aboutsummaryrefslogtreecommitdiff
path: root/yt_dlp/extractor/bpb.py
diff options
context:
space:
mode:
Diffstat (limited to 'yt_dlp/extractor/bpb.py')
-rw-r--r--yt_dlp/extractor/bpb.py65
1 files changed, 26 insertions, 39 deletions
diff --git a/yt_dlp/extractor/bpb.py b/yt_dlp/extractor/bpb.py
index 7fe089944..d7bf58b36 100644
--- a/yt_dlp/extractor/bpb.py
+++ b/yt_dlp/extractor/bpb.py
@@ -1,35 +1,20 @@
-import functools
import re
from .common import InfoExtractor
from ..utils import (
clean_html,
extract_attributes,
- get_element_text_and_html_by_tag,
- get_elements_by_class,
join_nonempty,
js_to_json,
mimetype2ext,
unified_strdate,
url_or_none,
urljoin,
- variadic,
)
-from ..utils.traversal import traverse_obj
-
-
-def html_get_element(tag=None, cls=None):
- assert tag or cls, 'One of tag or class is required'
-
- if cls:
- func = functools.partial(get_elements_by_class, cls, tag=tag)
- else:
- func = functools.partial(get_element_text_and_html_by_tag, tag)
-
- def html_get_element_wrapper(html):
- return variadic(func(html))[0]
-
- return html_get_element_wrapper
+from ..utils.traversal import (
+ find_element,
+ traverse_obj,
+)
class BpbIE(InfoExtractor):
@@ -41,12 +26,12 @@ class BpbIE(InfoExtractor):
'info_dict': {
'id': '297',
'ext': 'mp4',
- 'creator': 'Kooperative Berlin',
- 'description': 'md5:f4f75885ba009d3e2b156247a8941ce6',
- 'release_date': '20160115',
+ 'creators': ['Kooperative Berlin'],
+ 'description': r're:Joachim Gauck, .*\n\nKamera: .*',
+ 'release_date': '20150716',
'series': 'Interview auf dem Geschichtsforum 1989 | 2009',
- 'tags': ['Friedliche Revolution', 'Erinnerungskultur', 'Vergangenheitspolitik', 'DDR 1949 - 1990', 'Freiheitsrecht', 'BStU', 'Deutschland'],
- 'thumbnail': 'https://www.bpb.de/cache/images/7/297_teaser_16x9_1240.jpg?8839D',
+ 'tags': [],
+ 'thumbnail': r're:https?://www\.bpb\.de/cache/images/7/297_teaser_16x9_1240\.jpg.*',
'title': 'Joachim Gauck zu 1989 und die Erinnerung an die DDR',
'uploader': 'Bundeszentrale für politische Bildung',
},
@@ -55,11 +40,12 @@ class BpbIE(InfoExtractor):
'info_dict': {
'id': '522184',
'ext': 'mp4',
- 'creator': 'Institute for Strategic Dialogue Germany gGmbH (ISD)',
+ 'creators': ['Institute for Strategic Dialogue Germany gGmbH (ISD)'],
'description': 'md5:f83c795ff8f825a69456a9e51fc15903',
'release_date': '20230621',
- 'tags': ['Desinformation', 'Ukraine', 'Russland', 'Geflüchtete'],
- 'thumbnail': 'https://www.bpb.de/cache/images/4/522184_teaser_16x9_1240.png?EABFB',
+ 'series': 'Narrative über den Krieg Russlands gegen die Ukraine (NUK)',
+ 'tags': [],
+ 'thumbnail': r're:https://www\.bpb\.de/cache/images/4/522184_teaser_16x9_1240\.png.*',
'title': 'md5:9b01ccdbf58dbf9e5c9f6e771a803b1c',
'uploader': 'Bundeszentrale für politische Bildung',
},
@@ -68,11 +54,12 @@ class BpbIE(InfoExtractor):
'info_dict': {
'id': '518789',
'ext': 'mp4',
- 'creator': 'Institute for Strategic Dialogue Germany gGmbH (ISD)',
+ 'creators': ['Institute for Strategic Dialogue Germany gGmbH (ISD)'],
'description': 'md5:85228aed433e84ff0ff9bc582abd4ea8',
'release_date': '20230302',
- 'tags': ['Desinformation', 'Ukraine', 'Russland', 'Geflüchtete'],
- 'thumbnail': 'https://www.bpb.de/cache/images/9/518789_teaser_16x9_1240.jpeg?56D0D',
+ 'series': 'Narrative über den Krieg Russlands gegen die Ukraine (NUK)',
+ 'tags': [],
