1package Pod::Simple;
2use strict;
3use warnings;
4use Carp ();
5BEGIN           { *DEBUG = sub () {0} unless defined &DEBUG }
6use integer;
7use Pod::Escapes 1.04 ();
8use Pod::Simple::LinkSection ();
9use Pod::Simple::BlackBox ();
10use Pod::Simple::TiedOutFH;
11#use utf8;
12
13our @ISA = ('Pod::Simple::BlackBox');
14our $VERSION = '3.45';
15
16our @Known_formatting_codes = qw(I B C L E F S X Z);
17our %Known_formatting_codes = map(($_=>1), @Known_formatting_codes);
18our @Known_directives       = qw(head1 head2 head3 head4 head5 head6 item over back);
19our %Known_directives       = map(($_=>'Plain'), @Known_directives);
20our $NL = $/ unless defined $NL;
21
22#-----------------------------------------------------------------------------
23# Set up some constants:
24
25BEGIN {
26  if(defined &ASCII)    { }
27  elsif(chr(65) eq 'A') { *ASCII = sub () {1}  }
28  else                  { *ASCII = sub () {''} }
29
30  unless(defined &MANY_LINES) { *MANY_LINES = sub () {20} }
31  DEBUG > 4 and print STDERR "MANY_LINES is ", MANY_LINES(), "\n";
32  unless(MANY_LINES() >= 1) {
33    die "MANY_LINES is too small (", MANY_LINES(), ")!\nAborting";
34  }
35  if(defined &UNICODE) { }
36  elsif($] >= 5.008)   { *UNICODE = sub() {1}  }
37  else                 { *UNICODE = sub() {''} }
38}
39if(DEBUG > 2) {
40  print STDERR "# We are ", ASCII ? '' : 'not ', "in ASCII-land\n";
41  print STDERR "# We are under a Unicode-safe Perl.\n";
42}
43
44# The NO BREAK SPACE and SOFT HYHPEN are used in several submodules.
45if ($] ge 5.007_003) {  # On sufficiently modern Perls we can handle any
46                        # character set
47  $Pod::Simple::nbsp = chr utf8::unicode_to_native(0xA0);
48  $Pod::Simple::shy  = chr utf8::unicode_to_native(0xAD);
49}
50elsif (Pod::Simple::ASCII) {  # Hard code ASCII early Perl
51  $Pod::Simple::nbsp = "\xA0";
52  $Pod::Simple::shy  = "\xAD";
53}
54else { # EBCDIC on early Perl.  We know what the values are for the code
55        # pages supported then.
56  $Pod::Simple::nbsp = "\x41";
57  $Pod::Simple::shy  = "\xCA";
58}
59
60# Design note:
61# This is a parser for Pod.  It is not a parser for the set of Pod-like
62#  languages which happens to contain Pod -- it is just for Pod, plus possibly
63#  some extensions.
64
65# @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @
66#@ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @
67#@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
68
69__PACKAGE__->_accessorize(
70  '_output_is_for_JustPod', # For use only by Pod::Simple::JustPod,
71                       # If non-zero, don't expand Z<> E<> S<> L<>,
72                       # and count how many brackets in format codes
73  'nbsp_for_S',        # Whether to map S<...>'s to \xA0 characters
74  'source_filename',   # Filename of the source, for use in warnings
75  'source_dead',       # Whether to consider this parser's source dead
76
77  'output_fh',         # The filehandle we're writing to, if applicable.
78                       # Used only in some derived classes.
79
80  'hide_line_numbers', # For some dumping subclasses: whether to pointedly
81                       # suppress the start_line attribute
82
83  'line_count',        # the current line number
84  'pod_para_count',    # count of pod paragraphs seen so far
85
86  'no_whining',        # whether to suppress whining
87  'no_errata_section', # whether to suppress the errata section
88  'complain_stderr',   # whether to complain to stderr
89
90  'doc_has_started',   # whether we've fired the open-Document event yet
91
92  'bare_output',       # For some subclasses: whether to prepend
93                       #  header-code and postpend footer-code
94
95  'keep_encoding_directive',  # whether to emit =encoding
96  'nix_X_codes',       # whether to ignore X<...> codes
97  'merge_text',        # whether to avoid breaking a single piece of
98                       #  text up into several events
99
100  'preserve_whitespace', # whether to try to keep whitespace as-is
101  'strip_verbatim_indent', # What indent to strip from verbatim
102  'expand_verbatim_tabs',  # 0: preserve tabs in verbatim blocks
103                           # n: expand tabs to stops every n columns
104
105  'parse_characters',  # Whether parser should expect chars rather than octets
106
107 'content_seen',      # whether we've seen any real Pod content
108 'errors_seen',       # TODO: document.  whether we've seen any errors (fatal or not)
109
110 'codes_in_verbatim', # for PseudoPod extensions
111
112 'code_handler',      # coderef to call when a code (non-pod) line is seen
113 'cut_handler',       # ... when a =cut line is seen
114 'pod_handler',       # ... when a =pod line is seen
115 'whiteline_handler', # ... when a line with only whitespace is seen
116 #Called like:
117 # $code_handler->($line, $self->{'line_count'}, $self) if $code_handler;
118 #  $cut_handler->($line, $self->{'line_count'}, $self) if $cut_handler;
119 #  $pod_handler->($line, $self->{'line_count'}, $self) if $pod_handler;
120 #   $wl_handler->($line, $self->{'line_count'}, $self) if $wl_handler;
121 'parse_empty_lists', # whether to acknowledge empty =over/=back blocks
122 'raw_mode',          # to report entire raw lines instead of Pod elements
123);
124
125#@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
126
127sub any_errata_seen {  # good for using as an exit() value...
128  return shift->{'errors_seen'} || 0;
129}
130
131sub errata_seen {
132  return shift->{'all_errata'} || {};
133}
134
135# Returns the encoding only if it was recognized as being handled and set
136sub detected_encoding {
137  return shift->{'detected_encoding'};
138}
139
140sub encoding {
141  my $this = shift;
142  return $this->{'encoding'} unless @_;  # GET.
143
144  $this->_handle_encoding_line("=encoding $_[0]");
145  if ($this->{'_processed_encoding'}) {
146    delete $this->{'_processed_encoding'};
147    if(! $this->{'encoding_command_statuses'} ) {
148      DEBUG > 2 and print STDERR " CRAZY ERROR: encoding wasn't really handled?!\n";
149    } elsif( $this->{'encoding_command_statuses'}[-1] ) {
150      $this->scream( "=encoding $_[0]",
151         sprintf "Couldn't do %s: %s",
152         $this->{'encoding_command_reqs'  }[-1],
153         $this->{'encoding_command_statuses'}[-1],
154      );
155    } else {
156      DEBUG > 2 and print STDERR " (encoding successfully handled.)\n";
157    }
158    return $this->{'encoding'};
159  } else {
160    return undef;
161  }
162}
163
164#@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
165# Pull in some functions that, for some reason, I expect to see here too:
166BEGIN {
167  *pretty        = \&Pod::Simple::BlackBox::pretty;
168  *stringify_lol = \&Pod::Simple::BlackBox::stringify_lol;
169  *my_qr         = \&Pod::Simple::BlackBox::my_qr;
170}
171
172#@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
173
174sub version_report {
175  my $class = ref($_[0]) || $_[0];
176  if($class eq __PACKAGE__) {
177    return "$class $VERSION";
178  } else {
179    my $v = $class->VERSION;
180    return "$class $v (" . __PACKAGE__ . " $VERSION)";
181  }
182}
183
184#@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
185
186#sub curr_open { # read-only list accessor
187#  return @{ $_[0]{'curr_open'} || return() };
188#}
189#sub _curr_open_listref { $_[0]{'curr_open'} ||= [] }
190
191
192sub output_string {
193  # Works by faking out output_fh.  Simplifies our code.
194  #
195  my $this = shift;
196  return $this->{'output_string'} unless @_;  # GET.
