diff options
author | Jérémy Benoist <j0k3r@users.noreply.github.com> | 2019-05-28 14:00:22 +0200 |
---|---|---|
committer | GitHub <noreply@github.com> | 2019-05-28 14:00:22 +0200 |
commit | 48d136d3a08d7f4ca8e0d434d8104c746d31957d (patch) | |
tree | 308e15326adc658702f257693338f101d36e1af8 /src/Wallabag/CoreBundle | |
parent | 92a66835624acf6fd14f5adc5f8aab399658592e (diff) | |
parent | 6e68417f0356d0045d7a2aa3832507d362ddcfe8 (diff) | |
download | wallabag-48d136d3a08d7f4ca8e0d434d8104c746d31957d.tar.gz wallabag-48d136d3a08d7f4ca8e0d434d8104c746d31957d.tar.zst wallabag-48d136d3a08d7f4ca8e0d434d8104c746d31957d.zip |
Merge pull request #3390 from aaa2000/httplug
Use httplug and graby 2.0
Diffstat (limited to 'src/Wallabag/CoreBundle')
-rw-r--r-- | src/Wallabag/CoreBundle/Helper/ContentProxy.php | 35 | ||||
-rw-r--r-- | src/Wallabag/CoreBundle/Helper/DownloadImages.php | 23 | ||||
-rw-r--r-- | src/Wallabag/CoreBundle/Helper/HttpClientFactory.php | 51 | ||||
-rw-r--r-- | src/Wallabag/CoreBundle/Resources/config/services.yml | 11 |
4 files changed, 68 insertions, 52 deletions
diff --git a/src/Wallabag/CoreBundle/Helper/ContentProxy.php b/src/Wallabag/CoreBundle/Helper/ContentProxy.php index ca01dec8..c6fa0d98 100644 --- a/src/Wallabag/CoreBundle/Helper/ContentProxy.php +++ b/src/Wallabag/CoreBundle/Helper/ContentProxy.php | |||
@@ -54,7 +54,11 @@ class ContentProxy | |||
54 | 54 | ||
55 | if ((empty($content) || false === $this->validateContent($content)) && false === $disableContentUpdate) { | 55 | if ((empty($content) || false === $this->validateContent($content)) && false === $disableContentUpdate) { |
56 | $fetchedContent = $this->graby->fetchContent($url); | 56 | $fetchedContent = $this->graby->fetchContent($url); |
57 | $fetchedContent['title'] = $this->sanitizeContentTitle($fetchedContent['title'], $fetchedContent['content_type']); | 57 | |
58 | $fetchedContent['title'] = $this->sanitizeContentTitle( | ||
59 | $fetchedContent['title'], | ||
60 | isset($fetchedContent['headers']['content-type']) ? $fetchedContent['headers']['content-type'] : '' | ||
61 | ); | ||
58 | 62 | ||
59 | // when content is imported, we have information in $content | 63 | // when content is imported, we have information in $content |
60 | // in case fetching content goes bad, we'll keep the imported information instead of overriding them | 64 | // in case fetching content goes bad, we'll keep the imported information instead of overriding them |
@@ -188,8 +192,8 @@ class ContentProxy | |||
188 | /** | 192 | /** |
189 | * Try to sanitize the title of the fetched content from wrong character encodings and invalid UTF-8 character. | 193 | * Try to sanitize the title of the fetched content from wrong character encodings and invalid UTF-8 character. |
190 | * | 194 | * |
191 | * @param $title | 195 | * @param string $title |
192 | * @param $contentType | 196 | * @param string $contentType |
193 | * | 197 | * |
194 | * @return string | 198 | * @return string |
195 | */ | 199 | */ |
@@ -253,16 +257,14 @@ class ContentProxy | |||
253 | 257 | ||
254 | if (!empty($content['title'])) { | 258 | if (!empty($content['title'])) { |
255 | $entry->setTitle($content['title']); | 259 | $entry->setTitle($content['title']); |
256 | } elseif (!empty($content['open_graph']['og_title'])) { | ||
257 | $entry->setTitle($content['open_graph']['og_title']); | ||
258 | } | 260 | } |
259 | 261 | ||
260 | if (empty($content['html'])) { | 262 | if (empty($content['html'])) { |
