scripts/update_wiki.py

import argparse
import json
from typing import Any

import requests
from _danbooru import DEFAULT_RATE, RateLimiter, get_json
from _env import credentials
from _paths import DANBOORU_DB_PATH
from rich.progress import (
    BarColumn,
    Progress,
    TaskProgressColumn,
    TextColumn,
    TimeElapsedColumn,
)
from sqlalchemy import Table, text
from sqlalchemy.engine import Engine

from db import get_engine, metadata, t_artists, t_tag_aliases, t_tag_implications, t_tags, t_wiki_pages

BASE_URL = "https://danbooru.donmai.us"
PAGE_LIMIT = 1000

ENDPOINTS: list[tuple[str, Table]] = [
    ("wiki_pages", t_wiki_pages),
    ("tags", t_tags),
    ("tag_aliases", t_tag_aliases),
    ("tag_implications", t_tag_implications),
    ("artists", t_artists),
]


def auth_params(api_key: str | None, login: str | None) -> dict[str, str]:
    # 仅在提供凭据时附加认证参数,否则匿名访问
    if api_key and login:
        return {"api_key": api_key, "login": login}
    return {}


def describe(auth: dict[str, str]) -> str:
    # 只回显 login。匿名和已认证的限流差一个数量级,跑之前得能一眼看出走的哪条路。
    return f"authenticated as {auth['login']}" if auth else "anonymous (heavier rate limiting)"


def normalize_row(row: dict[str, Any], columns: list[str]) -> dict[str, Any]:
    out: dict[str, Any] = {column: row.get(column) for column in columns}
    if isinstance(out.get("other_names"), list):
        out["other_names"] = json.dumps(out["other_names"], ensure_ascii=False)
    return out


def fetch_page(
    session: requests.Session,
    endpoint: str,
    params: dict[str, Any],
    limiter: RateLimiter,
) -> list[dict[str, Any]]:
    return get_json(session, f"{BASE_URL}/{endpoint}.json", params, limiter)


def fetch_max_id(session: requests.Session, endpoint: str, auth: dict[str, str], limiter: RateLimiter) -> int:
    # page=b{很大的 id} 返回现存最大 id 的一行(id 降序),对所有 endpoint 通用
    rows = fetch_page(session, endpoint, {"limit": 1, "page": "b1000000000", **auth}, limiter)
    return int(rows[0]["id"]) if rows else 0


def sync_endpoint(
    session: requests.Session,
    engine: Engine,
    endpoint: str,
    table: Table,
    auth: dict[str, str],
    progress: Progress,
    limiter: RateLimiter,
) -> int:
    columns = [column.name for column in table.columns]
    max_id = fetch_max_id(session, endpoint, auth, limiter)
    task = progress.add_task(endpoint, total=max_id)
    cursor = 0
    total_rows = 0
    while True:
        # 节奏由 limiter 统一控制,这里不再自己 sleep —— 两处限速会互相叠加,
        # 而且撞到 429 时只有 limiter 那份会退避。
        rows = fetch_page(session, endpoint, {"limit": PAGE_LIMIT, "page": f"a{cursor}", **auth}, limiter)
        if not rows:
            break
        payload = [normalize_row(row, columns) for row in rows]
        with engine.connect() as conn:
            conn.execute(table.insert().prefix_with("OR REPLACE"), payload)
            conn.commit()
        cursor = max(int(row["id"]) for row in rows)
        total_rows += len(rows)
        progress.update(task, completed=min(cursor, max_id))
    progress.update(task, completed=max_id)
    return total_rows


def parse_args() -> argparse.Namespace:
    parser = argparse.ArgumentParser(
        description="Archive Danbooru wiki pages and tag metadata into SQLite."
    )
    parser.add_argument("--api_key", type=str)
    parser.add_argument("--login", type=str)
    parser.add_argument("--db_path", type=str, default=str(DANBOORU_DB_PATH))
    parser.add_argument(
        "--only",
        type=str,
        default=None,
        help="只同步指定 endpoint(逗号分隔),如 --only artists 或 --only wiki_pages,tags",
    )
    parser.add_argument(
        "--rate",
        type=float,
        default=DEFAULT_RATE,
        help=f"起步请求速率(req/s,默认 {DEFAULT_RATE})。撞到 429 会自动减半再慢慢恢复。",
    )
    return parser.parse_args()


def select_endpoints(only: str | None) -> list[tuple[str, Table]]:
    if only is None:
        return ENDPOINTS
    wanted = {name.strip() for name in only.split(",") if name.strip()}
    known = {name for name, _ in ENDPOINTS}
    unknown = wanted - known
    if unknown:
        raise SystemExit(f"unknown endpoint(s): {', '.join(sorted(unknown))}; available: {', '.join(sorted(known))}")
    return [(name, table) for name, table in ENDPOINTS if name in wanted]


def main() -> None:
    args = parse_args()
    auth = auth_params(*credentials(args.api_key, args.login))
    endpoints = select_endpoints(args.only)
    limiter = RateLimiter(args.rate)
    print(f"syncing {', '.join(name for name, _ in endpoints)} -- {describe(auth)} at {args.rate:g} req/s")
    engine = get_engine(args.db_path)
    metadata.create_all(engine, tables=[table for _, table in endpoints])

    session = requests.Session()
    counts: dict[str, int] = {}
    with Progress(
        TextColumn("[progress.description]{task.description}"),
        BarColumn(),
        TaskProgressColumn(),
        TimeElapsedColumn(),
    ) as progress:
        for endpoint, table in endpoints:
            counts[endpoint] = sync_endpoint(session, engine, endpoint, table, auth, progress, limiter)

    with engine.connect() as conn:
        for endpoint, table in endpoints:
            row_count = conn.execute(text(f"SELECT COUNT(*) FROM {table.name}")).scalar()
            print(f"{table.name}: fetched {counts[endpoint]}, table now has {row_count} rows")
    # 收尾时的速率就是这次实际能跑到的配额。明显低于起步值说明撞过 429,下次可以
    # 直接用 --rate 从那个数字起步,省掉一路试探。
    print(f"final rate: {limiter.rate:.2f} req/s (started at {args.rate:g})")


if __name__ == "__main__":
    main()