[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.
This commit is contained in:
Marc Abonce Seguin 2019-03-26 20:33:36 -06:00
parent a80a2d05d1
commit f2d49a6971
2 changed files with 87 additions and 139 deletions

View file

@ -8,7 +8,8 @@
# @stable no # @stable no
# @parse url, title, content, publishedDate, thumbnail, embedded # @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.engines.xpath import extract_text
from searx.utils import list_get from searx.utils import list_get
from searx.url_utils import quote_plus 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=' 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 # do search-request
def request(query, params): def request(query, params):
@ -63,27 +50,38 @@ def request(query, params):
def response(resp): def response(resp):
results = [] 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 results_json = loads(results_data) if results_data else {}
for result in dom.xpath(results_xpath): sections = results_json.get('contents', {})\
videoid = list_get(result.xpath('@data-context-item-id'), 0) .get('twoColumnSearchResultsRenderer', {})\
if videoid is not None: .get('primaryContents', {})\
url = base_youtube_url + videoid .get('sectionListRenderer', {})\
thumbnail = 'https://i.ytimg.com/vi/' + videoid + '/hqdefault.jpg' .get('contents', [])
title = extract_text_from_dom(result, title_xpath) or videoid for section in sections:
content = extract_text_from_dom(result, content_xpath) 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,
# append result 'title': title,
results.append({'url': url, 'content': content,
'title': title, 'template': 'videos.html',
'content': content, 'embedded': embedded,
'template': 'videos.html', 'thumbnail': thumbnail})
'embedded': embedded,
'thumbnail': thumbnail})
# return results # return results
return results return results

View file

@ -46,121 +46,71 @@ class TestYoutubeNoAPIEngine(SearxTestCase):
self.assertEqual(youtube_noapi.response(response), []) self.assertEqual(youtube_noapi.response(response), [])
html = """ html = """
<ol id="item-section-063864" class="item-section"> <div></div>
<li> <script>
<div class="yt-lockup yt-lockup-tile yt-lockup-video vve-check clearfix yt-uix-tile" window["ytInitialData"] = {
data-context-item-id="DIVZCPfAOeM" "contents": {
data-visibility-tracking="CBgQ3DAYACITCPGXnYau6sUCFZEIHAod-VQASCj0JECx_-GK5uqMpcIB"> "twoColumnSearchResultsRenderer": {
<div class="yt-lockup-dismissable"><div class="yt-lockup-thumbnail contains-addto"> "primaryContents": {
<a aria-hidden="true" href="/watch?v=DIVZCPfAOeM" class=" yt-uix-sessionlink pf-link" "sectionListRenderer": {
data-sessionlink="itct=CBgQ3DAYACITCPGXnYau6sUCFZEIHAod-VQASCj0JFIEdGVzdA"> "contents": [
<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> "itemSectionRenderer": {
<span class="thumb-menu dark-overflow-action-menu video-actions"> "contents": [
</span> {
</div> "videoRenderer": {
<div class="yt-lockup-content"> "videoId": "DIVZCPfAOeM",
<h3 class="yt-lockup-title"> "title": {
<a href="/watch?v=DIVZCPfAOeM" "simpleText": "Title"
class="yt-uix-tile-link yt-ui-ellipsis yt-ui-ellipsis-2 yt-uix-sessionlink spf-link" },
data-sessionlink="itct=CBgQ3DAYACITCPGXnYau6sUCFZEIHAod-VQASCj0JFIEdGVzdA" "descriptionSnippet": {
title="Top Speed Test Kawasaki Ninja H2 (Thailand) By. MEHAY SUPERBIKE" "runs": [
aria-describedby="description-id-259079" rel="spf-prefetch" dir="ltr"> {
Title "text": "Des"
</a> },
<span class="accessible-description" id="description-id-259079"> - Durée : 11:35.</span> {
</h3> "text": "cription"
<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> "videoRenderer": {
</ul> "videoId": "9C_HReR_McQ",
</div> "title": {
<div class="yt-lockup-description yt-ui-ellipsis yt-ui-ellipsis-2" dir="ltr"> "simpleText": "Title"
Description },
</div> "descriptionSnippet": {
<div class="yt-lockup-badges"> "simpleText": "Description"
<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> </script>
</div>
</div>
</li>
</ol>
""" """
response = mock.Mock(text=html) response = mock.Mock(text=html)
results = youtube_noapi.response(response) results = youtube_noapi.response(response)
self.assertEqual(type(results), list) 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]['title'], 'Title')
self.assertEqual(results[0]['url'], 'https://www.youtube.com/watch?v=DIVZCPfAOeM') self.assertEqual(results[0]['url'], 'https://www.youtube.com/watch?v=DIVZCPfAOeM')
self.assertEqual(results[0]['content'], 'Description') self.assertEqual(results[0]['content'], 'Description')
self.assertEqual(results[0]['thumbnail'], 'https://i.ytimg.com/vi/DIVZCPfAOeM/hqdefault.jpg') self.assertEqual(results[0]['thumbnail'], 'https://i.ytimg.com/vi/DIVZCPfAOeM/hqdefault.jpg')
self.assertTrue('DIVZCPfAOeM' in results[0]['embedded']) self.assertTrue('DIVZCPfAOeM' in results[0]['embedded'])
self.assertEqual(results[1]['title'], 'Title')
html = """ self.assertEqual(results[1]['url'], 'https://www.youtube.com/watch?v=9C_HReR_McQ')
<ol id="item-section-063864" class="item-section"> self.assertEqual(results[1]['content'], 'Description')
<li> self.assertEqual(results[1]['thumbnail'], 'https://i.ytimg.com/vi/9C_HReR_McQ/hqdefault.jpg')
<div class="yt-lockup yt-lockup-tile yt-lockup-video vve-check clearfix yt-uix-tile" self.assertTrue('9C_HReR_McQ' in results[1]['embedded'])
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)
html = """ html = """
<ol id="item-section-063864" class="item-section"> <ol id="item-section-063864" class="item-section">