PublishDateExtractor::getDateFromSchemaOrg() - Code Metrics - Inspection of "Fix return type" - scotteh/php-goose - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Branch — master (803fe5)

by Andrew

created 2018-02-14 12:28 UTC

PublishDateExtractor::getDateFromSchemaOrg() D

↳ Parent: PublishDateExtractor

Complexity

Conditions	9
Paths	48

Size

Total Lines	45
Code Lines	22

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
dl	0
loc	45
c	0
b	0
f	0
rs	4.909
cc	9
eloc	22
nc	48
nop	0

<?php declare(strict_types=1);

namespace Goose\Modules\Extractors;

use Goose\Article;
use Goose\Traits\ArticleMutatorTrait;
use Goose\Modules\{AbstractModule, ModuleInterface};
use DOMWrap\Element;

/**
 * Publish Date Extractor
 *
 * @package Goose\Modules\Extractors
 * @license http://www.apache.org/licenses/LICENSE-2.0 Apache License 2.0
 */
class PublishDateExtractor extends AbstractModule implements ModuleInterface {
    use ArticleMutatorTrait;

    /** @inheritdoc  */
    public function run(Article $article): self {
        $this->article($article);

        $dt = null;


        $dt = $this->getDateFromSchemaOrg();

        if (is_null($dt)) {
            $dt = $this->getDateFromOpenGraph();
        }

        if (is_null($dt)) {
            $dt = $this->getDateFromURL();
        }

        if (is_null($dt)) {
            $dt = $this->getDateFromDublinCore();
        }

        if (is_null($dt)) {
            $dt = $this->getDateFromParsely();
        }

        $article->setPublishDate($dt);

        return $this;
    }

    /**
     * @return \DateTime|null
     */
    private function getDateFromURL(): ?\DateTime {
        // Determine date based on URL
        if (preg_match('@(?:[\d]{4})(?<delimiter>[/-])(?:[\d]{2})\k<delimiter>(?:[\d]{2})@U', $this->article()->getFinalUrl(), $matches)) {
            $dt = \DateTime::createFromFormat('Y' . $matches['delimiter'] . 'm' . $matches['delimiter'] . 'd', $matches[0]);
            $dt->setTime(0, 0, 0);

            if ($dt === false) {
                return null;
            }

            return $dt;
        }

        /** @todo Add more date detection methods */

        return null;
    }

    /**
     * Check for and determine dates from Schema.org's datePublished property.
     *
     * Checks HTML tags (e.g. <meta>, <time>, etc.) and JSON-LD.
     *
     * @return \DateTime|null
     *
     * @see https://schema.org/datePublished
     */
    private function getDateFromSchemaOrg(): ?\DateTime {
        $dt = null;

        // Check for HTML tags (<meta>, <time>, etc.)
        $nodes = $this->article()->getRawDoc()->find('*[itemprop="datePublished"]');

        /* @var $node Element */
        foreach ($nodes as $node) {
            try {
                if ($node->hasAttribute('datetime')) {
                    $dt = new \DateTime($node->getAttribute('datetime'));
                    break;
                }
                if ($node->hasAttribute('content')) {
                    $dt = new \DateTime($node->getAttribute('content'));
                    break;
                }
            }
            catch (\Exception $e) {
                // Do nothing here in case the node has unrecognizable date information.
            }
        }

        if (!is_null($dt)) {
            return $dt;
        }

        // Check for JSON-LD
        $nodes = $this->article()->getRawDoc()->find('script[type="application/ld+json"]');