197
198  my $x = (defined($_[0]) and ref($_[0])) ? $_[0] : \( $_[0] );
199  $$x = '' unless defined $$x;
200  DEBUG > 4 and print STDERR "# Output string set to $x ($$x)\n";
201  $this->{'output_fh'} = Pod::Simple::TiedOutFH->handle_on($_[0]);
202  return
203    $this->{'output_string'} = $_[0];
204    #${ ${ $this->{'output_fh'} } };
205}
206
207sub abandon_output_string { $_[0]->abandon_output_fh; delete $_[0]{'output_string'} }
208sub abandon_output_fh     { $_[0]->output_fh(undef) }
209# These don't delete the string or close the FH -- they just delete our
210#  references to it/them.
211# TODO: document these
212
213#@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
214
215sub new {
216  # takes no parameters
217  my $class = ref($_[0]) || $_[0];
218  #Carp::croak(__PACKAGE__ . " is a virtual base class -- see perldoc "
219  #  . __PACKAGE__ );
220  my $obj = bless {
221    'accept_codes'      => { map( ($_=>$_), @Known_formatting_codes ) },
222    'accept_directives' => { %Known_directives },
223    'accept_targets'    => {},
224  }, $class;
225
226  $obj->expand_verbatim_tabs(8);
227  return $obj;
228}
229
230
231
232# TODO: an option for whether to interpolate E<...>'s, or just resolve to codes.
233
234#@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
235
236sub _handle_element_start {     # OVERRIDE IN DERIVED CLASS
237  my($self, $element_name, $attr_hash_r) = @_;
238  return;
239}
240
241sub _handle_element_end {       # OVERRIDE IN DERIVED CLASS
242  my($self, $element_name) = @_;
243  return;
244}
245
246sub _handle_text          {     # OVERRIDE IN DERIVED CLASS
247  my($self, $text) = @_;
248  return;
249}
250
251#@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
252#
253# And now directives (not targets)
254
255sub accept_directive_as_verbatim  { shift->_accept_directives('Verbatim', @_) }
256sub accept_directive_as_data      { shift->_accept_directives('Data',     @_) }
257sub accept_directive_as_processed { shift->_accept_directives('Plain',    @_) }
258
259sub _accept_directives {
260  my($this, $type) = splice @_,0,2;
261  foreach my $d (@_) {
262    next unless defined $d and length $d;
263    Carp::croak "\"$d\" isn't a valid directive name"
264     unless $d =~ m/^[a-zA-Z][a-zA-Z0-9]*$/s;
265    Carp::croak "\"$d\" is already a reserved Pod directive name"
266     if exists $Known_directives{$d};
267    $this->{'accept_directives'}{$d} = $type;
268    DEBUG > 2 and print STDERR "Learning to accept \"=$d\" as directive of type $type\n";
269  }
270  DEBUG > 6 and print STDERR "$this\'s accept_directives : ",
271   pretty($this->{'accept_directives'}), "\n";
272
273  return sort keys %{ $this->{'accept_directives'} } if wantarray;
274  return;
275}
276
277#--------------------------------------------------------------------------
278# TODO: document these:
279
280sub unaccept_directive { shift->unaccept_directives(@_) };
281
282sub unaccept_directives {
283  my $this = shift;
284  foreach my $d (@_) {
285    next unless defined $d and length $d;
286    Carp::croak "\"$d\" isn't a valid directive name"
287     unless $d =~ m/^[a-zA-Z][a-zA-Z0-9]*$/s;
288    Carp::croak "But you must accept \"$d\" directives -- it's a builtin!"
289     if exists $Known_directives{$d};
290    delete $this->{'accept_directives'}{$d};
291    DEBUG > 2 and print STDERR "OK, won't accept \"=$d\" as directive.\n";
292  }
293  return sort keys %{ $this->{'accept_directives'} } if wantarray;
294  return
295}
296
297#@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
298#
299# And now targets (not directives)
300
301sub accept_target         { shift->accept_targets(@_)         } # alias
302sub accept_target_as_text { shift->accept_targets_as_text(@_) } # alias
303
304
305sub accept_targets         { shift->_accept_targets('1', @_) }
306
307sub accept_targets_as_text { shift->_accept_targets('force_resolve', @_) }
308 # forces them to be processed, even when there's no ":".
309
310sub _accept_targets {
311  my($this, $type) = splice @_,0,2;
312  foreach my $t (@_) {
313    next unless defined $t and length $t;
314    # TODO: enforce some limitations on what a target name can be?
315    $this->{'accept_targets'}{$t} = $type;
316    DEBUG > 2 and print STDERR "Learning to accept \"$t\" as target of type $type\n";
317  }
318  return sort keys %{ $this->{'accept_targets'} } if wantarray;
319  return;
320}
321
322#--------------------------------------------------------------------------
323sub unaccept_target         { shift->unaccept_targets(@_) }
324
325sub unaccept_targets {
326  my $this = shift;
327  foreach my $t (@_) {
328    next unless defined $t and length $t;
329    # TODO: enforce some limitations on what a target name can be?
330    delete $this->{'accept_targets'}{$t};
331    DEBUG > 2 and print STDERR "OK, won't accept \"$t\" as target.\n";
332  }
333  return sort keys %{ $this->{'accept_targets'} } if wantarray;
334  return;
335}
336
337#@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
338#
339# And now codes (not targets or directives)
340
341# XXX Probably it is an error that the digit '9' is excluded from these re's.
342# Broken for early Perls on EBCDIC
343my $xml_name_re = my_qr('[^-.0-8:A-Z_a-z[:^ascii:]]', '9');
344$xml_name_re = qr/[\x00-\x2C\x2F\x39\x3B-\x40\x5B-\x5E\x60\x7B-\x7F]/
345                                                            unless $xml_name_re;
346
347sub accept_code { shift->accept_codes(@_) } # alias
348
349sub accept_codes {  # Add some codes
350  my $this = shift;
351
352  foreach my $new_code (@_) {
353    next unless defined $new_code and length $new_code;
354    # A good-enough check that it's good as an XML Name symbol:
355    Carp::croak "\"$new_code\" isn't a valid element name"
356      if $new_code =~ $xml_name_re
357          # Characters under 0x80 that aren't legal in an XML Name.
358      or $new_code =~ m/^[-\.0-9]/s
359      or $new_code =~ m/:[-\.0-9]/s;
360          # The legal under-0x80 Name characters that
361          #  an XML Name still can't start with.
362
363    $this->{'accept_codes'}{$new_code} = $new_code;
364
365    # Yes, map to itself -- just so that when we
366    #  see "=extend W [whatever] thatelementname", we say that W maps
367    #  to whatever $this->{accept_codes}{thatelementname} is,
368    #  i.e., "thatelementname".  Then when we go re-mapping,
369    #  a "W" in the treelet turns into "thatelementname".  We only
370    #  remap once.
371    # If we say we accept "W", then a "W" in the treelet simply turns
372    #  into "W".
373  }
374
375  return;
376}
377
378#--------------------------------------------------------------------------
379sub unaccept_code { shift->unaccept_codes(@_) }
380
381sub unaccept_codes { # remove some codes
382  my $this = shift;
383
384  foreach my $new_code (@_) {
385    next unless defined $new_code and length $new_code;
386    # A good-enough check that it's good as an XML Name symbol:
387    Carp::croak "\"$new_code\" isn't a valid element name"
388      if $new_code =~ $xml_name_re
389          # Characters under 0x80 that aren't legal in an XML Name.
390      or $new_code =~ m/^[-\.0-9]/s
391      or $new_code =~ m/:[-\.0-9]/s;
392          # The legal under-0x80 Name characters that
393          #  an XML Name still can't start with.
394
395    Carp::croak "But you must accept \"$new_code\" codes -- it's a builtin!"
396     if grep $new_code eq $_, @Known_formatting_codes;
397
398    delete $this->{'accept_codes'}{$new_code};
399
400    DEBUG > 2 and print STDERR "OK, won't accept the code $new_code<...>.\n";
401  }
402
403  return;
404}
405
406
407#@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
408#@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
409
410sub parse_string_document {
411  my $self = shift;
412  my @lines;
413  foreach my $line_group (@_) {
414    next unless defined $line_group and length $line_group;
415    pos($line_group) = 0;
416    while($line_group =~
417      m/([^\n\r]*)(\r?\n?)/g # supports \r, \n ,\r\n
418      #m/([^\n\r]*)((?:\r?\n)?)/g
419    ) {
420      #print(">> $1\n"),
421      $self->parse_lines($1)
422       if length($1) or length($2)
423        or pos($line_group) != length($line_group);
424       # I.e., unless it's a zero-length "empty line" at the very
425       #  end of "foo\nbar\n" (i.e., between the \n and the EOS).
