Text-to-SQL with LangChain (1) - Chain

pip install --upgrade --quiet  langchain langchain-community langchain-openai

import os

# Set API key as an environment variable
with open('.openai') as f:
    os.environ['OPENAI_API_KEY'] = f.read().strip()

import sqlite3

# Database file name
db_file = 'Chinook.db'
sql_file = 'Chinook_Sqlite.sql'

# Open database connection
with sqlite3.connect(db_file) as conn:
    cursor = conn.cursor()
    
    # Read SQL file
    with open(sql_file, 'r', encoding='utf-8') as file:
        sql_script = file.read()

    # Execute SQL script
    cursor.executescript(sql_script)
    
    # Commit and save database
    conn.commit()

from langchain_community.utilities import SQLDatabase

# Specify the database to connect to
db = SQLDatabase.from_uri("sqlite:///Chinook.db")

print(db.dialect)  # Database type
print(db.get_usable_table_names())  # List of usable table names

# Execute an SQL statement
db.run("SELECT * FROM Artist LIMIT 10;")

from langchain.chains import create_sql_query_chain
from langchain_openai import ChatOpenAI

llm = ChatOpenAI(model="gpt-3.5-turbo", temperature=0)
chain = create_sql_query_chain(llm, db)

response = chain.invoke({"question": "How many employees are there?"})
print(response)

db.run(response)

from langchain_community.tools.sql_database.tool import QuerySQLDataBaseTool

execute_query = QuerySQLDataBaseTool(db=db)
write_query = create_sql_query_chain(llm, db)
chain = write_query | execute_query
chain.invoke({"question": "How many employees are there?"})

from operator import itemgetter
from langchain_core.output_parsers import StrOutputParser
from langchain_core.prompts import PromptTemplate
from langchain_core.runnables import RunnablePassthrough

answer_prompt = PromptTemplate.from_template(
    """Based on the user's question, the corresponding SQL query, and the SQL result, answer the user's question.

Question: {question}
SQLQuery: {query}
SQLResult: {result}
Answer: """
)

# Chain to generate answers using the above prompt
answer = answer_prompt | llm | StrOutputParser()

# Chain to pass the SQL query result to the answer chain
chain = (
    RunnablePassthrough.assign(
        query=write_query
    ).assign(
        result=itemgetter("query") | execute_query
    )
    | answer
)

chain.invoke({"question": "How many employees are there?"})

def create_sql_query_chain(
    llm: BaseLanguageModel,
    db: SQLDatabase,
    prompt: Optional[BasePromptTemplate] = None,
    k: int = 5,
) -> Runnable[Union[SQLInput, SQLInputWithTables, Dict[str, Any]], str]:
    """Create a chain that generates SQL queries.

    *Security Note*: This chain generates SQL queries for the given database.

        The SQLDatabase class provides a get_table_info method that can be used
        to get column information as well as sample data from the table.

        To mitigate risk of leaking sensitive data, limit permissions
        to read and scope to the tables that are needed.

        Optionally, use the SQLInputWithTables input type to specify which tables
        are allowed to be accessed.

        Control access to who can submit requests to this chain.

        See https://python.langchain.com/docs/security for more information.

    Args:
        llm: The language model to use.
        db: The SQLDatabase to generate the query for.
        prompt: The prompt to use. If none is provided, will choose one
            based on dialect. Defaults to None. See Prompt section below for more.
        k: The number of results per select statement to return. Defaults to 5.

    Returns:
        A chain that takes in a question and generates a SQL query that answers
        that question.

