on master help sign in
Merge pull request #1544 from MarcAbonce/youtube_fix
ca09d91Alexandre Flament
searx/engines/youtube_noapi.py+34−36
@@ -8,7 +8,8 @@
8# @stable no
8# @stable no
9# @parse url, title, content, publishedDate, thumbnail, embedded
9# @parse url, title, content, publishedDate, thumbnail, embedded
10
10
11−from lxml import html
11+from functools import reduce
12+from json import loads
12from searx.engines.xpath import extract_text
13from searx.engines.xpath import extract_text
13from searx.utils import list_get
14from searx.utils import list_get
14from searx.url_utils import quote_plus
15from searx.url_utils import quote_plus
@@ -34,20 +35,6 @@ embedded_url = '<iframe width="540" height="304" ' +\
34
35
35base_youtube_url = 'https://www.youtube.com/watch?v='
36base_youtube_url = 'https://www.youtube.com/watch?v='
36
37
37−# specific xpath variables
38−results_xpath = "//ol/li/div[contains(@class, 'yt-lockup yt-lockup-tile yt-lockup-video vve-check')]"
39−url_xpath = './/h3/a/@href'
40−title_xpath = './/div[@class="yt-lockup-content"]/h3/a'
41−content_xpath = './/div[@class="yt-lockup-content"]/div[@class="yt-lockup-description yt-ui-ellipsis yt-ui-ellipsis-2"]'
42−
43−
44−# returns extract_text on the first result selected by the xpath or None
45−def extract_text_from_dom(result, xpath):
46− r = result.xpath(xpath)
47− if len(r) > 0:
48− return extract_text(r[0])
49− return None
50−
51
38
52# do search-request
39# do search-request
53def request(query, params):
40def request(query, params):
@@ -63,27 +50,38 @@ def request(query, params):
63def response(resp):
50def response(resp):
64 results = []
51 results = []
65
52
66− dom = html.fromstring(resp.text)
53+ results_data = resp.text[resp.text.find('ytInitialData'):]
67−
54+ results_data = results_data[results_data.find('{'):results_data.find(';\n')]
68− # parse results
55+
69− for result in dom.xpath(results_xpath):
56+ results_json = loads(results_data) if results_data else {}
70− videoid = list_get(result.xpath('@data-context-item-id'), 0)
57+ sections = results_json.get('contents', {})\
58+ .get('twoColumnSearchResultsRenderer', {})\
59+ .get('primaryContents', {})\
60+ .get('sectionListRenderer', {})\
61+ .get('contents', [])
62+
63+ for section in sections:
64+ for video_container in section.get('itemSectionRenderer', {}).get('contents', []):
65+ video = video_container.get('videoRenderer', {})
66+ videoid = video.get('videoId')
71− if videoid is not None:
67+ if videoid is not None:
72− url = base_youtube_url + videoid
68+ url = base_youtube_url + videoid
73− thumbnail = 'https://i.ytimg.com/vi/' + videoid + '/hqdefault.jpg'
69+ thumbnail = 'https://i.ytimg.com/vi/' + videoid + '/hqdefault.jpg'
74−
70+ title = video.get('title', {}).get('simpleText', videoid)
75− title = extract_text_from_dom(result, title_xpath) or videoid
71+ description_snippet = video.get('descriptionSnippet', {})
76− content = extract_text_from_dom(result, content_xpath)
72+ if 'runs' in description_snippet:
77−
73+ content = reduce(lambda a, b: a + b.get('text', ''), description_snippet.get('runs'), '')
74+ else:
75+ content = description_snippet.get('simpleText', '')
78− embedded = embedded_url.format(videoid=videoid)
76+ embedded = embedded_url.format(videoid=videoid)
79−
77+
80− # append result
78+ # append result
81− results.append({'url': url,
79+ results.append({'url': url,
82− 'title': title,
80+ 'title': title,
83− 'content': content,
81+ 'content': content,
84− 'template': 'videos.html',
82+ 'template': 'videos.html',
85− 'embedded': embedded,
83+ 'embedded': embedded,
86− 'thumbnail': thumbnail})
84+ 'thumbnail': thumbnail})
