| blob | 30a80b4f249ebec6516350de99478781a9d52889 |
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:
99 # Cleared with clearState():
108 # Temporary
109 # These are variables reset at least once per parse regardless of $clearState
118 /**#@-*/
120 /**
121 * Constructor
122 *
123 * @public
124 */
140 // PECL extension that conflicts with the core DOM extension (bug 13770)
147 }
150 }
152 /**
153 * Reduce memory usage to reduce the impact of circular references
154 */
158 }
161 }
162 }
164 /**
165 * Do various kinds of initialisation on the first call of the parser
166 */
170 }
181 }
183 /**
184 * Clear Parser state
185 *
186 * @private
187 */
192 }
206 /**
207 * Prefix for temporary replacement strings for the multipass parser.
208 * \x07 should never appear in input as it's disallowed in XML.
209 * Using it at the front also gives us a little extra robustness
210 * since it shouldn't match when butted up against identifier-like
211 * string constructs.
212 *
213 * Must not consist of all title characters, or else it will change
214 * the behaviour of <nowiki> in a link.
215 */
216 #$this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
217 # Changed to \x7f to allow XML double-parsing -- TS
221 # Clear these on every parse, bug 4549
229 );
236 # Fix cloning
239 }
243 }
247 // Shortcut alias
252 );
253 }
255 /**
256 * Set the context title
257 */
261 }
263 # Strip the fragment to avoid various odd effects
268 }
269 }
271 /**
272 * Accessor for mUniqPrefix.
273 *
274 * @public
275 */
278 // @fixme this is probably *horribly wrong*
279 // LanguageConverter seems to want $wgParser's uniqPrefix, however
280 // if this is called for a parser cache hit, the parser may not
281 // have ever been initialized in the first place.
282 // Not really sure what the heck is supposed to be going on here.
284 //throw new MWException( "Accessing uninitialized mUniqPrefix" );
285 }
287 }
289 /**
290 * Convert wikitext to HTML
291 * Do not call this function recursively.
292 *
293 * @param $text String: text we want to parse
294 * @param $title A title object
295 * @param $options ParserOptions
296 * @param $linestart boolean
297 * @param $clearState boolean
298 * @param $revid Int: number to pass in {{REVISIONID}}
299 * @return ParserOutput a ParserOutput
300 */
301 public function parse( $text, Title $title, ParserOptions $options, $linestart = true, $clearState = true, $revid = null ) {
302 /**
303 * First pass--just handle <nowiki> sections, pass the rest off
304 * to internalParse() which does all the real work.
305 */
314 }
323 }
326 # No more strip!
331 # Clean up special characters, only run once, next-to-last before doBlockLevels
333 # french spaces, last one Guillemet-left
334 # only if there is something before the space
336 # french spaces, Guillemet-right
339 );
346 # the position of the parserConvert() call should not be changed. it
347 # assumes that the links are all replaced and the only thing left
348 # is the <nowiki> mark.
349 # Side-effects: this calls $this->mOutput->setTitleText()
356 //!JF Move to its own function
371 }
373 }
381 # attempt to sanitize at least some nesting problems
382 # (bug #2702 and quite a few others)
384 # ''Something [http://www.cool.com cool''] -->
385 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
388 # fix up an anchor inside another anchor, only
389 # at least for a single single nested link (bug 3695)
392 # fix div inside inline elements- doBlockLevels won't wrap a line which
393 # contains a div, so fix it up here; replace
394 # div with escaped text
397 # remove empty italic or bold tag pairs, some
398 # introduced by rules above
400 );
406 }
409 $this->limitationWarn( 'expensive-parserfunction', $this->mExpensiveFunctionCount, $wgExpensiveParserFunctionLimit );
410 }
414 # Information on include size limits, for the benefit of users who try to skirt them
418 $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/$wgExpensiveParserFunctionLimit\n";
427 }
435 }
437 /**
438 * Recursive parser entry point that can be called from an extension tag
439 * hook.
440 *
441 * If $frame is not provided, then template variables (e.g., {{{1}}}) within $text are not expanded
442 *
443 * @param $text String: text extension wants to have parsed
444 * @param PPFrame $frame: The frame to use for expanding any template variables
445 */
453 }
455 /**
456 * Expand templates and variables in the text, producing valid, static wikitext.
457 * Also removes comments.
458 */
467 }
474 }
476 /**
477 * Get a random string
478 *
479 * @private
480 * @static
481 */
484 }
500 }
501 }
503 /**
504 * Get a preprocessor object
505 */
510 }
512 }
514 /**
515 * Replaces all occurrences of HTML-style comments and the given tags
516 * in the text with a random marker and returns the next text. The output
517 * parameter $matches will be an associative array filled with data in
518 * the form:
519 * 'UNIQ-xxxxx' => array(
520 * 'element',
521 * 'tag content',
522 * array( 'param' => 'x' ),
523 * '<element param="x">tag content</element>' ) )
524 *
525 * @param $elements list of element names. Comments are always extracted.
526 * @param $text Source text string.
527 * @param $uniq_prefix
528 *
529 * @public
530 * @static
531 */
545 }
547 // comment
553 // tag
558 }
564 // Empty element tag, <tag />
573 }
577 # No end tag -- let it run out to the end of the text.
583 }
584 }
590 }
592 }
594 /**
595 * Get a list of strippable XML-like elements
596 */
602 }
605 }
607 }
609 /**
610 * @deprecated use replaceVariables
611 */
614 }
616 /**
617 * Restores pre, math, and other extensions removed by strip()
618 *
619 * always call unstripNoWiki() after this one
620 * @private
621 * @deprecated use $this->mStripState->unstrip()
622 */
625 }
627 /**
628 * Always call this after unstrip() to preserve the order
629 *
630 * @private
631 * @deprecated use $this->mStripState->unstrip()
632 */
635 }
637 /**
638 * @deprecated use $this->mStripState->unstripBoth()
639 */
642 }
644 /**
645 * Add an item to the strip state
646 * Returns the unique tag which must be inserted into the stripped text
647 * The tag will be replaced with the original text in unstrip()
648 *
649 * @private
650 */
656 }
658 /**
659 * Interface with html tidy
660 * @deprecated Use MWTidy::tidy()
661 */
665 }
667 /**
668 * parse the wiki syntax used to render tables
669 *
670 * @private
671 */
690 }
695 // First check if we are starting a new table
708 // Don't do any of the following
712 // We are ending a table
718 }
722 }
726 }
730 // Now we have a table row
733 // Whats after the tag is now only attributes
746 }
750 }
756 }
757 else if ( $first_character === '|' || $first_character === '!' || substr ( $line , 0 , 2 ) === '|+' ) {
758 // This might be cell elements, td, th or captions
762 }
768 }
770 // Split up multiple cells on the same line.
771 // FIXME : This can result in improper nesting of tags processed
772 // by earlier parser steps, but should avoid splitting up eg
773 // attribute values containing literal "||".
778 // Loop through each table cell
780 {
783 {
787 }
792 }
798 }
808 }
812 // A cell could contain both parameters and data
815 // Bug 553: Note that a '|' inside an invalid link should not
816 // be mistaken as delimiting cell parameters
825 }
829 }
830 }
832 }
834 // Closing open td, tr && table
836 {
839 }
842 }
845 }
848 }
850 // Remove trailing line-ending (b/c)
853 }
855 // special case: don't return empty table
858 }
863 }
865 /**
866 * Helper function for parse() that transforms wiki markup into
867 * HTML. Only called for $mOutputType == self::OT_HTML.
