| 123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591 |
- # -*- coding: utf-8 -*-
- # Author : Charley
- # Python : 3.12.10
- # Date : 2026/07/15
- """
- SCP Auctions (catalogs.scpauctions.com) 公用模块:HTTP 配置、拍卖会列表解析、
- 场次内 lot 列表分页抓取、详情页解析。被 scp_history.py / scp_spider.py 复用。
- 目标网站: https://catalogs.scpauctions.com/auctions/past
- 逻辑要点:
- 1. 站点为 Bidsquare 托管平台,页面服务端渲染(SSR),无独立数据 API,
- 全部数据直接在 HTML 里,用 parsel 解析即可(本地开了 VPN,请求统一走代理)。
- 2. 拍卖会列表页 /auctions/past 分页:?page=N(每页 24 场),翻到空页为止。
- 每张拍卖会卡片是 <div class="gtm-visible_event" data-event_id=...>:
- event_id(去重键)、标题、类型、起止时间、主页/图录链接、封面图。
- 3. 图录(lot 列表)页 catalog_url?page=N&limit=120(limit 上限 120):
- lot 卡片是 <div id="stl-{item_id}" data-item_id=...>,含 Lot 号 / 标题 /
- 详情链接 / 缩略图 / 出价数 / 成交状态与成交价。翻到空页为止。
- 4. lot 详情页 /online-auctions/{house}/{slug}-{item_id}:
- 标题取 og:title;成交价在 .bidding-price(流拍则为空);
- 属性(Collection/Sport/Type/Athlete/Team 等)在 .item-attributes li,
- label/span 成对,字段数量不固定(2~5 个,可能缺 Athlete/Team),
- 故按 label 建字典再按需取值,缺失置空;
- 多图为 s1.img.bidsquare.com/item/{l|s}/... 归一化到 large 尺寸去重。
- """
- import re
- import random
- from datetime import datetime
- from loguru import logger
- from parsel import Selector
- from curl_cffi import requests
- from curl_cffi.requests import BrowserType
- from urllib.parse import urljoin
- from tenacity import retry, stop_after_attempt, wait_fixed, retry_if_not_exception_type
- class LotNotFound(Exception):
- """lot 详情页不存在(HTTP 404 或站点软 404「Nothing Found」页)。
- 该异常表示拍品已被撤下/移除,是永久性错误,不应重试;
- 捕获后应把记录标记为终态 state=3(不再进入补抓队列)。
- """
- # —— 站点常量 ——
- BASE_URL = "https://catalogs.scpauctions.com" # 站点根,用于拼接相对链接
- PAST_URL = "https://catalogs.scpauctions.com/auctions/past" # 拍卖会(历史)列表页
- CATALOG_PAGE_SIZE = 120 # 图录页每页 lot 数(limit 参数,站点上限 120)
- IMG_ITEM_RE = r'https://s1\.img\.bidsquare\.com/item/[ls]/\d+/\d+\.jpe?g(?:\?t=[^"\'\s\\]+)?' # 拍品图 URL
- # 直接用库内置的所有浏览器指纹(伪装真实浏览器 TLS/UA,规避潜在的指纹风控)
- client_identifier_list = [b.value for b in BrowserType]
- # 不写死 user-agent:curl_cffi 会按 impersonate 指纹自动带匹配的 UA,
- # 写死会造成 TLS 指纹与 UA 头矛盾,反而更易被识别。只保留通用、不冲突的头。
- headers = {
- "accept": "text/html,application/xhtml+xml,application/xml;q=0.9,image/avif,image/webp,image/apng,*/*;q=0.8,application/signed-exchange;v=b3;q=0.7",
- "accept-language": "en-US,en;q=0.9",
- }
- def after_log(retry_state):
- """tenacity retry 回调,统一打印重试日志。
- Args:
- retry_state: tenacity.RetryCallState,retry 框架自动传入。
- Returns:
- None: 仅打印日志,无返回值。
- """
- if retry_state.args and len(retry_state.args) > 0:
- log = retry_state.args[0]
- else:
- log = logger
- if retry_state.outcome.failed:
- log.warning(f"Function '{retry_state.fn.__name__}', Attempt {retry_state.attempt_number} Times")
- else:
- log.info(f"Function '{retry_state.fn.__name__}', Attempt {retry_state.attempt_number} succeeded")
- @retry(stop=stop_after_attempt(5), wait=wait_fixed(2), after=after_log)
- def get_proxys(log):
- """获取代理字典,全部请求方法的 proxies 参数都走这里。
- 本地开了 VPN,从大陆直连站点不通,请求必须走代理。默认用住宅代理(北美出口,
