1 # Pod::Text -- Convert POD data to formatted ASCII text.
3 # Copyright 1999, 2000, 2001, 2002, 2004, 2006, 2008
4 # Russ Allbery <rra@stanford.edu>
6 # This program is free software; you may redistribute it and/or modify it
7 # under the same terms as Perl itself.
9 # This module converts POD to formatted text. It replaces the old Pod::Text
10 # module that came with versions of Perl prior to 5.6.0 and attempts to match
11 # its output except for some specific circumstances where other decisions
12 # seemed to produce better output. It uses Pod::Parser and is designed to be
13 # very easy to subclass.
15 # Perl core hackers, please note that this module is also separately
16 # maintained outside of the Perl core as part of the podlators. Please send
17 # me any patches at the address above in addition to sending them to the
18 # standard Perl mailing lists.
20 ##############################################################################
21 # Modules and declarations
22 ##############################################################################
29 use vars qw(@ISA @EXPORT %ESCAPES $VERSION);
31 use Carp qw(carp croak);
35 @ISA = qw(Pod::Simple Exporter);
37 # We have to export pod2text for backward compatibility.
38 @EXPORT = qw(pod2text);
42 ##############################################################################
44 ##############################################################################
46 # This function handles code blocks. It's registered as a callback to
47 # Pod::Simple and therefore doesn't work as a regular method call, but all it
48 # does is call output_code with the line.
50 my ($line, $number, $parser) = @_;
51 $parser->output_code ($line . "\n");
54 # Initialize the object and set various Pod::Simple options that we need.
55 # Here, we also process any additional options passed to the constructor or
56 # set up defaults if none were given. Note that all internal object keys are
57 # in all-caps, reserving all lower-case object keys for Pod::Simple and user
61 my $self = $class->SUPER::new;
63 # Tell Pod::Simple to handle S<> by automatically inserting .
64 $self->nbsp_for_S (1);
66 # Tell Pod::Simple to keep whitespace whenever possible.
67 if ($self->can ('preserve_whitespace')) {
68 $self->preserve_whitespace (1);
70 $self->fullstop_space_harden (1);
73 # The =for and =begin targets that we accept.
74 $self->accept_targets (qw/text TEXT/);
76 # Ensure that contiguous blocks of code are merged together. Otherwise,
77 # some of the guesswork heuristics don't work right.
78 $self->merge_text (1);
80 # Pod::Simple doesn't do anything useful with our arguments, but we want
81 # to put them in our object as hash keys and values. This could cause
82 # problems if we ever clash with Pod::Simple's own internal class
85 my @opts = map { ("opt_$_", $opts{$_}) } keys %opts;
86 %$self = (%$self, @opts);
88 # Initialize various things from our parameters.
89 $$self{opt_alt} = 0 unless defined $$self{opt_alt};
90 $$self{opt_indent} = 4 unless defined $$self{opt_indent};
91 $$self{opt_margin} = 0 unless defined $$self{opt_margin};
92 $$self{opt_loose} = 0 unless defined $$self{opt_loose};
93 $$self{opt_sentence} = 0 unless defined $$self{opt_sentence};
94 $$self{opt_width} = 76 unless defined $$self{opt_width};
96 # Figure out what quotes we'll be using for C<> text.
97 $$self{opt_quotes} ||= '"';
98 if ($$self{opt_quotes} eq 'none') {
99 $$self{LQUOTE} = $$self{RQUOTE} = '';
100 } elsif (length ($$self{opt_quotes}) == 1) {
101 $$self{LQUOTE} = $$self{RQUOTE} = $$self{opt_quotes};
102 } elsif ($$self{opt_quotes} =~ /^(.)(.)$/
103 || $$self{opt_quotes} =~ /^(..)(..)$/) {
107 croak qq(Invalid quote specification "$$self{opt_quotes}");
110 # If requested, do something with the non-POD text.
111 $self->code_handler (\&handle_code) if $$self{opt_code};
113 # Return the created object.
