Article::setRawDoc() - Code Metrics - scotteh/php-goose - Measure and Improve Code Quality continuously with Scrutinizer

Article::setRawDoc() A
last analyzed 2023-09-05 13:38 UTC

↳ Parent: Article

Complexity

Conditions	1
Paths	1

Size

Total Lines	4
Code Lines	2

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
dl	0
loc	4
c	0
b	0
f	0
rs	10
cc	1
eloc	2
nc	1
nop	1

<?php declare(strict_types=1);

namespace Goose;

use Goose\Images\Image;
use DOMWrap\{Element, Document};

/**
 * Article
 *
 * @package Goose
 * @license http://www.apache.org/licenses/LICENSE-2.0 Apache License 2.0
 */
class Article {
    /**
     * Language of the article
     *
     * @var string|null
     */
    protected $language;

    /**
     * @param string $language
     *
     * @return self
     */
    public function setLanguage(string $language = null): self {
        $this->language = $language;

        return $this;
    }

    /** @return string|null */
    public function getLanguage(): ?string {
        return $this->language;
    }

    /**
     * OpenGraph meta data
     *
     * @var string[]
     */
    protected $openGraph = [];

    /**
     * @param string[] $openGraph
     *
     * @return self
     */
    public function setOpenGraph(array $openGraph): self {
        $this->openGraph = $openGraph;

        return $this;
    }

    /** @return string[] */
    public function getOpenGraph(): array {
        return $this->openGraph;
    }

    /**
     * Title of the article
     *
     * @var string|null
     */
    protected $title;

    /**
     * @param string $title
     *
     * @return self
     */
    public function setTitle(string $title = null): self {
        $this->title = $title;

        return $this;
    }

    /** @return string|null */
    public function getTitle(): ?string {
        return $this->title;
    }

    /**
     * Stores the lovely, pure text from the article, stripped of html, formatting, etc...
     * just raw text with paragraphs separated by newlines. This is probably what you want to use.
     *
     * @var string|null
     */
    protected $cleanedArticleText;

    /**
     * @param string $cleanedArticleText
     *
     * @return self
     */
    public function setCleanedArticleText(string $cleanedArticleText = null): self {
        $this->cleanedArticleText = $cleanedArticleText;

        return $this;
    }

    /** @return string|null */
    public function getCleanedArticleText(): ?string {
        return $this->cleanedArticleText;
    }

    /**
     * Article with the originals HTML tags (<p>, <a>, ..)
     *
     * @var string|null
     */
    protected $htmlArticle;

    /**
     * @param string $htmlArticle
     *
     * @return self
     */
    public function setHtmlArticle(string $htmlArticle = null): self {
        $this->htmlArticle = $htmlArticle;

        return $this;
    }

    /** @return string|null */
    public function getHtmlArticle(): ?string {
        return $this->htmlArticle;
    }

    /**
     * Meta description field in HTML source
     *
     * @var string|null
     */
    protected $metaDescription;

    /**
     * @param string $metaDescription
     *
     * @return self
     */
    public function setMetaDescription(string $metaDescription = null): self {
        $this->metaDescription = $metaDescription;

        return $this;
    }

    /** @return string|null */
    public function getMetaDescription(): ?string {
        return $this->metaDescription;
    }

    /**
     * Meta keywords field in the HTML source
     *
     * @var string|null
     */
    protected $metaKeywords;

    /**
     * @param string $metaKeywords
     *
     * @return self
     */
    public function setMetaKeywords(string $metaKeywords = null): self {
        $this->metaKeywords = $metaKeywords;

        return $this;
    }

    /** @return string */
    public function getMetaKeywords(): ?string {
        return $this->metaKeywords;
    }

    /**
     * The canonical link of this article if found in the meta data
     *
     * @var string|null
     */
    protected $canonicalLink;

    /**
     * @param string $canonicalLink
     *
     * @return self
     */
    public function setCanonicalLink(string $canonicalLink = null): self {
        $this->canonicalLink = $canonicalLink;

        return $this;
    }

    /** @return string|null */
    public function getCanonicalLink(): ?string {
        return $this->canonicalLink;
    }

