From f2d49a697124b8f4c6a4df68626b3d29ec959e70 Mon Sep 17 00:00:00 2001
From: Marc Abonce Seguin <marc-abonce@mailbox.org>
Date: Tue, 26 Mar 2019 20:33:36 -0600
Subject: [PATCH] [fix] get youtube results from js object

Results are not appearing in the html document anymore,
instead they are found inside an object embedded in a script.
---
 searx/engines/youtube_noapi.py           |  64 +++++----
 tests/unit/engines/test_youtube_noapi.py | 162 ++++++++---------------
 2 files changed, 87 insertions(+), 139 deletions(-)

diff --git a/searx/engines/youtube_noapi.py b/searx/engines/youtube_noapi.py
index 9f01841f..3bf25932 100644
--- a/searx/engines/youtube_noapi.py
+++ b/searx/engines/youtube_noapi.py
@@ -8,7 +8,8 @@
 # @stable      no
 # @parse       url, title, content, publishedDate, thumbnail, embedded
 
-from lxml import html
+from functools import reduce
+from json import loads
 from searx.engines.xpath import extract_text
 from searx.utils import list_get
 from searx.url_utils import quote_plus
@@ -34,20 +35,6 @@ embedded_url = '<iframe width="540" height="304" ' +\
 
 base_youtube_url = 'https://www.youtube.com/watch?v='
 
-# specific xpath variables
-results_xpath = "//ol/li/div[contains(@class, 'yt-lockup yt-lockup-tile yt-lockup-video vve-check')]"
-url_xpath = './/h3/a/@href'
-title_xpath = './/div[@class="yt-lockup-content"]/h3/a'
-content_xpath = './/div[@class="yt-lockup-content"]/div[@class="yt-lockup-description yt-ui-ellipsis yt-ui-ellipsis-2"]'
-
-
-# returns extract_text on the first result selected by the xpath or None
-def extract_text_from_dom(result, xpath):
-    r = result.xpath(xpath)
-    if len(r) > 0:
-        return extract_text(r[0])
-    return None
-
 
 # do search-request
 def request(query, params):
@@ -63,27 +50,38 @@ def request(query, params):
 def response(resp):
     results = []
 
-    dom = html.fromstring(resp.text)
+    results_data = resp.text[resp.text.find('ytInitialData'):]
+    results_data = results_data[results_data.find('{'):results_data.find(';\n')]
 
-    # parse results
-    for result in dom.xpath(results_xpath):
-        videoid = list_get(result.xpath('@data-context-item-id'), 0)
-        if videoid is not None:
-            url = base_youtube_url + videoid
-            thumbnail = 'https://i.ytimg.com/vi/' + videoid + '/hqdefault.jpg'
+    results_json = loads(results_data) if results_data else {}
+    sections = results_json.get('contents', {})\
+                           .get('twoColumnSearchResultsRenderer', {})\
+                           .get('primaryContents', {})\
+                           .get('sectionListRenderer', {})\
+                           .get('contents', [])
 
-            title = extract_text_from_dom(result, title_xpath) or videoid
-            content = extract_text_from_dom(result, content_xpath)
+    for section in sections:
+        for video_container in section.get('itemSectionRenderer', {}).get('contents', []):
+            video = video_container.get('videoRenderer', {})
+            videoid = video.get('videoId')
+            if videoid is not None:
+                url = base_youtube_url + videoid
+                thumbnail = 'https://i.ytimg.com/vi/' + videoid + '/hqdefault.jpg'
+                title = video.get('title', {}).get('simpleText', videoid)
+                description_snippet = video.get('descriptionSnippet', {})
+                if 'runs' in description_snippet:
+                    content = reduce(lambda a, b: a + b.get('text', ''), description_snippet.get('runs'), '')
+                else:
+                    content = description_snippet.get('simpleText', '')
+                embedded = embedded_url.format(videoid=videoid)
 
-            embedded = embedded_url.format(videoid=videoid)
-
-            # append result
-            results.append({'url': url,
-                            'title': title,
-                            'content': content,
-                            'template': 'videos.html',
-                            'embedded': embedded,
-                            'thumbnail': thumbnail})
+                # append result
+                results.append({'url': url,
+                                'title': title,
+                                'content': content,
+                                'template': 'videos.html',
+                                'embedded': embedded,
+                                'thumbnail': thumbnail})
 
