feedwordpress / syndicatedpostxpathquery.class.php

syndicatedpostxpathquery.class.php in FeedWordPress 2020.0118, at syndicatedpostxpathquery.class.php

463 lines 12.1 KB

Raw Download Zip

1	<?php
2	/**
3	* class SyndicatedPostXPathQuery: implements an XPath-like syntax used to query
4	* arbitrary elements within the syndicated item.
5	*
6	*/
7	class SyndicatedPostXPathQuery {
8	private $path;
9	private $parsedPath;
10	private $feed_type;
11	private $xmlns;
12	private $urlHash = array();
13
14	/**
15	* SyndicatedPostXPathQuery::__construct
16	*
17	* @param array $args
18	* @uses wp_parse_args
19	*
20	*/
21	public function __construct ($args = array()) {
22	if (is_string($args)) :
23	$args = array("path" => $args);
24	endif;
25
26	$args = wp_parse_args($args, array(
27	"path" => "",
28	));
29
30	$this->setPath($args['path']);
31	} /* SyndicatedPostXPathQuery::__construct() */
32
33	/**
34	* SyndicatedPostXPathQuery::getPath
35	*
36	* @param array $args
37	* @return mixed
38	*/
39	public function getPath ($args = array()) {
40	$args = wp_parse_args($args, array(
41	"parsed" => false,
42	));
43
44	return ($args['parsed'] ? $this->parsedPath : $this->path);
45	} /* SyndicatedPostXPathQuery::getPath () */
46
47	/**
48	* SyndicatedPostXPathQuery::setPath
49	*
50	* @param string $path
51	*/
52	public function setPath ($path) {
53	$this->urlHash = array();
54
55	$this->path = $path;
56
57	// Allow {url} notation for namespaces. URLs will contain : and /, so...
58	preg_match_all('/{([^}]+)}/', $path, $match, PREG_SET_ORDER);
59	foreach ($match as $ref) :
60	$this->urlHash[md5($ref[1])] = $ref[1];
61	endforeach;
62
63	foreach ($this->urlHash as $hash => $url) :
64	$path = str_replace('{'.$url.'}', '{#'.$hash.'}', $path);
65	endforeach;
66
67	$path = $this->parsePath(/cur=/ $path, /orig=/ $path);
68
69	$this->parsedPath = $path;
70
71	} /* SyndicatedPostXPathQuery::setPath() */
72
73	/**
74	* SyndicatedPostXPathQuery::snipSlug
75	*
76	* @return string
77	*/
78	protected function snipSlug ($path, $start, $n) {
79	$slug = substr($path, $start, ($n-$start));
80	if (strlen($slug) > 0) :
81	if (preg_match('/{#([^}]+)}/', $slug, $ref)) :
82	if (isset($this->urlHash[$ref[1]])) :
83	$slug = str_replace(
84	'{#'.$ref[1].'}',
85	'{'.$this->urlHash[$ref[1]].'}',
86	$slug
87	);
88	endif;
89	endif;
90	endif;
91	return $slug;
92	} /* SyndicatedPostXPathQuery::snipSlug () */
93
94	/**
95	* SyndicatedPostXPathQuery::parsePath ()
96	*
97	* @param mixed $path
98	* @param mixed $rootPath
99	* @return array\|object
100	*/
101	public function parsePath ($path, $rootPath) {
102	if (is_array($path)) :
103	// This looks like it's already been parsed.
104	$pp = $path;
105	else :
106	$pp = array();
107
108	// Okay let's parse this thing.
109	$n = 0; $start = 0; $state = 'slug';
110	while ($state != '$') :
111	switch ($state) :
112	case 'slash' :
113	$slug = $this->snipSlug($path, $start, $n);
114	if (strlen($slug) > 0) :
115	$pp[] = $slug;
116	endif;
117
118	$n++;
119	// don't include the slash in our next slug
120	$start = $n;
121
122	$state = (($n < strlen($path)) ? 'slug' : '$');
123
124	break;
125	case 'brackets' :
126
127	// first, snip off what we've consumed so far
128	$slug = $this->snipSlug($path, $start, $n);
129	if (strlen($slug) > 0) :
130	$pp[] = $slug;
131	endif;
132
133	// now, chase the ]
134	$depth = 1;
135	$n++; $start = $n;
136
137	// find the end of the [square-bracketed] expression
138	while ($depth > 0 and $n != '') :
139	$tok = ((strlen($path) > $n) ? $path[$n] : '');
140	switch ($tok) :
141	case '' :
142	// ERROR STATE: syntax error
143	$depth = -1;
144	$state = 'syntax-error';
145	break;
146	case '[' :
147	$depth++;
148	break;
149	case ']' :
150	$depth--;
151	break;
152	default :
153	// NOOP
154	endswitch;
155	$n++;
156	endwhile;
157
158	if ($state != 'syntax-error') :
159	$bracketed = substr($path, $start, ($n-$start)-1);
160
161	// recursive parsing
162	$oFilter = new stdClass;
163	$oFilter->verb = 'has';
164	$oFilter->query = $this->parsePath($bracketed, $rootPath);
165	$pp[] = $oFilter;
166
167	$start = $n;
168
169	$state = 'slash-expected';
170	endif;
171	break;
172
173	case 'slash-expected' :
