aboutsummaryrefslogtreecommitdiffhomepage
path: root/inc/3rdparty/libraries/MOBIClass/readability
diff options
context:
space:
mode:
authorThomas Citharel <tcit@tcit.fr>2014-12-14 00:16:39 +0100
committerThomas Citharel <tcit@tcit.fr>2014-12-14 00:16:39 +0100
commit1256e4c645c86f4165d052bf8116b957b373b837 (patch)
tree3cc3bd960df403a6e0645afcd9e4f5bbb68bfc0e /inc/3rdparty/libraries/MOBIClass/readability
parent9c55ed0923273da886497cb62302a79622e34a74 (diff)
parentdc69d3e8d895c87f9d26c2d1309e40d6090d4c8d (diff)
downloadwallabag-1256e4c645c86f4165d052bf8116b957b373b837.tar.gz
wallabag-1256e4c645c86f4165d052bf8116b957b373b837.tar.zst
wallabag-1256e4c645c86f4165d052bf8116b957b373b837.zip
Merge branch 'tcitworld-ebook' into dev
Diffstat (limited to 'inc/3rdparty/libraries/MOBIClass/readability')
-rw-r--r--inc/3rdparty/libraries/MOBIClass/readability/JSLikeHTMLElement.php110
-rw-r--r--inc/3rdparty/libraries/MOBIClass/readability/Readability.php1069
2 files changed, 1179 insertions, 0 deletions
diff --git a/inc/3rdparty/libraries/MOBIClass/readability/JSLikeHTMLElement.php b/inc/3rdparty/libraries/MOBIClass/readability/JSLikeHTMLElement.php
new file mode 100644
index 00000000..1a8ec88c
--- /dev/null
+++ b/inc/3rdparty/libraries/MOBIClass/readability/JSLikeHTMLElement.php
@@ -0,0 +1,110 @@
1<?php
2/**
3* JavaScript-like HTML DOM Element
4*
5* This class extends PHP's DOMElement to allow
6* users to get and set the innerHTML property of
7* HTML elements in the same way it's done in
8* JavaScript.
9*
10* Example usage:
11* @code
12* require_once 'JSLikeHTMLElement.php';
13* header('Content-Type: text/plain');
14* $doc = new DOMDocument();
15* $doc->registerNodeClass('DOMElement', 'JSLikeHTMLElement');
16* $doc->loadHTML('<div><p>Para 1</p><p>Para 2</p></div>');
17* $elem = $doc->getElementsByTagName('div')->item(0);
18*
19* // print innerHTML
20* echo $elem->innerHTML; // prints '<p>Para 1</p><p>Para 2</p>'
21* echo "\n\n";
22*
23* // set innerHTML
24* $elem->innerHTML = '<a href="http://fivefilters.org">FiveFilters.org</a>';
25* echo $elem->innerHTML; // prints '<a href="http://fivefilters.org">FiveFilters.org</a>'
26* echo "\n\n";
27*
28* // print document (with our changes)
29* echo $doc->saveXML();
30* @endcode
31*
32* @author Keyvan Minoukadeh - http://www.keyvan.net - keyvan@keyvan.net
33* @see http://fivefilters.org (the project this was written for)
34*/
35class JSLikeHTMLElement extends DOMElement
36{
37 /**
38 * Used for setting innerHTML like it's done in JavaScript:
39 * @code
40 * $div->innerHTML = '<h2>Chapter 2</h2><p>The story begins...</p>';
41 * @endcode
42 */
43 public function __set($name, $value) {
44 if ($name == 'innerHTML') {
45 // first, empty the element
46 for ($x=$this->childNodes->length-1; $x>=0; $x--) {
47 $this->removeChild($this->childNodes->item($x));
48 }
49 // $value holds our new inner HTML
50 if ($value != '') {
51 $f = $this->ownerDocument->createDocumentFragment();
52 // appendXML() expects well-formed markup (XHTML)
53 $result = @$f->appendXML($value); // @ to suppress PHP warnings
54 if ($result) {
55 if ($f->hasChildNodes()) $this->appendChild($f);
56 } else {
57 // $value is probably ill-formed
58 $f = new DOMDocument();
59 $value = mb_convert_encoding($value, 'HTML-ENTITIES', 'UTF-8');
60 // Using <htmlfragment> will generate a warning, but so will bad HTML
61 // (and by this point, bad HTML is what we've got).
62 // We use it (and suppress the warning) because an HTML fragment will
63 // be wrapped around <html><body> tags which we don't really want to keep.
64 // Note: despite the warning, if loadHTML succeeds it will return true.
65 $result = @$f->loadHTML('<htmlfragment>'.$value.'</htmlfragment>');
66 if ($result) {
67 $import = $f->getElementsByTagName('htmlfragment')->item(0);
68 foreach ($import->childNodes as $child) {
69 $importedNode = $this->ownerDocument->importNode($child, true);
70 $this->appendChild($importedNode);
71 }
72 } else {
73 // oh well, we tried, we really did. :(
74 // this element is now empty
75 }
76 }
77 }
78 } else {
79 $trace = debug_backtrace();
80 trigger_error('Undefined property via __set(): '.$name.' in '.$trace[0]['file'].' on line '.$trace[0]['line'], E_USER_NOTICE);
81 }
82 }
83
84 /**
85 * Used for getting innerHTML like it's done in JavaScript:
86 * @code
87 * $string = $div->innerHTML;
88 * @endcode
89 */
90 public function __get($name)
91 {
92 if ($name == 'innerHTML') {
93 $inner = '';
94 foreach ($this->childNodes as $child) {
95 $inner .= $this->ownerDocument->saveXML($child);
96 }
97 return $inner;
98 }
99
100 $trace = debug_backtrace();
101 trigger_error('Undefined property via __get(): '.$name.' in '.$trace[0]['file'].' on line '.$trace[0]['line'], E_USER_NOTICE);
102 return null;
103 }
104
105 public function __toString()
106 {
107 return '['.$this->tagName.']';
108 }
109}
110?> \ No newline at end of file
diff --git a/inc/3rdparty/libraries/MOBIClass/readability/Readability.php b/inc/3rdparty/libraries/MOBIClass/readability/Readability.php
new file mode 100644
index 00000000..91554243
--- /dev/null
+++ b/inc/3rdparty/libraries/MOBIClass/readability/Readability.php
@@ -0,0 +1,1069 @@
1<?php
2/**
3* Arc90's Readability ported to PHP for FiveFilters.org
4* Based on readability.js version 1.7.1 (without multi-page support)
5* ------------------------------------------------------
6* Original URL: http://lab.arc90.com/experiments/readability/js/readability.js
7* Arc90's project URL: http://lab.arc90.com/experiments/readability/
8* JS Source: http://code.google.com/p/arc90labs-readability
9* Ported by: Keyvan Minoukadeh, http://www.keyvan.net
10* More information: http://fivefilters.org/content-only/
11* License: Apache License, Version 2.0
12* Requires: PHP5
13* Date: 2010-10-29
14*
15* Differences between the PHP port and the original
16* ------------------------------------------------------
17* Arc90's Readability is designed to run in the browser. It works on the DOM
18* tree (the parsed HTML) after the page's CSS styles have been applied and
19* Javascript code executed. This PHP port does not run inside a browser.