868 *
869 * @private
870 */
876 # Hook to suspend the parser in this state
877 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
880 }
882 // if $frame is provided, then use $frame for replacing any variables
884 // use frame depth to infer how include/noinclude tags should be handled
885 // depth=0 means this is the top-level document; otherwise it's an included document
888 else
892 }
893 // if $frame is not provided, then use old-style replaceVariables
896 }
898 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
901 // Tables need to come after variable replacement for things to work
902 // properly; putting them before other transformations should keep
903 // exciting things like link expansions from showing up in surprising
904 // places.
914 }
919 # replaceInternalLinks may sometimes leave behind
920 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
928 }
930 /**
931 * Replace special strings like "ISBN xxx" and "RFC xxx" with
932 * magic external links.
933 *
934 * DML
935 * @private
936 */
942 '!(?: # Start cases
943 (<a.*?</a>) | # m[1]: Skip link text
946 (?:RFC|PMID)\s+([0-9]+) | # m[4]: RFC or PMID, capture number
948 (?: 97[89] [\ \-]? )? # optional 13-digit ISBN prefix
949 (?: [0-9] [\ \-]? ){9} # 9 digits with opt. delimiters
950 [0-9Xx] # check digit
955 }
959 # Skip anchor
962 # Skip HTML element
965 # Free external link
968 # RFC or PMID
983 }
989 # ISBN
995 ));
1002 }
1003 }
1005 /**
1006 * Make a free external link, given a user-supplied URL
1007 * @return HTML
1008 * @private
1009 */
1017 # The characters '<' and '>' (which were escaped by
1018 # removeHTMLtags()) should not be included in
1019 # URLs, per RFC 2396.
1024 }
1026 # Move trailing punctuation to $trail
1028 # If there is no left bracket, then consider right brackets fair game too
1031 }
1037 }
1041 # Is this an external image?
1044 # Not an image, make a link
1047 # Register it in the output object...
1048 # Replace unnecessary URL escape codes with their equivalent characters
1051 }
1054 }
1057 /**
1058 * Parse headers and return html
1059 *
1060 * @private
1061 */
1068 }
1071 }
1073 /**
1074 * Replace single quotes with HTML markup
1075 * @private
1076 * @return string the altered text
1077 */
1084 }
1088 }
1090 /**
1091 * Helper function for doAllQuotes()
1092 */
1097 else
1098 {
1099 # First, do some preliminary work. This may shift some apostrophes from
1100 # being mark-up to being text. It also counts the number of occurrences
1101 # of bold and italics mark-ups.
1106 {
1108 {
1109 # If there are ever four apostrophes, assume the first is supposed to
1110 # be text, and the remaining three constitute mark-up for bold text.
1112 {
1115 }
1116 # If there are more than 5 apostrophes in a row, assume they're all
1117 # text except for the last 5.
1119 {
1122 }
1123 # Count the number of occurrences of bold and italics mark-ups.
1124 # We are not counting sequences of five apostrophes.
1128 }
1130 }
1132 # If there is an odd number of both bold and italics, it is likely
1133 # that one of the bold ones was meant to be an apostrophe followed
1134 # by italics. Which one we cannot know for certain, but it is more
1135 # likely to be one that has a single-letter word before it.
1137 {
1143 {
1145 {
1154 }
1155 }
1157 }
1159 # If there is a single-letter word, use it!
1161 {
1164 }
1165 # If not, but there's a multi-letter word, use that one.
1167 {
1170 }
1171 # ... otherwise use the first one that has neither.
1172 # (notice that it is possible for all three to be -1 if, for example,
1173 # there is only one pentuple-apostrophe in the line)
1175 {
1178 }
1179 }
1181 # Now let's actually convert our apostrophic mush to HTML!
1187 {
1189 {
1192 else
1194 }
1195 else
1196 {
1198 {
1209 }
1211 {
1222 }
1224 {
1237 }
1238 }
1240 }
1241 # Now close all remaining tags. Notice that the order is important.
1248 # There might be lonely ''''', so make sure we have a buffer
1252 }
1253 }
1255 /**
1256 * Replace external links (REL)
1257 *
1258 * Note: this is all very hackish and the order of execution matters a lot.
1259 * Make sure to run maintenance/parserTests.php if you change this code.
1260 *
1261 * @private
1262 */
1279 # The characters '<' and '>' (which were escaped by
1280 # removeHTMLtags()) should not be included in
1281 # URLs, per RFC 2396.
1286 }
1288 # If the link text is an image URL, replace it with an <img> tag
1289 # This happened by accident in the original parser, but some people used it extensively
1293 }
1297 # Set linktype for CSS - if URL==text, link is essentially free
1300 # No link text, e.g. [http://domain.tld/some.link]
1302 # Autonumber if allowed. See bug #5918
1308 # Otherwise just use the URL
1311 }
1313 # Have link text, e.g. [http://domain.tld/some.link text]s
1314 # Check for trail
1316 }
1322 # Use the encoded URL
1323 # This means that users can paste URLs directly into the text
1324 # Funny characters like ö aren't valid in URLs anyway
1325 # This was changed in August 2004
1329 # Register link in the output object.
1330 # Replace unnecessary URL escape codes with the referenced character
1331 # This prevents spammers from hiding links from the filters
1334 }
1338 }
1340 /**
1341 * Get an associative array of additional HTML attributes appropriate for a
1342 * particular external link. This currently may include rel => nofollow
1343 * (depending on configuration, namespace, and the URL's domain) and/or a
1344 * target attribute (depending on configuration).
1345 *
1346 * @param string $url Optional URL, to extract the domain from for rel =>
1347 * nofollow if appropriate
1348 * @return array Associative array of HTML attributes
1349 */
1366 }
1367 }
1368 }
1369 }
1370 }
1373 }
1375 }
1378 /**
1379 * Replace unusual URL escape codes with their equivalent characters
1380 * @param string
1381 * @return string
1382 * @static
1383 * @todo This can merge genuinely required bits in the path or query string,
1384 * breaking legit URLs. A proper fix would treat the various parts of
1385 * the URL differently; as a workaround, just use the output for
1386 * statistical records, not for actual linking/output.
1387 */
1391 }
1393 /**
1394 * Callback function used in replaceUnusualEscapes().
1395 * Replaces unusual URL escape codes with their equivalent character
1396 * @static
1397 * @private
1398 */
1402 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1404 // No, shouldn't be escaped
1407 // Yes, leave it escaped
1409 }
1410 }
1412 /**
1413 * make an image if it's allowed, either through the global
1414 * option, through the exception, or through the on-wiki whitelist
1415 * @private
1416 */
1422 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1429 }
1430 }
1435 }
1439 # Image found
1441 }
1442 }
1447 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1451 # Image matches a whitelist entry
1454 }
1455 }
1456 }
1458 }
1460 /**
1461 * Process [[ ]] wikilinks
1462 * @return processed text
1463 *
1464 * @private
1465 */
1469 }
1471 /**
1472 * Process [[ ]] wikilinks (RIL)
1473 * @return LinkHolderArray
1474 *
1475 * @private
1476 */
1484 # the % is needed to support urlencoded titles as well
1487 # Match a link having the form [[namespace:link|alternate]]trail
1489 # Match cases where there is no "]]", which might still be images
1491 }
1496 #split the entire text string on occurences of [[
1498 #get the first element (all text up to first [[), and remove the space we added
1507 # Match the end of a line for a word that's not followed by whitespace,
1508 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1510 }
1516 }
1525 }
1528 }
1534 }
1538 # Loop for each link
1540 # Check for excessive memory usage
1542 # Too big
1543 # Do the existence check, replace the link holders and clear the array
1546 }
1555 }
1556 # first link
1560 }
1562 }
1569 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1570 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1571 # the real problem is with the $e1 regex
1572 # See bug 1300.
