rss_converter_dilbert.com.xsl: make the feed validate with feedvalidator.org

[tweeper.git] / tweeper.php

diff --git a/tweeper.php b/tweeper.php
index c2c5a04..4be5c7d 100644 (file)
--- a/tweeper.php
+++ b/tweeper.php
@@ -2,7 +2,7 @@
 /*
  * tweeper - a Twitter to RSS web scraper
  * 
- * Copyright (C) 2013  Antonio Ospite <ospite@studenti.unina.it>
+ * Copyright (C) 2013-2014  Antonio Ospite <ao2@ao2.it>
  * 
  * This program is free software: you can redistribute it and/or modify
  * it under the terms of the GNU General Public License as published by
@@ -52,6 +52,7 @@ class Tweeper {
     $ch = curl_init($url);
     curl_setopt_array($ch, array(
       CURLOPT_HEADER => FALSE,
+      CURLOPT_FOLLOWLOCATION => TRUE,  // follow http redirects to get the real URL
       CURLOPT_RETURNTRANSFER => TRUE,
       CURLOPT_SSL_VERIFYHOST => FALSE,
       CURLOPT_SSL_VERIFYPEER => FALSE,
@@ -90,6 +91,7 @@ class Tweeper {
       "audio/aac",
       "audio/mp4",
       "audio/mpeg",
+      "audio/ogg",
       "audio/vorbis",
       "audio/wav",
       "audio/webm",
@@ -121,11 +123,48 @@ class Tweeper {
     return $dom->saveXML($enc);
   }
 
+  /* Mimic the message from libxml.c::php_libxml_ctx_error_level() */
+  private function log_xml_error($error) {
+    $output = "";
+
+    switch ($error->level) {
+    case LIBXML_ERR_WARNING:
+      $output .= "Warning $error->code: ";
+      break;
+    case LIBXML_ERR_ERROR:
+      $output .= "Error $error->code: ";
+      break;
+    case LIBXML_ERR_FATAL:
+      $output .= "Fatal Error $error->code: ";
+      break;
+    }
+
+    $output .= trim($error->message);
+
+    if ($error->file) {
+      $output .= " in $error->file";
+    } else {
+      $output .= " in Entity,";
+    }
+
+    $output .=" line $error->line";
+
+    error_log($output);
+  }
+
   public function tweep($uri) {
     $html = Tweeper::get_contents($uri);
 
     $xmlDoc = new DOMDocument();
+
+    // Handle warnings and errors when loading invalid HTML.
+    $xml_errors_value = libxml_use_internal_errors(true);
     $xmlDoc->loadHTML($html);
+    foreach (libxml_get_errors() as $xml_error) {
+      $this->log_xml_error($xml_error);
+    }
+    libxml_clear_errors();
+    libxml_use_internal_errors($xml_errors_value);
 
     $output = $this->xsltProcessor->transformToXML($xmlDoc);
 
@@ -167,7 +206,8 @@ function parse_options_cli($argv, $argc)
       echo usage($argv);
       exit(0);
     default:
-      die(usage($argv));
+      fwrite(STDERR, usage($argv));
+      exit(1);
     }
   }
 
@@ -192,21 +232,30 @@ function parse_options_query_string()
 }
 
 
-if (php_sapi_name() != 'cli')
+if (php_sapi_name() != 'cli') {
   $options = parse_options_query_string();
-else
+  $ERROR_STREAM = fopen('php://output', 'w');
+} else {
   $options = parse_options_cli($argv, $argc);
+  $ERROR_STREAM = fopen('php://stderr', 'w');
+}
 
-if (!isset($options['src_url']))
-  die(usage($argv));
+if (!isset($options['src_url'])) {
+  fwrite($ERROR_STREAM, usage($argv));
+  exit(1);
+}
 
 $url = parse_url($options['src_url']);
-if (FALSE === $url || empty($url["host"]))
-  die("Invalid url: ${options['src_url']}\n");
+if (FALSE === $url || empty($url["host"])) {
+  fwrite($ERROR_STREAM, "Invalid url: ${options['src_url']}\n");
+  exit(1);
+}
 
 $stylesheet = "file://" . __DIR__ . "/rss_converter_" . $url["host"] . ".xsl";
-if (FALSE === file_exists($stylesheet))
-  die("Conversion to RSS not supported: {$url["host"]}\n");
+if (FALSE === file_exists($stylesheet)) {
+  fwrite($ERROR_STREAM, "Conversion to RSS not supported: {$url["host"]}\n");
+  exit(1);
+}
 
 $tweeper = new Tweeper($stylesheet, $options['generate_enclosure']);
 echo $tweeper->tweep($options['src_url']);