 Use this parameter to change the types of block braces for which the
2316 B<-bli> flag applies; see L<Specifying Block Types>. For example,
2317 B<-blil='if elsif else'> would apply it to only C<if/elsif/else> blocks.
2319 =item B<-bar>, B<--opening-brace-always-on-right>
2321 The default style, B<-nbl> places the opening code block brace on a new
2322 line if it does not fit on the same line as the opening keyword, like
2325 if ( $bigwasteofspace1 && $bigwasteofspace2
2326 || $bigwasteofspace3 && $bigwasteofspace4 )
2328 big_waste_of_time();
2331 To force the opening brace to always be on the right, use the B<-bar>
2332 flag. In this case, the above example becomes
2334 if ( $bigwasteofspace1 && $bigwasteofspace2
2335 || $bigwasteofspace3 && $bigwasteofspace4 ) {
2336 big_waste_of_time();
2339 A conflict occurs if both B<-bl> and B<-bar> are specified.
2341 =item B<-otr>, B<--opening-token-right> and related flags
2343 The B<-otr> flag is a hint that perltidy should not place a break between a
2344 comma and an opening token. For example:
2346 # default formatting
2347 push @{ $self->{$module}{$key} },
2349 accno => $ref->{accno},
2350 description => $ref->{description}
2354 push @{ $self->{$module}{$key} }, {
2355 accno => $ref->{accno},
2356 description => $ref->{description}
2359 The flag B<-otr> is actually an abbreviation for three other flags
2360 which can be used to control parens, hash braces, and square brackets
2361 separately if desired:
2363 -opr or --opening-paren-right
2364 -ohbr or --opening-hash-brace-right
2365 -osbr or --opening-square-bracket-right
2367 =item B<-bbhb=n>, B<--break-before-hash-brace=n> and related flags
2369 When a list of items spans multiple lines, the default formatting is to place
2370 the opening brace (or other container token) at the end of the starting line,
2380 This flag can change the default behavior to cause a line break to be placed
2381 before the opening brace according to the value given to the integer B<n>:
2383 -bbhb=0 never break [default]
2384 -bbhb=1 stable: break if the input script had a break
2385 -bbhb=2 break if list is 'complex' (see note below)
2386 -bbhb=3 always break
2399 There are several points to note about this flag:
2405 This parameter only applies if the opening brace is preceded by an '='
2410 This parameter only applies if the contents of the container looks like a list.
2411 The contents need to contain some commas or '=>'s at the next interior level to
2412 be considered a list.
2416 For the B<n=2> option, a list is considered 'complex' if it is part of a nested list
2417 structure which spans multiple lines in the input file.
2421 If multiple opening tokens have been 'welded' together with the B<-wn> parameter, then
2422 this parameter has no effect.
2426 The indentation of the braces will normally be one level of continuation
2427 indentation by default. This can be changed with the parameter
2428 B<-bbhbi=n> in the next section.
2432 Similar flags for controlling parens and square brackets are given in the subsequent section.
2436 =item B<-bbhbi=n>, B<--break-before-hash-brace-and-indent=n>
2438 This flag is a companion to B<-bbhb=n> for controlling the indentation of an opening hash brace
2439 which is placed on a new line by that parameter. The indentation is as follows:
2441 -bbhbi=0 one continuation level [default]
2442 -bbhbi=1 outdent by one continuation level
2443 -bbhbi=2 indent one full indentation level
2447 # perltidy -bbhb=3 -bbhbi=1
2456 # perltidy -bbhb=3 -bbhbi=2
2465 Note that this parameter has no effect unless B<-bbhb=n> is also set.
2467 =item B<-bbsb=n>, B<--break-before-square-bracket=n>
2469 This flag is similar to the flag described above, except it applies to lists contained within square brackets.
2471 -bbsb=0 never break [default]
2472 -bbsb=1 stable: break if the input script had a break
2473 -bbsb=2 break if list is 'complex' (part of nested list structure)
2474 -bbsb=3 always break
2476 =item B<-bbsbi=n>, B<--break-before-square-bracket-and-indent=n>
2478 This flag is a companion to B<-bbsb=n> for controlling the indentation of an opening square bracket
2479 which is placed on a new line by that parameter. The indentation is as follows:
2481 -bbsbi=0 one continuation level [default]
2482 -bbsbi=1 outdent by one continuation level
2483 -bbsbi=2 indent one full indentation level
2485 =item B<-bbp=n>, B<--break-before-paren=n>
2487 This flag is similar to B<-bbhb=n>, described above, except it applies to lists contained within parens.
2489 -bbp=0 never break [default]
2490 -bbp=1 stable: break if the input script had a break
2491 -bpb=2 break if list is 'complex' (part of nested list structure)
2494 =item B<-bbpi=n>, B<--break-before-paren-and-indent=n>
2496 This flag is a companion to B<-bbp=n> for controlling the indentation of an opening paren
2497 which is placed on a new line by that parameter. The indentation is as follows:
2499 -bbpi=0 one continuation level [default]
2500 -bbpi=1 outdent by one continuation level
2501 -bbpi=2 indent one full indentation level
2503 =item B<-wn>, B<--weld-nested-containers>
2505 The B<-wn> flag causes closely nested pairs of opening and closing container
2506 symbols (curly braces, brackets, or parens) to be "welded" together, meaning
2507 that they are treated as if combined into a single unit, with the indentation
2508 of the innermost code reduced to be as if there were just a single container
2513 # default formatting
2523 } } until $x++ > $z;
2525 When this flag is set perltidy makes a preliminary pass through the file and
2526 identifies all nested pairs of containers. To qualify as a nested pair, the
2527 closing container symbols must be immediately adjacent and the opening symbols
2528 must either (1) be adjacent as in the above example, or (2) have an anonymous
2529 sub declaration following an outer opening container symbol which is not a
2530 code block brace, or (3) have an outer opening paren separated from the inner
2531 opening symbol by any single non-container symbol or something that looks like
2532 a function evaluation, as illustrated in the next examples.
2534 Any container symbol may serve as both the inner container of one pair and as
2535 the outer container of an adjacent pair. Consequently, any number of adjacent
2536 opening or closing symbols may join together in weld. For example, here are
2537 three levels of wrapped function calls:
2539 # default formatting
2540 my (@date_time) = Localtime(
2543 $year, $month, $day, $hour, $minute, $second,
2544 '0', $offset, '0', '0'
2550 my (@date_time) = Localtime( Date_to_Time( Add_Delta_DHMS(
2551 $year, $month, $day, $hour, $minute, $second,
2552 '0', $offset, '0', '0'
2555 Notice how the indentation of the inner lines are reduced by two levels in this
2556 case. This example also shows the typical result of this formatting, namely it
2557 is a sandwich consisting of an initial opening layer, a central section of any
2558 complexity forming the "meat" of the sandwich, and a final closing layer. This
2559 predictable structure helps keep the compacted structure readable.
2561 The inner sandwich layer is required to be at least one line thick. If this
2562 cannot be achieved, welding does not occur. This constraint can cause
2563 formatting to take a couple of iterations to stabilize when it is first applied
2564 to a script. The B<-conv> flag can be used to insure that the final format is
2565 achieved in a single run.
2567 Here is an example illustrating a welded container within a welded containers:
2569 # default formatting
2574 $sx * int( $xr->numify() ) & $sy * int( $yr->numify() )
2584 $sx * int( $xr->numify() ) & $sy * int( $yr->numify() )
2589 The welded closing tokens are by default on a separate line but this can be
2590 modified with the B<-vtc=n> flag (described in the next section). For example,
2591 the same example adding B<-vtc=2> is
2593 # perltidy -wn -vtc=2
2596 $sx * int( $xr->numify() ) & $sy * int( $yr->numify() ) ) ),
2599 This format option is quite general but there are some limitations.
2601 One limitation is that any line length limit still applies and can cause long
2602 welded sections to be broken into multiple lines.
2604 Another limitation is that an opening symbol which delimits quoted text cannot
2605 be included in a welded pair. This is because quote delimiters are treated
2606 specially in perltidy.
2608 Finally, the stacking of containers defined by this flag have priority over
2609 any other container stacking flags. This is because any welding is done first.
2611 =item B<-wnxl=s>, B<--weld-nested-exclusion-list>
2613 The B<-wnxl=s> flag provides some control over the types of containers which
2614 can be welded. The B<-wn> flag by default is "greedy" in welding adjacent
2615 containers. If it welds more types of containers than desired, this flag
2616 provides a capability to reduce the amount of welding by specifying a list
2617 of things which should B<not> be welded.
2619 The logic in perltidy to apply this is straightforward. As each container
2620 token is being considered for joining a weld, any exclusion rules are consulted
2621 and used to reject the weld if necessary.
2623 This list is a string with space-separated items. Each item consists of up to
2624 three pieces of information: (1) an optional position, (2) an optional
2625 preceding type, and (3) a container type.
2627 The only required piece of information is a container type, which is one of
2628 '(', '[', '{' or 'q'. The first three of these are container tokens and the
2629 last represents a quoted list. For example the string
2633 means do B<NOT> include square-bracets, braces, or quotes in any welds. The only unspecified
2634 container is '(', so this string means that only welds involving parens will be made.
2636 To illustrate, following welded snippet consists of a chain of three welded
2637 containers with types '(' '[' and 'q':
2647 Even though the qw term uses parens as the quote delimiter, it has a special
2648 type 'q' here. If it appears in a weld it always appears at the end of the
2651 Any of the container types '[', '{', and '(' may be prefixed with a position
2652 indicator which is either '^', to indicate the first token of a welded
2653 sequence, or '.', to indicate an interior token of a welded sequence. (Since
2654 a quoted string 'q' always ends a chain it does need a position indicator).
2656 For example, if we do not want a sequence of welded containers to start with a
2657 square bracket we could use
2661 In the above snippet, there is a square bracket but it does not start the chain,
2662 so the formatting would be unchanged if it were formatted with this restriction.
2664 A third optional item of information which can be given is an alphanumeric
2665 letter which is used to limit the selection further depending on the type of
2666 token immediately before the container. If given, it goes just before the
2667 container symbol. The possible letters are currently 'k', 'K', 'f', 'F',
2668 'w', and 'W', with these meanings:
2670 'k' matches if the previous nonblank token is a perl builtin keyword (such as 'if', 'while'),
2671 'K' matches if 'k' does not, meaning that the previous token is not a keyword.
2672 'f' matches if the previous token is a function other than a keyword.
2673 'F' matches if 'f' does not.
2674 'w' matches if either 'k' or 'f' match.
2675 'W' matches if 'w' does not.
