signalpilot-ai-internal 0.4.6__py3-none-any.whl → 0.4.7__py3-none-any.whl

signalpilot_ai_internal/_version.py

@@ -1,4 +1,4 @@
 # This file is auto-generated by Hatchling. As such, do not:
 #   - modify
 #   - track in version control e.g. be sure to add to .gitignore
-__version__ = VERSION = '0.4.6'
+__version__ = VERSION = '0.4.7'

signalpilot_ai_internal/handlers.py

@@ -12,6 +12,7 @@ from .cache_service import get_cache_service
 from .cache_handlers import ChatHistoriesHandler, AppValuesHandler, CacheInfoHandler
 from .unified_database_schema_service import UnifiedDatabaseSchemaHandler, UnifiedDatabaseQueryHandler
 from .snowflake_schema_service import SnowflakeSchemaHandler, SnowflakeQueryHandler
+from .schema_search_service import SchemaSearchHandler
 
 
 class HelloWorldHandler(APIHandler):
@@ -279,6 +280,7 @@ def setup_handlers(web_app):
     # Database service endpoints
     database_schema_route = url_path_join(base_url, "signalpilot-ai-internal", "database", "schema")
     database_query_route = url_path_join(base_url, "signalpilot-ai-internal", "database", "query")
+    database_schema_search_route = url_path_join(base_url, "signalpilot-ai-internal", "database", "schema-search")
     
     # MySQL service endpoints
     mysql_schema_route = url_path_join(base_url, "signalpilot-ai-internal", "mysql", "schema")
@@ -309,6 +311,7 @@ def setup_handlers(web_app):
         # Database service endpoints (unified for PostgreSQL and MySQL)
         (database_schema_route, UnifiedDatabaseSchemaHandler),
         (database_query_route, UnifiedDatabaseQueryHandler),
+        (database_schema_search_route, SchemaSearchHandler),
         
         # MySQL service endpoints (use unified handler)
         (mysql_schema_route, UnifiedDatabaseSchemaHandler),
@@ -339,7 +342,8 @@ def setup_handlers(web_app):
     print(f"  - Cache Info: {cache_info_route}")
     print(f"  - Database Schema: {database_schema_route}")
     print(f"  - Database Query: {database_query_route}")
+    print(f"  - Database Schema Search: {database_schema_search_route}")
     print(f"  - MySQL Schema: {mysql_schema_route}")
     print(f"  - MySQL Query: {mysql_query_route}")
     print(f"  - Snowflake Schema: {snowflake_schema_route}")
-    print(f"  - Snowflake Query: {snowflake_query_route}")
+    print(f"  - Snowflake Query: {snowflake_query_route}")

signalpilot_ai_internal/schema_search_config.yml

@@ -0,0 +1,32 @@
+logging:
+  level: "WARNING"
+
+embedding:
+  location: "memory"
+  model: "multi-qa-MiniLM-L6-cos-v1"
+  metric: "cosine"
+  batch_size: 32
+  show_progress: false
+  cache_dir: "/tmp/.schema_search_cache"
+
+chunking:
+  strategy: "raw"
+  max_tokens: 256
+  overlap_tokens: 50
+  model: "gpt-4o-mini"
+
+search:
+  strategy: "hybrid"
+  initial_top_k: 20
+  rerank_top_k: 5
+  semantic_weight: 0.67
+  hops: 1
+
+reranker:
+  model: null
+
+schema:
+  include_columns: true
+  include_indices: true
+  include_foreign_keys: true
+  include_constraints: true

