PlomWiki: Zur Start-Seite Suche Letzte Änderungen (Feed) Letzte Kommentare (Feed)
Impressum Datenschutz-Erklärung


Ansicht Bearbeiten Anzeige-Titel setzen Versions-Geschichte Seiten-Passwort setzen AutoLink-Anzeige ein-/ausschalten
<?php if (!defined('PmWiki')) exit(); 
/*  AutomaticLinks2 / autolink2 version 0.2 
    Large parts by Christian Heller / 
	But a lot of this code and the whole foundation is from AutomaticLinks / autolink.php  
	by Karl Loncarek: 
	This file is part of PmWiki; you can redistribute it and/or modify 
	it under the terms of the GNU General Public License as published 
	by the Free Software Foundation; either version 2 of the License, or 
	(at your option) any later version.  See pmwiki.php for full details. 
	Automatically links phrases in wiki page texts to pages that have a sufficiently similar name, giving a certain 
	tolerance to different word endings, German umlauts and a certain case insensitivity. 
	Also provides for a (:autolink2backlinks:) directive that outputs a linked list of pages that seem to automatically 
	backlink to the current page in autolink2 fashion. 
	autolink2.php reads the names of all the files in wiki.d/ that belong to groups given to the Autolink2Activate function 
	and creates an entry for each of them in a file .autolink2-$group and an array $AutolinkListFull of the format 
	$id of a page =>  
	   ($pagename in the format $group.something,  
	    $regexp to match the non-group part of this pagename,  
		$references (an array of $ids) to other pages in the page text of $pagename) 
	On a given wiki page of $current_id in the $AutolinkLIstFull array, this data is used to match (via $regexps 
	for the $ids given in $references) and markup otherwise non-linked phrases in the page text to create links  
	to other wiki pages.  
	(see functions Autolink2Active, AutolinkSet) 
	The $references have previously been collected from reading .autolink2-group and during the updating of 
	$AutolinkListFull (which, of course, also writes back to .autolink2-group). 
	(see functions AutolinkListFullRead, AutolinkListFullUpdate and CollectPossibleReferences) 
	The $regexp is created with a certain tolerance towards other ways to write the phrase matching the pagename, 
	allowing for case insensitivity, different ways to write umlauts and, to a certain degree, different word endings.  
	It's double-checked by 1) matching the phrase in the page text in the Markup function, which opens the function  
	AutolinkSet for link creation which 2) tries to translate the phrase found via $regexp back into a pagename to link; 
	this may lead to cases where the $regexp of one $id matches a phrase that is found out by AutolinkSet to more 
	closely match another $id; a desired effect. 
	(see functions GenerateAutolinkRegexp, AutolinkSet) 
	There's also a Markup to output a list of all the pages that have $current_id in their list of references,  
	callable via the (:autolink2backlinks:) directive.  
	(see functions Autolink2Activate, GenerateBacklinks) 
	v.0.2 / 2009-12-18 Fixed a bug whereby empty reference IDs caused Markups with ergo empty regexps to markup a lot of space 
	                   not supposed to be marked up--by checking for an ID > 0 before the generation of autolink2markups. 
	TO DO: 
	- certain number words (i.e. word divisions that follow a word that ends in a number don't get matched yet) 
	- French accents etc. 
	- words like "ATripToParis" with one-letter-words 
	- uppercase-umlaut letter rules 
	- limits on the word-lengths that the tolerances for different word endings are effective for 
	- currently too large pages lead to page generation times over the 30s limit imposed by plomlompom's server; autolink2 can be 
	  deactivated for pages like this via $ExcludePagesFromAutolink2links, but a better solution would be to reduce page generation 
#set predefined values 
$AutolinkListFull = array(); 
$AutolinkListShort = array(); 
$AutolinkTime = 0; 
$AutolinkFmt = '<a class="autolink" href="$LinkUrl">$LinkText</a>'; 
function AutolinkListFullRead($group) { 
# Read the group-specific .autolink file and translate it into $AutolinkListFull. 
	global $AutolinkListFull,$AutolinkTime,$WorkDir; 
	$filename_autolink = $WorkDir.'/.autolink2-'.$group; 
	if (file_exists($filename_autolink)) { 
		$AutolinkTime = filemtime($filename_autolink); 
		if  ($fp = file_get_contents($filename_autolink, "r")) { 
			$entries = explode('!!!', $fp); # delimiter between entries is string "!!!" 