426    }
427  }
428  $self->parse_lines(undef); # to signal EOF
429  return $self;
430}
431
432sub _init_fh_source {
433  my($self, $source) = @_;
434
435  #DEBUG > 1 and print STDERR "Declaring $source as :raw for starters\n";
436  #$self->_apply_binmode($source, ':raw');
437  #binmode($source, ":raw");
438
439  return;
440}
441
442#:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.
443#
444
445sub parse_file {
446  my($self, $source) = (@_);
447
448  if(!defined $source) {
449    Carp::croak("Can't use empty-string as a source for parse_file");
450  } elsif(ref(\$source) eq 'GLOB') {
451    $self->{'source_filename'} = '' . ($source);
452  } elsif(ref $source) {
453    $self->{'source_filename'} = '' . ($source);
454  } elsif(!length $source) {
455    Carp::croak("Can't use empty-string as a source for parse_file");
456  } else {
457    {
458      local *PODSOURCE;
459      open(PODSOURCE, "<$source") || Carp::croak("Can't open $source: $!");
460      $self->{'source_filename'} = $source;
461      $source = *PODSOURCE{IO};
462    }
463    $self->_init_fh_source($source);
464  }
465  # By here, $source is a FH.
466
467  $self->{'source_fh'} = $source;
468
469  my($i, @lines);
470  until( $self->{'source_dead'} ) {
471    splice @lines;
472
473    for($i = MANY_LINES; $i--;) {  # read those many lines at a time
474      local $/ = $NL;
475      push @lines, scalar(<$source>);  # readline
476      last unless defined $lines[-1];
477       # but pass thru the undef, which will set source_dead to true
478    }
479
480    my $at_eof = ! $lines[-1]; # keep track of the undef
481    pop @lines if $at_eof; # silence warnings
482
483    # be eol agnostic
484    s/\r\n?/\n/g for @lines;
485
486    # make sure there are only one line elements for parse_lines
487    @lines = split(/(?<=\n)/, join('', @lines));
488
489    # push the undef back after popping it to set source_dead to true
490    push @lines, undef if $at_eof;
491
492    $self->parse_lines(@lines);
493  }
494  delete($self->{'source_fh'}); # so it can be GC'd
495  return $self;
496}
497
498#:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.
499
500sub parse_from_file {
501  # An emulation of Pod::Parser's interface, for the sake of Perldoc.
502  # Basically just a wrapper around parse_file.
503
504  my($self, $source, $to) = @_;
505  $self = $self->new unless ref($self); # so we tolerate being a class method
506
507  if(!defined $source)             { $source = *STDIN{IO}
508  } elsif(ref(\$source) eq 'GLOB') { # stet
509  } elsif(ref($source)           ) { # stet
510  } elsif(!length $source
511     or $source eq '-' or $source =~ m/^<&(?:STDIN|0)$/i
512  ) {
513    $source = *STDIN{IO};
514  }
515
516  if(!defined $to) {             $self->output_fh( *STDOUT{IO}   );
517  } elsif(ref(\$to) eq 'GLOB') { $self->output_fh( $to );
518  } elsif(ref($to)) {            $self->output_fh( $to );
519  } elsif(!length $to
520     or $to eq '-' or $to =~ m/^>&?(?:STDOUT|1)$/i
521  ) {
522    $self->output_fh( *STDOUT{IO} );
523  } elsif($to =~ m/^>&(?:STDERR|2)$/i) {
524    $self->output_fh( *STDERR{IO} );
525  } else {
526    require Symbol;
527    my $out_fh = Symbol::gensym();
528    DEBUG and print STDERR "Write-opening to $to\n";
529    open($out_fh, ">$to")  or  Carp::croak "Can't write-open $to: $!";
530    binmode($out_fh)
531     if $self->can('write_with_binmode') and $self->write_with_binmode;
532    $self->output_fh($out_fh);
533  }
534
535  return $self->parse_file($source);
536}
537
538#-----------------------------------------------------------------------------
539
540sub whine {
541  #my($self,$line,$complaint) = @_;
542  my $self = shift(@_);
543  ++$self->{'errors_seen'};
544  if($self->{'no_whining'}) {
545    DEBUG > 9 and print STDERR "Discarding complaint (at line $_[0]) $_[1]\n because no_whining is on.\n";
546    return;
547  }
548  push @{$self->{'all_errata'}{$_[0]}}, $_[1];
549  return $self->_complain_warn(@_) if $self->{'complain_stderr'};
550  return $self->_complain_errata(@_);
551}
552
553sub scream {    # like whine, but not suppressible
554  #my($self,$line,$complaint) = @_;
555  my $self = shift(@_);
556  ++$self->{'errors_seen'};
557  push @{$self->{'all_errata'}{$_[0]}}, $_[1];
558  return $self->_complain_warn(@_) if $self->{'complain_stderr'};
559  return $self->_complain_errata(@_);
560}
561
562sub _complain_warn {
563  my($self,$line,$complaint) = @_;
564  return printf STDERR "%s around line %s: %s\n",
565    $self->{'source_filename'} || 'Pod input', $line, $complaint;
566}
567
568sub _complain_errata {
569  my($self,$line,$complaint) = @_;
570  if( $self->{'no_errata_section'} ) {
571    DEBUG > 9 and print STDERR "Discarding erratum (at line $line) $complaint\n because no_errata_section is on.\n";
572  } else {
573    DEBUG > 9 and print STDERR "Queuing erratum (at line $line) $complaint\n";
574    push @{$self->{'errata'}{$line}}, $complaint
575      # for a report to be generated later!
576  }
577  return 1;
578}
579
580#@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
581
582sub _get_initial_item_type {
583  # A hack-wrapper here for when you have like "=over\n\n=item 456\n\n"
584  my($self, $para) = @_;
585  return $para->[1]{'~type'}  if $para->[1]{'~type'};
586
587  return $para->[1]{'~type'} = 'text'
588   if join("\n", @{$para}[2 .. $#$para]) =~ m/^\s*(\d+)\.?\s*$/s and $1 ne '1';
589  # Else fall thru to the general case:
590  return $self->_get_item_type($para);
591}
592
593
594
595sub _get_item_type {       # mutates the item!!
596  my($self, $para) = @_;
597  return $para->[1]{'~type'} if $para->[1]{'~type'};
598
599
600  # Otherwise we haven't yet been to this node.  Maybe alter it...
601
602  my $content = join "\n", @{$para}[2 .. $#$para];
603
604  if($content =~ m/^\s*\*\s*$/s or $content =~ m/^\s*$/s) {
605    # Like: "=item *", "=item   *   ", "=item"
606    splice @$para, 2; # so it ends up just being ['=item', { attrhash } ]
607    $para->[1]{'~orig_content'} = $content;
608    return $para->[1]{'~type'} = 'bullet';
609
610  } elsif($content =~ m/^\s*\*\s+(.+)/s) {  # tolerance
611
612    # Like: "=item * Foo bar baz";
613    $para->[1]{'~orig_content'}      = $content;
614    $para->[1]{'~_freaky_para_hack'} = $1;
615    DEBUG > 2 and print STDERR " Tolerating $$para[2] as =item *\\n\\n$1\n";
616    splice @$para, 2; # so it ends up just being ['=item', { attrhash } ]
617    return $para->[1]{'~type'} = 'bullet';
618
619  } elsif($content =~ m/^\s*(\d+)\.?\s*$/s) {
620    # Like: "=item 1.", "=item    123412"
621
622    $para->[1]{'~orig_content'} = $content;
623    $para->[1]{'number'} = $1;  # Yes, stores the number there!
