parser.php

来自「php 开发的内容管理系统」· PHP 代码 · 共 2,095 行 · 第 1/5 页

		return $retVal;
	}

	/**
	 * Render a forced-blue link inline; protect against double expansion of
	 * URLs if we're in a mode that prepends full URL prefixes to internal links.
	 * Since this little disaster has to split off the trail text to avoid
	 * breaking URLs in the following text without breaking trails on the
	 * wiki links, it's been made into a horrible function.
	 *
	 * @param Title $nt
	 * @param string $text
	 * @param string $query
	 * @param string $trail
	 * @param string $prefix
	 * @return string HTML-wikitext mix oh yuck
	 */
	function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
		list( $inside, $trail ) = Linker::splitTrail( $trail );
		$sk =& $this->mOptions->getSkin();
		$link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
		return $this->armorLinks( $link ) . $trail;
	}

	/**
	 * Insert a NOPARSE hacky thing into any inline links in a chunk that's
	 * going to go through further parsing steps before inline URL expansion.
	 *
	 * In particular this is important when using action=render, which causes
	 * full URLs to be included.
	 *
	 * Oh man I hate our multi-layer parser!
	 *
	 * @param string more-or-less HTML
	 * @return string less-or-more HTML with NOPARSE bits
	 */
	function armorLinks( $text ) {
		return preg_replace( "/\b(" . wfUrlProtocols() . ')/',
			"{$this->mUniqPrefix}NOPARSE$1", $text );
	}

	/**
	 * Return true if subpage links should be expanded on this page.
	 * @return bool
	 */
	function areSubpagesAllowed() {
		# Some namespaces don't allow subpages
		global $wgNamespacesWithSubpages;
		return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
	}

	/**
	 * Handle link to subpage if necessary
	 * @param string $target the source of the link
	 * @param string &$text the link text, modified as necessary
	 * @return string the full name of the link
	 * @private
	 */
	function maybeDoSubpageLink($target, &$text) {
		# Valid link forms:
		# Foobar -- normal
		# :Foobar -- override special treatment of prefix (images, language links)
		# /Foobar -- convert to CurrentPage/Foobar
		# /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
		# ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
		# ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage

		$fname = 'Parser::maybeDoSubpageLink';
		wfProfileIn( $fname );
		$ret = $target; # default return value is no change

		# Some namespaces don't allow subpages,
		# so only perform processing if subpages are allowed
		if( $this->areSubpagesAllowed() ) {
			# Look at the first character
			if( $target != '' && $target{0} == '/' ) {
				# / at end means we don't want the slash to be shown
				if( substr( $target, -1, 1 ) == '/' ) {
					$target = substr( $target, 1, -1 );
					$noslash = $target;
				} else {
					$noslash = substr( $target, 1 );
				}

				$ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash);
				if( '' === $text ) {
					$text = $target;
				} # this might be changed for ugliness reasons
			} else {
				# check for .. subpage backlinks
				$dotdotcount = 0;
				$nodotdot = $target;
				while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
					++$dotdotcount;
					$nodotdot = substr( $nodotdot, 3 );
				}
				if($dotdotcount > 0) {
					$exploded = explode( '/', $this->mTitle->GetPrefixedText() );
					if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
						$ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
						# / at the end means don't show full path
						if( substr( $nodotdot, -1, 1 ) == '/' ) {
							$nodotdot = substr( $nodotdot, 0, -1 );
							if( '' === $text ) {
								$text = $nodotdot;
							}
						}
						$nodotdot = trim( $nodotdot );
						if( $nodotdot != '' ) {
							$ret .= '/' . $nodotdot;
						}
					}
				}
			}
		}

		wfProfileOut( $fname );
		return $ret;
	}

	/**#@+
	 * Used by doBlockLevels()
	 * @private
	 */
	/* private */ function closeParagraph() {
		$result = '';
		if ( '' != $this->mLastSection ) {
			$result = '</' . $this->mLastSection  . ">\n";
		}
		$this->mInPre = false;
		$this->mLastSection = '';
		return $result;
	}
	# getCommon() returns the length of the longest common substring
	# of both arguments, starting at the beginning of both.
	#
	/* private */ function getCommon( $st1, $st2 ) {
		$fl = strlen( $st1 );
		$shorter = strlen( $st2 );
		if ( $fl < $shorter ) { $shorter = $fl; }

