Commit | Line | Data |
---|---|---|
94bdecf9 | 1 | #!/usr/bin/perl -w |
00d68f0d KW |
2 | |
3 | use Text::Tabs; | |
a9a7c81e | 4 | # |
6294c161 DM |
5 | # Unconditionally regenerate: |
6 | # | |
7 | # pod/perlintern.pod | |
8 | # pod/perlapi.pod | |
9 | # | |
10 | # from information stored in | |
11 | # | |
12 | # embed.fnc | |
9b983db7 | 13 | # plus all the core .c, .h, and .pod files listed in MANIFEST |
e10306bf KW |
14 | # plus %extra_input_pods |
15 | ||
16 | my %extra_input_pods = ( 'dist/ExtUtils-ParseXS/lib/perlxs.pod' => 1 ); | |
17 | ||
6294c161 | 18 | # Has an optional arg, which is the directory to chdir to before reading |
9b983db7 | 19 | # MANIFEST and the files |
6294c161 | 20 | # |
52a9d53b | 21 | # This script is invoked as part of 'make all' |
151c3fe5 | 22 | # |
9b983db7 KW |
23 | # The generated pod consists of sections of related elements, functions, |
24 | # macros, and variables. The keys of %valid_sections give the current legal | |
25 | # ones. Just add a new key to add a section. | |
26 | # | |
27 | # Throughout the files read by this script are lines like | |
28 | # | |
29 | # =for apidoc_section Section Name | |
3f620621 | 30 | # =for apidoc_section $section_name_variable |
9b983db7 KW |
31 | # |
32 | # "Section Name" (after having been stripped of leading space) must be one of | |
3f620621 KW |
33 | # the legal section names, or an error is thrown. $section_name_variable must |
34 | # be one of the legal section name variables defined below; these expand to | |
35 | # legal section names. This form is used so that minor wording changes in | |
36 | # these titles can be confied to this file. All the names of the variables | |
37 | # end in '_scn'; this suffix is optional in the apidoc_section lines. | |
38 | # | |
39 | # All API elements defined between this line and the next 'apidoc_section' | |
40 | # line will go into the section "Section Name" (or $section_name_variable), | |
41 | # sorted by dictionary order within it. perlintern and perlapi are parallel | |
42 | # documents, each potentially with a section "Section Name". Each element is | |
43 | # marked as to which document it goes into. If there are none for a | |
44 | # particular section in perlapi, that section is omitted. | |
9b983db7 KW |
45 | # |
46 | # Also, in .[ch] files, there may be | |
47 | # | |
48 | # =head1 Section Name | |
49 | # | |
50 | # lines in comments. These are also used by this program to switch to section | |
51 | # "Section Name". The difference is that if there are any lines after the | |
52 | # =head1, inside the same comment, and before any =for apidoc-ish lines, they | |
53 | # are used as a heading for section "Section Name" (in both perlintern and | |
54 | # perlapi). This includes any =head[2-5]. If more than one '=head1 Section | |
55 | # Name' line has content, they appear in the generated pod in an undefined | |
3f620621 | 56 | # order. Note that you can't use a $section_name_variable in =head1 lines |
9b983db7 KW |
57 | # |
58 | # The next =head1, =for apidoc_section, or file end terminates what goes into | |
59 | # the current section | |
60 | # | |
61 | # The %valid_sections hash below also can have header content, which will | |
62 | # appear before any =head1 content. The hash can also have footer content | |
63 | # content, which will appear at the end of the section, after all the | |
64 | # elements. | |
65 | # | |
66 | # The lines that define the actual functions, etc are documented in embed.fnc, | |
67 | # because they have flags which must be kept in sync with that file. | |
1fcde0e9 | 68 | |
56a0c332 | 69 | use strict; |
04adaf08 | 70 | use warnings; |
a64c954a | 71 | |
3dc8ec40 KW |
72 | # 80 column terminal - 2 for pager adding 2 columns; -4 for indent for |
73 | # non-heading lines; | |
74 | my $max_width = 80 - 2 - 4; | |
e4f5ddf4 | 75 | |
7882b24a NC |
76 | if (@ARGV) { |
77 | my $workdir = shift; | |
78 | chdir $workdir | |
79 | or die "Couldn't chdir to '$workdir': $!"; | |
80 | } | |
3d7c117d MB |
81 | require './regen/regen_lib.pl'; |
82 | require './regen/embed_lib.pl'; | |
7882b24a | 83 | |
a026e430 | 84 | my %described_elsewhere; |
b87d9527 | 85 | |
94bdecf9 | 86 | # |
346f75ff | 87 | # See database of global and static function prototypes in embed.fnc |
94bdecf9 JH |
88 | # This is used to generate prototype headers under various configurations, |
89 | # export symbols lists for different platforms, and macros to provide an | |
90 | # implicit interpreter context argument. | |
91 | # | |
92 | ||
6a235718 | 93 | my %docs; |
df6bd76f | 94 | my %seen; |
5ce57792 | 95 | my %funcflags; |
5ce57792 | 96 | my %missing; |
00d68f0d KW |
97 | my %missing_macros; |
98 | ||
99 | my $link_text = "Described in"; | |
100 | ||
783469eb KW |
101 | my $description_indent = 4; |
102 | my $usage_indent = 3; # + initial blank yields 4 total | |
103 | ||
3e294c2f | 104 | my $AV_scn = 'AV Handling'; |
3951f999 KW |
105 | my $callback_scn = 'Callback Functions'; |
106 | my $casting_scn = 'Casting'; | |
107 | my $casing_scn = 'Character case changing'; | |
108 | my $classification_scn = 'Character classification'; | |
109 | my $names_scn = 'Character names'; | |
110 | my $scope_scn = 'Compile-time scope hooks'; | |
111 | my $compiler_scn = 'Compiler and Preprocessor information'; | |
112 | my $directives_scn = 'Compiler directives'; | |
113 | my $concurrency_scn = 'Concurrency'; | |
240d081e | 114 | my $COP_scn = 'COPs and Hint Hashes'; |
3951f999 KW |
115 | my $CV_scn = 'CV Handling'; |
116 | my $custom_scn = 'Custom Operators'; | |
d1b9805e KW |
117 | my $debugging_scn = 'Debugging'; |
118 | my $display_scn = 'Display functions'; | |
3951f999 KW |
119 | my $embedding_scn = 'Embedding and Interpreter Cloning'; |
120 | my $errno_scn = 'Errno'; | |
121 | my $exceptions_scn = 'Exception Handling (simple) Macros'; | |
122 | my $filesystem_scn = 'Filesystem configuration values'; | |
b6f27e58 | 123 | my $filters_scn = 'Source Filters'; |
3951f999 KW |
124 | my $floating_scn = 'Floating point configuration values'; |
125 | my $formats_scn = 'Formats'; | |
126 | my $genconfig_scn = 'General Configuration'; | |
127 | my $globals_scn = 'Global Variables'; | |
128 | my $GV_scn = 'GV Handling'; | |
129 | my $hook_scn = 'Hook manipulation'; | |
130 | my $HV_scn = 'HV Handling'; | |
131 | my $io_scn = 'Input/Output'; | |
132 | my $integer_scn = 'Integer configuration values'; | |
133 | my $lexer_scn = 'Lexer interface'; | |
3e294c2f | 134 | my $locale_scn = 'Locales'; |
3951f999 KW |
135 | my $magic_scn = 'Magic'; |
136 | my $memory_scn = 'Memory Management'; | |
3e294c2f | 137 | my $MRO_scn = 'MRO'; |
3951f999 KW |
138 | my $multicall_scn = 'Multicall Functions'; |
139 | my $numeric_scn = 'Numeric Functions'; | |
72957953 KW |
140 | |
141 | # Now combined, as unclear which functions go where, but separate names kept | |
142 | # to avoid 1) other code changes; 2) in case it seems better to split again | |
143 | my $optrees_scn = 'Optrees'; | |
144 | my $optree_construction_scn = $optrees_scn; # Was 'Optree construction'; | |
145 | my $optree_manipulation_scn = $optrees_scn; # Was 'Optree Manipulation Functions' | |
3951f999 KW |
146 | my $pack_scn = 'Pack and Unpack'; |
147 | my $pad_scn = 'Pad Data Structures'; | |
148 | my $password_scn = 'Password and Group access'; | |
149 | my $paths_scn = 'Paths to system commands'; | |
3951f999 KW |
150 | my $prototypes_scn = 'Prototype information'; |
151 | my $regexp_scn = 'REGEXP Functions'; | |
152 | my $signals_scn = 'Signals'; | |
153 | my $site_scn = 'Site configuration'; | |
154 | my $sockets_scn = 'Sockets configuration values'; | |
3951f999 KW |
155 | my $stack_scn = 'Stack Manipulation Macros'; |
156 | my $string_scn = 'String Handling'; | |
157 | my $SV_flags_scn = 'SV Flags'; | |
158 | my $SV_scn = 'SV Handling'; | |
d8e799d8 | 159 | my $tainting_scn = 'Tainting'; |
3951f999 KW |
160 | my $time_scn = 'Time'; |
161 | my $typedefs_scn = 'Typedef names'; | |
162 | my $unicode_scn = 'Unicode Support'; | |
163 | my $utility_scn = 'Utility Functions'; | |
164 | my $versioning_scn = 'Versioning'; | |
165 | my $warning_scn = 'Warning and Dieing'; | |
3e294c2f | 166 | my $XS_scn = 'XS'; |
3951f999 | 167 | |
55d5a9fe KW |
168 | # Kept separate at end |
169 | my $undocumented_scn = 'Undocumented elements'; | |
170 | ||
00d68f0d | 171 | my %valid_sections = ( |
3e294c2f | 172 | $AV_scn => {}, |
3951f999 KW |
173 | $callback_scn => {}, |
174 | $casting_scn => {}, | |
175 | $casing_scn => {}, | |
176 | $classification_scn => {}, | |
177 | $scope_scn => {}, | |
178 | $compiler_scn => {}, | |
179 | $directives_scn => {}, | |
180 | $concurrency_scn => {}, | |
181 | $COP_scn => {}, | |
89222628 KW |
182 | $CV_scn => { |
183 | header => <<~'EOT', | |
184 | This section documents functions to manipulate CVs which are | |
185 | code-values, meaning subroutines. For more information, see | |
186 | L<perlguts>. | |
187 | EOT | |
188 | }, | |
189 | ||
3951f999 | 190 | $custom_scn => {}, |
d1b9805e KW |
191 | $debugging_scn => {}, |
192 | $display_scn => {}, | |
3951f999 KW |
193 | $embedding_scn => {}, |
194 | $errno_scn => {}, | |
195 | $exceptions_scn => {}, | |
196 | $filesystem_scn => { | |
6927cd44 KW |
197 | header => <<~'EOT', |
198 | Also see L</List of capability HAS_foo symbols>. | |
199 | EOT | |
200 | }, | |
b6f27e58 | 201 | $filters_scn => {}, |
3951f999 | 202 | $floating_scn => { |
00d68f0d | 203 | header => <<~'EOT', |
6927cd44 KW |
204 | Also L</List of capability HAS_foo symbols> lists capabilities |
205 | that arent in this section. For example C<HAS_ASINH>, for the | |
206 | hyperbolic sine function. | |
207 | EOT | |
208 | }, | |
3951f999 | 209 | $formats_scn => { |
6927cd44 KW |
210 | header => <<~'EOT', |
211 | These are used for formatting the corresponding type For example, | |
212 | instead of saying | |
213 | ||
214 | Perl_newSVpvf(pTHX_ "Create an SV with a %d in it\n", iv); | |
215 | ||
216 | use | |
217 | ||
218 | Perl_newSVpvf(pTHX_ "Create an SV with a " IVdf " in it\n", iv); | |
219 | ||