174	$tok = ((strlen($path) > $n) ? $path[$n] : '');
175	if ($tok == '/' or $tok == '') :
176	$state = 'slash';
177	else :
178	$state = 'syntax-error';
179	endif;
180	break;
181	case 'syntax-error' :
182	$pp = new WP_Error('xpath', __("Syntax error", "feedwordpress"));
183	$state = '$';
184	break;
185	case 'slug' :
186	default :
187	$tok = ((strlen($path) > $n) ? $path[$n] : '');
188	switch ($tok) :
189	case '' :
190	case '/' :
191	$state = 'slash';
192	break;
193	case '[' :
194	$state = 'brackets';
195	break;
196	default :
197	$n++;
198	endswitch;
199	endswitch;
200	endwhile;
201	endif;
202	return $pp;
203	} /* SyndicatedPostXPathQuery::parsePath() */
204
205	/**
206	* SyndicatedPostXPathQuery::match
207	*
208	* @param string $path
209	* @return array
210	*/
211	public function match ($r = array()) {
212	$path = $this->parsedPath;
213
214	$r = wp_parse_args($r, array(
215	"type" => SIMPLEPIE_TYPE_ATOM_10,
216	"xmlns" => array(),
217	"map" => array(),
218	"context" => array(),
219	"parent" => array(),
220	"format" => "string",
221	));
222
223	$this->feed_type = $r['type'];
224	$this->xmlns = $r['xmlns'];
225
226	// Start out with a get_item_tags query.
227	$node = '';
228	while (strlen($node)==0 and !is_null($node)) :
229	$node = array_shift($path);
230	endwhile;
231
232	if (is_string($node) and isset($r['map'][$node])) :
233	$data = $r['map'][$node];
234	$node = array_shift($path);
235	else :
236	$data = $r['map']['/'];
237	endif;
238
239	$matches = $data;
240	while (!is_null($node)) :
241	if (is_object($node) OR strlen($node) > 0) :
242	list($axis, $element) = $this->xpath_name_and_axis($node);
243	if ('self'==$axis) :
244	if (is_object($element) and property_exists($element, 'verb')) :
245
246	$subq = new self(array("path" => $element->query));
247	$result = $subq->match(array(
248	"type" => $r['type'],
249	"xmlns" => $r['xmlns'],
250	"map" => array(
251	"/" => $matches,
252	),
253	"context" => $matches,
254	"parent" => $r['parent'],
255	"format" => "object",
256	));
257
258	// when format = 'object' we should get back
259	// a sparse array of arrays, with indices = indices
260	// from the input array, each element = an array of
261	// one or more matching elements
262
263	if ($element->verb = 'has' and is_array($result)) :
264
265	$results = array();
266	foreach (array_keys($result) as $a) :
267	$results[$a] = $matches[$a];
268	endforeach;
269
270	$matches = $results;
271	$data = $matches;
272	endif;
273
274	elseif (is_numeric($node)) :
275
276	// according to W3C, sequence starts at position 1, not 0
277	// so subtract 1 to line up with PHP array starting at 0
278	$idx = intval($element) - 1;
279	if (isset($matches[$idx])) :
280	$data = array($idx => $matches[$idx]);
281	else :
282	$data = array();
283	endif;
284
285	$matches = array($idx => $data);
286	endif;
287
288	else :
289	$matches = array();
290
291	foreach ($data as $idx => $datum) :
292	if (!is_string($datum) and isset($datum[$axis])) :
293	foreach ($datum[$axis] as $ns => $elements) :
294	if (isset($elements[$element])) :
295	// Potential match.
296	// Check namespace.
297	if (is_string($elements[$element])) : // Attribute
298	$addenda = array($elements[$element]);
299	$contexts = array($datum);
300
301	// Element
302	else :
303	$addenda = $elements[$element];
304	$contexts = $elements[$element];
305	endif;
306
307	foreach ($addenda as $index => $addendum) :
308	$context = $contexts[$index];
309
310	$namespaces = $this->xpath_possible_namespaces($node, $context);
311	if (in_array($ns, $namespaces)) :
312	$matches[] = $addendum;
313	endif;
314	endforeach;
315	endif;
316	endforeach;
317	endif;
318	endforeach;
319
320	$data = $matches;
321	endif;
322	endif;
323	$node = array_shift($path);
324	endwhile;
325
326	$matches = array();
327	foreach ($data as $idx => $datum) :
328	if ($r['format'] == 'string') :
329	if (is_string($datum)) :
330	$matches[] = $datum;
331	elseif (isset($datum['data'])) :
332	$matches[] = $datum['data'];
333	endif;
334	else :
335	$matches[$idx] = $datum;
336	endif;
337	endforeach;
338
339	return $matches;
340	} /* SyndicatedPostXPathQuery::match() */
341
342	public function xpath_default_namespace () {
343	// Get the default namespace.