2677 For example, compare
2680 if ( defined( $_Cgi_Query{
2681 $Config{'methods'}{'authentication'}{'remote'}{'cgi'}{'username'}
2686 # perltidy -wn -wnxl='^K( {'
2688 $_Cgi_Query{ $Config{'methods'}{'authentication'}{'remote'}{'cgi'}
2692 The first case does maximum welding. In the second case the leading paren is
2693 retained by the rule (it would have been rejected if preceded by a non-keyword)
2694 but the curly brace is rejected by the rule.
2696 Here are some additional example strings and their meanings:
2698 '^(' - the weld must not start with a paren
2699 '.(' - the second and later tokens may not be parens
2700 '.w(' - the second and later tokens may not keyword or function call parens
2701 '(' - no parens in a weld
2702 '^K(' - exclude a leading paren preceded by a non-keyword
2703 '.k(' - exclude a secondary paren preceded by a keyword
2704 '[ {' - exclude all brackets and braces
2705 '[ ( ^K{' - exclude everything except nested structures like do {{ ... }}
2707 =item B<Vertical tightness> of non-block curly braces, parentheses, and square brackets.
2709 These parameters control what shall be called vertical tightness. Here are the
2716 Opening tokens (except for block braces) are controlled by B<-vt=n>, or
2717 B<--vertical-tightness=n>, where
2719 -vt=0 always break a line after opening token (default).
2720 -vt=1 do not break unless this would produce more than one
2721 step in indentation in a line.
2722 -vt=2 never break a line after opening token
2726 You must also use the B<-lp> flag when you use the B<-vt> flag; the
2727 reason is explained below.
2731 Closing tokens (except for block braces) are controlled by B<-vtc=n>, or
2732 B<--vertical-tightness-closing=n>, where
2734 -vtc=0 always break a line before a closing token (default),
2735 -vtc=1 do not break before a closing token which is followed
2736 by a semicolon or another closing token, and is not in
2738 -vtc=2 never break before a closing token.
2739 -vtc=3 Like -vtc=1 except always break before a closing token
2740 if the corresponding opening token follows an = or =>.
2742 The rules for B<-vtc=1> and B<-vtc=3> are designed to maintain a reasonable
2743 balance between tightness and readability in complex lists.
2747 Different controls may be applied to different token types,
2748 and it is also possible to control block braces; see below.
2752 Finally, please note that these vertical tightness flags are merely
2753 hints to the formatter, and it cannot always follow them. Things which
2754 make it difficult or impossible include comments, blank lines, blocks of
2755 code within a list, and possibly the lack of the B<-lp> parameter.
2756 Also, these flags may be ignored for very small lists (2 or 3 lines in
2761 Here are some examples:
2763 # perltidy -lp -vt=0 -vtc=0
2771 # perltidy -lp -vt=1 -vtc=0
2772 %romanNumerals = ( one => 'I',
2778 # perltidy -lp -vt=1 -vtc=1
2779 %romanNumerals = ( one => 'I',
2799 In the last example for B<-vtc=3>, the opening paren is preceded by an equals
2800 so the closing paren is placed on a new line.
2802 The difference between B<-vt=1> and B<-vt=2> is shown here:
2804 # perltidy -lp -vt=1
2806 mysprintf( "(void)find_threadsv(%s);",
2807 cstring( $threadsv_names[ $op->targ ] )
2811 # perltidy -lp -vt=2
2812 $init->add( mysprintf( "(void)find_threadsv(%s);",
2813 cstring( $threadsv_names[ $op->targ ] )
2817 With B<-vt=1>, the line ending in C<add(> does not combine with the next
2818 line because the next line is not balanced. This can help with
2819 readability, but B<-vt=2> can be used to ignore this rule.
2821 The tightest, and least readable, code is produced with both C<-vt=2> and
2824 # perltidy -lp -vt=2 -vtc=2
2825 $init->add( mysprintf( "(void)find_threadsv(%s);",
2826 cstring( $threadsv_names[ $op->targ ] ) ) );
2828 Notice how the code in all of these examples collapses vertically as
2829 B<-vt> increases, but the indentation remains unchanged. This is
2830 because perltidy implements the B<-vt> parameter by first formatting as
2831 if B<-vt=0>, and then simply overwriting one output line on top of the
2832 next, if possible, to achieve the desired vertical tightness. The
2833 B<-lp> indentation style has been designed to allow this vertical
2834 collapse to occur, which is why it is required for the B<-vt> parameter.
2836 The B<-vt=n> and B<-vtc=n> parameters apply to each type of container
2837 token. If desired, vertical tightness controls can be applied
2838 independently to each of the closing container token types.
2840 The parameters for controlling parentheses are B<-pvt=n> or
2841 B<--paren-vertical-tightness=n>, and B<-pvtc=n> or
2842 B<--paren-vertical-tightness-closing=n>.
2844 Likewise, the parameters for square brackets are B<-sbvt=n> or
2845 B<--square-bracket-vertical-tightness=n>, and B<-sbvtc=n> or
2846 B<--square-bracket-vertical-tightness-closing=n>.
2848 Finally, the parameters for controlling non-code block braces are
2849 B<-bvt=n> or B<--brace-vertical-tightness=n>, and B<-bvtc=n> or
2850 B<--brace-vertical-tightness-closing=n>.
2852 In fact, the parameter B<-vt=n> is actually just an abbreviation for
2853 B<-pvt=n -bvt=n sbvt=n>, and likewise B<-vtc=n> is an abbreviation
2854 for B<-pvtc=n -bvtc=n -sbvtc=n>.
2856 =item B<-bbvt=n> or B<--block-brace-vertical-tightness=n>
2858 The B<-bbvt=n> flag is just like the B<-vt=n> flag but applies
2859 to opening code block braces.
2861 -bbvt=0 break after opening block brace (default).
2862 -bbvt=1 do not break unless this would produce more than one
2863 step in indentation in a line.
2864 -bbvt=2 do not break after opening block brace.
2866 It is necessary to also use either B<-bl> or B<-bli> for this to work,
2867 because, as with other vertical tightness controls, it is implemented by
2868 simply overwriting a line ending with an opening block brace with the
2869 subsequent line. For example:
2871 # perltidy -bli -bbvt=0
2872 if ( open( FILE, "< $File" ) )
2874 while ( $File = <FILE> )
2882 # perltidy -bli -bbvt=1
2883 if ( open( FILE, "< $File" ) )
2884 { while ( $File = <FILE> )
2891 By default this applies to blocks associated with keywords B<if>,
2892 B<elsif>, B<else>, B<unless>, B<for>, B<foreach>, B<sub>, B<while>,
2893 B<until>, and also with a preceding label. This can be changed with
2894 the parameter B<-bbvtl=string>, or
2895 B<--block-brace-vertical-tightness-list=string>, where B<string> is a
2896 space-separated list of block types. For more information on the
2897 possible values of this string, see L<Specifying Block Types>
2899 For example, if we want to just apply this style to C<if>,
2900 C<elsif>, and C<else> blocks, we could use
2901 C<perltidy -bli -bbvt=1 -bbvtl='if elsif else'>.
2903 There is no vertical tightness control for closing block braces; with
2904 one exception they will be placed on separate lines.
2905 The exception is that a cascade of closing block braces may
2906 be stacked on a single line. See B<-scbb>.
2908 =item B<-sot>, B<--stack-opening-tokens> and related flags
2910 The B<-sot> flag tells perltidy to "stack" opening tokens
2911 when possible to avoid lines with isolated opening tokens.
2916 $opt_c = Text::CSV_XS->new(
2925 $opt_c = Text::CSV_XS->new( {
2932 For detailed control of individual closing tokens the following
2933 controls can be used:
2935 -sop or --stack-opening-paren
2936 -sohb or --stack-opening-hash-brace
2937 -sosb or --stack-opening-square-bracket
2938 -sobb or --stack-opening-block-brace
2940 The flag B<-sot> is an abbreviation for B<-sop -sohb -sosb>.
2942 The flag B<-sobb> is an abbreviation for B<-bbvt=2 -bbvtl='*'>. This
2943 will case a cascade of opening block braces to appear on a single line,
2944 although this an uncommon occurrence except in test scripts.
2946 =item B<-sct>, B<--stack-closing-tokens> and related flags
2948 The B<-sct> flag tells perltidy to "stack" closing tokens
2949 when possible to avoid lines with isolated closing tokens.
2954 $opt_c = Text::CSV_XS->new(
2963 $opt_c = Text::CSV_XS->new(
2970 The B<-sct> flag is somewhat similar to the B<-vtc> flags, and in some
2971 cases it can give a similar result. The difference is that the B<-vtc>
2972 flags try to avoid lines with leading opening tokens by "hiding" them at
2973 the end of a previous line, whereas the B<-sct> flag merely tries to
2974 reduce the number of lines with isolated closing tokens by stacking them
2975 but does not try to hide them. For example:
2978 $opt_c = Text::CSV_XS->new(
2982 always_quote => 1, } );
2984 For detailed control of the stacking of individual closing tokens the
2985 following controls can be used:
2987 -scp or --stack-closing-paren
2988 -schb or --stack-closing-hash-brace
2989 -scsb or --stack-closing-square-bracket
2990 -scbb or --stack-closing-block-brace
2992 The flag B<-sct> is an abbreviation for stacking the non-block closing
2993 tokens, B<-scp -schb -scsb>.
2995 Stacking of closing block braces, B<-scbb>, causes a cascade of isolated
2996 closing block braces to be combined into a single line as in the following
3004 push( @lines, "$w1 $w2 $w3 $w4\n" );
3007 To simplify input even further for the case in which both opening and closing
3008 non-block containers are stacked, the flag B<-sac> or B<--stack-all-containers>
3009 is an abbreviation for B<-sot -sct>.
3011 Please note that if both opening and closing tokens are to be stacked, then the
3012 newer flag B<-weld-nested-containers> may be preferable because it insures that
3013 stacking is always done symmetrically. It also removes an extra level of
3014 unnecessary indentation within welded containers. It is able to do this
3015 because it works on formatting globally rather than locally, as the B<-sot> and
3018 =item B<-dnl>, B<--delete-old-newlines>
3020 By default, perltidy first deletes all old line break locations, and then it
3021 looks for good break points to match the desired line length. Use B<-ndnl>
3022 or B<--nodelete-old-newlines> to force perltidy to retain all old line break
3025 =item B<-anl>, B<--add-newlines>
3027 By default, perltidy will add line breaks when necessary to create
3028 continuations of long lines and to improve the script appearance. Use
3029 B<-nanl> or B<--noadd-newlines> to prevent any new line breaks.
3031 This flag does not prevent perltidy from eliminating existing line
3032 breaks; see B<--freeze-newlines> to completely prevent changes to line
3035 =item B<Controlling whether perltidy breaks before or after operators>
3037 Four command line parameters provide some control over whether
3038 a line break should be before or after specific token types.
