Commit | Line | Data |
3fea05b9 |
1 | |
2 | require 5; |
3 | package Pod::Simple; |
4 | use strict; |
5 | use Carp (); |
6 | BEGIN { *DEBUG = sub () {0} unless defined &DEBUG } |
7 | use integer; |
8 | use Pod::Escapes 1.04 (); |
9 | use Pod::Simple::LinkSection (); |
10 | use Pod::Simple::BlackBox (); |
11 | #use utf8; |
12 | |
13 | use vars qw( |
14 | $VERSION @ISA |
15 | @Known_formatting_codes @Known_directives |
16 | %Known_formatting_codes %Known_directives |
17 | $NL |
18 | ); |
19 | |
20 | @ISA = ('Pod::Simple::BlackBox'); |
21 | $VERSION = '3.10'; |
22 | |
23 | @Known_formatting_codes = qw(I B C L E F S X Z); |
24 | %Known_formatting_codes = map(($_=>1), @Known_formatting_codes); |
25 | @Known_directives = qw(head1 head2 head3 head4 item over back); |
26 | %Known_directives = map(($_=>'Plain'), @Known_directives); |
27 | $NL = $/ unless defined $NL; |
28 | |
29 | #----------------------------------------------------------------------------- |
30 | # Set up some constants: |
31 | |
32 | BEGIN { |
33 | if(defined &ASCII) { } |
34 | elsif(chr(65) eq 'A') { *ASCII = sub () {1} } |
35 | else { *ASCII = sub () {''} } |
36 | |
37 | unless(defined &MANY_LINES) { *MANY_LINES = sub () {20} } |
38 | DEBUG > 4 and print "MANY_LINES is ", MANY_LINES(), "\n"; |
39 | unless(MANY_LINES() >= 1) { |
40 | die "MANY_LINES is too small (", MANY_LINES(), ")!\nAborting"; |
41 | } |
42 | if(defined &UNICODE) { } |
43 | elsif($] >= 5.008) { *UNICODE = sub() {1} } |
44 | else { *UNICODE = sub() {''} } |
45 | } |
46 | if(DEBUG > 2) { |
47 | print "# We are ", ASCII ? '' : 'not ', "in ASCII-land\n"; |
48 | print "# We are under a Unicode-safe Perl.\n"; |
49 | } |
50 | |
51 | # Design note: |
52 | # This is a parser for Pod. It is not a parser for the set of Pod-like |
53 | # languages which happens to contain Pod -- it is just for Pod, plus possibly |
54 | # some extensions. |
55 | |
56 | # @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ |
57 | #@ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ |
58 | #@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ |
59 | |
60 | __PACKAGE__->_accessorize( |
61 | 'nbsp_for_S', # Whether to map S<...>'s to \xA0 characters |
62 | 'source_filename', # Filename of the source, for use in warnings |
63 | 'source_dead', # Whether to consider this parser's source dead |
64 | |
65 | 'output_fh', # The filehandle we're writing to, if applicable. |
66 | # Used only in some derived classes. |
67 | |
68 | 'hide_line_numbers', # For some dumping subclasses: whether to pointedly |
69 | # suppress the start_line attribute |
70 | |
71 | 'line_count', # the current line number |
72 | 'pod_para_count', # count of pod paragraphs seen so far |
73 | |
74 | 'no_whining', # whether to suppress whining |
75 | 'no_errata_section', # whether to suppress the errata section |
76 | 'complain_stderr', # whether to complain to stderr |
77 | |
78 | 'doc_has_started', # whether we've fired the open-Document event yet |
79 | |
80 | 'bare_output', # For some subclasses: whether to prepend |
81 | # header-code and postpend footer-code |
82 | |
83 | 'fullstop_space_harden', # Whether to turn ". " into ".[nbsp] "; |
84 | |
85 | 'nix_X_codes', # whether to ignore X<...> codes |
86 | 'merge_text', # whether to avoid breaking a single piece of |
87 | # text up into several events |
88 | |
89 | 'preserve_whitespace', # whether to try to keep whitespace as-is |
90 | 'strip_verbatim_indent', # What indent to strip from verbatim |
91 | |
92 | 'content_seen', # whether we've seen any real Pod content |
93 | 'errors_seen', # TODO: document. whether we've seen any errors (fatal or not) |
94 | |
95 | 'codes_in_verbatim', # for PseudoPod extensions |
96 | |
97 | 'code_handler', # coderef to call when a code (non-pod) line is seen |
98 | 'cut_handler', # coderef to call when a =cut line is seen |
99 | #Called like: |
100 | # $code_handler->($line, $self->{'line_count'}, $self) if $code_handler; |
101 | # $cut_handler->($line, $self->{'line_count'}, $self) if $cut_handler; |
102 | |
103 | ); |
104 | |
105 | #@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ |
106 | |
107 | sub any_errata_seen { # good for using as an exit() value... |
108 | return shift->{'errors_seen'} || 0; |
109 | } |
110 | |
111 | #@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ |
112 | # Pull in some functions that, for some reason, I expect to see here too: |
113 | BEGIN { |
114 | *pretty = \&Pod::Simple::BlackBox::pretty; |
115 | *stringify_lol = \&Pod::Simple::BlackBox::stringify_lol; |
116 | } |
117 | |
118 | #@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ |
119 | |
120 | sub version_report { |
121 | my $class = ref($_[0]) || $_[0]; |
122 | if($class eq __PACKAGE__) { |
123 | return "$class $VERSION"; |
124 | } else { |
125 | my $v = $class->VERSION; |
126 | return "$class $v (" . __PACKAGE__ . " $VERSION)"; |
127 | } |
128 | } |
129 | |
130 | #@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ |
131 | |
132 | #sub curr_open { # read-only list accessor |
133 | # return @{ $_[0]{'curr_open'} || return() }; |
134 | #} |
135 | #sub _curr_open_listref { $_[0]{'curr_open'} ||= [] } |
136 | |
137 | |
138 | sub output_string { |
139 | # Works by faking out output_fh. Simplifies our code. |
140 | # |
141 | my $this = shift; |
142 | return $this->{'output_string'} unless @_; # GET. |
143 | |
144 | require Pod::Simple::TiedOutFH; |
145 | my $x = (defined($_[0]) and ref($_[0])) ? $_[0] : \( $_[0] ); |
146 | $$x = '' unless defined $$x; |
147 | DEBUG > 4 and print "# Output string set to $x ($$x)\n"; |
148 | $this->{'output_fh'} = Pod::Simple::TiedOutFH->handle_on($_[0]); |
149 | return |
150 | $this->{'output_string'} = $_[0]; |
151 | #${ ${ $this->{'output_fh'} } }; |
152 | } |
153 | |
154 | sub abandon_output_string { $_[0]->abandon_output_fh; delete $_[0]{'output_string'} } |
155 | sub abandon_output_fh { $_[0]->output_fh(undef) } |
156 | # These don't delete the string or close the FH -- they just delete our |
157 | # references to it/them. |
158 | # TODO: document these |
159 | |
160 | #@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ |
161 | |
162 | sub new { |
163 | # takes no parameters |
164 | my $class = ref($_[0]) || $_[0]; |
165 | #Carp::croak(__PACKAGE__ . " is a virtual base class -- see perldoc " |
166 | # . __PACKAGE__ ); |
167 | return bless { |
168 | 'accept_codes' => { map( ($_=>$_), @Known_formatting_codes ) }, |
169 | 'accept_directives' => { %Known_directives }, |
170 | 'accept_targets' => {}, |
171 | }, $class; |
172 | } |
173 | |
174 | |
175 | |
176 | # TODO: an option for whether to interpolate E<...>'s, or just resolve to codes. |
177 | |
178 | #@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ |
179 | |
180 | sub _handle_element_start { # OVERRIDE IN DERIVED CLASS |
181 | my($self, $element_name, $attr_hash_r) = @_; |
182 | return; |
183 | } |
184 | |
185 | sub _handle_element_end { # OVERRIDE IN DERIVED CLASS |
186 | my($self, $element_name) = @_; |
