| blob | 2f4d7027bcae3b770095bd9ba01ddb6040061ed7 |
1 <?php
2 /**
3 * @defgroup Parser Parser
4 *
5 * @file
6 * @ingroup Parser
7 * File for Parser and related classes
8 */
11 /**
12 * PHP Parser - Processes wiki markup (which uses a more user-friendly
13 * syntax, such as "[[link]]" for making links), and provides a one-way
14 * transformation of that wiki markup it into XHTML output / markup
15 * (which in turn the browser understands, and can display).
16 *
17 * <pre>
18 * There are five main entry points into the Parser class:
19 * parse()
20 * produces HTML output
21 * preSaveTransform().
22 * produces altered wiki markup.
23 * preprocess()
24 * removes HTML comments and expands templates
25 * cleanSig()
26 * Cleans a signature before saving it to preferences
27 * extractSections()
28 * Extracts sections from an article for section editing
29 *
30 * Globals used:
31 * objects: $wgLang, $wgContLang
32 *
33 * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
34 *
35 * settings:
36 * $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
37 * $wgNamespacesWithSubpages, $wgAllowExternalImages*,
38 * $wgLocaltimezone, $wgAllowSpecialInclusion*,
39 * $wgMaxArticleSize*
40 *
41 * * only within ParserOptions
42 * </pre>
43 *
44 * @ingroup Parser
45 */
46 class Parser
47 {
48 /**
49 * Update this version number when the ParserOutput format
50 * changes in an incompatible way, so the parser cache
51 * can automatically discard old data.
52 */
55 # Flags for Parser::setFunctionHook
56 # Also available as global constants from Defines.php
60 # Constants needed for external link processing
61 # Everything except bracket, space, or control characters
66 // State constants for the definition list colon extraction
76 // Flags for preprocessToDom
79 // Allowed values for $this->mOutputType
80 // Parameter to startExternalParse().
86 // Marker Suffix needs to be accessible staticly.
89 /**#@+
90 * @private
91 */
92 # Persistent:
98 # Cleared with clearState():
107 # Temporary
108 # These are variables reset at least once per parse regardless of $clearState
117 /**#@-*/
119 /**
120 * Constructor
121 *
122 * @public
123 */
138 // PECL extension that conflicts with the core DOM extension (bug 13770)
145 }
148 }
150 /**
151 * Reduce memory usage to reduce the impact of circular references
152 */
156 }
159 }
160 }
162 /**
163 * Do various kinds of initialisation on the first call of the parser
164 */
168 }
179 }
181 /**
182 * Clear Parser state
183 *
184 * @private
185 */
190 }
203 /**
204 * Prefix for temporary replacement strings for the multipass parser.
205 * \x07 should never appear in input as it's disallowed in XML.
206 * Using it at the front also gives us a little extra robustness
207 * since it shouldn't match when butted up against identifier-like
208 * string constructs.
209 *
210 * Must not consist of all title characters, or else it will change
211 * the behaviour of <nowiki> in a link.
212 */
213 #$this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
214 # Changed to \x7f to allow XML double-parsing -- TS
218 # Clear these on every parse, bug 4549
226 );
233 # Fix cloning
236 }
240 }
244 // Shortcut alias
249 );
250 }
252 /**
253 * Set the context title
254 */
258 }
260 # Strip the fragment to avoid various odd effects
265 }
266 }
268 /**
269 * Accessor for mUniqPrefix.
270 *
271 * @public
272 */
275 // @fixme this is probably *horribly wrong*
276 // LanguageConverter seems to want $wgParser's uniqPrefix, however
277 // if this is called for a parser cache hit, the parser may not
278 // have ever been initialized in the first place.
279 // Not really sure what the heck is supposed to be going on here.
281 //throw new MWException( "Accessing uninitialized mUniqPrefix" );
282 }
284 }
286 /**
287 * Convert wikitext to HTML
288 * Do not call this function recursively.
289 *
290 * @param $text String: text we want to parse
291 * @param $title A title object
292 * @param $options ParserOptions
293 * @param $linestart boolean
294 * @param $clearState boolean
295 * @param $revid Int: number to pass in {{REVISIONID}}
296 * @return ParserOutput a ParserOutput
297 */
298 public function parse( $text, Title $title, ParserOptions $options, $linestart = true, $clearState = true, $revid = null ) {
299 /**
300 * First pass--just handle <nowiki> sections, pass the rest off
301 * to internalParse() which does all the real work.
302 */
311 }
320 }
323 # No more strip!
328 # Clean up special characters, only run once, next-to-last before doBlockLevels
330 # french spaces, last one Guillemet-left
331 # only if there is something before the space
333 # french spaces, Guillemet-right
336 );
343 # the position of the parserConvert() call should not be changed. it
344 # assumes that the links are all replaced and the only thing left
345 # is the <nowiki> mark.
346 # Side-effects: this calls $this->mOutput->setTitleText()
353 //!JF Move to its own function
368 }
370 }
378 # attempt to sanitize at least some nesting problems
379 # (bug #2702 and quite a few others)
381 # ''Something [http://www.cool.com cool''] -->
382 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
385 # fix up an anchor inside another anchor, only
386 # at least for a single single nested link (bug 3695)
389 # fix div inside inline elements- doBlockLevels won't wrap a line which
390 # contains a div, so fix it up here; replace
391 # div with escaped text
394 # remove empty italic or bold tag pairs, some
395 # introduced by rules above
397 );
403 }
406 $this->limitationWarn( 'expensive-parserfunction', $this->mExpensiveFunctionCount, $wgExpensiveParserFunctionLimit );
407 }
411 # Information on include size limits, for the benefit of users who try to skirt them
415 $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/$wgExpensiveParserFunctionLimit\n";
424 }
432 }
434 /**
435 * Recursive parser entry point that can be called from an extension tag
436 * hook.
437 */
445 }
447 /**
448 * Expand templates and variables in the text, producing valid, static wikitext.
449 * Also removes comments.
450 */
459 }
466 }
468 /**
469 * Get a random string
470 *
471 * @private
472 * @static
473 */
476 }
492 }
493 }
495 /**
496 * Get a preprocessor object
497 */
502 }
504 }
506 /**
507 * Replaces all occurrences of HTML-style comments and the given tags
508 * in the text with a random marker and returns the next text. The output
509 * parameter $matches will be an associative array filled with data in
510 * the form:
511 * 'UNIQ-xxxxx' => array(
512 * 'element',
513 * 'tag content',
514 * array( 'param' => 'x' ),
515 * '<element param="x">tag content</element>' ) )
516 *
517 * @param $elements list of element names. Comments are always extracted.
518 * @param $text Source text string.