3039 Two parameters give detailed control:
3041 B<-wba=s> or B<--want-break-after=s>, and
3043 B<-wbb=s> or B<--want-break-before=s>.
3045 These parameters are each followed by a quoted string, B<s>, containing
3046 a list of token types (separated only by spaces). No more than one of each
3047 of these parameters should be specified, because repeating a
3048 command-line parameter always overwrites the previous one before
3049 perltidy ever sees it.
3051 By default, perltidy breaks B<after> these token types:
3052 % + - * / x != == >= <= =~ !~ < > | &
3053 = **= += *= &= <<= &&= -= /= |= >>= ||= //= .= %= ^= x=
3055 And perltidy breaks B<before> these token types by default:
3058 To illustrate, to cause a break after a concatenation operator, C<'.'>,
3059 rather than before it, the command line would be
3063 As another example, the following command would cause a break before
3064 math operators C<'+'>, C<'-'>, C<'/'>, and C<'*'>:
3068 These commands should work well for most of the token types that perltidy uses
3069 (use B<--dump-token-types> for a list). Also try the B<-D> flag on a short
3070 snippet of code and look at the .DEBUG file to see the tokenization. However,
3071 for a few token types there may be conflicts with hardwired logic which cause
3072 unexpected results. One example is curly braces, which should be controlled
3073 with the parameter B<bl> provided for that purpose.
3075 B<WARNING> Be sure to put these tokens in quotes to avoid having them
3076 misinterpreted by your command shell.
3078 Two additional parameters are available which, though they provide no further
3079 capability, can simplify input are:
3081 B<-baao> or B<--break-after-all-operators>,
3083 B<-bbao> or B<--break-before-all-operators>.
3085 The -baao sets the default to be to break after all of the following operators:
3087 % + - * / x != == >= <= =~ !~ < > | &
3088 = **= += *= &= <<= &&= -= /= |= >>= ||= //= .= %= ^= x=
3089 . : ? && || and or err xor
3091 and the B<-bbao> flag sets the default to break before all of these operators.
3092 These can be used to define an initial break preference which can be fine-tuned
3093 with the B<-wba> and B<-wbb> flags. For example, to break before all operators
3094 except an B<=> one could use --bbao -wba='=' rather than listing every
3095 single perl operator except B<=> on a -wbb flag.
3099 =head2 Controlling List Formatting
3101 Perltidy attempts to format lists of comma-separated values in tables which
3102 look good. Its default algorithms usually work well, but sometimes they don't.
3103 In this case, there are several methods available to control list formatting.
3105 A very simple way to prevent perltidy from changing the line breaks
3106 within a comma-separated list of values is to insert a blank line,
3107 comment, or side-comment anywhere between the opening and closing
3108 parens (or braces or brackets). This causes perltidy to skip
3109 over its list formatting logic. (The reason is that any of
3110 these items put a constraint on line breaks, and perltidy
3111 needs complete control over line breaks within a container to
3112 adjust a list layout). For example, let us consider
3120 The default formatting, which allows a maximum line length of 80,
3121 will flatten this down to one line:
3123 # perltidy (default)
3124 my @list = ( 1, 1, 1, 1, 2, 1, 1, 3, 3, 1, 1, 4, 6, 4, 1, );
3126 This formatting loses important information. If we place a side comment on one
3127 of the lines, for example, we get the following result with with default formatting
3131 1, # a side comment, comment, or blank keeps list intact
3138 We could achieve the same result with a blank line or full comment
3139 anywhere between the opening and closing parens.
3141 For another possibility see
3142 the -fs flag in L<Skipping Selected Sections of Code>.
3146 =item B<-boc>, B<--break-at-old-comma-breakpoints>
3148 The B<-boc> flag is another way to prevent comma-separated lists from being
3149 reformatted. Using B<-boc> on the above example, plus additional flags to retain
3150 the original style, yields
3152 # perltidy -boc -lp -pt=2 -vt=1 -vtc=1
3159 A disadvantage of this flag is that all tables in the file
3160 must already be nicely formatted.
3163 =item B<-mft=n>, B<--maximum-fields-per-table=n>
3165 If the computed number of fields for any table exceeds B<n>, then it
3166 will be reduced to B<n>. The default value for B<n> is a large number,
3167 40. While this value should probably be left unchanged as a general
3168 rule, it might be used on a small section of code to force a list to
3169 have a particular number of fields per line, and then either the B<-boc>
3170 flag could be used to retain this formatting, or a single comment could
3171 be introduced somewhere to freeze the formatting in future applications
3184 =item B<-cab=n>, B<--comma-arrow-breakpoints=n>
3186 A comma which follows a comma arrow, '=>', is given special
3187 consideration. In a long list, it is common to break at all such
3188 commas. This parameter can be used to control how perltidy breaks at
3189 these commas. (However, it will have no effect if old comma breaks are
3190 being forced because B<-boc> is used). The possible values of B<n> are:
3192 n=0 break at all commas after =>
3193 n=1 stable: break at all commas after => if container is open,
3194 EXCEPT FOR one-line containers
3195 n=2 break at all commas after =>, BUT try to form the maximum
3196 one-line container lengths
3197 n=3 do not treat commas after => specially at all
3198 n=4 break everything: like n=0 but ALSO break a short container with
3199 a => not followed by a comma when -vt=0 is used
3200 n=5 stable: like n=1 but ALSO break at open one-line containers when
3201 -vt=0 is used (default)
3203 For example, given the following single line, perltidy by default will
3204 not add any line breaks because it would break the existing one-line
3207 bless { B => $B, Root => $Root } => $package;
3209 Using B<-cab=0> will force a break after each comma-arrow item:
3217 If perltidy is subsequently run with this container broken, then by
3218 default it will break after each '=>' because the container is now
3219 broken. To reform a one-line container, the parameter B<-cab=2> could
3222 The flag B<-cab=3> can be used to prevent these commas from being
3223 treated specially. In this case, an item such as "01" => 31 is
3224 treated as a single item in a table. The number of fields in this table
3225 will be determined by the same rules that are used for any other table.
3230 "01" => 31, "02" => 29, "03" => 31, "04" => 30,
3231 "05" => 31, "06" => 30, "07" => 31, "08" => 31,
3232 "09" => 30, "10" => 31, "11" => 30, "12" => 31
3237 =head2 Retaining or Ignoring Existing Line Breaks
3239 Several additional parameters are available for controlling the extent
3240 to which line breaks in the input script influence the output script.
3241 In most cases, the default parameter values are set so that, if a choice
3242 is possible, the output style follows the input style. For example, if
3243 a short logical container is broken in the input script, then the
3244 default behavior is for it to remain broken in the output script.
3246 Most of the parameters in this section would only be required for a
3247 one-time conversion of a script from short container lengths to longer
3248 container lengths. The opposite effect, of converting long container
3249 lengths to shorter lengths, can be obtained by temporarily using a short
3250 maximum line length.
3254 =item B<-bol>, B<--break-at-old-logical-breakpoints>
3256 By default, if a logical expression is broken at a C<&&>, C<||>, C<and>,
3257 or C<or>, then the container will remain broken. Also, breaks
3258 at internal keywords C<if> and C<unless> will normally be retained.
3259 To prevent this, and thus form longer lines, use B<-nbol>.
3261 Please note that this flag does not duplicate old logical breakpoints. They
3262 are merely used as a hint with this flag that a statement should remain
3263 broken. Without this flag, perltidy will normally try to combine relatively
3264 short expressions into a single line.
3266 For example, given this snippet:
3268 return unless $cmd = $cmd || ($dot
3269 && $Last_Shell) || &prompt('|');
3271 # perltidy -bol [default]
3279 return unless $cmd = $cmd || ( $dot && $Last_Shell ) || &prompt('|');
3281 =item B<-bom>, B<--break-at-old-method-breakpoints>
3283 By default, a method call arrow C<-E<gt>> is considered a candidate for
3284 a breakpoint, but method chains will fill to the line width before a break is
3285 considered. With B<-bom>, breaks before the arrow are preserved, so if you
3286 have preformatted a method chain:
3289 ->related_resultset('CDs')
3290 ->related_resultset('Tracks')
3292 'track.id' => {-ident => 'none_search.id'},
3295 It will B<keep> these breaks, rather than become this:
3297 my $q = $rs->related_resultset('CDs')->related_resultset('Tracks')->search({
3298 'track.id' => {-ident => 'none_search.id'},
3301 This flag will also look for and keep a 'cuddled' style of calls,
3302 in which lines begin with a closing paren followed by a call arrow,
3306 my $q = $rs->related_resultset(
3308 )->related_resultset(
3311 'track.id' => { -ident => 'none_search.id' },
3314 You may want to include the B<-weld-nested-containers> flag in this case to keep
3315 nested braces and parens together, as in the last line.
3317 =item B<-bos>, B<--break-at-old-semicolon-breakpoints>
3319 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
3320 the following two-line input snippet:
3322 $z = sqrt($x**2 + $y**2)
3325 The default formatting will be:
3327 $z = sqrt( $x**2 + $y**2 );
3329 The result using B<perltidy -bos> keeps the isolated semicolon:
3331 $z = sqrt( $x**2 + $y**2 )
3334 The default is not to do this, B<-nbos>.
3337 =item B<-bok>, B<--break-at-old-keyword-breakpoints>
3339 By default, perltidy will retain a breakpoint before keywords which may
3340 return lists, such as C<sort> and <map>. This allows chains of these
3341 operators to be displayed one per line. Use B<-nbok> to prevent
3342 retaining these breakpoints.
3344 =item B<-bot>, B<--break-at-old-ternary-breakpoints>
3346 By default, if a conditional (ternary) operator is broken at a C<:>,
3347 then it will remain broken. To prevent this, and thereby
3348 form longer lines, use B<-nbot>.
3350 =item B<-boa>, B<--break-at-old-attribute-breakpoints>
3352 By default, if an attribute list is broken at a C<:> in the source file, then
3353 it will remain broken. For example, given the following code, the line breaks
3354 at the ':'s will be retained:
3359 : Get('Name' => 'foo') : Set('Name');
3361 If the attributes are on a single line in the source code then they will remain
3362 on a single line if possible.
3364 To prevent this, and thereby always form longer lines, use B<-nboa>.
3366 =item B<Keeping old breakpoints at specific token types>
3368 Two command line parameters provide detailed control over whether
3369 perltidy should keep an old line break before or after a specific
3372 B<-kbb=s> or B<--keep-old-breakpoints-before=s>, and
3374 B<-kba=s> or B<--keep-old-breakpoints-after=s>
3376 These parameters are each followed by a quoted string, B<s>, containing
3377 a list of token types (separated only by spaces). No more than one of each
3378 of these parameters should be specified, because repeating a
3379 command-line parameter always overwrites the previous one before
3380 perltidy ever sees it.