117 ##############################################################################
119 ##############################################################################
121 # This is the glue that connects the code below with Pod::Simple itself. The
122 # goal is to convert the event stream coming from the POD parser into method
123 # calls to handlers once the complete content of a tag has been seen. Each
124 # paragraph or POD command will have textual content associated with it, and
125 # as soon as all of a paragraph or POD command has been seen, that content
126 # will be passed in to the corresponding method for handling that type of
127 # object. The exceptions are handlers for lists, which have opening tag
128 # handlers and closing tag handlers that will be called right away.
130 # The internal hash key PENDING is used to store the contents of a tag until
131 # all of it has been seen. It holds a stack of open tags, each one
132 # represented by a tuple of the attributes hash for the tag and the contents
135 # Add a block of text to the contents of the current node, formatting it
136 # according to the current formatting instructions as we do.
138 my ($self, $text) = @_;
139 my $tag = $$self{PENDING}[-1];
143 # Given an element name, get the corresponding method name.
144 sub method_for_element {
145 my ($self, $element) = @_;
147 $element =~ tr/A-Z/a-z/;
148 $element =~ tr/_a-z0-9//cd;
152 # Handle the start of a new element. If cmd_element is defined, assume that
153 # we need to collect the entire tree for this element before passing it to the
154 # element method, and create a new tree into which we'll collect blocks of
155 # text and nested elements. Otherwise, if start_element is defined, call it.
156 sub _handle_element_start {
157 my ($self, $element, $attrs) = @_;
158 my $method = $self->method_for_element ($element);
160 # If we have a command handler, we need to accumulate the contents of the
161 # tag before calling it.
162 if ($self->can ("cmd_$method")) {
163 push (@{ $$self{PENDING} }, [ $attrs, '' ]);
164 } elsif ($self->can ("start_$method")) {
165 my $method = 'start_' . $method;
166 $self->$method ($attrs, '');
170 # Handle the end of an element. If we had a cmd_ method for this element,
171 # this is where we pass along the text that we've accumulated. Otherwise, if
172 # we have an end_ method for the element, call that.
173 sub _handle_element_end {
174 my ($self, $element) = @_;
175 my $method = $self->method_for_element ($element);
177 # If we have a command handler, pull off the pending text and pass it to
178 # the handler along with the saved attribute hash.
179 if ($self->can ("cmd_$method")) {
180 my $tag = pop @{ $$self{PENDING} };
181 my $method = 'cmd_' . $method;
182 my $text = $self->$method (@$tag);
184 if (@{ $$self{PENDING} } > 1) {
185 $$self{PENDING}[-1][1] .= $text;
187 $self->output ($text);
190 } elsif ($self->can ("end_$method")) {
191 my $method = 'end_' . $method;
196 ##############################################################################
198 ##############################################################################
200 # Wrap a line, indenting by the current left margin. We can't use Text::Wrap
201 # because it plays games with tabs. We can't use formline, even though we'd
202 # really like to, because it screws up non-printing characters. So we have to
203 # do the wrapping ourselves.
208 my $spaces = ' ' x $$self{MARGIN};
209 my $width = $$self{opt_width} - $$self{MARGIN};
210 while (length > $width) {
211 if (s/^([^\n]{0,$width})\s+// || s/^([^\n]{$width})//) {
212 $output .= $spaces . $1 . "\n";
217 $output .= $spaces . $_;
218 $output =~ s/\s+$/\n\n/;
222 # Reformat a paragraph of text for the current margin. Takes the text to
223 # reformat and returns the formatted text.
228 # If we're trying to preserve two spaces after sentences, do some munging
229 # to support that. Otherwise, smash all repeated whitespace.
230 if ($$self{opt_sentence}) {
238 return $self->wrap ($_);
241 # Output text to the output device.
243 my ($self, $text) = @_;
244 $text =~ tr/\240\255/ /d;
245 print { $$self{output_fh} } $text;
248 # Output a block of code (something that isn't part of the POD text). Called
249 # by preprocess_paragraph only if we were given the code option. Exists here
250 # only so that it can be overridden by subclasses.
251 sub output_code { $_[0]->output ($_[1]) }
253 ##############################################################################
254 # Document initialization
255 ##############################################################################
257 # Set up various things that have to be initialized on a per-document basis.
260 my $margin = $$self{opt_indent} + $$self{opt_margin};
262 # Initialize a few per-document variables.
