internal-links / core / linkbuilder.php

linkbuilder.php in Internal Link Juicer: SEO Auto Linker for WordPress trunk, at core/linkbuilder.php

430 lines 11.3 KB

Raw Download Zip

1	<?php
2	namespace ILJ\Core;
3
4	use ILJ\Core\Links\Text_To_Link_Converter_Interface;
5	use ILJ\Core\Links\Timeout_Monitor_Layer;
6	use ILJ\Core\Options;
7	use ILJ\Helper\Encoding;
8	use ILJ\Type\Ruleset;
9	use ILJ\Database\Linkindex;
10	use ILJ\Helper\Replacement;
11	use ILJ\Backend\Editor;
12	use ILJ\Database\LinkindexIndividualTemp;
13	use ILJ\Database\LinkindexTemp;
14	use ILJ\Helper\IndexAsset;
15	use ILJ\Helper\LinkBuilding;
16
17	/**
18	* The below line sets the `ticks` directive to prevent timeout occurring on the frontend.
19	* This is set for the complete file, @see https://www.php.net/manual/en/control-structures.declare.php
20	* This is needed for {@link Timeout_Monitor_Layer} to work, You might wonder why I didn't set this
21	* on {@link Timeout_Monitor_Layer}. The reason for this is that the ticks won't be propagated to code in other files.
22	* For example,
23	*
24	* `declare(ticks=1000) {
25	* $a = 20;
26	* $foo = new Foo()
27	* $foo->bar();
28	* }`
29	*
30	* You might expect all statements inside bar() method would be tickable, but it won't behave like that.
31	* You would need to set the ticks directive on the top of Foo.php in order to get
32	* the same behavior.
33	*/
34	declare(ticks=1000);
35
36	/**
37	* The main LinkBuilder class
38	*
39	* Is responsible for editing a piece of content and setting links within by a given Ruleset
40	*
41	* @package ILJ\Core
42	* @since 1.0.0
43	*/
44	class LinkBuilder implements Text_To_Link_Converter_Interface {
45
46	/**
47	* ID (post id / term id)
48	*
49	* @var int
50	* @since 1.0.0
51	*/
52	private $id = null;
53
54	/**
55	* Type (post/term)
56	*
57	* @var string
58	* @since 1.0.1
59	*/
60	private $type = null;
61
62	/**
63	* Link Ruleset
64	*
65	* @var Ruleset
66	* @since 1.0.0
67	*/
68	private $link_ruleset = null;
69
70	/**
71	* Replace Ruleset
72	*
73	* @var Ruleset
74	* @since 1.0.0
75	*/
76	private $replace_ruleset = null;
77
78	/**
79	* Content
80	*
81	* @var string
82	* @since 1.0.0
83	*/
84	private $content = '';
85
86	/**
87	* Link page count
88	*
89	* @var int
90	*/
91	private $link_page_count = 0;
92
93	/**
94	* Link target count
95	*
96	* @var array
97	*/
98	private $link_target_count = array();
99
100	/**
101	* Used pattern
102	*
103	* @var array
104	*/
105	private $used_pattern = array();
106
107	/**
108	* Link count per paragraph
109	*
110	* @var int
111	*/
112	private $linkcount_per_paragraph = 0;
113
114	/**
115	* Meta Option. Determines if greedy mode is on/off. Enables linking as often as possible.
116	*
117	* @var bool
118	*/
119	private $multi_keyword_mode = false;
120
121	/**
122	* Meta Option. Maximum number of outgoing links to be created in the current post/page. Zero means no limit
123	*
124	* @var int
125	*/
126	private $links_per_page = 0;
127
128	/**
129	* Meta option. Maximum number of links to be created in the current target. Zero means no limit
130	*
131	* @var int
132	*/
133	private $links_per_target = 0;
134
135	/**
136	* Meta option. Determine if it should go deeper into the paragraph or just continue
137	*
138	* @var int
139	*/
140	private $links_per_paragraph_switch = 0;
141
142	/**
143	* Maximum number of links to be created in the current paragraph. Zero means no limit
144	*
145	* @var int
146	*/
147	private $links_per_paragraph = 0;
148
149	/**
150	* The content type in which link replacements take place, defaults to html.
151	*
152	* @var boolean
153	*/
154	private $content_type;
155
156	/**
157	* Constructor of ILJ_LinkBuilder
158	*
159	* @param int $id The ID of the current subject
160	* @param string $type The type of the current subject
161	* @param string $build_type
162	* @param boolean $content_type The content type in which link replacements take place, defaults to html.
