Add maxlaw PC spider and shared proxy limiter

common_sites/dls.py

@@ -24,7 +24,7 @@ from request.proxy_config import get_proxies, report_proxy_status
 urllib3.disable_warnings(urllib3.exceptions.InsecureRequestWarning)
 
 from Db import Db
-from utils.rate_limiter import wait_for_request
+from utils.rate_limiter import request_slot
 
 DOMAIN = "大律师"
 LIST_TEMPLATE = "https://m.maxlaw.cn/law/{pinyin}?page={page}"
@@ -108,17 +108,16 @@ class DlsSpider:
 
     def _get(self, url: str, max_retries: int = 3, headers: Optional[Dict[str, str]] = None) -> Optional[str]:
         """发送 GET 请求，带重试机制"""
-        wait_for_request()
-        
         for attempt in range(max_retries):
             try:
                 # 使用更长的超时时间，分别设置连接和读取超时
-                resp = self.session.get(
-                    url, 
-                    timeout=(10, 30),  # (connect_timeout, read_timeout)
-                    verify=False,
-                    headers=headers,
-                )
+                with request_slot():
+                    resp = self.session.get(
+                        url,
+                        timeout=(10, 30),  # (connect_timeout, read_timeout)
+                        verify=False,
+                        headers=headers,
+                    )
                 status_code = resp.status_code
                 content = resp.text
                 resp.close()

common_sites/dls_fresh.py

@@ -22,7 +22,7 @@ if project_root not in sys.path:
     sys.path.append(project_root)
 
 from request.requests_client import RequestClientError, RequestsClient
-from utils.rate_limiter import wait_for_request
+from utils.rate_limiter import request_slot
 from Db import Db
 
 urllib3.disable_warnings(urllib3.exceptions.InsecureRequestWarning)
@@ -107,9 +107,9 @@ class DlsFreshCrawler:
     def _get_text(self, url: str, timeout: int = 20, max_retries: int = 3) -> str:
         last_error: Optional[Exception] = None
         for attempt in range(max_retries):
-            wait_for_request()
             try:
-                resp = self.client.get_text(url, timeout=timeout, verify=False)
+                with request_slot():
+                    resp = self.client.get_text(url, timeout=timeout, verify=False)
                 code = resp.status_code
                 if code == 403:
                     if attempt < max_retries - 1:

common_sites/dls_pc.py

@@ -0,0 +1,438 @@
+import json
+import os
+import random
+import re
+import sys
+import time
+from typing import Dict, List, Optional, Set, Tuple
+from urllib.parse import urljoin
+
+current_dir = os.path.dirname(os.path.abspath(__file__))
+project_root = os.path.dirname(current_dir)
+request_dir = os.path.join(project_root, "request")
+if request_dir not in sys.path:
+    sys.path.insert(0, request_dir)
+if project_root not in sys.path:
+    sys.path.append(project_root)
+
+import requests
+from requests.adapters import HTTPAdapter
+from urllib3.util.retry import Retry
+import urllib3
+from bs4 import BeautifulSoup
+
+from request.proxy_config import get_proxies, report_proxy_status
+from utils.rate_limiter import request_slot
+from Db import Db
+
+urllib3.disable_warnings(urllib3.exceptions.InsecureRequestWarning)
+
+DOMAIN = "大律师"
+SITE_BASE = "https://www.maxlaw.cn"
+LIST_URL_TEMPLATE = SITE_BASE + "/law/{pinyin}?page={page}"
+PROVINCE_API = "https://js.maxlaw.cn/js/ajax/common/getprovice.js"
+CITY_API_TEMPLATE = "https://js.maxlaw.cn/js/ajax/common/getcity_{province_id}.js"
+
+PHONE_RE = re.compile(r"1[3-9]\d{9}")
+REPLY_RE = re.compile(r"已回复[:：]?\s*(\d+)")
+AREA_PREFIX_RE = re.compile(r"^[A-Za-z]\s*")
+
+
+def normalize_phone(text: str) -> str:
+    compact = re.sub(r"\D", "", text or "")
+    match = PHONE_RE.search(compact)
+    return match.group(0) if match else ""
+
+
+def clean_area_name(text: str) -> str:
+    value = AREA_PREFIX_RE.sub("", (text or "").strip())
+    return value.strip()
+
+
+def normalize_region_text(text: str) -> str:
+    value = (text or "").strip()
+    value = value.replace("\xa0", " ")
+    value = value.replace("－", "-").replace("—", "-").replace("–", "-")
+    value = re.sub(r"\s*-\s*", "-", value)
+    value = re.sub(r"\s+", "", value)
+    return value
+
+
+class DlsPcSpider:
+    def __init__(self, db_connection):
+        self.db = db_connection
+        self.session = self._build_session()
+        self.max_pages = int(os.getenv("MAXLAW_PC_MAX_PAGES", "100"))
+        self.areas = self._load_areas()
+
+    def _build_session(self) -> requests.Session:
+        report_proxy_status()
+        session = requests.Session()
+        session.trust_env = False
+        proxies = get_proxies()
+        if proxies:
+            session.proxies.update(proxies)
+        else:
+            session.proxies.clear()
+
+        retries = Retry(
+            total=3,
+            backoff_factor=1,
+            status_forcelist=(429, 500, 502, 503, 504),
+            allowed_methods=frozenset(["GET"]),
+            raise_on_status=False,
+        )
+        adapter = HTTPAdapter(max_retries=retries)
+        session.mount("https://", adapter)
+        session.mount("http://", adapter)
+        session.headers.update({
+            "User-Agent": (
+                "Mozilla/5.0 (Windows NT 10.0; Win64; x64) "
+                "AppleWebKit/537.36 (KHTML, like Gecko) "
+                "Chrome/136.0.0.0 Safari/537.36"
+            ),
+            "Accept": "text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8",
+            "Accept-Language": "zh-CN,zh;q=0.9,en;q=0.8",
+            "Connection": "close",
+        })
+        return session
+
+    def _refresh_session(self) -> None:
+        try:
+            self.session.close()
+        except Exception:
+            pass
+        self.session = self._build_session()
+
+    def _get(self, url: str, max_retries: int = 3, headers: Optional[Dict[str, str]] = None) -> Optional[str]:
+        for attempt in range(max_retries):
+            try:
+                with request_slot():
+                    resp = self.session.get(url, timeout=(10, 25), verify=False, headers=headers)
+                status_code = resp.status_code
+                text = resp.text
+                resp.close()
+                if status_code == 403:
+                    if attempt < max_retries - 1:
+                        wait_time = 2 ** attempt + random.uniform(0.3, 1.0)
+                        print(f"403被拦截，{wait_time:.1f}秒后重试 ({attempt + 1}/{max_retries}): {url}")
+                        self._refresh_session()
+                        time.sleep(wait_time)
+                        continue
+                    print(f"请求失败 {url}: 403 Forbidden")
+                    return None
+                if status_code >= 400:
+                    raise requests.exceptions.HTTPError(f"{status_code} Error: {url}")
+                return text
+            except requests.exceptions.RequestException as exc:
+                if attempt < max_retries - 1:
+                    wait_time = 2 ** attempt + random.uniform(0.3, 1.0)
+                    print(f"请求失败，{wait_time:.1f}秒后重试 ({attempt + 1}/{max_retries}): {url} -> {exc}")
+                    time.sleep(wait_time)
+                    continue
+                print(f"请求失败 {url}: {exc}")
+                return None
+        return None
+
+    def _get_json(self, url: str) -> Optional[Dict]:
+        text = self._get(url)
+        if not text:
+            return None
+        try:
+            return json.loads(text.strip().lstrip("\ufeff"))
+        except ValueError as exc:
+            print(f"解析JSON失败 {url}: {exc}")
+            return None
+
+    def _load_areas(self) -> List[Dict[str, str]]:
+        areas = self._load_areas_from_site()
