Index: contrib/emaudio/providers/archive.inc
===================================================================
RCS file: contrib/emaudio/providers/archive.inc
diff -N contrib/emaudio/providers/archive.inc
--- /dev/null	1 Jan 1970 00:00:00 -0000
+++ contrib/emaudio/providers/archive.inc	1 Jan 1970 00:00:00 -0000
@@ -0,0 +1,366 @@
+<?php
+// $Id$
+
+/**
+ * @file
+ *  This is an archive.org provider include file for Embedded Media Audio.
+ */
+
+/**
+ *  This is the main URL for your provider.
+ */
+define('EMAUDIO_ARCHIVE_MAIN_URL', 'http://www.archive.org/');
+
+/**
+ *  This is the URL to the API of your provider, if this exists.
+ */
+define('EMAUDIO_ARCHIVE_API_URL', 'http://www.google.com/search?q=archive.org+api');
+
+/**
+ *  This defines the version of the content data array that we serialize
+ *  in emvideo_archive_data(). If we change the expected keys of that array,
+ *  we must increment this value, which will allow older content to be updated
+ *  to the new version automatically.
+ */
+define('EMAUDIO_ARCHIVE_DATA_VERSION', 1);
+
+/**
+ * hook emaudio_PROVIDER_info
+ * This returns information relevant to a specific 3rd party audio provider.
+ *
+ * @return
+ *   A keyed array of strings requested by various admin and other forms.
+ *    'provider' => The machine name of the provider. This must be the same as
+ *      the base name of this filename, before the .inc extension.
+ *    'name' => The translated name of the provider.
+ *    'url' => The url to the main page for the provider.
+ *    'settings_description' => A description of the provider that will be
+ *      posted in the admin settings form.
+ *    'supported_features' => An array of rows describing the state of certain
+ *      supported features by the provider. These will be rendered in a table,
+ *      with the columns being 'Feature', 'Supported', 'Notes'. In general,
+ *      the 'Feature' column will give the name of the feature, 'Supported'
+ *      will be Yes or No, and 'Notes' will give an optional description or
+ *      caveats to the feature.
+ */
+function emaudio_archive_info() {
+  $features = array(
+    array(t('Autoplay'), t('Yes'), ''),
+    array(t('RSS Attachment'), t('Yes'), ''),
+    array(t('Thumbnails'), t('Yes'), t('')),
+  );
+  return array(
+    'provider' => 'archive',
+    'name' => t('archive.org'),
+    'url' => EMAUDIO_ARCHIVE_MAIN_URL,
+    'settings_description' => t('These settings specifically affect audio streamed from !archive. You can also read more about its !api.', array('!archive' => l(t('Archive.org'), EMaudio_ARCHIVE_MAIN_URL), '!api' => l(t("developer's API"), EMaudio_ARCHIVE_API_URL))),
+    'supported_features' => $features,
+  );
+}
+
+/**
+ *  hook emaudio_PROVIDER_settings
+ *  This should return a subform to be added to the emaudio_settings() admin
+ *  settings page.
+ *
+ *  Note that a form field set will already be provided at $form['archive'],
+ *  so if you want specific provider settings within that field set, you should
+ *  add the elements to that form array element.
+ */
+function emaudio_archive_settings() {
+  $form = array();
+
+  return $form; 
+}
+
+/**
+ *  hook emaudio_PROVIDER_extract
+ *
+ *  This is called to extract the audio code from a pasted URL or embed code.
+ *
+ *  We'll be passed a URL or the embed code from a audio when an editor pastes
+ *  that in the field's textfield. We'll need to either pass back an array of
+ *  regex expressions to match, or do the matching ourselves and return the
+ *  resulting audio code.
+ *
+ *  @param $parse
+ *    An optional string with the pasted URL or embed code.
+ *  @return
+ *    Either an array of regex expressions to be tested, or a string with the
+ *    audio code to be used. If the hook tests the code itself, it should
+ *    return either the string of the audio code (if matched), or an empty
+ *    array. Otherwise, the calling function will handle testing the embed code
+ *    against each regex string in the returned array.
+ */
+function emaudio_archive_extract($parse = '') {
+  // Here we assume that a URL will be passed in the form of
+  // http://www.archive.org/audio/text-audio-title
+  // or embed code in the form of <object value="http://www.archive.org/embed...".
+
+  // We'll simply return an array of regular expressions for Embedded Media
+  // Field to handle for us.
+  return array(
+    // In this expression, we're looking first for text matching the expression
+    // between the @ signs. The 'i' at the end means we don't care about the
+    // case. Thus, if someone enters http://www.Archive.com, it will still
+    // match. We escape periods as \., as otherwise they match any character.
+    // The text in parentheses () will be returned as the provider audio code,
+    // if there's a match for the entire expression. In this particular case,
+    // ([^?]+) means to match one more more characters (+) that are not a
+    // question mark ([^\?]), which would denote a query in the URL.
