Use XPath to find the elements of interest

[GM_direct_download_links.git] / direct_download_links.user.js

diff --git a/direct_download_links.user.js b/direct_download_links.user.js
index 9f42a3f..9344012 100644 (file)
--- a/direct_download_links.user.js
+++ b/direct_download_links.user.js
@@ -1,6 +1,6 @@
 // direct_download_links - Add direct download links
 // direct_download_links - Add direct download links

-// version 0.2
-// 2011-11-14
+// version 0.3
+// 2011-12-23

 // Copyright (C) 2011  Antonio Ospite <ospite@studenti.unina.it>
 // Released under the GPL license
 // http://www.gnu.org/copyleft/gpl.html
 // Copyright (C) 2011  Antonio Ospite <ospite@studenti.unina.it>
 // Released under the GPL license
 // http://www.gnu.org/copyleft/gpl.html


@@ -28,31 +28,37 @@
 // @include        http://trovacinema.repubblica.it/*
 // @include        http://www.kataweb.it/tvzap/*
 // @include        http://www.rai.tv/*
 // @include        http://trovacinema.repubblica.it/*
 // @include        http://www.kataweb.it/tvzap/*
 // @include        http://www.rai.tv/*

+// @include        http://soundcloud.com/*

 // ==/UserScript==
 //
 // ==/UserScript==
 //

-// NOTE, for rai.tv to work you need to install a script like:
-// http://git.ao2.it/smooth-dl.git/blob_plain/HEAD:/scripts/SilverSpoof.user.js

 
 /*
  * TODO:
  *  - find a way to use the same string as in the @include lines to match the
  *    current window.location. Look for something like GM_testUrl() which builds
  *    the regexp starting from a glob line.
 
 /*
  * TODO:
  *  - find a way to use the same string as in the @include lines to match the
  *    current window.location. Look for something like GM_testUrl() which builds
  *    the regexp starting from a glob line.

- *  - use xpath instead of regexp like in http://a32.me/2009/11/greasemonkey/

  *  - use jquery, like shown in http://a32.me/2009/11/greasemonkey/
  *  - use jquery, like shown in http://a32.me/2009/11/greasemonkey/

+ *  - Support the "download" attribute for anchors:
+ *    http://www.whatwg.org/specs/web-apps/current-work/multipage/links.html#downloading-resources

  */
 
 /* Fields supported by the "site" object.
  *
  * Manadatory fields:
  */
 
 /* Fields supported by the "site" object.
  *
  * Manadatory fields:

- *   pageURL: the URL of the page we are modifying
- *   urlContainer: the element containing the URL to link
+ *   locationRegExp: the regexp describing the URL of the page we are modifying
+ *   urlContainerXPath: the XPath of the element containing the URL to link

  *   urlRegexp: the regular expression for finding the URL, the first
  *              sub-pattern is taken as the URL
  *   urlRegexp: the regular expression for finding the URL, the first
  *              sub-pattern is taken as the URL

- *   linkDest: the element where to place the Direct Download link
+ *   linkDestXPath: the XPath of the element where to place the Direct Download link

  *
  *
  * Optional fields:
  *
  *
  * Optional fields:

+ *
+ *   initCommand: a function called before the regExp is matched, this can
+ *                be useful in cases when some action needs to be done in
+ *                order to make the element containing the regExp be actually
+ *                rendered. It must accept  a 'site' parameter.
+ *

  *   onEvent: used to delay the urlRegexp matching to a certain event like
  *            'DOMNodeInserted' useful when the URL is added by some javascript
  *            library. It has two fields:
  *   onEvent: used to delay the urlRegexp matching to a certain event like
  *            'DOMNodeInserted' useful when the URL is added by some javascript
  *            library. It has two fields:


@@ -71,35 +77,47 @@
 var supported_sites = [
   {
     locationRegexp: /^http:\/\/video\.repubblica\.it\/.*$/,
 var supported_sites = [
   {
     locationRegexp: /^http:\/\/video\.repubblica\.it\/.*$/,

-    urlContainer: 'contA',
+    urlContainerXPath: '//div[@id="contA"]',

     urlRegexp: /'pcUrl', '((http|mms):\/\/[^']*)'/,
     urlRegexp: /'pcUrl', '((http|mms):\/\/[^']*)'/,

