Advanced text to sql sample rows

[osamadel]

Add NLSQLRetrieverWithSampleRows and NLSQLTableQueryEngineWithSampleRows to sql_query.py

Description

This is a very minor addition to the sql_query.py module, adding two classes NLSQLRetrieverWithSampleRows and NLSQLTableQueryEngineWithSampleRows which inherit from NLSQLRetriever and BaseSQLTableQueryEngine respectively. This addition is based on the great tutorial by Jerry Liu which then was updated to Workflows in this notebook. The motivation is that, in both the tutorial and the notebook, there is an implementation to add sample rows inside the context to improve the response of the LLM, by giving it a taste of the content of each retrieved table, which makes the SQL more resilient against slight variations in the values to be queried using SQL, for example, which value to use in a "where" clause under a Leave Type column, is it "Business Trip", "BUSINESS", or "TRIP" or who knows?! Giving a sample of the rows inside the context improves the response of the LLM in that regard. The implementation in the tutorial is done using Workflows or QueryPipeline, where most of it (missing only the sample rows) is already implemented in the NLSQLTableQueryEngine class, so why not add the sample rows to it?!

Fixes # Missing adding sample rows to the context in the NLSQLTableQueryEngine class

New Package?

Did I fill in the tool.llamahub section in the pyproject.toml and provide a detailed README.md for my new integration or package?

• Yes
• No

Version Bump?

Did I bump the version in the pyproject.toml file of the package I am updating? (Except for the llama-index-core package)

• Yes
• No

Type of Change

Please delete options that are not relevant.

• New feature (non-breaking change which adds functionality)
• This change requires a documentation update

How Has This Been Tested?

Your pull-request will likely not be merged unless it is covered by some form of impactful unit testing.

• I added new unit tests to cover this change
• I believe this change is already covered by existing unit tests

Suggested Checklist:

• I have performed a self-review of my own code
• I have commented my code, particularly in hard-to-understand areas
• I have made corresponding changes to the documentation
• I have added Google Colab support for the newly added notebooks.
• My changes generate no new warnings
• I have added tests that prove my fix is effective or that my feature works
• New and existing unit tests pass locally with my changes
• I ran make format; make lint to appease the lint gods

[logan-markewich]

ooc, does this actually require a brand new class? We can't just use the existing class and add an arg for the retriever?

[osamadel]

Okay, in a nutshell, we can, but the issue is that both NLSQLTableQueryEngine and SQLTableRetrieverQueryEngine use the NLSQLRetriever with its defaults arguments.
For example:

self._sql_retriever = NLSQLRetriever(
            sql_database,
            llm=llm,
            text_to_sql_prompt=text_to_sql_prompt,
            context_query_kwargs=context_query_kwargs,
            tables=tables,
            sql_parser_mode=SQLParserMode.PGVECTOR,
            context_str_prefix=context_str_prefix,
            sql_only=sql_only,
            callback_manager=callback_manager,

So in order to make the distinction clear, I opted for a separate class.
If we are to go with the argument option, we'll probably still need the query engine class that internally instantiate a NLSQLRetriever with the sample rows argument used.

What do you think?