261 | $content['html'] = $this->fetchingErrorMessage; | 263 | $content['html'] = $this->fetchingErrorMessage; |
262 | 264 | ||
263 | if (!empty($content['open_graph']['og_description'])) { | 265 | if (!empty($content['description'])) { |
264 | $content['html'] .= '<p><i>But we found a short description: </i></p>'; | 266 | $content['html'] .= '<p><i>But we found a short description: </i></p>'; |
265 | $content['html'] .= $content['open_graph']['og_description']; | 267 | $content['html'] .= $content['description']; |
266 | } | 268 | } |
267 | } | 269 | } |
268 | 270 | ||
@@ -277,8 +279,8 @@ class ContentProxy | |||
277 | $entry->setPublishedBy($content['authors']); | 279 | $entry->setPublishedBy($content['authors']); |
278 | } | 280 | } |
279 | 281 | ||
280 | if (!empty($content['all_headers']) && $this->storeArticleHeaders) { | 282 | if (!empty($content['headers'])) { |
281 | $entry->setHeaders($content['all_headers']); | 283 | $entry->setHeaders($content['headers']); |
282 | } | 284 | } |
283 | 285 | ||
284 | if (!empty($content['date'])) { | 286 | if (!empty($content['date'])) { |
@@ -290,28 +292,29 @@ class ContentProxy | |||
290 | } | 292 | } |
291 | 293 | ||
292 | $previewPictureUrl = ''; | 294 | $previewPictureUrl = ''; |
293 | if (!empty($content['open_graph']['og_image'])) { | 295 | if (!empty($content['image'])) { |
294 | $previewPictureUrl = $content['open_graph']['og_image']; | 296 | $previewPictureUrl = $content['image']; |
295 | } | 297 | } |
296 | 298 | ||
297 | // if content is an image, define it as a preview too | 299 | // if content is an image, define it as a preview too |
298 | if (!empty($content['content_type']) && \in_array($this->mimeGuesser->guess($content['content_type']), ['jpeg', 'jpg', 'gif', 'png'], true)) { | 300 | if (!empty($content['headers']['content-type']) && \in_array($this->mimeGuesser->guess($content['headers']['content-type']), ['jpeg', 'jpg', 'gif', 'png'], true)) { |
299 | $previewPictureUrl = $content['url']; | 301 | $previewPictureUrl = $content['url']; |
300 | } elseif (empty($previewPictureUrl)) { | 302 | } elseif (empty($previewPictureUrl)) { |
301 | $this->logger->debug('Extracting images from content to provide a default preview picture'); | 303 | $this->logger->debug('Extracting images from content to provide a default preview picture'); |
302 | $imagesUrls = DownloadImages::extractImagesUrlsFromHtml($content['html']); | 304 | $imagesUrls = DownloadImages::extractImagesUrlsFromHtml($content['html']); |
303 | $this->logger->debug(\count($imagesUrls) . ' pictures found'); | 305 | $this->logger->debug(\count($imagesUrls) . ' pictures found'); |
306 | |||
304 | if (!empty($imagesUrls)) { | 307 | if (!empty($imagesUrls)) { |
305 | $previewPictureUrl = $imagesUrls[0]; | 308 | $previewPictureUrl = $imagesUrls[0]; |
306 | } | 309 | } |
307 | } | 310 | } |
308 | 311 | ||
309 | if (!empty($previewPictureUrl)) { | 312 | if (!empty($content['headers']['content-type'])) { |
310 | $this->updatePreviewPicture($entry, $previewPictureUrl); | 313 | $entry->setMimetype($content['headers']['content-type']); |
311 | } | 314 | } |
312 | 315 | ||
313 | if (!empty($content['content_type'])) { | 316 | if (!empty($previewPictureUrl)) { |
314 | $entry->setMimetype($content['content_type']); | 317 | $this->updatePreviewPicture($entry, $previewPictureUrl); |
315 | } | 318 | } |
316 | 319 | ||
317 | try { | 320 | try { |