187 | return; |
188 | } |
189 | |
190 | sub _handle_text { # OVERRIDE IN DERIVED CLASS |
191 | my($self, $text) = @_; |
192 | return; |
193 | } |
194 | |
195 | #@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ |
196 | # |
197 | # And now directives (not targets) |
198 | |
199 | sub accept_directive_as_verbatim { shift->_accept_directives('Verbatim', @_) } |
200 | sub accept_directive_as_data { shift->_accept_directives('Data', @_) } |
201 | sub accept_directive_as_processed { shift->_accept_directives('Plain', @_) } |
202 | |
203 | sub _accept_directives { |
204 | my($this, $type) = splice @_,0,2; |
205 | foreach my $d (@_) { |
206 | next unless defined $d and length $d; |
207 | Carp::croak "\"$d\" isn't a valid directive name" |
208 | unless $d =~ m/^[a-zA-Z][a-zA-Z0-9]*$/s; |
209 | Carp::croak "\"$d\" is already a reserved Pod directive name" |
210 | if exists $Known_directives{$d}; |
211 | $this->{'accept_directives'}{$d} = $type; |
212 | DEBUG > 2 and print "Learning to accept \"=$d\" as directive of type $type\n"; |
213 | } |
214 | DEBUG > 6 and print "$this\'s accept_directives : ", |
215 | pretty($this->{'accept_directives'}), "\n"; |
216 | |
217 | return sort keys %{ $this->{'accept_directives'} } if wantarray; |
218 | return; |
219 | } |
220 | |
221 | #-------------------------------------------------------------------------- |
222 | # TODO: document these: |
223 | |
224 | sub unaccept_directive { shift->unaccept_directives(@_) }; |
225 | |
226 | sub unaccept_directives { |
227 | my $this = shift; |
228 | foreach my $d (@_) { |
229 | next unless defined $d and length $d; |
230 | Carp::croak "\"$d\" isn't a valid directive name" |
231 | unless $d =~ m/^[a-zA-Z][a-zA-Z0-9]*$/s; |
232 | Carp::croak "But you must accept \"$d\" directives -- it's a builtin!" |
233 | if exists $Known_directives{$d}; |
234 | delete $this->{'accept_directives'}{$d}; |
235 | DEBUG > 2 and print "OK, won't accept \"=$d\" as directive.\n"; |
236 | } |
237 | return sort keys %{ $this->{'accept_directives'} } if wantarray; |
238 | return |
239 | } |
240 | |
241 | #@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ |
242 | # |
243 | # And now targets (not directives) |
244 | |
245 | sub accept_target { shift->accept_targets(@_) } # alias |
246 | sub accept_target_as_text { shift->accept_targets_as_text(@_) } # alias |
247 | |
248 | |
249 | sub accept_targets { shift->_accept_targets('1', @_) } |
250 | |
251 | sub accept_targets_as_text { shift->_accept_targets('force_resolve', @_) } |
252 | # forces them to be processed, even when there's no ":". |
253 | |
254 | sub _accept_targets { |
255 | my($this, $type) = splice @_,0,2; |
256 | foreach my $t (@_) { |
257 | next unless defined $t and length $t; |
258 | # TODO: enforce some limitations on what a target name can be? |
259 | $this->{'accept_targets'}{$t} = $type; |
260 | DEBUG > 2 and print "Learning to accept \"$t\" as target of type $type\n"; |
261 | } |
262 | return sort keys %{ $this->{'accept_targets'} } if wantarray; |
263 | return; |
264 | } |
265 | |
266 | #-------------------------------------------------------------------------- |
267 | sub unaccept_target { shift->unaccept_targets(@_) } |
268 | |
269 | sub unaccept_targets { |
270 | my $this = shift; |
271 | foreach my $t (@_) { |
272 | next unless defined $t and length $t; |
273 | # TODO: enforce some limitations on what a target name can be? |
274 | delete $this->{'accept_targets'}{$t}; |
275 | DEBUG > 2 and print "OK, won't accept \"$t\" as target.\n"; |
276 | } |
277 | return sort keys %{ $this->{'accept_targets'} } if wantarray; |
278 | return; |
279 | } |
280 | |
281 | #@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ |
282 | # |
283 | # And now codes (not targets or directives) |
284 | |
285 | sub accept_code { shift->accept_codes(@_) } # alias |
286 | |
287 | sub accept_codes { # Add some codes |
288 | my $this = shift; |
289 | |
290 | foreach my $new_code (@_) { |
291 | next unless defined $new_code and length $new_code; |
292 | if(ASCII) { |
293 | # A good-enough check that it's good as an XML Name symbol: |
294 | Carp::croak "\"$new_code\" isn't a valid element name" |
295 | if $new_code =~ |
296 | m/[\x00-\x2C\x2F\x39\x3B-\x40\x5B-\x5E\x60\x7B-\x7F]/ |
297 | # Characters under 0x80 that aren't legal in an XML Name. |
298 | or $new_code =~ m/^[-\.0-9]/s |
299 | or $new_code =~ m/:[-\.0-9]/s; |
300 | # The legal under-0x80 Name characters that |
301 | # an XML Name still can't start with. |
302 | } |
303 | |
304 | $this->{'accept_codes'}{$new_code} = $new_code; |
305 | |
306 | # Yes, map to itself -- just so that when we |
307 | # see "=extend W [whatever] thatelementname", we say that W maps |
308 | # to whatever $this->{accept_codes}{thatelementname} is, |
309 | # i.e., "thatelementname". Then when we go re-mapping, |
310 | # a "W" in the treelet turns into "thatelementname". We only |
311 | # remap once. |
312 | # If we say we accept "W", then a "W" in the treelet simply turns |
313 | # into "W". |
314 | } |
315 | |
316 | return; |
317 | } |
318 | |
319 | #-------------------------------------------------------------------------- |
320 | sub unaccept_code { shift->unaccept_codes(@_) } |
321 | |
322 | sub unaccept_codes { # remove some codes |
323 | my $this = shift; |
324 | |
325 | foreach my $new_code (@_) { |
326 | next unless defined $new_code and length $new_code; |
327 | if(ASCII) { |
328 | # A good-enough check that it's good as an XML Name symbol: |
329 | Carp::croak "\"$new_code\" isn't a valid element name" |
330 | if $new_code =~ |
331 | m/[\x00-\x2C\x2F\x39\x3B-\x40\x5B-\x5E\x60\x7B-\x7F]/ |
332 | # Characters under 0x80 that aren't legal in an XML Name. |
333 | or $new_code =~ m/^[-\.0-9]/s |
334 | or $new_code =~ m/:[-\.0-9]/s; |
335 | # The legal under-0x80 Name characters that |
336 | # an XML Name still can't start with. |
337 | } |
338 | |
339 | Carp::croak "But you must accept \"$new_code\" codes -- it's a builtin!" |
340 | if grep $new_code eq $_, @Known_formatting_codes; |
341 | |
342 | delete $this->{'accept_codes'}{$new_code}; |
343 | |
344 | DEBUG > 2 and print "OK, won't accept the code $new_code<...>.\n"; |
345 | } |
346 | |
347 | return; |
348 | } |
349 | |
350 | |
351 | #@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ |
352 | #@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ |
353 | |
354 | sub parse_string_document { |
355 | my $self = shift; |
356 | my @lines; |
357 | foreach my $line_group (@_) { |
358 | next unless defined $line_group and length $line_group; |
359 | pos($line_group) = 0; |
360 | while($line_group =~ |
361 | m/([^\n\r]*)((?:\r?\n)?)/g |
362 | ) { |
363 | #print(">> $1\n"), |
364 | $self->parse_lines($1) |
365 | if length($1) or length($2) |
366 | or pos($line_group) != length($line_group); |
367 | # I.e., unless it's a zero-length "empty line" at the very |
368 | # end of "foo\nbar\n" (i.e., between the \n and the EOS). |
369 | } |
370 | } |