- 与 SCP 面向的美区一致)。若要改用本地 VPN 客户端的 HTTP 代理端口(如 Clash 的
- 127.0.0.1:7890),把 http_proxy / https_proxy 换成对应地址即可。
- Args:
- log: logger 对象。
- Returns:
- dict: requests 风格代理字典 {"http": ..., "https": ...}。
- Raises:
- Exception: 透传内部异常以便 tenacity 触发重试。
- """
- # 住宅 ip 池 北美(与参考项目 rea_spider 一致);如需切本地 VPN 端口在此改
- http_proxy = "http://u1952150085001297:sJMHl4qc4bM0@proxy.123proxy.cn:36927"
- https_proxy = "http://u1952150085001297:sJMHl4qc4bM0@proxy.123proxy.cn:36927"
- # 本地 VPN(Clash)示例,二选一:
- # http_proxy = https_proxy = "http://127.0.0.1:7890"
- try:
- return {
- "http": http_proxy,
- "https": https_proxy,
- }
- except Exception as e:
- log.error(f"Error getting proxy: {e}")
- raise e
- # retry_if_not_exception_type(LotNotFound):404 是永久性错误,直接抛出不重试,
- # 避免对已下架的 lot 做 5×3 次徒劳请求
- @retry(stop=stop_after_attempt(5), wait=wait_fixed(2), after=after_log,
- retry=retry_if_not_exception_type(LotNotFound))
- def _get_selector(log, session, impersonate, url):
- """GET 一个页面并返回 parsel Selector(带重试,走代理)。
- Args:
- log: logger 对象。
- session (requests.Session): curl_cffi 会话对象。
- impersonate (str): 浏览器指纹标识。
- url (str): 目标绝对 URL。
- Returns:
- Selector: 响应 HTML 的 parsel 解析对象。
- Raises:
- LotNotFound: 页面 HTTP 404/410 或软 404「Nothing Found」时抛出(不重试)。
- Exception: 其他 HTTP 非 2xx(如 5xx/超时)时透传以触发重试。
- """
- resp = session.get(url, headers=headers, impersonate=impersonate,
- proxies=get_proxys(log), timeout=30)
- # 页面不存在:404/410 直接判定;部分站点软 404 返回 200 但页面是「Nothing Found」
- if resp.status_code in (404, 410) or (
- resp.status_code == 200 and "can not be found" in resp.text):
- raise LotNotFound(f"页面不存在({resp.status_code}): {url}")
- resp.raise_for_status()
- return Selector(resp.text)
- def _parse_datetime(raw):
- """把站点时间原文解析成 datetime(丢弃时区名,保留东部墙钟时间)。
- Args:
- raw (str): 时间原文,如 "May 20, 2026 01:00PM EDT" 或带 "Start:" 前缀。
- Returns:
- datetime | None: 解析成功返回 datetime;失败或空返回 None。
- """
- if not raw:
- return None
- # 去掉 Start:/End: 前缀与末尾时区名(EDT/EST/…),只留 "May 20, 2026 01:00PM"
- text = re.sub(r'^\s*(Start|End)\s*:\s*', '', raw.strip(), flags=re.I)
- m = re.search(r'([A-Za-z]{3,9}\s+\d{1,2},\s*\d{4}\s+\d{1,2}:\d{2}\s*[AP]M)', text, re.I)
- if not m:
- return None
- try:
- return datetime.strptime(m.group(1).replace(" ", " ").upper(), "%b %d, %Y %I:%M%p")
- except ValueError:
- return None
- def parse_auction_list(selector):
- """解析 /auctions/past 单页的拍卖会卡片列表。
- Args:
- selector (Selector): 拍卖会列表页某一页 GET 响应的 parsel 解析对象。
- Returns:
- list[dict]: 每场一个 dict,字段:event_id, auction_name, auction_type,
- event_status, start_time_raw, end_time_raw, start_time, end_time,
- auction_url, catalog_url, cover_img;无卡片返回空列表。
- """
- auctions = []
- for card in selector.css('div.gtm-visible_event[data-event_id]'):
- event_id = card.attrib.get("data-event_id", "").strip()
- if not event_id:
- continue
- # 标题 + 拍卖会主页链接
- title_a = card.css('.list-top h1 a')
- auction_name = (title_a.xpath('normalize-space(.)').get() or
- card.attrib.get("data-event_name", "")).strip()
