Mercurial > hg
view tests/test-convert-cvsnt-mergepoints @ 12252:4481f8a93c7a stable
convert/darcs: handle non-ASCII metadata in darcs changelog (issue2354)
Given a commit author or message with non-ASCII characters in a darcs
repo, convert would raise a UnicodeEncodeError when adding changesets
to the hg changelog.
This happened because etree returns back unicode objects for any text
it can't encode into ASCII. convert was passing these objects to
changelog.add(), which would then attempt encoding.fromlocal() on
them.
This patch ensures converter_source.recode() is called on each piece
of commit data returned by etree.
(Also note that darcs is currently encoding agnostic and will print
out whatever is in a patch's metadata byte-for-byte, even in the XML
changelog.)
author | Brodie Rao <brodie@bitheap.org> |
---|---|
date | Fri, 10 Sep 2010 09:30:50 -0500 |
parents | 56a5f80556f5 |
children |
line wrap: on
line source
#!/bin/sh "$TESTDIR/hghave" cvs || exit 80 filterpath() { eval "$@" | sed "s:$CVSROOT:*REPO*:g" } cvscall() { echo cvs -f "$@" cvs -f "$@" } # output of 'cvs ci' varies unpredictably, so discard most of it # -- just keep the part that matters cvsci() { echo cvs -f ci -f "$@" cvs -f ci -f "$@" 2>&1 | egrep "^(new|initial) revision:" } hgcat() { hg --cwd src-hg cat -r tip "$1" } echo "[extensions]" >> $HGRCPATH echo "convert = " >> $HGRCPATH echo "graphlog = " >> $HGRCPATH echo "% create cvs repository" mkdir cvsmaster cd cvsmaster CVSROOT=`pwd` export CVSROOT CVS_OPTIONS=-f export CVS_OPTIONS cd .. filterpath cvscall -Q -d "$CVSROOT" init echo "% checkout #1: add foo.txt" cvscall -Q checkout -d cvsworktmp . cd cvsworktmp mkdir foo cvscall -Q add foo cd foo echo foo > foo.txt cvscall -Q add foo.txt cvsci -m "add foo.txt" foo.txt cd ../.. rm -rf cvsworktmp echo "% checkout #2: create MYBRANCH1 and modify foo.txt on it" cvscall -Q checkout -d cvswork foo cd cvswork cvscall -q rtag -b -R MYBRANCH1 foo cvscall -Q update -P -r MYBRANCH1 echo bar > foo.txt cvsci -m "bar" foo.txt echo baz > foo.txt cvsci -m "baz" foo.txt echo "% create MYBRANCH1_2 and modify foo.txt some more" cvscall -q rtag -b -R -r MYBRANCH1 MYBRANCH1_2 foo cvscall -Q update -P -r MYBRANCH1_2 echo bazzie > foo.txt cvsci -m "bazzie" foo.txt echo "% create MYBRANCH1_1 and modify foo.txt yet again" cvscall -q rtag -b -R MYBRANCH1_1 foo cvscall -Q update -P -r MYBRANCH1_1 echo quux > foo.txt cvsci -m "quux" foo.txt echo "% merge MYBRANCH1 to MYBRANCH1_1" filterpath cvscall -Q update -P -jMYBRANCH1 # carefully placed sleep to dodge cvs bug (optimization?) where it # sometimes ignores a "commit" command if it comes too fast (the -f # option in cvsci seems to work for all the other commits in this # script) sleep 1 echo xyzzy > foo.txt cvsci -m "merge1+clobber" foo.txt echo "% return to trunk and merge MYBRANCH1_2" cvscall -Q update -P -A filterpath cvscall -Q update -P -jMYBRANCH1_2 cvsci -m "merge2" foo.txt REALCVS=`which cvs` echo "for x in \$*; do if [ \"\$x\" = \"rlog\" ]; then echo \"RCS file: $CVSROOT/foo/foo.txt,v\"; cat $TESTDIR/test-convert-cvsnt-mergepoints.rlog; exit 0; fi; done; $REALCVS \$*" > ../cvs chmod +x ../cvs PATH=..:${PATH} hg debugcvsps --parents foo | sed -e 's/Author:.*/Author:/' -e 's/Date:.*/Date:/' cd ..