Merge pull request #2679 from jcharaoui/fix-2658

Fix content from imported entried being discarded when URL goes bad

8 files changed, 19 insertions, 16 deletions

diff --git a/src/Wallabag/CoreBundle/Helper/ContentProxy.php b/src/Wallabag/CoreBundle/Helper/ContentProxy.php
index fd059325..0130bd2b 100644
--- a/src/Wallabag/CoreBundle/Helper/ContentProxy.php
+++ b/src/Wallabag/CoreBundle/Helper/ContentProxy.php
@@ -21,14 +21,16 @@ class ContentProxy
     protected $logger;
     protected $tagRepository;
     protected $mimeGuesser;
+    protected $fetchingErrorMessage;
 
-    public function __construct(Graby $graby, RuleBasedTagger $tagger, TagRepository $tagRepository, LoggerInterface $logger)
+    public function __construct(Graby $graby, RuleBasedTagger $tagger, TagRepository $tagRepository, LoggerInterface $logger, $fetchingErrorMessage)
     {
         $this->graby = $graby;
         $this->tagger = $tagger;
         $this->logger = $logger;
         $this->tagRepository = $tagRepository;
         $this->mimeGuesser = new MimeTypeExtensionGuesser();
+        $this->fetchingErrorMessage = $fetchingErrorMessage;
     }
 
     /**
@@ -48,7 +50,13 @@ class ContentProxy
     {
         // do we have to fetch the content or the provided one is ok?
         if (empty($content) || false === $this->validateContent($content)) {
-            $content = $this->graby->fetchContent($url);
+            $fetchedContent = $this->graby->fetchContent($url);
+
+            // when content is imported, we have information in $content
+            // in case fetching content goes bad, we'll keep the imported information instead of overriding them
+            if (empty($content) || $fetchedContent['html'] !== $this->fetchingErrorMessage) {
+                $content = $fetchedContent;
+            }
         }
 
         $title = $content['title'];
@@ -58,7 +66,7 @@ class ContentProxy
 
         $html = $content['html'];
         if (false === $html) {
-            $html = '<p>Unable to retrieve readable content.</p>';
+            $html = $this->fetchingErrorMessage;
 
             if (isset($content['open_graph']['og_description'])) {
                 $html .= '<p><i>But we found a short description: </i></p>';
@@ -71,8 +79,8 @@ class ContentProxy
         $entry->setContent($html);
         $entry->setHttpStatus(isset($content['status']) ? $content['status'] : '');
 
-        $entry->setLanguage($content['language']);
-        $entry->setMimetype($content['content_type']);
+        $entry->setLanguage(isset($content['language']) ? $content['language'] : '');
+        $entry->setMimetype(isset($content['content_type']) ? $content['content_type'] : '');
         $entry->setReadingTime(Utils::getReadingTime($html));
 
         $domainName = parse_url($entry->getUrl(), PHP_URL_HOST);
@@ -85,7 +93,7 @@ class ContentProxy
         }
 
         // if content is an image define as a preview too
-        if (in_array($this->mimeGuesser->guess($content['content_type']), ['jpeg', 'jpg', 'gif', 'png'], true)) {
+        if (isset($content['content_type']) && in_array($this->mimeGuesser->guess($content['content_type']), ['jpeg', 'jpg', 'gif', 'png'], true)) {
             $entry->setPreviewPicture($content['url']);
         }
 
diff --git a/src/Wallabag/CoreBundle/Resources/config/services.yml b/src/Wallabag/CoreBundle/Resources/config/services.yml
index bcf0c9ca..fadd5e49 100644
--- a/src/Wallabag/CoreBundle/Resources/config/services.yml
+++ b/src/Wallabag/CoreBundle/Resources/config/services.yml
@@ -86,6 +86,7 @@ services:
             - "@wallabag_core.rule_based_tagger"
             - "@wallabag_core.tag_repository"
             - "@logger"
+            - '%wallabag_core.fetching_error_message%'
 
