]>
Commit | Line | Data |
---|---|---|
d1a77684 AG |
1 | import itertools |
2 | import json | |
ac668111 | 3 | import urllib.parse |
d1a77684 AG |
4 | |
5 | from .common import InfoExtractor | |
d1a77684 | 6 | from ..utils import ( |
d1a77684 | 7 | ExtractorError, |
ac668111 | 8 | clean_html, |
d1a77684 AG |
9 | int_or_none, |
10 | str_to_int, | |
11 | url_or_none, | |
12 | ) | |
13 | ||
14 | ||
15 | class ChingariBaseIE(InfoExtractor): | |
16 | def _get_post(self, id, post_data): | |
17 | media_data = post_data['mediaLocation'] | |
18 | base_url = media_data['base'] | |
19 | author_data = post_data.get('authorData', {}) | |
20 | song_data = post_data.get('song', {}) # revist this in future for differentiating b/w 'art' and 'author' | |
21 | ||
22 | formats = [{ | |
23 | 'format_id': frmt, | |
24 | 'width': str_to_int(frmt[1:]), | |
25 | 'url': base_url + frmt_path, | |
26 | } for frmt, frmt_path in media_data.get('transcoded', {}).items()] | |
27 | ||
28 | if media_data.get('path'): | |
29 | formats.append({ | |
30 | 'format_id': 'original', | |
31 | 'format_note': 'Direct video.', | |
32 | 'url': base_url + '/apipublic' + media_data['path'], | |
33 | 'quality': 10, | |
34 | }) | |
d1a77684 AG |
35 | timestamp = str_to_int(post_data.get('created_at')) |
36 | if timestamp: | |
37 | timestamp = int_or_none(timestamp, 1000) | |
38 | ||
39 | thumbnail, uploader_url = None, None | |
40 | if media_data.get('thumbnail'): | |
41 | thumbnail = base_url + media_data.get('thumbnail') | |
42 | if author_data.get('username'): | |
43 | uploader_url = 'https://chingari.io/' + author_data.get('username') | |
44 | ||
45 | return { | |
46 | 'id': id, | |
58391068 | 47 | 'extractor_key': ChingariIE.ie_key(), |
48 | 'extractor': 'Chingari', | |
ac668111 | 49 | 'title': urllib.parse.unquote_plus(clean_html(post_data.get('caption'))), |
50 | 'description': urllib.parse.unquote_plus(clean_html(post_data.get('caption'))), | |
d1a77684 AG |
51 | 'duration': media_data.get('duration'), |
52 | 'thumbnail': url_or_none(thumbnail), | |
53 | 'like_count': post_data.get('likeCount'), | |
54 | 'view_count': post_data.get('viewsCount'), | |
55 | 'comment_count': post_data.get('commentCount'), | |
56 | 'repost_count': post_data.get('shareCount'), | |
57 | 'timestamp': timestamp, | |
58 | 'uploader_id': post_data.get('userId') or author_data.get('_id'), | |
59 | 'uploader': author_data.get('name'), | |
60 | 'uploader_url': url_or_none(uploader_url), | |
61 | 'track': song_data.get('title'), | |
62 | 'artist': song_data.get('author'), | |
63 | 'formats': formats, | |
64 | } | |
65 | ||
66 | ||
67 | class ChingariIE(ChingariBaseIE): | |
73f035e1 | 68 | _VALID_URL = r'https?://(?:www\.)?chingari\.io/share/post\?id=(?P<id>[^&/#?]+)' |
d1a77684 AG |
69 | _TESTS = [{ |
70 | 'url': 'https://chingari.io/share/post?id=612f8f4ce1dc57090e8a7beb', | |
71 | 'info_dict': { | |
72 | 'id': '612f8f4ce1dc57090e8a7beb', | |
73 | 'ext': 'mp4', | |
74 | 'title': 'Happy birthday Srila Prabhupada', | |
75 | 'description': 'md5:c7080ebfdfeb06016e638c286d6bc3fa', | |
76 | 'duration': 0, | |
77 | 'thumbnail': 'https://media.chingari.io/uploads/c41d30e2-06b6-4e3b-9b4b-edbb929cec06-1630506826911/thumbnail/198f993f-ce87-4623-82c6-cd071bd6d4f4-1630506828016.jpg', | |
78 | 'like_count': int, | |
79 | 'view_count': int, | |
80 | 'comment_count': int, | |
81 | 'repost_count': int, | |
82 | 'timestamp': 1630506828, | |
83 | 'upload_date': '20210901', | |
84 | 'uploader_id': '5f0403982c8bd344f4813f8c', | |
85 | 'uploader': 'ISKCON,Inc.', | |
86 | 'uploader_url': 'https://chingari.io/iskcon,inc', | |
87 | 'track': None, | |
88 | 'artist': None, | |
89 | }, | |
90 | 'params': {'skip_download': True} | |
91 | }] | |
92 | ||
93 | def _real_extract(self, url): | |
94 | id = self._match_id(url) | |
95 | post_json = self._download_json(f'https://api.chingari.io/post/post_details/{id}', id) | |
96 | if post_json['code'] != 200: | |
97 | raise ExtractorError(post_json['message'], expected=True) | |
98 | post_data = post_json['data'] | |
99 | return self._get_post(id, post_data) | |
100 | ||
101 | ||
102 | class ChingariUserIE(ChingariBaseIE): | |
73f035e1 | 103 | _VALID_URL = r'https?://(?:www\.)?chingari\.io/(?!share/post)(?P<id>[^/?]+)' |
d1a77684 AG |
104 | _TESTS = [{ |