-    linkDest: 'contA',
+    linkDestXPath: '//div[@id="contA"]',

   },
   {
     locationRegexp: /^http:\/\/tv\.repubblica\.it\/.*$/,
   },
   {
     locationRegexp: /^http:\/\/tv\.repubblica\.it\/.*$/,

-    urlContainer: 'boxPlayer',
+    urlContainerXPath: '//div[@id="boxPlayer"]',

     urlRegexp: /'pcUrl', '((http|mms):\/\/[^']*)'/,
     linkDest: 'box_embed',
     urlRegexp: /'pcUrl', '((http|mms):\/\/[^']*)'/,
     linkDest: 'box_embed',

+    linkDestXPath: '//div[@id="box_embed"]',

   },
   {
     locationRegexp: /^http:\/\/trovacinema\.repubblica\.it\/.*$/,
   },
   {
     locationRegexp: /^http:\/\/trovacinema\.repubblica\.it\/.*$/,

-    urlContainer: 'col-center',
+    urlContainerXPath: '//div[@id="col-center"]',

     urlRegexp: /'flvUrl', '((http|mms):\/\/[^']*)'/,
     urlRegexp: /'flvUrl', '((http|mms):\/\/[^']*)'/,

-    linkDest: 'col-center',
+    linkDestXPath: '//div[@id="col-center"]',

   },
   {
     locationRegexp: /^http:\/\/www\.kataweb\.it\/tvzap\/.*$/,
   },
   {
     locationRegexp: /^http:\/\/www\.kataweb\.it\/tvzap\/.*$/,

-    urlContainer: 'tvzap_video',
+    urlContainerXPath: '//div[@id="tvzap_video"]',

     urlRegexp: /'pcUrl', '((http|mms):\/\/[^']*)'/,
     urlRegexp: /'pcUrl', '((http|mms):\/\/[^']*)'/,

-    linkDest: 'playerCont',
+    linkDestXPath: '//div[@id="playerCont"]',

   },
   {
     locationRegexp: /^http:\/\/www\.rai\.tv\/.*$/,
   },
   {
     locationRegexp: /^http:\/\/www\.rai\.tv\/.*$/,

-    urlContainer: 'Player',
+      initCommand: function(site) {
+        unsafeWindow.Silverlight.isInstalled = function(version) {
+          return true;
+        };
+    },
+    urlContainerXPath: '//div[@id="Player"]',

     urlRegexp: /mediaUri=(http:\/\/[^,]*)/,
     onEvent: { evt: 'DOMNodeInserted', targetElement: 'object' },
     processURL: _rai_get_actual_url,
     urlRegexp: /mediaUri=(http:\/\/[^,]*)/,
     onEvent: { evt: 'DOMNodeInserted', targetElement: 'object' },
     processURL: _rai_get_actual_url,

-    linkDest: 'Player',
+    linkDestXPath: '//div[@id="Player"]',
+  },
+  {
+    locationRegexp: /^http:\/\/soundcloud.com\/.*$/,
+    urlContainerXPath: '//div[@id="main-content-inner"]',
+    urlRegexp: /"streamUrl":"([^"]*)"/,
+    linkDestXPath: '//div[@id="main-content-inner"]',

   },
 ];
 
   },
 ];
 


@@ -109,10 +127,17 @@ for (i = 0; i < supported_sites.length; i++) {
 
   var result = window.location.href.match(site.locationRegexp);
   if (result) {
 
   var result = window.location.href.match(site.locationRegexp);
   if (result) {

+    if (site.initCommand) {
+      site.initCommand(site);
+    }

     direct_download_link_add(window.location.href, site);
   }
 }
 
     direct_download_link_add(window.location.href, site);
   }
 }
 

+function getElementByXPath(query, root) {
+  return document.evaluate(query, root || document, null, XPathResult.ANY_UNORDERED_NODE_TYPE, null).singleNodeValue;
+}
+

 /**
  * Add a Direct Download link on the page for the specified URL
  *
 /**
  * Add a Direct Download link on the page for the specified URL
  *


