Files
matitos_news/app_urls/fetcher/src/fetch_missing_kids.py
2025-04-07 09:37:16 +02:00

44 lines
1.9 KiB
Python

from .db_utils import DB_Handler
from ..models import Search, Source
import os
import requests
import json
import traceback
from .logger import get_logger
logger = get_logger()
class FetchMissingKids():
def __init__(self) -> None:
logger.debug("Initializing Fetcher MissingKids")
def run(self, number_pages=-1):
try:
logger.debug("Starting MissingKids.run(), processing #{} pages".format(number_pages))
# Get source object
obj_source, created = Source.objects.get_or_create(source="missingkids.org")
# Get search object
obj_search, created = Search.objects.get_or_create(search="missingkids.org/poster", type=Search.TYPE_ENUM.URL_HOST)
try:
# Missing kids fetching endpoint, parameter number of pages to fetch
missingkids_fetch_endpoint = os.path.join(os.getenv("SELENIUM_ENDPOINT", "http://localhost:80"), "get_missing_kids/?pages={}".format(number_pages))
# Timeout
if (number_pages > 15) or (number_pages == -1):
timeout = 60*90 # 1.5h
else:
timeout = 60*10 # 10 min
logger.debug("Running request on MissingKids endpoint: {}".format(missingkids_fetch_endpoint))
# Request
r = requests.get(missingkids_fetch_endpoint, timeout=timeout)
# Decode
urls_fetched = json.loads(r.text).get("list_urls", [])
except Exception as e:
logger.warning("Timeout on request: {}. {}".format(missingkids_fetch_endpoint, str(e)))
urls_fetched = []
# Write to DB
DB_Handler().insert_raw_urls(urls_fetched, obj_source, obj_search)
except Exception as e:
logger.warning("Exception in MissingKids.run(): {}\n{}".format(e, traceback.format_exc()))