xref: /dokuwiki/inc/parserutils.php (revision 16999ed115e8a9df3ce15f1efe7e4423b925c1d2)
1<?php
2
3/**
4 * Utilities for accessing the parser
5 *
6 * @license    GPL 2 (http://www.gnu.org/licenses/gpl.html)
7 * @author     Harry Fuecks <hfuecks@gmail.com>
8 * @author     Andreas Gohr <andi@splitbrain.org>
9 */
10
11use dokuwiki\Cache\CacheInstructions;
12use dokuwiki\Cache\CacheRenderer;
13use dokuwiki\ChangeLog\PageChangeLog;
14use dokuwiki\Extension\PluginController;
15use dokuwiki\Extension\Event;
16use dokuwiki\Parsing\Handler;
17use dokuwiki\Parsing\ModeRegistry;
18use dokuwiki\Parsing\Parser;
19
20/**
21 * How many pages shall be rendered for getting metadata during one request
22 * at maximum? Note that this limit isn't respected when METADATA_RENDER_UNLIMITED
23 * is passed as render parameter to p_get_metadata.
24 */
25if (!defined('P_GET_METADATA_RENDER_LIMIT')) define('P_GET_METADATA_RENDER_LIMIT', 5);
26
27/** Don't render metadata even if it is outdated or doesn't exist */
28define('METADATA_DONT_RENDER', 0);
29/**
30 * Render metadata when the page is really newer or the metadata doesn't exist.
31 * Uses just a simple check, but should work pretty well for loading simple
32 * metadata values like the page title and avoids rendering a lot of pages in
33 * one request. The P_GET_METADATA_RENDER_LIMIT is used in this mode.
34 * Use this if it is unlikely that the metadata value you are requesting
35 * does depend e.g. on pages that are included in the current page using
36 * the include plugin (this is very likely the case for the page title, but
37 * not for relation references).
38 */
39define('METADATA_RENDER_USING_SIMPLE_CACHE', 1);
40/**
41 * Render metadata using the metadata cache logic. The P_GET_METADATA_RENDER_LIMIT
42 * is used in this mode. Use this mode when you are requesting more complex
43 * metadata. Although this will cause rendering more often it might actually have
44 * the effect that less current metadata is returned as it is more likely than in
45 * the simple cache mode that metadata needs to be rendered for all pages at once
46 * which means that when the metadata for the page is requested that actually needs
47 * to be updated the limit might have been reached already.
48 */
49define('METADATA_RENDER_USING_CACHE', 2);
50/**
51 * Render metadata without limiting the number of pages for which metadata is
52 * rendered. Use this mode with care, normally it should only be used in places
53 * like the indexer or in cli scripts where the execution time normally isn't
54 * limited. This can be combined with the simple cache using
55 * METADATA_RENDER_USING_CACHE | METADATA_RENDER_UNLIMITED.
56 */
57define('METADATA_RENDER_UNLIMITED', 4);
58
59/**
60 * Returns the parsed Wikitext in XHTML for the given id and revision.
