X-Git-Url: https://git.ao2.it/tweeper.git/blobdiff_plain/547175cc33bb2aa612c706eac783d8f506fe9e7e..3db7ab4d8e5244f1c5c00ac966b2d615511f235b:/tweeper.php?ds=sidebyside

diff --git a/tweeper.php b/tweeper.php
index d084398..fc3a3ef 100644
--- a/tweeper.php
+++ b/tweeper.php
@@ -1,6 +1,7 @@
 <?php
-/*
- * tweeper - a Twitter to RSS web scraper
+/**
+ * @file
+ * Tweeper - a Twitter to RSS web scraper.
  *
  * Copyright (C) 2013-2015  Antonio Ospite <ao2@ao2.it>
  *
@@ -18,38 +19,74 @@
  * along with this program.  If not, see <http://www.gnu.org/licenses/>.
  */
 
+require_once 'Symfony/Component/Serializer/autoload.php';
+
+use Symfony\Component\Serializer\Serializer;
+use Symfony\Component\Serializer\Encoder\XmlEncoder;
+use Symfony\Component\Serializer\Normalizer\ObjectNormalizer;
+
 date_default_timezone_set('UTC');
 
+/**
+ * Scrape supported websites and perform conversion to RSS.
+ */
 class Tweeper {
 
-  private static $USER_AGENT = "Mozilla/5.0 (Windows NT 6.1; rv:22.0) Gecko/20130405 Firefox/22.0";
+  private static $userAgent = "Mozilla/5.0 (Windows NT 6.1; rv:22.0) Gecko/20130405 Firefox/22.0";
 
+  /**
+   * Constructor sets up {@link $generate_enclosure}.
+   */
   public function __construct($generate_enclosure = FALSE) {
     $this->generate_enclosure = $generate_enclosure;
   }
 
-  public static function epoch_to_gmdate($timestamp)
-  {
-    return gmdate('D, d M Y H:i:s', $timestamp) . ' GMT';
+  /**
+   * Convert numeric Epoch to the date format expected in a RSS document.
+   */
+  public static function epochToRssDate($timestamp) {
+    if (!is_numeric($timestamp) || is_nan($timestamp)) {
+      $timestamp = 0;
+    }
+
+    return gmdate(DATE_RSS, $timestamp);
   }
 
-  public static function str_to_gmdate($date)
-  {
+  /**
+   * Convert generic date string to the date format expected in a RSS document.
+   */
+  public static function strToRssDate($date) {
     $timestamp = strtotime($date);
-    return Tweeper::epoch_to_gmdate($timestamp);
+    if (FALSE === $timestamp) {
+      $timestamp = 0;
+    }
+
+    return Tweeper::epochToRssDate($timestamp);
   }
 
-  private static function get_contents($url)
-  {
+  /**
+   * Convert string to UpperCamelCase.
+   */
+  public static function toUpperCamelCase($str, $delim = ' ') {
+    $str_upper = ucwords($str, $delim);
+    $str_camel_case = str_replace($delim, '', $str_upper);
+    return $str_camel_case;
+  }
+
+  /**
+   * Get the contents from a URL.
+   */
+  private static function getUrlContents($url) {
     $ch = curl_init($url);
     curl_setopt_array($ch, array(
       CURLOPT_HEADER => FALSE,
-      CURLOPT_FOLLOWLOCATION => TRUE,  // follow http redirects to get the real URL
+      // Follow http redirects to get the real URL.
+      CURLOPT_FOLLOWLOCATION => TRUE,
       CURLOPT_RETURNTRANSFER => TRUE,
       CURLOPT_SSL_VERIFYHOST => FALSE,
       CURLOPT_SSL_VERIFYPEER => FALSE,
       CURLOPT_HTTPHEADER => array('Accept-language: en'),
-      CURLOPT_USERAGENT => Tweeper::$USER_AGENT,
+      CURLOPT_USERAGENT => Tweeper::$userAgent,
     ));
     $contents = curl_exec($ch);
     curl_close($ch);
@@ -57,17 +94,20 @@ class Tweeper {
     return $contents;
   }
 