263 $$self{INDENTS} = []; # Stack of indentations.
264 $$self{MARGIN} = $margin; # Default left margin.
265 $$self{PENDING} = [[]]; # Pending output.
270 ##############################################################################
272 ##############################################################################
274 # This method is called whenever an =item command is complete (in other words,
275 # we've seen its associated paragraph or know for certain that it doesn't have
276 # one). It gets the paragraph associated with the item as an argument. If
277 # that argument is empty, just output the item tag; if it contains a newline,
278 # output the item tag followed by the newline. Otherwise, see if there's
279 # enough room for us to output the item tag in the margin of the text or if we
280 # have to put it on a separate line.
282 my ($self, $text) = @_;
283 my $tag = $$self{ITEM};
284 unless (defined $tag) {
285 carp "Item called without tag";
290 # Calculate the indentation and margin. $fits is set to true if the tag
291 # will fit into the margin of the paragraph given our indentation level.
292 my $indent = $$self{INDENTS}[-1];
293 $indent = $$self{opt_indent} unless defined $indent;
294 my $margin = ' ' x $$self{opt_margin};
295 my $fits = ($$self{MARGIN} - $indent >= length ($tag) + 1);
297 # If the tag doesn't fit, or if we have no associated text, print out the
298 # tag separately. Otherwise, put the tag in the margin of the paragraph.
299 if (!$text || $text =~ /^\s+$/ || !$fits) {
300 my $realindent = $$self{MARGIN};
301 $$self{MARGIN} = $indent;
302 my $output = $self->reformat ($tag);
303 $output =~ s/^$margin /$margin:/ if ($$self{opt_alt} && $indent > 0);
304 $output =~ s/\n*$/\n/;
306 # If the text is just whitespace, we have an empty item paragraph;
307 # this can result from =over/=item/=back without any intermixed
308 # paragraphs. Insert some whitespace to keep the =item from merging
309 # into the next paragraph.
310 $output .= "\n" if $text && $text =~ /^\s*$/;
312 $self->output ($output);
313 $$self{MARGIN} = $realindent;
314 $self->output ($self->reformat ($text)) if ($text && $text =~ /\S/);
316 my $space = ' ' x $indent;
317 $space =~ s/^$margin /$margin:/ if $$self{opt_alt};
318 $text = $self->reformat ($text);
319 $text =~ s/^$margin /$margin:/ if ($$self{opt_alt} && $indent > 0);
320 my $tagspace = ' ' x length $tag;
321 $text =~ s/^($space)$tagspace/$1$tag/ or warn "Bizarre space in item";
322 $self->output ($text);
326 # Handle a basic block of text. The only tricky thing here is that if there
327 # is a pending item tag, we need to format this as an item paragraph.
329 my ($self, $attrs, $text) = @_;
331 if (defined $$self{ITEM}) {
332 $self->item ($text . "\n");
334 $self->output ($self->reformat ($text . "\n"));
339 # Handle a verbatim paragraph. Just print it out, but indent it according to
342 my ($self, $attrs, $text) = @_;
343 $self->item if defined $$self{ITEM};
344 return if $text =~ /^\s*$/;
345 $text =~ s/^(\n*)(\s*\S+)/$1 . (' ' x $$self{MARGIN}) . $2/gme;
346 $text =~ s/\s*$/\n\n/;
347 $self->output ($text);
351 # Handle literal text (produced by =for and similar constructs). Just output
352 # it with the minimum of changes.
354 my ($self, $attrs, $text) = @_;
356 $text =~ s/\n{0,2}$/\n/;
357 $self->output ($text);
361 ##############################################################################
363 ##############################################################################
365 # The common code for handling all headers. Takes the header text, the
366 # indentation, and the surrounding marker for the alt formatting method.
368 my ($self, $text, $indent, $marker) = @_;
369 $self->item ("\n\n") if defined $$self{ITEM};
371 if ($$self{opt_alt}) {
372 my $closemark = reverse (split (//, $marker));
373 my $margin = ' ' x $$self{opt_margin};
374 $self->output ("\n" . "$margin$marker $text $closemark" . "\n\n");
376 $text .= "\n" if $$self{opt_loose};
377 my $margin = ' ' x ($$self{opt_margin} + $indent);
378 $self->output ($margin . $text . "\n");
383 # First level heading.