			foreach ($entries as $entry) { 
				$data = explode('%250a', $entry); # delimiter between data inside of entries is "%250a" 
				$id = $data[0]; 
				$name = $data[1]; 
				$regexp = $data[2]; 
				$references = explode(',', $data[3]); # delimiter between references is "," 
				$AutolinkListFull[$id] = array($name, $regexp, $references); 
function AutolinkListFullUpdate($group) { 
# Check if files have been added, removed or changed since the last update.  
# Update $AutolinkListFull and the group-specific .autolink file accordingly. 
	global $AutolinkTime,$AutolinkListFull,$WorkDir,$SearchPatterns; 
	# Fill $matches with all pagenames in wiki.d/ to consider for autolink2. 
	$filename_autolink =  $WorkDir.'/.autolink2-'.$group; 
	if ($handle = opendir($WorkDir)) { # get filenames that are matching 
		while (false !== ($file = readdir($handle))) { 
			if ($file != '.' && $file!= '..' && substr($file,0,strlen($group)+1) == $group.'.' && 0 == preg_match('/,del-[0-9]*$/', $file)) { # only accept group-specific files and ignore deleted pages 
				if (isset($SearchPatterns['default'])) { 
					foreach($SearchPatterns['default'] as $v) { #check for in search excluded files 
						$count += preg_match($v,$file); 
				if ($count==0) { #save only not excluded files 
	# Check which pages are new and which have been modified. 
	# Create a filename => id array called $modify for modified pages 
	# and a filename array called $new_pages for new ones. 
	$list_filenames_ids = array(); 
	foreach ($AutolinkListFull as $id => $pageData) { 
		$list_filenames_ids[$pageData[0]] = $id; 
	$modify = array(); 
	$new_pages = array(); 
	foreach ($matches as $filename) { 
		$list_filenames = array_keys($list_filenames_ids); 
		if (in_array($filename, $list_filenames)) { 
			if (filemtime($WorkDir.'/'.$filename) >= $AutolinkTime) { 
				$modify[$filename] = $list_filenames_ids[$filename]; 
		else { 
			$new_pages[] = $filename; 
	# If there are new pages ...	 
	if (count($new_pages) > 0) { 
		# ... first create new IDs for these, higher than all existing ones. 
		# Build a filename => id array named $create with these. 
		$list_IDs = array_keys($AutolinkListFull); 
		$highest_id = 0; 
		if ($list_IDs != array()) { 
			$highest_id = max(max($list_IDs), 0); 
		foreach ($new_pages as $filename) { 
			$highest_id = $highest_id + 1; 
			$create[$filename] = $highest_id; 
		# Create entries for the new pages in $AutolinkListFull. 
		# Their only content at first will be their name and the regular expression to find matches to them. 
		# We will count their possible autolink references to other pages at a later point, together with  
		# the merely modified ones, when we have all new pages together in $AutolinkListFull. 
		foreach ($create as $filename => $id) { 
			$regexp = GenerateAutolinkRegexp($filename); 
			$AutolinkListFull[$id] = array($filename, $regexp, array()); 
		$modify = array_merge($modify, $create); 
		# Now that $AutolinkListFull has all new pages added, re-sort it by filename-length, from large to small. 
		$filenames_with_id = array(); 
		foreach($AutolinkListFull as $id => $pageData) { 
			$filename = $pageData[0]; 
			$filenames_with_id[$filename] = $id;		 
		$filenames = array_keys($filenames_with_id); 
		uasort($filenames,'sizersort'); # sort array on size, largest entries first 
		$AutolinkListFullNew = array(); 
		foreach($filenames as $filename) { 
			$id = $filenames_with_id[$filename]; 
			$AutolinkListFullNew[$id] = $AutolinkListFull[$id]; 
		$AutolinkListFull = $AutolinkListFullNew; 
