Used for merging a bunch of code from a project together into one file with context; useful for feeding LLMs.

merge_code.sh

function merge_files() {
    # Default values
    local output_file=""   # empty => write to stdout (default)
    local to_clipboard=false   # -c => copy to clipboard instead of stdout
    local extensions=()
    local extra_excluded_dirs=()
    local extra_excluded_files=()

    # Robust default ignore list. These are matched by NAME at any depth
    # (recursively), so e.g. a "test" or "node_modules" folder is dropped no
    # matter how deep it sits in a monorepo.
    local default_excluded_dirs=(
        # version control
        .git .svn .hg
        # JavaScript / Node / Bun / Yarn / pnpm + bundler output
        node_modules bower_components .yarn .pnp .npm
        .next .nuxt .svelte-kit .turbo .parcel-cache .cache
        dist build out coverage
        # Python / Poetry / virtualenvs
        .venv venv env __pycache__ .pytest_cache .mypy_cache .ruff_cache
        .tox .eggs "*.egg-info" .ipynb_checkpoints site-packages
        # R / renv / packrat
        renv packrat .Rproj.user
        # Swift / Xcode
        .build .swiftpm DerivedData "*.xcodeproj" "*.xcworkspace" Pods Carthage
        # other language build/output dirs
        target vendor .gradle .terraform
        # editors / OS
        .idea .vscode
    )

    # File basenames (globs allowed) to skip even when their extension matches.
    # Mostly lockfiles and minified bundles that add noise, not signal.
    local default_excluded_files=(
        package-lock.json yarn.lock pnpm-lock.yaml bun.lockb
        poetry.lock Pipfile.lock composer.lock Cargo.lock renv.lock
        "*.min.js" "*.min.css"
    )

    # Help message
    local usage="Usage: merge_files -e EXTENSIONS [OPTIONS]
Required:
    -e, --extensions EXT    Comma-separated list of extensions (e.g., py,js,ts)
Options:
    -o, --output FILE       Write to FILE instead of stdout (default: stdout)
    -c, --clipboard         Copy merged output to the clipboard (via pbcopy)
    -x, --exclude DIRS      Comma-separated dir NAMES to ignore recursively
                            (added to the built-in defaults)
    -X, --exclude-files F   Comma-separated file names/globs to ignore
                            (added to the built-in defaults)
    -h, --help              Show this help message
By default output goes to stdout (pipe it, or use -o to save a file).
Examples:
    merge_files -e py,js                       # print merged code to stdout
    merge_files -e py -o all.txt               # save to a file
    merge_files -e py,js,ts -c                  # copy straight to the clipboard
    merge_files -e R,r -x test,fixtures        # ignore every test/ dir recursively
    merge_files -e py,js,ts | pbcopy           # straight to clipboard"

    # Helper: split a comma-separated value into an array, appending to $1.
    _mf_split_into() {
        local __name="$1" __val="$2"
        if [[ -n "$ZSH_VERSION" ]]; then
            eval "${__name}+=(\"\${(@s:,:)__val}\")"
        else
            local __oldIFS=$IFS part
            IFS=','
            for part in $__val; do
                eval "${__name}+=(\"\$part\")"
            done
            IFS=$__oldIFS
        fi
    }

    # Check if no arguments provided
    if [[ $# -eq 0 ]]; then
        echo "Error: No arguments provided" >&2
        echo "$usage" >&2
        return 1
    fi

    # Parse arguments
    local extensions_provided=false
    while [[ $# -gt 0 ]]; do
        case $1 in
            -o|--output)
                output_file="$2"
                shift 2
                ;;
            -c|--clipboard)
                to_clipboard=true
                shift
                ;;
            -e|--extensions)
                _mf_split_into extensions "$2"
                extensions_provided=true
                shift 2
                ;;
            -x|--exclude)
                _mf_split_into extra_excluded_dirs "$2"
                shift 2
                ;;
            -X|--exclude-files)
                _mf_split_into extra_excluded_files "$2"
                shift 2
                ;;
            -h|--help)
                echo "$usage"
                return 0
                ;;
            *)
                echo "Error: Unknown option: $1" >&2
                echo "$usage" >&2
                return 1
                ;;
        esac
    done