-  private static function get_info($url)
-  {
+  /**
+   * Get the headers from a URL.
+   */
+  private static function getUrlInfo($url) {
     $ch = curl_init($url);
     curl_setopt_array($ch, array(
       CURLOPT_HEADER => TRUE,
       CURLOPT_NOBODY => TRUE,
-      CURLOPT_FOLLOWLOCATION => TRUE,  // follow http redirects to get the real URL
+      // Follow http redirects to get the real URL.
+      CURLOPT_FOLLOWLOCATION => TRUE,
       CURLOPT_RETURNTRANSFER => TRUE,
       CURLOPT_SSL_VERIFYHOST => FALSE,
       CURLOPT_SSL_VERIFYPEER => FALSE,
-      CURLOPT_USERAGENT => Tweeper::$USER_AGENT,
+      CURLOPT_USERAGENT => Tweeper::$userAgent,
     ));
     curl_exec($ch);
     $url_info = curl_getinfo($ch);
@@ -76,10 +116,14 @@ class Tweeper {
     return $url_info;
   }
 
-  public static function generate_enclosure($url)
-  {
+  /**
+   * Generate an RSS <enclosure/> element.
+   */
+  public static function generateEnclosure($url) {
     $supported_content_types = array(
+      "application/octet-stream",
       "application/ogg",
+      "application/pdf",
       "audio/aac",
       "audio/mp4",
       "audio/mpeg",
@@ -90,17 +134,14 @@ class Tweeper {
       "audio/x-midi",
       "image/gif",
       "image/jpeg",
+      "image/png",
       "video/avi",
       "video/mp4",
       "video/mpeg",
       "video/ogg",
     );
 
-    // The RSS specification says that the enclosure element url must be http.
-    // See http://sourceforge.net/p/feedvalidator/bugs/72/
-    $http_url = preg_replace("/^https/", "http", $url);
-
-    $url_info = Tweeper::get_info($http_url);
+    $url_info = Tweeper::getUrlInfo($url);
 
     $supported = in_array($url_info['content_type'], $supported_content_types);
     if (!$supported) {
@@ -108,74 +149,129 @@ class Tweeper {
       return '';
     }
 
-    $dom = new DomDocument();
+    // The RSS specification says that the enclosure element URL must be http.
+    // See http://sourceforge.net/p/feedvalidator/bugs/72/
+    $http_url = preg_replace("/^https/", "http", $url_info['url']);
+
+    $dom = new DOMDocument();
     $enc = $dom->createElement('enclosure');
-    $enc->setAttribute('url', $url_info['url']);
+    $enc->setAttribute('url', $http_url);
     $enc->setAttribute('length', $url_info['download_content_length']);
     $enc->setAttribute('type', $url_info['content_type']);
 
-    $dom->appendChild($enc);
-
-    return $dom->saveXML($enc);
+    return $enc;
   }
 
-  /* Mimic the message from libxml.c::php_libxml_ctx_error_level() */
-  private function log_xml_error($error) {
+  /**
+   * Mimic the message from libxml.c::php_libxml_ctx_error_level()
+   */
+  private static function logXmlError($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;
+      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 {
+    }
+    else {
       $output .= " in Entity,";
     }
 
-    $output .=" line $error->line";
+    $output .= " line $error->line";
 
     error_log($output);
   }
 
-  private function load_stylesheet($host) {
-    $stylesheet = "file://" . __DIR__ . "/rss_converter_" . $host . ".xsl";
-    if (FALSE === file_exists($stylesheet)) {
-      trigger_error("Conversion to RSS not supported: $host", E_USER_ERROR);
+  /**
+   * Convert json to XML.
+   */
+  private static function jsonToXml($json, $root_node_name) {
+    // Apparently the ObjectNormalizer used afterwards is not able to handle
+    // the stdClass object created by json_decode() with the default setting
+    // $assoc = false; so use $assoc = true.
+    $data = json_decode($json, $assoc = TRUE);
+    if (!$data) {
       return NULL;
     }
 
-    $stylesheet_contents = $this->get_contents($stylesheet);
+    $encoder = new XmlEncoder();
+    $normalizer = new ObjectNormalizer();
+    $serializer = new Serializer(array($normalizer), array($encoder));
 