519 * @param $uniq_prefix
520 *
521 * @public
522 * @static
523 */
537 }
539 // comment
545 // tag
550 }
556 // Empty element tag, <tag />
565 }
569 # No end tag -- let it run out to the end of the text.
575 }
576 }
582 }
584 }
586 /**
587 * Get a list of strippable XML-like elements
588 */
594 }
597 }
599 }
601 /**
602 * @deprecated use replaceVariables
603 */
606 }
608 /**
609 * Restores pre, math, and other extensions removed by strip()
610 *
611 * always call unstripNoWiki() after this one
612 * @private
613 * @deprecated use $this->mStripState->unstrip()
614 */
617 }
619 /**
620 * Always call this after unstrip() to preserve the order
621 *
622 * @private
623 * @deprecated use $this->mStripState->unstrip()
624 */
627 }
629 /**
630 * @deprecated use $this->mStripState->unstripBoth()
631 */
634 }
636 /**
637 * Add an item to the strip state
638 * Returns the unique tag which must be inserted into the stripped text
639 * The tag will be replaced with the original text in unstrip()
640 *
641 * @private
642 */
648 }
650 /**
651 * Interface with html tidy, used if $wgUseTidy = true.
652 * If tidy isn't able to correct the markup, the original will be
653 * returned in all its glory with a warning comment appended.
654 *
655 * Either the external tidy program or the in-process tidy extension
656 * will be used depending on availability. Override the default
657 * $wgTidyInternal setting to disable the internal if it's not working.
658 *
659 * @param string $text Hideous HTML input
660 * @return string Corrected HTML output
661 * @public
662 * @static
663 */
673 }
677 }
679 }
681 /**
682 * Spawn an external HTML tidy process and get corrected markup back from it.
683 *
684 * @private
685 * @static
686 */
698 );
701 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
703 // Theoretically, this style of communication could cause a deadlock
704 // here. If the stdout buffer fills up, then writes to stdin could
705 // block. This doesn't appear to happen with tidy, because tidy only
706 // writes to stdout after it's finished reading from stdin. Search
707 // for tidyParseStdin and tidySaveStdout in console/tidy.c
712 }
715 }
716 }
721 // Some kind of error happened, so we couldn't get the corrected text.
722 // Just give up; we'll use the source text and append a warning.
726 }
727 }
729 /**
730 * Use the HTML tidy PECL extension to use the tidy library in-process,
731 * saving the overhead of spawning a new process.
732 *
733 * 'pear install tidy' should be able to compile the extension module.
734 *
735 * @private
736 * @static
737 */
746 // 2 is magic number for fatal error
747 // http://www.php.net/manual/en/function.tidy-get-status.php
751 }
756 }
760 }
762 /**
763 * parse the wiki syntax used to render tables
764 *
765 * @private
766 */
785 }
790 // First check if we are starting a new table
803 // Don't do any of the following
807 // We are ending a table
813 }
817 }
821 }
825 // Now we have a table row
828 // Whats after the tag is now only attributes
841 }
845 }
851 }
852 else if ( $first_character === '|' || $first_character === '!' || substr ( $line , 0 , 2 ) === '|+' ) {
853 // This might be cell elements, td, th or captions
857 }
863 }
865 // Split up multiple cells on the same line.
866 // FIXME : This can result in improper nesting of tags processed
867 // by earlier parser steps, but should avoid splitting up eg
868 // attribute values containing literal "||".
873 // Loop through each table cell
875 {
878 {
882 }
887 }
893 }
903 }
907 // A cell could contain both parameters and data
910 // Bug 553: Note that a '|' inside an invalid link should not
911 // be mistaken as delimiting cell parameters
920 }
924 }
925 }
927 }
929 // Closing open td, tr && table
931 {
934 }
937 }
940 }
943 }
945 // Remove trailing line-ending (b/c)
948 }
950 // special case: don't return empty table
953 }
958 }
960 /**
961 * Helper function for parse() that transforms wiki markup into
962 * HTML. Only called for $mOutputType == self::OT_HTML.
963 *
964 * @private
965 */
970 # Hook to suspend the parser in this state
971 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
974 }
977 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
980 // Tables need to come after variable replacement for things to work
981 // properly; putting them before other transformations should keep
982 // exciting things like link expansions from showing up in surprising
983 // places.
993 }
998 # replaceInternalLinks may sometimes leave behind
999 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1007 }
1009 /**
1010 * Replace special strings like "ISBN xxx" and "RFC xxx" with
1011 * magic external links.
1012 *
1013 * DML
1014 * @private
1015 */
1021 '!(?: # Start cases
1022 (<a.*?</a>) | # m[1]: Skip link text
1025 (?:RFC|PMID)\s+([0-9]+) | # m[4]: RFC or PMID, capture number
1027 (?: 97[89] [\ \-]? )? # optional 13-digit ISBN prefix
1028 (?: [0-9] [\ \-]? ){9} # 9 digits with opt. delimiters
1029 [0-9Xx] # check digit
1034 }
1038 # Skip anchor
1041 # Skip HTML element
1044 # Free external link
1047 # RFC or PMID
1059 }
1065 # ISBN
1071 ));
1078 }
1079 }
1081 /**
1082 * Make a free external link, given a user-supplied URL
1083 * @return HTML
1084 * @private
1085 */
1093 # The characters '<' and '>' (which were escaped by
1094 # removeHTMLtags()) should not be included in
1095 # URLs, per RFC 2396.
1100 }
1102 # Move trailing punctuation to $trail
1104 # If there is no left bracket, then consider right brackets fair game too
1107 }
1113 }
1117 # Is this an external image?
1120 # Not an image, make a link
1121 $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free', $this->mTitle->getNamespace() );
1122 # Register it in the output object...
1123 # Replace unnecessary URL escape codes with their equivalent characters
1126 }
1129 }
1132 /**
1133 * Parse headers and return html
1134 *
1135 * @private
1136 */
1143 }
1146 }
1148 /**
1149 * Replace single quotes with HTML markup
1150 * @private
1151 * @return string the altered text
1152 */
1159 }
1163 }
1165 /**
1166 * Helper function for doAllQuotes()
1167 */
1172 else
1173 {
1174 # First, do some preliminary work. This may shift some apostrophes from
1175 # being mark-up to being text. It also counts the number of occurrences
1176 # of bold and italics mark-ups.
1181 {
1183 {
1184 # If there are ever four apostrophes, assume the first is supposed to
1185 # be text, and the remaining three constitute mark-up for bold text.
1187 {
1190 }
1191 # If there are more than 5 apostrophes in a row, assume they're all
1192 # text except for the last 5.
1194 {
1197 }
1198 # Count the number of occurrences of bold and italics mark-ups.
1199 # We are not counting sequences of five apostrophes.
