]>
Commit | Line | Data |
---|---|---|
1d485a1a | 1 | import concurrent.futures |
19a03940 | 2 | import contextlib |
adbc4ec4 THD |
3 | import http.client |
4 | import json | |
5 | import math | |
95d8f7ea | 6 | import os |
ac668111 | 7 | import struct |
95d8f7ea | 8 | import time |
ac668111 | 9 | import urllib.error |
95d8f7ea S |
10 | |
11 | from .common import FileDownloader | |
12 | from .http import HttpFD | |
1d3586d0 | 13 | from ..aes import aes_cbc_decrypt_bytes, unpad_pkcs7 |
ac668111 | 14 | from ..compat import compat_os_name |
95d8f7ea | 15 | from ..utils import ( |
4c7853de | 16 | DownloadError, |
95d8f7ea | 17 | encodeFilename, |
f8271158 | 18 | error_to_compat_str, |
69035555 | 19 | sanitized_Request, |
a539f065 | 20 | traverse_obj, |
95d8f7ea S |
21 | ) |
22 | ||
23 | ||
24 | class HttpQuietDownloader(HttpFD): | |
25 | def to_screen(self, *args, **kargs): | |
26 | pass | |
27 | ||
998a3cae | 28 | to_console_title = to_screen |
1d485a1a | 29 | |
95d8f7ea S |
30 | |
31 | class FragmentFD(FileDownloader): | |
32 | """ | |
33 | A base file downloader class for fragmented media (e.g. f4m/m3u8 manifests). | |
16a8b798 S |
34 | |
35 | Available options: | |
36 | ||
9603b660 S |
37 | fragment_retries: Number of times to retry a fragment for HTTP error (DASH |
38 | and hlsnative only) | |
39 | skip_unavailable_fragments: | |
40 | Skip unavailable fragments (DASH and hlsnative only) | |
0eee52f3 S |
41 | keep_fragments: Keep downloaded fragments on disk after downloading is |
42 | finished | |
59a7a13e | 43 | concurrent_fragment_downloads: The number of threads to use for native hls and dash downloads |
e8e73840 | 44 | _no_ytdl_file: Don't use .ytdl file |
290f64db | 45 | |
7a5c1cfe | 46 | For each incomplete fragment download yt-dlp keeps on disk a special |
290f64db | 47 | bookkeeping file with download state and metadata (in future such files will |
7a5c1cfe | 48 | be used for any incomplete download handled by yt-dlp). This file is |
290f64db S |
49 | used to properly handle resuming, check download file consistency and detect |
50 | potential errors. The file has a .ytdl extension and represents a standard | |
51 | JSON file of the following format: | |
52 | ||
53 | extractor: | |
54 | Dictionary of extractor related data. TBD. | |
55 | ||
56 | downloader: | |
57 | Dictionary of downloader related data. May contain following data: | |
58 | current_fragment: | |
59 | Dictionary with current (being downloaded) fragment data: | |
85f6de25 | 60 | index: 0-based index of current fragment among all fragments |
290f64db S |
61 | fragment_count: |
62 | Total count of fragments | |
50534b71 | 63 | |
85f6de25 | 64 | This feature is experimental and file format may change in future. |
95d8f7ea S |
65 | """ |
66 | ||
75a24854 | 67 | def report_retry_fragment(self, err, frag_index, count, retries): |
721f26b8 | 68 | self.to_screen( |
4c7853de | 69 | '\r[download] Got server HTTP error: %s. Retrying fragment %d (attempt %d of %s) ...' |
75a24854 | 70 | % (error_to_compat_str(err), frag_index, count, self.format_retries(retries))) |
