1 # Pod::Man -- Convert POD data to formatted *roff input.
2 # $Id: Man.pm,v 2.4 2005/03/19 19:40:01 eagle Exp $
4 # Copyright 1999, 2000, 2001, 2002, 2003, 2004, 2005
5 # Russ Allbery <rra@stanford.edu>
6 # Substantial contributions by Sean Burke <sburke@cpan.org>
8 # This program is free software; you may redistribute it and/or modify it
9 # under the same terms as Perl itself.
11 # This module translates POD documentation into *roff markup using the man
12 # macro set, and is intended for converting POD documents written as Unix
13 # manual pages to manual pages that can be read by the man(1) command. It is
14 # a replacement for the pod2man command distributed with versions of Perl
17 # Perl core hackers, please note that this module is also separately
18 # maintained outside of the Perl core as part of the podlators. Please send
19 # me any patches at the address above in addition to sending them to the
20 # standard Perl mailing lists.
22 ##############################################################################
23 # Modules and declarations
24 ##############################################################################
31 use subs qw(makespace);
32 use vars qw(@ISA %ESCAPES $PREAMBLE $VERSION);
36 use POSIX qw(strftime);
38 @ISA = qw(Pod::Simple);
40 # Don't use the CVS revision as the version, since this module is also in Perl
41 # core and too many things could munge CVS magic revision strings. This
42 # number should ideally be the same as the CVS revision in podlators, however.
45 # Set the debugging level. If someone has inserted a debug function into this
46 # class already, use that. Otherwise, use any Pod::Simple debug function
47 # that's defined, and failing that, define a debug level of 10.
49 my $parent = defined (&Pod::Simple::DEBUG) ? \&Pod::Simple::DEBUG : undef;
50 unless (defined &DEBUG) {
51 *DEBUG = $parent || sub () { 10 };
55 # Import the ASCII constant from Pod::Simple. This is true iff we're in an
56 # ASCII-based universe (including such things as ISO 8859-1 and UTF-8), and is
57 # generally only false for EBCDIC.
58 BEGIN { *ASCII = \&Pod::Simple::ASCII }
60 # Pretty-print a data structure. Only used for debugging.
61 BEGIN { *pretty = \&Pod::Simple::pretty }
63 ##############################################################################
64 # Object initialization
65 ##############################################################################
67 # Initialize the object and set various Pod::Simple options that we need.
68 # Here, we also process any additional options passed to the constructor or
69 # set up defaults if none were given. Note that all internal object keys are
70 # in all-caps, reserving all lower-case object keys for Pod::Simple and user
74 my $self = $class->SUPER::new;
76 # Tell Pod::Simple to handle S<> by automatically inserting .
77 $self->nbsp_for_S (1);
79 # Tell Pod::Simple to keep whitespace whenever possible.
80 if ($self->can ('preserve_whitespace')) {
81 $self->preserve_whitespace (1);
83 $self->fullstop_space_harden (1);
86 # The =for and =begin targets that we accept.
87 $self->accept_targets (qw/man MAN roff ROFF/);
89 # Ensure that contiguous blocks of code are merged together. Otherwise,
90 # some of the guesswork heuristics don't work right.
91 $self->merge_text (1);
93 # Pod::Simple doesn't do anything useful with our arguments, but we want
94 # to put them in our object as hash keys and values. This could cause
95 # problems if we ever clash with Pod::Simple's own internal class
97 %$self = (%$self, @_);
99 # Initialize various other internal constants based on our arguments.
104 # For right now, default to turning on all of the magic.
105 $$self{MAGIC_CPP} = 1;
106 $$self{MAGIC_EMDASH} = 1;
107 $$self{MAGIC_FUNC} = 1;
108 $$self{MAGIC_MANREF} = 1;
109 $$self{MAGIC_SMALLCAPS} = 1;
110 $$self{MAGIC_VARS} = 1;
115 # Translate a font string into an escape.
116 sub toescape { (length ($_[0]) > 1 ? '\f(' : '\f') . $_[0] }
118 # Determine which fonts the user wishes to use and store them in the object.
119 # Regular, italic, bold, and bold-italic are constants, but the fixed width
120 # fonts may be set by the user. Sets the internal hash key FONTS which is
121 # used to map our internal font escapes to actual *roff sequences later.
125 # Figure out the fixed-width font. If user-supplied, make sure that they
126 # are the right length.
127 for (qw/fixed fixedbold fixeditalic fixedbolditalic/) {
128 my $font = $$self{$_};
129 if (defined ($font) && (length ($font) < 1 || length ($font) > 2)) {
130 croak qq(roff font should be 1 or 2 chars, not "$font");
134 # Set the default fonts. We can't be sure portably across different
135 # implementations what fixed bold-italic may be called (if it's even
136 # available), so default to just bold.
137 $$self{fixed} ||= 'CW';
138 $$self{fixedbold} ||= 'CB';
139 $$self{fixeditalic} ||= 'CI';
140 $$self{fixedbolditalic} ||= 'CB';
142 # Set up a table of font escapes. First number is fixed-width, second is
143 # bold, third is italic.
144 $$self{FONTS} = { '000' => '\fR', '001' => '\fI',
145 '010' => '\fB', '011' => '\f(BI',
146 '100' => toescape ($$self{fixed}),
147 '101' => toescape ($$self{fixeditalic}),
148 '110' => toescape ($$self{fixedbold}),
149 '111' => toescape ($$self{fixedbolditalic}) };
152 # Initialize the quotes that we'll be using for C<> text. This requires some
153 # special handling, both to parse the user parameter if given and to make sure
154 # that the quotes will be safe against *roff. Sets the internal hash keys
159 $$self{quotes} ||= '"';
160 if ($$self{quotes} eq 'none') {
161 $$self{LQUOTE} = $$self{RQUOTE} = '';
162 } elsif (length ($$self{quotes}) == 1) {
163 $$self{LQUOTE} = $$self{RQUOTE} = $$self{quotes};
164 } elsif ($$self{quotes} =~ /^(.)(.)$/
165 || $$self{quotes} =~ /^(..)(..)$/) {
169 croak(qq(Invalid quote specification "$$self{quotes}"))
172 # Double the first quote; note that this should not be s///g as two double
173 # quotes is represented in *roff as three double quotes, not four. Weird,
175 $$self{LQUOTE} =~ s/\"/\"\"/;
176 $$self{RQUOTE} =~ s/\"/\"\"/;
179 # Initialize the page title information and indentation from our arguments.
183 # We used to try first to get the version number from a local binary, but
184 # we shouldn't need that any more. Get the version from the running Perl.
185 # Work a little magic to handle subversions correctly under both the
186 # pre-5.6 and the post-5.6 version numbering schemes.
187 my @version = ($] =~ /^(\d+)\.(\d{3})(\d{0,3})$/);
189 $version[2] *= 10 ** (3 - length $version[2]);
190 for (@version) { $_ += 0 }
191 my $version = join ('.', @version);
193 # Set the defaults for page titles and indentation if the user didn't
195 $$self{center} = 'User Contributed Perl Documentation'
196 unless defined $$self{center};
197 $$self{release} = 'perl v' . $version
198 unless defined $$self{release};
200 unless defined $$self{indent};
202 # Double quotes in things that will be quoted.
203 for (qw/center release/) {
204 $$self{$_} =~ s/\"/\"\"/g if $$self{$_};
208 ##############################################################################
210 ##############################################################################
212 # This is the glue that connects the code below with Pod::Simple itself. The
213 # goal is to convert the event stream coming from the POD parser into method
214 # calls to handlers once the complete content of a tag has been seen. Each
215 # paragraph or POD command will have textual content associated with it, and
216 # as soon as all of a paragraph or POD command has been seen, that content
217 # will be passed in to the corresponding method for handling that type of
218 # object. The exceptions are handlers for lists, which have opening tag
219 # handlers and closing tag handlers that will be called right away.
