How to create SQLagent using django models

[thondeboer]

Checked other resources

• I added a very descriptive title to this question.
• I searched the LangChain documentation with the integrated search.
• I used the GitHub search to find a similar question and didn't find it.

Commit to Help

• I commit to help with one of those options 👆

Example Code

No code used

Description

I am trying to learn how to use Django model and their definitions to enable searching and reasoning over DJANGO models directly, not using postgresql language or if so, how to automatically extract the database information, including the ForeignKey definitions.

I have a Django model definition for about 30 tables, which have very elaborate help_texts that could be useful for a LLM model to understand the contents of the table and they also contain many foreignkeys to link to other database, as per usual in a SQL database.

I can extract those fields and attempt to make a SQLSchema, but the names of the fields in a django model, don't always correspond with the column names, especially when ForeignKeys are involved. A field called "concept" that is a foreign_key to a another model, will be column "concept_id" in the table, so this complicates the SQLschema definition (can ofcourse program it to add _id, for type==foreignkey etc.

But there are also other complicated many_to_many relationships with linker tables etc. that DJANGO querying can directly use....

So, I guess I am looking for a LLM that can create DJANGO query languge code to execute or another method to efficiently do DJANGO-like querying.

Is SQLagent using direct access to postgres DB my only viable option (Other than me training a model specifically in DJANGO query langugage)

System Info

System Information

OS: Linux
OS Version: #40~22.04.1-Ubuntu SMP PREEMPT_DYNAMIC Thu Nov 16 10:53:04 UTC 2
Python Version: 3.10.12 (main, Nov 20 2023, 15:14:05) [GCC 11.4.0]

Package Information

langchain_core: 0.1.45
langchain: 0.1.16
langchain_community: 0.0.34
langsmith: 0.1.50
langchain_fireworks: 0.1.2
langchain_mistralai: 0.1.2
langchain_openai: 0.1.3
langchain_text_splitters: 0.0.1
langchainhub: 0.1.15

Packages not installed (Not Necessarily a Problem)

The following packages were not found:

langgraph
langserve

[dosubot[bot]]

To achieve Django-like querying and interaction with your Django models directly, leveraging the LangChain framework's create_sql_agent function and the SQLDatabaseToolkit could be a viable solution. These tools are designed to facilitate the interaction with SQL databases, including handling ForeignKey definitions and the discrepancies between Django model field names and actual database column names.

The SQLDatabaseToolkit specifically provides functionalities for executing queries, retrieving database schema information, and ensuring compatibility with Django's ORM system. This means you can configure the SQLDatabase instance to connect to Django's database settings and adapt the toolkit to work with Django model fields.

For handling many-to-many relationships and linker tables, which are directly utilized in Django querying, the documentation and examples provided in the LangChain framework should offer guidance on configuring custom prompt templates. These templates can help in translating Django-like querying syntax into actual SQL queries that the SQLDatabaseToolkit can execute.

In summary, using the LangChain framework's create_sql_agent and SQLDatabaseToolkit, you can efficiently perform Django-like querying without relying solely on direct PostgreSQL database access. This approach allows you to leverage the elaborate help_texts and ForeignKey relationships defined in your Django models for more intuitive and effective database interactions.

Details

Sources

• How to build a LangChain agents that can interact with data from a postgresql database in the django rest framework backend
• libs/community/langchain_community/agent_toolkits/sql/base.py
• libs/community/langchain_community/tools/sql_database/tool.py
• libs/community/langchain_community/agent_toolkits/sql/toolkit.py
• libs/langchain/langchain/chains/sql_database/prompt.py
• libs/langchain/langchain/chains/sql_database/query.py
• docs/docs/integrations/toolkits/sql_database.ipynb

About Dosu
This response is meant to be useful and save you time. It is not meant to be a precise solution, but rather a starting point for your own research.

Help me be more useful! Please leave a 👍 if this is helpful and 👎 if it is irrelevant.

To continue the conversation, mention @dosu.

[gt732]

@thondeboer Did you make any progress with this? I'm in a similar boat.