youtube_dl_exe=YOUTUBE_DL,
finished_warcs_dir=os.environ["FINISHED_WARCS_DIR"],
warc_max_size=WARC_MAX_SIZE,
monitor_disk=WPULL_MONITOR_DISK,
monitor_memory=WPULL_MONITOR_MEMORY,
)
pipeline = Pipeline(
CheckIP(), CheckLocalWebserver(),
GetItemFromQueue(control,
pipeline_id,
downloader,
ao_only=env.get('AO_ONLY'),
large=env.get('LARGE'),
version_check=(VERSION, pipeline_version)),
StartHeartbeat(control), SetFetchDepth(), PreparePaths(), WriteInfo(),
DownloadUrlFile(control),
Wpull(wpull_args,
accept_on_exit_code=AcceptAny(),
env={
'ITEM_IDENT': ItemInterpolation('%(ident)s'),
'LOG_KEY': ItemInterpolation('%(log_key)s'),
'REDIS_URL': REDIS_URL,
'PATH': os.environ['PATH']
}), RelabelIfAborted(control), CompressLogIfFailed(), WriteInfo(),
MoveFiles(target_directory=os.environ["FINISHED_WARCS_DIR"]),
StopHeartbeat(), MarkItemAsDone(control, EXPIRE_TIME))
def stop_control():
#control.flag_logging_thread_for_termination()
_, _, _, pipeline_id = monitoring.pipeline_id()
wpull_args = WpullArgs(default_user_agent=DEFAULT_USER_AGENT,
wpull_exe=WPULL_EXE,
youtube_dl_exe=YOUTUBE_DL,
phantomjs_exe=PHANTOMJS,
finished_warcs_dir=os.environ["FINISHED_WARCS_DIR"],
warc_max_size=WARC_MAX_SIZE)
pipeline = Pipeline(
CheckIP(),
GetItemFromQueue(control,
pipeline_id,
downloader,
ao_only=env.get('AO_ONLY'),
large=env.get('LARGE')), StartHeartbeat(control),
SetFetchDepth(), PreparePaths(), WriteInfo(), DownloadUrlFile(control),
WgetDownload(wpull_args,
accept_on_exit_code=AcceptAny(),
env={
'ITEM_IDENT': ItemInterpolation('%(ident)s'),
'LOG_KEY': ItemInterpolation('%(log_key)s'),
'REDIS_URL': REDIS_URL,
'PATH': os.environ['PATH']
}), RelabelIfAborted(control), WriteInfo(), MoveFiles(),
LimitConcurrent(
2,
RsyncUpload(target=RSYNC_URL,
target_source_path=ItemInterpolation("%(data_dir)s"),
files=ItemValue("all_target_files"),
extra_args=['--partial', '--partial-dir', '.rsync-tmp'])),
args.extend(phantomjs_args)
return args
@classmethod
def add_args(cls, args, names, item):
for name in names:
value = name % item
if value:
args.append(value)
_, _, _, pipeline_id = monitoring.pipeline_id()
pipeline = Pipeline(
GetItemFromQueue(control, pipeline_id, ao_only=env.get('AO_ONLY')),
StartHeartbeat(control),
SetFetchDepth(),
PreparePaths(),
WriteInfo(),
DownloadUrlFile(control),
WgetDownload(WpullArgs(),
accept_on_exit_code=AcceptAny(),
env={
'ITEM_IDENT': ItemInterpolation('%(ident)s'),
'LOG_KEY': ItemInterpolation('%(log_key)s'),
'REDIS_URL': REDIS_URL,
'PATH': os.environ['PATH']
}),
RelabelIfAborted(control),
WriteInfo(),
MoveFiles(),