- auction_url = title_a.attrib.get("href", "").strip()
- # 类型(Timed Auction 等)
- auction_type = (card.css('.list-top label::text').get() or "").strip()
- # 起止时间:.list-top 下多个 <span>,按前缀区分
- start_raw = end_raw = ""
- for sp in card.css('.list-top span'):
- t = (sp.xpath('normalize-space(.)').get() or "").strip()
- if t.lower().startswith("start"):
- start_raw = t
- elif t.lower().startswith("end"):
- end_raw = t
- # 图录链接(去掉锚点 #catalog)
- catalog_url = (card.css('a.view-catalog-btn::attr(href)').get() or "").strip()
- catalog_url = catalog_url.split("#")[0]
- cover_img = (card.css('.slider-for img::attr(src)').get() or "").strip()
- start_dt = _parse_datetime(start_raw)
- end_dt = _parse_datetime(end_raw)
- auctions.append({
- "event_id": event_id,
- "auction_name": auction_name,
- "auction_type": auction_type,
- "event_status": card.attrib.get("data-event_status", "").strip(),
- "start_time_raw": re.sub(r'^\s*Start\s*:\s*', '', start_raw, flags=re.I).strip(),
- "end_time_raw": re.sub(r'^\s*End\s*:\s*', '', end_raw, flags=re.I).strip(),
- "start_time": start_dt.strftime("%Y-%m-%d %H:%M:%S") if start_dt else None,
- "end_time": end_dt.strftime("%Y-%m-%d %H:%M:%S") if end_dt else None,
- "auction_url": urljoin(BASE_URL, auction_url) if auction_url else None,
- "catalog_url": urljoin(BASE_URL, catalog_url) if catalog_url else None,
- "cover_img": cover_img or None,
- })
- return auctions
- @retry(stop=stop_after_attempt(3), wait=wait_fixed(2), after=after_log)
- def get_auction_list(log, session, impersonate, only_first_page=False):
- """翻页抓取 /auctions/past 全部(或仅首页)拍卖会。
- Args:
- log: logger 对象。
- session (requests.Session): curl_cffi 会话对象。
- impersonate (str): 浏览器指纹标识。
- only_first_page (bool, optional): True 仅取首页(增量用,最近场次在首页顶部);
- False 翻完全部页(全量用)。Defaults to False。
- Returns:
- list[dict]: 拍卖会 dict 列表,已按 event_id 去重。
- """
- scope = "首页" if only_first_page else "全部"
- log.info(f"获取{scope}拍卖会列表")
- result, seen = [], set()
- page = 1
- while True:
- url = f"{PAST_URL}?page={page}"
- sel = _get_selector(log, session, impersonate, url)
- page_auctions = parse_auction_list(sel)
- if not page_auctions:
- break # 空页 = 已翻过末页
- new = [a for a in page_auctions if a["event_id"] not in seen]
- for a in new:
- seen.add(a["event_id"])
- result.append(a)
- if not new:
- break # 本页全重复(越界回卷兜底),停止
- log.info(f" 第 {page} 页解析到 {len(page_auctions)} 场(新增 {len(new)},累计 {len(result)})")
- if only_first_page:
- break
- page += 1
- log.info(f"共解析到 {len(result)} 场拍卖会")
- return result
- def _clean_price(text):
- """把成交价文本清洗成纯数字字符串。
- Args:
- text (str): 原始价格文本,如 "$90,000"。
- Returns:
- str: 去掉 $ 和千分位逗号后的数字串,如 "90000";无价格返回空串。
- """
- if not text:
- return ""
- m = re.search(r'\$?\s*([\d,]+)', text)
- return m.group(1).replace(",", "") if m else ""
- def parse_lot_cards(selector, auction):
- """解析图录页单页的 lot 卡片。
- Args:
- selector (Selector): 图录页某一页 GET 响应的 parsel 解析对象。
- auction (dict): 当前场次 dict(含 event_id / auction_name),回填到每条 lot。
- Returns:
- list[dict]: 每条 lot 一个 dict,字段:item_id, event_id, auction_name,
- lot_number, title, detail_url, list_img, sold_status, sold_for, bids;
- 无卡片返回空列表。
- """
- rows = []
- for card in selector.css('div[id^="stl-"][data-item_id]'):
- item_id = card.attrib.get("data-item_id", "").strip()
- if not item_id:
- continue
- title_a = card.css('.lot_title a')
- title = (title_a.xpath('normalize-space(.)').get() or "").strip()
- detail_url = (title_a.attrib.get("href") or "").strip()
- # Lot 号:"Lot 1" → "1"
- lot_raw = (card.css('.lot_Num::text').get() or "").strip()
- lot_number = (re.search(r'(\d+)', lot_raw) or [None, ""])[1] if lot_raw else ""
- list_img = (card.css('.catalog_img img::attr(src)').get() or "").strip()
- # 底部区:"9 BidsSold for$90,000" / "24 BidsUnsold"
- bottom = card.css('.item-list-bottom')
- bottom_text = (bottom.xpath('normalize-space(.)').get() or "") if bottom else ""
- bids_m = re.search(r'(\d+)\s*Bids?', bottom_text, re.I)
- bids = bids_m.group(1) if bids_m else ""
- if re.search(r'Unsold', bottom_text, re.I):
- sold_status, sold_for = "unsold", ""
- elif re.search(r'Sold\s*for', bottom_text, re.I):
- sold_status = "sold"
- sold_for = _clean_price(re.sub(r'.*Sold\s*for', '', bottom_text, flags=re.I))
- else:
- sold_status, sold_for = "", ""
- rows.append({
- "item_id": item_id,
- "event_id": auction["event_id"],
- "auction_name": auction.get("auction_name"),
- "lot_number": lot_number,
- "title": title,
- "detail_url": urljoin(BASE_URL, detail_url) if detail_url else None,
- "list_img": list_img or None,
- "sold_status": sold_status or None,
- "sold_for": sold_for or None,
- "bids": bids or None,
- })
- return rows
- @retry(stop=stop_after_attempt(5), wait=wait_fixed(2), after=after_log)
- def _fetch_lot_page(log, session, impersonate, catalog_url, page):
- """抓取图录页的某一页 lot 卡片。
- Args:
- log: logger 对象。
- session (requests.Session): curl_cffi 会话对象。
- impersonate (str): 浏览器指纹标识。
- catalog_url (str): 图录页绝对 URL(不含 query)。
- page (int): 页码,从 1 开始。
- Returns:
- Selector: 该页响应的 parsel 解析对象。
- """
- url = f"{catalog_url}?page={page}&limit={CATALOG_PAGE_SIZE}"
- return _get_selector(log, session, impersonate, url)
- def fetch_auction_lots(log, session, impersonate, auction):
- """抓取一个场次下的全部 lot 列表(翻页,去重 item_id)。
- Args:
- log: logger 对象。
- session (requests.Session): curl_cffi 会话对象。
- impersonate (str): 浏览器指纹标识。
- auction (dict): 场次 dict(含 event_id / auction_name / catalog_url)。
- Returns:
- list[dict]: 该场次全部 lot 列表(阶段一字段,不含详情)。
- """
- catalog_url = auction.get("catalog_url")
- if not catalog_url:
- log.warning(f"场次 {auction['event_id']} 无 catalog_url,跳过")
- return []
- log.info(f"开始抓取场次 {auction['event_id']}({auction.get('auction_name')}) 的 lot 列表")
- all_lots, seen = [], set()
- page = 1
- while True:
- sel = _fetch_lot_page(log, session, impersonate, catalog_url, page)
- lots = parse_lot_cards(sel, auction)
- if not lots:
- break # 空页 = 已翻过末页
- new = [x for x in lots if x["item_id"] not in seen]
- for x in new:
- seen.add(x["item_id"])