1573 #
1574 # Still some problems for cases where the ] is meant to be outside punctuation,
1575 # and no image is in sight. See bug 2095.
1576 #
1580 )
1581 {
1584 }
1585 # fix up urlencoded title texts
1587 # Should anchors '#' also be rejected?
1589 }
1591 } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1596 }
1602 }
1606 # Don't allow internal links to pages containing
1607 # PROTO: where PROTO is a valid URL protocol; these
1608 # should be external links.
1613 }
1615 # Make subpage if necessary
1620 }
1624 # Strip off leading ':'
1626 }
1635 }
1646 #look at the next 'line' to see if we can close it there
1651 }
1654 # the first ]] closes the inner link, the second the image
1660 #if there's exactly one ]] that's fine, we'll keep looking
1663 #if $next_line is invalid too, we need look no further
1666 }
1667 }
1669 # we couldn't find the end of this imageLink, so output it raw
1670 #but don't ignore what might be perfectly normal links in the text we've examined
1673 # note: no $trail, because without an end, there *is* no trail
1676 }
1679 # note: no $trail, because without an end, there *is* no trail
1682 }
1684 }
1689 # Link not escaped by : , create the various objects
1692 # Interwikis
1694 if ( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1700 }
1707 # if no parameters were passed, $text
1708 # becomes something like "File:Foo.png",
1709 # which we don't want to pass on to the
1710 # image generator
1713 # recursively parse links inside the image caption
1714 # actually, this will parse them in any other parameters, too,
1715 # but it might be hard to fix that, and it doesn't matter ATM
1718 }
1719 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1721 }
1726 }
1736 }
1742 /**
1743 * Strip the whitespace Category links produce, see bug 87
1744 * @todo We might want to use trim($tmp, "\n") here.
1745 */
1750 }
1751 }
1753 # Self-link checking
1758 }
1759 }
1761 # NS_MEDIA is a pseudo-namespace for linking directly to a file
1762 # FIXME: Should do batch file existence checks, see comment below
1765 # Give extensions a chance to select the file revision for us
1772 }
1773 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1778 }
1781 # Some titles, such as valid special pages or files in foreign repos, should
1782 # be shown as bluelinks even though they're not included in the page table
1783 #
1784 # FIXME: isAlwaysKnown() can be expensive for file links; we should really do
1785 # batch file existence checks for NS_FILE and NS_MEDIA
1790 # Links will be added to the output link list after checking
1792 }
1794 }
1797 }
1799 /**
1800 * Make a link placeholder. The text returned can be later resolved to a real link with
1801 * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1802 * parsing of interwiki links, and secondly to allow all existence checks and
1803 * article length checks (for stub links) to be bundled into a single query.
1804 *
1805 * @deprecated
1806 */
1809 }
1811 /**
1812 * Render a forced-blue link inline; protect against double expansion of
1813 * URLs if we're in a mode that prepends full URL prefixes to internal links.
1814 * Since this little disaster has to split off the trail text to avoid
1815 * breaking URLs in the following text without breaking trails on the
1816 * wiki links, it's been made into a horrible function.
1817 *
1818 * @param Title $nt
1819 * @param string $text
1820 * @param string $query
1821 * @param string $trail
1822 * @param string $prefix
1823 * @return string HTML-wikitext mix oh yuck
1824 */
1828 // FIXME: use link() instead of deprecated makeKnownLinkObj()
1831 }
1833 /**
1834 * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1835 * going to go through further parsing steps before inline URL expansion.
1836 *
1837 * Not needed quite as much as it used to be since free links are a bit
1838 * more sensible these days. But bracketed links are still an issue.
1839 *
1840 * @param string more-or-less HTML
1841 * @return string less-or-more HTML with NOPARSE bits
1842 */
1846 }
1848 /**
1849 * Return true if subpage links should be expanded on this page.
1850 * @return bool
1851 */
1853 # Some namespaces don't allow subpages
1855 }
1857 /**
1858 * Handle link to subpage if necessary
1859 * @param string $target the source of the link
1860 * @param string &$text the link text, modified as necessary
1861 * @return string the full name of the link
1862 * @private
1863 */
1866 }
1868 /**#@+
1869 * Used by doBlockLevels()
1870 * @private
1871 */
1876 }
1880 }
1881 # getCommon() returns the length of the longest common substring
1882 # of both arguments, starting at the beginning of both.
1883 #
1891 }
1893 }
1894 # These next three functions open, continue, and close the list
1895 # element appropriate to the prefix character passed into them.
1896 #
1906 }
1910 }
1923 }
1924 }
1926 }
1937 }
1938 }
1941 }
1942 /**#@-*/
1944 /**
1945 * Make lists from lines starting with ':', '*', '#', etc. (DBL)
1946 *
1947 * @param $linestart bool whether or not this is at the start of a line.
1948 * @private
1949 * @return string the lists rendered as HTML
1950 */
1954 # Parsing through the text line by line. The main thing
1955 # happening here is handling of block-level elements p, pre,
1956 # and making lists from lines starting with * # : etc.
1957 #
1966 # Fix up $linestart
1971 }
1972 // * = ul
1973 // # = ol
1974 // ; = dt
1975 // : = dd
1980 // If not in a <pre> element, scan for and figure out what prefixes are there.
1982 # Multiple prefixes may abut each other for nested lists.
1986 # eh?
1987 // ; and : are both from definition-lists, so they're equivalent
1988 // for the purposes of determining whether or not we need to open/close
1989 // elements.
1994 # Don't interpret any other prefixes in preformatted text
1998 }
2000 # List generation
2002 # Same as the last item, so no need to deal with nesting or opening stuff
2007 # The one nasty exception: definition lists work like this:
2008 # ; title : definition text
2009 # So we check for : in the remainder text to split up the
2010 # title and definition, without b0rking links.
2015 }
2016 }
2018 // We need to open or close prefixes, or both.
2020 # Either open or close a level...
2024 // Close all the prefixes which aren't shared.
2028 }
2030 // Continue the current prefix if appropriate.
2033 }
2035 // Open prefixes where appropriate.
2041 # FIXME: This is dupe of code above
2045 }
2046 }
2048 }
2050 }
2052 // If we have no prefixes, go to paragraph mode.
2055 # No prefix (not in list)--go to paragraph mode
2056 // XXX: use a stack for nestable elements like span, table and div
2057 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2060 '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2063 #Â TODO bug 5718: paragraph closed
2067 }
2072 }
2075 // pre
2080 }
2083 // paragraph
2096 }
2097 }
2106 }
2107 }
2108 }
2109 }
2111 }
2112 // somewhere above we forget to get out of pre block (bug 785)
2115 }
2118 }
2119 }
2123 }
2127 }
2131 }
2133 /**
2134 * Split up a string on ':', ignoring any occurences inside tags
2135 * to prevent illegal overlapping.
