# 项目文件：crawler_project/src/catalogcrawl/transport.py
# 这里只重试读取 GET；状态失败与网络等待失败保留 URL 和尝试次数。
from asyncio import sleep
import logging
import math

import httpx

from .errors import CrawlError

logger = logging.getLogger(__name__)
RETRY_STATUS = {429, 500, 502, 503, 504}


def retry_delay(response, attempt):
    fallback = 0.05 * 2 ** (attempt - 1)
    value = response.headers.get("Retry-After") if response is not None else None
    try:
        seconds = float(value)
    except (TypeError, ValueError):
        return fallback
    return seconds if math.isfinite(seconds) and seconds >= 0 else fallback


async def get_response(client, url, gate, attempts=3):
    if attempts < 1:
        raise ValueError("请求尝试次数应至少为 1")
    for attempt in range(1, attempts + 1):
        await gate.wait()
        response = None
        try:
            response = await client.get(url)
            response.raise_for_status()
            return response
        except httpx.HTTPStatusError as cause:
            retryable = cause.response.status_code in RETRY_STATUS
            failure = cause
        except (
            httpx.ConnectError,
            httpx.ReadTimeout,
            httpx.RemoteProtocolError,
        ) as cause:
            retryable, failure = True, cause
        except httpx.RequestError as cause:
            raise CrawlError(f"{url}：{type(cause).__name__} {cause}") from cause
        if not retryable or attempt == attempts:
            raise CrawlError(f"{url}：第 {attempt} 次读取失败，{failure}") from failure
        delay = retry_delay(response, attempt)
        logger.warning("准备重试 url=%s attempt=%s wait=%.3f", url, attempt, delay)
        # 退避让出循环；下一次尝试仍经过共享的 StartGate。
        await sleep(delay)

# 项目文件：demos/retry_get.py
# 服务前两次返回 503，第三次成功；重试范围覆盖读取，不覆盖 HTML 字段解析。
import asyncio

import httpx
from crawlsite import running_site
from catalogcrawl.pacing import StartGate
from catalogcrawl.transport import get_response


async def main(base_url):
    async with httpx.AsyncClient(timeout=1, trust_env=False) as client:
        response = await get_response(
            client, base_url + "/books/b103", StartGate(0), attempts=3
        )
        print(response.status_code)


if __name__ == "__main__":
    with running_site() as site:
        site.state.failures["/books/b103"] = 2
        asyncio.run(main(site.base_url))
        print("尝试次数", site.state.counts["/books/b103"])