385 my ($self, $attrs, $text) = @_;
386 $self->heading ($text, 0, '====');
389 # Second level heading.
391 my ($self, $attrs, $text) = @_;
392 $self->heading ($text, $$self{opt_indent} / 2, '== ');
395 # Third level heading.
397 my ($self, $attrs, $text) = @_;
398 $self->heading ($text, $$self{opt_indent} * 2 / 3 + 0.5, '= ');
401 # Fourth level heading.
403 my ($self, $attrs, $text) = @_;
404 $self->heading ($text, $$self{opt_indent} * 3 / 4 + 0.5, '- ');
407 ##############################################################################
409 ##############################################################################
411 # Handle the beginning of an =over block. Takes the type of the block as the
412 # first argument, and then the attr hash. This is called by the handlers for
413 # the four different types of lists (bullet, number, text, and block).
414 sub over_common_start {
415 my ($self, $attrs) = @_;
416 $self->item ("\n\n") if defined $$self{ITEM};
418 # Find the indentation level.
419 my $indent = $$attrs{indent};
420 unless (defined ($indent) && $indent =~ /^\s*[-+]?\d{1,4}\s*$/) {
421 $indent = $$self{opt_indent};
424 # Add this to our stack of indents and increase our current margin.
425 push (@{ $$self{INDENTS} }, $$self{MARGIN});
426 $$self{MARGIN} += ($indent + 0);
430 # End an =over block. Takes no options other than the class pointer. Output
431 # any pending items and then pop one level of indentation.
432 sub over_common_end {
434 $self->item ("\n\n") if defined $$self{ITEM};
435 $$self{MARGIN} = pop @{ $$self{INDENTS} };
439 # Dispatch the start and end calls as appropriate.
440 sub start_over_bullet { $_[0]->over_common_start ($_[1]) }
441 sub start_over_number { $_[0]->over_common_start ($_[1]) }
442 sub start_over_text { $_[0]->over_common_start ($_[1]) }
443 sub start_over_block { $_[0]->over_common_start ($_[1]) }
444 sub end_over_bullet { $_[0]->over_common_end }
445 sub end_over_number { $_[0]->over_common_end }
446 sub end_over_text { $_[0]->over_common_end }
447 sub end_over_block { $_[0]->over_common_end }
449 # The common handler for all item commands. Takes the type of the item, the
450 # attributes, and then the text of the item.
452 my ($self, $type, $attrs, $text) = @_;
453 $self->item if defined $$self{ITEM};
455 # Clean up the text. We want to end up with two variables, one ($text)
456 # which contains any body text after taking out the item portion, and
457 # another ($item) which contains the actual item text. Note the use of
458 # the internal Pod::Simple attribute here; that's a potential land mine.
461 if ($type eq 'bullet') {
463 } elsif ($type eq 'number') {
464 $item = $$attrs{'~orig_content'};
467 $item =~ s/\s*\n\s*/ /g;
470 $$self{ITEM} = $item;
472 # If body text for this item was included, go ahead and output that now.
480 # Dispatch the item commands to the appropriate place.
481 sub cmd_item_bullet { my $self = shift; $self->item_common ('bullet', @_) }
482 sub cmd_item_number { my $self = shift; $self->item_common ('number', @_) }
483 sub cmd_item_text { my $self = shift; $self->item_common ('text', @_) }
484 sub cmd_item_block { my $self = shift; $self->item_common ('block', @_) }
486 ##############################################################################
488 ##############################################################################
491 sub cmd_b { return $_[0]{alt} ? "``$_[2]''" : $_[2] }
492 sub cmd_f { return $_[0]{alt} ? "\"$_[2]\"" : $_[2] }
493 sub cmd_i { return '*' . $_[2] . '*' }
494 sub cmd_x { return '' }
496 # Apply a whole bunch of messy heuristics to not quote things that don't
497 # benefit from being quoted. These originally come from Barrie Slaymaker and
498 # largely duplicate code in Pod::Man.
500 my ($self, $attrs, $text) = @_;
502 # A regex that matches the portion of a variable reference that's the
503 # array or hash index, separated out just because we want to use it in
504 # several places in the following regex.