		# There are no references yet to the new pages in the old ones,  
		# and later we will only do a total recount in those that are new or have been modified. 
		# Therefore it becomes necessary to go through the page texts of all unmodified old pages  
		# to find possible references to the new ones. 
		$IDsToBeIgnored = array_values($modify); 
		foreach ($create as $filename_new => $id_new) { 
			$regexp = $AutolinkListFull[$id_new][1]; 
			foreach ($AutolinkListFull as $id => $pageData) { 
				if (!(in_array($id, $IDsToBeIgnored))) { 
					$filename = $pageData[0]; 
					$page = ReadPage($filename); 
					$pagetext = $page['text']; 
					if (1 == preg_match('/'.$regexp.'/i', $pagetext)) { 
						$AutolinkListFull[$id][2][] = $id_new; 
	# For all pages mentioned in $modify (i.e. the modified as well as the new ones):  
	# Analyze their text to find possible references to other pages via regular expressions. 
	foreach ($modify as $filename => $id) { 
		$AutolinkListFull[$id][2] = CollectPossibleReferences($filename); 
	# check whether files do exist, if not: remove from filename/title list 
	foreach($AutolinkListFull as $id => $pageData) { 
		$filename = $pageData[0]; 
		if (!file_exists("$WorkDir/$filename")) { 
	# Now write the $AutolinkListFull content into the group-specific .autolink file! 
	if ($modify != array()) { 
		$string_part = ''; 
		$array_to_string = array(); 
		foreach($AutolinkListFull as $id => $pageData) { 
			$string_part = $id.'%250a'.$pageData[0].'%250a'.$pageData[1].'%250a'.implode(',', $pageData[2]); 
			$array_to_string[] = $string_part; 
		$string = implode('!!!', $array_to_string); 
		file_put_contents($filename_autolink, $string); 
function GenerateAutolinkRegexp($filename) { 
# Generate a regular expression compatible with the needs of autolink2 out of a pagename to be autolinked. 
# The rules for the regexp generation are the following: 
# 1) strip the group part of the filename 
# 2) find occurences of CamelCase and therefore multiple words contained in the filename; 
#    allow for any number of occurences of [.,- ] in between multiple words thus found; 
#    also give a certain tolerance for more or less or different characters in the suffixes 
#    of words preceding uppercase CamelCase words 
# 	 a) do this first for words where there are two lowercase characters preceding uppercase characters 
#    b) do this second for the words left where there is only one lowercase character preceding an uppercase character 
#    This rule combination makes sense for German umlauts (single characters) which have been transformed into two latin characters, 
#    for which the path is opened in the next step: 
# 3) allow two-characters-combinations typical for transformation from one German umlaut character into two latin characters 
#    to be read as either these two latin characters or as one German umlaut character 
	$filename = preg_replace('/([A-Za-z0-9]*)\.([A-Za-z0-9]*)/', '$2', $filename); # we strip the group part of the filename 
	$regexp = preg_replace('/[a-z][a-z][-]?([A-Z0-9])/', '[a-zäöüß]?[a-zäöüß]?[a-zäöüß]?[.,\-: ]*$1', $filename); 
	$regexp = preg_replace('/[a-z][-]?([A-Z0-9])/', '[a-zäöüß]?[a-zäöüß]?[.,\-: ]*$1', $regexp); 
	$searcharray = array(  'ae',       'Ae',       'oe',       'Oe',       'ue',       'Ue',       'ss'); 
	$replacearray = array( '((ae)|ä)', '((Ae)|Ä)', '((oe)|ö)', '((Oe)|ö)', '((ue)|ü)', '((Ue)|ü)', '((ss)|ß)'); 
	$regexp = str_replace($searcharray,$replacearray,$regexp); 
	return $regexp; 
function sizersort($a,$b) { 
# sorting helper function to sort regarding textsize in mind, from longest to smallest 
	if (strlen($a)==strlen($b)) { 
		return 0; 
	return (strlen($a) > strlen($b)) ? -1 : 1; 
function CollectPossibleReferences($filename) { 
# Search $AutolinkListFull for pagenames that seem to be referenced in the page text. 
	global $AutolinkListFull; 
	$page = ReadPage($filename); 
	$pagetext = $page['text']; 
	$array_references = array(); 
	foreach ($AutolinkListFull as $id => $pageData) { 
		$regexp = $pageData[1]; 
		if (1 == preg_match('/'.$regexp.'/i', $pagetext)) { 
			$array_references[] = $id; 
	return $array_references; 
