[TASK] Merge submodule dbal into core
[Packages/TYPO3.CMS.git] / typo3 / sysext / openid / lib / php-openid / Auth / OpenID / Parse.php
1 <?php
2
3 /**
4 * This module implements a VERY limited parser that finds <link> tags
5 * in the head of HTML or XHTML documents and parses out their
6 * attributes according to the OpenID spec. It is a liberal parser,
7 * but it requires these things from the data in order to work:
8 *
9 * - There must be an open <html> tag
10 *
11 * - There must be an open <head> tag inside of the <html> tag
12 *
13 * - Only <link>s that are found inside of the <head> tag are parsed
14 * (this is by design)
15 *
16 * - The parser follows the OpenID specification in resolving the
17 * attributes of the link tags. This means that the attributes DO
18 * NOT get resolved as they would by an XML or HTML parser. In
19 * particular, only certain entities get replaced, and href
20 * attributes do not get resolved relative to a base URL.
21 *
22 * From http://openid.net/specs.bml:
23 *
24 * - The openid.server URL MUST be an absolute URL. OpenID consumers
25 * MUST NOT attempt to resolve relative URLs.
26 *
27 * - The openid.server URL MUST NOT include entities other than &amp;,
28 * &lt;, &gt;, and &quot;.
29 *
30 * The parser ignores SGML comments and <![CDATA[blocks]]>. Both kinds
31 * of quoting are allowed for attributes.
32 *
33 * The parser deals with invalid markup in these ways:
34 *
35 * - Tag names are not case-sensitive
36 *
37 * - The <html> tag is accepted even when it is not at the top level
38 *
39 * - The <head> tag is accepted even when it is not a direct child of
40 * the <html> tag, but a <html> tag must be an ancestor of the
41 * <head> tag
42 *
43 * - <link> tags are accepted even when they are not direct children
44 * of the <head> tag, but a <head> tag must be an ancestor of the
45 * <link> tag
46 *
47 * - If there is no closing tag for an open <html> or <head> tag, the
48 * remainder of the document is viewed as being inside of the
49 * tag. If there is no closing tag for a <link> tag, the link tag is
50 * treated as a short tag. Exceptions to this rule are that <html>
51 * closes <html> and <body> or <head> closes <head>
52 *
53 * - Attributes of the <link> tag are not required to be quoted.
54 *
55 * - In the case of duplicated attribute names, the attribute coming
56 * last in the tag will be the value returned.
57 *
58 * - Any text that does not parse as an attribute within a link tag
59 * will be ignored. (e.g. <link pumpkin rel='openid.server' /> will
60 * ignore pumpkin)
61 *
62 * - If there are more than one <html> or <head> tag, the parser only
63 * looks inside of the first one.
64 *
65 * - The contents of <script> tags are ignored entirely, except
66 * unclosed <script> tags. Unclosed <script> tags are ignored.
67 *
68 * - Any other invalid markup is ignored, including unclosed SGML
69 * comments and unclosed <![CDATA[blocks.
70 *
71 * PHP versions 4 and 5
72 *
73 * LICENSE: See the COPYING file included in this distribution.
74 *
75 * @access private
76 * @package OpenID
77 * @author JanRain, Inc. <openid@janrain.com>
78 * @copyright 2005-2008 Janrain, Inc.
79 * @license http://www.apache.org/licenses/LICENSE-2.0 Apache
80 */
81
82 /**
83 * Require Auth_OpenID::arrayGet().
84 */
85 require_once "Auth/OpenID.php";
86
87 class Auth_OpenID_Parse {
88
89 /**
90 * Specify some flags for use with regex matching.