23326151 | 71 | self.sleep_retry('fragment', count) |
721f26b8 | 72 | |
b4b855eb | 73 | def report_skip_fragment(self, frag_index, err=None): |
74 | err = f' {err};' if err else '' | |
75 | self.to_screen(f'[download]{err} Skipping fragment {frag_index:d} ...') | |
9603b660 | 76 | |
69035555 S |
77 | def _prepare_url(self, info_dict, url): |
78 | headers = info_dict.get('http_headers') | |
79 | return sanitized_Request(url, None, headers) if headers else url | |
80 | ||
3ba7740d | 81 | def _prepare_and_start_frag_download(self, ctx, info_dict): |
95d8f7ea | 82 | self._prepare_frag_download(ctx) |
3ba7740d | 83 | self._start_frag_download(ctx, info_dict) |
95d8f7ea | 84 | |
e8e73840 | 85 | def __do_ytdl_file(self, ctx): |
adbc4ec4 | 86 | return ctx['live'] is not True and ctx['tmpfilename'] != '-' and not self.params.get('_no_ytdl_file') |
adb4b03c | 87 | |
d3f0687c | 88 | def _read_ytdl_file(self, ctx): |
500a86a5 | 89 | assert 'ytdl_corrupt' not in ctx |
205a0654 | 90 | stream, _ = self.sanitize_open(self.ytdl_filename(ctx['filename']), 'r') |
500a86a5 | 91 | try: |
4d49884c F |
92 | ytdl_data = json.loads(stream.read()) |
93 | ctx['fragment_index'] = ytdl_data['downloader']['current_fragment']['index'] | |
94 | if 'extra_state' in ytdl_data['downloader']: | |
95 | ctx['extra_state'] = ytdl_data['downloader']['extra_state'] | |
500a86a5 S |
96 | except Exception: |
97 | ctx['ytdl_corrupt'] = True | |
98 | finally: | |
99 | stream.close() | |
d3f0687c S |
100 | |
101 | def _write_ytdl_file(self, ctx): | |
205a0654 | 102 | frag_index_stream, _ = self.sanitize_open(self.ytdl_filename(ctx['filename']), 'w') |
ad3dc496 | 103 | try: |
104 | downloader = { | |
105 | 'current_fragment': { | |
106 | 'index': ctx['fragment_index'], | |
107 | }, | |
108 | } | |
109 | if 'extra_state' in ctx: | |
110 | downloader['extra_state'] = ctx['extra_state'] | |
111 | if ctx.get('fragment_count') is not None: | |
112 | downloader['fragment_count'] = ctx['fragment_count'] | |
113 | frag_index_stream.write(json.dumps({'downloader': downloader})) | |
114 | finally: | |
115 | frag_index_stream.close() | |
d3f0687c | 116 | |
273762c8 | 117 | def _download_fragment(self, ctx, frag_url, info_dict, headers=None, request_data=None): |
d3f0687c | 118 | fragment_filename = '%s-Frag%d' % (ctx['tmpfilename'], ctx['fragment_index']) |
38d70284 | 119 | fragment_info_dict = { |
75a24854 RA |
120 | 'url': frag_url, |
121 | 'http_headers': headers or info_dict.get('http_headers'), | |
273762c8 | 122 | 'request_data': request_data, |
bd50a52b | 123 | 'ctx_id': ctx.get('ctx_id'), |
38d70284 | 124 | } |
3fe75fdc | 125 | success, _ = ctx['dl'].download(fragment_filename, fragment_info_dict) |
75a24854 | 126 | if not success: |
d71fd412 | 127 | return False |
38d70284 | 128 | if fragment_info_dict.get('filetime'): |
129 | ctx['fragment_filetime'] = fragment_info_dict.get('filetime') | |
4c7853de | 130 | ctx['fragment_filename_sanitized'] = fragment_filename |
d71fd412 | 131 | return True |
4c7853de | 132 | |
133 | def _read_fragment(self, ctx): | |
e06bd880 | 134 | if not ctx.get('fragment_filename_sanitized'): |