221 # The internal hash key PENDING is used to store the contents of a tag until
222 # all of it has been seen. It holds a stack of open tags, each one
223 # represented by a tuple of the attributes hash for the tag, formatting
224 # options for the tag (which are inherited), and the contents of the tag.
226 # Add a block of text to the contents of the current node, formatting it
227 # according to the current formatting instructions as we do.
229 my ($self, $text) = @_;
230 DEBUG > 3 and print "== $text\n";
231 my $tag = $$self{PENDING}[-1];
232 $$tag[2] .= $self->format_text ($$tag[1], $text);
235 # Given an element name, get the corresponding method name.
236 sub method_for_element {
237 my ($self, $element) = @_;
239 $element =~ tr/A-Z/a-z/;
240 $element =~ tr/_a-z0-9//cd;
244 # Handle the start of a new element. If cmd_element is defined, assume that
245 # we need to collect the entire tree for this element before passing it to the
246 # element method, and create a new tree into which we'll collect blocks of
247 # text and nested elements. Otherwise, if start_element is defined, call it.
248 sub _handle_element_start {
249 my ($self, $element, $attrs) = @_;
250 DEBUG > 3 and print "++ $element (<", join ('> <', %$attrs), ">)\n";
251 my $method = $self->method_for_element ($element);
253 # If we have a command handler, we need to accumulate the contents of the
254 # tag before calling it. Turn off IN_NAME for any command other than
255 # <Para> so that IN_NAME isn't still set for the first heading after the
257 if ($self->can ("cmd_$method")) {
258 DEBUG > 2 and print "<$element> starts saving a tag\n";
259 $$self{IN_NAME} = 0 if ($element ne 'Para');
261 # How we're going to format embedded text blocks depends on the tag
262 # and also depends on our parent tags. Thankfully, inside tags that
263 # turn off guesswork and reformatting, nothing else can turn it back
264 # on, so this can be strictly inherited.
265 my $formatting = $$self{PENDING}[-1][1];
266 $formatting = $self->formatting ($formatting, $element);
267 push (@{ $$self{PENDING} }, [ $attrs, $formatting, '' ]);
268 DEBUG > 4 and print "Pending: [", pretty ($$self{PENDING}), "]\n";
269 } elsif ($self->can ("start_$method")) {
270 my $method = 'start_' . $method;
271 $self->$method ($attrs, '');
273 DEBUG > 2 and print "No $method start method, skipping\n";
277 # Handle the end of an element. If we had a cmd_ method for this element,
278 # this is where we pass along the tree that we built. Otherwise, if we have
279 # an end_ method for the element, call that.
280 sub _handle_element_end {
281 my ($self, $element) = @_;
282 DEBUG > 3 and print "-- $element\n";
283 my $method = $self->method_for_element ($element);
285 # If we have a command handler, pull off the pending text and pass it to
286 # the handler along with the saved attribute hash.
287 if ($self->can ("cmd_$method")) {
288 DEBUG > 2 and print "</$element> stops saving a tag\n";
289 my $tag = pop @{ $$self{PENDING} };
290 DEBUG > 4 and print "Popped: [", pretty ($tag), "]\n";
291 DEBUG > 4 and print "Pending: [", pretty ($$self{PENDING}), "]\n";
292 my $method = 'cmd_' . $method;
293 my $text = $self->$method ($$tag[0], $$tag[2]);
295 if (@{ $$self{PENDING} } > 1) {
296 $$self{PENDING}[-1][2] .= $text;
298 $self->output ($text);
301 } elsif ($self->can ("end_$method")) {
302 my $method = 'end_' . $method;
305 DEBUG > 2 and print "No $method end method, skipping\n";
309 ##############################################################################
311 ##############################################################################
313 # Return formatting instructions for a new block. Takes the current
314 # formatting and the new element. Formatting inherits negatively, in the
315 # sense that if the parent has turned off guesswork, all child elements should
316 # leave it off. We therefore return a copy of the same formatting
317 # instructions but possibly with more things turned off depending on the
320 my ($self, $current, $element) = @_;
323 %options = %$current;
325 %options = (guesswork => 1, cleanup => 1, convert => 1);
327 if ($element eq 'Data') {
328 $options{guesswork} = 0;
329 $options{cleanup} = 0;
330 $options{convert} = 0;
331 } elsif ($element eq 'X') {
332 $options{guesswork} = 0;
333 $options{cleanup} = 0;
334 } elsif ($element eq 'Verbatim' || $element eq 'C') {
335 $options{guesswork} = 0;
340 # Format a text block. Takes a hash of formatting options and the text to
341 # format. Currently, the only formatting options are guesswork, cleanup, and
342 # convert, all of which are boolean.
344 my ($self, $options, $text) = @_;
345 my $guesswork = $$options{guesswork} && !$$self{IN_NAME};
346 my $cleanup = $$options{cleanup};
347 my $convert = $$options{convert};
349 # Normally we do character translation, but we won't even do that in
353 $text =~ s/(\\|[^\x00-\x7F])/$ESCAPES{ord ($1)} || "X"/eg;
355 $text =~ s/(\\)/$ESCAPES{ord ($1)} || "X"/eg;
359 # Cleanup just tidies up a few things, telling *roff that the hyphens are
360 # hard and putting a bit of space between consecutive underscores.
363 $text =~ s/_(?=_)/_\\|/g;
366 # If guesswork is asked for, do that. This involves more substantial
367 # formatting based on various heuristics that may only be appropriate for
368 # particular documents.
370 $text = $self->guesswork ($text);
376 # Handles C<> text, deciding whether to put \*C` around it or not. This is a
377 # whole bunch of messy heuristics to try to avoid overquoting, originally from
378 # Barrie Slaymaker. This largely duplicates similar code in Pod::Text.
383 # A regex that matches the portion of a variable reference that's the
384 # array or hash index, separated out just because we want to use it in
385 # several places in the following regex.
386 my $index = '(?: \[.*\] | \{.*\} )?';
388 # Check for things that we don't want to quote, and if we find any of
389 # them, return the string with just a font change and no quoting.
393 ( [\'\`\"] ) .* \1 # already quoted
394 | \` .* \' # `quoted'
395 | \$+ [\#^]? \S $index # special ($^Foo, $")
396 | [\$\@%&*]+ \#? [:\'\w]+ $index # plain var or func
397 | [\$\@%&*]* [:\'\w]+ (?: -> )? \(\s*[^\s,]\s*\) # 0/1-arg func call
398 | [-+]? ( \d[\d.]* | \.\d+ ) (?: [eE][-+]?\d+ )? # a number
399 | 0x [a-fA-F\d]+ # a hex constant
402 }xso and return '\f(FS' . $_ . '\f(FE';
404 # If we didn't return, go ahead and quote the text.
405 return '\f(FS\*(C`' . $_ . "\\*(C'\\f(FE";
408 # Takes a text block to perform guesswork on. Returns the text block with
409 # formatting codes added. This is the code that marks up various Perl
410 # constructs and things commonly used in man pages without requiring the user
411 # to add any explicit markup, and is applied to all non-literal text. We're
412 # guaranteed that the text we're applying guesswork to does not contain any
413 # *roff formatting codes. Note that the inserted font sequences must be
414 # treated later with mapfonts or textmapfonts.
