File size: 8,870 Bytes
3a7b7ae
 
 
 
 
 
6e94c23
3a7b7ae
 
 
 
 
 
 
 
 
6e94c23
 
 
3a7b7ae
 
 
 
 
 
6e94c23
 
 
 
3a7b7ae
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
6e94c23
3a7b7ae
6e94c23
 
 
 
 
 
 
 
 
 
 
 
 
 
 
3a7b7ae
6e94c23
 
 
 
 
3a7b7ae
6e94c23
 
 
 
 
 
3a7b7ae
 
6e94c23
3a7b7ae
 
6e94c23
3a7b7ae
 
6e94c23
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
3a7b7ae
6e94c23
3a7b7ae
 
 
 
 
 
6e94c23
3a7b7ae
6e94c23
 
 
 
 
 
 
 
 
 
3a7b7ae
 
6e94c23
 
3a7b7ae
 
 
 
 
 
 
 
 
 
 
6e94c23
3a7b7ae
 
6e94c23
 
 
 
 
 
 
 
 
 
 
 
3a7b7ae
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
from playwright.async_api import async_playwright
from langchain_community.agent_toolkits import PlayWrightBrowserToolkit
from dotenv import load_dotenv
import os
import requests
from langchain.agents import Tool
from langchain.tools import StructuredTool
from langchain_community.agent_toolkits import FileManagementToolkit
from langchain_community.tools.wikipedia.tool import WikipediaQueryRun
from langchain_experimental.tools import PythonREPLTool
from langchain_community.utilities import GoogleSerperAPIWrapper
from langchain_community.utilities.wikipedia import WikipediaAPIWrapper
from langchain.agents.agent_toolkits import SQLDatabaseToolkit
from langchain.sql_database import SQLDatabase
import uuid
import sqlite3
from langchain_openai import ChatOpenAI
from pydantic import BaseModel, Field
    
load_dotenv(override=True)
pushover_token = os.getenv("PUSHOVER_TOKEN")
pushover_user = os.getenv("PUSHOVER_USER")
pushover_url = "https://api.pushover.net/1/messages.json"
serper = GoogleSerperAPIWrapper()

class GetUserHistoryInput(BaseModel):
    username: str = Field(..., description="The username to fetch history for")
    query_text: str = Field(None, description="Natural language query about history to retrieve")

async def playwright_tools():
    playwright = await async_playwright().start()
    browser = await playwright.chromium.launch(headless=False)
    toolkit = PlayWrightBrowserToolkit.from_browser(async_browser=browser)
    return toolkit.get_tools(), browser, playwright


def push(text: str):
    """Send a push notification to the user"""
    requests.post(pushover_url, data = {"token": pushover_token, "user": pushover_user, "message": text})
    return "success"


def get_file_tools():
    toolkit = FileManagementToolkit(root_dir="sandbox")
    return toolkit.get_tools()


def format_search_history(formatted_results, success_criteria=None):

    if not formatted_results:
        return "No search history found."
    
    
    prompt = f"""

    You are a formatting expert. Format the following search history data:

    

    {formatted_results}

    

    Format this data according to these requirements:

    {success_criteria if success_criteria else "Make it readable and well-organized."}

    

    Use plain text only (no HTML or Markdown). Return ONLY the formatted output without any explanation.

    You can use Unicode symbols, ASCII art, or emojis if appropriate.

    """

    llm = ChatOpenAI(model="gpt-4o-mini", temperature=0.3)
    response = llm.invoke(prompt)
    return response.content
    
def get_user_history(query_text: str = None, username: str = None):
    """

    Retrieves search history from the database based on natural language query.

    Converts natural language to SQL, executes the query, and formats results.

