[downloader/fragment] Output ad fragment count
[ytdl] / youtube_dl / downloader / fragment.py
1 from __future__ import division, unicode_literals
2
3 import os
4 import time
5 import json
6
7 from .common import FileDownloader
8 from .http import HttpFD
9 from ..utils import (
10     error_to_compat_str,
11     encodeFilename,
12     sanitize_open,
13     sanitized_Request,
14 )
15
16
17 class HttpQuietDownloader(HttpFD):
18     def to_screen(self, *args, **kargs):
19         pass
20
21
22 class FragmentFD(FileDownloader):
23     """
24     A base file downloader class for fragmented media (e.g. f4m/m3u8 manifests).
25
26     Available options:
27
28     fragment_retries:   Number of times to retry a fragment for HTTP error (DASH
29                         and hlsnative only)
30     skip_unavailable_fragments:
31                         Skip unavailable fragments (DASH and hlsnative only)
32     keep_fragments:     Keep downloaded fragments on disk after downloading is
33                         finished
34
35     For each incomplete fragment download youtube-dl keeps on disk a special
36     bookkeeping file with download state and metadata (in future such files will
37     be used for any incomplete download handled by youtube-dl). This file is
38     used to properly handle resuming, check download file consistency and detect
39     potential errors. The file has a .ytdl extension and represents a standard
40     JSON file of the following format:
41
42     extractor:
43         Dictionary of extractor related data. TBD.
44
45     downloader:
46         Dictionary of downloader related data. May contain following data:
47             current_fragment:
48                 Dictionary with current (being downloaded) fragment data:
49                 index:  0-based index of current fragment among all fragments
50             fragment_count:
51                 Total count of fragments
52
53     This feature is experimental and file format may change in future.
54     """
55
56     def report_retry_fragment(self, err, frag_index, count, retries):
57         self.to_screen(
58             '[download] Got server HTTP error: %s. Retrying fragment %d (attempt %d of %s)...'
59             % (error_to_compat_str(err), frag_index, count, self.format_retries(retries)))
60
61     def report_skip_fragment(self, frag_index):
62         self.to_screen('[download] Skipping fragment %d...' % frag_index)
63
64     def _prepare_url(self, info_dict, url):
65         headers = info_dict.get('http_headers')
66         return sanitized_Request(url, None, headers) if headers else url
67
68     def _prepare_and_start_frag_download(self, ctx):
69         self._prepare_frag_download(ctx)
70         self._start_frag_download(ctx)
71
72     @staticmethod
73     def __do_ytdl_file(ctx):
74         return not ctx['live'] and not ctx['tmpfilename'] == '-'
75
76     def _read_ytdl_file(self, ctx):
77         stream, _ = sanitize_open(self.ytdl_filename(ctx['filename']), 'r')
78         ctx['fragment_index'] = json.loads(stream.read())['downloader']['current_fragment']['index']
79         stream.close()
80
81     def _write_ytdl_file(self, ctx):
82         frag_index_stream, _ = sanitize_open(self.ytdl_filename(ctx['filename']), 'w')
83         downloader = {
84             'current_fragment': {
85                 'index': ctx['fragment_index'],
86             },
87         }
88         if ctx.get('fragment_count') is not None:
89             downloader['fragment_count'] = ctx['fragment_count']
90         frag_index_stream.write(json.dumps({'downloader': downloader}))
91         frag_index_stream.close()
92
93     def _download_fragment(self, ctx, frag_url, info_dict, headers=None):
94         fragment_filename = '%s-Frag%d' % (ctx['tmpfilename'], ctx['fragment_index'])
95         success = ctx['dl'].download(fragment_filename, {
96             'url': frag_url,
97             'http_headers': headers or info_dict.get('http_headers'),
98         })
99         if not success:
100             return False, None
101         down, frag_sanitized = sanitize_open(fragment_filename, 'rb')
102         ctx['fragment_filename_sanitized'] = frag_sanitized
103         frag_content = down.read()
104         down.close()
105         return True, frag_content
106
107     def _append_fragment(self, ctx, frag_content):
108         try:
109             ctx['dest_stream'].write(frag_content)
110         finally:
111             if self.__do_ytdl_file(ctx):
112                 self._write_ytdl_file(ctx)
113             if not self.params.get('keep_fragments', False):
114                 os.remove(ctx['fragment_filename_sanitized'])
115             del ctx['fragment_filename_sanitized']
116
117     def _prepare_frag_download(self, ctx):
118         if 'live' not in ctx:
119             ctx['live'] = False
120         if not ctx['live']:
121             total_frags_str = '%d' % ctx['total_frags']
122             ad_frags = ctx.get('ad_frags', 0)
123             if ad_frags:
124                 total_frags_str += ' (not including %d ad)' % ad_frags
125         else:
126             total_frags_str = 'unknown (live)'
127         self.to_screen(
128             '[%s] Total fragments: %s' % (self.FD_NAME, total_frags_str))
129         self.report_destination(ctx['filename'])
130         dl = HttpQuietDownloader(
131             self.ydl,
132             {
133                 'continuedl': True,
134                 'quiet': True,
135                 'noprogress': True,
136                 'ratelimit': self.params.get('ratelimit'),
137                 'retries': self.params.get('retries', 0),
138                 'nopart': self.params.get('nopart', False),
139                 'test': self.params.get('test', False),
140             }
141         )
142         tmpfilename = self.temp_name(ctx['filename'])
143         open_mode = 'wb'
144         resume_len = 0
145
146         # Establish possible resume length
147         if os.path.isfile(encodeFilename(tmpfilename)):
148             open_mode = 'ab'
149             resume_len = os.path.getsize(encodeFilename(tmpfilename))
150
151         # Should be initialized before ytdl file check
152         ctx.update({
153             'tmpfilename': tmpfilename,
154             'fragment_index': 0,
155         })
156
157         if self.__do_ytdl_file(ctx):
158             if os.path.isfile(encodeFilename(self.ytdl_filename(ctx['filename']))):
159                 self._read_ytdl_file(ctx)
160                 if ctx['fragment_index'] > 0 and resume_len == 0:
161                     self.report_error(
162                         'Inconsistent state of incomplete fragment download. '
163                         'Restarting from the beginning...')