2136 * @param string $str the string to split
2137 * @param string &$before set to everything before the ':'
2138 * @param string &$after set to everything after the ':'
2139 * return string the position of the ':', or false if none found
2140 */
2146 // Nothing to find!
2149 }
2153 // Easy; no tag nesting to worry about
2158 }
2160 // Ugly state machine to walk through avoiding tags.
2168 // (Using the number is a performance hack for common cases)
2172 // Could be either a <start> tag or an </end> tag
2177 // We found it!
2182 }
2183 // Embedded in a tag; don't break it.
2186 // Skip ahead looking for something interesting
2189 // Nothing else interesting
2192 }
2196 // We found it!
2201 }
2202 }
2204 // Nothing else interesting to find; abort!
2205 // We're nested, but there's no close tags left. Abort!
2207 }
2208 // Skip ahead to next tag start
2211 }
2214 // In a <tag>
2221 // Slash may be followed by >?
2225 // ignore
2226 }
2237 // Illegal early close? This shouldn't happen D:
2242 }
2245 // In a </tag>
2252 }
2254 }
2258 // Yes, a self-closed tag <blah/>
2261 // Probably we're jumping the gun, and this is an attribute
2263 }
2268 }
2275 }
2282 }
2286 }
2287 }
2291 }
2294 }
2296 /**
2297 * Return value of a magic variable (like PAGENAME)
2298 *
2299 * @private
2300 */
2305 /**
2306 * Some of these require message or data lookups and can be
2307 * expensive to check many times.
2308 */
2312 }
2313 }
2318 # Use the time zone
2323 }
2338 }
2414 }
2422 }
2433 // Let the edit saving system know we should parse the page
2434 // *after* a revision ID has been assigned.
2440 // Let the edit saving system know we should parse the page
2441 // *after* a revision ID has been assigned. This is for null edits.
2447 // Let the edit saving system know we should parse the page
2448 // *after* a revision ID has been assigned. This is for null edits.
2454 // Let the edit saving system know we should parse the page
2455 // *after* a revision ID has been assigned. This is for null edits.
2461 // Let the edit saving system know we should parse the page
2462 // *after* a revision ID has been assigned. This is for null edits.
2468 // Let the edit saving system know we should parse the page
2469 // *after* a revision ID has been assigned. This is for null edits.
2475 // Let the edit saving system know we should parse the page
2476 // *after* a revision ID has been assigned. This is for null edits.
2512 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2513 // int to remove the padding
2532 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2533 // int to remove the padding
2587 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret, &$frame ) ) )
2589 else
2591 }
2597 }
2599 /**
2600 * initialise the magic variables (like CURRENTMONTHNAME)
2601 *
2602 * @private
2603 */
2610 }
2612 /**
2613 * Preprocess some wikitext and return the document tree.
2614 * This is the ghost of replace_variables().
2615 *
2616 * @param string $text The text to parse
2617 * @param integer flags Bitwise combination of:
2618 * self::PTD_FOR_INCLUSION Handle <noinclude>/<includeonly> as if the text is being
2619 * included. Default is to assume a direct page view.
2620 *
2621 * The generated DOM tree must depend only on the input text and the flags.
2622 * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2623 *
2624 * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2625 * change in the DOM tree for a given text, must be passed through the section identifier
2626 * in the section edit link and thus back to extractSections().
2627 *
2628 * The output of this function is currently only cached in process memory, but a persistent
2629 * cache may be implemented at a later date which takes further advantage of these strict
2630 * dependency requirements.
2631 *
2632 * @private
2633 */
2637 }
2639 /*
2640 * Return a three-element array: leading whitespace, string contents, trailing whitespace
2641 */
2651 }
2653 }
2655 /**
2656 * Replace magic variables, templates, and template arguments
2657 * with the appropriate text. Templates are substituted recursively,
2658 * taking care to avoid infinite loops.
2659 *
2660 * Note that the substitution depends on value of $mOutputType:
2661 * self::OT_WIKI: only {{subst:}} templates
2662 * self::OT_PREPROCESS: templates but not extension tags
2663 * self::OT_HTML: all templates and extension tags
2664 *
2665 * @param string $tex The text to transform
2666 * @param PPFrame $frame Object describing the arguments passed to the template.
2667 * Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2668 * Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2669 * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2670 * @private
2671 */
2673 # Is there any text? Also, Prevent too big inclusions!
2676 }
2682 wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
2684 }
2692 }
2694 /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2707 }
2710 }
2711 }
2712 }
2715 }
2717 /**
2718 * Warn the user when a parser limitation is reached
2719 * Will warn at most once the user per limitation type
2720 *
2721 * @param string $limitationType, should be one of:
2722 * 'expensive-parserfunction' (corresponding messages: 'expensive-parserfunction-warning', 'expensive-parserfunction-category')
2723 * 'post-expand-template-argument' (corresponding messages: 'post-expand-template-argument-warning', 'post-expand-template-argument-category')
2724 * 'post-expand-template-inclusion' (corresponding messages: 'post-expand-template-inclusion-warning', 'post-expand-template-inclusion-category')
2725 * @params int $current, $max When an explicit limit has been
2726 * exceeded, provide the values (optional)
2727 */
2729 //does no harm if $current and $max are present but are unnecessary for the message
2730 $warning = wfMsgExt( "$limitationType-warning", array( 'parsemag', 'escape' ), $current, $max );
2733 }
2735 /**
2736 * Return the text of a template, after recursively
2737 * replacing any variables or templates within the template.
2738 *
2739 * @param array $piece The parts of the template
2740 * $piece['title']: the title, i.e. the part before the |
2741 * $piece['parts']: the parameter array
2742 * $piece['lineStart']: whether the brace was at the start of a line
2743 * @param PPFrame The current frame, contains template arguments
2744 * @return string the text of the template
2745 * @private
2746 */
2752 # Flags
2760 # Title object, where $text came from
2763 # $part1 is the bit before the first |, and must contain only title characters.
2764 # Various prefixes will be stripped from it later.
2769 # Original title text preserved for various purposes
2772 # $args is a list of argument nodes, starting from index 0, not including $part1
2776 # SUBST
2781 # One of two possibilities is true:
2782 # 1) Found SUBST but not in the PST phase
2783 # 2) Didn't find SUBST and in the PST phase
2784 # In either case, return without further processing
2788 }
2789 }
2791 # Variables
2799 }
2800 }
2802 # MSG, MSGNW and RAW
2804 # Check for MSGNW:
2809 # Remove obsolete MSG:
2812 }
2814 # Check for RAW:
2818 }
2819 }
2822 # Parser functions
2828 # Case sensitive functions
2833 # Case insensitive functions
2839 }
2840 }
2846 # Add a frame parameter, and pass the arguments as an array
2851 }
2854 # Convert arguments to plain text
2857 }
2859 }
2861 # Workaround for PHP bug 35229 and similar
2866 }
2876 }
2878 // Extract flags into the local scope
2879 // This allows callers to set flags such as nowiki, found, etc.