1203 }
1205 }
1207 # If there is an odd number of both bold and italics, it is likely
1208 # that one of the bold ones was meant to be an apostrophe followed
1209 # by italics. Which one we cannot know for certain, but it is more
1210 # likely to be one that has a single-letter word before it.
1212 {
1218 {
1220 {
1229 }
1230 }
1232 }
1234 # If there is a single-letter word, use it!
1236 {
1239 }
1240 # If not, but there's a multi-letter word, use that one.
1242 {
1245 }
1246 # ... otherwise use the first one that has neither.
1247 # (notice that it is possible for all three to be -1 if, for example,
1248 # there is only one pentuple-apostrophe in the line)
1250 {
1253 }
1254 }
1256 # Now let's actually convert our apostrophic mush to HTML!
1262 {
1264 {
1267 else
1269 }
1270 else
1271 {
1273 {
1284 }
1286 {
1297 }
1299 {
1312 }
1313 }
1315 }
1316 # Now close all remaining tags. Notice that the order is important.
1323 # There might be lonely ''''', so make sure we have a buffer
1327 }
1328 }
1330 /**
1331 * Replace external links (REL)
1332 *
1333 * Note: this is all very hackish and the order of execution matters a lot.
1334 * Make sure to run maintenance/parserTests.php if you change this code.
1335 *
1336 * @private
1337 */
1354 # The characters '<' and '>' (which were escaped by
1355 # removeHTMLtags()) should not be included in
1356 # URLs, per RFC 2396.
1361 }
1363 # If the link text is an image URL, replace it with an <img> tag
1364 # This happened by accident in the original parser, but some people used it extensively
1368 }
1372 # Set linktype for CSS - if URL==text, link is essentially free
1375 # No link text, e.g. [http://domain.tld/some.link]
1377 # Autonumber if allowed. See bug #5918
1382 # Otherwise just use the URL
1385 }
1387 # Have link text, e.g. [http://domain.tld/some.link text]s
1388 # Check for trail
1390 }
1396 # Use the encoded URL
1397 # This means that users can paste URLs directly into the text
1398 # Funny characters like ö aren't valid in URLs anyway
1399 # This was changed in August 2004
1400 $s .= $sk->makeExternalLink( $url, $text, false, $linktype, $this->mTitle->getNamespace() ) . $dtrail . $trail;
1402 # Register link in the output object.
1403 # Replace unnecessary URL escape codes with the referenced character
1404 # This prevents spammers from hiding links from the filters
1407 }
1411 }
1413 /**
1414 * Replace unusual URL escape codes with their equivalent characters
1415 * @param string
1416 * @return string
1417 * @static
1418 * @todo This can merge genuinely required bits in the path or query string,
1419 * breaking legit URLs. A proper fix would treat the various parts of
1420 * the URL differently; as a workaround, just use the output for
1421 * statistical records, not for actual linking/output.
1422 */
1426 }
1428 /**
1429 * Callback function used in replaceUnusualEscapes().
1430 * Replaces unusual URL escape codes with their equivalent character
1431 * @static
1432 * @private
1433 */
1437 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1439 // No, shouldn't be escaped
1442 // Yes, leave it escaped
1444 }
1445 }
1447 /**
1448 * make an image if it's allowed, either through the global
1449 * option, through the exception, or through the on-wiki whitelist
1450 * @private
1451 */
1457 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1464 }
1465 }
1470 }
1474 # Image found
1476 }
1477 }
1482 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1486 # Image matches a whitelist entry
1489 }
1490 }
1491 }
1493 }
1495 /**
1496 * Process [[ ]] wikilinks
1497 * @return processed text
1498 *
1499 * @private
1500 */
1504 }
1506 /**
1507 * Process [[ ]] wikilinks (RIL)
1508 * @return LinkHolderArray
1509 *
1510 * @private
1511 */
1519 # the % is needed to support urlencoded titles as well
1522 # Match a link having the form [[namespace:link|alternate]]trail
1524 # Match cases where there is no "]]", which might still be images
1526 }
1531 #split the entire text string on occurences of [[
1533 #get the first element (all text up to first [[), and remove the space we added
1542 # Match the end of a line for a word that's not followed by whitespace,
1543 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1545 }
1551 }
1560 }
1563 }
1569 }
1573 # Loop for each link
1575 # Check for excessive memory usage
1577 # Too big
1578 # Do the existence check, replace the link holders and clear the array
1581 }
1590 }
1591 # first link
1595 }
1597 }
1604 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1605 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1606 # the real problem is with the $e1 regex
1607 # See bug 1300.
1608 #
1609 # Still some problems for cases where the ] is meant to be outside punctuation,
1610 # and no image is in sight. See bug 2095.
1611 #
1615 )
1616 {
1619 }
1620 # fix up urlencoded title texts
1622 # Should anchors '#' also be rejected?
1624 }
1626 } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1631 }
1637 }
1641 # Don't allow internal links to pages containing
1642 # PROTO: where PROTO is a valid URL protocol; these
1643 # should be external links.
1648 }
1650 # Make subpage if necessary
1655 }
1659 # Strip off leading ':'
1661 }
1670 }
1681 #look at the next 'line' to see if we can close it there
1686 }
1689 # the first ]] closes the inner link, the second the image
1695 #if there's exactly one ]] that's fine, we'll keep looking
1698 #if $next_line is invalid too, we need look no further
1701 }
1702 }
1704 # we couldn't find the end of this imageLink, so output it raw
1705 #but don't ignore what might be perfectly normal links in the text we've examined
1708 # note: no $trail, because without an end, there *is* no trail
1711 }
1714 # note: no $trail, because without an end, there *is* no trail
1717 }
1719 }
1724 # Link not escaped by : , create the various objects
1727 # Interwikis
1729 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1735 }
1741 # recursively parse links inside the image caption
1742 # actually, this will parse them in any other parameters, too,
1743 # but it might be hard to fix that, and it doesn't matter ATM
1747 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1749 }
1754 }
1764 }
1770 /**
1771 * Strip the whitespace Category links produce, see bug 87
1772 * @todo We might want to use trim($tmp, "\n") here.
1773 */
1778 }
1779 }
1781 # Self-link checking
1786 }
1787 }
1789 # Special and Media are pseudo-namespaces; no pages actually exist in them
1791 # Give extensions a chance to select the file revision for us
1798 }
1799 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1808 }
1813 // Force a blue link if the file exists; may be a remote
1814 // upload on the shared repository, and we want to see its
1815 // auto-generated page.
1819 }
1820 }
1822 }
1825 }
1827 /**
1828 * Make a link placeholder. The text returned can be later resolved to a real link with
1829 * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1830 * parsing of interwiki links, and secondly to allow all existence checks and
1831 * article length checks (for stub links) to be bundled into a single query.
1832 *
1833 * @deprecated
1834 */
1837 }
1839 /**
1840 * Render a forced-blue link inline; protect against double expansion of
1841 * URLs if we're in a mode that prepends full URL prefixes to internal links.