163	*
164	* @return void
165	* @since 1.0.1
166	*/
167	public function __construct($id, $type, $build_type = null, $content_type = 'html') {
168	$this->id = $id;
169	$this->type = $type;
170	$this->replace_ruleset = new Ruleset();
171	$this->multi_keyword_mode = (bool) Options::getOption(\ILJ\Core\Options\MultipleKeywords::getKey());
172	$this->links_per_page = (false === $this->multi_keyword_mode) ? Options::getOption(\ILJ\Core\Options\LinksPerPage::getKey()) : 0;
173	$this->links_per_target = (false === $this->multi_keyword_mode) ? Options::getOption(\ILJ\Core\Options\LinksPerTarget::getKey()) : 0;
174
175	$this->content_type = $content_type;
176
177	$this->setupInLinks($build_type);
178
179	}
180
181	/**
182	* Loads all ingoing links to current content from linkindex table and sets a new Ruleset type with it
183	*
184	* @since 1.0.0
185	* @param string $build_type
186	* @return void
187	*/
188	private function setupInLinks($build_type = null) {
189	$this->link_ruleset = new Ruleset();
190	$this->link_page_count = 0;
191	$this->link_target_count = array();
192	$this->used_pattern = array();
193
194
195
196	if (is_null($build_type)) {
197	$post_rules = Linkindex::getRules($this->id, $this->type);
198	} elseif (IndexAsset::ILJ_FULL_BUILD == $build_type) {
199	$post_rules = LinkindexTemp::getRules($this->id, $this->type);
200	} elseif (IndexAsset::ILJ_INDIVIDUAL_BUILD == $build_type) {
201	$post_rules = LinkindexIndividualTemp::getRules($this->id, $this->type);
202	}
203
204	foreach ($post_rules as $post_rule) {
205	$this->link_ruleset->addRule($post_rule->anchor, $post_rule->link_to, $post_rule->type_to);
206	}
207
208	}
209
210	/**
211	* Return the linked content.
212	*
213	* @param mixed $content The content value passed to cache link value into
214	* @return mixed Mixed return value to cater different content compatibilities
215	*/
216	public function link_content($content) {
217	return $this->linkContent($content);
218	}
219
220	/**
221	* Applies the link rules to a given piece of content
222	*
223	* @deprecated
224	* @since 1.0.0
225	* @param string $content The content of the post, where the rules get applied
226	* @return string
227	*/
228	public function linkContent($content) {
229	if (!LinkBuilding::is_filter_needed($this->id, $this->type)) {
230	return $content;
231	}
232	$this->content = $content;
233	$this->replace_ruleset = Replacement::mask($this->content);
234	if ('' != $this->content) {
235	$this->maskLinkRules();
236	$this->applyReplaceRules();
237	}
238	return $this->content;
239	}
240
241	/**
242	* Applies the given LinkRuleset for the document through masking within the content.
243	*
244	* @since 1.0.0
245	* @return void
246	*/
247	private function maskLinkRules() {
248	$link_page_count = 0;
249	$link_target_count = array();
250	$outgoing_links_limit = 0;
251	if ($this->links_per_page > 0) {
252	$outgoing_links_limit = $this->links_per_page;
253	}
254
255	{
256	$temp_values = $this->createLinkIndex($this->content, 0, $link_page_count, $link_target_count, 0, false, $outgoing_links_limit);
257	$full_content = $temp_values['content'];
258	$this->link_ruleset->reset();
259	}
260	$this->content = $full_content;
261	}
262
263	/**
264	* Runs Logical Checks For Linking Rules and settings
265	*
266	* @since 1.2.15
267	* @param mixed $content
268	* @param mixed $index
269	* @param mixed $link_page_count
270	* @param mixed $link_target_count
271	* @param mixed $linkcount_per_paragraph
272	* @param mixed $loop_thru_per_paragraph
273	* @param mixed $outgoing_links_limit
274	* @return array
275	*/
276	protected function createLinkIndex($content, $index, $link_page_count, $link_target_count, $linkcount_per_paragraph, $loop_thru_per_paragraph, $outgoing_links_limit) {
277	while ($this->link_ruleset->hasRule()) {
278	$link_rule = $this->link_ruleset->getRule();
279
280	if (0 != $outgoing_links_limit && 0 < $outgoing_links_limit) {
281	if ($link_page_count == $outgoing_links_limit) {
282	break;
283	}
284	}
285
286	if (($this->links_per_target > 0 && array_key_exists($link_rule->value, $link_target_count) && $link_target_count[$link_rule->value] >= $this->links_per_target)
287	\|\| (!$this->multi_keyword_mode && in_array($link_rule->pattern, $this->used_pattern))