220 | This keeps you from having to know if, say an IV, needs to be | |
221 | printed as C<%d>, C<%ld>, or something else. | |
00d68f0d KW |
222 | EOT |
223 | }, | |
3951f999 | 224 | $genconfig_scn => { |
00d68f0d KW |
225 | header => <<~'EOT', |
226 | This section contains configuration information not otherwise | |
227 | found in the more specialized sections of this document. At the | |
228 | end is a list of C<#defines> whose name should be enough to tell | |
229 | you what they do, and a list of #defines which tell you if you | |
230 | need to C<#include> files to get the corresponding functionality. | |
231 | EOT | |
232 | ||
233 | footer => <<~'EOT', | |
234 | ||
235 | =head2 List of capability C<HAS_I<foo>> symbols | |
236 | ||
6927cd44 KW |
237 | This is a list of those symbols that dont appear elsewhere in ths |
238 | document that indicate if the current platform has a certain | |
239 | capability. Their names all begin with C<HAS_>. Only those | |
240 | symbols whose capability is directly derived from the name are | |
241 | listed here. All others have their meaning expanded out elsewhere | |
242 | in this document. This (relatively) compact list is because we | |
243 | think that the expansion would add little or no value and take up | |
244 | a lot of space (because there are so many). If you think certain | |
245 | ones should be expanded, send email to | |
246 | L<perl5-porters@perl.org|mailto:perl5-porters@perl.org>. | |
00d68f0d KW |
247 | |
248 | Each symbol here will be C<#define>d if and only if the platform | |
249 | has the capability. If you need more detail, see the | |
6927cd44 KW |
250 | corresponding entry in F<config.h>. For convenience, the list is |
251 | split so that the ones that indicate there is a reentrant version | |
252 | of a capability are listed separately | |
00d68f0d KW |
253 | |
254 | __HAS_LIST__ | |
255 | ||
6927cd44 KW |
256 | And, the reentrant capabilities: |
257 | ||
258 | __HAS_R_LIST__ | |
259 | ||
00d68f0d KW |
260 | Example usage: |
261 | ||
262 | =over | |
263 | ||
264 | #ifdef HAS_STRNLEN | |
265 | use strnlen() | |
266 | #else | |
267 | use an alternative implementation | |
268 | #endif | |
269 | ||
270 | =back | |
271 | ||
272 | =head2 List of C<#include> needed symbols | |
273 | ||
274 | This list contains symbols that indicate if certain C<#include> | |
275 | files are present on the platform. If your code accesses the | |
276 | functionality that one of these is for, you will need to | |
277 | C<#include> it if the symbol on this list is C<#define>d. For | |
278 | more detail, see the corresponding entry in F<config.h>. | |
279 | ||
280 | __INCLUDE_LIST__ | |
281 | ||
282 | Example usage: | |
283 | ||
284 | =over | |
285 | ||
286 | #ifdef I_WCHAR | |
287 | #include <wchar.h> | |
288 | #endif | |
289 | ||
290 | =back | |
291 | EOT | |
292 | }, | |
3951f999 KW |
293 | $globals_scn => {}, |
294 | $GV_scn => {}, | |
295 | $hook_scn => {}, | |
296 | $HV_scn => {}, | |
297 | $io_scn => {}, | |
298 | $integer_scn => {}, | |
299 | $lexer_scn => {}, | |
3e294c2f | 300 | $locale_scn => {}, |
3951f999 KW |
301 | $magic_scn => {}, |
302 | $memory_scn => {}, | |
3e294c2f | 303 | $MRO_scn => {}, |
3951f999 KW |
304 | $multicall_scn => {}, |
305 | $numeric_scn => {}, | |
72957953 | 306 | $optrees_scn => {}, |
3951f999 KW |
307 | $optree_construction_scn => {}, |
308 | $optree_manipulation_scn => {}, | |
309 | $pack_scn => {}, | |
310 | $pad_scn => {}, | |
311 | $password_scn => {}, | |
312 | $paths_scn => {}, | |
3951f999 KW |
313 | $prototypes_scn => {}, |
314 | $regexp_scn => {}, | |
315 | $signals_scn => {}, | |
316 | $site_scn => { | |
6927cd44 KW |
317 | header => <<~'EOT', |
318 | These variables give details as to where various libraries, | |
319 | installation destinations, I<etc.>, go, as well as what various | |
320 | installation options were selected | |
321 | EOT | |
322 | }, | |
3951f999 | 323 | $sockets_scn => {}, |
3951f999 KW |
324 | $stack_scn => {}, |
325 | $string_scn => { | |
89222628 KW |
326 | header => <<~EOT, |
327 | See also C<L</$unicode_scn>>. | |
6927cd44 KW |
328 | EOT |
329 | }, | |
3951f999 KW |
330 | $SV_flags_scn => {}, |
331 | $SV_scn => {}, | |
d8e799d8 | 332 | $tainting_scn => {}, |
3951f999 KW |
333 | $time_scn => {}, |
334 | $typedefs_scn => {}, | |
89222628 KW |
335 | $unicode_scn => { |
336 | header => <<~EOT, | |
337 | L<perlguts/Unicode Support> has an introduction to this API. | |
338 | ||
339 | See also C<L</$classification_scn>>, | |
340 | C<L</$casing_scn>>, | |
341 | and C<L</$string_scn>>. | |
342 | Various functions outside this section also work specially with | |
343 | Unicode. Search for the string "utf8" in this document. | |
344 | EOT | |
345 | }, | |
3951f999 KW |
346 | $utility_scn => {}, |
347 | $versioning_scn => {}, | |
348 | $warning_scn => {}, | |
3e294c2f | 349 | $XS_scn => {}, |
00d68f0d | 350 | ); |
94bdecf9 | 351 | |
9e019b59 KW |
352 | # Somewhat loose match for an apidoc line so we can catch minor typos. |
353 | # Parentheses are used to capture portions so that below we verify | |
354 | # that things are the actual correct syntax. | |
355 | my $apidoc_re = qr/ ^ (\s*) # $1 | |
356 | (=?) # $2 | |
357 | (\s*) # $3 | |
358 | for (\s*) # $4 | |
359 | apidoc (_item)? # $5 | |
360 | (\s*) # $6 | |
361 | (.*?) # $7 | |
362 | \s* \n /x; | |
183f6e59 | 363 | # Only certain flags, dealing with display, are acceptable for apidoc_item |
b8324975 | 364 | my $display_flags = "fFnDopsTx"; |
9e019b59 KW |
365 | |
366 | sub check_api_doc_line ($$) { | |
367 | my ($file, $in) = @_; | |
368 | ||
369 | return unless $in =~ $apidoc_re; | |
370 | ||
371 | my $is_item = defined $5; | |
372 | my $is_in_proper_form = length $1 == 0 | |
373 | && length $2 > 0 | |
374 | && length $3 == 0 | |
375 | && length $4 > 0 | |
6e93196a KW |
376 | && length $7 > 0 |
377 | && ( length $6 > 0 | |
378 | || ($is_item && substr($7, 0, 1) eq '|')); | |
9e019b59 KW |
379 | my $proto_in_file = $7; |
380 | my $proto = $proto_in_file; | |
381 | $proto = "||$proto" if $proto !~ /\|/; | |
382 | my ($flags, $ret_type, $name, @args) = split /\s*\|\s*/, $proto; | |
383 | ||
384 | $name && $is_in_proper_form or die <<EOS; | |
385 | Bad apidoc at $file line $.: | |
386 | $in | |
387 | Expected: | |
388 | =for apidoc flags|returntype|name|arg|arg|... | |
389 | =for apidoc flags|returntype|name | |
390 | =for apidoc name | |
391 | (or 'apidoc_item') | |
392 | EOS | |
393 | ||
56aac0d0 | 394 | die "Only [$display_flags] allowed in apidoc_item:\n$in" |
783469eb KW |
395 | if $is_item && $flags =~ /[^$display_flags]/; |
396 | ||
9e019b59 KW |
397 | return ($name, $flags, $ret_type, $is_item, $proto_in_file, @args); |
398 | } | |
94bdecf9 | 399 | |
4be3b9ca KW |
400 | sub embed_override($) { |
401 | my ($element_name) = shift; | |
402 | ||
403 | # If the entry is also in embed.fnc, it should be defined | |
404 | # completely there, but not here | |
405 | my $embed_docref = delete $funcflags{$element_name}; | |
406 | ||
407 | return unless $embed_docref and %$embed_docref; | |
408 | ||
409 | my $flags = $embed_docref->{'flags'}; | |
410 | warn "embed.fnc entry '$element_name' missing 'd' flag" | |
411 | unless $flags =~ /d/; | |
412 | ||
413 | return ($flags, $embed_docref->{'ret_type'}, $embed_docref->{args}->@*); | |
414 | } | |
415 | ||
89222628 KW |
416 | # The section that is in effect at the beginning of the given file. If not |
417 | # listed here, an apidoc_section line must precede any apidoc lines. | |
418 | # This allows the files listed here that generally are single-purpose, to not | |
419 | # have to worry about the autodoc section | |
420 | my %initial_file_section = ( | |
421 | 'av.c' => $AV_scn, | |
422 | 'av.h' => $AV_scn, | |
423 | 'cv.h' => $CV_scn, | |
189b9288 | 424 | 'dist/ExtUtils-ParseXS/lib/perlxs.pod' => $XS_scn, |
89222628 KW |
425 | 'doio.c' => $io_scn, |
426 | 'gv.c' => $GV_scn, | |
427 | 'gv.h' => $GV_scn, | |
428 | 'hv.h' => $HV_scn, | |
429 | 'locale.c' => $locale_scn, | |
430 | 'malloc.c' => $memory_scn, | |
431 | 'numeric.c' => $numeric_scn, | |
432 | 'opnames.h' => $optree_construction_scn, | |
433 | 'pad.h'=> $pad_scn, | |
434 | 'patchlevel.h' => $versioning_scn, | |
435 | 'perlio.h' => $io_scn, | |
436 | 'pod/perlapio.pod' => $io_scn, | |
437 | 'pod/perlcall.pod' => $callback_scn, | |
438 | 'pod/perlembed.pod' => $embedding_scn, | |
439 | 'pod/perlfilter.pod' => $filters_scn, | |
440 | 'pod/perliol.pod' => $io_scn, | |
441 | 'pod/perlmroapi.pod' => $MRO_scn, | |
442 | 'pod/perlreguts.pod' => $regexp_scn, | |
443 | 'pp_pack.c' => $pack_scn, | |
444 | 'pp_sort.c' => $SV_scn, | |
445 | 'regcomp.c' => $regexp_scn, | |
446 | 'regexp.h' => $regexp_scn, | |
d63f0f68 KW |
447 | 'sv.h' => $SV_scn, |
448 | 'sv.c' => $SV_scn, | |
072be79c | 449 | 'sv_inline.h' => $SV_scn, |
d8e799d8 | 450 | 'taint.c' => $tainting_scn, |
89222628 KW |
451 | 'unicode_constants.h' => $unicode_scn, |
452 | 'utf8.c' => $unicode_scn, | |
453 | 'utf8.h' => $unicode_scn, | |
454 | 'vutil.c' => $versioning_scn, | |
455 | ); | |
456 | ||
94bdecf9 JH |
457 | sub autodoc ($$) { # parse a file and extract documentation info |
458 | my($fh,$file) = @_; | |
63da8140 | 459 | my($in, $line_num, $header, $section); |
f554dfc5 | 460 | |
89222628 KW |
461 | $section = $initial_file_section{$file} |
462 | if defined $initial_file_section{$file}; | |
463 | ||
3cfa66c4 KW |
464 | my $file_is_C = $file =~ / \. [ch] $ /x; |
465 | ||
f554dfc5 | 466 | # Count lines easier |
b1cb8a49 | 467 | my $get_next_line = sub { $line_num++; return <$fh> }; |
f554dfc5 | 468 | |
63da8140 KW |
469 | # Read the file |
470 | while ($in = $get_next_line->()) { | |
471 | last unless defined $in; | |
3cfa66c4 | 472 | |
63da8140 KW |
473 | next unless ( $in =~ / ^ =for [ ]+ apidoc /x |
474 | # =head1 lines only have effect in C files | |
475 | || ($file_is_C && $in =~ /^=head1/)); | |
476 | ||
477 | # Here, the line introduces a portion of the input that we care about. | |
478 | # Either it is for an API element, or heading text which we expect | |
