feat: add get schema tool

backend/app/agent/nl2sql.py

@@ -4,12 +4,15 @@ import os
 import json
 import time
 import threading
+import logging
 from pathlib import Path
 from typing import List, Optional, Dict, Any, Callable, Awaitable
 from pydantic import BaseModel, Field
 import duckdb
 import pandas as pd
 
+logger = logging.getLogger(__name__)
+
 # Add project root to sys.path to allow importing nanobot
 PROJECT_ROOT = Path(__file__).resolve().parents[3]
 if str(PROJECT_ROOT) not in sys.path:
@@ -221,8 +224,11 @@ async def _check_connection_with_cache(source: str, connector: Any) -> bool:
     try:
         ok = await asyncio.wait_for(
             asyncio.to_thread(connector.test_connection),
-            timeout=10.0
+            timeout=15.0
         )
+    except asyncio.TimeoutError:
+        print("Connection test failed or timed out: Timeout after 15 seconds")
+        ok = False
     except Exception as e:
         print(f"Connection test failed or timed out: {e}")
         ok = False
@@ -300,8 +306,10 @@ async def process_nl2sql(
             try:
                 schema = await asyncio.wait_for(
                     asyncio.to_thread(connector.get_schema),
-                    timeout=30.0
+                    timeout=120.0
                 )
+            except asyncio.TimeoutError:
+                return NL2SQLResponse(sql="", result=[], error="Failed to fetch schema: Timeout after 120 seconds. Data source might be too large or network is slow.")
             except Exception as e:
                 return NL2SQLResponse(sql="", result=[], error=f"Failed to fetch schema: {e}")
             
@@ -449,7 +457,13 @@ Language: Chinese (Simplified)
                 # Fallback if LLM doesn't return valid JSON despite instructions
                 sql_query = content
                 
-            await emit_progress(f"SQL 生成完成 ({time.perf_counter() - llm_started:.2f}s)")
+            logger.info(f"Generated SQL for query '{request.query}':\n{sql_query}")
+            
+            # 格式化单行 SQL 用于在前端进度中展示
+            formatted_sql = sql_query.replace('\n', ' ')
+            if len(formatted_sql) > 150:
+                formatted_sql = formatted_sql[:147] + "..."
+            await emit_progress(f"SQL 生成完成: {formatted_sql}")
 
         except Exception as e:
             return NL2SQLResponse(sql=sql_query, result=[], error=f"LLM generation failed: {e}")

backend/app/connectors/postgres.py

@@ -29,7 +29,44 @@ class PostgresConnector:
             # Default schema for postgres is 'public', sqlite is None
             schema_name = 'public' if self.engine.dialect.name == 'postgresql' else None
             
-            for table_name in inspector.get_table_names(schema=schema_name):
+            table_names = inspector.get_table_names(schema=schema_name)
+            
+            # Use SQLAlchemy 2.0+ multi-fetch to avoid N+1 queries issue, especially over remote networks
+            if hasattr(inspector, 'get_multi_columns'):
+                multi_columns = inspector.get_multi_columns(schema=schema_name)
+                multi_pk = inspector.get_multi_pk_constraint(schema=schema_name)
+                multi_fk = inspector.get_multi_foreign_keys(schema=schema_name)
+                
+                for table_name in table_names:
+                    key = (schema_name, table_name)
+                    
+                    columns = []
+                    for col in multi_columns.get(key, []):
+                        columns.append({
+                            "name": col['name'], 
+                            "type": str(col['type'])
+                        })
+                        
+                    pk_constraint = multi_pk.get(key)
+                    pks = pk_constraint.get('constrained_columns', []) if pk_constraint else []
+                    
+                    foreign_keys = []
+                    for fk in multi_fk.get(key, []):
+                        foreign_keys.append({
+                            "constrained_columns": fk['constrained_columns'],
+                            "referred_table": fk['referred_table'],
+                            "referred_columns": fk['referred_columns']
+                        })
+                        
+                    schema[table_name] = {
+                        "columns": columns,
+                        "primary_keys": pks,
+                        "foreign_keys": foreign_keys
+                    }
+                return schema
+
+            # Fallback for older SQLAlchemy versions
+            for table_name in table_names:
                 columns = []
                 # get columns
                 for col in inspector.get_columns(table_name, schema=schema_name):
@@ -59,8 +96,10 @@ class PostgresConnector:
                 }
             return schema
         except Exception as e:
+            import traceback
+            traceback.print_exc()
             print(f"Error getting schema: {e}")
-            return {}
+            raise e
 
     def test_connection(self) -> bool:
         try:

backend/app/core/nanobot.py

@@ -108,8 +108,10 @@ class NanobotIntegration:
     def _register_custom_tools(self, agent: AgentLoop):
         from app.tools.nl2sql import NL2SQLTool
         from app.tools.visualization import VisualizationTool
+        from app.tools.get_schema import GetDatabaseSchemaTool
         agent.tools.register(NL2SQLTool())
         agent.tools.register(VisualizationTool())
+        agent.tools.register(GetDatabaseSchemaTool())
 
