diffjson: compare JSON documents for the differences

diffjson

#!/bin/bash

CONV_JSON=/tmp/convjson.$$
SRC1=/tmp/diffjson.1.$$.json
SRC2=/tmp/diffjson.2.$$.json
SRC3=/tmp/diffjson.3.$$.json

PROGRAM_NAME=$(basename $0)
VERSION_STRING="0.1"
PYTHON=

cleanup() {
    rm -f "$CONV_JSON" "$SRC1" "$SRC2" "$SRC3"
}

trap "cleanup; exit 1" INT HUP TERM QUIT

debug() {
    [ "$option_debug" ] && echo "${RED}debug: $*${CLR}" 1>&2
}

error() {
    local ecode
    ecode=$1
    shift
    echo "$PROGRAM_NAME: $*" 1>&2
    [ "$ecode" -ne 0 ] && exit "$ecode"
    return 0
}

mk_convjson() {
    sed '1,/^__END__/d' "$0" >$CONV_JSON
    chmod +x "$CONV_JSON"
}

check_python2() {
    local py="$PYTHON"

    if [ -x "$py" ]; then
        if "$py" --version 2>&1 | grep ' 2\.[67]' >/dev/null; then
            echo "$py"
            return 0
        fi
    fi
    
    py="$(which python 2>/dev/null)"
    if [ -x "$py" ]; then
        if "$py" --version 2>&1 | grep ' 2\.[67]' >/dev/null; then
            echo "$py"
            return 0
        fi
    fi
    py="$(which python2.7 2>/dev/null)"
    if [ -x "$py" ]; then
        if "$py" --version 2>&1 | grep ' 2\.[67]' >/dev/null; then
            echo "$py"
            return 0
        fi
    fi
    py="$(which python2.6 2>/dev/null)"
    if [ -x "$py" ]; then
        if "$py" --version 2>&1 | grep ' 2\.[67]' >/dev/null; then
            echo "$py"
            return 0
        fi
    fi
    error 1 "python 2.x not found"
}

help_and_exit() {
    cat <<EOF
Usage: $PROGRAM_NAME [OPTION...] FILE1 FILE2
Usage: $PROGRAM_NAME [OPTION...] FILE1 FILE2 -- [DIFF-OPTION...]
Usage: $PROGRAM_NAME [OPTION...] FILE1 FILE2 FILE3 -- [DIFF-OPTION...]
Compare two JSON files for difference

    -d PROGRAM   set the program for the diff.
                 By default, $PROGRAM_NAME tries to use colordiff.  If not, diff(1) will be used.
                 if PROGRAM is 'emacs', then emacsclient with ediff will be launched.

    -h           show help message and exit
    -v           display version string and exit

    -D           show debug information

Any remaining argument in DIFF-OPTION position is passed to the diff program as options.
If there is no DIFF-OPTION, then $PROGRAM_NAME will insert these options:

    diff         -w -c
    colordiff    -w -c
    vimdiff      --cmd 'set diffopt+=iwhite' --cmd 'set diffopt+=iwhite'

Copyright (C) 2016 Seong-Kook Shin  <cinsky at gmail.com>
License WTFPLv2:  Do What The Fuck You Want To Public License <http://www.wtfpl.net/>.

EOF
    exit 0
}


PYTHON=$(check_python2)
mk_convjson

option_diff_program=$(which diff 2>/dev/null)

while getopts "hvDd:" opt; do
    case "$opt" in
        v)
            echo "$PROGRAM_NAME version $VERSION_STRING"
            exit 0
            ;;
        h)
            help_and_exit
            ;;
        d)
            option_diff_program=$OPTARG
            ;;
        D)
            option_debug=1
            ;;
        *)
            echo "option: $opt"
            ;;
    esac
done
shift $((OPTIND - 1))

debug "PYTHON=$PYTHON"
debug "CONV_JSON=$CONV_JSON"
debug "DIFF=$option_diff_program"

if [ "$#" -lt 2 ]; then
    error 0 "argument(s) required"
    error 1 "Try with \`-h' for more"
fi