505 my $index = '(?: \[.*\] | \{.*\} )?';
507 # Check for things that we don't want to quote, and if we find any of
508 # them, return the string with just a font change and no quoting.
512 ( [\'\`\"] ) .* \1 # already quoted
513 | \` .* \' # `quoted'
514 | \$+ [\#^]? \S $index # special ($^Foo, $")
515 | [\$\@%&*]+ \#? [:\'\w]+ $index # plain var or func
516 | [\$\@%&*]* [:\'\w]+ (?: -> )? \(\s*[^\s,]\s*\) # 0/1-arg func call
517 | [+-]? ( \d[\d.]* | \.\d+ ) (?: [eE][+-]?\d+ )? # a number
518 | 0x [a-fA-F\d]+ # a hex constant
523 # If we didn't return, go ahead and quote the text.
524 return $$self{opt_alt}
526 : "$$self{LQUOTE}$text$$self{RQUOTE}";
529 # Links reduce to the text that we're given, wrapped in angle brackets if it's
532 my ($self, $attrs, $text) = @_;
533 return $$attrs{type} eq 'url' ? "<$text>" : $text;
536 ##############################################################################
537 # Backwards compatibility
538 ##############################################################################
540 # The old Pod::Text module did everything in a pod2text() function. This
541 # tries to provide the same interface for legacy applications.
545 # This is really ugly; I hate doing option parsing in the middle of a
546 # module. But the old Pod::Text module supported passing flags to its
547 # entry function, so handle -a and -<number>.
548 while ($_[0] =~ /^-/) {
550 if ($flag eq '-a') { push (@args, alt => 1) }
551 elsif ($flag =~ /^-(\d+)$/) { push (@args, width => $1) }
558 # Now that we know what arguments we're using, create the parser.
559 my $parser = Pod::Text->new (@args);
561 # If two arguments were given, the second argument is going to be a file
562 # handle. That means we want to call parse_from_filehandle(), which means
563 # we need to turn the first argument into a file handle. Magic open will
564 # handle the <&STDIN case automagically.
568 unless (open (IN, $fhs[0])) {
569 croak ("Can't open $fhs[0] for reading: $!\n");
573 $parser->output_fh ($fhs[1]);
574 my $retval = $parser->parse_file ($fhs[0]);
575 my $fh = $parser->output_fh ();
579 return $parser->parse_file (@_);
583 # Reset the underlying Pod::Simple object between calls to parse_from_file so
584 # that the same object can be reused to convert multiple pages.
585 sub parse_from_file {
589 # Fake the old cutting option to Pod::Parser. This fiddings with internal
590 # Pod::Simple state and is quite ugly; we need a better approach.
591 if (ref ($_[0]) eq 'HASH') {
593 if (defined ($$opts{-cutting}) && !$$opts{-cutting}) {
595 $$self{last_was_blank} = 1;
600 my $retval = $self->Pod::Simple::parse_from_file (@_);
602 # Flush output, since Pod::Simple doesn't do this. Ideally we should also
603 # close the file descriptor if we had to open one, but we can't easily
605 my $fh = $self->output_fh ();
606 my $oldfh = select $fh;
615 # Pod::Simple failed to provide this backward compatibility function, so
616 # implement it ourselves. File handles are one of the inputs that
617 # parse_from_file supports.
618 sub parse_from_filehandle {
620 $self->parse_from_file (@_);
623 ##############################################################################
624 # Module return value and documentation
625 ##############################################################################
632 Pod::Text - Convert POD data to formatted ASCII text
635 alt Allbery Sean Burke's
640 my $parser = Pod::Text->new (sentence => 0, width => 78);
642 # Read POD from STDIN and write to STDOUT.
643 $parser->parse_from_filehandle;
645 # Read POD from file.pod and write to file.txt.
646 $parser->parse_from_file ('file.pod', 'file.txt');
650 Pod::Text is a module that can convert documentation in the POD format (the
651 preferred language for documenting Perl) into formatted ASCII. It uses no
652 special formatting controls or codes whatsoever, and its output is therefore
653 suitable for nearly any device.
