Add option `--file-access-retries` (#2066)

Closes #517
Authored by: ehoogeveen-medweb

README.md

@@ -463,6 +463,8 @@ You can also fork the project on github and run your fork's [build workflow](.gi
                                      video data is re-extracted (e.g. 100K)
     -R, --retries RETRIES            Number of retries (default is 10), or
                                      "infinite"
+    --file-access-retries RETRIES    Number of times to retry on file access error
+                                     (default is 10), or "infinite"
     --fragment-retries RETRIES       Number of retries for a fragment (default
                                      is 10), or "infinite" (DASH, hlsnative and
                                      ISM)

yt_dlp/YoutubeDL.py

@@ -450,8 +450,8 @@ class YoutubeDL(object):
     The following parameters are not used by YoutubeDL itself, they are used by
     the downloader (see yt_dlp/downloader/common.py):
     nopart, updatetime, buffersize, ratelimit, throttledratelimit, min_filesize,
-    max_filesize, test, noresizebuffer, retries, fragment_retries, continuedl,
-    noprogress, xattr_set_filesize, hls_use_mpegts, http_chunk_size,
+    max_filesize, test, noresizebuffer, retries, file_access_retries, fragment_retries,
+    continuedl, noprogress, xattr_set_filesize, hls_use_mpegts, http_chunk_size,
     external_downloader_args, concurrent_fragment_downloads.
 
     The following options are used by the post processors:

yt_dlp/__init__.py

@@ -222,6 +222,8 @@ def _real_main(argv=None):
         return parsed_retries
     if opts.retries is not None:
         opts.retries = parse_retries(opts.retries)
+    if opts.file_access_retries is not None:
+        opts.file_access_retries = parse_retries(opts.file_access_retries, 'file access ')
     if opts.fragment_retries is not None:
         opts.fragment_retries = parse_retries(opts.fragment_retries, 'fragment ')
     if opts.extractor_retries is not None:
@@ -673,6 +675,7 @@ def _real_main(argv=None):
         'throttledratelimit': opts.throttledratelimit,
         'overwrites': opts.overwrites,
         'retries': opts.retries,
+        'file_access_retries': opts.file_access_retries,
         'fragment_retries': opts.fragment_retries,
         'extractor_retries': opts.extractor_retries,
         'skip_unavailable_fragments': opts.skip_unavailable_fragments,

yt_dlp/downloader/common.py

@@ -4,12 +4,14 @@ import os
 import re
 import time
 import random
+import errno
 
 from ..utils import (
     decodeArgument,
     encodeFilename,
     error_to_compat_str,
     format_bytes,
+    sanitize_open,
     shell_quote,
     timeconvert,
     timetuple_from_msec,
@@ -39,6 +41,7 @@ class FileDownloader(object):
     ratelimit:          Download speed limit, in bytes/sec.
     throttledratelimit: Assume the download is being throttled below this speed (bytes/sec)
     retries:            Number of times to retry for HTTP error 5xx
+    file_access_retries:   Number of times to retry on file access error
     buffersize:         Size of download buffer in bytes.
     noresizebuffer:     Do not automatically resize the download buffer.
     continuedl:         Try to continue downloads if possible.
@@ -207,6 +210,21 @@ class FileDownloader(object):
     def ytdl_filename(self, filename):
         return filename + '.ytdl'
 
+    def sanitize_open(self, filename, open_mode):
+        file_access_retries = self.params.get('file_access_retries', 10)
+        retry = 0
+        while True:
+            try:
+                return sanitize_open(filename, open_mode)
+            except (IOError, OSError) as err:
+                retry = retry + 1
+                if retry > file_access_retries or err.errno not in (errno.EACCES,):
+                    raise
+                self.to_screen(
+                    '[download] Got file access error. Retrying (attempt %d of %s) ...'
+                    % (retry, self.format_retries(file_access_retries)))
+                time.sleep(0.01)
+
     def try_rename(self, old_filename, new_filename):
         if old_filename == new_filename:
             return

yt_dlp/downloader/external.py

@@ -22,7 +22,6 @@ from ..utils import (
     handle_youtubedl_headers,
     check_executable,
     Popen,
-    sanitize_open,
 )
 
