yakka/library/YakkaTextProcessor.php

<?php

class YakkaTextProcessor {
        var $engine;

        var $pattern;

        var $tagState;
        var $tagCount;

        function YakkaTextProcessor(&$engine) {
                $this->engine = &$engine;

                $this->pattern = "/(\%\%.*?\%\%|\"\".*?\"\"|\[\[.*?\]\]|<|>|&|\*\*.*?\*\*|\/\/.*?\/\/|__.*?__|\#\#.*?\#\#|======.*?======|=====.*?=====|====.*?====|===.*?===|==.*?==|----|---|\n[\t\|]+(-|[0-9,a-z,A-Z]+\))?|\{\{.*?\}\}|\b[A-Z][A-Z,a-z]+[:]([A-Z][a-z]+[A-Z,0-9][A-Z,a-z,0-9]*)\b|\b([A-Z][a-z]+[A-Z,0-9][A-Z,a-z,0-9]*)\b|\n)/ms";
        }

        function Process($source) {
                $source = str_replace("\r", "", $source);
                $source = trim($source)."\n";
                $source = $this->ExecuteBuilder($source);
                $source = preg_replace("/<br \/>$/", "", trim($source));
                return $source;
        }


        function ExecuteBuilder($source) {
                return preg_replace_callback($this->pattern, array(&$this, "Build"), $source);
        }

        function Build($things) {
                $thing = $things[1];

                static $oldIndentLevel = 0;
                static $indentClosers = array();
                static $br = 1;

                global $wakka;

                // convert HTML thingies
                if ($thing == "<")
                        return "&lt;";
                else if ($thing == ">")
                        return "&gt;";
                else if ($thing == "&")
                        return "&amp;";
                // bold
                else if (preg_match("/^\*\*(.*)\*\*$/s", $thing, $matches))
                {
                        return "<b>".$this->ExecuteBuilder($matches[1])."</b>";
                }
                // italic
                else if (preg_match("/^\/\/(.*)\/\/$/s", $thing, $matches))
                {
                        return "<i>".$this->ExecuteBuilder($matches[1])."</i>";
                }
                // underlinue
                else if (preg_match("/^__(.*)__$/s", $thing, $matches))
                {
                        return "<u>".$this->ExecuteBuilder($matches[1])."</u>";
                }
                // monospace
                else if (preg_match("/^\#\#(.*)\#\#$/s", $thing, $matches))
                {
                        return "<tt>".$this->ExecuteBuilder($matches[1])."</tt>";
                }
                // headers
                else if (preg_match("/^======(.*)======$/s", $thing, $matches))
                {
                        $br = 0;
                        return "<h1>".$this->ExecuteBuilder($matches[1])."</h1>";
                }
                else if (preg_match("/^=====(.*)=====$/s", $thing, $matches))
                {
                        $br = 0;
                        return "<h2>".$this->ExecuteBuilder($matches[1])."</h2>";
                }
                else if (preg_match("/^====(.*)====$/s", $thing, $matches))
                {
                        $br = 0;
                        return "<h3>".$this->ExecuteBuilder($matches[1])."</h3>";
                }
                else if (preg_match("/^===(.*)===$/s", $thing, $matches))
                {
                        $br = 0;
                        return "<h4>".$this->ExecuteBuilder($matches[1])."</h4>";
                }
                else if (preg_match("/^==(.*)==$/s", $thing, $matches))
                {
                        $br = 0;
                        return "<h5>".$this->ExecuteBuilder($matches[1])."</h5>";
                }
                // separators
                else if ($thing == "----")
                {
                        // TODO: This could probably be improved for situations where someone puts text on the same line as a separator.
                        //       Which is a stupid thing to do anyway! HAW HAW! Ahem.
                        $br = 0;
                        return "<hr/>";
                }
                // forced line breaks
                else if ($thing == "---")
                {
                        return "<br />";
                }
                // escaped text
                else if (preg_match("/^\"\"(.*)\"\"$/s", $thing, $matches))
                {
                        return $matches[1];
                }
                // code text
                else if (preg_match("/^\%\%(.*)\%\%$/s", $thing, $matches))
                {
                        // check if a language has been specified
                        $code = $matches[1];
                        if (preg_match("/^\((.+?)\)(.*)$/s", $code, $matches))
                        {
                                list(, $language, $code) = $matches;
                        }
                        switch ($language)
                        {
                        case "php":
                                $formatter = "php";
                                break;
                        default:
                                $formatter = "code";
                        }

                        $output = "<div class=\"code\">";
                        $output .= "CODE";//$wakka->Format(trim($code), $formatter);
                        $output .= "</div>";

