LHDiff — A Hybrid, Language-Independent Line Differencing Tool

LHDiff is a Python implementation of the algorithm described in “LHDiff: A Language-Independent Hybrid Approach for Code Differencing” by Asaduzzaman et al. (2013).

This tool detects line-level changes between two text files (source code or plain text) using a hybrid strategy that combines:

• Context Similarity (Cosine Similarity over surrounding lines)
• Content Similarity (Levenshtein Distance)
• Efficient Hashing (Simhash)

The result is a robust algorithm capable of identifying matches, modified lines, and even moved lines within files.

Features

Context-Aware Matching

Prevents mismatching identical but unrelated lines by analyzing the similarity of surrounding lines using cosine similarity.

Hybrid Matching Approach

Uses:

• Simhash → Generates candidates quickly
• Levenshtein Distance → Scores content similarity precisely

This reduces false matches while maintaining fast performance.

Move Detection

Detects when a line appears in both files but at different locations.

Prerequisites

• Python 3.x
• simhash library

Installation

Install the required dependency:

pip install simhash 

Usage

Run the LHDiff script from your terminal:

python LHdiff.py <path_to_file_A> <path_to_file_B>

Example

python LHdiff.py New_File_Versions/NewFile1.txt Old_File_Versions/OldFile1.txt

Example Output

[(1, 1), (2, 3), (3, 2), (4, 4), (5, 5), (6, 6)]

Interpretation

(1, 1) → Line 1 in File A matches Line 1 in File B

(2, 3) → Line 2 in File A corresponds to Line 3 in File B (moved)

(3, 2) → Line 3 in File A corresponds to Line 2 in File B (moved)

(n, m) pairs show all detected matches, moves, and alignments