[YoutubeDL] Use a progress hook for progress reporting
[youtube-dl] / youtube_dl / downloader / f4m.py
index c460c167a2db78be7a1cdb8e81a3efe89e677ed1..5a1f8e680054a5db2967327f1abd9340e2e8a3f0 100644 (file)
@@ -1,4 +1,4 @@
-from __future__ import unicode_literals
+from __future__ import division, unicode_literals
 
 import base64
 import io
@@ -177,13 +177,12 @@ def build_fragments_list(boot_info):
     """ Return a list of (segment, fragment) for each fragment in the video """
     res = []
     segment_run_table = boot_info['segments'][0]
-    # I've only found videos with one segment
-    segment_run_entry = segment_run_table['segment_run'][0]
-    n_frags = segment_run_entry[1]
     fragment_run_entry_table = boot_info['fragments'][0]['fragments']
     first_frag_number = fragment_run_entry_table[0]['first']
-    for (i, frag_number) in zip(range(1, n_frags + 1), itertools.count(first_frag_number)):
-        res.append((1, frag_number))
+    fragments_counter = itertools.count(first_frag_number)
+    for segment, fragments_count in segment_run_table['segment_run']:
+        for _ in range(fragments_count):
+            res.append((segment, next(fragments_counter)))
     return res
 
 
@@ -231,25 +230,32 @@ class F4mFD(FileDownloader):
     A downloader for f4m manifests or AdobeHDS.
     """
 
+    def _get_unencrypted_media(self, doc):
+        media = doc.findall(_add_ns('media'))
+        if not media:
+            self.report_error('No media found')
+        for e in (doc.findall(_add_ns('drmAdditionalHeader')) +
+                  doc.findall(_add_ns('drmAdditionalHeaderSet'))):
+            # If id attribute is missing it's valid for all media nodes
+            # without drmAdditionalHeaderId or drmAdditionalHeaderSetId attribute
+            if 'id' not in e.attrib:
+                self.report_error('Missing ID in f4m DRM')
+        media = list(filter(lambda e: 'drmAdditionalHeaderId' not in e.attrib and
+                                      'drmAdditionalHeaderSetId' not in e.attrib,
+                            media))
+        if not media:
+            self.report_error('Unsupported DRM')
+        return media
+
     def real_download(self, filename, info_dict):
         man_url = info_dict['url']
         requested_bitrate = info_dict.get('tbr')
         self.to_screen('[download] Downloading f4m manifest')
         manifest = self.ydl.urlopen(man_url).read()
-        self.report_destination(filename)
-        http_dl = HttpQuietDownloader(
-            self.ydl,
-            {
-                'continuedl': True,
-                'quiet': True,
-                'noprogress': True,
-                'ratelimit': self.params.get('ratelimit', None),
-                'test': self.params.get('test', False),
-            }
-        )
 
         doc = etree.fromstring(manifest)
-        formats = [(int(f.attrib.get('bitrate', -1)), f) for f in doc.findall(_add_ns('media'))]
+        formats = [(int(f.attrib.get('bitrate', -1)), f)
+                   for f in self._get_unencrypted_media(doc)]
         if requested_bitrate is None:
             # get the best format
             formats = sorted(formats, key=lambda f: f[0])
@@ -281,39 +287,67 @@ class F4mFD(FileDownloader):
         # For some akamai manifests we'll need to add a query to the fragment url
         akamai_pv = xpath_text(doc, _add_ns('pv-2.0'))
 
+        self.report_destination(filename)
+        http_dl = HttpQuietDownloader(
+            self.ydl,
+            {
+                'continuedl': True,
+                'quiet': True,
+                'noprogress': True,
+                'ratelimit': self.params.get('ratelimit', None),
+                'test': self.params.get('test', False),
+            }
+        )
         tmpfilename = self.temp_name(filename)
         (dest_stream, tmpfilename) = sanitize_open(tmpfilename, 'wb')
+
         write_flv_header(dest_stream)
         write_metadata_tag(dest_stream, metadata)
 
         # This dict stores the download progress, it's updated by the progress
         # hook
         state = {
+            'status': 'downloading',
             'downloaded_bytes': 0,
-            'frag_counter': 0,
+            'frag_index': 0,
+            'frag_count': total_frags,
+            'filename': filename,
+            'tmpfilename': tmpfilename,
         }
         start = time.time()
 
-        def frag_progress_hook(status):
-            frag_total_bytes = status.get('total_bytes', 0)
-            estimated_size = (state['downloaded_bytes'] +
-                              (total_frags - state['frag_counter']) * frag_total_bytes)
-            if status['status'] == 'finished':
+        def frag_progress_hook(s):
+            if s['status'] not in ('downloading', 'finished'):
+                return
+
+            frag_total_bytes = s.get('total_bytes', 0)
+            if s['status'] == 'finished':
                 state['downloaded_bytes'] += frag_total_bytes
-                state['frag_counter'] += 1
-                progress = self.calc_percent(state['frag_counter'], total_frags)
+                state['frag_index'] += 1
+
+            estimated_size = (
+                (state['downloaded_bytes'] + frag_total_bytes)
+                / (state['frag_index'] + 1) * total_frags)
+            time_now = time.time()
+            state['total_bytes_estimate'] = estimated_size
+            state['elapsed'] = time_now - start
+
+            if s['status'] == 'finished':
+                progress = self.calc_percent(state['frag_index'], total_frags)
                 byte_counter = state['downloaded_bytes']
             else:
-                frag_downloaded_bytes = status['downloaded_bytes']
+                frag_downloaded_bytes = s['downloaded_bytes']
                 byte_counter = state['downloaded_bytes'] + frag_downloaded_bytes
                 frag_progress = self.calc_percent(frag_downloaded_bytes,
                                                   frag_total_bytes)
-                progress = self.calc_percent(state['frag_counter'], total_frags)
+                progress = self.calc_percent(state['frag_index'], total_frags)
                 progress += frag_progress / float(total_frags)
 
-            eta = self.calc_eta(start, time.time(), estimated_size, byte_counter)
-            self.report_progress(progress, format_bytes(estimated_size),
-                                 status.get('speed'), eta)
+                state['eta'] = self.calc_eta(
+                    start, time_now, estimated_size, state['downloaded_bytes'] + frag_downloaded_bytes)
+                state['speed'] = s.get('speed')
+            self._hook_progress(state)
+
         http_dl.add_progress_hook(frag_progress_hook)
 
         frags_filenames = []
@@ -337,8 +371,8 @@ class F4mFD(FileDownloader):
             frags_filenames.append(frag_filename)
 
         dest_stream.close()
-        self.report_finish(format_bytes(state['downloaded_bytes']), time.time() - start)
 
+        elapsed = time.time() - start
         self.try_rename(tmpfilename, filename)
         for frag_file in frags_filenames:
             os.remove(frag_file)
@@ -349,6 +383,7 @@ class F4mFD(FileDownloader):
             'total_bytes': fsize,
             'filename': filename,
             'status': 'finished',
+            'elapsed': elapsed,
         })
 
         return True