Ashish/yt-dlp
1
1
Fork 0
mirror of https://github.com/yt-dlp/yt-dlp.git synced 2024-06-28 03:25:27 +00:00
Code Issues 7 Projects Releases Wiki Activity

[ie/Dropbox] Fix extractor (#7926)

Browse Source 
Closes #7005, Closes #7696
Authored by: nathantouze, bashonly, denhotte
This commit is contained in:
Nathan Touzé 2023-08-28 23:33:48 +02:00 committed by GitHub
parent c2d8ee0000
commit b9f2bc2dbe
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 30 additions and 12 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

42
yt_dlp/extractor/dropbox.py
View File

@ -1,3 +1,4 @@
import base64
import os.path import os.path
import re import re
@ -5,14 +6,13 @@ from .common import InfoExtractor
from ..compat import compat_urllib_parse_unquote from ..compat import compat_urllib_parse_unquote
from ..utils import ( from ..utils import (
ExtractorError, ExtractorError,
traverse_obj, update_url_query,
try_get,
url_basename, url_basename,
) )
class DropboxIE(InfoExtractor): class DropboxIE(InfoExtractor):
_VALID_URL = r'https?://(?:www\.)?dropbox[.]com/sh?/(?P<id>[a-zA-Z0-9]{15})/.*' _VALID_URL = r'https?://(?:www\.)?dropbox\.com/(?:(?:e/)?scl/fi|sh?)/(?P<id>\w+)'
_TESTS = [ _TESTS = [
{ {
'url': 'https://www.dropbox.com/s/nelirfsxnmcfbfh/youtube-dl%20test%20video%20%27%C3%A4%22BaW_jenozKc.mp4?dl=0', 'url': 'https://www.dropbox.com/s/nelirfsxnmcfbfh/youtube-dl%20test%20video%20%27%C3%A4%22BaW_jenozKc.mp4?dl=0',
@ -22,7 +22,16 @@ class DropboxIE(InfoExtractor):
'title': 'youtube-dl test video \'ä"BaW_jenozKc' 'title': 'youtube-dl test video \'ä"BaW_jenozKc'
} }
}, { }, {
'url': 'https://www.dropbox.com/sh/662glsejgzoj9sr/AAByil3FGH9KFNZ13e08eSa1a/Pregame%20Ceremony%20Program%20PA%2020140518.m4v', 'url': 'https://www.dropbox.com/s/nelirfsxnmcfbfh',
'only_matching': True,
}, {
'url': 'https://www.dropbox.com/sh/2mgpiuq7kv8nqdf/AABy-fW4dkydT4GmWi2mdOUDa?dl=0&preview=Drone+Shot.mp4',
'only_matching': True,
}, {
'url': 'https://www.dropbox.com/scl/fi/r2kd2skcy5ylbbta5y1pz/DJI_0003.MP4?dl=0&rlkey=wcdgqangn7t3lnmmv6li9mu9h',
'only_matching': True,
}, {
'url': 'https://www.dropbox.com/e/scl/fi/r2kd2skcy5ylbbta5y1pz/DJI_0003.MP4?dl=0&rlkey=wcdgqangn7t3lnmmv6li9mu9h',
'only_matching': True, 'only_matching': True,
}, },
] ]
@ -53,16 +62,25 @@ class DropboxIE(InfoExtractor):
else: else:
raise ExtractorError('Password protected video, use --video-password <password>', expected=True) raise ExtractorError('Password protected video, use --video-password <password>', expected=True)
info_json = self._search_json(r'InitReact\.mountComponent\(.*?,', webpage, 'mountComponent', video_id, formats, subtitles, has_anonymous_download = [], {}, False
contains_pattern=r'{.+?"preview".+?}', end_pattern=r'\)')['props'] for encoded in reversed(re.findall(r'registerStreamedPrefetch\s*\(\s*"[\w/+=]+"\s*,\s*"([\w/+=]+)"', webpage)):
transcode_url = traverse_obj(info_json, ((None, 'preview'), 'file', 'preview', 'content', 'transcode_url'), get_all=False) decoded = base64.b64decode(encoded).decode('utf-8', 'ignore')
formats, subtitles = self._extract_m3u8_formats_and_subtitles(transcode_url, video_id) transcode_url = self._search_regex(
r'\n\x03(https://[^\x12\x03\n]+\.m3u8)', decoded, 'transcode url', default=None)
if not transcode_url:
continue
formats, subtitles = self._extract_m3u8_formats_and_subtitles(transcode_url, video_id)
has_anonymous_download = self._search_regex(r'(anonymous:\tanonymous)', decoded, 'anonymous', default=False)
break
# downloads enabled we can get the original file # downloads enabled we can get the original file
if 'anonymous' in (try_get(info_json, lambda x: x['sharePermission']['canDownloadRoles']) or []): if has_anonymous_download:
video_url = re.sub(r'[?&]dl=0', '', url) formats.append({
video_url += ('?' if '?' not in video_url else '&') + 'dl=1' 'url': update_url_query(url, {'dl': '1'}),
formats.append({'url': video_url, 'format_id': 'original', 'format_note': 'Original', 'quality': 1}) 'format_id': 'original',
'format_note': 'Original',
'quality': 1
})
return { return {
'id': video_id, 'id': video_id,