signalpilot_ai_internal/schema_search_service.py

@@ -0,0 +1,109 @@
+import json
+import os
+import subprocess
+import sys
+from pathlib import Path
+from typing import Optional
+
+from jupyter_server.base.handlers import APIHandler
+import tornado
+from schema_search import SchemaSearch
+from sqlalchemy import create_engine
+
+
+class SchemaSearchHandler(APIHandler):
+    CONFIG_PATH = Path(__file__).with_name("schema_search_config.yml")
+
+    def _get_database_url(self, explicit: Optional[str]) -> Optional[str]:
+        if isinstance(explicit, str) and explicit.strip():
+            return explicit.strip()
+
+        for key, value in os.environ.items():
+            if key.endswith("_CONNECTION_JSON") and isinstance(value, str) and value.strip().startswith("{"):
+                config = json.loads(value)
+                url = config.get("connectionUrl")
+                if url:
+                    return url
+        return os.environ.get("DB_URL")
+
+    @tornado.web.authenticated
+    async def post(self):
+        body = self.get_json_body() or {}
+        queries = body.get("queries")
+        if isinstance(queries, str):
+            queries = [queries]
+
+        if not isinstance(queries, list):
+            self.set_status(400)
+            self.finish(json.dumps({"error": "queries parameter must be a list of strings"}))
+            return
+
+        queries = [q.strip() for q in queries if isinstance(q, str) and q.strip()]
+
+        if not queries:
+            self.set_status(400)
+            self.finish(json.dumps({"error": "queries parameter is required"}))
+            return
+
+        db_url = self._get_database_url(body.get("dbUrl"))
+        if not db_url:
+            self.set_status(400)
+            self.finish(json.dumps({"error": "Database connection URL is not configured"}))
+            return
+
+        db_url = db_url.strip()
+        db_url_lower = db_url.lower()
+
+        if db_url_lower.startswith("mysql://"):
+            db_url = "mysql+pymysql://" + db_url[len("mysql://"):]
+            db_url_lower = db_url.lower()
+
+        if db_url_lower.startswith("snowflake://"):
+            self._ensure_snowflake_dependencies()
+        elif db_url_lower.startswith("postgresql") or db_url_lower.startswith("postgres") or db_url_lower.startswith("mysql+pymysql"):
+            pass
+        else:
+            self.set_status(400)
+            self.finish(json.dumps({"error": "Schema search currently supports PostgreSQL, MySQL, or Snowflake connections"}))
+            return
+
+        engine = None
+        try:
+            engine = create_engine(db_url)
+            schema_search = SchemaSearch(engine=engine, config_path=str(self.CONFIG_PATH))
+            schema_search.index()
+
+            limit = body.get("limit")
+            if limit is not None:
+                limit = max(1, min(int(limit), 10))
+            else:
+                limit = 5
+
+            query_results = []
+            for query in queries:
+                result = schema_search.search(query, limit=limit)
+                query_results.append({
+                    "query": query,
+                    "results": result
+                })
+
+            self.finish(json.dumps({"results": query_results}))
+        except Exception as error:
+            self.set_status(500)
+            self.finish(json.dumps({"error": f"Schema search failed: {error}"}))
+        finally:
+            if engine is not None:
+                try:
+                    engine.dispose()
+                except Exception:
+                    pass
+
+    def _install_package(self, package: str) -> None:
+        subprocess.check_call([sys.executable, "-m", "pip", "install", package])
+
+    def _ensure_snowflake_dependencies(self) -> None:
+        try:
+            import snowflake.sqlalchemy  # type: ignore  # noqa: F401
+        except ImportError:
+            self._install_package("snowflake-sqlalchemy")
+            import snowflake.sqlalchemy  # type: ignore  # noqa: F401

{signalpilot_ai_internal-0.4.6.data → signalpilot_ai_internal-0.4.7.data}/data/share/jupyter/labextensions/signalpilot-ai-internal/package.json

@@ -1,6 +1,6 @@
 {
   "name": "signalpilot-ai-internal",
-  "version": "0.4.6",
+  "version": "0.4.7",
   "description": "SignalPilot Agent - Your Jupyter Notebook Assistant",
   "keywords": [
     "jupyter",
@@ -133,7 +133,7 @@
     "outputDir": "signalpilot_ai_internal/labextension",
     "schemaDir": "schema",
     "_build": {
-      "load": "static/remoteEntry.61ef7b3a050c161d7a88.js",
+      "load": "static/remoteEntry.2a797d447eeb725a28cf.js",
       "extension": "./extension",
       "style": "./style"
     }

{signalpilot_ai_internal-0.4.6.data → signalpilot_ai_internal-0.4.7.data}/data/share/jupyter/labextensions/signalpilot-ai-internal/schemas/signalpilot-ai-internal/package.json.orig

@@ -1,6 +1,6 @@
 {
   "name": "signalpilot-ai-internal",
-  "version": "0.4.6",
+  "version": "0.4.7",
   "description": "SignalPilot Agent - Your Jupyter Notebook Assistant",
   "keywords": [
     "jupyter",