[extractor/iqiyi] Fix `Iq` JS regex (#5922)

Closes #5702
Authored by: bashonly
This commit is contained in:
bashonly 2023-01-02 05:50:37 +00:00 committed by GitHub
parent 13f930abc0
commit d7f9871469
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 12 additions and 9 deletions

View File

@ -527,11 +527,14 @@ class IqIE(InfoExtractor):
webpack_js_url = self._proto_relative_url(self._search_regex(
r'<script src="((?:https?)?//stc.iqiyipic.com/_next/static/chunks/webpack-\w+\.js)"', webpage, 'webpack URL'))
webpack_js = self._download_webpage(webpack_js_url, video_id, note='Downloading webpack JS', errnote='Unable to download webpack JS')
webpack_map1, webpack_map2 = [self._parse_json(js_map, video_id, transform_source=js_to_json) for js_map in self._search_regex(
r'\(({[^}]*})\[\w+\][^\)]*\)\s*\+\s*["\']\.["\']\s*\+\s*({[^}]*})\[\w+\]\+["\']\.js', webpack_js, 'JS locations', group=(1, 2))]
for module_index in reversed(list(webpack_map2.keys())):
webpack_map = self._search_json(
r'["\']\s*\+\s*', webpack_js, 'JS locations', video_id,
contains_pattern=r'{\s*(?:\d+\s*:\s*["\'][\da-f]+["\']\s*,?\s*)+}',
end_pattern=r'\[\w+\]\+["\']\.js', transform_source=js_to_json)
for module_index in reversed(webpack_map):
module_js = self._download_webpage(
f'https://stc.iqiyipic.com/_next/static/chunks/{webpack_map1.get(module_index, module_index)}.{webpack_map2[module_index]}.js',
f'https://stc.iqiyipic.com/_next/static/chunks/{module_index}.{webpack_map[module_index]}.js',
video_id, note=f'Downloading #{module_index} module JS', errnote='Unable to download module JS', fatal=False) or ''
if 'vms request' in module_js:
self.cache.store('iq', 'player_js', module_js)
@ -543,11 +546,11 @@ class IqIE(InfoExtractor):
self._extract_vms_player_js(webpage, video_id), 'signature function')
def _update_bid_tags(self, webpage, video_id):
extracted_bid_tags = self._parse_json(
self._search_regex(
r'arguments\[1\][^,]*,\s*function\s*\([^\)]*\)\s*{\s*"use strict";?\s*var \w=({.+}})\s*,\s*\w\s*=\s*{\s*getNewVd',
self._extract_vms_player_js(webpage, video_id), 'video tags', default=''),
video_id, transform_source=js_to_json, fatal=False)
extracted_bid_tags = self._search_json(
r'function\s*\([^)]*\)\s*\{\s*"use strict";?\s*var \w\s*=\s*',
self._extract_vms_player_js(webpage, video_id), 'video tags', video_id,
contains_pattern=r'{\s*\d+\s*:\s*\{\s*nbid\s*:.+}\s*}',
end_pattern=r'\s*,\s*\w\s*=\s*\{\s*getNewVd', fatal=False, transform_source=js_to_json)
if not extracted_bid_tags:
return
self._BID_TAGS = {