                        return $output;
                }
                // forced links
                else if (preg_match("/^\[\[(\S*)(\s+(.+))?\]\]$/", $thing, $matches))
                {
                        list (, $url, , $text) = $matches;
                        if ($url)
                        {
                                if (!$text) $text = $url;
                                return $this->CreateLink($url, "", $text);
                        }
                        else
                        {
                                return "";
                        }
                }
                // indented text
                else if (preg_match("/\n([\t\|]+)(-|([0-9,a-z,A-Z]+)\))?/s", $thing, $matches))
                {
                        // new line
                        $result .= ($br ? "<br />\n" : "\n");

                        // we definitely want no line break in this one.
                        $br = 0;

                        // find out which indent type we want
                        $newIndentType = $matches[2];
                        if (!$newIndentType) { $opener = "<blockquote>"; $closer = "</blockquote>"; $br = 1; }
                        else if ($newIndentType == "-") { $opener = "<ul>"; $closer = "</ul>"; $li = 1; }
                        else { $opener = "<ol type=\"".$newIndentType."\">"; $closer = "</ol>"; $li = 1; }

                        // get new indent level
                        $newIndentLevel = strlen($matches[1]);
                        if ($newIndentLevel > $oldIndentLevel)
                        {
                                for ($i = 0; $i < $newIndentLevel - $oldIndentLevel; $i++)
                                {
                                        $result .= $opener;
                                        array_push($indentClosers, $closer);
                                }
                        }
                        else if ($newIndentLevel < $oldIndentLevel)
                        {
                                for ($i = 0; $i < $oldIndentLevel - $newIndentLevel; $i++)
                                {
                                        $result .= array_pop($indentClosers);
                                }
                        }

                        $oldIndentLevel = $newIndentLevel;

                        if ($li) $result .= "<li/>";

                        return $result;
                }
                // new lines
                else if ($thing == "\n")
                {
                        // if we got here, there was no tab in the next line; this means that we can close all open indents.
                        $c = count($indentClosers);
                        for ($i = 0; $i < $c; $i++)
                        {
                                $result .= array_pop($indentClosers);
                                $br = 0;
                        }
                        $oldIndentLevel = 0;

                        $result .= ($br ? "<br />\n" : "\n");
                        $br = 1;
                        return $result;
                }
                // events
                else if (preg_match("/^\{\{(.*?)\}\}$/s", $thing, $matches))
                {
                        if ($matches[1])
                                return $this->engine->runAction($matches[1]);
                        else
                                return "{{}}";
                }
                // interwiki links!
                else if (preg_match("/^[A-Z][A-Z,a-z]+[:]([A-Z][a-z]+[A-Z,0-9][A-Z,a-z,0-9]*)$/s", $thing))
                {
                        return $this->CreateLink($thing);
                }
                // wakka links!
                else if (preg_match("/^[A-Z][a-z]+[A-Z,0-9][A-Z,a-z,0-9]*$/s", $thing))
                {
                        return $this->CreateLink($thing);
                }
                // if we reach this point, it must have been an accident.
                return $thing;
        }

        function CreateLink($tag, $method = null, $text = null) {
                $track = 1;
                if (!$text) $text = $tag;
                
                // is this an interwiki link?
                if (preg_match("/^([A-Z][A-Z,a-z]+)[:]([A-Z][a-z]+[A-Z,0-9][A-Z,a-z,0-9]*)$/", $tag, $matches))
                {
                        //$tag = $this->GetInterWikiUrl($matches[1], $matches[2]);
                        //return "<a href=\"$tag\">$text</a>";
                        return "--interwikiurl--";
                }
                // is this a full link? ie, does it contain alpha-numeric characters?
                else if (preg_match("/[^[:alnum:]]/", $tag))
                {
                        // check for email addresses
                        if (preg_match("/^.+\@.+$/", $tag))
                        {
                                $tag = "mailto:".$tag;
                        }
                        // check for protocol-less URLs
                        else if (!preg_match("/:/", $tag))
                        {
                                $tag = "http://".$tag;
                        }
                        return "<a href=\"$tag\">$text</a>";
                }
                else
                {
                        // it's a Wakka link!
                        return ($this->engine->pageStorage->hasPage($tag) ? "<a href=\"".$this->engine->buildCommand($tag, $method)."\">".$text."</a>" : "<span class=\"missingpage\">".$text."<a href=\"".$this->engine->buildCommand($tag, "Create")."\">?</a></span>");
                }
        }

