109 lines
3.5 KiB
Python
109 lines
3.5 KiB
Python
from scheduler import job
|
|
|
|
from .src.fetch_feed import FetchFeeds
|
|
from .src.fetch_parser import FetchParser
|
|
from .src.fetch_search import FetchSearcher
|
|
from .src.db_utils import DB_Handler
|
|
'''
|
|
from src.missing_kids_fetch import MissingKidsFetch
|
|
'''
|
|
|
|
from .src.logger import get_logger
|
|
logger = get_logger()
|
|
|
|
@job('default')
|
|
def fetch_feeds():
|
|
task = "Fetch Feeds"
|
|
logger.info("Task triggered: {}".format(task))
|
|
FetchFeeds().run()
|
|
logger.info("Task completed: {}".format(task))
|
|
|
|
@job('default')
|
|
def fetch_parser():
|
|
task = "Fetch Parser"
|
|
logger.info("Task triggered: {}".format(task))
|
|
FetchParser().run()
|
|
logger.info("Task completed: {}".format(task))
|
|
|
|
@job('default')
|
|
def fetch_search():
|
|
task = "Fetch Search"
|
|
logger.info("Task triggered: {}".format(task))
|
|
FetchSearcher().run()
|
|
logger.info("Task completed: {}".format(task))
|
|
|
|
# TODO: fetch_missing_kids()
|
|
|
|
@job('default')
|
|
def process_raw_urls(batch_size=50):
|
|
task = "Process raw URLs"
|
|
logger.info("Task triggered: {}".format(task))
|
|
DB_Handler().process_raw_urls(batch_size=batch_size)
|
|
logger.info("Task completed: {}".format(task))
|
|
|
|
@job('default')
|
|
def process_error_urls(batch_size=50):
|
|
task = "Process error URLs"
|
|
logger.info("Task triggered: {}".format(task))
|
|
DB_Handler().process_error_urls(batch_size=batch_size)
|
|
logger.info("Task completed: {}".format(task))
|
|
|
|
@job('default')
|
|
def process_missing_kids_urls(batch_size=50):
|
|
task = "Process Missing Kids URLs"
|
|
logger.info("Task triggered: {}".format(task))
|
|
DB_Handler().process_missing_kids_urls(batch_size=batch_size)
|
|
logger.info("Task completed: {}".format(task))
|
|
|
|
@job('default')
|
|
def process_missing_kids_urls_all(batch_size=None):
|
|
task = "Process Missing Kids URLs ALL"
|
|
logger.info("Task triggered: {}".format(task))
|
|
DB_Handler().process_missing_kids_urls(batch_size=batch_size)
|
|
logger.info("Task completed: {}".format(task))
|
|
|
|
|
|
|
|
|
|
|
|
@job('default')
|
|
def background_task(process_type: str):
|
|
logger.info("Task triggered: {}".format(process_type))
|
|
|
|
try:
|
|
if (process_type == "fetch_feeds"):
|
|
FetchFeeds().run()
|
|
elif (process_type == "fetch_parser"):
|
|
FetchParser().run()
|
|
elif (process_type == "fetch_search"):
|
|
FetchSearcher().run()
|
|
#elif (process_type == "fetch_missingkids"):
|
|
# FetchMissingKids().run()
|
|
elif ("process_" in process_type):
|
|
# Batch size encoded in URL
|
|
try:
|
|
batch_size = int(process_type.split("_")[-1])
|
|
except Exception as e:
|
|
batch_size = None
|
|
# Task type
|
|
if ("process_raw_urls" in process_type):
|
|
DB_Handler().process_raw_urls(batch_size=batch_size)
|
|
elif ("process_error_urls" in process_type):
|
|
DB_Handler().process_error_urls(batch_size=batch_size)
|
|
elif ("process_missing_kids_urls" in process_type):
|
|
DB_Handler().process_missing_kids_urls(batch_size=batch_size)
|
|
else:
|
|
logger.info("Task unknown!: {}".format(process_type))
|
|
|
|
'''
|
|
# Selenium based
|
|
elif (process_type == "fetch_missing_kids_reduced"):
|
|
MissingKidsFetch(db_handler, num_pages=4).run()
|
|
elif (process_type == "fetch_missing_kids_full"):
|
|
MissingKidsFetch(db_handler, num_pages=100000).run()
|
|
'''
|
|
|
|
logger.info("Task completed: {}".format(process_type))
|
|
except Exception as e:
|
|
logger.error(e)
|