1 # Pod::Man -- Convert POD data to formatted *roff input.
3 # Copyright 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008
4 # Russ Allbery <rra@stanford.edu>
5 # Substantial contributions by Sean Burke <sburke@cpan.org>
7 # This program is free software; you may redistribute it and/or modify it
8 # under the same terms as Perl itself.
10 # This module translates POD documentation into *roff markup using the man
11 # macro set, and is intended for converting POD documents written as Unix
12 # manual pages to manual pages that can be read by the man(1) command. It is
13 # a replacement for the pod2man command distributed with versions of Perl
16 # Perl core hackers, please note that this module is also separately
17 # maintained outside of the Perl core as part of the podlators. Please send
18 # me any patches at the address above in addition to sending them to the
19 # standard Perl mailing lists.
21 ##############################################################################
22 # Modules and declarations
23 ##############################################################################
30 use subs qw(makespace);
31 use vars qw(@ISA %ESCAPES $PREAMBLE $VERSION);
35 use POSIX qw(strftime);
37 @ISA = qw(Pod::Simple);
41 # Set the debugging level. If someone has inserted a debug function into this
42 # class already, use that. Otherwise, use any Pod::Simple debug function
43 # that's defined, and failing that, define a debug level of 10.
45 my $parent = defined (&Pod::Simple::DEBUG) ? \&Pod::Simple::DEBUG : undef;
46 unless (defined &DEBUG) {
47 *DEBUG = $parent || sub () { 10 };
51 # Import the ASCII constant from Pod::Simple. This is true iff we're in an
52 # ASCII-based universe (including such things as ISO 8859-1 and UTF-8), and is
53 # generally only false for EBCDIC.
54 BEGIN { *ASCII = \&Pod::Simple::ASCII }
56 # Pretty-print a data structure. Only used for debugging.
57 BEGIN { *pretty = \&Pod::Simple::pretty }
59 ##############################################################################
60 # Object initialization
61 ##############################################################################
63 # Initialize the object and set various Pod::Simple options that we need.
64 # Here, we also process any additional options passed to the constructor or
65 # set up defaults if none were given. Note that all internal object keys are
66 # in all-caps, reserving all lower-case object keys for Pod::Simple and user
70 my $self = $class->SUPER::new;
72 # Tell Pod::Simple not to handle S<> by automatically inserting .
73 $self->nbsp_for_S (1);
75 # Tell Pod::Simple to keep whitespace whenever possible.
76 if ($self->can ('preserve_whitespace')) {
77 $self->preserve_whitespace (1);
79 $self->fullstop_space_harden (1);
82 # The =for and =begin targets that we accept.
83 $self->accept_targets (qw/man MAN roff ROFF/);
85 # Ensure that contiguous blocks of code are merged together. Otherwise,
86 # some of the guesswork heuristics don't work right.
87 $self->merge_text (1);
89 # Pod::Simple doesn't do anything useful with our arguments, but we want
90 # to put them in our object as hash keys and values. This could cause
91 # problems if we ever clash with Pod::Simple's own internal class
93 %$self = (%$self, @_);
95 # Send errors to stderr if requested.
97 $self->no_errata_section (1);
98 $self->complain_stderr (1);
99 delete $$self{stderr};
102 # Initialize various other internal constants based on our arguments.
107 # For right now, default to turning on all of the magic.
108 $$self{MAGIC_CPP} = 1;
109 $$self{MAGIC_EMDASH} = 1;
110 $$self{MAGIC_FUNC} = 1;
111 $$self{MAGIC_MANREF} = 1;
112 $$self{MAGIC_SMALLCAPS} = 1;
113 $$self{MAGIC_VARS} = 1;
118 # Translate a font string into an escape.
119 sub toescape { (length ($_[0]) > 1 ? '\f(' : '\f') . $_[0] }
121 # Determine which fonts the user wishes to use and store them in the object.
122 # Regular, italic, bold, and bold-italic are constants, but the fixed width
123 # fonts may be set by the user. Sets the internal hash key FONTS which is
124 # used to map our internal font escapes to actual *roff sequences later.
128 # Figure out the fixed-width font. If user-supplied, make sure that they
129 # are the right length.
130 for (qw/fixed fixedbold fixeditalic fixedbolditalic/) {
131 my $font = $$self{$_};
132 if (defined ($font) && (length ($font) < 1 || length ($font) > 2)) {
133 croak qq(roff font should be 1 or 2 chars, not "$font");
137 # Set the default fonts. We can't be sure portably across different
138 # implementations what fixed bold-italic may be called (if it's even
139 # available), so default to just bold.
140 $$self{fixed} ||= 'CW';
141 $$self{fixedbold} ||= 'CB';
142 $$self{fixeditalic} ||= 'CI';
143 $$self{fixedbolditalic} ||= 'CB';
145 # Set up a table of font escapes. First number is fixed-width, second is
146 # bold, third is italic.
147 $$self{FONTS} = { '000' => '\fR', '001' => '\fI',
148 '010' => '\fB', '011' => '\f(BI',
149 '100' => toescape ($$self{fixed}),
150 '101' => toescape ($$self{fixeditalic}),
151 '110' => toescape ($$self{fixedbold}),
152 '111' => toescape ($$self{fixedbolditalic}) };
155 # Initialize the quotes that we'll be using for C<> text. This requires some
156 # special handling, both to parse the user parameter if given and to make sure
157 # that the quotes will be safe against *roff. Sets the internal hash keys
162 $$self{quotes} ||= '"';
163 if ($$self{quotes} eq 'none') {
164 $$self{LQUOTE} = $$self{RQUOTE} = '';
165 } elsif (length ($$self{quotes}) == 1) {
166 $$self{LQUOTE} = $$self{RQUOTE} = $$self{quotes};
167 } elsif ($$self{quotes} =~ /^(.)(.)$/
168 || $$self{quotes} =~ /^(..)(..)$/) {
172 croak(qq(Invalid quote specification "$$self{quotes}"))
175 # Double the first quote; note that this should not be s///g as two double
176 # quotes is represented in *roff as three double quotes, not four. Weird,
178 $$self{LQUOTE} =~ s/\"/\"\"/;
179 $$self{RQUOTE} =~ s/\"/\"\"/;
182 # Initialize the page title information and indentation from our arguments.
186 # We used to try first to get the version number from a local binary, but
187 # we shouldn't need that any more. Get the version from the running Perl.
188 # Work a little magic to handle subversions correctly under both the
189 # pre-5.6 and the post-5.6 version numbering schemes.
190 my @version = ($] =~ /^(\d+)\.(\d{3})(\d{0,3})$/);
192 $version[2] *= 10 ** (3 - length $version[2]);
193 for (@version) { $_ += 0 }
194 my $version = join ('.', @version);
196 # Set the defaults for page titles and indentation if the user didn't
198 $$self{center} = 'User Contributed Perl Documentation'
199 unless defined $$self{center};
200 $$self{release} = 'perl v' . $version
201 unless defined $$self{release};
203 unless defined $$self{indent};
205 # Double quotes in things that will be quoted.
206 for (qw/center release/) {
207 $$self{$_} =~ s/\"/\"\"/g if $$self{$_};
211 ##############################################################################
213 ##############################################################################
215 # This is the glue that connects the code below with Pod::Simple itself. The
216 # goal is to convert the event stream coming from the POD parser into method
217 # calls to handlers once the complete content of a tag has been seen. Each
218 # paragraph or POD command will have textual content associated with it, and
219 # as soon as all of a paragraph or POD command has been seen, that content
220 # will be passed in to the corresponding method for handling that type of
221 # object. The exceptions are handlers for lists, which have opening tag
222 # handlers and closing tag handlers that will be called right away.
224 # The internal hash key PENDING is used to store the contents of a tag until
225 # all of it has been seen. It holds a stack of open tags, each one
226 # represented by a tuple of the attributes hash for the tag, formatting
227 # options for the tag (which are inherited), and the contents of the tag.
229 # Add a block of text to the contents of the current node, formatting it
230 # according to the current formatting instructions as we do.
