]> asedeno.scripts.mit.edu Git - youtube-dl.git/blob - youtube_dl/extractor/gedidigital.py
[twitch:clips] Add access token query to download URLs (closes #29136)
[youtube-dl.git] / youtube_dl / extractor / gedidigital.py
1 # coding: utf-8
2 from __future__ import unicode_literals
3
4 import re
5
6 from .common import InfoExtractor
7 from ..utils import (
8     determine_ext,
9     int_or_none,
10 )
11
12
13 class GediDigitalIE(InfoExtractor):
14     _VALID_URL = r'''(?x)https?://video\.
15         (?:
16             (?:
17                 (?:espresso\.)?repubblica
18                 |lastampa
19                 |ilsecoloxix
20             )|
21             (?:
22                 iltirreno
23                 |messaggeroveneto
24                 |ilpiccolo
25                 |gazzettadimantova
26                 |mattinopadova
27                 |laprovinciapavese
28                 |tribunatreviso
29                 |nuovavenezia
30                 |gazzettadimodena
31                 |lanuovaferrara
32                 |corrierealpi
33                 |lasentinella
34             )\.gelocal
35         )\.it(?:/[^/]+){2,3}?/(?P<id>\d+)(?:[/?&#]|$)'''
36     _TESTS = [{
37         'url': 'https://video.lastampa.it/politica/il-paradosso-delle-regionali-la-lega-vince-ma-sembra-aver-perso/121559/121683',
38         'md5': '84658d7fb9e55a6e57ecc77b73137494',
39         'info_dict': {
40             'id': '121559',
41             'ext': 'mp4',
42             'title': 'Il paradosso delle Regionali: ecco perché la Lega vince ma sembra aver perso',
43             'description': 'md5:de7f4d6eaaaf36c153b599b10f8ce7ca',
44             'thumbnail': r're:^https://www\.repstatic\.it/video/photo/.+?-thumb-full-.+?\.jpg$',
45             'duration': 125,
46         },
47     }, {
48         'url': 'https://video.espresso.repubblica.it/embed/tutti-i-video/01-ted-villa/14772/14870&width=640&height=360',
49         'only_matching': True,
50     }, {
51         'url': 'https://video.repubblica.it/motori/record-della-pista-a-spa-francorchamps-la-pagani-huayra-roadster-bc-stupisce/367415/367963',
52         'only_matching': True,
53     }, {
54         'url': 'https://video.ilsecoloxix.it/sport/cassani-e-i-brividi-azzurri-ai-mondiali-di-imola-qui-mi-sono-innamorato-del-ciclismo-da-ragazzino-incredibile-tornarci-da-ct/66184/66267',
55         'only_matching': True,
56     }, {
57         'url': 'https://video.iltirreno.gelocal.it/sport/dentro-la-notizia-ferrari-cosa-succede-a-maranello/141059/142723',
58         'only_matching': True,
59     }, {
60         'url': 'https://video.messaggeroveneto.gelocal.it/locale/maria-giovanna-elmi-covid-vaccino/138155/139268',
61         'only_matching': True,
62     }, {
63         'url': 'https://video.ilpiccolo.gelocal.it/dossier/big-john/dinosauro-big-john-al-via-le-visite-guidate-a-trieste/135226/135751',
64         'only_matching': True,
65     }, {
66         'url': 'https://video.gazzettadimantova.gelocal.it/locale/dal-ponte-visconteo-di-valeggio-l-and-8217sos-dei-ristoratori-aprire-anche-a-cena/137310/137818',
67         'only_matching': True,
68     }, {
69         'url': 'https://video.mattinopadova.gelocal.it/dossier/coronavirus-in-veneto/covid-a-vo-un-anno-dopo-un-cuore-tricolore-per-non-dimenticare/138402/138964',
70         'only_matching': True,
71     }, {
72         'url': 'https://video.laprovinciapavese.gelocal.it/locale/mede-zona-rossa-via-alle-vaccinazioni-per-gli-over-80/137545/138120',
73         'only_matching': True,
74     }, {
75         'url': 'https://video.tribunatreviso.gelocal.it/dossier/coronavirus-in-veneto/ecco-le-prima-vaccinazioni-di-massa-nella-marca/134485/135024',
76         'only_matching': True,
77     }, {
78         'url': 'https://video.nuovavenezia.gelocal.it/locale/camion-troppo-alto-per-il-ponte-ferroviario-perde-il-carico/135734/136266',
79         'only_matching': True,
80     }, {