20* We use PHP's ability to parse HTML to build our DOM tree, but we cannot
21* rely on CSS or Javascript support. As such, the results will not always
22* match Arc90's Readability. (For example, if a web page contains CSS style
23* rules or Javascript code which hide certain HTML elements from display,
24* Arc90's Readability will dismiss those from consideration but our PHP port,
25* unable to understand CSS or Javascript, will not know any better.)
26*
27* Another significant difference is that the aim of Arc90's Readability is
28* to re-present the main content block of a given web page so users can
29* read it more easily in their browsers. Correct identification, clean up,
30* and separation of the content block is only a part of this process.
31* This PHP port is only concerned with this part, it does not include code
32* that relates to presentation in the browser - Arc90 already do
33* that extremely well, and for PDF output there's FiveFilters.org's
34* PDF Newspaper: http://fivefilters.org/pdf-newspaper/.
35*
36* Finally, this class contains methods that might be useful for developers
37* working on HTML document fragments. So without deviating too much from
38* the original code (which I don't want to do because it makes debugging
39* and updating more difficult), I've tried to make it a little more
40* developer friendly. You should be able to use the methods here on
41* existing DOMElement objects without passing an entire HTML document to
42* be parsed.
43*/
44
45// This class allows us to do JavaScript like assignements to innerHTML
46require_once(dirname(__FILE__).'/JSLikeHTMLElement.php');
47
48// Alternative usage (for testing only!)
49// uncomment the lins below and call Readability.php in your browser
50// passing it the URL of the page you'd like content from, e.g.:
51// Readability.php?url=http://medialens.org/alerts/09/090615_the_guardian_climate.php
52
53/*
54if (!isset($_GET['url']) || $_GET['url'] == '') {
55 die('Please pass a URL to the script. E.g. Readability.php?url=bla.com/story.html');
56}
57$url = $_GET['url'];
58if (!preg_match('!^https?://!i', $url)) $url = 'http://'.$url;
59$html = file_get_contents($url);
60$r = new Readability($html, $url);
61$r->init();
62echo $r->articleContent->innerHTML;
63*/
64
65class Readability
66{
67 public $version = '1.7.1-without-multi-page';
68 public $convertLinksToFootnotes = false;
69 public $revertForcedParagraphElements = true;
70 public $articleTitle;
71 public $articleContent;
72 public $dom;
73 public $url = null; // optional - URL where HTML was retrieved
74 public $debug = false;
75 protected $body = null; //
76 protected $bodyCache = null; // Cache the body HTML in case we need to re-use it later
77 protected $flags = 7; // 1 | 2 | 4; // Start with all flags set.
78 protected $success = false; // indicates whether we were able to extract or not
79
80 /**
81 * All of the regular expressions in use within readability.
82 * Defined up here so we don't instantiate them repeatedly in loops.
83 **/
84 public $regexps = array(
85 'unlikelyCandidates' => '/combx|comment|community|disqus|extra|foot|header|menu|remark|rss|shoutbox|sidebar|sponsor|ad-break|agegate|pagination|pager|popup|tweet|twitter/i',
86 'okMaybeItsACandidate' => '/and|article|body|column|main|shadow/i',
87 'positive' => '/article|body|content|entry|hentry|main|page|pagination|post|text|blog|story/i',
88 'negative' => '/combx|comment|com-|contact|foot|footer|footnote|masthead|media|meta|outbrain|promo|related|scroll|shoutbox|sidebar|sponsor|shopping|tags|tool|widget/i',
89 'divToPElements' => '/<(a|blockquote|dl|div|img|ol|p|pre|table|ul)/i',
90 'replaceBrs' => '/(<br[^>]*>[ \n\r\t]*){2,}/i',
91 'replaceFonts' => '/<(\/?)font[^>]*>/i',
92 // 'trimRe' => '/^\s+|\s+$/g', // PHP has trim()
93 'normalize' => '/\s{2,}/',
94 'killBreaks' => '/(<br\s*\/?>(\s|&nbsp;?)*){1,}/',
95 'video' => '/http:\/\/(www\.)?(youtube|vimeo)\.com/i',
96 'skipFootnoteLink' => '/^\s*(\[?[a-z0-9]{1,2}\]?|^|edit|citation needed)\s*$/i'
97 );
98
99 /* constants */
100 const FLAG_STRIP_UNLIKELYS = 1;
101 const FLAG_WEIGHT_CLASSES = 2;
102 const FLAG_CLEAN_CONDITIONALLY = 4;
103
104 /**
105 * Create instance of Readability
106 * @param string UTF-8 encoded string
107 * @param string (optional) URL associated with HTML (used for footnotes)
108 */
109 function __construct($html, $url=null)
110 {
111 /* Turn all double br's into p's */
112 /* Note, this is pretty costly as far as processing goes. Maybe optimize later. */
113 $html = preg_replace($this->regexps['replaceBrs'], '</p><p>', $html);
114 $html = preg_replace($this->regexps['replaceFonts'], '<$1span>', $html);
115 $html = mb_convert_encoding($html, 'HTML-ENTITIES', "UTF-8");
116 $this->dom = new DOMDocument();
117 $this->dom->preserveWhiteSpace = false;
118 $this->dom->registerNodeClass('DOMElement', 'JSLikeHTMLElement');
119 @$this->dom->loadHTML($html);
120 $this->url = $url;
121 }
122
123 /**
124 * Get article title element
125 * @return DOMElement
126 */
127 public function getTitle() {
128 return $this->articleTitle;
129 }
130
131 /**
132 * Get article content element
133 * @return DOMElement
134 */
135 public function getContent() {
136 return $this->articleContent;
137 }
138
139 /**
140 * Runs readability.
