for f in old/$file new/$file; do sed -ie ' 1 { : N $ { s@$]\+id=\n\?"$$id[a-z0-9]\+$$"[^>]*>$@\1a_idN\3@g s@$]\+name=\n\?"$$id[a-z0-9]\+$$"[^>]*>$@\1a_nameN\3@g s@$]\+href="#$$[^"]\+$$"[^>]*>$@\1href_anchor\3@g s@$]\+id="$$[a-z0-9]\+$$"[^>]*>$@\1div_idN\3@g } N b N }' $f & done wait } check_compressed_file() { local file=$1 local ext=$2 local tmpdir=`mktemp -d` local ftype local ret=0 echo "$ext file with odd filename: $file" if test -n "$tmpdir"; then mkdir $tmpdir/{old,new} cp --parents --dereference old/$file $tmpdir/ cp --parents --dereference new/$file $tmpdir/ if pushd $tmpdir > /dev/null ; then case "$ext" in bz2) mv old/$file{,.bz2} mv new/$file{,.bz2} bzip2 -d old/$file.bz2 & bzip2 -d new/$file.bz2 & wait ;; gzip) mv old/$file{,.gz} mv new/$file{,.gz} gzip -d old/$file.gz & gzip -d new/$file.gz & wait ;; xz) mv old/$file{,.xz} mv new/$file{,.xz} xz -d old/$file.xz & xz -d new/$file.xz & wait ;; esac ftype=`/usr/bin/file old/$file | sed 's@^[^:]\+:[[:blank:]]*@@'` case $ftype in POSIX\ tar\ archive) echo "$ext content is: $ftype" mv old/$file{,.tar} mv new/$file{,.tar} if ! check_single_file ${file}.tar; then ret=1 fi ;; ASCII\ cpio\ archive\ *) echo "$ext content is: $ftype" mv old/$file{,.cpio} mv new/$file{,.cpio} if ! check_single_file ${file}.cpio; then ret=1 fi ;; fifo*pipe*) ftype_new="`/usr/bin/file new/$file | sed -e 's@^[^:]\+:[[:blank:]]*@@' -e 's@[[:blank:]]*$@@'`" if [ "$ftype_new" = "$ftype" ]; then return 0 fi return 1 ;; *) echo "unhandled $ext content: $ftype" if ! diff_two_files; then ret=1 fi ;; esac popd > /dev/null fi rm -rf "$tmpdir" fi return $ret } check_single_file() { local file="$1" # If the two files are the same, return at once. if [ -f old/$file -a -f new/$file ]; then if cmp -s old/$file new/$file; then return 0 fi fi case $file in *.spec) sed -i -e "s,Release:.*$release1,Release: @RELEASE@," old/$file sed -i -e "s,Release:.*$release2,Release: @RELEASE@," new/$file ;; *.exe.mdb|*.dll.mdb) # Just debug information, we can skip them echo "$file skipped as debug file." return 0 ;; *.a) flist=`ar t new/$file` pwd=$PWD fdir=`dirname $file` cd old/$fdir ar x `basename $file` cd $pwd/new/$fdir ar x `basename $file` cd $pwd for f in $flist; do if ! check_single_file $fdir/$f; then return 1 fi done return 0 ;; *.cpio) flist=`cpio --quiet --list --force-local < "new/$file" | sort` pwd=$PWD fdir=$file.extract.$PPID.$$ mkdir old/$fdir new/$fdir cd old/$fdir cpio --quiet --extract --force-local < "../${file##*/}" cd $pwd/new/$fdir cpio --quiet --extract --force-local < "../${file##*/}" cd $pwd local ret=0 for f in $flist; do if ! check_single_file $fdir/$f; then ret=1 if test -z "$check_all"; then break fi fi done rm -rf old/$fdir new/$fdir return $ret ;; *.squashfs) flist=`unsquashfs -no-progress -ls -dest '' "new/$file" | grep -Ev '^(Parallel unsquashfs:|[0-9]+ inodes )' | sort` fdir=$file.extract.$PPID.