		for ( $i = 0; $i < $shorter; ++$i ) {
			if ( $st1{$i} != $st2{$i} ) { break; }
		}
		return $i;
	}
	# These next three functions open, continue, and close the list
	# element appropriate to the prefix character passed into them.
	#
	/* private */ function openList( $char ) {
		$result = $this->closeParagraph();

		if ( '*' == $char ) { $result .= '<ul><li>'; }
		else if ( '#' == $char ) { $result .= '<ol><li>'; }
		else if ( ':' == $char ) { $result .= '<dl><dd>'; }
		else if ( ';' == $char ) {
			$result .= '<dl><dt>';
			$this->mDTopen = true;
		}
		else { $result = '<!-- ERR 1 -->'; }

		return $result;
	}

	/* private */ function nextItem( $char ) {
		if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
		else if ( ':' == $char || ';' == $char ) {
			$close = '</dd>';
			if ( $this->mDTopen ) { $close = '</dt>'; }
			if ( ';' == $char ) {
				$this->mDTopen = true;
				return $close . '<dt>';
			} else {
				$this->mDTopen = false;
				return $close . '<dd>';
			}
		}
		return '<!-- ERR 2 -->';
	}

	/* private */ function closeList( $char ) {
		if ( '*' == $char ) { $text = '</li></ul>'; }
		else if ( '#' == $char ) { $text = '</li></ol>'; }
		else if ( ':' == $char ) {
			if ( $this->mDTopen ) {
				$this->mDTopen = false;
				$text = '</dt></dl>';
			} else {
				$text = '</dd></dl>';
			}
		}
		else {	return '<!-- ERR 3 -->'; }
		return $text."\n";
	}
	/**#@-*/

	/**
	 * Make lists from lines starting with ':', '*', '#', etc.
	 *
	 * @private
	 * @return string the lists rendered as HTML
	 */
	function doBlockLevels( $text, $linestart ) {
		$fname = 'Parser::doBlockLevels';
		wfProfileIn( $fname );

		# Parsing through the text line by line.  The main thing
		# happening here is handling of block-level elements p, pre,
		# and making lists from lines starting with * # : etc.
		#
		$textLines = explode( "\n", $text );

		$lastPrefix = $output = '';
		$this->mDTopen = $inBlockElem = false;
		$prefixLength = 0;
		$paragraphStack = false;

		if ( !$linestart ) {
			$output .= array_shift( $textLines );
		}
		foreach ( $textLines as $oLine ) {
			$lastPrefixLength = strlen( $lastPrefix );
			$preCloseMatch = preg_match('/<\\/pre/i', $oLine );
			$preOpenMatch = preg_match('/<pre/i', $oLine );
			if ( !$this->mInPre ) {
				# Multiple prefixes may abut each other for nested lists.
				$prefixLength = strspn( $oLine, '*#:;' );
				$pref = substr( $oLine, 0, $prefixLength );

				# eh?
				$pref2 = str_replace( ';', ':', $pref );
				$t = substr( $oLine, $prefixLength );
				$this->mInPre = !empty($preOpenMatch);
			} else {
				# Don't interpret any other prefixes in preformatted text
				$prefixLength = 0;
				$pref = $pref2 = '';
				$t = $oLine;
			}

			# List generation
			if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
				# Same as the last item, so no need to deal with nesting or opening stuff
				$output .= $this->nextItem( substr( $pref, -1 ) );
				$paragraphStack = false;

				if ( substr( $pref, -1 ) == ';') {
					# The one nasty exception: definition lists work like this:
					# ; title : definition text
					# So we check for : in the remainder text to split up the
					# title and definition, without b0rking links.
					$term = $t2 = '';
					if ($this->findColonNoLinks($t, $term, $t2) !== false) {
						$t = $t2;
						$output .= $term . $this->nextItem( ':' );
					}
				}
			} elseif( $prefixLength || $lastPrefixLength ) {
				# Either open or close a level...
				$commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
				$paragraphStack = false;

				while( $commonPrefixLength < $lastPrefixLength ) {
					$output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
					--$lastPrefixLength;
				}
				if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
					$output .= $this->nextItem( $pref{$commonPrefixLength-1} );
				}
				while ( $prefixLength > $commonPrefixLength ) {
					$char = substr( $pref, $commonPrefixLength, 1 );
					$output .= $this->openList( $char );

