1 # Pod::Man -- Convert POD data to formatted *roff input.
2 # $Id: Man.pm,v 1.30 2001/11/28 01:14:28 eagle Exp $
4 # Copyright 1999, 2000, 2001 by Russ Allbery <rra@stanford.edu>
6 # This program is free software; you may redistribute it and/or modify it
7 # under the same terms as Perl itself.
9 # This module translates POD documentation into *roff markup using the man
10 # macro set, and is intended for converting POD documents written as Unix
11 # manual pages to manual pages that can be read by the man(1) command. It is
12 # a replacement for the pod2man command distributed with versions of Perl
15 # Perl core hackers, please note that this module is also separately
16 # maintained outside of the Perl core as part of the podlators. Please send
17 # me any patches at the address above in addition to sending them to the
18 # standard Perl mailing lists.
20 ##############################################################################
21 # Modules and declarations
22 ##############################################################################
28 use Carp qw(carp croak);
29 use Pod::ParseLink qw(parselink);
33 use subs qw(makespace);
34 use vars qw(@ISA %ESCAPES $PREAMBLE $VERSION);
36 @ISA = qw(Pod::Parser);
38 # Don't use the CVS revision as the version, since this module is also in Perl
39 # core and too many things could munge CVS magic revision strings. This
40 # number should ideally be the same as the CVS revision in podlators, however.
44 ##############################################################################
45 # Preamble and *roff output tables
46 ##############################################################################
48 # The following is the static preamble which starts all *roff output we
49 # generate. It's completely static except for the font to use as a
50 # fixed-width font, which is designed by @CFONT@, and the left and right
51 # quotes to use for C<> text, designated by @LQOUTE@ and @RQUOTE@. $PREAMBLE
52 # should therefore be run through s/\@CFONT\@/<font>/g before output.
53 $PREAMBLE = <<'----END OF PREAMBLE----';
54 .de Sh \" Subsection heading
62 .de Sp \" Vertical space (when we can't use .PP)
66 .de Vb \" Begin verbatim text
71 .de Ve \" End verbatim text
76 .\" Set up some character translations and predefined strings. \*(-- will
77 .\" give an unbreakable dash, \*(PI will give pi, \*(L" will give a left
78 .\" double quote, and \*(R" will give a right double quote. | will give a
79 .\" real vertical bar. \*(C+ will give a nicer C++. Capital omega is used to
80 .\" do unbreakable dashes and therefore won't be available. \*(C` and \*(C'
81 .\" expand to `' in nroff, nothing in troff, for use with C<>.
83 .ds C+ C\v'-.1v'\h'-1p'\s-2+\h'-1p'+\s0\v'.1v'\h'-1p'
87 . if (\n(.H=4u)&(1m=24u) .ds -- \(*W\h'-12u'\(*W\h'-12u'-\" diablo 10 pitch
88 . if (\n(.H=4u)&(1m=20u) .ds -- \(*W\h'-12u'\(*W\h'-8u'-\" diablo 12 pitch
101 .\" If the F register is turned on, we'll generate index entries on stderr for
102 .\" titles (.TH), headers (.SH), subsections (.Sh), items (.Ip), and index
103 .\" entries marked with X<> in POD. Of course, you'll have to process the
104 .\" output yourself in some meaningful fashion.
107 . tm Index:\\$1\t\\n%\t"\\$2"
113 .\" For nroff, turn off justification. Always turn off hyphenation; it makes
114 .\" way too many mistakes in technical documents.
118 .\" Accent mark definitions (@(#)ms.acc 1.5 88/02/08 SMI; from UCB 4.2).
119 .\" Fear. Run. Save yourself. No user-serviceable parts.
120 . \" fudge factors for nroff and troff
129 . ds #H ((1u-(\\\\n(.fu%2u))*.13m)
135 . \" simple accents for nroff and troff
145 . ds ' \\k:\h'-(\\n(.wu*8/10-\*(#H)'\'\h"|\\n:u"
146 . ds ` \\k:\h'-(\\n(.wu*8/10-\*(#H)'\`\h'|\\n:u'
147 . ds ^ \\k:\h'-(\\n(.wu*10/11-\*(#H)'^\h'|\\n:u'
148 . ds , \\k:\h'-(\\n(.wu*8/10)',\h'|\\n:u'
149 . ds ~ \\k:\h'-(\\n(.wu-\*(#H-.1m)'~\h'|\\n:u'
150 . ds / \\k:\h'-(\\n(.wu*8/10-\*(#H)'\z\(sl\h'|\\n:u'
152 . \" troff and (daisy-wheel) nroff accents
153 .ds : \\k:\h'-(\\n(.wu*8/10-\*(#H+.1m+\*(#F)'\v'-\*(#V'\z.\h'.2m+\*(#F'.\h'|\\n:u'\v'\*(#V'
154 .ds 8 \h'\*(#H'\(*b\h'-\*(#H'
155 .ds o \\k:\h'-(\\n(.wu+\w'\(de'u-\*(#H)/2u'\v'-.3n'\*(#[\z\(de\v'.3n'\h'|\\n:u'\*(#]
156 .ds d- \h'\*(#H'\(pd\h'-\w'~'u'\v'-.25m'\f2\(hy\fP\v'.25m'\h'-\*(#H'
157 .ds D- D\\k:\h'-\w'D'u'\v'-.11m'\z\(hy\v'.11m'\h'|\\n:u'
158 .ds th \*(#[\v'.3m'\s+1I\s-1\v'-.3m'\h'-(\w'I'u*2/3)'\s-1o\s+1\*(#]
159 .ds Th \*(#[\s+2I\s-2\h'-\w'I'u*3/5'\v'-.3m'o\v'.3m'\*(#]
160 .ds ae a\h'-(\w'a'u*4/10)'e
161 .ds Ae A\h'-(\w'A'u*4/10)'E
162 . \" corrections for vroff
163 .if v .ds ~ \\k:\h'-(\\n(.wu*9/10-\*(#H)'\s-2\u~\d\s+2\h'|\\n:u'
164 .if v .ds ^ \\k:\h'-(\\n(.wu*10/11-\*(#H)'\v'-.4m'^\v'.4m'\h'|\\n:u'
165 . \" for low resolution devices (crt and lpr)
166 .if \n(.H>23 .if \n(.V>19 \
179 ----END OF PREAMBLE----
182 # This table is taken nearly verbatim from Tom Christiansen's pod2man. It
183 # assumes that the standard preamble has already been printed, since that's
184 # what defines all of the accent marks. Note that some of these are quoted
185 # with double quotes since they contain embedded single quotes, so use \\
186 # uniformly for backslash for readability.