    /**
     * Domain of the article we're parsing
     *
     * @var string|null
     */
    protected $domain;

    /**
     * @param string $domain
     *
     * @return self
     */
    public function setDomain(string $domain = null): self {
        $this->domain = $domain;

        return $this;
    }

    /** @return string|null */
    public function getDomain(): ?string {
        return $this->domain;
    }

    /**
     * Top Element we think is a candidate for the main body of the article
     *
     * @var Element|null
     */
    protected $topNode;

    /**
     * @param Element|null $topNode
     *
     * @return self
     */
    public function setTopNode(Element $topNode = null): self {
        $this->topNode = $topNode;

        return $this;
    }

    /** @return Element|null */
    public function getTopNode(): ?Element {
        return $this->topNode;
    }

    /**
     * Top Image object that we think represents this article
     *
     * @var Image|null
     */
    protected $topImage;

    /**
     * @param Image|null $topImage
     *
     * @return self
     */
    public function setTopImage(Image $topImage = null): self {
        $this->topImage = $topImage;

        return $this;
    }

    /** @return Image|null */
    public function getTopImage(): ?Image {
        return $this->topImage;
    }

    /**
     * All image candidates from article
     *
     * @var Image[]
     */
    protected $allImages = [];

    /**
     * @param Image[] $allImages
     *
     * @return self
     */
    public function setAllImages(array $allImages = []): self {
        $this->allImages = $allImages;

        return $this;
    }

    /** @return Image[] */
    public function getAllImages(): array {
        return $this->allImages;
    }

    /**
     * Tags that may have been in the article, these are not meta keywords
     *
     * @var string[]
     */
    protected $tags = [];

    /**
     * @param string[] $tags
     *
     * @return self
     */
    public function setTags(array $tags): self {
        $this->tags = $tags;

        return $this;
    }

    /** @return string[] */
    public function getTags(): array {
        return $this->tags;
    }

    /**
     *  List of links in the article
     *
     * @var string[]
     */
    protected $links = [];

    /**
     * @param string[] $links
     *
     * @return self
     */
    public function setLinks(array $links): self {
        $this->links = $links;

        return $this;
    }

    /** @return string[] */
    public function getLinks(): array {
        return $this->links;
    }

    /**
     * List of any videos we found on the page like youtube, vimeo
     *
     * @var string[]
     */
    protected $videos = [];

    /**
     * @param string[] $videos
     *
     * @return self
     */
    public function setVideos(array $videos): self {
        $this->videos = $videos;

        return $this;
    }

    /** @return string[] */
    public function getVideos(): array {
        return $this->videos;
    }

    /**
     * Final URL that we're going to try and fetch content against, this would be expanded if any
     * escaped fragments were found in the starting url
     *
     * @var string|null
     */
    protected $finalUrl;

    /**
     * @param string $finalUrl
     *
     * @return self
     */
    public function setFinalUrl(string $finalUrl = null): self {
        $this->finalUrl = $finalUrl;

        return $this;
    }

    /** @return string|null */
    public function getFinalUrl(): ?string {
        return $this->finalUrl;
    }

    /**
     * MD5 hash of the url to use for various identification tasks
     *
     * @var string|null
     */
    protected $linkhash;

    /**
     * @param string $linkhash
     *
     * @return self
     */
    public function setLinkhash(string $linkhash = null): self {
        $this->linkhash = $linkhash;

        return $this;
    }

    /** @return string */
    public function getLinkhash(): ?string {
        return $this->linkhash;
    }

    /**
     * Raw HTML straight from the network connection
     *
     * @var string|null
     */
    protected $rawHtml;

    /**
     * @param string $rawHtml
     *
     * @return self
     */
    public function setRawHtml(string $rawHtml = null): self {
        $this->rawHtml = $rawHtml;

        return $this;
    }

    /** @return string|null */
    public function getRawHtml(): ?string {
        return $this->rawHtml;
    }

