more

yt_dlp/extractor/pbs.py

@@ -1,15 +1,20 @@
+import functools
 import re
 
 from .common import InfoExtractor
 from ..compat import compat_str
 from ..utils import (
     ExtractorError,
+    OnDemandPagedList,
     determine_ext,
     int_or_none,
     float_or_none,
     js_to_json,
     clean_html,
     get_elements_html_by_class,
+    get_element_html_by_class,
+    get_element_by_id,
+extract_attributes,
     orderedSet,
     strip_jsonp,
     strip_or_none,
@@ -795,9 +800,10 @@ class PBSShowIE(InfoExtractor):
     }]
 
     _JSON_SEARCH = r'<script[^>]+id="content-strip-data" type="application/json">'
+    _SHOW_JSON_SEARCH = r'GTMDataLayer\.push\('
+    PAGE_SIZE = 40
 
     _TOKEN = 'zyG6tQcGPE5swyAEFLqKUwMuMMuF6IO2DZ6ZDQjGfsL0e4dcTLwqkTTul05Jdve7'
-    PAGE_SIZE = 25
     HTML_CLASS_NAMES = {
         'channel': {
             'container': 'channel-videos-container',
@@ -816,22 +822,48 @@ class PBSShowIE(InfoExtractor):
     def _make_url(playlist_id):
         return f'https://watch.opb.org/show/{playlist_id}'
 
+    def _fetch_season_page(self, playlist_id, page_num):
+        playlist_url = self._make_url(playlist_id)
+        season_id = f'{playlist_id}-{page_num}'
+
+        season_page = self._download_webpage(f'{playlist_url}/episodes/season/{page_num}', video_id=season_id)
+        season_data = get_elements_html_by_class("video-summary", season_page)
+
     def _real_extract(self, url):
         playlist_id = self._match_valid_url(url).group('id')
         playlist_url = self._make_url(playlist_id)
         webpage = self._download_webpage(self._make_url(playlist_id), playlist_id)
         show_data = self._search_json(self._JSON_SEARCH, webpage, 'seasons', playlist_id)
+        # show_metadata = self._search_json(self._SHOW_JSON_SEARCH, webpage, 'show metadata', playlist_id)
 
-        for show_season_metadata in sorted(show_data.get('episodes_data', {}).get('seasons', []), key=lambda x: x.get('ordinal', 0), reverse=True):
-            season_ordinal = show_season_metadata.get('ordinal', 0)
-            if season_ordinal == 0:
-                continue
+        playlist_description = clean_html(get_element_html_by_class("show-hero__description--long is-hidden", webpage))
+        show_metadata = extract_attributes(get_element_html_by_class("show-hero__my-list btn--mylist--placeholder", webpage))
 
-            season_id = f'{playlist_id}-{season_ordinal}'
+        playlist_title = show_metadata['data-gtml-label']
+        clean_html(playlist_description[0])
+
+
+
+        return self.playlist_result(
+            OnDemandPagedList(
+                pagefunc=functools.partial(self._fetch_season_page, playlist_id),
+                pagesize=self.PAGE_SIZE
+            ),
+            playlist_id=playlist_id,
+            playlist_title=playlist_title,
+            # playlist_title=
+        )
 
-            season_page = self._download_webpage(f'{playlist_url}/episodes/season/{season_ordinal}', video_id=season_id)
-            season_data = get_elements_html_by_class("video-summary", season_page)
-            pass
+        # for show_season_metadata in sorted(show_data.get('episodes_data', {}).get('seasons', []), key=lambda x: x.get('ordinal', 0), reverse=True):
+        #     season_ordinal = show_season_metadata.get('ordinal', 0)
+        #     if season_ordinal == 0:
+        #         continue
+        #
+        #     season_id = f'{playlist_id}-{season_ordinal}'
+        #
+        #     season_page = self._download_webpage(f'{playlist_url}/episodes/season/{season_ordinal}', video_id=season_id)
+        #     season_data = get_elements_html_by_class("video-summary", season_page)
+        #     pass
 
 
         return