+        if areas:
+            print(f"[大律师PC] 地区来源: site, 地区数: {len(areas)}")
+            return areas
+
+        areas = self._load_areas_from_db()
+        if areas:
+            print(f"[大律师PC] 地区来源: db, 地区数: {len(areas)}")
+            return areas
+
+        print("[大律师PC] 无地区数据")
+        return []
+
+    def _load_areas_from_site(self) -> List[Dict[str, str]]:
+        data = self._get_json(PROVINCE_API)
+        if not data or str(data.get("status")) != "1":
+            return []
+
+        result: List[Dict[str, str]] = []
+        seen_pinyin: Set[str] = set()
+
+        for province in data.get("ds", []) or []:
+            province_id = province.get("id")
+            province_name = clean_area_name(province.get("name", ""))
+            province_pinyin = (province.get("py_code") or "").strip()
+
+            city_rows = []
+            if province_id:
+                city_data = self._get_json(CITY_API_TEMPLATE.format(province_id=province_id))
+                if city_data and str(city_data.get("status")) == "1":
+                    city_rows = city_data.get("ds", []) or []
+
+            if not city_rows and province_pinyin and province_pinyin not in seen_pinyin:
+                seen_pinyin.add(province_pinyin)
+                result.append({
+                    "province": province_name,
+                    "city": province_name,
+                    "pinyin": province_pinyin,
+                })
+                continue
+
+            for city in city_rows:
+                city_name = clean_area_name(city.get("name", ""))
+                city_pinyin = (city.get("py_code") or "").strip()
+                if not city_pinyin or city_pinyin in seen_pinyin:
+                    continue
+                seen_pinyin.add(city_pinyin)
+                result.append({
+                    "province": province_name,
+                    "city": city_name or province_name,
+                    "pinyin": city_pinyin,
+                })
+
+        return result
+
+    def _load_areas_from_db(self) -> List[Dict[str, str]]:
+        tables = ("area_new", "area", "area2")
+        last_error = None
+        for table in tables:
+            try:
+                rows = self.db.select_data(
+                    table,
+                    "province, city, pinyin",
+                    "domain='maxlaw' AND level=2",
+                ) or []
+            except Exception as exc:
+                last_error = exc
+                continue
+
+            if rows:
+                return rows
+
+        if last_error:
+            print(f"[大律师PC] 加载数据库地区失败: {last_error}")
+        return []
+
+    def _existing_phones(self, phones: List[str]) -> Set[str]:
+        if not phones:
+            return set()
+        existing: Set[str] = set()
+        cur = self.db.db.cursor()
+        try:
+            chunk_size = 500
+            for i in range(0, len(phones), chunk_size):
+                chunk = phones[i:i + chunk_size]
+                placeholders = ",".join(["%s"] * len(chunk))
+                sql = f"SELECT phone FROM lawyer WHERE domain=%s AND phone IN ({placeholders})"
+                cur.execute(sql, [DOMAIN, *chunk])
+                for row in cur.fetchall():
+                    existing.add(row[0])
+        finally:
+            cur.close()
+        return existing
+
+    def _build_list_url(self, pinyin: str, page: int) -> str:
+        return LIST_URL_TEMPLATE.format(pinyin=pinyin, page=page)
+
+    def _parse_location_line(
+        self,
+        text: str,
+        fallback_province: str,
+        fallback_city: str,
+    ) -> Tuple[str, str, str]:
+        raw = (text or "").replace("\xa0", " ")
+        raw = re.sub(r"\s+", " ", raw).strip()
+        if not raw:
+            return fallback_province, fallback_city or fallback_province, ""
+
+        parts = raw.split(" ", 1)
+        area_text = parts[0].strip()
+        law_firm = parts[1].strip() if len(parts) > 1 else ""
+
+        province = fallback_province
+        city = fallback_city or fallback_province
+        if "-" in area_text:
+            area_parts = [item.strip() for item in area_text.split("-", 1)]
+            if area_parts[0]:
+                province = area_parts[0]
+            if len(area_parts) > 1 and area_parts[1]:
+                city = area_parts[1]
+        elif area_text:
+            province = area_text
+            city = area_text
+
+        return province, city, law_firm
+
+    def _extract_page_region(self, soup: BeautifulSoup) -> str:
+        button = soup.select_one(".filter .filter-btn")
+        if button:
+            return normalize_region_text(button.get_text(" ", strip=True))
+        title = soup.select_one(".findLawyer-title h1")
+        if title:
+            return normalize_region_text(title.get_text(strip=True).replace("律师", ""))
+        return ""
+
+    def _page_matches_area(self, soup: BeautifulSoup, province: str, city: str) -> Tuple[bool, str]:
+        current_region = self._extract_page_region(soup)
+        if not current_region:
+            return True, current_region
+        if "全国" in current_region:
+            return False, current_region
+
+        norm_province = normalize_region_text(province)
+        norm_city = normalize_region_text(city or province)
+
+        if norm_city and norm_city != norm_province:
+            matched = norm_province in current_region and norm_city in current_region
+        else:
+            matched = norm_province in current_region
+
+        if matched:
+            return True, current_region
+
+        title = soup.select_one(".findLawyer-title h1")
+        title_text = ""
+        if title:
+            title_text = normalize_region_text(title.get_text(strip=True).replace("律师", ""))
+
+        if norm_city and norm_city != norm_province:
+            matched = norm_city in title_text
+        else:
+            matched = norm_province in title_text
+
+        return matched, current_region or title_text
+
+    def _parse_list(self, html: str, province: str, city: str, list_url: str, area_pinyin: str) -> Tuple[bool, int, int]:
+        soup = BeautifulSoup(html, "html.parser")
+        matched, current_region = self._page_matches_area(soup, province, city)
+        if not matched:
+            print(f"  页面地区不匹配，停止分页: 目标={province}-{city} 当前={current_region or '未知'}")
+            return False, 0, 0
+
+        cards = []
+        seen_page_phone: Set[str] = set()
+
+        for item in soup.select("ul.findLawyer-list > li.clearfix"):
+            name_link = item.select_one(".findLawyer-list-detail-name a[href]")
+            phone_tag = item.select_one(".findLawyer-list-detail-name span")
+            if not name_link or not phone_tag:
+                continue
+
+            phone = normalize_phone(phone_tag.get_text(" ", strip=True))
+            if not phone or phone in seen_page_phone:
+                continue
+            seen_page_phone.add(phone)
+
+            name = name_link.get_text(strip=True)
+            detail_url = urljoin(SITE_BASE, name_link.get("href", "").strip())
+
+            location_tag = item.select_one(".findLawyer-list-detail-the")
+            card_province, card_city, law_firm = self._parse_location_line(
+                location_tag.get_text(" ", strip=True) if location_tag else "",
+                province,
+                city,
+            )
+
+            specialties = []
+            for dd in item.select(".findLawyer-list-detail-fields dd"):
+                text = dd.get_text(strip=True)
+                if text:
+                    specialties.append(text)
+
+            reply_count = None
+            reply_tag = item.select_one(".findLawyer-list-detail-other a")
+            if reply_tag:
+                match = REPLY_RE.search(reply_tag.get_text(" ", strip=True))
+                if match:
+                    reply_count = int(match.group(1))
+
+            cards.append({
+                "name": name,
+                "law_firm": law_firm,
+                "province": card_province or province,
+                "city": card_city or city or province,
+                "phone": phone,
+                "url": detail_url,
+                "domain": DOMAIN,
+                "create_time": int(time.time()),
+                "params": json.dumps({
+                    "area_pinyin": area_pinyin,
+                    "source": list_url,
+                    "specialties": specialties,
+                    "reply_count": reply_count,
+                }, ensure_ascii=False),
+            })
+
+        if not cards:
+            return True, 0, 0
+
+        phones = [item["phone"] for item in cards if item.get("phone")]
+        existing = self._existing_phones(phones)
+        inserted = 0
+
+        for item in cards:
+            phone = item.get("phone")
+            if not phone:
+                continue
+            if phone in existing:
+                print(f"    -- 已存在: {item['name']} ({phone})")
+                continue
+            try:
+                self.db.insert_data("lawyer", item)
+                inserted += 1
+                print(f"    -> 新增: {item['name']} ({phone})")
+            except Exception as exc:
+                print(f"    插入失败 {item.get('url')}: {exc}")
+
+        return True, inserted, len(cards)
+
+    def run(self):
+        print("启动大律师 PC 站采集...")
+        if not self.areas:
+            print("无地区数据")
+            return
+
+        for area in self.areas:
+            province = (area.get("province") or "").strip()
+            city = (area.get("city") or province).strip()
+            pinyin = (area.get("pinyin") or "").strip()
+            if not province or not pinyin:
+                continue
+
+            area_label = province if not city or city == province else f"{province}-{city}"
+            print(f"采集地区: {area_label} ({pinyin})")
+
+            for page in range(1, self.max_pages + 1):
+                list_url = self._build_list_url(pinyin, page)
+                print(f"  第 {page} 页: {list_url}")
+                html = self._get(list_url, headers={"Referer": SITE_BASE + "/law"})
+                if not html:
+                    break
+
+                page_ok, inserted, parsed_count = self._parse_list(html, province, city, list_url, pinyin)
+                if not page_ok:
+                    break
+                if parsed_count == 0:
+                    print("  当前页无律师卡片，停止")
+                    break
+
+                if inserted == 0:
+                    print("  当前页无新增数据")
+
+                time.sleep(0.5)
+
+        print("大律师 PC 站采集完成")
+
+
+if __name__ == "__main__":
+    with Db() as db:
+        spider = DlsPcSpider(db)
+        spider.run()