    # Check if extensions were provided
    if ! $extensions_provided || [[ ${#extensions[@]} -eq 0 ]]; then
        echo "Error: No extensions provided. Use -e or --extensions option." >&2
        echo "$usage" >&2
        return 1
    fi

    # -o and -c are mutually exclusive destinations.
    if $to_clipboard && [[ -n "$output_file" ]]; then
        echo "Error: -c/--clipboard cannot be combined with -o/--output." >&2
        return 1
    fi
    if $to_clipboard && ! command -v pbcopy >/dev/null 2>&1; then
        echo "Error: -c/--clipboard requires pbcopy, which was not found." >&2
        return 1
    fi

    # Merge user-supplied excludes onto the defaults (additive, not replacing).
    local excluded_dirs=("${default_excluded_dirs[@]}" "${extra_excluded_dirs[@]}")
    local excluded_files=("${default_excluded_files[@]}" "${extra_excluded_files[@]}")

    # Comment map
    declare -A comment_map=(
        [py]="#"
        [js]="//"
        [ts]="//"
        [sh]="#"
        [cpp]="//"
        [java]="//"
        [rb]="#"
        [php]="#"
        [pl]="#"
        [R]="#"
        [r]="#"
        [swift]="//"
        [go]="//"
        [rs]="//"
        [c]="//"
        [h]="//"
        [yaml]="#"
        [yml]="#"
        [ini]="#"
        [conf]="#"
        [md]="<!--"
        [html]="<!--"
        [css]="/*"
    )
    local DEFAULT_COMMENT="#"

    # Build the find expression.
    # 1) Prune directories matched by NAME (recursive at any depth).
    local prune_expr=("(" "-type" "d" "(")
    local first=true dir
    for dir in "${excluded_dirs[@]}"; do
        $first || prune_expr+=("-o")
        prune_expr+=("-name" "$dir")
        first=false
    done
    prune_expr+=(")" "-prune" ")")

    # 2) Match files by extension.
    local name_expr=("(")
    first=true
    local ext
    for ext in "${extensions[@]}"; do
        $first || name_expr+=("-o")
        name_expr+=("-name" "*.$ext")
        first=false
    done
    name_expr+=(")")

    # 3) Exclude noisy files by basename/glob, plus the output file itself so
    #    it never gets merged into itself.
    local notfile_expr=()
    local f
    for f in "${excluded_files[@]}"; do
        notfile_expr+=("!" "-name" "$f")
    done
    [[ -n "$output_file" ]] && notfile_expr+=("!" "-name" "${output_file##*/}")

    # Emit merged content to stdout; caller redirects to a file if requested.
    _mf_emit() {
        find . "${prune_expr[@]}" -o \( -type f "${name_expr[@]}" "${notfile_expr[@]}" -print \) \
            | sort | while IFS= read -r file; do
            local extension="${file##*.}"
            local comment_prefix="${comment_map[$extension]}"
            [[ -z "$comment_prefix" ]] && comment_prefix="$DEFAULT_COMMENT"

            printf '%s %s\n\n' "$comment_prefix" "$file"
            cat "$file"
            printf '\n'
        done
    }

    if [[ -n "$output_file" ]]; then
        _mf_emit > "$output_file"
        if [[ ! -s "$output_file" ]]; then
            rm -f "$output_file"
            echo "Warning: No files found matching extensions: ${extensions[*]}" >&2
            return 0
        fi
        echo "Merged [${extensions[*]}] into $output_file" >&2
    elif $to_clipboard; then
        local merged
        merged="$(_mf_emit)"
        if [[ -z "$merged" ]]; then
            echo "Warning: No files found matching extensions: ${extensions[*]}" >&2
            return 0
        fi
        printf '%s' "$merged" | pbcopy
        echo "Merged [${extensions[*]}] into the clipboard" >&2
    else
        _mf_emit
    fi
}