61 *
62 * If $excuse is true an explanation is returned if the file
63 * wasn't found
64 *
65 * @param string $id page id
66 * @param string|int $rev revision timestamp or empty string
67 * @param bool $excuse
68 * @param string $date_at
69 *
70 * @return null|string
71 * @author Andreas Gohr <andi@splitbrain.org>
72 *
73 */
74function p_wiki_xhtml($id, $rev = '', $excuse = true, $date_at = '')
75{
76    $file = wikiFN($id, $rev);
77    $ret = '';
78
79    //ensure $id is in global $ID (needed for parsing)
80    global $ID;
81    $keep = $ID;
82    $ID = $id;
83
84    if ($rev || $date_at) {
85        if (file_exists($file)) {
86            //no caching on old revisions
87            $ret = p_render('xhtml', p_get_instructions(io_readWikiPage($file, $id, $rev)), $info, $date_at);
88        } elseif ($excuse) {
89            $ret = p_locale_xhtml('norev');
90        }
91    } elseif (file_exists($file)) {
92        $ret = p_cached_output($file, 'xhtml', $id);
93    } elseif ($excuse) {
94        //check if the page once existed
95        $changelog = new PageChangeLog($id);
96        if ($changelog->hasRevisions()) {
97            $ret = p_locale_xhtml('onceexisted');
98        } else {
99            $ret = p_locale_xhtml('newpage');
100        }
101    }
102
103    //restore ID (just in case)
104    $ID = $keep;
105
106    return $ret;
107}
108
109/**
110 * Returns the specified local text in parsed format
111 *
112 * @param string $id page id
113 * @return null|string
114 * @author Andreas Gohr <andi@splitbrain.org>
115 *
116 */
117function p_locale_xhtml($id)
118{
119    //fetch parsed locale
120    $data = ['id' => $id, 'html' => ''];
121
122    $event = new Event('PARSER_LOCALE_XHTML', $data);
123    if ($event->advise_before()) {
124        $data['html'] = p_cached_output(localeFN($data['id']));
125    }
126    $event->advise_after();
127
128    return $data['html'];
129}
130
131/**
132 * Returns the given file parsed into the requested output format
133 *
134 * @param string $file filename, path to file
135 * @param string $format
136 * @param string $id page id
137 * @param string|null $syntax syntax flavour to parse under; null uses the
138 *     configured $conf['syntax']. When passed explicitly it also enters the
139 *     cache key so the same file rendered under two syntaxes never collides.
140 *     See p_get_instructions().
141 * @return null|string
142 * @author Andreas Gohr <andi@splitbrain.org>
143 * @author Chris Smith <chris@jalakai.co.uk>
144 *
145 */
146function p_cached_output($file, $format = 'xhtml', $id = '', $syntax = null)
147{
148    global $conf;
149
150    $cache = new CacheRenderer($id, $file, $format, $syntax);
151    if ($cache->useCache()) {
152        $parsed = $cache->retrieveCache(false);
153        if ($conf['allowdebug'] && $format == 'xhtml') {
154            $parsed .= "\n<!-- cachefile {$cache->cache} used -->\n";
155        }
156    } else {
157        $parsed = p_render($format, p_cached_instructions($file, false, $id, $syntax), $info);
158
159        if (!empty($info['cache']) && $cache->storeCache($parsed)) { // storeCache() attempts to save cachefile
160            if ($conf['allowdebug'] && $format == 'xhtml') {
161                $parsed .= "\n<!-- no cachefile used, but created {$cache->cache} -->\n";
162            }
163        } else {
164            $cache->removeCache(); //try to delete cachefile
165            if ($conf['allowdebug'] && $format == 'xhtml') {
166                $parsed .= "\n<!-- no cachefile used, caching forbidden -->\n";
167            }
168        }
169    }
170
171    return $parsed;
172}
173
174/**
175 * Returns the render instructions for a file
176 *
177 * Uses and creates a serialized cache file
178 *
179 * @param string $file filename, path to file
180 * @param bool $cacheonly
181 * @param string $id page id
182 * @param string|null $syntax syntax flavour to parse under; null uses the
183 *     configured $conf['syntax']. See p_get_instructions().
184 * @return array|null
185 * @author Andreas Gohr <andi@splitbrain.org>
186 *
187 */
188function p_cached_instructions($file, $cacheonly = false, $id = '', $syntax = null)
189{
190    static $run = null;
191    if (is_null($run)) $run = [];
192
193    // The in-request memo and the on-disk cache are both keyed on $syntax so
194    // the same file rendered under two syntaxes in one request (e.g. a plugin
195    // forcing 'dw' on a doc whose configured syntax is 'md') does not collide.
