Prompt Manager

Usage

prompt_manager.get_model_list()

Description

Returns a list of available models in local ollama installation.

Returns:

List of model names

Note

The models are expected to be installed locally and accessible via the ollama python library. The function returns a list of model names that can be used for generating SQL queries.

You can check if ollama is installed by running the following command in your terminal:

ollama list

NAME                    ID              SIZE    MODIFIED    

starcoder2:7b           1550ab21b10d    4.0 GB  5 weeks ago     

codellama:7b            8fdf8f752f6e    3.8 GB  5 weeks ago     

sqlcoder:7b             77ac14348387    4.1 GB  5 weeks ago     

duckdb-nsql:latest      3ed734989690    3.8 GB  5 weeks ago

prompt_manager.get_system_prompt()

Description

Returns a system prompt template for SQL query generation. The system prompt provides context to the model about its role and the task it needs to perform. The prompt includes a placeholder for metadata, which will be replaced with the actual database schema information. The metadata is expected to be a list of tuples, each containing information about a table and its columns. The system prompt also includes guidelines for the model to follow when generating the SQL query.

prompt_manager.get_guideline_prompts()

Description

Returns a set of guidelines for the model to follow when generating SQL queries. The guidelines specify that the result should be a valid SQL query that can be executed on the provided schema. The model is instructed to return only the SQL query as a string. The guidelines are designed to ensure that the generated SQL query is syntactically correct and adheres to SQL standards. The guidelines are appended to the system prompt to provide additional context to the model.

prompt_manager.generate_sql(prompt: str, model: str)

Description

Generates a SQL query from a natural language prompt using the specified model. The function first retrieves the metadata from the database using the get_all_metadata function. The metadata is then used to replace the placeholder in the system prompt template. The model is then called to generate the SQL query based on the provided prompt and system prompt. The generated SQL query is returned along with the metadata used for generation.

Parameters:

• prompt (str) – The natural language prompt.

• model (str) – The model to use for query generation (default: duckdb-nsql:latest).

Returns:

A tuple containing the generated SQL query and the metadata used for generation.

Raises:

Exception – If the model is not found or if there is an error during SQL generation.

Installation

Ollama Installation: https://ollama.com/download

Ollama CLI commands: https://github.com/ollama/ollama?tab=readme-ov-file#quickstart

ollama
Usage
    ollama [flags]
    ollama [command]

Available Commands
serve       Start ollama
create      Create a model from a Modelfile
show        Show information for a model
run         Run a model
pull        Pull a model from a registry
push        Push a model to a registry
list        List models
ps          List running models
cp          Copy a model
rm          Remove a model
help        Help about any command

Flags:
-h, --help      help for ollama
-v, --version   Show version information

Use "ollama [command] --help" for more information about a command.

Ollama Python Package

pip install ollama