FeedFetcher::fetch() - Code Metrics - Inspection of "step back as maintainer" - owncloud/news - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Push — master ( 150a3d...219536 )

unknown

created 2016-06-14 16:10 UTC

FeedFetcher::fetch() C

↳ Parent: FeedFetcher

Complexity

Conditions	7
Paths	53

Size

Total Lines	46
Code Lines	29

Duplication

Lines	0
Ratio	0 %

Importance

Changes	2
Bugs	0	Features	1

Metric	Value
c	2
b	0
f	1
dl	0
loc	46
rs	6.7272
cc	7
eloc	29
nc	53
nop	7

<?php
/**
 * ownCloud - News
 *
 * This file is licensed under the Affero General Public License version 3 or
 * later. See the COPYING file.
 *
 * @author Alessandro Cosentino <[email protected]>
 * @author Bernhard Posselt <[email protected]>
 * @copyright Alessandro Cosentino 2012
 * @copyright Bernhard Posselt 2012, 2014
 */

namespace OCA\News\Fetcher;

use Exception;

use PicoFeed\Parser\MalFormedXmlException;
use PicoFeed\Reader\Reader;
use PicoFeed\Parser\Parser;
use PicoFeed\Reader\SubscriptionNotFoundException;
use PicoFeed\Reader\UnsupportedFeedFormatException;
use PicoFeed\Client\InvalidCertificateException;
use PicoFeed\Client\InvalidUrlException;
use PicoFeed\Client\MaxRedirectException;
use PicoFeed\Client\MaxSizeException;
use PicoFeed\Client\TimeoutException;
use PicoFeed\Client\ForbiddenException;
use PicoFeed\Client\UnauthorizedException;

use OCP\IL10N;
use OCP\AppFramework\Utility\ITimeFactory;

use OCA\News\Db\Item;
use OCA\News\Db\Feed;
use OCA\News\Utility\PicoFeedFaviconFactory;
use OCA\News\Utility\PicoFeedReaderFactory;

class FeedFetcher implements IFeedFetcher {

    private $faviconFactory;
    private $reader;
    private $l10n;
    private $time;

    public function __construct(Reader $reader,
                                PicoFeedFaviconFactory $faviconFactory,
                                IL10N $l10n,
                                ITimeFactory $time) {
        $this->faviconFactory = $faviconFactory;
        $this->reader = $reader;
        $this->time = $time;
        $this->l10n = $l10n;
    }


    /**
     * This fetcher handles all the remaining urls therefore always returns true
     */
    public function canHandle($url) {
        return true;
    }


    /**
     * Fetch a feed from remote
     * @param string $url remote url of the feed
     * @param boolean $getFavicon if the favicon should also be fetched,
     * defaults to true
     * @param string $lastModified a last modified value from an http header
     * defaults to false. If lastModified matches the http header from the feed
     * no results are fetched
     * @param string $etag an etag from an http header.
     * If lastModified matches the http header from the feed
     * no results are fetched
     * @param bool fullTextEnabled if true tells the fetcher to enhance the
     * articles by fetching custom enhanced content
     * @param string $basicAuthUser if given, basic auth is set for this feed
     * @param string $basicAuthPassword if given, basic auth is set for this
     * feed. Ignored if user is null or an empty string
     * @throws FetcherException if it fails
     * @return array an array containing the new feed and its items, first
     * element being the Feed and second element being an array of Items
     */
    public function fetch($url, $getFavicon = true, $lastModified = null,
                          $etag = null, $fullTextEnabled = false,
                          $basicAuthUser = null, $basicAuthPassword = null) {
        try {
            if ($basicAuthUser !== null && trim($basicAuthUser) !== '') {
                $resource = $this->reader->discover($url, $lastModified, $etag,
                    $basicAuthUser,
                    $basicAuthPassword);
            } else {
                $resource = $this->reader->discover($url, $lastModified, $etag);
            }

            if (!$resource->isModified()) {
                return [null, null];
            }

            $location = $resource->getUrl();
            $etag = $resource->getEtag();
            $content = $resource->getContent();
            $encoding = $resource->getEncoding();
            $lastModified = $resource->getLastModified();

            $parser = $this->reader->getParser($location, $content, $encoding);

            if ($fullTextEnabled) {
                $parser->enableContentGrabber();
            }

            $parsedFeed = $parser->execute();

            $feed = $this->buildFeed(
                $parsedFeed, $url, $getFavicon, $lastModified, $etag, $location
            );