1842 * Since this little disaster has to split off the trail text to avoid
1843 * breaking URLs in the following text without breaking trails on the
1844 * wiki links, it's been made into a horrible function.
1845 *
1846 * @param Title $nt
1847 * @param string $text
1848 * @param string $query
1849 * @param string $trail
1850 * @param string $prefix
1851 * @return string HTML-wikitext mix oh yuck
1852 */
1858 }
1860 /**
1861 * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1862 * going to go through further parsing steps before inline URL expansion.
1863 *
1864 * Not needed quite as much as it used to be since free links are a bit
1865 * more sensible these days. But bracketed links are still an issue.
1866 *
1867 * @param string more-or-less HTML
1868 * @return string less-or-more HTML with NOPARSE bits
1869 */
1873 }
1875 /**
1876 * Return true if subpage links should be expanded on this page.
1877 * @return bool
1878 */
1880 # Some namespaces don't allow subpages
1882 }
1884 /**
1885 * Handle link to subpage if necessary
1886 * @param string $target the source of the link
1887 * @param string &$text the link text, modified as necessary
1888 * @return string the full name of the link
1889 * @private
1890 */
1892 # Valid link forms:
1893 # Foobar -- normal
1894 # :Foobar -- override special treatment of prefix (images, language links)
1895 # /Foobar -- convert to CurrentPage/Foobar
1896 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1897 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1898 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1903 # Some namespaces don't allow subpages,
1904 # so only perform processing if subpages are allowed
1912 }
1913 # bug 7425
1915 # Look at the first character
1917 # / at end means we don't want the slash to be shown
1924 }
1931 # check for .. subpage backlinks
1937 }
1942 # / at the end means don't show full path
1947 }
1948 }
1952 }
1954 }
1955 }
1956 }
1957 }
1961 }
1963 /**#@+
1964 * Used by doBlockLevels()
1965 * @private
1966 */
1971 }
1975 }
1976 # getCommon() returns the length of the longest common substring
1977 # of both arguments, starting at the beginning of both.
1978 #
1986 }
1988 }
1989 # These next three functions open, continue, and close the list
1990 # element appropriate to the prefix character passed into them.
1991 #
2001 }
2005 }
2018 }
2019 }
2021 }
2032 }
2033 }
2036 }
2037 /**#@-*/
2039 /**
2040 * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2041 *
2042 * @private
2043 * @return string the lists rendered as HTML
2044 */
2048 # Parsing through the text line by line. The main thing
2049 # happening here is handling of block-level elements p, pre,
2050 # and making lists from lines starting with * # : etc.
2051 #
2060 # Fix up $linestart
2065 }
2071 # Multiple prefixes may abut each other for nested lists.
2075 # eh?
2080 # Don't interpret any other prefixes in preformatted text
2084 }
2086 # List generation
2088 # Same as the last item, so no need to deal with nesting or opening stuff
2093 # The one nasty exception: definition lists work like this:
2094 # ; title : definition text
2095 # So we check for : in the remainder text to split up the
2096 # title and definition, without b0rking links.
2101 }
2102 }
2104 # Either open or close a level...
2111 }
2114 }
2120 # FIXME: This is dupe of code above
2124 }
2125 }
2127 }
2129 }
2132 # No prefix (not in list)--go to paragraph mode
2133 // XXX: use a stack for nestable elements like span, table and div
2134 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2137 '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2140 #Â TODO bug 5718: paragraph closed
2144 }
2149 }
2152 // pre
2157 }
2160 // paragraph
2173 }
2174 }
2183 }
2184 }
2185 }
2186 }
2188 }
2189 // somewhere above we forget to get out of pre block (bug 785)
2192 }
2195 }
2196 }
2200 }
2204 }
2208 }
2210 /**
2211 * Split up a string on ':', ignoring any occurences inside tags
2212 * to prevent illegal overlapping.
2213 * @param string $str the string to split
2214 * @param string &$before set to everything before the ':'
2215 * @param string &$after set to everything after the ':'
2216 * return string the position of the ':', or false if none found
2217 */
2223 // Nothing to find!
2226 }
2230 // Easy; no tag nesting to worry about
2235 }
2237 // Ugly state machine to walk through avoiding tags.
2245 // (Using the number is a performance hack for common cases)
2249 // Could be either a <start> tag or an </end> tag
2254 // We found it!
2259 }
2260 // Embedded in a tag; don't break it.
2263 // Skip ahead looking for something interesting
2266 // Nothing else interesting
2269 }
2273 // We found it!
2278 }
2279 }
2281 // Nothing else interesting to find; abort!
2282 // We're nested, but there's no close tags left. Abort!
2284 }
2285 // Skip ahead to next tag start
2288 }
2291 // In a <tag>
2298 // Slash may be followed by >?
2302 // ignore
2303 }
2314 // Illegal early close? This shouldn't happen D:
2319 }
2322 // In a </tag>
2329 }
2331 }
2335 // Yes, a self-closed tag <blah/>
2338 // Probably we're jumping the gun, and this is an attribute
2340 }
2345 }
2352 }
2359 }
2363 }
2364 }
2368 }
2371 }
2373 /**
2374 * Return value of a magic variable (like PAGENAME)
2375 *
2376 * @private
2377 */
2381 /**
2382 * Some of these require message or data lookups and can be
2383 * expensive to check many times.
2384 */
2388 }
2389 }
2394 # Use the time zone
2399 }
2413 }
2463 }
2470 }
2478 // Let the edit saving system know we should parse the page
2479 // *after* a revision ID has been assigned.
2484 // Let the edit saving system know we should parse the page
2485 // *after* a revision ID has been assigned. This is for null edits.
2490 // Let the edit saving system know we should parse the page
2491 // *after* a revision ID has been assigned. This is for null edits.
2496 // Let the edit saving system know we should parse the page
2497 // *after* a revision ID has been assigned. This is for null edits.
2502 // Let the edit saving system know we should parse the page
2503 // *after* a revision ID has been assigned. This is for null edits.
2508 // Let the edit saving system know we should parse the page
2509 // *after* a revision ID has been assigned. This is for null edits.
2530 return $this->mVarCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2534 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2535 // int to remove the padding
2548 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2549 // int to remove the padding
2586 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret ) ) )
2588 else
2590 }
2591 }
2593 /**
2594 * initialise the magic variables (like CURRENTMONTHNAME)
2595 *
2596 * @private
2597 */
2604 }
2606 /**
2607 * Preprocess some wikitext and return the document tree.
2608 * This is the ghost of replace_variables().
2609 *
2610 * @param string $text The text to parse
2611 * @param integer flags Bitwise combination of:
2612 * self::PTD_FOR_INCLUSION Handle <noinclude>/<includeonly> as if the text is being
2613 * included. Default is to assume a direct page view.