diff --git a/src/Wallabag/CoreBundle/Helper/DownloadImages.php b/src/Wallabag/CoreBundle/Helper/DownloadImages.php index c1645e45..7a39a2e4 100644 --- a/src/Wallabag/CoreBundle/Helper/DownloadImages.php +++ b/src/Wallabag/CoreBundle/Helper/DownloadImages.php | |||
@@ -2,8 +2,13 @@ | |||
2 | 2 | ||
3 | namespace Wallabag\CoreBundle\Helper; | 3 | namespace Wallabag\CoreBundle\Helper; |
4 | 4 | ||
5 | use GuzzleHttp\Client; | 5 | use Http\Client\Common\HttpMethodsClient; |
6 | use GuzzleHttp\Message\Response; | 6 | use Http\Client\Common\Plugin\ErrorPlugin; |
7 | use Http\Client\Common\PluginClient; | ||
8 | use Http\Client\HttpClient; | ||
9 | use Http\Discovery\MessageFactoryDiscovery; | ||
10 | use Http\Message\MessageFactory; | ||
11 | use Psr\Http\Message\ResponseInterface; | ||
7 | use Psr\Log\LoggerInterface; | 12 | use Psr\Log\LoggerInterface; |
8 | use Symfony\Component\DomCrawler\Crawler; | 13 | use Symfony\Component\DomCrawler\Crawler; |
9 | use Symfony\Component\Finder\Finder; | 14 | use Symfony\Component\Finder\Finder; |
@@ -19,9 +24,9 @@ class DownloadImages | |||
19 | private $mimeGuesser; | 24 | private $mimeGuesser; |
20 | private $wallabagUrl; | 25 | private $wallabagUrl; |
21 | 26 | ||
22 | public function __construct(Client $client, $baseFolder, $wallabagUrl, LoggerInterface $logger) | 27 | public function __construct(HttpClient $client, $baseFolder, $wallabagUrl, LoggerInterface $logger, MessageFactory $messageFactory = null) |
23 | { | 28 | { |
24 | $this->client = $client; | 29 | $this->client = new HttpMethodsClient(new PluginClient($client, [new ErrorPlugin()]), $messageFactory ?: MessageFactoryDiscovery::find()); |
25 | $this->baseFolder = $baseFolder; | 30 | $this->baseFolder = $baseFolder; |
26 | $this->wallabagUrl = rtrim($wallabagUrl, '/'); | 31 | $this->wallabagUrl = rtrim($wallabagUrl, '/'); |
27 | $this->logger = $logger; | 32 | $this->logger = $logger; |
@@ -135,7 +140,7 @@ class DownloadImages | |||
135 | $localPath = $folderPath . '/' . $hashImage . '.' . $ext; | 140 | $localPath = $folderPath . '/' . $hashImage . '.' . $ext; |
136 | 141 | ||
137 | try { | 142 | try { |
138 | $im = imagecreatefromstring($res->getBody()); | 143 | $im = imagecreatefromstring((string) $res->getBody()); |
139 | } catch (\Exception $e) { | 144 | } catch (\Exception $e) { |
140 | $im = false; | 145 | $im = false; |
141 | } | 146 | } |
@@ -306,14 +311,14 @@ class DownloadImages | |||
306 | /** | 311 | /** |
307 | * Retrieve and validate the extension from the response of the url of the image. | 312 | * Retrieve and validate the extension from the response of the url of the image. |
308 | * | 313 | * |
309 | * @param Response $res Guzzle Response | 314 | * @param ResponseInterface $res Http Response |
310 | * @param string $imagePath Path from the src image from the content (used for log only) | 315 | * @param string $imagePath Path from the src image from the content (used for log only) |
311 | * | 316 | * |
312 | * @return string|false Extension name or false if validation failed | 317 | * @return string|false Extension name or false if validation failed |
313 | */ | 318 | */ |
314 | private function getExtensionFromResponse(Response $res, $imagePath) | 319 | private function getExtensionFromResponse(ResponseInterface $res, $imagePath) |
315 | { | 320 | { |
316 | $ext = $this->mimeGuesser->guess($res->getHeader('content-type')); | 321 | $ext = $this->mimeGuesser->guess(current($res->getHeader('content-type'))); |