function AutolinkSet($pattern) { 
# AutolinkSet is called by the Markup function on a text pattern to decide  
# whether it fits $AutolinkListShort well enough to turn it into a link; 
# and if so, which link to turn it into. 
# A sequence of comparison rules of growing tolerance is traversed for each 
# pagename in $AutolinkListShort in combination with $pattern. If a comparison 
# is successful, but the rule was rather lax, the pagenames in  
# $AutolinkListShort that have not been compared yet to $pattern are also  
# tried out according to stricter rules; thus not only the first, but also 
# the highest similarity is matched. 
	global $AutolinkListShort,$pagename,$AutolinkFmt; 
	$local_page = $pagename; # There's so many $pagenameSomethingSomething variables after this, so we rename this one to avoid confusion. 
	# Some formatting on $pattern to eliminate umlauts. 
	# (We assume that in pagenames umlauts have already been transformed to non-umlauts in a way comparable to this ruleset.) 
	$pattern_temp = preg_replace('/ä/', 'ae', $pattern); 
	$pattern_temp = preg_replace('/Ä/', 'Ae', $pattern_temp); 
	$pattern_temp = preg_replace('/ö/', 'oe', $pattern_temp); 
	$pattern_temp = preg_replace('/Ö/', 'Oe', $pattern_temp); 
	$pattern_temp = preg_replace('/ü/', 'ue', $pattern_temp); 
	$pattern_temp = preg_replace('/Ü/', 'Ue', $pattern_temp); 
	$pattern_temp = preg_replace('/ß/', 'ss', $pattern_temp); 
	# We create different versions of $pattern to accomodate different comparison rules. 
	# $pattern_HyphenPossible cuts out occurences of '.', ',' ':' and ' ', but keeps '-' (unlike $pattern_NoHyphen).	 
	# $pattern_divided identifies single words and replaces everything (or the nothing) between them with '!!!' to unambiguously divide them. 
	# (Rule for this: single words are separated where uppercase characters or numbers follow lowercase letters and/or by occurences of '.', ',', '-', ':', ' '.) 
	# $pattern_parts translates $pattern_divided into an array of the single words identified in $pattern. 
	$pattern_HyphenPossible = preg_replace('/[.,: ]*/', '', $pattern_temp); 
	$pattern_HyphenPossible = strtolower($pattern_HyphenPossible); 
	$pattern_NoHyphen = preg_replace('/[-]*/', '', $pattern_HyphenPossible); 
	$pattern_divided = preg_replace('/([a-z])[.,\-: ]*([A-Z0-9])/', '$1!!!$2', $pattern_temp); 
	$pattern_divided = preg_replace('/([a-z])[.,\-: ]+([a-z0-9])/', '$1!!!$2', $pattern_divided); 
	$pattern_divided = strtolower($pattern_divided); 
	$pattern_parts = explode('!!!', $pattern_divided); 
	$number_pattern_parts = count($pattern_parts); 
	# $FuzzyMatch is to be filled if a fuzzy match between a pattern and a linkable pagename is found. 
	# It will be used in case no better match is found. 
	# $FuzzyMatch[1] contains the linkable pagename. 
	# $FuzzyMatch[0] contains an integer indicating a prioritization; the higher the integer, the better the match. 
	$FuzzyMatch = array(0, ''); 
	# Traverse $AutolinkListShort for pagenames (as $pagenameAbsolute, later to be shortened to $pagename) to compare $pattern to. 