188 'amp' => '&', # ampersand
189 'apos' => "'", # apostrophe
190 'lt' => '<', # left chevron, less-than
191 'gt' => '>', # right chevron, greater-than
192 'quot' => '"', # double quote
193 'sol' => '/', # solidus (forward slash)
194 'verbar' => '|', # vertical bar
196 'Aacute' => "A\\*'", # capital A, acute accent
197 'aacute' => "a\\*'", # small a, acute accent
198 'Acirc' => 'A\\*^', # capital A, circumflex accent
199 'acirc' => 'a\\*^', # small a, circumflex accent
200 'AElig' => '\*(AE', # capital AE diphthong (ligature)
201 'aelig' => '\*(ae', # small ae diphthong (ligature)
202 'Agrave' => "A\\*`", # capital A, grave accent
203 'agrave' => "A\\*`", # small a, grave accent
204 'Aring' => 'A\\*o', # capital A, ring
205 'aring' => 'a\\*o', # small a, ring
206 'Atilde' => 'A\\*~', # capital A, tilde
207 'atilde' => 'a\\*~', # small a, tilde
208 'Auml' => 'A\\*:', # capital A, dieresis or umlaut mark
209 'auml' => 'a\\*:', # small a, dieresis or umlaut mark
210 'Ccedil' => 'C\\*,', # capital C, cedilla
211 'ccedil' => 'c\\*,', # small c, cedilla
212 'Eacute' => "E\\*'", # capital E, acute accent
213 'eacute' => "e\\*'", # small e, acute accent
214 'Ecirc' => 'E\\*^', # capital E, circumflex accent
215 'ecirc' => 'e\\*^', # small e, circumflex accent
216 'Egrave' => 'E\\*`', # capital E, grave accent
217 'egrave' => 'e\\*`', # small e, grave accent
218 'ETH' => '\\*(D-', # capital Eth, Icelandic
219 'eth' => '\\*(d-', # small eth, Icelandic
220 'Euml' => 'E\\*:', # capital E, dieresis or umlaut mark
221 'euml' => 'e\\*:', # small e, dieresis or umlaut mark
222 'Iacute' => "I\\*'", # capital I, acute accent
223 'iacute' => "i\\*'", # small i, acute accent
224 'Icirc' => 'I\\*^', # capital I, circumflex accent
225 'icirc' => 'i\\*^', # small i, circumflex accent
226 'Igrave' => 'I\\*`', # capital I, grave accent
227 'igrave' => 'i\\*`', # small i, grave accent
228 'Iuml' => 'I\\*:', # capital I, dieresis or umlaut mark
229 'iuml' => 'i\\*:', # small i, dieresis or umlaut mark
230 'Ntilde' => 'N\*~', # capital N, tilde
231 'ntilde' => 'n\*~', # small n, tilde
232 'Oacute' => "O\\*'", # capital O, acute accent
233 'oacute' => "o\\*'", # small o, acute accent
234 'Ocirc' => 'O\\*^', # capital O, circumflex accent
235 'ocirc' => 'o\\*^', # small o, circumflex accent
236 'Ograve' => 'O\\*`', # capital O, grave accent
237 'ograve' => 'o\\*`', # small o, grave accent
238 'Oslash' => 'O\\*/', # capital O, slash
239 'oslash' => 'o\\*/', # small o, slash
240 'Otilde' => 'O\\*~', # capital O, tilde
241 'otilde' => 'o\\*~', # small o, tilde
242 'Ouml' => 'O\\*:', # capital O, dieresis or umlaut mark
243 'ouml' => 'o\\*:', # small o, dieresis or umlaut mark
244 'szlig' => '\*8', # small sharp s, German (sz ligature)
245 'THORN' => '\\*(Th', # capital THORN, Icelandic
246 'thorn' => '\\*(th', # small thorn, Icelandic
247 'Uacute' => "U\\*'", # capital U, acute accent
248 'uacute' => "u\\*'", # small u, acute accent
249 'Ucirc' => 'U\\*^', # capital U, circumflex accent
250 'ucirc' => 'u\\*^', # small u, circumflex accent
251 'Ugrave' => 'U\\*`', # capital U, grave accent
252 'ugrave' => 'u\\*`', # small u, grave accent
253 'Uuml' => 'U\\*:', # capital U, dieresis or umlaut mark
254 'uuml' => 'u\\*:', # small u, dieresis or umlaut mark
255 'Yacute' => "Y\\*'", # capital Y, acute accent
256 'yacute' => "y\\*'", # small y, acute accent
257 'yuml' => 'y\\*:', # small y, dieresis or umlaut mark
259 'nbsp' => '\\ ', # non-breaking space
260 'shy' => '', # soft (discretionary) hyphen
264 ##############################################################################
265 # Static helper functions
266 ##############################################################################
268 # Protect leading quotes and periods against interpretation as commands. Also
269 # protect anything starting with a backslash, since it could expand or hide
270 # something that *roff would interpret as a command. This is overkill, but
271 # it's much simpler than trying to parse *roff here.
274 s/^([.\'\\])/\\&$1/mg;
278 # Translate a font string into an escape.
279 sub toescape { (length ($_[0]) > 1 ? '\f(' : '\f') . $_[0] }
282 ##############################################################################
284 ##############################################################################
286 # Initialize the object. Here, we also process any additional options passed
287 # to the constructor or set up defaults if none were given. center is the
288 # centered title, release is the version number, and date is the date for the
289 # documentation. Note that we can't know what file name we're processing due
290 # to the architecture of Pod::Parser, so that *has* to either be passed to the
291 # constructor or set separately with Pod::Man::name().
295 # Figure out the fixed-width font. If user-supplied, make sure that they
296 # are the right length.
297 for (qw/fixed fixedbold fixeditalic fixedbolditalic/) {
298 if (defined $$self{$_}) {
299 if (length ($$self{$_}) < 1 || length ($$self{$_}) > 2) {
300 croak qq(roff font should be 1 or 2 chars,)
301 . qq( not "$$self{$_}");
308 # Set the default fonts. We can't be sure what fixed bold-italic is going
309 # to be called, so default to just bold.
