monolith/sources/ingest.py

import asyncio
from os import getenv

import orjson

import db
import util
from processing import process

SOURCES = ["4ch", "irc", "dis"]
# DEBUG CODE REMOVE ME
# SOURCES.remove("4ch")
# SOURCES.remove("dis")
# DEBUG CODE REMOVE ME
KEYPREFIX = "queue."

# Chunk size per source (divide by len(SOURCES) for total)
CHUNK_SIZE = int(getenv("MONOLITH_INGEST_CHUNK_SIZE", "900"))
ITER_DELAY = float(getenv("MONOLITH_INGEST_ITER_DELAY", "0.5"))

log = util.get_logger("ingest")


class Ingest(object):
    def __init__(self):
        name = self.__class__.__name__
        self.log = util.get_logger(name)
        self.log.info(
            (
                "Starting ingest handler for chunk size of "
                f"{CHUNK_SIZE} every {ITER_DELAY} seconds."
            )
        )

    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
            for item in chunk:
                item = orjson.loads(item)
                items.append(item)
        if items:
            await process.spawn_processing_threads(items)