common_sites/findlaw.py

@@ -16,6 +16,7 @@ if project_root not in sys.path:
 import requests
 from request.proxy_config import get_proxies, report_proxy_status
 from Db import Db
+from utils.rate_limiter import request_slot
 
 DOMAIN = "找法网"
 LIST_TEMPLATE = "https://m.findlaw.cn/{pinyin}/q_lawyer/p{page}?ajax=1&order=0&sex=-1"
@@ -59,7 +60,8 @@ class FindlawSpider:
         headers = {"Referer": referer}
         for attempt in range(max_retries):
             try:
-                resp = self.session.get(url, timeout=15, verify=verify, headers=headers)
+                with request_slot():
+                    resp = self.session.get(url, timeout=15, verify=verify, headers=headers)
                 status_code = resp.status_code
                 text = resp.text
                 resp.close()

common_sites/hualv.py

@@ -20,6 +20,7 @@ from request.proxy_config import get_proxies, report_proxy_status
 
 from Db import Db
 from config import HEADERS
+from utils.rate_limiter import request_slot
 
 LIST_URL = "https://m.66law.cn/findlawyer/rpc/loadlawyerlist/"
 DOMAIN = "华律"
@@ -100,7 +101,8 @@ class HualvSpider:
     def _post(self, data: Dict[str, str], max_retries: int = 3) -> Optional[Dict]:
         for attempt in range(max_retries):
             try:
-                resp = self.session.post(LIST_URL, data=data, timeout=20, verify=False)
+                with request_slot():
+                    resp = self.session.post(LIST_URL, data=data, timeout=20, verify=False)
                 status_code = resp.status_code
                 text = resp.text
                 resp.close()
@@ -272,7 +274,8 @@ class HualvSpider:
     def _get_detail(self, url: str, max_retries: int = 3) -> Optional[str]:
         for attempt in range(max_retries):
             try:
-                resp = self.session.get(url, timeout=15, verify=False)
+                with request_slot():
+                    resp = self.session.get(url, timeout=15, verify=False)
                 status_code = resp.status_code
                 text = resp.text
                 resp.close()

common_sites/lawtime.py

@@ -26,6 +26,7 @@ urllib3.disable_warnings(urllib3.exceptions.InsecureRequestWarning)
 
 from Db import Db
 from config import LAWTIME_CONFIG
+from utils.rate_limiter import request_slot
 