135 | return None | |
d71fd412 LNO |
136 | try: |
137 | down, frag_sanitized = self.sanitize_open(ctx['fragment_filename_sanitized'], 'rb') | |
138 | except FileNotFoundError: | |
139 | if ctx.get('live'): | |
140 | return None | |
141 | raise | |
d3f0687c S |
142 | ctx['fragment_filename_sanitized'] = frag_sanitized |
143 | frag_content = down.read() | |
75a24854 | 144 | down.close() |
4c7853de | 145 | return frag_content |
75a24854 RA |
146 | |
147 | def _append_fragment(self, ctx, frag_content): | |
d3f0687c S |
148 | try: |
149 | ctx['dest_stream'].write(frag_content) | |
593f2f79 | 150 | ctx['dest_stream'].flush() |
d3f0687c | 151 | finally: |
adb4b03c | 152 | if self.__do_ytdl_file(ctx): |
d3f0687c | 153 | self._write_ytdl_file(ctx) |
0eee52f3 | 154 | if not self.params.get('keep_fragments', False): |
45806d44 | 155 | self.try_remove(encodeFilename(ctx['fragment_filename_sanitized'])) |
d3f0687c | 156 | del ctx['fragment_filename_sanitized'] |
75a24854 | 157 | |
95d8f7ea | 158 | def _prepare_frag_download(self, ctx): |
5fa1702c S |
159 | if 'live' not in ctx: |
160 | ctx['live'] = False | |
5efaf43c S |
161 | if not ctx['live']: |
162 | total_frags_str = '%d' % ctx['total_frags'] | |
163 | ad_frags = ctx.get('ad_frags', 0) | |
164 | if ad_frags: | |
165 | total_frags_str += ' (not including %d ad)' % ad_frags | |
166 | else: | |
167 | total_frags_str = 'unknown (live)' | |
86e5f3ed | 168 | self.to_screen(f'[{self.FD_NAME}] Total fragments: {total_frags_str}') |
95d8f7ea | 169 | self.report_destination(ctx['filename']) |
666c36d5 | 170 | dl = HttpQuietDownloader(self.ydl, { |
171 | **self.params, | |
172 | 'noprogress': True, | |
173 | 'test': False, | |
174 | }) | |
95d8f7ea | 175 | tmpfilename = self.temp_name(ctx['filename']) |
75a24854 RA |
176 | open_mode = 'wb' |
177 | resume_len = 0 | |
d3f0687c | 178 | |
75a24854 RA |
179 | # Establish possible resume length |
180 | if os.path.isfile(encodeFilename(tmpfilename)): | |
181 | open_mode = 'ab' | |
182 | resume_len = os.path.getsize(encodeFilename(tmpfilename)) | |
d3f0687c | 183 | |
adb4b03c S |
184 | # Should be initialized before ytdl file check |
185 | ctx.update({ | |
186 | 'tmpfilename': tmpfilename, | |
187 | 'fragment_index': 0, | |
188 | }) | |
d3f0687c | 189 | |
adb4b03c S |
190 | if self.__do_ytdl_file(ctx): |
191 | if os.path.isfile(encodeFilename(self.ytdl_filename(ctx['filename']))): | |
192 | self._read_ytdl_file(ctx) | |
500a86a5 S |
193 | is_corrupt = ctx.get('ytdl_corrupt') is True |
194 | is_inconsistent = ctx['fragment_index'] > 0 and resume_len == 0 | |
195 | if is_corrupt or is_inconsistent: | |
196 | message = ( | |
197 | '.ytdl file is corrupt' if is_corrupt else | |
198 | 'Inconsistent state of incomplete fragment download') | |
6f3b4a98 | 199 | self.report_warning( |
5ef7d9bd | 200 | '%s. Restarting from the beginning ...' % message) |
e7c3e334 | 201 | ctx['fragment_index'] = resume_len = 0 |
500a86a5 S |
202 | if 'ytdl_corrupt' in ctx: |
203 | del ctx['ytdl_corrupt'] | |
e7c3e334 | 204 | self._write_ytdl_file(ctx) |
adb4b03c S |
