Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -1,24 +1,36 @@
|
|
1 |
-
from flask import Flask, render_template, request, redirect, url_for,
|
2 |
from flask_socketio import SocketIO
|
3 |
-
import os
|
4 |
import threading
|
|
|
5 |
from dotenv import load_dotenv
|
|
|
6 |
from werkzeug.utils import secure_filename
|
|
|
7 |
|
8 |
# LangChain and agent imports
|
9 |
-
from
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
10 |
from langchain_core.messages import AIMessage, ToolMessage
|
11 |
from pydantic import BaseModel, Field
|
12 |
from typing_extensions import TypedDict
|
13 |
-
from langgraph.graph import END,
|
14 |
from langgraph.graph.message import AnyMessage, add_messages
|
15 |
from langchain_core.runnables import RunnableLambda, RunnableWithFallbacks
|
16 |
from langgraph.prebuilt import ToolNode
|
17 |
from langchain_core.prompts import ChatPromptTemplate
|
18 |
from langchain_community.utilities import SQLDatabase
|
19 |
-
from langchain_community.agent_toolkits import SQLDatabaseToolkit
|
20 |
-
from langchain_core.tools import tool
|
21 |
-
import traceback
|
22 |
|
23 |
# Load environment variables
|
24 |
load_dotenv()
|
@@ -27,121 +39,123 @@ load_dotenv()
|
|
27 |
UPLOAD_FOLDER = os.path.join(os.getcwd(), "uploads")
|
28 |
os.makedirs(UPLOAD_FOLDER, exist_ok=True)
|
29 |
BASE_DIR = os.path.abspath(os.path.dirname(__file__))
|
|
|
|
|
30 |
|
31 |
# API Keys from .env file
|
|
|
32 |
os.environ["GROQ_API_KEY"] = os.getenv("GROQ_API_KEY")
|
33 |
os.environ["MISTRAL_API_KEY"] = os.getenv("MISTRAL_API_KEY")
|
34 |
|
35 |
-
#
|
36 |
-
flask_app = Flask(__name__)
|
37 |
-
flask_app.config['UPLOAD_FOLDER'] = UPLOAD_FOLDER
|
38 |
-
# Set secret key for flash messages:
|
39 |
-
flask_app.config['SECRET_KEY'] = os.getenv("FLASK_SECRET_KEY", "mysecretkey")
|
40 |
-
socketio = SocketIO(flask_app, cors_allowed_origins="*")
|
41 |
-
|
42 |
-
# Global state
|
43 |
agent_app = None
|
44 |
abs_file_path = None
|
|
|
45 |
|
|
|
|
|
|
|
46 |
def create_agent_app(db_path: str):
|
47 |
-
|
48 |
-
|
49 |
-
|
50 |
-
|
51 |
-
|
52 |
-
|
53 |
-
|
54 |
-
|
55 |
-
try:
|
56 |
-
db_instance = SQLDatabase.from_uri(f"sqlite:///{abs_db_path}")
|
57 |
-
except Exception as e:
|
58 |
-
flash(f"[ERROR]: Failed to connect to DB: {e}", "error")
|
59 |
-
raise
|
60 |
-
|
61 |
-
@tool
|
62 |
def db_query_tool(query: str) -> str:
|
63 |
"""
|
64 |
-
|
65 |
-
If the query is invalid or returns no result, an error message will be returned.
|
66 |
-
In case of an error, the user is advised to rewrite the query and try again.