232 my ($self, $text) = @_;
233 DEBUG > 3 and print "== $text\n";
234 my $tag = $$self{PENDING}[-1];
235 $$tag[2] .= $self->format_text ($$tag[1], $text);
238 # Given an element name, get the corresponding method name.
239 sub method_for_element {
240 my ($self, $element) = @_;
242 $element =~ tr/A-Z/a-z/;
243 $element =~ tr/_a-z0-9//cd;
247 # Handle the start of a new element. If cmd_element is defined, assume that
248 # we need to collect the entire tree for this element before passing it to the
249 # element method, and create a new tree into which we'll collect blocks of
250 # text and nested elements. Otherwise, if start_element is defined, call it.
251 sub _handle_element_start {
252 my ($self, $element, $attrs) = @_;
253 DEBUG > 3 and print "++ $element (<", join ('> <', %$attrs), ">)\n";
254 my $method = $self->method_for_element ($element);
256 # If we have a command handler, we need to accumulate the contents of the
257 # tag before calling it. Turn off IN_NAME for any command other than
258 # <Para> so that IN_NAME isn't still set for the first heading after the
260 if ($self->can ("cmd_$method")) {
261 DEBUG > 2 and print "<$element> starts saving a tag\n";
262 $$self{IN_NAME} = 0 if ($element ne 'Para');
264 # How we're going to format embedded text blocks depends on the tag
265 # and also depends on our parent tags. Thankfully, inside tags that
266 # turn off guesswork and reformatting, nothing else can turn it back
267 # on, so this can be strictly inherited.
268 my $formatting = $$self{PENDING}[-1][1];
269 $formatting = $self->formatting ($formatting, $element);
270 push (@{ $$self{PENDING} }, [ $attrs, $formatting, '' ]);
271 DEBUG > 4 and print "Pending: [", pretty ($$self{PENDING}), "]\n";
272 } elsif ($self->can ("start_$method")) {
273 my $method = 'start_' . $method;
274 $self->$method ($attrs, '');
276 DEBUG > 2 and print "No $method start method, skipping\n";
280 # Handle the end of an element. If we had a cmd_ method for this element,
281 # this is where we pass along the tree that we built. Otherwise, if we have
282 # an end_ method for the element, call that.
283 sub _handle_element_end {
284 my ($self, $element) = @_;
285 DEBUG > 3 and print "-- $element\n";
286 my $method = $self->method_for_element ($element);
288 # If we have a command handler, pull off the pending text and pass it to
289 # the handler along with the saved attribute hash.
290 if ($self->can ("cmd_$method")) {
291 DEBUG > 2 and print "</$element> stops saving a tag\n";
292 my $tag = pop @{ $$self{PENDING} };
293 DEBUG > 4 and print "Popped: [", pretty ($tag), "]\n";
294 DEBUG > 4 and print "Pending: [", pretty ($$self{PENDING}), "]\n";
295 my $method = 'cmd_' . $method;
296 my $text = $self->$method ($$tag[0], $$tag[2]);
298 if (@{ $$self{PENDING} } > 1) {
299 $$self{PENDING}[-1][2] .= $text;
301 $self->output ($text);
304 } elsif ($self->can ("end_$method")) {
305 my $method = 'end_' . $method;
308 DEBUG > 2 and print "No $method end method, skipping\n";
312 ##############################################################################
314 ##############################################################################
316 # Return formatting instructions for a new block. Takes the current
317 # formatting and the new element. Formatting inherits negatively, in the
318 # sense that if the parent has turned off guesswork, all child elements should
319 # leave it off. We therefore return a copy of the same formatting
320 # instructions but possibly with more things turned off depending on the
323 my ($self, $current, $element) = @_;
326 %options = %$current;
328 %options = (guesswork => 1, cleanup => 1, convert => 1);
330 if ($element eq 'Data') {
331 $options{guesswork} = 0;
332 $options{cleanup} = 0;
333 $options{convert} = 0;
334 } elsif ($element eq 'X') {
335 $options{guesswork} = 0;
336 $options{cleanup} = 0;
337 } elsif ($element eq 'Verbatim' || $element eq 'C') {
338 $options{guesswork} = 0;
339 $options{literal} = 1;
344 # Format a text block. Takes a hash of formatting options and the text to
345 # format. Currently, the only formatting options are guesswork, cleanup, and
346 # convert, all of which are boolean.
348 my ($self, $options, $text) = @_;
349 my $guesswork = $$options{guesswork} && !$$self{IN_NAME};
350 my $cleanup = $$options{cleanup};
351 my $convert = $$options{convert};
352 my $literal = $$options{literal};
354 # Cleanup just tidies up a few things, telling *roff that the hyphens are
355 # hard, putting a bit of space between consecutive underscores, and
356 # escaping backslashes. Be careful not to mangle our character
357 # translations by doing this before processing character translation.
361 $text =~ s/_(?=_)/_\\|/g;
364 # Normally we do character translation, but we won't even do that in
365 # <Data> blocks or if UTF-8 output is desired.
366 if ($convert && !$$self{utf8} && ASCII) {
367 $text =~ s/([^\x00-\x7F])/$ESCAPES{ord ($1)} || "X"/eg;
370 # Ensure that *roff doesn't convert literal quotes to UTF-8 single quotes,
371 # but don't mess up our accept escapes.
373 $text =~ s/(?<!\\\*)\'/\\*\(Aq/g;
374 $text =~ s/(?<!\\\*)\`/\\\`/g;
377 # If guesswork is asked for, do that. This involves more substantial
378 # formatting based on various heuristics that may only be appropriate for
379 # particular documents.
381 $text = $self->guesswork ($text);
387 # Handles C<> text, deciding whether to put \*C` around it or not. This is a
388 # whole bunch of messy heuristics to try to avoid overquoting, originally from
389 # Barrie Slaymaker. This largely duplicates similar code in Pod::Text.
394 # A regex that matches the portion of a variable reference that's the
395 # array or hash index, separated out just because we want to use it in
396 # several places in the following regex.
397 my $index = '(?: \[.*\] | \{.*\} )?';
399 # Check for things that we don't want to quote, and if we find any of
400 # them, return the string with just a font change and no quoting.
404 ( [\'\`\"] ) .* \1 # already quoted
405 | \\\*\(Aq .* \\\*\(Aq # quoted and escaped
406 | \\?\` .* ( \' | \\\*\(Aq ) # `quoted'
407 | \$+ [\#^]? \S $index # special ($^Foo, $")
408 | [\$\@%&*]+ \#? [:\'\w]+ $index # plain var or func
409 | [\$\@%&*]* [:\'\w]+ (?: -> )? \(\s*[^\s,]\s*\) # 0/1-arg func call
410 | [-+]? ( \d[\d.]* | \.\d+ ) (?: [eE][-+]?\d+ )? # a number
411 | 0x [a-fA-F\d]+ # a hex constant
414 }xso and return '\f(FS' . $_ . '\f(FE';
416 # If we didn't return, go ahead and quote the text.
417 return '\f(FS\*(C`' . $_ . "\\*(C'\\f(FE";
420 # Takes a text block to perform guesswork on. Returns the text block with
421 # formatting codes added. This is the code that marks up various Perl
422 # constructs and things commonly used in man pages without requiring the user
423 # to add any explicit markup, and is applied to all non-literal text. We're
424 # guaranteed that the text we're applying guesswork to does not contain any
425 # *roff formatting codes. Note that the inserted font sequences must be
426 # treated later with mapfonts or textmapfonts.
428 # This method is very fragile, both in the regular expressions it uses and in
429 # the ordering of those modifications. Care and testing is required when
434 DEBUG > 5 and print " Guesswork called on [$_]\n";
436 # By the time we reach this point, all hypens will be escaped by adding a
437 # backslash. We want to undo that escaping if they're part of regular
438 # words and there's only a single dash, since that's a real hyphen that
439 # *roff gets to consider a possible break point. Make sure that a dash
440 # after the first character of a word stays non-breaking, however.