 LIST_BASE = "https://m.lawtime.cn/{pinyin}/lawyer/?page={page}"
 DETAIL_BASE = "https://m.lawtime.cn"
@@ -123,7 +124,8 @@ class LawtimeSpider:
     def _get_with_session(self, session: requests.Session, url: str, max_retries: int = 3, is_thread: bool = False) -> Optional[str]:
         for attempt in range(max_retries):
             try:
-                resp = session.get(url, timeout=15, verify=False)
+                with request_slot():
+                    resp = session.get(url, timeout=15, verify=False)
                 status_code = resp.status_code
                 text = resp.text
                 resp.close()

common_sites/six4365.py

@@ -23,6 +23,7 @@ from request.proxy_config import get_proxies, report_proxy_status
 urllib3.disable_warnings(urllib3.exceptions.InsecureRequestWarning)
 
 from Db import Db
+from utils.rate_limiter import request_slot
 
 DOMAIN = "律图"
 LIST_URL = "https://m.64365.com/findLawyer/rpc/FindLawyer/LawyerRecommend/"
@@ -144,7 +145,8 @@ class Six4365Spider:
     def _post(self, payload: Dict[str, str], max_retries: int = 3) -> Optional[str]:
         for attempt in range(max_retries):
             try:
-                resp = self.session.post(LIST_URL, data=payload, timeout=10, verify=False)
+                with request_slot():
+                    resp = self.session.post(LIST_URL, data=payload, timeout=10, verify=False)
                 status_code = resp.status_code
                 text = resp.text
                 resp.close()
@@ -301,7 +303,8 @@ class Six4365Spider:
         session = self._get_thread_session()
         for attempt in range(max_retries):
             try:
-                resp = session.get(url, timeout=10, verify=False)
+                with request_slot():
+                    resp = session.get(url, timeout=10, verify=False)
                 status_code = resp.status_code
                 text = resp.text
                 resp.close()

common_sites/start.sh

@@ -5,9 +5,20 @@ set -euo pipefail
 cd "$(dirname "$0")"
 
 echo "使用 request/proxy_settings.json 读取代理配置"
+export PROXY_MAX_REQUESTS_PER_SECOND="${PROXY_MAX_REQUESTS_PER_SECOND:-5}"
 
-nohup python ../common_sites/dls.py > dls.log 2>&1 & # 大律师
-nohup python ../common_sites/findlaw.py > findlaw.log 2>&1 & # 找法网
-nohup python ../common_sites/lawtime.py > lawtime.log 2>&1 &  # 法律快车
-nohup python ../common_sites/six4365.py > six4365.log 2>&1 & # 律图
-nohup python ../common_sites/hualv.py > hualv.log 2>&1 & # 华律
+start_job() {
+  local script="$1"
+  local log_file="$2"
+  local label="$3"
+  nohup python "../common_sites/${script}" > "${log_file}" 2>&1 &
+  echo "启动 ${label}: ${script} -> ${log_file}"
+  sleep 1
+}
+
+start_job "dls.py" "dls.log" "大律师"
+start_job "dls_pc.py" "dls_pc.log" "大律师PC站"
+start_job "findlaw.py" "findlaw.log" "找法网"
+start_job "lawtime.py" "lawtime.log" "法律快车"
+start_job "six4365.py" "six4365.log" "律图"
+start_job "hualv.py" "hualv.log" "华律"

utils/rate_limiter.py

@@ -1,76 +1,191 @@
 """
 全局请求速率限制器
-确保代理每秒不超过5次请求
+
+默认按“所有爬虫进程共享一个桶”来限流，避免 `bash start.sh`
+同时启动多个进程时，每个进程各自 5 次/秒，叠加后把代理冲爆。
 """
+from contextlib import contextmanager
+import json
+import os
+import tempfile
 import time
 import threading
-from collections import deque
+from pathlib import Path
+from uuid import uuid4
+
+import fcntl
 