    Example:

        .. code-block:: python

            # pip install -U langchain langchain-community langchain-openai
            from langchain_openai import ChatOpenAI
            from langchain.chains import create_sql_query_chain
            from langchain_community.utilities import SQLDatabase

            db = SQLDatabase.from_uri("sqlite:///Chinook.db")
            llm = ChatOpenAI(model="gpt-3.5-turbo", temperature=0)
            chain = create_sql_query_chain(llm, db)
            response = chain.invoke({"question": "How many employees are there"})

    Prompt:
        If no prompt is provided, a default prompt is selected based on the SQLDatabase dialect. If one is provided, it must support input variables:
            * input: The user question plus suffix "\nSQLQuery: " is passed here.
            * top_k: The number of results per select statement (the `k` argument to
                this function) is passed in here.
            * table_info: Table definitions and sample rows are passed in here. If the
                user specifies "table_names_to_use" when invoking chain, only those
                will be included. Otherwise, all tables are included.
            * dialect (optional): If dialect input variable is in prompt, the db
                dialect will be passed in here.

        Here's an example prompt:

        .. code-block:: python

            from langchain_core.prompts import PromptTemplate

            template = '''Given an input question, first create a syntactically correct {dialect} query to run, then look at the results of the query and return the answer.
            Use the following format:

            Question: "Question here"
            SQLQuery: "SQL Query to run"
            SQLResult: "Result of the SQLQuery"
            Answer: "Final answer here"

            Only use the following tables:

            {table_info}.

            Question: {input}'''
            prompt = PromptTemplate.from_template(template)
    """  # noqa: E501
    if prompt is not None:
        prompt_to_use = prompt
    elif db.dialect in SQL_PROMPTS:
        prompt_to_use = SQL_PROMPTS[db.dialect]
    else:
        prompt_to_use = PROMPT
    if {"input", "top_k", "table_info"}.difference(prompt_to_use.input_variables):
        raise ValueError(
            f"Prompt must have input variables: 'input', 'top_k', "
            f"'table_info'. Received prompt with input variables: "
            f"{prompt_to_use.input_variables}. Full prompt:\n\n{prompt_to_use}"
        )
    if "dialect" in prompt_to_use.input_variables:
        prompt_to_use = prompt_to_use.partial(dialect=db.dialect)

    inputs = {
        "input": lambda x: x["question"] + "\nSQLQuery: ",
        "table_info": lambda x: db.get_table_info(
            table_names=x.get("table_names_to_use")
        ),
    }
    return (
        RunnablePassthrough.assign(**inputs)  # type: ignore
        | (
            lambda x: {
                k: v
                for k, v in x.items()
                if k not in ("question", "table_names_to_use")
            }
        )
        | prompt_to_use.partial(top_k=str(k))
        | llm.bind(stop=["\nSQLResult:"])
        | StrOutputParser()
        | _strip
    )

import os
from operator import itemgetter

from langchain.chains import create_sql_query_chain
from langchain_community.tools.sql_database.tool import QuerySQLDataBaseTool
from langchain_community.utilities import SQLDatabase
from langchain_core.output_parsers import StrOutputParser
from langchain_core.prompts import PromptTemplate
from langchain_core.runnables import RunnablePassthrough
from langchain_openai import ChatOpenAI

# Set API key as an environment variable
with open(".openai") as f:
    os.environ["OPENAI_API_KEY"] = f.read().strip()

# Specify the database to connect to
db = SQLDatabase.from_uri("sqlite:///Chinook.db")

# LLM settings
llm = ChatOpenAI(model="gpt-3.5-turbo", temperature=0)

# Tool to execute SQL statements
execute_query = QuerySQLDataBaseTool(db=db)

# Chain to generate SQL statements
write_query = create_sql_query_chain(llm, db)

# Prompt template to generate answers in natural language
answer_prompt = PromptTemplate.from_template(
    """Based on the user's question, the corresponding SQL query, and the SQL result, answer the user's question.

Question: {question}
SQLQuery: {query}
SQLResult: {result}
Answer: """
)

# Chain to generate answers using the above prompt
answer = answer_prompt | llm | StrOutputParser()

# Chain to pass the SQL query result to the answer chain
chain = (
    RunnablePassthrough.assign(
        query=write_query
    ).assign(
        result=itemgetter("query") | execute_query
    )
    | answer
)

# Execute the query
chain.invoke({"question": "How many employees are there?"})