src/Tweeper.php: add a retry mechanism for cURL sessions

[tweeper.git] / src / Tweeper.php

diff --git a/src/Tweeper.php b/src/Tweeper.php
index 22d16ab..d58b529 100644 (file)
--- a/src/Tweeper.php
+++ b/src/Tweeper.php
@@ -37,6 +37,8 @@ date_default_timezone_set('UTC');
 class Tweeper {
 
   private static $userAgent = "Mozilla/5.0 (Windows NT 6.1; WOW64; rv:60.0) Gecko/20100101 Firefox/60.0";
 class Tweeper {
 
   private static $userAgent = "Mozilla/5.0 (Windows NT 6.1; WOW64; rv:60.0) Gecko/20100101 Firefox/60.0";

+  private static $maxConnectionTimeout = 5;
+  private static $maxConnectionRetries = 5;

 
   /**
    * Create a new Tweeper object controlling optional settings.
 
   /**
    * Create a new Tweeper object controlling optional settings.


@@ -86,12 +88,32 @@ class Tweeper {
   }
 
   /**
   }
 
   /**

+   * Perform a cURL session multiple times when it fails with a timeout.
+   *
+   * @param resource $ch
+   *   a cURL session handle.
+   */
+  private static function curlExec($ch) {
+    $ret = FALSE;
+    $attempt = 0;
+    do {
+      $ret = curl_exec($ch);
+      if (FALSE === $ret) {
+        trigger_error(curl_error($ch), E_USER_WARNING);
+      }
+    } while (curl_errno($ch) == CURLE_OPERATION_TIMEDOUT && ++$attempt < Tweeper::$maxConnectionRetries);
+
+    return $ret;
+  }
+
+  /**

    * Get the contents from a URL.
    */
   private static function getUrlContents($url) {
     $ch = curl_init($url);
     curl_setopt_array($ch, array(
       CURLOPT_HEADER => FALSE,
    * Get the contents from a URL.
    */
   private static function getUrlContents($url) {
     $ch = curl_init($url);
     curl_setopt_array($ch, array(
       CURLOPT_HEADER => FALSE,

+      CURLOPT_CONNECTTIMEOUT => Tweeper::$maxConnectionTimeout,

       // Follow http redirects to get the real URL.
       CURLOPT_FOLLOWLOCATION => TRUE,
       CURLOPT_RETURNTRANSFER => TRUE,
       // Follow http redirects to get the real URL.
       CURLOPT_FOLLOWLOCATION => TRUE,
       CURLOPT_RETURNTRANSFER => TRUE,


@@ -100,10 +122,7 @@ class Tweeper {
       CURLOPT_HTTPHEADER => array('Accept-language: en'),
       CURLOPT_USERAGENT => Tweeper::$userAgent,
     ));
       CURLOPT_HTTPHEADER => array('Accept-language: en'),
       CURLOPT_USERAGENT => Tweeper::$userAgent,
     ));

-    $contents = curl_exec($ch);
-    if (FALSE === $contents) {
-      trigger_error(curl_error($ch));
-    }
+    $contents = Tweeper::curlExec($ch);

     curl_close($ch);
 
     return $contents;
     curl_close($ch);
 
     return $contents;


@@ -117,6 +136,7 @@ class Tweeper {
     curl_setopt_array($ch, array(
       CURLOPT_HEADER => TRUE,
       CURLOPT_NOBODY => TRUE,
     curl_setopt_array($ch, array(
       CURLOPT_HEADER => TRUE,
       CURLOPT_NOBODY => TRUE,

+      CURLOPT_CONNECTTIMEOUT => Tweeper::$maxConnectionTimeout,

       // Follow http redirects to get the real URL.
       CURLOPT_FOLLOWLOCATION => TRUE,
       CURLOPT_RETURNTRANSFER => TRUE,
       // Follow http redirects to get the real URL.
       CURLOPT_FOLLOWLOCATION => TRUE,
       CURLOPT_RETURNTRANSFER => TRUE,


@@ -125,16 +145,15 @@ class Tweeper {
       CURLOPT_USERAGENT => Tweeper::$userAgent,
     ));
 
       CURLOPT_USERAGENT => Tweeper::$userAgent,
     ));
 

-    $ret = curl_exec($ch);
+    $ret = Tweeper::curlExec($ch);

     if (FALSE === $ret) {
     if (FALSE === $ret) {

-      trigger_error(curl_error($ch));

       curl_close($ch);
       return FALSE;
     }
 
     $url_info = curl_getinfo($ch);
     if (FALSE === $url_info) {
       curl_close($ch);
       return FALSE;
     }
 
     $url_info = curl_getinfo($ch);
     if (FALSE === $url_info) {

-      trigger_error(curl_error($ch));
+      trigger_error(curl_error($ch), E_USER_WARNING);

     }
     curl_close($ch);
 
     }
     curl_close($ch);
 


@@ -168,13 +187,13 @@ class Tweeper {
 
     $url_info = Tweeper::getUrlInfo($url);
     if (FALSE === $url_info) {
 
     $url_info = Tweeper::getUrlInfo($url);
     if (FALSE === $url_info) {

-      error_log("Failed to retrieve info for URL: " . $url);
+      trigger_error("Failed to retrieve info for URL: " . $url, E_USER_WARNING);

       return '';
     }
 
     $supported = in_array($url_info['content_type'], $supported_content_types);
     if (!$supported) {
       return '';
     }
 