     # return results
     return results
diff --git a/tests/unit/engines/test_youtube_noapi.py b/tests/unit/engines/test_youtube_noapi.py
index 41dcbb74..cbf7b9bc 100644
--- a/tests/unit/engines/test_youtube_noapi.py
+++ b/tests/unit/engines/test_youtube_noapi.py
@@ -46,121 +46,71 @@ class TestYoutubeNoAPIEngine(SearxTestCase):
         self.assertEqual(youtube_noapi.response(response), [])
 
         html = """
-        <ol id="item-section-063864" class="item-section">
-            <li>
-                <div class="yt-lockup yt-lockup-tile yt-lockup-video vve-check clearfix yt-uix-tile"
-                data-context-item-id="DIVZCPfAOeM"
-                data-visibility-tracking="CBgQ3DAYACITCPGXnYau6sUCFZEIHAod-VQASCj0JECx_-GK5uqMpcIB">
-                <div class="yt-lockup-dismissable"><div class="yt-lockup-thumbnail contains-addto">
-                <a aria-hidden="true" href="/watch?v=DIVZCPfAOeM" class=" yt-uix-sessionlink pf-link"
-                data-sessionlink="itct=CBgQ3DAYACITCPGXnYau6sUCFZEIHAod-VQASCj0JFIEdGVzdA">
-                <div class="yt-thumb video-thumb"><img src="//i.ytimg.com/vi/DIVZCPfAOeM/mqdefault.jpg"
-                width="196" height="110"/></div><span class="video-time" aria-hidden="true">11:35</span></a>
-                <span class="thumb-menu dark-overflow-action-menu video-actions">
-                </span>
-                </div>
-                <div class="yt-lockup-content">
-                <h3 class="yt-lockup-title">
-                <a href="/watch?v=DIVZCPfAOeM"
-                class="yt-uix-tile-link yt-ui-ellipsis yt-ui-ellipsis-2 yt-uix-sessionlink spf-link"
-                data-sessionlink="itct=CBgQ3DAYACITCPGXnYau6sUCFZEIHAod-VQASCj0JFIEdGVzdA"
-                title="Top Speed Test Kawasaki Ninja H2 (Thailand) By. MEHAY SUPERBIKE"
-                aria-describedby="description-id-259079" rel="spf-prefetch" dir="ltr">
-                Title
-                </a>
-                <span class="accessible-description" id="description-id-259079"> - Durée : 11:35.</span>
-                </h3>
-                <div class="yt-lockup-byline">de
-                <a href="/user/mheejapan" class=" yt-uix-sessionlink spf-link g-hovercard"
-                data-sessionlink="itct=CBgQ3DAYACITCPGXnYau6sUCFZEIHAod-VQASCj0JA" data-ytid="UCzEesu54Hjs0uRKmpy66qeA"
-                data-name="">MEHAY SUPERBIKE</a></div><div class="yt-lockup-meta">
-                <ul class="yt-lockup-meta-info">
-                    <li>il y a 20 heures</li>
-                    <li>8 424 vues</li>
-                </ul>
-                </div>
-                <div class="yt-lockup-description yt-ui-ellipsis yt-ui-ellipsis-2" dir="ltr">
-                Description
-                </div>
-                <div class="yt-lockup-badges">
-                <ul class="yt-badge-list ">
-                    <li class="yt-badge-item" >
-                        <span class="yt-badge">Nouveauté</span>
-                    </li>
-                    <li class="yt-badge-item" ><span class="yt-badge " >HD</span></li>
-                </ul>
-                </div>
-                <div class="yt-lockup-action-menu yt-uix-menu-container">
-                <div class="yt-uix-menu yt-uix-videoactionmenu hide-until-delayloaded"
-                data-video-id="DIVZCPfAOeM" data-menu-content-id="yt-uix-videoactionmenu-menu">
-                </div>
-                </div>
-                </div>
-                </div>
-                </div>
-            </li>
-        </ol>
+        <div></div>
+        <script>
+            window["ytInitialData"] = {
+                "contents": {
+                    "twoColumnSearchResultsRenderer": {
+                        "primaryContents": {
+                            "sectionListRenderer": {
+                                "contents": [
+                                    {
+                                        "itemSectionRenderer": {
+                                            "contents": [
+                                                {
+                                                    "videoRenderer": {
+                                                        "videoId": "DIVZCPfAOeM",
+                                                        "title": {
+                                                            "simpleText": "Title"
+                                                        },
+                                                        "descriptionSnippet": {
+                                                            "runs": [
+                                                                {
+                                                                    "text": "Des"
+                                                                },
+                                                                {
+                                                                    "text": "cription"
+                                                                }