					if ( ';' == $char ) {
						# FIXME: This is dupe of code above
						if ($this->findColonNoLinks($t, $term, $t2) !== false) {
							$t = $t2;
							$output .= $term . $this->nextItem( ':' );
						}
					}
					++$commonPrefixLength;
				}
				$lastPrefix = $pref2;
			}
			if( 0 == $prefixLength ) {
				wfProfileIn( "$fname-paragraph" );
				# No prefix (not in list)--go to paragraph mode
				// XXX: use a stack for nestable elements like span, table and div
				$openmatch = preg_match('/(<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/center|<\\/tr|<\\/td|<\\/th)/iS', $t );
				$closematch = preg_match(
					'/(<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
					'<td|<th|<div|<\\/div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<center)/iS', $t );
				if ( $openmatch or $closematch ) {
					$paragraphStack = false;
					#聽TODO bug 5718: paragraph closed
					$output .= $this->closeParagraph();
					if ( $preOpenMatch and !$preCloseMatch ) {
						$this->mInPre = true;
					}
					if ( $closematch ) {
						$inBlockElem = false;
					} else {
						$inBlockElem = true;
					}
				} else if ( !$inBlockElem && !$this->mInPre ) {
					if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
						// pre
						if ($this->mLastSection != 'pre') {
							$paragraphStack = false;
							$output .= $this->closeParagraph().'<pre>';
							$this->mLastSection = 'pre';
						}
						$t = substr( $t, 1 );
					} else {
						// paragraph
						if ( '' == trim($t) ) {
							if ( $paragraphStack ) {
								$output .= $paragraphStack.'<br />';
								$paragraphStack = false;
								$this->mLastSection = 'p';
							} else {
								if ($this->mLastSection != 'p' ) {
									$output .= $this->closeParagraph();
									$this->mLastSection = '';
									$paragraphStack = '<p>';
								} else {
									$paragraphStack = '</p><p>';
								}
							}
						} else {
							if ( $paragraphStack ) {
								$output .= $paragraphStack;
								$paragraphStack = false;
								$this->mLastSection = 'p';
							} else if ($this->mLastSection != 'p') {
								$output .= $this->closeParagraph().'<p>';
								$this->mLastSection = 'p';
							}
						}
					}
				}
				wfProfileOut( "$fname-paragraph" );
			}
			// somewhere above we forget to get out of pre block (bug 785)
			if($preCloseMatch && $this->mInPre) {
				$this->mInPre = false;
			}
			if ($paragraphStack === false) {
				$output .= $t."\n";
			}
		}
		while ( $prefixLength ) {
			$output .= $this->closeList( $pref2{$prefixLength-1} );
			--$prefixLength;
		}
		if ( '' != $this->mLastSection ) {
			$output .= '</' . $this->mLastSection . '>';
			$this->mLastSection = '';
		}

		wfProfileOut( $fname );
		return $output;
	}

	/**
	 * Split up a string on ':', ignoring any occurences inside tags
	 * to prevent illegal overlapping.
	 * @param string $str the string to split
	 * @param string &$before set to everything before the ':'
	 * @param string &$after set to everything after the ':'
	 * return string the position of the ':', or false if none found
	 */
	function findColonNoLinks($str, &$before, &$after) {
		$fname = 'Parser::findColonNoLinks';
		wfProfileIn( $fname );
		
		$pos = strpos( $str, ':' );
		if( $pos === false ) {
			// Nothing to find!
			wfProfileOut( $fname );
			return false;
		}
		
		$lt = strpos( $str, '<' );
		if( $lt === false || $lt > $pos ) {
			// Easy; no tag nesting to worry about
			$before = substr( $str, 0, $pos );
			$after = substr( $str, $pos+1 );
			wfProfileOut( $fname );
			return $pos;
		}
		
		// Ugly state machine to walk through avoiding tags.
		$state = MW_COLON_STATE_TEXT;
		$stack = 0;
		$len = strlen( $str );
		for( $i = 0; $i < $len; $i++ ) {
			$c = $str{$i};
			
			switch( $state ) {
			// (Using the number is a performance hack for common cases)
			case 0: // MW_COLON_STATE_TEXT:
				switch( $c ) {
				case "<":
					// Could be either a <start> tag or an </end> tag
					$state = MW_COLON_STATE_TAGSTART;
					break;
				case ":":
					if( $stack == 0 ) {
						// We found it!
						$before = substr( $str, 0, $i );
						$after = substr( $str, $i + 1 );
						wfProfileOut( $fname );
						return $i;
					}
					// Embedded in a tag; don't break it.
					break;
				default:
					// Skip ahead looking