}

?>
1	<?php
2
3	class YakkaTextProcessor {
4	var $engine;
5
6	var $pattern;
7
8	var $tagState;
9	var $tagCount;
10
11	function YakkaTextProcessor(&$engine) {
12	$this->engine = &$engine;
13
14	$this->pattern = "/(\%\%.?\%\%\|\"\".?\"\"\|\[\[.?\]\]\|<\|>\|&\|\\.?\\\|\/\/.?\/\/\|__.?__\|\#\#.?\#\#\|======.?======\|=====.?=====\|====.?====\|===.?===\|==.?==\|----\|---\|\n[\t\\|]+(-\|[0-9,a-z,A-Z]+\))?\|\{\{.?\}\}\|\b[A-Z][A-Z,a-z]+[:]([A-Z][a-z]+[A-Z,0-9][A-Z,a-z,0-9])\b\|\b([A-Z][a-z]+[A-Z,0-9][A-Z,a-z,0-9]*)\b\|\n)/ms";
15	}
16
17	function Process($source) {
18	$source = str_replace("\r", "", $source);
19	$source = trim($source)."\n";
20	$source = $this->ExecuteBuilder($source);
21	$source = preg_replace("/<br \/>$/", "", trim($source));
22	return $source;
23	}
24
25
26	function ExecuteBuilder($source) {
27	return preg_replace_callback($this->pattern, array(&$this, "Build"), $source);
28	}
29
30	function Build($things) {
31	$thing = $things[1];
32
33	static $oldIndentLevel = 0;
34	static $indentClosers = array();
35	static $br = 1;
36
37	global $wakka;
38
39	// convert HTML thingies
40	if ($thing == "<")
41	return "<";
42	else if ($thing == ">")
43	return ">";
44	else if ($thing == "&")
45	return "&";
46	// bold
47	else if (preg_match("/^\\(.)\\*$/s", $thing, $matches))
48	{
49	return "<b>".$this->ExecuteBuilder($matches[1])."</b>";
50	}
51	// italic
52	else if (preg_match("/^\/\/(.*)\/\/$/s", $thing, $matches))
53	{
54	return "<i>".$this->ExecuteBuilder($matches[1])."</i>";
55	}
56	// underlinue
57	else if (preg_match("/^__(.*)__$/s", $thing, $matches))
58	{
59	return "<u>".$this->ExecuteBuilder($matches[1])."</u>";
60	}
61	// monospace
62	else if (preg_match("/^\#\#(.*)\#\#$/s", $thing, $matches))
63	{
64	return "<tt>".$this->ExecuteBuilder($matches[1])."</tt>";
65	}
66	// headers
67	else if (preg_match("/^======(.*)======$/s", $thing, $matches))
68	{
69	$br = 0;
70	return "<h1>".$this->ExecuteBuilder($matches[1])."</h1>";
71	}
72	else if (preg_match("/^=====(.*)=====$/s", $thing, $matches))
73	{
74	$br = 0;
75	return "<h2>".$this->ExecuteBuilder($matches[1])."</h2>";
76	}
77	else if (preg_match("/^====(.*)====$/s", $thing, $matches))
78	{
79	$br = 0;
80	return "<h3>".$this->ExecuteBuilder($matches[1])."</h3>";
81	}
82	else if (preg_match("/^===(.*)===$/s", $thing, $matches))
83	{
84	$br = 0;
85	return "<h4>".$this->ExecuteBuilder($matches[1])."</h4>";
86	}
87	else if (preg_match("/^==(.*)==$/s", $thing, $matches))
88	{
89	$br = 0;
90	return "<h5>".$this->ExecuteBuilder($matches[1])."</h5>";
91	}
92	// separators
93	else if ($thing == "----")
94	{
95	// TODO: This could probably be improved for situations where someone puts text on the same line as a separator.
96	// Which is a stupid thing to do anyway! HAW HAW! Ahem.
97	$br = 0;
98	return "<hr/>";
99	}
100	// forced line breaks
101	else if ($thing == "---")
102	{
103	return "<br />";
104	}
105	// escaped text
106	else if (preg_match("/^\"\"(.*)\"\"$/s", $thing, $matches))
107	{
108	return $matches[1];
109	}
110	// code text
111	else if (preg_match("/^\%\%(.*)\%\%$/s", $thing, $matches))
112	{
113	// check if a language has been specified
114	$code = $matches[1];
115	if (preg_match("/^\((.+?)\)(.*)$/s", $code, $matches))
116	{
117	list(, $language, $code) = $matches;
118	}
119	switch ($language)
120	{
121	case "php":
122	$formatter = "php";
123	break;
124	default:
125	$formatter = "code";
126	}
127
128	$output = "<div class=\"code\">";
129	$output .= "CODE";//$wakka->Format(trim($code), $formatter);
130	$output .= "</div>";
131
132	return $output;
133	}
134	// forced links