$$ unsquashfs -no-progress -dest old/$fdir "old/$file" unsquashfs -no-progress -dest new/$fdir "new/$file" local ret=0 for f in $flist; do if ! check_single_file $fdir/$f; then ret=1 if test -z "$check_all"; then break fi fi done rm -rf old/$fdir new/$fdir return $ret ;; *.tar|*.tar.bz2|*.tar.gz|*.tgz|*.tbz2) flist=`tar tf new/$file` pwd=$PWD fdir=`dirname $file` cd old/$fdir tar xf `basename $file` cd $pwd/new/$fdir tar xf `basename $file` cd $pwd local ret=0 for f in $flist; do if ! check_single_file $fdir/$f; then ret=1 if test -z "$check_all"; then break fi fi done return $ret ;; *.zip|*.jar|*.war) cd old unjar_l ./$file |sort > flist # 10-05-2010 14:39 sed -i -e "s, [0-9][0-9]-[0-9][0-9]-[0-9]\+ [0-9][0-9]:[0-9][0-9] , date ," flist # 2012-02-03 07:59 sed -i -e "s, 20[0-9][0-9]-[0-9][0-9]-[0-9][0-9] [0-9][0-9]:[0-9][0-9] , date ," flist cd ../new unjar_l ./$file |sort> flist sed -i -e "s, [0-9][0-9]-[0-9][0-9]-[0-9]\+ [0-9][0-9]:[0-9][0-9] , date ,; " flist sed -i -e "s, 20[0-9][0-9]-[0-9][0-9]-[0-9][0-9] [0-9][0-9]:[0-9][0-9] , date ," flist cd .. if ! cmp -s old/flist new/flist; then echo "$file has different file list" diff -u old/flist new/flist return 1 fi flist=`grep date new/flist | sed -e 's,.* date ,,'` pwd=$PWD fdir=`dirname $file` cd old/$fdir unjar `basename $file` cd $pwd/new/$fdir unjar `basename $file` cd $pwd local ret=0 for f in $flist; do if test -f new/$fdir/$f && ! check_single_file $fdir/$f; then ret=1 if test -z "$check_all"; then break fi fi done return $ret;; *.pyc|*.pyo) perl -e "open fh, '+<', 'old/$file'; seek fh, 4, SEEK_SET; print fh '0000';" perl -e "open fh, '+<', 'new/$file'; seek fh, 4, SEEK_SET; print fh '0000';" ;; *.dvi) # Opcodes 247: pre; i[1], num[4], den[4], mag[4], k[1], x[k] perl -e " my \$rec; open fh, '+<', 'old/$file'; my \$dummy = read fh, \$rec, 15; (\$pre, \$i, \$num, \$den, \$mag, \$k) = unpack('C2 N3 C', \$rec); seek fh, 15, SEEK_SET; while (\$k > 0) { print fh '0'; \$k--; } " perl -e " my \$rec; open fh, '+<', 'new/$file'; my \$dummy = read fh, \$rec, 15; (\$pre, \$i, \$num, \$den, \$mag, \$k) = unpack('C2 N3 C', \$rec); seek fh, 15, SEEK_SET; while (\$k > 0) { print fh '0'; \$k--; } " ;; *.bz2) bunzip2 -c old/$file > old/${file/.bz2/} bunzip2 -c new/$file > new/${file/.bz2/} check_single_file ${file/.bz2/} return $? ;; *.gz) gunzip -c old/$file > old/${file/.gz/} gunzip -c new/$file > new/${file/.gz/} check_single_file ${file/.gz/} return $? ;; *.rpm) $self_script -a old/$file new/$file return $? ;; *png) # Try to remove timestamps, only if convert from ImageMagick is installed if [[ $(type -p convert) ]]; then convert old/$file +set date:create +set date:modify old/${file}.$PPID.$$ convert new/$file +set date:create +set date:modify new/${file}.$PPID.$$ mv -f old/${file}.$PPID.$$ old/${file} mv -f new/${file}.$PPID.$$ new/${file} if ! diff_two_files; then return 1 fi return 0 fi ;; /usr/share/locale/*/LC_MESSAGES/*.mo|/usr/share/locale-bundle/*/LC_MESSAGES/*.mo) for f in old/$file new/$file; do sed -i -e "s,POT-Creation-Date: ....-..-.. ..:..