+ 'thumbnail': r're:https://www\.bpb\.de/cache/images/9/518789_teaser_16x9_1240\.jpeg.*',
'title': 'md5:3e956f264bb501f6383f10495a401da4',
'uploader': 'Bundeszentrale für politische Bildung',
},
@@ -84,12 +71,12 @@ class BpbIE(InfoExtractor):
'info_dict': {
'id': '315813',
'ext': 'mp3',
- 'creator': 'Axel Schröder',
+ 'creators': ['Axel Schröder'],
'description': 'md5:eda9d1af34e5912efef5baf54fba4427',
'release_date': '20200921',
'series': 'Auf Endlagersuche. Der deutsche Weg zu einem sicheren Atommülllager',
'tags': ['Atomenergie', 'Endlager', 'hoch-radioaktiver Abfall', 'Endlagersuche', 'Atommüll', 'Atomendlager', 'Gorleben', 'Deutschland'],
- 'thumbnail': 'https://www.bpb.de/cache/images/3/315813_teaser_16x9_1240.png?92A94',
+ 'thumbnail': r're:https://www\.bpb\.de/cache/images/3/315813_teaser_16x9_1240\.png.*',
'title': 'Folge 1: Eine Einführung',
'uploader': 'Bundeszentrale für politische Bildung',
},
@@ -98,12 +85,12 @@ class BpbIE(InfoExtractor):
'info_dict': {
'id': '517806',
'ext': 'mp3',
- 'creator': 'Bundeszentrale für politische Bildung',
+ 'creators': ['Bundeszentrale für politische Bildung'],
'description': 'md5:594689600e919912aade0b2871cc3fed',
'release_date': '20230127',
'series': 'Vorträge des Fachtags "Modernisierer. Grenzgänger. Anstifter. Sechs Jahrzehnte \'Neue Rechte\'"',
'tags': ['Rechtsextremismus', 'Konservatismus', 'Konservativismus', 'neue Rechte', 'Rechtspopulismus', 'Schnellroda', 'Deutschland'],
- 'thumbnail': 'https://www.bpb.de/cache/images/6/517806_teaser_16x9_1240.png?7A7A0',
+ 'thumbnail': r're:https://www\.bpb\.de/cache/images/6/517806_teaser_16x9_1240\.png.*',
'title': 'Die Weltanschauung der "Neuen Rechten"',
'uploader': 'Bundeszentrale für politische Bildung',
},
@@ -147,7 +134,7 @@ class BpbIE(InfoExtractor):
video_id = self._match_id(url)
webpage = self._download_webpage(url, video_id)
- title_result = traverse_obj(webpage, ({html_get_element(cls='opening-header__title')}, {self._TITLE_RE.match}))
+ title_result = traverse_obj(webpage, ({find_element(cls='opening-header__title')}, {self._TITLE_RE.match}))
json_lds = list(self._yield_json_ld(webpage, video_id, fatal=False))
return {
@@ -156,15 +143,15 @@ class BpbIE(InfoExtractor):
# This metadata could be interpreted otherwise, but it fits "series" the most
'series': traverse_obj(title_result, ('series', {str.strip})) or None,
'description': join_nonempty(*traverse_obj(webpage, [(
- {html_get_element(cls='opening-intro')},
- [{html_get_element(tag='bpb-accordion-item')}, {html_get_element(cls='text-content')}],
+ {find_element(cls='opening-intro')},
+ [{find_element(tag='bpb-accordion-item')}, {find_element(cls='text-content')}],
), {clean_html}]), delim='\n\n') or None,
- 'creator': self._html_search_meta('author', webpage),
+ 'creators': traverse_obj(self._html_search_meta('author', webpage), all),
'uploader': self._html_search_meta('publisher', webpage),
'release_date': unified_strdate(self._html_search_meta('date', webpage)),
'tags': traverse_obj(json_lds, (..., 'keywords', {lambda x: x.split(',')}, ...)),
**traverse_obj(self._parse_vue_attributes('bpb-player', webpage, video_id), {
'formats': (':sources', ..., {self._process_source}),
- 'thumbnail': ('poster', {lambda x: urljoin(url, x)}),
+ 'thumbnail': ('poster', {urljoin(url)}),
}),
}
generated by cgit v1.2.3 (git 2.26.2) at 2025-08-20 03:59:00 +0000