2883 }
2887 }
2888 }
2889 }
2891 }
2893 # Finish mangling title and then check for loops.
2894 # Set $title to a Title object and $titleText to the PDBK
2897 # Split the title into page and subpage
2902 }
2906 # Check for language variants if the template is not found
2909 }
2910 # Do recursion depth check
2914 $text = '<span class="error">' . wfMsgForContent( 'parser-template-recursion-depth-warning', $limit ) . '</span>';
2915 }
2916 }
2917 }
2919 # Load from database
2923 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
2929 }
2930 } else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
2938 }
2939 }
2941 # If the title is valid but undisplayable, make a link to it
2945 }
2947 // Interwiki transclusion
2953 // Preprocess it like a template
2956 }
2958 }
2960 # Do infinite loop check
2961 # This has to be done after redirect resolution to avoid infinite loops via redirects
2964 $text = '<span class="error">' . wfMsgForContent( 'parser-template-loop-warning', $titleText ) . '</span>';
2966 }
2968 }
2970 # If we haven't found text to substitute by now, we're done
2971 # Recover the source wikitext and return it
2976 }
2978 # Expand DOM-style return values in a child frame
2980 # Clean up argument array
2986 # Expansion is eligible for the empty-frame cache
2992 }
2994 # Uncached expansion
2996 }
2997 }
3001 }
3003 # Replace raw HTML by a placeholder
3004 # Add a blank line preceding, to prevent it from mucking up
3005 # immediately preceding headings
3008 }
3009 # Escape nowiki-style return values
3012 }
3013 # Bug 529: if the template begins with a table or block-level
3014 # element, it should be treated as beginning a new line.
3015 # This behaviour is somewhat controversial.
3016 elseif ( is_string( $text ) && !$piece['lineStart'] && preg_match('/^(?:{\\||:|;|#|\*)/', $text)) /*}*/{
3018 }
3021 # Error, oversize inclusion
3023 $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3025 }
3031 }
3035 }
3037 /**
3038 * Get the semi-parsed DOM representation of a template with a given title,
3039 * and its redirect destination title. Cached.
3040 */
3049 }
3052 }
3054 // Cache miss, go to the database
3060 }
3068 }
3071 }
3073 /**
3074 * Fetch the unparsed text of a template and register a reference to it.
3075 */
3084 }
3085 }
3087 }
3092 }
3094 /**
3095 * Static function to get a template
3096 * Can be overridden via ParserOptions::setTemplateCallback().
3097 */
3103 // Loop to fetch the article, with up to 1 redirect
3105 # Give extensions a chance to select the revision instead
3116 }
3119 // If there is no current revision, there is no page
3123 }
3139 }
3142 }
3145 }
3146 // Redirect?
3149 }
3154 }
3156 /**
3157 * Transclude an interwiki link.
3158 */
3170 }
3180 }
3192 }
3195 /**
3196 * Triple brace replacement -- used for template arguments
3197 * @private
3198 */
3209 && (
3213 )
3214 ) {
3215 # No match in frame, use the supplied default
3217 }
3221 }
3224 # No match anywhere
3226 }
3229 }
3234 }
3238 }
3240 /**
3241 * Return the text to be used for a given extension tag.
3242 * This is the ghost of strip().
3243 *
3244 * @param array $params Associative array of parameters:
3245 * name PPNode for the tag name
3246 * attr PPNode for unparsed text where tag attributes are thought to be
3247 * attributes Optional associative array of parsed attributes
3248 * inner Contents of extension element
3249 * noClose Original text did not have a close tag
3250 * @param PPFrame $frame
3251 */
3259 $marker = "{$this->mUniqPrefix}-$name-" . sprintf('%08X', $this->mMarkerIndex++) . self::MARKER_SUFFIX;
3268 }
3276 }
3292 }
3293 /* else let a tag hook handle it (bug 21222) */
3296 # Workaround for PHP bug 35229 and similar
3299 }
3312 }
3313 }
3317 }
3322 }
3323 }
3329 }
3330 }
3338 }
3340 }
3342 /**
3343 * Increment an include size counter
3344 *
3345 * @param string $type The type of expansion
3346 * @param integer $size The size of the text
3347 * @return boolean False if this inclusion would take it over the maximum, true otherwise
3348 */
3355 }
3356 }
3358 /**
3359 * Increment the expensive function count
3360 *
3361 * @return boolean False if the limit has been exceeded
3362 */
3368 }
3370 }
3372 /**
3373 * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3374 * Fills $this->mDoubleUnderscores, returns the modified text
3375 */
3378 // The position of __TOC__ needs to be recorded
3384 // Set a placeholder. At the end we'll fill it in with the TOC.
3387 // Only keep the first one.
3389 }
3391 // Now match and remove the rest of them
3397 }
3400 }
3401 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3404 }
3405 # (bug 8068) Allow control over whether robots index a page.
3406 #
3407 # FIXME (bug 14899): __INDEX__ always overrides __NOINDEX__ here! This
3408 # is not desirable, the last one on the page should win.
3412 }
3416 }
3419 }
3421 /**
3422 * Add a tracking category, getting the title from a system message,
3423 * or print a debug message if the title is invalid.
3424 * @param $msg String message key
3425 * @return Bool whether the addition was successful
3426 */
3430 # Allow tracking categories to be disabled by setting them to "-"
3440 }
3441 }
3443 /**
3444 * This function accomplishes several tasks:
3445 * 1) Auto-number headings if that option is enabled
3446 * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3447 * 3) Add a Table of contents on the top for users who have enabled the option
3448 * 4) Auto-anchor headings
3449 *
3450 * It loops through all headlines, collects the necessary data, then splits up the
3451 * string and re-inserts the newly formatted headlines.
3452 *
3453 * @param string $text
3454 * @param string $origText Original, untouched wikitext
3455 * @param boolean $isMain
3456 * @private
3457 */
3464 // Do not call quickUserCan unless necessary
3467 }
3469 # Inhibit editsection links if requested in the page
3470 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) || $this->mOptions->getIsPrintable() ) {
3472 }
3474 # Get all headlines for numbering them and adding funky stuff like [edit]
3475 # links - this is for later, but we need the number of headlines right now
3477 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3479 # if there are fewer than 4 headlines in the article, do not show TOC
3480 # unless it's been explicitly enabled.
3484 # Allow user to stipulate that a page should have a "new section"
3485 # link added via __NEWSECTIONLINK__
3488 }
3490 # Allow user to remove the "new section"
3491 # link via __NONEWSECTIONLINK__
3494 }
3496 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3497 # override above conditions and always show TOC above first header
3501 }
3503 # We need this to perform operations on the HTML
3506 # headline counter
3510 # Ugh .. the TOC should have neat indentation levels which can be
3511 # passed to the skin functions. These are determined here
3543 }
3548 }
3552 # Increase TOC level
3559 }
3560 }
3562 # Decrease TOC level, find level to jump to
3566 # Found last matching level
3569 }
3571 # Found first matching level below current level
3574 }
3575 }
3579 # Unindent only if the previous toc level was shown :p
3584 }
3585 }
3586 }
3588 # No change in level, end TOC line
3591 }
3592 }
3596 # count number of headlines for each level
3603 }
3606 }
3607 }
3609 # The safe header is a version of the header text safe to use for links
3610 # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3613 # Remove link placeholders by the link text.
