Extract all source codes (filter by file ext) into one txt file for LLM context

source_extract.sh

#!/bin/bash

print_usage() {
    echo "Usage: $0 -i <input_directory> -o <output_directory> -x <extensions>"
    echo "Example: $0 -i input_dir -o output_dir -x kt java xml"
}

while getopts "i:o:x:" opt; do
    case $opt in
        i) input_directory="$OPTARG" ;;
        o) output_directory="$OPTARG" ;;
        x) extensions="$OPTARG" ;;
        \?) print_usage
            exit 1 ;;
    esac
done

if [ -z "$input_directory" ] || [ -z "$output_directory" ] || [ -z "$extensions" ]; then
    print_usage
    exit 1
fi

mkdir -p "$output_directory"
output_file="$output_directory/merged.txt"

> "$output_file"

IFS=' ' read -r -a ext_array <<< "$extensions"

for ext in "${ext_array[@]}"; do
    find "$input_directory" -type f -name "*.$ext" | while read -r file; do
        echo -e "\n// file: $(basename "$file")\n" >> "$output_file"
        cat "$file" >> "$output_file"
    done
done

echo "All merged into $output_file ✅"