310 $$self{fixed} ||= 'CW';
311 $$self{fixedbold} ||= 'CB';
312 $$self{fixeditalic} ||= 'CI';
313 $$self{fixedbolditalic} ||= 'CB';
315 # Set up a table of font escapes. First number is fixed-width, second is
316 # bold, third is italic.
317 $$self{FONTS} = { '000' => '\fR', '001' => '\fI',
318 '010' => '\fB', '011' => '\f(BI',
319 '100' => toescape ($$self{fixed}),
320 '101' => toescape ($$self{fixeditalic}),
321 '110' => toescape ($$self{fixedbold}),
322 '111' => toescape ($$self{fixedbolditalic})};
324 # Extra stuff for page titles.
325 $$self{center} = 'User Contributed Perl Documentation'
326 unless defined $$self{center};
327 $$self{indent} = 4 unless defined $$self{indent};
329 # We used to try first to get the version number from a local binary, but
330 # we shouldn't need that any more. Get the version from the running Perl.
331 # Work a little magic to handle subversions correctly under both the
332 # pre-5.6 and the post-5.6 version numbering schemes.
333 if (!defined $$self{release}) {
334 my @version = ($] =~ /^(\d+)\.(\d{3})(\d{0,3})$/);
336 $version[2] *= 10 ** (3 - length $version[2]);
337 for (@version) { $_ += 0 }
338 $$self{release} = 'perl v' . join ('.', @version);
341 # Double quotes in things that will be quoted.
342 for (qw/center date release/) {
343 $$self{$_} =~ s/\"/\"\"/g if $$self{$_};
346 # Figure out what quotes we'll be using for C<> text.
347 $$self{quotes} ||= '"';
348 if ($$self{quotes} eq 'none') {
349 $$self{LQUOTE} = $$self{RQUOTE} = '';
350 } elsif (length ($$self{quotes}) == 1) {
351 $$self{LQUOTE} = $$self{RQUOTE} = $$self{quotes};
352 } elsif ($$self{quotes} =~ /^(.)(.)$/
353 || $$self{quotes} =~ /^(..)(..)$/) {
357 croak qq(Invalid quote specification "$$self{quotes}");
360 # Double the first quote; note that this should not be s///g as two double
361 # quotes is represented in *roff as three double quotes, not four. Weird,
363 $$self{LQUOTE} =~ s/\"/\"\"/;
364 $$self{RQUOTE} =~ s/\"/\"\"/;
366 $self->SUPER::initialize;
369 # For each document we process, output the preamble first.
373 # Try to figure out the name and section from the file name.
374 my $section = $$self{section} || 1;
375 my $name = $$self{name};
376 if (!defined $name) {
377 $name = $self->input_file;
378 $section = 3 if (!$$self{section} && $name =~ /\.pm\z/i);
379 $name =~ s/\.p(od|[lm])\z//i;
380 if ($section !~ /^3/) {
381 require File::Basename;
382 $name = uc File::Basename::basename ($name);
384 # Assume that we're dealing with a module. We want to figure out
385 # the full module name from the path to the file, but we don't
386 # want to include too much of the path into the module name. Lose
387 # everything up to the first of:
389 # */lib/*perl*/ standard or site_perl module
390 # */*perl*/lib/ from -Dprefix=/opt/perl
391 # */*perl*/ random module hierarchy
393 # which works. Also strip off a leading site or site_perl
394 # component, any OS-specific component, and any version number
395 # component, and strip off an initial component of "lib" or
396 # "blib/lib" since that's what ExtUtils::MakeMaker creates.
397 # splitdir requires at least File::Spec 0.8.
399 my ($volume, $dirs, $file) = File::Spec->splitpath ($name);
400 my @dirs = File::Spec->splitdir ($dirs);
403 for ($i = 0; $i < scalar @dirs; $i++) {
404 if ($dirs[$i] eq 'lib' && $dirs[$i + 1] =~ /perl/) {
407 } elsif ($dirs[$i] =~ /perl/) {
409 $cut++ if $dirs[$i + 1] eq 'lib';
414 splice (@dirs, 0, $cut);
415 shift @dirs if ($dirs[0] =~ /^site(_perl)?$/);
416 shift @dirs if ($dirs[0] =~ /^[\d.]+$/);
417 shift @dirs if ($dirs[0] =~ /^(.*-$^O|$^O-.*|$^O)$/);
419 shift @dirs if $dirs[0] eq 'lib';
420 splice (@dirs, 0, 2) if ($dirs[0] eq 'blib' && $dirs[1] eq 'lib');
422 # Remove empty directories when building the module name; they
423 # occur too easily on Unix by doubling slashes.
424 $name = join ('::', (grep { $_ ? $_ : () } @dirs), $file);
428 # If $name contains spaces, quote it; this mostly comes up in the case of
430 $name = '"' . $name . '"' if ($name =~ /\s/);
432 # Modification date header. Try to use the modification time of our
434 if (!defined $$self{date}) {
435 my $time = (stat $self->input_file)[9] || time;
436 my ($day, $month, $year) = (localtime $time)[3,4,5];
439 $$self{date} = sprintf ('%4d-%02d-%02d', $year, $month, $day);
442 # Now, print out the preamble and the title. The meaning of the arguments
443 # to .TH unfortunately vary by system; some systems consider the fourth
444 # argument to be a "source" and others use it as a version number.
445 # Generally it's just presented as the left-side footer, though, so it
446 # doesn't matter too much if a particular system gives it another
449 # The order of date and release used to be reversed in older versions of
450 # this module, but this order is correct for both Solaris and Linux.
451 local $_ = $PREAMBLE;
452 s/\@CFONT\@/$$self{fixed}/;
453 s/\@LQUOTE\@/$$self{LQUOTE}/;
454 s/\@RQUOTE\@/$$self{RQUOTE}/;
456 my $pversion = $Pod::Parser::VERSION;
457 print { $self->output_handle } <<"----END OF HEADER----";
458 .\\" Automatically generated by Pod::Man v$VERSION, Pod::Parser v$pversion
460 .\\" Standard preamble:
461 .\\" ========================================================================
463 .\\" ========================================================================
465 .IX Title "$name $section"
466 .TH $name $section "$$self{date}" "$$self{release}" "$$self{center}"
468 ----END OF HEADER----
470 # Initialize a few per-file variables.