    /**
     * DOM Document object
     *
     * @var Document
     */
    protected $doc;

    /**
     * @param Document $doc
     *
     * @return self
     */
    public function setDoc(Document $doc): self {
        $this->doc = $doc;

        return $this;
    }

    /** @return Document */
    public function getDoc(): Document {
        return $this->doc;
    }

    /**
     * Original DOM document that contains a pure object from the original HTML without any cleaning
     * options done on it
     *
     * @var Document
     */
    protected $rawDoc;

    /**
     * @param Document $rawDoc
     *
     * @return self
     */
    public function setRawDoc(Document $rawDoc): self {
        $this->rawDoc = $rawDoc;

        return $this;
    }

    /** @return Document */
    public function getRawDoc(): Document {
        return $this->rawDoc;
    }

    /**
     * Original psr7 response object
     *
     * @var \Psr\Http\Message\ResponseInterface
     */
    protected $rawResponse;

    /**
     * @param \Psr\Http\Message\ResponseInterface|null $rawResponse
     *
     * @return self
     */
    public function setRawResponse(\Psr\Http\Message\ResponseInterface $rawResponse): self {
        $this->rawResponse = $rawResponse;

        return $this;
    }

    /** @return \Psr\Http\Message\ResponseInterface */
    public function getRawResponse(): \Psr\Http\Message\ResponseInterface {
        return $this->rawResponse;
    }

    /**
     * Sometimes useful to try and know when the publish date of an article was
     *
     * @var \DateTime|null
     */
    protected $publishDate;

    /**
     * @param \DateTime|null $publishDate
     *
     * @return self
     */
    public function setPublishDate(\DateTime $publishDate = null): self {
        $this->publishDate = $publishDate;

        return $this;
    }

    /** @return \DateTime|null */
    public function getPublishDate(): ?\DateTime {
        return $this->publishDate;
    }

    /**
     * A property bucket for consumers of goose to store custom data extractions.
     *
     * @var string|null
     */
    protected $additionalData;

    /**
     * @param string|null $additionalData
     *
     * @return self
     */
    public function setAdditionalData(string $additionalData = null): self {
        $this->additionalData = $additionalData;

        return $this;
    }

    /** @return string|null */
    public function getAdditionalData(): ?string {
        return $this->additionalData;
    }

    /**
     * Facebook Open Graph data that that is found in Article Meta tags
     *
     * @var string|null
     */
    protected $openGraphData;

    /**
     * @param string $openGraphData
     *
     * @return self
     */
    public function setOpenGraphData(string $openGraphData = null): self {
        $this->openGraphData = $openGraphData;

        return $this;
    }

    /** @return string|null */
    public function getOpenGraphData(): ?string {
        return $this->openGraphData;
    }

    /**
     * Most popular words used in the lovely article
     *
     * @var string[]
     */
    protected $popularWords = [];

    /**
     * @param string[] $popularWords
     *
     * @return self
     */
    public function setPopularWords(array $popularWords): self {
        $this->popularWords = $popularWords;

        return $this;
    }

    /** @return string[] */
    public function getPopularWords(): array {
        return $this->popularWords;
    }
}