91 */
92 var $_re_flags = "si";
93
94 /**
95 * Stuff to remove before we start looking for tags
96 */
97 var $_removed_re =
98 "<!--.*?-->|<!\[CDATA\[.*?\]\]>|<script\b(?!:)[^>]*>.*?<\/script>";
99
100 /**
101 * Starts with the tag name at a word boundary, where the tag name
102 * is not a namespace
103 */
104 var $_tag_expr = "<%s\b(?!:)([^>]*?)(?:\/>|>(.*)(?:<\/?%s\s*>|\Z))";
105
106 var $_attr_find = '\b(\w+)=("[^"]*"|\'[^\']*\'|[^\'"\s\/<>]+)';
107
108 var $_open_tag_expr = "<%s\b";
109 var $_close_tag_expr = "<((\/%s\b)|(%s[^>\/]*\/))>";
110
111 function Auth_OpenID_Parse()
112 {
113 $this->_link_find = sprintf("/<link\b(?!:)([^>]*)(?!<)>/%s",
114 $this->_re_flags);
115
116 $this->_entity_replacements = array(
117 'amp' => '&',
118 'lt' => '<',
119 'gt' => '>',
120 'quot' => '"'
121 );
122
123 $this->_attr_find = sprintf("/%s/%s",
124 $this->_attr_find,
125 $this->_re_flags);
126
127 $this->_removed_re = sprintf("/%s/%s",
128 $this->_removed_re,
129 $this->_re_flags);
130
131 $this->_ent_replace =
132 sprintf("&(%s);", implode("|",
133 $this->_entity_replacements));
134 }
135
136 /**
137 * Returns a regular expression that will match a given tag in an
138 * SGML string.
139 */
140 function tagMatcher($tag_name, $close_tags = null)
141 {
142 $expr = $this->_tag_expr;
143
144 if ($close_tags) {
145 $options = implode("|", array_merge(array($tag_name), $close_tags));
146 $closer = sprintf("(?:%s)", $options);
147 } else {
148 $closer = $tag_name;
149 }
150
151 $expr = sprintf($expr, $tag_name, $closer);
152 return sprintf("/%s/%s", $expr, $this->_re_flags);
153 }
154
155 function openTag($tag_name)
156 {
157 $expr = sprintf($this->_open_tag_expr, $tag_name);
158 return sprintf("/%s/%s", $expr, $this->_re_flags);
159 }
160
161 function closeTag($tag_name)
162 {
163 $expr = sprintf($this->_close_tag_expr, $tag_name, $tag_name);
164 return sprintf("/%s/%s", $expr, $this->_re_flags);
165 }
166
167 function htmlBegin($s)
168 {
169 $matches = array();
170 $result = preg_match($this->openTag('html'), $s,
171 $matches, PREG_OFFSET_CAPTURE);
172 if ($result === false || !$matches) {
173 return false;
174 }
175 // Return the offset of the first match.
176 return $matches[0][1];
177 }
178
179 function htmlEnd($s)
180 {
181 $matches = array();
182 $result = preg_match($this->closeTag('html'), $s,
183 $matches, PREG_OFFSET_CAPTURE);
184 if ($result === false || !$matches) {
185 return false;
186 }
187 // Return the offset of the first match.
188 return $matches[count($matches) - 1][1];
189 }
190
191 function headFind()
192 {
193 return $this->tagMatcher('head', array('body', 'html'));
194 }
195
196 function replaceEntities($str)
197 {
198 foreach ($this->_entity_replacements as $old => $new) {
199 $str = preg_replace(sprintf("/&%s;/", $old), $new, $str);
200 }
201 return $str;
202 }
203
204 function removeQuotes($str)
205 {
206 $matches = array();
207 $double = '/^"(.*)"$/';
208 $single = "/^\'(.*)\'$/";
209
210 if (preg_match($double, $str, $matches)) {
211 return $matches[1];
212 } else if (preg_match($single, $str, $matches)) {
213 return $matches[1];
214 } else {
215 return $str;
216 }
217 }
218
219 function match($regexp, $text, &$match)
220 {
221 if (!is_callable('mb_ereg_search_init')) {
222 if (!preg_match($regexp, $text, $match)) {
223 return false;
224 }
225 $match = $match[0];
226 return true;
227 }
228
229 $regexp = substr($regexp, 1, strlen($regexp) - 2 - strlen($this->_re_flags));
230 mb_ereg_search_init($text);
231 if (!mb_ereg_search($regexp)) {
232 return false;
233 }
234 $match = mb_ereg_search_getregs();
235 return true;
236 }
237
238 /**
239 * Find all link tags in a string representing a HTML document and
240 * return a list of their attributes.
241 *
242 * @todo This is quite ineffective and may fail with the default
243 * pcre.backtrack_limit of 100000 in PHP 5.2, if $html is big.