|
67 |
"""
|
68 |
try:
|
69 |
result = db_instance.run_no_throw(query)
|
70 |
-
return result
|
71 |
except Exception as e:
|
72 |
-
flash(f"[ERROR]: Exception during query execution: {e}", "error")
|
73 |
return f"Error: {str(e)}"
|
74 |
|
|
|
|
|
|
|
75 |
class SubmitFinalAnswer(BaseModel):
|
76 |
-
final_answer: str = Field(
|
77 |
|
|
|
|
|
|
|
78 |
class State(TypedDict):
|
79 |
messages: Annotated[list[AnyMessage], add_messages]
|
80 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
81 |
try:
|
82 |
-
|
83 |
-
Double check the SQLite query for common mistakes, including:
|
84 |
-
- Using NOT IN with NULL values
|
85 |
-
- Using UNION when UNION ALL should have been used
|
86 |
-
- Using BETWEEN for exclusive ranges
|
87 |
-
- Data type mismatch in predicates
|
88 |
-
- Properly quoting identifiers
|
89 |
-
- Using the correct number of arguments for functions
|
90 |
-
- Casting to the correct data type
|
91 |
-
- Using the proper columns for joins
|
92 |
-
|
93 |
-
If there are any of the above mistakes, rewrite the query. If there are no mistakes, just reproduce the original query.
|
94 |
-
|
95 |
-
You will call the appropriate tool to execute the query after running this check.
|
96 |
-
"""
|
97 |
-
|
98 |
-
query_check = ChatPromptTemplate.from_messages([
|
99 |
-
("system", query_check_system),
|
100 |
-
("placeholder", "{messages}")
|
101 |
-
]) | llm.bind_tools([db_query_tool])
|
102 |
-
|
103 |
-
query_gen_system = """You are a SQL expert with a strong attention to detail.
|
104 |
-
|
105 |
-
Given an input question, output a syntactically correct SQLite query to run, then look at the results of the query and return the answer.
|
106 |
-
|
107 |
-
DO NOT call any tool besides SubmitFinalAnswer to submit the final answer.
|
108 |
-
|
109 |
-
When generating the query:
|
110 |
-
|
111 |
-
Output the SQL query that answers the input question without a tool call.
|
112 |
-
|
113 |
-
Unless the user specifies a specific number of examples they wish to obtain, always limit your query to at most 5 results.
|
114 |
-
You can order the results by a relevant column to return the most interesting examples in the database.
|
115 |
-
Never query for all the columns from a specific table, only ask for the relevant columns given the question.
|
116 |
-
|
117 |
-
If you get an error while executing a query, rewrite the query and try again.
|
118 |
-
|
119 |
-
If you get an empty result set, you should try to rewrite the query to get a non-empty result set.
|
120 |
-
NEVER make stuff up if you don't have enough information to answer the query... just say you don't have enough information.
|
121 |
-
|
122 |
-
If you have enough information to answer the input question, simply invoke the appropriate tool to submit the final answer to the user.
|
123 |
-
|
124 |
-
DO NOT make any DML statements (INSERT, UPDATE, DELETE, DROP etc.) to the database. Do not return any sql query except answer.
|
125 |
-
"""
|
126 |
-
|
127 |
-
|
128 |
-
query_gen = ChatPromptTemplate.from_messages([
|
129 |
-
("system", query_gen_system),
|
130 |
-
("placeholder", "{messages}")
|
131 |
-
]) | llm.bind_tools([SubmitFinalAnswer])
|
132 |
-
except Exception as e:
|
133 |
-
flash(f"[ERROR]: Failed to create prompt templates: {e}", "error")
|
134 |
-
raise
|
135 |
-
|
136 |
-
try:
|
137 |
-
toolkit = SQLDatabaseToolkit(db=db_instance, llm=llm)
|
138 |
-
tools_instance = toolkit.get_tools()
|
139 |
except Exception as e:
|
140 |
-
|
141 |
-
|
142 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
143 |
def first_tool_call(state: State):
|
144 |
-
return {"messages": [AIMessage(content="", tool_calls=[{"name": "sql_db_list_tables", "args": {}, "id": "tool_abcd123"}])]}
|
145 |
|
146 |
def handle_tool_error(state: State):
|
147 |
tool_calls = state["messages"][-1].tool_calls
|
@@ -156,8 +170,7 @@ def create_agent_app(db_path: str):
|
|
156 |
try:
|
157 |
message = query_gen.invoke(state)
|
158 |
except Exception as e:
|
159 |
-
|
160 |
-
raise
|
161 |
tool_messages = []
|
162 |
if message.tool_calls:
|
163 |
for tc in message.tool_calls:
|
@@ -168,8 +181,9 @@ def create_agent_app(db_path: str):
|
|
168 |
))
|
169 |
return {"messages": [message] + tool_messages}
|
170 |
|
171 |
-
def should_continue(state: State):
|
172 |
-
|
|
|
173 |
if getattr(last_message, "tool_calls", None):
|
174 |
return END
|
175 |
if last_message.content.startswith("Error:"):
|
@@ -179,16 +193,18 @@ def create_agent_app(db_path: str):
|
|
179 |
def model_check_query(state: State):
|
180 |
return {"messages": [query_check.invoke({"messages": [state["messages"][-1]]})]}
|
181 |
|
182 |
-
|
|
|
|
|
|
|
183 |
schema_tool = next((t for t in tools_instance if t.name == "sql_db_schema"), None)
|
184 |
model_get_schema = llm.bind_tools([schema_tool])
|
185 |
|
186 |
workflow = StateGraph(State)
|
187 |
workflow.add_node("first_tool_call", first_tool_call)
|
188 |
-
workflow.add_node("list_tables_tool", create_tool_node_with_fallback([
|
189 |
workflow.add_node("get_schema_tool", create_tool_node_with_fallback([schema_tool]))
|
190 |
-
|
191 |
-
workflow.add_node("model_get_schema", lambda s: {"messages": [model_get_schema.invoke(s["messages"])]})
|
192 |
workflow.add_node("query_gen", query_gen_node)
|
193 |
workflow.add_node("correct_query", model_check_query)
|
194 |
workflow.add_node("execute_query", create_tool_node_with_fallback([db_query_tool]))
|
@@ -204,121 +220,111 @@ def create_agent_app(db_path: str):
|
|
204 |
|
205 |
return workflow.compile()
|
206 |
|
207 |
-
|
208 |
-
|
209 |
-
|
210 |
-
|
211 |
-
|
212 |
-
|
213 |
-
|
214 |
-
|
215 |
-
|
216 |
-
|
217 |
-
|
218 |
-
|
219 |
-
|
220 |
-
|
221 |
-
socketio.emit("log", {"message": "[ERROR]: No DB file uploaded."})
|
222 |
-
socketio.emit("final", {"message": "No database available. Please upload one and try again."})
|
223 |
-
flash("No database available. Please upload one and try again.", "error")
|
224 |
-
return
|
225 |
-
try:
|
226 |
-
# Lazy agent initialization: use the previously uploaded DB.
|
227 |
-
if agent_app is None:
|
228 |
-
socketio.emit("log", {"message": "[INFO]: Initializing agent for the first time..."})
|
229 |
-
agent_app = create_agent_app(abs_file_path)
|
230 |
-
socketio.emit("log", {"message": "[INFO]: Agent initialized."})
|
231 |
-
flash("Agent initialized.", "info")
|
232 |
-
query = {"messages": [("user", prompt)]}
|
233 |
-
result = agent_app.invoke(query)
|
234 |
try:
|
235 |
-
|
236 |
except Exception as e:
|
237 |
-
|
238 |
-
|
239 |
-
|
240 |
-
|
241 |
-
|
242 |
-
|
243 |
-
|
244 |
-
|
245 |
-
|
246 |
-
|
247 |
-
|
248 |
-
|
249 |
-
|
250 |
-
|
251 |
-
|
252 |
-
|
253 |
-
|
254 |
-
|
255 |
-
|
256 |
-
|
257 |
-
|
258 |
-
|
259 |
-
|
260 |
-
|
261 |
-
|
262 |
-
|
263 |
-
|
264 |
-
|
265 |
-
|
266 |
-
|
267 |
-
|
268 |
-
|
269 |
-
|
270 |
-
|
271 |
-
|
272 |
-
|
273 |
-
|
274 |
-
|
275 |
-
|
276 |
-
|
277 |
-
|
278 |
-
|
279 |
-
|
280 |
-
|
281 |