471 $$self{INDENT} = 0; # Current indentation level.
472 $$self{INDENTS} = []; # Stack of indentations.
473 $$self{INDEX} = []; # Index keys waiting to be printed.
474 $$self{ITEMS} = 0; # The number of consecutive =items.
475 $$self{SHIFTWAIT} = 0; # Whether there is a shift waiting.
476 $$self{SHIFTS} = []; # Stack of .RS shifts.
480 ##############################################################################
482 ##############################################################################
484 # Called for each command paragraph. Gets the command, the associated
485 # paragraph, the line number, and a Pod::Paragraph object. Just dispatches
486 # the command to a method named the same as the command. =cut is handled
487 # internally by Pod::Parser.
491 return if $command eq 'pod';
492 return if ($$self{EXCLUDE} && $command ne 'end');
493 if ($self->can ('cmd_' . $command)) {
494 $command = 'cmd_' . $command;
495 $self->$command (@_);
497 my ($text, $line, $paragraph) = @_;
499 ($file, $line) = $paragraph->file_line;
501 $text = " $text" if ($text =~ /^\S/);
502 warn qq($file:$line: Unknown command paragraph "=$command$text"\n);
507 # Called for a verbatim paragraph. Gets the paragraph, the line number, and a
508 # Pod::Paragraph object. Rofficate backslashes, untabify, put a zero-width
509 # character at the beginning of each line to protect against commands, and
513 return if $$self{EXCLUDE};
517 my $lines = tr/\n/\n/;
518 1 while s/^(.*?)(\t+)/$1 . ' ' x (length ($2) * 8 - length ($1) % 8)/me;
520 s/^(\s*\S)/'\&' . $1/gme;
522 $self->output (".Vb $lines\n$_.Ve\n");
523 $$self{NEEDSPACE} = 0;
526 # Called for a regular text block. Gets the paragraph, the line number, and a
527 # Pod::Paragraph object. Perform interpolation and output the results.
530 return if $$self{EXCLUDE};
531 $self->output ($_[0]), return if $$self{VERBATIM};
533 # Parse the tree. collapse knows about references to scalars as well as
534 # scalars and does the right thing with them. Tidy up any trailing
537 $text = $self->parse ($text, @_);
538 $text =~ s/\n\s*$/\n/;
540 # Output the paragraph. We also have to handle =over without =item. If
541 # there's an =over without =item, NEWINDENT will be set, and we need to
542 # handle creation of the indent here. Set WEIRDINDENT so that it will be
543 # cleaned up on =back.
545 if ($$self{SHIFTWAIT}) {
546 $self->output (".RS $$self{INDENT}\n");
547 push (@{ $$self{SHIFTS} }, $$self{INDENT});
548 $$self{SHIFTWAIT} = 0;
550 $self->output (protect $self->textmapfonts ($text));
552 $$self{NEEDSPACE} = 1;
555 # Called for a formatting code. Takes a Pod::InteriorSequence object and
556 # returns a reference to a scalar. This scalar is the final formatted text.
557 # It's returned as a reference to an array so that other formatting codes
558 # above us know that the text has already been processed.
560 my ($self, $seq) = @_;
561 my $command = $seq->cmd_name;
563 # We have to defer processing of the inside of an L<> formatting code. If
564 # this code is nested inside an L<> code, return the literal raw text of
566 my $parent = $seq->nested;
567 while (defined $parent) {
568 return $seq->raw_text if ($parent->cmd_name eq 'L');
569 $parent = $parent->nested;
572 # Zero-width characters.
573 return [ '\&' ] if ($command eq 'Z');
575 # C<>, L<>, X<>, and E<> don't apply guesswork to their contents. C<>
576 # needs some additional special handling.
577 my $literal = ($command =~ /^[CELX]$/);
578 local $_ = $self->collapse ($seq->parse_tree, $literal, $command eq 'C');
580 # Handle E<> escapes. Numeric escapes that match one of the supported ISO
581 # 8859-1 characters don't work at present.
582 if ($command eq 'E') {
585 } elsif (exists $ESCAPES{$_}) {
586 return [ $ESCAPES{$_} ];
588 my ($file, $line) = $seq->file_line;
589 warn "$file:$line: Unknown escape E<$_>\n";
594 # For all the other codes, empty content produces no output.
595 return '' if $_ eq '';
597 # Handle simple formatting codes.
598 if ($command eq 'B') {
599 return [ '\f(BS' . $_ . '\f(BE' ];
600 } elsif ($command eq 'F' || $command eq 'I') {
601 return [ '\f(IS' . $_ . '\f(IE' ];
602 } elsif ($command eq 'C') {
603 return [ $self->quote_literal ($_) ];
607 if ($command eq 'L') {
608 my ($text, $type) = (parselink ($_))[1,4];
609 return '' unless $text;
610 my ($file, $line) = $seq->file_line;
611 $text = $self->parse ($text, $line);
612 $text = '<' . $text . '>' if $type eq 'url';
616 # Whitespace protection replaces whitespace with "\ ".
617 if ($command eq 'S') {
622 # Add an index entry to the list of ones waiting to be output.
623 if ($command eq 'X') {
624 push (@{ $$self{INDEX} }, $_);
628 # Anything else is unknown.
629 my ($file, $line) = $seq->file_line;
630 warn "$file:$line: Unknown formatting code $command<$_>\n";
634 ##############################################################################
636 ##############################################################################
638 # All command paragraphs take the paragraph and the line number.
640 # First level heading. We can't output .IX in the NAME section due to a bug
641 # in some versions of catman, so don't output a .IX for that section. .SH
642 # already uses small caps, so remove \s1 and \s-1.
645 local $_ = $self->parse (@_);
649 if ($$self{ITEMS} > 1) {
651 $self->output (".PD\n");
653 $self->output ($self->switchquotes ('.SH', $self->mapfonts ($_)));
654 $self->outindex (($_ eq 'NAME') ? () : ('Header', $_));
655 $$self{NEEDSPACE} = 0;
658 # Second level heading.