+....,POT-Creation-Date: 1970-01-01 00:00+0000," $f done ;; */rdoc/files/*.html) # ruby documentation # Mon Sep 20 19:02:43 +0000 2010 for f in old/$file new/$file; do sed -i -e 's%[A-Z][a-z][a-z] [A-Z][a-z][a-z] [0-9]\+ [0-9]\+:[0-9]\+:[0-9]\+ +0000 201[0-9]%Mon Sep 20 19:02:43 +0000 2010%g' $f done strip_numbered_anchors ;; /usr/share/doc/kde/HTML/*/*/index.cache|/usr/share/doc/kde/HTML/*/*/*/index.cache|\ /usr/share/gtk-doc/html/*/*.html|/usr/share/gtk-doc/html/*/*.devhelp2) # various kde and gtk packages strip_numbered_anchors ;; /usr/share/doc/packages/*/*.html|\ /usr/share/doc/packages/*/*/*.html|\ /usr/share/doc/*/html/*.html|\ /usr/share/doc/kde/HTML/*/*/*.html) for f in old/$file new/$file; do sed -i -e ' s|META NAME="Last-modified" CONTENT="[^"]\+"|META NAME="Last-modified" CONTENT="Thu Mar 3 10:32:44 2016"| s||| s| # # # # sed -i -e ' /^/{ : next n /^<\/head>/{ b end_head } s/^// t next s/^/\1 some-date-removed-by-build-compare \5/ t next s/^/\1 some-date-removed-by-build-compare \3/ t next s/^// b next } : end_head ' $f # Gjdoc HtmlDoclet: sed -i -e 's%Generated by Gjdoc HtmlDoclet [0-9,.]*, part of GNU Classpath Tools, on .*, 20.. [0-9]*:..:.. $a\|p$\.m\. GMT.%Generated by Gjdoc.%' $f sed -i -e 's%, on [A-Z][a-z]* [0-9]*, 20?? [0-9]*:??:?? $a|p$\.m\. GMT.

%, on January 1, 2009 0:00:00 a.m. GMT.

%' $f sed -i -e 's%\), on [a-zA-Z]* [0-9][0-9], 20.. [0-9]*:..:.. $a\|p$\.m\. GMT.

%%' $f # deprecated-list is randomly ordered, sort it for comparison case $f in */deprecated-list.html) sort -o $f $f ;; esac done ;; /usr/share/javadoc/gjdoc.properties |\ /usr/share/javadoc/*/gjdoc.properties) for f in old/$file new/$file; do sed -i -e 's|^#[A-Z][a-z]\{2\} [A-Z][a-z]\{2\} [0-9]\{2\} ..:..:.. GMT 20..$|#Fri Jan 01 11:27:36 GMT 2009|' $f done ;; */fonts.scale|*/fonts.dir|*/encodings.dir) for f in old/$file new/$file; do # sort files before comparing sort -o $f $f done ;; /var/adm/perl-modules/*) for f in old/$file new/$file; do sed -i -e 's|^=head2 ... ... .. ..:..:.. ....: C|=head2 Wed Jul 1 00:00:00 2009: C|' $f done ;; /usr/share/man/man3/*3pm) for f in old/$file new/$file; do sed -i -e 's| 3 "20..-..-.." "perl v5....." "User Contributed Perl Documentation"$| 3 "2009-01-01" "perl v5.10.0" "User Contributed Perl Documentation"|' $f trim_man_TH $f trim_man_first_line $f done ;; */share/man/*|/usr/lib/texmf/doc/man/*/*) for f in old/$file new/$file; do trim_man_TH $f trim_man_first_line $f # generated by docbook xml: #.\" Date: 09/13/2010 sed -i -e 's|Date: [0-1][0-9]/[0-9][0-9]/201[0-9]|Date: 09/13/2010|' $f done ;; *.elc) # emacs lisp files for f in old/$file new/$file; do sed -i -e ' s|^;;; .ompiled by abuild@.* on ... ... .. ..:..:.. ....|;;; compiled by abuild@buildhost on Wed Jul 01 00:00:00 2009| s|^;;; from file .*\.el|;;; from file /home/abuild/rpmbuild/BUILD/anthy-9100h/src-util/elc.8411/anthy-azik.el| s|^;;; emacs version .*|;;; emacs version 21.5 (beta34) "kale" XEmacs Lucid.| s|^;;; bytecomp version .*|;;; bytecomp version 2.28 XEmacs; 2009-08-09.