diffdocx

#! /usr/bin/env bash

set -ueE -o pipefail

prog="$(basename "$0")"

function usage {
    [ "${*-}" ] && echo "$prog: $*" 1>&2
    cat 1>&2 <<EOF
Usage: $prog file1 file2
Usage: $prog -r dir1 dir1

Show the text differences in two Word files (.doc or .docx), or
show the text differences in the WOrd files in two directories.
EOF
    exit 1
}

function errordie {
    [ "${*-}" ] && echo "$prog: $*" 1>&2
    exit 1
}

if ! type -p pandoc > /dev/null; then
    errordie "pandoc is not installed"
fi

if ! type -p diff-pdf > /dev/null; then
    errordie "diff-pdf is not installed"
fi

recurse=

while [ $# -gt 0 ]; do
    case $1 in
        -r) recurse=$1 ;;
        --) usage "unknown argument: $1" ;;
        *)  break
	    ;;
    esac
    shift
done

[ $# -eq 2 ] || usage

function mydiff {
    diff <(pandoc --to org "$1") <(pandoc --to org "$2")
}

{
    if [ "$recurse" ]; then
        # both arguments must be directories
        [ -d "$1" ] || usage "first argument not directory"
        [ -d "$2" ] || usage "second argument not directory"
        while IFS=$'\n' read -r line; do
            if [[ $line =~ ^Files\ (.*)\ and\ (.*)\ differ$ ]]; then
                left=${BASH_REMATCH[1]}
                right=${BASH_REMATCH[1]}
                #echo "ext=${left##*.}"
                case ${left##*.} in
                    doc|docx)
                        echo "+ diff $left $right"
                        mydiff "$left" "$right"
                        ;;
                    pdf)
                        if diff-pdf "$left" "$right"; then
                            echo "+ PDF UNCHANGED: $left $right"
                        else
                            echo "+ PDF DIFFERENT: $left $right"
                        fi
                        ;;
                    *)  echo "NOTE: will not check $left and $right"
                        ;;
                esac
            elif [[ $line =~ ^Only\ in ]]; then
                # ignore
                :
            else
                errordie "could not parse: $line"
            fi
        done <<< "$(diff -qr "$1" "$2")"
    else
        mydiff "$1" "$2"
    fi
    exit 0
}