KakoIndex.py

import asyncio
from io import BytesIO
import pathlib
import re
from urllib.parse import urljoin, urlparse

import httpx
import msgspec
import typer
import tqdm
from bs4 import BeautifulSoup
from loguru import logger
from dateutil.parser import parse
from dateutil import tz
import aiofile

app = typer.Typer()

JST = tz.gettz("Japan/Tokyo")


class ThreadData(msgspec.Struct):
    id: int
    title: str
    region: str  # Root Server
    city: str  # Channel/Category/etc.
    replies: int


class Message(msgspec.Struct):
    name: str
    mail: str | None
    dateid: str | None
    body: str
    title: str | None

class BrokenMessage(msgspec.Struct):
    blocks:list[str]

class FilledThreadData(ThreadData):
    messages: list[Message]


date_refixer = re.compile(
    r"\((?:[月火水木金土日]|Sat|Sun|Mon|Tue|Wed|Thr|Fri)\)", flags=re.IGNORECASE
)


def parse_date_id(dateid_string: str) -> tuple[float, str | None]:
    """
    Parses a string to extract a timestamp and message IDs.

    Args:
        dateid_string: The input string potentially containing date, ID, and BE ID.

    Returns:
        A tuple containing the Unix timestamp (float) or None and the formatted
        message ID string or None.
    """
    # Standardize the input string
    cleaned_id = dateid_string.removesuffix(".net").strip()

    # Use regular expressions to find all ID parts
    # This is more robust than splitting
    id_parts = re.findall(r"(ID|BE):([\w\.\/]+)", cleaned_id)
    message_ids = [f"{kind}:{value}" for kind, value in id_parts]
    message_id_str = " | ".join(message_ids) if message_ids else None

    # Extract the date part by removing the ID parts
    date_part = re.sub(r"\s*(ID|BE):[\w\.\/+-=?!()★]+", "", cleaned_id).strip()

    timestamp = -1
    if date_part:
        # Assuming date_refixer is a compiled regex object like:
        # For demonstration, we'll define a simple one.
        try:
            clean_date_str = date_refixer.sub("", date_part) + " JST"
            timestamp = parse(clean_date_str, tzinfos={"JST": JST}).timestamp()
        except ValueError as e:
            logger.error(f"{clean_date_str}|{dateid_string}|{e}")
            # Handle cases where parsing might fail
            timestamp = -1

    return timestamp, message_id_str


def parse_splitter(line: str):
    if not line:
        return
    blocks = line.split("<>")
    if len(blocks) > 5:
        print(blocks)
        logger.warning(f"{blocks}\nBlock count validation failed. Please check.")
        return BrokenMessage(blocks)
    name, email, dateid, body, *rest = blocks
    if body == "<em>■ このスレッドは過去ログ倉庫に格納されています</em>":
        return
    rest = [i for i in rest if i]

    if dateid.lower().startswith(("停止", "あぼーん", "over 1000 ", "移転")):
        return
    # Remove
    # XXX: not used and to be processed seperately. Because as it turns out, it's kind of hard!
    # _, _ = parse_date_id(dateid)
    # logger.debug(f"{jp_date, message_id}")
    return Message(
        name,
        email if email else None,
        dateid,
        body,
        "<>".join(rest) if rest else None,
    )


def parse_messages(text: str) -> list[Message]:
    messages = []
    for message in text.split("\n"):
        try:
            specmsg = parse_splitter(message)
            messages.append(specmsg)
        except Exception as e:
            logger.exception(e)
    return messages


async def retry_error(session: httpx.AsyncClient, url: str):
    while True:
        try:
            response = await session.get(url)
            response.raise_for_status()
            return response
        except Exception as e:
            logger.warning(f"Failed to fetch {url} | {e} | Trying again")
            await asyncio.sleep(5)


async def main(output_dump: pathlib.Path):
    session = httpx.AsyncClient()
    session.headers["user-agent"] = ""

    loop = asyncio.get_running_loop()

    boards = await session.get("https://2ch.sc/bbsmenu.html")
    boards.raise_for_status()
    soup = BeautifulSoup(boards.text, "lxml")

    encoder = msgspec.json.Encoder()

