monolith/processing/process.py

106 lines
3.3 KiB
Python
Raw Normal View History

from concurrent.futures import ProcessPoolExecutor
import asyncio
import os
import ujson
from siphashc import siphash
import db
import util
# 4chan schema
from schemas.ch4_s import ATTRMAP
# For key generation
import string
import random
# For timestamp processing
import datetime
# For 4chan message parsing
from bs4 import BeautifulSoup
from numpy import array_split
from math import ceil
log = util.get_logger("process")
# Maximum number of CPU threads to use for post processing
CPU_THREADS = os.cpu_count()
p = ProcessPoolExecutor(CPU_THREADS)
def get_hash_key():
hash_key = db.r.get("hashing_key")
if not hash_key:
letters = string.ascii_lowercase
hash_key = "".join(random.choice(letters) for i in range(16))
log.debug(f"Created new hash key: {hash_key}")
db.r.set("hashing_key", hash_key)
else:
hash_key = hash_key.decode("ascii")
log.debug(f"Decoded hash key: {hash_key}")
return hash_key
hash_key = get_hash_key()
async def spawn_processing_threads(data):
print("SPAWN", data)
if len(data) < CPU_THREADS:
split_data = [data]
else:
msg_per_core = int(len(data) / CPU_THREADS)
print("MSG PER CORE", msg_per_core)
split_data = array_split(data, ceil(len(data) / msg_per_core))
print("SPLIT DATA", split_data)
for split in split_data:
print("DELEGATING TO THREAD", len(split))
await process_data_thread(split)
@asyncio.coroutine
def process_data_thread(data):
"""
Helper to spawn threads to process a list of data.
"""
loop = asyncio.get_event_loop()
yield from loop.run_in_executor(p, process_data, data)
def process_data(data):
print("PROCESSING DATA", data)
for index, msg in enumerate(data):
#print("PROCESSING", msg)
if msg["src"] == "4ch":
board = msg["net"]
thread = msg["channel"]
# Calculate hash for post
post_normalised = ujson.dumps(msg, sort_keys=True)
hash = siphash(hash_key, post_normalised)
hash = str(hash)
redis_key = f"cache.{board}.{thread}.{msg['no']}"
key_content = db.r.get(redis_key)
if key_content:
key_content = key_content.decode("ascii")
if key_content == hash:
continue
else:
data[index][index]["type"] = "update"
db.r.set(redis_key, hash)
for key2, value in list(msg.items()):
if key2 in ATTRMAP:
msg[ATTRMAP[key2]] = data[index][key2]
del data[index][key2]
if "ts" in msg:
old_time = data[index]["ts"]
# '08/30/22(Tue)02:25:37'
time_spl = old_time.split(":")
if len(time_spl) == 3:
old_ts = datetime.strptime(old_time, "%m/%d/%y(%a)%H:%M:%S")
else:
old_ts = datetime.strptime(old_time, "%m/%d/%y(%a)%H:%M")
# new_ts = old_ts.isoformat()
new_ts = int(old_ts.timestamp())
data[index]["ts"] = new_ts
if "msg" in msg:
soup = BeautifulSoup(data[index]["msg"], "html.parser")
msg = soup.get_text(separator="\n")
data[index]["msg"] = msg