205 | else: |
206 | self._write_ytdl_file(ctx) | |
e7c3e334 | 207 | assert ctx['fragment_index'] == 0 |
d3f0687c | 208 | |
205a0654 | 209 | dest_stream, tmpfilename = self.sanitize_open(tmpfilename, open_mode) |
75a24854 | 210 | |
95d8f7ea S |
211 | ctx.update({ |
212 | 'dl': dl, | |
213 | 'dest_stream': dest_stream, | |
214 | 'tmpfilename': tmpfilename, | |
75a24854 RA |
215 | # Total complete fragments downloaded so far in bytes |
216 | 'complete_frags_downloaded_bytes': resume_len, | |
95d8f7ea S |
217 | }) |
218 | ||
3ba7740d | 219 | def _start_frag_download(self, ctx, info_dict): |
3bce4ff7 | 220 | resume_len = ctx['complete_frags_downloaded_bytes'] |
95d8f7ea | 221 | total_frags = ctx['total_frags'] |
bd50a52b | 222 | ctx_id = ctx.get('ctx_id') |
95d8f7ea S |
223 | # This dict stores the download progress, it's updated by the progress |
224 | # hook | |
225 | state = { | |
226 | 'status': 'downloading', | |
3bce4ff7 | 227 | 'downloaded_bytes': resume_len, |
3e0304fe RA |
228 | 'fragment_index': ctx['fragment_index'], |
229 | 'fragment_count': total_frags, | |
95d8f7ea S |
230 | 'filename': ctx['filename'], |
231 | 'tmpfilename': ctx['tmpfilename'], | |
b83b782d S |
232 | } |
233 | ||
234 | start = time.time() | |
235 | ctx.update({ | |
236 | 'started': start, | |
bd93fd5d | 237 | 'fragment_started': start, |
709185a2 S |
238 | # Amount of fragment's bytes downloaded by the time of the previous |
239 | # frag progress hook invocation | |
b83b782d S |
240 | 'prev_frag_downloaded_bytes': 0, |
241 | }) | |
95d8f7ea S |
242 | |
243 | def frag_progress_hook(s): | |
244 | if s['status'] not in ('downloading', 'finished'): | |
245 | return | |
246 | ||
36195c44 M |
247 | if not total_frags and ctx.get('fragment_count'): |
248 | state['fragment_count'] = ctx['fragment_count'] | |
249 | ||
bd50a52b THD |
250 | if ctx_id is not None and s.get('ctx_id') != ctx_id: |
251 | return | |
252 | ||
253 | state['max_progress'] = ctx.get('max_progress') | |
254 | state['progress_idx'] = ctx.get('progress_idx') | |
255 | ||
5fa1702c | 256 | time_now = time.time() |
2c2f1efd | 257 | state['elapsed'] = time_now - start |
3c91e416 | 258 | frag_total_bytes = s.get('total_bytes') or 0 |
3ba7740d | 259 | s['fragment_info_dict'] = s.pop('info_dict', {}) |
5fa1702c S |
260 | if not ctx['live']: |
261 | estimated_size = ( | |
3089bc74 S |
262 | (ctx['complete_frags_downloaded_bytes'] + frag_total_bytes) |
263 | / (state['fragment_index'] + 1) * total_frags) | |
5fa1702c | 264 | state['total_bytes_estimate'] = estimated_size |
95d8f7ea | 265 | |
709185a2 | 266 | if s['status'] == 'finished': |
3e0304fe RA |
267 | state['fragment_index'] += 1 |
268 | ctx['fragment_index'] = state['fragment_index'] | |
b83b782d S |
269 | state['downloaded_bytes'] += frag_total_bytes - ctx['prev_frag_downloaded_bytes'] |
270 | ctx['complete_frags_downloaded_bytes'] = state['downloaded_bytes'] | |
bd93fd5d | 271 | ctx['speed'] = state['speed'] = self.calc_speed( |
272 | ctx['fragment_started'], time_now, frag_total_bytes) | |
273 | ctx['fragment_started'] = time.time() | |