624
625    splice @$para, 2; # so it ends up just being ['=item', { attrhash } ]
626    return $para->[1]{'~type'} = 'number';
627
628  } else {
629    # It's anything else.
630    return $para->[1]{'~type'} = 'text';
631
632  }
633}
634
635#-----------------------------------------------------------------------------
636
637sub _make_treelet {
638  my $self = shift;  # and ($para, $start_line)
639  my $treelet;
640  if(!@_) {
641    return [''];
642  } if(ref $_[0] and ref $_[0][0] and $_[0][0][0] eq '~Top') {
643    # Hack so we can pass in fake-o pre-cooked paragraphs:
644    #  just have the first line be a reference to a ['~Top', {}, ...]
645    # We use this feechure in gen_errata and stuff.
646
647    DEBUG and print STDERR "Applying precooked treelet hack to $_[0][0]\n";
648    $treelet = $_[0][0];
649    splice @$treelet, 0, 2;  # lop the top off
650    return $treelet;
651  } else {
652    $treelet = $self->_treelet_from_formatting_codes(@_);
653  }
654
655  if( ! $self->{'_output_is_for_JustPod'}   # Retain these as-is for pod output
656     && $self->_remap_sequences($treelet) )
657  {
658    $self->_treat_Zs($treelet);  # Might as well nix these first
659    $self->_treat_Ls($treelet);  # L has to precede E and S
660    $self->_treat_Es($treelet);
661    $self->_treat_Ss($treelet);  # S has to come after E
662    $self->_wrap_up($treelet); # Nix X's and merge texties
663
664  } else {
665    DEBUG and print STDERR "Formatless treelet gets fast-tracked.\n";
666     # Very common case!
667  }
668
669  splice @$treelet, 0, 2;  # lop the top off
670
671  return $treelet;
672}
673
674#:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.
675
676sub _wrap_up {
677  my($self, @stack) = @_;
678  my $nixx  = $self->{'nix_X_codes'};
679  my $merge = $self->{'merge_text' };
680  return unless $nixx or $merge;
681
682  DEBUG > 2 and print STDERR "\nStarting _wrap_up traversal.\n",
683   $merge ? (" Merge mode on\n") : (),
684   $nixx  ? (" Nix-X mode on\n") : (),
685  ;
686
687
688  my($i, $treelet);
689  while($treelet = shift @stack) {
690    DEBUG > 3 and print STDERR " Considering children of this $treelet->[0] node...\n";
691    for($i = 2; $i < @$treelet; ++$i) { # iterate over children
692      DEBUG > 3 and print STDERR " Considering child at $i ", pretty($treelet->[$i]), "\n";
693      if($nixx and ref $treelet->[$i] and $treelet->[$i][0] eq 'X') {
694        DEBUG > 3 and print STDERR "   Nixing X node at $i\n";
695        splice(@$treelet, $i, 1); # just nix this node (and its descendants)
696        # no need to back-update the counter just yet
697        redo;
698
699      } elsif($merge and $i != 2 and  # non-initial
700         !ref $treelet->[$i] and !ref $treelet->[$i - 1]
701      ) {
702        DEBUG > 3 and print STDERR "   Merging ", $i-1,
703         ":[$treelet->[$i-1]] and $i\:[$treelet->[$i]]\n";
704        $treelet->[$i-1] .= ( splice(@$treelet, $i, 1) )[0];
705        DEBUG > 4 and print STDERR "    Now: ", $i-1, ":[$treelet->[$i-1]]\n";
706        --$i;
707        next;
708        # since we just pulled the possibly last node out from under
709        #  ourselves, we can't just redo()
710
711      } elsif( ref $treelet->[$i] ) {
712        DEBUG > 4 and print STDERR "  Enqueuing ", pretty($treelet->[$i]), " for traversal.\n";
713        push @stack, $treelet->[$i];
714
715        if($treelet->[$i][0] eq 'L') {
716          my $thing;
717          foreach my $attrname ('section', 'to') {
718            if(defined($thing = $treelet->[$i][1]{$attrname}) and ref $thing) {
719              unshift @stack, $thing;
720              DEBUG > 4 and print STDERR "  +Enqueuing ",
721               pretty( $treelet->[$i][1]{$attrname} ),
722               " as an attribute value to tweak.\n";
723            }
724          }
725        }
726      }
727    }
728  }
729  DEBUG > 2 and print STDERR "End of _wrap_up traversal.\n\n";
730
731  return;
732}
733
734#:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.
735
736sub _remap_sequences {
737  my($self,@stack) = @_;
738
739  if(@stack == 1 and @{ $stack[0] } == 3 and !ref $stack[0][2]) {
740    # VERY common case: abort it.
741    DEBUG and print STDERR "Skipping _remap_sequences: formatless treelet.\n";
742    return 0;
743  }
744
745  my $map = ($self->{'accept_codes'} || die "NO accept_codes in $self?!?");
746
747  my $start_line = $stack[0][1]{'start_line'};
748  DEBUG > 2 and printf
749   "\nAbout to start _remap_sequences on treelet from line %s.\n",
750   $start_line || '[?]'
751  ;
752  DEBUG > 3 and print STDERR " Map: ",
753    join('; ', map "$_=" . (
754        ref($map->{$_}) ? join(",", @{$map->{$_}}) : $map->{$_}
755      ),
756      sort keys %$map ),
757    ("B~C~E~F~I~L~S~X~Z" eq join '~', sort keys %$map)
758     ? "  (all normal)\n" : "\n"
759  ;
760
761  # A recursive algorithm implemented iteratively!  Whee!
762
763  my($is, $was, $i, $treelet); # scratch
764  while($treelet = shift @stack) {
765    DEBUG > 3 and print STDERR " Considering children of this $treelet->[0] node...\n";
766    for($i = 2; $i < @$treelet; ++$i) { # iterate over children
767      next unless ref $treelet->[$i];  # text nodes are uninteresting
768
769      DEBUG > 4 and print STDERR "  Noting child $i : $treelet->[$i][0]<...>\n";
770
771      $is = $treelet->[$i][0] = $map->{ $was = $treelet->[$i][0] };
772      if( DEBUG > 3 ) {
773        if(!defined $is) {
774          print STDERR "   Code $was<> is UNKNOWN!\n";
775        } elsif($is eq $was) {
776          DEBUG > 4 and print STDERR "   Code $was<> stays the same.\n";
777        } else  {
778          print STDERR "   Code $was<> maps to ",
779           ref($is)
780            ? ( "tags ", map("$_<", @$is), '...', map('>', @$is), "\n" )
781            : "tag $is<...>.\n";
782        }
783      }
784
785      if(!defined $is) {
786        $self->whine($start_line, "Deleting unknown formatting code $was<>");
787        $is = $treelet->[$i][0] = '1';  # But saving the children!
788        # I could also insert a leading "$was<" and tailing ">" as
789        # children of this node, but something about that seems icky.
790      }
791      if(ref $is) {
792        my @dynasty = @$is;
793        DEBUG > 4 and print STDERR "    Renaming $was node to $dynasty[-1]\n";
794        $treelet->[$i][0] = pop @dynasty;
795        my $nugget;
796        while(@dynasty) {
797          DEBUG > 4 and printf
798           "    Grafting a new %s node between %s and %s\n",
799           $dynasty[-1], $treelet->[0], $treelet->[$i][0],
800          ;
801
802          #$nugget = ;
803          splice @$treelet, $i, 1, [pop(@dynasty), {}, $treelet->[$i]];
804            # relace node with a new parent
805        }
806      } elsif($is eq '0') {
807        splice(@$treelet, $i, 1); # just nix this node (and its descendants)
808        --$i;  # back-update the counter
809      } elsif($is eq '1') {
810        splice(@$treelet, $i, 1 # replace this node with its children!