416 # This method is very fragile, both in the regular expressions it uses and in
417 # the ordering of those modifications. Care and testing is required when
422 DEBUG > 5 and print " Guesswork called on [$_]\n";
424 # By the time we reach this point, all hypens will be escaped by adding a
425 # backslash. We want to do that escaping if they're part of regular words
426 # and there's only a single dash, since that's a real hyphen that *roff
427 # gets to consider a possible break point. Make sure that a dash after
428 # the first character of a word stays non-breaking, however.
430 # Note that this is not user-controllable; we pretty much have to do this
431 # transformation or *roff will mangle the output in unacceptable ways.
433 ( (?:\G|^|\s) [a-zA-Z] ) ( \\- )?
434 ( (?: [a-zA-Z]+ \\-)+ )
435 ( [a-zA-Z]+ ) (?=\s|\Z|\\\ )
438 my ($prefix, $hyphen, $main, $suffix) = ($1, $2, $3, $4);
441 $prefix . $hyphen . $main . $suffix;
444 # Translate "--" into a real em-dash if it's used like one. This means
445 # that it's either surrounded by whitespace, it follows a regular word, or
446 # it occurs between two regular words.
447 if ($$self{MAGIC_EMDASH}) {
448 s{ (\s) \\-\\- (\s) } { $1 . '\*(--' . $2 }egx;
449 s{ (\b[a-zA-Z]+) \\-\\- (\s|\Z|[a-zA-Z]+\b) } { $1 . '\*(--' . $2 }egx;
452 # Make words in all-caps a little bit smaller; they look better that way.
453 # However, we don't want to change Perl code (like @ARGV), nor do we want
454 # to fix the MIME in MIME-Version since it looks weird with the
457 # We change only a string of all caps (2) either at the beginning of the
458 # line or following regular punctuation (like quotes) or whitespace (1),
459 # and followed by either similar punctuation, an em-dash, or the end of
461 if ($$self{MAGIC_SMALLCAPS}) {
463 ( ^ | [\s\(\"\'\`\[\{<>] | \\\ ) # (1)
464 ( [A-Z] [A-Z] (?: [/A-Z+:\d_\$&] | \\- )* ) # (2)
465 (?= [\s>\}\]\(\)\'\".?!,;] | \\*\(-- | \\\ | $ ) # (3)
467 $1 . '\s-1' . $2 . '\s0'
471 # Note that from this point forward, we have to adjust for \s-1 and \s-0
472 # strings inserted around things that we've made small-caps if later
473 # transforms should work on those strings.
475 # Italize functions in the form func(), including functions that are in
476 # all capitals, but don't italize if there's anything between the parens.
477 # The function must start with an alphabetic character or underscore and
478 # then consist of word characters or colons.
479 if ($$self{MAGIC_FUNC}) {
482 ( [A-Za-z_] ([:\w] | \\s-?[01])+ \(\) )
484 $1 . '\f(IS' . $2 . '\f(IE'
488 # Change references to manual pages to put the page name in italics but
489 # the number in the regular font, with a thin space between the name and
490 # the number. Only recognize func(n) where func starts with an alphabetic
491 # character or underscore and contains only word characters, periods (for
492 # configuration file man pages), or colons, and n is a single digit,
493 # optionally followed by some number of lowercase letters. Note that this
494 # does not recognize man page references like perl(l) or socket(3SOCKET).
495 if ($$self{MAGIC_MANREF}) {
498 ( [A-Za-z_] (?:[.:\w] | \\- | \\s-?[01])+ )
501 $1 . '\f(IS' . $2 . '\f(IE\|' . $3
505 # Convert simple Perl variable references to a fixed-width font. Be
506 # careful not to convert functions, though; there are too many subtleties
507 # with them to want to perform this transformation.
508 if ($$self{MAGIC_VARS}) {
514 $1 . '\f(FS' . $2 . '\f(FE'
518 # Fix up double quotes. Unfortunately, we miss this transformation if the
519 # quoted text contains any code with formatting codes and there's not much
520 # we can effectively do about that, which makes it somewhat unclear if
521 # this is really a good idea.
522 s{ \" ([^\"]+) \" } { '\*(L"' . $1 . '\*(R"' }egx;
524 # Make C++ into \*(C+, which is a squinched version.
525 if ($$self{MAGIC_CPP}) {
526 s{ \b C\+\+ } {\\*\(C+}gx;
530 DEBUG > 5 and print " Guesswork returning [$_]\n";
534 ##############################################################################
536 ##############################################################################
538 # When building up the *roff code, we don't use real *roff fonts. Instead, we
539 # embed font codes of the form \f(<font>[SE] where <font> is one of B, I, or
540 # F, S stands for start, and E stands for end. This method turns these into
541 # the right start and end codes.
543 # We add this level of complexity because the old pod2man didn't get code like
544 # B<someI<thing> else> right; after I<> it switched back to normal text rather
545 # than bold. We take care of this by using variables that state whether bold,
546 # italic, or fixed are turned on as a combined pointer to our current font
547 # sequence, and set each to the number of current nestings of start tags for
550 # \fP changes to the previous font, but only one previous font is kept. We
551 # don't know what the outside level font is; normally it's R, but if we're
552 # inside a heading it could be something else. So arrange things so that the
553 # outside font is always the "previous" font and end with \fP instead of \fR.
554 # Idea from Zack Weinberg.
556 my ($self, $text) = @_;
557 my ($fixed, $bold, $italic) = (0, 0, 0);
558 my %magic = (F => \$fixed, B => \$bold, I => \$italic);
565 if ($last ne '\fR') { $sequence = '\fP' }
566 ${ $magic{$1} } += ($2 eq 'S') ? 1 : -1;
567 $f = $$self{FONTS}{ ($fixed && 1) . ($bold && 1) . ($italic && 1) };
571 if ($f ne '\fR') { $sequence .= $f }
579 # Unfortunately, there is a bug in Solaris 2.6 nroff (not present in GNU
580 # groff) where the sequence \fB\fP\f(CW\fP leaves the font set to B rather
581 # than R, presumably because \f(CW doesn't actually do a font change. To work
582 # around this, use a separate textmapfonts for text blocks where the default
583 # font is always R and only use the smart mapfonts for headings.
585 my ($self, $text) = @_;
586 my ($fixed, $bold, $italic) = (0, 0, 0);
587 my %magic = (F => \$fixed, B => \$bold, I => \$italic);
591 ${ $magic{$1} } += ($2 eq 'S') ? 1 : -1;
592 $$self{FONTS}{ ($fixed && 1) . ($bold && 1) . ($italic && 1) };
597 # Given a command and a single argument that may or may not contain double
598 # quotes, handle double-quote formatting for it. If there are no double
599 # quotes, just return the command followed by the argument in double quotes.
600 # If there are double quotes, use an if statement to test for nroff, and for
601 # nroff output the command followed by the argument in double quotes with
602 # embedded double quotes doubled. For other formatters, remap paired double
603 # quotes to LQUOTE and RQUOTE.
605 my ($self, $command, $text, $extra) = @_;
606 $text =~ s/\\\*\([LR]\"/\"/g;
608 # We also have to deal with \*C` and \*C', which are used to add the
609 # quotes around C<> text, since they may expand to " and if they do this
610 # confuses the .SH macros and the like no end. Expand them ourselves.
611 # Also separate troff from nroff if there are any fixed-width fonts in use
612 # to work around problems with Solaris nroff.