87
85
88 # return results
86 # return results
89 return results
87 return results
tests/unit/engines/test_youtube_noapi.py+56−106
@@ -46,121 +46,71 @@ class TestYoutubeNoAPIEngine(SearxTestCase):
46 self.assertEqual(youtube_noapi.response(response), [])
46 self.assertEqual(youtube_noapi.response(response), [])
47
47
48 html = """
48 html = """
49− <ol id="item-section-063864" class="item-section">
49+ <div></div>
50− <li>
50+ <script>
51− <div class="yt-lockup yt-lockup-tile yt-lockup-video vve-check clearfix yt-uix-tile"
51+ window["ytInitialData"] = {
52− data-context-item-id="DIVZCPfAOeM"
52+ "contents": {
53− data-visibility-tracking="CBgQ3DAYACITCPGXnYau6sUCFZEIHAod-VQASCj0JECx_-GK5uqMpcIB">
53+ "twoColumnSearchResultsRenderer": {
54− <div class="yt-lockup-dismissable"><div class="yt-lockup-thumbnail contains-addto">
54+ "primaryContents": {
55− <a aria-hidden="true" href="/watch?v=DIVZCPfAOeM" class=" yt-uix-sessionlink pf-link"
55+ "sectionListRenderer": {
56− data-sessionlink="itct=CBgQ3DAYACITCPGXnYau6sUCFZEIHAod-VQASCj0JFIEdGVzdA">
56+ "contents": [
57− <div class="yt-thumb video-thumb"><img src="//i.ytimg.com/vi/DIVZCPfAOeM/mqdefault.jpg"
57+ {
58− width="196" height="110"/></div><span class="video-time" aria-hidden="true">11:35</span></a>
58+ "itemSectionRenderer": {
59− <span class="thumb-menu dark-overflow-action-menu video-actions">
59+ "contents": [
60− </span>
60+ {
61− </div>
61+ "videoRenderer": {
62− <div class="yt-lockup-content">
62+ "videoId": "DIVZCPfAOeM",
63− <h3 class="yt-lockup-title">
63+ "title": {
64− <a href="/watch?v=DIVZCPfAOeM"
64+ "simpleText": "Title"
65− class="yt-uix-tile-link yt-ui-ellipsis yt-ui-ellipsis-2 yt-uix-sessionlink spf-link"
65+ },
66− data-sessionlink="itct=CBgQ3DAYACITCPGXnYau6sUCFZEIHAod-VQASCj0JFIEdGVzdA"
66+ "descriptionSnippet": {
67− title="Top Speed Test Kawasaki Ninja H2 (Thailand) By. MEHAY SUPERBIKE"
67+ "runs": [
68− aria-describedby="description-id-259079" rel="spf-prefetch" dir="ltr">
68+ {
69− Title
69+ "text": "Des"
70− </a>
70+ },
71− <span class="accessible-description" id="description-id-259079"> - Durée : 11:35.</span>
71+ {
72− </h3>
72+ "text": "cription"
73− <div class="yt-lockup-byline">de
73+ }
74− <a href="/user/mheejapan" class=" yt-uix-sessionlink spf-link g-hovercard"
74+ ]
75− data-sessionlink="itct=CBgQ3DAYACITCPGXnYau6sUCFZEIHAod-VQASCj0JA" data-ytid="UCzEesu54Hjs0uRKmpy66qeA"
75+ }
76− data-name="">MEHAY SUPERBIKE</a></div><div class="yt-lockup-meta">
76+ }
77− <ul class="yt-lockup-meta-info">
77+ },
78− <li>il y a 20 heures</li>
78+ {
79− <li>8 424 vues</li>
79+ "videoRenderer": {
80− </ul>
80+ "videoId": "9C_HReR_McQ",
81− </div>
81+ "title": {
82− <div class="yt-lockup-description yt-ui-ellipsis yt-ui-ellipsis-2" dir="ltr">
82+ "simpleText": "Title"
83+ },
84+ "descriptionSnippet": {
83− Description
85+ "simpleText": "Description"
84− </div>
86+ }
85− <div class="yt-lockup-badges">
87+ }
86− <ul class="yt-badge-list ">
88+ }
87− <li class="yt-badge-item" >
89+ ]
88− <span class="yt-badge">Nouveauté</span>
90+ }
89− </li>
91+ }
90− <li class="yt-badge-item" ><span class="yt-badge " >HD</span></li>
92+ ]
91− </ul>
93+ }
92− </div>
94+ }
93− <div class="yt-lockup-action-menu yt-uix-menu-container">
95+ }
94− <div class="yt-uix-menu yt-uix-videoactionmenu hide-until-delayloaded"
96+ }
95− data-video-id="DIVZCPfAOeM" data-menu-content-id="yt-uix-videoactionmenu-menu">
97+ };
96− </div>
98+ </script>
97− </div>
98− </div>
99− </div>
100− </div>
101− </li>
102− </ol>
103 """
99 """
104 response = mock.Mock(text=html)
100 response = mock.Mock(text=html)
105 results = youtube_noapi.response(response)