3382 For example, -kbb='=>' means that if an input line begins with a '=>' then the
3383 output script should also have a line break before that token.
3385 For example, given the script:
3390 my ( $self, $x, $y ) = ( shift, @_ );
3394 # perltidy [default]
3395 method 'foo' => [ Int, Int ] => sub {
3396 my ( $self, $x, $y ) = ( shift, @_ );
3400 # perltidy -kbb='=>'
3404 my ( $self, $x, $y ) = ( shift, @_ );
3408 =item B<-iob>, B<--ignore-old-breakpoints>
3410 Use this flag to tell perltidy to ignore existing line breaks to the
3411 maximum extent possible. This will tend to produce the longest possible
3412 containers, regardless of type, which do not exceed the line length
3413 limit. But please note that this parameter has priority over all
3414 other parameters requesting that certain old breakpoints be kept.
3416 =item B<-kis>, B<--keep-interior-semicolons>
3418 Use the B<-kis> flag to prevent breaking at a semicolon if
3419 there was no break there in the input file. Normally
3420 perltidy places a newline after each semicolon which
3421 terminates a statement unless several statements are
3422 contained within a one-line brace block. To illustrate,
3423 consider the following input lines:
3425 dbmclose(%verb_delim); undef %verb_delim;
3426 dbmclose(%expanded); undef %expanded;
3428 The default is to break after each statement, giving
3430 dbmclose(%verb_delim);
3432 dbmclose(%expanded);
3435 With B<perltidy -kis> the multiple statements are retained:
3437 dbmclose(%verb_delim); undef %verb_delim;
3438 dbmclose(%expanded); undef %expanded;
3440 The statements are still subject to the specified value
3441 of B<maximum-line-length> and will be broken if this
3442 maximum is exceeded.
3446 =head2 Blank Line Control
3448 Blank lines can improve the readability of a script if they are carefully
3449 placed. Perltidy has several commands for controlling the insertion,
3450 retention, and removal of blank lines.
3454 =item B<-fbl>, B<--freeze-blank-lines>
3456 Set B<-fbl> if you want to the blank lines in your script to
3457 remain exactly as they are. The rest of the parameters in
3458 this section may then be ignored. (Note: setting the B<-fbl> flag
3459 is equivalent to setting B<-mbl=0> and B<-kbl=2>).
3461 =item B<-bbc>, B<--blanks-before-comments>
3463 A blank line will be introduced before a full-line comment. This is the
3464 default. Use B<-nbbc> or B<--noblanks-before-comments> to prevent
3465 such blank lines from being introduced.
3467 =item B<-blbs=n>, B<--blank-lines-before-subs=n>
3469 The parameter B<-blbs=n> requests that least B<n> blank lines precede a sub
3470 definition which does not follow a comment and which is more than one-line
3471 long. The default is <-blbs=1>. B<BEGIN> and B<END> blocks are included.
3473 The requested number of blanks statement will be inserted regardless of the
3474 value of B<--maximum-consecutive-blank-lines=n> (B<-mbl=n>) with the exception
3475 that if B<-mbl=0> then no blanks will be output.
3477 This parameter interacts with the value B<k> of the parameter B<--maximum-consecutive-blank-lines=k> (B<-mbl=k>) as follows:
3479 1. If B<-mbl=0> then no blanks will be output. This allows all blanks to be suppressed with a single parameter. Otherwise,
3481 2. If the number of old blank lines in the script is less than B<n> then
3482 additional blanks will be inserted to make the total B<n> regardless of the
3485 3. If the number of old blank lines in the script equals or exceeds B<n> then
3486 this parameter has no effect, however the total will not exceed
3487 value specified on the B<-mbl=k> flag.
3490 =item B<-blbp=n>, B<--blank-lines-before-packages=n>
3492 The parameter B<-blbp=n> requests that least B<n> blank lines precede a package
3493 which does not follow a comment. The default is B<-blbp=1>.
3495 This parameter interacts with the value B<k> of the parameter
3496 B<--maximum-consecutive-blank-lines=k> (B<-mbl=k>) in the same way as described
3497 for the previous item B<-blbs=n>.
3500 =item B<-bbs>, B<--blanks-before-subs>
3502 For compatibility with previous versions, B<-bbs> or B<--blanks-before-subs>
3503 is equivalent to F<-blbp=1> and F<-blbs=1>.
3505 Likewise, B<-nbbs> or B<--noblanks-before-subs>
3506 is equivalent to F<-blbp=0> and F<-blbs=0>.
3508 =item B<-bbb>, B<--blanks-before-blocks>
3510 A blank line will be introduced before blocks of coding delimited by
3511 B<for>, B<foreach>, B<while>, B<until>, and B<if>, B<unless>, in the following
3518 The block is not preceded by a comment.
3522 The block is not a one-line block.
3526 The number of consecutive non-blank lines at the current indentation depth is at least B<-lbl>
3531 This is the default. The intention of this option is to introduce
3532 some space within dense coding.
3533 This is negated with B<-nbbb> or B<--noblanks-before-blocks>.
3535 =item B<-lbl=n> B<--long-block-line-count=n>
3537 This controls how often perltidy is allowed to add blank lines before
3538 certain block types (see previous section). The default is 8. Entering
3539 a value of B<0> is equivalent to entering a very large number.
3541 =item B<-blao=i> or B<--blank-lines-after-opening-block=i>
3543 This control places a minimum of B<i> blank lines B<after> a line which B<ends>
3544 with an opening block brace of a specified type. By default, this only applies
3545 to the block of a named B<sub>, but this can be changed (see B<-blaol> below).
3546 The default is not to do this (B<i=0>).
3548 Please see the note below on using the B<-blao> and B<-blbc> options.
3550 =item B<-blbc=i> or B<--blank-lines-before-closing-block=i>
3552 This control places a minimum of B<i> blank lines B<before> a line which
3553 B<begins> with a closing block brace of a specified type. By default, this
3554 only applies to the block of a named B<sub>, but this can be changed (see
3555 B<-blbcl> below). The default is not to do this (B<i=0>).
3557 =item B<-blaol=s> or B<--blank-lines-after-opening-block-list=s>
3559 The parameter B<s> is a list of block type keywords to which the flag B<-blao>
3560 should apply. The section L<"Specifying Block Types"> explains how to list
3563 =item B<-blbcl=s> or B<--blank-lines-before-closing-block-list=s>
3565 This parameter is a list of block type keywords to which the flag B<-blbc>
3566 should apply. The section L<"Specifying Block Types"> explains how to list
3569 =item B<Note on using the> B<-blao> and B<-blbc> options.
3571 These blank line controls introduce a certain minimum number of blank lines in
3572 the text, but the final number of blank lines may be greater, depending on
3573 values of the other blank line controls and the number of old blank lines. A
3574 consequence is that introducing blank lines with these and other controls
3575 cannot be exactly undone, so some experimentation with these controls is
3576 recommended before using them.
3578 For example, suppose that for some reason we decide to introduce one blank
3579 space at the beginning and ending of all blocks. We could do
3582 perltidy -blao=2 -blbc=2 -blaol='*' -blbcl='*' filename
3584 Now suppose the script continues to be developed, but at some later date we
3585 decide we don't want these spaces after all. We might expect that running with
3586 the flags B<-blao=0> and B<-blbc=0> will undo them. However, by default
3587 perltidy retains single blank lines, so the blank lines remain.
3589 We can easily fix this by telling perltidy to ignore old blank lines by
3590 including the added parameter B<-kbl=0> and rerunning. Then the unwanted blank
3591 lines will be gone. However, this will cause all old blank lines to be
3592 ignored, perhaps even some that were added by hand to improve formatting. So
3593 please be cautious when using these parameters.
3595 =item B<-mbl=n> B<--maximum-consecutive-blank-lines=n>
3597 This parameter specifies the maximum number of consecutive blank lines which
3598 will be output within code sections of a script. The default is n=1. If the
3599 input file has more than n consecutive blank lines, the number will be reduced
3600 to n except as noted above for the B<-blbp> and B<-blbs> parameters. If B<n=0>
3601 then no blank lines will be output (unless all old blank lines are retained
3602 with the B<-kbl=2> flag of the next section).
3604 This flag obviously does not apply to pod sections,
3605 here-documents, and quotes.
3607 =item B<-kbl=n>, B<--keep-old-blank-lines=n>
3609 The B<-kbl=n> flag gives you control over how your existing blank lines are
3612 The possible values of B<n> are:
3614 n=0 ignore all old blank lines
3615 n=1 stable: keep old blanks, but limited by the value of the B<-mbl=n> flag
3616 n=2 keep all old blank lines, regardless of the value of the B<-mbl=n> flag
3618 The default is B<n=1>.
3620 =item B<-sob>, B<--swallow-optional-blank-lines>
3622 This is equivalent to B<kbl=0> and is included for compatibility with
3625 =item B<-nsob>, B<--noswallow-optional-blank-lines>
3627 This is equivalent to B<kbl=1> and is included for compatibility with
3632 B<Controls for blank lines around lines of consecutive keywords>
3634 The parameters in this section provide some control over the placement of blank
3635 lines within and around groups of statements beginning with selected keywords.
3636 These blank lines are called here B<keyword group blanks>, and all of the
3637 parameters begin with B<--keyword-group-blanks*>, or B<-kgb*> for short. The
3638 default settings do not employ these controls but they can be enabled with the
3639 following parameters:
3641 B<-kgbl=s> or B<--keyword-group-blanks-list=s>; B<s> is a quoted string of keywords
3643 B<-kgbs=s> or B<--keyword-group-blanks-size=s>; B<s> gives the number of keywords required to form a group.
3645 B<-kgbb=n> or B<--keyword-group-blanks-before=n>; B<n> = (0, 1, or 2) controls a leading blank
3647 B<-kgba=n> or B<--keyword-group-blanks-after=n>; B<n> = (0, 1, or 2) controls a trailing blank
3649 B<-kgbi> or B<--keyword-group-blanks-inside> is a switch for adding blanks between subgroups
3651 B<-kgbd> or B<--keyword-group-blanks-delete> is a switch for removing initial blank lines between keywords
3653 B<-kgbr=n> or B<--keyword-group-blanks-repeat-count=n> can limit the number of times this logic is applied
3655 In addition, the following abbreviations are available to for simplified usage:
3657 B<-kgb> or B<--keyword-group-blanks> is short for B<-kgbb=2 -kgba=2 kgbi>
3659 B<-nkgb> or B<--nokeyword-group-blanks>, is short for B<-kgbb=1 -kgba=1 nkgbi>
3661 Before describing the meaning of the parameters in detail let us look at an
3662 example which is formatted with default parameter settings.