371 | $self->parse_lines(undef); # to signal EOF |
372 | return $self; |
373 | } |
374 | |
375 | sub _init_fh_source { |
376 | my($self, $source) = @_; |
377 | |
378 | #DEBUG > 1 and print "Declaring $source as :raw for starters\n"; |
379 | #$self->_apply_binmode($source, ':raw'); |
380 | #binmode($source, ":raw"); |
381 | |
382 | return; |
383 | } |
384 | |
385 | #:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:. |
386 | # |
387 | |
388 | sub parse_file { |
389 | my($self, $source) = (@_); |
390 | |
391 | if(!defined $source) { |
392 | Carp::croak("Can't use empty-string as a source for parse_file"); |
393 | } elsif(ref(\$source) eq 'GLOB') { |
394 | $self->{'source_filename'} = '' . ($source); |
395 | } elsif(ref $source) { |
396 | $self->{'source_filename'} = '' . ($source); |
397 | } elsif(!length $source) { |
398 | Carp::croak("Can't use empty-string as a source for parse_file"); |
399 | } else { |
400 | { |
401 | local *PODSOURCE; |
402 | open(PODSOURCE, "<$source") || Carp::croak("Can't open $source: $!"); |
403 | $self->{'source_filename'} = $source; |
404 | $source = *PODSOURCE{IO}; |
405 | } |
406 | $self->_init_fh_source($source); |
407 | } |
408 | # By here, $source is a FH. |
409 | |
410 | $self->{'source_fh'} = $source; |
411 | |
412 | my($i, @lines); |
413 | until( $self->{'source_dead'} ) { |
414 | splice @lines; |
415 | for($i = MANY_LINES; $i--;) { # read those many lines at a time |
416 | local $/ = $NL; |
417 | push @lines, scalar(<$source>); # readline |
418 | last unless defined $lines[-1]; |
419 | # but pass thru the undef, which will set source_dead to true |
420 | } |
421 | $self->parse_lines(@lines); |
422 | } |
423 | delete($self->{'source_fh'}); # so it can be GC'd |
424 | return $self; |
425 | } |
426 | |
427 | #:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:. |
428 | |
429 | sub parse_from_file { |
430 | # An emulation of Pod::Parser's interface, for the sake of Perldoc. |
431 | # Basically just a wrapper around parse_file. |
432 | |
433 | my($self, $source, $to) = @_; |
434 | $self = $self->new unless ref($self); # so we tolerate being a class method |
435 | |
436 | if(!defined $source) { $source = *STDIN{IO} |
437 | } elsif(ref(\$source) eq 'GLOB') { # stet |
438 | } elsif(ref($source) ) { # stet |
439 | } elsif(!length $source |
440 | or $source eq '-' or $source =~ m/^<&(STDIN|0)$/i |
441 | ) { |
442 | $source = *STDIN{IO}; |
443 | } |
444 | |
445 | if(!defined $to) { $self->output_fh( *STDOUT{IO} ); |
446 | } elsif(ref(\$to) eq 'GLOB') { $self->output_fh( $to ); |
447 | } elsif(ref($to)) { $self->output_fh( $to ); |
448 | } elsif(!length $to |
449 | or $to eq '-' or $to =~ m/^>&?(?:STDOUT|1)$/i |
450 | ) { |
451 | $self->output_fh( *STDOUT{IO} ); |
452 | } else { |
453 | require Symbol; |
454 | my $out_fh = Symbol::gensym(); |
455 | DEBUG and print "Write-opening to $to\n"; |
456 | open($out_fh, ">$to") or Carp::croak "Can't write-open $to: $!"; |
457 | binmode($out_fh) |
458 | if $self->can('write_with_binmode') and $self->write_with_binmode; |
459 | $self->output_fh($out_fh); |
460 | } |
461 | |
462 | return $self->parse_file($source); |
463 | } |
464 | |
465 | #----------------------------------------------------------------------------- |
466 | |
467 | sub whine { |
468 | #my($self,$line,$complaint) = @_; |
469 | my $self = shift(@_); |
470 | ++$self->{'errors_seen'}; |
471 | if($self->{'no_whining'}) { |
472 | DEBUG > 9 and print "Discarding complaint (at line $_[0]) $_[1]\n because no_whining is on.\n"; |
473 | return; |
474 | } |
475 | return $self->_complain_warn(@_) if $self->{'complain_stderr'}; |
476 | return $self->_complain_errata(@_); |
477 | } |
478 | |
479 | sub scream { # like whine, but not suppressable |
480 | #my($self,$line,$complaint) = @_; |
481 | my $self = shift(@_); |
482 | ++$self->{'errors_seen'}; |
483 | return $self->_complain_warn(@_) if $self->{'complain_stderr'}; |
484 | return $self->_complain_errata(@_); |
485 | } |
486 | |
487 | sub _complain_warn { |
488 | my($self,$line,$complaint) = @_; |
489 | return printf STDERR "%s around line %s: %s\n", |
490 | $self->{'source_filename'} || 'Pod input', $line, $complaint; |
491 | } |
492 | |
493 | sub _complain_errata { |
494 | my($self,$line,$complaint) = @_; |
495 | if( $self->{'no_errata_section'} ) { |
496 | DEBUG > 9 and print "Discarding erratum (at line $line) $complaint\n because no_errata_section is on.\n"; |
497 | } else { |
498 | DEBUG > 9 and print "Queuing erratum (at line $line) $complaint\n"; |
499 | push @{$self->{'errata'}{$line}}, $complaint |
500 | # for a report to be generated later! |
501 | } |
502 | return 1; |
503 | } |
504 | |
505 | #@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ |
506 | |
507 | sub _get_initial_item_type { |
508 | # A hack-wrapper here for when you have like "=over\n\n=item 456\n\n" |
509 | my($self, $para) = @_; |
510 | return $para->[1]{'~type'} if $para->[1]{'~type'}; |
511 | |
512 | return $para->[1]{'~type'} = 'text' |
513 | if join("\n", @{$para}[2 .. $#$para]) =~ m/^\s*(\d+)\.?\s*$/s and $1 ne '1'; |
514 | # Else fall thru to the general case: |
515 | return $self->_get_item_type($para); |
516 | } |
517 | |
518 | |
519 | |
520 | sub _get_item_type { # mutates the item!! |
521 | my($self, $para) = @_; |
522 | return $para->[1]{'~type'} if $para->[1]{'~type'}; |
523 | |
524 | |
525 | # Otherwise we haven't yet been to this node. Maybe alter it... |
526 | |
527 | my $content = join "\n", @{$para}[2 .. $#$para]; |
528 | |
529 | if($content =~ m/^\s*\*\s*$/s or $content =~ m/^\s*$/s) { |
530 | # Like: "=item *", "=item * ", "=item" |
531 | splice @$para, 2; # so it ends up just being ['=item', { attrhash } ] |
532 | $para->[1]{'~orig_content'} = $content; |
533 | return $para->[1]{'~type'} = 'bullet'; |
534 | |
535 | } elsif($content =~ m/^\s*\*\s+(.+)/s) { # tolerance |
536 | |
537 | # Like: "=item * Foo bar baz"; |
538 | $para->[1]{'~orig_content'} = $content; |
539 | $para->[1]{'~_freaky_para_hack'} = $1; |
540 | DEBUG > 2 and print " Tolerating $$para[2] as =item *\\n\\n$1\n"; |
541 | splice @$para, 2; # so it ends up just being ['=item', { attrhash } ] |
542 | return $para->[1]{'~type'} = 'bullet'; |
543 | |
544 | } elsif($content =~ m/^\s*(\d+)\.?\s*$/s) { |
545 | # Like: "=item 1.", "=item 123412" |
546 | |
547 | $para->[1]{'~orig_content'} = $content; |
548 | $para->[1]{'number'} = $1; # Yes, stores the number there! |
549 | |
550 | splice @$para, 2; # so it ends up just being ['=item', { attrhash } ] |
551 | return $para->[1]{'~type'} = 'number'; |
552 | |
553 | } else { |
554 | # It's anything else. |
555 | return $para->[1]{'~type'} = 'text'; |
556 | |
557 | } |
558 | } |
559 | |
560 | #----------------------------------------------------------------------------- |
561 | |
562 | sub _make_treelet { |
563 | my $self = shift; # and ($para, $start_line) |
564 | my $treelet; |
565 | if(!@_) { |
566 | return ['']; |
567 | } if(ref $_[0] and ref $_[0][0] and $_[0][0][0] eq '~Top') { |