164                     ctx['fragment_index'] = resume_len = 0
165                     self._write_ytdl_file(ctx)
166             else:
167                 self._write_ytdl_file(ctx)
168                 assert ctx['fragment_index'] == 0
169
170         dest_stream, tmpfilename = sanitize_open(tmpfilename, open_mode)
171
172         ctx.update({
173             'dl': dl,
174             'dest_stream': dest_stream,
175             'tmpfilename': tmpfilename,
176             # Total complete fragments downloaded so far in bytes
177             'complete_frags_downloaded_bytes': resume_len,
178         })
179
180     def _start_frag_download(self, ctx):
181         total_frags = ctx['total_frags']
182         # This dict stores the download progress, it's updated by the progress
183         # hook
184         state = {
185             'status': 'downloading',
186             'downloaded_bytes': ctx['complete_frags_downloaded_bytes'],
187             'fragment_index': ctx['fragment_index'],
188             'fragment_count': total_frags,
189             'filename': ctx['filename'],
190             'tmpfilename': ctx['tmpfilename'],
191         }
192
193         start = time.time()
194         ctx.update({
195             'started': start,
196             # Amount of fragment's bytes downloaded by the time of the previous
197             # frag progress hook invocation
198             'prev_frag_downloaded_bytes': 0,
199         })
200
201         def frag_progress_hook(s):
202             if s['status'] not in ('downloading', 'finished'):
203                 return
204
205             time_now = time.time()
206             state['elapsed'] = time_now - start
207             frag_total_bytes = s.get('total_bytes') or 0
208             if not ctx['live']:
209                 estimated_size = (
210                     (ctx['complete_frags_downloaded_bytes'] + frag_total_bytes) /
211                     (state['fragment_index'] + 1) * total_frags)
212                 state['total_bytes_estimate'] = estimated_size
213
214             if s['status'] == 'finished':
215                 state['fragment_index'] += 1
216                 ctx['fragment_index'] = state['fragment_index']
217                 state['downloaded_bytes'] += frag_total_bytes - ctx['prev_frag_downloaded_bytes']
218                 ctx['complete_frags_downloaded_bytes'] = state['downloaded_bytes']
219                 ctx['prev_frag_downloaded_bytes'] = 0
220             else:
221                 frag_downloaded_bytes = s['downloaded_bytes']
222                 state['downloaded_bytes'] += frag_downloaded_bytes - ctx['prev_frag_downloaded_bytes']
223                 if not ctx['live']:
224                     state['eta'] = self.calc_eta(
225                         start, time_now, estimated_size,
226                         state['downloaded_bytes'])
227                 state['speed'] = s.get('speed') or ctx.get('speed')
228                 ctx['speed'] = state['speed']
229                 ctx['prev_frag_downloaded_bytes'] = frag_downloaded_bytes
230             self._hook_progress(state)
231
232         ctx['dl'].add_progress_hook(frag_progress_hook)
233
234         return start
235
236     def _finish_frag_download(self, ctx):
237         ctx['dest_stream'].close()
238         if self.__do_ytdl_file(ctx):
239             ytdl_filename = encodeFilename(self.ytdl_filename(ctx['filename']))
240             if os.path.isfile(ytdl_filename):
241                 os.remove(ytdl_filename)
242         elapsed = time.time() - ctx['started']
243         self.try_rename(ctx['tmpfilename'], ctx['filename'])
244         fsize = os.path.getsize(encodeFilename(ctx['filename']))
245
246         self._hook_progress({
247             'downloaded_bytes': fsize,
248             'total_bytes': fsize,
249             'filename': ctx['filename'],
250             'status': 'finished',
251             'elapsed': elapsed,
252         })