2614 *
2615 * The generated DOM tree must depend only on the input text and the flags.
2616 * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2617 *
2618 * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2619 * change in the DOM tree for a given text, must be passed through the section identifier
2620 * in the section edit link and thus back to extractSections().
2621 *
2622 * The output of this function is currently only cached in process memory, but a persistent
2623 * cache may be implemented at a later date which takes further advantage of these strict
2624 * dependency requirements.
2625 *
2626 * @private
2627 */
2631 }
2633 /*
2634 * Return a three-element array: leading whitespace, string contents, trailing whitespace
2635 */
2645 }
2647 }
2649 /**
2650 * Replace magic variables, templates, and template arguments
2651 * with the appropriate text. Templates are substituted recursively,
2652 * taking care to avoid infinite loops.
2653 *
2654 * Note that the substitution depends on value of $mOutputType:
2655 * self::OT_WIKI: only {{subst:}} templates
2656 * self::OT_PREPROCESS: templates but not extension tags
2657 * self::OT_HTML: all templates and extension tags
2658 *
2659 * @param string $tex The text to transform
2660 * @param PPFrame $frame Object describing the arguments passed to the template.
2661 * Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2662 * Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2663 * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2664 * @private
2665 */
2667 # Prevent too big inclusions
2670 }
2677 wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
2679 }
2687 }
2689 /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2702 }
2705 }
2706 }
2707 }
2710 }
2712 /**
2713 * Warn the user when a parser limitation is reached
2714 * Will warn at most once the user per limitation type
2715 *
2716 * @param string $limitationType, should be one of:
2717 * 'expensive-parserfunction' (corresponding messages: 'expensive-parserfunction-warning', 'expensive-parserfunction-category')
2718 * 'post-expand-template-argument' (corresponding messages: 'post-expand-template-argument-warning', 'post-expand-template-argument-category')
2719 * 'post-expand-template-inclusion' (corresponding messages: 'post-expand-template-inclusion-warning', 'post-expand-template-inclusion-category')
2720 * @params int $current, $max When an explicit limit has been
2721 * exceeded, provide the values (optional)
2722 */
2725 //does no harm if $current and $max are present but are unnecessary for the message
2731 }
2732 }
2734 /**
2735 * Return the text of a template, after recursively
2736 * replacing any variables or templates within the template.
2737 *
2738 * @param array $piece The parts of the template
2739 * $piece['title']: the title, i.e. the part before the |
2740 * $piece['parts']: the parameter array
2741 * $piece['lineStart']: whether the brace was at the start of a line
2742 * @param PPFrame The current frame, contains template arguments
2743 * @return string the text of the template
2744 * @private
2745 */
2751 # Flags
2759 # Title object, where $text came from
2762 # $part1 is the bit before the first |, and must contain only title characters.
2763 # Various prefixes will be stripped from it later.
2768 # Original title text preserved for various purposes
2771 # $args is a list of argument nodes, starting from index 0, not including $part1
2775 # SUBST
2780 # One of two possibilities is true:
2781 # 1) Found SUBST but not in the PST phase
2782 # 2) Didn't find SUBST and in the PST phase
2783 # In either case, return without further processing
2787 }
2788 }
2790 # Variables
2798 }
2799 }
2801 # MSG, MSGNW and RAW
2803 # Check for MSGNW:
2808 # Remove obsolete MSG:
2811 }
2813 # Check for RAW:
2817 }
2818 }
2821 # Parser functions
2827 # Case sensitive functions
2832 # Case insensitive functions
2838 }
2839 }
2845 # Add a frame parameter, and pass the arguments as an array
2850 }
2853 # Convert arguments to plain text
2856 }
2858 }
2860 # Workaround for PHP bug 35229 and similar
2863 }
2873 }
2875 // Extract flags into the local scope
2876 // This allows callers to set flags such as nowiki, found, etc.
2880 }
2884 }
2885 }
2886 }
2888 }
2890 # Finish mangling title and then check for loops.
2891 # Set $title to a Title object and $titleText to the PDBK
2894 # Split the title into page and subpage
2899 }
2903 # Check for language variants if the template is not found
2906 }
2907 # Do infinite loop check
2912 }
2913 # Do recursion depth check
2918 }
2919 }
2920 }
2922 # Load from database
2926 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
2932 }
2933 } else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
2941 }
2942 }
2944 # If the title is valid but undisplayable, make a link to it
2948 }
2950 // Interwiki transclusion
2956 // Preprocess it like a template
2959 }
2961 }
2963 }
2965 # If we haven't found text to substitute by now, we're done
2966 # Recover the source wikitext and return it
2971 }
2973 # Expand DOM-style return values in a child frame
2975 # Clean up argument array
2981 # Expansion is eligible for the empty-frame cache
2987 }
2989 # Uncached expansion
2991 }
2992 }
2996 }
2998 # Replace raw HTML by a placeholder
2999 # Add a blank line preceding, to prevent it from mucking up
3000 # immediately preceding headings
3003 }
3004 # Escape nowiki-style return values
3007 }
3008 # Bug 529: if the template begins with a table or block-level
3009 # element, it should be treated as beginning a new line.
3010 # This behaviour is somewhat controversial.
3011 elseif ( is_string( $text ) && !$piece['lineStart'] && preg_match('/^(?:{\\||:|;|#|\*)/', $text)) /*}*/{
3013 }
3016 # Error, oversize inclusion
3018 $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3020 }
3026 }
3030 }
3032 /**
3033 * Get the semi-parsed DOM representation of a template with a given title,
3034 * and its redirect destination title. Cached.
3035 */
3044 }
3047 }
3049 // Cache miss, go to the database
3055 }
3063 }
3066 }
3068 /**
3069 * Fetch the unparsed text of a template and register a reference to it.
3070 */
3079 }
3080 }
3082 }
3087 }
3089 /**
3090 * Static function to get a template
3091 * Can be overridden via ParserOptions::setTemplateCallback().
3092 */
3098 // Loop to fetch the article, with up to 1 redirect
3100 # Give extensions a chance to select the revision instead
3111 }
3114 // If there is no current revision, there is no page
3118 }
3134 }
3137 }
3140 }
3141 // Redirect?
3144 }
3149 }
3151 /**
3152 * Transclude an interwiki link.
3153 */
3164 return "<div class=\"mw-iw-transclusion\">\n" . $this->fetchScaryTemplateMaybeFromCache($url) . "</div>\n";
3165 }
3177 }
3178 }
3190 }
3193 /**
3194 * Triple brace replacement -- used for template arguments
3195 * @private
3196 */
3207 && (
3211 )
3212 ) {
3213 # No match in frame, use the supplied default
3215 }
3219 }
3222 # No match anywhere
3224 }
3227 }
3232 }
3236 }
3238 /**
3239 * Return the text to be used for a given extension tag.