568 | # Hack so we can pass in fake-o pre-cooked paragraphs: |
569 | # just have the first line be a reference to a ['~Top', {}, ...] |
570 | # We use this feechure in gen_errata and stuff. |
571 | |
572 | DEBUG and print "Applying precooked treelet hack to $_[0][0]\n"; |
573 | $treelet = $_[0][0]; |
574 | splice @$treelet, 0, 2; # lop the top off |
575 | return $treelet; |
576 | } else { |
577 | $treelet = $self->_treelet_from_formatting_codes(@_); |
578 | } |
579 | |
580 | if( $self->_remap_sequences($treelet) ) { |
581 | $self->_treat_Zs($treelet); # Might as well nix these first |
582 | $self->_treat_Ls($treelet); # L has to precede E and S |
583 | $self->_treat_Es($treelet); |
584 | $self->_treat_Ss($treelet); # S has to come after E |
585 | |
586 | $self->_wrap_up($treelet); # Nix X's and merge texties |
587 | |
588 | } else { |
589 | DEBUG and print "Formatless treelet gets fast-tracked.\n"; |
590 | # Very common case! |
591 | } |
592 | |
593 | splice @$treelet, 0, 2; # lop the top off |
594 | |
595 | return $treelet; |
596 | } |
597 | |
598 | #:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:. |
599 | |
600 | sub _wrap_up { |
601 | my($self, @stack) = @_; |
602 | my $nixx = $self->{'nix_X_codes'}; |
603 | my $merge = $self->{'merge_text' }; |
604 | return unless $nixx or $merge; |
605 | |
606 | DEBUG > 2 and print "\nStarting _wrap_up traversal.\n", |
607 | $merge ? (" Merge mode on\n") : (), |
608 | $nixx ? (" Nix-X mode on\n") : (), |
609 | ; |
610 | |
611 | |
612 | my($i, $treelet); |
613 | while($treelet = shift @stack) { |
614 | DEBUG > 3 and print " Considering children of this $treelet->[0] node...\n"; |
615 | for($i = 2; $i < @$treelet; ++$i) { # iterate over children |
616 | DEBUG > 3 and print " Considering child at $i ", pretty($treelet->[$i]), "\n"; |
617 | if($nixx and ref $treelet->[$i] and $treelet->[$i][0] eq 'X') { |
618 | DEBUG > 3 and print " Nixing X node at $i\n"; |
619 | splice(@$treelet, $i, 1); # just nix this node (and its descendants) |
620 | # no need to back-update the counter just yet |
621 | redo; |
622 | |
623 | } elsif($merge and $i != 2 and # non-initial |
624 | !ref $treelet->[$i] and !ref $treelet->[$i - 1] |
625 | ) { |
626 | DEBUG > 3 and print " Merging ", $i-1, |
627 | ":[$treelet->[$i-1]] and $i\:[$treelet->[$i]]\n"; |
628 | $treelet->[$i-1] .= ( splice(@$treelet, $i, 1) )[0]; |
629 | DEBUG > 4 and print " Now: ", $i-1, ":[$treelet->[$i-1]]\n"; |
630 | --$i; |
631 | next; |
632 | # since we just pulled the possibly last node out from under |
633 | # ourselves, we can't just redo() |
634 | |
635 | } elsif( ref $treelet->[$i] ) { |
636 | DEBUG > 4 and print " Enqueuing ", pretty($treelet->[$i]), " for traversal.\n"; |
637 | push @stack, $treelet->[$i]; |
638 | |
639 | if($treelet->[$i][0] eq 'L') { |
640 | my $thing; |
641 | foreach my $attrname ('section', 'to') { |
642 | if(defined($thing = $treelet->[$i][1]{$attrname}) and ref $thing) { |
643 | unshift @stack, $thing; |
644 | DEBUG > 4 and print " +Enqueuing ", |
645 | pretty( $treelet->[$i][1]{$attrname} ), |
646 | " as an attribute value to tweak.\n"; |
647 | } |
648 | } |
649 | } |
650 | } |
651 | } |
652 | } |
653 | DEBUG > 2 and print "End of _wrap_up traversal.\n\n"; |
654 | |
655 | return; |
656 | } |
657 | |
658 | #:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:. |
659 | |
660 | sub _remap_sequences { |
661 | my($self,@stack) = @_; |
662 | |
663 | if(@stack == 1 and @{ $stack[0] } == 3 and !ref $stack[0][2]) { |
664 | # VERY common case: abort it. |
665 | DEBUG and print "Skipping _remap_sequences: formatless treelet.\n"; |
666 | return 0; |
667 | } |
668 | |
669 | my $map = ($self->{'accept_codes'} || die "NO accept_codes in $self?!?"); |
670 | |
671 | my $start_line = $stack[0][1]{'start_line'}; |
672 | DEBUG > 2 and printf |
673 | "\nAbout to start _remap_sequences on treelet from line %s.\n", |
674 | $start_line || '[?]' |
675 | ; |
676 | DEBUG > 3 and print " Map: ", |
677 | join('; ', map "$_=" . ( |
678 | ref($map->{$_}) ? join(",", @{$map->{$_}}) : $map->{$_} |
679 | ), |
680 | sort keys %$map ), |
681 | ("B~C~E~F~I~L~S~X~Z" eq join '~', sort keys %$map) |
682 | ? " (all normal)\n" : "\n" |
683 | ; |
684 | |
685 | # A recursive algorithm implemented iteratively! Whee! |
686 | |
687 | my($is, $was, $i, $treelet); # scratch |
688 | while($treelet = shift @stack) { |
689 | DEBUG > 3 and print " Considering children of this $treelet->[0] node...\n"; |
690 | for($i = 2; $i < @$treelet; ++$i) { # iterate over children |
691 | next unless ref $treelet->[$i]; # text nodes are uninteresting |
692 | |
693 | DEBUG > 4 and print " Noting child $i : $treelet->[$i][0]<...>\n"; |
694 | |
695 | $is = $treelet->[$i][0] = $map->{ $was = $treelet->[$i][0] }; |
696 | if( DEBUG > 3 ) { |
697 | if(!defined $is) { |
698 | print " Code $was<> is UNKNOWN!\n"; |
699 | } elsif($is eq $was) { |
700 | DEBUG > 4 and print " Code $was<> stays the same.\n"; |
701 | } else { |
702 | print " Code $was<> maps to ", |
703 | ref($is) |
704 | ? ( "tags ", map("$_<", @$is), '...', map('>', @$is), "\n" ) |
705 | : "tag $is<...>.\n"; |
706 | } |
707 | } |
708 | |
709 | if(!defined $is) { |
710 | $self->whine($start_line, "Deleting unknown formatting code $was<>"); |
711 | $is = $treelet->[$i][0] = '1'; # But saving the children! |
712 | # I could also insert a leading "$was<" and tailing ">" as |
713 | # children of this node, but something about that seems icky. |
714 | } |
715 | if(ref $is) { |
716 | my @dynasty = @$is; |
717 | DEBUG > 4 and print " Renaming $was node to $dynasty[-1]\n"; |
718 | $treelet->[$i][0] = pop @dynasty; |
719 | my $nugget; |
720 | while(@dynasty) { |
721 | DEBUG > 4 and printf |
722 | " Grafting a new %s node between %s and %s\n", |
723 | $dynasty[-1], $treelet->[0], $treelet->[$i][0], |
724 | ; |
725 | |
726 | #$nugget = ; |
727 | splice @$treelet, $i, 1, [pop(@dynasty), {}, $treelet->[$i]]; |
728 | # relace node with a new parent |
729 | } |
730 | } elsif($is eq '0') { |
731 | splice(@$treelet, $i, 1); # just nix this node (and its descendants) |
732 | --$i; # back-update the counter |
733 | } elsif($is eq '1') { |
734 | splice(@$treelet, $i, 1 # replace this node with its children! |
735 | => splice @{ $treelet->[$i] },2 |
736 | # (not catching its first two (non-child) items) |
737 | ); |
738 | --$i; # back up for new stuff |
739 | } else { |
740 | # otherwise it's unremarkable |
741 | unshift @stack, $treelet->[$i]; # just recurse |
742 | } |
743 | } |
744 | } |
745 | |
746 | DEBUG > 2 and print "End of _remap_sequences traversal.\n\n"; |
747 | |
748 | if(@_ == 2 and @{ $_[1] } == 3 and !ref $_[1][2]) { |
749 | DEBUG and print "Noting that the treelet is now formatless.\n"; |
750 | return 0; |
751 | } |
752 | return 1; |
753 | } |
754 | |
755 | # . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . |
756 | |
757 | sub _ponder_extend { |
758 | |
759 | # "Go to an extreme, move back to a more comfortable place" |