479 | # will be used for elements later in the file | |
480 | ||
481 | my ($text, $element_name, $flags, $ret_type, $is_item, $proto_in_file); | |
183f6e59 | 482 | my (@args, @items); |
63da8140 KW |
483 | |
484 | # If the line starts a new section ... | |
485 | if ($in=~ /^ = (?: for [ ]+ apidoc_section | head1 ) [ ]+ (.*) /x) { | |
183f6e59 | 486 | |
b1cb8a49 | 487 | $section = $1; |
3f620621 KW |
488 | if ($section =~ / ^ \$ /x) { |
489 | $section .= '_scn' unless $section =~ / _scn $ /; | |
490 | $section = eval "$section"; | |
491 | die "Unknown \$section variable '$section' in $file: $@" if $@; | |
492 | } | |
6927cd44 KW |
493 | die "Unknown section name '$section' in $file near line $.\n" |
494 | unless defined $valid_sections{$section}; | |
495 | ||
3cfa66c4 | 496 | } |
63da8140 KW |
497 | elsif ($in=~ /^ =for [ ]+ apidoc \B /x) { # Otherwise better be a |
498 | # plain apidoc line | |
183f6e59 KW |
499 | die "Unkown apidoc-type line '$in'" unless $in=~ /^=for apidoc_item/; |
500 | die "apidoc_item doesn't immediately follow an apidoc entry: '$in'"; | |
63da8140 KW |
501 | } |
502 | else { # Plain apidoc | |
503 | ||
504 | ($element_name, $flags, $ret_type, $is_item, $proto_in_file, @args) | |
505 | = check_api_doc_line($file, $in); | |
4be3b9ca KW |
506 | # Override this line with any info in embed.fnc |
507 | my ($embed_flags, $embed_ret_type, @embed_args) | |
508 | = embed_override($element_name); | |
509 | if ($embed_ret_type) { | |
63da8140 | 510 | warn "embed.fnc entry overrides redundant information in" |
4be3b9ca KW |
511 | . " '$proto_in_file' in $file" |
512 | if $flags || $ret_type || @args; | |
513 | $flags = $embed_flags; | |
514 | $ret_type = $embed_ret_type; | |
515 | @args = @embed_args; | |
516 | } | |
9e728159 KW |
517 | elsif ($flags !~ /[my]/) { # Not in embed.fnc, is missing if not |
518 | # a macro or typedef | |
63da8140 KW |
519 | $missing{$element_name} = $file; |
520 | } | |
3cfa66c4 | 521 | |
56aac0d0 | 522 | die "flag '$1' is not legal (for function $element_name (from $file))" |
ed18bf14 | 523 | if $flags =~ / ( [^AabCDdEeFfGhiIMmNnTOoPpRrSsUuWXxy] ) /x; |
151c3fe5 | 524 | |
9e728159 KW |
525 | die "'u' flag must also have 'm' or 'y' flags' for $element_name" |
526 | if $flags =~ /u/ && $flags !~ /[my]/; | |
63da8140 KW |
527 | warn ("'$element_name' not \\w+ in '$proto_in_file' in $file") |
528 | if $flags !~ /N/ && $element_name !~ / ^ [_[:alpha:]] \w* $ /x; | |
f554dfc5 | 529 | |
63da8140 KW |
530 | if (exists $seen{$element_name} && $flags !~ /h/) { |
531 | die ("'$element_name' in $file was already documented in $seen{$element_name}"); | |
532 | } | |
533 | else { | |
534 | $seen{$element_name} = $file; | |
151c3fe5 | 535 | } |
94bdecf9 | 536 | } |
df6bd76f | 537 | |
63da8140 KW |
538 | # Here we have processed the initial line in the heading text or API |
539 | # element, and have saved the important information from it into the | |
540 | # corresponding variables. Now accumulate the text that applies to it | |
541 | # up to a terminating line, which is one of: | |
542 | # 1) =cut | |
543 | # 2) =head (in a C file only =head1) | |
544 | # 3) an end comment line in a C file: m:^\s*\*/: | |
183f6e59 | 545 | # 4) =for apidoc... (except apidoc_item lines) |
63da8140 KW |
546 | $text = ""; |
547 | my $head_ender_num = ($file_is_C) ? 1 : ""; | |
548 | while (defined($in = $get_next_line->())) { | |
549 | ||
550 | last if $in =~ /^=cut/x; | |
551 | last if $in =~ /^=head$head_ender_num/; | |
552 | ||
553 | if ($file_is_C && $in =~ m: ^ \s* \* / $ :x) { | |
554 | ||
555 | # End of comment line in C files is a fall-back terminator, | |
556 | # but warn only if there actually is some accumulated text | |
557 | warn "=cut missing? $file:$line_num:$in" if $text =~ /\S/; | |
558 | last; | |
df6bd76f | 559 | } |
df6bd76f | 560 | |
63da8140 KW |
561 | if ($in !~ / ^ =for [ ]+ apidoc /x) { |
562 | $text .= $in; | |
563 | next; | |
564 | } | |
565 | ||
183f6e59 | 566 | # Here, the line is an apidoc line. All but apidoc_item terminate |
63da8140 | 567 | # the text being accumulated. |
183f6e59 KW |
568 | last if $in =~ / ^ =for [ ]+ apidoc_section /x; |
569 | ||
570 | my ($item_name, $item_flags, $item_ret_type, $is_item, | |
4be3b9ca | 571 | $item_proto, @item_args) = check_api_doc_line($file, $in); |
183f6e59 KW |
572 | last unless $is_item; |
573 | ||
574 | # Here, is an apidoc_item_line; They can only come within apidoc | |
575 | # paragraphs. | |
4be3b9ca KW |
576 | die "Unexpected api_doc_item line '$item_proto'" |
577 | unless $element_name; | |
183f6e59 KW |
578 | |
579 | # We accept blank lines between these, but nothing else; | |
56aac0d0 KW |
580 | die "apidoc_item lines must immediately follow apidoc lines for " |
581 | . " '$element_name' in $file" | |
183f6e59 | 582 | if $text =~ /\S/; |
4be3b9ca KW |
583 | # Override this line with any info in embed.fnc |
584 | my ($embed_flags, $embed_ret_type, @embed_args) | |
585 | = embed_override($item_name); | |
586 | if ($embed_ret_type) { | |
587 | warn "embed.fnc entry overrides redundant information in" | |
588 | . " '$item_proto' in $file" | |
589 | if $item_flags || $item_ret_type || @item_args; | |
590 | ||
591 | $item_flags = $embed_flags; | |
592 | $item_ret_type = $embed_ret_type; | |
593 | @item_args = @embed_args; | |
594 | } | |
183f6e59 KW |
595 | |
596 | # Use the base entry flags if none for this item; otherwise add in | |
597 | # any non-display base entry flags. | |
598 | if ($item_flags) { | |
599 | $item_flags .= $flags =~ s/[$display_flags]//rg; | |
600 | } | |
601 | else { | |
602 | $item_flags = $flags; | |
603 | } | |
604 | $item_ret_type = $ret_type unless $item_ret_type; | |
605 | @item_args = @args unless @item_args; | |
606 | push @items, { name => $item_name, | |
607 | ret_type => $item_ret_type, | |
608 | flags => $item_flags, | |
609 | args => [ @item_args ], | |
610 | }; | |
611 | ||
612 | # This line shows that this element is documented. | |
613 | delete $funcflags{$item_name}; | |
a9a7c81e | 614 | } |
3cfa66c4 | 615 | |
63da8140 KW |
616 | # Here, are done accumulating the text for this item. Trim it |
617 | $text =~ s/ ^ \s* //x; | |
618 | $text =~ s/ \s* $ //x; | |
619 | $text .= "\n" if $text ne ""; | |
6523e108 | 620 | |
63da8140 KW |
621 | # And treat all-spaces as nothing at all |
622 | undef $text unless $text =~ /\S/; | |
df6bd76f | 623 | |
63da8140 | 624 | if ($element_name) { |
df6bd76f | 625 | |
63da8140 KW |
626 | # Here, we have accumulated into $text, the pod for $element_name |
627 | my $where = $flags =~ /A/ ? 'api' : 'guts'; | |
5ce57792 | 628 | |
63da8140 KW |
629 | die "No =for apidoc_section nor =head1 in $file for '$element_name'\n" |
630 | unless defined $section; | |
631 | if (exists $docs{$where}{$section}{$element_name}) { | |
632 | warn "$0: duplicate API entry for '$element_name' in" | |
633 | . " $where/$section\n"; | |
634 | next; | |
20046047 | 635 | } |
5ce57792 | 636 | |
63da8140 KW |
637 | # Override the text with just a link if the flags call for that |
638 | my $is_link_only = ($flags =~ /h/); | |
639 | if ($is_link_only) { | |
640 | if ($file_is_C) { | |
56aac0d0 | 641 | die "Can't currently handle link with items to it:\n$in" if @items; |
63da8140 KW |
642 | redo; # Don't put anything if C source |
643 | } | |
644 | ||
645 | # Here, is an 'h' flag in pod. We add a reference to the pod (and | |
646 | # nothing else) to perlapi/intern. (It would be better to add a | |
647 | # reference to the correct =item,=header, but something that makes | |
648 | # it harder is that it that might be a duplicate, like '=item *'; | |
649 | # so that is a future enhancement XXX. Another complication is | |
650 | # there might be more than one deserving candidates.) | |
651 | my $podname = $file =~ s!.*/!!r; # Rmv directory name(s) | |
652 | $podname =~ s/\.pod//; | |
653 | $text = "Described in L<$podname>.\n"; | |
654 | ||
655 | # Don't output a usage example for linked to documentation if | |
656 | # it is trivial (has no arguments) and we aren't to add a | |
657 | # semicolon | |
658 | $flags .= 'U' if $flags =~ /n/ && $flags !~ /[Us]/; | |
659 | ||
660 | # Keep track of all the pod files that we refer to. | |
661 | push $described_elsewhere{$podname}->@*, $podname; | |
662 | } | |
5ce57792 | 663 | |
5bf02caf KW |
664 | $docs{$where}{$section}{$element_name}{flags} = $flags; |
665 | $docs{$where}{$section}{$element_name}{pod} = $text; | |
666 | $docs{$where}{$section}{$element_name}{file} = $file; | |
667 | $docs{$where}{$section}{$element_name}{ret_type} = $ret_type; | |
668 | push $docs{$where}{$section}{$element_name}{args}->@*, @args; | |
183f6e59 | 669 | push $docs{$where}{$section}{$element_name}{items}->@*, @items; |
a9a7c81e | 670 | } |
63da8140 KW |
671 | elsif ($text) { |
672 | $valid_sections{$section}{header} = "" unless | |
673 | defined $valid_sections{$section}{header}; | |
674 | $valid_sections{$section}{header} .= "\n$text"; | |
a9a7c81e | 675 | } |
63da8140 KW |
676 | |
677 | # We already have the first line of what's to come in $in | |
678 | redo; | |
679 | ||
680 | } # End of loop through input | |
94bdecf9 JH |
681 | } |
682 | ||
00d68f0d KW |
683 | my %configs; |
684 | my @has_defs; | |
685 | my @has_r_defs; # Reentrant symbols | |
686 | my @include_defs; | |
63447426 | 687 | |
00d68f0d KW |
688 | sub parse_config_h { |
689 | use re '/aa'; # Everthing is ASCII in this file | |
690 | ||
691 | # Process config.h | |
d3871820 KW |
692 | my $config_h = 'config.h'; |
693 | $config_h = 'win32/config.h' unless -e $config_h; | |
694 | die "Can't find $config_h" unless -e $config_h; | |
695 | open my $fh, '<', $config_h or die "Can't open $config_h: $!"; | |
00d68f0d KW |
696 | while (<$fh>) { |
697 | ||
698 | # Look for lines like /* FOO_BAR: | |
699 | # By convention all config.h descriptions begin like that | |
700 | if (m[ ^ /\* [ ] ( [[:alpha:]] \w+ ) : \s* $ ]ax) { | |
701 | my $name = $1; | |
702 | ||
703 | # Here we are starting the description for $name in config.h. We | |
704 | # accumulate the entire description for it into @description. | |