141 *
142 * Workflow:
143 * 1. Prep the document by removing script tags, css, etc.
144 * 2. Build readability's DOM tree.
145 * 3. Grab the article content from the current dom tree.
146 * 4. Replace the current DOM tree with the new one.
147 * 5. Read peacefully.
148 *
149 * @return boolean true if we found content, false otherwise
150 **/
151 public function init()
152 {
153 $this->removeScripts($this->dom);
154
155 // Assume successful outcome
156 $this->success = true;
157
158 $bodyElems = $this->dom->getElementsByTagName('body');
159 if ($bodyElems->length > 0) {
160 if ($this->bodyCache == null) {
161 $this->bodyCache = $bodyElems->item(0)->innerHTML;
162 }
163 if ($this->body == null) {
164 $this->body = $bodyElems->item(0);
165 }
166 }
167
168 $this->prepDocument();
169
170 //die($this->dom->documentElement->parentNode->nodeType);
171 //$this->setInnerHTML($this->dom->documentElement, $this->getInnerHTML($this->dom->documentElement));
172 //die($this->getInnerHTML($this->dom->documentElement));
173
174 /* Build readability's DOM tree */
175 $overlay = $this->dom->createElement('div');
176 $innerDiv = $this->dom->createElement('div');
177 $articleTitle = $this->getArticleTitle();
178 $articleContent = $this->grabArticle();
179
180 if (!$articleContent) {
181 $this->success = false;
182 $articleContent = $this->dom->createElement('div');
183 $articleContent->setAttribute('id', 'readability-content');
184 $articleContent->innerHTML = '<p>Sorry, Readability was unable to parse this page for content.</p>';
185 }
186
187 $overlay->setAttribute('id', 'readOverlay');
188 $innerDiv->setAttribute('id', 'readInner');
189
190 /* Glue the structure of our document together. */
191 $innerDiv->appendChild($articleTitle);
192 $innerDiv->appendChild($articleContent);
193 $overlay->appendChild($innerDiv);
194
195 /* Clear the old HTML, insert the new content. */
196 $this->body->innerHTML = '';
197 $this->body->appendChild($overlay);
198 //document.body.insertBefore(overlay, document.body.firstChild);
199 $this->body->removeAttribute('style');
200
201 $this->postProcessContent($articleContent);
202
203 // Set title and content instance variables
204 $this->articleTitle = $articleTitle;
205 $this->articleContent = $articleContent;
206
207 return $this->success;
208 }
209
210 /**
211 * Debug
212 */
213 protected function dbg($msg) {
214 if ($this->debug) echo '* ',$msg, '<br />', "\n";
215 }
216
217 /**
218 * Run any post-process modifications to article content as necessary.
219 *
220 * @param DOMElement
221 * @return void
222 */
223 public function postProcessContent($articleContent) {
224 if ($this->convertLinksToFootnotes && !preg_match('/wikipedia\.org/', @$this->url)) {
225 $this->addFootnotes($articleContent);
226 }
227 }
228
229 /**
230 * Get the article title as an H1.
231 *
232 * @return DOMElement
233 */
234 protected function getArticleTitle() {
235 $curTitle = '';
236 $origTitle = '';
237
238 try {
239 $curTitle = $origTitle = $this->getInnerText($this->dom->getElementsByTagName('title')->item(0));
240 } catch(Exception $e) {}
241
242 if (preg_match('/ [\|\-] /', $curTitle))
243 {
244 $curTitle = preg_replace('/(.*)[\|\-] .*/i', '$1', $origTitle);
245
246 if (count(explode(' ', $curTitle)) < 3) {
247 $curTitle = preg_replace('/[^\|\-]*[\|\-](.*)/i', '$1', $origTitle);
248 }
249 }
250 else if (strpos($curTitle, ': ') !== false)
251 {
252 $curTitle = preg_replace('/.*:(.*)/i', '$1', $origTitle);
253
254 if (count(explode(' ', $curTitle)) < 3) {
255 $curTitle = preg_replace('/[^:]*[:](.*)/i','$1', $origTitle);
256 }
257 }
258 else if(strlen($curTitle) > 150 || strlen($curTitle) < 15)
259 {
260 $hOnes = $this->dom->getElementsByTagName('h1');
261 if($hOnes->length == 1)
262 {
263 $curTitle = $this->getInnerText($hOnes->item(0));
264 }
265 }
266
267 $curTitle = trim($curTitle);
268
269 if (count(explode(' ', $curTitle)) <= 4) {
270 $curTitle = $origTitle;
271 }
272
273 $articleTitle = $this->dom->createElement('h1');
274 $articleTitle->innerHTML = $curTitle;
275
276 return $articleTitle;
277 }
278
279 /**
280 * Prepare the HTML document for readability to scrape it.
281 * This includes things like stripping javascript, CSS, and handling terrible markup.
282 *
283 * @return void
284 **/
285 protected function prepDocument() {
286 /**
287 * In some cases a body element can't be found (if the HTML is totally hosed for example)
288 * so we create a new body node and append it to the document.
289 */
290 if($this->dom->documentElement == null){
291 throw new Exception("No document element");
292 }
293 if ($this->body == null)
294 {
295 $this->body = $this->dom->createElement('body');
296 $this->dom->documentElement->appendChild($this->body);
297 }
298
299 $this->body->setAttribute('id', 'readabilityBody');
300
301 /* Remove all style tags in head */
302 $styleTags = $this->dom->getElementsByTagName('style');
303 for ($i = $styleTags->length-1; $i >= 0; $i--)
304 {
305 $styleTags->item($i)->parentNode->removeChild($styleTags->item($i));
306 }
307
308 /* Turn all double br's into p's */
309 /* Note, this is pretty costly as far as processing goes. Maybe optimize later. */
310 //document.body.innerHTML = document.body.innerHTML.replace(readability.regexps.replaceBrs, '</p><p>').replace(readability.regexps.replaceFonts, '<$1span>');
311 // We do this in the constructor for PHP as that's when we have raw HTML - before parsing it into a DOM tree.