            $items = [];
            foreach ($parsedFeed->getItems() as $item) {
                $items[] = $this->buildItem($item, $parsedFeed);
            }

            return [$feed, $items];

        } catch (Exception $ex) {
            $this->handleError($ex, $url);
        }

    }


    private function handleError(Exception $ex, $url) {
        $msg = $ex->getMessage();

        if ($ex instanceof MalFormedXmlException) {
            $msg = $this->l10n->t('Feed contains invalid XML');
        } else if ($ex instanceof SubscriptionNotFoundException) {
            $msg = $this->l10n->t('Feed not found: either the website ' .
                'does not provide a feed or blocks access. To rule out ' .
                'blocking, try to download the feed on your server\'s ' .
                'command line using curl: curl ' . $url);
        } else if ($ex instanceof UnsupportedFeedFormatException) {
            $msg = $this->l10n->t('Detected feed format is not supported');
        } else if ($ex instanceof InvalidCertificateException) {
            $msg = $this->buildCurlSslErrorMessage($ex->getCode());
        } else if ($ex instanceof InvalidUrlException) {
            $msg = $this->l10n->t('Website not found');
        } else if ($ex instanceof MaxRedirectException) {
            $msg = $this->l10n->t('More redirects than allowed, aborting');
        } else if ($ex instanceof MaxSizeException) {
            $msg = $this->l10n->t('Bigger than maximum allowed size');
        } else if ($ex instanceof TimeoutException) {
            $msg = $this->l10n->t('Request timed out');
        } else if ($ex instanceof UnauthorizedException) {
            $msg = $this->l10n->t('Required credentials for feed were ' .
                'either missing or incorrect');
        } else if ($ex instanceof ForbiddenException) {
            $msg = $this->l10n->t('Forbidden to access feed');
        }

        throw new FetcherException($msg);
    }

    private function buildCurlSslErrorMessage($errorCode) {
        switch ($errorCode) {
            case 35: // CURLE_SSL_CONNECT_ERROR
                return $this->l10n->t(
                    'Certificate error: A problem occurred ' .
                    'somewhere in the SSL/TLS handshake. Could be ' .
                    'certificates (file formats, paths, permissions), ' .
                    'passwords, and others.'
                );
            case 51: // CURLE_PEER_FAILED_VERIFICATION
                return $this->l10n->t(
                    'Certificate error: The remote server\'s SSL ' .
                    'certificate or SSH md5 fingerprint was deemed not OK.'
                );
            case 58: // CURLE_SSL_CERTPROBLEM
                return $this->l10n->t(
                    'Certificate error: Problem with the local client ' .
                    'certificate.'
                );
            case 59: // CURLE_SSL_CIPHER
                return $this->l10n->t(
                    'Certificate error: Couldn\'t use specified cipher.'
                );
            case 60: // CURLE_SSL_CACERT
                return $this->l10n->t(
                    'Certificate error: Peer certificate cannot be ' .
                    'authenticated with known CA certificates.'
                );
            case 64: // CURLE_USE_SSL_FAILED
                return $this->l10n->t(
                    'Certificate error: Requested FTP SSL level failed.'
                );
            case 66: // CURLE_SSL_ENGINE_INITFAILED
                return $this->l10n->t(
                    'Certificate error: Initiating the SSL Engine failed.'
                );
            case 77: // CURLE_SSL_CACERT_BADFILE
                return $this->l10n->t(
                    'Certificate error: Problem with reading the SSL CA ' .
                    'cert (path? access rights?)'
                );
            case 83: // CURLE_SSL_ISSUER_ERROR
                return $this->l10n->t(
                    'Certificate error: Issuer check failed'
                );
            default:
                return $this->l10n->t('Unknown SSL certificate error!');
        }
    }

    private function decodeTwice($string) {
        return html_entity_decode(
            html_entity_decode(
                $string, ENT_QUOTES | ENT_HTML5, 'UTF-8'
            ),
            ENT_QUOTES | ENT_HTML5, 'UTF-8'
        );
    }


    protected function determineRtl($parsedItem, $parsedFeed) {
        $itemLang = $parsedItem->getLanguage();
        $feedLang = $parsedFeed->getLanguage();

        if ($itemLang) {
            return Parser::isLanguageRTL($itemLang);
        } else {
            return Parser::isLanguageRTL($feedLang);
        }
    }


    protected function buildItem($parsedItem, $parsedFeed) {
        $item = new Item();
        $item->setUnread();
        $item->setUrl($parsedItem->getUrl());
        $item->setGuid($parsedItem->getId());
        $item->setGuidHash($item->getGuid());
        $item->setPubDate($parsedItem->getDate()->getTimestamp());
        $item->setLastModified($this->time->getTime());
        $item->setRtl($this->determineRtl($parsedItem, $parsedFeed));

