@hupan56/wlkj 2.7.12 → 3.0.0

package/templates/qoder/scripts/{kg_semantic.py → kg/kg_semantic.py}

@@ -1,150 +1,156 @@
-# -*- coding: utf-8 -*-
-"""
-LLM semantic labeling layer for the knowledge graph.
-
-After kg_build.py extracts structural data (AST-level), this module:
-  1. Finds entities that lack semantic labels (cn == dir, no business_domain)
-  2. Outputs a labeling prompt for the AI (QoderWork) to fill in
-  3. Provides a write-back function to store labels in DuckDB
-
-The AI (QoderWork's LLM) sees the prompt during session-start or when
-kg_build reports new modules, and labels them — zero human trigger.
-
-Usage (called by AI automatically):
-  python kg_semantic.py --check       # check what needs labeling
-  python kg_semantic.py --label '{"assets": {"cn": "资产管理", "domain": "固定资产"}}'
-"""
-import json
-import os
-import sys
-
-SCRIPTS = os.path.dirname(os.path.abspath(__file__))
-sys.path.insert(0, SCRIPTS)
-from common.paths import DATA_INDEX_DIR
-from kg_duckdb import get_db
-
-DB_PATH = str(DATA_INDEX_DIR / 'kg.duckdb')
-
-
-def check_unlabeled():
-    """Find modules that need semantic labeling (cn == dir or cn is empty)."""
-    con = get_db()
-    # Find FEATURE entities where cn == dir (unlabeled) or cn is empty
-    rows = con.execute("""
-        SELECT id, canonical, cn FROM entities
-        WHERE type = 'FEATURE'
-          AND (cn = canonical OR cn = '' OR cn IS NULL)
-        ORDER BY canonical
-    """).fetchall()
-
-    if not rows:
-        return None
-
-    # Build labeling prompt
-    modules = []
-    for eid, dirname, cn in rows:
-        # Gather context: what endpoints/pages does this module have?
-        ep_count = con.execute(
-            "SELECT COUNT(*) FROM edges WHERE from_id = ? AND edge_type = 'has_endpoint'",
-            [eid]
-        ).fetchone()[0]
-        btn_count = con.execute(
-            "SELECT COUNT(*) FROM edges WHERE from_id = ? AND edge_type = 'has_button'",
-            [eid]
-        ).fetchone()[0]
-        # Sample endpoints for context
-        sample_eps = con.execute("""
-            SELECT e.to_id FROM edges e
-            WHERE e.from_id = ? AND e.edge_type = 'has_endpoint' LIMIT 5
-        """, [eid]).fetchall()
-
-        modules.append({
-            'dir': dirname,
-            'endpoints': ep_count,
-            'buttons': btn_count,
-            'sample_apis': [r[0].replace('E:', '') for r in sample_eps],
-        })
-
-    con.close()
-
-    prompt = {
-        'action': 'semantic_labeling_needed',
-        'description': '以下模块需要语义标注（中文名+业务领域）。请根据目录名和API端点推断。',
-        'modules': modules,
-        'instructions': (
-            '对每个模块返回: {"目录名": {"cn": "中文名", "domain": "业务领域"}}\n'
-            '示例: {"assets": {"cn": "资产管理", "domain": "固定资产管理"}}\n'
-            '然后用 kg_semantic.py --label 写回'
-        ),
-    }
-    return prompt
-
-
-def apply_labels(labels_json):
-    """Write semantic labels back to DuckDB.