661 local $_ = $self->parse (@_);
664 if ($$self{ITEMS} > 1) {
666 $self->output (".PD\n");
668 $self->output ($self->switchquotes ('.Sh', $self->mapfonts ($_)));
669 $self->outindex ('Subsection', $_);
670 $$self{NEEDSPACE} = 0;
673 # Third level heading.
676 local $_ = $self->parse (@_);
679 if ($$self{ITEMS} > 1) {
681 $self->output (".PD\n");
684 $self->output ($self->textmapfonts ('\f(IS' . $_ . '\f(IE') . "\n");
685 $self->outindex ('Subsection', $_);
686 $$self{NEEDSPACE} = 1;
689 # Fourth level heading.
692 local $_ = $self->parse (@_);
695 if ($$self{ITEMS} > 1) {
697 $self->output (".PD\n");
700 $self->output ($self->textmapfonts ($_) . "\n");
701 $self->outindex ('Subsection', $_);
702 $$self{NEEDSPACE} = 1;
705 # Start a list. For indents after the first, wrap the outside indent in .RS
706 # so that hanging paragraph tags will be correct.
710 unless (/^[-+]?\d+\s+$/) { $_ = $$self{indent} }
711 if (@{ $$self{SHIFTS} } < @{ $$self{INDENTS} }) {
712 $self->output (".RS $$self{INDENT}\n");
713 push (@{ $$self{SHIFTS} }, $$self{INDENT});
715 push (@{ $$self{INDENTS} }, $$self{INDENT});
716 $$self{INDENT} = ($_ + 0);
717 $$self{SHIFTWAIT} = 1;
720 # End a list. If we've closed an embedded indent, we've mangled the hanging
721 # paragraph indent, so temporarily replace it with .RS and set WEIRDINDENT.
722 # We'll close that .RS at the next =back or =item.
725 $$self{INDENT} = pop @{ $$self{INDENTS} };
726 unless (defined $$self{INDENT}) {
727 my ($file, $line, $paragraph) = @_;
728 ($file, $line) = $paragraph->file_line;
729 warn "$file:$line: Unmatched =back\n";
732 if (@{ $$self{SHIFTS} } > @{ $$self{INDENTS} }) {
733 $self->output (".RE\n");
734 pop @{ $$self{SHIFTS} };
736 if (@{ $$self{INDENTS} } > 0) {
737 $self->output (".RE\n");
738 $self->output (".RS $$self{INDENT}\n");
740 $$self{NEEDSPACE} = 1;
741 $$self{SHIFTWAIT} = 0;
744 # An individual list item. Emit an index entry for anything that's
745 # interesting, but don't emit index entries for things like bullets and
746 # numbers. rofficate bullets too while we're at it (so for nice output, use *
747 # for your lists rather than o or . or - or some other thing). Newlines in an
748 # item title are turned into spaces since *roff can't handle them embedded.
751 local $_ = $self->parse (@_);
755 if (/\w/ && !/^\w[.\)]\s*$/) {
757 $index =~ s/^\s*[-*+o.]?(?:\s+|\Z)//;
760 s/^\*(\s|\Z)/\\\(bu$1/;
761 if (@{ $$self{SHIFTS} } == @{ $$self{INDENTS} }) {
762 $self->output (".RE\n");
763 pop @{ $$self{SHIFTS} };
765 $_ = $self->textmapfonts ($_);
766 $self->output (".PD 0\n") if ($$self{ITEMS} == 1);
767 $self->output ($self->switchquotes ('.IP', $_, $$self{INDENT}));
768 $self->outindex ($index ? ('Item', $index) : ());
769 $$self{NEEDSPACE} = 0;
771 $$self{SHIFTWAIT} = 0;
774 # Begin a block for a particular translator. Setting VERBATIM triggers
775 # special handling in textblock().
779 my ($kind) = /^(\S+)/ or return;
780 if ($kind eq 'man' || $kind eq 'roff') {
781 $$self{VERBATIM} = 1;
787 # End a block for a particular translator. We assume that all =begin/=end
788 # pairs are properly closed.
792 $$self{VERBATIM} = 0;
795 # One paragraph for a particular translator. Ignore it unless it's intended
796 # for man or roff, in which case we output it verbatim.
800 return unless s/^(?:man|roff)\b[ \t]*\n?//;
805 ##############################################################################
806 # Escaping and fontification
807 ##############################################################################
809 # At this point, we'll have embedded font codes of the form \f(<font>[SE]
810 # where <font> is one of B, I, or F. Turn those into the right font start or
811 # end codes. The old pod2man didn't get B<someI<thing> else> right; after I<>
812 # it switched back to normal text rather than bold. We take care of this by
813 # using variables as a combined pointer to our current font sequence, and set
814 # each to the number of current nestings of start tags for that font. Use
815 # them as a vector to look up what font sequence to use.
817 # \fP changes to the previous font, but only one previous font is kept. We
818 # don't know what the outside level font is; normally it's R, but if we're
819 # inside a heading it could be something else. So arrange things so that the
820 # outside font is always the "previous" font and end with \fP instead of \fR.
821 # Idea from Zack Weinberg.
826 my ($fixed, $bold, $italic) = (0, 0, 0);
827 my %magic = (F => \$fixed, B => \$bold, I => \$italic);
832 if ($last ne '\fR') { $sequence = '\fP' }
833 ${ $magic{$1} } += ($2 eq 'S') ? 1 : -1;
834 $f = $$self{FONTS}{($fixed && 1) . ($bold && 1) . ($italic && 1)};
838 if ($f ne '\fR') { $sequence .= $f }
846 # Unfortunately, there is a bug in Solaris 2.6 nroff (not present in GNU
847 # groff) where the sequence \fB\fP\f(CW\fP leaves the font set to B rather
848 # than R, presumably because \f(CW doesn't actually do a font change. To work
849 # around this, use a separate textmapfonts for text blocks where the default
850 # font is always R and only use the smart mapfonts for headings.
855 my ($fixed, $bold, $italic) = (0, 0, 0);
856 my %magic = (F => \$fixed, B => \$bold, I => \$italic);
858 ${ $magic{$1} } += ($2 eq 'S') ? 1 : -1;
859 $$self{FONTS}{($fixed && 1) . ($bold && 1) . ($italic && 1)};
865 ##############################################################################
866 # *roff-specific parsing and magic
867 ##############################################################################
869 # Called instead of parse_text, calls parse_text with the right flags.