    # with BytesIO() as fout:
    for links in soup.find_all("a", attrs={"href": True}):
        link = links["href"]
        if (
            "/be/" in link
            or "//be." in link
            or "//be." in link
            or link
            in [
                "//2ch.sc/",
                "//info.2ch.sc/guide/",
                "//sp.2ch.sc/",
                "//sweet.2ch.sc/headline/",
                "//find.2ch.sc/",
                "//irc.2ch.sc/",
            ]
            or link.endswith(".html")
        ):
            continue
        if "2ch.sc" not in link:
            continue
        if not isinstance(link, str):
            raise Exception()

        decomposed = urlparse(link)
        city = decomposed.path.strip("/")
        region = decomposed.hostname.replace(".2ch.sc", "")

        dumpFile = output_dump.with_stem(f"{output_dump.stem}-{region}_{city}")
        if dumpFile.is_file() or dumpFile.with_suffix(".jsonl.zst").is_file():
            continue
        
        resolved_base = urljoin("https://", link)
        warehouse_catalog = urljoin(resolved_base, "kako/subject.txt")
        warehouses = await session.get(warehouse_catalog, timeout=None)
        if warehouses.status_code == 404:
            logger.warning(f"{warehouse_catalog} returned 404")
            continue

        warehouses.raise_for_status()
        warehouse_ids = [
            i.split("<>")[0] for i in warehouses.text.split("\n") if i.startswith("o")
        ]
        threadsDump = []
        concurrent = asyncio.Semaphore(128)

        async def fetch_subjects(warehouse_id: str):
            async with concurrent:
                subject_id = urljoin(resolved_base, f"kako/{warehouse_id}/subject.txt")
                subjects = [
                    i
                    for i in (await retry_error(session, subject_id)).text.split("\n")
                    if i
                ]
                for subject in subjects:
                    subject = subject.strip()
                    thread_dat, *rest = subject.split("<>")
                    *thread_title, replies = "<>".join(rest).split("(")

                    thread = ThreadData(
                        int(thread_dat.split(".")[0]),
                        "[".join(thread_title).rstrip(" "),
                        region,
                        city,
                        int(replies.rstrip(")")),
                    )
                    threadsDump.append(thread)

        threads = [
            asyncio.create_task(fetch_subjects(subject_id))
            for subject_id in warehouse_ids
        ]
        for completed in tqdm.tqdm(
            asyncio.as_completed(threads), total=len(warehouse_ids)
        ):
            tid = await completed
        
        if len(threadsDump) <= 1_000_000:
            logger.warning(f"{warehouse_catalog} has less than 1,000,000 archived threads. Skipping.")
            continue
        
        logger.info("Requesting threads.")
        taskQueue = asyncio.Queue(maxsize=1024 * 128)
        responseQueue = asyncio.Queue(maxsize=1024 * 128)
        totalThreads = len(threadsDump)
        logger.info(f"total: {totalThreads}")

        async def writer_task():
            nonlocal dumpFile
            rootStem = dumpFile.stem
            i = 0
            with tqdm.tqdm(total=totalThreads) as pbar:
                fout = await aiofile.async_open(dumpFile, "wb")
                while True:
                    tid = await responseQueue.get()
                    if tid is None:
                        break
                    data = await loop.run_in_executor(None,encoder.encode_lines,[tid])
                    # logger.debug(data)
                    await fout.write(data)
                    pbar.update(1)
                    if fout.tell() > (1024**3) * 10:
                        await fout.close()
                        i += 1
                        dumpFile = dumpFile.with_stem(f"{rootStem}-{i}")
                        fout = await aiofile.async_open(dumpFile, "wb")

        async def fetch_thread_task():
            while True:
                thread_data = await taskQueue.get()
                if thread_data is None:
                    break
                thread_dat = urljoin(resolved_base, f"dat/{thread_data.id}.dat")
                tries = 10
                while tries > 0:
                    try:
                        response = await session.get(thread_dat)
                        response.raise_for_status()
                        break
                    except Exception as e:
                        tries -= 1
                        logger.warning(
                            f"Failed to fetch {thread_dat} | {e} | Trying again"
                        )
                await responseQueue.put(
                    FilledThreadData(
                        thread_data.id,
                        thread_data.title,
                        thread_data.region,
                        thread_data.city,
                        thread_data.replies,
                        await asyncio.to_thread(parse_messages, response.text),
                    )
                )

        logger.info("Starting workers.")
        
        workers: list[asyncio.Task] = [
            loop.create_task(fetch_thread_task()) for _ in range(128)
        ]
        writer = loop.create_task(writer_task())

        for thread in threadsDump:
            await taskQueue.put(thread)
        for _ in range(128):
            await taskQueue.put(None)
        while workers:
            workers = [i for i in workers if not i.done()]
            await asyncio.sleep(0.01)
        await responseQueue.put(None)
        await writer


@app.command()
def dump_2ch_kako(output_catalogs: pathlib.Path):
    asyncio.run(main(output_catalogs))


if __name__ == "__main__":
    app()