811          => splice @{ $treelet->[$i] },2
812              # (not catching its first two (non-child) items)
813        );
814        --$i;  # back up for new stuff
815      } else {
816        # otherwise it's unremarkable
817        unshift @stack, $treelet->[$i];  # just recurse
818      }
819    }
820  }
821
822  DEBUG > 2 and print STDERR "End of _remap_sequences traversal.\n\n";
823
824  if(@_ == 2 and @{ $_[1] } == 3 and !ref $_[1][2]) {
825    DEBUG and print STDERR "Noting that the treelet is now formatless.\n";
826    return 0;
827  }
828  return 1;
829}
830
831# . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
832
833sub _ponder_extend {
834
835  # "Go to an extreme, move back to a more comfortable place"
836  #  -- /Oblique Strategies/,  Brian Eno and Peter Schmidt
837
838  my($self, $para) = @_;
839  my $content = join ' ', splice @$para, 2;
840  $content =~ s/^\s+//s;
841  $content =~ s/\s+$//s;
842
843  DEBUG > 2 and print STDERR "Ogling extensor: =extend $content\n";
844
845  if($content =~
846    m/^
847      (\S+)         # 1 : new item
848      \s+
849      (\S+)         # 2 : fallback(s)
850      (?:\s+(\S+))? # 3 : element name(s)
851      \s*
852      $
853    /xs
854  ) {
855    my $new_letter = $1;
856    my $fallbacks_one = $2;
857    my $elements_one;
858    $elements_one = defined($3) ? $3 : $1;
859
860    DEBUG > 2 and print STDERR "Extensor has good syntax.\n";
861
862    unless($new_letter =~ m/^[A-Z]$/s or $new_letter) {
863      DEBUG > 2 and print STDERR " $new_letter isn't a valid thing to entend.\n";
864      $self->whine(
865        $para->[1]{'start_line'},
866        "You can extend only formatting codes A-Z, not like \"$new_letter\""
867      );
868      return;
869    }
870
871    if(grep $new_letter eq $_, @Known_formatting_codes) {
872      DEBUG > 2 and print STDERR " $new_letter isn't a good thing to extend, because known.\n";
873      $self->whine(
874        $para->[1]{'start_line'},
875        "You can't extend an established code like \"$new_letter\""
876      );
877
878      #TODO: or allow if last bit is same?
879
880      return;
881    }
882
883    unless($fallbacks_one =~ m/^[A-Z](,[A-Z])*$/s  # like "B", "M,I", etc.
884      or $fallbacks_one eq '0' or $fallbacks_one eq '1'
885    ) {
886      $self->whine(
887        $para->[1]{'start_line'},
888        "Format for second =extend parameter must be like"
889        . " M or 1 or 0 or M,N or M,N,O but you have it like "
890        . $fallbacks_one
891      );
892      return;
893    }
894
895    unless($elements_one =~ m/^[^ ,]+(,[^ ,]+)*$/s) { # like "B", "M,I", etc.
896      $self->whine(
897        $para->[1]{'start_line'},
898        "Format for third =extend parameter: like foo or bar,Baz,qu:ux but not like "
899        . $elements_one
900      );
901      return;
902    }
903
904    my @fallbacks  = split ',', $fallbacks_one,  -1;
905    my @elements   = split ',', $elements_one, -1;
906
907    foreach my $f (@fallbacks) {
908      next if exists $Known_formatting_codes{$f} or $f eq '0' or $f eq '1';
909      DEBUG > 2 and print STDERR "  Can't fall back on unknown code $f\n";
910      $self->whine(
911        $para->[1]{'start_line'},
912        "Can't use unknown formatting code '$f' as a fallback for '$new_letter'"
913      );
914      return;
915    }
916
917    DEBUG > 3 and printf STDERR "Extensor: Fallbacks <%s> Elements <%s>.\n",
918     @fallbacks, @elements;
919
920    my $canonical_form;
921    foreach my $e (@elements) {
922      if(exists $self->{'accept_codes'}{$e}) {
923        DEBUG > 1 and print STDERR " Mapping '$new_letter' to known extension '$e'\n";
924        $canonical_form = $e;
925        last; # first acceptable elementname wins!
926      } else {
927        DEBUG > 1 and print STDERR " Can't map '$new_letter' to unknown extension '$e'\n";
928      }
929    }
930
931
932    if( defined $canonical_form ) {
933      # We found a good N => elementname mapping
934      $self->{'accept_codes'}{$new_letter} = $canonical_form;
935      DEBUG > 2 and print
936       "Extensor maps $new_letter => known element $canonical_form.\n";
937    } else {
938      # We have to use the fallback(s), which might be '0', or '1'.
939      $self->{'accept_codes'}{$new_letter}
940        = (@fallbacks == 1) ? $fallbacks[0] : \@fallbacks;
941      DEBUG > 2 and print
942       "Extensor maps $new_letter => fallbacks @fallbacks.\n";
943    }
944
945  } else {
946    DEBUG > 2 and print STDERR "Extensor has bad syntax.\n";
947    $self->whine(
948      $para->[1]{'start_line'},
949      "Unknown =extend syntax: $content"
950    )
951  }
952  return;
953}
954
955
956#:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.
957
958sub _treat_Zs {  # Nix Z<...>'s
959  my($self,@stack) = @_;
960
961  my($i, $treelet);
962  my $start_line = $stack[0][1]{'start_line'};
963
964  # A recursive algorithm implemented iteratively!  Whee!
965
966  while($treelet = shift @stack) {
967    for($i = 2; $i < @$treelet; ++$i) { # iterate over children
968      next unless ref $treelet->[$i];  # text nodes are uninteresting
969      unless($treelet->[$i][0] eq 'Z') {
970        unshift @stack, $treelet->[$i]; # recurse
971        next;
972      }
973
974      DEBUG > 1 and print STDERR "Nixing Z node @{$treelet->[$i]}\n";
975
976      # bitch UNLESS it's empty
977      unless(  @{$treelet->[$i]} == 2
978           or (@{$treelet->[$i]} == 3 and $treelet->[$i][2] eq '')
979      ) {
980        $self->whine( $start_line, "A non-empty Z<>" );
981      }      # but kill it anyway
982
983      splice(@$treelet, $i, 1); # thereby just nix this node.
984      --$i;
985
986    }
987  }
988
989  return;
990}
991
992# . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
993
994# Quoting perlpodspec:
995
996# In parsing an L<...> code, Pod parsers must distinguish at least four
997# attributes:
998
999############# Not used.  Expressed via the element children plus
1000#############  the value of the "content-implicit" flag.
1001# First:
1002# The link-text. If there is none, this must be undef. (E.g., in "L<Perl
1003# Functions|perlfunc>", the link-text is "Perl Functions". In
1004# "L<Time::HiRes>" and even "L<|Time::HiRes>", there is no link text. Note
1005# that link text may contain formatting.)
1006#
1007
1008############# The element children
1009# Second:
1010# The possibly inferred link-text -- i.e., if there was no real link text,
1011# then this is the text that we'll infer in its place. (E.g., for
1012# "L<Getopt::Std>", the inferred link text is "Getopt::Std".)
1013#
1014
1015############# The "to" attribute (which might be text, or a treelet)
1016# Third:
1017# The name or URL, or undef if none. (E.g., in "L<Perl
1018# Functions|perlfunc>", the name -- also sometimes called the page -- is
1019# "perlfunc". In "L</CAVEATS>", the name is undef.)
1020#
1021
1022############# The "section" attribute (which might be next, or a treelet)
1023# Fourth:
1024# The section (AKA "item" in older perlpods), or undef if none. E.g., in
1025# Getopt::Std/DESCRIPTION, "DESCRIPTION" is the section. (Note that this
1026# is not the same as a manpage section like the "5" in "man 5 crontab".
1027# "Section Foo" in the Pod sense means the part of the text that's
1028# introduced by the heading or item whose text is "Foo".)
1029#
1030# Pod parsers may also note additional attributes including:
1031#
1032
1033############# The "type" attribute.
1034# Fifth:
1035# A flag for whether item 3 (if present) is a URL (like
1036# "http://lists.perl.org" is), in which case there should be no section
1037# attribute; a Pod name (like "perldoc" and "Getopt::Std" are); or
1038# possibly a man page name (like "crontab(5)" is).
1039#
1040
1041############# The "raw" attribute that is already there.
1042# Sixth:
1043# The raw original L<...> content, before text is split on "|", "/", etc,
1044# and before E<...> codes are expanded.