872 $self->parse_text ({ -expand_seq => 'sequence',
873 -expand_ptree => 'collapse' }, @_);
876 # Takes a parse tree, a flag saying whether or not to treat it as literal text
877 # (not call guesswork on it), and a flag saying whether or not to clean some
878 # things up for *roff, and returns the concatenation of all of the text
879 # strings in that parse tree. If the literal flag isn't true, guesswork()
880 # will be called on all plain scalars in the parse tree. Otherwise, just
881 # escape backslashes in the normal case. If collapse is being called on a C<>
882 # code, $cleanup should be set to true and some additional cleanup will be
883 # done. Assumes that everything in the parse tree is either a scalar or a
884 # reference to a scalar.
886 my ($self, $ptree, $literal, $cleanup) = @_;
887 return join ('', map {
898 $self->guesswork ($_);
903 # Takes a text block to perform guesswork on; this is guaranteed not to
904 # contain any formatting codes. Returns the text block with remapping done.
909 # rofficate backslashes.
912 # Ensure double underbars have a tiny space between them.
915 # Leave hyphens only if they're part of regular words and there is only
916 # one dash at a time. Leave a dash after the first character as a regular
917 # non-breaking dash, but don't let it mark the rest of the word invalid
921 ( (?:\G|^|\s) [a-zA-Z] ) ( \\- )?
922 ( (?: [a-zA-Z]+ \\-)+ )
923 ( [a-zA-Z]+ ) (?=\s|\Z)
926 my ($prefix, $hyphen, $main, $suffix) = ($1, $2, $3, $4);
929 $prefix . $hyphen . $main . $suffix;
932 # Translate -- into a real em dash if it's used like one.
933 s{ (\s) \\-\\- (\s) } { $1 . '\*(--' . $2 }egx;
934 s{ (\b[a-zA-Z]+) \\-\\- (\s|\Z|[a-zA-Z]+\b) } { $1 . '\*(--' . $2 }egx;
936 # Make all caps a little smaller. Be careful here, since we don't want to
937 # make @ARGV into small caps, nor do we want to fix the MIME in
938 # MIME-Version, since it looks weird with the full-height V.
940 ( ^ | [\s\(\"\'\`\[\{<>] )
941 ( [A-Z] [A-Z] (?: [/A-Z+:\d_\$&] | \\- )* )
942 (?= [\s>\}\]\(\)\'\".?!,;] | \\*\(-- | $ )
943 } { $1 . '\s-1' . $2 . '\s0' }egx;
945 # Italize functions in the form func().
949 [A-Za-z_] ([:\w]|\\s-?[01])+ \(\)
951 } { $1 . '\f(IS' . $2 . '\f(IE' }egx;
953 # func(n) is a reference to a manual page. Make it \fIfunc\fR\|(n).
956 ( [A-Za-z_] (?:[.:\w]|\\-|\\s-?[01])+ )
960 } { $1 . '\f(IS' . $2 . '\f(IE\|' . $3 }egx;
962 # Convert simple Perl variable references to a fixed-width font.
967 } { $1 . '\f(FS' . $2 . '\f(FE'}egx;
969 # Fix up double quotes.
970 s{ \" ([^\"]+) \" } { '\*(L"' . $1 . '\*(R"' }egx;
972 # Make C++ into \*(C+, which is a squinched version.
973 s{ \b C\+\+ } {\\*\(C+}gx;
979 # Handles C<> text, deciding whether to put \*C` around it or not. This is a
980 # whole bunch of messy heuristics to try to avoid overquoting, originally from
981 # Barrie Slaymaker. This largely duplicates similar code in Pod::Text.
986 # A regex that matches the portion of a variable reference that's the
987 # array or hash index, separated out just because we want to use it in
988 # several places in the following regex.
989 my $index = '(?: \[.*\] | \{.*\} )?';
991 # Check for things that we don't want to quote, and if we find any of
992 # them, return the string with just a font change and no quoting.
996 ( [\'\`\"] ) .* \1 # already quoted
997 | \` .* \' # `quoted'
998 | \$+ [\#^]? \S $index # special ($^Foo, $")
999 | [\$\@%&*]+ \#? [:\'\w]+ $index # plain var or func
1000 | [\$\@%&*]* [:\'\w]+ (?: -> )? \(\s*[^\s,]\s*\) # 0/1-arg func call
1001 | [+-]? [\d.]+ (?: [eE] [+-]? \d+ )? # a number
1002 | 0x [a-fA-F\d]+ # a hex constant
1005 }xo && return '\f(FS' . $_ . '\f(FE';
1007 # If we didn't return, go ahead and quote the text.
1008 return '\f(FS\*(C`' . $_ . "\\*(C'\\f(FE";
1012 ##############################################################################
1014 ##############################################################################
1016 # Make vertical whitespace.
1019 $self->output (".PD\n") if ($$self{ITEMS} > 1);
1021 $self->output ($$self{INDENT} > 0 ? ".Sp\n" : ".PP\n")
1022 if $$self{NEEDSPACE};
1025 # Output any pending index entries, and optionally an index entry given as an
1026 # argument. Support multiple index entries in X<> separated by slashes, and
1027 # strip special escapes from index entries.
1029 my ($self, $section, $index) = @_;
1030 my @entries = map { split m%\s*/\s*% } @{ $$self{INDEX} };
1031 return unless ($section || @entries);
1035 push (@output, [ 'Xref', join (' ', @entries) ]);
1038 $index =~ s/\\-/-/g;
1039 $index =~ s/\\(?:s-?\d|.\(..|.)//g;
1040 push (@output, [ $section, $index ]);
1043 my ($type, $entry) = @$_;
1044 $entry =~ s/\"/\"\"/g;
1045 $self->output (".IX $type " . '"' . $entry . '"' . "\n");
1049 # Output text to the output device.
1050 sub output { print { $_[0]->output_handle } $_[1] }
1052 # Given a command and a single argument that may or may not contain double
1053 # quotes, handle double-quote formatting for it. If there are no double
1054 # quotes, just return the command followed by the argument in double quotes.
1055 # If there are double quotes, use an if statement to test for nroff, and for
1056 # nroff output the command followed by the argument in double quotes with
1057 # embedded double quotes doubled. For other formatters, remap paired double
1058 # quotes to LQUOTE and RQUOTE.