196    $runKey = $file . '|' . ($syntax ?? '');
197
198    $cache = new CacheInstructions($id, $file, $syntax);
199
200    if ($cacheonly || $cache->useCache() || (isset($run[$runKey]) && !defined('DOKU_UNITTEST'))) {
201        return $cache->retrieveCache();
202    } elseif (file_exists($file)) {
203        // no cache - do some work
204        $ins = p_get_instructions(io_readWikiPage($file, $id), $syntax);
205        if ($cache->storeCache($ins)) {
206            $run[$runKey] = true; // we won't rebuild these instructions in the same run again
207        } else {
208            msg('Unable to save cache file. Hint: disk full; file permissions; safe_mode setting.', -1);
209        }
210        return $ins;
211    }
212
213    return null;
214}
215
216/**
217 * turns a page into a list of instructions
218 *
219 * This is the one place in the parser pipeline where the configured wiki
220 * syntax preference ($conf['syntax']) is read. From here on the syntax is
221 * a parameter carried by the ModeRegistry, never a global lookup.
222 *
223 * @param string $text raw wiki syntax text
224 * @param string|null $syntax syntax flavour to parse under: 'dw', 'md',
225 *     'dw+md' or 'md+dw'. null (the default) means "use the configured
226 *     $conf['syntax']" — appropriate for user content. Locale/bundled
227 *     callers pass 'dw' explicitly.
228 * @return array a list of instruction arrays
229 * @author Harry Fuecks <hfuecks@gmail.com>
230 * @author Andreas Gohr <andi@splitbrain.org>
231 *
232 */
233function p_get_instructions($text, $syntax = null)
234{
235    global $conf;
236
237    $registry = new ModeRegistry($syntax ?? $conf['syntax']);
238
239    // Create the parser and handler
240    $Handler = new Handler($registry);
241    $Parser = new Parser($Handler, $registry);
242
243    //add modes to parser
244    foreach ($registry->getModes() as $mode) {
245        $Parser->addMode($mode['mode'], $mode['obj']);
246    }
247
248    // Do the parsing
249    Event::createAndTrigger('PARSER_WIKITEXT_PREPROCESS', $text);
250    return $Parser->parse($text);
251}
252
253/**
254 * returns the metadata of a page
255 *
256 * @param string $id The id of the page the metadata should be returned from
257 * @param string $key The key of the metdata value that shall be read (by default everything)
258 *                        separate hierarchies by " " like "date created"
259 * @param int $render If the page should be rendererd - possible values:
260 *     METADATA_DONT_RENDER, METADATA_RENDER_USING_SIMPLE_CACHE, METADATA_RENDER_USING_CACHE
261 *     METADATA_RENDER_UNLIMITED (also combined with the previous two options),
262 *     default: METADATA_RENDER_USING_CACHE
263 * @return mixed The requested metadata fields
264 *
265 * @author Esther Brunner <esther@kaffeehaus.ch>
266 * @author Michael Hamann <michael@content-space.de>
267 */
268function p_get_metadata($id, $key = '', $render = METADATA_RENDER_USING_CACHE)
269{
270    global $ID;
271    static $render_count = 0;
272    // track pages that have already been rendered in order to avoid rendering the same page
273    // again
274    static $rendered_pages = [];
275
276    // cache the current page
277    // Benchmarking shows the current page's metadata is generally the only page metadata
278    // accessed several times. This may catch a few other pages, but that shouldn't be an issue.