b83b782d | 274 | ctx['prev_frag_downloaded_bytes'] = 0 |
709185a2 S |
275 | else: |
276 | frag_downloaded_bytes = s['downloaded_bytes'] | |
b83b782d | 277 | state['downloaded_bytes'] += frag_downloaded_bytes - ctx['prev_frag_downloaded_bytes'] |
5fa1702c S |
278 | if not ctx['live']: |
279 | state['eta'] = self.calc_eta( | |
3bce4ff7 | 280 | start, time_now, estimated_size - resume_len, |
281 | state['downloaded_bytes'] - resume_len) | |
bd93fd5d | 282 | ctx['speed'] = state['speed'] = self.calc_speed( |
283 | ctx['fragment_started'], time_now, frag_downloaded_bytes) | |
b83b782d | 284 | ctx['prev_frag_downloaded_bytes'] = frag_downloaded_bytes |
3ba7740d | 285 | self._hook_progress(state, info_dict) |
95d8f7ea S |
286 | |
287 | ctx['dl'].add_progress_hook(frag_progress_hook) | |
288 | ||
289 | return start | |
290 | ||
3ba7740d | 291 | def _finish_frag_download(self, ctx, info_dict): |
95d8f7ea | 292 | ctx['dest_stream'].close() |
adb4b03c S |
293 | if self.__do_ytdl_file(ctx): |
294 | ytdl_filename = encodeFilename(self.ytdl_filename(ctx['filename'])) | |
295 | if os.path.isfile(ytdl_filename): | |
45806d44 | 296 | self.try_remove(ytdl_filename) |
95d8f7ea | 297 | elapsed = time.time() - ctx['started'] |
0ff2c1ec S |
298 | |
299 | if ctx['tmpfilename'] == '-': | |
300 | downloaded_bytes = ctx['complete_frags_downloaded_bytes'] | |
301 | else: | |
302 | self.try_rename(ctx['tmpfilename'], ctx['filename']) | |
38d70284 | 303 | if self.params.get('updatetime', True): |
304 | filetime = ctx.get('fragment_filetime') | |
305 | if filetime: | |
19a03940 | 306 | with contextlib.suppress(Exception): |
38d70284 | 307 | os.utime(ctx['filename'], (time.time(), filetime)) |
0ff2c1ec | 308 | downloaded_bytes = os.path.getsize(encodeFilename(ctx['filename'])) |
95d8f7ea S |
309 | |
310 | self._hook_progress({ | |
0ff2c1ec S |
311 | 'downloaded_bytes': downloaded_bytes, |
312 | 'total_bytes': downloaded_bytes, | |
95d8f7ea S |
313 | 'filename': ctx['filename'], |
314 | 'status': 'finished', | |
315 | 'elapsed': elapsed, | |
bd50a52b THD |
316 | 'ctx_id': ctx.get('ctx_id'), |
317 | 'max_progress': ctx.get('max_progress'), | |
318 | 'progress_idx': ctx.get('progress_idx'), | |
3ba7740d | 319 | }, info_dict) |
5219cb3e | 320 | |
321 | def _prepare_external_frag_download(self, ctx): | |
322 | if 'live' not in ctx: | |
323 | ctx['live'] = False | |
324 | if not ctx['live']: | |
325 | total_frags_str = '%d' % ctx['total_frags'] | |
326 | ad_frags = ctx.get('ad_frags', 0) | |
327 | if ad_frags: | |
328 | total_frags_str += ' (not including %d ad)' % ad_frags | |
329 | else: | |
330 | total_frags_str = 'unknown (live)' | |
86e5f3ed | 331 | self.to_screen(f'[{self.FD_NAME}] Total fragments: {total_frags_str}') |
5219cb3e | 332 | |
333 | tmpfilename = self.temp_name(ctx['filename']) | |
334 | ||
335 | # Should be initialized before ytdl file check | |
336 | ctx.update({ | |
337 | 'tmpfilename': tmpfilename, | |
338 | 'fragment_index': 0, | |
339 | }) | |
4c7853de | 340 | |
1009f67c | 341 | def decrypter(self, info_dict): |
342 | _key_cache = {} | |
343 | ||
344 | def _get_key(url): | |