760 | # -- /Oblique Strategies/, Brian Eno and Peter Schmidt |
761 | |
762 | my($self, $para) = @_; |
763 | my $content = join ' ', splice @$para, 2; |
764 | $content =~ s/^\s+//s; |
765 | $content =~ s/\s+$//s; |
766 | |
767 | DEBUG > 2 and print "Ogling extensor: =extend $content\n"; |
768 | |
769 | if($content =~ |
770 | m/^ |
771 | (\S+) # 1 : new item |
772 | \s+ |
773 | (\S+) # 2 : fallback(s) |
774 | (?:\s+(\S+))? # 3 : element name(s) |
775 | \s* |
776 | $ |
777 | /xs |
778 | ) { |
779 | my $new_letter = $1; |
780 | my $fallbacks_one = $2; |
781 | my $elements_one; |
782 | $elements_one = defined($3) ? $3 : $1; |
783 | |
784 | DEBUG > 2 and print "Extensor has good syntax.\n"; |
785 | |
786 | unless($new_letter =~ m/^[A-Z]$/s or $new_letter) { |
787 | DEBUG > 2 and print " $new_letter isn't a valid thing to entend.\n"; |
788 | $self->whine( |
789 | $para->[1]{'start_line'}, |
790 | "You can extend only formatting codes A-Z, not like \"$new_letter\"" |
791 | ); |
792 | return; |
793 | } |
794 | |
795 | if(grep $new_letter eq $_, @Known_formatting_codes) { |
796 | DEBUG > 2 and print " $new_letter isn't a good thing to extend, because known.\n"; |
797 | $self->whine( |
798 | $para->[1]{'start_line'}, |
799 | "You can't extend an established code like \"$new_letter\"" |
800 | ); |
801 | |
802 | #TODO: or allow if last bit is same? |
803 | |
804 | return; |
805 | } |
806 | |
807 | unless($fallbacks_one =~ m/^[A-Z](,[A-Z])*$/s # like "B", "M,I", etc. |
808 | or $fallbacks_one eq '0' or $fallbacks_one eq '1' |
809 | ) { |
810 | $self->whine( |
811 | $para->[1]{'start_line'}, |
812 | "Format for second =extend parameter must be like" |
813 | . " M or 1 or 0 or M,N or M,N,O but you have it like " |
814 | . $fallbacks_one |
815 | ); |
816 | return; |
817 | } |
818 | |
819 | unless($elements_one =~ m/^[^ ,]+(,[^ ,]+)*$/s) { # like "B", "M,I", etc. |
820 | $self->whine( |
821 | $para->[1]{'start_line'}, |
822 | "Format for third =extend parameter: like foo or bar,Baz,qu:ux but not like " |
823 | . $elements_one |
824 | ); |
825 | return; |
826 | } |
827 | |
828 | my @fallbacks = split ',', $fallbacks_one, -1; |
829 | my @elements = split ',', $elements_one, -1; |
830 | |
831 | foreach my $f (@fallbacks) { |
832 | next if exists $Known_formatting_codes{$f} or $f eq '0' or $f eq '1'; |
833 | DEBUG > 2 and print " Can't fall back on unknown code $f\n"; |
834 | $self->whine( |
835 | $para->[1]{'start_line'}, |
836 | "Can't use unknown formatting code '$f' as a fallback for '$new_letter'" |
837 | ); |
838 | return; |
839 | } |
840 | |
841 | DEBUG > 3 and printf "Extensor: Fallbacks <%s> Elements <%s>.\n", |
842 | @fallbacks, @elements; |
843 | |
844 | my $canonical_form; |
845 | foreach my $e (@elements) { |
846 | if(exists $self->{'accept_codes'}{$e}) { |
847 | DEBUG > 1 and print " Mapping '$new_letter' to known extension '$e'\n"; |
848 | $canonical_form = $e; |
849 | last; # first acceptable elementname wins! |
850 | } else { |
851 | DEBUG > 1 and print " Can't map '$new_letter' to unknown extension '$e'\n"; |
852 | } |
853 | } |
854 | |
855 | |
856 | if( defined $canonical_form ) { |
857 | # We found a good N => elementname mapping |
858 | $self->{'accept_codes'}{$new_letter} = $canonical_form; |
859 | DEBUG > 2 and print |
860 | "Extensor maps $new_letter => known element $canonical_form.\n"; |
861 | } else { |
862 | # We have to use the fallback(s), which might be '0', or '1'. |
863 | $self->{'accept_codes'}{$new_letter} |
864 | = (@fallbacks == 1) ? $fallbacks[0] : \@fallbacks; |
865 | DEBUG > 2 and print |
866 | "Extensor maps $new_letter => fallbacks @fallbacks.\n"; |
867 | } |
868 | |
869 | } else { |
870 | DEBUG > 2 and print "Extensor has bad syntax.\n"; |
871 | $self->whine( |
872 | $para->[1]{'start_line'}, |
873 | "Unknown =extend syntax: $content" |
874 | ) |
875 | } |
876 | return; |
877 | } |
878 | |
879 | |
880 | #:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:. |
881 | |
882 | sub _treat_Zs { # Nix Z<...>'s |
883 | my($self,@stack) = @_; |
884 | |
885 | my($i, $treelet); |
886 | my $start_line = $stack[0][1]{'start_line'}; |
887 | |
888 | # A recursive algorithm implemented iteratively! Whee! |
889 | |
890 | while($treelet = shift @stack) { |
891 | for($i = 2; $i < @$treelet; ++$i) { # iterate over children |
892 | next unless ref $treelet->[$i]; # text nodes are uninteresting |
893 | unless($treelet->[$i][0] eq 'Z') { |
894 | unshift @stack, $treelet->[$i]; # recurse |
895 | next; |
896 | } |
897 | |
898 | DEBUG > 1 and print "Nixing Z node @{$treelet->[$i]}\n"; |
899 | |
900 | # bitch UNLESS it's empty |
901 | unless( @{$treelet->[$i]} == 2 |
902 | or (@{$treelet->[$i]} == 3 and $treelet->[$i][2] eq '') |
903 | ) { |
904 | $self->whine( $start_line, "A non-empty Z<>" ); |
905 | } # but kill it anyway |
906 | |
907 | splice(@$treelet, $i, 1); # thereby just nix this node. |
908 | --$i; |
909 | |
910 | } |
911 | } |
912 | |
913 | return; |
914 | } |
915 | |
916 | # . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . |
917 | |
918 | # Quoting perlpodspec: |
919 | |
920 | # In parsing an L<...> code, Pod parsers must distinguish at least four |
921 | # attributes: |
922 | |
923 | ############# Not used. Expressed via the element children plus |
924 | ############# the value of the "content-implicit" flag. |
925 | # First: |
926 | # The link-text. If there is none, this must be undef. (E.g., in "L<Perl |
927 | # Functions|perlfunc>", the link-text is "Perl Functions". In |
928 | # "L<Time::HiRes>" and even "L<|Time::HiRes>", there is no link text. Note |
929 | # that link text may contain formatting.) |
930 | # |
931 | |
932 | ############# The element children |
933 | # Second: |
934 | # The possibly inferred link-text -- i.e., if there was no real link text, |
935 | # then this is the text that we'll infer in its place. (E.g., for |
936 | # "L<Getopt::Std>", the inferred link text is "Getopt::Std".) |
937 | # |
938 | |
939 | ############# The "to" attribute (which might be text, or a treelet) |
940 | # Third: |
941 | # The name or URL, or undef if none. (E.g., in "L<Perl |
942 | # Functions|perlfunc>", the name -- also sometimes called the page -- is |
943 | # "perlfunc". In "L</CAVEATS>", the name is undef.) |
944 | # |
945 | |
946 | ############# The "section" attribute (which might be next, or a treelet) |
947 | # Fourth: |
948 | # The section (AKA "item" in older perlpods), or undef if none. E.g., in |
949 | # Getopt::Std/DESCRIPTION, "DESCRIPTION" is the section. (Note that this |
950 | # is not the same as a manpage section like the "5" in "man 5 crontab". |
951 | # "Section Foo" in the Pod sense means the part of the text that's |
952 | # introduced by the heading or item whose text is "Foo".) |
953 | # |
954 | # Pod parsers may also note additional attributes including: |
955 | # |
956 | |
957 | ############# The "type" attribute. |
958 | # Fifth: |
959 | # A flag for whether item 3 (if present) is a URL (like |