-
-    labels_json: {"assets": {"cn": "资产管理", "domain": "固定资产"}, ...}
-    """
-    labels = json.loads(labels_json) if isinstance(labels_json, str) else labels_json
-    con = get_db()
-    updated = 0
-
-    for dirname, info in labels.items():
-        cn = info.get('cn', dirname)
-        domain = info.get('domain', '')
-
-        # Find the FEATURE entity by canonical name (dir)
-        rows = con.execute(
-            "SELECT id FROM entities WHERE type='FEATURE' AND canonical = ?",
-            [dirname]
-        ).fetchall()
-
-        for row in rows:
-            eid = row[0]
-            # Update cn
-            con.execute(
-                "UPDATE entities SET cn = ? WHERE id = ?",
-                [cn, eid]
-            )
-            # Update props with domain
-            props_row = con.execute(
-                "SELECT props FROM entities WHERE id = ?", [eid]
-            ).fetchone()
-            if props_row:
-                try:
-                    props = json.loads(props_row[0]) if props_row[0] else {}
-                except:
-                    props = {}
-                props['business_domain'] = domain
-                props['labeled'] = True
-                con.execute(
-                    "UPDATE entities SET props = ? WHERE id = ?",
-                    [json.dumps(props, ensure_ascii=False), eid]
-                )
-            # Update alias
-            con.execute(
-                "INSERT OR REPLACE INTO aliases VALUES (?, ?)",
-                [cn.lower(), eid]
-            )
-            updated += 1
-
-    con.close()
-    return updated
-
-
-if __name__ == '__main__':
-    if len(sys.argv) > 1 and sys.argv[1] == '--check':
-        result = check_unlabeled()
-        if result:
-            print(json.dumps(result, ensure_ascii=False, indent=2))
-        else:
-            print('{"status": "all_labeled", "message": "所有模块已有语义标注"}')
-
-    elif len(sys.argv) > 2 and sys.argv[1] == '--label':
-        count = apply_labels(sys.argv[2])
-        print('Updated %d entities with semantic labels' % count)
-
-    else:
-        print('Usage:')
-        print('  python kg_semantic.py --check    # 查看待标注模块')
-        print('  python kg_semantic.py --label \'{"assets":{"cn":"资产管理","domain":"固定资产"}}\'')
+# -*- coding: utf-8 -*-
+# v3.0 路径自举: 引导到 common/bootstrap, 统一 sys.path 逻辑
+import os as _o, sys as _s
+_cp = _o.path.join(_o.path.dirname(_o.path.dirname(_o.path.abspath(__file__))), 'common')
+if _cp not in _s.path: _s.path.insert(0, _cp)
+from bootstrap import setup; setup()
+
+"""
+LLM semantic labeling layer for the knowledge graph.
+
+After kg_build.py extracts structural data (AST-level), this module:
+  1. Finds entities that lack semantic labels (cn == dir, no business_domain)
+  2. Outputs a labeling prompt for the AI (QoderWork) to fill in
+  3. Provides a write-back function to store labels in DuckDB
+
+The AI (QoderWork's LLM) sees the prompt during session-start or when
+kg_build reports new modules, and labels them — zero human trigger.
+
+Usage (called by AI automatically):
+  python kg_semantic.py --check       # check what needs labeling
+  python kg_semantic.py --label '{"assets": {"cn": "资产管理", "domain": "固定资产"}}'
+"""
+import json
+import os
+import sys
+
+SCRIPTS = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))  # 子包→scripts/根
+sys.path.insert(0, SCRIPTS)
+from common.paths import DATA_INDEX_DIR
+from kg_duckdb import get_db
+
+DB_PATH = str(DATA_INDEX_DIR / 'kg.duckdb')
+
+
+def check_unlabeled():
+    """Find modules that need semantic labeling (cn == dir or cn is empty)."""
+    con = get_db()
+    # Find FEATURE entities where cn == dir (unlabeled) or cn is empty
+    rows = con.execute("""
+        SELECT id, canonical, cn FROM entities
+        WHERE type = 'FEATURE'
+          AND (cn = canonical OR cn = '' OR cn IS NULL)
+        ORDER BY canonical
+    """).fetchall()
+
+    if not rows:
+        return None
+
+    # Build labeling prompt
+    modules = []
+    for eid, dirname, cn in rows:
+        # Gather context: what endpoints/pages does this module have?
