#!/bin/bash

sep='<no-filename-ever-contains-this>'

safe_types_regexp=('^(inode|text|image|video)/')
safe_types_string=('application/pdf' 'application/postscript')
safe_files_regexp=('/\.(git|hg|svn)/')
safe_files_string=()

# don't care about files less than 3 bytes.
min_size=3

normalize_filename() {
	local cwd="`pwd`"
	readlink -m -- "$1"|sed "s|^$cwd/|./|"
}

matches_string() {
	local needle=$1
	shift
	for straw in "$@"; do
		if [[ "$needle" = "$straw" ]]; then
			return 0
		fi
	done
	return 1
}

matches_regexp() {
	local needle=$1
	shift
	for straw in "$@"; do
		if [[ "$needle" =~ $straw ]]; then
			return 0
		fi
	done
	return 1
}

print-human() {
	libremessages warning "The source directory `pwd` contains binary files:"
	sed 's/./  -> &/'
}

print-machine() {
	cat
}

main() {
	format=human
	# Parse arguments
	for arg in "$@"; do
		case "$arg" in
			-m) format=machine;;
			*) safe_files_string+=("$(normalize_filename "$arg")");;
		esac
	done

	# Init
	unsafe_files="$(mktemp)"
	cleanup() { rm -f -- "$unsafe_files"; }
	trap cleanup EXIT

	# Heavy lifting
	find . -type f -printf '%s %h/%f\n' | # find all files
	while read -r size file; do # filter out files smaller than $min_size
		[[ $size < $min_size ]] || printf '%s\n' "$file"
	done |
	xargs -d'\n' file --mime-type -r -F "$sep" | # identify the filetypes
	sed -r "s@(.*)${sep}\s*(.*)@\2:\1@" |  # reformat the output to be easier to parse
	while IFS=: read -r type file; do
		file="$(normalize_filename "$file")"

		if \
			matches_string "$file" "${safe_files_string[@]}" || \
			matches_string "$type" "${safe_types_string[@]}" || \
			matches_regexp "$file" "${safe_files_regexp[@]}" || \
			matches_regexp "$type" "${safe_types_regexp[@]}" ; then
			: # do nothing
		else
			printf "%s\n" "$file"
		fi
	done > "$unsafe_files"

	if [[ -n "$(cat "$unsafe_files")" ]]; then
		<"$unsafe_files" sort | print-$format
		exit 1
	fi
}

main "$@"