442 # Note that this is not user-controllable; we pretty much have to do this
443 # transformation or *roff will mangle the output in unacceptable ways.
445 ( (?:\G|^|\s) [\(\"]* [a-zA-Z] ) ( \\- )?
446 ( (?: [a-zA-Z\']+ \\-)+ )
447 ( [a-zA-Z\']+ ) (?= [\)\".?!,;:]* (?:\s|\Z|\\\ ) )
450 my ($prefix, $hyphen, $main, $suffix) = ($1, $2, $3, $4);
453 $prefix . $hyphen . $main . $suffix;
456 # Translate "--" into a real em-dash if it's used like one. This means
457 # that it's either surrounded by whitespace, it follows a regular word, or
458 # it occurs between two regular words.
459 if ($$self{MAGIC_EMDASH}) {
460 s{ (\s) \\-\\- (\s) } { $1 . '\*(--' . $2 }egx;
461 s{ (\b[a-zA-Z]+) \\-\\- (\s|\Z|[a-zA-Z]+\b) } { $1 . '\*(--' . $2 }egx;
464 # Make words in all-caps a little bit smaller; they look better that way.
465 # However, we don't want to change Perl code (like @ARGV), nor do we want
466 # to fix the MIME in MIME-Version since it looks weird with the
469 # We change only a string of all caps (2) either at the beginning of the
470 # line or following regular punctuation (like quotes) or whitespace (1),
471 # and followed by either similar punctuation, an em-dash, or the end of
473 if ($$self{MAGIC_SMALLCAPS}) {
475 ( ^ | [\s\(\"\'\`\[\{<>] | \\\ ) # (1)
476 ( [A-Z] [A-Z] (?: [/A-Z+:\d_\$&] | \\- )* ) # (2)
477 (?= [\s>\}\]\(\)\'\".?!,;] | \\*\(-- | \\\ | $ ) # (3)
479 $1 . '\s-1' . $2 . '\s0'
483 # Note that from this point forward, we have to adjust for \s-1 and \s-0
484 # strings inserted around things that we've made small-caps if later
485 # transforms should work on those strings.
487 # Italize functions in the form func(), including functions that are in
488 # all capitals, but don't italize if there's anything between the parens.
489 # The function must start with an alphabetic character or underscore and
490 # then consist of word characters or colons.
491 if ($$self{MAGIC_FUNC}) {
494 ( [A-Za-z_] ([:\w] | \\s-?[01])+ \(\) )
496 $1 . '\f(IS' . $2 . '\f(IE'
500 # Change references to manual pages to put the page name in italics but
501 # the number in the regular font, with a thin space between the name and
502 # the number. Only recognize func(n) where func starts with an alphabetic
503 # character or underscore and contains only word characters, periods (for
504 # configuration file man pages), or colons, and n is a single digit,
505 # optionally followed by some number of lowercase letters. Note that this
506 # does not recognize man page references like perl(l) or socket(3SOCKET).
507 if ($$self{MAGIC_MANREF}) {
510 ( [A-Za-z_] (?:[.:\w] | \\- | \\s-?[01])+ )
513 $1 . '\f(IS' . $2 . '\f(IE\|' . $3
517 # Convert simple Perl variable references to a fixed-width font. Be
518 # careful not to convert functions, though; there are too many subtleties
519 # with them to want to perform this transformation.
520 if ($$self{MAGIC_VARS}) {
526 $1 . '\f(FS' . $2 . '\f(FE'
530 # Fix up double quotes. Unfortunately, we miss this transformation if the
531 # quoted text contains any code with formatting codes and there's not much
532 # we can effectively do about that, which makes it somewhat unclear if
533 # this is really a good idea.
534 s{ \" ([^\"]+) \" } { '\*(L"' . $1 . '\*(R"' }egx;
536 # Make C++ into \*(C+, which is a squinched version.
537 if ($$self{MAGIC_CPP}) {
538 s{ \b C\+\+ } {\\*\(C+}gx;
542 DEBUG > 5 and print " Guesswork returning [$_]\n";
546 ##############################################################################
548 ##############################################################################
550 # When building up the *roff code, we don't use real *roff fonts. Instead, we
551 # embed font codes of the form \f(<font>[SE] where <font> is one of B, I, or
552 # F, S stands for start, and E stands for end. This method turns these into
553 # the right start and end codes.
555 # We add this level of complexity because the old pod2man didn't get code like
556 # B<someI<thing> else> right; after I<> it switched back to normal text rather
557 # than bold. We take care of this by using variables that state whether bold,
558 # italic, or fixed are turned on as a combined pointer to our current font
559 # sequence, and set each to the number of current nestings of start tags for
562 # \fP changes to the previous font, but only one previous font is kept. We
563 # don't know what the outside level font is; normally it's R, but if we're
564 # inside a heading it could be something else. So arrange things so that the
565 # outside font is always the "previous" font and end with \fP instead of \fR.
566 # Idea from Zack Weinberg.
568 my ($self, $text) = @_;
569 my ($fixed, $bold, $italic) = (0, 0, 0);
570 my %magic = (F => \$fixed, B => \$bold, I => \$italic);
577 if ($last ne '\fR') { $sequence = '\fP' }
578 ${ $magic{$1} } += ($2 eq 'S') ? 1 : -1;
579 $f = $$self{FONTS}{ ($fixed && 1) . ($bold && 1) . ($italic && 1) };
583 if ($f ne '\fR') { $sequence .= $f }
591 # Unfortunately, there is a bug in Solaris 2.6 nroff (not present in GNU
592 # groff) where the sequence \fB\fP\f(CW\fP leaves the font set to B rather
593 # than R, presumably because \f(CW doesn't actually do a font change. To work
594 # around this, use a separate textmapfonts for text blocks where the default
595 # font is always R and only use the smart mapfonts for headings.
597 my ($self, $text) = @_;
598 my ($fixed, $bold, $italic) = (0, 0, 0);
599 my %magic = (F => \$fixed, B => \$bold, I => \$italic);
603 ${ $magic{$1} } += ($2 eq 'S') ? 1 : -1;
604 $$self{FONTS}{ ($fixed && 1) . ($bold && 1) . ($italic && 1) };
609 # Given a command and a single argument that may or may not contain double
610 # quotes, handle double-quote formatting for it. If there are no double
611 # quotes, just return the command followed by the argument in double quotes.
612 # If there are double quotes, use an if statement to test for nroff, and for
613 # nroff output the command followed by the argument in double quotes with
614 # embedded double quotes doubled. For other formatters, remap paired double
615 # quotes to LQUOTE and RQUOTE.
617 my ($self, $command, $text, $extra) = @_;
618 $text =~ s/\\\*\([LR]\"/\"/g;
620 # We also have to deal with \*C` and \*C', which are used to add the
621 # quotes around C<> text, since they may expand to " and if they do this
622 # confuses the .SH macros and the like no end. Expand them ourselves.
623 # Also separate troff from nroff if there are any fixed-width fonts in use
624 # to work around problems with Solaris nroff.
625 my $c_is_quote = ($$self{LQUOTE} =~ /\"/) || ($$self{RQUOTE} =~ /\"/);
626 my $fixedpat = join '|', @{ $$self{FONTS} }{'100', '101', '110', '111'};
627 $fixedpat =~ s/\\/\\\\/g;
628 $fixedpat =~ s/\(/\\\(/g;
629 if ($text =~ m/\"/ || $text =~ m/$fixedpat/) {
630 $text =~ s/\"/\"\"/g;
633 $troff =~ s/\"\"([^\"]*)\"\"/\`\`$1\'\'/g;
634 if ($c_is_quote and $text =~ m/\\\*\(C[\'\`]/) {
635 $nroff =~ s/\\\*\(C\`/$$self{LQUOTE}/g;
636 $nroff =~ s/\\\*\(C\'/$$self{RQUOTE}/g;
637 $troff =~ s/\\\*\(C[\'\`]//g;
639 $nroff = qq("$nroff") . ($extra ? " $extra" : '');
640 $troff = qq("$troff") . ($extra ? " $extra" : '');
642 # Work around the Solaris nroff bug where \f(CW\fP leaves the font set
643 # to Roman rather than the actual previous font when used in headings.