	# If $pagenameAbsolute equals the local page, ignore it. 
	# Else, strip the group name part from $pagenameAbsolute to generate $pagename. 
	# (Rule for this: Cut out the pagename part (alphabetical characters, numbers and hyphens) that ends in a dot.) 
	# $pagename_lowercase finally will make it comparable to the lowercase versions of $pattern. 
	foreach ($AutolinkListShort as $id => $pageData) { 
		$pagenameAbsolute = $pageData[0]; 
		if ($pagenameAbsolute != $local_page) { 
			$pagenameRelative = preg_replace('/([A-Za-z0-9\-]*)\.([A-Za-z0-9]*)/', '$2', $pagenameAbsolute); 
			$pagename_lowercase = strtolower($pagenameRelative); 
			# COMPARISON TOLERANCE LEVEL 0: success in the case of an absolute match. 
			# If the lowercase version of the remaining string fits $pattern_HyphenPossible, return a link straight away. 
			if ($pagename_lowercase == $pattern_HyphenPossible ) { 
			# If the tolerance comparison 0 failed, continue comparisons on the current $pagename if no previous fuzzy (non-absolute) match of priority 2 has been found. 
			if ($FuzzyMatch[0] < 2) { 
				# COMPARISON TOLERANCE LEVEL 1: success in the case of an absolute match, but ignoring hyphens. 
				$pagename_NoHyphen = preg_replace('/-/', '', $pagename_lowercase); 
				if ($pagename_NoHyphen == $pattern_NoHyphen) { 
					$FuzzyMatch[0] = 2; 
					$FuzzyMatch[1] = $pagenameAbsolute; 
				# Since tolerance comparison 1 failed, further continue comparisons on the current $pagename if 
				# 1) no previous fuzzy (non-absolute) match has been found. 
				# 2) $pattern consists of more than one word (check $number_pattern_parts). 
				# 3) $pagename consists of at least as many words. 
				# For condition 3) we try to split $pagename into single words like we did previously with $pattern. 
				# (Rule for this: single words are separated where uppercase characters or numbers follow lowercase letters and/or by occurences of '-'. 
				# Notice that this rule is a little simpler than the one for $pattern. We assume less tolerance for the naming of pages.) 
				elseif (($FuzzyMatch[0] == 0) and ($number_pattern_parts > 1)) { 
					$pagename_divided = preg_replace('/([a-z])[-]?([A-Z0-9])/', '$1!!!$2', $pagenameRelative); 
					$pagename_divided = strtolower($pagename_divided); 
					$pagename_parts = explode('!!!', $pagename_divided); 
					$number_pagename_parts = count($pagename_parts); 
					if ($number_pagename_parts == $number_pattern_parts) { 
						# COMPARISON TOLERANCE LEVEL 2: check if the words would fit if they had different suffixes. 
						# We traverse the arrays $pagename_parts and $pattern_parts to compare parts. 
						# While doing this, we give a certain indifference to word endings: 
						# Compare not only $pattern_part and $pagename_part, but also versions of each truncated at the end. 
						# If the comparison is successful, remember $pagenameAbsolute as a $FuzzyMatch: 
						# This is to be used in case no better match is found in the remaining iterations of $AutolinkListShort. 
						$FuzzyMatch = array(1, $pagenameAbsolute); 
						$counter = 0; 
						while (($Parts_PagenameAndPattern_Similar !== 0) and ($counter < $number_pattern_parts)) { 
							$pagename_part = $pagename_parts[$counter]; 
							$pagename_part_short = substr($pagename_part, 0, -1); 
							$pattern_part = $pattern_parts[$counter]; 
							$pattern_part_short = substr($pattern_part, 0, -1); 
							if (($pagename_part !== $pattern_part) and ($pagename_part_short !== $pattern_part_short) and ($pagename_part_short !== $pattern_part) and ($pagename_part !== $pattern_part_short)) { 
									$FuzzyMatch = array(0, ''); 
							$counter = $counter + 1; 
	# If nothing but a fuzzy match has been found for the pattern, and this is not $local_page, return it as the link. 
	if (($FuzzyMatch[0] > 0) and ($FuzzyMatch[1] !== $local_page)) { 
	# If all else fails, at least return $pattern without a link. 
	return $pattern; 
function GenerateBacklinks($current_id) { 
# Generate a list in pmWiki markup format with links for all the pages  
# that reference a given page (identified by $current_id)  
	global $AutolinkListFull; 