        /* @var $node Element */
        foreach ($nodes as $node) {
            try {
                $json = json_decode($node->text());
                if (isset($json->datePublished)) {
                    $dt = new \DateTime($json->datePublished);
                    break;
                }
            }
            catch (\Exception $e) {
                // Do nothing here in case the node has unrecognizable date information.
            }
        }

        return $dt;
    }

    /**
     * Check for and determine dates based on Dublin Core standards.
     *
     * @return \DateTime|null
     *
     * @see http://dublincore.org/documents/dcmi-terms/#elements-date
     * @see http://dublincore.org/documents/2000/07/16/usageguide/qualified-html.shtml
     */
    private function getDateFromDublinCore(): ?\DateTime {
        $dt = null;
        $nodes = $this->article()->getRawDoc()->find('*[name="dc.date"], *[name="dc.date.issued"], *[name="DC.date.issued"]');

        /* @var $node Element */
        foreach ($nodes as $node) {
            try {
                if ($node->hasAttribute('content')) {
                    $dt = new \DateTime($node->getAttribute('content'));
                    break;
                }
            }
            catch (\Exception $e) {
                // Do nothing here in case the node has unrecognizable date information.
            }
        }

        if (!is_null($dt)) {
            return $dt;
        }

        return $dt;
    }

    /**
     * Check for and determine dates based on OpenGraph standards.
     *
     * @return \DateTime|null
     *
     * @see http://ogp.me/
     * @see http://ogp.me/#type_article
     */
    private function getDateFromOpenGraph(): ?\DateTime {
        $dt = null;

        $og_data = $this->article()->getOpenGraph();

        try {
            if (isset($og_data['published_time'])) {
                $dt = new \DateTime($og_data['published_time']);
            }
            if (is_null($dt) && isset($og_data['pubdate'])) {
                $dt = new \DateTime($og_data['pubdate']);
            }
        }
        catch (\Exception $e) {
            // Do nothing here in case the node has unrecognizable date information.
        }

        return $dt;
    }

    /**
     * Check for and determine dates based on Parsely metadata.
     *
     * Checks JSON-LD, <meta> tags and parsely-page.
     *
     * @return \DateTime|null
     *
     * @see https://www.parsely.com/help/integration/jsonld/
     * @see https://www.parsely.com/help/integration/metatags/
     * @see https://www.parsely.com/help/integration/ppage/
     */
    private function getDateFromParsely(): ?\DateTime {
        $dt = null;

        // JSON-LD
        $nodes = $this->article()->getRawDoc()->find('script[type="application/ld+json"]');

        /* @var $node Element */
        foreach ($nodes as $node) {
            try {
                $json = json_decode($node->text());
                if (isset($json->dateCreated)) {
                    $dt = new \DateTime($json->dateCreated);
                    break;
                }
            }
            catch (\Exception $e) {
                // Do nothing here in case the node has unrecognizable date information.
            }
        }

        if (!is_null($dt)) {
            return $dt;
        }

        // <meta> tags
        $nodes = $this->article()->getRawDoc()->find('meta[name="parsely-pub-date"]');

        /* @var $node Element */
        foreach ($nodes as $node) {
            try {
                if ($node->hasAttribute('content')) {
                    $dt = new \DateTime($node->getAttribute('content'));
                    break;
                }
            }
            catch (\Exception $e) {
                // Do nothing here in case the node has unrecognizable date information.
            }
        }

        if (!is_null($dt)) {
            return $dt;
        }

        // parsely-page
        $nodes = $this->article()->getRawDoc()->find('meta[name="parsely-page"]');

        /* @var $node Element */
        foreach ($nodes as $node) {
            try {
                if ($node->hasAttribute('content')) {
                    $json = json_decode($node->getAttribute('content'));
                    if (isset($json->pub_date)) {
                        $dt = new \DateTime($json->pub_date);
                        break;
                    }
                }
            }
            catch (\Exception $e) {
                // Do nothing here in case the node has unrecognizable date information.
            }
        }