312 // Manipulating innerHTML as it's done in JS is not possible in PHP.
313 }
314
315 /**
316 * For easier reading, convert this document to have footnotes at the bottom rather than inline links.
317 * @see http://www.roughtype.com/archives/2010/05/experiments_in.php
318 *
319 * @return void
320 **/
321 public function addFootnotes($articleContent) {
322 $footnotesWrapper = $this->dom->createElement('div');
323 $footnotesWrapper->setAttribute('id', 'readability-footnotes');
324 $footnotesWrapper->innerHTML = '<h3>References</h3>';
325
326 $articleFootnotes = $this->dom->createElement('ol');
327 $articleFootnotes->setAttribute('id', 'readability-footnotes-list');
328 $footnotesWrapper->appendChild($articleFootnotes);
329
330 $articleLinks = $articleContent->getElementsByTagName('a');
331
332 $linkCount = 0;
333 for ($i = 0; $i < $articleLinks->length; $i++)
334 {
335 $articleLink = $articleLinks->item($i);
336 $footnoteLink = $articleLink->cloneNode(true);
337 $refLink = $this->dom->createElement('a');
338 $footnote = $this->dom->createElement('li');
339 $linkDomain = @parse_url($footnoteLink->getAttribute('href'), PHP_URL_HOST);
340 if (!$linkDomain && isset($this->url)) $linkDomain = @parse_url($this->url, PHP_URL_HOST);
341 //linkDomain = footnoteLink.host ? footnoteLink.host : document.location.host,
342 $linkText = $this->getInnerText($articleLink);
343
344 if ((strpos($articleLink->getAttribute('class'), 'readability-DoNotFootnote') !== false) || preg_match($this->regexps['skipFootnoteLink'], $linkText)) {
345 continue;
346 }
347
348 $linkCount++;
349
350 /** Add a superscript reference after the article link */
351 $refLink->setAttribute('href', '#readabilityFootnoteLink-' . $linkCount);
352 $refLink->innerHTML = '<small><sup>[' . $linkCount . ']</sup></small>';
353 $refLink->setAttribute('class', 'readability-DoNotFootnote');
354 $refLink->setAttribute('style', 'color: inherit;');
355
356 //TODO: does this work or should we use DOMNode.isSameNode()?
357 if ($articleLink->parentNode->lastChild == $articleLink) {
358 $articleLink->parentNode->appendChild($refLink);
359 } else {
360 $articleLink->parentNode->insertBefore($refLink, $articleLink->nextSibling);
361 }
362
363 $articleLink->setAttribute('style', 'color: inherit; text-decoration: none;');
364 $articleLink->setAttribute('name', 'readabilityLink-' . $linkCount);
365
366 $footnote->innerHTML = '<small><sup><a href="#readabilityLink-' . $linkCount . '" title="Jump to Link in Article">^</a></sup></small> ';
367
368 $footnoteLink->innerHTML = ($footnoteLink->getAttribute('title') != '' ? $footnoteLink->getAttribute('title') : $linkText);
369 $footnoteLink->setAttribute('name', 'readabilityFootnoteLink-' . $linkCount);
370
371 $footnote->appendChild($footnoteLink);
372 if ($linkDomain) $footnote->innerHTML = $footnote->innerHTML . '<small> (' . $linkDomain . ')</small>';
373
374 $articleFootnotes->appendChild($footnote);
375 }
376
377 if ($linkCount > 0) {
378 $articleContent->appendChild($footnotesWrapper);
379 }
380 }
381
382 /**
383 * Reverts P elements with class 'readability-styled'
384 * to text nodes - which is what they were before.
385 *
386 * @param DOMElement
387 * @return void
388 */
389 function revertReadabilityStyledElements($articleContent) {
390 $xpath = new DOMXPath($articleContent->ownerDocument);
391 $elems = $xpath->query('.//p[@class="readability-styled"]', $articleContent);
392 //$elems = $articleContent->getElementsByTagName('p');
393 for ($i = $elems->length-1; $i >= 0; $i--) {
394 $e = $elems->item($i);
395 $e->parentNode->replaceChild($articleContent->ownerDocument->createTextNode($e->textContent), $e);
396 //if ($e->hasAttribute('class') && $e->getAttribute('class') == 'readability-styled') {
397 // $e->parentNode->replaceChild($this->dom->createTextNode($e->textContent), $e);
398 //}
399 }
400 }
401
402 /**
403 * Prepare the article node for display. Clean out any inline styles,
404 * iframes, forms, strip extraneous <p> tags, etc.
405 *
406 * @param DOMElement
407 * @return void
408 */
409 function prepArticle($articleContent) {
410 $this->cleanStyles($articleContent);
411 $this->killBreaks($articleContent);
412 if ($this->revertForcedParagraphElements) {
413 $this->revertReadabilityStyledElements($articleContent);
414 }
415
416 /* Clean out junk from the article content */
417 $this->cleanConditionally($articleContent, 'form');
418 $this->clean($articleContent, 'object');
419 $this->clean($articleContent, 'h1');
420
421 /**
422 * If there is only one h2, they are probably using it
423 * as a header and not a subheader, so remove it since we already have a header.