613 my $c_is_quote = ($$self{LQUOTE} =~ /\"/) || ($$self{RQUOTE} =~ /\"/);
614 my $fixedpat = join '|', @{ $$self{FONTS} }{'100', '101', '110', '111'};
615 $fixedpat =~ s/\\/\\\\/g;
616 $fixedpat =~ s/\(/\\\(/g;
617 if ($text =~ m/\"/ || $text =~ m/$fixedpat/) {
618 $text =~ s/\"/\"\"/g;
621 $troff =~ s/\"\"([^\"]*)\"\"/\`\`$1\'\'/g;
622 if ($c_is_quote and $text =~ m/\\\*\(C[\'\`]/) {
623 $nroff =~ s/\\\*\(C\`/$$self{LQUOTE}/g;
624 $nroff =~ s/\\\*\(C\'/$$self{RQUOTE}/g;
625 $troff =~ s/\\\*\(C[\'\`]//g;
627 $nroff = qq("$nroff") . ($extra ? " $extra" : '');
628 $troff = qq("$troff") . ($extra ? " $extra" : '');
630 # Work around the Solaris nroff bug where \f(CW\fP leaves the font set
631 # to Roman rather than the actual previous font when used in headings.
632 # troff output may still be broken, but at least we can fix nroff by
633 # just switching the font changes to the non-fixed versions.
634 $nroff =~ s/\Q$$self{FONTS}{100}\E(.*)\\f[PR]/$1/g;
635 $nroff =~ s/\Q$$self{FONTS}{101}\E(.*)\\f([PR])/\\fI$1\\f$2/g;
636 $nroff =~ s/\Q$$self{FONTS}{110}\E(.*)\\f([PR])/\\fB$1\\f$2/g;
637 $nroff =~ s/\Q$$self{FONTS}{111}\E(.*)\\f([PR])/\\f\(BI$1\\f$2/g;
639 # Now finally output the command. Bother with .ie only if the nroff
640 # and troff output aren't the same.
641 if ($nroff ne $troff) {
642 return ".ie n $command $nroff\n.el $command $troff\n";
644 return "$command $nroff\n";
647 $text = qq("$text") . ($extra ? " $extra" : '');
648 return "$command $text\n";
652 # Protect leading quotes and periods against interpretation as commands. Also
653 # protect anything starting with a backslash, since it could expand or hide
654 # something that *roff would interpret as a command. This is overkill, but
655 # it's much simpler than trying to parse *roff here.
657 my ($self, $text) = @_;
658 $text =~ s/^([.\'\\])/\\&$1/mg;
662 # Make vertical whitespace if NEEDSPACE is set, appropriate to the indentation
663 # level the situation. This function is needed since in *roff one has to
664 # create vertical whitespace after paragraphs and between some things, but
665 # other macros create their own whitespace. Also close out a sequence of
666 # repeated =items, since calling makespace means we're about to begin the item
670 $self->output (".PD\n") if $$self{ITEMS} > 1;
672 $self->output ($$self{INDENT} > 0 ? ".Sp\n" : ".PP\n")
673 if $$self{NEEDSPACE};
676 # Output any pending index entries, and optionally an index entry given as an
677 # argument. Support multiple index entries in X<> separated by slashes, and
678 # strip special escapes from index entries.
680 my ($self, $section, $index) = @_;
681 my @entries = map { split m%\s*/\s*% } @{ $$self{INDEX} };
682 return unless ($section || @entries);
684 # We're about to output all pending entries, so clear our pending queue.
687 # Build the output. Regular index entries are marked Xref, and headings
688 # pass in their own section. Undo some *roff formatting on headings.
691 push @output, [ 'Xref', join (' ', @entries) ];
695 $index =~ s/\\(?:s-?\d|.\(..|.)//g;
696 push @output, [ $section, $index ];
699 # Print out the .IX commands.
701 my ($type, $entry) = @$_;
702 $entry =~ s/\"/\"\"/g;
703 $self->output (".IX $type " . '"' . $entry . '"' . "\n");
707 # Output some text, without any additional changes.
709 my ($self, @text) = @_;
710 print { $$self{output_fh} } @text;
713 ##############################################################################
714 # Document initialization
715 ##############################################################################
717 # Handle the start of the document. Here we handle empty documents, as well
718 # as setting up our basic macros in a preamble and building the page title.
720 my ($self, $attrs) = @_;
721 if ($$attrs{contentless} && !$$self{ALWAYS_EMIT_SOMETHING}) {
722 DEBUG and print "Document is contentless\n";
723 $$self{CONTENTLESS} = 1;
727 # Determine information for the preamble and then output it.
728 my ($name, $section);
729 if (defined $$self{name}) {
730 $name = $$self{name};
731 $section = $$self{section} || 1;
733 ($name, $section) = $self->devise_title;
735 my $date = $$self{date} || $self->devise_date;
736 $self->preamble ($name, $section, $date)
737 unless $self->bare_output or DEBUG > 9;
739 # Initialize a few per-document variables.
740 $$self{INDENT} = 0; # Current indentation level.
741 $$self{INDENTS} = []; # Stack of indentations.
742 $$self{INDEX} = []; # Index keys waiting to be printed.
743 $$self{IN_NAME} = 0; # Whether processing the NAME section.
744 $$self{ITEMS} = 0; # The number of consecutive =items.
745 $$self{ITEMTYPES} = []; # Stack of =item types, one per list.
746 $$self{SHIFTWAIT} = 0; # Whether there is a shift waiting.
747 $$self{SHIFTS} = []; # Stack of .RS shifts.
748 $$self{PENDING} = [[]]; # Pending output.
751 # Handle the end of the document. This does nothing but print out a final
752 # comment at the end of the document under debugging.
755 return if $self->bare_output;
756 return if ($$self{CONTENTLESS} && !$$self{ALWAYS_EMIT_SOMETHING});
757 $self->output (q(.\" [End document]) . "\n") if DEBUG;
760 # Try to figure out the name and section from the file name and return them as
761 # a list, returning an empty name and section 1 if we can't find any better
762 # information. Uses File::Basename and File::Spec as necessary.
765 my $name = $self->source_filename || '';
766 my $section = $$self{section} || 1;
767 $section = 3 if (!$$self{section} && $name =~ /\.pm\z/i);
768 $name =~ s/\.p(od|[lm])\z//i;
770 # If the section isn't 3, then the name defaults to just the basename of
771 # the file. Otherwise, assume we're dealing with a module. We want to
772 # figure out the full module name from the path to the file, but we don't
773 # want to include too much of the path into the module name. Lose
774 # anything up to the first off:
776 # */lib/*perl*/ standard or site_perl module
777 # */*perl*/lib/ from -Dprefix=/opt/perl
778 # */*perl*/ random module hierarchy
780 # which works. Also strip off a leading site, site_perl, or vendor_perl
781 # component, any OS-specific component, and any version number component,
782 # and strip off an initial component of "lib" or "blib/lib" since that's
783 # what ExtUtils::MakeMaker creates. splitdir requires at least File::Spec
785 if ($section !~ /^3/) {
786 require File::Basename;
787 $name = uc File::Basename::basename ($name);
790 my ($volume, $dirs, $file) = File::Spec->splitpath ($name);
791 my @dirs = File::Spec->splitdir ($dirs);
794 for ($i = 0; $i < scalar @dirs; $i++) {
795 if ($dirs[$i] eq 'lib' && $dirs[$i + 1] =~ /perl/) {
798 } elsif ($dirs[$i] =~ /perl/) {
800 $cut++ if $dirs[$i + 1] eq 'lib';
805 splice (@dirs, 0, $cut);
806 shift @dirs if ($dirs[0] =~ /^(site|vendor)(_perl)?$/);
807 shift @dirs if ($dirs[0] =~ /^[\d.]+$/);
808 shift @dirs if ($dirs[0] =~ /^(.*-$^O|$^O-.*|$^O)$/);
810 shift @dirs if $dirs[0] eq 'lib';
811 splice (@dirs, 0, 2) if ($dirs[0] eq 'blib' && $dirs[1] eq 'lib');
813 # Remove empty directories when building the module name; they
814 # occur too easily on Unix by doubling slashes.