+    '@archive\.org\/details\/([^\"\&]+)@i',
+
+    // Now we test for embedded audio code, which is similar in this case to
+    // the above expression, except that we can safely assume we won't have a
+    // query in the URL, and that the URL will be surrounded by quotation marks,
+    // and have /embed/ rather than /audio/ in the URL. Note that regular
+    // expressions will be tested for matches in the order provided, so you
+    // may need to move this value above the other in some cases. Obviously,
+    // in the case of this archive provider, you could easily improve the
+    // regular expression to match against either a URL or embed code with
+    // one expression, such as '@archive\.com/[watch|embed]/([^"\?]+)@i'.
+    // However, many other providers have more complex requirements, so
+    // we split them up for this demonstration.
+    '@archive\.org/download/([^/]+)=@i',
+  );
+}
+
+/**
+ *  hook emaudio_PROVIDER_data
+ *
+ *  Provides an array to be serialised and made available with $item elsewhere.
+ *
+ *  This data can be used to store any extraneous information available
+ *  specifically to the archive provider.
+ */
+function emaudio_archive_data($field, $item, $error_field = '') {
+
+  // Initialize the data array.
+  $data = array();
+
+  // Create some version control. Thus if we make changes to the data array
+  // down the road, we can respect older content. If allowed by Embedded Media
+  // Field, any older content will automatically update this array as needed.
+  // In any case, you should account for the version if you increment it.
+  $data['emaudio_archive_version'] = EMaudio_ARCHIVE_DATA_VERSION;
+
+  // Add the thumbnail data. 
+  $data['thumbnail'] = 'http://www.archive.org/download/'. $item['value'] .'/format=Thumbnail?.jpg';
+
+  // Get the path to media files and XML files.
+  // This is real kludgy, but without a real API there's no other way.
+  // First we need to get the user facing HTML page.  
+  $html_page = drupal_http_request('http://www.archive.org/details/'. $item['value']);
+  if ($html_page->error || $html_page->code != 200) {
+    form_set_error($error_field, 'The HTML page for the item at archive.org could not be retrieved: ', $html_page->code .': '. $html_page->error);
+    return $data;
+  }
+  
+  // Scrape this page and find the path to the data directory.
+  // A regex expert would be able to do all this with a single regex statement, but that stuff is dark arts.
+  $html_chunks = explode('All Files: ', $html_page->data);
+  $html_chunks = explode('HTTP', $html_chunks[1]);
+  $html_chunks = explode('href=', $html_chunks[0]);
+  preg_match("/http:\/\/(.*)\"/", array_pop($html_chunks), $matches);
+  if (empty($matches)) {
+    form_set_error($error_field, 'The data directory for the item at archive.org could not be retrieved.');
+    return $data;
+  }
+  $data_url = rtrim($matches[0], '"');
+  
+  // In this directory should be two XML files, one for the list of files, one for metadata.
+  $xml_files_url = $data_url .'/'. $item['value'] . '_files.xml';
+  $xml_meta_url = $data_url .'/'. $item['value'] . '_meta.xml';
+  
+  // Retreive the XML files.
+  $xml_files = emfield_request_xml('archive', $xml_files_url, array(), TRUE, TRUE, $item['value'] . '_files');
+  $xml_meta = emfield_request_xml('archive', $xml_meta_url, array(), TRUE, TRUE, $item['value'] . '_meta');
+  if ($xml_meta['stat'] == 'error' || empty($xml_meta)) {
+    drupal_set_message('Additional information about the item at archive.org could not be retrieved.  The audio can still be displayed.');
+  }
+  else {
+    $data['metadata'] = $xml_meta['METADATA'];
+  }
+  if ($xml_files['stat'] == 'error' || empty($xml_files)) {
+    form_set_error($error_field, 'The list of files for the item at archive.org could not be retrieved.  The audio can not be displayed.');
+    return $data;
+  }
+  
+  // There are a tonne of useless thumbnails in this list that we don't want.  
+  // We also need to sort through the available derivitives and choose the best one.
+  $derivative_files = array('mp3' => '', 'ogg' => '');
+  foreach($xml_files as $file_name => $file_data) { 
+    $extension = substr($file_name, -3);
+    if (array_key_exists($extension, $derivative_files)) {
+      $derivative_files[$extension] = $file_name;
+    }
+  }
+  foreach($derivative_files as $file_name) {
+    if ($file_name) {
+      $data['url'] = 'http://www.archive.org/download/'. $item['value'] .'/'. $file_name;
+      break;
+    }
+  }
+
+  return $data;
+}
+
+/**
+ *  hook emaudio_PROVIDER_rss
+ *
+ *  This attaches a file to an RSS feed.
+ */
+function emaudio_archive_rss($item, $teaser = NULL) {
+  if ($item['value']) {
+    $file['thumbnail']['filepath'] = $item['data']['thumbnail'];
+
+    return $file;
+  }
+}
+
+/**
+ * hook emaudio_PROVIDER_embedded_link($audio_code)
+ * returns a link to view the audio at the provider's site.