644 # troff output may still be broken, but at least we can fix nroff by
645 # just switching the font changes to the non-fixed versions.
646 $nroff =~ s/\Q$$self{FONTS}{100}\E(.*?)\\f[PR]/$1/g;
647 $nroff =~ s/\Q$$self{FONTS}{101}\E(.*?)\\f([PR])/\\fI$1\\f$2/g;
648 $nroff =~ s/\Q$$self{FONTS}{110}\E(.*?)\\f([PR])/\\fB$1\\f$2/g;
649 $nroff =~ s/\Q$$self{FONTS}{111}\E(.*?)\\f([PR])/\\f\(BI$1\\f$2/g;
651 # Now finally output the command. Bother with .ie only if the nroff
652 # and troff output aren't the same.
653 if ($nroff ne $troff) {
654 return ".ie n $command $nroff\n.el $command $troff\n";
656 return "$command $nroff\n";
659 $text = qq("$text") . ($extra ? " $extra" : '');
660 return "$command $text\n";
664 # Protect leading quotes and periods against interpretation as commands. Also
665 # protect anything starting with a backslash, since it could expand or hide
666 # something that *roff would interpret as a command. This is overkill, but
667 # it's much simpler than trying to parse *roff here.
669 my ($self, $text) = @_;
670 $text =~ s/^([.\'\\])/\\&$1/mg;
674 # Make vertical whitespace if NEEDSPACE is set, appropriate to the indentation
675 # level the situation. This function is needed since in *roff one has to
676 # create vertical whitespace after paragraphs and between some things, but
677 # other macros create their own whitespace. Also close out a sequence of
678 # repeated =items, since calling makespace means we're about to begin the item
682 $self->output (".PD\n") if $$self{ITEMS} > 1;
684 $self->output ($$self{INDENT} > 0 ? ".Sp\n" : ".PP\n")
685 if $$self{NEEDSPACE};
688 # Output any pending index entries, and optionally an index entry given as an
689 # argument. Support multiple index entries in X<> separated by slashes, and
690 # strip special escapes from index entries.
692 my ($self, $section, $index) = @_;
693 my @entries = map { split m%\s*/\s*% } @{ $$self{INDEX} };
694 return unless ($section || @entries);
696 # We're about to output all pending entries, so clear our pending queue.
699 # Build the output. Regular index entries are marked Xref, and headings
700 # pass in their own section. Undo some *roff formatting on headings.
703 push @output, [ 'Xref', join (' ', @entries) ];
707 $index =~ s/\\(?:s-?\d|.\(..|.)//g;
708 push @output, [ $section, $index ];
711 # Print out the .IX commands.
713 my ($type, $entry) = @$_;
714 $entry =~ s/\"/\"\"/g;
715 $self->output (".IX $type " . '"' . $entry . '"' . "\n");
719 # Output some text, without any additional changes.
721 my ($self, @text) = @_;
722 print { $$self{output_fh} } @text;
725 ##############################################################################
726 # Document initialization
727 ##############################################################################
729 # Handle the start of the document. Here we handle empty documents, as well
730 # as setting up our basic macros in a preamble and building the page title.
732 my ($self, $attrs) = @_;
733 if ($$attrs{contentless} && !$$self{ALWAYS_EMIT_SOMETHING}) {
734 DEBUG and print "Document is contentless\n";
735 $$self{CONTENTLESS} = 1;
739 # Determine information for the preamble and then output it.
740 my ($name, $section);
741 if (defined $$self{name}) {
742 $name = $$self{name};
743 $section = $$self{section} || 1;
745 ($name, $section) = $self->devise_title;
747 my $date = $$self{date} || $self->devise_date;
748 $self->preamble ($name, $section, $date)
749 unless $self->bare_output or DEBUG > 9;
751 # Initialize a few per-document variables.
752 $$self{INDENT} = 0; # Current indentation level.
753 $$self{INDENTS} = []; # Stack of indentations.
754 $$self{INDEX} = []; # Index keys waiting to be printed.
755 $$self{IN_NAME} = 0; # Whether processing the NAME section.
756 $$self{ITEMS} = 0; # The number of consecutive =items.
757 $$self{ITEMTYPES} = []; # Stack of =item types, one per list.
758 $$self{SHIFTWAIT} = 0; # Whether there is a shift waiting.
759 $$self{SHIFTS} = []; # Stack of .RS shifts.
760 $$self{PENDING} = [[]]; # Pending output.
763 # Handle the end of the document. This does nothing but print out a final
764 # comment at the end of the document under debugging.
767 return if $self->bare_output;
768 return if ($$self{CONTENTLESS} && !$$self{ALWAYS_EMIT_SOMETHING});
769 $self->output (q(.\" [End document]) . "\n") if DEBUG;
772 # Try to figure out the name and section from the file name and return them as
773 # a list, returning an empty name and section 1 if we can't find any better
774 # information. Uses File::Basename and File::Spec as necessary.
777 my $name = $self->source_filename || '';
778 my $section = $$self{section} || 1;
779 $section = 3 if (!$$self{section} && $name =~ /\.pm\z/i);
780 $name =~ s/\.p(od|[lm])\z//i;
782 # If the section isn't 3, then the name defaults to just the basename of
783 # the file. Otherwise, assume we're dealing with a module. We want to
784 # figure out the full module name from the path to the file, but we don't
785 # want to include too much of the path into the module name. Lose
786 # anything up to the first off:
788 # */lib/*perl*/ standard or site_perl module
789 # */*perl*/lib/ from -Dprefix=/opt/perl
790 # */*perl*/ random module hierarchy
792 # which works. Also strip off a leading site, site_perl, or vendor_perl
793 # component, any OS-specific component, and any version number component,
794 # and strip off an initial component of "lib" or "blib/lib" since that's
795 # what ExtUtils::MakeMaker creates. splitdir requires at least File::Spec
797 if ($section !~ /^3/) {
798 require File::Basename;
799 $name = uc File::Basename::basename ($name);
802 my ($volume, $dirs, $file) = File::Spec->splitpath ($name);
803 my @dirs = File::Spec->splitdir ($dirs);
806 for ($i = 0; $i < @dirs; $i++) {
807 if ($dirs[$i] =~ /perl/) {
809 $cut++ if ($dirs[$i + 1] && $dirs[$i + 1] eq 'lib');
814 splice (@dirs, 0, $cut);
815 shift @dirs if ($dirs[0] =~ /^(site|vendor)(_perl)?$/);
816 shift @dirs if ($dirs[0] =~ /^[\d.]+$/);
817 shift @dirs if ($dirs[0] =~ /^(.*-$^O|$^O-.*|$^O)$/);
819 shift @dirs if $dirs[0] eq 'lib';
820 splice (@dirs, 0, 2) if ($dirs[0] eq 'blib' && $dirs[1] eq 'lib');
822 # Remove empty directories when building the module name; they
823 # occur too easily on Unix by doubling slashes.
824 $name = join ('::', (grep { $_ ? $_ : () } @dirs), $file);
826 return ($name, $section);
829 # Determine the modification date and return that, properly formatted in ISO
830 # format. If we can't get the modification date of the input, instead use the
831 # current time. Pod::Simple returns a completely unuseful stringified file
832 # handle as the source_filename for input from a file handle, so we have to
833 # deal with that as well.
836 my $input = $self->source_filename;
839 $time = (stat $input)[9] || time;
843 return strftime ('%Y-%m-%d', localtime $time);
846 # Print out the preamble and the title. The meaning of the arguments to .TH
847 # unfortunately vary by system; some systems consider the fourth argument to
848 # be a "source" and others use it as a version number. Generally it's just
849 # presented as the left-side footer, though, so it doesn't matter too much if
850 # a particular system gives it another interpretation.
852 # The order of date and release used to be reversed in older versions of this
853 # module, but this order is correct for both Solaris and Linux.
