fix: bind knowledge maps to explicit regions

app/project_lifecycle.py

@@ -20,6 +20,7 @@ from app.config import settings
 SAFE_RESOURCE_ID = re.compile(r"^[A-Za-z0-9][A-Za-z0-9_.-]{0,127}$")
 SAFE_SCHEMA_IDENTIFIER = re.compile(r"^[A-Za-z_][A-Za-z0-9_]{0,127}$")
 SAFE_SCHEMA_VERSION = re.compile(r"^[A-Za-z0-9][A-Za-z0-9._-]{0,63}$")
+SAFE_ADCODE = re.compile(r"^\d{6}$")
 
 TYPE_ALIASES = {
     "str": "string",
@@ -51,6 +52,15 @@ SUPPORTED_FIELD_TYPES = {
 GRAPH_IMPORT_BATCH_THRESHOLD = 500
 GRAPH_IMPORT_BATCH_SIZE = 250
 
+SPATIAL_REGION_LEVELS = {"province", "city", "district"}
+MAP_ENTITY_TYPES = {
+    "FoodPlace",
+    "Hotel",
+    "ScenicSpot",
+    "TransitFacility",
+    "BusStop",
+}
+
 
 class ProjectValidationError(ValueError):
     """Raised when a provision payload is invalid before any write occurs."""
@@ -69,6 +79,48 @@ class GraphImportError(RuntimeError):
         self.cleanup_required = cleanup_error is not None
 
 
+def _adcode_belongs_to_region(adcode: str, region_adcode: str, level: str) -> bool:
+    if not SAFE_ADCODE.fullmatch(adcode) or not SAFE_ADCODE.fullmatch(region_adcode):
+        return False
+    if level == "province":
+        return adcode[:2] == region_adcode[:2]
+    if level == "city":
+        return adcode[:4] == region_adcode[:4]
+    return adcode == region_adcode
+
+
+def _region_names_match(value: Any, region_name: str) -> bool:
+    candidate = str(value or "").strip()
+    expected = str(region_name or "").strip()
+    return bool(candidate and expected and (candidate in expected or expected in candidate))
+
+
+def _is_spatial_map_node(node: Mapping[str, Any]) -> bool:
+    properties = node.get("properties")
+    if not isinstance(properties, Mapping):
+        return False
+    has_coordinates = (
+        properties.get("lng") is not None
+        and properties.get("lat") is not None
+    ) or (
+        properties.get("longitude") is not None
+        and properties.get("latitude") is not None
+    )
+    if not has_coordinates:
+        return False
+    labels = {
+        str(label)
+        for label in (node.get("labels") or [])
+        if str(label).strip()
+    }
+    labels.add(str(node.get("type") or ""))
+    return bool(
+        labels & MAP_ENTITY_TYPES
+        or str(properties.get("type_label") or "").strip()
+        or str(properties.get("place_type") or "").strip()
+    )
+
+
 def _as_mapping(value: Any) -> dict[str, Any]:
     return dict(value) if isinstance(value, Mapping) else {}
 
@@ -246,6 +298,20 @@ def normalize_provision_payload(body: Any) -> dict[str, Any]:
     map_scope = str(raw_spatial_map.get("scope") or "libo").strip().lower()
     if map_scope not in {"libo", "guizhou"}:
         errors.append("spatial_map.scope 仅支持 libo 或 guizhou")
+    map_region_name = str(raw_spatial_map.get("region_name") or "").strip()
+    map_region_adcode = str(raw_spatial_map.get("region_adcode") or "").strip()
+    map_region_level = str(raw_spatial_map.get("region_level") or "").strip().lower()
+    if map_enabled:
+        if not map_region_name:
+            errors.append("启用知识地图时必须选择地图区域")
+        if not SAFE_ADCODE.fullmatch(map_region_adcode):
+            errors.append("spatial_map.region_adcode 必须是 6 位行政区划代码")
+        if map_region_level not in SPATIAL_REGION_LEVELS:
+            errors.append("spatial_map.region_level 仅支持 province、city 或 district")
+        if SAFE_ADCODE.fullmatch(map_region_adcode):
+            expected_scope = "libo" if map_region_adcode == "522722" else "guizhou"
+            if map_scope != expected_scope:
+                errors.append("spatial_map.scope 与所选地图区域不一致")
 
     if not project_id:
         errors.append("项目英文标识不能为空")
@@ -399,6 +465,42 @@ def normalize_provision_payload(body: Any) -> dict[str, Any]:
             normalized_node["labels"] = normalized_labels
         nodes.append(normalized_node)
 
+    if map_enabled:
+        map_nodes = [node for node in nodes if _is_spatial_map_node(node)]
+        if not map_nodes:
+            errors.append("启用知识地图时，图谱数据至少需要一个带经纬度和业态的空间节点")
+        elif SAFE_ADCODE.fullmatch(map_region_adcode) and map_region_level in SPATIAL_REGION_LEVELS:
+            unverified_region = 0
+            mismatched_adcodes: set[str] = set()
+            for node in map_nodes:
+                properties = node["properties"]
+                adcode = str(properties.get("adcode") or "").strip()
+                if adcode and not _adcode_belongs_to_region(
+                    adcode,
+                    map_region_adcode,
+                    map_region_level,
+                ):
+                    mismatched_adcodes.add(adcode)
+                if adcode:
+                    continue
+                region_field = {
+                    "province": properties.get("province"),
+                    "city": properties.get("city"),
+                    "district": properties.get("district"),
+                }[map_region_level]
+                if not _region_names_match(region_field, map_region_name):
+                    unverified_region += 1
+            if unverified_region:
+                errors.append(
+                    f"知识地图空间节点有 {unverified_region} 条既无可用 adcode，"
+                    "也没有与所选区域匹配的省市区县字段"
+                )
+            if mismatched_adcodes:
+                examples = "、".join(sorted(mismatched_adcodes)[:5])
+                errors.append(
+                    f"知识地图数据与所选区域 {map_region_name} 不一致，发现行政区划代码：{examples}"
+                )
+
     relations: list[dict[str, Any]] = []
     for index, raw_relation in enumerate(raw_relations):
         path = f"graph_data.relations[{index}]"
@@ -462,7 +564,13 @@ def normalize_provision_payload(body: Any) -> dict[str, Any]:
         "display_name": display_name,
         "description": str(body.get("description") or "").strip() or None,
         "graph_name": graph_name,
-        "spatial_map": {"enabled": map_enabled, "scope": map_scope},
+        "spatial_map": {
+            "enabled": map_enabled,
+            "scope": map_scope,
+            "region_name": map_region_name,
+            "region_adcode": map_region_adcode,
+            "region_level": map_region_level or "province",
+        },
         "schema": schema,
         "graph_data": {"nodes": nodes, "relations": relations},
         "counts": {