345 | if url not in _key_cache: | |
346 | _key_cache[url] = self.ydl.urlopen(self._prepare_url(info_dict, url)).read() | |
347 | return _key_cache[url] | |
348 | ||
349 | def decrypt_fragment(fragment, frag_content): | |
350 | decrypt_info = fragment.get('decrypt_info') | |
351 | if not decrypt_info or decrypt_info['METHOD'] != 'AES-128': | |
352 | return frag_content | |
ac668111 | 353 | iv = decrypt_info.get('IV') or struct.pack('>8xq', fragment['media_sequence']) |
1009f67c | 354 | decrypt_info['KEY'] = decrypt_info.get('KEY') or _get_key(info_dict.get('_decryption_key_url') or decrypt_info['URI']) |
355 | # Don't decrypt the content in tests since the data is explicitly truncated and it's not to a valid block | |
356 | # size (see https://github.com/ytdl-org/youtube-dl/pull/27660). Tests only care that the correct data downloaded, | |
357 | # not what it decrypts to. | |
358 | if self.params.get('test', False): | |
359 | return frag_content | |
1d3586d0 | 360 | return unpad_pkcs7(aes_cbc_decrypt_bytes(frag_content, decrypt_info['KEY'], iv)) |
1009f67c | 361 | |
362 | return decrypt_fragment | |
363 | ||
bd50a52b THD |
364 | def download_and_append_fragments_multiple(self, *args, pack_func=None, finish_func=None): |
365 | ''' | |
366 | @params (ctx1, fragments1, info_dict1), (ctx2, fragments2, info_dict2), ... | |
367 | all args must be either tuple or list | |
368 | ''' | |
adbc4ec4 | 369 | interrupt_trigger = [True] |
bd50a52b THD |
370 | max_progress = len(args) |
371 | if max_progress == 1: | |
372 | return self.download_and_append_fragments(*args[0], pack_func=pack_func, finish_func=finish_func) | |
adbc4ec4 | 373 | max_workers = self.params.get('concurrent_fragment_downloads', 1) |
49a57e70 | 374 | if max_progress > 1: |
375 | self._prepare_multiline_status(max_progress) | |
a539f065 | 376 | is_live = any(traverse_obj(args, (..., 2, 'is_live'), default=[])) |
bd50a52b THD |
377 | |
378 | def thread_func(idx, ctx, fragments, info_dict, tpe): | |
379 | ctx['max_progress'] = max_progress | |
380 | ctx['progress_idx'] = idx | |
adbc4ec4 THD |
381 | return self.download_and_append_fragments( |
382 | ctx, fragments, info_dict, pack_func=pack_func, finish_func=finish_func, | |
383 | tpe=tpe, interrupt_trigger=interrupt_trigger) | |
bd50a52b THD |
384 | |
385 | class FTPE(concurrent.futures.ThreadPoolExecutor): | |
386 | # has to stop this or it's going to wait on the worker thread itself | |
387 | def __exit__(self, exc_type, exc_val, exc_tb): | |
388 | pass | |
389 | ||
adbc4ec4 | 390 | if compat_os_name == 'nt': |
a44ca5a4 | 391 | def future_result(future): |
a539f065 LNO |
392 | while True: |
393 | try: | |
394 | return future.result(0.1) | |
395 | except KeyboardInterrupt: | |
396 | raise | |
397 | except concurrent.futures.TimeoutError: | |
398 | continue | |
399 | else: | |
a44ca5a4 | 400 | def future_result(future): |
a539f065 LNO |
401 | return future.result() |
402 | ||
f0734e11 L |
403 | def interrupt_trigger_iter(fg): |
404 | for f in fg: | |
405 | if not interrupt_trigger[0]: | |
406 | break | |
407 | yield f | |