855 my ($self, $name, $section, $date) = @_;
856 my $preamble = $self->preamble_template (!$$self{utf8});
858 # Build the index line and make sure that it will be syntactically valid.
859 my $index = "$name $section";
860 $index =~ s/\"/\"\"/g;
862 # If name or section contain spaces, quote them (section really never
863 # should, but we may as well be cautious).
864 for ($name, $section) {
871 # Double quotes in date, since it will be quoted.
872 $date =~ s/\"/\"\"/g;
874 # Substitute into the preamble the configuration options.
875 $preamble =~ s/\@CFONT\@/$$self{fixed}/;
876 $preamble =~ s/\@LQUOTE\@/$$self{LQUOTE}/;
877 $preamble =~ s/\@RQUOTE\@/$$self{RQUOTE}/;
880 # Get the version information.
881 my $version = $self->version_report;
883 # Finally output everything.
884 $self->output (<<"----END OF HEADER----");
885 .\\" Automatically generated by $version
887 .\\" Standard preamble:
888 .\\" ========================================================================
890 .\\" ========================================================================
893 .TH $name $section "$date" "$$self{release}" "$$self{center}"
894 .\\" For nroff, turn off justification. Always turn off hyphenation; it makes
895 .\\" way too many mistakes in technical documents.
898 ----END OF HEADER----
899 $self->output (".\\\" [End of preamble]\n") if DEBUG;
902 ##############################################################################
904 ##############################################################################
906 # Handle a basic block of text. The only tricky part of this is if this is
907 # the first paragraph of text after an =over, in which case we have to change
908 # indentations for *roff.
910 my ($self, $attrs, $text) = @_;
911 my $line = $$attrs{start_line};
913 # Output the paragraph. We also have to handle =over without =item. If
914 # there's an =over without =item, SHIFTWAIT will be set, and we need to
915 # handle creation of the indent here. Add the shift to SHIFTS so that it
916 # will be cleaned up on =back.
918 if ($$self{SHIFTWAIT}) {
919 $self->output (".RS $$self{INDENT}\n");
920 push (@{ $$self{SHIFTS} }, $$self{INDENT});
921 $$self{SHIFTWAIT} = 0;
924 # Add the line number for debugging, but not in the NAME section just in
925 # case the comment would confuse apropos.
926 $self->output (".\\\" [At source line $line]\n")
927 if defined ($line) && DEBUG && !$$self{IN_NAME};
929 # Force exactly one newline at the end and strip unwanted trailing
930 # whitespace at the end.
933 # Output the paragraph.
934 $self->output ($self->protect ($self->textmapfonts ($text)));
936 $$self{NEEDSPACE} = 1;
940 # Handle a verbatim paragraph. Put a null token at the beginning of each line
941 # to protect against commands and wrap in .Vb/.Ve (which we define in our
944 my ($self, $attrs, $text) = @_;
946 # Ignore an empty verbatim paragraph.
947 return unless $text =~ /\S/;
949 # Force exactly one newline at the end and strip unwanted trailing
950 # whitespace at the end.
953 # Get a count of the number of lines before the first blank line, which
954 # we'll pass to .Vb as its parameter. This tells *roff to keep that many
955 # lines together. We don't want to tell *roff to keep huge blocks
957 my @lines = split (/\n/, $text);
963 $unbroken = 10 if ($unbroken > 12 && !$$self{MAGIC_VNOPAGEBREAK_LIMIT});
965 # Prepend a null token to each line.
968 # Output the results.
970 $self->output (".Vb $unbroken\n$text.Ve\n");
971 $$self{NEEDSPACE} = 1;
975 # Handle literal text (produced by =for and similar constructs). Just output
976 # it with the minimum of changes.
978 my ($self, $attrs, $text) = @_;
980 $text =~ s/\n{0,2}$/\n/;
981 $self->output ($text);
985 ##############################################################################
987 ##############################################################################
989 # Common code for all headings. This is called before the actual heading is
990 # output. It returns the cleaned up heading text (putting the heading all on
991 # one line) and may do other things, like closing bad =item blocks.
993 my ($self, $text, $line) = @_;
995 $text =~ s/\s*\n\s*/ /g;
997 # This should never happen; it means that we have a heading after =item
998 # without an intervening =back. But just in case, handle it anyway.
999 if ($$self{ITEMS} > 1) {
1001 $self->output (".PD\n");
1004 # Output the current source line.
1005 $self->output ( ".\\\" [At source line $line]\n" )
1006 if defined ($line) && DEBUG;
1010 # First level heading. We can't output .IX in the NAME section due to a bug
1011 # in some versions of catman, so don't output a .IX for that section. .SH
1012 # already uses small caps, so remove \s0 and \s-1. Maintain IN_NAME as
1015 my ($self, $attrs, $text) = @_;
1016 $text =~ s/\\s-?\d//g;
1017 $text = $self->heading_common ($text, $$attrs{start_line});
1018 my $isname = ($text eq 'NAME' || $text =~ /\(NAME\)/);
1019 $self->output ($self->switchquotes ('.SH', $self->mapfonts ($text)));
1020 $self->outindex ('Header', $text) unless $isname;
1021 $$self{NEEDSPACE} = 0;
1022 $$self{IN_NAME} = $isname;
1026 # Second level heading.
1028 my ($self, $attrs, $text) = @_;
1029 $text = $self->heading_common ($text, $$attrs{start_line});
1030 $self->output ($self->switchquotes ('.SS', $self->mapfonts ($text)));
1031 $self->outindex ('Subsection', $text);
1032 $$self{NEEDSPACE} = 0;
1036 # Third level heading. *roff doesn't have this concept, so just put the
1037 # heading in italics as a normal paragraph.
1039 my ($self, $attrs, $text) = @_;
1040 $text = $self->heading_common ($text, $$attrs{start_line});
1042 $self->output ($self->textmapfonts ('\f(IS' . $text . '\f(IE') . "\n");
1043 $self->outindex ('Subsection', $text);
1044 $$self{NEEDSPACE} = 1;
1048 # Fourth level heading. *roff doesn't have this concept, so just put the
1049 # heading as a normal paragraph.
1051 my ($self, $attrs, $text) = @_;
1052 $text = $self->heading_common ($text, $$attrs{start_line});
1054 $self->output ($self->textmapfonts ($text) . "\n");
1055 $self->outindex ('Subsection', $text);
1056 $$self{NEEDSPACE} = 1;
1060 ##############################################################################
1062 ##############################################################################
1064 # All of the formatting codes that aren't handled internally by the parser,
1065 # other than L<> and X<>.
1066 sub cmd_b { return '\f(BS' . $_[2] . '\f(BE' }
1067 sub cmd_i { return '\f(IS' . $_[2] . '\f(IE' }
1068 sub cmd_f { return '\f(IS' . $_[2] . '\f(IE' }
1069 sub cmd_c { return $_[0]->quote_literal ($_[2]) }
1071 # Index entries are just added to the pending entries.
1073 my ($self, $attrs, $text) = @_;
1074 push (@{ $$self{INDEX} }, $text);
1078 # Links reduce to the text that we're given, wrapped in angle brackets if it's
1081 my ($self, $attrs, $text) = @_;
1082 return $$attrs{type} eq 'url' ? "<$text>" : $text;
1085 ##############################################################################
1087 ##############################################################################
1089 # Handle the beginning of an =over block. Takes the type of the block as the
1090 # first argument, and then the attr hash. This is called by the handlers for
1091 # the four different types of lists (bullet, number, text, and block).
1092 sub over_common_start {
1093 my ($self, $type, $attrs) = @_;
1094 my $line = $$attrs{start_line};
1095 my $indent = $$attrs{indent};
1096 DEBUG > 3 and print " Starting =over $type (line $line, indent ",
1097 ($indent || '?'), "\n";
1099 # Find the indentation level.