1061 my $command = shift;
1064 s/\\\*\([LR]\"/\"/g;
1066 # We also have to deal with \*C` and \*C', which are used to add the
1067 # quotes around C<> text, since they may expand to " and if they do this
1068 # confuses the .SH macros and the like no end. Expand them ourselves.
1069 # Also separate troff from nroff if there are any fixed-width fonts in use
1070 # to work around problems with Solaris nroff.
1071 my $c_is_quote = ($$self{LQUOTE} =~ /\"/) || ($$self{RQUOTE} =~ /\"/);
1072 my $fixedpat = join ('|', @{ $$self{FONTS} }{'100', '101', '110', '111'});
1073 $fixedpat =~ s/\\/\\\\/g;
1074 $fixedpat =~ s/\(/\\\(/g;
1075 if (/\"/ || /$fixedpat/) {
1079 $troff =~ s/\"\"([^\"]*)\"\"/\`\`$1\'\'/g;
1080 if ($c_is_quote && /\\\*\(C[\'\`]/) {
1081 $nroff =~ s/\\\*\(C\`/$$self{LQUOTE}/g;
1082 $nroff =~ s/\\\*\(C\'/$$self{RQUOTE}/g;
1083 $troff =~ s/\\\*\(C[\'\`]//g;
1085 $nroff = qq("$nroff") . ($extra ? " $extra" : '');
1086 $troff = qq("$troff") . ($extra ? " $extra" : '');
1088 # Work around the Solaris nroff bug where \f(CW\fP leaves the font set
1089 # to Roman rather than the actual previous font when used in headings.
1090 # troff output may still be broken, but at least we can fix nroff by
1091 # just switching the font changes to the non-fixed versions.
1092 $nroff =~ s/\Q$$self{FONTS}{100}\E(.*)\\f[PR]/$1/g;
1093 $nroff =~ s/\Q$$self{FONTS}{101}\E(.*)\\f([PR])/\\fI$1\\f$2/g;
1094 $nroff =~ s/\Q$$self{FONTS}{110}\E(.*)\\f([PR])/\\fB$1\\f$2/g;
1095 $nroff =~ s/\Q$$self{FONTS}{111}\E(.*)\\f([PR])/\\f\(BI$1\\f$2/g;
1097 # Now finally output the command. Only bother with .ie if the nroff
1098 # and troff output isn't the same.
1099 if ($nroff ne $troff) {
1100 return ".ie n $command $nroff\n.el $command $troff\n";
1102 return "$command $nroff\n";
1105 $_ = qq("$_") . ($extra ? " $extra" : '');
1106 return "$command $_\n";
1112 ##############################################################################
1114 ##############################################################################
1118 Pod::Man - Convert POD data to formatted *roff input
1123 my $parser = Pod::Man->new (release => $VERSION, section => 8);
1125 # Read POD from STDIN and write to STDOUT.
1126 $parser->parse_from_filehandle;
1128 # Read POD from file.pod and write to file.1.
1129 $parser->parse_from_file ('file.pod', 'file.1');
1133 Pod::Man is a module to convert documentation in the POD format (the
1134 preferred language for documenting Perl) into *roff input using the man
1135 macro set. The resulting *roff code is suitable for display on a terminal
1136 using L<nroff(1)>, normally via L<man(1)>, or printing using L<troff(1)>.
1137 It is conventionally invoked using the driver script B<pod2man>, but it can
1138 also be used directly.
1140 As a derived class from Pod::Parser, Pod::Man supports the same methods and
1141 interfaces. See L<Pod::Parser> for all the details; briefly, one creates a
1142 new parser with C<< Pod::Man->new() >> and then calls either
1143 parse_from_filehandle() or parse_from_file().
1145 new() can take options, in the form of key/value pairs that control the
1146 behavior of the parser. See below for details.
1148 If no options are given, Pod::Man uses the name of the input file with any
1149 trailing C<.pod>, C<.pm>, or C<.pl> stripped as the man page title, to
1150 section 1 unless the file ended in C<.pm> in which case it defaults to
1151 section 3, to a centered title of "User Contributed Perl Documentation", to
1152 a centered footer of the Perl version it is run with, and to a left-hand
1153 footer of the modification date of its input (or the current date if given
1156 Pod::Man assumes that your *roff formatters have a fixed-width font named
1157 CW. If yours is called something else (like CR), use the C<fixed> option to
1158 specify it. This generally only matters for troff output for printing.
1159 Similarly, you can set the fonts used for bold, italic, and bold italic
1162 Besides the obvious pod conversions, Pod::Man also takes care of formatting
1163 func(), func(3), and simple variable references like $foo or @bar so you
1164 don't have to use code escapes for them; complex expressions like
1165 C<$fred{'stuff'}> will still need to be escaped, though. It also translates
1166 dashes that aren't used as hyphens into en dashes, makes long dashes--like
1167 this--into proper em dashes, fixes "paired quotes," makes C++ look right,
1168 puts a little space between double underbars, makes ALLCAPS a teeny bit
1169 smaller in B<troff>, and escapes stuff that *roff treats as special so that
1172 The recognized options to new() are as follows. All options take a single
1179 Sets the centered page header to use instead of "User Contributed Perl
1184 Sets the left-hand footer. By default, the modification date of the input
1185 file will be used, or the current date if stat() can't find that file (the
1186 case if the input is from STDIN), and the date will be formatted as
1191 The fixed-width font to use for vertabim text and code. Defaults to CW.
1192 Some systems may want CR instead. Only matters for B<troff> output.
1196 Bold version of the fixed-width font. Defaults to CB. Only matters for
1201 Italic version of the fixed-width font (actually, something of a misnomer,
1202 since most fixed-width fonts only have an oblique version, not an italic
1203 version). Defaults to CI. Only matters for B<troff> output.
1205 =item fixedbolditalic
1207 Bold italic (probably actually oblique) version of the fixed-width font.
1208 Pod::Man doesn't assume you have this, and defaults to CB. Some systems
1209 (such as Solaris) have this font available as CX. Only matters for B<troff>
1214 Set the name of the manual page. Without this option, the manual name is
1215 set to the uppercased base name of the file being converted unless the
1216 manual section is 3, in which case the path is parsed to see if it is a Perl
1217 module path. If it is, a path like C<.../lib/Pod/Man.pm> is converted into
1218 a name like C<Pod::Man>. This option, if given, overrides any automatic
1219 determination of the name.