-    $xslDoc = new DOMDocument();
-    $xslDoc->loadXML($stylesheet_contents);
+    $serializer_options = array(
+      'xml_encoding' => "UTF-8",
+      'xml_format_output' => TRUE,
+      'xml_root_node_name' => $root_node_name,
+    );
 
-    $xsltProcessor = new XSLTProcessor();
-    $xsltProcessor->registerPHPFunctions();
-    $xsltProcessor->setParameter('', 'generateEnclosure', $this->generate_enclosure);
-    $xsltProcessor->importStylesheet($xslDoc);
+    $xml_data = $serializer->serialize($data, 'xml', $serializer_options);
+    if (!$xml_data) {
+      trigger_error("Cannot serialize data", E_USER_ERROR);
+      return NULL;
+    }
 
-    return $xsltProcessor;
+    return $xml_data;
   }
 
-  private function html_to_xml($html) {
+  /**
+   * Convert the Instagram content to XML.
+   */
+  private function getXmlInstagramCom($html) {
+    // Extract the json data from the html code.
+    $json_match_expr = '/window._sharedData = (.*);/';
+    $ret = preg_match($json_match_expr, $html, $matches);
+    if ($ret !== 1) {
+      trigger_error("Cannot match expression: $json_match_expr\n", E_USER_ERROR);
+      return NULL;
+    }
+
+    return Tweeper::jsonToXml($matches[1], 'instagram');
+  }
+
+  /**
+   * Make the Facebook HTML processable.
+   */
+  private function preprocessHtmlFacebookCom($html) {
+    $html = str_replace('<!--', '', $html);
+    $html = str_replace('-->', '', $html);
+    return $html;
+  }
+
+  /**
+   * Convert the HTML retrieved from the site to XML.
+   */
+  private function htmlToXml($html, $host) {
     $xmlDoc = new DOMDocument();
 
     // Handle warnings and errors when loading invalid HTML.
-    $xml_errors_value = libxml_use_internal_errors(true);
-    $xmlDoc->loadHTML($html);
+    $xml_errors_value = libxml_use_internal_errors(TRUE);
+
+    // If there is a host-specific method to get the XML data, use it!
+    $get_xml_host_method = 'getXml' . Tweeper::toUpperCamelCase($host, '.');
+    if (method_exists($this, $get_xml_host_method)) {
+      $xml_data = call_user_func_array(array($this, $get_xml_host_method), array($html));
+      $xmlDoc->loadXML($xml_data);
+    }
+    else {
+      $xmlDoc->loadHTML($html);
+    }
+
     foreach (libxml_get_errors() as $xml_error) {
-      $this->log_xml_error($xml_error);
+      Tweeper::logXmlError($xml_error);
     }
     libxml_clear_errors();
     libxml_use_internal_errors($xml_errors_value);
@@ -183,24 +279,58 @@ class Tweeper {
     return $xmlDoc;
   }
 
+  /**
+   * Load a stylesheet if the web site is supported.
+   */
+  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);
+      return NULL;
+    }
+
+    $stylesheet_contents = Tweeper::getUrlContents($stylesheet);
+
+    $xslDoc = new DOMDocument();
+    $xslDoc->loadXML($stylesheet_contents);
+
+    $xsltProcessor = new XSLTProcessor();
+    $xsltProcessor->registerPHPFunctions();
+    $xsltProcessor->setParameter('', 'generate-enclosure', $this->generate_enclosure);
+    $xsltProcessor->importStylesheet($xslDoc);
+
+    return $xsltProcessor;
+  }
+
+  /**
+   * Convert the site content to RSS.
+   */
   public function tweep($src_url) {
     $url = parse_url($src_url);
     if (FALSE === $url || empty($url["host"])) {
-      trigger_error("Invalid url: $src_url", E_USER_ERROR);
+      trigger_error("Invalid URL: $src_url", E_USER_ERROR);
       return NULL;
     }
 
-    $xsltProcessor = $this->load_stylesheet($url["host"]);
+    // Strip the leading www. to be more forgiving on input URLs.
+    $host = preg_replace('/^www\./', '', $url["host"]);
+
+    $xsltProcessor = $this->loadStylesheet($host);
     if (NULL === $xsltProcessor) {
       return NULL;
     }
 