279    $cache = ($ID == $id);
280    $meta = p_read_metadata($id, $cache);
281
282    if (!is_numeric($render)) {
283        if ($render) {
284            $render = METADATA_RENDER_USING_SIMPLE_CACHE;
285        } else {
286            $render = METADATA_DONT_RENDER;
287        }
288    }
289
290    // prevent recursive calls in the cache
291    static $recursion = false;
292    if (!$recursion && $render != METADATA_DONT_RENDER && !isset($rendered_pages[$id]) && page_exists($id)) {
293        $recursion = true;
294
295        $cachefile = new CacheRenderer($id, wikiFN($id), 'metadata');
296
297        $do_render = false;
298        if ($render & METADATA_RENDER_UNLIMITED || $render_count < P_GET_METADATA_RENDER_LIMIT) {
299            if ($render & METADATA_RENDER_USING_SIMPLE_CACHE) {
300                $pagefn = wikiFN($id);
301                $metafn = metaFN($id, '.meta');
302                if (!file_exists($metafn) || @filemtime($pagefn) > @filemtime($cachefile->cache)) {
303                    $do_render = true;
304                }
305            } elseif (!$cachefile->useCache()) {
306                $do_render = true;
307            }
308        }
309        if ($do_render) {
310            if (!defined('DOKU_UNITTEST')) {
311                ++$render_count;
312                $rendered_pages[$id] = true;
313            }
314            $old_meta = $meta;
315            $meta = p_render_metadata($id, $meta);
316            // only update the file when the metadata has been changed
317            if ($meta == $old_meta || p_save_metadata($id, $meta)) {
318                // store a timestamp in order to make sure that the cachefile is touched
319                // this timestamp is also stored when the meta data is still the same
320                $cachefile->storeCache(time());
321            } else {
322                msg('Unable to save metadata file. Hint: disk full; file permissions; safe_mode setting.', -1);
323            }
324        }
325
326        $recursion = false;
327    }
328
329    $val = $meta['current'] ?? null;
330
331    // filter by $key
332    foreach (preg_split('/\s+/', $key, 2, PREG_SPLIT_NO_EMPTY) as $cur_key) {
333        if (!isset($val[$cur_key])) {
334            return null;
335        }
336        $val = $val[$cur_key];
337    }
338    return $val;
339}
340
341/**
342 * sets metadata elements of a page
343 *
344 * @see http://www.dokuwiki.org/devel:metadata#functions_to_get_and_set_metadata
345 *
346 * @param string $id is the ID of a wiki page
347 * @param array $data is an array with key ⇒ value pairs to be set in the metadata
348 * @param boolean $render whether or not the page metadata should be generated with the renderer
349 * @param boolean $persistent indicates whether or not the particular metadata value will persist through
350 *                            the next metadata rendering.
351 * @return boolean true on success
352 *
353 * @author Esther Brunner <esther@kaffeehaus.ch>
354 * @author Michael Hamann <michael@content-space.de>
355 */
356function p_set_metadata($id, $data, $render = false, $persistent = true)
357{
358    if (!is_array($data)) return false;
359
360    global $ID, $METADATA_RENDERERS;
361
362    // if there is currently a renderer change the data in the renderer instead
363    if (isset($METADATA_RENDERERS[$id])) {
364        $orig =& $METADATA_RENDERERS[$id];
365        $meta = $orig;
366    } else {
367        // cache the current page
368        $cache = ($ID == $id);
369        $orig = p_read_metadata($id, $cache);
370
371        // render metadata first?
372        $meta = $render ? p_render_metadata($id, $orig) : $orig;
373    }
374
375    // now add the passed metadata
376    $protected = ['description', 'date', 'contributor'];
377    foreach ($data as $key => $value) {
378        // be careful with sub-arrays of $meta['relation']
379        if ($key == 'relation') {
380            foreach ($value as $subkey => $subvalue) {
381                if (isset($meta['current'][$key][$subkey]) && is_array($meta['current'][$key][$subkey])) {
382                    $meta['current'][$key][$subkey] = array_replace($meta['current'][$key][$subkey], (array)$subvalue);
383                } else {
384                    $meta['current'][$key][$subkey] = $subvalue;
385                }
386                if ($persistent) {
387                    if (isset($meta['persistent'][$key][$subkey]) && is_array($meta['persistent'][$key][$subkey])) {
388                        $meta['persistent'][$key][$subkey] = array_replace(
389                            $meta['persistent'][$key][$subkey],
390                            (array)$subvalue
391                        );
392                    } else {
393                        $meta['persistent'][$key][$subkey] = $subvalue;
394                    }
395                }
396            }
397
398            // be careful with some senisitive arrays of $meta
399        } elseif (in_array($key, $protected, true)) {
400            // these keys, must have subkeys - a legitimate value must be an array
401            if (is_array($value)) {
402                $meta['current'][$key] = empty($meta['current'][$key]) ?