705 | # Flowing text from one input line to another is appended into the | |
706 | # same array element to make a single flowing line element, but | |
707 | # verbatim lines are kept as separate elements in @description. | |
708 | # This will facilitate later doing pattern matching without regard | |
709 | # to line boundaries on non-verbatim text. | |
710 | ||
711 | die "Multiple config.h entries for '$name'" | |
712 | if defined $configs{$name}{description}; | |
713 | ||
714 | # Get first line of description | |
715 | $_ = <$fh>; | |
716 | ||
717 | # Each line in the description begins with blanks followed by '/*' | |
718 | # and some spaces. | |
719 | die "Unexpected config.h initial line for $name: '$_'" | |
720 | unless s/ ^ ( \s* \* \s* ) //x; | |
721 | my $initial_text = $1; | |
722 | ||
723 | # Initialize the description with this first line (after having | |
724 | # stripped the prefix text) | |
725 | my @description = $_; | |
726 | ||
727 | # The first line is used as a template for how much indentation | |
728 | # each normal succeeding line has. Lines indented further | |
729 | # will be considered as intended to be verbatim. But, empty lines | |
730 | # likely won't have trailing blanks, so just strip the whole thing | |
731 | # for them. | |
732 | my $strip_initial_qr = qr! \s* \* \s* $ | |
733 | | \Q$initial_text\E | |
734 | !x; | |
735 | $configs{$name}{verbatim} = 0; | |
736 | ||
737 | # Read in the remainder of the description | |
738 | while (<$fh>) { | |
739 | last if s| ^ \s* \* / ||x; # A '*/' ends it | |
740 | ||
741 | die "Unexpected config.h description line for $name: '$_'" | |
742 | unless s/$strip_initial_qr//; | |
743 | ||
744 | # Fix up the few flawed lines in config.h wherein a new | |
745 | # sentence begins with a tab (and maybe a space after that). | |
746 | # Although none of them currently do, let it recognize | |
747 | # something like | |
748 | # | |
749 | # "... text"). The next sentence ... | |
750 | # | |
751 | s/ ( \w "? \)? \. ) \t \s* ( [[:alpha:]] ) /$1 $2/xg; | |
752 | ||
753 | # If this line has extra indentation or looks to have columns, | |
754 | # it should be treated as verbatim. Columns are indicated by | |
755 | # use of interior: tabs, 3 spaces in a row, or even 2 spaces | |
756 | # not preceded by punctuation. | |
757 | if ($_ !~ m/ ^ \s | |
758 | | \S (?: \t | |
759 | | \s{3} | |
760 | | (*nlb:[[:punct:]]) \s{2} | |
761 | ) | |
762 | /x) | |
763 | { | |
764 | # But here, is not a verbatim line. Add an empty line if | |
765 | # this is the first non-verbatim after a run of verbatims | |
766 | if ($description[-1] =~ /^\s/) { | |
767 | push @description, "\n", $_; | |
768 | } | |
769 | else { # Otherwise, append this flowing line to the | |
770 | # current flowing line | |
771 | $description[-1] .= $_; | |
772 | } | |
773 | } | |
774 | else { | |
775 | $configs{$name}{verbatim} = 1; | |
776 | ||
777 | # The first verbatim line in a run of them is separated by an | |
778 | # empty line from the flowing lines above it | |
779 | push @description, "\n" if $description[-1] =~ /^\S/; | |
780 | ||
781 | $_ = Text::Tabs::expand($_); | |
782 | ||
783 | # Only a single space so less likely to wrap | |
784 | s/ ^ \s* / /x; | |
785 | ||
786 | push @description, $_; | |
787 | } | |
788 | } | |
789 | ||
790 | push $configs{$name}{description}->@*, @description | |
791 | ||
792 | } # Not a description; see if it is a macro definition. | |
793 | elsif (m! ^ | |
794 | (?: / \* )? # Optional commented-out | |
795 | # indication | |
796 | \# \s* define \s+ ( \w+ ) # $1 is the name | |
797 | ( \s* ) # $2 indicates if args or not | |
798 | ( .*? ) # $3 is any definition | |
799 | (?: / \s* \* \* / )? # Optional trailing /**/ or / **/ | |
800 | $ | |
801 | !x) | |
802 | { | |
803 | my $name = $1; | |
804 | ||
805 | # There can be multiple definitions for a name. We want to know | |
806 | # if any of them has arguments, and if any has a body. | |
807 | $configs{$name}{has_args} //= $2 eq ""; | |
808 | $configs{$name}{has_args} ||= $2 eq ""; | |
809 | $configs{$name}{has_defn} //= $3 ne ""; | |
810 | $configs{$name}{has_defn} ||= $3 ne ""; | |
811 | } | |
812 | } | |
813 | ||
814 | # We now have stored the description and information about every #define | |
815 | # in the file. The description is in a form convenient to operate on to | |
816 | # convert to pod. Do that now. | |
817 | foreach my $name (keys %configs) { | |
818 | next unless defined $configs{$name}{description}; | |
819 | ||
820 | # All adjacent non-verbatim lines of the description are appended | |
821 | # together in a single element in the array. This allows the patterns | |
822 | # to work across input line boundaries. | |
823 | ||
824 | my $pod = ""; | |
825 | while (defined ($_ = shift $configs{$name}{description}->@*)) { | |
826 | chomp; | |
827 | ||
828 | if (/ ^ \S /x) { # Don't edit verbatim lines | |
829 | ||
830 | # Enclose known file/path names not already so enclosed | |
831 | # with <...>. (Some entries in config.h are already | |
832 | # '<path/to/file>') | |
833 | my $file_name_qr = qr! [ \w / ]+ \. | |
834 | (?: c | h | xs | p [lm] | pmc | PL | |
835 | | sh | SH | exe ) \b | |
836 | !xx; | |
837 | my $path_name_qr = qr! (?: / \w+ )+ !x; | |
838 | for my $re ($file_name_qr, $path_name_qr) { | |
839 | s! (*nlb:[ < \w / ]) ( $re ) !<$1>!gxx; | |
840 | } | |
841 | ||
842 | # Enclose <... file/path names with F<...> (but no double | |
843 | # angle brackets) | |
844 | for my $re ($file_name_qr, $path_name_qr) { | |
845 | s! < ( $re ) > !F<$1>!gxx; | |
846 | } | |
847 | ||
848 | # Explain metaconfig units | |
849 | s/ ( \w+ \. U \b ) /$1 (part of metaconfig)/gx; | |
850 | ||
851 | # Convert "See foo" to "See C<L</foo>>" if foo is described in | |
852 | # this file. Also create a link to the known file INSTALL. | |
853 | # And, to be more general, handle "See also foo and bar", and | |
854 | # "See also foo, bar, and baz" | |
855 | while (m/ \b [Ss]ee \s+ | |
856 | (?: also \s+ )? ( \w+ ) | |
857 | (?: , \s+ ( \w+ ) )? | |
858 | (?: ,? \s+ and \s+ ( \w+ ) )? /xg) { | |
859 | my @links = $1; | |
860 | push @links, $2 if defined $2; | |
861 | push @links, $3 if defined $3; | |
862 | foreach my $link (@links) { | |
863 | if ($link eq 'INSTALL') { | |
864 | s/ \b INSTALL \b /C<L<INSTALL>>/xg; | |
865 | } | |
866 | elsif (grep { $link =~ / \b $_ \b /x } keys %configs) { | |
867 | s| \b $link \b |C<L</$link>>|xg; | |
868 | $configs{$link}{linked} = 1; | |
869 | $configs{$name}{linked} = 1; | |
870 | } | |
871 | } | |
872 | } | |
873 | ||
874 | # Enclose what we think are symbols with C<...>. | |
875 | no warnings 'experimental::vlb'; | |
876 | s/ (*nlb:<) | |
877 | ( | |
878 | # Any word followed immediately with parens or | |
879 | # brackets | |
880 | \b \w+ (?: \( [^)]* \) # parameter list | |
881 | | \[ [^]]* \] # or array reference | |
882 | ) | |
883 | | (*plb: ^ | \s ) -D \w+ # Also -Dsymbols. | |
884 | | \b (?: struct | union ) \s \w+ | |
885 | ||
886 | # Words that contain underscores (which are | |
887 | # definitely not text) or three uppercase letters in | |
888 | # a row. Length two ones, like IV, aren't enclosed, | |
889 | # because they often don't look as nice. | |
890 | | \b \w* (?: _ | [[:upper:]]{3,} ) \w* \b | |
891 | ) | |
892 | (*nla:>) | |
893 | /C<$1>/xg; | |
894 | ||
895 | # These include foo when the name is HAS_foo. This is a | |
896 | # heuristic which works in most cases. | |
897 | if ($name =~ / ^ HAS_ (.*) /x) { | |
898 | my $symbol = lc $1; | |
899 | ||
900 | # Don't include path components, nor things already in | |
901 | # <>, or with trailing '(', '[' | |
902 | s! \b (*nlb:[/<]) $symbol (*nla:[[/>(]) \b !C<$symbol>!xg; | |
903 | } | |
904 | } | |
905 | ||
906 | $pod .= "$_\n"; | |
907 | } | |
908 | delete $configs{$name}{description}; | |
909 | ||
910 | $configs{$name}{pod} = $pod; | |
911 | } | |
912 | ||
913 | # Now have converted the description to pod. We also now have enough | |
914 | # information that we can do cross checking to find definitions without | |
915 | # corresponding pod, and see if they are mentioned in some description; | |
916 | # otherwise they aren't documented. | |
917 | NAME: | |
918 | foreach my $name (keys %configs) { | |
919 | ||
920 | # A definition without pod | |
921 | if (! defined $configs{$name}{pod}) { | |
922 | ||
923 | # Leading/trailing underscore means internal to config.h, e.g., | |
924 | # _GNU_SOURCE | |
925 | next if $name =~ / ^ _ /x; | |
926 | next if $name =~ / _ $ /x; | |
927 | ||
928 | # MiXeD case names are internal to config.h; the first 4 | |
929 | # characters are sufficient to determine this | |
930 | next if $name =~ / ^ [[:upper:]] [[:lower:]] | |
931 | [[:upper:]] [[:lower:]] | |
932 | /x; | |
933 | ||
934 | # Here, not internal to config.h. Look to see if this symbol is | |
935 | # mentioned in the pod of some other. If so, assume it is | |
936 | # documented. | |
937 | foreach my $check_name (keys %configs) { | |
938 | my $this_element = $configs{$check_name}; | |
939 | my $this_pod = $this_element->{pod}; | |
940 | if (defined $this_pod) { | |
941 | next NAME if $this_pod =~ / \b $name \b /x; | |
942 | } | |
943 | } | |
944 | ||
945 | warn "$name has no documentation\n"; | |
946 | $missing_macros{$name} = 'config.h'; | |
947 | ||
948 | next; | |
949 | } | |
950 | ||
951 | my $has_defn = $configs{$name}{has_defn}; | |
952 | my $has_args = $configs{$name}{has_args}; | |
953 | ||
954 | # Check if any section already has an entry for this element. | |
955 | # If so, it better be a placeholder, in which case we replace it | |
956 | # with this entry. | |
957 | foreach my $section (keys $docs{'api'}->%*) { | |
958 | if (exists $docs{'api'}{$section}{$name}) { | |
959 | my $was = $docs{'api'}{$section}{$name}->{pod}; | |
960 | $was = "" unless $was; | |
961 | chomp $was; | |
962 | if ($was ne "" && $was !~ m/$link_text/) { | |
963 | die "Multiple descriptions for $name\n" | |
964 | . "$section contained '$was'"; | |
965 | } | |
966 | $docs{'api'}{$section}{$name}->{pod} = $configs{$name}{pod}; | |