1045
1046
1047# For L<...> codes without a "name|" part, only E<...> and Z<> codes may
1048# occur -- no other formatting codes. That is, authors should not use
1049# "L<B<Foo::Bar>>".
1050#
1051# Note, however, that formatting codes and Z<>'s can occur in any and all
1052# parts of an L<...> (i.e., in name, section, text, and url).
1053
1054sub _treat_Ls {  # Process our dear dear friends, the L<...> sequences
1055
1056  # L<name>
1057  # L<name/"sec"> or L<name/sec>
1058  # L</"sec"> or L</sec> or L<"sec">
1059  # L<text|name>
1060  # L<text|name/"sec"> or L<text|name/sec>
1061  # L<text|/"sec"> or L<text|/sec> or L<text|"sec">
1062  # L<scheme:...>
1063  # L<text|scheme:...>
1064
1065  my($self,@stack) = @_;
1066
1067  my($i, $treelet);
1068  my $start_line = $stack[0][1]{'start_line'};
1069
1070  # A recursive algorithm implemented iteratively!  Whee!
1071
1072  while($treelet = shift @stack) {
1073    for(my $i = 2; $i < @$treelet; ++$i) {
1074      # iterate over children of current tree node
1075      next unless ref $treelet->[$i];  # text nodes are uninteresting
1076      unless($treelet->[$i][0] eq 'L') {
1077        unshift @stack, $treelet->[$i]; # recurse
1078        next;
1079      }
1080
1081
1082      # By here, $treelet->[$i] is definitely an L node
1083      my $ell = $treelet->[$i];
1084      DEBUG > 1 and print STDERR "Ogling L node " . pretty($ell) . "\n";
1085
1086      # bitch if it's empty or is just '/'
1087      if (@{$ell} == 3 and $ell->[2] =~ m!\A\s*/\s*\z!) {
1088        $self->whine( $start_line, "L<> contains only '/'" );
1089        $treelet->[$i] = 'L</>';  # just make it a text node
1090        next;  # and move on
1091      }
1092      if(  @{$ell} == 2
1093       or (@{$ell} == 3 and $ell->[2] eq '')
1094      ) {
1095        $self->whine( $start_line, "An empty L<>" );
1096        $treelet->[$i] = 'L<>';  # just make it a text node
1097        next;  # and move on
1098      }
1099
1100      if( (! ref $ell->[2]  && $ell->[2] =~ /\A\s/)
1101        ||(! ref $ell->[-1] && $ell->[-1] =~ /\s\z/)
1102      ) {
1103        $self->whine( $start_line, "L<> starts or ends with whitespace" );
1104      }
1105
1106      # Catch URLs:
1107
1108      # there are a number of possible cases:
1109      # 1) text node containing url: http://foo.com
1110      #   -> [ 'http://foo.com' ]
1111      # 2) text node containing url and text: foo|http://foo.com
1112      #   -> [ 'foo|http://foo.com' ]
1113      # 3) text node containing url start: mailto:xE<at>foo.com
1114      #   -> [ 'mailto:x', [ E ... ], 'foo.com' ]
1115      # 4) text node containing url start and text: foo|mailto:xE<at>foo.com
1116      #   -> [ 'foo|mailto:x', [ E ... ], 'foo.com' ]
1117      # 5) other nodes containing text and url start: OE<39>Malley|http://foo.com
1118      #   -> [ 'O', [ E ... ], 'Malley', '|http://foo.com' ]
1119      # ... etc.
1120
1121      # anything before the url is part of the text.
1122      # anything after it is part of the url.
1123      # the url text node itself may contain parts of both.
1124
1125      if (my ($url_index, $text_part, $url_part) =
1126        # grep is no good here; we want to bail out immediately so that we can
1127        # use $1, $2, etc. without having to do the match twice.
1128        sub {
1129          for (2..$#$ell) {
1130            next if ref $ell->[$_];
1131            next unless $ell->[$_] =~ m/^(?:([^|]*)\|)?(\w+:[^:\s]\S*)$/s;
1132            return ($_, $1, $2);
1133          }
1134          return;
1135        }->()
1136      ) {
1137        $ell->[1]{'type'} = 'url';
1138
1139        my @text = @{$ell}[2..$url_index-1];
1140        push @text, $text_part if defined $text_part;
1141
1142        my @url  = @{$ell}[$url_index+1..$#$ell];
1143        unshift @url, $url_part;
1144
1145        unless (@text) {
1146          $ell->[1]{'content-implicit'} = 'yes';
1147          @text = @url;
1148        }
1149
1150        $ell->[1]{to} = Pod::Simple::LinkSection->new(
1151          @url == 1
1152          ? $url[0]
1153          : [ '', {}, @url ],
1154        );
1155
1156        splice @$ell, 2, $#$ell, @text;
1157
1158        next;
1159      }
1160
1161      # Catch some very simple and/or common cases
1162      if(@{$ell} == 3 and ! ref $ell->[2]) {
1163        my $it = $ell->[2];
1164        if($it =~ m{^[^/|]+[(][-a-zA-Z0-9]+[)]$}s) { # man sections
1165          # Hopefully neither too broad nor too restrictive a RE
1166          DEBUG > 1 and print STDERR "Catching \"$it\" as manpage link.\n";
1167          $ell->[1]{'type'} = 'man';
1168          # This's the only place where man links can get made.
1169          $ell->[1]{'content-implicit'} = 'yes';
1170          $ell->[1]{'to'  } =
1171            Pod::Simple::LinkSection->new( $it ); # treelet!
1172
1173          next;
1174        }
1175        if($it =~ m/^[^\/\|,\$\%\@\ \"\<\>\:\#\&\*\{\}\[\]\(\)]+(\:\:[^\/\|,\$\%\@\ \"\<\>\:\#\&\*\{\}\[\]\(\)]+)*$/s) {
1176          # Extremely forgiving idea of what constitutes a bare
1177          #  modulename link like L<Foo::Bar> or even L<Thing::1.0::Docs::Tralala>
1178          DEBUG > 1 and print STDERR "Catching \"$it\" as ho-hum L<Modulename> link.\n";
1179          $ell->[1]{'type'} = 'pod';
1180          $ell->[1]{'content-implicit'} = 'yes';
1181          $ell->[1]{'to'  } =
1182            Pod::Simple::LinkSection->new( $it ); # treelet!
1183          next;
1184        }
1185        # else fall thru...
1186      }
1187
1188
1189
1190      # ...Uhoh, here's the real L<...> parsing stuff...
1191      # "With the ill behavior, with the ill behavior, with the ill behavior..."
1192
1193      DEBUG > 1 and print STDERR "Running a real parse on this non-trivial L\n";
1194
1195
1196      my $link_text; # set to an arrayref if found
1197      my @ell_content = @$ell;
1198      splice @ell_content,0,2; # Knock off the 'L' and {} bits
1199
1200      DEBUG > 3 and print STDERR " Ell content to start: ",
1201       pretty(@ell_content), "\n";
1202
1203
1204      # Look for the "|" -- only in CHILDREN (not all underlings!)
1205      # Like L<I like the strictness|strict>
1206      DEBUG > 3 and
1207         print STDERR "  Peering at L content for a '|' ...\n";
1208      for(my $j = 0; $j < @ell_content; ++$j) {
1209        next if ref $ell_content[$j];
1210        DEBUG > 3 and
1211         print STDERR "    Peering at L-content text bit \"$ell_content[$j]\" for a '|'.\n";
1212
1213        if($ell_content[$j] =~ m/^([^\|]*)\|(.*)$/s) {
1214          my @link_text = ($1);   # might be 0-length
1215          $ell_content[$j] = $2;  # might be 0-length
1216
1217          DEBUG > 3 and
1218           print STDERR "     FOUND a '|' in it.  Splitting into [$1] + [$2]\n";
1219
1220          if ($link_text[0] =~ m{[|/]}) {
1221            $self->whine(
1222              $start_line,
1223              "alternative text '$link_text[0]' contains non-escaped | or /"
1224            );
1225          }
1226
1227          unshift @link_text, splice @ell_content, 0, $j;
1228            # leaving only things at J and after
1229          @ell_content =  grep ref($_)||length($_), @ell_content ;
1230          $link_text   = [grep ref($_)||length($_), @link_text  ];
1231          DEBUG > 3 and printf
1232           "  So link text is %s\n  and remaining ell content is %s\n",
1233            pretty($link_text), pretty(@ell_content);
1234          last;
1235        }
1236      }
1237
1238
1239      # Now look for the "/" -- only in CHILDREN (not all underlings!)