3614 # <!--LINK number-->
3615 # turns into
3616 # link text with suffix
3619 # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3623 $safeHeadline
3624 );
3627 # For the anchor, strip out HTML-y stuff period
3632 # Save headline for section edit hint before it's escaped
3640 # For reverse compatibility, provide an id that's
3641 # HTML4-compatible, like we used to.
3642 #
3643 # It may be worth noting, academically, that it's possible for
3644 # the legacy anchor to conflict with a non-legacy headline
3645 # anchor on the page. In this case likely the "correct" thing
3646 # would be to either drop the legacy anchors or make sure
3647 # they're numbered first. However, this would require people
3648 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
3649 # manually, so let's not bother worrying about it.
3655 # No reason to have both (in fact, we can't)
3659 # The legacy id is invalid XML. We used to allow this, but
3660 # there's no reason to do so anymore. Backward
3661 # compatibility will fail slightly in this case, but it's
3662 # no big deal.
3664 }
3665 }
3667 # HTML names must be case-insensitively unique (bug 10721). FIXME:
3668 # Does this apply to Unicode characters? Because we aren't
3669 # handling those here.
3675 }
3677 # count how many in assoc. array so we can track dupes in anchors
3682 }
3687 }
3689 # Don't number the heading if it is the only one (looks silly)
3691 # the two are different if the line contains a link
3693 }
3695 # Create the anchor for linking from the TOC to the section
3700 }
3703 }
3707 }
3709 # Add the section to the section tree
3710 # Find the DOM node for this header
3716 }
3720 }
3730 );
3732 # give headline the correct <h#> tag
3735 # Put a T flag in the section identifier, to indicate to extractSections()
3736 # that sections inside <includeonly> should be counted.
3740 }
3743 }
3749 }
3753 # Never ever show TOC if no headers
3756 }
3761 }
3764 }
3768 }
3770 # split up and insert constructed headlines
3777 # This is the [edit] link that appears for the top block of text when
3778 # section editing is enabled
3780 # Disabled because it broke block formatting
3781 # For example, a bullet point in the top line
3782 # $full .= $sk->editSectionLink(0);
3783 }
3786 # Top anchor now in skin
3788 }
3792 }
3794 }
3799 }
3800 }
3802 /**
3803 * Merge $tree2 into $tree1 by replacing the section with index
3804 * $section in $tree1 and its descendants with the sections in $tree2.
3805 * Note that in the returned section tree, only the 'index' and
3806 * 'byteoffset' fields are guaranteed to be correct.
3807 * @param $tree1 array Section tree from ParserOutput::getSectons()
3808 * @param $tree2 array Section tree
3809 * @param $section int Section index
3810 * @param $title Title Title both section trees come from
3811 * @param $len2 int Length of the original wikitext for $tree2
3812 * @return array Merged section tree
3813 */
3826 // We've skipped enough
3828 else
3830 }
3836 // Rewrite index, byteoffset and number
3841 }
3848 // We're at $section
3849 // Insert sections from $tree2 here
3851 // Rewrite the fields in $s2
3852 // before inserting it
3865 }
3866 // Skip all descendants of $section in $tree1
3869 }
3870 }
3872 }
3874 /**
3875 * Increment a section number. Helper function for mergeSectionTrees()
3876 * @param $number array Array representing a section number
3877 * @param $level int Current TOC level (depth)
3878 * @param $lastLevel int Level of previous TOC entry
3879 */
3890 }
3892 /**
3893 * Transform wiki markup when saving a page by doing \r\n -> \n
3894 * conversion, substitting signatures, {{subst:}} templates, etc.
3895 *
3896 * @param string $text the text to transform
3897 * @param Title &$title the Title object for the current article
3898 * @param User $user the User object describing the current user
3899 * @param ParserOptions $options parsing options
3900 * @param bool $clearState whether to clear the parser state first
3901 * @return string the altered wiki markup
3902 * @public
3903 */
3911 }
3915 );
3920 }
3922 /**
3923 * Pre-save transform helper function
3924 * @private
3925 */
3929 /* Note: This is the timestamp saved as hardcoded wikitext to
3930 * the database, we use $wgContLang here in order to give
3931 * everyone the same signature and use the default one rather
3932 * than the one selected in each user's preferences.
3933 *
3934 * (see also bug 12815)
3935 */
3945 /* Allow translation of timezones trough wiki. date() can return
3946 * whatever crap the system uses, localised or not, so we cannot
3947 * ship premade translations.
3948 */
3954 }
3958 # Variable replacement
3959 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3962 # Signatures
3968 ) );
3970 # Context links: [[|name]] and [[name (context)|]]
3971 #
3978 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/"; # [[ns:page (context), context|]]
3981 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
3993 # if there's no context, don't bother duplicating the title
3995 }
3997 # Trim trailing whitespace
4001 }
4003 /**
4004 * Fetch the user's signature text, if any, and normalize to
4005 * validated, ready-to-insert wikitext.
4006 * If you have pre-fetched the nickname or the fancySig option, you can
4007 * specify them here to save a database query.
4008 *
4009 * @param User $user
4010 * @return string
4011 */
4017 // If not given, retrieve from the user object.
4030 # Sig. might contain markup; validate this
4032 # Validated; clean up (if needed) and return it
4035 # Failed to validate; fall back to the default
4038 }
4039 }
4041 // Make sure nickname doesnt get a sig in a sig
4044 # If we're still here, make it a link to the user page
4051 }
4052 }
4054 /**
4055 * Check that the user's signature contains no bad XML
4056 *
4057 * @param string $text
4058 * @return mixed An expanded string, or false if invalid.
4059 */
4062 }
4064 /**
4065 * Clean up signature text
4066 *
4067 * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4068 * 2) Substitute all transclusions
4069 *
4070 * @param string $text
4071 * @param $parsing Whether we're cleaning (preferences save) or parsing
4072 * @return string Signature text
4073 */
4081 }
4083 # Option to disable this feature
4086 }
4088 # FIXME: regex doesn't respect extension tags or nowiki
4089 # => Move this logic to braceSubstitution()
4091 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4102 }
4105 }
4107 /**
4108 * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4109 * @param string $text
4110 * @return string Signature text with /~{3,5}/ removed
4111 */
4115 }
4117 /**
4118 * Set up some variables which are usually set up in parse()
4119 * so that an external function can call some class members with confidence
4120 * @public
4121 */
4128 }
4129 }
4131 /**
4132 * Wrapper for preprocess()
4133 *
4134 * @param string $text the text to preprocess
4135 * @param ParserOptions $options options
4136 * @return string
4137 * @public
4138 */
4143 # Guard against infinite recursion
4146 }
4155 }
4157 /**
4158 * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4159 * The callback should have the following form:
4160 * function myParserHook( $text, $params, &$parser ) { ... }
4161 *
4162 * Transform and return $text. Use $parser for any required context, e.g. use
4163 * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4164 *
4165 * @public
4166 *
4167 * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
4168 * @param mixed $callback The callback function (and object) to use for the tag
4169 *
4170 * @return The old value of the mTagHooks array associated with the hook
4171 */
4178 }
4181 }
4185 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4189 }
4191 /**
4192 * Remove all tag hooks
4193 */
4197 }
4199 /**
4200 * Create a function, e.g. {{sum:1|2|3}}
4201 * The callback function should have the form:
4202 * function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4203 *
4204 * Or with SFH_OBJECT_ARGS:
4205 * function myParserFunction( $parser, $frame, $args ) { ... }
4206 *
4207 * The callback may either return the text result of the function, or an array with the text
4208 * in element 0, and a number of flags in the other elements. The names of the flags are
4209 * specified in the keys. Valid flags are:
4210 * found The text returned is valid, stop processing the template. This
4211 * is on by default.