        // unescape content because angularjs helps against XSS
        $item->setTitle($this->decodeTwice($parsedItem->getTitle()));
        $item->setAuthor($this->decodeTwice($parsedItem->getAuthor()));

        // purification is done in the service layer
        $body = $parsedItem->getContent();
        $body = mb_convert_encoding($body, 'HTML-ENTITIES',
            mb_detect_encoding($body));
        $item->setBody($body);

        $enclosureUrl = $parsedItem->getEnclosureUrl();
        if ($enclosureUrl) {
            $enclosureType = $parsedItem->getEnclosureType();
            if (stripos($enclosureType, 'audio/') !== false ||
                stripos($enclosureType, 'video/') !== false
            ) {
                $item->setEnclosureMime($enclosureType);
                $item->setEnclosureLink($enclosureUrl);
            }
        }

        $item->generateSearchIndex();

        return $item;
    }


    protected function buildFeed($parsedFeed, $url, $getFavicon, $modified,
                                 $etag, $location) {
        $feed = new Feed();

        $link = $parsedFeed->getSiteUrl();

        if (!$link) {
            $link = $location;
        }

        // unescape content because angularjs helps against XSS
        $title = strip_tags($this->decodeTwice($parsedFeed->getTitle()));
        $feed->setTitle($title);
        $feed->setUrl($url);  // the url used to add the feed
        $feed->setLocation($location);  // the url where the feed was found
        $feed->setLink($link);  // <link> attribute in the feed
        $feed->setHttpLastModified($modified);
        $feed->setHttpEtag($etag);
        $feed->setAdded($this->time->getTime());

        if ($getFavicon) {
            $faviconFetcher = $this->faviconFactory->build();
            $favicon = $faviconFetcher->find($feed->getLink());
            $feed->setFaviconLink($favicon);
        }