960 | # "http://lists.perl.org" is), in which case there should be no section |
961 | # attribute; a Pod name (like "perldoc" and "Getopt::Std" are); or |
962 | # possibly a man page name (like "crontab(5)" is). |
963 | # |
964 | |
965 | ############# Not implemented, I guess. |
966 | # Sixth: |
967 | # The raw original L<...> content, before text is split on "|", "/", etc, |
968 | # and before E<...> codes are expanded. |
969 | |
970 | |
971 | # For L<...> codes without a "name|" part, only E<...> and Z<> codes may |
972 | # occur -- no other formatting codes. That is, authors should not use |
973 | # "L<B<Foo::Bar>>". |
974 | # |
975 | # Note, however, that formatting codes and Z<>'s can occur in any and all |
976 | # parts of an L<...> (i.e., in name, section, text, and url). |
977 | |
978 | sub _treat_Ls { # Process our dear dear friends, the L<...> sequences |
979 | |
980 | # L<name> |
981 | # L<name/"sec"> or L<name/sec> |
982 | # L</"sec"> or L</sec> or L<"sec"> |
983 | # L<text|name> |
984 | # L<text|name/"sec"> or L<text|name/sec> |
985 | # L<text|/"sec"> or L<text|/sec> or L<text|"sec"> |
986 | # L<scheme:...> |
987 | # Ltext|scheme:...> |
988 | |
989 | my($self,@stack) = @_; |
990 | |
991 | my($i, $treelet); |
992 | my $start_line = $stack[0][1]{'start_line'}; |
993 | |
994 | # A recursive algorithm implemented iteratively! Whee! |
995 | |
996 | while($treelet = shift @stack) { |
997 | for(my $i = 2; $i < @$treelet; ++$i) { |
998 | # iterate over children of current tree node |
999 | next unless ref $treelet->[$i]; # text nodes are uninteresting |
1000 | unless($treelet->[$i][0] eq 'L') { |
1001 | unshift @stack, $treelet->[$i]; # recurse |
1002 | next; |
1003 | } |
1004 | |
1005 | |
1006 | # By here, $treelet->[$i] is definitely an L node |
1007 | my $ell = $treelet->[$i]; |
1008 | DEBUG > 1 and print "Ogling L node $ell\n"; |
1009 | |
1010 | # bitch if it's empty |
1011 | if( @{$ell} == 2 |
1012 | or (@{$ell} == 3 and $ell->[2] eq '') |
1013 | ) { |
1014 | $self->whine( $start_line, "An empty L<>" ); |
1015 | $treelet->[$i] = 'L<>'; # just make it a text node |
1016 | next; # and move on |
1017 | } |
1018 | |
1019 | # Catch URLs: |
1020 | |
1021 | # there are a number of possible cases: |
1022 | # 1) text node containing url: http://foo.com |
1023 | # -> [ 'http://foo.com' ] |
1024 | # 2) text node containing url and text: foo|http://foo.com |
1025 | # -> [ 'foo|http://foo.com' ] |
1026 | # 3) text node containing url start: mailto:xE<at>foo.com |
1027 | # -> [ 'mailto:x', [ E ... ], 'foo.com' ] |
1028 | # 4) text node containing url start and text: foo|mailto:xE<at>foo.com |
1029 | # -> [ 'foo|mailto:x', [ E ... ], 'foo.com' ] |
1030 | # 5) other nodes containing text and url start: OE<39>Malley|http://foo.com |
1031 | # -> [ 'O', [ E ... ], 'Malley', '|http://foo.com' ] |
1032 | # ... etc. |
1033 | |
1034 | # anything before the url is part of the text. |
1035 | # anything after it is part of the url. |
1036 | # the url text node itself may contain parts of both. |
1037 | |
1038 | if (my ($url_index, $text_part, $url_part) = |
1039 | # grep is no good here; we want to bail out immediately so that we can |
1040 | # use $1, $2, etc. without having to do the match twice. |
1041 | sub { |
1042 | for (2..$#$ell) { |
1043 | next if ref $ell->[$_]; |
1044 | next unless $ell->[$_] =~ m/^(?:([^|]*)\|)?(\w+:[^:\s]\S*)$/s; |
1045 | return ($_, $1, $2); |
1046 | } |
1047 | return; |
1048 | }->() |
1049 | ) { |
1050 | $ell->[1]{'type'} = 'url'; |
1051 | |
1052 | my @text = @{$ell}[2..$url_index-1]; |
1053 | push @text, $text_part if defined $text_part; |
1054 | |
1055 | my @url = @{$ell}[$url_index+1..$#$ell]; |
1056 | unshift @url, $url_part; |
1057 | |
1058 | unless (@text) { |
1059 | $ell->[1]{'content-implicit'} = 'yes'; |
1060 | @text = @url; |
1061 | } |
1062 | |
1063 | $ell->[1]{to} = Pod::Simple::LinkSection->new( |
1064 | @url == 1 |
1065 | ? $url[0] |
1066 | : [ '', {}, @url ], |
1067 | ); |
1068 | |
1069 | splice @$ell, 2, $#$ell, @text; |
1070 | |
1071 | next; |
1072 | } |
1073 | |
1074 | # Catch some very simple and/or common cases |
1075 | if(@{$ell} == 3 and ! ref $ell->[2]) { |
1076 | my $it = $ell->[2]; |
1077 | if($it =~ m/^[-a-zA-Z0-9]+\([-a-zA-Z0-9]+\)$/s) { # man sections |
1078 | # Hopefully neither too broad nor too restrictive a RE |
1079 | DEBUG > 1 and print "Catching \"$it\" as manpage link.\n"; |
1080 | $ell->[1]{'type'} = 'man'; |
1081 | # This's the only place where man links can get made. |
1082 | $ell->[1]{'content-implicit'} = 'yes'; |
1083 | $ell->[1]{'to' } = |
1084 | Pod::Simple::LinkSection->new( $it ); # treelet! |
1085 | |
1086 | next; |
1087 | } |
1088 | if($it =~ m/^[^\/\|,\$\%\@\ \"\<\>\:\#\&\*\{\}\[\]\(\)]+(\:\:[^\/\|,\$\%\@\ \"\<\>\:\#\&\*\{\}\[\]\(\)]+)*$/s) { |
1089 | # Extremely forgiving idea of what constitutes a bare |
1090 | # modulename link like L<Foo::Bar> or even L<Thing::1.0::Docs::Tralala> |
1091 | DEBUG > 1 and print "Catching \"$it\" as ho-hum L<Modulename> link.\n"; |
1092 | $ell->[1]{'type'} = 'pod'; |
1093 | $ell->[1]{'content-implicit'} = 'yes'; |
1094 | $ell->[1]{'to' } = |
1095 | Pod::Simple::LinkSection->new( $it ); # treelet! |
1096 | next; |
1097 | } |
1098 | # else fall thru... |
1099 | } |
1100 | |
1101 | |
1102 | |
1103 | # ...Uhoh, here's the real L<...> parsing stuff... |
1104 | # "With the ill behavior, with the ill behavior, with the ill behavior..." |
1105 | |
1106 | DEBUG > 1 and print "Running a real parse on this non-trivial L\n"; |
1107 | |
1108 | |
1109 | my $link_text; # set to an arrayref if found |
1110 | my @ell_content = @$ell; |
1111 | splice @ell_content,0,2; # Knock off the 'L' and {} bits |
1112 | |
1113 | DEBUG > 3 and print " Ell content to start: ", |
1114 | pretty(@ell_content), "\n"; |
1115 | |
1116 | |
1117 | # Look for the "|" -- only in CHILDREN (not all underlings!) |
1118 | # Like L<I like the strictness|strict> |
1119 | DEBUG > 3 and |
1120 | print " Peering at L content for a '|' ...\n"; |
1121 | for(my $j = 0; $j < @ell_content; ++$j) { |
1122 | next if ref $ell_content[$j]; |
1123 | DEBUG > 3 and |
1124 | print " Peering at L-content text bit \"$ell_content[$j]\" for a '|'.\n"; |
1125 | |
1126 | if($ell_content[$j] =~ m/^([^\|]*)\|(.*)$/s) { |
1127 | my @link_text = ($1); # might be 0-length |
1128 | $ell_content[$j] = $2; # might be 0-length |
1129 | |
1130 | DEBUG > 3 and |
1131 | print " FOUND a '|' in it. Splitting into [$1] + [$2]\n"; |
1132 | |
1133 | unshift @link_text, splice @ell_content, 0, $j; |
1134 | # leaving only things at J and after |
1135 | @ell_content = grep ref($_)||length($_), @ell_content ; |
1136 | $link_text = [grep ref($_)||length($_), @link_text ]; |
1137 | DEBUG > 3 and printf |
1138 | " So link text is %s\n and remaining ell content is %s\n", |
1139 | pretty($link_text), pretty(@ell_content); |
1140 | last; |
1141 | } |
1142 | } |
1143 | |
1144 | |
1145 | # Now look for the "/" -- only in CHILDREN (not all underlings!) |