4212 * nowiki Wiki markup in the return value should be escaped
4213 * isHTML The returned text is HTML, armour it against wikitext transformation
4214 *
4215 * @public
4216 *
4217 * @param string $id The magic word ID
4218 * @param mixed $callback The callback function (and object) to use
4219 * @param integer $flags a combination of the following flags:
4220 * SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4221 *
4222 * SFH_OBJECT_ARGS Pass the template arguments as PPNode objects instead of text. This
4223 * allows for conditional expansion of the parse tree, allowing you to eliminate dead
4224 * branches and thus speed up parsing. It is also possible to analyse the parse tree of
4225 * the arguments, and to control the way they are expanded.
4226 *
4227 * The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4228 * arguments, for instance:
4229 * $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4230 *
4231 * For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4232 * future versions. Please call $frame->expand() on it anyway so that your code keeps
4233 * working if/when this is changed.
4234 *
4235 * If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4236 * expansion.
4237 *
4238 * Please read the documentation in includes/parser/Preprocessor.php for more information
4239 * about the methods available in PPFrame and PPNode.
4240 *
4241 * @return The old callback function for this name, if any
4242 */
4249 # Add to function cache
4258 # Case
4261 }
4262 # Add leading hash
4265 }
4266 # Remove trailing colon
4269 }
4271 }
4273 }
4275 /**
4276 * Get all registered function hook identifiers
4277 *
4278 * @return array
4279 */
4282 }
4284 /**
4285 * Create a tag function, e.g. <test>some stuff</test>.
4286 * Unlike tag hooks, tag functions are parsed at preprocessor level.
4287 * Unlike parser functions, their content is not preprocessed.
4288 */
4297 }
4300 }
4302 /**
4303 * FIXME: update documentation. makeLinkObj() is deprecated.
4304 * Replace <!--LINK--> link placeholders with actual links, in the buffer
4305 * Placeholders created in Skin::makeLinkObj()
4306 * Returns an array of link CSS classes, indexed by PDBK.
4307 */
4310 }
4312 /**
4313 * Replace <!--LINK--> link placeholders with plain text of links
4314 * (not HTML-formatted).
4315 * @param string $text
4316 * @return string
4317 */
4320 }
4322 /**
4323 * Tag hook handler for 'pre'.
4324 */
4326 // Backwards-compatibility hack
4333 }
4335 /**
4336 * Tag hook handler for 'a'. Renders a HTML <a> tag, allowing most attributes, filtering href against
4337 * allowed protocols and spam blacklist.
4338 **/
4342 }
4354 # Non-link <a> tag
4356 }
4362 }
4364 /**
4365 * Renders an image gallery from a text with one line per image.
4366 * text labels may be given by using |-style alternative text. E.g.
4367 * Image:one.jpg|The number "1"
4368 * Image:tree.jpg|A tree
4369 * given as text will return the HTML of a gallery with two images,
4370 * labeled 'The number "1"' and
4371 * 'A tree'.
4372 */
4389 }
4392 }
4395 }
4398 }
4404 # match lines like these:
4405 # Image:someimage.jpg|This is some image
4408 # Skip empty lines
4411 }
4418 # Bogus title. Ignore these so we don't bomb out later.
4420 }
4425 }
4431 # Only add real images (bug #5586)
4434 }
4435 }
4437 }
4444 }
4446 // Initialise static lists
4453 );
4461 }
4462 }
4463 }
4465 // Add handler params
4471 }
4472 }
4475 }
4476 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4477 }
4479 /**
4480 * Parse image options text and use it to make an image
4481 * @param Title $title
4482 * @param string $options
4483 * @param LinkHolderArray $holders
4484 */
4486 # Check if the options text is of the form "options|alt text"
4487 # Options are:
4488 # * thumbnail make a thumbnail with enlarge-icon and caption, alignment depends on lang
4489 # * left no resizing, just left align. label is used for alt= only
4490 # * right same, but right aligned
4491 # * none same, but not aligned
4492 # * ___px scale to ___ pixels width, no aligning. e.g. use in taxobox
4493 # * center center the image
4494 # * frame Keep original image size, no magnify-button.
4495 # * framed Same as "frame"
4496 # * frameless like 'thumb' but without a frame. Keeps user preferences for width
4497 # * upright reduce width for upright images, rounded to full __0 px
4498 # * border draw a 1px border around the image
4499 # * alt Text for HTML alt attribute (defaults to empty)
4500 # * link Set the target of the image link. Can be external, interwiki, or local
4501 # vertical-align values (no % or length right now):
4502 # * baseline
4503 # * sub
4504 # * super
4505 # * top
4506 # * text-top
4507 # * middle
4508 # * bottom
4509 # * text-bottom
4514 # Give extensions a chance to select the file revision for us
4516 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time, &$descQuery ) );
4520 }
4522 # Get the file
4525 # Get parameter map
4530 # Process the input parameters
4541 // Special case; width and height come in one variable together
4544 # (bug 13500) In both cases (width/height and width only),
4545 # permit trailing "px" for backward compatibility.
4552 }
4556 }
4562 }
4566 # Validate handler parameter
4569 # Validate internal parameters
4573 // @fixme - possibly check validity here for
4574 // manualthumb? downstream behavior seems odd with
4575 // missing manual thumbs.
4591 }
4599 }
4600 }
4603 // Most other things appear to be empty or numeric...
4605 }
4606 }
4610 }
4611 }
4612 }
4615 }
4616 }
4618 # Process alignment parameters
4621 }
4624 }
4628 # Will the image be presented in a frame, with the caption below?
4634 # In the old days, [[Image:Foo|text...]] would set alt text. Later it
4635 # came to also set the caption, ordinary text after the image -- which
4636 # makes no sense, because that just repeats the text multiple times in
4637 # screen readers. It *also* came to set the title attribute.
4638 #
4639 # Now that we have an alt attribute, we should not set the alt text to
4640 # equal the caption: that's worse than useless, it just repeats the
4641 # text. This is the framed/thumbnail case. If there's no caption, we
4642 # use the unnamed parameter for alt text as well, just for the time be-
4643 # ing, if the unnamed param is set and the alt param is not.
4644 #
4645 # For the future, we need to figure out if we want to tweak this more,
4646 # e.g., introducing a title= parameter for the title; ignoring the un-
4647 # named parameter entirely for images without a caption; adding an ex-
4648 # plicit caption= parameter and preserving the old magic unnamed para-
4649 # meter for BC; ...