424 ***/
425 if ($articleContent->getElementsByTagName('h2')->length == 1) {
426 $this->clean($articleContent, 'h2');
427 }
428 $this->clean($articleContent, 'iframe');
429
430 $this->cleanHeaders($articleContent);
431
432 /* Do these last as the previous stuff may have removed junk that will affect these */
433 $this->cleanConditionally($articleContent, 'table');
434 $this->cleanConditionally($articleContent, 'ul');
435 $this->cleanConditionally($articleContent, 'div');
436
437 /* Remove extra paragraphs */
438 $articleParagraphs = $articleContent->getElementsByTagName('p');
439 for ($i = $articleParagraphs->length-1; $i >= 0; $i--)
440 {
441 $imgCount = $articleParagraphs->item($i)->getElementsByTagName('img')->length;
442 $embedCount = $articleParagraphs->item($i)->getElementsByTagName('embed')->length;
443 $objectCount = $articleParagraphs->item($i)->getElementsByTagName('object')->length;
444
445 if ($imgCount === 0 && $embedCount === 0 && $objectCount === 0 && $this->getInnerText($articleParagraphs->item($i), false) == '')
446 {
447 $articleParagraphs->item($i)->parentNode->removeChild($articleParagraphs->item($i));
448 }
449 }
450
451 try {
452 $articleContent->innerHTML = preg_replace('/<br[^>]*>\s*<p/i', '<p', $articleContent->innerHTML);
453 //articleContent.innerHTML = articleContent.innerHTML.replace(/<br[^>]*>\s*<p/gi, '<p');
454 }
455 catch (Exception $e) {
456 $this->dbg("Cleaning innerHTML of breaks failed. This is an IE strict-block-elements bug. Ignoring.: " . $e);
457 }
458 }
459
460 /**
461 * Initialize a node with the readability object. Also checks the
462 * className/id for special names to add to its score.
463 *
464 * @param Element
465 * @return void
466 **/
467 protected function initializeNode($node) {
468 $readability = $this->dom->createAttribute('readability');
469 $readability->value = 0; // this is our contentScore
470 $node->setAttributeNode($readability);
471
472 switch (strtoupper($node->tagName)) { // unsure if strtoupper is needed, but using it just in case
473 case 'DIV':
474 $readability->value += 5;
475 break;
476
477 case 'PRE':
478 case 'TD':
479 case 'BLOCKQUOTE':
480 $readability->value += 3;
481 break;
482
483 case 'ADDRESS':
484 case 'OL':
485 case 'UL':
486 case 'DL':
487 case 'DD':
488 case 'DT':
489 case 'LI':
490 case 'FORM':
491 $readability->value -= 3;
492 break;
493
494 case 'H1':
495 case 'H2':
496 case 'H3':
497 case 'H4':
498 case 'H5':
499 case 'H6':
500 case 'TH':
501 $readability->value -= 5;
502 break;
503 }
504 $readability->value += $this->getClassWeight($node);
505 }
506
507 /***
508 * grabArticle - Using a variety of metrics (content score, classname, element types), find the content that is
509 * most likely to be the stuff a user wants to read. Then return it wrapped up in a div.
510 *
511 * @return DOMElement
512 **/
513 protected function grabArticle($page=null) {
514 $stripUnlikelyCandidates = $this->flagIsActive(self::FLAG_STRIP_UNLIKELYS);
515 if (!$page) $page = $this->dom;
516 $allElements = $page->getElementsByTagName('*');
517 /**
518 * First, node prepping. Trash nodes that look cruddy (like ones with the class name "comment", etc), and turn divs
519 * into P tags where they have been used inappropriately (as in, where they contain no other block level elements.)
520 *
521 * Note: Assignment from index for performance. See http://www.peachpit.com/articles/article.aspx?p=31567&seqNum=5
522 * TODO: Shouldn't this be a reverse traversal?
523 **/
524 $node = null;
525 $nodesToScore = array();
526 for ($nodeIndex = 0; ($node = $allElements->item($nodeIndex)); $nodeIndex++) {
527 //for ($nodeIndex=$targetList->length-1; $nodeIndex >= 0; $nodeIndex--) {
528 //$node = $targetList->item($nodeIndex);
529 $tagName = strtoupper($node->tagName);
530 /* Remove unlikely candidates */
531 if ($stripUnlikelyCandidates) {
532 $unlikelyMatchString = $node->getAttribute('class') . $node->getAttribute('id');
533 if (
534 preg_match($this->regexps['unlikelyCandidates'], $unlikelyMatchString) &&
535 !preg_match($this->regexps['okMaybeItsACandidate'], $unlikelyMatchString) &&
536 $tagName != 'BODY'
537 )
538 {
539 $this->dbg('Removing unlikely candidate - ' . $unlikelyMatchString);
540 //$nodesToRemove[] = $node;
541 $node->parentNode->removeChild($node);
542 $nodeIndex--;
543 continue;
544 }
545 }
546
547 if ($tagName == 'P' || $tagName == 'TD' || $tagName == 'PRE') {
548 $nodesToScore[] = $node;
549 }
550
551 /* Turn all divs that don't have children block level elements into p's */
552 if ($tagName == 'DIV') {
553 if (!preg_match($this->regexps['divToPElements'], $node->innerHTML)) {
554 //$this->dbg('Altering div to p');
555 $newNode = $this->dom->createElement('p');
556 try {
557 $newNode->innerHTML = $node->innerHTML;
558 //$nodesToReplace[] = array('new'=>$newNode, 'old'=>$node);
559 $node->parentNode->replaceChild($newNode, $node);
560 $nodeIndex--;
561 $nodesToScore[] = $node; // or $newNode?
562 }
563 catch(Exception $e) {
564 $this->dbg('Could not alter div to p, reverting back to div.: ' . $e);
565 }
566 }
567 else
568 {
569 /* EXPERIMENTAL */
570 // TODO: change these p elements back to text nodes after processing
571 for ($i = 0, $il = $node->childNodes->length; $i < $il; $i++) {
572 $childNode = $node->childNodes->item($i);
573 if ($childNode->nodeType == 3) { // XML_TEXT_NODE
574 //$this->dbg('replacing text node with a p tag with the same content.');
575 $p = $this->dom->createElement('p');
576 $p->innerHTML = $childNode->nodeValue;
577 $p->setAttribute('style', 'display: inline;');
578 $p->setAttribute('class', 'readability-styled');
579 $childNode->parentNode->replaceChild($p, $childNode);
580 }
581 }
582 }
583 }
584 }
585
586 /**
587 * Loop through all paragraphs, and assign a score to them based on how content-y they look.
588 * Then add their score to their parent node.
589 *
590 * A score is determined by things like number of commas, class names, etc. Maybe eventually link density.