3664 print "Entering test 2\n";
3666 use Encode qw(from_to encode decode
3667 encode_utf8 decode_utf8
3668 find_encoding is_utf8);
3669 use charnames qw(greek);
3670 my @encodings = grep( /iso-?8859/, Encode::encodings() );
3671 my @character_set = ( '0' .. '9', 'A' .. 'Z', 'a' .. 'z' );
3672 my @source = qw(ascii iso8859-1 cp1250);
3673 my @destiny = qw(cp1047 cp37 posix-bc);
3674 my @ebcdic_sets = qw(cp1047 cp37 posix-bc);
3675 my $str = join( '', map( chr($_), 0x20 .. 0x7E ) );
3676 return unless ($str);
3678 using B<perltidy -kgb> gives:
3680 print "Entering test 2\n";
3681 <----------this blank controlled by -kgbb
3683 use Encode qw(from_to encode decode
3684 encode_utf8 decode_utf8
3685 find_encoding is_utf8);
3686 use charnames qw(greek);
3687 <---------this blank controlled by -kgbi
3688 my @encodings = grep( /iso-?8859/, Encode::encodings() );
3689 my @character_set = ( '0' .. '9', 'A' .. 'Z', 'a' .. 'z' );
3690 my @source = qw(ascii iso8859-1 cp1250);
3691 my @destiny = qw(cp1047 cp37 posix-bc);
3692 my @ebcdic_sets = qw(cp1047 cp37 posix-bc);
3693 my $str = join( '', map( chr($_), 0x20 .. 0x7E ) );
3694 <----------this blank controlled by -kgba
3695 return unless ($str);
3697 Blank lines have been introduced around the B<my> and B<use> sequences. What
3698 happened is that the default keyword list includes B<my> and B<use> but not
3699 B<print> and B<return>. So a continuous sequence of nine B<my> and B<use>
3700 statements was located. This number exceeds the default threshold of five, so
3701 blanks were placed before and after the entire group. Then, since there was
3702 also a subsequence of six B<my> lines, a blank line was introduced to separate
3705 Finer control over blank placement can be achieved by using the individual
3706 parameters rather than the B<-kgb> flag. The individual controls are as follows.
3708 B<-kgbl=s> or B<--keyword-group-blanks-list=s>, where B<s> is a quoted string,
3709 defines the set of keywords which will be formed into groups. The string is a
3710 space separated list of keywords. The default set is B<s="use require local
3711 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 B<BC>. To include static block comments (which normally begin with '##'), include the symbol B<SBC>.
3713 B<-kgbs=s> or B<--keyword-group-blanks-size=s>, where B<s> is a string
3714 describing the number of consecutive keyword statements forming a group. If
3715 B<s> is an integer then it is the minimum number required for a group. A
3716 maximum value may also be given with the format B<s=min.max>, where B<min> is
3717 the minimum number and B<max> is the maximum number, and the min and max values
3718 are separated by one or more dots. No groups will be found if the maximum is
3719 less than the minimum. The maximum is unlimited if not given. The default is
3720 B<s=5>. Some examples:
3722 s min max number for group
3723 3 3 unlimited 3 or more
3729 B<-kgbb=n> or B<--keyword-group-blanks-before=n> specifies whether
3730 a blank should appear before the first line of the group, as follows:
3732 n=0 => (delete) an existing blank line will be removed
3733 n=1 => (stable) no change to the input file is made [DEFAULT]
3734 n=2 => (insert) a blank line is introduced if possible
3736 B<-kgba=n> or B<--keyword-group-blanks-after=n> likewise specifies
3737 whether a blank should appear after the last line of the group, using the same
3738 scheme (0=delete, 1=stable, 2=insert).
3740 B<-kgbi> or B<--keyword-group-blanks-inside> controls
3741 the insertion of blank lines between the first and last statement of the entire
3742 group. If there is a continuous run of a single statement type with more than
3743 the minimum threshold number (as specified with B<-kgbs=s>) then this
3744 switch causes a blank line be inserted between this
3745 subgroup and the others. In the example above this happened between the
3746 B<use> and B<my> statements.
3748 B<-kgbd> or B<--keyword-group-blanks-delete> controls the deletion of any
3749 blank lines that exist in the the group when it is first scanned. When
3750 statements are initially scanned, any existing blank lines are included in the
3751 collection. Any such orignial blank lines will be deleted before any other
3752 insertions are made when the parameter B<-kgbd> is set. The default is not to
3755 B<-kgbr=n> or B<--keyword-group-blanks-repeat-count=n> specifies B<n>, the
3756 maximum number of times this logic will be applied to any file. The special
3757 value B<n=0> is the same as n=infinity which means it will be applied to an
3758 entire script [Default]. A value B<n=1> could be used to make it apply just
3759 one time for example. This might be useful for adjusting just the B<use>
3760 statements in the top part of a module for example.
3762 B<-kgb> or B<--keyword-group-blanks> is an abbreviation equivalent to setting
3763 B<-kgbb=1 -kgba=1 -kgbi>. This turns on keyword group formatting with a set of
3766 B<-nkgb> or B<--nokeyword-group-blanks> is equivalent to B<-kgbb=0 -kgba
3767 nkgbi>. This flag turns off keyword group blank lines and is the default
3770 Here are a few notes about the functioning of this technique.
3776 These parameters are probably more useful as part of a major code reformatting
3777 operation rather than as a routine formatting operation.
3779 In particular, note that deleting old blank lines with B<-kgbd> is an
3780 irreversible operation so it should be applied with care. Existing blank lines
3781 may be serving an important role in controlling vertical alignment.
3785 Conflicts which arise among these B<kgb*> parameters and other blank line
3786 controls are generally resolved by producing the maximum number of blank lines
3787 implied by any parameter.
3789 For example, if the flags B<--freeze-blank-lines>, or
3790 B<--keep-old-blank-lines=2>, are set, then they have priority over any blank
3791 line deletion implied by the B<-kgb> flags of this section, so no blank lines
3794 For another example, if a keyword group ends at a B<sub> and the flag B<kgba=0> requests no blank line there, but we also have B<--blank-lines-before-subs=2>, then two blank lines will still be introduced before the sub.
3798 The introduction of blank lines does not occur if it would conflict with other
3799 input controls or code validity. For example, a blank line will not be placed
3800 within a here-doc or within a section of code marked with format skipping
3801 comments. And in general, a blank line will only be introduced at the end of a
3802 group if the next statement is a line of code.
3806 The count which is used to determine the group size is not the number of lines
3807 but rather the total number of keywords which are found. Individual statements
3808 with a certain leading keyword may continue on multiple lines, but if any of
3809 these lines is nested more than one level deep then that group will be ended.
3813 The search for groups of lines with similar leading keywords is based on the
3814 input source, not the final formatted source. Consequently, if the source code
3815 is badly formatted, it would be best to make a first formatting pass without
3822 A style refers to a convenient collection of existing parameters.
3826 =item B<-gnu>, B<--gnu-style>
3828 B<-gnu> gives an approximation to the GNU Coding Standards (which do
3829 not apply to perl) as they are sometimes implemented. At present, this
3830 style overrides the default style with the following parameters:
3832 -lp -bl -noll -pt=2 -bt=2 -sbt=2 -icp
3834 =item B<-pbp>, B<--perl-best-practices>
3836 B<-pbp> is an abbreviation for the parameters in the book B<Perl Best Practices>
3839 -l=78 -i=4 -ci=4 -st -se -vt=2 -cti=0 -pt=1 -bt=1 -sbt=1 -bbt=1 -nsfs -nolq
3840 -wbb="% + - * / x != == >= <= =~ !~ < > | & =
3841 **= += *= &= <<= &&= -= /= |= >>= ||= //= .= %= ^= x="
3843 Please note that this parameter set includes -st and -se flags, which make
3844 perltidy act as a filter on one file only. These can be overridden by placing
3845 B<-nst> and/or B<-nse> after the -pbp parameter.
3847 Also note that the value of continuation indentation, -ci=4, is equal to the
3848 value of the full indentation, -i=4. It is recommended that the either (1) the
3849 parameter B<-ci=2> be used instead, or the flag B<-xci> be set. This will help
3850 show structure, particularly when there are ternary statements. The following
3851 snippet illustrates these options.
3856 : $type eq 'item' ? "the $section entry"
3857 : "the section on $section"
3861 ? ( $section ? ' in ' : '' ) . "the $page$page_ext manpage"
3862 : ' elsewhere in this document'
3865 # perltidy -pbp -ci=2
3868 : $type eq 'item' ? "the $section entry"
3869 : "the section on $section"
3873 ? ( $section ? ' in ' : '' ) . "the $page$page_ext manpage"
3874 : ' elsewhere in this document'
3877 # perltidy -pbp -xci
3880 : $type eq 'item' ? "the $section entry"
3881 : "the section on $section"
3884 ? ( $section ? ' in ' : '' ) . "the $page$page_ext manpage"
3885 : ' elsewhere in this document'
3888 The B<-xci> flag was developed after the B<-pbp> parameters were published so you need
3889 to include it separately.
3891 =item B<One-line blocks>
3893 There are a few points to note regarding one-line blocks. A one-line
3894 block is something like this,
3897 where the contents within the curly braces is short enough to fit
3900 With few exceptions, perltidy retains existing one-line blocks, if it
3901 is possible within the line-length constraint, but it does not attempt
3902 to form new ones. In other words, perltidy will try to follow the
3903 one-line block style of the input file.
3905 If an existing one-line block is longer than the maximum line length,
3906 however, it will be broken into multiple lines. When this happens, perltidy
3907 checks for and adds any optional terminating semicolon (unless the B<-nasc>
3908 option is used) if the block is a code block.
3910 The main exception is that perltidy will attempt to form new one-line
3911 blocks following the keywords C<map>, C<eval>, and C<sort>, because
3912 these code blocks are often small and most clearly displayed in a single
3915 One-line block rules can conflict with the cuddled-else option. When
3916 the cuddled-else option is used, perltidy retains existing one-line
3917 blocks, even if they do not obey cuddled-else formatting.
3919 Occasionally, when one-line blocks get broken because they exceed the
3920 available line length, the formatting will violate the requested brace style.
3921 If this happens, reformatting the script a second time should correct
3924 Sometimes it might be desirable to convert a script to have one-line blocks
3925 whenever possible. Although there is currently no flag for this, a simple
3926 workaround is to execute perltidy twice, once with the flag B<-noadd-newlines>
3927 and then once again with normal parameters, like this:
3929 cat infile | perltidy -nanl | perltidy >outfile
3931 When executed on this snippet
3934 die "failed to execute: $!\n";
3937 print "Had enough.\n";
3938 die "failed to execute: $!\n";
3943 if ( $? == -1 ) { die "failed to execute: $!\n"; }
3945 print "Had enough.\n";
3946 die "failed to execute: $!\n";
3949 This shows that blocks with a single statement become one-line blocks.
3951 =item B<-olbs=n>, B<--one-line-block-semicolons=n>
3953 This flag controls the placement of semicolons at the end of one-line blocks.
3954 Semicolons are optional before a closing block brace, and frequently they are
3955 omitted at the end of a one-line block containing just a single statement.