-    $html = $this->get_contents($src_url);
+    $html = Tweeper::getUrlContents($src_url);
     if (FALSE === $html) {
       return NULL;
     }
 
-    $xmlDoc = $this->html_to_xml($html);
+    $preprocess_html_host_method = 'preprocessHtml' . Tweeper::toUpperCamelCase($host, '.');
+    if (method_exists($this, $preprocess_html_host_method)) {
+      $html = call_user_func_array(array($this, $preprocess_html_host_method), array($html));
+    }
+
+    $xmlDoc = $this->htmlToXml($html, $host);
     if (NULL === $xmlDoc) {
       return NULL;
     }
@@ -213,40 +343,57 @@ class Tweeper {
     }
     return $output;
   }
+
 }
 
-function usage($argv)
-{
-  if (php_sapi_name() != 'cli')
-    $usage = htmlentities("{$_SERVER['SCRIPT_NAME']}?src_url=<src_url>&generate_enclosure=<0|1>");
-  else
+/**
+ * Check if the script is being run from the command line.
+ */
+function is_cli() {
+  return (php_sapi_name() === "cli");
+}
+
+/**
+ * Show the script usage.
+ */
+function usage($argv) {
+  if (is_cli()) {
     $usage = "{$argv[0]} [-e|-h|--help] <src_url>\n";
+  }
+  else {
+    $usage = htmlentities("{$_SERVER['SCRIPT_NAME']}?src_url=<src_url>&generate_enclosure=<0|1>");
+  }
 
   return "usage: $usage";
 }
 
-function parse_options_cli($argv, $argc)
-{
+/**
+ * Parse command line options.
+ */
+function parse_options_cli($argv, $argc) {
   $options = array(
-    'generate_enclosure' => FALSE
+    'generate_enclosure' => FALSE,
   );
 
-  if ($argc < 2)
+  if ($argc < 2) {
     return $options;
+  }
 
   $cli_options = getopt("eh", array("help"));
   foreach ($cli_options as $opt => $val) {
     switch ($opt) {
-    case 'e':
-      $options['generate_enclosure'] = TRUE;
-      break;
-    case 'h':
-    case 'help':
-      echo usage($argv);
-      exit(0);
-    default:
-      fwrite(STDERR, usage($argv));
-      exit(1);
+      case 'e':
+        $options['generate_enclosure'] = TRUE;
+        break;
+
+      case 'h':
+      case 'help':
+        echo usage($argv);
+        exit(0);
+
+      default:
+        fwrite(STDERR, usage($argv));
+        exit(1);
     }
   }
 
@@ -255,32 +402,36 @@ function parse_options_cli($argv, $argc)
   return $options;
 }
 
-function parse_options_query_string()
-{
+/**
+ * Parse options passed from a query string.
+ */
+function parse_options_query_string() {
   $options = array(
-    'generate_enclosure' => FALSE
+    'generate_enclosure' => FALSE,
   );
 
-  if (isset($_GET['src_url']))
+  if (isset($_GET['src_url'])) {
     $options['src_url'] = $_GET['src_url'];
+  }
 
-  if (isset($_GET['generate_enclosure']))
+  if (isset($_GET['generate_enclosure'])) {
     $options['generate_enclosure'] = $_GET['generate_enclosure'] == 1;
+  }
 
   return $options;
 }
 
-
-if (php_sapi_name() != 'cli') {
-  $options = parse_options_query_string();
-  $ERROR_STREAM = fopen('php://output', 'w');
-} else {
+if (is_cli()) {
   $options = parse_options_cli($argv, $argc);
-  $ERROR_STREAM = fopen('php://stderr', 'w');
+  $error_stream = fopen('php://stderr', 'w');
+}
+else {
+  $options = parse_options_query_string();
+  $error_stream = fopen('php://output', 'w');
 }
 
 if (!isset($options['src_url'])) {
-  fwrite($ERROR_STREAM, usage($argv));
+  fwrite($error_stream, usage(is_cli() ? $argv : NULL));
   exit(1);
 }