815 $name = join ('::', (grep { $_ ? $_ : () } @dirs), $file);
817 return ($name, $section);
820 # Determine the modification date and return that, properly formatted in ISO
821 # format. If we can't get the modification date of the input, instead use the
825 my $input = $self->source_filename;
826 my $time = ($input ? (stat $input)[9] : time);
827 return strftime ('%Y-%m-%d', localtime $time);
830 # Print out the preamble and the title. The meaning of the arguments to .TH
831 # unfortunately vary by system; some systems consider the fourth argument to
832 # be a "source" and others use it as a version number. Generally it's just
833 # presented as the left-side footer, though, so it doesn't matter too much if
834 # a particular system gives it another interpretation.
836 # The order of date and release used to be reversed in older versions of this
837 # module, but this order is correct for both Solaris and Linux.
839 my ($self, $name, $section, $date) = @_;
840 my $preamble = $self->preamble_template;
842 # Build the index line and make sure that it will be syntactically valid.
843 my $index = "$name $section";
844 $index =~ s/\"/\"\"/g;
846 # If name or section contain spaces, quote them (section really never
847 # should, but we may as well be cautious).
848 for ($name, $section) {
855 # Double quotes in date, since it will be quoted.
856 $date =~ s/\"/\"\"/g;
858 # Substitute into the preamble the configuration options.
859 $preamble =~ s/\@CFONT\@/$$self{fixed}/;
860 $preamble =~ s/\@LQUOTE\@/$$self{LQUOTE}/;
861 $preamble =~ s/\@RQUOTE\@/$$self{RQUOTE}/;
864 # Get the version information.
865 my $version = $self->version_report;
867 # Finally output everything.
868 $self->output (<<"----END OF HEADER----");
869 .\\" Automatically generated by $version
871 .\\" Standard preamble:
872 .\\" ========================================================================
874 .\\" ========================================================================
877 .TH $name $section "$date" "$$self{release}" "$$self{center}"
878 ----END OF HEADER----
879 $self->output (".\\\" [End of preamble]\n") if DEBUG;
882 ##############################################################################
884 ##############################################################################
886 # Handle a basic block of text. The only tricky part of this is if this is
887 # the first paragraph of text after an =over, in which case we have to change
888 # indentations for *roff.
890 my ($self, $attrs, $text) = @_;
891 my $line = $$attrs{start_line};
893 # Output the paragraph. We also have to handle =over without =item. If
894 # there's an =over without =item, SHIFTWAIT will be set, and we need to
895 # handle creation of the indent here. Add the shift to SHIFTS so that it
896 # will be cleaned up on =back.
898 if ($$self{SHIFTWAIT}) {
899 $self->output (".RS $$self{INDENT}\n");
900 push (@{ $$self{SHIFTS} }, $$self{INDENT});
901 $$self{SHIFTWAIT} = 0;
904 # Add the line number for debugging, but not in the NAME section just in
905 # case the comment would confuse apropos.
906 $self->output (".\\\" [At source line $line]\n")
907 if defined ($line) && DEBUG && !$$self{IN_NAME};
909 # Force exactly one newline at the end and strip unwanted trailing
910 # whitespace at the end.
913 # Output the paragraph.
914 $self->output ($self->protect ($self->textmapfonts ($text)));
916 $$self{NEEDSPACE} = 1;
920 # Handle a verbatim paragraph. Put a null token at the beginning of each line
921 # to protect against commands and wrap in .Vb/.Ve (which we define in our
924 my ($self, $attrs, $text) = @_;
926 # Ignore an empty verbatim paragraph.
927 return unless $text =~ /\S/;
929 # Force exactly one newline at the end and strip unwanted trailing
930 # whitespace at the end.
933 # Get a count of the number of lines before the first blank line, which
934 # we'll pass to .Vb as its parameter. This tells *roff to keep that many
935 # lines together. We don't want to tell *roff to keep huge blocks
937 my @lines = split (/\n/, $text);
943 $unbroken = 10 if ($unbroken > 12 && !$$self{MAGIC_VNOPAGEBREAK_LIMIT});
945 # Prepend a null token to each line.
948 # Output the results.
950 $self->output (".Vb $unbroken\n$text.Ve\n");
951 $$self{NEEDSPACE} = 1;
955 # Handle literal text (produced by =for and similar constructs). Just output
956 # it with the minimum of changes.
958 my ($self, $attrs, $text) = @_;
960 $text =~ s/\n{0,2}$/\n/;
961 $self->output ($text);
965 ##############################################################################
967 ##############################################################################
969 # Common code for all headings. This is called before the actual heading is
970 # output. It returns the cleaned up heading text (putting the heading all on
971 # one line) and may do other things, like closing bad =item blocks.
973 my ($self, $text, $line) = @_;
975 $text =~ s/\s*\n\s*/ /g;
977 # This should never happen; it means that we have a heading after =item
978 # without an intervening =back. But just in case, handle it anyway.
979 if ($$self{ITEMS} > 1) {
981 $self->output (".PD\n");
984 # Output the current source line.
985 $self->output ( ".\\\" [At source line $line]\n" )
986 if defined ($line) && DEBUG;
990 # First level heading. We can't output .IX in the NAME section due to a bug
991 # in some versions of catman, so don't output a .IX for that section. .SH
992 # already uses small caps, so remove \s0 and \s-1. Maintain IN_NAME as
995 my ($self, $attrs, $text) = @_;
996 $text =~ s/\\s-?\d//g;
997 $text = $self->heading_common ($text, $$attrs{start_line});
998 my $isname = ($text eq 'NAME' || $text =~ /\(NAME\)/);
999 $self->output ($self->switchquotes ('.SH', $self->mapfonts ($text)));
1000 $self->outindex ('Header', $text) unless $isname;
1001 $$self{NEEDSPACE} = 0;
1002 $$self{IN_NAME} = $isname;
1006 # Second level heading.
1008 my ($self, $attrs, $text) = @_;
1009 $text = $self->heading_common ($text, $$attrs{start_line});
1010 $self->output ($self->switchquotes ('.Sh', $self->mapfonts ($text)));
1011 $self->outindex ('Subsection', $text);
1012 $$self{NEEDSPACE} = 0;
1016 # Third level heading. *roff doesn't have this concept, so just put the
1017 # heading in italics as a normal paragraph.
1019 my ($self, $attrs, $text) = @_;
1020 $text = $self->heading_common ($text, $$attrs{start_line});
1022 $self->output ($self->textmapfonts ('\f(IS' . $text . '\f(IE') . "\n");
1023 $self->outindex ('Subsection', $text);
1024 $$self{NEEDSPACE} = 1;
1028 # Fourth level heading. *roff doesn't have this concept, so just put the
1029 # heading as a normal paragraph.
1031 my ($self, $attrs, $text) = @_;
1032 $text = $self->heading_common ($text, $$attrs{start_line});
1034 $self->output ($self->textmapfonts ($text) . "\n");
1035 $self->outindex ('Subsection', $text);
1036 $$self{NEEDSPACE} = 1;
1040 ##############################################################################
1042 ##############################################################################
1044 # All of the formatting codes that aren't handled internally by the parser,
1045 # other than L<> and X<>.
1046 sub cmd_b { return '\f(BS' . $_[2] . '\f(BE' }
1047 sub cmd_i { return '\f(IS' . $_[2] . '\f(IE' }
1048 sub cmd_f { return '\f(IS' . $_[2] . '\f(IE' }
1049 sub cmd_c { return $_[0]->quote_literal ($_[2]) }
1051 # Index entries are just added to the pending entries.
