commit-msg

#!/bin/bash

# This script is used by the commit-msg hook to check the commit message
# and to update the commit message if necessary with the correct format.
# There is based to https://www.conventionalcommits.org/en/v1.0.0/
#
# Format:
# <type>[optional scope]: <subject>
# (max 80 chars, no upper case, no final dot)
#
# [optional description]
#
# [footer(s)]
# (* && Signed-off-by: <name> <email>)

if [[ ${BASH_SOURCE[0]} -ef "$0" ]]; then
    # source config and common file
    source "${0%/*}/components/config"
    source "${0%/*}/components/common"
else
    # source from unit test
    source "./components/config"
    source "./components/common"
fi

# shellcheck disable=SC2206
## CM_TYPE is an array of commit message type,
## TYPE_EXTRA can be used to add extra commit message type
CM_TYPE=(
    "build"
    "chore"
    "ci"
    "docs"
    "feat"
    "fix"
    "perf"
    "refactor"
    "style"
    "test"
    "merge"
    "wip"
    ${TYPE_EXTRA[*]}
)

# shellcheck disable=SC2206
## CM_EMOJI is an array of emoji type, there is mapped to the commit type
## TYPE_EXTRA_EMOJI can be used to add extra emoji type
CM_EMOJI=(
    "🏗️"
    "🙈"
    "🚀"
    "📝"
    "✨"
    "🐛"
    "⚡️"
    "♻️"
    "💄"
    "🧪"
    "🔀"
    "🚧"
    ${TYPE_EXTRA_EMOJI[*]}
)

## RGX_MESSAGE and RGX_MESSAGE_EMOJI are the regex for commit message header
BISF="$IFS"
IFS="|"
RGX_EMOJI="(?<emoji>${CM_EMOJI[*]})"
RGX_HEADER="^(?:${RGX_EMOJI} )?(?<type>${CM_TYPE[*]})(?:\((?<scope>[^)A-Z]+)\))?(?<breaking>!)?:"
RGX_MESSAGE="^${RGX_HEADER}\s(?<subject>[^A-Z].+[^.])$"
RGX_MESSAGE_EMOJI="^${RGX_EMOJI}(?:\((?<scope>[^)A-Z]+)\))?(?<breaking>!)?:?\s(?<subject>[^A-Z].+[^.])$"
IFS="$BISF"
unset BISF

## CM_FOOTER is the regex for commit message footer
CM_FOOTER="(?:[^:]+:\s.+\n)*(?:Signed-off-by:\s(?<name>.+)\s<(?<email>[^@]+@.+)>)"

## Format description
HEADER_RULE="Commit message format: <type>(optional scope): <subject> (max $HEADER_MAX_LENGTH chars, no upper case, no final dot, with '!' before the colon for breaking change)"
FOOTER_RULE="Commit footer format: Signed-off-by: <name> <email>"

## print_available_type is a function to list all available commit type
print_available_type() {
    local _types=()
    for i in "${!CM_TYPE[@]}"; do
        if [[ ${CM_EMOJI[$i]} == "" ]]; then
            _types+=("${CM_TYPE[$i]}")
        else
            _types+=("${CM_TYPE[$i]}: ${CM_EMOJI[$i]}")
        fi
    done

    info "Available commit type :"
    info "$(printf "%s, " "${_types[@]}")"
}

## format_checker is the function to check and fix the commit message format
format_checker() {
    local _section=1
    local _empty_line=$FALSE

    local _content=()
    while read -r line; do
        # skip comment line
        if [[ $line == "#"* ]]; then
            continue
        fi

        if [[ $line == "" ]]; then
            # two empty lines are not allowed we can safely ignore it
            if [[ $_empty_line -eq $TRUE ]]; then
                continue
            fi

            # first line can't be empty
            if [[ ${_content[$_section]} == "" ]]; then
                error "A section can't be empty, probably a missing header"
                return $FAILURE
            fi

            # an empty line is expect after a section
            _section=$((_section + 1))
            _empty_line=$TRUE

            continue
        fi

        # reset empty line flag
        _empty_line=$FALSE

        # save section content
        if [[ ${_content[$_section]} == "" ]]; then
            _content[$_section]="$line"
        else
            _content[$_section]+=$'\n'"$line"
        fi

        continue
    done <<<"$1"

    if [[ $_section -gt 1 ]]; then
        # if current section is not empty and not contain signedoff add a section,
        # but if is empty and the previous one contains signedoff delete the last one
        if [[ -n ${_content[$_section]} ]] && ! grep -sPq "Signed-off-by" <<<"${_content[$_section]}"; then
            _section=$((_section + 1))
        elif [[ -z ${_content[$_section]} ]] && grep -sPq "Signed-off-by" <<<"${_content[$((_section - 1))]}"; then
            _section=$((_section - 1))
        fi
    else
        # in case of only one section are found
        _section=$((_section + 1))
    fi

    # check and fix the header format
    if ! _content[1]=$(format_header "${_content[1]}"); then
        return 1
    fi

    # check the footer format and required signature
    if ! _content[$_section]=$(format_footer "${_content[$_section]}"); then
        return 1
    fi

    # return the result with an empty line between each sections
    for id in "${!_content[@]}"; do
        if [[ $id -ne 1 ]]; then
            echo ""
        fi

        echo "${_content[$id]}"
    done

}

## format_header is the function to check and fix the commit message header format
format_header() {
    local _error=$FALSE

    # check if required arguments are provided
    if [[ -z $1 ]]; then
        return $FAILURE
    fi

    local _line="$1"

    # check if there is multiple line
    if [[ $_line == *$'\n'* ]]; then
        error "Multiple line are not allowed in the header"
        return $FAILURE
    fi