403                    $value :
404                    array_replace((array)$meta['current'][$key], $value);
405
406                if ($persistent) {
407                    $meta['persistent'][$key] = empty($meta['persistent'][$key]) ?
408                        $value :
409                        array_replace((array)$meta['persistent'][$key], $value);
410                }
411            }
412
413            // no special treatment for the rest
414        } else {
415            $meta['current'][$key] = $value;
416            if ($persistent) $meta['persistent'][$key] = $value;
417        }
418    }
419
420    // save only if metadata changed
421    if ($meta == $orig) return true;
422
423    if (isset($METADATA_RENDERERS[$id])) {
424        // set both keys individually as the renderer has references to the individual keys
425        $METADATA_RENDERERS[$id]['current'] = $meta['current'];
426        $METADATA_RENDERERS[$id]['persistent'] = $meta['persistent'];
427        return true;
428    } else {
429        return p_save_metadata($id, $meta);
430    }
431}
432
433/**
434 * Purges the non-persistant part of the meta data
435 * used on page deletion
436 *
437 * @param string $id page id
438 * @return bool  success / fail
439 * @author Michael Klier <chi@chimeric.de>
440 *
441 */
442function p_purge_metadata($id)
443{
444    $meta = p_read_metadata($id);
445    foreach ($meta['current'] as $key => $value) {
446        if (isset($meta[$key]) && is_array($meta[$key])) {
447            $meta['current'][$key] = [];
448        } else {
449            $meta['current'][$key] = '';
450        }
451    }
452    return p_save_metadata($id, $meta);
453}
454
455/**
456 * read the metadata from source/cache for $id
457 * (internal use only - called by p_get_metadata & p_set_metadata)
458 *
459 * @param string $id absolute wiki page id
460 * @param bool $cache whether or not to cache metadata in memory
461 *                             (only use for metadata likely to be accessed several times)
462 *
463 * @return   array             metadata
464 * @author   Christopher Smith <chris@jalakai.co.uk>
465 *
466 */
467function p_read_metadata($id, $cache = false)
468{
469    global $cache_metadata;
470
471    if (isset($cache_metadata[(string)$id])) return $cache_metadata[(string)$id];
472
473    $file = metaFN($id, '.meta');
474    $meta = file_exists($file) ?
475        unserialize(io_readFile($file, false)) :
476        ['current' => [], 'persistent' => []];
477
478    if ($cache) {
479        $cache_metadata[(string)$id] = $meta;
480    }
481
482    return $meta;
483}
484
485/**
486 * This is the backend function to save a metadata array to a file
487 *
488 * @param string $id absolute wiki page id
489 * @param array $meta metadata
490 *
491 * @return   bool              success / fail
492 */
493function p_save_metadata($id, $meta)
494{
495    // sync cached copies, including $INFO metadata
496    global $cache_metadata, $INFO;
497
498    if (isset($cache_metadata[$id])) $cache_metadata[$id] = $meta;
499    if (!empty($INFO) && isset($INFO['id']) && ($id == $INFO['id'])) {
500        $INFO['meta'] = $meta['current'];
501    }
502
503    return io_saveFile(metaFN($id, '.meta'), serialize($meta));
504}
505
506/**
507 * renders the metadata of a page
508 *
509 * @param string $id page id
510 * @param array $orig the original metadata
511 * @return array|null array('current'=> array,'persistent'=> array);
512 * @author Esther Brunner <esther@kaffeehaus.ch>
513 *
514 */
515function p_render_metadata($id, $orig)
516{
517    // make sure the correct ID is in global ID
518    global $ID, $METADATA_RENDERERS;
519
520    // avoid recursive rendering processes for the same id
521    if (isset($METADATA_RENDERERS[$id])) {
522        return $orig;
523    }
524
525    // store the original metadata in the global $METADATA_RENDERERS so p_set_metadata can use it
526    $METADATA_RENDERERS[$id] =& $orig;
527
528    $keep = $ID;