     wallabag_core.rule_based_tagger:
         class: Wallabag\CoreBundle\Helper\RuleBasedTagger
diff --git a/src/Wallabag/ImportBundle/Import/ChromeImport.php b/src/Wallabag/ImportBundle/Import/ChromeImport.php
index d7620bcb..1a324934 100644
--- a/src/Wallabag/ImportBundle/Import/ChromeImport.php
+++ b/src/Wallabag/ImportBundle/Import/ChromeImport.php
@@ -37,7 +37,7 @@ class ChromeImport extends BrowserImport
     {
         $data = [
             'title' => $entry['name'],
-            'html' => '',
+            'html' => false,
             'url' => $entry['url'],
             'is_archived' => $this->markAsRead,
             'tags' => '',
diff --git a/src/Wallabag/ImportBundle/Import/FirefoxImport.php b/src/Wallabag/ImportBundle/Import/FirefoxImport.php
index e010f5a4..d3f99770 100644
--- a/src/Wallabag/ImportBundle/Import/FirefoxImport.php
+++ b/src/Wallabag/ImportBundle/Import/FirefoxImport.php
@@ -37,7 +37,7 @@ class FirefoxImport extends BrowserImport
     {
         $data = [
             'title' => $entry['title'],
-            'html' => '',
+            'html' => false,
             'url' => $entry['uri'],
             'is_archived' => $this->markAsRead,
             'tags' => '',
diff --git a/src/Wallabag/ImportBundle/Import/InstapaperImport.php b/src/Wallabag/ImportBundle/Import/InstapaperImport.php
index cf4c785c..70a53f1a 100644
--- a/src/Wallabag/ImportBundle/Import/InstapaperImport.php
+++ b/src/Wallabag/ImportBundle/Import/InstapaperImport.php
@@ -74,8 +74,7 @@ class InstapaperImport extends AbstractImport
                 'status' => $data[3],
                 'is_archived' => $data[3] === 'Archive' || $data[3] === 'Starred',
                 'is_starred' => $data[3] === 'Starred',
-                'content_type' => '',
-                'language' => '',
+                'html' => false,
             ];
         }
         fclose($handle);
diff --git a/src/Wallabag/ImportBundle/Import/PinboardImport.php b/src/Wallabag/ImportBundle/Import/PinboardImport.php
index 9bcfbc36..d9865534 100644
--- a/src/Wallabag/ImportBundle/Import/PinboardImport.php
+++ b/src/Wallabag/ImportBundle/Import/PinboardImport.php
@@ -98,8 +98,6 @@ class PinboardImport extends AbstractImport
         $data = [
             'title' => $importedEntry['description'],
             'url' => $importedEntry['href'],
-            'content_type' => '',
-            'language' => '',
             'is_archived' => ('no' === $importedEntry['toread']) || $this->markAsRead,
             'is_starred' => false,
             'created_at' => $importedEntry['time'],
diff --git a/src/Wallabag/ImportBundle/Import/ReadabilityImport.php b/src/Wallabag/ImportBundle/Import/ReadabilityImport.php
index b8c0f777..de320d23 100644
--- a/src/Wallabag/ImportBundle/Import/ReadabilityImport.php
+++ b/src/Wallabag/ImportBundle/Import/ReadabilityImport.php
@@ -98,11 +98,10 @@ class ReadabilityImport extends AbstractImport
         $data = [
             'title' => $importedEntry['article__title'],
             'url' => $importedEntry['article__url'],
-            'content_type' => '',
-            'language' => '',
             'is_archived' => $importedEntry['archive'] || $this->markAsRead,
             'is_starred' => $importedEntry['favorite'],
             'created_at' => $importedEntry['date_added'],
+            'html' => false,
         ];
 
         $entry = new Entry($this->user);
diff --git a/src/Wallabag/ImportBundle/Import/WallabagV1Import.php b/src/Wallabag/ImportBundle/Import/WallabagV1Import.php
index 4f001062..59e3ce02 100644
--- a/src/Wallabag/ImportBundle/Import/WallabagV1Import.php
+++ b/src/Wallabag/ImportBundle/Import/WallabagV1Import.php
@@ -37,8 +37,6 @@ class WallabagV1Import extends WallabagImport
             'title' => $entry['title'],
             'html' => $entry['content'],
             'url' => $entry['url'],
-            'content_type' => '',
-            'language' => '',
             'is_archived' => $entry['is_read'] || $this->markAsRead,
             'is_starred' => $entry['is_fav'],
             'tags' => '',