@@ -121,9 +146,9 @@ for (i = 0; i < supported_sites.length; i++) {
  * @return: null on error, true on success
  */
 function direct_download_link_add(pageURL, site) {
  * @return: null on error, true on success
  */
 function direct_download_link_add(pageURL, site) {

-  var element = document.getElementById(site.urlContainer);
+  var element = getElementByXPath(site.urlContainerXPath);

   if (!element) {
   if (!element) {

-    DDL_log('DirectDL (' + site.pageURL  + '): Cannot find the element ' + site.urlContainer + ' containing the URL.');
+    DDL_log('DirectDL (' + site.pageURL  + '): Cannot find the element ' + site.urlContainerXPath + ' containing the URL.');

     return null;
   }
 
     return null;
   }
 


@@ -140,6 +165,7 @@ function direct_download_link_add(pageURL, site) {
       }
      _get_URL(site, element);
     }, false);
       }
      _get_URL(site, element);
     }, false);

+    return;

   }
 
   _get_URL(site, element);
   }
 
   _get_URL(site, element);


@@ -168,8 +194,8 @@ function _get_URL(site, element) {
     return;
   }
 
     return;
   }
 

-  var evt = document.createEvent('Event');  
-  evt.initEvent('UrlFetched', true, true);  
+  var evt = document.createEvent('Event');
+  evt.initEvent('UrlFetched', true, true);

   evt.site = site;
   evt.URL = URL;
   document.dispatchEvent(evt);
   evt.site = site;
   evt.URL = URL;
   document.dispatchEvent(evt);


@@ -179,7 +205,7 @@ function _add_link(e) {
   var site = e.site;
   var URL = e.URL;;
 
   var site = e.site;
   var URL = e.URL;;
 

-  var destination = document.getElementById(site.linkDest);
+  var destination = getElementByXPath(site.linkDestXPath);

   if (!destination) {
     DDL_log('DirectDl (' + site.pageURL + '): Cannot add the direct download link.');
     return;
   if (!destination) {
     DDL_log('DirectDl (' + site.pageURL + '): Cannot add the direct download link.');
     return;


@@ -199,6 +225,7 @@ function _add_link(e) {
     style += ' border: 2px solid red;'
     style += ' float: right; font-size: large;';
     style += ' padding: .5em; margin: 1em;'
     style += ' border: 2px solid red;'
     style += ' float: right; font-size: large;';
     style += ' padding: .5em; margin: 1em;'

+    style += ' position: relative; z-index: 1000;'

     download_link.setAttribute('style', style);
 
     destination.insertBefore(download_link, destination.firstChild);
     download_link.setAttribute('style', style);
 
     destination.insertBefore(download_link, destination.firstChild);


@@ -215,6 +242,17 @@ function DDL_log(message) {
 }
 
 function _rai_get_actual_url(site, URL) {
 }
 
 function _rai_get_actual_url(site, URL) {

+
+  // SmoothStreaming manifest files get added without processing, for now:
+  if (URL.match(/.*\.csm$/)) {
+    var evt = document.createEvent('Event');
+    evt.initEvent('UrlFetched', true, true);
+    evt.site = site;
+    evt.URL = URL;
+    document.dispatchEvent(evt);
+    return;
+  }
+

   // http://www.neaveru.com/wordpress/index.php/2008/05/09/greasemonkey-bug-domnodeinserted-event-doesnt-allow-gm_xmlhttprequest/
   setTimeout( function() {
     GM_xmlhttpRequest({
   // http://www.neaveru.com/wordpress/index.php/2008/05/09/greasemonkey-bug-domnodeinserted-event-doesnt-allow-gm_xmlhttprequest/
   setTimeout( function() {
     GM_xmlhttpRequest({


@@ -233,8 +271,8 @@ function _rai_get_actual_url(site, URL) {
         if (ref.length > 0) {
           href = ref[0].getAttribute('HREF');;
 
         if (ref.length > 0) {
           href = ref[0].getAttribute('HREF');;
 

-          var evt = document.createEvent('Event');  
-          evt.initEvent('UrlFetched', true, true);  
+          var evt = document.createEvent('Event');
+          evt.initEvent('UrlFetched', true, true);

           evt.site = site;
           evt.URL = href;
           document.dispatchEvent(evt);
           evt.site = site;
           evt.URL = href;
           document.dispatchEvent(evt);