    

    Args:

        username: The username to fetch history for (required)

        query_text: Natural language description of what history to retrieve (e.g. "show my searches about AI")

    """
    try:
        
        conn = sqlite3.connect("query_log.db")
        cursor = conn.cursor()
        
        cursor.execute("SELECT name FROM sqlite_master WHERE type='table' AND name='search_history'")
        if not cursor.fetchone():
            return "Error: The search history table does not exist yet. Try making some searches first."
        
        cursor.execute("PRAGMA table_info(search_history)")
        schema_info = cursor.fetchall()
        schema_description = "\n".join([f"- {col[1]} ({col[2]})" for col in schema_info])
        
        default_query = "SELECT * FROM search_history WHERE username = ? ORDER BY timestamp DESC"
        if query_text:
            llm = ChatOpenAI(model="gpt-4o-mini",temperature=0)
            
            prompt = f"""

            Table name: search_history

            Table schema:

            {schema_description}

            

            User question: {query_text}

            Username filter: {username} (Always filter by this username for security)

            

            Generate a SQL query that answers the user's question while ALWAYS including 'WHERE username = ?' 

            in the query for security. Return ONLY the SQL query without explanation or backticks.

            """
            
            max_tries = 3
            sql_query = None
            
            for attempt in range(max_tries):
                try:
                    #print(f"Generating SQL (attempt {attempt+1}/{max_tries})...")
                    sql_response = llm.invoke(prompt)
                    generated_query = sql_response.content.strip()
                    
                    if "WHERE username = ?" not in generated_query and "where username = ?" not in generated_query:
                        generated_query = f"{generated_query} WHERE username = ?"
                    
                    #print(f"Generated query: {generated_query}")
                    
                    if generated_query.count("WHERE username = ?") > 1 or generated_query.count("where username = ?") > 1:
                        generated_query = generated_query.replace("WHERE username = ?", "").replace("where username = ?", "")
                        if "WHERE" not in generated_query and "where" not in generated_query:
                            generated_query = f"{generated_query} WHERE username = ?"
                        else:
                            generated_query = generated_query.replace("WHERE", "WHERE username = ? AND").replace("where", "where username = ? AND")
                    
                    conn.execute("EXPLAIN " + generated_query, (username,))
                    
                    sql_query = generated_query
                    break
                except Exception as e:
                    error_message = str(e)
                    print(f"SQL error on attempt {attempt+1}: {error_message}")
                    
                    prompt += f"\nPrevious attempt failed with error: {error_message}\nPlease fix and try again."
                    
                    if attempt == max_tries - 1:
                        print("Falling back to default query")
                        sql_query = default_query
        else:
            sql_query = default_query
        
        print(f"Executing query: {sql_query}")
        cursor.execute(sql_query, (username,))
        results = cursor.fetchall()
        
        formatted_results = []
        for row in results:
            record = {}
            for i, col in enumerate(cursor.description):
                record[col[0]] = row[i]
            formatted_results.append(record)
        
        conn.close()
        
        if not formatted_results:
            return "You don't have any search history records matching your query."
        
        output = "Here's your search history:\n\n"
        output += format_search_history(formatted_results)
        

        return output
        
    except Exception as e:
        print(f"[ERROR] Database error: {str(e)}")
        return f"Sorry, I encountered an error while retrieving your search history: {str(e)}"
        

async def other_tools():
    push_tool = Tool(name="send_push_notification", func=push, description="Use this tool when you want to send a push notification")
    file_tools = get_file_tools()

    tool_search =Tool(
        name="search",
        func=serper.run,
        description="Use this tool when you want to get the results of an online web search"
    )
    
    get_history_tool = StructuredTool(
                            name="get_user_history",
                            func=get_user_history,
                            description="""Use this tool when the user wants to view or query their search history.

                            

                            Examples of when to use this tool:

                            - User asks about their search history

                            - User wants to see previous searches

                            - User asks for specific searches (like "show my searches about AI")

                            

                            Required parameters:

                            - query_text: The user's natural language question about their history

                            - username: The username of the current user

                            """,
                            args_schema=GetUserHistoryInput
                        )
    wikipedia = WikipediaAPIWrapper()
    wiki_tool = WikipediaQueryRun(api_wrapper=wikipedia)

    python_repl = PythonREPLTool()
    
    return file_tools + [push_tool, tool_search, python_repl,  wiki_tool, get_history_tool]