def handle(self, *args, **options):
if settings.CENTRAL_SERVER:
raise CommandError("videoscan should be run on the distributed server only.")
caching_enabled = (settings.CACHE_TIME != 0)
touched_video_ids = []
# Filesystem
files = glob.glob(os.path.join(settings.CONTENT_ROOT, "*.mp4"))
youtube_ids_in_filesystem = set([os.path.splitext(os.path.basename(f))[0] for f in files])
# Database
videos_marked_at_all = set([video.youtube_id for video in VideoFile.objects.all()])
def delete_objects_for_incomplete_videos():
# delete VideoFile objects that are not marked as in progress, but are neither 0% nor 100% done; they're broken
video_files_to_delete = VideoFile.objects.filter(download_in_progress=False, percent_complete__gt=0, percent_complete__lt=100)
deleted_video_ids = [i18n.get_video_id(video_file.youtube_id) for video_file in video_files_to_delete]
video_files_to_delete.delete()
if deleted_video_ids:
self.stdout.write("Deleted %d VideoFile models (to mark them as not downloaded, since they were in a bad state)\n" % len(deleted_video_ids))
return deleted_video_ids
touched_video_ids += delete_objects_for_incomplete_videos()
def add_missing_objects_to_db(youtube_ids_in_filesystem, videos_marked_at_all):
# Files that exist, but are not in the DB, should be assumed to be good videos,
# and just needing to be added to the DB. Add them to the DB in this way,
# so that these files also trigger the update code below (and trigger cache invalidation)
youtube_ids_needing_model_creation = list(youtube_ids_in_filesystem - videos_marked_at_all)
new_video_files = []
if youtube_ids_needing_model_creation:
for lang_code, youtube_ids in divide_videos_by_language(youtube_ids_needing_model_creation).iteritems():
# OK to do bulk_create; cache invalidation triggered via save download
lang_video_files = [VideoFile(youtube_id=id, percent_complete=100, download_in_progress=False, language=lang_code) for id in youtube_ids]
VideoFile.objects.bulk_create(lang_video_files)
new_video_files += lang_video_files
caching.invalidate_all_caches() # Do this within the loop, to update users ASAP
self.stdout.write("Created %d VideoFile models (and marked them as complete, since the files exist)\n" % len(new_video_files))
return [i18n.get_video_id(video_file.youtube_id) for video_file in new_video_files]
touched_video_ids += add_missing_objects_to_db(youtube_ids_in_filesystem, videos_marked_at_all)
def update_objects_to_be_complete(youtube_ids_in_filesystem):
# Files that exist, are in the DB, but have percent_complete=0, download_in_progress=False
updated_video_ids = []
for chunk in break_into_chunks(youtube_ids_in_filesystem):
video_files_needing_model_update = VideoFile.objects.filter(percent_complete=0, download_in_progress=False, youtube_id__in=chunk)
video_files_needing_model_update.update(percent_complete=100, flagged_for_download=False)
updated_video_ids += [i18n.get_video_id(video_file.youtube_id) for video_file in video_files_needing_model_update]
if updated_video_ids:
caching.invalidate_all_caches()
self.stdout.write("Updated %d VideoFile models (to mark them as complete, since the files exist)\n" % len(updated_video_ids))
return updated_video_ids
touched_video_ids += update_objects_to_be_complete(youtube_ids_in_filesystem)
def delete_objects_for_missing_videos(youtube_ids_in_filesystem, videos_marked_at_all):
# VideoFile objects say they're available, but that don't actually exist.
deleted_video_ids = []
videos_flagged_for_download = set([video.youtube_id for video in VideoFile.objects.filter(flagged_for_download=True)])
videos_needing_model_deletion_chunked = break_into_chunks(videos_marked_at_all - youtube_ids_in_filesystem - videos_flagged_for_download)
for chunk in videos_needing_model_deletion_chunked:
video_files_needing_model_deletion = VideoFile.objects.filter(youtube_id__in=chunk)
video_files_needing_model_deletion.delete()
deleted_video_ids += [video_file.video_id for video_file in video_files_needing_model_deletion]
if deleted_video_ids:
self.stdout.write("Deleted %d VideoFile models (because the videos didn't exist in the filesystem)\n" % len(deleted_video_ids))
return deleted_video_ids
touched_video_ids += delete_objects_for_missing_videos(youtube_ids_in_filesystem, videos_marked_at_all)
if options["auto_cache"] and caching_enabled and touched_video_ids:
caching.regenerate_all_pages_related_to_videos(video_ids=list(set(touched_video_ids)))
def handle(self, *args, **options):
if settings.CENTRAL_SERVER:
raise CommandError(
"videoscan should be run on the distributed server only.")
