youtube-dl

commit 0571ffda7dd12fc1067c0344f3ce4ce47b39edb0
parent 5556047465e0601d2bdee0e5a436cee64b745851
Author: Sergey M․ <dstftw@gmail.com>
Date:   Sat, 30 Apr 2016 01:43:39 +0600

[rtlnl] Improve extraction (Closes #9329)

* Make hls extraction non fatal and revert ext
* Extract progressive formats' metadata from corresponding hls formats

Diffstat:
M
youtube_dl/extractor/rtlnl.py
 | 
55
+++++++++++++++++++++++++++++++++----------------------

1 file changed, 33 insertions(+), 22 deletions(-)
diff --git a/youtube_dl/extractor/rtlnl.py b/youtube_dl/extractor/rtlnl.py
@@ -94,33 +94,44 @@ class RtlNlIE(InfoExtractor):
         videopath = material['videopath']
         m3u8_url = meta.get('videohost', 'http://manifest.us.rtl.nl') + videopath
 
-        formats = self._extract_m3u8_formats(m3u8_url, uuid)
+        formats = self._extract_m3u8_formats(
+            m3u8_url, uuid, 'mp4', m3u8_id='hls', fatal=False)
 
         video_urlpart = videopath.split('/adaptive/')[1][:-5]
         PG_URL_TEMPLATE = 'http://pg.us.rtl.nl/rtlxl/network/%s/progressive/%s.mp4'
 
-        formats.extend([
-            {
-                'url': PG_URL_TEMPLATE % ('a2t', video_urlpart),
-                'format_id': 'a2t',
-                'width': 512,
-                'height': 288,
-            },
-            {
-                'url': PG_URL_TEMPLATE % ('a3t', video_urlpart),
-                'format_id': 'a3t',
-                'width': 704,
-                'height': 400,
-                'quality': 0,
-            },
-            {
-                'url': PG_URL_TEMPLATE % ('nettv', video_urlpart),
-                'format_id': 'nettv',
-                'width': 1280,
-                'height': 720,
-                'quality': 0,
+        PG_FORMATS = (
+            ('a2t', 512, 288),
+            ('a3t', 704, 400),
+            ('nettv', 1280, 720),
+        )
+
+        def pg_format(format_id, width, height):
+            return {
+                'url': PG_URL_TEMPLATE % (format_id, video_urlpart),
+                'format_id': 'pg-%s' % format_id,
+                'protocol': 'http',
+                'width': width,
+                'height': height,
             }
-        ])
+
+        if not formats:
+            formats = [pg_format(*pg_tuple) for pg_tuple in PG_FORMATS]
+        else:
+            pg_formats = []
+            for format_id, width, height in PG_FORMATS:
+                try:
+                    # Find hls format with the same width and height corresponding
+                    # to progressive format and copy metadata from it.
+                    f = next(f for f in formats
+                             if f.get('width') == width and f.get('height') == height).copy()
+                    f.update(pg_format(format_id, width, height))
+                    pg_formats.append(f)
+                except StopIteration:
+                    # Missing hls format does mean that no progressive format with
+                    # such width and height exists either.
+                    pass
+            formats.extend(pg_formats)
         self._sort_formats(formats)
 
         thumbnails = []