1240      # And afterward, anything left in @ell_content will be the raw name
1241      # Like L<Foo::Bar/Object Methods>
1242      my $section_name;  # set to arrayref if found
1243      DEBUG > 3 and print STDERR "  Peering at L-content for a '/' ...\n";
1244      for(my $j = 0; $j < @ell_content; ++$j) {
1245        next if ref $ell_content[$j];
1246        DEBUG > 3 and
1247         print STDERR "    Peering at L-content text bit \"$ell_content[$j]\" for a '/'.\n";
1248
1249        if($ell_content[$j] =~ m/^([^\/]*)\/(.*)$/s) {
1250          my @section_name = ($2); # might be 0-length
1251          $ell_content[$j] =  $1;  # might be 0-length
1252
1253          DEBUG > 3 and
1254           print STDERR "     FOUND a '/' in it.",
1255             "  Splitting to page [...$1] + section [$2...]\n";
1256
1257          push @section_name, splice @ell_content, 1+$j;
1258            # leaving only things before and including J
1259
1260          @ell_content  = grep ref($_)||length($_), @ell_content  ;
1261          @section_name = grep ref($_)||length($_), @section_name ;
1262
1263          # Turn L<.../"foo"> into L<.../foo>
1264          if(@section_name
1265            and !ref($section_name[0]) and !ref($section_name[-1])
1266            and $section_name[ 0] =~ m/^\"/s
1267            and $section_name[-1] =~ m/\"$/s
1268            and !( # catch weird degenerate case of L<"> !
1269              @section_name == 1 and $section_name[0] eq '"'
1270            )
1271          ) {
1272            $section_name[ 0] =~ s/^\"//s;
1273            $section_name[-1] =~ s/\"$//s;
1274            DEBUG > 3 and
1275             print STDERR "     Quotes removed: ", pretty(@section_name), "\n";
1276          } else {
1277            DEBUG > 3 and
1278             print STDERR "     No need to remove quotes in ", pretty(@section_name), "\n";
1279          }
1280
1281          $section_name = \@section_name;
1282          last;
1283        }
1284      }
1285
1286      # Turn L<"Foo Bar"> into L</Foo Bar>
1287      if(!$section_name and @ell_content
1288         and !ref($ell_content[0]) and !ref($ell_content[-1])
1289         and $ell_content[ 0] =~ m/^\"/s
1290         and $ell_content[-1] =~ m/\"$/s
1291         and !( # catch weird degenerate case of L<"> !
1292           @ell_content == 1 and $ell_content[0] eq '"'
1293         )
1294      ) {
1295        $section_name = [splice @ell_content];
1296        $section_name->[ 0] =~ s/^\"//s;
1297        $section_name->[-1] =~ s/\"$//s;
1298        $ell->[1]{'~tolerated'} = 1;
1299      }
1300
1301      # Turn L<Foo Bar> into L</Foo Bar>.
1302      if(!$section_name and !$link_text and @ell_content
1303         and grep !ref($_) && m/ /s, @ell_content
1304      ) {
1305        $section_name = [splice @ell_content];
1306        $ell->[1]{'~deprecated'} = 1;
1307        # That's support for the now-deprecated syntax.
1308        # Note that it deliberately won't work on L<...|Foo Bar>
1309      }
1310
1311
1312      # Now make up the link_text
1313      # L<Foo>     -> L<Foo|Foo>
1314      # L</Bar>    -> L<"Bar"|Bar>
1315      # L<Foo/Bar> -> L<"Bar" in Foo/Foo>
1316      unless($link_text) {
1317        $ell->[1]{'content-implicit'} = 'yes';
1318        $link_text = [];
1319        push @$link_text, '"', @$section_name, '"' if $section_name;
1320
1321        if(@ell_content) {
1322          $link_text->[-1] .= ' in ' if $section_name;
1323          push @$link_text, @ell_content;
1324        }
1325      }
1326
1327
1328      # And the E resolver will have to deal with all our treeletty things:
1329
1330      if(@ell_content == 1 and !ref($ell_content[0])
1331         and $ell_content[0] =~ m{^[^/]+[(][-a-zA-Z0-9]+[)]$}s
1332      ) {
1333        $ell->[1]{'type'}    = 'man';
1334        DEBUG > 3 and print STDERR "Considering this ($ell_content[0]) a man link.\n";
1335      } else {
1336        $ell->[1]{'type'}    = 'pod';
1337        DEBUG > 3 and print STDERR "Considering this a pod link (not man or url).\n";
1338      }
1339
1340      if( defined $section_name ) {
1341        $ell->[1]{'section'} = Pod::Simple::LinkSection->new(
1342          ['', {}, @$section_name]
1343        );
1344        DEBUG > 3 and print STDERR "L-section content: ", pretty($ell->[1]{'section'}), "\n";
1345      }
1346
1347      if( @ell_content ) {
1348        $ell->[1]{'to'} = Pod::Simple::LinkSection->new(
1349          ['', {}, @ell_content]
1350        );
1351        DEBUG > 3 and print STDERR "L-to content: ", pretty($ell->[1]{'to'}), "\n";
1352      }
1353
1354      # And update children to be the link-text:
1355      @$ell = (@$ell[0,1], defined($link_text) ? splice(@$link_text) : '');
1356
1357      DEBUG > 2 and print STDERR "End of L-parsing for this node " . pretty($treelet->[$i]) . "\n";
1358
1359      unshift @stack, $treelet->[$i]; # might as well recurse
1360    }
1361  }
1362
1363  return;
1364}
1365
1366# . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1367
1368sub _treat_Es {
1369  my($self,@stack) = @_;
1370
1371  my($i, $treelet, $content, $replacer, $charnum);
1372  my $start_line = $stack[0][1]{'start_line'};
1373
1374  # A recursive algorithm implemented iteratively!  Whee!
1375
1376
1377  # Has frightening side effects on L nodes' attributes.
1378
1379  #my @ells_to_tweak;
1380
1381  while($treelet = shift @stack) {
1382    for(my $i = 2; $i < @$treelet; ++$i) { # iterate over children
1383      next unless ref $treelet->[$i];  # text nodes are uninteresting
1384      if($treelet->[$i][0] eq 'L') {
1385        # SPECIAL STUFF for semi-processed L<>'s
1386
1387        my $thing;
1388        foreach my $attrname ('section', 'to') {
1389          if(defined($thing = $treelet->[$i][1]{$attrname}) and ref $thing) {
1390            unshift @stack, $thing;
1391            DEBUG > 2 and print STDERR "  Enqueuing ",
1392             pretty( $treelet->[$i][1]{$attrname} ),
1393             " as an attribute value to tweak.\n";
1394          }
1395        }
1396
1397        unshift @stack, $treelet->[$i]; # recurse
1398        next;
1399      } elsif($treelet->[$i][0] ne 'E') {
1400        unshift @stack, $treelet->[$i]; # recurse
1401        next;
1402      }
1403
1404      DEBUG > 1 and print STDERR "Ogling E node ", pretty($treelet->[$i]), "\n";
1405
1406      # bitch if it's empty
1407      if(  @{$treelet->[$i]} == 2
1408       or (@{$treelet->[$i]} == 3 and $treelet->[$i][2] eq '')
1409      ) {
1410        $self->whine( $start_line, "An empty E<>" );
1411        $treelet->[$i] = 'E<>'; # splice in a literal
1412        next;
1413      }
1414
1415      # bitch if content is weird
1416      unless(@{$treelet->[$i]} == 3 and !ref($content = $treelet->[$i][2])) {
1417        $self->whine( $start_line, "An E<...> surrounding strange content" );
1418        $replacer = $treelet->[$i]; # scratch
1419        splice(@$treelet, $i, 1,   # fake out a literal
1420          'E<',
1421          splice(@$replacer,2), # promote its content
1422          '>'
1423        );
1424        # Don't need to do --$i, as the 'E<' we just added isn't interesting.