| ' $f done ;; /var/lib/texmf/web2c/*/*fmt |\ /var/lib/texmf/web2c/metafont/*.base|\ /var/lib/texmf/web2c/metapost/*.mem) # binary dump of TeX and Metafont formats, we can ignore them for good echo "difference in $file ignored." return 0 ;; */libtool) for f in old/$file new/$file; do sed -i -e 's|^# Libtool was configured on host [A-Za-z0-9]*:$|# Libtool was configured on host x42:|' $f done ;; /etc/mail/*cf|/etc/sendmail.cf) # from sendmail package for f in old/$file new/$file; do # - ##### built by abuild@build33 on Thu May 6 11:21:17 UTC 2010 sed -i -e 's|built by abuild@[a-z0-9]* on ... ... [0-9]* [0-9]*:[0-9][0-9]:[0-9][0-9] .* 20[0-9][0-9]|built by abuild@build42 on Thu May 6 11:21:17 UTC 2010|' $f done ;; */created.rid) # ruby documentation # file just contains a timestamp and nothing else, so ignore it echo "Ignore $file" return 0 ;; */Linux*Env.Set.sh) # LibreOffice files, contains: # Generated on: Mon Apr 18 13:19:22 UTC 2011 for f in old/$file new/$file; do sed -i -e 's%^# Generated on:.*UTC 201[0-9] *$%# Generated on: Sometime%g' $f done ;; /usr/lib/libreoffice/solver/inc/*/deliver.log) # LibreOffice log file echo "Ignore $file" return 0 ;; /var/adm/update-messages/*|/var/adm/update-scripts/*) # fetchmsttfonts embeds the release number in the update shell script. sed -i "s/${name_ver_rel_old_regex_l}/@NAME_VER_REL@/" old/$file sed -i "s/${name_ver_rel_new_regex_l}/@NAME_VER_REL@/" new/$file ;; *.ps) for f in "old/$file" "new/$file"; do sed -i -e ' /^%%CreationDate:[[:blank:]]/d /^%%Creator:[[:blank:]]groff[[:blank:]]version[[:blank:]]/d /^%DVIPSSource:[[:blank:]]/d ' "$f" done ;; *pdf) # PDF files contain a unique ID, remove it # Format of the ID is: # /ID [<9ACE247A70CF9BEAFEE15E116259BD6D> <9ACE247A70CF9BEAFEE15E116259BD6D>] # with optional spaces. pdftex creates also: # /CreationDate (D:20120103083206Z) # /ModDate (D:20120103083206Z) # and possibly XML metadata as well for f in "old/$file" "new/$file"; do sed -i \ '/obj/,/endobj/{ s%/ID \?\[ \?<[^>]\+> \?<[^>]\+> \?\]%/IDrandom%g; s%/CreationDate \?(D:[^)]*)%/CreationDate (D: XXX)%g; s%/ModDate \?(D:[^)]*)%/ModDate (D: XXX)%g; s%[^<]*%XXX%g; s%[^<]*%XXX%g; s%[^<]*%XXX%g; s%[^<]*%XXX%g; s%[^<]*%XXX%g; }' "$f" done ;; */linuxrc.config) echo "${file}" sed -i '/^InitrdID:/s@^.*@InitrdID: something@' "old/$file" sed -i '/^InitrdID:/s@^.*@InitrdID: something@' "new/$file" ;; */ld.so.cache|*/etc/machine-id) # packaged by libguestfs return 0 ;; */etc/hosts) # packaged by libguestfs sed -i 's/^127.0.0.1[[:blank:]].*/127.0.0.1 hst/' "old/$file" sed -i 's/^127.0.0.1[[:blank:]].*/127.0.0.1 hst/' "new/$file" ;; esac ftype=`/usr/bin/file old/$file | sed -e 's@^[^:]\+:[[:blank:]]*@@' -e 's@[[:blank:]]*$@@'` case $ftype in PE32\ executable*Mono\/\.Net\ assembly*) echo "PE32 Mono/.Net assembly: $file" if [ -x /usr/bin/monodis ] ; then monodis old/$file 2>/dev/null|sed -e 's/GUID = {.*}/GUID = { 42 }/;'> ${file1} monodis new/$file 2>/dev/null|sed -e 's/GUID = {.