4652 # No caption or alt text, add the filename as the alt text so
4653 # that screen readers at least get some description of the image
4655 }
4656 # Do not set $params['frame']['title'] because tooltips don't make sense
4657 # for framed images
4660 # No alt text, use the "caption" for the alt text
4664 # No caption, fall back to using the filename for the
4665 # alt text
4667 }
4668 }
4669 # Use the "caption" for the tooltip text
4671 }
4675 # Linker does the rest
4676 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'], $time, $descQuery );
4678 # Give the handler a chance to modify the parser object
4681 }
4684 }
4687 # Strip bad stuff out of the title (tooltip). We can't just use
4688 # replaceLinkHoldersText() here, because if this function is called
4689 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
4694 }
4696 # make sure there are no placeholders in thumbnail attributes
4697 # that are later expanded to html- so expand them now and
4698 # remove the tags
4703 }
4705 /**
4706 * Set a flag in the output object indicating that the content is dynamic and
4707 * shouldn't be cached.
4708 */
4712 }
4714 /**#@+
4715 * Callback from the Sanitizer for expanding items found in HTML attribute
4716 * values, so they can be safely tested and escaped.
4717 * @param string $text
4718 * @param PPFrame $frame
4719 * @return string
4720 * @private
4721 */
4726 }
4728 /**#@-*/
4730 /**#@+
4731 * Accessor/mutator
4732 */
4736 /**#@-*/
4738 /**#@+
4739 * Accessor
4740 */
4741 function getTags() { return array_merge( array_keys($this->mTransparentTagHooks), array_keys( $this->mTagHooks ) ); }
4742 /**#@-*/
4745 /**
4746 * Break wikitext input into sections, and either pull or replace
4747 * some particular section's text.
4748 *
4749 * External callers should use the getSection and replaceSection methods.
4750 *
4751 * @param string $text Page wikitext
4752 * @param string $section A section identifier string of the form:
4753 * <flag1> - <flag2> - ... - <section number>
4754 *
4755 * Currently the only recognised flag is "T", which means the target section number
4756 * was derived during a template inclusion parse, in other words this is a template
4757 * section edit link. If no flags are given, it was an ordinary section edit link.
4758 * This flag is required to avoid a section numbering mismatch when a section is
4759 * enclosed by <includeonly> (bug 6563).
4760 *
4761 * The section number 0 pulls the text before the first heading; other numbers will
4762 * pull the given section along with its lower-level subsections. If the section is
4763 * not found, $mode=get will return $newtext, and $mode=replace will return $text.
4764 *
4765 * @param string $mode One of "get" or "replace"
4766 * @param string $newText Replacement text for section data.
4767 * @return string for "get", the extracted section text.
4768 * for "replace", the whole page with the section replaced.
4769 */
4779 // Process section extraction flags
4786 }
4787 }
4788 // Preprocess the text
4791 // <h> nodes indicate section breaks
4792 // They can only occur at the top level, so we can find them by iterating the root's children
4795 // Find the target section
4797 // Section zero doesn't nest, level=big
4806 }
4807 }
4810 }
4812 }
4813 }
4816 // Not found
4821 }
4822 }
4824 // Find the end of the section, including nested sections
4831 }
4832 }
4835 }
4839 // Write out the remainder (in replace mode only)
4841 // Output the replacement text
4842 // Add two newlines on -- trailing whitespace in $newText is conventionally
4843 // stripped by the editor, so we need both newlines to restore the paragraph gap
4844 // Only add trailing whitespace if there is newText
4847 }
4852 }
4853 }
4856 // Re-insert stripped tags
4858 }
4861 }
4863 /**
4864 * This function returns the text of a section, specified by a number ($section).
4865 * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4866 * the first section before any such heading (section 0).
4867 *
4868 * If a section contains subsections, these are also returned.
4869 *
4870 * @param string $text text to look in
4871 * @param string $section section identifier
4872 * @param string $deftext default to return if section is not found
4873 * @return string text of the requested section
4874 */
4877 }
4881 }
4883 /**
4884 * Get the timestamp associated with the current revision, adjusted for
4885 * the default server-local timestamp
4886 */
4895 // Normalize timestamp to internal MW format for timezone processing.
4896 // This has the added side-effect of replacing a null value with
4897 // the current time, which gives us more sensible behavior for
4898 // previews.
4901 // The cryptic '' timezone parameter tells to use the site-default
4902 // timezone offset instead of the user settings.
4903 //
4904 // Since this value will be saved into the parser cache, served
4905 // to other users, and potentially even used inside links and such,
4906 // it needs to be consistent for all visitors.
4910 }
4912 }
4914 /**
4915 * Get the name of the user that edited the last revision
4916 */
4918 // if this template is subst: the revision id will be blank,
4919 // so just use the current user's name
4926 }
4928 }
4930 /**
4931 * Mutator for $mDefaultSort
4932 *
4933 * @param $sort New value
4934 */
4937 }
4939 /**
4940 * Accessor for $mDefaultSort
4941 * Will use the title/prefixed title if none is set
4942 *
4943 * @return string
4944 */
4954 }
4955 }
4957 /**
4958 * Accessor for $mDefaultSort
4959 * Unlike getDefaultSort(), will return false if none is set
4960 *
4961 * @return string or false
4962 */
4965 }
4967 /**
4968 * Try to guess the section anchor name based on a wikitext fragment
4969 * presumably extracted from a heading, for example "Header" from
4970 * "== Header ==".
4971 */
4973 # Strip out wikitext links(they break the anchor)
4976 # strip out HTML
4983 );
4988 }
4990 /**
4991 * Strips a text string of wikitext for use in a section anchor
4992 *
4993 * Accepts a text string and then removes all wikitext from the
4994 * string and leaves only the resultant text (i.e. the result of
4995 * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
4996 * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
4997 * to create valid section anchors by mimicing the output of the
4998 * parser when headings are parsed.
4999 *
5000 * @param $text string Text string to be stripped of wikitext
5001 * for use in a Section anchor
5002 * @return Filtered text string
5003 */
5005 # Strip internal link markup
5009 # Strip external link markup (FIXME: Not Tolerant to blank link text
5010 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
5011 # on how many empty links there are on the page - need to figure that out.
5014 # Parse wikitext quotes (italics & bold)
5017 # Strip HTML tags
5020 }
5024 }
5026 /**
5027 * strip/replaceVariables/unstrip for preprocessor regression testing
5028 */
5033 }
5041 }
5047 }
5049 }
5054 }
5056 }
5076 }
5077 }
5078 }
5080 }
5086 // First, find all strip markers, and store their
5087 // data in an array.
5090 while( ( $start_pos = strpos( $text, $this->mUniqPrefix, $pos ) ) && ( $end_pos = strpos( $text, self::MARKER_SUFFIX, $pos ) ) ) {
5102 }
5106 }
5109 // Now, find all of our links, and store THEIR
5110 // data in an array! :)
5114 // Internal links
5121 }
5126 }
5130 // Interwiki links
5136 }
5141 }
5143 function unserialiseHalfParsedText( $data, $intPrefix = null /* Unique identifying prefix */ ) {
5147 // First, extract the strip state.
5152 // Now, extract the text, and renumber links
5156 // Internal...
5163 }
5164 }
5166 // Interwiki...
5172 }
5174 // Should be good to go.
5176 }
5177 }
5179 /**
5180 * @todo document, briefly.
5181 * @ingroup Parser
5182 */
5189 }
5199 }
5209 }
5220 }
5221 }
5223 /**
5224 * @todo document, briefly.
5225 * @ingroup Parser
5226 */
5235 }
5236 }
5237 }