591 **/
592 $candidates = array();
593 for ($pt=0; $pt < count($nodesToScore); $pt++) {
594 $parentNode = $nodesToScore[$pt]->parentNode;
595 // $grandParentNode = $parentNode ? $parentNode->parentNode : null;
596 $grandParentNode = !$parentNode ? null : (($parentNode->parentNode instanceof DOMElement) ? $parentNode->parentNode : null);
597 $innerText = $this->getInnerText($nodesToScore[$pt]);
598
599 if (!$parentNode || !isset($parentNode->tagName)) {
600 continue;
601 }
602
603 /* If this paragraph is less than 25 characters, don't even count it. */
604 if(strlen($innerText) < 25) {
605 continue;
606 }
607
608 /* Initialize readability data for the parent. */
609 if (!$parentNode->hasAttribute('readability'))
610 {
611 $this->initializeNode($parentNode);
612 $candidates[] = $parentNode;
613 }
614
615 /* Initialize readability data for the grandparent. */
616 if ($grandParentNode && !$grandParentNode->hasAttribute('readability') && isset($grandParentNode->tagName))
617 {
618 $this->initializeNode($grandParentNode);
619 $candidates[] = $grandParentNode;
620 }
621
622 $contentScore = 0;
623
624 /* Add a point for the paragraph itself as a base. */
625 $contentScore++;
626
627 /* Add points for any commas within this paragraph */
628 $contentScore += count(explode(',', $innerText));
629
630 /* For every 100 characters in this paragraph, add another point. Up to 3 points. */
631 $contentScore += min(floor(strlen($innerText) / 100), 3);
632
633 /* Add the score to the parent. The grandparent gets half. */
634 $parentNode->getAttributeNode('readability')->value += $contentScore;
635
636 if ($grandParentNode) {
637 $grandParentNode->getAttributeNode('readability')->value += $contentScore/2;
638 }
639 }
640
641 /**
642 * After we've calculated scores, loop through all of the possible candidate nodes we found
643 * and find the one with the highest score.
644 **/
645 $topCandidate = null;
646 for ($c=0, $cl=count($candidates); $c < $cl; $c++)
647 {
648 /**
649 * Scale the final candidates score based on link density. Good content should have a
650 * relatively small link density (5% or less) and be mostly unaffected by this operation.
651 **/
652 $readability = $candidates[$c]->getAttributeNode('readability');
653 $readability->value = $readability->value * (1-$this->getLinkDensity($candidates[$c]));
654
655 $this->dbg('Candidate: ' . $candidates[$c]->tagName . ' (' . $candidates[$c]->getAttribute('class') . ':' . $candidates[$c]->getAttribute('id') . ') with score ' . $readability->value);
656
657 if (!$topCandidate || $readability->value > (int)$topCandidate->getAttribute('readability')) {
658 $topCandidate = $candidates[$c];
659 }
660 }
661
662 /**
663 * If we still have no top candidate, just use the body as a last resort.
664 * We also have to copy the body node so it is something we can modify.
665 **/
666 if ($topCandidate === null || strtoupper($topCandidate->tagName) == 'BODY')
667 {
668 $topCandidate = $this->dom->createElement('div');
669 $topCandidate->innerHTML = ($page instanceof DOMDocument) ? $page->saveXML($page->documentElement) : $page->innerHTML;
670 $page->innerHTML = '';
671 $page->appendChild($topCandidate);
672 $this->initializeNode($topCandidate);
673 }
674
675 /**
676 * Now that we have the top candidate, look through its siblings for content that might also be related.
677 * Things like preambles, content split by ads that we removed, etc.
678 **/
679 $articleContent = $this->dom->createElement('div');
680 $articleContent->setAttribute('id', 'readability-content');
681 $siblingScoreThreshold = max(10, ((int)$topCandidate->getAttribute('readability')) * 0.2);
682 $siblingNodes = $topCandidate->parentNode->childNodes;
683
684 for ($s=0, $sl=$siblingNodes->length; $s < $sl; $s++)
685 {
686 $siblingNode = $siblingNodes->item($s);
687 $append = false;
688
689 $this->dbg('Looking at sibling node: ' . $siblingNode->nodeName . (($siblingNode->nodeType === XML_ELEMENT_NODE && $siblingNode->hasAttribute('readability')) ? (' with score ' . $siblingNode->getAttribute('readability')) : ''));
690
691 //dbg('Sibling has score ' . ($siblingNode->readability ? siblingNode.readability.contentScore : 'Unknown'));
692
693 if ($siblingNode === $topCandidate)
694 // or if ($siblingNode->isSameNode($topCandidate))
695 {
696 $append = true;
697 }
698
699 $contentBonus = 0;
700 /* Give a bonus if sibling nodes and top candidates have the example same classname */
701 if ($siblingNode->nodeType === XML_ELEMENT_NODE && $siblingNode->getAttribute('class') == $topCandidate->getAttribute('class') && $topCandidate->getAttribute('class') != '') {
702 $contentBonus += ((int)$topCandidate->getAttribute('readability')) * 0.2;
703 }
704
705 if ($siblingNode->nodeType === XML_ELEMENT_NODE && $siblingNode->hasAttribute('readability') && (((int)$siblingNode->getAttribute('readability')) + $contentBonus) >= $siblingScoreThreshold)
706 {
707 $append = true;
708 }
709
710 if (strtoupper($siblingNode->nodeName) == 'P') {
711 $linkDensity = $this->getLinkDensity($siblingNode);
712 $nodeContent = $this->getInnerText($siblingNode);
713 $nodeLength = strlen($nodeContent);
714
715 if ($nodeLength > 80 && $linkDensity < 0.25)
716 {
717 $append = true;
718 }
719 else if ($nodeLength < 80 && $linkDensity === 0 && preg_match('/\.( |$)/', $nodeContent))
720 {
721 $append = true;
722 }
723 }
724
725 if ($append)
726 {
727 $this->dbg('Appending node: ' . $siblingNode->nodeName);
728
729 $nodeToAppend = null;
730 $sibNodeName = strtoupper($siblingNode->nodeName);
731 if ($sibNodeName != 'DIV' && $sibNodeName != 'P') {
732 /* We have a node that isn't a common block level element, like a form or td tag. Turn it into a div so it doesn't get filtered out later by accident. */
733
734 $this->dbg('Altering siblingNode of ' . $sibNodeName . ' to div.');
735 $nodeToAppend = $this->dom->createElement('div');
736 try {
737 $nodeToAppend->setAttribute('id', $siblingNode->getAttribute('id'));
738 $nodeToAppend->innerHTML = $siblingNode->innerHTML;
739 }
740 catch(Exception $e)
741 {
742 $this->dbg('Could not alter siblingNode to div, reverting back to original.');
743 $nodeToAppend = $siblingNode;
744 $s--;
745 $sl--;
746 }
747 } else {
748 $nodeToAppend = $siblingNode;
749 $s--;
750 $sl--;
751 }
752
753 /* To ensure a node does not interfere with readability styles, remove its classnames */
754 $nodeToAppend->removeAttribute('class');
755
756 /* Append sibling and subtract from our list because it removes the node when you append to another node */
757 $articleContent->appendChild($nodeToAppend);
758 }
759 }
760
761 /**
762 * So we have all of the content that we need. Now we clean it up for presentation.
