Plugin cleanup and tweaks

This commit is contained in:
2023-02-20 19:18:45 -06:00
parent 372e4ff3dc
commit 3ad9e1c7bb
1138 changed files with 48878 additions and 40445 deletions

View File

@@ -1,30 +1,23 @@
from __future__ import division, unicode_literals
import concurrent.futures
import contextlib
import http.client
import json
import math
import os
import struct
import time
try:
import concurrent.futures
can_threaded_download = True
except ImportError:
can_threaded_download = False
import urllib.error
from .common import FileDownloader
from .http import HttpFD
from ..aes import aes_cbc_decrypt_bytes, unpad_pkcs7
from ..compat import (
compat_os_name,
compat_urllib_error,
compat_struct_pack,
)
from ..compat import compat_os_name
from ..utils import (
DownloadError,
error_to_compat_str,
RetryManager,
encodeFilename,
sanitized_Request,
traverse_obj,
)
@@ -32,9 +25,7 @@ class HttpQuietDownloader(HttpFD):
def to_screen(self, *args, **kargs):
pass
def report_retry(self, err, count, retries):
super().to_screen(
f'[download] Got server HTTP error: {err}. Retrying (attempt {count} of {self.format_retries(retries)}) ...')
to_console_title = to_screen
class FragmentFD(FileDownloader):
@@ -74,9 +65,9 @@ class FragmentFD(FileDownloader):
"""
def report_retry_fragment(self, err, frag_index, count, retries):
self.to_screen(
'\r[download] Got server HTTP error: %s. Retrying fragment %d (attempt %d of %s) ...'
% (error_to_compat_str(err), frag_index, count, self.format_retries(retries)))
self.deprecation_warning('yt_dlp.downloader.FragmentFD.report_retry_fragment is deprecated. '
'Use yt_dlp.downloader.FileDownloader.report_retry instead')
return self.report_retry(err, count, retries, frag_index)
def report_skip_fragment(self, frag_index, err=None):
err = f' {err};' if err else ''
@@ -130,16 +121,23 @@ class FragmentFD(FileDownloader):
'request_data': request_data,
'ctx_id': ctx.get('ctx_id'),
}
success = ctx['dl'].download(fragment_filename, fragment_info_dict)
success, _ = ctx['dl'].download(fragment_filename, fragment_info_dict)
if not success:
return False, None
return False
if fragment_info_dict.get('filetime'):
ctx['fragment_filetime'] = fragment_info_dict.get('filetime')
ctx['fragment_filename_sanitized'] = fragment_filename
return True, self._read_fragment(ctx)
return True
def _read_fragment(self, ctx):
down, frag_sanitized = self.sanitize_open(ctx['fragment_filename_sanitized'], 'rb')
if not ctx.get('fragment_filename_sanitized'):
return None
try:
down, frag_sanitized = self.sanitize_open(ctx['fragment_filename_sanitized'], 'rb')
except FileNotFoundError:
if ctx.get('live'):
return None
raise
ctx['fragment_filename_sanitized'] = frag_sanitized
frag_content = down.read()
down.close()
@@ -153,7 +151,7 @@ class FragmentFD(FileDownloader):
if self.__do_ytdl_file(ctx):
self._write_ytdl_file(ctx)
if not self.params.get('keep_fragments', False):
os.remove(encodeFilename(ctx['fragment_filename_sanitized']))
self.try_remove(encodeFilename(ctx['fragment_filename_sanitized']))
del ctx['fragment_filename_sanitized']
def _prepare_frag_download(self, ctx):
@@ -166,21 +164,13 @@ class FragmentFD(FileDownloader):
total_frags_str += ' (not including %d ad)' % ad_frags
else:
total_frags_str = 'unknown (live)'
self.to_screen(
'[%s] Total fragments: %s' % (self.FD_NAME, total_frags_str))
self.to_screen(f'[{self.FD_NAME}] Total fragments: {total_frags_str}')
self.report_destination(ctx['filename'])
dl = HttpQuietDownloader(
self.ydl,
{
'continuedl': True,
'quiet': self.params.get('quiet'),
'noprogress': True,
'ratelimit': self.params.get('ratelimit'),
'retries': self.params.get('retries', 0),