317 | $this->logger->debug('DownloadImages: Checking extension', ['ext' => $ext, 'header' => $res->getHeader('content-type')]); | 322 | $this->logger->debug('DownloadImages: Checking extension', ['ext' => $ext, 'header' => $res->getHeader('content-type')]); |
318 | 323 | ||
319 | // ok header doesn't have the extension, try a different way | 324 | // ok header doesn't have the extension, try a different way |
diff --git a/src/Wallabag/CoreBundle/Helper/HttpClientFactory.php b/src/Wallabag/CoreBundle/Helper/HttpClientFactory.php index 4602a684..b8e95381 100644 --- a/src/Wallabag/CoreBundle/Helper/HttpClientFactory.php +++ b/src/Wallabag/CoreBundle/Helper/HttpClientFactory.php | |||
@@ -2,16 +2,18 @@ | |||
2 | 2 | ||
3 | namespace Wallabag\CoreBundle\Helper; | 3 | namespace Wallabag\CoreBundle\Helper; |
4 | 4 | ||
5 | use Graby\Ring\Client\SafeCurlHandler; | 5 | use GuzzleHttp\Client as GuzzleClient; |
6 | use GuzzleHttp\Client; | ||
7 | use GuzzleHttp\Cookie\CookieJar; | 6 | use GuzzleHttp\Cookie\CookieJar; |
8 | use GuzzleHttp\Event\SubscriberInterface; | 7 | use GuzzleHttp\Event\SubscriberInterface; |
8 | use Http\Adapter\Guzzle5\Client as GuzzleAdapter; | ||
9 | use Http\Client\HttpClient; | ||
10 | use Http\HttplugBundle\ClientFactory\ClientFactory; | ||
9 | use Psr\Log\LoggerInterface; | 11 | use Psr\Log\LoggerInterface; |
10 | 12 | ||
11 | /** | 13 | /** |
12 | * Builds and configures the Guzzle HTTP client. | 14 | * Builds and configures the HTTP client. |
13 | */ | 15 | */ |
14 | class HttpClientFactory | 16 | class HttpClientFactory implements ClientFactory |
15 | { | 17 | { |
16 | /** @var [\GuzzleHttp\Event\SubscriberInterface] */ | 18 | /** @var [\GuzzleHttp\Event\SubscriberInterface] */ |
17 | private $subscribers = []; | 19 | private $subscribers = []; |
@@ -37,35 +39,42 @@ class HttpClientFactory | |||
37 | } | 39 | } |
38 | 40 | ||
39 | /** | 41 | /** |
40 | * @return \GuzzleHttp\Client|null | 42 | * Adds a subscriber to the HTTP client. |
43 | * | ||
44 | * @param SubscriberInterface $subscriber | ||
45 | */ | ||
46 | public function addSubscriber(SubscriberInterface $subscriber) | ||
47 | { | ||
48 | $this->subscribers[] = $subscriber; | ||
49 | } | ||
50 | |||
51 | /** | ||
52 | * Input an array of configuration to be able to create a HttpClient. | ||
53 | * | ||
54 | * @param array $config | ||
55 | * | ||
56 | * @return HttpClient | ||
41 | */ | 57 | */ |
42 | public function buildHttpClient() | 58 | public function createClient(array $config = []) |
43 | { | 59 | { |
44 | $this->logger->log('debug', 'Restricted access config enabled?', ['enabled' => (int) $this->restrictedAccess]); | 60 | $this->logger->log('debug', 'Restricted access config enabled?', ['enabled' => (int) $this->restrictedAccess]); |
45 | 61 | ||
46 | if (0 === (int) $this->restrictedAccess) { | 62 | if (0 === (int) $this->restrictedAccess) { |
47 | return; | 63 | return new GuzzleAdapter(new GuzzleClient($config)); |
48 | } | 64 | } |
49 | 65 | ||
50 | // we clear the cookie to avoid websites who use cookies for analytics | 66 | // we clear the cookie to avoid websites who use cookies for analytics |
51 | $this->cookieJar->clear(); | 67 | $this->cookieJar->clear(); |
52 | // need to set the (shared) cookie jar | 68 | if (!isset($config['defaults']['cookies'])) { |
53 | $client = new Client(['handler' => new SafeCurlHandler(), 'defaults' => ['cookies' => $this->cookieJar]]); | 69 | // need to set the (shared) cookie jar |