-
|
282 |
-
|
283 |
-
|
284 |
-
|
285 |
-
|
286 |
-
|
287 |
-
|
288 |
-
|
289 |
-
|
290 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
291 |
file.save(db_path)
|
292 |
-
abs_file_path = os.path.abspath(db_path)
|
293 |
-
agent_app = None
|
294 |
-
|
295 |
socketio.emit("log", {"message": f"[INFO]: Database file '{filename}' uploaded."})
|
|
|
296 |
return redirect(url_for("index"))
|
297 |
-
|
298 |
-
|
299 |
-
|
300 |
-
|
301 |
-
|
302 |
-
|
303 |
-
|
304 |
-
|
305 |
-
|
306 |
-
|
307 |
-
|
308 |
-
|
309 |
-
|
310 |
-
|
311 |
-
@socketio.on("user_input")
|
312 |
-
def handle_user_input(data):
|
313 |
-
prompt = data.get("message")
|
314 |
-
if not prompt:
|
315 |
-
socketio.emit("log", {"message": "[ERROR]: Empty prompt."})
|
316 |
-
flash("Empty prompt.", "error")
|
317 |
-
return
|
318 |
-
run_agent(prompt, socketio)
|
319 |
-
|
320 |
-
# Expose the Flask app as "app" for Gunicorn
|
321 |
-
app = flask_app
|
322 |
|
323 |
if __name__ == "__main__":
|
324 |
-
|
|
|
1 |
+
from flask import Flask, render_template, request, redirect, url_for, send_from_directory, flash
|
2 |
from flask_socketio import SocketIO
|
|
|
3 |
import threading
|
4 |
+
import os
|
5 |
from dotenv import load_dotenv
|
6 |
+
import sqlite3
|
7 |
from werkzeug.utils import secure_filename
|
8 |
+
import traceback
|
9 |
|
10 |
# LangChain and agent imports
|
11 |
+
from langchain_community.chat_models.huggingface import ChatHuggingFace # if needed later
|
12 |
+
from langchain.agents import Tool
|
13 |
+
from langchain.agents.format_scratchpad import format_log_to_str
|
14 |
+
from langchain.agents.output_parsers import ReActJsonSingleInputOutputParser
|
15 |
+
from langchain_core.callbacks import CallbackManager, BaseCallbackHandler
|
16 |
+
from langchain_community.agent_toolkits.load_tools import load_tools
|
17 |
+
from langchain_core.tools import tool
|
18 |
+
from langchain_community.agent_toolkits import SQLDatabaseToolkit
|
19 |
+
from langchain.chains import LLMMathChain
|
20 |
+
from langchain import hub
|
21 |
+
from langchain_community.tools import DuckDuckGoSearchRun
|
22 |
+
|
23 |
+
# Agent requirements and type hints
|
24 |
+
from typing import Annotated, Literal, TypedDict, Any
|
25 |
from langchain_core.messages import AIMessage, ToolMessage
|
26 |
from pydantic import BaseModel, Field
|
27 |
from typing_extensions import TypedDict
|
28 |
+
from langgraph.graph import END, StateGraph, START
|
29 |
from langgraph.graph.message import AnyMessage, add_messages
|
30 |
from langchain_core.runnables import RunnableLambda, RunnableWithFallbacks
|
31 |
from langgraph.prebuilt import ToolNode
|
32 |
from langchain_core.prompts import ChatPromptTemplate
|
33 |
from langchain_community.utilities import SQLDatabase
|
|
|
|
|
|
|
34 |
|
35 |
# Load environment variables
|
36 |
load_dotenv()
|
|
|
39 |
UPLOAD_FOLDER = os.path.join(os.getcwd(), "uploads")
|
40 |
os.makedirs(UPLOAD_FOLDER, exist_ok=True)
|
41 |
BASE_DIR = os.path.abspath(os.path.dirname(__file__))
|
42 |
+
DATABASE_URI = f"sqlite:///{os.path.join(BASE_DIR, 'data', 'mydb.db')}"
|
43 |
+
print("DATABASE URI:", DATABASE_URI)
|
44 |
|
45 |
# API Keys from .env file
|
46 |
+
import os
|
47 |
os.environ["GROQ_API_KEY"] = os.getenv("GROQ_API_KEY")
|
48 |
os.environ["MISTRAL_API_KEY"] = os.getenv("MISTRAL_API_KEY")
|
49 |
|
50 |
+
# Global variables for dynamic agent and DB file path; initially None.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
