0.0.4

spider/work.py

@@ -57,6 +57,34 @@ CA_AREA_CODES = {
 }
 
 
+# 主要城市的区号（更精确的城市级约束）
+CITY_AREA_CODES = {
+    "Calgary": ["403", "587", "825"],
+    "Edmonton": ["780", "587", "825"],
+    "Vancouver": ["604", "778", "236", "672"],
+    "Halifax": ["902", "782"],
+    "Toronto": ["416", "647", "437"],
+}
+
+
+# 邮编首字母合法性映射（按省份缩写）
+POSTAL_PREFIXES = {
+    "AB": {"T"},
+    "BC": {"V"},
+    "MB": {"R"},
+    "NB": {"E"},
+    "NL": {"A"},
+    "NS": {"B"},
+    "ON": {"K", "L", "M"},
+    "PE": {"C"},
+    "QC": {"G", "H", "J"},
+    "SK": {"S"},
+    "NT": {"X"},
+    "NU": {"X"},
+    "YT": {"Y"},
+}
+
+
 REMOTE_PROVINCES = {"NL", "NT", "NU", "YT"}
 
 
@@ -261,6 +289,46 @@ def _random_phone(province_abbr: str) -> str:
     return f"({area}) {exchange}-{line}"
 
 
+def _random_phone_city(province_abbr: str, city: Optional[str]) -> str:
+    """
+    按城市优先选择区号，若城市未配置则回退到省份区号
+
+    参数:
+        province_abbr (str): 省份缩写
+        city (Optional[str]): 城市名
+
+    返回值:
+        str: 电话，例如 "(403) 555-1234"
+    """
+    codes = None
+    if city:
+        codes = CITY_AREA_CODES.get(city)
+    codes = codes or CA_AREA_CODES.get(province_abbr, ["000"])
+    area = random.choice(codes)
+    exchange = str(random.randint(200, 899)).zfill(3)
+    line = str(random.randint(1000, 9999)).zfill(4)
+    return f"(#{area}) {exchange}-{line}".replace("#", "")
+
+
+def _postal_valid_for_province(province_abbr: str, postcode: str) -> bool:
+    """
+    校验邮编首字母是否符合省份规范
+
+    参数:
+        province_abbr (str): 省份缩写
+        postcode (str): 邮编字符串
+
+    返回值:
+        bool: 合法返回 True，否则 False
+    """
+    if not postcode:
+        return False
+    prefixes = POSTAL_PREFIXES.get(province_abbr)
+    if not prefixes:
+        return True
+    return postcode[0].upper() in prefixes
+
+
 def generate_canada_info(province: str, city: Optional[str] = None, max_attempts: int = 15, sleep_sec: float = 0.6) -> Dict[str, str]:
     """
     随机生成加拿大个人与地址信息，可指定省份（全称或缩写）与可选城市
@@ -289,14 +357,14 @@ def generate_canada_info(province: str, city: Optional[str] = None, max_attempts
         address_str = _format_address(addr, prov_abbr)
         if prov_abbr in REMOTE_PROVINCES:
             break
-        if addr.get("house_number") and (addr.get("road") or addr.get("residential") or addr.get("footway")) and city_name:
+        if addr.get("house_number") and (addr.get("road") or addr.get("residential") or addr.get("footway")) and city_name and _postal_valid_for_province(prov_abbr, postcode):
             break
         time.sleep(sleep_sec)
 
     firstname, lastname = _random_name()
     full_name = f"{firstname} {lastname}"
     birthday = _random_birthday()
-    phone = _random_phone(prov_abbr)
+    phone = _random_phone_city(prov_abbr, city or chosen_city)
 
     return {
         "firstname": firstname,