3240 * This is the ghost of strip().
3241 *
3242 * @param array $params Associative array of parameters:
3243 * name PPNode for the tag name
3244 * attr PPNode for unparsed text where tag attributes are thought to be
3245 * attributes Optional associative array of parsed attributes
3246 * inner Contents of extension element
3247 * noClose Original text did not have a close tag
3248 * @param PPFrame $frame
3249 */
3257 $marker = "{$this->mUniqPrefix}-$name-" . sprintf('%08X', $this->mMarkerIndex++) . self::MARKER_SUFFIX;
3265 }
3273 }
3286 # Workaround for PHP bug 35229 and similar
3289 }
3295 }
3296 }
3300 }
3305 }
3306 }
3312 }
3313 }
3319 }
3321 }
3323 /**
3324 * Increment an include size counter
3325 *
3326 * @param string $type The type of expansion
3327 * @param integer $size The size of the text
3328 * @return boolean False if this inclusion would take it over the maximum, true otherwise
3329 */
3336 }
3337 }
3339 /**
3340 * Increment the expensive function count
3341 *
3342 * @return boolean False if the limit has been exceeded
3343 */
3349 }
3351 }
3353 /**
3354 * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3355 * Fills $this->mDoubleUnderscores, returns the modified text
3356 */
3358 // The position of __TOC__ needs to be recorded
3364 // Set a placeholder. At the end we'll fill it in with the TOC.
3367 // Only keep the first one.
3369 }
3371 // Now match and remove the rest of them
3377 }
3380 }
3381 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3384 $containerCategory = Title::makeTitleSafe( NS_CATEGORY, wfMsgForContent( 'hidden-category-category' ) );
3389 }
3390 }
3391 # (bug 8068) Allow control over whether robots index a page.
3392 #
3393 # FIXME (bug 14899): __INDEX__ always overrides __NOINDEX__ here! This
3394 # is not desirable, the last one on the page should win.
3399 }
3402 }
3404 /**
3405 * This function accomplishes several tasks:
3406 * 1) Auto-number headings if that option is enabled
3407 * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3408 * 3) Add a Table of contents on the top for users who have enabled the option
3409 * 4) Auto-anchor headings
3410 *
3411 * It loops through all headlines, collects the necessary data, then splits up the
3412 * string and re-inserts the newly formatted headlines.
3413 *
3414 * @param string $text
3415 * @param boolean $isMain
3416 * @private
3417 */
3426 }
3428 # Inhibit editsection links if requested in the page
3431 }
3433 # Get all headlines for numbering them and adding funky stuff like [edit]
3434 # links - this is for later, but we need the number of headlines right now
3436 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3438 # if there are fewer than 4 headlines in the article, do not show TOC
3439 # unless it's been explicitly enabled.
3443 # Allow user to stipulate that a page should have a "new section"
3444 # link added via __NEWSECTIONLINK__
3447 }
3449 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3450 # override above conditions and always show TOC above first header
3454 }
3456 # We need this to perform operations on the HTML
3459 # headline counter
3463 # Ugh .. the TOC should have neat indentation levels which can be
3464 # passed to the skin functions. These are determined here
3490 }
3495 }
3501 # Increase TOC level
3508 }
3509 }
3511 # Decrease TOC level, find level to jump to
3514 # Can only go down to level 1
3519 # Found last matching level
3522 }
3524 # Found first matching level below current level
3527 }
3528 }
3529 }
3532 # Unindent only if the previous toc level was shown :p
3537 }
3538 }
3539 }
3541 # No change in level, end TOC line
3544 }
3545 }
3549 # count number of headlines for each level
3556 }
3559 }
3560 }
3561 }
3563 # The safe header is a version of the header text safe to use for links
3564 # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3567 # Remove link placeholders by the link text.
3568 # <!--LINK number-->
3569 # turns into
3570 # link text with suffix
3573 # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3577 $safeHeadline
3578 );
3581 # For the anchor, strip out HTML-y stuff period
3585 # Save headline for section edit hint before it's escaped
3588 # HTML names must be case-insensitively unique (bug 10721)
3591 # count how many in assoc. array so we can track dupes in anchors
3595 # Don't number the heading if it is the only one (looks silly)
3597 # the two are different if the line contains a link
3599 }
3601 # Create the anchor for linking from the TOC to the section
3605 }
3608 $tocraw[] = array( 'toclevel' => $toclevel, 'level' => $level, 'line' => $tocline, 'number' => $numbering );
3609 }
3610 # give headline the correct <h#> tag
3613 # Put a T flag in the section identifier, to indicate to extractSections()
3614 # that sections inside <includeonly> should be counted.
3618 }
3621 }
3622 $head[$headlineCount] = $sk->makeHeadline( $level, $matches['attrib'][$headlineCount], $anchor, $headline, $editlink );
3625 }
3629 # Never ever show TOC if no headers
3632 }
3637 }
3639 }
3641 # split up and insert constructed headlines
3648 # This is the [edit] link that appears for the top block of text when
3649 # section editing is enabled
3651 # Disabled because it broke block formatting
3652 # For example, a bullet point in the top line
3653 # $full .= $sk->editSectionLink(0);
3654 }
3657 # Top anchor now in skin
3659 }
3663 }
3665 }
3670 }
3671 }
3673 /**
3674 * Transform wiki markup when saving a page by doing \r\n -> \n
3675 * conversion, substitting signatures, {{subst:}} templates, etc.
3676 *
3677 * @param string $text the text to transform
3678 * @param Title &$title the Title object for the current article
3679 * @param User &$user the User object describing the current user
3680 * @param ParserOptions $options parsing options
3681 * @param bool $clearState whether to clear the parser state first
3682 * @return string the altered wiki markup
3683 * @public
3684 */
3692 }
3696 );
3701 }
3703 /**
3704 * Pre-save transform helper function
3705 * @private
3706 */
3710 /* Note: This is the timestamp saved as hardcoded wikitext to
3711 * the database, we use $wgContLang here in order to give
3712 * everyone the same signature and use the default one rather
3713 * than the one selected in each user's preferences.
3714 *
3715 * (see also bug 12815)
3716 */
3726 }
3730 # Variable replacement
3731 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3734 # Signatures
3740 ) );
3742 # Context links: [[|name]] and [[name (context)|]]
3743 #
3750 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/"; # [[ns:page (context), context|]]
3753 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
3765 # if there's no context, don't bother duplicating the title
3767 }
3769 # Trim trailing whitespace
3773 }
3775 /**
3776 * Fetch the user's signature text, if any, and normalize to
3777 * validated, ready-to-insert wikitext.