    # check the header length
    if [[ ${#_line} -gt $HEADER_MAX_LENGTH ]]; then
        warning "Current header length is ${#_line} '${_line}'"
        error "The header is too long, max length is $HEADER_MAX_LENGTH"
        return $FAILURE
    fi

    # Check the commit header
    if ! (grep -sPq "${RGX_HEADER}.*" <<<"$_line"); then
        print_available_type
        error "Invalid header format in commit message"
        error "$_line"
        error "$HEADER_RULE"
        return $FAILURE
    fi

    # check the header format RGX_MESSAGE or RGX_MESSAGE_EMOJI
    (
        grep -sPq "${RGX_MESSAGE}" <<<"$_line" ||
            grep -sPq "${RGX_MESSAGE_EMOJI}" <<<"$_line"
    ) || _error=$TRUE

    # try to auto fix the header if needed
    if [[ $_error -eq $TRUE ]] && [[ $HEADER_AUTOFIX -eq $TRUE ]]; then
        # reset error flag
        _error=$FALSE

        # Convert the header to lowercase
        _line=$(tr '[:upper:]' '[:lower:]' <<<"$_line") || {
            warning "Line: '$_line'"
            error "Failed to convert the header to lowercase"
            return $FAILURE
        }

        # Remove dot at the end of the header if any
        if [[ $_line == *"." ]]; then
            _line=${_line::-1}
        fi

        # try to fix bad colon
        _line=$(sed -E 's/^([^:     ]+)\s*:\s*/\1: /' <<<"$_line") || {
            warning "Line: '$_line'"
            error "Failed to remove bad semicolon"
            return $FAILURE
        }

        # check the format again
        (grep -sPq "${RGX_MESSAGE}" <<<"$_line" || grep -sPq "${RGX_MESSAGE_EMOJI}" <<<"$_line") || {
            error "Invalid header format can't be fixed"
            error "$_line"
            error "$HEADER_RULE"
            return $FAILURE
        }
    fi

    if [[ $_error -eq $TRUE ]]; then
        error "Invalid header format"
        error "$_line"
        error "$HEADER_RULE"
        return $FAILURE
    fi

    # emoji
    _line="$(emoji_me "$_line")" || {
        error "Failed to apply emoji"
        return $FAILURE
    }

    # return the result
    echo "$_line"
    return $SUCCESS
}

## format_footer is the function to check and fix the commit message footer
format_footer() {
    local _line="$1"

    # if signoff is required
    if [[ $SIGNOFF_REQUIRED -eq $TRUE ]]; then
        local _error=$FALSE

        # check the footer format
        if ! (grep -sPq "${CM_FOOTER}" <<<"$_line"); then
            _error=$TRUE
        fi

        # autofix if needed
        if [[ $_error -eq $TRUE ]] && [[ $SIGNOFF_AUTOFIX -eq $TRUE ]]; then
            _line=$(sed -E 's/\n?Signed-off-by:.*//' <<<"$_line")

            if [[ -z $GIT_AUTHOR_EMAIL ]] || [[ -z $GIT_AUTHOR_NAME ]]; then
                error "Can't autofix the footer, missing author information"
                return $FAILURE
            fi

            if [[ $_line != *"Signed-off-by"* ]]; then
                _error=$FALSE

                if [[ -n $_line ]]; then
                    _line+=$'\n'
                fi

                _line+="Signed-off-by: $GIT_AUTHOR_NAME <$GIT_AUTHOR_EMAIL>"
            fi
        fi

        if [[ $_error -eq $TRUE ]]; then
            error "Invalid footer format Signed-off-by is missing or badly formatted"
            return $FAILURE
        fi
    fi

    # return the result
    echo "$_line"
    return $SUCCESS
}

## emoji_me is a function to apply emoji to a commit message
emoji_me() {
    # check if required arguments are provided
    if [[ -z $1 ]]; then
        return $FAILURE
    fi

    local _line="$1"

    # check if emoji_me is disable or if line already start with an emoji
    if [[ $TYPE_TO_EMOJI -eq $FALSE ]] ||
        (grep -sPq "^${RGX_EMOJI}" <<<"$_line"); then
        echo "$_line"
        return $SUCCESS
    fi

    # find the commit type in CM_TYPE array
    local _type_id=-1
    for i in "${!CM_TYPE[@]}"; do
        if (grep -sPq "^${CM_TYPE[$i]}" <<<"$_line"); then
            _type_id="$i"
            break
        fi
    done

    if [[ _type_id -eq -1 ]]; then
        error "No allowed commit type found in the commit message: $_line"
        return $FAILURE
    fi

    # check if an emoji is available at this id
    if [[ ${CM_EMOJI[$_type_id]} == "" ]]; then
        error "No emoji found for this commit type: ${CM_TYPE[$_type_id]}"
        return $FAILURE
    fi

    # apply the emoji
    case $TYPE_TO_EMOJI in
    "$TRUE")
        echo "$_line" | sed -E "s/${CM_TYPE[$_type_id]}:? ?/${CM_EMOJI[$_type_id]} /"
        ;;
    "$SUPERPOSITION")
        echo "${CM_EMOJI[$_type_id]} ${_line}"
        ;;
    *)
        return $FAILURE
        ;;
    esac
}

## MAIN
# do not execute if this script is sourced required for unit testing
if [[ ${BASH_SOURCE[0]} -ef "$0" ]]; then
    # check if arg is a regular file
    if [[ ! -f $1 ]]; then
        warning "Argument: $1"
        error "Gived argument is not a regular file"
        exit $FAILURE
    fi

    # process the file
    _result="$(format_checker "$(<"$1")")" && {
        echo "$_result" >"$1"
        exit $SUCCESS
    }

    exit $FAILURE
fi