From 47fb5ba64791d57a3ba1fafee6bdba7394f0306f Mon Sep 17 00:00:00 2001 From: Synarp Date: Wed, 12 Jun 2024 01:46:57 +0200 Subject: [PATCH] workaround for the 1000 item API-limit --- yt_dlp/extractor/lbry.py | 34 +++++++++++++++++++++++++++++++--- 1 file changed, 31 insertions(+), 3 deletions(-) diff --git a/yt_dlp/extractor/lbry.py b/yt_dlp/extractor/lbry.py index dcb44d07f..e0ffbf479 100644 --- a/yt_dlp/extractor/lbry.py +++ b/yt_dlp/extractor/lbry.py @@ -96,6 +96,30 @@ class LBRYBaseIE(InfoExtractor): 'url': self._permanent_url(url, item['name'], item['claim_id']), } + def _metapage_entries(self, display_id, url, params): + if 'release_time' in params: + raise ExtractorError('release_time isn\'t allowed because _metapage_entires needs to specify it.') + if not ('order_by' in params and params['order_by'] == ['release_time']): + raise ExtractorError('videos must be sorted by release_time for _metapage_entries to work.') + + last_metapage = [] + metapage = OnDemandPagedList( + functools.partial(self._fetch_page, display_id, url, params), + self._PAGE_SIZE).getslice() + + while len(metapage) > 0: + yield from metapage + + next_metapage_params = { + **params, + "release_time": "<=%s" % metapage[-1]["release_timestamp"] + } + last_metapage = metapage + metapage = OnDemandPagedList( + functools.partial(self._fetch_page, display_id, url, next_metapage_params), + self._PAGE_SIZE).getslice() + metapage = [x for x in metapage if x not in last_metapage] + def _playlist_entries(self, url, display_id, claim_param, metadata): qs = parse_qs(url) content = qs.get('content', [None])[0] @@ -123,9 +147,13 @@ class LBRYBaseIE(InfoExtractor): languages.append('none') params['any_languages'] = languages - entries = OnDemandPagedList( - functools.partial(self._fetch_page, display_id, url, params), - self._PAGE_SIZE) + if qs.get('order', ['new'])[0] == 'new': + entries = self._metapage_entries(display_id, url, params) + else: + self.report_warning("Extraction is limited to 1000 Videos when not sorting by newest.") + entries = OnDemandPagedList( + functools.partial(self._fetch_page, display_id, url, params), + self._PAGE_SIZE) return self.playlist_result( entries, display_id, **traverse_obj(metadata, ('value', {