.lessfilter

#!/usr/bin/env bash

# exit 1 if no 1st argument present
[[ -z "${1}" ]] && exit 1

# Source clean-hash function from kdm-bash-env, or exit 2
# shellcheck disable=SC1091
. "${HOME}/.kdm/rc.d/00-base" || exit 2

# if pygmentize is not present, exit now
hash pygmentize || exit 0


_debugLog() {
	# Comment below line to enable debug output
	return 0

	echo "1                : '${1}'"
	echo "FILE_NAME        : '${FILE_NAME}'"
	echo "FILE_MIME_TYPE   : '${FILE_MIME_TYPE}'"
	echo "EXTENSION_DETECT : '${EXTENSION_DETECT}'"
	echo "LEXER_OVERRIDE   : '${LEXER_OVERRIDE}'"
	echo
}


# Pygments formatter and style
PYG_FMT="terminal16m"
PYG_STYLE="monokai"

# Get base file name
FILE_NAME="$(basename "${1}")"

# Force lowercase filename
FILE_NAME="${FILE_NAME,,}"


LEXER_OVERRIDE=""

# If file type was detected based on file extension
EXTENSION_DETECT="false"


# Try to detect file type purely based on file name extension
case "${FILE_NAME}" in
	# Fairly common file types normally supported by pygmentize
	*.[ch]|*.[ch]pp|*.[ch]xx|*.ad[asb]|*.asm|*.awk|*.axp|*.bat|*.c|*.cc|*.cmd|*.cnf|*.conf|*.config|*.cpp|*.css|*.diff|*.ebuild|*.eclass|*.groff|*.h|*.hh|*.htm|*.html|*.inc|*.java|*.l|*.lsp|*.m4|*.md|*.p|*.pas|*.patch|*.pl|*.pm|*.pod|*.pov|*.ppd|*.rb|*.sass|*.scss|*.sh|*.sls|*.sql|*.svg|*.xml|*.xps|*.xsd|*.xsl|*.yaml|*.yml)
		EXTENSION_DETECT="true"
	;;
esac

# Call pygmentize without lexer set, to attempt auto-detection based on file extension
if [[ "${EXTENSION_DETECT}" == "true" ]]; then
	_debugLog "${1}"
	pygmentize -f "${PYG_FMT}" -O "style=${PYG_STYLE}" "${1}"
	exit "${?}"
fi


# Normal file extension-based detection failed, add some manual effort
if [[ -z "${LEXER_OVERRIDE}" ]]; then
	case "${FILE_NAME}" in
		# .bash_logout, .bash_profile, .bashrc, .profile, etc
		.bash*|.profile) LEXER_OVERRIDE="sh" ;;

		# systemd files
		*.mount|*.service|*.socket|*.target|*.timer) LEXER_OVERRIDE="dosini" ;;

		*.ino)   LEXER_OVERRIDE="c"      ;;
		*.json)  LEXER_OVERRIDE="json"   ;;
		*.php)   LEXER_OVERRIDE="php"    ;;
		*.plist) LEXER_OVERRIDE="xml"    ;;
		*.py)    LEXER_OVERRIDE="python" ;;
		*.xdf)   LEXER_OVERRIDE="xml"    ;;
	esac
fi


# grep for specific shebang, if it passes, set lexer value
if [[ -z "${LEXER_OVERRIDE}" ]]; then
	SHEBANG_LINE="$(head -n 1 "${1}")"
	echo "${SHEBANG_LINE}" | grep -Eq '#!.*./bin.*.node'     && LEXER_OVERRIDE="javascript"
	echo "${SHEBANG_LINE}" | grep -Eq '#!.*./bin.*.php.*'    && LEXER_OVERRIDE="php"
	echo "${SHEBANG_LINE}" | grep -Eq '#!.*./bin.*.python.*' && LEXER_OVERRIDE="python"
	echo "${SHEBANG_LINE}" | grep -Eq '#!.*./bin.*.sh'       && LEXER_OVERRIDE="sh"
fi


# Shebang grep failed, try detecting based on mime type
if [[ -z "${LEXER_OVERRIDE}" ]]; then
	FILE_MIME_TYPE="$(file --brief --dereference --no-pad --no-buffer --mime-type "${FILE_PATH}" 2> /dev/null | head -n 1)"

	case "${FILE_MIME_TYPE}" in
		application/javascript) LEXER_OVERRIDE="javascript" ;;
		application/json)       LEXER_OVERRIDE="json"       ;;
		text/html)              LEXER_OVERRIDE="html"       ;;
		text/x-c)               LEXER_OVERRIDE="c"          ;;
		text/x-php)             LEXER_OVERRIDE="php"        ;;
		text/x-script.python)   LEXER_OVERRIDE="python"     ;;
		text/x-shellscript)     LEXER_OVERRIDE="sh"         ;;
		text/xml)               LEXER_OVERRIDE="xml"        ;;
	esac
fi


_debugLog "${1}"


if [[ -n "${LEXER_OVERRIDE}" ]]; then
	# JSON : Format with jq, sort output, use tab indendation
	if [[ "${LEXER_OVERRIDE}" == "json" ]] && hash jq; then
		jq -S --tab '.' "${1}" | pygmentize -f "${PYG_FMT}" -O "style=${PYG_STYLE}" -l "${LEXER_OVERRIDE}"
		exit "${?}"
	fi

	pygmentize -f "${PYG_FMT}" -O "style=${PYG_STYLE}" -l "${LEXER_OVERRIDE}" "${1}"
	exit "${?}"
fi


# Last-ditch effort
# -g = Guess the lexer from the file contents, or pass through as plain text if nothing can be guessed
pygmentize -g -f "${PYG_FMT}" -O "style=${PYG_STYLE}" "${1}"