1100 unless (defined ($indent) && $indent =~ /^[-+]?\d{1,4}\s*$/) {
1101 $indent = $$self{indent};
1104 # If we've gotten multiple indentations in a row, we need to emit the
1105 # pending indentation for the last level that we saw and haven't acted on
1106 # yet. SHIFTS is the stack of indentations that we've actually emitted
1108 if (@{ $$self{SHIFTS} } < @{ $$self{INDENTS} }) {
1109 $self->output (".RS $$self{INDENT}\n");
1110 push (@{ $$self{SHIFTS} }, $$self{INDENT});
1113 # Now, do record-keeping. INDENTS is a stack of indentations that we've
1114 # seen so far, and INDENT is the current level of indentation. ITEMTYPES
1115 # is a stack of list types that we've seen.
1116 push (@{ $$self{INDENTS} }, $$self{INDENT});
1117 push (@{ $$self{ITEMTYPES} }, $type);
1118 $$self{INDENT} = $indent + 0;
1119 $$self{SHIFTWAIT} = 1;
1122 # End an =over block. Takes no options other than the class pointer.
1123 # Normally, once we close a block and therefore remove something from INDENTS,
1124 # INDENTS will now be longer than SHIFTS, indicating that we also need to emit
1125 # *roff code to close the indent. This isn't *always* true, depending on the
1126 # circumstance. If we're still inside an indentation, we need to emit another
1127 # .RE and then a new .RS to unconfuse *roff.
1128 sub over_common_end {
1130 DEBUG > 3 and print " Ending =over\n";
1131 $$self{INDENT} = pop @{ $$self{INDENTS} };
1132 pop @{ $$self{ITEMTYPES} };
1134 # If we emitted code for that indentation, end it.
1135 if (@{ $$self{SHIFTS} } > @{ $$self{INDENTS} }) {
1136 $self->output (".RE\n");
1137 pop @{ $$self{SHIFTS} };
1140 # If we're still in an indentation, *roff will have now lost track of the
1141 # right depth of that indentation, so fix that.
1142 if (@{ $$self{INDENTS} } > 0) {
1143 $self->output (".RE\n");
1144 $self->output (".RS $$self{INDENT}\n");
1146 $$self{NEEDSPACE} = 1;
1147 $$self{SHIFTWAIT} = 0;
1150 # Dispatch the start and end calls as appropriate.
1151 sub start_over_bullet { my $s = shift; $s->over_common_start ('bullet', @_) }
1152 sub start_over_number { my $s = shift; $s->over_common_start ('number', @_) }
1153 sub start_over_text { my $s = shift; $s->over_common_start ('text', @_) }
1154 sub start_over_block { my $s = shift; $s->over_common_start ('block', @_) }
1155 sub end_over_bullet { $_[0]->over_common_end }
1156 sub end_over_number { $_[0]->over_common_end }
1157 sub end_over_text { $_[0]->over_common_end }
1158 sub end_over_block { $_[0]->over_common_end }
1160 # The common handler for all item commands. Takes the type of the item, the
1161 # attributes, and then the text of the item.
1163 # Emit an index entry for anything that's interesting, but don't emit index
1164 # entries for things like bullets and numbers. Newlines in an item title are
1165 # turned into spaces since *roff can't handle them embedded.
1167 my ($self, $type, $attrs, $text) = @_;
1168 my $line = $$attrs{start_line};
1169 DEBUG > 3 and print " $type item (line $line): $text\n";
1171 # Clean up the text. We want to end up with two variables, one ($text)
1172 # which contains any body text after taking out the item portion, and
1173 # another ($item) which contains the actual item text.
1176 if ($type eq 'bullet') {
1178 $text =~ s/\n*$/\n/;
1179 } elsif ($type eq 'number') {
1180 $item = $$attrs{number} . '.';
1183 $item =~ s/\s*\n\s*/ /g;
1185 $index = $item if ($item =~ /\w/);
1188 # Take care of the indentation. If shifts and indents are equal, close
1189 # the top shift, since we're about to create an indentation with .IP.
1190 # Also output .PD 0 to turn off spacing between items if this item is
1191 # directly following another one. We only have to do that once for a
1192 # whole chain of items so do it for the second item in the change. Note
1193 # that makespace is what undoes this.
1194 if (@{ $$self{SHIFTS} } == @{ $$self{INDENTS} }) {
1195 $self->output (".RE\n");
1196 pop @{ $$self{SHIFTS} };
1198 $self->output (".PD 0\n") if ($$self{ITEMS} == 1);
1200 # Now, output the item tag itself.
1201 $item = $self->textmapfonts ($item);
1202 $self->output ($self->switchquotes ('.IP', $item, $$self{INDENT}));
1203 $$self{NEEDSPACE} = 0;
1205 $$self{SHIFTWAIT} = 0;
1207 # If body text for this item was included, go ahead and output that now.
1209 $text =~ s/\s*$/\n/;
1211 $self->output ($self->protect ($self->textmapfonts ($text)));
1212 $$self{NEEDSPACE} = 1;
1214 $self->outindex ($index ? ('Item', $index) : ());
1217 # Dispatch the item commands to the appropriate place.
1218 sub cmd_item_bullet { my $self = shift; $self->item_common ('bullet', @_) }
1219 sub cmd_item_number { my $self = shift; $self->item_common ('number', @_) }
1220 sub cmd_item_text { my $self = shift; $self->item_common ('text', @_) }
1221 sub cmd_item_block { my $self = shift; $self->item_common ('block', @_) }
1223 ##############################################################################
1224 # Backward compatibility
1225 ##############################################################################
1227 # Reset the underlying Pod::Simple object between calls to parse_from_file so
1228 # that the same object can be reused to convert multiple pages.
1229 sub parse_from_file {
1233 # Fake the old cutting option to Pod::Parser. This fiddings with internal
1234 # Pod::Simple state and is quite ugly; we need a better approach.
1235 if (ref ($_[0]) eq 'HASH') {
1236 my $opts = shift @_;
1237 if (defined ($$opts{-cutting}) && !$$opts{-cutting}) {
1239 $$self{last_was_blank} = 1;
1244 my $retval = $self->SUPER::parse_from_file (@_);
1246 # Flush output, since Pod::Simple doesn't do this. Ideally we should also
1247 # close the file descriptor if we had to open one, but we can't easily
1249 my $fh = $self->output_fh ();
1250 my $oldfh = select $fh;
1259 # Pod::Simple failed to provide this backward compatibility function, so
1260 # implement it ourselves. File handles are one of the inputs that
1261 # parse_from_file supports.
1262 sub parse_from_filehandle {
1264 $self->parse_from_file (@_);
1267 ##############################################################################
1268 # Translation tables
1269 ##############################################################################
1271 # The following table is adapted from Tom Christiansen's pod2man. It assumes
1272 # that the standard preamble has already been printed, since that's what
1273 # defines all of the accent marks. We really want to do something better than
1274 # this when *roff actually supports other character sets itself, since these
1275 # results are pretty poor.
1277 # This only works in an ASCII world. What to do in a non-ASCII world is very
1278 # unclear -- hopefully we can assume UTF-8 and just leave well enough alone.
1279 @ESCAPES{0xA0 .. 0xFF} = (
1280 "\\ ", undef, undef, undef, undef, undef, undef, undef,
1281 undef, undef, undef, undef, undef, "\\%", undef, undef,
1283 undef, undef, undef, undef, undef, undef, undef, undef,
1284 undef, undef, undef, undef, undef, undef, undef, undef,
1286 "A\\*`", "A\\*'", "A\\*^", "A\\*~", "A\\*:", "A\\*o", "\\*(AE", "C\\*,",
1287 "E\\*`", "E\\*'", "E\\*^", "E\\*:", "I\\*`", "I\\*'", "I\\*^", "I\\*:",
1289 "\\*(D-", "N\\*~", "O\\*`", "O\\*'", "O\\*^", "O\\*~", "O\\*:", undef,
1290 "O\\*/", "U\\*`", "U\\*'", "U\\*^", "U\\*:", "Y\\*'", "\\*(Th", "\\*8",
1292 "a\\*`", "a\\*'", "a\\*^", "a\\*~", "a\\*:", "a\\*o", "\\*(ae", "c\\*,",
1293 "e\\*`", "e\\*'", "e\\*^", "e\\*:", "i\\*`", "i\\*'", "i\\*^", "i\\*:",
1295 "\\*(d-", "n\\*~", "o\\*`", "o\\*'", "o\\*^", "o\\*~", "o\\*:", undef,
1296 "o\\*/" , "u\\*`", "u\\*'", "u\\*^", "u\\*:", "y\\*'", "\\*(th", "y\\*:",
1299 ##############################################################################
1301 ##############################################################################
1303 # The following is the static preamble which starts all *roff output we
1304 # generate. Most is static except for the font to use as a fixed-width font,
1305 # which is designed by @CFONT@, and the left and right quotes to use for C<>
1306 # text, designated by @LQOUTE@ and @RQUOTE@. However, the second part, which
1307 # defines the accent marks, is only used if $escapes is set to true.
