Code Duplication in 'gpt-engineer/steps.py'

[Madhav-MKNC]

Description The code contains sections with similar logic or repeated patterns, such as the gen_spec, gen_unit_tests, and gen_code functions. This duplication violates the DRY (Don't Repeat Yourself) principle, which advocates for avoiding redundant code and promoting code reuse.

Here's an overview of the shared logic:

1. All three functions start with a similar sequence of messages being initialized, which includes the system prompt and instructions.
2. They make use of the ai.next() method to generate AI responses based on the provided messages.
3. They retrieve or assign values to the dbs.memory dictionary using specific keys.
4. They save the generated content to the workspace using the to_files() function.

Based on this analysis, we can extract the shared logic into a separate function or utility. Here's an example of how it can be refactored:

def perform_ai_task(ai: AI, dbs: DBs, messages: List[Dict[str, str]], output_key: str) -> List[Dict[str, str]]:
    """
    Perform an AI task by generating responses based on the given messages.
    Save the output to the workspace using the specified output_key.
    Return the updated messages.
    """
    messages = messages + [ai.fassistant(dbs.workspace[output_key])]
    messages = ai.next(messages, dbs.identity["use_qa"])
    dbs.workspace[output_key] = messages[-1]["content"]
    to_files(dbs.workspace[output_key], dbs.workspace)
    return messages

The updated gen_spec(), gen_unit_tests() and gen_code() functions:

def gen_spec(ai: AI, dbs: DBs):
    messages = [
        ai.fsystem(f"Instructions: {dbs.input['main_prompt']}"),
    ]
    return perform_ai_task(ai, dbs, messages, "specification")

def gen_unit_tests(ai: AI, dbs: DBs):
    messages = [
        ai.fsystem(f"Instructions: {dbs.input['main_prompt']}"),
        ai.fsystem(f"Specification:\n\n{dbs.memory['specification']}"),
    ]
    return perform_ai_task(ai, dbs, messages, "unit_tests")

def gen_code(ai: AI, dbs: DBs):
    messages = [
        ai.fsystem(f"Instructions: {dbs.input['main_prompt']}"),
        ai.fsystem(f"Specification:\n\n{dbs.memory['specification']}"),
        ai.fsystem(f"Unit tests:\n\n{dbs.memory['unit_tests']}"),
    ]
    return perform_ai_task(ai, dbs, messages, "code")

Shall I submit a PR for this?