        return $dt;
    }
}


1			<?php declare(strict_types=1);
2
3			namespace Goose\Modules\Extractors;
4
5			use Goose\Article;
6			use Goose\Traits\ArticleMutatorTrait;
7			use Goose\Modules\{AbstractModule, ModuleInterface};
8			use DOMWrap\Element;
9
10			/**
11			* Publish Date Extractor
12			*
13			* @package Goose\Modules\Extractors
14			* @license http://www.apache.org/licenses/LICENSE-2.0 Apache License 2.0
15			*/
16			class PublishDateExtractor extends AbstractModule implements ModuleInterface {
17			use ArticleMutatorTrait;
18
19			/** @inheritdoc */
20			public function run(Article $article): self {
21			$this->article($article);
22
23			$dt = null;
			0 ignored issues – show Unused Code introduced 2018-02-14 12:31 UTC by Report Bug Copy Issue Report The assignment to `$dt` is dead and can be removed. Loading history...
24
25			$dt = $this->getDateFromSchemaOrg();
26
27			if (is_null($dt)) {
28			$dt = $this->getDateFromOpenGraph();
29			}
30
31			if (is_null($dt)) {
32			$dt = $this->getDateFromURL();
33			}
34
35			if (is_null($dt)) {
36			$dt = $this->getDateFromDublinCore();
37			}
38
39			if (is_null($dt)) {
40			$dt = $this->getDateFromParsely();
41			}
42
43			$article->setPublishDate($dt);
44
45			return $this;
46			}
47
48			/**
49			* @return \DateTime\|null
50			*/
51			private function getDateFromURL(): ?\DateTime {
52			// Determine date based on URL
53			if (preg_match('@(?:[\d]{4})(?<delimiter>[/-])(?:[\d]{2})\k<delimiter>(?:[\d]{2})@U', $this->article()->getFinalUrl(), $matches)) {
54			$dt = \DateTime::createFromFormat('Y' . $matches['delimiter'] . 'm' . $matches['delimiter'] . 'd', $matches[0]);
55			$dt->setTime(0, 0, 0);
56
57			if ($dt === false) {
58			return null;
59			}
60
61			return $dt;
62			}
63
64			/** @todo Add more date detection methods */
65
66			return null;
67			}
68
69			/**
70			* Check for and determine dates from Schema.org's datePublished property.
71			*
72			* Checks HTML tags (e.g. <meta>, <time>, etc.) and JSON-LD.
73			*
74			* @return \DateTime\|null
75			*
76			* @see https://schema.org/datePublished
77			*/
78			private function getDateFromSchemaOrg(): ?\DateTime {
79			$dt = null;
80
81			// Check for HTML tags (<meta>, <time>, etc.)
82			$nodes = $this->article()->getRawDoc()->find('*[itemprop="datePublished"]');
83
84			/* @var $node Element */
85			foreach ($nodes as $node) {
86			try {
87			if ($node->hasAttribute('datetime')) {
88			$dt = new \DateTime($node->getAttribute('datetime'));
89			break;
90			}
91			if ($node->hasAttribute('content')) {
92			$dt = new \DateTime($node->getAttribute('content'));
93			break;
94			}
95			}
96			catch (\Exception $e) {
97			// Do nothing here in case the node has unrecognizable date information.
98			}
99			}
100
101			if (!is_null($dt)) {
102			return $dt;
103			}
104
105			// Check for JSON-LD
106			$nodes = $this->article()->getRawDoc()->find('script[type="application/ld+json"]');
107
108			/* @var $node Element */
109			foreach ($nodes as $node) {
110			try {
111			$json = json_decode($node->text());
112			if (isset($json->datePublished)) {
113			$dt = new \DateTime($json->datePublished);
114			break;
115			}
116			}
117			catch (\Exception $e) {
118			// Do nothing here in case the node has unrecognizable date information.
119			}
120			}
121
122			return $dt;
123			}
124
125			/**
126			* Check for and determine dates based on Dublin Core standards.
127			*
128			* @return \DateTime\|null
129			*
130			* @see http://dublincore.org/documents/dcmi-terms/#elements-date
131			* @see http://dublincore.org/documents/2000/07/16/usageguide/qualified-html.shtml