caching_enabled = (settings.CACHE_TIME != 0)
touched_video_ids = []
# Filesystem
files = glob.glob(os.path.join(settings.CONTENT_ROOT, "*.mp4"))
youtube_ids_in_filesystem = set(
[os.path.splitext(os.path.basename(f))[0] for f in files])
# Database
videos_marked_at_all = set(
[video.youtube_id for video in VideoFile.objects.all()])
def delete_objects_for_incomplete_videos():
# delete VideoFile objects that are not marked as in progress, but are neither 0% nor 100% done; they're broken
video_files_to_delete = VideoFile.objects.filter(
download_in_progress=False,
percent_complete__gt=0,
percent_complete__lt=100)
deleted_video_ids = [
i18n.get_video_id(video_file.youtube_id)
for video_file in video_files_to_delete
]
video_files_to_delete.delete()
if deleted_video_ids:
self.stdout.write(
"Deleted %d VideoFile models (to mark them as not downloaded, since they were in a bad state)\n"
% len(deleted_video_ids))
return deleted_video_ids
touched_video_ids += delete_objects_for_incomplete_videos()
def add_missing_objects_to_db(youtube_ids_in_filesystem,
videos_marked_at_all):
# Files that exist, but are not in the DB, should be assumed to be good videos,
# and just needing to be added to the DB. Add them to the DB in this way,
# so that these files also trigger the update code below (and trigger cache invalidation)
youtube_ids_needing_model_creation = list(
youtube_ids_in_filesystem - videos_marked_at_all)
new_video_files = []
if youtube_ids_needing_model_creation:
for lang_code, youtube_ids in divide_videos_by_language(
youtube_ids_needing_model_creation).iteritems():
# OK to do bulk_create; cache invalidation triggered via save download
lang_video_files = [
VideoFile(youtube_id=id,
percent_complete=100,
download_in_progress=False,
language=lang_code) for id in youtube_ids
]
VideoFile.objects.bulk_create(lang_video_files)
new_video_files += lang_video_files
caching.invalidate_all_caches(
) # Do this within the loop, to update users ASAP
self.stdout.write(
"Created %d VideoFile models (and marked them as complete, since the files exist)\n"
% len(new_video_files))
return [
i18n.get_video_id(video_file.youtube_id)
for video_file in new_video_files
]
touched_video_ids += add_missing_objects_to_db(
youtube_ids_in_filesystem, videos_marked_at_all)
def update_objects_to_be_complete(youtube_ids_in_filesystem):
# Files that exist, are in the DB, but have percent_complete=0, download_in_progress=False
updated_video_ids = []
for chunk in break_into_chunks(youtube_ids_in_filesystem):
video_files_needing_model_update = VideoFile.objects.filter(
percent_complete=0,
download_in_progress=False,
youtube_id__in=chunk)
video_files_needing_model_update.update(
percent_complete=100, flagged_for_download=False)
updated_video_ids += [
i18n.get_video_id(video_file.youtube_id)
for video_file in video_files_needing_model_update
]
if updated_video_ids:
caching.invalidate_all_caches()
self.stdout.write(
"Updated %d VideoFile models (to mark them as complete, since the files exist)\n"
% len(updated_video_ids))
return updated_video_ids
touched_video_ids += update_objects_to_be_complete(
youtube_ids_in_filesystem)
def delete_objects_for_missing_videos(youtube_ids_in_filesystem,
videos_marked_at_all):
# VideoFile objects say they're available, but that don't actually exist.
deleted_video_ids = []
videos_flagged_for_download = set([
video.youtube_id for video in VideoFile.objects.filter(
flagged_for_download=True)
])
videos_needing_model_deletion_chunked = break_into_chunks(
videos_marked_at_all - youtube_ids_in_filesystem -
videos_flagged_for_download)
for chunk in videos_needing_model_deletion_chunked:
video_files_needing_model_deletion = VideoFile.objects.filter(
youtube_id__in=chunk)
video_files_needing_model_deletion.delete()
deleted_video_ids += [
video_file.video_id
for video_file in video_files_needing_model_deletion
]
if deleted_video_ids:
self.stdout.write(
"Deleted %d VideoFile models (because the videos didn't exist in the filesystem)\n"
% len(deleted_video_ids))
return deleted_video_ids
touched_video_ids += delete_objects_for_missing_videos(
youtube_ids_in_filesystem, videos_marked_at_all)
if options["auto_cache"] and caching_enabled and touched_video_ids:
caching.regenerate_all_pages_related_to_videos(
video_ids=list(set(touched_video_ids)))
def handle(self, *args, **options):
self.video = None
handled_youtube_ids = [] # stored to deal with caching
failed_youtube_ids = [] # stored to avoid requerying failures.