967 | $configs{$name}{section} = $section; | |
968 | last; | |
969 | } | |
970 | } | |
971 | ||
972 | my $handled = 0; # Haven't handled this yet | |
973 | ||
974 | if (defined $configs{$name}{'section'}) { | |
975 | # This has been taken care of elsewhere. | |
976 | $handled = 1; | |
977 | } | |
978 | else { | |
9e728159 | 979 | my $flags = ""; |
00d68f0d KW |
980 | if ($has_defn && ! $has_args) { |
981 | $configs{$name}{args} = 1; | |
982 | } | |
983 | ||
984 | # Symbols of the form I_FOO are for #include files. They have | |
985 | # special usage information | |
986 | if ($name =~ / ^ I_ ( .* ) /x) { | |
987 | my $file = lc $1 . '.h'; | |
988 | $configs{$name}{usage} = <<~"EOT"; | |
989 | #ifdef $name | |
990 | #include <$file> | |
991 | #endif | |
992 | EOT | |
993 | } | |
994 | ||
995 | # Compute what section this variable should go into. This | |
996 | # heuristic was determined by manually inspecting the current | |
997 | # things in config.h, and should be adjusted as necessary as | |
998 | # deficiencies are found. | |
999 | # | |
1000 | # This is the default section for macros with a definiton but | |
1001 | # no arguments, meaning it is replaced unconditionally | |
1002 | # | |
1003 | my $sb = qr/ _ | \b /x; # segment boundary | |
1004 | my $dash_or_spaces = qr/ - | \s+ /x; | |
1005 | my $pod = $configs{$name}{pod}; | |
1006 | if ($name =~ / ^ USE_ /x) { | |
3951f999 | 1007 | $configs{$name}{'section'} = $site_scn; |
00d68f0d KW |
1008 | } |
1009 | elsif ($name =~ / SLEEP | (*nlb:SYS_) TIME | TZ | $sb TM $sb /x) | |
1010 | { | |
3951f999 | 1011 | $configs{$name}{'section'} = $time_scn; |
00d68f0d | 1012 | } |
c86b25ed KW |
1013 | elsif ( $name =~ / ^ [[:alpha:]]+ f $ /x |
1014 | && $configs{$name}{pod} =~ m/ \b format \b /ix) | |
1015 | { | |
1016 | $configs{$name}{'section'} = $formats_scn; | |
1017 | } | |
00d68f0d KW |
1018 | elsif ($name =~ / DOUBLE | FLOAT | LONGDBL | LDBL | ^ NV |
1019 | | $sb CASTFLAGS $sb | |
1020 | | QUADMATH | |
1021 | | $sb (?: IS )? NAN | |
1022 | | $sb (?: IS )? FINITE | |
1023 | /x) | |
1024 | { | |
1025 | $configs{$name}{'section'} = | |
3951f999 | 1026 | $floating_scn; |
00d68f0d KW |
1027 | } |
1028 | elsif ($name =~ / (?: POS | OFF | DIR ) 64 /x) { | |
3951f999 | 1029 | $configs{$name}{'section'} = $filesystem_scn; |
00d68f0d KW |
1030 | } |
1031 | elsif ( $name =~ / $sb (?: BUILTIN | CPP ) $sb | ^ CPP /x | |
1032 | || $configs{$name}{pod} =~ m/ \b align /x) | |
1033 | { | |
3951f999 | 1034 | $configs{$name}{'section'} = $compiler_scn; |
00d68f0d KW |
1035 | } |
1036 | elsif ($name =~ / ^ [IU] [ \d V ] | |
1037 | | ^ INT | SHORT | LONG | QUAD | 64 | 32 /xx) | |
1038 | { | |
3951f999 | 1039 | $configs{$name}{'section'} = $integer_scn; |
00d68f0d KW |
1040 | } |
1041 | elsif ($name =~ / $sb t $sb /x) { | |
3951f999 | 1042 | $configs{$name}{'section'} = $typedefs_scn; |
9e728159 | 1043 | $flags .= 'y'; |
00d68f0d KW |
1044 | } |
1045 | elsif ( $name =~ / ^ PERL_ ( PRI | SCN ) | $sb FORMAT $sb /x | |
1046 | && $configs{$name}{pod} =~ m/ \b format \b /ix) | |
1047 | { | |
3951f999 | 1048 | $configs{$name}{'section'} = $formats_scn; |
00d68f0d KW |
1049 | } |
1050 | elsif ($name =~ / BACKTRACE /x) { | |
d1b9805e | 1051 | $configs{$name}{'section'} = $debugging_scn; |
00d68f0d KW |
1052 | } |
1053 | elsif ($name =~ / ALLOC $sb /x) { | |
3951f999 | 1054 | $configs{$name}{'section'} = $memory_scn; |
00d68f0d KW |
1055 | } |
1056 | elsif ( $name =~ / STDIO | FCNTL | EOF | FFLUSH | |
1057 | | $sb FILE $sb | |
1058 | | $sb DIR $sb | |
1059 | | $sb LSEEK | |
1060 | | $sb INO $sb | |
1061 | | $sb OPEN | |
1062 | | $sb CLOSE | |
1063 | | ^ DIR | |
1064 | | ^ INO $sb | |
1065 | | DIR $ | |
1066 | | FILENAMES | |
1067 | /x | |
1068 | || $configs{$name}{pod} =~ m! I/O | stdio | |
1069 | | file \s+ descriptor | |
1070 | | file \s* system | |
1071 | | statfs | |
1072 | !x) | |
1073 | { | |
3951f999 | 1074 | $configs{$name}{'section'} = $filesystem_scn; |
00d68f0d KW |
1075 | } |
1076 | elsif ($name =~ / ^ SIG | SIGINFO | signal /ix) { | |
3951f999 | 1077 | $configs{$name}{'section'} = $signals_scn; |
00d68f0d KW |
1078 | } |
1079 | elsif ($name =~ / $sb ( PROTO (?: TYPE)? S? ) $sb /x) { | |
3951f999 | 1080 | $configs{$name}{'section'} = $prototypes_scn; |
00d68f0d KW |
1081 | } |
1082 | elsif ( $name =~ / ^ LOC_ /x | |
1083 | || $configs{$name}{pod} =~ /full path/i) | |
1084 | { | |
3951f999 | 1085 | $configs{$name}{'section'} = $paths_scn; |
00d68f0d KW |
1086 | } |
1087 | elsif ($name =~ / $sb LC_ | LOCALE | langinfo /xi) { | |
3e294c2f | 1088 | $configs{$name}{'section'} = $locale_scn; |
00d68f0d KW |
1089 | } |
1090 | elsif ($configs{$name}{pod} =~ / GCC | C99 | C\+\+ /xi) { | |
3951f999 | 1091 | $configs{$name}{'section'} = $compiler_scn; |
00d68f0d KW |
1092 | } |
1093 | elsif ($name =~ / PASSW (OR)? D | ^ PW | ( PW | GR ) ENT /x) | |
1094 | { | |
3951f999 | 1095 | $configs{$name}{'section'} = $password_scn; |
00d68f0d KW |
1096 | } |
1097 | elsif ($name =~ / SOCKET | $sb SOCK /x) { | |
3951f999 | 1098 | $configs{$name}{'section'} = $sockets_scn; |
00d68f0d KW |
1099 | } |
1100 | elsif ( $name =~ / THREAD | MULTIPLICITY /x | |
1101 | || $configs{$name}{pod} =~ m/ \b pthread /ix) | |
1102 | { | |
3951f999 | 1103 | $configs{$name}{'section'} = $concurrency_scn; |
00d68f0d KW |
1104 | } |
1105 | elsif ($name =~ / PERL | ^ PRIV | SITE | ARCH | BIN | |
1106 | | VENDOR | ^ USE | |
1107 | /x) | |
1108 | { | |
3951f999 | 1109 | $configs{$name}{'section'} = $site_scn; |
00d68f0d KW |
1110 | } |
1111 | elsif ( $pod =~ / \b floating $dash_or_spaces point \b /ix | |
63447426 KW |
1112 | || $pod =~ / \b (double | single) $dash_or_spaces precision \b /ix |
1113 | || $pod =~ / \b doubles \b /ix | |
1114 | || $pod =~ / \b (?: a | the | long ) \s+ (?: double | NV ) \b /ix) | |
00d68f0d KW |
1115 | { |
1116 | $configs{$name}{'section'} = | |
3951f999 | 1117 | $floating_scn; |
00d68f0d KW |
1118 | } |
1119 | else { | |
1120 | # Above are the specific sections. The rest go into a | |
1121 | # grab-bag of general configuration values. However, we put | |
1122 | # two classes of them into lists of their names, without their | |
1123 | # descriptions, when we think that the description doesn't add | |
1124 | # any real value. One list contains the #include variables: | |
1125 | # the description is basically boiler plate for each of these. | |
1126 | # The other list contains the very many things that are of the | |
1127 | # form HAS_foo, and \bfoo\b is contained in its description, | |
1128 | # and there is no verbatim text in the pod or links to/from it | |
1129 | # (which would add value). That means that it is likely the | |
1130 | # intent of the variable can be gleaned from just its name, | |
1131 | # and unlikely the description adds signficant value, so just | |
1132 | # listing them suffices. Giving their descriptions would | |
1133 | # expand this pod significantly with little added value. | |
1134 | if ( ! $has_defn | |
1135 | && ! $configs{$name}{verbatim} | |
1136 | && ! $configs{$name}{linked}) | |
1137 | { | |
1138 | if ($name =~ / ^ I_ ( .* ) /x) { | |
1139 | push @include_defs, $name; | |
1140 | next; | |
1141 | } | |
1142 | elsif ($name =~ / ^ HAS_ ( .* ) /x) { | |
1143 | my $canonical_name = $1; | |
1144 | $canonical_name =~ s/_//g; | |
1145 | ||
1146 | my $canonical_pod = $configs{$name}{pod}; | |
1147 | $canonical_pod =~ s/_//g; | |
1148 | ||
1149 | if ($canonical_pod =~ / \b $canonical_name \b /xi) { | |
1150 | if ($name =~ / $sb R $sb /x) { | |
1151 | push @has_r_defs, $name; | |
1152 | } | |
1153 | else { | |
1154 | push @has_defs, $name; | |
1155 | } | |
1156 | next; | |
1157 | } | |
1158 | } | |
1159 | } | |
1160 | ||
3951f999 | 1161 | $configs{$name}{'section'} = $genconfig_scn; |
00d68f0d KW |
1162 | } |
1163 | ||
1164 | my $section = $configs{$name}{'section'}; | |
6927cd44 KW |
1165 | die "Internal error: '$section' not in \%valid_sections" |
1166 | unless grep { $_ eq $section } keys %valid_sections; | |
9e728159 | 1167 | $flags .= 'AdmnT'; |
00d68f0d | 1168 | $flags .= 'U' unless defined $configs{$name}{usage}; |
9e728159 | 1169 | |
63447426 | 1170 | # All the information has been gathered; save it |
00d68f0d KW |
1171 | $docs{'api'}{$section}{$name}{flags} = $flags; |
1172 | $docs{'api'}{$section}{$name}{pod} = $configs{$name}{pod}; | |
1173 | $docs{'api'}{$section}{$name}{ret_type} = ""; | |
1174 | $docs{'api'}{$section}{$name}{file} = 'config.h'; | |
1175 | $docs{'api'}{$section}{$name}{usage} | |
1176 | = $configs{$name}{usage} if defined $configs{$name}{usage}; | |
1177 | push $docs{'api'}{$section}{$name}{args}->@*, (); | |
1178 | push $docs{'api'}{$section}{$name}{items}->@*, (); | |
1179 | } | |
1180 | } | |
1181 | } | |
1182 | ||
c7a2af98 KW |
1183 | sub format_pod_indexes($) { |
1184 | my $entries_ref = shift; | |
1185 | ||
1186 | # Output the X<> references to the names, packed since they don't get | |
1187 | # displayed, but not too many per line so that when someone is editing the | |
1188 | # file, it doesn't run on | |
1189 | ||
1190 | my $text =""; | |
1191 | my $line_length = 0; | |
1192 | for my $name (sort dictionary_order $entries_ref->@*) { | |
1193 | my $entry = "X<$name>"; | |
1194 | my $entry_length = length $entry; | |
1195 | ||
1196 | # Don't loop forever if we have a verrry long name, and don't go too | |
1197 | # far to the right. | |
1198 | if ($line_length > 0 && $line_length + $entry_length > $max_width) { | |
1199 | $text .= "\n"; | |
1200 | $line_length = 0; | |
1201 | } | |
1202 | ||
1203 | $text .= $entry; | |
1204 | $line_length += $entry_length; | |
1205 | } | |
1206 | ||
1207 | return $text; | |
1208 | } | |
1209 | ||
b63f8ed4 | 1210 | sub docout ($$$) { # output the docs for one function group |
b1cb8a49 | 1211 | my($fh, $element_name, $docref) = @_; |
5bf02caf | 1212 | |
783469eb KW |
1213 | # Trim trailing space |
1214 | $element_name =~ s/\s*$//; | |
1215 | ||
5bf02caf | 1216 | my $flags = $docref->{flags}; |
63da8140 | 1217 | my $pod = $docref->{pod} // ""; |
5bf02caf | 1218 | my $file = $docref->{file}; |