344	$type = $this->feed_type;
345	if ($type & SIMPLEPIE_TYPE_ATOM_10) :
346	$defaultNS = SIMPLEPIE_NAMESPACE_ATOM_10;
347	elseif ($type & SIMPLEPIE_TYPE_ATOM_03) :
348	$defaultNS = SIMPLEPIE_NAMESPACE_ATOM_03;
349	elseif ($type & SIMPLEPIE_TYPE_RSS_090) :
350	$defaultNS = SIMPLEPIE_NAMESPACE_RSS_090;
351	elseif ($type & SIMPLEPIE_TYPE_RSS_10) :
352	$defaultNS = SIMPLEPIE_NAMESPACE_RSS_10;
353	elseif ($type & SIMPLEPIE_TYPE_RSS_20) :
354	$defaultNS = SIMPLEPIE_NAMESPACE_RSS_20;
355	else :
356	$defaultNS = SIMPLEPIE_NAMESPACE_RSS_20;
357	endif;
358	return $defaultNS;
359	} /* SyndicatedPostXPathQuery::xpath_default_namespace() */
360
361	public function xpath_name_and_axis ($node) {
362	$ns = NULL; $element = NULL;
363
364	$axis = 'child'; // "In effect, `child` is the default axis."
365	if (is_object($node) and property_exists($node, 'verb')):
366	if ('has'==$node->verb) :
367	$axis = 'self';
368	endif;
369	elseif (strpos($node, '::') !== false) :
370	list($axis, $node) = explode("::", $node, 2);
371	if ($axis=='attribute') :
372	$axis = 'attribs'; // map from W3C to SimplePie's idiosyncratic notation
373	endif;
374	elseif (substr($node, 0, 1)=='@') :
375	$axis = 'attribs'; $node = substr($node, 1);
376	elseif (is_numeric($node)) :
377	$axis = 'self';
378	elseif (substr($node, 0, 1)=='/') :
379	// FIXME: properly, we should check for // and if we have it,
380	// treat that as short for /descendent-or-self::node()/
381	$axis = 'child'; $node = substr($node, 1);
382	else :
383	// NOOP
384	endif;
385
386	if (is_string($node) and preg_match('/^{([^}])}(.)$/', $node, $ref)) :
387	$element = $ref[2];
388	elseif (is_string($node) and strpos($node, ':') !== FALSE) :
389	list($xmlns, $element) = explode(':', $node, 2);
390	else :
391	$element = $node;
392	endif;
393	return array($axis, $element);
394	} /* SyndicatedPostXPathQuery::xpath_name_and_axis () */
395
396	public function xpath_possible_namespaces ($node, $datum = array()) {
397	$ns = NULL; $element = NULL;
398
399	if (substr($node, 0, 1)=='@') :
400	$attr = '@'; $node = substr($node, 1);
401	else :
402	$attr = '';
403	endif;
404
405	if (preg_match('/^{([^}])}(.)$/', $node, $ref)) :
406	$ns = array($ref[1]);
407	elseif (strpos($node, ':') !== FALSE) :
408	list($xmlns, $element) = explode(':', $node, 2);
409
410	if (isset($this->xmlns['reverse'][$xmlns])) :
411	$ns = $this->xmlns['reverse'][$xmlns];
412	else :
413	$ns = array($xmlns);
414	endif;
415
416	// Fucking SimplePie. For attributes in default xmlns.
417	$defaultNS = $this->xpath_default_namespace();
418	if (isset($this->xmlns['forward'][$defaultNS])
419	and ($xmlns==$this->xmlns['forward'][$defaultNS])) :
420	$ns[] = '';
421	endif;
422
423	if (isset($datum['xmlns'])) :
424	if (isset($datum['xmlns'][$xmlns])) :
425	$ns[] = $datum['xmlns'][$xmlns];
426	endif;
427	endif;
428	else :
429	// Often in SimplePie, the default namespace gets stored
430	// as an empty string rather than a URL.
431	$ns = array($this->xpath_default_namespace(), '');
432	endif;
433	return array_unique($ns);
434	} /* SyndicatedPostXPathQuery::xpath_possible_namespaces() */
435
436	} /* class SyndicatedPostXPathQuery */
437
438	// When called directly, run through and perform some tests.
439	if (basename($_SERVER['SCRIPT_FILENAME'])==basename(__FILE__)) :
440	# some day when I am a grown-up developer I might include
441	# some test cases in this here section
442	# we need to implement wp_parse_args(), __(), and class WP_Error ...
443	#function wp_parse_args ($r, $defaults) {
444	# return array_merge($defaults, $r);
445	#}
446	#function __($text, $domain) {
447	# return $text;
448	#}
449	#class WP_Error {
450	# public function __construct ( $slug, $message ) {
451	# /DBG/ echo $slug;
452	# /DBG/ echo ": ";
453	# /DBG/ echo $message;
454	# }
455	#}
456	#
457	#header("Content-type: text/plain");
458	#
459	#$spxq = new SyndicatedPostXPathQuery(array("path" => $_REQUEST['p']));
460	#
461	#var_dump($spxq);
462	endif;
463