1			<?php declare(strict_types=1);
2
3			namespace Goose;
4
5			use Goose\Images\Image;
6			use DOMWrap\{Element, Document};
7
8			/**
9			* Article
10			*
11			* @package Goose
12			* @license http://www.apache.org/licenses/LICENSE-2.0 Apache License 2.0
13			*/
14			class Article {
15			/**
16			* Language of the article
17			*
18			* @var string\|null
19			*/
20			protected $language;
21
22			/**
23			* @param string $language
24			*
25			* @return self
26			*/
27			public function setLanguage(string $language = null): self {
28			$this->language = $language;
29
30			return $this;
31			}
32
33			/** @return string\|null */
34			public function getLanguage(): ?string {
35			return $this->language;
36			}
37
38			/**
39			* OpenGraph meta data
40			*
41			* @var string[]
42			*/
43			protected $openGraph = [];
44
45			/**
46			* @param string[] $openGraph
47			*
48			* @return self
49			*/
50			public function setOpenGraph(array $openGraph): self {
51			$this->openGraph = $openGraph;
52
53			return $this;
54			}
55
56			/** @return string[] */
57			public function getOpenGraph(): array {
58			return $this->openGraph;
59			}
60
61			/**
62			* Title of the article
63			*
64			* @var string\|null
65			*/
66			protected $title;
67
68			/**
69			* @param string $title
70			*
71			* @return self
72			*/
73			public function setTitle(string $title = null): self {
74			$this->title = $title;
75
76			return $this;
77			}
78
79			/** @return string\|null */
80			public function getTitle(): ?string {
81			return $this->title;
82			}
83
84			/**
85			* Stores the lovely, pure text from the article, stripped of html, formatting, etc...
86			* just raw text with paragraphs separated by newlines. This is probably what you want to use.
87			*
88			* @var string\|null
89			*/
90			protected $cleanedArticleText;
91
92			/**
93			* @param string $cleanedArticleText
94			*
95			* @return self
96			*/
97			public function setCleanedArticleText(string $cleanedArticleText = null): self {
98			$this->cleanedArticleText = $cleanedArticleText;
99
100			return $this;
101			}
102
103			/** @return string\|null */
104			public function getCleanedArticleText(): ?string {
105			return $this->cleanedArticleText;
106			}
107
108			/**
109			* Article with the originals HTML tags (<p>, <a>, ..)
110			*
111			* @var string\|null
112			*/
113			protected $htmlArticle;
114
115			/**
116			* @param string $htmlArticle
117			*
118			* @return self
119			*/
120			public function setHtmlArticle(string $htmlArticle = null): self {
121			$this->htmlArticle = $htmlArticle;
122
123			return $this;
124			}
125
126			/** @return string\|null */
127			public function getHtmlArticle(): ?string {
128			return $this->htmlArticle;
129			}
130
131			/**
132			* Meta description field in HTML source
133			*
134			* @var string\|null
135			*/
136			protected $metaDescription;
137
138			/**
139			* @param string $metaDescription
140			*
141			* @return self
142			*/
143			public function setMetaDescription(string $metaDescription = null): self {
144			$this->metaDescription = $metaDescription;
145
146			return $this;
147			}
148
149			/** @return string\|null */
150			public function getMetaDescription(): ?string {
151			return $this->metaDescription;
152			}
153
154			/**
155			* Meta keywords field in the HTML source
156			*
157			* @var string\|null
158			*/
159			protected $metaKeywords;
160
161			/**
162			* @param string $metaKeywords
163			*
164			* @return self
165			*/
166			public function setMetaKeywords(string $metaKeywords = null): self {
167			$this->metaKeywords = $metaKeywords;
168
169			return $this;
170			}
171
172			/** @return string */
173			public function getMetaKeywords(): ?string {
174			return $this->metaKeywords;
175			}
176
177			/**
178			* The canonical link of this article if found in the meta data
179			*
180			* @var string\|null
181			*/
182			protected $canonicalLink;
183
184			/**
185			* @param string $canonicalLink
186			*
187			* @return self
188			*/
189			public function setCanonicalLink(string $canonicalLink = null): self {
190			$this->canonicalLink = $canonicalLink;
191
192			return $this;
193			}
194
195			/** @return string\|null */
196			public function getCanonicalLink(): ?string {
197			return $this->canonicalLink;
198			}
199
200			/**
201			* Domain of the article we're parsing