        return $feed;
    }

}


1			<?php
2			/**
3			* ownCloud - News
4			*
5			* This file is licensed under the Affero General Public License version 3 or
6			* later. See the COPYING file.
7			*
8			* @author Alessandro Cosentino <[email protected]>
9			* @author Bernhard Posselt <[email protected]>
10			* @copyright Alessandro Cosentino 2012
11			* @copyright Bernhard Posselt 2012, 2014
12			*/
13
14			namespace OCA\News\Fetcher;
15
16			use Exception;
17
18			use PicoFeed\Parser\MalFormedXmlException;
19			use PicoFeed\Reader\Reader;
20			use PicoFeed\Parser\Parser;
21			use PicoFeed\Reader\SubscriptionNotFoundException;
22			use PicoFeed\Reader\UnsupportedFeedFormatException;
23			use PicoFeed\Client\InvalidCertificateException;
24			use PicoFeed\Client\InvalidUrlException;
25			use PicoFeed\Client\MaxRedirectException;
26			use PicoFeed\Client\MaxSizeException;
27			use PicoFeed\Client\TimeoutException;
28			use PicoFeed\Client\ForbiddenException;
29			use PicoFeed\Client\UnauthorizedException;
30
31			use OCP\IL10N;
32			use OCP\AppFramework\Utility\ITimeFactory;
33
34			use OCA\News\Db\Item;
35			use OCA\News\Db\Feed;
36			use OCA\News\Utility\PicoFeedFaviconFactory;
37			use OCA\News\Utility\PicoFeedReaderFactory;
38
39			class FeedFetcher implements IFeedFetcher {
40
41			private $faviconFactory;
42			private $reader;
43			private $l10n;
44			private $time;
45
46			public function __construct(Reader $reader,
47			PicoFeedFaviconFactory $faviconFactory,
48			IL10N $l10n,
49			ITimeFactory $time) {
50			$this->faviconFactory = $faviconFactory;
51			$this->reader = $reader;
52			$this->time = $time;
53			$this->l10n = $l10n;
54			}
55
56
57			/**
58			* This fetcher handles all the remaining urls therefore always returns true
59			*/
60			public function canHandle($url) {
61			return true;
62			}
63
64
65			/**
66			* Fetch a feed from remote
67			* @param string $url remote url of the feed
68			* @param boolean $getFavicon if the favicon should also be fetched,
69			* defaults to true
70			* @param string $lastModified a last modified value from an http header
71			* defaults to false. If lastModified matches the http header from the feed
72			* no results are fetched
73			* @param string $etag an etag from an http header.
74			* If lastModified matches the http header from the feed
75			* no results are fetched
76			* @param bool fullTextEnabled if true tells the fetcher to enhance the
77			* articles by fetching custom enhanced content
78			* @param string $basicAuthUser if given, basic auth is set for this feed
79			* @param string $basicAuthPassword if given, basic auth is set for this
80			* feed. Ignored if user is null or an empty string
81			* @throws FetcherException if it fails
82			* @return array an array containing the new feed and its items, first
83			* element being the Feed and second element being an array of Items
84			*/
85			public function fetch($url, $getFavicon = true, $lastModified = null,
86			$etag = null, $fullTextEnabled = false,
87			$basicAuthUser = null, $basicAuthPassword = null) {
88			try {
89			if ($basicAuthUser !== null && trim($basicAuthUser) !== '') {
90			$resource = $this->reader->discover($url, $lastModified, $etag,
91			$basicAuthUser,
92			$basicAuthPassword);
93			} else {
94			$resource = $this->reader->discover($url, $lastModified, $etag);
95			}
96
97			if (!$resource->isModified()) {
98			return [null, null];
99			}
100
101			$location = $resource->getUrl();
102			$etag = $resource->getEtag();
103			$content = $resource->getContent();
104			$encoding = $resource->getEncoding();
105			$lastModified = $resource->getLastModified();
106
107			$parser = $this->reader->getParser($location, $content, $encoding);
108
109			if ($fullTextEnabled) {
110			$parser->enableContentGrabber();
111			}
112
113			$parsedFeed = $parser->execute();
114
115			$feed = $this->buildFeed(
116			$parsedFeed, $url, $getFavicon, $lastModified, $etag, $location
117			);
118
119			$items = [];
120			foreach ($parsedFeed->getItems() as $item) {
121			$items[] = $this->buildItem($item, $parsedFeed);
122			}
123
124			return [$feed, $items];
125
126			} catch (Exception $ex) {
127			$this->handleError($ex, $url);
128			}
129
130			}
131
132
133			private function handleError(Exception $ex, $url) {
134			$msg = $ex->getMessage();
135
136			if ($ex instanceof MalFormedXmlException) {
137			$msg = $this->l10n->t('Feed contains invalid XML');
138			} else if ($ex instanceof SubscriptionNotFoundException) {
139			$msg = $this->l10n->t('Feed not found: either the website ' .
140			'does not provide a feed or blocks access. To rule out ' .
141			'blocking, try to download the feed on your server\'s ' .
142			'command line using curl: curl ' . $url);
143			} else if ($ex instanceof UnsupportedFeedFormatException) {
144			$msg = $this->l10n->t('Detected feed format is not supported');
145			} else if ($ex instanceof InvalidCertificateException) {
146			$msg = $this->buildCurlSslErrorMessage($ex->getCode());
147			} else if ($ex instanceof InvalidUrlException) {
148			$msg = $this->l10n->t('Website not found');
149			} else if ($ex instanceof MaxRedirectException) {
150			$msg = $this->l10n->t('More redirects than allowed, aborting');