783469eb | 1219 | |
183f6e59 | 1220 | my @items = $docref->{items}->@*; |
5bf02caf | 1221 | |
783469eb KW |
1222 | # Make the main element the first of the items. This allows uniform |
1223 | # treatment below | |
1224 | unshift @items, { name => $element_name, | |
1225 | flags => $flags, | |
1226 | ret_type => $docref->{ret_type}, | |
1227 | args => [ $docref->{args}->@* ], | |
1228 | }; | |
94bdecf9 | 1229 | |
7a997878 | 1230 | warn("Empty pod for $element_name (from $file)") unless $pod =~ /\S/; |
9639517b | 1231 | |
783469eb KW |
1232 | print $fh "\n=over $description_indent\n"; |
1233 | print $fh "\n=item C<$_->{name}>\n" for @items; | |
63da8140 | 1234 | |
783469eb KW |
1235 | # If we're printing only a link to an element, this isn't the major entry, |
1236 | # so no X<> here. | |
1237 | if ($flags !~ /h/) { | |
1238 | print $fh "X<$_->{name}>" for @items; | |
1239 | print $fh "\n"; | |
d4e99c76 | 1240 | } |
63da8140 | 1241 | |
d1681de8 KW |
1242 | my @deprecated; |
1243 | my @experimental; | |
783469eb | 1244 | for my $item (@items) { |
d1681de8 KW |
1245 | push @deprecated, "C<$item->{name}>" if $item->{flags} =~ /D/; |
1246 | push @experimental, "C<$item->{name}>" if $item->{flags} =~ /x/; | |
1247 | } | |
54c193ae | 1248 | |
d1681de8 KW |
1249 | for my $which (\@deprecated, \@experimental) { |
1250 | if ($which->@*) { | |
1251 | my $is; | |
1252 | my $it; | |
1253 | my $list; | |
1254 | ||
1255 | if ($which->@* == 1) { | |
1256 | $is = 'is'; | |
1257 | $it = 'it'; | |
1258 | $list = $which->[0]; | |
1259 | } | |
1260 | elsif ($which->@* == @items) { | |
1261 | $is = 'are'; | |
1262 | $it = 'them'; | |
1263 | $list = (@items == 2) | |
1264 | ? "both forms" | |
1265 | : "all these forms"; | |
1266 | } | |
1267 | else { | |
1268 | $is = 'are'; | |
1269 | $it = 'them'; | |
1270 | my $final = pop $which->@*; | |
1271 | $list = "the " . join ", ", $which->@*; | |
1272 | $list .= "," if $which->@* > 1; | |
1273 | $list .= " and $final forms"; | |
1274 | } | |
54c193ae | 1275 | |
d1681de8 KW |
1276 | if ($which == \@deprecated) { |
1277 | print $fh <<~"EOT"; | |
1278 | ||
1279 | C<B<DEPRECATED!>> It is planned to remove $list | |
1280 | from a future release of Perl. Do not use $it for | |
1281 | new code; remove $it from existing code. | |
1282 | EOT | |
1283 | } | |
1284 | else { | |
1285 | print $fh <<~"EOT"; | |
1286 | ||
1287 | NOTE: $list $is B<experimental> and may change or be | |
1288 | removed without notice. | |
1289 | EOT | |
1290 | } | |
783469eb | 1291 | } |
54c193ae | 1292 | } |
94bdecf9 | 1293 | |
783469eb KW |
1294 | chomp $pod; # Make sure prints pod with a single trailing \n |
1295 | print $fh "\n", $pod, "\n"; | |
3cfa66c4 | 1296 | |
783469eb KW |
1297 | for my $item (@items) { |
1298 | my $item_flags = $item->{flags}; | |
1299 | my $item_name = $item->{name}; | |
3cfa66c4 | 1300 | |
783469eb KW |
1301 | print $fh "\nNOTE: the C<perl_$item_name()> form is B<deprecated>.\n" |
1302 | if $item_flags =~ /O/; | |
1303 | # Is Perl_, but no #define foo # Perl_foo | |
1304 | if (($item_flags =~ /p/ && $item_flags =~ /o/ && $item_flags !~ /M/) | |
1305 | ||
1306 | # Can't handle threaded varargs | |
1307 | || ($item_flags =~ /f/ && $item_flags !~ /T/)) | |
1308 | { | |
1309 | $item->{name} = "Perl_$item_name"; | |
1310 | print $fh <<~"EOT"; | |
1311 | ||
1312 | NOTE: C<$item_name> must be explicitly called as | |
1313 | C<$item->{name}> | |
1314 | EOT | |
1315 | print $fh "with an C<aTHX_> parameter" if $item_flags !~ /T/; | |
1316 | print $fh ".\n"; | |
1317 | } | |
1318 | } | |
94bdecf9 | 1319 | |
9e728159 KW |
1320 | if ($flags =~ /[Uy]/) { # no usage; typedefs are considered simple enough |
1321 | # to never warrant a usage line | |
1322 | warn("U and s flags are incompatible") | |
1323 | if $flags =~ /U/ && $flags =~ /s/; | |
20046047 | 1324 | # nothing |
05ca4832 | 1325 | } else { |
783469eb KW |
1326 | |
1327 | print $fh "\n=over $usage_indent\n"; | |
1328 | ||
00d68f0d KW |
1329 | if (defined $docref->{usage}) { # An override of the usage section |
1330 | print $fh "\n", ($docref->{usage} =~ s/^/ /mrg), "\n"; | |
1331 | } | |
783469eb KW |
1332 | else { |
1333 | ||
8d1c4423 KW |
1334 | # Add the thread context formal parameter on expanded-out names |
1335 | for my $item (@items) { | |
1336 | unshift $item->{args}->@*, (($item->{args}->@*) | |
1337 | ? "pTHX_" | |
1338 | : "pTHX") | |
1339 | if $item->{flags} !~ /T/ | |
1340 | && $item->{name} =~ /^Perl_/; | |
1341 | } | |
1342 | ||
783469eb | 1343 | # Look through all the items in this entry. If all have the same |
4be56a7d KW |
1344 | # return type and arguments (including thread context), only the |
1345 | # main entry is displayed. | |
783469eb KW |
1346 | # Also, find the longest return type and longest name so that if |
1347 | # multiple ones are shown, they can be vertically aligned nicely | |
783469eb | 1348 | my $need_individual_usage = 0; |
53c9fbf2 | 1349 | my $longest_name_length = length $items[0]->{name}; |
783469eb | 1350 | my $base_ret_type = $items[0]->{ret_type}; |
53c9fbf2 | 1351 | my $longest_ret = length $base_ret_type; |
783469eb | 1352 | my @base_args = $items[0]->{args}->@*; |
4be56a7d | 1353 | my $base_thread_context = $items[0]->{flags} =~ /T/; |
53c9fbf2 | 1354 | for (my $i = 1; $i < @items; $i++) { |
783469eb | 1355 | no warnings 'experimental::smartmatch'; |
53c9fbf2 | 1356 | my $item = $items[$i]; |
783469eb KW |
1357 | $need_individual_usage = 1 |
1358 | if $item->{ret_type} ne $base_ret_type | |
4be56a7d KW |
1359 | || ! ($item->{args}->@* ~~ @base_args) |
1360 | || ( $item->{flags} =~ /T/ | |
1361 | != $base_thread_context); | |
783469eb KW |
1362 | my $ret_length = length $item->{ret_type}; |
1363 | $longest_ret = $ret_length if $ret_length > $longest_ret; | |
1364 | my $name_length = length $item->{name}; | |
1365 | $longest_name_length = $name_length | |
1366 | if $name_length > $longest_name_length; | |
1ded1f42 | 1367 | } |
783469eb KW |
1368 | |
1369 | # If we're only showing one entry, only its length matters. | |
1370 | $longest_name_length = length($items[0]->{name}) | |
1371 | unless $need_individual_usage; | |
1372 | print $fh "\n"; | |
1373 | ||
1374 | my $indent = 1; # 1 is sufficient for verbatim; =over is used | |
1375 | # for more | |
1376 | my $ret_name_sep_length = 2; # spaces between return type and name | |
1377 | my $name_indent = $indent + $longest_ret; | |
1378 | $name_indent += $ret_name_sep_length if $longest_ret; | |
1379 | ||
ca711307 KW |
1380 | my $this_max_width = |
1381 | $max_width - $description_indent - $usage_indent; | |
1382 | ||
783469eb KW |
1383 | for my $item (@items) { |
1384 | my $ret_type = $item->{ret_type}; | |
1385 | my @args = $item->{args}->@*; | |
1386 | my $name = $item->{name}; | |
1387 | my $item_flags = $item->{flags}; | |
1388 | ||
783469eb KW |
1389 | # The return type |
1390 | print $fh (" " x $indent), $ret_type; | |
1391 | ||
1392 | print $fh " " x ( $ret_name_sep_length | |
1393 | + $longest_ret - length $ret_type); | |
1394 | print $fh $name; | |
1395 | ||
1396 | if ($item_flags =~ /n/) { # no args | |
1397 | warn("$file: $element_name: n flag without m") | |
1398 | unless $item_flags =~ /m/; | |
1399 | warn("$file: $name: n flag but apparently has args") | |
1400 | if @args; | |
1401 | } | |
1402 | else { | |
1403 | # +1 for the '(' | |
1404 | my $arg_indent = $name_indent + $longest_name_length + 1; | |
1405 | ||
1406 | # Align the argument lists of the items | |
1407 | print $fh " " x ($longest_name_length - length($name)); | |
1408 | print $fh "("; | |
1409 | ||
1410 | # Display as many of the arguments on the same line as | |
1411 | # will fit. | |
1412 | my $total_length = $arg_indent; | |
1413 | my $first_line = 1; | |
1414 | for (my $i = 0; $i < @args; $i++) { | |
1415 | my $arg = $args[$i]; | |
1416 | my $arg_length = length($arg); | |
1417 | ||
1418 | # All but the first arg are preceded by a blank | |
1419 | my $use_blank = $i > 0; | |
1420 | ||
1421 | # +1 here and below because either the argument has a | |
1422 | # trailing comma or trailing ')' | |
1423 | $total_length += $arg_length + $use_blank + 1; | |
1424 | ||
1425 | # We want none of the arguments to be positioned so | |
1426 | # they extend too far to the right. Ideally, they | |
1427 | # should all start in the same column as the arguments | |
1428 | # on the first line of the function display do. But, if | |
1429 | # necessary, outdent them so that they all start in | |
1430 | # another column, with the longest ending at the right | |
1431 | # margin, like so: | |
1432 | # void function_name(pTHX_ short1, | |
1433 | # short2, | |
1434 | # very_long_argument, | |
1435 | # short3) | |
ca711307 | 1436 | if ($total_length > $this_max_width) { |
783469eb KW |
1437 | |
1438 | # If this is the first continuation line, | |
1439 | # calculate the longest argument; this will be the | |
1440 | # one we may have to outdent for. | |
1441 | if ($first_line) { | |
1442 | $first_line = 0; | |
1443 | ||
1444 | # We will need at least as much as the current | |
1445 | # argument | |
1446 | my $longest_arg_length = $arg_length | |
1447 | + $use_blank + 1; | |
1448 | ||
1449 | # Look through the rest of the args to see if | |
1450 | # any are longer than this one. | |
1451 | for (my $j = $i + 1; $j < @args; $j++) { | |
1452 | ||
1453 | # Include the trailing ',' or ')' in the | |
1454 | # length. No need to concern ourselves | |
1455 | # with a leading blank, as the argument | |
1456 | # would be positioned first on the next | |
1457 | # line | |
1458 | my $peek_arg_length = length ($args[$j]) | |
1459 | + 1; | |
1460 | $longest_arg_length = $peek_arg_length | |
1461 | if $peek_arg_length > $longest_arg_length; | |
1462 | } | |
1463 | ||
1464 | # Calculate the new indent if necessary. | |
ca711307 KW |
1465 | $arg_indent = |
1466 | $this_max_width - $longest_arg_length | |
783469eb | 1467 | if $arg_indent + $longest_arg_length |
ca711307 | 1468 | > $this_max_width; |
783469eb KW |
1469 | } |
1470 | ||
1471 | print $fh "\n", (" " x $arg_indent); | |