529    $ID = $id;
530
531    // add an extra key for the event - to tell event handlers the page whose metadata this is
532    $orig['page'] = $id;
533    $evt = new Event('PARSER_METADATA_RENDER', $orig);
534    if ($evt->advise_before()) {
535        // get instructions
536        $instructions = p_cached_instructions(wikiFN($id), false, $id);
537        if (is_null($instructions)) {
538            $ID = $keep;
539            unset($METADATA_RENDERERS[$id]);
540            return null; // something went wrong with the instructions
541        }
542
543        // set up the renderer
544        $renderer = new Doku_Renderer_metadata();
545        $renderer->meta =& $orig['current'];
546        $renderer->persistent =& $orig['persistent'];
547
548        // loop through the instructions
549        foreach ($instructions as $instruction) {
550            // execute the callback against the renderer
551            call_user_func_array([&$renderer, $instruction[0]], (array)$instruction[1]);
552        }
553
554        $evt->result = ['current' => &$renderer->meta, 'persistent' => &$renderer->persistent];
555    }
556    $evt->advise_after();
557
558    // clean up
559    $ID = $keep;
560    unset($METADATA_RENDERERS[$id]);
561    return $evt->result;
562}
563
564/**
565 * Renders a list of instruction to the specified output mode
566 *
567 * In the $info array is information from the renderer returned
568 *
569 * @param string $mode
570 * @param array|null|false $instructions
571 * @param array $info returns render info like enabled toc and cache
572 * @param string $date_at
573 * @return null|string rendered output
574 * @author Andreas Gohr <andi@splitbrain.org>
575 *
576 * @author Harry Fuecks <hfuecks@gmail.com>
577 */
578function p_render($mode, $instructions, &$info, $date_at = '')
579{
580    if (is_null($instructions)) return '';
581    if ($instructions === false) return '';
582
583    $Renderer = p_get_renderer($mode);
584    if (is_null($Renderer)) return null;
585
586    $Renderer->reset();
587
588    if ($date_at) {
589        $Renderer->date_at = $date_at;
590    }
591
592    $Renderer->smileys = getSmileys();
593    $Renderer->entities = getEntities();
594    $Renderer->acronyms = getAcronyms();
595    $Renderer->interwiki = getInterwiki();
596
597    // Loop through the instructions
598    foreach ($instructions as $instruction) {
599        // Execute the callback against the Renderer
600        if (method_exists($Renderer, $instruction[0])) {
601            call_user_func_array([&$Renderer, $instruction[0]], $instruction[1] ?: []);
602        }
603    }
604
605    //set info array
606    $info = $Renderer->info;
607
608    // Post process and return the output
609    $data = [$mode, & $Renderer->doc];
610    Event::createAndTrigger('RENDERER_CONTENT_POSTPROCESS', $data);
611    return $Renderer->doc;
612}
613
614/**
615 * Figure out the correct renderer class to use for $mode,
616 * instantiate and return it
617 *
618 * @param string $mode Mode of the renderer to get
619 * @return null|Doku_Renderer The renderer
620 *
621 * @author Christopher Smith <chris@jalakai.co.uk>
622 */
623function p_get_renderer($mode)
624{
625    /** @var PluginController $plugin_controller */
626    global $conf, $plugin_controller;
627
628    $rname = empty($conf['renderer_' . $mode]) ? $mode : $conf['renderer_' . $mode];
629    $rclass = "Doku_Renderer_$rname";
630
631    // if requested earlier or a bundled renderer
632    if (class_exists($rclass)) {
633        return new $rclass();
634    }
635
636    // not bundled, see if its an enabled renderer plugin & when $mode is 'xhtml', the renderer can supply that format.
637    /** @var Doku_Renderer $Renderer */
638    $Renderer = $plugin_controller->load('renderer', $rname);
639    if ($Renderer && is_a($Renderer, 'Doku_Renderer') && ($mode != 'xhtml' || $mode == $Renderer->getFormat())) {
640        return $Renderer;
641    }
642
643    // there is a configuration error!