288	) {
289	$this->link_ruleset->nextRule();
290	continue;
291	}
292
293
294
295	$pattern = wptexturize($link_rule->pattern);
296	$pattern = Encoding::escape_ascii($pattern);
297	if (Options::getOption(Options\Case_Sensitive_Mode_Switch::getKey())) {
298	preg_match_all('/' . Encoding::mask_pattern($pattern) . '/u', $content, $rule_match);
299	} else {
300	preg_match_all('/' . Encoding::mask_pattern($pattern) . '/ui', $content, $rule_match);
301	}
302
303	if (!isset($rule_match['phrase']) \|\| !count($rule_match['phrase'])) {
304	$this->link_ruleset->nextRule();
305	continue;
306	}
307	$phrases = $rule_match['phrase'];
308	foreach ($phrases as $rule) {
309	if (($this->links_per_target > 0
310	&& array_key_exists($link_rule->value, $link_target_count)
311	&& $link_target_count[$link_rule->value] == $this->links_per_target)
312	\|\| (!$this->multi_keyword_mode && in_array($link_rule->pattern, $this->used_pattern))
313	) {
314	$this->link_ruleset->nextRule();
315	continue 2;
316	}
317
318
319	$rule_id = 'ilj_' . uniqid('', true);
320	$link = $this->generateLink($link_rule, esc_html($rule));
321	if (!$link) {
322	$this->link_ruleset->nextRule();
323	continue;
324	}
325	$rule = Encoding::escape_ascii($rule);
326	if (Options::getOption(Options\Case_Sensitive_Mode_Switch::getKey())) {
327	$content = preg_replace('/' . Encoding::mask_pattern($rule) . '/u', $rule_id, $content, 1);
328	} else {
329	$content = preg_replace('/' . Encoding::mask_pattern($rule) . '/ui', $rule_id, $content, 1);
330	}
331	$this->replace_ruleset->addRule($rule_id, $link);
332
333	if (!array_key_exists($link_rule->value, $this->link_target_count)) {
334	$link_target_count[$link_rule->value] = 0;
335	}
336
337	$this->used_pattern[] = $link_rule->pattern;
338	$link_target_count[$link_rule->value]++;
339	$link_page_count++;
340
341	}
342	$this->link_ruleset->nextRule();
343	}
344	$obj = array(
345	'content' => $content,
346	'link_target_count' => $link_target_count,
347	'link_page_count' => $link_page_count,
348	);
349	return $obj;
350	}
351
352	/**
353	* Applies the configured masks and replaces the placeholders with the generated links.
354	*
355	* @since 1.0.0
356	* @return void
357	*/
358	private function applyReplaceRules() {
359	while ($this->replace_ruleset->hasRule()) {
360	$replace_rule = $this->replace_ruleset->getRule();
361	$this->content = str_replace($replace_rule->pattern, $replace_rule->value, $this->content);
362	$this->replace_ruleset->nextRule();
363	}
364
365	if (preg_match('/ilj\_[a-z0-9]{14}\.[0-9]{8}/', $this->content)) {
366	$this->replace_ruleset->reset();
367	$this->applyReplaceRules();
368	}
369	}
370
371	/**
372	* Returns the template for link output
373	*
374	* @since 1.0.0
375	* @return string
376	*/
377	private function getLinkTemplate() {
378	$default_template = \ILJ\Core\Options\LinkOutputInternal::getDefault();
379	$template = Options::getOption(\ILJ\Core\Options\LinkOutputInternal::getKey());
380
381	if ('' == $template) {
382	return $default_template;
383	}
384	return wp_specialchars_decode($template, \ENT_QUOTES);
385	}
386
387	/**
388	* Generates the link markup
389	*
390	* @since 1.0.0
391	* @param string $link_rule The post where the link should point to
392	* @param string $anchor The anchortext for the link
393	* @return bool\|string
394	*/
395	private function generateLink($link_rule, $anchor) {
396	$template = $this->getLinkTemplate();
397	$nofollow = (bool) Options::getOption(\ILJ\Core\Options\InternalNofollow::getKey());
398	$link_attrs = array();
399
400	if ('post' == $link_rule->type) {
401	if (get_post_status($link_rule->value) != 'publish') {
402	return false;
403	}
404	$url = get_the_permalink($link_rule->value);
405	}
406
407
408
409	$link = str_replace('{{url}}', (isset($url) ? $url : '#'), $template);
410	$link = str_replace('{{anchor}}', $anchor, $link);
411
412
413
414	$check_nofollow = true;
415
416
417
418	if ($check_nofollow && $nofollow) {
419	$link = str_replace('<a ', '<a rel="nofollow" ', $link);
420	}
421
422	if ('json' === $this->content_type) {
423	// If the content is json, the link should be escaped before replacement.
424	$link = trim(wp_json_encode($link), '"');
425	}
426
427	return $link;
428	}
429	}
430