202			*
203			* @var string\|null
204			*/
205			protected $domain;
206
207			/**
208			* @param string $domain
209			*
210			* @return self
211			*/
212			public function setDomain(string $domain = null): self {
213			$this->domain = $domain;
214
215			return $this;
216			}
217
218			/** @return string\|null */
219			public function getDomain(): ?string {
220			return $this->domain;
221			}
222
223			/**
224			* Top Element we think is a candidate for the main body of the article
225			*
226			* @var Element\|null
227			*/
228			protected $topNode;
229
230			/**
231			* @param Element\|null $topNode
232			*
233			* @return self
234			*/
235			public function setTopNode(Element $topNode = null): self {
236			$this->topNode = $topNode;
237
238			return $this;
239			}
240
241			/** @return Element\|null */
242			public function getTopNode(): ?Element {
243			return $this->topNode;
244			}
245
246			/**
247			* Top Image object that we think represents this article
248			*
249			* @var Image\|null
250			*/
251			protected $topImage;
252
253			/**
254			* @param Image\|null $topImage
255			*
256			* @return self
257			*/
258			public function setTopImage(Image $topImage = null): self {
259			$this->topImage = $topImage;
260
261			return $this;
262			}
263
264			/** @return Image\|null */
265			public function getTopImage(): ?Image {
266			return $this->topImage;
267			}
268
269			/**
270			* All image candidates from article
271			*
272			* @var Image[]
273			*/
274			protected $allImages = [];
275
276			/**
277			* @param Image[] $allImages
278			*
279			* @return self
280			*/
281			public function setAllImages(array $allImages = []): self {
282			$this->allImages = $allImages;
283
284			return $this;
285			}
286
287			/** @return Image[] */
288			public function getAllImages(): array {
289			return $this->allImages;
290			}
291
292			/**
293			* Tags that may have been in the article, these are not meta keywords
294			*
295			* @var string[]
296			*/
297			protected $tags = [];
298
299			/**
300			* @param string[] $tags
301			*
302			* @return self
303			*/
304			public function setTags(array $tags): self {
305			$this->tags = $tags;
306
307			return $this;
308			}
309
310			/** @return string[] */
311			public function getTags(): array {
312			return $this->tags;
313			}
314
315			/**
316			* List of links in the article
317			*
318			* @var string[]
319			*/
320			protected $links = [];
321
322			/**
323			* @param string[] $links
324			*
325			* @return self
326			*/
327			public function setLinks(array $links): self {
328			$this->links = $links;
329
330			return $this;
331			}
332
333			/** @return string[] */
334			public function getLinks(): array {
335			return $this->links;
336			}
337
338			/**
339			* List of any videos we found on the page like youtube, vimeo
340			*
341			* @var string[]
342			*/
343			protected $videos = [];
344
345			/**
346			* @param string[] $videos
347			*
348			* @return self
349			*/
350			public function setVideos(array $videos): self {
351			$this->videos = $videos;
352
353			return $this;
354			}
355
356			/** @return string[] */
357			public function getVideos(): array {
358			return $this->videos;
359			}
360
361			/**
362			* Final URL that we're going to try and fetch content against, this would be expanded if any
363			* escaped fragments were found in the starting url
364			*
365			* @var string\|null
366			*/
367			protected $finalUrl;
368
369			/**
370			* @param string $finalUrl
371			*
372			* @return self
373			*/
374			public function setFinalUrl(string $finalUrl = null): self {
375			$this->finalUrl = $finalUrl;
376
377			return $this;
378			}
379
380			/** @return string\|null */
381			public function getFinalUrl(): ?string {
382			return $this->finalUrl;
383			}
384
385			/**
386			* MD5 hash of the url to use for various identification tasks
387			*
388			* @var string\|null
389			*/
390			protected $linkhash;
391
392			/**
393			* @param string $linkhash
394			*
395			* @return self
396			*/
397			public function setLinkhash(string $linkhash = null): self {
398			$this->linkhash = $linkhash;
399
400			return $this;
401			}
402
403			/** @return string */
404			public function getLinkhash(): ?string {
405			return $this->linkhash;
406			}
407
408			/**