1223 Sets the quote marks used to surround CE<lt>> text. If the value is a
1224 single character, it is used as both the left and right quote; if it is two
1225 characters, the first character is used as the left quote and the second as
1226 the right quoted; and if it is four characters, the first two are used as
1227 the left quote and the second two as the right quote.
1229 This may also be set to the special value C<none>, in which case no quote
1230 marks are added around CE<lt>> text (but the font is still changed for troff
1235 Set the centered footer. By default, this is the version of Perl you run
1236 Pod::Man under. Note that some system an macro sets assume that the
1237 centered footer will be a modification date and will prepend something like
1238 "Last modified: "; if this is the case, you may want to set C<release> to
1239 the last modified date and C<date> to the version number.
1243 Set the section for the C<.TH> macro. The standard section numbering
1244 convention is to use 1 for user commands, 2 for system calls, 3 for
1245 functions, 4 for devices, 5 for file formats, 6 for games, 7 for
1246 miscellaneous information, and 8 for administrator commands. There is a lot
1247 of variation here, however; some systems (like Solaris) use 4 for file
1248 formats, 5 for miscellaneous information, and 7 for devices. Still others
1249 use 1m instead of 8, or some mix of both. About the only section numbers
1250 that are reliably consistent are 1, 2, and 3.
1252 By default, section 1 will be used unless the file ends in .pm in which case
1253 section 3 will be selected.
1257 The standard Pod::Parser method parse_from_filehandle() takes up to two
1258 arguments, the first being the file handle to read POD from and the second
1259 being the file handle to write the formatted output to. The first defaults
1260 to STDIN if not given, and the second defaults to STDOUT. The method
1261 parse_from_file() is almost identical, except that its two arguments are the
1262 input and output disk files instead. See L<Pod::Parser> for the specific
1269 =item roff font should be 1 or 2 chars, not "%s"
1271 (F) You specified a *roff font (using C<fixed>, C<fixedbold>, etc.) that
1272 wasn't either one or two characters. Pod::Man doesn't support *roff fonts
1273 longer than two characters, although some *roff extensions do (the canonical
1274 versions of B<nroff> and B<troff> don't either).
1276 =item Invalid link %s
1278 (W) The POD source contained a C<LE<lt>E<gt>> formatting code that
1279 Pod::Man was unable to parse. You should never see this error message; it
1280 probably indicates a bug in Pod::Man.
1282 =item Invalid quote specification "%s"
1284 (F) The quote specification given (the quotes option to the constructor) was
1285 invalid. A quote specification must be one, two, or four characters long.
1287 =item %s:%d: Unknown command paragraph "%s".
1289 (W) The POD source contained a non-standard command paragraph (something of
1290 the form C<=command args>) that Pod::Man didn't know about. It was ignored.
1292 =item %s:%d: Unknown escape EE<lt>%sE<gt>
1294 (W) The POD source contained an C<EE<lt>E<gt>> escape that Pod::Man didn't
1295 know about. C<EE<lt>%sE<gt>> was printed verbatim in the output.
1297 =item %s:%d: Unknown formatting code %s
1299 (W) The POD source contained a non-standard formatting code (something of
1300 the form C<XE<lt>E<gt>>) that Pod::Man didn't know about. It was ignored.
1302 =item %s:%d: Unmatched =back
1304 (W) Pod::Man encountered a C<=back> command that didn't correspond to an
1311 Eight-bit input data isn't handled at all well at present. The correct
1312 approach would be to map EE<lt>E<gt> escapes to the appropriate UTF-8
1313 characters and then do a translation pass on the output according to the
1314 user-specified output character set. Unfortunately, we can't send eight-bit
1315 data directly to the output unless the user says this is okay, since some
1316 vendor *roff implementations can't handle eight-bit data. If the *roff
1317 implementation can, however, that's far superior to the current hacked
1318 characters that only work under troff.
1320 There is currently no way to turn off the guesswork that tries to format
1321 unmarked text appropriately, and sometimes it isn't wanted (particularly
1322 when using POD to document something other than Perl).
1324 The NAME section should be recognized specially and index entries emitted
1325 for everything in that section. This would have to be deferred until the
1326 next section, since extraneous things in NAME tends to confuse various man
1329 Pod::Man doesn't handle font names longer than two characters. Neither do
1330 most B<troff> implementations, but GNU troff does as an extension. It would
1331 be nice to support as an option for those who want to use it.
1333 The preamble added to each output file is rather verbose, and most of it is
1334 only necessary in the presence of EE<lt>E<gt> escapes for non-ASCII
1335 characters. It would ideally be nice if all of those definitions were only
1336 output if needed, perhaps on the fly as the characters are used.
1338 Pod::Man is excessively slow.
1342 The handling of hyphens and em dashes is somewhat fragile, and one may get
1343 the wrong one under some circumstances. This should only matter for
1346 When and whether to use small caps is somewhat tricky, and Pod::Man doesn't
1347 necessarily get it right.
1351 L<Pod::Parser>, L<perlpod(1)>, L<pod2man(1)>, L<nroff(1)>, L<troff(1)>,
1352 L<man(1)>, L<man(7)>
1354 Ossanna, Joseph F., and Brian W. Kernighan. "Troff User's Manual,"
1355 Computing Science Technical Report No. 54, AT&T Bell Laboratories. This is
1356 the best documentation of standard B<nroff> and B<troff>. At the time of
1357 this writing, it's available at
1358 L<http://www.cs.bell-labs.com/cm/cs/cstr.html>.
1360 The man page documenting the man macro set may be L<man(5)> instead of
1361 L<man(7)> on your system. Also, please see L<pod2man(1)> for extensive
1362 documentation on writing manual pages if you've not done it before and
1363 aren't familiar with the conventions.
1367 Russ Allbery <rra@stanford.edu>, based I<very> heavily on the original
1368 B<pod2man> by Tom Christiansen <tchrist@mox.perl.com>.
1370 =head1 COPYRIGHT AND LICENSE
1372 Copyright 1999, 2000, 2001 by Russ Allbery <rra@stanford.edu>.
1374 This program is free software; you may redistribute it and/or modify it
1375 under the same terms as Perl itself.