1425        next;
1426      }
1427
1428      DEBUG > 1 and print STDERR "Ogling E<$content>\n";
1429
1430      # XXX E<>'s contents *should* be a valid char in the scope of the current
1431      # =encoding directive. Defaults to iso-8859-1, I believe. Fix this in the
1432      # future sometime.
1433
1434      $charnum  = Pod::Escapes::e2charnum($content);
1435      DEBUG > 1 and print STDERR " Considering E<$content> with char ",
1436        defined($charnum) ? $charnum : "undef", ".\n";
1437
1438      if(!defined( $charnum )) {
1439        DEBUG > 1 and print STDERR "I don't know how to deal with E<$content>.\n";
1440        $self->whine( $start_line, "Unknown E content in E<$content>" );
1441        $replacer = "E<$content>"; # better than nothing
1442      } elsif($charnum >= 255 and !UNICODE) {
1443        $replacer = ASCII ? "\xA4" : "?";
1444        DEBUG > 1 and print STDERR "This Perl version can't handle ",
1445          "E<$content> (chr $charnum), so replacing with $replacer\n";
1446      } else {
1447        $replacer = Pod::Escapes::e2char($content);
1448        DEBUG > 1 and print STDERR " Replacing E<$content> with $replacer\n";
1449      }
1450
1451      splice(@$treelet, $i, 1, $replacer); # no need to back up $i, tho
1452    }
1453  }
1454
1455  return;
1456}
1457
1458
1459# . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1460
1461sub _treat_Ss {
1462  my($self,$treelet) = @_;
1463
1464  _change_S_to_nbsp($treelet,0) if $self->{'nbsp_for_S'};
1465
1466  # TODO: or a change_nbsp_to_S
1467  #  Normalizing nbsp's to S is harder: for each text node, make S content
1468  #  out of anything matching m/([^ \xA0]*(?:\xA0+[^ \xA0]*)+)/
1469
1470
1471  return;
1472}
1473
1474sub _change_S_to_nbsp { #  a recursive function
1475  # Sanely assumes that the top node in the excursion won't be an S node.
1476  my($treelet, $in_s) = @_;
1477
1478  my $is_s = ('S' eq $treelet->[0]);
1479  $in_s ||= $is_s; # So in_s is on either by this being an S element,
1480                   #  or by an ancestor being an S element.
1481
1482  for(my $i = 2; $i < @$treelet; ++$i) {
1483    if(ref $treelet->[$i]) {
1484      if( _change_S_to_nbsp( $treelet->[$i], $in_s ) ) {
1485        my $to_pull_up = $treelet->[$i];
1486        splice @$to_pull_up,0,2;   # ...leaving just its content
1487        splice @$treelet, $i, 1, @$to_pull_up;  # Pull up content
1488        $i +=  @$to_pull_up - 1;   # Make $i skip the pulled-up stuff
1489      }
1490    } else {
1491      $treelet->[$i] =~ s/\s/$Pod::Simple::nbsp/g if $in_s;
1492
1493       # Note that if you apply nbsp_for_S to text, and so turn
1494       # "foo S<bar baz> quux" into "foo bar&#160;faz quux", you
1495       # end up with something that fails to say "and don't hyphenate
1496       # any part of 'bar baz'".  However, hyphenation is such a vexing
1497       # problem anyway, that most Pod renderers just don't render it
1498       # at all.  But if you do want to implement hyphenation, I guess
1499       # that you'd better have nbsp_for_S off.
1500    }
1501  }
1502
1503  return $is_s;
1504}
1505
1506#-----------------------------------------------------------------------------
1507
1508sub _accessorize {  # A simple-minded method-maker
1509  no strict 'refs';
1510  foreach my $attrname (@_) {
1511    next if $attrname =~ m/::/; # a hack
1512    *{caller() . '::' . $attrname} = sub {
1513      use strict;
1514      $Carp::CarpLevel = 1,  Carp::croak(
1515       "Accessor usage: \$obj->$attrname() or \$obj->$attrname(\$new_value)"
1516      ) unless (@_ == 1 or @_ == 2) and ref $_[0];
1517
1518      (@_ == 1) ?  $_[0]->{$attrname}
1519                : ($_[0]->{$attrname} = $_[1]);
1520    };
1521  }
1522  # Ya know, they say accessories make the ensemble!
1523  return;
1524}
1525
1526# . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1527# . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1528#=============================================================================
1529
1530sub filter {
1531  my($class, $source) = @_;
1532  my $new = $class->new;
1533  $new->output_fh(*STDOUT{IO});
1534
1535  if(ref($source || '') eq 'SCALAR') {
1536    $new->parse_string_document( $$source );
1537  } elsif(ref($source)) {  # it's a file handle
1538    $new->parse_file($source);
1539  } else {  # it's a filename
1540    $new->parse_file($source);
1541  }
1542
1543  return $new;
1544}
1545
1546
1547#-----------------------------------------------------------------------------
1548
1549sub _out {
1550  # For use in testing: Class->_out($source)
1551  #  returns the transformation of $source
1552
1553  my $class = shift(@_);
1554
1555  my $mutor = shift(@_) if @_ and ref($_[0] || '') eq 'CODE';
1556
1557  DEBUG and print STDERR "\n\n", '#' x 76,
1558   "\nAbout to parse source: {{\n$_[0]\n}}\n\n";
1559
1560
1561  my $parser = ref $class && $class->isa(__PACKAGE__) ? $class : $class->new;
1562  $parser->hide_line_numbers(1);
1563
1564  my $out = '';
1565  $parser->output_string( \$out );
1566  DEBUG and print STDERR " _out to ", \$out, "\n";
1567
1568  $mutor->($parser) if $mutor;
1569
1570  $parser->parse_string_document( $_[0] );
1571  # use Data::Dumper; print STDERR Dumper($parser), "\n";
1572  return $out;
1573}
1574
1575
1576sub _duo {
1577  # For use in testing: Class->_duo($source1, $source2)
1578  #  returns the parse trees of $source1 and $source2.
1579  # Good in things like: &ok( Class->duo(... , ...) );
1580
1581  my $class = shift(@_);
1582
1583  Carp::croak "But $class->_duo is useful only in list context!"
1584   unless wantarray;
1585
1586  my $mutor = shift(@_) if @_ and ref($_[0] || '') eq 'CODE';
1587
1588  Carp::croak "But $class->_duo takes two parameters, not: @_"
1589   unless @_ == 2;
1590
1591  my(@out);
1592
1593  while( @_ ) {
1594    my $parser = $class->new;
1595
1596    push @out, '';
1597    $parser->output_string( \( $out[-1] ) );
1598
1599    DEBUG and print STDERR " _duo out to ", $parser->output_string(),
1600      " = $parser->{'output_string'}\n";
1601
1602    $parser->hide_line_numbers(1);
1603    $mutor->($parser) if $mutor;
1604    $parser->parse_string_document( shift( @_ ) );
1605    # use Data::Dumper; print STDERR Dumper($parser), "\n";
1606  }
1607
1608  return @out;
1609}
1610
1611
1612
1613#-----------------------------------------------------------------------------
16141;
1615__END__
1616
1617TODO:
1618A start_formatting_code and end_formatting_code methods, which in the
1619base class call start_L, end_L, start_C, end_C, etc., if they are
1620defined.
1621
1622have the POD FORMATTING ERRORS section note the localtime, and the
1623version of Pod::Simple.
1624
1625option to delete all E<shy>s?
1626option to scream if under-0x20 literals are found in the input, or
1627under-E<32> E codes are found in the tree. And ditto \x7f-\x9f
1628
1629Option to turn highbit characters into their compromised form? (applies
1630to E parsing too)
1631
1632TODO: BOM/encoding things.
1633
1634TODO: ascii-compat things in the XML classes?
1635
1636