409			* Raw HTML straight from the network connection
410			*
411			* @var string\|null
412			*/
413			protected $rawHtml;
414
415			/**
416			* @param string $rawHtml
417			*
418			* @return self
419			*/
420			public function setRawHtml(string $rawHtml = null): self {
421			$this->rawHtml = $rawHtml;
422
423			return $this;
424			}
425
426			/** @return string\|null */
427			public function getRawHtml(): ?string {
428			return $this->rawHtml;
429			}
430
431			/**
432			* DOM Document object
433			*
434			* @var Document
435			*/
436			protected $doc;
437
438			/**
439			* @param Document $doc
440			*
441			* @return self
442			*/
443			public function setDoc(Document $doc): self {
444			$this->doc = $doc;
445
446			return $this;
447			}
448
449			/** @return Document */
450			public function getDoc(): Document {
451			return $this->doc;
452			}
453
454			/**
455			* Original DOM document that contains a pure object from the original HTML without any cleaning
456			* options done on it
457			*
458			* @var Document
459			*/
460			protected $rawDoc;
461
462			/**
463			* @param Document $rawDoc
464			*
465			* @return self
466			*/
467			public function setRawDoc(Document $rawDoc): self {
468			$this->rawDoc = $rawDoc;
469
470			return $this;
471			}
472
473			/** @return Document */
474			public function getRawDoc(): Document {
475			return $this->rawDoc;
476			}
477
478			/**
479			* Original psr7 response object
480			*
481			* @var \Psr\Http\Message\ResponseInterface
482			*/
483			protected $rawResponse;
484
485			/**
486			* @param \Psr\Http\Message\ResponseInterface\|null $rawResponse
487			*
488			* @return self
489			*/
490			public function setRawResponse(\Psr\Http\Message\ResponseInterface $rawResponse): self {
491			$this->rawResponse = $rawResponse;
492
493			return $this;
494			}
495
496			/** @return \Psr\Http\Message\ResponseInterface */
497			public function getRawResponse(): \Psr\Http\Message\ResponseInterface {
498			return $this->rawResponse;
499			}
500
501			/**
502			* Sometimes useful to try and know when the publish date of an article was
503			*
504			* @var \DateTime\|null
505			*/
506			protected $publishDate;
507
508			/**
509			* @param \DateTime\|null $publishDate
510			*
511			* @return self
512			*/
513			public function setPublishDate(\DateTime $publishDate = null): self {
514			$this->publishDate = $publishDate;
515
516			return $this;
517			}
518
519			/** @return \DateTime\|null */
520			public function getPublishDate(): ?\DateTime {
521			return $this->publishDate;
522			}
523
524			/**
525			* A property bucket for consumers of goose to store custom data extractions.
526			*
527			* @var string\|null
528			*/
529			protected $additionalData;
530
531			/**
532			* @param string\|null $additionalData
533			*
534			* @return self
535			*/
536			public function setAdditionalData(string $additionalData = null): self {
537			$this->additionalData = $additionalData;
538
539			return $this;
540			}
541
542			/** @return string\|null */
543			public function getAdditionalData(): ?string {
544			return $this->additionalData;
545			}
546
547			/**
548			* Facebook Open Graph data that that is found in Article Meta tags
549			*
550			* @var string\|null
551			*/
552			protected $openGraphData;
553
554			/**
555			* @param string $openGraphData
556			*
557			* @return self
558			*/
559			public function setOpenGraphData(string $openGraphData = null): self {
560			$this->openGraphData = $openGraphData;
561
562			return $this;
563			}
564
565			/** @return string\|null */
566			public function getOpenGraphData(): ?string {
567			return $this->openGraphData;
568			}
569
570			/**
571			* Most popular words used in the lovely article
572			*
573			* @var string[]
574			*/
575			protected $popularWords = [];
576
577			/**
578			* @param string[] $popularWords
579			*
580			* @return self
581			*/
582			public function setPopularWords(array $popularWords): self {
583			$this->popularWords = $popularWords;
584
585			return $this;
586			}
587
588			/** @return string[] */
589			public function getPopularWords(): array {
590			return $this->popularWords;
591			}
592			}

scotteh / php-goose

Article::setRawDoc() A last analyzed 2023-09-05 13:38 UTC

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like

Article::setRawDoc() A
last analyzed 2023-09-05 13:38 UTC