     $supported = in_array($url_info['content_type'], $supported_content_types);
     if (!$supported) {

-      error_log("Unsupported enclosure content type \"" . $url_info['content_type'] . "\" for URL: " . $url_info['url']);
+      trigger_error("Unsupported enclosure content type \"" . $url_info['content_type'] . "\" for URL: " . $url_info['url'], E_USER_WARNING);

       return '';
     }
 
       return '';
     }
 


@@ -222,7 +241,7 @@ class Tweeper {
 
     $output .= " line $error->line";
 
 
     $output .= " line $error->line";
 

-    error_log($output);
+    trigger_error($output, E_USER_WARNING);

   }
 
   /**
   }
 
   /**


@@ -249,7 +268,7 @@ class Tweeper {
 
     $xml_data = $serializer->serialize($data, 'xml', $serializer_options);
     if (!$xml_data) {
 
     $xml_data = $serializer->serialize($data, 'xml', $serializer_options);
     if (!$xml_data) {

-      trigger_error("Cannot serialize data", E_USER_ERROR);
+      trigger_error("Cannot serialize data", E_USER_WARNING);

       return NULL;
     }
 
       return NULL;
     }
 


@@ -264,7 +283,7 @@ class Tweeper {
     $json_match_expr = '/window._sharedData = (.*);/';
     $ret = preg_match($json_match_expr, $html, $matches);
     if ($ret !== 1) {
     $json_match_expr = '/window._sharedData = (.*);/';
     $ret = preg_match($json_match_expr, $html, $matches);
     if ($ret !== 1) {

-      trigger_error("Cannot match expression: $json_match_expr\n", E_USER_ERROR);
+      trigger_error("Cannot match expression: $json_match_expr\n", E_USER_WARNING);

       return NULL;
     }
 
       return NULL;
     }
 


@@ -326,12 +345,13 @@ class Tweeper {
   private function loadStylesheet($host) {
     $stylesheet = "file://" . __DIR__ . "/rss_converter_" . $host . ".xsl";
     if (FALSE === file_exists($stylesheet)) {
   private function loadStylesheet($host) {
     $stylesheet = "file://" . __DIR__ . "/rss_converter_" . $host . ".xsl";
     if (FALSE === file_exists($stylesheet)) {

-      trigger_error("Conversion to RSS not supported for $host ($stylesheet not found)", E_USER_ERROR);
+      trigger_error("Conversion to RSS not supported for $host ($stylesheet not found)", E_USER_WARNING);

       return NULL;
     }
 
     $stylesheet_contents = Tweeper::getUrlContents($stylesheet);
     if (FALSE === $stylesheet_contents) {
       return NULL;
     }
 
     $stylesheet_contents = Tweeper::getUrlContents($stylesheet);
     if (FALSE === $stylesheet_contents) {

+      trigger_error("Cannot open $stylesheet", E_USER_WARNING);

       return NULL;
     }
 
       return NULL;
     }
 


@@ -353,14 +373,14 @@ class Tweeper {
   public function tweep($src_url, $host=NULL, $validate_scheme=TRUE) {
     $url = parse_url($src_url);
     if (FALSE === $url) {
   public function tweep($src_url, $host=NULL, $validate_scheme=TRUE) {
     $url = parse_url($src_url);
     if (FALSE === $url) {

-      trigger_error("Invalid URL: $src_url", E_USER_ERROR);
+      trigger_error("Invalid URL: $src_url", E_USER_WARNING);

       return NULL;
     }
 
     if (TRUE === $validate_scheme) {
       $scheme = $url["scheme"];
       if (!in_array($scheme, array("http", "https"))) {
       return NULL;
     }
 
     if (TRUE === $validate_scheme) {
       $scheme = $url["scheme"];
       if (!in_array($scheme, array("http", "https"))) {

-        trigger_error("unsupported scheme: $scheme", E_USER_ERROR);
+        trigger_error("unsupported scheme: $scheme", E_USER_WARNING);

         return NULL;
       }
     }
         return NULL;
       }
     }


@@ -368,7 +388,7 @@ class Tweeper {
     // if the host is not given derive it from the URL
     if (NULL === $host) {
       if (empty($url["host"])) {
     // if the host is not given derive it from the URL
     if (NULL === $host) {
       if (empty($url["host"])) {

-        trigger_error("Invalid host in URL: $src_url", E_USER_ERROR);
+        trigger_error("Invalid host in URL: $src_url", E_USER_WARNING);

         return NULL;
       }
       // Strip the leading www. to be more forgiving on input URLs.
         return NULL;
       }
       // Strip the leading www. to be more forgiving on input URLs.


@@ -382,6 +402,7 @@ class Tweeper {
 
     $html = Tweeper::getUrlContents($src_url);
     if (FALSE === $html) {
 
     $html = Tweeper::getUrlContents($src_url);
     if (FALSE === $html) {

+      trigger_error("Failed to retrieve $src_url", E_USER_WARNING);

       return NULL;
     }
 
       return NULL;
     }
 


@@ -397,7 +418,7 @@ class Tweeper {
 
     $output = $xsltProcessor->transformToXML($xmlDoc);
     if (FALSE === $output) {
 
     $output = $xsltProcessor->transformToXML($xmlDoc);
     if (FALSE === $output) {

-      trigger_error('XSL transformation failed.', E_USER_ERROR);
+      trigger_error('XSL transformation failed.', E_USER_WARNING);

       return NULL;
     }
 
       return NULL;
     }