105 | 'url': 'https://chingari.io/dada1023', | |
d1a77684 AG |
106 | 'info_dict': { |
107 | 'id': 'dada1023', | |
108 | }, | |
58391068 | 109 | 'params': {'playlistend': 3}, |
110 | 'playlist': [{ | |
d1a77684 AG |
111 | 'url': 'https://chingari.io/share/post?id=614781f3ade60b3a0bfff42a', |
112 | 'info_dict': { | |
113 | 'id': '614781f3ade60b3a0bfff42a', | |
114 | 'ext': 'mp4', | |
115 | 'title': '#chingaribappa ', | |
116 | 'description': 'md5:d1df21d84088770468fa63afe3b17857', | |
117 | 'duration': 7, | |
118 | 'thumbnail': 'https://media.chingari.io/uploads/346d86d4-abb2-474e-a164-ffccf2bbcb72-1632076273717/thumbnail/b0b3aac2-2b86-4dd1-909d-9ed6e57cf77c-1632076275552.jpg', | |
119 | 'like_count': int, | |
120 | 'view_count': int, | |
121 | 'comment_count': int, | |
122 | 'repost_count': int, | |
123 | 'timestamp': 1632076275, | |
124 | 'upload_date': '20210919', | |
125 | 'uploader_id': '5efc4b12cca35c3d1794c2d3', | |
126 | 'uploader': 'dada (girish) dhawale', | |
127 | 'uploader_url': 'https://chingari.io/dada1023', | |
128 | 'track': None, | |
129 | 'artist': None | |
130 | }, | |
131 | 'params': {'skip_download': True} | |
132 | }, { | |
133 | 'url': 'https://chingari.io/share/post?id=6146b132bcbf860959e12cba', | |
134 | 'info_dict': { | |
135 | 'id': '6146b132bcbf860959e12cba', | |
136 | 'ext': 'mp4', | |
137 | 'title': 'Tactor harvesting', | |
138 | 'description': 'md5:8403f12dce68828b77ecee7eb7e887b7', | |
139 | 'duration': 59.3, | |
140 | 'thumbnail': 'https://media.chingari.io/uploads/b353ca70-7a87-400d-93a6-fa561afaec86-1632022814584/thumbnail/c09302e3-2043-41b1-a2fe-77d97e5bd676-1632022834260.jpg', | |
141 | 'like_count': int, | |
142 | 'view_count': int, | |
143 | 'comment_count': int, | |
144 | 'repost_count': int, | |
145 | 'timestamp': 1632022834, | |
146 | 'upload_date': '20210919', | |
147 | 'uploader_id': '5efc4b12cca35c3d1794c2d3', | |
148 | 'uploader': 'dada (girish) dhawale', | |
149 | 'uploader_url': 'https://chingari.io/dada1023', | |
150 | 'track': None, | |
151 | 'artist': None | |
152 | }, | |
153 | 'params': {'skip_download': True} | |
154 | }, { | |
155 | 'url': 'https://chingari.io/share/post?id=6145651b74cb030a64c40b82', | |
156 | 'info_dict': { | |
157 | 'id': '6145651b74cb030a64c40b82', | |
158 | 'ext': 'mp4', | |
159 | 'title': '#odiabhajan ', | |
160 | 'description': 'md5:687ea36835b9276cf2af90f25e7654cb', | |
161 | 'duration': 56.67, | |
162 | 'thumbnail': 'https://media.chingari.io/uploads/6cbf216b-babc-4cce-87fe-ceaac8d706ac-1631937782708/thumbnail/8855754f-6669-48ce-b269-8cc0699ed6da-1631937819522.jpg', | |
163 | 'like_count': int, | |
164 | 'view_count': int, | |
165 | 'comment_count': int, | |
166 | 'repost_count': int, | |
167 | 'timestamp': 1631937819, | |
168 | 'upload_date': '20210918', | |
169 | 'uploader_id': '5efc4b12cca35c3d1794c2d3', | |
170 | 'uploader': 'dada (girish) dhawale', | |
171 | 'uploader_url': 'https://chingari.io/dada1023', | |
172 | 'track': None, | |
173 | 'artist': None | |
174 | }, | |
175 | 'params': {'skip_download': True} | |
176 | }], | |
177 | }, { | |
178 | 'url': 'https://chingari.io/iskcon%2Cinc', | |
179 | 'playlist_mincount': 1025, | |
180 | 'info_dict': { | |
181 | 'id': 'iskcon%2Cinc', | |
182 | }, | |
183 | }] | |
184 | ||
185 | def _entries(self, id): | |
186 | skip = 0 | |
187 | has_more = True | |
188 | for page in itertools.count(): | |
189 | posts = self._download_json('https://api.chingari.io/users/getPosts', id, | |
190 | data=json.dumps({'userId': id, 'ownerId': id, 'skip': skip, 'limit': 20}).encode(), | |
191 | headers={'content-type': 'application/json;charset=UTF-8'}, | |
192 | note='Downloading page %s' % page) | |
193 | for post in posts.get('data', []): | |
194 | post_data = post['post'] | |
195 | yield self._get_post(post_data['_id'], post_data) | |
196 | skip += 20 | |
197 | has_more = posts['hasMoreData'] | |
198 | if not has_more: | |
199 | break | |
200 | ||
201 | def _real_extract(self, url): | |
202 | alt_id = self._match_id(url) | |
203 | post_json = self._download_json(f'https://api.chingari.io/user/{alt_id}', alt_id) | |
204 | if post_json['code'] != 200: | |
205 | raise ExtractorError(post_json['message'], expected=True) | |
206 | id = post_json['data']['_id'] | |
207 | return self.playlist_result(self._entries(id), playlist_id=alt_id) |