1472 | $total_length = $arg_indent + $arg_length + 1; | |
1473 | $use_blank = 0; | |
1474 | } | |
1475 | ||
1476 | # Display this argument | |
1477 | print $fh " " if $use_blank; | |
1478 | print $fh $arg; | |
1479 | print $fh "," if $i < @args - 1 && $args[$i] ne 'pTHX_'; | |
1480 | ||
1481 | } # End of loop through args | |
1482 | ||
1483 | print $fh ")"; | |
1ded1f42 | 1484 | } |
783469eb KW |
1485 | |
1486 | print $fh ";" if $item_flags =~ /s/; # semicolon: "dTHR;" | |
1487 | print $fh "\n"; | |
1488 | ||
1489 | # Only the first entry is normally displayed | |
1490 | last unless $need_individual_usage; | |
1ded1f42 | 1491 | } |
1ded1f42 | 1492 | } |
783469eb KW |
1493 | |
1494 | print $fh "\n=back\n"; | |
94bdecf9 | 1495 | } |
783469eb KW |
1496 | |
1497 | print $fh "\n=back\n"; | |
63da8140 | 1498 | print $fh "\n=for hackers\nFound in file $file\n"; |
94bdecf9 JH |
1499 | } |
1500 | ||
02c5183e KW |
1501 | sub construct_missings_section { |
1502 | my ($pod_name, $missings_ref) = @_; | |
1503 | my $text = ""; | |
1504 | ||
1505 | return $text unless $missings_ref->@*; | |
1506 | ||
1507 | $text .= <<~EOT; | |
d5c04b1e | 1508 | |
55d5a9fe | 1509 | =head1 $undocumented_scn |
02c5183e KW |
1510 | |
1511 | EOT | |
1512 | if ($pod_name eq 'perlapi') { | |
1513 | $text .= <<~'EOT'; | |
1514 | The following functions have been flagged as part of the public | |
1515 | API, but are currently undocumented. Use them at your own risk, | |
1516 | as the interfaces are subject to change. Functions that are not | |
1517 | listed in this document are not intended for public use, and | |
1518 | should NOT be used under any circumstances. | |
1519 | ||
1520 | If you feel you need to use one of these functions, first send | |
1521 | email to L<perl5-porters@perl.org|mailto:perl5-porters@perl.org>. | |
1522 | It may be that there is a good reason for the function not being | |
1523 | documented, and it should be removed from this list; or it may | |
1524 | just be that no one has gotten around to documenting it. In the | |
1525 | latter case, you will be asked to submit a patch to document the | |
1526 | function. Once your patch is accepted, it will indicate that the | |
1527 | interface is stable (unless it is explicitly marked otherwise) and | |
1528 | usable by you. | |
1529 | EOT | |
1530 | } | |
1531 | else { | |
1532 | $text .= <<~'EOT'; | |
1533 | The following functions are currently undocumented. If you use | |
1534 | one of them, you may wish to consider creating and submitting | |
1535 | documentation for it. | |
1536 | EOT | |
1537 | } | |
1538 | ||
c7a2af98 | 1539 | $text .= "\n" . format_pod_indexes($missings_ref); |
e4f5ddf4 KW |
1540 | |
1541 | # Sort the elements. | |
1542 | my @missings = sort dictionary_order $missings_ref->@*; | |
1543 | ||
e4f5ddf4 | 1544 | |
c7a2af98 | 1545 | $text .= "\n"; |
e4f5ddf4 KW |
1546 | |
1547 | use integer; | |
1548 | ||
1549 | # Look through all the elements in the list and see how many columns we | |
1550 | # could place them in the output what will fit in the available width. | |
1551 | my $min_spacer = 2; # Need this much space between columns | |
1552 | my $columns; | |
1553 | my $rows; | |
1554 | my @col_widths; | |
1555 | ||
1556 | COLUMN: | |
1557 | # We start with more columns, and work down until we find a number that | |
1558 | # can accommodate all the data. This algorithm doesn't require the | |
1559 | # resulting columns to all have the same width. This can allow for | |
1560 | # as tight of packing as the data will possibly allow. | |
1561 | for ($columns = 7; $columns > 1; $columns--) { | |
1562 | ||
1563 | # For this many columns, we will need this many rows (final row might | |
1564 | # not be completely filled) | |
1565 | $rows = (@missings + $columns - 1) / $columns; | |
1566 | ||
1567 | my $row_width = 0; | |
1568 | my $i = 0; # Which missing element | |
1569 | ||
1570 | # For each column ... | |
1571 | for my $col (0 .. $columns - 1) { | |
1572 | ||
1573 | # Calculate how wide the column needs to be, which is based on the | |
1574 | # widest element in it | |
1575 | $col_widths[$col] = 0; | |
1576 | ||
1577 | # Look through all the rows to find the widest element | |
1578 | for my $row (0 .. $rows - 1) { | |
1579 | ||
1580 | # Skip if this row doesn't have an entry for this column | |
1581 | last if $i >= @missings; | |
1582 | ||
1583 | # This entry occupies this many bytes. | |
1584 | my $this_width = length $missings[$i]; | |
1585 | ||
1586 | # All but the final column need a spacer between it and the | |
1587 | # next column over. | |
1588 | $this_width += $min_spacer if $col < $columns - 1; | |
1589 | ||
1590 | ||
1591 | # This column will need to have enough width to accommodate | |
1592 | # this element | |
1593 | if ($this_width > $col_widths[$col]) { | |
1594 | ||
1595 | # We can't have this many columns if the total width | |
1596 | # exceeds the available; bail now and try fewer columns | |
1597 | next COLUMN if $row_width + $this_width > $max_width; | |
1598 | ||
1599 | $col_widths[$col] = $this_width; | |
1600 | } | |
1601 | ||
1602 | $i++; # The next row will contain the next item | |
1603 | } | |
1604 | ||
1605 | $row_width += $col_widths[$col]; | |
1606 | next COLUMN if $row_width > $max_width; | |
1607 | } | |
1608 | ||
1609 | # If we get this far, this many columns works | |
1610 | last; | |
1611 | } | |
1612 | ||
1613 | # Here, have calculated the number of rows ($rows) and columns ($columns) | |
1614 | # required to list the elements. @col_widths contains the width of each | |
1615 | # column. | |
1616 | ||
1617 | $text .= "\n\n=over $description_indent\n\n"; | |
1618 | ||
1619 | # Assemble the output | |
1620 | for my $row (0 .. $rows - 1) { | |
1621 | for my $col (0 .. $columns - 1) { | |
1622 | $text .= " " if $col == 0; # Indent one to mark as verbatim | |
1623 | ||
1624 | my $index = $row + $rows * $col; # Convert 2 dimensions to 1 | |
1625 | ||
1626 | # Skip if this row doesn't have an entry for this column | |
1627 | next if $index >= @missings; | |
1628 | ||
1629 | my $element = $missings[$index]; | |
1630 | $text .= $element; | |
1631 | ||
1632 | # Add alignment spaces for all but final column | |
1633 | $text .= " " x ($col_widths[$col] - length $element) | |
1634 | if $col < $columns - 1; | |
1635 | } | |
02c5183e | 1636 | |
e4f5ddf4 | 1637 | $text .= "\n"; # End of row |
02c5183e KW |
1638 | } |
1639 | ||
1640 | $text .= "\n=back\n"; | |
1641 | ||
1642 | return $text; | |
1643 | } | |
1644 | ||
1645 | sub dictionary_order { | |
f83c6033 KW |
1646 | # Do a case-insensitive dictionary sort, with only alphabetics |
1647 | # significant, falling back to using everything for determinancy | |
1354d57e | 1648 | return (uc($a =~ s/[[:^alpha:]]//r) cmp uc($b =~ s/[[:^alpha:]]//r)) |
f83c6033 KW |
1649 | || uc($a) cmp uc($b) |
1650 | || $a cmp $b; | |
1651 | } | |
1652 | ||
7b73ff98 | 1653 | sub output { |
d5c04b1e | 1654 | my ($podname, $header, $dochash, $missings_ref, $footer) = @_; |
6a4c4cd4 DM |
1655 | # |
1656 | # strip leading '|' from each line which had been used to hide | |
1657 | # pod from pod checkers. | |
1658 | s/^\|//gm for $header, $footer; | |
1659 | ||
7882b24a | 1660 | my $fh = open_new("pod/$podname.pod", undef, |
20046047 | 1661 | {by => "$0 extracting documentation", |
f1f44974 | 1662 | from => 'the C source files'}, 1); |
e0492643 | 1663 | |
63da8140 | 1664 | print $fh $header, "\n"; |
e0492643 | 1665 | |
02c5183e | 1666 | for my $section_name (sort dictionary_order keys %valid_sections) { |
7a997878 | 1667 | my $section_info = $dochash->{$section_name}; |
36ea2d2f | 1668 | |
5781dcac KW |
1669 | # We allow empty sections in perlintern. |
1670 | if (! $section_info && $podname eq 'perlapi') { | |
36ea2d2f | 1671 | warn "Empty section '$section_name'; skipped"; |
36ea2d2f KW |
1672 | next; |
1673 | } | |
1674 | ||
63da8140 KW |
1675 | print $fh "\n=head1 $section_name\n"; |
1676 | ||
5781dcac KW |
1677 | if ($podname eq 'perlapi') { |
1678 | print $fh "\n", $valid_sections{$section_name}{header}, "\n" | |
1679 | if defined $valid_sections{$section_name}{header}; | |
151c3fe5 | 1680 | |
5781dcac KW |
1681 | # Output any heading-level documentation and delete so won't get in |
1682 | # the way later | |
1683 | if (exists $section_info->{""}) { | |
1684 | print $fh "\n", $section_info->{""}, "\n"; | |
1685 | delete $section_info->{""}; | |
1686 | } | |
151c3fe5 | 1687 | } |
151c3fe5 | 1688 | |
74ab7b79 | 1689 | if ($section_info && keys $section_info->%*) { |
02c5183e | 1690 | for my $function_name (sort dictionary_order keys %$section_info) { |
5781dcac KW |
1691 | docout($fh, $function_name, $section_info->{$function_name}); |
1692 | } | |
1693 | } | |
1694 | else { | |
1695 | print $fh "\nThere are only public API items currently in $section_name\n"; | |
20046047 | 1696 | } |
00b0fb83 KW |
1697 | |
1698 | print $fh "\n", $valid_sections{$section_name}{footer}, "\n" | |
5781dcac KW |
1699 | if $podname eq 'perlapi' |
1700 | && defined $valid_sections{$section_name}{footer}; | |
7b73ff98 NC |
1701 | } |
1702 | ||
d5c04b1e | 1703 | print $fh construct_missings_section($podname, $missings_ref); |
63da8140 KW |
1704 | |
1705 | print $fh "\n$footer\n=cut\n"; | |
5a0155e6 | 1706 | |
7882b24a | 1707 | read_only_bottom_close_and_rename($fh); |
cd093254 MM |
1708 | } |
1709 | ||
e8e591c9 NC |
1710 | foreach (@{(setup_embed())[0]}) { |
1711 | next if @$_ < 2; | |
b1cb8a49 | 1712 | my ($flags, $ret_type, $func, @args) = @$_; |
e8e591c9 | 1713 | s/\b(?:NN|NULLOK)\b\s+//g for @args; |
bc350081 | 1714 | |
5ce57792 | 1715 | $funcflags{$func} = { |
20046047 | 1716 | flags => $flags, |
b1cb8a49 | 1717 | ret_type => $ret_type, |
20046047 KE |
1718 | args => \@args, |
1719 | }; | |
5ce57792 NC |
1720 | } |
1721 | ||
5ce57792 NC |
1722 | # glob() picks up docs from extra .c or .h files that may be in unclean |
1723 | # development trees. | |
741c0772 NC |
1724 | open my $fh, '<', 'MANIFEST' |
1725 | or die "Can't open MANIFEST: $!"; | |
1726 | while (my $line = <$fh>) { | |