 
@@ -144,11 +143,11 @@ class ExternalFD(FragmentFD):
                 return -1
 
         decrypt_fragment = self.decrypter(info_dict)
-        dest, _ = sanitize_open(tmpfilename, 'wb')
+        dest, _ = self.sanitize_open(tmpfilename, 'wb')
         for frag_index, fragment in enumerate(info_dict['fragments']):
             fragment_filename = '%s-Frag%d' % (tmpfilename, frag_index)
             try:
-                src, _ = sanitize_open(fragment_filename, 'rb')
+                src, _ = self.sanitize_open(fragment_filename, 'rb')
             except IOError as err:
                 if skip_unavailable_fragments and frag_index > 1:
                     self.report_skip_fragment(frag_index, err)
@@ -290,7 +289,7 @@ class Aria2cFD(ExternalFD):
             for frag_index, fragment in enumerate(info_dict['fragments']):
                 fragment_filename = '%s-Frag%d' % (os.path.basename(tmpfilename), frag_index)
                 url_list.append('%s\n\tout=%s' % (fragment['url'], fragment_filename))
-            stream, _ = sanitize_open(url_list_file, 'wb')
+            stream, _ = self.sanitize_open(url_list_file, 'wb')
             stream.write('\n'.join(url_list).encode('utf-8'))
             stream.close()
             cmd += ['-i', url_list_file]

yt_dlp/downloader/fragment.py

@@ -24,7 +24,6 @@ from ..utils import (
     DownloadError,
     error_to_compat_str,
     encodeFilename,
-    sanitize_open,
     sanitized_Request,
 )
 
@@ -96,7 +95,7 @@ class FragmentFD(FileDownloader):
 
     def _read_ytdl_file(self, ctx):
         assert 'ytdl_corrupt' not in ctx
-        stream, _ = sanitize_open(self.ytdl_filename(ctx['filename']), 'r')
+        stream, _ = self.sanitize_open(self.ytdl_filename(ctx['filename']), 'r')
         try:
             ytdl_data = json.loads(stream.read())
             ctx['fragment_index'] = ytdl_data['downloader']['current_fragment']['index']
@@ -108,7 +107,7 @@ class FragmentFD(FileDownloader):
             stream.close()
 
     def _write_ytdl_file(self, ctx):
-        frag_index_stream, _ = sanitize_open(self.ytdl_filename(ctx['filename']), 'w')
+        frag_index_stream, _ = self.sanitize_open(self.ytdl_filename(ctx['filename']), 'w')
         try:
             downloader = {
                 'current_fragment': {
@@ -140,7 +139,7 @@ class FragmentFD(FileDownloader):
         return True, self._read_fragment(ctx)
 
     def _read_fragment(self, ctx):
-        down, frag_sanitized = sanitize_open(ctx['fragment_filename_sanitized'], 'rb')
+        down, frag_sanitized = self.sanitize_open(ctx['fragment_filename_sanitized'], 'rb')
         ctx['fragment_filename_sanitized'] = frag_sanitized
         frag_content = down.read()
         down.close()
@@ -216,7 +215,7 @@ class FragmentFD(FileDownloader):
                 self._write_ytdl_file(ctx)
                 assert ctx['fragment_index'] == 0
 
-        dest_stream, tmpfilename = sanitize_open(tmpfilename, open_mode)
+        dest_stream, tmpfilename = self.sanitize_open(tmpfilename, open_mode)
 
         ctx.update({
             'dl': dl,

yt_dlp/downloader/http.py

@@ -16,7 +16,6 @@ from ..utils import (
     ContentTooShortError,
     encodeFilename,
     int_or_none,
-    sanitize_open,
     sanitized_Request,
     ThrottledDownload,
     write_xattr,
@@ -263,7 +262,7 @@ class HttpFD(FileDownloader):
                 # Open destination file just in time
                 if ctx.stream is None:
                     try:
-                        ctx.stream, ctx.tmpfilename = sanitize_open(
+                        ctx.stream, ctx.tmpfilename = self.sanitize_open(
                             ctx.tmpfilename, ctx.open_mode)
                         assert ctx.stream is not None
                         ctx.filename = self.undo_temp_name(ctx.tmpfilename)

yt_dlp/options.py

@@ -681,6 +681,10 @@ def parseOpts(overrideArguments=None):
         '-R', '--retries',
         dest='retries', metavar='RETRIES', default=10,
         help='Number of retries (default is %default), or "infinite"')
+    downloader.add_option(
+        '--file-access-retries',
+        dest='file_access_retries', metavar='RETRIES', default=10,
+        help='Number of times to retry on file access error (default is %default), or "infinite"')
     downloader.add_option(
         '--fragment-retries',
         dest='fragment_retries', metavar='RETRIES', default=10,