 
 class RateLimiter:
     """
-    令牌桶算法实现的速率限制器
+    基于文件锁的跨进程滑动窗口限流器。
+
+    - 同一台机器上的多个 Python 进程会共享同一个状态文件
+    - 同一个进程内的多个线程也会一起走这个限流器
     """
-    def __init__(self, max_requests_per_second: int = 5):
-        """
-        初始化速率限制器
-        
-        Args:
-            max_requests_per_second: 每秒最大请求数
-        """
-        self.max_requests = max_requests_per_second
-        self.requests = deque()
-        self.lock = threading.RLock()
-    
-    def acquire(self):
-        """
-        获取请求权限，如果需要则等待
-        """
-        with self.lock:
-            now = time.time()
-            
-            # 清理超过1秒的请求记录
-            while self.requests and now - self.requests[0] >= 1.0:
-                self.requests.popleft()
-            
-            # 如果当前请求数已达上限，等待
-            if len(self.requests) >= self.max_requests:
-                # 计算需要等待的时间
-                wait_time = 1.0 - (now - self.requests[0])
-                if wait_time > 0:
-                    time.sleep(wait_time)
-                    return self.acquire()  # 递归调用以重新检查
-            
-            # 记录这次请求
-            self.requests.append(now)
-    
+
+    def __init__(
+        self,
+        max_requests_per_second: int = 5,
+        window_seconds: float = 1.0,
+        state_file: str | None = None,
+    ):
+        self.max_requests = max(1, int(max_requests_per_second))
+        self.max_concurrent = max(
+            1,
+            int(os.getenv("PROXY_MAX_CONCURRENT_REQUESTS", str(self.max_requests))),
+        )
+        self.window_seconds = max(0.1, float(window_seconds))
+        self.lease_seconds = max(
+            5.0,
+            float(os.getenv("PROXY_REQUEST_LEASE_SECONDS", "120")),
+        )
+        default_state = os.path.join(
+            tempfile.gettempdir(),
+            "lawyers_proxy_rate_limiter.json",
+        )
+        self.state_file = Path(
+            state_file or os.getenv("PROXY_RATE_LIMIT_FILE", default_state)
+        )
+        self.lock_file = self.state_file.with_suffix(self.state_file.suffix + ".lock")
+        self._thread_lock = threading.RLock()
+        self.state_file.parent.mkdir(parents=True, exist_ok=True)
+        self.lock_file.parent.mkdir(parents=True, exist_ok=True)
+
+    def _load_state(self) -> dict:
+        if not self.state_file.exists():
+            return {"timestamps": [], "leases": {}}
+        try:
+            raw = self.state_file.read_text(encoding="utf-8").strip()
+            if not raw:
+                return {"timestamps": [], "leases": {}}
+            data = json.loads(raw)
+            if isinstance(data, list):
+                return {
+                    "timestamps": [float(item) for item in data],
+                    "leases": {},
+                }
+            if not isinstance(data, dict):
+                return {"timestamps": [], "leases": {}}
+            timestamps = data.get("timestamps", []) or []
+            leases = data.get("leases", {}) or {}
+            return {
+                "timestamps": [float(item) for item in timestamps],
+                "leases": {str(key): float(value) for key, value in leases.items()},
+            }
+        except Exception:
+            return {"timestamps": [], "leases": {}}
+
+    def _save_state(self, state: dict) -> None:
+        payload = json.dumps(state, ensure_ascii=False)
+        self.state_file.write_text(payload, encoding="utf-8")
+
+    def _normalize_state(self, state: dict, now: float) -> dict:
+        timestamps = [
+            float(ts)
+            for ts in (state.get("timestamps", []) or [])
+            if now - float(ts) < self.window_seconds
+        ]
+        leases = {
+            str(key): float(value)
+            for key, value in (state.get("leases", {}) or {}).items()
+            if now - float(value) < self.lease_seconds
+        }
+        return {"timestamps": timestamps, "leases": leases}
+
+    def acquire(self) -> None:
+        token = None