1053 my ($self, $attrs, $text) = @_;
1054 push (@{ $$self{INDEX} }, $text);
1058 # Links reduce to the text that we're given, wrapped in angle brackets if it's
1061 my ($self, $attrs, $text) = @_;
1062 return $$attrs{type} eq 'url' ? "<$text>" : $text;
1065 ##############################################################################
1067 ##############################################################################
1069 # Handle the beginning of an =over block. Takes the type of the block as the
1070 # first argument, and then the attr hash. This is called by the handlers for
1071 # the four different types of lists (bullet, number, text, and block).
1072 sub over_common_start {
1073 my ($self, $type, $attrs) = @_;
1074 my $line = $$attrs{start_line};
1075 my $indent = $$attrs{indent};
1076 DEBUG > 3 and print " Starting =over $type (line $line, indent ",
1077 ($indent || '?'), "\n";
1079 # Find the indentation level.
1080 unless (defined ($indent) && $indent =~ /^[-+]?\d{1,4}\s*$/) {
1081 $indent = $$self{indent};
1084 # If we've gotten multiple indentations in a row, we need to emit the
1085 # pending indentation for the last level that we saw and haven't acted on
1086 # yet. SHIFTS is the stack of indentations that we've actually emitted
1088 if (@{ $$self{SHIFTS} } < @{ $$self{INDENTS} }) {
1089 $self->output (".RS $$self{INDENT}\n");
1090 push (@{ $$self{SHIFTS} }, $$self{INDENT});
1093 # Now, do record-keeping. INDENTS is a stack of indentations that we've
1094 # seen so far, and INDENT is the current level of indentation. ITEMTYPES
1095 # is a stack of list types that we've seen.
1096 push (@{ $$self{INDENTS} }, $$self{INDENT});
1097 push (@{ $$self{ITEMTYPES} }, $type);
1098 $$self{INDENT} = $indent + 0;
1099 $$self{SHIFTWAIT} = 1;
1102 # End an =over block. Takes no options other than the class pointer.
1103 # Normally, once we close a block and therefore remove something from INDENTS,
1104 # INDENTS will now be longer than SHIFTS, indicating that we also need to emit
1105 # *roff code to close the indent. This isn't *always* true, depending on the
1106 # circumstance. If we're still inside an indentation, we need to emit another
1107 # .RE and then a new .RS to unconfuse *roff.
1108 sub over_common_end {
1110 DEBUG > 3 and print " Ending =over\n";
1111 $$self{INDENT} = pop @{ $$self{INDENTS} };
1112 pop @{ $$self{ITEMTYPES} };
1114 # If we emitted code for that indentation, end it.
1115 if (@{ $$self{SHIFTS} } > @{ $$self{INDENTS} }) {
1116 $self->output (".RE\n");
1117 pop @{ $$self{SHIFTS} };
1120 # If we're still in an indentation, *roff will have now lost track of the
1121 # right depth of that indentation, so fix that.
1122 if (@{ $$self{INDENTS} } > 0) {
1123 $self->output (".RE\n");
1124 $self->output (".RS $$self{INDENT}\n");
1126 $$self{NEEDSPACE} = 1;
1127 $$self{SHIFTWAIT} = 0;
1130 # Dispatch the start and end calls as appropriate.
1131 sub start_over_bullet { my $s = shift; $s->over_common_start ('bullet', @_) }
1132 sub start_over_number { my $s = shift; $s->over_common_start ('number', @_) }
1133 sub start_over_text { my $s = shift; $s->over_common_start ('text', @_) }
1134 sub start_over_block { my $s = shift; $s->over_common_start ('block', @_) }
1135 sub end_over_bullet { $_[0]->over_common_end }
1136 sub end_over_number { $_[0]->over_common_end }
1137 sub end_over_text { $_[0]->over_common_end }
1138 sub end_over_block { $_[0]->over_common_end }
1140 # The common handler for all item commands. Takes the type of the item, the
1141 # attributes, and then the text of the item.
1143 # Emit an index entry for anything that's interesting, but don't emit index
1144 # entries for things like bullets and numbers. Newlines in an item title are
1145 # turned into spaces since *roff can't handle them embedded.
1147 my ($self, $type, $attrs, $text) = @_;
1148 my $line = $$attrs{start_line};
1149 DEBUG > 3 and print " $type item (line $line): $text\n";
1151 # Clean up the text. We want to end up with two variables, one ($text)
1152 # which contains any body text after taking out the item portion, and
1153 # another ($item) which contains the actual item text.
1156 if ($type eq 'bullet') {
1158 $text =~ s/\n*$/\n/;
1159 } elsif ($type eq 'number') {
1160 $item = $$attrs{number} . '.';
1163 $item =~ s/\s*\n\s*/ /g;
1165 $index = $item if ($item =~ /\w/);
1168 # Take care of the indentation. If shifts and indents are equal, close
1169 # the top shift, since we're about to create an indentation with .IP.
1170 # Also output .PD 0 to turn off spacing between items if this item is
1171 # directly following another one. We only have to do that once for a
1172 # whole chain of items so do it for the second item in the change. Note
1173 # that makespace is what undoes this.
1174 if (@{ $$self{SHIFTS} } == @{ $$self{INDENTS} }) {
1175 $self->output (".RE\n");
1176 pop @{ $$self{SHIFTS} };
1178 $self->output (".PD 0\n") if ($$self{ITEMS} == 1);
1180 # Now, output the item tag itself.
1181 $item = $self->textmapfonts ($item);
1182 $self->output ($self->switchquotes ('.IP', $item, $$self{INDENT}));
1183 $$self{NEEDSPACE} = 0;
1185 $$self{SHIFTWAIT} = 0;
1187 # If body text for this item was included, go ahead and output that now.
1189 $text =~ s/\s*$/\n/;
1191 $self->output ($self->protect ($self->textmapfonts ($text)));
1192 $$self{NEEDSPACE} = 1;
1194 $self->outindex ($index ? ('Item', $index) : ());
1197 # Dispatch the item commands to the appropriate place.
1198 sub cmd_item_bullet { my $self = shift; $self->item_common ('bullet', @_) }
1199 sub cmd_item_number { my $self = shift; $self->item_common ('number', @_) }
1200 sub cmd_item_text { my $self = shift; $self->item_common ('text', @_) }
1201 sub cmd_item_block { my $self = shift; $self->item_common ('block', @_) }
1203 ##############################################################################
1204 # Translation tables
1205 ##############################################################################
1207 # The following table is adapted from Tom Christiansen's pod2man. It assumes
1208 # that the standard preamble has already been printed, since that's what
1209 # defines all of the accent marks. We really want to do something better than
1210 # this when *roff actually supports other character sets itself, since these
1211 # results are pretty poor.
1213 # This only works in an ASCII world. What to do in a non-ASCII world is very
1215 @ESCAPES{0xA0 .. 0xFF} = (
1216 "\\ ", undef, undef, undef, undef, undef, undef, undef,
1217 undef, undef, undef, undef, undef, "\\%", undef, undef,
1219 undef, undef, undef, undef, undef, undef, undef, undef,
1220 undef, undef, undef, undef, undef, undef, undef, undef,
1222 "A\\*`", "A\\*'", "A\\*^", "A\\*~", "A\\*:", "A\\*o", "\\*(AE", "C\\*,",
1223 "E\\*`", "E\\*'", "E\\*^", "E\\*:", "I\\*`", "I\\*'", "I\\*^", "I\\*:",
1225 "\\*(D-", "N\\*~", "O\\*`", "O\\*'", "O\\*^", "O\\*~", "O\\*:", undef,
1226 "O\\*/", "U\\*`", "U\\*'", "U\\*^", "U\\*:", "Y\\*'", "\\*(Th", "\\*8",
1228 "a\\*`", "a\\*'", "a\\*^", "a\\*~", "a\\*:", "a\\*o", "\\*(ae", "c\\*,",
1229 "e\\*`", "e\\*'", "e\\*^", "e\\*:", "i\\*`", "i\\*'", "i\\*^", "i\\*:",
1231 "\\*(d-", "n\\*~", "o\\*`", "o\\*'", "o\\*^", "o\\*~", "o\\*:", undef,
1232 "o\\*/" , "u\\*`", "u\\*'", "u\\*^", "u\\*:", "y\\*'", "\\*(th", "y\\*:",
1235 # Make sure that at least this works even outside of ASCII.
