Merge branch 'live'

lib/Bridge.php

@@ -71,6 +71,102 @@ abstract class BridgeAbstract implements BridgeInterface{
 
         return $this;
     }
+
+    /**
+     * Set default image SRC attribute to point on given server when none is provided (that's to say when image src starts with '/'
+     */
+    public function defaultImageSrcTo($content, $server) {
+        foreach($content->find('img') as $image) {
+            if(strpos($image->src, '/')==0) {
+                $image->src = $server.$image->src;
+            }
+        }
+    }
+}
+
+/**
+ * Extension of BridgeAbstract allowing caching of files downloaded over http files.
+ * This is specially useful for sites from Gawker or Liberation networks, which allow pages excerpts top be viewed together on index, while full pages have to be downloaded
+ * separately.
+ * This class mainly provides a get_cached method which will will download the file from its remote location.
+ * TODO allow file cache invalidation by touching files on access, and removing files/directories which have not been touched since ... a long time
+ * After all, rss-bridge is not respaw, isn't it ?
+ */
+abstract class HttpCachingBridgeAbstract extends BridgeAbstract {
+    
+    /**
+     * Maintain locally cached versions of pages to download to avoid multiple doiwnloads.
+     * A file name is generated by replacing all "/" by "_", and the file is saved below this bridge cache
+     * @param url url to cache
+     * @return content of file as string
+     */
+    public function get_cached($url) {
+        $simplified_url = str_replace(["http://", "https://", "?", "&", "="], ["", "", "/", "/", "/"], $url);
+		// TODO build this from the variable given to Cache
+		$pageCacheDir = __DIR__ . '/../cache/'."pages/";
+        $filename =  $pageCacheDir.$simplified_url;
+        if (substr($filename, -1) == '/') {
+            $filename = $filename."index.html";
+        }
+        if(file_exists($filename)) {
+//            $this->message("loading cached file from ".$filename." for page at url ".$url);
+			// TODO touch file and its parent, and try to do neighbour deletion
+            $this->refresh_in_cache($pageCacheDir, $filename);
+		} else {
+//            $this->message("we have no local copy of ".$url." Downloading to ".$filename);
+            $dir = substr($filename, 0, strrpos($filename, '/'));
+            if(!is_dir($dir)) {
+//				$this->message("creating directories for ".$dir);
+                mkdir($dir, 0777, true);
+            }
+            $this->download_remote($url, $filename);
+        }
+        return file_get_contents($filename);
+    }
+    
+    private function refresh_in_cache($pageCacheDir, $filename) {
+		$currentPath = $filename;
+		while(!$pageCacheDir==$currentPath) {
+			touch($currentPath);
+			$currentPath = dirname($currentPath);
+		}
+    }
+
+    public function download_remote($url , $save_path) {
+        $f = fopen( $save_path , 'w+');
+        if($f) {
+            $handle = fopen($url , "rb");
+            if($handle) {
+                while (!feof($handle)) {
+                    $contents = fread($handle, 8192);
+                    if($contents) {
+                        fwrite($f , $contents);
+                    }
+                }
+                fclose($handle);
+            }
+            fclose($f);
+        }
+    }
+    
+    public function remove_from_cache($url) {
+        $simplified_url = str_replace(["http://", "https://", "?", "&", "="], ["", "", "/", "/", "/"], $url);
+    	// TODO build this from the variable given to Cache
+		$pageCacheDir = __DIR__ . '/../cache/'."pages/";
+        $filename =  realpath($pageCacheDir.$simplified_url);
+        $this->message("removing from cache \"".$filename."\" WELL, NOT REALLY");
+        // filename is NO GOOD
+//        unlink($filename);
+    }
+    
+    public function message($text) {
+        $backtrace = debug_backtrace(DEBUG_BACKTRACE_IGNORE_ARGS, 3);
+        $calling = $backtrace[2];
+        $message = $calling["file"].":".$calling["line"]
+            ." class ".get_class($this)."->".$calling["function"]
+            ." - ".$text;
+        error_log($message);
+    }
 }
 
 class Bridge{