408 | ||
a539f065 | 409 | spins = [] |
bd50a52b | 410 | for idx, (ctx, fragments, info_dict) in enumerate(args): |
adbc4ec4 | 411 | tpe = FTPE(math.ceil(max_workers / max_progress)) |
f0734e11 | 412 | job = tpe.submit(thread_func, idx, ctx, interrupt_trigger_iter(fragments), info_dict, tpe) |
bd50a52b THD |
413 | spins.append((tpe, job)) |
414 | ||
415 | result = True | |
416 | for tpe, job in spins: | |
417 | try: | |
a44ca5a4 | 418 | result = result and future_result(job) |
adbc4ec4 THD |
419 | except KeyboardInterrupt: |
420 | interrupt_trigger[0] = False | |
bd50a52b THD |
421 | finally: |
422 | tpe.shutdown(wait=True) | |
a539f065 | 423 | if not interrupt_trigger[0] and not is_live: |
adbc4ec4 | 424 | raise KeyboardInterrupt() |
a539f065 LNO |
425 | # we expect the user wants to stop and DO WANT the preceding postprocessors to run; |
426 | # so returning a intermediate result here instead of KeyboardInterrupt on live | |
819e0531 | 427 | return result |
bd50a52b | 428 | |
adbc4ec4 THD |
429 | def download_and_append_fragments( |
430 | self, ctx, fragments, info_dict, *, pack_func=None, finish_func=None, | |
431 | tpe=None, interrupt_trigger=None): | |
432 | if not interrupt_trigger: | |
433 | interrupt_trigger = (True, ) | |
434 | ||
4c7853de | 435 | fragment_retries = self.params.get('fragment_retries', 0) |
adbc4ec4 THD |
436 | is_fatal = ( |
437 | ((lambda _: False) if info_dict.get('is_live') else (lambda idx: idx == 0)) | |
438 | if self.params.get('skip_unavailable_fragments', True) else (lambda _: True)) | |
439 | ||
4c7853de | 440 | if not pack_func: |
441 | pack_func = lambda frag_content, _: frag_content | |
442 | ||
443 | def download_fragment(fragment, ctx): | |
a539f065 | 444 | if not interrupt_trigger[0]: |
d71fd412 | 445 | return |
a539f065 | 446 | |
4c7853de | 447 | frag_index = ctx['fragment_index'] = fragment['frag_index'] |
185bf310 | 448 | ctx['last_error'] = None |
d9d8b857 | 449 | headers = info_dict.get('http_headers', {}).copy() |
4c7853de | 450 | byte_range = fragment.get('byte_range') |
451 | if byte_range: | |
452 | headers['Range'] = 'bytes=%d-%d' % (byte_range['start'], byte_range['end'] - 1) | |
453 | ||
454 | # Never skip the first fragment | |
d71fd412 | 455 | fatal, count = is_fatal(fragment.get('index') or (frag_index - 1)), 0 |
4c7853de | 456 | while count <= fragment_retries: |
457 | try: | |
36195c44 | 458 | ctx['fragment_count'] = fragment.get('fragment_count') |
d71fd412 LNO |
459 | if self._download_fragment(ctx, fragment['url'], info_dict, headers): |
460 | break | |
461 | return | |
ac668111 | 462 | except (urllib.error.HTTPError, http.client.IncompleteRead) as err: |
4c7853de | 463 | # Unavailable (possibly temporary) fragments may be served. |
464 | # First we try to retry then either skip or abort. | |
465 | # See https://github.com/ytdl-org/youtube-dl/issues/10165, | |
466 | # https://github.com/ytdl-org/youtube-dl/issues/10448). | |
467 | count += 1 | |
185bf310 | 468 | ctx['last_error'] = err |
4c7853de | 469 | if count <= fragment_retries: |
470 | self.report_retry_fragment(err, frag_index, count, fragment_retries) | |