+ *  @param $audio_code
+ *    The string containing the audio to watch.
+ *  @return
+ *    A string containing the URL to view the audio at the original provider's site.
+ */
+function emaudio_archive_embedded_link($audio_code) {
+  return 'http://www.archive.org/details/'. $audio_code;
+}
+
+/**
+ * The embedded flash displaying the archive audio.
+ */
+function theme_emaudio_archive_flash($item, $width, $height, $autoplay) {
+  $output = '';
+  $title = addslashes($item['data']['metadata']['TITLE'][0]);
+  if ($item['embed']) {
+    $output = <<<EOD
+      <embed type="application/x-shockwave-flash" width="$width"   height="$height"  allowfullscreen="true"  
+        allowscriptaccess="always"  src="http://www.archive.org/flow/flowplayer.commercial-3.0.5.swf"   
+        w3c="true"  flashvars='config={
+          "key":"#\$b6eb72a0f2f1e29f3d4",
+          "playlist":[
+            {
+              "url":"{$item['data']['url']}",
+              "autoPlay":false
+            }
+          ],
+          "clip":{
+            "autoPlay":true
+          },
+          "canvas":{
+            "backgroundColor":"0x000000",
+            "backgroundGradient":"none"
+          },
+          "plugins":{
+            "audio":{
+              "url":"http://www.archive.org/flow/flowplayer.audio-3.0.3-dev.swf"
+            },
+            "controls":{
+              "playlist":false,
+              "fullscreen":false,
+              "gloss":"high",
+              "backgroundColor":"0x000000",
+              "backgroundGradient":"medium",
+              "sliderColor":"0x777777",
+              "progressColor":"0x777777",
+              "timeColor":"0xeeeeee",
+              "durationColor":"0x01DAFF",
+              "buttonColor":"0x333333",
+              "buttonOverColor":"0x505050"
+            }
+          },
+          "contextMenu":[
+            {
+              "{$title}":"function()"
+            },
+            "-",
+            "Flowplayer 3.0.5"
+          ]
+        }'> </embed>
+EOD;
+  }
+  return $output;
+}
+
+/**
+ * hook emaudio_PROVIDER_thumbnail
+ * Returns the external url for a thumbnail of a specific audio.
+ *  @param $field
+ *    The field of the requesting node.
+ *  @param $item
+ *    The actual content of the field from the requesting node.
+ *  @return
+ *    A URL pointing to the thumbnail.
+ */
+function emaudio_archive_thumbnail($field, $item, $formatter, $node, $width, $height) {
+  // In this demonstration, we previously retrieved a thumbnail using oEmbed
+  // during the data hook.
+  return $item['data']['thumbnail'];
+}
+
+/**
+ *  hook emaudio_PROVIDER_audio
+ *  This actually displays the full/normal-sized audio we want, usually on the
+ *  default page view.
+ *  @param $embed
+ *    The audio code for the audio to embed.
+ *  @param $width
+ *    The width to display the audio.
+ *  @param $height
+ *    The height to display the audio.
+ *  @param $field
+ *    The field info from the requesting node.
+ *  @param $item
+ *    The actual content from the field.
+ *  @return
+ *    The html of the embedded audio.
+ */
+function emaudio_archive_audio($embed, $width, $height, $field, $item, &$node, $autoplay) {
+  $output = theme('emaudio_archive_flash', $item, $width, $height, $autoplay);
+  return $output;
+}
+
+/**
+ *  hook emaudio_PROVIDER_audio
+ *
+ *  This actually displays the preview-sized audio we want, commonly for the
+ *  teaser.
+ *  @param $embed
+ *    The audio code for the audio to embed.
+ *  @param $width
+ *    The width to display the audio.
+ *  @param $height
+ *    The height to display the audio.
+ *  @param $field
+ *    The field info from the requesting node.
+ *  @param $item
+ *    The actual content from the field.
+ *  @return
+ *    The html of the embedded audio.
+ */
+function emaudio_archive_preview($embed, $width, $height, $field, $item, &$node, $autoplay) {
+  $output = theme('emaudio_archive_flash', $item, $width, $height, $autoplay);
+  return $output;
+}
+
+/**
+ *  Implementation of hook_emfield_subtheme.
+ *  This returns any theme functions defined by this provider.
+ */
+function emaudio_archive_emfield_subtheme() {
+  $themes = array(
+    'emaudio_archive_flash'  => array(
+      'arguments' => array('item' => NULL, 'width' => NULL, 'height' => NULL, 'autoplay' => NULL),
+      'file' => 'providers/archive.inc',
+      // If you don't provide a 'path' value, then it will default to
+      // the emaudio.module path. Obviously, replace 'emarchive' with
+      // the actual name of your custom module.
+      //'path' => drupal_get_path('module', 'emarchive'),
+    )
+  );
+  return $themes;
+}