3778 *
3779 * @param User $user
3780 * @return string
3781 * @private
3782 */
3794 # Sig. might contain markup; validate this
3796 # Validated; clean up (if needed) and return it
3799 # Failed to validate; fall back to the default
3802 }
3803 }
3805 // Make sure nickname doesnt get a sig in a sig
3808 # If we're still here, make it a link to the user page
3815 }
3816 }
3818 /**
3819 * Check that the user's signature contains no bad XML
3820 *
3821 * @param string $text
3822 * @return mixed An expanded string, or false if invalid.
3823 */
3826 }
3828 /**
3829 * Clean up signature text
3830 *
3831 * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
3832 * 2) Substitute all transclusions
3833 *
3834 * @param string $text
3835 * @param $parsing Whether we're cleaning (preferences save) or parsing
3836 * @return string Signature text
3837 */
3845 }
3847 # Option to disable this feature
3850 }
3852 # FIXME: regex doesn't respect extension tags or nowiki
3853 # => Move this logic to braceSubstitution()
3855 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3866 }
3869 }
3871 /**
3872 * Strip ~~~, ~~~~ and ~~~~~ out of signatures
3873 * @param string $text
3874 * @return string Signature text with /~{3,5}/ removed
3875 */
3879 }
3881 /**
3882 * Set up some variables which are usually set up in parse()
3883 * so that an external function can call some class members with confidence
3884 * @public
3885 */
3892 }
3893 }
3895 /**
3896 * Wrapper for preprocess()
3897 *
3898 * @param string $text the text to preprocess
3899 * @param ParserOptions $options options
3900 * @return string
3901 * @public
3902 */
3907 # Guard against infinite recursion
3910 }
3919 }
3921 /**
3922 * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3923 * The callback should have the following form:
3924 * function myParserHook( $text, $params, &$parser ) { ... }
3925 *
3926 * Transform and return $text. Use $parser for any required context, e.g. use
3927 * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
3928 *
3929 * @public
3930 *
3931 * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3932 * @param mixed $callback The callback function (and object) to use for the tag
3933 *
3934 * @return The old value of the mTagHooks array associated with the hook
3935 */
3942 }
3945 }
3949 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
3953 }
3955 /**
3956 * Remove all tag hooks
3957 */
3961 }
3963 /**
3964 * Create a function, e.g. {{sum:1|2|3}}
3965 * The callback function should have the form:
3966 * function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
3967 *
3968 * Or with SFH_OBJECT_ARGS:
3969 * function myParserFunction( $parser, $frame, $args ) { ... }
3970 *
3971 * The callback may either return the text result of the function, or an array with the text
3972 * in element 0, and a number of flags in the other elements. The names of the flags are
3973 * specified in the keys. Valid flags are:
3974 * found The text returned is valid, stop processing the template. This
3975 * is on by default.
3976 * nowiki Wiki markup in the return value should be escaped
3977 * isHTML The returned text is HTML, armour it against wikitext transformation
3978 *
3979 * @public
3980 *
3981 * @param string $id The magic word ID
3982 * @param mixed $callback The callback function (and object) to use
3983 * @param integer $flags a combination of the following flags:
3984 * SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
3985 *
3986 * SFH_OBJECT_ARGS Pass the template arguments as PPNode objects instead of text. This
3987 * allows for conditional expansion of the parse tree, allowing you to eliminate dead
3988 * branches and thus speed up parsing. It is also possible to analyse the parse tree of
3989 * the arguments, and to control the way they are expanded.
3990 *
3991 * The $frame parameter is a PPFrame. This can be used to produce expanded text from the
3992 * arguments, for instance:
3993 * $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
3994 *
3995 * For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
3996 * future versions. Please call $frame->expand() on it anyway so that your code keeps
3997 * working if/when this is changed.
3998 *
3999 * If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4000 * expansion.
4001 *
4002 * Please read the documentation in includes/parser/Preprocessor.php for more information
4003 * about the methods available in PPFrame and PPNode.
4004 *
4005 * @return The old callback function for this name, if any
4006 */
4011 # Add to function cache
4020 # Case
4023 }
4024 # Add leading hash
4027 }
4028 # Remove trailing colon
4031 }
4033 }
4035 }
4037 /**
4038 * Get all registered function hook identifiers
4039 *
4040 * @return array
4041 */
4044 }
4046 /**
4047 * Replace <!--LINK--> link placeholders with actual links, in the buffer
4048 * Placeholders created in Skin::makeLinkObj()
4049 * Returns an array of link CSS classes, indexed by PDBK.
4050 */
4053 }
4055 /**
4056 * Replace <!--LINK--> link placeholders with plain text of links
4057 * (not HTML-formatted).
4058 * @param string $text
4059 * @return string
4060 */
4063 }
4065 /**
4066 * Tag hook handler for 'pre'.
4067 */
4069 // Backwards-compatibility hack
4076 }
4078 /**
4079 * Renders an image gallery from a text with one line per image.
4080 * text labels may be given by using |-style alternative text. E.g.
4081 * Image:one.jpg|The number "1"
4082 * Image:tree.jpg|A tree
4083 * given as text will return the HTML of a gallery with two images,
4084 * labeled 'The number "1"' and
4085 * 'A tree'.
4086 */
4103 }
4106 }
4109 }
4112 }
4118 # match lines like these:
4119 # Image:someimage.jpg|This is some image
4122 # Skip empty lines
4125 }
4132 # Bogus title. Ignore these so we don't bomb out later.
4134 }
4139 }
4145 # Only add real images (bug #5586)
4148 }
4149 }
4151 }
4158 }
4160 // Initialise static lists
4167 );
4175 }
4176 }
4177 }
4179 // Add handler params
4185 }
4186 }
4189 }
4190 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4191 }
4193 /**
4194 * Parse image options text and use it to make an image
4195 * @param Title $title
4196 * @param string $options
4197 * @param LinkHolderArray $holders
4198 */
4200 # Check if the options text is of the form "options|alt text"
4201 # Options are:
4202 # * thumbnail make a thumbnail with enlarge-icon and caption, alignment depends on lang
4203 # * left no resizing, just left align. label is used for alt= only
4204 # * right same, but right aligned
4205 # * none same, but not aligned
4206 # * ___px scale to ___ pixels width, no aligning. e.g. use in taxobox
4207 # * center center the image
4208 # * framed Keep original image size, no magnify-button.
4209 # * frameless like 'thumb' but without a frame. Keeps user preferences for width
4210 # * upright reduce width for upright images, rounded to full __0 px
4211 # * border draw a 1px border around the image
4212 # vertical-align values (no % or length right now):
4213 # * baseline
4214 # * sub
4215 # * super
4216 # * top
4217 # * text-top
4218 # * middle
4219 # * bottom
4220 # * text-bottom
4225 # Give extensions a chance to select the file revision for us
4227 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time, &$descQuery ) );
4231 }
4233 # Get parameter map
4239 # Process the input parameters
4250 // Special case; width and height come in one variable together
4253 # (bug 13500) In both cases (width/height and width only),
4254 # permit trailing "px" for backward compatibility.