+        while True:
+            token = self.try_acquire_slot()
+            if token:
+                self.release(token)
+                return
+            time.sleep(0.05)
+
+    def try_acquire_slot(self) -> str | None:
+        while True:
+            wait_time = 0.0
+            with self._thread_lock:
+                with open(self.lock_file, "a+", encoding="utf-8") as lock_fp:
+                    fcntl.flock(lock_fp.fileno(), fcntl.LOCK_EX)
+                    now = time.time()
+                    state = self._normalize_state(self._load_state(), now)
+                    timestamps = state["timestamps"]
+                    leases = state["leases"]
+
+                    if len(timestamps) < self.max_requests and len(leases) < self.max_concurrent:
+                        token = uuid4().hex
+                        timestamps.append(now)
+                        leases[token] = now
+                        self._save_state(state)
+                        fcntl.flock(lock_fp.fileno(), fcntl.LOCK_UN)
+                        return token
+
+                    wait_candidates = []
+                    if len(timestamps) >= self.max_requests and timestamps:
+                        wait_candidates.append(self.window_seconds - (now - timestamps[0]))
+                    if len(leases) >= self.max_concurrent:
+                        wait_candidates.append(0.05)
+                    wait_time = max(0.05, min([item for item in wait_candidates if item > 0] or [0.05]))
+                    fcntl.flock(lock_fp.fileno(), fcntl.LOCK_UN)
+
+            time.sleep(wait_time)
+
+    def release(self, token: str | None) -> None:
+        if not token:
+            return
+        with self._thread_lock:
+            with open(self.lock_file, "a+", encoding="utf-8") as lock_fp:
+                fcntl.flock(lock_fp.fileno(), fcntl.LOCK_EX)
+                now = time.time()
+                state = self._normalize_state(self._load_state(), now)
+                leases = state["leases"]
+                if token in leases:
+                    leases.pop(token, None)
+                    self._save_state(state)
+                else:
+                    self._save_state(state)
+                fcntl.flock(lock_fp.fileno(), fcntl.LOCK_UN)
+
     def can_make_request(self) -> bool:
-        """
-        检查是否可以立即发起请求（非阻塞）
-        """
-        with self.lock:
-            now = time.time()
-            
-            # 清理超过1秒的请求记录
-            while self.requests and now - self.requests[0] >= 1.0:
-                self.requests.popleft()
-            
-            return len(self.requests) < self.max_requests
+        with self._thread_lock:
+            with open(self.lock_file, "a+", encoding="utf-8") as lock_fp:
+                fcntl.flock(lock_fp.fileno(), fcntl.LOCK_EX)
+                now = time.time()
+                state = self._normalize_state(self._load_state(), now)
+                self._save_state(state)
+                allowed = (
+                    len(state["timestamps"]) < self.max_requests
+                    and len(state["leases"]) < self.max_concurrent
+                )
+                fcntl.flock(lock_fp.fileno(), fcntl.LOCK_UN)
+                return allowed
 
 
-# 全局速率限制器实例
-global_rate_limiter = RateLimiter(max_requests_per_second=5)
+global_rate_limiter = RateLimiter(
+    max_requests_per_second=int(os.getenv("PROXY_MAX_REQUESTS_PER_SECOND", "5"))
+)
 
 
 def wait_for_request():
-    """
-    等待直到可以发起请求
-    """
+    """等待直到可以发起请求。"""
     global_rate_limiter.acquire()
 
 
 def can_request_now() -> bool:
-    """
-    检查是否可以立即发起请求
-    """
+    """检查是否可以立即发起请求。"""
     return global_rate_limiter.can_make_request()
+
+
+@contextmanager
+def request_slot():
+    """
+    申请一个跨进程共享的请求槽位，请求结束后自动释放。
+
+    这样既能限制“每秒启动多少请求”，也能限制“同时在飞多少请求”。
+    """
+    token = global_rate_limiter.try_acquire_slot()
+    try:
+        yield
+    finally:
+        global_rate_limiter.release(token)