1146 | # And afterward, anything left in @ell_content will be the raw name |
1147 | # Like L<Foo::Bar/Object Methods> |
1148 | my $section_name; # set to arrayref if found |
1149 | DEBUG > 3 and print " Peering at L-content for a '/' ...\n"; |
1150 | for(my $j = 0; $j < @ell_content; ++$j) { |
1151 | next if ref $ell_content[$j]; |
1152 | DEBUG > 3 and |
1153 | print " Peering at L-content text bit \"$ell_content[$j]\" for a '/'.\n"; |
1154 | |
1155 | if($ell_content[$j] =~ m/^([^\/]*)\/(.*)$/s) { |
1156 | my @section_name = ($2); # might be 0-length |
1157 | $ell_content[$j] = $1; # might be 0-length |
1158 | |
1159 | DEBUG > 3 and |
1160 | print " FOUND a '/' in it.", |
1161 | " Splitting to page [...$1] + section [$2...]\n"; |
1162 | |
1163 | push @section_name, splice @ell_content, 1+$j; |
1164 | # leaving only things before and including J |
1165 | |
1166 | @ell_content = grep ref($_)||length($_), @ell_content ; |
1167 | @section_name = grep ref($_)||length($_), @section_name ; |
1168 | |
1169 | # Turn L<.../"foo"> into L<.../foo> |
1170 | if(@section_name |
1171 | and !ref($section_name[0]) and !ref($section_name[-1]) |
1172 | and $section_name[ 0] =~ m/^\"/s |
1173 | and $section_name[-1] =~ m/\"$/s |
1174 | and !( # catch weird degenerate case of L<"> ! |
1175 | @section_name == 1 and $section_name[0] eq '"' |
1176 | ) |
1177 | ) { |
1178 | $section_name[ 0] =~ s/^\"//s; |
1179 | $section_name[-1] =~ s/\"$//s; |
1180 | DEBUG > 3 and |
1181 | print " Quotes removed: ", pretty(@section_name), "\n"; |
1182 | } else { |
1183 | DEBUG > 3 and |
1184 | print " No need to remove quotes in ", pretty(@section_name), "\n"; |
1185 | } |
1186 | |
1187 | $section_name = \@section_name; |
1188 | last; |
1189 | } |
1190 | } |
1191 | |
1192 | # Turn L<"Foo Bar"> into L</Foo Bar> |
1193 | if(!$section_name and @ell_content |
1194 | and !ref($ell_content[0]) and !ref($ell_content[-1]) |
1195 | and $ell_content[ 0] =~ m/^\"/s |
1196 | and $ell_content[-1] =~ m/\"$/s |
1197 | and !( # catch weird degenerate case of L<"> ! |
1198 | @ell_content == 1 and $ell_content[0] eq '"' |
1199 | ) |
1200 | ) { |
1201 | $section_name = [splice @ell_content]; |
1202 | $section_name->[ 0] =~ s/^\"//s; |
1203 | $section_name->[-1] =~ s/\"$//s; |
1204 | } |
1205 | |
1206 | # Turn L<Foo Bar> into L</Foo Bar>. |
1207 | if(!$section_name and !$link_text and @ell_content |
1208 | and grep !ref($_) && m/ /s, @ell_content |
1209 | ) { |
1210 | $section_name = [splice @ell_content]; |
1211 | # That's support for the now-deprecated syntax. |
1212 | # (Maybe generate a warning eventually?) |
1213 | # Note that it deliberately won't work on L<...|Foo Bar> |
1214 | } |
1215 | |
1216 | |
1217 | # Now make up the link_text |
1218 | # L<Foo> -> L<Foo|Foo> |
1219 | # L</Bar> -> L<"Bar"|Bar> |
1220 | # L<Foo/Bar> -> L<"Bar" in Foo/Foo> |
1221 | unless($link_text) { |
1222 | $ell->[1]{'content-implicit'} = 'yes'; |
1223 | $link_text = []; |
1224 | push @$link_text, '"', @$section_name, '"' if $section_name; |
1225 | |
1226 | if(@ell_content) { |
1227 | $link_text->[-1] .= ' in ' if $section_name; |
1228 | push @$link_text, @ell_content; |
1229 | } |
1230 | } |
1231 | |
1232 | |
1233 | # And the E resolver will have to deal with all our treeletty things: |
1234 | |
1235 | if(@ell_content == 1 and !ref($ell_content[0]) |
1236 | and $ell_content[0] =~ m/^[-a-zA-Z0-9]+\([-a-zA-Z0-9]+\)$/s |
1237 | ) { |
1238 | $ell->[1]{'type'} = 'man'; |
1239 | DEBUG > 3 and print "Considering this ($ell_content[0]) a man link.\n"; |
1240 | } else { |
1241 | $ell->[1]{'type'} = 'pod'; |
1242 | DEBUG > 3 and print "Considering this a pod link (not man or url).\n"; |
1243 | } |
1244 | |
1245 | if( defined $section_name ) { |
1246 | $ell->[1]{'section'} = Pod::Simple::LinkSection->new( |
1247 | ['', {}, @$section_name] |
1248 | ); |
1249 | DEBUG > 3 and print "L-section content: ", pretty($ell->[1]{'section'}), "\n"; |
1250 | } |
1251 | |
1252 | if( @ell_content ) { |
1253 | $ell->[1]{'to'} = Pod::Simple::LinkSection->new( |
1254 | ['', {}, @ell_content] |
1255 | ); |
1256 | DEBUG > 3 and print "L-to content: ", pretty($ell->[1]{'to'}), "\n"; |
1257 | } |
1258 | |
1259 | # And update children to be the link-text: |
1260 | @$ell = (@$ell[0,1], defined($link_text) ? splice(@$link_text) : ''); |
1261 | |
1262 | DEBUG > 2 and print "End of L-parsing for this node $treelet->[$i]\n"; |
1263 | |
1264 | unshift @stack, $treelet->[$i]; # might as well recurse |
1265 | } |
1266 | } |
1267 | |
1268 | return; |
1269 | } |
1270 | |
1271 | # . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . |
1272 | |
1273 | sub _treat_Es { |
1274 | my($self,@stack) = @_; |
1275 | |
1276 | my($i, $treelet, $content, $replacer, $charnum); |
1277 | my $start_line = $stack[0][1]{'start_line'}; |
1278 | |
1279 | # A recursive algorithm implemented iteratively! Whee! |
1280 | |
1281 | |
1282 | # Has frightening side effects on L nodes' attributes. |
1283 | |
1284 | #my @ells_to_tweak; |
1285 | |
1286 | while($treelet = shift @stack) { |
1287 | for(my $i = 2; $i < @$treelet; ++$i) { # iterate over children |
1288 | next unless ref $treelet->[$i]; # text nodes are uninteresting |
1289 | if($treelet->[$i][0] eq 'L') { |
1290 | # SPECIAL STUFF for semi-processed L<>'s |
1291 | |
1292 | my $thing; |
1293 | foreach my $attrname ('section', 'to') { |
1294 | if(defined($thing = $treelet->[$i][1]{$attrname}) and ref $thing) { |
1295 | unshift @stack, $thing; |
1296 | DEBUG > 2 and print " Enqueuing ", |
1297 | pretty( $treelet->[$i][1]{$attrname} ), |
1298 | " as an attribute value to tweak.\n"; |
1299 | } |
1300 | } |
1301 | |
1302 | unshift @stack, $treelet->[$i]; # recurse |
1303 | next; |
1304 | } elsif($treelet->[$i][0] ne 'E') { |
1305 | unshift @stack, $treelet->[$i]; # recurse |
1306 | next; |
1307 | } |
1308 | |
1309 | DEBUG > 1 and print "Ogling E node ", pretty($treelet->[$i]), "\n"; |
1310 | |
1311 | # bitch if it's empty |
1312 | if( @{$treelet->[$i]} == 2 |
1313 | or (@{$treelet->[$i]} == 3 and $treelet->[$i][2] eq '') |
1314 | ) { |
1315 | $self->whine( $start_line, "An empty E<>" ); |
1316 | $treelet->[$i] = 'E<>'; # splice in a literal |
1317 | next; |
1318 | } |
1319 | |
1320 | # bitch if content is weird |
1321 | unless(@{$treelet->[$i]} == 3 and !ref($content = $treelet->[$i][2])) { |
1322 | $self->whine( $start_line, "An E<...> surrounding strange content" ); |
1323 | $replacer = $treelet->[$i]; # scratch |
1324 | splice(@$treelet, $i, 1, # fake out a literal |
1325 | 'E<', |
1326 | splice(@$replacer,2), # promote its content |
1327 | '>' |
1328 | ); |
1329 | # Don't need to do --$i, as the 'E<' we just added isn't interesting. |
1330 | next; |
1331 | } |
1332 | |
1333 | DEBUG > 1 and print "Ogling E<$content>\n"; |
1334 | |
1335 | $charnum = Pod::Escapes::e2charnum($content); |
1336 | DEBUG > 1 and print " Considering E<$content> with char ", |
1337 | defined($charnum) ? $charnum : "undef", ".\n"; |
1338 | |
1339 | if(!defined( $charnum )) { |