	# Look up $current_id in the reference lists of all the page entries in $AutolinkListFull. 
	# If found, add these to the $backlinks list. 
	$backlinks = array(); 
	foreach ($AutolinkListFull as $id => $pageData) { 
		if (in_array($current_id, $pageData[2])) { 
			$backlinks[] = $pageData[0]; 
	# Sort the backlinks list alphabetically and turn it into a string $string_backlinks 
	# that formats them as a list of links into the pmWiki markup language. Return it. 
	$string_backlinks = ''; 
	foreach ($backlinks as $pagenameAbsolute) { 
		$string_backlinks = $string_backlinks.'* [['.$pagenameAbsolute.']] (:nl:)'; 
	return $string_backlinks; 
function Autolink2Activate($groups, $ExcludePagesFromAutolink2links) { 
	global $AutolinkListFull,$AutolinkListShort,$pagename; 
	# Read and update all group-specific .autolink files as well as the $AutolinkListFull array. 
	$group = explode (" ",$groups); 
	foreach ($group as $v) { 
	# Stop here if the current page ($pagename) is included in $ExcludePagesFromAutolink2links. 
	foreach ($ExcludePagesFromAutolink2links as $pattern) { 
		if (1 == preg_match($pattern, $pagename)) { 
	# Try to find the current page in the AutolinkList. Stop here if it cannot be found. 
	$current_id = ''; 
	foreach($AutolinkListFull as $id => $pageData) { 
		if ($pageData[0] == $pagename) { 
			$current_id = $id; 
	if ($current_id != '') { 
		# $current_references is a list of the IDs of all the pages that seem to be referenced in this page's text. 
		# Shorten $AutolinkListFull to $AutolinkListShort, which only contains entries for these very pages. 
		# Notice that, even though the list is shortened, the kept entries' relative order to each other  
		# (hopefully top-down from the longest pagenames to the shortest ones due to AutolinkListFullUpdate function)  
		# is conserved. 
		$current_references = $AutolinkListFull[$current_id][2]; 
		foreach ($AutolinkListFull as $id => $pageData) { 
			if (in_array($id, $current_references)) { 
				$AutolinkListShort[$id] = $AutolinkListFull[$id]; 
		# Create a Markup function for every assumed Autolink > 0 (don't catch empty references!) reference, to be identified via regular expressions. 
		# The best place for this sequence in the order of pmWiki Markups seems to be after 'style'. 
		$LastMarkupName = 'style'; 
		foreach ($current_references as $referenced_id) { 
			if ($referenced_id > 0) { 
				$regexp = $AutolinkListFull[$referenced_id][1]; 
				$NewMarkupName = 'autolink2links'.$referenced_id; 
				# Try to match only Autolink regular expressions outside of HTML tags (i.e. which are not followed  
				# in any distance by a '>' if there's no '<' or line-end in-between). 
				# AutolinkSet has the final responsibility for deciding whether the Markup is to do anything,  
				# i.e. create a link and pointing to what. 
				Markup($NewMarkupName, '>'.$LastMarkupName, '/('.$regexp.')(?=[^>]*($|<(?!\/a)))/ie', 'Keep(AutolinkSet("$1"), "L")'); 
				$LastMarkupName = $NewMarkupName; 
		# Markup to output a backlink list via the GenerateBacklinks function. 
		$backlinks = GenerateBacklinks($current_id); 
		Markup('autolink2backlinks', '<nl0', '/\(:autolink2backlinks:\)/', $backlinks); 
# $groups gives the names of the groups this plugin is supposed to work for. 
# $ExcludePagesFromAutolink2linksMarkup gives regexp patterns for pagenames on which the autolink2links Markup is not to be used. 
# (Notice that this does not mean that they won't be taken into account for autolink2backlinks generation.) 
$groups = 'Mind'; 
$ExcludePagesFromAutolink2links = array('/Mind.RecentChanges/', '/Mind.TwitterHistory-/'); 
Autolink2Activate($groups, $ExcludePagesFromAutolink2links); 


Keine Kommentare zu dieser Seite.

Schreibe deinen eigenen Kommentar

Kommentar-Schreiben derzeit nicht möglich: Kein Captcha gesetzt.

PlomWiki-Engine lizensiert unter der AGPLv3. Quellcode verfügbar auf GitHub.