monolith/sources/ingest.py

import asyncio

import ujson

import db
import util

from processing import process

SOURCES = ["irc", "dis", "4ch"]
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.get_chunk()
            await asyncio.sleep(ITER_DELAY)

    async def get_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)
        if items:
            print("PROCESSING", len(items))
            await process.spawn_processing_threads(items)
            print("DONE WITH PROCESSING", len(items))
            await db.store_kafka_batch(items)