tools/diff-zip.sh · typeable/pandoc

git.iko.soy/typeable/pandoc

33 lines 876 B history raw

1#!/bin/sh

3# This script allows you to compare two epub, odt, or docx

4# containers, ignoring insignificant formatting differences

5# in the XML contents.

7UNAME=$(uname)

8if [ "$UNAME" = "Darwin" ]; then

9 FIND="find -E"

10else

11 FIND="find -regextype posix-extended"

14f1="$1"

15f2="$2"

16test -f "$f1" -a -f "$f2" || {

17 echo "Usage: diff-zip firstfile secondfile" && exit 1

19WORKDIR=$(mktemp -d -t diff-zip.XXX)

20trap "{ rm -r $WORKDIR; }" EXIT

21unzip -q -d "$WORKDIR/a" "$f1"

22unzip -q -d "$WORKDIR/b" "$f2"

23cd "$WORKDIR"

24mkdir tidy

25for x in a b; do

26 cp -r $x tidy/

27 $FIND $x -iregex '.*\.(xhtml|xml|rdf|rels)' -exec sh -c 'mkdir -p "$(dirname tidy/$1)" && tidy -q -xml -utf8 -i "$1" > "tidy/$1"' _ {} \;

28done

29cd tidy

30mkdir c

31cp -r a/* c/

32cp -r b/* c/

33find c -type f -exec sh -c 'echo -e "\033[1m=== ${1#*/} ===\033[0m" ; diff -u "a/${1#*/}" "b/${1#*/}" 2>&1' _ {} \;