"$PYTHON" "$CONV_JSON" "$1" > "$SRC1"
[ "$?" -ne 0 ] && exit 1
"$PYTHON" "$CONV_JSON" "$2" > "$SRC2"
[ "$?" -ne 0 ] && exit 1
SRC=("$SRC1" "$SRC2")
shift 2

if [ "$#" -gt 0 ]; then
    if [ "$1" != '--' ]; then
        "$PYTHON" "$CONV_JSON" "$1" > "$SRC3"
        [ "$?" -ne 0 ] && exit 1
        shift
        SRC=("${SRC[@]}" "$SRC3")
    fi
    shift
fi

option_diff_options=("$@")
shift
if [ "${#option_diff_options[@]}" -eq 0 ]; then
    case $(basename "$option_diff_program") in
        diff)
            option_diff_options=('-w' '-c')
            ;;
        colordiff)
            option_diff_options=('-w' '-c')
            ;;
        vimdiff)
            option_diff_options=('--cmd' 'set diffopt+=iwhite' '--cmd' 'set diffopt+=iwhite')
            ;;
        ediff)
            ;;
    esac
fi

if [ "${#SRC[@]}" -eq 3 -a $(basename "$option_diff_program") = "diff" ]; then
    if which diff3 2>/dev/null; then
        option_diff_program=$(which diff3)
        option_diff_options=()
    fi
fi

if [ "${#SRC[@]}" -eq 2 -a $(basename "$option_diff_program") = "diff" ]; then
    if which colordiff 2>/dev/null; then
        option_diff_program=$(which colordiff)
    fi
fi

if [ $(basename "$option_diff_program") = "emacs" ]; then
    quoted1=${SRC1//\\/\\\\}; quoted1=${quoted1//\"/\\\"}
    quoted2=${SRC2//\\/\\\\}; quoted2=${quoted2//\"/\\\"}
    quoted3=${SRC3//\\/\\\\}; quoted3=${quoted3//\"/\\\"}

    if [ "${#SRC[@]}" -eq 2 ]; then
        debug "emacsclient -a '' --eval \"(ediff3 \"$quoted1\" \"$quoted2\")\""
        emacsclient -a '' --eval "(ediff \"$quoted1\" \"$quoted2\")"
        ECODE="$?"
    else
        debug "emacsclient -a '' --eval \"(ediff3 \"$quoted1\" \"$quoted2\" \"$quoted3\")\""
        emacsclient -a '' --eval "(ediff3 \"$quoted1\" \"$quoted2\" \"$quoted3\")"
        ECODE="$?"
    fi
else
    debug "$option_diff_program ${option_diff_options[*]} ${SRC[*]}"
    "$option_diff_program" "${option_diff_options[@]}" "${SRC[@]}"
    ECODE="$?"
fi

cleanup

exit "$ECODE"
# ====================================================================
__END__
from StringIO import StringIO
import json
import sys
import types

def convert_json(outfd, jsobj, depth, postfix="", prefix=""):
    #indent = ("[%d]" % depth) + ".." * depth
    indent = "  " * depth
    #indent = ""
    
    if isinstance(jsobj, types.NoneType):
        outfd.write(indent)
        outfd.write("%s\n" % "null")
    elif isinstance(jsobj, types.BooleanType):
        outfd.write(indent)
        outfd.write("%s\n" % str(jsobj).lower())
    elif isinstance(jsobj, types.LongType):
        outfd.write(indent)
        outfd.write("%ld\n" % jsobj)
    elif isinstance(jsobj, types.IntType):
        outfd.write(indent)
        outfd.write("%d\n" % jsobj)
    elif isinstance(jsobj, types.FloatType):
        outfd.write(indent)
        outfd.write("%f\n" % jsobj)
    elif isinstance(jsobj, types.ListType):
        outfd.write(indent)
        outfd.write("[\n");
        for elm in sorted(jsobj):
            convert_json(outfd, elm, depth + 1)
        outfd.write(indent)
        outfd.write("]\n");
    elif isinstance(jsobj, types.DictType):
        outfd.write(indent)
        outfd.write("{\n");
        for elm in sorted(jsobj.keys()):
            if elm == "timestamp" and depth == 2:
                pass
            elif elm == "url" and depth == 4:
                pass
            else:
                convert_json(outfd, elm, depth + 1, ":")
                convert_json(outfd, jsobj[elm], depth + 2)
        outfd.write(indent)
        outfd.write("}\n");
    elif isinstance(jsobj, types.StringType):
        outfd.write(indent)
        outfd.write("\"%s\"%s\n" % (jsobj, postfix))
    elif isinstance(jsobj, types.UnicodeType):
        outfd.write(indent)
        outfd.write("\"%s\"%s\n" % (jsobj.encode("UTF-8"), postfix))
    else:
        sys.stderr.write("diffjson: cannot handle %s type\n" % type(jsobj))
        sys.exit(1)
        
for arg in sys.argv[1:]:
    try:
        fd = open(arg, "r")
        j = json.load(fd)
    except IOError as e:
        sys.stderr.write("diffjson: %s: %s\n" % (arg, e.strerror))
        sys.exit(1)
    except ValueError:
        sys.stderr.write("diffjson: %s: JSON parse error\n" % arg)
        sys.exit(1)
    convert_json(sys.stdout, j, 0)