471 | except DownloadError: | |
472 | # Don't retry fragment if error occurred during HTTP downloading | |
473 | # itself since it has own retry settings | |
474 | if not fatal: | |
475 | break | |
476 | raise | |
477 | ||
d71fd412 | 478 | if count > fragment_retries and fatal: |
4c7853de | 479 | ctx['dest_stream'].close() |
480 | self.report_error('Giving up after %s fragment retries' % fragment_retries) | |
4c7853de | 481 | |
4c7853de | 482 | def append_fragment(frag_content, frag_index, ctx): |
a44ca5a4 | 483 | if frag_content: |
484 | self._append_fragment(ctx, pack_func(frag_content, frag_index)) | |
485 | elif not is_fatal(frag_index - 1): | |
486 | self.report_skip_fragment(frag_index, 'fragment not found') | |
487 | else: | |
488 | ctx['dest_stream'].close() | |
489 | self.report_error(f'fragment {frag_index} not found, unable to continue') | |
490 | return False | |
4c7853de | 491 | return True |
492 | ||
1009f67c | 493 | decrypt_fragment = self.decrypter(info_dict) |
494 | ||
adbc4ec4 THD |
495 | max_workers = math.ceil( |
496 | self.params.get('concurrent_fragment_downloads', 1) / ctx.get('max_progress', 1)) | |
1d485a1a | 497 | if max_workers > 1: |
4c7853de | 498 | def _download_fragment(fragment): |
723d44b9 | 499 | ctx_copy = ctx.copy() |
d71fd412 LNO |
500 | download_fragment(fragment, ctx_copy) |
501 | return fragment, fragment['frag_index'], ctx_copy.get('fragment_filename_sanitized') | |
4c7853de | 502 | |
503 | self.report_warning('The download speed shown is only of one thread. This is a known issue and patches are welcome') | |
bd50a52b | 504 | with tpe or concurrent.futures.ThreadPoolExecutor(max_workers) as pool: |
1890fc63 | 505 | try: |
506 | for fragment, frag_index, frag_filename in pool.map(_download_fragment, fragments): | |
507 | ctx.update({ | |
508 | 'fragment_filename_sanitized': frag_filename, | |
509 | 'fragment_index': frag_index, | |
510 | }) | |
511 | if not append_fragment(decrypt_fragment(fragment, self._read_fragment(ctx)), frag_index, ctx): | |
512 | return False | |
513 | except KeyboardInterrupt: | |
514 | self._finish_multiline_status() | |
515 | self.report_error( | |
516 | 'Interrupted by user. Waiting for all threads to shutdown...', is_error=False, tb=False) | |
517 | pool.shutdown(wait=False) | |
518 | raise | |
4c7853de | 519 | else: |
520 | for fragment in fragments: | |
adbc4ec4 THD |
521 | if not interrupt_trigger[0]: |
522 | break | |
c854208c LNO |
523 | try: |
524 | download_fragment(fragment, ctx) | |
19a03940 | 525 | result = append_fragment( |
526 | decrypt_fragment(fragment, self._read_fragment(ctx)), fragment['frag_index'], ctx) | |
c854208c LNO |
527 | except KeyboardInterrupt: |
528 | if info_dict.get('is_live'): | |
529 | break | |
530 | raise | |
4c7853de | 531 | if not result: |
532 | return False | |
533 | ||
25a3f4f5 F |
534 | if finish_func is not None: |
535 | ctx['dest_stream'].write(finish_func()) | |
536 | ctx['dest_stream'].flush() | |
3ba7740d | 537 | self._finish_frag_download(ctx, info_dict) |
8e897ed2 | 538 | return True |