81         'url': 'https://video.gazzettadimodena.gelocal.it/locale/modena-scoperta-la-proteina-che-predice-il-livello-di-gravita-del-covid/139109/139796',
82         'only_matching': True,
83     }, {
84         'url': 'https://video.lanuovaferrara.gelocal.it/locale/due-bombole-di-gpl-aperte-e-abbandonate-i-vigili-bruciano-il-gas/134391/134957',
85         'only_matching': True,
86     }, {
87         'url': 'https://video.corrierealpi.gelocal.it/dossier/cortina-2021-i-mondiali-di-sci-alpino/mondiali-di-sci-il-timelapse-sulla-splendida-olympia/133760/134331',
88         'only_matching': True,
89     }, {
90         'url': 'https://video.lasentinella.gelocal.it/locale/vestigne-centra-un-auto-e-si-ribalta/138931/139466',
91         'only_matching': True,
92     }, {
93         'url': 'https://video.espresso.repubblica.it/tutti-i-video/01-ted-villa/14772',
94         'only_matching': True,
95     }]
96
97     def _real_extract(self, url):
98         video_id = self._match_id(url)
99
100         webpage = self._download_webpage(url, video_id)
101         title = self._html_search_meta(
102             ['twitter:title', 'og:title'], webpage, fatal=True)
103         player_data = re.findall(
104             r"PlayerFactory\.setParam\('(?P<type>format|param)',\s*'(?P<name>[^']+)',\s*'(?P<val>[^']+)'\);",
105             webpage)
106
107         formats = []
108         duration = thumb = None
109         for t, n, v in player_data:
110             if t == 'format':
111                 if n in ('video-hds-vod-ec', 'video-hls-vod-ec', 'video-viralize', 'video-youtube-pfp'):
112                     continue
113                 elif n.endswith('-vod-ak'):
114                     formats.extend(self._extract_akamai_formats(
115                         v, video_id, {'http': 'media.gedidigital.it'}))
116                 else:
117                     ext = determine_ext(v)
118                     if ext == 'm3u8':
119                         formats.extend(self._extract_m3u8_formats(
120                             v, video_id, 'mp4', 'm3u8_native', m3u8_id=n, fatal=False))
121                         continue
122                     f = {
123                         'format_id': n,
124                         'url': v,
125                     }
126                     if ext == 'mp3':
127                         abr = int_or_none(self._search_regex(
128                             r'-mp3-audio-(\d+)', v, 'abr', default=None))
129                         f.update({
130                             'abr': abr,
131                             'tbr': abr,
132                             'vcodec': 'none'
133                         })
134                     else:
135                         mobj = re.match(r'^video-rrtv-(\d+)(?:-(\d+))?$', n)
136                         if mobj:
137                             f.update({
138                                 'height': int(mobj.group(1)),
139                                 'vbr': int_or_none(mobj.group(2)),
140                             })
141                         if not f.get('vbr'):
142                             f['vbr'] = int_or_none(self._search_regex(
143                                 r'-video-rrtv-(\d+)', v, 'abr', default=None))
144                     formats.append(f)
145             elif t == 'param':
146                 if n in ['image_full', 'image']:
147                     thumb = v
148                 elif n == 'videoDuration':
149                     duration = int_or_none(v)
150
151         self._sort_formats(formats)
152
153         return {
154             'id': video_id,
155             'title': title,
156             'description': self._html_search_meta(
157                 ['twitter:description', 'og:description', 'description'], webpage),
158             'thumbnail': thumb or self._og_search_thumbnail(webpage),
159             'formats': formats,
160             'duration': duration,
161         }