3956 By default, perltidy follows the input file regarding these semicolons,
3957 but this behavior can be controlled by this flag. The values of n are:
3959 n=0 remove terminal semicolons in one-line blocks having a single statement
3960 n=1 stable; keep input file placement of terminal semicolons [DEFAULT ]
3961 n=2 add terminal semicolons in all one-line blocks
3963 Note that the B<n=2> option has no effect if adding semicolons is prohibited
3964 with the B<-nasc> flag. Also not that while B<n=2> adds missing semicolons to
3965 all one-line blocks, regardless of complexity, the B<n=0> option only removes
3966 ending semicolons which terminate one-line blocks containing just one
3967 semicolon. So these two options are not exact inverses.
3969 =item B<-olbn=n>, B<--one-line-block-nesting=n>
3971 Nested one-line blocks are lines with code blocks which themselves contain code
3972 blocks. For example, the following line is a nested one-line block.
3974 foreach (@list) { if ($_ eq $asked_for) { last } ++$found }
3976 The default behavior is to break such lines into multiple lines, but this
3977 behavior can be controlled with this flag. The values of n are:
3979 n=0 break nested one-line blocks into multiple lines [DEFAULT]
3980 n=1 stable: keep existing nested-one line blocks intact
3982 For the above example, the default formatting (B<-olbn=0>) is
3985 if ( $_ eq $asked_for ) { last }
3989 If the parameter B<-olbn=1> is given, then the line will be left intact if it
3990 is a single line in the source, or it will be broken into multiple lines if it
3991 is broken in multiple lines in the source.
3997 =head2 Controlling Vertical Alignment
3999 Vertical alignment refers to lining up certain symbols in a list of consecutive
4000 similar lines to improve readability. For example, the "fat commas" are
4001 aligned in the following statement:
4004 PeerAddr => join( ".", @port[ 0 .. 3 ] ),
4005 PeerPort => $port[4] * 256 + $port[5],
4009 Vertical alignment can be completely turned off using B<-novalign>, a flag
4010 mainly intended for debugging. However, vertical alignment can be forced to
4011 stop and restart by selectively introducing blank lines. For example, a blank
4012 has been inserted in the following code to keep somewhat similar things
4016 'format' => [ 'tidy', 'html', 'user' ],
4017 'output-line-ending' => [ 'dos', 'win', 'mac', 'unix' ],
4018 'character-encoding' => [ 'none', 'utf8' ],
4020 'block-brace-tightness' => [ 0, 2 ],
4021 'brace-tightness' => [ 0, 2 ],
4022 'paren-tightness' => [ 0, 2 ],
4023 'square-bracket-tightness' => [ 0, 2 ],
4026 Vertical alignment is implemented by locally increasing an existing blank space
4027 to produce alignment with an adjacent line. It cannot occur if there is no
4028 blank space to increase. So if a particular space is removed by one of the
4029 existing controls then vertical alignment cannot occur. Likewise, if a space is
4030 added with one of the controls, then vertical alignment might occur.
4034 # perltidy -nwls='=>'
4036 PeerAddr=> join( ".", @port[ 0 .. 3 ] ),
4037 PeerPort=> $port[4] * 256 + $port[5],
4041 =head2 Other Controls
4045 =item B<Deleting selected text>
4047 Perltidy can selectively delete comments and/or pod documentation. The
4048 command B<-dac> or B<--delete-all-comments> will delete all comments
4049 B<and> all pod documentation, leaving just code and any leading system
4052 The command B<-dp> or B<--delete-pod> will remove all pod documentation
4055 Two commands which remove comments (but not pod) are: B<-dbc> or
4056 B<--delete-block-comments> and B<-dsc> or B<--delete-side-comments>.
4057 (Hanging side comments will be deleted with side comments here.)
4059 The negatives of these commands also work, and are the defaults. When
4060 block comments are deleted, any leading 'hash-bang' will be retained.
4061 Also, if the B<-x> flag is used, any system commands before a leading
4062 hash-bang will be retained (even if they are in the form of comments).
4064 =item B<Writing selected text to a file>
4066 When perltidy writes a formatted text file, it has the ability to also
4067 send selected text to a file with a F<.TEE> extension. This text can
4068 include comments and pod documentation.
4070 The command B<-tac> or B<--tee-all-comments> will write all comments
4071 B<and> all pod documentation.
4073 The command B<-tp> or B<--tee-pod> will write all pod documentation (but
4076 The commands which write comments (but not pod) are: B<-tbc> or
4077 B<--tee-block-comments> and B<-tsc> or B<--tee-side-comments>.
4078 (Hanging side comments will be written with side comments here.)
4080 The negatives of these commands also work, and are the defaults.
4082 =item B<Using a F<.perltidyrc> command file>
4084 If you use perltidy frequently, you probably won't be happy until you
4085 create a F<.perltidyrc> file to avoid typing commonly-used parameters.
4086 Perltidy will first look in your current directory for a command file
4087 named F<.perltidyrc>. If it does not find one, it will continue looking
4088 for one in other standard locations.
4090 These other locations are system-dependent, and may be displayed with
4091 the command C<perltidy -dpro>. Under Unix systems, it will first look
4092 for an environment variable B<PERLTIDY>. Then it will look for a
4093 F<.perltidyrc> file in the home directory, and then for a system-wide
4094 file F</usr/local/etc/perltidyrc>, and then it will look for
4095 F</etc/perltidyrc>. Note that these last two system-wide files do not
4096 have a leading dot. Further system-dependent information will be found
4097 in the INSTALL file distributed with perltidy.
4099 Under Windows, perltidy will also search for a configuration file named perltidy.ini since Windows does not allow files with a leading period (.).
4100 Use C<perltidy -dpro> to see the possible locations for your system.
4101 An example might be F<C:\Documents and Settings\All Users\perltidy.ini>.
4103 Another option is the use of the PERLTIDY environment variable.
4104 The method for setting environment variables depends upon the version of
4105 Windows that you are using. Instructions for Windows 95 and later versions can
4108 http://www.netmanage.com/000/20021101_005_tcm21-6336.pdf
4110 Under Windows NT / 2000 / XP the PERLTIDY environment variable can be placed in
4111 either the user section or the system section. The later makes the
4112 configuration file common to all users on the machine. Be sure to enter the
4113 full path of the configuration file in the value of the environment variable.
4114 Ex. PERLTIDY=C:\Documents and Settings\perltidy.ini
4116 The configuration file is free format, and simply a list of parameters, just as
4117 they would be entered on a command line. Any number of lines may be used, with
4118 any number of parameters per line, although it may be easiest to read with one
4119 parameter per line. Comment text begins with a #, and there must
4120 also be a space before the # for side comments. It is a good idea to
4121 put complex parameters in either single or double quotes.
4123 Here is an example of a F<.perltidyrc> file:
4125 # This is a simple of a .perltidyrc configuration file
4126 # This implements a highly spaced style
4127 -se # errors to standard error output
4128 -w # show all warnings
4129 -bl # braces on new lines
4130 -pt=0 # parens not tight at all
4131 -bt=0 # braces not tight
4132 -sbt=0 # square brackets not tight
4134 The parameters in the F<.perltidyrc> file are installed first, so any
4135 parameters given on the command line will have priority over them.
4137 To avoid confusion, perltidy ignores any command in the .perltidyrc
4138 file which would cause some kind of dump and an exit. These are:
4140 -h -v -ddf -dln -dop -dsn -dtt -dwls -dwrs -ss
4142 There are several options may be helpful in debugging a F<.perltidyrc>
4149 A very helpful command is B<--dump-profile> or B<-dpro>. It writes a
4150 list of all configuration filenames tested to standard output, and
4151 if a file is found, it dumps the content to standard output before
4152 exiting. So, to find out where perltidy looks for its configuration
4153 files, and which one if any it selects, just enter
4159 It may be simplest to develop and test configuration files with
4160 alternative names, and invoke them with B<-pro=filename> on the command
4161 line. Then rename the desired file to F<.perltidyrc> when finished.
4165 The parameters in the F<.perltidyrc> file can be switched off with
4166 the B<-npro> option.
4170 The commands B<--dump-options>, B<--dump-defaults>, B<--dump-long-names>,
4171 and B<--dump-short-names>, all described below, may all be helpful.
4175 =item B<Creating a new abbreviation>
4177 A special notation is available for use in a F<.perltidyrc> file
4178 for creating an abbreviation for a group
4179 of options. This can be used to create a
4180 shorthand for one or more styles which are frequently, but not always,
4181 used. The notation is to group the options within curly braces which
4182 are preceded by the name of the alias (without leading dashes), like this:
4189 where B<newword> is the abbreviation, and B<opt1>, etc, are existing parameters
4190 I<or other abbreviations>. The main syntax requirement is that the new
4191 abbreviation along with its opening curly brace must begin on a new line.
4192 Space before and after the curly braces is optional.
4194 For a specific example, the following line
4196 oneliner { --maximum-line-length=0 --noadd-newlines --noadd-terminal-newline}
4198 or equivalently with abbreviations
4200 oneliner { -l=0 -nanl -natnl }
4202 could be placed in a F<.perltidyrc> file to temporarily override the maximum
4203 line length with a large value, to temporarily prevent new line breaks from
4204 being added, and to prevent an extra newline character from being added the
4205 file. All other settings in the F<.perltidyrc> file still apply. Thus it
4206 provides a way to format a long 'one liner' when perltidy is invoked with
4208 perltidy --oneliner ...
4210 (Either C<-oneliner> or C<--oneliner> may be used).
4212 =item Skipping leading non-perl commands with B<-x> or B<--look-for-hash-bang>
4214 If your script has leading lines of system commands or other text which
4215 are not valid perl code, and which are separated from the start of the
4216 perl code by a "hash-bang" line, ( a line of the form C<#!...perl> ),
4217 you must use the B<-x> flag to tell perltidy not to parse and format any
4218 lines before the "hash-bang" line. This option also invokes perl with a
4219 -x flag when checking the syntax. This option was originally added to
4220 allow perltidy to parse interactive VMS scripts, but it should be used
4221 for any script which is normally invoked with C<perl -x>.
4223 Please note: do not use this flag unless you are sure your script needs it.
4224 Parsing errors can occur if it does not have a hash-bang, or, for example, if
4225 the actual first hash-bang is in a here-doc. In that case a parsing error will
4226 occur because the tokenization will begin in the middle of the here-doc.