set_process_priority.lowest(logging=settings.LOG)
try:
while True: # loop until the method is aborted
# Grab any video that hasn't been tried yet
videos = VideoFile.objects \
.filter(flagged_for_download=True, download_in_progress=False) \
.exclude(youtube_id__in=failed_youtube_ids)
video_count = videos.count()
if video_count == 0:
self.stdout.write(
_("Nothing to download; exiting.") + "\n")
break
# Grab a video as OURS to handle, set fields to indicate to others that we're on it!
# Update the video logging
video = videos[0]
video.download_in_progress = True
video.percent_complete = 0
video.save()
self.stdout.write(
(_("Downloading video '%(youtube_id)s'...") + "\n") %
{"youtube_id": video.youtube_id})
# Update the progress logging
self.set_stages(
num_stages=video_count + len(handled_youtube_ids) +
len(failed_youtube_ids) + int(options["auto_cache"]))
if not self.started():
self.start(stage_name=video.youtube_id)
# Initiate the download process
try:
ensure_dir(settings.CONTENT_ROOT)
progress_callback = partial(
self.download_progress_callback, video)
try:
# Download via urllib
download_video(video.youtube_id,
callback=progress_callback)
except URLNotFound:
# Video was not found on amazon cloud service,
# either due to a KA mistake, or due to the fact
# that it's a dubbed video.
#
# We can use youtube-dl to get that video!!
logging.debug(
_("Retrieving youtube video %(youtube_id)s via youtube-dl"
) % {"youtube_id": video.youtube_id})
def youtube_dl_cb(stats, progress_callback, *args,
**kwargs):
if stats['status'] == "finished":
percent = 100.
elif stats['status'] == "downloading":
percent = 100. * stats[
'downloaded_bytes'] / stats['total_bytes']
else:
percent = 0.
progress_callback(percent=percent)
scrape_video(video.youtube_id,
quiet=not settings.DEBUG,
callback=partial(
youtube_dl_cb,
progress_callback=progress_callback))
# If we got here, we downloaded ... somehow :)
handled_youtube_ids.append(video.youtube_id)
self.stdout.write(_("Download is complete!") + "\n")
# caching.invalidate_all_caches() # Unnecessary; we have a database listener for this.
except DownloadCancelled:
# Cancellation event
video.percent_complete = 0
video.flagged_for_download = False
video.download_in_progress = False
video.save()
failed_youtube_ids.append(video.youtube_id)
except Exception as e:
# On error, report the error, mark the video as not downloaded,
# and allow the loop to try other videos.
msg = _(
"Error in downloading %(youtube_id)s: %(error_msg)s"
) % {
"youtube_id": video.youtube_id,
"error_msg": unicode(e)
}
self.stderr.write("%s\n" % msg)
# If a connection error, we should retry.
if isinstance(e, DownloadError):
connection_error = "[Errno 8]" in e.message
elif isinstance(e, IOError) and hasattr(e, "strerror"):
connection_error = e.strerror[0] == 8
else:
connection_error = False
video.download_in_progress = False
video.flagged_for_download = connection_error # Any error other than a connection error is fatal.
video.save()
# Rather than getting stuck on one video, continue to the next video.
self.update_stage(
stage_status="error",
notes=_("%(error_msg)s; continuing to next video.") %
{"error_msg": msg})
failed_youtube_ids.append(video.youtube_id)
continue
# This can take a long time, without any further update, so ... best to avoid.
if options["auto_cache"] and caching.caching_is_enabled(
) and handled_youtube_ids:
self.update_stage(
stage_name=self.video.youtube_id,
stage_percent=0,
notes=_("Generating all pages related to videos."))
caching.regenerate_all_pages_related_to_videos(video_ids=list(
set([
i18n.get_video_id(yid) or yid
for yid in handled_youtube_ids
])))
# Update
self.complete(
notes=
_("Downloaded %(num_handled_videos)s of %(num_total_videos)s videos successfully."