+        ep_count = con.execute(
+            "SELECT COUNT(*) FROM edges WHERE from_id = ? AND edge_type = 'has_endpoint'",
+            [eid]
+        ).fetchone()[0]
+        btn_count = con.execute(
+            "SELECT COUNT(*) FROM edges WHERE from_id = ? AND edge_type = 'has_button'",
+            [eid]
+        ).fetchone()[0]
+        # Sample endpoints for context
+        sample_eps = con.execute("""
+            SELECT e.to_id FROM edges e
+            WHERE e.from_id = ? AND e.edge_type = 'has_endpoint' LIMIT 5
+        """, [eid]).fetchall()
+
+        modules.append({
+            'dir': dirname,
+            'endpoints': ep_count,
+            'buttons': btn_count,
+            'sample_apis': [r[0].replace('E:', '') for r in sample_eps],
+        })
+
+    con.close()
+
+    prompt = {
+        'action': 'semantic_labeling_needed',
+        'description': '以下模块需要语义标注（中文名+业务领域）。请根据目录名和API端点推断。',
+        'modules': modules,
+        'instructions': (
+            '对每个模块返回: {"目录名": {"cn": "中文名", "domain": "业务领域"}}\n'
+            '示例: {"assets": {"cn": "资产管理", "domain": "固定资产管理"}}\n'
+            '然后用 kg_semantic.py --label 写回'
+        ),
+    }
+    return prompt
+
+
+def apply_labels(labels_json):
+    """Write semantic labels back to DuckDB.
+
+    labels_json: {"assets": {"cn": "资产管理", "domain": "固定资产"}, ...}
+    """
+    labels = json.loads(labels_json) if isinstance(labels_json, str) else labels_json
+    con = get_db()
+    updated = 0
+
+    for dirname, info in labels.items():
+        cn = info.get('cn', dirname)
+        domain = info.get('domain', '')
+
+        # Find the FEATURE entity by canonical name (dir)
+        rows = con.execute(
+            "SELECT id FROM entities WHERE type='FEATURE' AND canonical = ?",
+            [dirname]
+        ).fetchall()
+
+        for row in rows:
+            eid = row[0]
+            # Update cn
+            con.execute(
+                "UPDATE entities SET cn = ? WHERE id = ?",
+                [cn, eid]
+            )
+            # Update props with domain
+            props_row = con.execute(
+                "SELECT props FROM entities WHERE id = ?", [eid]
+            ).fetchone()
+            if props_row:
+                try:
+                    props = json.loads(props_row[0]) if props_row[0] else {}
+                except Exception:
+                    props = {}
+                props['business_domain'] = domain
+                props['labeled'] = True
+                con.execute(
+                    "UPDATE entities SET props = ? WHERE id = ?",
+                    [json.dumps(props, ensure_ascii=False), eid]
+                )
+            # Update alias
+            con.execute(
+                "INSERT OR REPLACE INTO aliases VALUES (?, ?)",
+                [cn.lower(), eid]
+            )
+            updated += 1
+
+    con.close()
+    return updated
+
+
+if __name__ == '__main__':
+    if len(sys.argv) > 1 and sys.argv[1] == '--check':
+        result = check_unlabeled()
+        if result:
+            print(json.dumps(result, ensure_ascii=False, indent=2))
+        else:
+            print('{"status": "all_labeled", "message": "所有模块已有语义标注"}')
+
+    elif len(sys.argv) > 2 and sys.argv[1] == '--label':
+        count = apply_labels(sys.argv[2])
+        print('Updated %d entities with semantic labels' % count)
+
+    else:
+        print('Usage:')
+        print('  python kg_semantic.py --check    # 查看待标注模块')
+        print('  python kg_semantic.py --label \'{"assets":{"cn":"资产管理","domain":"固定资产"}}\'')