763 **/
764 $this->prepArticle($articleContent);
765
766 /**
767 * Now that we've gone through the full algorithm, check to see if we got any meaningful content.
768 * If we didn't, we may need to re-run grabArticle with different flags set. This gives us a higher
769 * likelihood of finding the content, and the sieve approach gives us a higher likelihood of
770 * finding the -right- content.
771 **/
772 if (strlen($this->getInnerText($articleContent, false)) < 250)
773 {
774 $this->body->innerHTML = $this->bodyCache;
775
776 if ($this->flagIsActive(self::FLAG_STRIP_UNLIKELYS)) {
777 $this->removeFlag(self::FLAG_STRIP_UNLIKELYS);
778 return $this->grabArticle($this->body);
779 }
780 else if ($this->flagIsActive(self::FLAG_WEIGHT_CLASSES)) {
781 $this->removeFlag(self::FLAG_WEIGHT_CLASSES);
782 return $this->grabArticle($this->body);
783 }
784 else if ($this->flagIsActive(self::FLAG_CLEAN_CONDITIONALLY)) {
785 $this->removeFlag(self::FLAG_CLEAN_CONDITIONALLY);
786 return $this->grabArticle($this->body);
787 }
788 else {
789 return false;
790 }
791 }
792 return $articleContent;
793 }
794
795 /**
796 * Remove script tags from document
797 *
798 * @param DOMElement
799 * @return void
800 */
801 public function removeScripts($doc) {
802 $scripts = $doc->getElementsByTagName('script');
803 for($i = $scripts->length-1; $i >= 0; $i--)
804 {
805 $scripts->item($i)->parentNode->removeChild($scripts->item($i));
806 }
807 }
808
809 /**
810 * Get the inner text of a node.
811 * This also strips out any excess whitespace to be found.
812 *
813 * @param DOMElement $
814 * @param boolean $normalizeSpaces (default: true)
815 * @return string
816 **/
817 public function getInnerText($e, $normalizeSpaces=true) {
818 $textContent = '';
819
820 if (!isset($e->textContent) || $e->textContent == '') {
821 return '';
822 }
823
824 $textContent = trim($e->textContent);
825
826 if ($normalizeSpaces) {
827 return preg_replace($this->regexps['normalize'], ' ', $textContent);
828 } else {
829 return $textContent;
830 }
831 }
832
833 /**
834 * Get the number of times a string $s appears in the node $e.
835 *
836 * @param DOMElement $e
837 * @param string - what to count. Default is ","
838 * @return number (integer)
839 **/
840 public function getCharCount($e, $s=',') {
841 return substr_count($this->getInnerText($e), $s);
842 }
843
844 /**
845 * Remove the style attribute on every $e and under.
846 *
847 * @param DOMElement $e
848 * @return void
849 */
850 public function cleanStyles($e) {
851 $elems = $e->getElementsByTagName('*');
852 foreach ($elems as $elem) {
853 $elem->removeAttribute('style');
854 }
855 }
856
857 /**
858 * Get the density of links as a percentage of the content
859 * This is the amount of text that is inside a link divided by the total text in the node.
860 *
861 * @param DOMElement $e
862 * @return number (float)
863 */
864 public function getLinkDensity($e) {
865 $links = $e->getElementsByTagName('a');
866 $textLength = strlen($this->getInnerText($e));
867 $linkLength = 0;
868 for ($i=0, $il=$links->length; $i < $il; $i++)
869 {
870 $linkLength += strlen($this->getInnerText($links->item($i)));
871 }
872 if ($textLength > 0) {
873 return $linkLength / $textLength;
874 } else {
875 return 0;
876 }
877 }
878
879 /**
880 * Get an elements class/id weight. Uses regular expressions to tell if this
881 * element looks good or bad.
882 *
883 * @param DOMElement $e
884 * @return number (Integer)
885 */
886 public function getClassWeight($e) {
887 if(!$this->flagIsActive(self::FLAG_WEIGHT_CLASSES)) {
888 return 0;
889 }
890
891 $weight = 0;
892
893 /* Look for a special classname */
894 if ($e->hasAttribute('class') && $e->getAttribute('class') != '')
895 {
896 if (preg_match($this->regexps['negative'], $e->getAttribute('class'))) {
897 $weight -= 25;
898 }
899 if (preg_match($this->regexps['positive'], $e->getAttribute('class'))) {
900 $weight += 25;
901 }
902 }
903
904 /* Look for a special ID */
905 if ($e->hasAttribute('id') && $e->getAttribute('id') != '')
906 {
907 if (preg_match($this->regexps['negative'], $e->getAttribute('id'))) {
908 $weight -= 25;
909 }
910 if (preg_match($this->regexps['positive'], $e->getAttribute('id'))) {
911 $weight += 25;
912 }
913 }
914 return $weight;
915 }
916
917 /**
918 * Remove extraneous break tags from a node.
919 *
920 * @param DOMElement $node
921 * @return void
922 */
923 public function killBreaks($node) {
924 $html = $node->innerHTML;
925 $html = preg_replace($this->regexps['killBreaks'], '<br />', $html);
926 $node->innerHTML = $html;
927 }
928
929 /**
930 * Clean a node of all elements of type "tag".