151			} else if ($ex instanceof MaxSizeException) {
152			$msg = $this->l10n->t('Bigger than maximum allowed size');
153			} else if ($ex instanceof TimeoutException) {
154			$msg = $this->l10n->t('Request timed out');
155			} else if ($ex instanceof UnauthorizedException) {
156			$msg = $this->l10n->t('Required credentials for feed were ' .
157			'either missing or incorrect');
158			} else if ($ex instanceof ForbiddenException) {
159			$msg = $this->l10n->t('Forbidden to access feed');
160			}
161
162			throw new FetcherException($msg);
163			}
164
165			private function buildCurlSslErrorMessage($errorCode) {
166			switch ($errorCode) {
167			case 35: // CURLE_SSL_CONNECT_ERROR
168			return $this->l10n->t(
169			'Certificate error: A problem occurred ' .
170			'somewhere in the SSL/TLS handshake. Could be ' .
171			'certificates (file formats, paths, permissions), ' .
172			'passwords, and others.'
173			);
174			case 51: // CURLE_PEER_FAILED_VERIFICATION
175			return $this->l10n->t(
176			'Certificate error: The remote server\'s SSL ' .
177			'certificate or SSH md5 fingerprint was deemed not OK.'
178			);
179			case 58: // CURLE_SSL_CERTPROBLEM
180			return $this->l10n->t(
181			'Certificate error: Problem with the local client ' .
182			'certificate.'
183			);
184			case 59: // CURLE_SSL_CIPHER
185			return $this->l10n->t(
186			'Certificate error: Couldn\'t use specified cipher.'
187			);
188			case 60: // CURLE_SSL_CACERT
189			return $this->l10n->t(
190			'Certificate error: Peer certificate cannot be ' .
191			'authenticated with known CA certificates.'
192			);
193			case 64: // CURLE_USE_SSL_FAILED
194			return $this->l10n->t(
195			'Certificate error: Requested FTP SSL level failed.'
196			);
197			case 66: // CURLE_SSL_ENGINE_INITFAILED
198			return $this->l10n->t(
199			'Certificate error: Initiating the SSL Engine failed.'
200			);
201			case 77: // CURLE_SSL_CACERT_BADFILE
202			return $this->l10n->t(
203			'Certificate error: Problem with reading the SSL CA ' .
204			'cert (path? access rights?)'
205			);
206			case 83: // CURLE_SSL_ISSUER_ERROR
207			return $this->l10n->t(
208			'Certificate error: Issuer check failed'
209			);
210			default:
211			return $this->l10n->t('Unknown SSL certificate error!');
212			}
213			}
214
215			private function decodeTwice($string) {
216			return html_entity_decode(
217			html_entity_decode(
218			$string, ENT_QUOTES \| ENT_HTML5, 'UTF-8'
219			),
220			ENT_QUOTES \| ENT_HTML5, 'UTF-8'
221			);
222			}
223
224
225			protected function determineRtl($parsedItem, $parsedFeed) {
226			$itemLang = $parsedItem->getLanguage();
227			$feedLang = $parsedFeed->getLanguage();
228
229			if ($itemLang) {
230			return Parser::isLanguageRTL($itemLang);
231			} else {
232			return Parser::isLanguageRTL($feedLang);
233			}
234			}
235
236
237			protected function buildItem($parsedItem, $parsedFeed) {
238			$item = new Item();
239			$item->setUnread();
240			$item->setUrl($parsedItem->getUrl());
241			$item->setGuid($parsedItem->getId());
242			$item->setGuidHash($item->getGuid());
243			$item->setPubDate($parsedItem->getDate()->getTimestamp());
244			$item->setLastModified($this->time->getTime());
245			$item->setRtl($this->determineRtl($parsedItem, $parsedFeed));
246
247			// unescape content because angularjs helps against XSS
248			$item->setTitle($this->decodeTwice($parsedItem->getTitle()));
249			$item->setAuthor($this->decodeTwice($parsedItem->getAuthor()));
250
251			// purification is done in the service layer
252			$body = $parsedItem->getContent();
253			$body = mb_convert_encoding($body, 'HTML-ENTITIES',
254			mb_detect_encoding($body));
255			$item->setBody($body);
256
257			$enclosureUrl = $parsedItem->getEnclosureUrl();
258			if ($enclosureUrl) {
259			$enclosureType = $parsedItem->getEnclosureType();
260			if (stripos($enclosureType, 'audio/') !== false \|\|
261			stripos($enclosureType, 'video/') !== false
262			) {
263			$item->setEnclosureMime($enclosureType);
264			$item->setEnclosureLink($enclosureUrl);
265			}
266			}
267
268			$item->generateSearchIndex();
269
270			return $item;
271			}
272
273
274			protected function buildFeed($parsedFeed, $url, $getFavicon, $modified,
275			$etag, $location) {
276			$feed = new Feed();
277
278			$link = $parsedFeed->getSiteUrl();
279
280			if (!$link) {
281			$link = $location;
282			}
283
284			// unescape content because angularjs helps against XSS
285			$title = strip_tags($this->decodeTwice($parsedFeed->getTitle()));
286			$feed->setTitle($title);
287			$feed->setUrl($url); // the url used to add the feed
288			$feed->setLocation($location); // the url where the feed was found
289			$feed->setLink($link); // <link> attribute in the feed
290			$feed->setHttpLastModified($modified);
291			$feed->setHttpEtag($etag);
292			$feed->setAdded($this->time->getTime());
293
294			if ($getFavicon) {
295			$faviconFetcher = $this->faviconFactory->build();
296			$favicon = $faviconFetcher->find($feed->getLink());
297			$feed->setFaviconLink($favicon);
298			}
299
300			return $feed;
301			}
302
303			}
304

owncloud / news

Push — master ( 150a3d...219536 )

FeedFetcher::fetch() C

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like