#!/bin/bash
#
# irregular_cask_whitespace
#
# find irregular whitespace in Cask files
#
# notes
#
#     requires a recent-ish Perl with Unicode support, probably 5.14
#     or better.
#
# bugs
#
# todo
#

###
### settings
###

set -e
set -o pipefail
set +o histexpand
set -o nounset
shopt -s nocasematch
shopt -s nullglob

###
### functions
###

warn () {
    local message="$@"
    message="${message//\\t/$'\011'}"
    message="${message//\\n/$'\012'}"
    message="${message%"${message##*[![:space:]]}"}"
    printf "%s\n" "$message" 1>&2
}

die () {
    warn "$@"
    exit 1
}

###
### main
###

_irregular_cask_whitespace () {
    local directory="$1"
    cd "$directory" || die "Could not cd to '$directory'"

    printf "# No trailing newline at EOF\n"
    perl -C32 -0777 -ne 'print "    $ARGV\n" if m{[^\n]\z}s'      -- ./*.rb

    printf "\n# Extra trailing newline at EOF\n"
    perl -C32 -0777 -ne 'print "    $ARGV\n" if m{\n{2}\z}s'      -- ./*.rb

    printf "\n# Final 'end' indented\n"
    perl -C32 -0777 -ne 'print "    $ARGV\n" if m{ end\s+\z}s'    -- ./*.rb

    printf "\n# Extra newline before final end\n"
    perl -C32 -0777 -ne 'print "    $ARGV\n" if m{\n\nend\s+\z}s' -- ./*.rb

    printf "\n# Extra newline before header\n"
    perl -C32 -0777 -ne 'print "    $ARGV\n" if m{\n\ncask\s+:v\d\S*\s+=>}s' -- ./*.rb

    printf "\n# Extra newline after header\n"
    perl -C32 -0777 -ne 'print "    $ARGV\n" if m{(?:\A|\n)cask\s+:v\d\S*\s+=>[^\n]+\n\n\s*(\S+)}s and $1 ne "if"' -- ./*.rb

    printf "\n# No empty line before uninstall\n"
    perl -C32 -0777 -ne 'print "    $ARGV\n" if m{\n[^\n]+\n +uninstall }s'  -- ./*.rb

    # todo?
    # printf "\n# No empty line before caveats\n"
    # perl -C32 -0777 -ne 'print "    $ARGV\n" if m{\n[^\n]+\n +caveats }s'    -- ./*.rb

    printf "\n# Extra interior newlines\n"
    perl -C32 -0777 -ne 'print "    $ARGV\n" if m{\n{3,}}s'    -- ./*.rb

    printf "\n# Leading whitespace at BOF\n"
    perl -C32 -0777 -ne 'print "    $ARGV\n" if m{\A\s}s'      -- ./*.rb

    printf "\n# Trailing whitespace at EOL (includes Tab/CR)\n"
    perl -C32       -ne 'print "    $ARGV\n" if m{\s\n}s'      -- ./*.rb | sort | uniq

    printf "\n# Tabs\n"
    perl -C32 -0777 -ne 'print "    $ARGV\n" if m{\t}s'        -- ./*.rb

    printf "\n# Carriage Returns\n"
    perl -C32 -0777 -ne 'print "    $ARGV\n" if m{\r}s'        -- ./*.rb

    printf "\n# Misc Control Characters\n"
    perl -C32 -0777 -ne 'print "    $ARGV\n" if m{[\x00-\x08\x0B-\x0C\x0E\x1F]}s'                -- ./*.rb

    printf "\n# First indent not 2\n"
    perl -C32 -0777 -ne 's{\A(.*?\n)?cask\s+[^\n]+\n+}{}s; print "    $ARGV\n" unless m{\A  \S}s'  -- ./*.rb

    printf "\n# Indents not multiple of 2\n"
    perl -C32 -0777 -ne 'print "    $ARGV\n" if m{\n(?:  ){0,} [a-z]}s'                          -- ./*.rb

    printf "\n# Unicode Space Characters\n"
    # \x{0085}\x{0088}\x{0089}
    perl -C32 -0777 -ne 'print "    $ARGV\n" if m{[\x{008a}\x{00a0}\x{1680}\x{180e}\x{2000}\x{2001}\x{2002}\x{2003}\x{2004}\x{2005}\x{2006}\x{2007}\x{2008}\x{2009}\x{200a}\x{200b}\x{2028}\x{2029}\x{202f}\x{205f}\x{2060}\x{3000}\x{feff}\x{e0020}]}s' -- ./*.rb

}

###
### argument processing
###

if [[ "${1:-}" =~ ^-+h(elp)?$ ]]; then
   printf "irregular_cask_whitespace <dir>

   Find irregular whitespace in Cask files within <dir>

"
    exit
fi

if [ "$#" -ne 1 ]; then
    die "Single directory argument required"
elif ! [ -d "$1" ]; then
    die "No directory found at '$1'"
fi

###
### dispatch
###

_irregular_cask_whitespace "${@:-}"

#