931 * (Unless it's a youtube/vimeo video. People love movies.)
932 *
933 * @param DOMElement $e
934 * @param string $tag
935 * @return void
936 */
937 public function clean($e, $tag) {
938 $targetList = $e->getElementsByTagName($tag);
939 $isEmbed = ($tag == 'object' || $tag == 'embed');
940
941 for ($y=$targetList->length-1; $y >= 0; $y--) {
942 /* Allow youtube and vimeo videos through as people usually want to see those. */
943 if ($isEmbed) {
944 $attributeValues = '';
945 for ($i=0, $il=$targetList->item($y)->attributes->length; $i < $il; $i++) {
946 $attributeValues .= $targetList->item($y)->attributes->item($i)->value . '|'; // DOMAttr? (TODO: test)
947 }
948
949 /* First, check the elements attributes to see if any of them contain youtube or vimeo */
950 if (preg_match($this->regexps['video'], $attributeValues)) {
951 continue;
952 }
953
954 /* Then check the elements inside this element for the same. */
955 if (preg_match($this->regexps['video'], $targetList->item($y)->innerHTML)) {
956 continue;
957 }
958 }
959 $targetList->item($y)->parentNode->removeChild($targetList->item($y));
960 }
961 }
962
963 /**
964 * Clean an element of all tags of type "tag" if they look fishy.
965 * "Fishy" is an algorithm based on content length, classnames,
966 * link density, number of images & embeds, etc.
967 *
968 * @param DOMElement $e
969 * @param string $tag
970 * @return void
971 */
972 public function cleanConditionally($e, $tag) {
973 if (!$this->flagIsActive(self::FLAG_CLEAN_CONDITIONALLY)) {
974 return;
975 }
976
977 $tagsList = $e->getElementsByTagName($tag);
978 $curTagsLength = $tagsList->length;
979
980 /**
981 * Gather counts for other typical elements embedded within.
982 * Traverse backwards so we can remove nodes at the same time without effecting the traversal.
983 *
984 * TODO: Consider taking into account original contentScore here.
985 */
986 for ($i=$curTagsLength-1; $i >= 0; $i--) {
987 $weight = $this->getClassWeight($tagsList->item($i));
988 $contentScore = ($tagsList->item($i)->hasAttribute('readability')) ? (int)$tagsList->item($i)->getAttribute('readability') : 0;
989
990 $this->dbg('Cleaning Conditionally ' . $tagsList->item($i)->tagName . ' (' . $tagsList->item($i)->getAttribute('class') . ':' . $tagsList->item($i)->getAttribute('id') . ')' . (($tagsList->item($i)->hasAttribute('readability')) ? (' with score ' . $tagsList->item($i)->getAttribute('readability')) : ''));
991
992 if ($weight + $contentScore < 0) {
993 $tagsList->item($i)->parentNode->removeChild($tagsList->item($i));
994 }
995 else if ( $this->getCharCount($tagsList->item($i), ',') < 10) {
996 /**
997 * If there are not very many commas, and the number of
998 * non-paragraph elements is more than paragraphs or other ominous signs, remove the element.
999 **/
1000 $p = $tagsList->item($i)->getElementsByTagName('p')->length;
1001 $img = $tagsList->item($i)->getElementsByTagName('img')->length;
1002 $li = $tagsList->item($i)->getElementsByTagName('li')->length-100;
1003 $input = $tagsList->item($i)->getElementsByTagName('input')->length;
1004
1005 $embedCount = 0;
1006 $embeds = $tagsList->item($i)->getElementsByTagName('embed');
1007 for ($ei=0, $il=$embeds->length; $ei < $il; $ei++) {
1008 if (preg_match($this->regexps['video'], $embeds->item($ei)->getAttribute('src'))) {
1009 $embedCount++;
1010 }
1011 }
1012
1013 $linkDensity = $this->getLinkDensity($tagsList->item($i));
1014 $contentLength = strlen($this->getInnerText($tagsList->item($i)));
1015 $toRemove = false;
1016
1017 if ( $img > $p ) {
1018 $toRemove = true;
1019 } else if ($li > $p && $tag != 'ul' && $tag != 'ol') {
1020 $toRemove = true;
1021 } else if ( $input > floor($p/3) ) {
1022 $toRemove = true;
1023 } else if ($contentLength < 25 && ($img === 0 || $img > 2) ) {
1024 $toRemove = true;
1025 } else if($weight < 25 && $linkDensity > 0.2) {
1026 $toRemove = true;
1027 } else if($weight >= 25 && $linkDensity > 0.5) {
1028 $toRemove = true;
1029 } else if(($embedCount == 1 && $contentLength < 75) || $embedCount > 1) {
1030 $toRemove = true;
1031 }
1032
1033 if ($toRemove) {
1034 $tagsList->item($i)->parentNode->removeChild($tagsList->item($i));
1035 }
1036 }
1037 }
1038 }
1039
1040 /**
1041 * Clean out spurious headers from an Element. Checks things like classnames and link density.
1042 *
1043 * @param DOMElement $e
1044 * @return void
1045 */
1046 public function cleanHeaders($e) {
1047 for ($headerIndex = 1; $headerIndex < 3; $headerIndex++) {
1048 $headers = $e->getElementsByTagName('h' . $headerIndex);
1049 for ($i=$headers->length-1; $i >=0; $i--) {
1050 if ($this->getClassWeight($headers->item($i)) < 0 || $this->getLinkDensity($headers->item($i)) > 0.33) {
1051 $headers->item($i)->parentNode->removeChild($headers->item($i));
1052 }
1053 }
1054 }
1055 }
1056
1057 public function flagIsActive($flag) {
1058 return ($this->flags & $flag) > 0;
1059 }
1060
1061 public function addFlag($flag) {
1062 $this->flags = $this->flags | $flag;
1063 }
1064
1065 public function removeFlag($flag) {
1066 $this->flags = $this->flags & ~$flag;
1067 }
1068}
1069?> \ No newline at end of file
generated by cgit v1.2.3 (git 2.25.1) at 2024-11-16 17:04:58 +0000