*}/GUID = { 42 }/;'> ${file2} if ! cmp -s ${file1} ${file2}; then echo "$file differs ($ftype)" diff --speed-large-files -u ${file1} ${file2} return 1 fi else echo "Cannot compare, no monodis installed" return 1 fi ;; ELF*executable*|ELF*[LM]SB\ relocatable*|ELF*[LM]SB\ shared\ object*) $OBJDUMP -d --no-show-raw-insn old/$file > $file1 ret=$? $OBJDUMP -d --no-show-raw-insn new/$file > $file2 if test ${ret}$? != 00 ; then # objdump has no idea how to handle it if ! diff_two_files; then return 1 fi return 0 fi filter_disasm $file1 filter_disasm $file2 sed -i -e "s,old/,," $file1 sed -i -e "s,new/,," $file2 elfdiff= if ! diff --speed-large-files -u $file1 $file2 > $dfile; then echo "$file differs in assembler output" head -n 200 $dfile elfdiff="1" fi echo "" >$file1 echo "" >$file2 # Don't compare .build-id, .gnu_debuglink and .gnu_debugdata sections sections="$($OBJDUMP -s new/$file | grep "Contents of section .*:" | sed -r "s,.* (.*):,\1,g" | grep -v -e "\.build-id" -e "\.gnu_debuglink" -e "\.gnu_debugdata" | tr "\n" " ")" for section in $sections; do $OBJDUMP -s -j $section old/$file | sed "s,^old/,," > $file1 $OBJDUMP -s -j $section new/$file | sed "s,^new/,," > $file2 if ! diff -u $file1 $file2 > $dfile; then echo "$file differs in ELF section $section" head -n 200 $dfile elfdiff="1" fi done if test -z "$elfdiff"; then echo "$file: only difference was in build-id, gnu_debuglink or gnu_debugdata, GOOD." return 0 fi return 1 ;; *ASCII*|*text*) if ! cmp -s old/$file new/$file; then echo "$file differs ($ftype)" diff -u old/$file new/$file | head -n 200 return 1 fi ;; directory|setuid\ directory|setuid,\ directory|sticky,\ directory) # tar might package directories - ignore them here return 0 ;; bzip2\ compressed\ data*) if ! check_compressed_file "$file" "bz2"; then return 1 fi ;; gzip\ compressed\ data*) if ! check_compressed_file "$file" "gzip"; then return 1 fi ;; XZ\ compressed\ data*) if ! check_compressed_file "$file" "xz"; then return 1 fi ;; POSIX\ tar\ archive) mv old/$file{,.tar} mv new/$file{,.tar} if ! check_single_file ${file}.tar; then return 1 fi ;; cpio\ archive) mv old/$file{,.cpio} mv new/$file{,.cpio} if ! check_single_file ${file}.cpio; then return 1 fi ;; Squashfs\ filesystem,*) echo "$file ($ftype)" mv old/$file{,.squashfs} mv new/$file{,.squashfs} if ! check_single_file ${file}.squashfs; then return 1 fi ;; broken\ symbolic\ link\ to\ *|symbolic\ link\ to\ *) readlink "old/$file" > $file1 readlink "new/$file" > $file2 if ! diff -u $file1 $file2; then echo "symlink target for $file differs" return 1 fi ;; block\ special\ *) ;; character\ special\ *) ;; *) if ! diff_two_files; then return 1 fi ;; esac return 0 } # We need /proc mounted for some tests, so check that it's mounted and # complain if not. PROC_MOUNTED=0 if [ ! -d /proc/self/ ]; then echo "/proc is not mounted" mount -orw -n -tproc none /proc PROC_MOUNTED=1 fi # preserve cmp_spec result for check_all runs ret=$RES for file in $files; do if ! check_single_file $file; then ret=1 if test -z "$check_all"; then break fi fi done if [ "$PROC_MOUNTED" -eq "1" ]; then echo "Unmounting proc" umount /proc fi rm $file1 $file2 $dfile $rename_script rm -rf $dir if test "$ret" = 0; then echo "Package content is identical" fi exit $ret # vim: tw=666 ts=2 shiftwidth=2 et