Rewrote code & fix tests

[github/wallabag/wallabag.git] / src / Wallabag / CoreBundle / Helper / ContentProxy.php

diff --git a/src/Wallabag/CoreBundle/Helper/ContentProxy.php b/src/Wallabag/CoreBundle/Helper/ContentProxy.php
index 4b3e6fbb161fd176ad875b5f3b1d8857037d47df..3024fdc50455d94b1b0034f47a9d11863c528ca3 100644 (file)
--- a/src/Wallabag/CoreBundle/Helper/ContentProxy.php
+++ b/src/Wallabag/CoreBundle/Helper/ContentProxy.php
@@ -31,23 +31,28 @@ class ContentProxy
     }
 
     /**
-     * Fetch content using graby and hydrate given entry with results information.
-     * In case we couldn't find content, we'll try to use Open Graph data.
+     * Update entry using either fetched or provided content.
      *
-     * We can also force the content, in case of an import from the v1 for example, so the function won't
-     * fetch the content from the website but rather use information given with the $content parameter.
-     *
-     * @param Entry  $entry   Entry to update
-     * @param string $url     Url to grab content for
-     * @param array  $content An array with AT LEAST keys title, html, url, language & content_type to skip the fetchContent from the url
-     *
-     * @return Entry
+     * @param Entry  $entry                Entry to update
+     * @param string $url                  Url of the content
+     * @param array  $content              Array with content provided for import with AT LEAST keys title, html, url to skip the fetchContent from the url
+     * @param bool   $disableContentUpdate Whether to skip trying to fetch content using Graby
      */
-    public function updateEntry(Entry $entry, $url, array $content = [])
+    public function updateEntry(Entry $entry, $url, array $content = [], $disableContentUpdate = false)
     {
-        // do we have to fetch the content or the provided one is ok?
-        if (empty($content) || false === $this->validateContent($content)) {
-            $fetchedContent = $this->graby->fetchContent($url);
+        if (!empty($content['html'])) {
+            $content['html'] = $this->graby->cleanupHtml($content['html'], $url);
+        }
+
+        if ((empty($content) || false === $this->validateContent($content)) && false === $disableContentUpdate) {
+            try {
+                $fetchedContent = $this->graby->fetchContent($url);
+            } catch (\Exception $e) {
+                $this->logger->error('Error while trying to fetch content from URL.', [
+                    'entry_url' => $url,
+                    'error_msg' => $e->getMessage(),
+                ]);
+            }
 
             // 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
@@ -56,8 +61,24 @@ class ContentProxy
             }
         }
 
+        // be sure to keep the url in case of error
+        // so we'll be able to refetch it in the future
+        $content['url'] = !empty($content['url']) ? $content['url'] : $url;
+
+        $this->stockEntry($entry, $content);
+    }
+
+    /**
+     * Stock entry with fetched or imported content.
+     * Will fall back to OpenGraph data if available.
+     *
+     * @param Entry $entry   Entry to stock
+     * @param array $content Array with at least title and URL
+     */
+    private function stockEntry(Entry $entry, array $content)
+    {
         $title = $content['title'];
-        if (!$title && isset($content['open_graph']['og_title'])) {
+        if (!$title && !empty($content['open_graph']['og_title'])) {
             $title = $content['open_graph']['og_title'];
         }
 
@@ -65,19 +86,30 @@ class ContentProxy
         if (false === $html) {
             $html = $this->fetchingErrorMessage;
 
-            if (isset($content['open_graph']['og_description'])) {
+            if (!empty($content['open_graph']['og_description'])) {
                 $html .= '<p><i>But we found a short description: </i></p>';
                 $html .= $content['open_graph']['og_description'];
             }
         }
 
-        $entry->setUrl($content['url'] ?: $url);
+        $entry->setUrl($content['url']);
         $entry->setTitle($title);
         $entry->setContent($html);
         $entry->setHttpStatus(isset($content['status']) ? $content['status'] : '');
 
-        if (isset($content['date']) && null !== $content['date'] && '' !== $content['date']) {
-            $entry->setPublishedAt(new \DateTime($content['date']));
+        if (!empty($content['date'])) {
+            $date = $content['date'];
+
+            // is it a timestamp?
+            if (filter_var($date, FILTER_VALIDATE_INT) !== false) {
+                $date = '@'.$content['date'];
+            }
+
+            try {
+                $entry->setPublishedAt(new \DateTime($date));
+            } catch (\Exception $e) {
+                $this->logger->warning('Error while defining date', ['e' => $e, 'url' => $content['url'], 'date' => $content['date']]);
+            }
         }
 
         if (!empty($content['authors'])) {
@@ -97,12 +129,12 @@ class ContentProxy
             $entry->setDomainName($domainName);
         }
 
-        if (isset($content['open_graph']['og_image']) && $content['open_graph']['og_image']) {
+        if (!empty($content['open_graph']['og_image'])) {
             $entry->setPreviewPicture($content['open_graph']['og_image']);
         }
 
         // if content is an image define as a preview too
-        if (isset($content['content_type']) && in_array($this->mimeGuesser->guess($content['content_type']), ['jpeg', 'jpg', 'gif', 'png'], true)) {
+        if (!empty($content['content_type']) && in_array($this->mimeGuesser->guess($content['content_type']), ['jpeg', 'jpg', 'gif', 'png'], true)) {
             $entry->setPreviewPicture($content['url']);
         }
 
@@ -110,17 +142,14 @@ class ContentProxy
             $this->tagger->tag($entry);
         } catch (\Exception $e) {
             $this->logger->error('Error while trying to automatically tag an entry.', [
-                'entry_url' => $url,
+                'entry_url' => $content['url'],
                 'error_msg' => $e->getMessage(),
             ]);
         }
-
-        return $entry;
     }
 
     /**
-     * Validate that the given content as enough value to be used
-     * instead of fetch the content from the url.
+     * Validate that the given content has at least a title, an html and a url.
      *
      * @param array $content
      *
@@ -128,6 +157,6 @@ class ContentProxy
      */
     private function validateContent(array $content)
     {
-        return isset($content['title']) && isset($content['html']) && isset($content['url']) && isset($content['language']) && isset($content['content_type']);
+        return !empty($content['title']) && !empty($content['html']) && !empty($content['url']);
     }
 }