monolith/sources/ingest.py

34 lines
857 B
Python

import db
import util
import ujson
import asyncio
SOURCES = ["irc"]
KEYPREFIX = "queue."
CHUNK_SIZE = 1000
ITER_DELAY = 0.5
class Ingest(object):
def __init__(self):
name = self.__class__.__name__
self.log = util.get_logger(name)
async def run(self):
while True:
await self.process_chunk()
await asyncio.sleep(ITER_DELAY)
async def process_chunk(self):
items = []
for source in SOURCES:
key = f"{KEYPREFIX}{source}"
chunk = await db.ar.spop(key, CHUNK_SIZE)
if not chunk:
continue
self.log.info(f"Got chunk: {chunk}")
for item in chunk:
item = ujson.loads(item)
self.log.info(f"Got item: {item}")
items.append(item)
db.store_message_bulk(items)