70 | $config['defaults']['cookies'] = $this->cookieJar; | ||
71 | } | ||
54 | 72 | ||
73 | $guzzle = new GuzzleClient($config); | ||
55 | foreach ($this->subscribers as $subscriber) { | 74 | foreach ($this->subscribers as $subscriber) { |
56 | $client->getEmitter()->attach($subscriber); | 75 | $guzzle->getEmitter()->attach($subscriber); |
57 | } | 76 | } |
58 | 77 | ||
59 | return $client; | 78 | return new GuzzleAdapter($guzzle); |
60 | } | ||
61 | |||
62 | /** | ||
63 | * Adds a subscriber to the HTTP client. | ||
64 | * | ||
65 | * @param SubscriberInterface $subscriber | ||
66 | */ | ||
67 | public function addSubscriber(SubscriberInterface $subscriber) | ||
68 | { | ||
69 | $this->subscribers[] = $subscriber; | ||
70 | } | 79 | } |
71 | } | 80 | } |
diff --git a/src/Wallabag/CoreBundle/Resources/config/services.yml b/src/Wallabag/CoreBundle/Resources/config/services.yml index 280d779d..31986951 100644 --- a/src/Wallabag/CoreBundle/Resources/config/services.yml +++ b/src/Wallabag/CoreBundle/Resources/config/services.yml | |||
@@ -42,7 +42,7 @@ services: | |||
42 | - | 42 | - |
43 | error_message: '%wallabag_core.fetching_error_message%' | 43 | error_message: '%wallabag_core.fetching_error_message%' |
44 | error_message_title: '%wallabag_core.fetching_error_message_title%' | 44 | error_message_title: '%wallabag_core.fetching_error_message_title%' |
45 | - "@wallabag_core.guzzle.http_client" | 45 | - "@wallabag_core.http_client" |
46 | - "@wallabag_core.graby.config_builder" | 46 | - "@wallabag_core.graby.config_builder" |
47 | calls: | 47 | calls: |
48 | - [ setLogger, [ "@logger" ] ] | 48 | - [ setLogger, [ "@logger" ] ] |
@@ -55,9 +55,8 @@ services: | |||
55 | - {} | 55 | - {} |
56 | - "@logger" | 56 | - "@logger" |
57 | 57 | ||
58 | wallabag_core.guzzle.http_client: | 58 | wallabag_core.http_client: |
59 | class: GuzzleHttp\ClientInterface | 59 | alias: 'httplug.client.wallabag_core' |
60 | factory: ["@wallabag_core.guzzle.http_client_factory", buildHttpClient] | ||
61 | 60 | ||
62 | wallabag_core.guzzle_authenticator.config_builder: | 61 | wallabag_core.guzzle_authenticator.config_builder: |
63 | class: Wallabag\CoreBundle\GuzzleSiteAuthenticator\GrabySiteConfigBuilder | 62 | class: Wallabag\CoreBundle\GuzzleSiteAuthenticator\GrabySiteConfigBuilder |
@@ -73,7 +72,7 @@ services: | |||
73 | bd_guzzle_site_authenticator.site_config_builder: | 72 | bd_guzzle_site_authenticator.site_config_builder: |
74 | alias: wallabag_core.guzzle_authenticator.config_builder | 73 | alias: wallabag_core.guzzle_authenticator.config_builder |
75 | 74 | ||
76 | wallabag_core.guzzle.http_client_factory: | 75 | wallabag_core.http_client_factory: |
77 | class: Wallabag\CoreBundle\Helper\HttpClientFactory | 76 | class: Wallabag\CoreBundle\Helper\HttpClientFactory |
78 | arguments: | 77 | arguments: |
79 | - "@wallabag_core.guzzle.cookie_jar" | 78 | - "@wallabag_core.guzzle.cookie_jar" |
@@ -212,7 +211,7 @@ services: | |||
212 | - "@logger" | 211 | - "@logger" |
213 | 212 | ||
214 | wallabag_core.entry.download_images.client: | 213 | wallabag_core.entry.download_images.client: |
215 | class: GuzzleHttp\Client | 214 | alias: 'httplug.client.wallabag_core.entry.download_images' |
216 | 215 | ||
217 | wallabag_core.helper.crypto_proxy: | 216 | wallabag_core.helper.crypto_proxy: |
218 | class: Wallabag\CoreBundle\Helper\CryptoProxy | 217 | class: Wallabag\CoreBundle\Helper\CryptoProxy |