Code Check

Do you have a problem with duplicated functions in your coding project, such as Java? This may be the solution of finding them.

Specify the location of your repository in the local-config.properties file (just make a copy of config.properties) and run the code! Then you will get a file in the "results"-folder with all the places where functions match.

Features

• Query an LLM (large language model) on complicated situations where 2 or more functions could be similar.
• Configuration - it's easy to change the configurations.
• Comparison - matching through names of lines and functions.
• Tested on both UNIX and Windows.

Testing

To run the test files you can run this command with Gradle:

./gradlew test

To run tests directly in IntelliJ or Eclipse (or a program of your choice). Just run the testing files - and it should be working.

Getting started with large language model (LLM)

Download the language model

These are only suggestions, you may use any source and model as you want.

• Either download the bin-file directly from Hugging Face which we used for testing: ggml-model-gpt4all-falcon-q4_0.bin (4.06 GB)

• Or choose downloads from the website for gpt4all-falcon-ggml: https://huggingface.co/nomic-ai/gpt4all-falcon-ggml/tree/main

• You mou may also download any other model of your taste, and use that one instead.

Setting up the configuration

Create and update the local file local-config.properties (make a copy of config.properties and edit it.), then after running you will get a file in a "results" folder with places where functions/methods of different code are being similar or matching.

Adding the model and setting the configuration to use LLM

1. Move the downloaded LLM into a new folder called models in the main repository.
2. In local-config.properties, set LLM_FILE to the name of your downloaded model, for example model/ggml-model-gpt4all-falcon-q4_0.bin.
3. In local-config.properties, set RUN_WITH_LLM to true.

Results

By default results will be shown in the ./results/ folder. Also, by default, the naming would be result_{nr} where {nr} is represented by a number that is incremental (until RESULT_FILE_LIMIT).

Configurations

You can edit configurations in local-config.properties. However, the main configuration will still be loaded from Properties first. After that Local Properties and Testing Properties (the testing properties is reset every run).

Properties

Property	Default values if not set (if not required)	Type	Description
LOGGING_LEVEL	INFO	String	Which level to show the logging in.
PATH_TO_CODE	src/test/resources/	String	Required attribute. Where the code is located.
PATH_TO_RESULTS	results	String	Results directory.
RESULT_NAME_PREFIX	result_{nr}	String	Result name. {nr} will be replaced by an incremental number.
RESULTS_FILE_LIMIT	100	Integer	Since it's creating a new file each run, there is a limit.
EXCLUDED_PATHS	[]	List	Separated by commas. Files/Paths to exclude during a run.
RUN_WITH_LLM	false	String	Whether to use LLM during the run or not.
LLM_FILE	models/ggml-model-gpt4all-falcon-q4_0.bin	String	If running with LLM, this has to be set linking to a model.
LLM_THREADS	4	Integer	How many threads the LLM will be using.
TEMP_FILE_ENABLED	false	Boolean	Whether to use a temporary file.
TEMP_FILE	debugFile.txt	String	Where the temporary file is located.

LOGGING_LEVEL

The different options are the following:

	ERROR	WARNING	INFO	DEBUG	TRACE
NONE
ERROR	X
WARNING	X	X
INFO	X	X	X
DEBUG	X	X	X	X
TRACE	X	X	X	X	X