+                                                            ]
+                                                        }
+                                                    }
+                                                },
+                                                {
+                                                    "videoRenderer": {
+                                                        "videoId": "9C_HReR_McQ",
+                                                        "title": {
+                                                            "simpleText": "Title"
+                                                        },
+                                                        "descriptionSnippet": {
+                                                            "simpleText": "Description"
+                                                        }
+                                                    }
+                                                }
+                                            ]
+                                        }
+                                    }
+                                ]
+                            }
+                        }
+                    }
+                }
+            };
+        </script>
         """
         response = mock.Mock(text=html)
         results = youtube_noapi.response(response)
         self.assertEqual(type(results), list)
-        self.assertEqual(len(results), 1)
+        self.assertEqual(len(results), 2)
         self.assertEqual(results[0]['title'], 'Title')
         self.assertEqual(results[0]['url'], 'https://www.youtube.com/watch?v=DIVZCPfAOeM')
         self.assertEqual(results[0]['content'], 'Description')
         self.assertEqual(results[0]['thumbnail'], 'https://i.ytimg.com/vi/DIVZCPfAOeM/hqdefault.jpg')
         self.assertTrue('DIVZCPfAOeM' in results[0]['embedded'])
-
-        html = """
-        <ol id="item-section-063864" class="item-section">
-            <li>
-                <div class="yt-lockup yt-lockup-tile yt-lockup-video vve-check clearfix yt-uix-tile"
-                data-context-item-id="DIVZCPfAOeM"
-                data-visibility-tracking="CBgQ3DAYACITCPGXnYau6sUCFZEIHAod-VQASCj0JECx_-GK5uqMpcIB">
-                <div class="yt-lockup-dismissable"><div class="yt-lockup-thumbnail contains-addto">
-                <a aria-hidden="true" href="/watch?v=DIVZCPfAOeM" class=" yt-uix-sessionlink pf-link"
-                data-sessionlink="itct=CBgQ3DAYACITCPGXnYau6sUCFZEIHAod-VQASCj0JFIEdGVzdA">
-                <div class="yt-thumb video-thumb"><img src="//i.ytimg.com/vi/DIVZCPfAOeM/mqdefault.jpg"
-                width="196" height="110"/></div><span class="video-time" aria-hidden="true">11:35</span></a>
-                <span class="thumb-menu dark-overflow-action-menu video-actions">
-                </span>
-                </div>
-                <div class="yt-lockup-content">
-                <h3 class="yt-lockup-title">
-                <span class="accessible-description" id="description-id-259079"> - Durée : 11:35.</span>
-                </h3>
-                <div class="yt-lockup-byline">de
-                <a href="/user/mheejapan" class=" yt-uix-sessionlink spf-link g-hovercard"
-                data-sessionlink="itct=CBgQ3DAYACITCPGXnYau6sUCFZEIHAod-VQASCj0JA" data-ytid="UCzEesu54Hjs0uRKmpy66qeA"
-                data-name="">MEHAY SUPERBIKE</a></div><div class="yt-lockup-meta">
-                <ul class="yt-lockup-meta-info">
-                    <li>il y a 20 heures</li>
-                    <li>8 424 vues</li>
-                </ul>
-                </div>
-                <div class="yt-lockup-badges">
-                <ul class="yt-badge-list ">
-                    <li class="yt-badge-item" >
-                        <span class="yt-badge">Nouveauté</span>
-                    </li>
-                    <li class="yt-badge-item" ><span class="yt-badge " >HD</span></li>
-                </ul>
-                </div>
-                <div class="yt-lockup-action-menu yt-uix-menu-container">
-                <div class="yt-uix-menu yt-uix-videoactionmenu hide-until-delayloaded"
-                data-video-id="DIVZCPfAOeM" data-menu-content-id="yt-uix-videoactionmenu-menu">
-                </div>
-                </div>
-                </div>
-                </div>
-                </div>
-            </li>
-        </ol>
-        """
-        response = mock.Mock(text=html)
-        results = youtube_noapi.response(response)
-        self.assertEqual(type(results), list)
-        self.assertEqual(len(results), 1)
+        self.assertEqual(results[1]['title'], 'Title')
+        self.assertEqual(results[1]['url'], 'https://www.youtube.com/watch?v=9C_HReR_McQ')
+        self.assertEqual(results[1]['content'], 'Description')
+        self.assertEqual(results[1]['thumbnail'], 'https://i.ytimg.com/vi/9C_HReR_McQ/hqdefault.jpg')
+        self.assertTrue('9C_HReR_McQ' in results[1]['embedded'])
 
         html = """
         <ol id="item-section-063864" class="item-section">