51 |
agent_app = None
|
52 |
abs_file_path = None
|
53 |
+
db_path = None
|
54 |
|
55 |
+
# =============================================================================
|
56 |
+
# create_agent_app: Given a database path, initialize the agent workflow.
|
57 |
+
# =============================================================================
|
58 |
def create_agent_app(db_path: str):
|
59 |
+
# Use ChatGroq as our LLM here; you can swap to ChatMistralAI if preferred.
|
60 |
+
from langchain_groq import ChatGroq
|
61 |
+
llm = ChatGroq(model="llama3-70b-8192")
|
62 |
+
|
63 |
+
# -------------------------------------------------------------------------
|
64 |
+
# Define a tool for executing SQL queries, with an explicit description.
|
65 |
+
# -------------------------------------------------------------------------
|
66 |
+
@tool(description="Executes a SQL query on the connected SQLite database and returns the result.")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
67 |
def db_query_tool(query: str) -> str:
|
68 |
"""
|
69 |
+
Executes a SQL query on the connected SQLite database.
|
|
|
|
|
70 |
"""
|
71 |
try:
|
72 |
result = db_instance.run_no_throw(query)
|
73 |
+
return result if result else "Error: Query failed. Please rewrite your query and try again."
|
74 |
except Exception as e:
|
|
|
75 |
return f"Error: {str(e)}"
|
76 |
|
77 |
+
# -------------------------------------------------------------------------
|
78 |
+
# Pydantic model for final answer.
|
79 |
+
# -------------------------------------------------------------------------
|
80 |
class SubmitFinalAnswer(BaseModel):
|
81 |
+
final_answer: str = Field(..., description="The final answer to the user")
|
82 |
|
83 |
+
# -------------------------------------------------------------------------
|
84 |
+
# Define state type for our workflow.
|
85 |
+
# -------------------------------------------------------------------------
|
86 |
class State(TypedDict):
|
87 |
messages: Annotated[list[AnyMessage], add_messages]
|
88 |
|
89 |
+
# -------------------------------------------------------------------------
|
90 |
+
# Set up prompt templates for query checking and query generation.
|
91 |
+
# -------------------------------------------------------------------------
|
92 |
+
query_check_system = (
|
93 |
+
"You are a SQL expert with a strong attention to detail.\n"
|
94 |
+
"Double check the SQLite query for common mistakes, including:\n"
|
95 |
+
"- Using NOT IN with NULL values\n"
|
96 |
+
"- Using UNION when UNION ALL should have been used\n"
|
97 |
+
"- Using BETWEEN for exclusive ranges\n"
|
98 |
+
"- Data type mismatch in predicates\n"
|
99 |
+
"- Properly quoting identifiers\n"
|
100 |
+
"- Using the correct number of arguments for functions\n"
|
101 |
+
"- Casting to the correct data type\n"
|
102 |
+
"- Using the proper columns for joins\n\n"
|
103 |
+
"If there are any of the above mistakes, rewrite the query. If there are no mistakes, just reproduce the original query.\n"
|
104 |
+
"You will call the appropriate tool to execute the query after running this check."