- all_lots.extend(new)
- log.info(f" 第 {page} 页 {len(lots)} 条(新增 {len(new)},累计 {len(all_lots)})")
- if not new:
- break # 本页全重复,停止
- page += 1
- log.info(f"场次 {auction['event_id']} 共抓 {len(all_lots)} 条 lot")
- return all_lots
- def parse_lot_detail(selector):
- """解析 lot 详情页,抽取标题、成交价、属性、多图。
- 详情页属性区(.item-attributes)字段数量不固定:不同拍品可能是 2~5 个,
- Athlete/Team 等常缺失。故先把所有 li 的 label→span 建成字典,再按需取字段,
- 缺失字段返回 None,保证解析健壮。
- Args:
- selector (Selector): 详情页 GET 响应的 parsel 解析对象。
- Returns:
- dict: {
- "title": 标题, "sold_for": 成交价(纯数字串,流拍为空),
- "sold_status": "sold"/"unsold", "bids": 出价数,
- "collection", "sport", "item_type", "athlete", "team": 属性(缺失为 None),
- "imgs": 多图 URL 逗号拼接
- }。
- """
- # 标题:og:title 最干净(页面首个 h1 是拍卖会名,非拍品标题)
- title = (selector.css('meta[property="og:title"]::attr(content)').get() or "").strip()
- if not title:
- # 兜底:取非拍卖会名的那个 h1
- h1s = [t.strip() for t in selector.css('h1 ::text, h1::text').getall() if t.strip()]
- title = h1s[1] if len(h1s) > 1 else (h1s[0] if h1s else "")
- # 成交价:.bidding-price 内首个 $金额;流拍则为空
- price_text = selector.css('.bidding-price').xpath('normalize-space(.)').get() or ""
- sold_for = _clean_price(price_text)
- sold_status = "sold" if sold_for else "unsold"
- # 出价数:页面 "[9 Bids]"
- body_text = selector.xpath('normalize-space(//body)').get() or ""
- bids_m = re.search(r'\[?\s*(\d+)\s*Bids?\s*\]?', body_text, re.I)
- bids = bids_m.group(1) if bids_m else ""
- # 属性:.item-attributes li -> {label: value},字段数量不固定,按需取
- attrs = {}
- for li in selector.css('.item-attributes li'):
- label = (li.css('label::text').get() or "").strip().rstrip(":")
- value = (li.css('span').xpath('normalize-space(.)').get() or "").strip()
- if label:
- attrs[label.lower()] = value
- # 多图:/item/{l|s}/... 统一归一化到 large 尺寸并去重(保持出现顺序)
- imgs, seen = [], set()
- for u in selector.re(IMG_ITEM_RE):
- large = u.replace("/item/s/", "/item/l/")
- if large not in seen:
- seen.add(large)
- imgs.append(large)
- return {
- "title": title,
- "sold_for": sold_for or None,
- "sold_status": sold_status,
- "bids": bids or None,
- "collection": attrs.get("collection") or None,
- "sport": attrs.get("sport") or None,
- "item_type": attrs.get("type") or None,
- "athlete": attrs.get("athlete") or None,
- "team": attrs.get("team") or None,
- "imgs": ",".join(imgs) if imgs else None,
- }
- @retry(stop=stop_after_attempt(3), wait=wait_fixed(2), after=after_log,
- retry=retry_if_not_exception_type(LotNotFound))
- def fetch_lot_detail(log, session, impersonate, detail_url):
- """GET lot 详情页并解析。
- Args:
- log: logger 对象。
- session (requests.Session): curl_cffi 会话对象。
- impersonate (str): 浏览器指纹标识。
- detail_url (str): 详情页绝对 URL。
- Returns:
- dict: parse_lot_detail 的返回结果。
- """
- log.debug(f"获取详情 {detail_url}")
- sel = _get_selector(log, session, impersonate, detail_url)
- return parse_lot_detail(sel)
- def save_auction(log, sql_pool, auction):
- """把一场拍卖会写入 scp_auction_record(幂等,靠 event_id 唯一索引去重)。
- Args:
- log: logger 对象。
- sql_pool: MySQL 连接池;传 None 时不入库。