1236 $ESCAPES{ord("\\")} = "\\e";
1238 ##############################################################################
1240 ##############################################################################
1242 # The following is the static preamble which starts all *roff output we
1243 # generate. It's completely static except for the font to use as a
1244 # fixed-width font, which is designed by @CFONT@, and the left and right
1245 # quotes to use for C<> text, designated by @LQOUTE@ and @RQUOTE@.
1246 sub preamble_template {
1247 return <<'----END OF PREAMBLE----';
1248 .de Sh \" Subsection heading
1256 .de Sp \" Vertical space (when we can't use .PP)
1260 .de Vb \" Begin verbatim text
1265 .de Ve \" End verbatim text
1269 .\" Set up some character translations and predefined strings. \*(-- will
1270 .\" give an unbreakable dash, \*(PI will give pi, \*(L" will give a left
1271 .\" double quote, and \*(R" will give a right double quote. | will give a
1272 .\" real vertical bar. \*(C+ will give a nicer C++. Capital omega is used to
1273 .\" do unbreakable dashes and therefore won't be available. \*(C` and \*(C'
1274 .\" expand to `' in nroff, nothing in troff, for use with C<>.
1276 .ds C+ C\v'-.1v'\h'-1p'\s-2+\h'-1p'+\s0\v'.1v'\h'-1p'
1280 . if (\n(.H=4u)&(1m=24u) .ds -- \(*W\h'-12u'\(*W\h'-12u'-\" diablo 10 pitch
1281 . if (\n(.H=4u)&(1m=20u) .ds -- \(*W\h'-12u'\(*W\h'-8u'-\" diablo 12 pitch
1294 .\" If the F register is turned on, we'll generate index entries on stderr for
1295 .\" titles (.TH), headers (.SH), subsections (.Sh), items (.Ip), and index
1296 .\" entries marked with X<> in POD. Of course, you'll have to process the
1297 .\" output yourself in some meaningful fashion.
1300 . tm Index:\\$1\t\\n%\t"\\$2"
1306 .\" For nroff, turn off justification. Always turn off hyphenation; it makes
1307 .\" way too many mistakes in technical documents.
1311 .\" Accent mark definitions (@(#)ms.acc 1.5 88/02/08 SMI; from UCB 4.2).
1312 .\" Fear. Run. Save yourself. No user-serviceable parts.
1313 . \" fudge factors for nroff and troff
1322 . ds #H ((1u-(\\\\n(.fu%2u))*.13m)
1328 . \" simple accents for nroff and troff
1338 . ds ' \\k:\h'-(\\n(.wu*8/10-\*(#H)'\'\h"|\\n:u"
1339 . ds ` \\k:\h'-(\\n(.wu*8/10-\*(#H)'\`\h'|\\n:u'
1340 . ds ^ \\k:\h'-(\\n(.wu*10/11-\*(#H)'^\h'|\\n:u'
1341 . ds , \\k:\h'-(\\n(.wu*8/10)',\h'|\\n:u'
1342 . ds ~ \\k:\h'-(\\n(.wu-\*(#H-.1m)'~\h'|\\n:u'
1343 . ds / \\k:\h'-(\\n(.wu*8/10-\*(#H)'\z\(sl\h'|\\n:u'
1345 . \" troff and (daisy-wheel) nroff accents
1346 .ds : \\k:\h'-(\\n(.wu*8/10-\*(#H+.1m+\*(#F)'\v'-\*(#V'\z.\h'.2m+\*(#F'.\h'|\\n:u'\v'\*(#V'
1347 .ds 8 \h'\*(#H'\(*b\h'-\*(#H'
1348 .ds o \\k:\h'-(\\n(.wu+\w'\(de'u-\*(#H)/2u'\v'-.3n'\*(#[\z\(de\v'.3n'\h'|\\n:u'\*(#]
1349 .ds d- \h'\*(#H'\(pd\h'-\w'~'u'\v'-.25m'\f2\(hy\fP\v'.25m'\h'-\*(#H'
1350 .ds D- D\\k:\h'-\w'D'u'\v'-.11m'\z\(hy\v'.11m'\h'|\\n:u'
1351 .ds th \*(#[\v'.3m'\s+1I\s-1\v'-.3m'\h'-(\w'I'u*2/3)'\s-1o\s+1\*(#]
1352 .ds Th \*(#[\s+2I\s-2\h'-\w'I'u*3/5'\v'-.3m'o\v'.3m'\*(#]
1353 .ds ae a\h'-(\w'a'u*4/10)'e
1354 .ds Ae A\h'-(\w'A'u*4/10)'E
1355 . \" corrections for vroff
1356 .if v .ds ~ \\k:\h'-(\\n(.wu*9/10-\*(#H)'\s-2\u~\d\s+2\h'|\\n:u'
1357 .if v .ds ^ \\k:\h'-(\\n(.wu*10/11-\*(#H)'\v'-.4m'^\v'.4m'\h'|\\n:u'
1358 . \" for low resolution devices (crt and lpr)
1359 .if \n(.H>23 .if \n(.V>19 \
1371 .rm #[ #] #H #V #F C
1372 ----END OF PREAMBLE----
1376 ##############################################################################
1377 # Module return value and documentation
1378 ##############################################################################
1385 Pod::Man - Convert POD data to formatted *roff input
1390 my $parser = Pod::Man->new (release => $VERSION, section => 8);
1392 # Read POD from STDIN and write to STDOUT.
1393 $parser->parse_file (\*STDIN);
1395 # Read POD from file.pod and write to file.1.
1396 $parser->parse_from_file ('file.pod', 'file.1');
1400 Pod::Man is a module to convert documentation in the POD format (the
1401 preferred language for documenting Perl) into *roff input using the man
1402 macro set. The resulting *roff code is suitable for display on a terminal
1403 using L<nroff(1)>, normally via L<man(1)>, or printing using L<troff(1)>.
1404 It is conventionally invoked using the driver script B<pod2man>, but it can
1405 also be used directly.
1407 As a derived class from Pod::Simple, Pod::Man supports the same methods and
1408 interfaces. See L<Pod::Simple> for all the details.
1410 new() can take options, in the form of key/value pairs that control the
1411 behavior of the parser. See below for details.
1413 If no options are given, Pod::Man uses the name of the input file with any
1414 trailing C<.pod>, C<.pm>, or C<.pl> stripped as the man page title, to
1415 section 1 unless the file ended in C<.pm> in which case it defaults to
1416 section 3, to a centered title of "User Contributed Perl Documentation", to
1417 a centered footer of the Perl version it is run with, and to a left-hand
1418 footer of the modification date of its input (or the current date if given
1421 Pod::Man assumes that your *roff formatters have a fixed-width font named
1422 CW. If yours is called something else (like CR), use the C<fixed> option to
1423 specify it. This generally only matters for troff output for printing.