b87d9527 | 1727 | next unless my ($file) = $line =~ /^(\S+\.(?:[ch]|pod))\t/; |
5ce57792 | 1728 | |
e10306bf KW |
1729 | # Don't pick up pods from these. |
1730 | next if $file =~ m! ^ ( cpan | dist | ext ) / !x | |
1731 | && ! defined $extra_input_pods{$file}; | |
7cb0b6c4 | 1732 | |
1ae6ead9 | 1733 | open F, '<', $file or die "Cannot open $file for docs: $!\n"; |
5ce57792 NC |
1734 | autodoc(\*F,$file); |
1735 | close F or die "Error closing $file: $!\n"; | |
1736 | } | |
741c0772 | 1737 | close $fh or die "Error whilst reading MANIFEST: $!"; |
5ce57792 | 1738 | |
00d68f0d KW |
1739 | parse_config_h(); |
1740 | ||
5ce57792 NC |
1741 | for (sort keys %funcflags) { |
1742 | next unless $funcflags{$_}{flags} =~ /d/; | |
6523e108 | 1743 | next if $funcflags{$_}{flags} =~ /h/; |
573cf2be | 1744 | warn "no docs for $_\n"; |
bc350081 | 1745 | } |
94bdecf9 | 1746 | |
5ce57792 | 1747 | foreach (sort keys %missing) { |
5ce57792 | 1748 | warn "Function '$_', documented in $missing{$_}, not listed in embed.fnc"; |
94bdecf9 JH |
1749 | } |
1750 | ||
ff5af78d KW |
1751 | # List of funcs in the public API that aren't also marked as core-only, |
1752 | # experimental nor deprecated. | |
b87d9527 KW |
1753 | my @missing_api = grep $funcflags{$_}{flags} =~ /A/ |
1754 | && $funcflags{$_}{flags} !~ /[xD]/ | |
1755 | && !$docs{api}{$_}, keys %funcflags; | |
00d68f0d | 1756 | push @missing_api, keys %missing_macros; |
a026e430 | 1757 | |
e10306bf | 1758 | my @other_places = ( qw(perlclib ), keys %described_elsewhere ); |
ea7c1d7a KW |
1759 | my $places_other_than_intern = join ", ", |
1760 | map { "L<$_>" } sort dictionary_order 'perlapi', @other_places; | |
1761 | my $places_other_than_api = join ", ", | |
1762 | map { "L<$_>" } sort dictionary_order 'perlintern', @other_places; | |
a026e430 | 1763 | |
00d68f0d | 1764 | # The S< > makes things less densely packed, hence more readable |
02c5183e KW |
1765 | my $has_defs_text .= join ",S< > ", map { "C<$_>" } sort dictionary_order @has_defs; |
1766 | my $has_r_defs_text .= join ",S< > ", map { "C<$_>" } sort dictionary_order @has_r_defs; | |
3951f999 KW |
1767 | $valid_sections{$genconfig_scn}{footer} =~ s/__HAS_LIST__/$has_defs_text/; |
1768 | $valid_sections{$genconfig_scn}{footer} =~ s/__HAS_R_LIST__/$has_r_defs_text/; | |
00d68f0d | 1769 | |
02c5183e | 1770 | my $include_defs_text .= join ",S< > ", map { "C<$_>" } sort dictionary_order @include_defs; |
3951f999 | 1771 | $valid_sections{$genconfig_scn}{footer} =~ s/__INCLUDE_LIST__/$include_defs_text/; |
00d68f0d | 1772 | |
55d5a9fe KW |
1773 | my $section_list = join "\n\n", map { "=item L</$_>" } |
1774 | sort(dictionary_order keys %valid_sections), | |
1775 | $undocumented_scn; # Keep last | |
7260d4ba | 1776 | |
b87d9527 | 1777 | output('perlapi', <<"_EOB_", $docs{api}, \@missing_api, <<"_EOE_"); |
6a4c4cd4 DM |
1778 | |=encoding UTF-8 |
1779 | | | |
1780 | |=head1 NAME | |
1781 | | | |
1782 | |perlapi - autogenerated documentation for the perl public API | |
1783 | | | |
1784 | |=head1 DESCRIPTION | |
1785 | |X<Perl API> X<API> X<api> | |
1786 | | | |
b87d9527 KW |
1787 | |This file contains most of the documentation of the perl public API, as |
1788 | |generated by F<embed.pl>. Specifically, it is a listing of functions, | |
a026e430 KW |
1789 | |macros, flags, and variables that may be used by extension writers. Besides |
1790 | |L<perlintern> and F<config.h>, some items are listed here as being actually | |
1791 | |documented in another pod. | |
b87d9527 | 1792 | | |
55d5a9fe | 1793 | |L<At the end|/$undocumented_scn> is a list of functions which have yet |
b87d9527 KW |
1794 | |to be documented. Patches welcome! The interfaces of these are subject to |
1795 | |change without notice. | |
1796 | | | |
6927cd44 KW |
1797 | |Some of the functions documented here are consolidated so that a single entry |
1798 | |serves for multiple functions which all do basically the same thing, but have | |
1799 | |some slight differences. For example, one form might process magic, while | |
1800 | |another doesn't. The name of each variation is listed at the top of the | |
1801 | |single entry. But if all have the same signature (arguments and return type) | |
1802 | |except for their names, only the usage for the base form is shown. If any | |
1803 | |one of the forms has a different signature (such as returning C<const> or | |
1804 | |not) every function's signature is explicitly displayed. | |
7260d4ba | 1805 | | |
a026e430 KW |
1806 | |Anything not listed here or in the other mentioned pods is not part of the |
1807 | |public API, and should not be used by extension writers at all. For these | |
1808 | |reasons, blindly using functions listed in F<proto.h> is to be avoided when | |
1809 | |writing extensions. | |
6a4c4cd4 DM |
1810 | | |
1811 | |In Perl, unlike C, a string of characters may generally contain embedded | |
1812 | |C<NUL> characters. Sometimes in the documentation a Perl string is referred | |
1813 | |to as a "buffer" to distinguish it from a C string, but sometimes they are | |
1814 | |both just referred to as strings. | |
1815 | | | |
1816 | |Note that all Perl API global variables must be referenced with the C<PL_> | |
1817 | |prefix. Again, those not listed here are not to be used by extension writers, | |
1818 | |and can be changed or removed without notice; same with macros. | |
1819 | |Some macros are provided for compatibility with the older, | |
1820 | |unadorned names, but this support may be disabled in a future release. | |
1821 | | | |
1822 | |Perl was originally written to handle US-ASCII only (that is characters | |
1823 | |whose ordinal numbers are in the range 0 - 127). | |
1824 | |And documentation and comments may still use the term ASCII, when | |
1825 | |sometimes in fact the entire range from 0 - 255 is meant. | |
1826 | | | |
1827 | |The non-ASCII characters below 256 can have various meanings, depending on | |
1828 | |various things. (See, most notably, L<perllocale>.) But usually the whole | |
1829 | |range can be referred to as ISO-8859-1. Often, the term "Latin-1" (or | |
1830 | |"Latin1") is used as an equivalent for ISO-8859-1. But some people treat | |
1831 | |"Latin1" as referring just to the characters in the range 128 through 255, or | |
2de859ce | 1832 | |sometimes from 160 through 255. |
6a4c4cd4 DM |
1833 | |This documentation uses "Latin1" and "Latin-1" to refer to all 256 characters. |
1834 | | | |
1835 | |Note that Perl can be compiled and run under either ASCII or EBCDIC (See | |
1836 | |L<perlebcdic>). Most of the documentation (and even comments in the code) | |
1837 | |ignore the EBCDIC possibility. | |
1838 | |For almost all purposes the differences are transparent. | |
1839 | |As an example, under EBCDIC, | |
1840 | |instead of UTF-8, UTF-EBCDIC is used to encode Unicode strings, and so | |
1841 | |whenever this documentation refers to C<utf8> | |
1842 | |(and variants of that name, including in function names), | |
1843 | |it also (essentially transparently) means C<UTF-EBCDIC>. | |
1844 | |But the ordinals of characters differ between ASCII, EBCDIC, and | |
1845 | |the UTF- encodings, and a string encoded in UTF-EBCDIC may occupy a different | |
1846 | |number of bytes than in UTF-8. | |
1847 | | | |
6927cd44 KW |
1848 | |The organization of this document is tentative and subject to change. |
1849 | |Suggestions and patches welcome | |
1850 | |L<perl5-porters\@perl.org|mailto:perl5-porters\@perl.org>. | |
1851 | | | |
1852 | |The sections in this document currently are | |
1853 | | | |
1854 | |=over | |
1855 | ||
1856 | |$section_list | |
1857 | | | |
1858 | |=back | |
1859 | | | |
6a4c4cd4 | 1860 | |The listing below is alphabetical, case insensitive. |
94bdecf9 | 1861 | _EOB_ |
6a4c4cd4 DM |
1862 | |=head1 AUTHORS |
1863 | | | |
1864 | |Until May 1997, this document was maintained by Jeff Okamoto | |
7b1f0a98 | 1865 | |<okamoto\@corp.hp.com>. It is now maintained as part of Perl itself. |
6a4c4cd4 DM |
1866 | | |
1867 | |With lots of help and suggestions from Dean Roehrich, Malcolm Beattie, | |
1868 | |Andreas Koenig, Paul Hudson, Ilya Zakharevich, Paul Marquess, Neil | |
1869 | |Bowers, Matthew Green, Tim Bunce, Spider Boardman, Ulrich Pfeifer, | |
1870 | |Stephen McCamant, and Gurusamy Sarathy. | |
1871 | | | |
7b1f0a98 | 1872 | |API Listing originally by Dean Roehrich <roehrich\@cray.com>. |
6a4c4cd4 DM |
1873 | | |
1874 | |Updated to be autogenerated from comments in the source by Benjamin Stuhl. | |
1875 | | | |
1876 | |=head1 SEE ALSO | |
1877 | | | |
ea7c1d7a | 1878 | |F<config.h>, $places_other_than_api |
94bdecf9 JH |
1879 | _EOE_ |
1880 | ||
79fc8511 FC |
1881 | # List of non-static internal functions |
1882 | my @missing_guts = | |
9f589e47 | 1883 | grep $funcflags{$_}{flags} !~ /[AS]/ && !$docs{guts}{$_}, keys %funcflags; |
5a0155e6 | 1884 | |
7b1f0a98 | 1885 | output('perlintern', <<'_EOB_', $docs{guts}, \@missing_guts, <<"_EOE_"); |
6a4c4cd4 DM |
1886 | |=head1 NAME |
1887 | | | |
1888 | |perlintern - autogenerated documentation of purely B<internal> | |
20046047 | 1889 | |Perl functions |
6a4c4cd4 DM |
1890 | | |
1891 | |=head1 DESCRIPTION | |
1892 | |X<internal Perl functions> X<interpreter functions> | |
1893 | | | |
1894 | |This file is the autogenerated documentation of functions in the | |
1895 | |Perl interpreter that are documented using Perl's internal documentation | |
1896 | |format but are not marked as part of the Perl API. In other words, | |
1897 | |B<they are not for use in extensions>! | |
5781dcac KW |
1898 | |
1899 | |It has the same sections as L<perlapi>, though some may be empty. | |
6a4c4cd4 | 1900 | | |
7b1f0a98 | 1901 | _EOB_ |
6a4c4cd4 DM |
1902 | | |
1903 | |=head1 AUTHORS | |
1904 | | | |
1905 | |The autodocumentation system was originally added to the Perl core by | |
1906 | |Benjamin Stuhl. Documentation is by whoever was kind enough to | |
1907 | |document their functions. | |
1908 | | | |
1909 | |=head1 SEE ALSO | |
1910 | | | |
ea7c1d7a | 1911 | |F<config.h>, $places_other_than_intern |
7b1f0a98 | 1912 | _EOE_ |