1308 sub preamble_template {
1309 my ($self, $accents) = @_;
1310 my $preamble = <<'----END OF PREAMBLE----';
1311 .de Sp \" Vertical space (when we can't use .PP)
1315 .de Vb \" Begin verbatim text
1320 .de Ve \" End verbatim text
1324 .\" Set up some character translations and predefined strings. \*(-- will
1325 .\" give an unbreakable dash, \*(PI will give pi, \*(L" will give a left
1326 .\" double quote, and \*(R" will give a right double quote. \*(C+ will
1327 .\" give a nicer C++. Capital omega is used to do unbreakable dashes and
1328 .\" therefore won't be available. \*(C` and \*(C' expand to `' in nroff,
1329 .\" nothing in troff, for use with C<>.
1331 .ds C+ C\v'-.1v'\h'-1p'\s-2+\h'-1p'+\s0\v'.1v'\h'-1p'
1335 . if (\n(.H=4u)&(1m=24u) .ds -- \(*W\h'-12u'\(*W\h'-12u'-\" diablo 10 pitch
1336 . if (\n(.H=4u)&(1m=20u) .ds -- \(*W\h'-12u'\(*W\h'-8u'-\" diablo 12 pitch
1349 .\" Escape single quotes in literal strings from groff's Unicode transform.
1350 .ie \n(.g .ds Aq \(aq
1353 .\" If the F register is turned on, we'll generate index entries on stderr for
1354 .\" titles (.TH), headers (.SH), subsections (.SS), items (.Ip), and index
1355 .\" entries marked with X<> in POD. Of course, you'll have to process the
1356 .\" output yourself in some meaningful fashion.
1359 . tm Index:\\$1\t\\n%\t"\\$2"
1368 ----END OF PREAMBLE----
1371 $preamble .= <<'----END OF PREAMBLE----'
1373 .\" Accent mark definitions (@(#)ms.acc 1.5 88/02/08 SMI; from UCB 4.2).
1374 .\" Fear. Run. Save yourself. No user-serviceable parts.
1375 . \" fudge factors for nroff and troff
1384 . ds #H ((1u-(\\\\n(.fu%2u))*.13m)
1390 . \" simple accents for nroff and troff
1400 . ds ' \\k:\h'-(\\n(.wu*8/10-\*(#H)'\'\h"|\\n:u"
1401 . ds ` \\k:\h'-(\\n(.wu*8/10-\*(#H)'\`\h'|\\n:u'
1402 . ds ^ \\k:\h'-(\\n(.wu*10/11-\*(#H)'^\h'|\\n:u'
1403 . ds , \\k:\h'-(\\n(.wu*8/10)',\h'|\\n:u'
1404 . ds ~ \\k:\h'-(\\n(.wu-\*(#H-.1m)'~\h'|\\n:u'
1405 . ds / \\k:\h'-(\\n(.wu*8/10-\*(#H)'\z\(sl\h'|\\n:u'
1407 . \" troff and (daisy-wheel) nroff accents
1408 .ds : \\k:\h'-(\\n(.wu*8/10-\*(#H+.1m+\*(#F)'\v'-\*(#V'\z.\h'.2m+\*(#F'.\h'|\\n:u'\v'\*(#V'
1409 .ds 8 \h'\*(#H'\(*b\h'-\*(#H'
1410 .ds o \\k:\h'-(\\n(.wu+\w'\(de'u-\*(#H)/2u'\v'-.3n'\*(#[\z\(de\v'.3n'\h'|\\n:u'\*(#]
1411 .ds d- \h'\*(#H'\(pd\h'-\w'~'u'\v'-.25m'\f2\(hy\fP\v'.25m'\h'-\*(#H'
1412 .ds D- D\\k:\h'-\w'D'u'\v'-.11m'\z\(hy\v'.11m'\h'|\\n:u'
1413 .ds th \*(#[\v'.3m'\s+1I\s-1\v'-.3m'\h'-(\w'I'u*2/3)'\s-1o\s+1\*(#]
1414 .ds Th \*(#[\s+2I\s-2\h'-\w'I'u*3/5'\v'-.3m'o\v'.3m'\*(#]
1415 .ds ae a\h'-(\w'a'u*4/10)'e
1416 .ds Ae A\h'-(\w'A'u*4/10)'E
1417 . \" corrections for vroff
1418 .if v .ds ~ \\k:\h'-(\\n(.wu*9/10-\*(#H)'\s-2\u~\d\s+2\h'|\\n:u'
1419 .if v .ds ^ \\k:\h'-(\\n(.wu*10/11-\*(#H)'\v'-.4m'^\v'.4m'\h'|\\n:u'
1420 . \" for low resolution devices (crt and lpr)
1421 .if \n(.H>23 .if \n(.V>19 \
1433 .rm #[ #] #H #V #F C
1434 ----END OF PREAMBLE----
1440 ##############################################################################
1441 # Module return value and documentation
1442 ##############################################################################
1449 Pod::Man - Convert POD data to formatted *roff input
1452 en em ALLCAPS teeny fixedbold fixeditalic fixedbolditalic stderr utf8
1453 UTF-8 Allbery Sean Burke Ossanna Solaris formatters troff uppercased
1459 my $parser = Pod::Man->new (release => $VERSION, section => 8);
1461 # Read POD from STDIN and write to STDOUT.
1462 $parser->parse_file (\*STDIN);
1464 # Read POD from file.pod and write to file.1.
1465 $parser->parse_from_file ('file.pod', 'file.1');
1469 Pod::Man is a module to convert documentation in the POD format (the
1470 preferred language for documenting Perl) into *roff input using the man
1471 macro set. The resulting *roff code is suitable for display on a terminal
1472 using L<nroff(1)>, normally via L<man(1)>, or printing using L<troff(1)>.
1473 It is conventionally invoked using the driver script B<pod2man>, but it can
1474 also be used directly.
1476 As a derived class from Pod::Simple, Pod::Man supports the same methods and
1477 interfaces. See L<Pod::Simple> for all the details.
1479 new() can take options, in the form of key/value pairs that control the
1480 behavior of the parser. See below for details.
1482 If no options are given, Pod::Man uses the name of the input file with any
1483 trailing C<.pod>, C<.pm>, or C<.pl> stripped as the man page title, to
1484 section 1 unless the file ended in C<.pm> in which case it defaults to
1485 section 3, to a centered title of "User Contributed Perl Documentation", to
1486 a centered footer of the Perl version it is run with, and to a left-hand
1487 footer of the modification date of its input (or the current date if given
1488 C<STDIN> for input).
1490 Pod::Man assumes that your *roff formatters have a fixed-width font named
1491 C<CW>. If yours is called something else (like C<CR>), use the C<fixed>
1492 option to specify it. This generally only matters for troff output for
1493 printing. Similarly, you can set the fonts used for bold, italic, and
1494 bold italic fixed-width output.