244 * It should rather use stripos (in PHP5) or strpos()+strtoupper()
245 * in PHP4 to manage this.
246 *
247 * @param string $html The text to parse
248 * @return array $list An array of arrays of attributes, one for each
249 * link tag
250 */
251 function parseLinkAttrs($html)
252 {
253 $stripped = preg_replace($this->_removed_re,
254 "",
255 $html);
256
257 $html_begin = $this->htmlBegin($stripped);
258 $html_end = $this->htmlEnd($stripped);
259
260 if ($html_begin === false) {
261 return array();
262 }
263
264 if ($html_end === false) {
265 $html_end = strlen($stripped);
266 }
267
268 $stripped = substr($stripped, $html_begin,
269 $html_end - $html_begin);
270
271 // Workaround to prevent PREG_BACKTRACK_LIMIT_ERROR:
272 $old_btlimit = ini_set( 'pcre.backtrack_limit', -1 );
273
274 // Try to find the <HEAD> tag.
275 $head_re = $this->headFind();
276 $head_match = array();
277 if (!$this->match($head_re, $stripped, $head_match)) {
278 ini_set( 'pcre.backtrack_limit', $old_btlimit );
279 return array();
280 }
281
282 $link_data = array();
283 $link_matches = array();
284
285 if (!preg_match_all($this->_link_find, $head_match[0],
286 $link_matches)) {
287 ini_set( 'pcre.backtrack_limit', $old_btlimit );
288 return array();
289 }
290
291 foreach ($link_matches[0] as $link) {
292 $attr_matches = array();
293 preg_match_all($this->_attr_find, $link, $attr_matches);
294 $link_attrs = array();
295 foreach ($attr_matches[0] as $index => $full_match) {
296 $name = $attr_matches[1][$index];
297 $value = $this->replaceEntities(
298 $this->removeQuotes($attr_matches[2][$index]));
299
300 $link_attrs[strtolower($name)] = $value;
301 }
302 $link_data[] = $link_attrs;
303 }
304
305 ini_set( 'pcre.backtrack_limit', $old_btlimit );
306 return $link_data;
307 }
308
309 function relMatches($rel_attr, $target_rel)
310 {
311 // Does this target_rel appear in the rel_str?
312 // XXX: TESTME
313 $rels = preg_split("/\s+/", trim($rel_attr));
314 foreach ($rels as $rel) {
315 $rel = strtolower($rel);
316 if ($rel == $target_rel) {
317 return 1;
318 }
319 }
320
321 return 0;
322 }
323
324 function linkHasRel($link_attrs, $target_rel)
325 {
326 // Does this link have target_rel as a relationship?
327 // XXX: TESTME
328 $rel_attr = Auth_OpeniD::arrayGet($link_attrs, 'rel', null);
329 return ($rel_attr && $this->relMatches($rel_attr,
330 $target_rel));
331 }
332
333 function findLinksRel($link_attrs_list, $target_rel)
334 {
335 // Filter the list of link attributes on whether it has
336 // target_rel as a relationship.
337 // XXX: TESTME
338 $result = array();
339 foreach ($link_attrs_list as $attr) {
340 if ($this->linkHasRel($attr, $target_rel)) {
341 $result[] = $attr;
342 }
343 }
344
345 return $result;
346 }
347
348 function findFirstHref($link_attrs_list, $target_rel)
349 {
350 // Return the value of the href attribute for the first link
351 // tag in the list that has target_rel as a relationship.
352 // XXX: TESTME
353 $matches = $this->findLinksRel($link_attrs_list,
354 $target_rel);
355 if (!$matches) {
356 return null;
357 }
358 $first = $matches[0];
359 return Auth_OpenID::arrayGet($first, 'href', null);
360 }
361 }
362
363 function Auth_OpenID_legacy_discover($html_text, $server_rel,
364 $delegate_rel)
365 {
366 $p = new Auth_OpenID_Parse();
367
368 $link_attrs = $p->parseLinkAttrs($html_text);
369
370 $server_url = $p->findFirstHref($link_attrs,
371 $server_rel);
372
373 if ($server_url === null) {
374 return false;
375 } else {
376 $delegate_url = $p->findFirstHref($link_attrs,
377 $delegate_rel);
378 return array($delegate_url, $server_url);
379 }
380 }
381