) % {
"num_handled_videos":
len(handled_youtube_ids),
"num_total_videos":
len(handled_youtube_ids) + len(failed_youtube_ids),
})
except Exception as e:
self.cancel(stage_status="error",
notes=_("Error: %(error_msg)s") % {"error_msg": e})
raise
def handle(self, *args, **options):
self.video = None
handled_youtube_ids = [] # stored to deal with caching
failed_youtube_ids = [] # stored to avoid requerying failures.
set_process_priority.lowest(logging=settings.LOG)
try:
while True: # loop until the method is aborted
# Grab any video that hasn't been tried yet
videos = VideoFile.objects.filter(flagged_for_download=True, download_in_progress=False).exclude(
youtube_id__in=failed_youtube_ids
)
video_count = videos.count()
if video_count == 0:
self.stdout.write(_("Nothing to download; exiting.") + "\n")
break
# Grab a video as OURS to handle, set fields to indicate to others that we're on it!
# Update the video logging
video = videos[0]
video.download_in_progress = True
video.percent_complete = 0
video.save()
self.stdout.write(
(_("Downloading video '%(youtube_id)s'...") + "\n") % {"youtube_id": video.youtube_id}
)
# Update the progress logging
self.set_stages(
num_stages=video_count
+ len(handled_youtube_ids)
+ len(failed_youtube_ids)
+ int(options["auto_cache"])
)
if not self.started():
self.start(stage_name=video.youtube_id)
# Initiate the download process
try:
ensure_dir(settings.CONTENT_ROOT)
progress_callback = partial(self.download_progress_callback, video)
try:
# Download via urllib
download_video(video.youtube_id, callback=progress_callback)
except URLNotFound:
# Video was not found on amazon cloud service,
# either due to a KA mistake, or due to the fact
# that it's a dubbed video.
#
# We can use youtube-dl to get that video!!
logging.debug(
_("Retrieving youtube video %(youtube_id)s via youtube-dl")
% {"youtube_id": video.youtube_id}
)
def youtube_dl_cb(stats, progress_callback, *args, **kwargs):
if stats["status"] == "finished":
percent = 100.0
elif stats["status"] == "downloading":
percent = 100.0 * stats["downloaded_bytes"] / stats["total_bytes"]
else:
percent = 0.0
progress_callback(percent=percent)
scrape_video(
video.youtube_id,
quiet=not settings.DEBUG,
callback=partial(youtube_dl_cb, progress_callback=progress_callback),
)
# If we got here, we downloaded ... somehow :)
handled_youtube_ids.append(video.youtube_id)
self.stdout.write(_("Download is complete!") + "\n")
# caching.invalidate_all_caches() # Unnecessary; we have a database listener for this.
except DownloadCancelled:
# Cancellation event
video.percent_complete = 0
video.flagged_for_download = False
video.download_in_progress = False
video.save()
failed_youtube_ids.append(video.youtube_id)
except Exception as e:
# On error, report the error, mark the video as not downloaded,
# and allow the loop to try other videos.
msg = _("Error in downloading %(youtube_id)s: %(error_msg)s") % {
"youtube_id": video.youtube_id,
"error_msg": unicode(e),
}
self.stderr.write("%s\n" % msg)
# If a connection error, we should retry.
if isinstance(e, DownloadError):
connection_error = "[Errno 8]" in e.message
elif isinstance(e, IOError) and hasattr(e, "strerror"):
connection_error = e.strerror[0] == 8
else:
connection_error = False
video.download_in_progress = False
video.flagged_for_download = connection_error # Any error other than a connection error is fatal.
video.save()
# Rather than getting stuck on one video, continue to the next video.
self.update_stage(
stage_status="error", notes=_("%(error_msg)s; continuing to next video.") % {"error_msg": msg}
)
failed_youtube_ids.append(video.youtube_id)
continue
# This can take a long time, without any further update, so ... best to avoid.
if options["auto_cache"] and caching.caching_is_enabled() and handled_youtube_ids:
self.update_stage(
stage_name=self.video.youtube_id,
stage_percent=0,
notes=_("Generating all pages related to videos."),
)
caching.regenerate_all_pages_related_to_videos(
video_ids=list(set([i18n.get_video_id(yid) or yid for yid in handled_youtube_ids]))
)
# Update
self.complete(
notes=_("Downloaded %(num_handled_videos)s of %(num_total_videos)s videos successfully.")
% {
"num_handled_videos": len(handled_youtube_ids),
"num_total_videos": len(handled_youtube_ids) + len(failed_youtube_ids),
}
)
except Exception as e:
self.cancel(stage_status="error", notes=_("Error: %(error_msg)s") % {"error_msg": e})
raise