4228 =item B<Making a file unreadable>
4230 The goal of perltidy is to improve the readability of files, but there
4231 are two commands which have the opposite effect, B<--mangle> and
4232 B<--extrude>. They are actually
4233 merely aliases for combinations of other parameters. Both of these
4234 strip all possible whitespace, but leave comments and pod documents,
4235 so that they are essentially reversible. The
4236 difference between these is that B<--mangle> puts the fewest possible
4237 line breaks in a script while B<--extrude> puts the maximum possible.
4238 Note that these options do not provided any meaningful obfuscation, because
4239 perltidy can be used to reformat the files. They were originally
4240 developed to help test the tokenization logic of perltidy, but they
4242 One use for B<--mangle> is the following:
4244 perltidy --mangle myfile.pl -st | perltidy -o myfile.pl.new
4246 This will form the maximum possible number of one-line blocks (see next
4247 section), and can sometimes help clean up a badly formatted script.
4249 A similar technique can be used with B<--extrude> instead of B<--mangle>
4250 to make the minimum number of one-line blocks.
4252 Another use for B<--mangle> is to combine it with B<-dac> to reduce
4253 the file size of a perl script.
4257 The following flags are available for debugging:
4259 B<--dump-cuddled-block-list> or B<-dcbl> will dump to standard output the
4260 internal hash of cuddled block types created by a B<-cuddled-block-list> input
4263 B<--dump-defaults> or B<-ddf> will write the default option set to standard output and quit
4265 B<--dump-profile> or B<-dpro> will write the name of the current
4266 configuration file and its contents to standard output and quit.
4268 B<--dump-options> or B<-dop> will write current option set to standard
4271 B<--dump-long-names> or B<-dln> will write all command line long names (passed
4272 to Get_options) to standard output and quit.
4274 B<--dump-short-names> or B<-dsn> will write all command line short names
4275 to standard output and quit.
4277 B<--dump-token-types> or B<-dtt> will write a list of all token types
4278 to standard output and quit.
4280 B<--dump-want-left-space> or B<-dwls> will write the hash %want_left_space
4281 to standard output and quit. See the section on controlling whitespace
4284 B<--dump-want-right-space> or B<-dwrs> will write the hash %want_right_space
4285 to standard output and quit. See the section on controlling whitespace
4288 B<--no-memoize> or B<-nmem> will turn of memoizing.
4289 Memoization can reduce run time when running perltidy repeatedly in a
4290 single process. It is on by default but can be deactivated for
4291 testing with B<-nmem>.
4293 B<--no-timestamp> or B<-nts> will eliminate any time stamps in output files to prevent
4294 differences in dates from causing test installation scripts to fail. There are just
4295 a couple of places where timestamps normally occur. One is in the headers of
4296 html files, and another is when the B<-cscw> option is selected. The default is
4297 to allow timestamps (B<--timestamp> or B<-ts>).
4299 B<--file-size-order> or B<-fso> will cause files to be processed in order of
4300 increasing size, when multiple files are being processed. This is useful
4301 during program development, when large numbers of files with varying sizes are
4302 processed, because it can reduce virtual memory usage.
4304 B<--maximum-file-size-mb=n> or B<-maxfs=n> specifies the maximum file size in
4305 megabytes that perltidy will attempt to format. This parameter is provided to
4306 avoid causing system problems by accidentally attempting to format an extremely
4307 large data file. Most perl scripts are less than about 2 MB in size. The
4308 integer B<n> has a default value of 10, so perltidy will skip formatting files
4309 which have a size greater than 10 MB. The command to increase the limit to 20
4310 MB for example would be
4314 This only applies to files specified by filename on the command line.
4316 B<--maximum-level-errors=n> or B<-maxle=n> specifies the maximum number of
4317 indentation level errors are allowed before perltidy skips formatting and just
4318 outputs a file verbatim. The default is B<n=1>. This means that if the final
4319 indentation of a script differs from the starting indentation by more than 1
4320 levels, the file will be output verbatim. To avoid formatting if there are any
4321 indentation level errors use -maxle=0. To skip this check you can either set n
4322 equal to a large number, such as B<n=100>, or set B<n=-1>.
4324 For example, the following script has level error of 3 and will be output verbatim
4326 Input and default output:
4335 B<--maximum-unexpected-errors=n> or B<-maxue=n> specifies the maximum number of
4336 unexpected tokenization errors are allowed before formatting is skipped and a
4337 script is output verbatim. The intention is to avoid accidentally formatting
4338 a non-perl script, such as an html file for example. This check can be turned
4339 off by setting B<n=0>.
4341 A recommended value is B<n=3>. However, the default is B<n=0> (skip this check)
4342 to avoid causing problems with scripts which have extended syntaxes.
4344 B<-DEBUG> will write a file with extension F<.DEBUG> for each input file
4345 showing the tokenization of all lines of code.
4347 =item B<Working with MakeMaker, AutoLoader and SelfLoader>
4349 The first $VERSION line of a file which might be eval'd by MakeMaker
4350 is passed through unchanged except for indentation.
4351 Use B<--nopass-version-line>, or B<-npvl>, to deactivate this feature.
4353 If the AutoLoader module is used, perltidy will continue formatting
4354 code after seeing an __END__ line.
4355 Use B<--nolook-for-autoloader>, or B<-nlal>, to deactivate this feature.
4357 Likewise, if the SelfLoader module is used, perltidy will continue formatting
4358 code after seeing a __DATA__ line.
4359 Use B<--nolook-for-selfloader>, or B<-nlsl>, to deactivate this feature.
4361 =item B<Working around problems with older version of Perl>
4363 Perltidy contains a number of rules which help avoid known subtleties
4364 and problems with older versions of perl, and these rules always
4365 take priority over whatever formatting flags have been set. For example,
4366 perltidy will usually avoid starting a new line with a bareword, because
4367 this might cause problems if C<use strict> is active.
4369 There is no way to override these rules.
4377 =item The B<-html> master switch
4379 The flag B<-html> causes perltidy to write an html file with extension
4380 F<.html>. So, for example, the following command
4382 perltidy -html somefile.pl
4384 will produce a syntax-colored html file named F<somefile.pl.html>
4385 which may be viewed with a browser.
4387 B<Please Note>: In this case, perltidy does not do any formatting to the
4388 input file, and it does not write a formatted file with extension
4389 F<.tdy>. This means that two perltidy runs are required to create a
4390 fully reformatted, html copy of a script.
4392 =item The B<-pre> flag for code snippets
4394 When the B<-pre> flag is given, only the pre-formatted section, within
4395 the <PRE> and </PRE> tags, will be output. This simplifies inclusion
4396 of the output in other files. The default is to output a complete
4399 =item The B<-nnn> flag for line numbering
4401 When the B<-nnn> flag is given, the output lines will be numbered.
4403 =item The B<-toc>, or B<--html-table-of-contents> flag
4405 By default, a table of contents to packages and subroutines will be
4406 written at the start of html output. Use B<-ntoc> to prevent this.
4407 This might be useful, for example, for a pod document which contains a
4408 number of unrelated code snippets. This flag only influences the code
4409 table of contents; it has no effect on any table of contents produced by
4410 pod2html (see next item).
4412 =item The B<-pod>, or B<--pod2html> flag
4414 There are two options for formatting pod documentation. The default is
4415 to pass the pod through the Pod::Html module (which forms the basis of
4416 the pod2html utility). Any code sections are formatted by perltidy, and
4417 the results then merged. Note: perltidy creates a temporary file when
4418 Pod::Html is used; see L<"FILES">. Also, Pod::Html creates temporary
4419 files for its cache.
4421 NOTE: Perltidy counts the number of C<=cut> lines, and either moves the
4422 pod text to the top of the html file if there is one C<=cut>, or leaves
4423 the pod text in its original order (interleaved with code) otherwise.
4425 Most of the flags accepted by pod2html may be included in the perltidy
4426 command line, and they will be passed to pod2html. In some cases,
4427 the flags have a prefix C<pod> to emphasize that they are for the
4428 pod2html, and this prefix will be removed before they are passed to
4429 pod2html. The flags which have the additional C<pod> prefix are:
4431 --[no]podheader --[no]podindex --[no]podrecurse --[no]podquiet
4432 --[no]podverbose --podflush
4434 The flags which are unchanged from their use in pod2html are:
4436 --backlink=s --cachedir=s --htmlroot=s --libpods=s --title=s
4437 --podpath=s --podroot=s
4439 where 's' is an appropriate character string. Not all of these flags are
4440 available in older versions of Pod::Html. See your Pod::Html documentation for
4443 The alternative, indicated with B<-npod>, is not to use Pod::Html, but
4444 rather to format pod text in italics (or whatever the stylesheet
4445 indicates), without special html markup. This is useful, for example,
4446 if pod is being used as an alternative way to write comments.
4448 =item The B<-frm>, or B<--frames> flag
4450 By default, a single html output file is produced. This can be changed
4451 with the B<-frm> option, which creates a frame holding a table of
4452 contents in the left panel and the source code in the right side. This
4453 simplifies code browsing. Assume, for example, that the input file is
4454 F<MyModule.pm>. Then, for default file extension choices, these three
4455 files will be created:
4457 MyModule.pm.html - the frame
4458 MyModule.pm.toc.html - the table of contents
4459 MyModule.pm.src.html - the formatted source code
4461 Obviously this file naming scheme requires that output be directed to a real
4462 file (as opposed to, say, standard output). If this is not the
4463 case, or if the file extension is unknown, the B<-frm> option will be
4466 =item The B<-text=s>, or B<--html-toc-extension> flag
4468 Use this flag to specify the extra file extension of the table of contents file
4469 when html frames are used. The default is "toc".
4470 See L<Specifying File Extensions>.
4472 =item The B<-sext=s>, or B<--html-src-extension> flag
4474 Use this flag to specify the extra file extension of the content file when html
4475 frames are used. The default is "src".
4476 See L<Specifying File Extensions>.
4478 =item The B<-hent>, or B<--html-entities> flag
4480 This flag controls the use of Html::Entities for html formatting. By
4481 default, the module Html::Entities is used to encode special symbols.
4482 This may not be the right thing for some browser/language
4483 combinations. Use --nohtml-entities or -nhent to prevent this.
4485 =item B<Style Sheets>
4487 Style sheets make it very convenient to control and adjust the
4488 appearance of html pages. The default behavior is to write a page of
4489 html with an embedded style sheet.