'nopart': self.params.get('nopart', False),
'test': self.params.get('test', False),
}
)
dl = HttpQuietDownloader(self.ydl, {
**self.params,
'noprogress': True,
'test': False,
})
tmpfilename = self.temp_name(ctx['filename'])
open_mode = 'wb'
resume_len = 0
@@ -253,6 +243,9 @@ class FragmentFD(FileDownloader):
if s['status'] not in ('downloading', 'finished'):
return
if not total_frags and ctx.get('fragment_count'):
state['fragment_count'] = ctx['fragment_count']
if ctx_id is not None and s.get('ctx_id') != ctx_id:
return
@@ -299,21 +292,26 @@ class FragmentFD(FileDownloader):
if self.__do_ytdl_file(ctx):
ytdl_filename = encodeFilename(self.ytdl_filename(ctx['filename']))
if os.path.isfile(ytdl_filename):
os.remove(ytdl_filename)
self.try_remove(ytdl_filename)
elapsed = time.time() - ctx['started']
if ctx['tmpfilename'] == '-':
downloaded_bytes = ctx['complete_frags_downloaded_bytes']
to_file = ctx['tmpfilename'] != '-'
if to_file:
downloaded_bytes = os.path.getsize(encodeFilename(ctx['tmpfilename']))
else:
downloaded_bytes = ctx['complete_frags_downloaded_bytes']
if not downloaded_bytes:
if to_file:
self.try_remove(ctx['tmpfilename'])
self.report_error('The downloaded file is empty')
return False
elif to_file:
self.try_rename(ctx['tmpfilename'], ctx['filename'])
if self.params.get('updatetime', True):
filetime = ctx.get('fragment_filetime')
if filetime:
try:
os.utime(ctx['filename'], (time.time(), filetime))
except Exception:
pass
downloaded_bytes = os.path.getsize(encodeFilename(ctx['filename']))
filetime = ctx.get('fragment_filetime')
if self.params.get('updatetime', True) and filetime:
with contextlib.suppress(Exception):
os.utime(ctx['filename'], (time.time(), filetime))
self._hook_progress({
'downloaded_bytes': downloaded_bytes,
@@ -325,6 +323,7 @@ class FragmentFD(FileDownloader):
'max_progress': ctx.get('max_progress'),
'progress_idx': ctx.get('progress_idx'),
}, info_dict)
return True
def _prepare_external_frag_download(self, ctx):
if 'live' not in ctx:
@@ -336,8 +335,7 @@ class FragmentFD(FileDownloader):
total_frags_str += ' (not including %d ad)' % ad_frags
else:
total_frags_str = 'unknown (live)'
self.to_screen(
'[%s] Total fragments: %s' % (self.FD_NAME, total_frags_str))
self.to_screen(f'[{self.FD_NAME}] Total fragments: {total_frags_str}')
tmpfilename = self.temp_name(ctx['filename'])
@@ -356,11 +354,14 @@ class FragmentFD(FileDownloader):
return _key_cache[url]
def decrypt_fragment(fragment, frag_content):
if frag_content is None:
return
decrypt_info = fragment.get('decrypt_info')
if not decrypt_info or decrypt_info['METHOD'] != 'AES-128':
return frag_content
iv = decrypt_info.get('IV') or compat_struct_pack('>8xq', fragment['media_sequence'])
decrypt_info['KEY'] = decrypt_info.get('KEY') or _get_key(info_dict.get('_decryption_key_url') or decrypt_info['URI'])
iv = decrypt_info.get('IV') or struct.pack('>8xq', fragment['media_sequence'])
decrypt_info['KEY'] = (decrypt_info.get('KEY')
or _get_key(traverse_obj(info_dict, ('hls_aes', 'uri')) or decrypt_info['URI']))
# Don't decrypt the content in tests since the data is explicitly truncated and it's not to a valid block
# size (see https://github.com/ytdl-org/youtube-dl/pull/27660). Tests only care that the correct data downloaded,
# not what it decrypts to.
@@ -370,7 +371,7 @@ class FragmentFD(FileDownloader):
return decrypt_fragment
def download_and_append_fragments_multiple(self, *args, pack_func=None, finish_func=None):
def download_and_append_fragments_multiple(self, *args, **kwargs):
'''
@params (ctx1, fragments1, info_dict1), (ctx2, fragments2, info_dict2), ...