     def _make_provider(self, config: Config):
         # Logic adapted from nanobot/cli/commands.py

backend/app/tools/get_schema.py

@@ -0,0 +1,131 @@
+import json
+import logging
+from typing import Any
+import asyncio
+
+from nanobot.agent.tools.base import Tool
+from app.context import current_data_source, current_file_url, current_progress_callback
+from app.connectors.postgres import postgres_connector
+from app.connectors.clickhouse import clickhouse_connector
+from app.connectors.factory import get_connector
+from app.database import SessionLocal
+from app.models.datasource import DataSource
+
+# Import schema logic from nl2sql
+from app.agent.nl2sql import (
+    _get_cached_schema,
+    _set_cached_schema,
+    _check_connection_with_cache,
+    _get_upload_payload
+)
+
+logger = logging.getLogger(__name__)
+
+class GetDatabaseSchemaTool(Tool):
+    """
+    Tool for fetching the database schema directly without SQL generation.
+    """
+
+    @property
+    def name(self) -> str:
+        return "get_database_schema"
+
+    @property
+    def description(self) -> str:
+        return (
+            "Get the structural schema of the currently connected database or data source. "
+            "Use this tool when the user asks questions about metadata, such as 'what tables are there', "
+            "'show me the database structure', 'what are the columns in table X', etc. "
+            "It directly returns the schema without generating SQL."
+        )
+
+    @property
+    def parameters(self) -> dict[str, Any]:
+        return {
+            "type": "object",
+            "properties": {},
+            "required": [],
+        }
+
+    async def execute(self, **kwargs: Any) -> str:
+        source = current_data_source.get()
+        file_url = current_file_url.get()
+        on_progress = current_progress_callback.get()
+
+        async def emit_progress(msg: str):
+            if on_progress:
+                await on_progress(msg)
+
+        await emit_progress("正在获取数据源结构...")
+
+        connector = None
+        schema = {}
+
+        if not source:
+            return "Error: No data source connected."
+
+        if source == "postgres":
+            connector = postgres_connector
+        elif source == "clickhouse":
+            connector = clickhouse_connector
+        elif source == "upload":
+            try:
+                payload = await asyncio.to_thread(_get_upload_payload, file_url)
+                schema = payload["schema"]
+                await emit_progress("文件 Schema 获取完成")
+            except Exception as e:
+                return f"Failed to get upload schema: {e}"
+        elif source.startswith("ds:"):
+            try:
+                ds_id = int(source.split(":")[1])
+                def _get_ds_connector():
+                    db = SessionLocal()
+                    try:
+                        ds = db.query(DataSource).filter(DataSource.id == ds_id).first()
+                        if not ds: return None
+                        return get_connector(ds)
+                    finally:
+                        db.close()
+                connector = await asyncio.to_thread(_get_ds_connector)
+                if not connector:
+                    return f"Data source not found: {source}"
+            except Exception as e:
+                return f"Failed to load data source: {e}"
+        else:
+            return f"Unsupported data source: {source}"
+
+        if connector:
+            cached_schema = _get_cached_schema(source, connector)
+            if cached_schema is not None:
+                schema = cached_schema
+                await emit_progress(f"命中缓存，成功获取 {len(schema)} 张表结构")
+            else:
+                if not await _check_connection_with_cache(source, connector):
+                    return f"Failed to connect to {source}"
+                
+                try:
+                    schema = await asyncio.wait_for(
+                        asyncio.to_thread(connector.get_schema),
+                        timeout=120.0
+                    )
+                    _set_cached_schema(source, connector, schema)
+                    await emit_progress(f"成功获取 {len(schema)} 张表结构")
+                except asyncio.TimeoutError:
+                    return "Failed to fetch schema: Timeout after 120 seconds."
+                except Exception as e:
+                    return f"Failed to fetch schema: {e}"
+
+        # Format the output for the LLM to make it readable and token-efficient
+        lines = []
+        for table_name, table_info in schema.items():
+            if isinstance(table_info, list):
+                # Clickhouse/Upload format: [{"name": "col", "type": "type"}]
+                cols = ", ".join([f"{c['name']} ({c['type']})" for c in table_info])
+                lines.append(f"Table: {table_name}\n  Columns: {cols}")
+            elif isinstance(table_info, dict):
+                # Postgres format: {"columns": [...], "primary_keys": [...], "foreign_keys": [...]}
+                cols = ", ".join([f"{c['name']} ({c['type']})" for c in table_info.get("columns", [])])
+                pks = ", ".join(table_info.get("primary_keys", []))
+                lines.append(f"Table: {table_name}\n  Columns: {cols}\n  Primary Keys: {pks}")
+        
+        return "\n\n".join(lines) if lines else "No tables found in schema."