655 As a derived class from Pod::Simple, Pod::Text supports the same methods and
656 interfaces. See L<Pod::Simple> for all the details; briefly, one creates a
657 new parser with C<< Pod::Text->new() >> and then normally calls parse_file().
659 new() can take options, in the form of key/value pairs, that control the
660 behavior of the parser. The currently recognized options are:
666 If set to a true value, selects an alternate output format that, among other
667 things, uses a different heading style and marks C<=item> entries with a
668 colon in the left margin. Defaults to false.
672 If set to a true value, the non-POD parts of the input file will be included
673 in the output. Useful for viewing code documented with POD blocks with the
674 POD rendered and the code left intact.
678 The number of spaces to indent regular text, and the default indentation for
679 C<=over> blocks. Defaults to 4.
683 If set to a true value, a blank line is printed after a C<=head1> heading.
684 If set to false (the default), no blank line is printed after C<=head1>,
685 although one is still printed after C<=head2>. This is the default because
686 it's the expected formatting for manual pages; if you're formatting
687 arbitrary text documents, setting this to true may result in more pleasing
692 The width of the left margin in spaces. Defaults to 0. This is the margin
693 for all text, including headings, not the amount by which regular text is
694 indented; for the latter, see the I<indent> option. To set the right
695 margin, see the I<width> option.
699 Sets the quote marks used to surround CE<lt>> text. If the value is a
700 single character, it is used as both the left and right quote; if it is two
701 characters, the first character is used as the left quote and the second as
702 the right quoted; and if it is four characters, the first two are used as
703 the left quote and the second two as the right quote.
705 This may also be set to the special value C<none>, in which case no quote
706 marks are added around CE<lt>> text.
710 If set to a true value, Pod::Text will assume that each sentence ends in two
711 spaces, and will try to preserve that spacing. If set to false, all
712 consecutive whitespace in non-verbatim paragraphs is compressed into a
713 single space. Defaults to true.
717 The column at which to wrap text on the right-hand side. Defaults to 76.
721 The standard Pod::Simple method parse_file() takes one argument, the file or
722 file handle to read from, and writes output to standard output unless that
723 has been changed with the output_fh() method. See L<Pod::Simple> for the
724 specific details and for other alternative interfaces.
730 =item Bizarre space in item
732 =item Item called without tag
734 (W) Something has gone wrong in internal C<=item> processing. These
735 messages indicate a bug in Pod::Text; you should never see them.
737 =item Can't open %s for reading: %s
739 (F) Pod::Text was invoked via the compatibility mode pod2text() interface
740 and the input file it was given could not be opened.
742 =item Invalid quote specification "%s"
744 (F) The quote specification given (the quotes option to the constructor) was
745 invalid. A quote specification must be one, two, or four characters long.
751 This is a replacement for an earlier Pod::Text module written by Tom
752 Christiansen. It has a revamped interface, since it now uses Pod::Simple,
753 but an interface roughly compatible with the old Pod::Text::pod2text()
754 function is still available. Please change to the new calling convention,
757 The original Pod::Text contained code to do formatting via termcap
758 sequences, although it wasn't turned on by default and it was problematic to
759 get it to work at all. This rewrite doesn't even try to do that, but a
760 subclass of it does. Look for L<Pod::Text::Termcap>.
764 L<Pod::Simple>, L<Pod::Text::Termcap>, L<pod2text(1)>
766 The current version of this module is always available from its web site at
767 L<http://www.eyrie.org/~eagle/software/podlators/>. It is also part of the
768 Perl core distribution as of 5.6.0.
772 Russ Allbery <rra@stanford.edu>, based I<very> heavily on the original
773 Pod::Text by Tom Christiansen <tchrist@mox.perl.com> and its conversion to
774 Pod::Parser by Brad Appleton <bradapp@enteract.com>. Sean Burke's initial
775 conversion of Pod::Man to use Pod::Simple provided much-needed guidance on
776 how to use Pod::Simple.
778 =head1 COPYRIGHT AND LICENSE
780 Copyright 1999, 2000, 2001, 2002, 2004, 2006, 2008 Russ Allbery
783 This program is free software; you may redistribute it and/or modify it
784 under the same terms as Perl itself.