New upstream version 2019.09.01

[youtubedl] / youtube_dl / downloader / dash.py

diff --git a/youtube_dl/downloader/dash.py b/youtube_dl/downloader/dash.py
index 8b1b17c6ec5c0b05a93a63ef446d549c76951e0e..c6d674bc6c0690e00dee3c70879a97574c3911f3 100644 (file)
--- a/youtube_dl/downloader/dash.py
+++ b/youtube_dl/downloader/dash.py
@@ -1,12 +1,10 @@
 from __future__ import unicode_literals
 
-import os
-import re
-
 from .fragment import FragmentFD
+from ..compat import compat_urllib_error
 from ..utils import (
-    sanitize_open,
-    encodeFilename,
+    DownloadError,
+    urljoin,
 )
 
 
@@ -18,42 +16,65 @@ class DashSegmentsFD(FragmentFD):
     FD_NAME = 'dashsegments'
 
     def real_download(self, filename, info_dict):
-        base_url = info_dict['url']
-        segment_urls = [info_dict['segment_urls'][0]] if self.params.get('test', False) else info_dict['segment_urls']
-        initialization_url = info_dict.get('initialization_url')
+        fragment_base_url = info_dict.get('fragment_base_url')
+        fragments = info_dict['fragments'][:1] if self.params.get(
+            'test', False) else info_dict['fragments']
 
         ctx = {
             'filename': filename,
-            'total_frags': len(segment_urls) + (1 if initialization_url else 0),
+            'total_frags': len(fragments),
         }
 
         self._prepare_and_start_frag_download(ctx)
 
-        def combine_url(base_url, target_url):
-            if re.match(r'^https?://', target_url):
-                return target_url
-            return '%s%s%s' % (base_url, '' if base_url.endswith('/') else '/', target_url)
+        fragment_retries = self.params.get('fragment_retries', 0)
+        skip_unavailable_fragments = self.params.get('skip_unavailable_fragments', True)
 
-        segments_filenames = []
+        frag_index = 0
+        for i, fragment in enumerate(fragments):
+            frag_index += 1
+            if frag_index <= ctx['fragment_index']:
+                continue
+            # In DASH, the first segment contains necessary headers to
+            # generate a valid MP4 file, so always abort for the first segment
+            fatal = i == 0 or not skip_unavailable_fragments
+            count = 0
+            while count <= fragment_retries:
+                try:
+                    fragment_url = fragment.get('url')
+                    if not fragment_url:
+                        assert fragment_base_url
+                        fragment_url = urljoin(fragment_base_url, fragment['path'])
+                    success, frag_content = self._download_fragment(ctx, fragment_url, info_dict)
+                    if not success:
+                        return False
+                    self._append_fragment(ctx, frag_content)
+                    break
+                except compat_urllib_error.HTTPError as err:
+                    # YouTube may often return 404 HTTP error for a fragment causing the
+                    # whole download to fail. However if the same fragment is immediately
+                    # retried with the same request data this usually succeeds (1-2 attempts
+                    # is usually enough) thus allowing to download the whole file successfully.
+                    # To be future-proof we will retry all fragments that fail with any
+                    # HTTP error.
+                    count += 1
+                    if count <= fragment_retries:
+                        self.report_retry_fragment(err, frag_index, count, fragment_retries)
+                except DownloadError:
+                    # Don't retry fragment if error occurred during HTTP downloading
+                    # itself since it has own retry settings
+                    if not fatal:
+                        self.report_skip_fragment(frag_index)
+                        break
+                    raise
 
-        def append_url_to_file(target_url, target_filename):
-            success = ctx['dl'].download(target_filename, {'url': combine_url(base_url, target_url)})
-            if not success:
+            if count > fragment_retries:
+                if not fatal:
+                    self.report_skip_fragment(frag_index)
+                    continue
+                self.report_error('giving up after %s fragment retries' % fragment_retries)
                 return False
-            down, target_sanitized = sanitize_open(target_filename, 'rb')
-            ctx['dest_stream'].write(down.read())
-            down.close()
-            segments_filenames.append(target_sanitized)
-
-        if initialization_url:
-            append_url_to_file(initialization_url, ctx['tmpfilename'] + '-Init')
-        for i, segment_url in enumerate(segment_urls):
-            segment_filename = '%s-Seg%d' % (ctx['tmpfilename'], i)
-            append_url_to_file(segment_url, segment_filename)
 
         self._finish_frag_download(ctx)
 
-        for segment_file in segments_filenames:
-            os.remove(encodeFilename(segment_file))
-
         return True