backend/app/tools/nl2sql.py

@@ -74,14 +74,8 @@ class NL2SQLTool(Tool):
             # Call the core logic
             result = await process_nl2sql(request, on_progress=on_progress)
             
-            if result.error:
+            # Always save visualization payload to context so the chat stream can pick it up
-                return f"Error executing query: {result.error}"
+            # Even if there's an error, we want the frontend to see the generated SQL
-
-            # Save the result data to context for potential later use by VisualizationTool
-            if result.result:
-                current_data.set(result.result)
-
-            # Save visualization payload to context so the chat stream can pick it up
             viz_payload = _build_sql_chart_viz(result)
             existing_viz = current_viz_data.get()
             if isinstance(existing_viz, dict):
@@ -91,6 +85,13 @@ class NL2SQLTool(Tool):
             else:
                 current_viz_data.set(viz_payload)
 
+            if result.error:
+                return f"Error executing query: {result.error}\nGenerated SQL: {result.sql}"
+
+            # Save the result data to context for potential later use by VisualizationTool
+            if result.result:
+                current_data.set(result.result)
+
             # Build a summary string for the Agent to read
             row_count = len(result.result) if result.result else 0
             

backend/main.py

@@ -246,6 +246,7 @@ async def nanobot_chat_stream(request: ChatRequest):
                     else:
                         yield f"data: {json.dumps({'type': 'progress', 'content': progress}, ensure_ascii=False)}\n\n"
                 except asyncio.TimeoutError:
+                    yield ": keep-alive\n\n"
                     continue
 
             response = await current_task

backend/test_nl2sql.py

@@ -0,0 +1,12 @@
+import asyncio
+import json
+from app.agent.nl2sql import process_nl2sql, NL2SQLRequest
+
+async def main():
+    req = NL2SQLRequest(query="列出所有表", source="postgres", generate_chart=False)
+    res = await process_nl2sql(req)
+    print("SQL:", res.sql)
+    print("Error:", res.error)
+    print("Result:", res.result)
+
+asyncio.run(main())

frontend/src/components/ChatInterface.tsx

@@ -958,7 +958,8 @@ export function ChatInterface() {
             </div>
           ) : (
             <div className="max-w-3xl mx-auto px-4 py-8 space-y-8">
-              {messages.map((msg) => {
+              {messages.map((msg, msgIdx) => {
+                const isMessageGenerating = isLoading && msgIdx === messages.length - 1;
                 const { markdown, reportHtml } = splitReportHtml(msg.content);
                 const externalReportUrl = extractExternalReport(msg.content);
                 return (
@@ -992,17 +993,24 @@ export function ChatInterface() {
                         {msg.progressLogs && msg.progressLogs.length > 0 ? (
                           <div className="mb-2 rounded-xl border border-zinc-100 bg-zinc-50/70 px-3 py-2">
                             <div className="flex items-center gap-2 text-zinc-500 text-xs mb-1.5 pb-1.5 border-b border-zinc-100/50">
-                              {msg.awaitingFirstToken ? <Loader2 className="h-3.5 w-3.5 animate-spin" /> : <CheckCircle2 className="h-3.5 w-3.5 text-emerald-500" />}
+                              {isMessageGenerating ? <Loader2 className="h-3.5 w-3.5 animate-spin" /> : <CheckCircle2 className="h-3.5 w-3.5 text-emerald-500" />}
-                              <span>{msg.awaitingFirstToken ? t('processing') : t('processCompleted')}</span>
+                              <span>{isMessageGenerating ? t('processing') : t('processCompleted')}</span>
                             </div>
-                            <div className="space-y-1.5 max-h-[160px] overflow-y-auto pr-1">
+                            <div 
+                              className="space-y-1.5 max-h-[160px] overflow-y-auto pr-1"
+                              ref={(el) => {
+                                if (el && isMessageGenerating) {
+                                  el.scrollTop = el.scrollHeight;
+                                }
+                              }}
+                            >
                               {msg.progressLogs.map((log, idx, arr) => {
                                 const isLast = idx === arr.length - 1;
-                                // 如果是正在处理的会话，且当前日志是最后一条，或者是明确包含“正在”的日志，则显示 loading
+                                // 只有当是整个会话的最后一条消息，且当前日志是最后一条时，才显示 loading 动画
-                                const isLoadingLog = (isLast && msg.awaitingFirstToken) || log.includes(t('processingIndicator'));
+                                const isLoadingLog = isLast && isMessageGenerating;
                                 return (
                                   <div key={`${msg.id}-log-${idx}`} className="flex items-start gap-2 text-[12px] text-zinc-500 leading-5">
-                                    {isLoadingLog && msg.awaitingFirstToken ? (
+                                    {isLoadingLog ? (
                                       <Settings className="mt-0.5 h-3.5 w-3.5 text-amber-500 animate-spin shrink-0" />
                                     ) : (
                                       <CheckCircle2 className="mt-0.5 h-3.5 w-3.5 text-emerald-500 shrink-0" />