- auction (dict): parse_auction_list 产出的场次 dict。
- Returns:
- None: 仅入库,无返回值。
- """
- if sql_pool is None:
- return
- row = {k: auction.get(k) for k in (
- "event_id", "auction_name", "auction_type", "event_status",
- "start_time_raw", "end_time_raw", "start_time", "end_time",
- "auction_url", "catalog_url", "cover_img",
- )}
- sql_pool.insert_many(table="scp_auction_record", data_list=[row], ignore=True)
- def crawl_one_auction(log, sql_pool, session, impersonate, auction):
- """抓取单个场次的全部 lot 列表并入库(阶段一:只抓列表,不进详情页)。
- 两阶段设计:本函数只负责 lot 列表入库(state 默认 0),详情字段由后续
- update_details_for_pending 扫库 state != 1 的记录单独补抓。抓完把
- scp_auction_record.lots_state 置 1,标记该场列表已抓。
- Args:
- log: logger 对象。
- sql_pool: MySQL 连接池;传 None 时只返回数据,不入库。
- session (requests.Session): curl_cffi 会话对象。
- impersonate (str): 浏览器指纹标识。
- auction (dict): 场次 dict(含 event_id / auction_name / catalog_url)。
- Returns:
- list[dict]: 该场次全部 lot 列表数据(阶段一字段,不含详情)。
- """
- lots = fetch_auction_lots(log, session, impersonate, auction)
- if sql_pool is not None:
- if lots:
- sql_pool.insert_many(table="scp_lot_record", data_list=lots, ignore=True)
- # 标记该场 lot 列表已抓
- sql_pool.update_one_or_dict(
- table="scp_auction_record",
- data={"lots_state": 1},
- condition={"event_id": auction["event_id"]},
- )
- log.info(f"场次 {auction['event_id']}({auction.get('auction_name')}) 共抓 {len(lots)} 条 lot")
- return lots
- def get_details(log, detail_url, sql_pool, sql_id):
- """对单条已入库 lot 补抓详情(阶段二),写回 scp_lot_record。
- Args:
- log: logger 对象。
- detail_url (str): 详情页 URL。
- sql_pool: MySQL 连接池。
- sql_id: 数据库记录 id。
- Returns:
- None: 仅入库,无返回值。
- """
- log.info(f">>> 补抓详情 {detail_url}")
- impersonate = random.choice(client_identifier_list)
- with requests.Session() as session:
- detail = fetch_lot_detail(log, session, impersonate, detail_url)
- # 详情字段 + 标记已抓;成交价/状态/出价以详情页为准(覆盖列表阶段的值)
- data = {**detail, "state": 1}
- sql_pool.update_one_or_dict(
- table="scp_lot_record",
- data=data,
- condition={"id": sql_id},
- )
- def update_details_for_pending(log, sql_pool):
- """扫库里待抓 / 瞬时失败的 lot,逐条补抓详情。
- state 状态机:0 待抓,1 已抓成功,2 瞬时失败(可重试),3 页面不存在(404,永久跳过)。
- 只选 state in (0, 2)——即跳过已成功(1)与永久不存在(3),避免 404 记录每轮死循环重试。
- Args:
- log: logger 对象。
- sql_pool: MySQL 连接池。
- Returns:
- None: 仅入库,无返回值。
- """
- log.debug("Updating detail pages ...")
- rows = sql_pool.select_all(
- "select id, detail_url from scp_lot_record where state in (0, 2)"
- )
- for row in rows:
- sql_id, detail_url = row[0], row[1]
- try:
- get_details(log, detail_url, sql_pool, sql_id)
- except LotNotFound as e:
- # 页面已下架/不存在:置终态 3,不再进入补抓队列
- log.warning(f"lot 页面不存在,标记 state=3 不再重试: {detail_url} ({e})")
- sql_pool.update_one_or_dict(
- table="scp_lot_record",
- data={"state": 3},
- condition={"id": sql_id},
- )
- except Exception as e:
- # 瞬时错误(超时/5xx/代理抖动):置 2,下轮再试
- log.error(f"Error getting details for {detail_url}: {e}")
- sql_pool.update_one_or_dict(
- table="scp_lot_record",
- data={"state": 2},
- condition={"id": sql_id},
- )
|