1424 Similarly, you can set the fonts used for bold, italic, and bold italic
1427 Besides the obvious pod conversions, Pod::Man also takes care of formatting
1428 func(), func(3), and simple variable references like $foo or @bar so you
1429 don't have to use code escapes for them; complex expressions like
1430 C<$fred{'stuff'}> will still need to be escaped, though. It also translates
1431 dashes that aren't used as hyphens into en dashes, makes long dashes--like
1432 this--into proper em dashes, fixes "paired quotes," makes C++ look right,
1433 puts a little space between double underbars, makes ALLCAPS a teeny bit
1434 smaller in B<troff>, and escapes stuff that *roff treats as special so that
1437 The recognized options to new() are as follows. All options take a single
1444 Sets the centered page header to use instead of "User Contributed Perl
1449 Sets the left-hand footer. By default, the modification date of the input
1450 file will be used, or the current date if stat() can't find that file (the
1451 case if the input is from STDIN), and the date will be formatted as
1456 The fixed-width font to use for vertabim text and code. Defaults to CW.
1457 Some systems may want CR instead. Only matters for B<troff> output.
1461 Bold version of the fixed-width font. Defaults to CB. Only matters for
1466 Italic version of the fixed-width font (actually, something of a misnomer,
1467 since most fixed-width fonts only have an oblique version, not an italic
1468 version). Defaults to CI. Only matters for B<troff> output.
1470 =item fixedbolditalic
1472 Bold italic (probably actually oblique) version of the fixed-width font.
1473 Pod::Man doesn't assume you have this, and defaults to CB. Some systems
1474 (such as Solaris) have this font available as CX. Only matters for B<troff>
1479 Set the name of the manual page. Without this option, the manual name is
1480 set to the uppercased base name of the file being converted unless the
1481 manual section is 3, in which case the path is parsed to see if it is a Perl
1482 module path. If it is, a path like C<.../lib/Pod/Man.pm> is converted into
1483 a name like C<Pod::Man>. This option, if given, overrides any automatic
1484 determination of the name.
1488 Sets the quote marks used to surround CE<lt>> text. If the value is a
1489 single character, it is used as both the left and right quote; if it is two
1490 characters, the first character is used as the left quote and the second as
1491 the right quoted; and if it is four characters, the first two are used as
1492 the left quote and the second two as the right quote.
1494 This may also be set to the special value C<none>, in which case no quote
1495 marks are added around CE<lt>> text (but the font is still changed for troff
1500 Set the centered footer. By default, this is the version of Perl you run
1501 Pod::Man under. Note that some system an macro sets assume that the
1502 centered footer will be a modification date and will prepend something like
1503 "Last modified: "; if this is the case, you may want to set C<release> to
1504 the last modified date and C<date> to the version number.
1508 Set the section for the C<.TH> macro. The standard section numbering
1509 convention is to use 1 for user commands, 2 for system calls, 3 for
1510 functions, 4 for devices, 5 for file formats, 6 for games, 7 for
1511 miscellaneous information, and 8 for administrator commands. There is a lot
1512 of variation here, however; some systems (like Solaris) use 4 for file
1513 formats, 5 for miscellaneous information, and 7 for devices. Still others
1514 use 1m instead of 8, or some mix of both. About the only section numbers
1515 that are reliably consistent are 1, 2, and 3.
1517 By default, section 1 will be used unless the file ends in .pm in which case
1518 section 3 will be selected.
1522 The standard Pod::Simple method parse_file() takes one argument naming the
1523 POD file to read from. By default, the output is sent to STDOUT, but this
1524 can be changed with the output_fd() method.
1526 The standard Pod::Simple method parse_from_file() takes up to two
1527 arguments, the first being the input file to read POD from and the second
1528 being the file to write the formatted output to.
1530 You can also call parse_lines() to parse an array of lines or
1531 parse_string_document() to parse a document already in memory. To put the
1532 output into a string instead of a file handle, call the output_string()
1533 method. See L<Pod::Simple> for the specific details.
1539 =item roff font should be 1 or 2 chars, not "%s"
1541 (F) You specified a *roff font (using C<fixed>, C<fixedbold>, etc.) that
1542 wasn't either one or two characters. Pod::Man doesn't support *roff fonts
1543 longer than two characters, although some *roff extensions do (the canonical
1544 versions of B<nroff> and B<troff> don't either).
1546 =item Invalid quote specification "%s"
1548 (F) The quote specification given (the quotes option to the constructor) was
1549 invalid. A quote specification must be one, two, or four characters long.
1555 Eight-bit input data isn't handled at all well at present. The correct
1556 approach would be to map EE<lt>E<gt> escapes to the appropriate UTF-8
1557 characters and then do a translation pass on the output according to the
1558 user-specified output character set. Unfortunately, we can't send eight-bit
1559 data directly to the output unless the user says this is okay, since some
1560 vendor *roff implementations can't handle eight-bit data. If the *roff
1561 implementation can, however, that's far superior to the current hacked
1562 characters that only work under troff.
1564 There is currently no way to turn off the guesswork that tries to format
1565 unmarked text appropriately, and sometimes it isn't wanted (particularly
1566 when using POD to document something other than Perl). Most of the work
1567 towards fixing this has now been done, however, and all that's still needed
1568 is a user interface.
1570 The NAME section should be recognized specially and index entries emitted
1571 for everything in that section. This would have to be deferred until the
1572 next section, since extraneous things in NAME tends to confuse various man
1573 page processors. Currently, no index entries are emitted for anything in
1576 Pod::Man doesn't handle font names longer than two characters. Neither do
1577 most B<troff> implementations, but GNU troff does as an extension. It would
1578 be nice to support as an option for those who want to use it.
1580 The preamble added to each output file is rather verbose, and most of it
1581 is only necessary in the presence of non-ASCII characters. It would
1582 ideally be nice if all of those definitions were only output if needed,
1583 perhaps on the fly as the characters are used.
1585 Pod::Man is excessively slow.
1589 The handling of hyphens and em dashes is somewhat fragile, and one may get
1590 the wrong one under some circumstances. This should only matter for
1593 When and whether to use small caps is somewhat tricky, and Pod::Man doesn't
1594 necessarily get it right.
1596 Converting neutral double quotes to properly matched double quotes doesn't
1597 work unless there are no formatting codes between the quote marks. This
1598 only matters for troff output.
1602 Russ Allbery <rra@stanford.edu>, based I<very> heavily on the original
1603 B<pod2man> by Tom Christiansen <tchrist@mox.perl.com>. The modifications to
1604 work with Pod::Simple instead of Pod::Parser were originally contributed by
1605 Sean Burke (but I've since hacked them beyond recognition and all bugs are
1608 =head1 COPYRIGHT AND LICENSE
1610 Copyright 1999, 2000, 2001, 2002, 2003, 2004, 2005
1611 by Russ Allbery <rra@stanford.edu>.
1613 This program is free software; you may redistribute it and/or modify it
1614 under the same terms as Perl itself.
1618 L<Pod::Simple>, L<perlpod(1)>, L<pod2man(1)>, L<nroff(1)>, L<troff(1)>,
1619 L<man(1)>, L<man(7)>
1621 Ossanna, Joseph F., and Brian W. Kernighan. "Troff User's Manual,"
1622 Computing Science Technical Report No. 54, AT&T Bell Laboratories. This is
1623 the best documentation of standard B<nroff> and B<troff>. At the time of
1624 this writing, it's available at
1625 L<http://www.cs.bell-labs.com/cm/cs/cstr.html>.
1627 The man page documenting the man macro set may be L<man(5)> instead of
1628 L<man(7)> on your system. Also, please see L<pod2man(1)> for extensive
1629 documentation on writing manual pages if you've not done it before and
1630 aren't familiar with the conventions.
1632 The current version of this module is always available from its web site at
1633 L<http://www.eyrie.org/~eagle/software/podlators/>. It is also part of the
1634 Perl core distribution as of 5.6.0.