1340 | DEBUG > 1 and print "I don't know how to deal with E<$content>.\n"; |
1341 | $self->whine( $start_line, "Unknown E content in E<$content>" ); |
1342 | $replacer = "E<$content>"; # better than nothing |
1343 | } elsif($charnum >= 255 and !UNICODE) { |
1344 | $replacer = ASCII ? "\xA4" : "?"; |
1345 | DEBUG > 1 and print "This Perl version can't handle ", |
1346 | "E<$content> (chr $charnum), so replacing with $replacer\n"; |
1347 | } else { |
1348 | $replacer = Pod::Escapes::e2char($content); |
1349 | DEBUG > 1 and print " Replacing E<$content> with $replacer\n"; |
1350 | } |
1351 | |
1352 | splice(@$treelet, $i, 1, $replacer); # no need to back up $i, tho |
1353 | } |
1354 | } |
1355 | |
1356 | return; |
1357 | } |
1358 | |
1359 | |
1360 | # . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . |
1361 | |
1362 | sub _treat_Ss { |
1363 | my($self,$treelet) = @_; |
1364 | |
1365 | _change_S_to_nbsp($treelet,0) if $self->{'nbsp_for_S'}; |
1366 | |
1367 | # TODO: or a change_nbsp_to_S |
1368 | # Normalizing nbsp's to S is harder: for each text node, make S content |
1369 | # out of anything matching m/([^ \xA0]*(?:\xA0+[^ \xA0]*)+)/ |
1370 | |
1371 | |
1372 | return; |
1373 | } |
1374 | |
1375 | |
1376 | sub _change_S_to_nbsp { # a recursive function |
1377 | # Sanely assumes that the top node in the excursion won't be an S node. |
1378 | my($treelet, $in_s) = @_; |
1379 | |
1380 | my $is_s = ('S' eq $treelet->[0]); |
1381 | $in_s ||= $is_s; # So in_s is on either by this being an S element, |
1382 | # or by an ancestor being an S element. |
1383 | |
1384 | for(my $i = 2; $i < @$treelet; ++$i) { |
1385 | if(ref $treelet->[$i]) { |
1386 | if( _change_S_to_nbsp( $treelet->[$i], $in_s ) ) { |
1387 | my $to_pull_up = $treelet->[$i]; |
1388 | splice @$to_pull_up,0,2; # ...leaving just its content |
1389 | splice @$treelet, $i, 1, @$to_pull_up; # Pull up content |
1390 | $i += @$to_pull_up - 1; # Make $i skip the pulled-up stuff |
1391 | } |
1392 | } else { |
1393 | $treelet->[$i] =~ s/\s/\xA0/g if ASCII and $in_s; |
1394 | # (If not in ASCIIland, we can't assume that \xA0 == nbsp.) |
1395 | |
1396 | # Note that if you apply nbsp_for_S to text, and so turn |
1397 | # "foo S<bar baz> quux" into "foo bar faz quux", you |
1398 | # end up with something that fails to say "and don't hyphenate |
1399 | # any part of 'bar baz'". However, hyphenation is such a vexing |
1400 | # problem anyway, that most Pod renderers just don't render it |
1401 | # at all. But if you do want to implement hyphenation, I guess |
1402 | # that you'd better have nbsp_for_S off. |
1403 | } |
1404 | } |
1405 | |
1406 | return $is_s; |
1407 | } |
1408 | |
1409 | #----------------------------------------------------------------------------- |
1410 | |
1411 | sub _accessorize { # A simple-minded method-maker |
1412 | no strict 'refs'; |
1413 | foreach my $attrname (@_) { |
1414 | next if $attrname =~ m/::/; # a hack |
1415 | *{caller() . '::' . $attrname} = sub { |
1416 | use strict; |
1417 | $Carp::CarpLevel = 1, Carp::croak( |
1418 | "Accessor usage: \$obj->$attrname() or \$obj->$attrname(\$new_value)" |
1419 | ) unless (@_ == 1 or @_ == 2) and ref $_[0]; |
1420 | (@_ == 1) ? $_[0]->{$attrname} |
1421 | : ($_[0]->{$attrname} = $_[1]); |
1422 | }; |
1423 | } |
1424 | # Ya know, they say accessories make the ensemble! |
1425 | return; |
1426 | } |
1427 | |
1428 | # . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . |
1429 | # . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . |
1430 | #============================================================================= |
1431 | |
1432 | sub filter { |
1433 | my($class, $source) = @_; |
1434 | my $new = $class->new; |
1435 | $new->output_fh(*STDOUT{IO}); |
1436 | |
1437 | if(ref($source || '') eq 'SCALAR') { |
1438 | $new->parse_string_document( $$source ); |
1439 | } elsif(ref($source)) { # it's a file handle |
1440 | $new->parse_file($source); |
1441 | } else { # it's a filename |
1442 | $new->parse_file($source); |
1443 | } |
1444 | |
1445 | return $new; |
1446 | } |
1447 | |
1448 | |
1449 | #----------------------------------------------------------------------------- |
1450 | |
1451 | sub _out { |
1452 | # For use in testing: Class->_out($source) |
1453 | # returns the transformation of $source |
1454 | |
1455 | my $class = shift(@_); |
1456 | |
1457 | my $mutor = shift(@_) if @_ and ref($_[0] || '') eq 'CODE'; |
1458 | |
1459 | DEBUG and print "\n\n", '#' x 76, |
1460 | "\nAbout to parse source: {{\n$_[0]\n}}\n\n"; |
1461 | |
1462 | |
1463 | my $parser = ref $class && $class->isa(__PACKAGE__) ? $class : $class->new; |
1464 | $parser->hide_line_numbers(1); |
1465 | |
1466 | my $out = ''; |
1467 | $parser->output_string( \$out ); |
1468 | DEBUG and print " _out to ", \$out, "\n"; |
1469 | |
1470 | $mutor->($parser) if $mutor; |
1471 | |
1472 | $parser->parse_string_document( $_[0] ); |
1473 | # use Data::Dumper; print Dumper($parser), "\n"; |
1474 | return $out; |
1475 | } |
1476 | |
1477 | |
1478 | sub _duo { |
1479 | # For use in testing: Class->_duo($source1, $source2) |
1480 | # returns the parse trees of $source1 and $source2. |
1481 | # Good in things like: &ok( Class->duo(... , ...) ); |
1482 | |
1483 | my $class = shift(@_); |
1484 | |
1485 | Carp::croak "But $class->_duo is useful only in list context!" |
1486 | unless wantarray; |
1487 | |
1488 | my $mutor = shift(@_) if @_ and ref($_[0] || '') eq 'CODE'; |
1489 | |
1490 | Carp::croak "But $class->_duo takes two parameters, not: @_" |
1491 | unless @_ == 2; |
1492 | |
1493 | my(@out); |
1494 | |
1495 | while( @_ ) { |
1496 | my $parser = $class->new; |
1497 | |
1498 | push @out, ''; |
1499 | $parser->output_string( \( $out[-1] ) ); |
1500 | |
1501 | DEBUG and print " _duo out to ", $parser->output_string(), |
1502 | " = $parser->{'output_string'}\n"; |
1503 | |
1504 | $parser->hide_line_numbers(1); |
1505 | $mutor->($parser) if $mutor; |
1506 | $parser->parse_string_document( shift( @_ ) ); |
1507 | # use Data::Dumper; print Dumper($parser), "\n"; |
1508 | } |
1509 | |
1510 | return @out; |
1511 | } |
1512 | |
1513 | |
1514 | |
1515 | #----------------------------------------------------------------------------- |
1516 | 1; |
1517 | __END__ |
1518 | |
1519 | TODO: |
1520 | A start_formatting_code and end_formatting_code methods, which in the |
1521 | base class call start_L, end_L, start_C, end_C, etc., if they are |
1522 | defined. |
1523 | |
1524 | have the POD FORMATTING ERRORS section note the localtime, and the |
1525 | version of Pod::Simple. |
1526 | |
1527 | option to delete all E<shy>s? |
1528 | option to scream if under-0x20 literals are found in the input, or |
1529 | under-E<32> E codes are found in the tree. And ditto \x7f-\x9f |
1530 | |
1531 | Option to turn highbit characters into their compromised form? (applies |
1532 | to E parsing too) |
1533 | |
1534 | TODO: BOM/encoding things. |
1535 | |
1536 | TODO: ascii-compat things in the XML classes? |
1537 | |