|
105 |
+
)
|
106 |
+
query_check_prompt = ChatPromptTemplate.from_messages([
|
107 |
+
("system", query_check_system),
|
108 |
+
("placeholder", "{messages}")
|
109 |
+
])
|
110 |
+
query_check = query_check_prompt | llm.bind_tools([db_query_tool])
|
111 |
+
|
112 |
+
query_gen_system = (
|
113 |
+
"You are a SQL expert with a strong attention to detail.\n\n"
|
114 |
+
"Given an input question, output a syntactically correct SQLite query to run, then look at the results of the query and return the answer.\n\n"
|
115 |
+
"DO NOT call any tool besides SubmitFinalAnswer to submit the final answer.\n\n"
|
116 |
+
"When generating the query:\n"
|
117 |
+
"Output the SQL query that answers the input question without a tool call.\n"
|
118 |
+
"Unless the user specifies a specific number of examples they wish to obtain, always limit your query to at most 5 results.\n"
|
119 |
+
"You can order the results by a relevant column to return the most interesting examples in the database.\n"
|
120 |
+
"Never query for all the columns from a specific table, only ask for the relevant columns given the question.\n\n"
|
121 |
+
"If you get an error while executing a query, rewrite the query and try again.\n"
|
122 |
+
"If you get an empty result set, you should try to rewrite the query to get a non-empty result set.\n"
|
123 |
+
"NEVER make stuff up if you don't have enough information to answer the query... just say you don't have enough information.\n\n"
|
124 |
+
"If you have enough information to answer the input question, simply invoke the appropriate tool to submit the final answer to the user.\n"
|
125 |
+
"DO NOT make any DML statements (INSERT, UPDATE, DELETE, DROP etc.) to the database. Do not return any SQL query except answer."
|
126 |
+
)
|
127 |
+
query_gen_prompt = ChatPromptTemplate.from_messages([
|
128 |
+
("system", query_gen_system),
|
129 |
+
("placeholder", "{messages}")
|
130 |
+
])
|
131 |
+
query_gen = query_gen_prompt | llm.bind_tools([SubmitFinalAnswer])
|
132 |
+
|
133 |
+
# -------------------------------------------------------------------------
|
134 |
+
# Update database URI, create SQLDatabase connection.
|
135 |
+
# -------------------------------------------------------------------------
|
136 |
+
abs_db_path_local = os.path.abspath(db_path)
|
137 |
+
global DATABASE_URI
|
138 |
+
DATABASE_URI = abs_db_path_local
|
139 |
+
db_uri = f"sqlite:///{abs_db_path_local}"
|
140 |
+
print("db_uri", db_uri)
|
141 |
+
|
142 |
try:
|
143 |
+
db_instance = SQLDatabase.from_uri(db_uri)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
144 |
except Exception as e:
|
145 |
+
raise Exception(f"Failed to create SQLDatabase connection: {e}")
|
146 |
+
print("db_instance----->", db_instance)
|
147 |
+
|
148 |
+
# -------------------------------------------------------------------------
|
149 |
+
# Create SQL toolkit.
|
150 |
+
# -------------------------------------------------------------------------
|
151 |
+
toolkit_instance = SQLDatabaseToolkit(db=db_instance, llm=llm)
|
152 |
+
tools_instance = toolkit_instance.get_tools()
|
153 |
+
|
154 |
+
# -------------------------------------------------------------------------
|
155 |
+
# Define workflow nodes and fallback functions.
|
156 |
+
# -------------------------------------------------------------------------
|
157 |
def first_tool_call(state: State):
|
158 |
+
return {"messages": [AIMessage(content="", tool_calls=[{"name": "sql_db_list_tables", "args": {}, "id": "tool_abcd123"}])]}
|
159 |
|
160 |
def handle_tool_error(state: State):
|
161 |
tool_calls = state["messages"][-1].tool_calls
|
|
|
170 |
try:
|
171 |
message = query_gen.invoke(state)
|
172 |
except Exception as e:
|
173 |
+
raise Exception(f"Exception in query_gen_node: {e}")
|
|
|
174 |
tool_messages = []
|
175 |
if message.tool_calls:
|
176 |
for tc in message.tool_calls:
|
|
|
181 |
))
|
182 |
return {"messages": [message] + tool_messages}
|
183 |
|
184 |
+
def should_continue(state: State) -> Literal[END, "correct_query", "query_gen"]:
|
185 |
+
messages = state["messages"]
|
186 |
+
last_message = messages[-1]
|
187 |
if getattr(last_message, "tool_calls", None):
|
188 |
return END
|
189 |
if last_message.content.startswith("Error:"):
|
|
|
193 |
def model_check_query(state: State):