4491 An alternative to an embedded style sheet is to create a page with a
4492 link to an external style sheet. This is indicated with the
4493 B<-css=filename>, where the external style sheet is F<filename>. The
4494 external style sheet F<filename> will be created if and only if it does
4495 not exist. This option is useful for controlling multiple pages from a
4498 To cause perltidy to write a style sheet to standard output and exit,
4499 use the B<-ss>, or B<--stylesheet>, flag. This is useful if the style
4500 sheet could not be written for some reason, such as if the B<-pre> flag
4501 was used. Thus, for example,
4503 perltidy -html -ss >mystyle.css
4505 will write a style sheet with the default properties to file
4508 The use of style sheets is encouraged, but a web page without a style
4509 sheets can be created with the flag B<-nss>. Use this option if you
4510 must to be sure that older browsers (roughly speaking, versions prior to
4511 4.0 of Netscape Navigator and Internet Explorer) can display the
4512 syntax-coloring of the html files.
4514 =item B<Controlling HTML properties>
4516 Note: It is usually more convenient to accept the default properties
4517 and then edit the stylesheet which is produced. However, this section
4518 shows how to control the properties with flags to perltidy.
4520 Syntax colors may be changed from their default values by flags of the either
4521 the long form, B<-html-color-xxxxxx=n>, or more conveniently the short form,
4522 B<-hcx=n>, where B<xxxxxx> is one of the following words, and B<x> is the
4523 corresponding abbreviation:
4526 ---------- -------- --
4529 identifier identifier i
4530 bareword, function bareword w
4532 quite, pattern quote q
4533 here doc text here-doc-text h
4534 here doc target here-doc-target hh
4535 punctuation punctuation pu
4537 structural braces structure s
4538 semicolon semicolon sc
4542 sub definition name subroutine m
4543 pod text pod-text pd
4545 A default set of colors has been defined, but they may be changed by providing
4546 values to any of the following parameters, where B<n> is either a 6 digit
4547 hex RGB color value or an ascii name for a color, such as 'red'.
4549 To illustrate, the following command will produce an html
4550 file F<somefile.pl.html> with "aqua" keywords:
4552 perltidy -html -hck=00ffff somefile.pl
4554 and this should be equivalent for most browsers:
4556 perltidy -html -hck=aqua somefile.pl
4558 Perltidy merely writes any non-hex names that it sees in the html file.
4559 The following 16 color names are defined in the HTML 3.2 standard:
4578 Many more names are supported in specific browsers, but it is safest
4579 to use the hex codes for other colors. Helpful color tables can be
4580 located with an internet search for "HTML color tables".
4582 Besides color, two other character attributes may be set: bold, and italics.
4583 To set a token type to use bold, use the flag
4584 B<--html-bold-xxxxxx> or B<-hbx>, where B<xxxxxx> or B<x> are the long
4585 or short names from the above table. Conversely, to set a token type to
4586 NOT use bold, use B<--nohtml-bold-xxxxxx> or B<-nhbx>.
4588 Likewise, to set a token type to use an italic font, use the flag
4589 B<--html-italic-xxxxxx> or B<-hix>, where again B<xxxxxx> or B<x> are the
4590 long or short names from the above table. And to set a token type to
4591 NOT use italics, use B<--nohtml-italic-xxxxxx> or B<-nhix>.
4593 For example, to use bold braces and lime color, non-bold, italics keywords the
4594 following command would be used:
4596 perltidy -html -hbs -hck=00FF00 -nhbk -hik somefile.pl
4598 The background color can be specified with B<--html-color-background=n>,
4599 or B<-hcbg=n> for short, where n is a 6 character hex RGB value. The
4600 default color of text is the value given to B<punctuation>, which is
4603 Here are some notes and hints:
4605 1. If you find a preferred set of these parameters, you may want
4606 to create a F<.perltidyrc> file containing them. See the perltidy man
4607 page for an explanation.
4609 2. Rather than specifying values for these parameters, it is probably
4610 easier to accept the defaults and then edit a style sheet. The style
4611 sheet contains comments which should make this easy.
4613 3. The syntax-colored html files can be very large, so it may be best to
4614 split large files into smaller pieces to improve download times.
4618 =head1 SOME COMMON INPUT CONVENTIONS
4620 =head2 Specifying Block Types
4622 Several parameters which refer to code block types may be customized by also
4623 specifying an associated list of block types. The type of a block is the name
4624 of the keyword which introduces that block, such as B<if>, B<else>, or B<sub>.
4625 An exception is a labeled block, which has no keyword, and should be specified
4626 with just a colon. To specify all blocks use B<'*'>.
4628 The keyword B<sub> indicates a named sub. For anonymous subs, use the special
4631 For example, the following parameter specifies C<sub>, labels, C<BEGIN>, and
4634 -cscl="sub : BEGIN END"
4636 (the meaning of the -cscl parameter is described above.) Note that
4637 quotes are required around the list of block types because of the
4638 spaces. For another example, the following list specifies all block types
4639 for vertical tightness:
4643 =head2 Specifying File Extensions
4645 Several parameters allow default file extensions to be overridden. For
4646 example, a backup file extension may be specified with B<-bext=ext>,
4647 where B<ext> is some new extension. In order to provides the user some
4648 flexibility, the following convention is used in all cases to decide if
4649 a leading '.' should be used. If the extension C<ext> begins with
4650 C<A-Z>, C<a-z>, or C<0-9>, then it will be appended to the filename with
4651 an intermediate '.' (or perhaps a '_' on VMS systems). Otherwise, it
4652 will be appended directly.
4654 For example, suppose the file is F<somefile.pl>. For C<-bext=old>, a '.' is
4655 added to give F<somefile.pl.old>. For C<-bext=.old>, no additional '.' is
4656 added, so again the backup file is F<somefile.pl.old>. For C<-bext=~>, then no
4657 dot is added, and the backup file will be F<somefile.pl~> .
4659 =head1 SWITCHES WHICH MAY BE NEGATED
4661 The following list shows all short parameter names which allow a prefix
4662 'n' to produce the negated form:
4664 D anl asbl asc ast asu atnl aws b baa
4665 baao bar bbao bbb bbc bbs bl bli boa boc
4666 bok bol bom bos bot cblx ce conv cs csc
4667 cscb cscw dac dbc dcbl dcsc ddf dln dnl dop
4668 dp dpro dsc dsm dsn dtt dwls dwrs dws f
4669 fll fpva frm fs fso gcs hbc hbcm hbco hbh
4670 hbhh hbi hbj hbk hbm hbn hbp hbpd hbpu hbq
4671 hbs hbsc hbv hbw hent hic hicm hico hih hihh
4672 hii hij hik him hin hip hipd hipu hiq his
4673 hisc hiv hiw hsc html ibc icb icp iob isbc
4674 iscl kgb kgbd kgbi kis lal log lop lp lsl
4675 mem nib ohbr okw ola olc oll olq opr opt
4676 osbc osbr otr ple pod pvl q sac sbc sbl
4677 scbb schb scp scsb sct se sfp sfs skp sob
4678 sobb sohb sop sosb sot ssc st sts t tac
4679 tbc toc tp tqw trp ts tsc tso vmll w
4682 Equivalently, the prefix 'no' or 'no-' on the corresponding long names may be
4689 =item B<Parsing Limitations>
4691 Perltidy should work properly on most perl scripts. It does a lot of
4692 self-checking, but still, it is possible that an error could be
4693 introduced and go undetected. Therefore, it is essential to make
4694 careful backups and to test reformatted scripts.
4696 The main current limitation is that perltidy does not scan modules
4697 included with 'use' statements. This makes it necessary to guess the
4698 context of any bare words introduced by such modules. Perltidy has good
4699 guessing algorithms, but they are not infallible. When it must guess,
4700 it leaves a message in the log file.
4702 If you encounter a bug, please report it.
4704 =item B<What perltidy does not parse and format>
4706 Perltidy indents but does not reformat comments and C<qw> quotes.
4707 Perltidy does not in any way modify the contents of here documents or
4708 quoted text, even if they contain source code. (You could, however,
4709 reformat them separately). Perltidy does not format 'format' sections
4710 in any way. And, of course, it does not modify pod documents.
4718 =item B<Temporary files>
4720 Under the -html option with the default --pod2html flag, a temporary file is
4721 required to pass text to Pod::Html. Unix systems will try to use the POSIX
4722 tmpnam() function. Otherwise the file F<perltidy.TMP> will be temporarily
4723 created in the current working directory.
4725 =item B<Special files when standard input is used>
4727 When standard input is used, the log file, if saved, is F<perltidy.LOG>,
4728 and any errors are written to F<perltidy.ERR> unless the B<-se> flag is
4729 set. These are saved in the current working directory.
4731 =item B<Files overwritten>
4733 The following file extensions are used by perltidy, and files with these
4734 extensions may be overwritten or deleted: F<.ERR>, F<.LOG>, F<.TEE>,
4735 and/or F<.tdy>, F<.html>, and F<.bak>, depending on the run type and
4738 =item B<Files extensions limitations>
4740 Perltidy does not operate on files for which the run could produce a file with
4741 a duplicated file extension. These extensions include F<.LOG>, F<.ERR>,
4742 F<.TEE>, and perhaps F<.tdy> and F<.bak>, depending on the run type. The
4743 purpose of this rule is to prevent generating confusing filenames such as
4744 F<somefile.tdy.tdy.tdy>.
4748 =head1 ERROR HANDLING
4750 An exit value of 0, 1, or 2 is returned by perltidy to indicate the status of the result.
4752 A exit value of 0 indicates that perltidy ran to completion with no error messages.
4754 A non-zero exit value indicates some kind of problem was detected.
4756 An exit value of 1 indicates that perltidy terminated prematurely, usually due
4757 to some kind of errors in the input parameters. This can happen for example if
4758 a parameter is misspelled or given an invalid value. Error messages in the
4759 standard error output will indicate the cause of any problem. If perltidy
4760 terminates prematurely then no output files will be produced.
4762 An exit value of 2 indicates that perltidy was able to run to completion but
4763 there there are (1) warning messages in the standard error output related to
4764 parameter errors or problems and/or (2) warning messages in the perltidy error
4765 file(s) relating to possible syntax errors in one or more of the source
4766 script(s) being tidied. When multiple files are being processed, an error
4767 detected in any single file will produce this type of exit condition.
4771 perlstyle(1), Perl::Tidy(3)
4775 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'.
4779 This man page documents perltidy version 20210717
4783 A list of current bugs and issues can be found at the CPAN site L<https://rt.cpan.org/Public/Dist/Display.html?Name=Perl-Tidy>
4785 To report a new bug or problem, use the link on this page.
4787 The source code repository is at L<https://github.com/perltidy/perltidy>.
4791 Copyright (c) 2000-2021 by Steve Hancock
4795 This package is free software; you can redistribute it and/or modify it
4796 under the terms of the "GNU General Public License".
4798 Please refer to the file "COPYING" for details.
4802 This package is distributed in the hope that it will be useful,
4803 but WITHOUT ANY WARRANTY; without even the implied warranty of
4804 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
4806 See the "GNU General Public License" for more details.
projects / perltidy.git / blob