135	else if (preg_match("/^\[\[(\S*)(\s+(.+))?\]\]$/", $thing, $matches))
136	{
137	list (, $url, , $text) = $matches;
138	if ($url)
139	{
140	if (!$text) $text = $url;
141	return $this->CreateLink($url, "", $text);
142	}
143	else
144	{
145	return "";
146	}
147	}
148	// indented text
149	else if (preg_match("/\n([\t\\|]+)(-\|([0-9,a-z,A-Z]+)\))?/s", $thing, $matches))
150	{
151	// new line
152	$result .= ($br ? "<br />\n" : "\n");
153
154	// we definitely want no line break in this one.
155	$br = 0;
156
157	// find out which indent type we want
158	$newIndentType = $matches[2];
159	if (!$newIndentType) { $opener = "<blockquote>"; $closer = "</blockquote>"; $br = 1; }
160	else if ($newIndentType == "-") { $opener = "<ul>"; $closer = "</ul>"; $li = 1; }
161	else { $opener = "<ol type=\"".$newIndentType."\">"; $closer = "</ol>"; $li = 1; }
162
163	// get new indent level
164	$newIndentLevel = strlen($matches[1]);
165	if ($newIndentLevel > $oldIndentLevel)
166	{
167	for ($i = 0; $i < $newIndentLevel - $oldIndentLevel; $i++)
168	{
169	$result .= $opener;
170	array_push($indentClosers, $closer);
171	}
172	}
173	else if ($newIndentLevel < $oldIndentLevel)
174	{
175	for ($i = 0; $i < $oldIndentLevel - $newIndentLevel; $i++)
176	{
177	$result .= array_pop($indentClosers);
178	}
179	}
180
181	$oldIndentLevel = $newIndentLevel;
182
183	if ($li) $result .= "<li/>";
184
185	return $result;
186	}
187	// new lines
188	else if ($thing == "\n")
189	{
190	// if we got here, there was no tab in the next line; this means that we can close all open indents.
191	$c = count($indentClosers);
192	for ($i = 0; $i < $c; $i++)
193	{
194	$result .= array_pop($indentClosers);
195	$br = 0;
196	}
197	$oldIndentLevel = 0;
198
199	$result .= ($br ? "<br />\n" : "\n");
200	$br = 1;
201	return $result;
202	}
203	// events
204	else if (preg_match("/^\{\{(.*?)\}\}$/s", $thing, $matches))
205	{
206	if ($matches[1])
207	return $this->engine->runAction($matches[1]);
208	else
209	return "{{}}";
210	}
211	// interwiki links!
212	else if (preg_match("/^[A-Z][A-Z,a-z]+[:]([A-Z][a-z]+[A-Z,0-9][A-Z,a-z,0-9]*)$/s", $thing))
213	{
214	return $this->CreateLink($thing);
215	}
216	// wakka links!
217	else if (preg_match("/^[A-Z][a-z]+[A-Z,0-9][A-Z,a-z,0-9]*$/s", $thing))
218	{
219	return $this->CreateLink($thing);
220	}
221	// if we reach this point, it must have been an accident.
222	return $thing;
223	}
224
225	function CreateLink($tag, $method = null, $text = null) {
226	$track = 1;
227	if (!$text) $text = $tag;
228
229	// is this an interwiki link?
230	if (preg_match("/^([A-Z][A-Z,a-z]+)[:]([A-Z][a-z]+[A-Z,0-9][A-Z,a-z,0-9]*)$/", $tag, $matches))
231	{
232	//$tag = $this->GetInterWikiUrl($matches[1], $matches[2]);
233	//return "<a href=\"$tag\">$text</a>";
234	return "--interwikiurl--";
235	}
236	// is this a full link? ie, does it contain alpha-numeric characters?
237	else if (preg_match("/[^[:alnum:]]/", $tag))
238	{
239	// check for email addresses
240	if (preg_match("/^.+\@.+$/", $tag))
241	{
242	$tag = "mailto:".$tag;
243	}
244	// check for protocol-less URLs
245	else if (!preg_match("/:/", $tag))
246	{
247	$tag = "http://".$tag;
248	}
249	return "<a href=\"$tag\">$text</a>";
250	}
251	else
252	{
253	// it's a Wakka link!
254	return ($this->engine->pageStorage->hasPage($tag) ? "<a href=\"".$this->engine->buildCommand($tag, $method)."\">".$text."</a>" : "<span class=\"missingpage\">".$text."<a href=\"".$this->engine->buildCommand($tag, "Create")."\">?</a></span>");
255	}
256	}
257
258	}
259
260	?>
MailToCvsAdmin">MailToCvsAdmin	ViewVC Help
Powered by ViewVC 1.1.26