|
194 |
return {"messages": [query_check.invoke({"messages": [state["messages"][-1]]})]}
|
195 |
|
196 |
+
# -------------------------------------------------------------------------
|
197 |
+
# Get tools for listing tables and fetching schema.
|
198 |
+
# -------------------------------------------------------------------------
|
199 |
+
list_tables_tool = next((t for t in tools_instance if t.name == "sql_db_list_tables"), None)
|
200 |
schema_tool = next((t for t in tools_instance if t.name == "sql_db_schema"), None)
|
201 |
model_get_schema = llm.bind_tools([schema_tool])
|
202 |
|
203 |
workflow = StateGraph(State)
|
204 |
workflow.add_node("first_tool_call", first_tool_call)
|
205 |
+
workflow.add_node("list_tables_tool", create_tool_node_with_fallback([list_tables_tool]))
|
206 |
workflow.add_node("get_schema_tool", create_tool_node_with_fallback([schema_tool]))
|
207 |
+
workflow.add_node("model_get_schema", lambda state: {"messages": [model_get_schema.invoke(state["messages"])]})
|
|
|
208 |
workflow.add_node("query_gen", query_gen_node)
|
209 |
workflow.add_node("correct_query", model_check_query)
|
210 |
workflow.add_node("execute_query", create_tool_node_with_fallback([db_query_tool]))
|
|
|
220 |
|
221 |
return workflow.compile()
|
222 |
|
223 |
+
# =============================================================================
|
224 |
+
# create_app: The application factory.
|
225 |
+
# =============================================================================
|
226 |
+
def create_app():
|
227 |
+
flask_app = Flask(__name__, static_url_path='/uploads', static_folder='uploads')
|
228 |
+
socketio = SocketIO(flask_app, cors_allowed_origins="*")
|
229 |
+
|
230 |
+
if not os.path.exists(UPLOAD_FOLDER):
|
231 |
+
os.makedirs(UPLOAD_FOLDER)
|
232 |
+
flask_app.config['UPLOAD_FOLDER'] = UPLOAD_FOLDER
|
233 |
+
flask_app.config['SECRET_KEY'] = os.getenv("FLASK_SECRET_KEY", "mysecretkey")
|
234 |
+
|
235 |
+
@flask_app.route("/files/<path:filename>")
|
236 |
+
def uploaded_file(filename):
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
237 |
try:
|
238 |
+
return send_from_directory(flask_app.config['UPLOAD_FOLDER'], filename)
|
239 |
except Exception as e:
|
240 |
+
flash(f"Could not send file: {str(e)}", "error")
|
241 |
+
return redirect(url_for("index"))
|
242 |
+
|
243 |
+
def run_agent(prompt, socketio):
|
244 |
+
global agent_app, abs_file_path, db_path
|
245 |
+
if not abs_file_path:
|
246 |
+
socketio.emit("log", {"message": "[ERROR]: No DB file uploaded."})
|
247 |
+
socketio.emit("final", {"message": "No database available. Please upload one and try again."})
|
248 |
+
flash("No database available. Please upload one and try again.", "error")
|
249 |
+
return
|
250 |
+
try:
|
251 |
+
if agent_app is None:
|
252 |
+
socketio.emit("log", {"message": "[INFO]: Initializing agent for the first time..."})
|
253 |
+
agent_app = create_agent_app(abs_file_path)
|
254 |
+
socketio.emit("log", {"message": "[INFO]: Agent initialized."})
|
255 |
+
flash("Agent initialized.", "info")
|
256 |
+
query = {"messages": [("user", prompt)]}
|
257 |
+
result = agent_app.invoke(query)
|
258 |
+
try:
|
259 |
+
result = result["messages"][-1].tool_calls[0]["args"]["final_answer"]
|
260 |
+
except Exception as e:
|
261 |
+
result = "Query failed or no valid answer found."