101 results = youtube_noapi.response(response)
106 self.assertEqual(type(results), list)
102 self.assertEqual(type(results), list)
107− self.assertEqual(len(results), 1)
103+ self.assertEqual(len(results), 2)
108 self.assertEqual(results[0]['title'], 'Title')
104 self.assertEqual(results[0]['title'], 'Title')
109 self.assertEqual(results[0]['url'], 'https://www.youtube.com/watch?v=DIVZCPfAOeM')
105 self.assertEqual(results[0]['url'], 'https://www.youtube.com/watch?v=DIVZCPfAOeM')
110 self.assertEqual(results[0]['content'], 'Description')
106 self.assertEqual(results[0]['content'], 'Description')
111 self.assertEqual(results[0]['thumbnail'], 'https://i.ytimg.com/vi/DIVZCPfAOeM/hqdefault.jpg')
107 self.assertEqual(results[0]['thumbnail'], 'https://i.ytimg.com/vi/DIVZCPfAOeM/hqdefault.jpg')
112 self.assertTrue('DIVZCPfAOeM' in results[0]['embedded'])
108 self.assertTrue('DIVZCPfAOeM' in results[0]['embedded'])
113−
109+ self.assertEqual(results[1]['title'], 'Title')
114− html = """
110+ self.assertEqual(results[1]['url'], 'https://www.youtube.com/watch?v=9C_HReR_McQ')
115− <ol id="item-section-063864" class="item-section">
111+ self.assertEqual(results[1]['content'], 'Description')
116− <li>
112+ self.assertEqual(results[1]['thumbnail'], 'https://i.ytimg.com/vi/9C_HReR_McQ/hqdefault.jpg')
117− <div class="yt-lockup yt-lockup-tile yt-lockup-video vve-check clearfix yt-uix-tile"
113+ self.assertTrue('9C_HReR_McQ' in results[1]['embedded'])
118− data-context-item-id="DIVZCPfAOeM"
119− data-visibility-tracking="CBgQ3DAYACITCPGXnYau6sUCFZEIHAod-VQASCj0JECx_-GK5uqMpcIB">
120− <div class="yt-lockup-dismissable"><div class="yt-lockup-thumbnail contains-addto">
121− <a aria-hidden="true" href="/watch?v=DIVZCPfAOeM" class=" yt-uix-sessionlink pf-link"
122− data-sessionlink="itct=CBgQ3DAYACITCPGXnYau6sUCFZEIHAod-VQASCj0JFIEdGVzdA">
123− <div class="yt-thumb video-thumb"><img src="//i.ytimg.com/vi/DIVZCPfAOeM/mqdefault.jpg"
124− width="196" height="110"/></div><span class="video-time" aria-hidden="true">11:35</span></a>
125− <span class="thumb-menu dark-overflow-action-menu video-actions">
126− </span>
127− </div>
128− <div class="yt-lockup-content">
129− <h3 class="yt-lockup-title">
130− <span class="accessible-description" id="description-id-259079"> - Durée : 11:35.</span>
131− </h3>
132− <div class="yt-lockup-byline">de
133− <a href="/user/mheejapan" class=" yt-uix-sessionlink spf-link g-hovercard"
134− data-sessionlink="itct=CBgQ3DAYACITCPGXnYau6sUCFZEIHAod-VQASCj0JA" data-ytid="UCzEesu54Hjs0uRKmpy66qeA"
135− data-name="">MEHAY SUPERBIKE</a></div><div class="yt-lockup-meta">
136− <ul class="yt-lockup-meta-info">
137− <li>il y a 20 heures</li>
138− <li>8 424 vues</li>
139− </ul>
140− </div>
141− <div class="yt-lockup-badges">
142− <ul class="yt-badge-list ">
143− <li class="yt-badge-item" >
144− <span class="yt-badge">Nouveauté</span>
145− </li>
146− <li class="yt-badge-item" ><span class="yt-badge " >HD</span></li>
147− </ul>
148− </div>
149− <div class="yt-lockup-action-menu yt-uix-menu-container">
150− <div class="yt-uix-menu yt-uix-videoactionmenu hide-until-delayloaded"
151− data-video-id="DIVZCPfAOeM" data-menu-content-id="yt-uix-videoactionmenu-menu">
152− </div>
153− </div>
154− </div>
155− </div>
156− </div>
157− </li>
158− </ol>
159− """
160− response = mock.Mock(text=html)
161− results = youtube_noapi.response(response)
162− self.assertEqual(type(results), list)
163− self.assertEqual(len(results), 1)
164
114
165 html = """
115 html = """
166 <ol id="item-section-063864" class="item-section">
116 <ol id="item-section-063864" class="item-section">