1496 Besides the obvious pod conversions, Pod::Man also takes care of
1497 formatting func(), func(3), and simple variable references like $foo or
1498 @bar so you don't have to use code escapes for them; complex expressions
1499 like C<$fred{'stuff'}> will still need to be escaped, though. It also
1500 translates dashes that aren't used as hyphens into en dashes, makes long
1501 dashes--like this--into proper em dashes, fixes "paired quotes," makes C++
1502 look right, puts a little space between double underscores, makes ALLCAPS
1503 a teeny bit smaller in B<troff>, and escapes stuff that *roff treats as
1504 special so that you don't have to.
1506 The recognized options to new() are as follows. All options take a single
1513 Sets the centered page header to use instead of "User Contributed Perl
1518 Sets the left-hand footer. By default, the modification date of the input
1519 file will be used, or the current date if stat() can't find that file (the
1520 case if the input is from C<STDIN>), and the date will be formatted as
1525 The fixed-width font to use for verbatim text and code. Defaults to
1526 C<CW>. Some systems may want C<CR> instead. Only matters for B<troff>
1531 Bold version of the fixed-width font. Defaults to C<CB>. Only matters
1532 for B<troff> output.
1536 Italic version of the fixed-width font (actually, something of a misnomer,
1537 since most fixed-width fonts only have an oblique version, not an italic
1538 version). Defaults to C<CI>. Only matters for B<troff> output.
1540 =item fixedbolditalic
1542 Bold italic (probably actually oblique) version of the fixed-width font.
1543 Pod::Man doesn't assume you have this, and defaults to C<CB>. Some
1544 systems (such as Solaris) have this font available as C<CX>. Only matters
1545 for B<troff> output.
1549 Set the name of the manual page. Without this option, the manual name is
1550 set to the uppercased base name of the file being converted unless the
1551 manual section is 3, in which case the path is parsed to see if it is a Perl
1552 module path. If it is, a path like C<.../lib/Pod/Man.pm> is converted into
1553 a name like C<Pod::Man>. This option, if given, overrides any automatic
1554 determination of the name.
1558 Sets the quote marks used to surround CE<lt>> text. If the value is a
1559 single character, it is used as both the left and right quote; if it is two
1560 characters, the first character is used as the left quote and the second as
1561 the right quoted; and if it is four characters, the first two are used as
1562 the left quote and the second two as the right quote.
1564 This may also be set to the special value C<none>, in which case no quote
1565 marks are added around CE<lt>> text (but the font is still changed for troff
1570 Set the centered footer. By default, this is the version of Perl you run
1571 Pod::Man under. Note that some system an macro sets assume that the
1572 centered footer will be a modification date and will prepend something like
1573 "Last modified: "; if this is the case, you may want to set C<release> to
1574 the last modified date and C<date> to the version number.
1578 Set the section for the C<.TH> macro. The standard section numbering
1579 convention is to use 1 for user commands, 2 for system calls, 3 for
1580 functions, 4 for devices, 5 for file formats, 6 for games, 7 for
1581 miscellaneous information, and 8 for administrator commands. There is a lot
1582 of variation here, however; some systems (like Solaris) use 4 for file
1583 formats, 5 for miscellaneous information, and 7 for devices. Still others
1584 use 1m instead of 8, or some mix of both. About the only section numbers
1585 that are reliably consistent are 1, 2, and 3.
1587 By default, section 1 will be used unless the file ends in C<.pm> in which
1588 case section 3 will be selected.
1592 Send error messages about invalid POD to standard error instead of
1593 appending a POD ERRORS section to the generated *roff output.
1597 By default, Pod::Man produces the most conservative possible *roff output
1598 to try to ensure that it will work with as many different *roff
1599 implementations as possible. Many *roff implementations cannot handle
1600 non-ASCII characters, so this means all non-ASCII characters are converted
1601 either to a *roff escape sequence that tries to create a properly accented
1602 character (at least for troff output) or to C<X>.
1604 If this option is set, Pod::Man will instead output UTF-8. If your *roff
1605 implementation can handle it, this is the best output format to use and
1606 avoids corruption of documents containing non-ASCII characters. However,
1607 be warned that *roff source with literal UTF-8 characters is not supported
1608 by many implementations and may even result in segfaults and other bad
1613 The standard Pod::Simple method parse_file() takes one argument naming the
1614 POD file to read from. By default, the output is sent to C<STDOUT>, but
1615 this can be changed with the output_fd() method.
1617 The standard Pod::Simple method parse_from_file() takes up to two
1618 arguments, the first being the input file to read POD from and the second
1619 being the file to write the formatted output to.
1621 You can also call parse_lines() to parse an array of lines or
1622 parse_string_document() to parse a document already in memory. To put the
1623 output into a string instead of a file handle, call the output_string()
1624 method. See L<Pod::Simple> for the specific details.
1630 =item roff font should be 1 or 2 chars, not "%s"
1632 (F) You specified a *roff font (using C<fixed>, C<fixedbold>, etc.) that
1633 wasn't either one or two characters. Pod::Man doesn't support *roff fonts
1634 longer than two characters, although some *roff extensions do (the canonical
1635 versions of B<nroff> and B<troff> don't either).
1637 =item Invalid quote specification "%s"
1639 (F) The quote specification given (the quotes option to the constructor) was
1640 invalid. A quote specification must be one, two, or four characters long.
1646 There is currently no way to turn off the guesswork that tries to format
1647 unmarked text appropriately, and sometimes it isn't wanted (particularly
1648 when using POD to document something other than Perl). Most of the work
1649 towards fixing this has now been done, however, and all that's still needed
1650 is a user interface.
1652 The NAME section should be recognized specially and index entries emitted
1653 for everything in that section. This would have to be deferred until the
1654 next section, since extraneous things in NAME tends to confuse various man
1655 page processors. Currently, no index entries are emitted for anything in
1658 Pod::Man doesn't handle font names longer than two characters. Neither do
1659 most B<troff> implementations, but GNU troff does as an extension. It would
1660 be nice to support as an option for those who want to use it.
1662 The preamble added to each output file is rather verbose, and most of it
1663 is only necessary in the presence of non-ASCII characters. It would
1664 ideally be nice if all of those definitions were only output if needed,
1665 perhaps on the fly as the characters are used.
1667 Pod::Man is excessively slow.
1671 The handling of hyphens and em dashes is somewhat fragile, and one may get
1672 the wrong one under some circumstances. This should only matter for
1675 When and whether to use small caps is somewhat tricky, and Pod::Man doesn't
1676 necessarily get it right.
1678 Converting neutral double quotes to properly matched double quotes doesn't
1679 work unless there are no formatting codes between the quote marks. This
1680 only matters for troff output.
1684 Russ Allbery <rra@stanford.edu>, based I<very> heavily on the original
1685 B<pod2man> by Tom Christiansen <tchrist@mox.perl.com>. The modifications to
1686 work with Pod::Simple instead of Pod::Parser were originally contributed by
1687 Sean Burke (but I've since hacked them beyond recognition and all bugs are
1690 =head1 COPYRIGHT AND LICENSE
1692 Copyright 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008
1693 Russ Allbery <rra@stanford.edu>.
1695 This program is free software; you may redistribute it and/or modify it
1696 under the same terms as Perl itself.
1700 L<Pod::Simple>, L<perlpod(1)>, L<pod2man(1)>, L<nroff(1)>, L<troff(1)>,
1701 L<man(1)>, L<man(7)>
1703 Ossanna, Joseph F., and Brian W. Kernighan. "Troff User's Manual,"
1704 Computing Science Technical Report No. 54, AT&T Bell Laboratories. This is
1705 the best documentation of standard B<nroff> and B<troff>. At the time of
1706 this writing, it's available at
1707 L<http://www.cs.bell-labs.com/cm/cs/cstr.html>.
1709 The man page documenting the man macro set may be L<man(5)> instead of
1710 L<man(7)> on your system. Also, please see L<pod2man(1)> for extensive
1711 documentation on writing manual pages if you've not done it before and
1712 aren't familiar with the conventions.
1714 The current version of this module is always available from its web site at
1715 L<http://www.eyrie.org/~eagle/software/podlators/>. It is also part of the
1716 Perl core distribution as of 5.6.0.