|
262 |
+
flash("Query failed or no valid answer found.", "warning")
|
263 |
+
print("final_answer------>", result)
|
264 |
+
socketio.emit("final", {"message": result})
|
265 |
+
except Exception as e:
|
266 |
+
error_message = f"Generation failed: {str(e)}"
|
267 |
+
socketio.emit("log", {"message": f"[ERROR]: {error_message}"})
|
268 |
+
socketio.emit("final", {"message": "Generation failed."})
|
269 |
+
flash(error_message, "error")
|
270 |
+
traceback.print_exc()
|
271 |
+
|
272 |
+
@flask_app.route("/")
|
273 |
+
def index():
|
274 |
+
return render_template("index.html")
|
275 |
+
|
276 |
+
@flask_app.route("/generate", methods=["POST"])
|
277 |
+
def generate():
|
278 |
+
try:
|
279 |
+
socketio.emit("log", {"message": "[STEP]: Entering query generation..."})
|
280 |
+
data = request.json
|
281 |
+
prompt = data.get("prompt", "")
|
282 |
+
socketio.emit("log", {"message": f"[INFO]: Received prompt: {prompt}"})
|
283 |
+
thread = threading.Thread(target=run_agent, args=(prompt, socketio))
|
284 |
+
socketio.emit("log", {"message": f"[INFO]: Starting thread: {thread}"})
|
285 |
+
thread.start()
|
286 |
+
flash("Query submitted successfully.", "info")
|
287 |
+
return "OK", 200
|
288 |
+
except Exception as e:
|
289 |
+
error_message = f"[ERROR]: {str(e)}"
|
290 |
+
socketio.emit("log", {"message": error_message})
|
291 |
+
flash(error_message, "error")
|
292 |
+
return "ERROR", 500
|
293 |
+
|
294 |
+
@flask_app.route("/upload", methods=["GET", "POST"])
|
295 |
+
def upload():
|
296 |
+
global abs_file_path, agent_app, db_path
|
297 |
+
try:
|
298 |
+
if request.method == "POST":
|
299 |
+
file = request.files.get("file")
|
300 |
+
if not file:
|
301 |
+
flash("No file uploaded.", "error")
|
302 |
+
return "No file uploaded", 400
|
303 |
+
filename = secure_filename(file.filename)
|
304 |
+
if filename.endswith('.db'):
|
305 |
+
db_path = os.path.join(flask_app.config['UPLOAD_FOLDER'], "uploaded.db")
|
306 |
+
print("Saving file to:", db_path)
|
307 |
file.save(db_path)
|
308 |
+
abs_file_path = os.path.abspath(db_path)
|
309 |
+
agent_app = None # Reset the agent so it is lazily reinitialized on next query.
|
310 |
+
print(f"[INFO]: File '{filename}' uploaded. Agent will be initialized on first query.")
|
311 |
socketio.emit("log", {"message": f"[INFO]: Database file '{filename}' uploaded."})
|
312 |
+
flash(f"Database file '{filename}' uploaded successfully.", "info")
|
313 |
return redirect(url_for("index"))
|
314 |
+
return render_template("upload.html")
|
315 |
+
except Exception as e:
|
316 |
+
error_message = f"[ERROR]: {str(e)}"
|
317 |
+
print(error_message)
|
318 |
+
flash(error_message, "error")
|
319 |
+
socketio.emit("log", {"message": error_message})
|
320 |
+
return render_template("upload.html")
|
321 |
+
|
322 |
+
return flask_app, socketio
|
323 |
+
|
324 |
+
# =============================================================================
|
325 |
+
# Create the app for Gunicorn compatibility.
|
326 |
+
# =============================================================================
|
327 |
+
app, socketio_instance = create_app()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
328 |
|
329 |
if __name__ == "__main__":
|
330 |
+
socketio_instance.run(app, debug=True)
|