all args must be either tuple or list
@@ -378,63 +379,76 @@ class FragmentFD(FileDownloader):
interrupt_trigger = [True]
max_progress = len(args)
if max_progress == 1:
return self.download_and_append_fragments(*args[0], pack_func=pack_func, finish_func=finish_func)
return self.download_and_append_fragments(*args[0], **kwargs)
max_workers = self.params.get('concurrent_fragment_downloads', 1)
if max_progress > 1:
self._prepare_multiline_status(max_progress)
is_live = any(traverse_obj(args, (..., 2, 'is_live')))
def thread_func(idx, ctx, fragments, info_dict, tpe):
ctx['max_progress'] = max_progress
ctx['progress_idx'] = idx
return self.download_and_append_fragments(
ctx, fragments, info_dict, pack_func=pack_func, finish_func=finish_func,
tpe=tpe, interrupt_trigger=interrupt_trigger)
ctx, fragments, info_dict, **kwargs, tpe=tpe, interrupt_trigger=interrupt_trigger)
class FTPE(concurrent.futures.ThreadPoolExecutor):
# has to stop this or it's going to wait on the worker thread itself
def __exit__(self, exc_type, exc_val, exc_tb):
pass
spins = []
if compat_os_name == 'nt':
self.report_warning('Ctrl+C does not work on Windows when used with parallel threads. '
'This is a known issue and patches are welcome')
def future_result(future):
while True:
try:
return future.result(0.1)
except KeyboardInterrupt:
raise
except concurrent.futures.TimeoutError:
continue
else:
def future_result(future):
return future.result()
def interrupt_trigger_iter(fg):
for f in fg:
if not interrupt_trigger[0]:
break
yield f
spins = []
for idx, (ctx, fragments, info_dict) in enumerate(args):
tpe = FTPE(math.ceil(max_workers / max_progress))
job = tpe.submit(thread_func, idx, ctx, fragments, info_dict, tpe)
job = tpe.submit(thread_func, idx, ctx, interrupt_trigger_iter(fragments), info_dict, tpe)
spins.append((tpe, job))
result = True
for tpe, job in spins:
try:
result = result and job.result()
result = result and future_result(job)
except KeyboardInterrupt:
interrupt_trigger[0] = False
finally:
tpe.shutdown(wait=True)
if not interrupt_trigger[0]:
if not interrupt_trigger[0] and not is_live:
raise KeyboardInterrupt()
# we expect the user wants to stop and DO WANT the preceding postprocessors to run;
# so returning a intermediate result here instead of KeyboardInterrupt on live
return result
def download_and_append_fragments(
self, ctx, fragments, info_dict, *, pack_func=None, finish_func=None,
tpe=None, interrupt_trigger=None):
if not interrupt_trigger:
interrupt_trigger = (True, )
self, ctx, fragments, info_dict, *, is_fatal=(lambda idx: False),
pack_func=(lambda content, idx: content), finish_func=None,
tpe=None, interrupt_trigger=(True, )):
fragment_retries = self.params.get('fragment_retries', 0)
is_fatal = (
((lambda _: False) if info_dict.get('is_live') else (lambda idx: idx == 0))
if self.params.get('skip_unavailable_fragments', True) else (lambda _: True))
if not pack_func:
pack_func = lambda frag_content, _: frag_content
if not self.params.get('skip_unavailable_fragments', True):
is_fatal = lambda _: True
def download_fragment(fragment, ctx):
if not interrupt_trigger[0]:
return
frag_index = ctx['fragment_index'] = fragment['frag_index']
ctx['last_error'] = None
if not interrupt_trigger[0]:
return False, frag_index
headers = info_dict.get('http_headers', {}).copy()
byte_range = fragment.get('byte_range')
if byte_range:
@@ -442,82 +456,78 @@ class FragmentFD(FileDownloader):
# Never skip the first fragment
fatal = is_fatal(fragment.get('index') or (frag_index - 1))
count, frag_content = 0, None
while count <= fragment_retries:
try:
success, frag_content = self._download_fragment(ctx, fragment['url'], info_dict, headers)
if not success:
return False, frag_index
break
except (compat_urllib_error.HTTPError, http.client.IncompleteRead) as err:
# Unavailable (possibly temporary) fragments may be served.
# First we try to retry then either skip or abort.
# See https://github.com/ytdl-org/youtube-dl/issues/10165,
# https://github.com/ytdl-org/youtube-dl/issues/10448).
count += 1
ctx['last_error'] = err
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:
break
raise
if count > fragment_retries:
if not fatal:
return False, frag_index
ctx['dest_stream'].close()
self.report_error('Giving up after %s fragment retries' % fragment_retries)
return False, frag_index
return frag_content, frag_index
def error_callback(err, count, retries):
if fatal and count > retries:
ctx['dest_stream'].close()
self.report_retry(err, count, retries, frag_index, fatal)
ctx['last_error'] = err
for retry in RetryManager(self.params.get('fragment_retries'), error_callback):
try:
ctx['fragment_count'] = fragment.get('fragment_count')
if not self._download_fragment(ctx, fragment['url'], info_dict, headers):
return
except (urllib.error.HTTPError, http.client.IncompleteRead) as err:
retry.error = err
continue
except DownloadError: # has own retry settings
if fatal:
raise
def append_fragment(frag_content, frag_index, ctx):
if not frag_content:
if not is_fatal(frag_index - 1):
self.report_skip_fragment(frag_index, 'fragment not found')
return True
else:
ctx['dest_stream'].close()
self.report_error(
'fragment %s not found, unable to continue' % frag_index)
return False
self._append_fragment(ctx, pack_func(frag_content, frag_index))
if frag_content:
self._append_fragment(ctx, pack_func(frag_content, frag_index))
elif not is_fatal(frag_index - 1):
self.report_skip_fragment(frag_index, 'fragment not found')
else:
ctx['dest_stream'].close()
self.report_error(f'fragment {frag_index} not found, unable to continue')
return False
return True
decrypt_fragment = self.decrypter(info_dict)
max_workers = math.ceil(
self.params.get('concurrent_fragment_downloads', 1) / ctx.get('max_progress', 1))
if can_threaded_download and max_workers > 1:
if max_workers > 1:
def _download_fragment(fragment):
ctx_copy = ctx.copy()
frag_content, frag_index = download_fragment(fragment, ctx_copy)
return fragment, frag_content, frag_index, ctx_copy.get('fragment_filename_sanitized')
download_fragment(fragment, ctx_copy)
return fragment, fragment['frag_index'], ctx_copy.get('fragment_filename_sanitized')
self.report_warning('The download speed shown is only of one thread. This is a known issue and patches are welcome')
with tpe or concurrent.futures.ThreadPoolExecutor(max_workers) as pool:
for fragment, frag_content, frag_index, frag_filename in pool.map(_download_fragment, fragments):
if not interrupt_trigger[0]:
break
ctx['fragment_filename_sanitized'] = frag_filename
ctx['fragment_index'] = frag_index
result = append_fragment(decrypt_fragment(fragment, frag_content), frag_index, ctx)
if not result:
return False
try:
for fragment, frag_index, frag_filename in pool.map(_download_fragment, fragments):
ctx.update({
'fragment_filename_sanitized': frag_filename,
'fragment_index': frag_index,
})
if not append_fragment(decrypt_fragment(fragment, self._read_fragment(ctx)), frag_index, ctx):
return False
except KeyboardInterrupt:
self._finish_multiline_status()
self.report_error(
'Interrupted by user. Waiting for all threads to shutdown...', is_error=False, tb=False)
pool.shutdown(wait=False)
raise
else:
for fragment in fragments:
if not interrupt_trigger[0]:
break
frag_content, frag_index = download_fragment(fragment, ctx)
result = append_fragment(decrypt_fragment(fragment, frag_content), frag_index, ctx)
try:
download_fragment(fragment, ctx)
result = append_fragment(
decrypt_fragment(fragment, self._read_fragment(ctx)), fragment['frag_index'], ctx)
except KeyboardInterrupt:
if info_dict.get('is_live'):
break
raise
if not result:
return False
if finish_func is not None:
ctx['dest_stream'].write(finish_func())
ctx['dest_stream'].flush()
self._finish_frag_download(ctx, info_dict)
return True
return self._finish_frag_download(ctx, info_dict)