4261 }
4265 }
4271 }
4275 # Validate handler parameter
4278 # Validate internal parameters
4281 /// @fixme - possibly check validity here?
4282 /// downstream behavior seems odd with missing manual thumbs.
4286 // Most other things appear to be empty or numeric...
4288 }
4289 }
4293 }
4294 }
4295 }
4298 }
4299 }
4301 # Process alignment parameters
4304 }
4307 }
4309 # Strip bad stuff out of the alt text
4310 # We can't just use replaceLinkHoldersText() here, because if this function
4311 # is called from replaceInternalLinks2(), mLinkHolders won't be up to date.
4316 }
4318 # make sure there are no placeholders in thumbnail attributes
4319 # that are later expanded to html- so expand them now and
4320 # remove the tags
4329 # Linker does the rest
4330 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'], $time, $descQuery );
4332 # Give the handler a chance to modify the parser object
4335 }
4338 }
4340 /**
4341 * Set a flag in the output object indicating that the content is dynamic and
4342 * shouldn't be cached.
4343 */
4347 }
4349 /**#@+
4350 * Callback from the Sanitizer for expanding items found in HTML attribute
4351 * values, so they can be safely tested and escaped.
4352 * @param string $text
4353 * @param PPFrame $frame
4354 * @return string
4355 * @private
4356 */
4361 }
4363 /**#@-*/
4365 /**#@+
4366 * Accessor/mutator
4367 */
4371 /**#@-*/
4373 /**#@+
4374 * Accessor
4375 */
4376 function getTags() { return array_merge( array_keys($this->mTransparentTagHooks), array_keys( $this->mTagHooks ) ); }
4377 /**#@-*/
4380 /**
4381 * Break wikitext input into sections, and either pull or replace
4382 * some particular section's text.
4383 *
4384 * External callers should use the getSection and replaceSection methods.
4385 *
4386 * @param string $text Page wikitext
4387 * @param string $section A section identifier string of the form:
4388 * <flag1> - <flag2> - ... - <section number>
4389 *
4390 * Currently the only recognised flag is "T", which means the target section number
4391 * was derived during a template inclusion parse, in other words this is a template
4392 * section edit link. If no flags are given, it was an ordinary section edit link.
4393 * This flag is required to avoid a section numbering mismatch when a section is
4394 * enclosed by <includeonly> (bug 6563).
4395 *
4396 * The section number 0 pulls the text before the first heading; other numbers will
4397 * pull the given section along with its lower-level subsections. If the section is
4398 * not found, $mode=get will return $newtext, and $mode=replace will return $text.
4399 *
4400 * @param string $mode One of "get" or "replace"
4401 * @param string $newText Replacement text for section data.
4402 * @return string for "get", the extracted section text.
4403 * for "replace", the whole page with the section replaced.
4404 */
4414 // Process section extraction flags
4421 }
4422 }
4423 // Preprocess the text
4426 // <h> nodes indicate section breaks
4427 // They can only occur at the top level, so we can find them by iterating the root's children
4430 // Find the target section
4432 // Section zero doesn't nest, level=big
4441 }
4442 }
4445 }
4447 }
4448 }
4451 // Not found
4456 }
4457 }
4459 // Find the end of the section, including nested sections
4466 }
4467 }
4470 }
4474 // Write out the remainder (in replace mode only)
4476 // Output the replacement text
4477 // Add two newlines on -- trailing whitespace in $newText is conventionally
4478 // stripped by the editor, so we need both newlines to restore the paragraph gap
4483 }
4484 }
4487 // Re-insert stripped tags
4489 }
4492 }
4494 /**
4495 * This function returns the text of a section, specified by a number ($section).
4496 * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4497 * the first section before any such heading (section 0).
4498 *
4499 * If a section contains subsections, these are also returned.
4500 *
4501 * @param string $text text to look in
4502 * @param string $section section identifier
4503 * @param string $deftext default to return if section is not found
4504 * @return string text of the requested section
4505 */
4508 }
4512 }
4514 /**
4515 * Get the timestamp associated with the current revision, adjusted for
4516 * the default server-local timestamp
4517 */
4526 // Normalize timestamp to internal MW format for timezone processing.
4527 // This has the added side-effect of replacing a null value with
4528 // the current time, which gives us more sensible behavior for
4529 // previews.
4532 // The cryptic '' timezone parameter tells to use the site-default
4533 // timezone offset instead of the user settings.
4534 //
4535 // Since this value will be saved into the parser cache, served
4536 // to other users, and potentially even used inside links and such,
4537 // it needs to be consistent for all visitors.
4541 }
4543 }
4545 /**
4546 * Mutator for $mDefaultSort
4547 *
4548 * @param $sort New value
4549 */
4552 }
4554 /**
4555 * Accessor for $mDefaultSort
4556 * Will use the title/prefixed title if none is set
4557 *
4558 * @return string
4559 */
4569 }
4570 }
4572 /**
4573 * Try to guess the section anchor name based on a wikitext fragment
4574 * presumably extracted from a heading, for example "Header" from
4575 * "== Header ==".
4576 */
4578 # Strip out wikitext links(they break the anchor)
4581 # strip out HTML
4588 );
4593 }
4595 /**
4596 * Strips a text string of wikitext for use in a section anchor
4597 *
4598 * Accepts a text string and then removes all wikitext from the
4599 * string and leaves only the resultant text (i.e. the result of
4600 * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
4601 * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
4602 * to create valid section anchors by mimicing the output of the
4603 * parser when headings are parsed.
4604 *
4605 * @param $text string Text string to be stripped of wikitext
4606 * for use in a Section anchor
4607 * @return Filtered text string
4608 */
4610 # Strip internal link markup
4614 # Strip external link markup (FIXME: Not Tolerant to blank link text
4615 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
4616 # on how many empty links there are on the page - need to figure that out.
4619 # Parse wikitext quotes (italics & bold)
4622 # Strip HTML tags
4625 }
4629 }
4631 /**
4632 * strip/replaceVariables/unstrip for preprocessor regression testing
4633 */
4638 }
4646 }
4652 }
4654 }
4659 }
4661 }
4681 }
4682 }
4683 }
4685 }
4686 }
4688 /**
4689 * @todo document, briefly.
4690 * @ingroup Parser
4691 */
4698 }
4708 }
4718 }
4729 }
4730 }
4732 /**
4733 * @todo document, briefly.
4734 * @ingroup Parser
4735 */
4744 }
4745 }
4746 }