[pbs] Revert prefer portalplayer

[youtube-dl] / youtube_dl / extractor / pbs.py

diff --git a/youtube_dl/extractor/pbs.py b/youtube_dl/extractor/pbs.py
index 744e4a09aa6facf850aeda790bebf687c90a14a4..0d4d709de969dc30fc9e349c46b544e189f90390 100644 (file)
--- a/youtube_dl/extractor/pbs.py
+++ b/youtube_dl/extractor/pbs.py
@@ -4,10 +4,12 @@ from __future__ import unicode_literals
 import re
 
 from .common import InfoExtractor
+from ..compat import compat_HTTPError
 from ..utils import (
     ExtractorError,
     determine_ext,
     int_or_none,
+    js_to_json,
     strip_jsonp,
     unified_strdate,
     US_RATINGS,
@@ -16,7 +18,7 @@ from ..utils import (
 
 class PBSIE(InfoExtractor):
     _STATIONS = (
-        (r'(?:video|www)\.pbs\.org', 'PBS: Public Broadcasting Service'),  # http://www.pbs.org/
+        (r'(?:video|www|player)\.pbs\.org', 'PBS: Public Broadcasting Service'),  # http://www.pbs.org/
         (r'video\.aptv\.org', 'APT - Alabama Public Television (WBIQ)'),  # http://aptv.org/
         (r'video\.gpb\.org', 'GPB/Georgia Public Broadcasting (WGTV)'),  # http://www.gpb.org/
         (r'video\.mpbonline\.org', 'Mississippi Public Broadcasting (WMPN)'),  # http://www.mpbonline.org
@@ -432,9 +434,21 @@ class PBSIE(InfoExtractor):
                 for vid_id in video_id]
             return self.playlist_result(entries, display_id)
 
-        info = self._download_json(
-            'http://player.pbs.org/videoInfo/%s?format=json&type=partner' % video_id,
-            display_id)
+        try:
+            info = self._download_json(
+                'http://player.pbs.org/videoInfo/%s?format=json&type=partner' % video_id,
+                display_id, 'Downloading video info JSON')
+        except ExtractorError as e:
+            if not isinstance(e.cause, compat_HTTPError) or e.cause.code != 404:
+                raise
+            # videoInfo API may not work for some videos, fallback to portalplayer API
+            player = self._download_webpage(
+                'http://player.pbs.org/portalplayer/%s' % video_id, display_id)
+            info = self._parse_json(
+                self._search_regex(
+                    r'(?s)PBS\.videoData\s*=\s*({.+?});\n',
+                    player, 'video data', default='{}'),
+                display_id, transform_source=js_to_json, fatal=False)
 
         formats = []
         for encoding_name in ('recommended_encoding', 'alternate_encoding'):