644    // not bundled, not a valid enabled plugin, use $mode to try to fallback to a bundled renderer
645    $rclass = "Doku_Renderer_$mode";
646    if (class_exists($rclass)) {
647        // viewers should see renderered output, so restrict the warning to admins only
648        $msg = "No renderer '$rname' found for mode '$mode', check your plugins";
649        if ($mode == 'xhtml') {
650            $msg .= " and the 'renderer_xhtml' config setting";
651        }
652        $msg .= ".<br/>Attempting to fallback to the bundled renderer.";
653        msg($msg, -1, '', '', MSG_ADMINS_ONLY);
654
655        $Renderer = new $rclass();
656        $Renderer->nocache();     // fallback only (and may include admin alerts), don't cache
657        return $Renderer;
658    }
659
660    // fallback failed, alert the world
661    msg("No renderer '$rname' found for mode '$mode'", -1);
662    return null;
663}
664
665/**
666 * Gets the first heading from a file
667 *
668 * @param string $id dokuwiki page id
669 * @param int $render rerender if first heading not known
670 *                             default: METADATA_RENDER_USING_SIMPLE_CACHE
671 *                             Possible values: METADATA_DONT_RENDER,
672 *                                              METADATA_RENDER_USING_SIMPLE_CACHE,
673 *                                              METADATA_RENDER_USING_CACHE,
674 *                                              METADATA_RENDER_UNLIMITED
675 * @return string|null The first heading
676 *
677 * @author Andreas Gohr <andi@splitbrain.org>
678 * @author Michael Hamann <michael@content-space.de>
679 */
680function p_get_first_heading($id, $render = METADATA_RENDER_USING_SIMPLE_CACHE)
681{
682    return p_get_metadata(cleanID($id), 'title', $render);
683}
684
685/**
686 * Wrapper for GeSHi Code Highlighter, provides caching of its output
687 *
688 * @param string $code source code to be highlighted
689 * @param string $language language to provide highlighting
690 * @param string $wrapper html element to wrap the returned highlighted text
691 * @return string xhtml code
692 *
693 * @author Christopher Smith <chris@jalakai.co.uk>
694 * @author Andreas Gohr <andi@splitbrain.org>
695 */
696function p_xhtml_cached_geshi($code, $language, $wrapper = 'pre', ?array $options = null)
697{
698    global $conf, $config_cascade, $INPUT;
699    $language = strtolower($language);
700
701    // remove any leading or trailing blank lines
702    $code = preg_replace('/^\s*?\n|\s*?\n$/', '', $code);
703
704    $optionsmd5 = md5(serialize($options));
705    $cache = getCacheName($language . $code . $optionsmd5, ".code");
706    $ctime = @filemtime($cache);
707    if (
708        $ctime && !$INPUT->bool('purge') &&
709        $ctime > filemtime(DOKU_INC . 'vendor/composer/installed.json') &&  // libraries changed
710        $ctime > filemtime(reset($config_cascade['main']['default']))
711    ) { // dokuwiki changed
712        $highlighted_code = io_readFile($cache, false);
713    } else {
714        $geshi = new GeSHi($code, $language);
715        $geshi->set_encoding('utf-8');
716        $geshi->enable_classes();
717        $geshi->set_header_type(GESHI_HEADER_PRE);
718        $geshi->set_link_target($conf['target']['extern']);
719        if ($options !== null) {
720            foreach ($options as $function => $params) {
721                if (is_callable([$geshi, $function])) {
722                    $geshi->$function($params);
723                }
724            }
725        }
726
727        // remove GeSHi's wrapper element (we'll replace it with our own later)
728        // we need to use a GeSHi wrapper to avoid <BR> throughout the highlighted text
729        $highlighted_code = trim(preg_replace('!^<pre[^>]*>|</pre>$!', '', $geshi->parse_code()), "\n\r");
730        io_saveFile($cache, $highlighted_code);
731    }
732
733    // add a wrapper element if required
734    if ($wrapper) {
735        return "<$wrapper class=\"code $language\">$highlighted_code</$wrapper>";
736    } else {
737        return $highlighted_code;
738    }
739}
740