132			*/
133			private function getDateFromDublinCore(): ?\DateTime {
134			$dt = null;
135			$nodes = $this->article()->getRawDoc()->find('[name="dc.date"], [name="dc.date.issued"], *[name="DC.date.issued"]');
136
137			/* @var $node Element */
138			foreach ($nodes as $node) {
139			try {
140			if ($node->hasAttribute('content')) {
141			$dt = new \DateTime($node->getAttribute('content'));
142			break;
143			}
144			}
145			catch (\Exception $e) {
146			// Do nothing here in case the node has unrecognizable date information.
147			}
148			}
149
150			if (!is_null($dt)) {
151			return $dt;
152			}
153
154			return $dt;
155			}
156
157			/**
158			* Check for and determine dates based on OpenGraph standards.
159			*
160			* @return \DateTime\|null
161			*
162			* @see http://ogp.me/
163			* @see http://ogp.me/#type_article
164			*/
165			private function getDateFromOpenGraph(): ?\DateTime {
166			$dt = null;
167
168			$og_data = $this->article()->getOpenGraph();
169
170			try {
171			if (isset($og_data['published_time'])) {
172			$dt = new \DateTime($og_data['published_time']);
173			}
174			if (is_null($dt) && isset($og_data['pubdate'])) {
175			$dt = new \DateTime($og_data['pubdate']);
176			}
177			}
178			catch (\Exception $e) {
179			// Do nothing here in case the node has unrecognizable date information.
180			}
181
182			return $dt;
183			}
184
185			/**
186			* Check for and determine dates based on Parsely metadata.
187			*
188			* Checks JSON-LD, <meta> tags and parsely-page.
189			*
190			* @return \DateTime\|null
191			*
192			* @see https://www.parsely.com/help/integration/jsonld/
193			* @see https://www.parsely.com/help/integration/metatags/
194			* @see https://www.parsely.com/help/integration/ppage/
195			*/
196			private function getDateFromParsely(): ?\DateTime {
197			$dt = null;
198
199			// JSON-LD
200			$nodes = $this->article()->getRawDoc()->find('script[type="application/ld+json"]');
201
202			/* @var $node Element */
203			foreach ($nodes as $node) {
204			try {
205			$json = json_decode($node->text());
206			if (isset($json->dateCreated)) {
207			$dt = new \DateTime($json->dateCreated);
208			break;
209			}
210			}
211			catch (\Exception $e) {
212			// Do nothing here in case the node has unrecognizable date information.
213			}
214			}
215
216			if (!is_null($dt)) {
217			return $dt;
218			}
219
220			// <meta> tags
221			$nodes = $this->article()->getRawDoc()->find('meta[name="parsely-pub-date"]');
222
223			/* @var $node Element */
224			foreach ($nodes as $node) {
225			try {
226			if ($node->hasAttribute('content')) {
227			$dt = new \DateTime($node->getAttribute('content'));
228			break;
229			}
230			}
231			catch (\Exception $e) {
232			// Do nothing here in case the node has unrecognizable date information.
233			}
234			}
235
236			if (!is_null($dt)) {
237			return $dt;
238			}
239
240			// parsely-page
241			$nodes = $this->article()->getRawDoc()->find('meta[name="parsely-page"]');
242
243			/* @var $node Element */
244			foreach ($nodes as $node) {
245			try {
246			if ($node->hasAttribute('content')) {
247			$json = json_decode($node->getAttribute('content'));
248			if (isset($json->pub_date)) {
249			$dt = new \DateTime($json->pub_date);
250			break;
251			}
252			}
253			}
254			catch (\Exception $e) {
255			// Do nothing here in case the node has unrecognizable date information.
256			}
257			}
258
259			return $dt;
260			}
261			}
262

scotteh / php-goose

Branch — master (803fe5)

PublishDateExtractor::getDateFromSchemaOrg() D

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like