ali@0: #!/bin/sh
ali@0: if [ $# -ne 2 ]; then
ali@0:     echo "Usage: sf2bz <sf-project-export>.xml <bz-import>.xml" >&2
ali@0:     exit 1
ali@0: fi
ali@0: input=$1
ali@0: output=$2
ali@0: topdir=/home/ali/wk/slashem/web.scripts
ali@0: cd $topdir
ali@0: cache_dir=$topdir/cache
ali@0: tagsoup=tagsoup/tagsoup-1.0.1.jar
ali@0: #tmpdir=`mktemp -d /tmp/XXXXXXXX` || exit 1
ali@0: tmpdir=/tmp/sf2bz; rm -rf $tmpdir; mkdir $tmpdir
ali@0: mkdir -p $cache_dir/sf/artifacts
ali@0: # Sourceforge's export is pretty good, but misses a few things.
ali@0: # Deal with one of these (descriptions for attached files) here.
ali@0: echo '<?xml version="1.0"?>' > $tmpdir/descriptions.xml
ali@0: echo '<artifacts xmlns:xi="http://www.w3.org/2001/XInclude">' \
ali@0:   >> $tmpdir/descriptions.xml
ali@0: ./get_attached_files $input | cut -s "-d " -f 1,2 | sort -n | uniq | \
ali@0:   while read atid aid; do
ali@0:     mkdir -p $cache_dir/sf/artifacts/$atid
ali@0:     if [ ! -r $cache_dir/sf/artifacts/$atid/$aid.html ]; then
ali@0: 	wget -q -O $cache_dir/sf/artifacts/$atid/$aid.html \
ali@0: 	  "http://sourceforge.net/tracker/index.php?func=detail&aid=$aid&group_id=9746&atid=$atid"
ali@0:     fi
ali@0:     mkdir -p $cache_dir/artifacts/$atid
ali@0:     if [ ! -r $cache_dir/artifacts/$atid/$aid.xhtml ]; then
ali@0: 	java -jar $tagsoup $cache_dir/sf/artifacts/$atid/$aid.html \
ali@0: 	  > $cache_dir/artifacts/$atid/$aid.xhtml
ali@0:     fi
ali@0:     echo "<xi:include href=\"$cache_dir/artifacts/$atid/$aid.xhtml\"/>" >> $tmpdir/descriptions.xml
ali@0: done
ali@0: echo '</artifacts>' >> $tmpdir/descriptions.xml
ali@0: xsltproc --xinclude attachment_descs.xsl $tmpdir/descriptions.xml \
ali@0:   > $tmpdir/attachment-descriptions.xml
ali@0: # Some details of attachments are included in the project export.
ali@0: xsltproc attachment_details.xsl $input > $tmpdir/attachment-details.xml
ali@0: # Then convert the attachments to XML (as required by bugzilla)
ali@0: # pulling in the various bits of information we have gathered.
ali@0: mkdir -p $cache_dir/sf/attachments $cache_dir/attachments
ali@0: mkdir -p $tmpdir/jail
ali@0: ./get_attached_files artifacts.xml | while read atid aid file_id; do
ali@0:     if [ ! -r $cache_dir/sf/attachments/$file_id ]; then
ali@0: 	wget -N -O $cache_dir/sf/attachments/$file_id "http://sourceforge.net/tracker/download.php?group_id=9746&atid=$atid&file_id=$file_id&aid=$aid"
ali@0:     fi
ali@0:     if [ ! -r $cache_dir/attachments/$file_id.xml ]; then
ali@0: 	type=`file -bi $cache_dir/sf/attachments/$file_id | sed 's/;.*//'`
ali@0: 	case "$type" in
ali@0: 	    text/*)
ali@0: 		(cd $tmpdir/jail; patch --dry-run -s -f \
ali@0: 		  < $cache_dir/sf/attachments/$file_id > /dev/null 2>&1)
ali@0: 		if [ $? -lt 2 ]; then
ali@0: 		    pflag="-p"
ali@0: 		else
ali@0: 		    pflag=""
ali@0: 		fi
ali@0: 		;;
ali@0: 	    *)
ali@0: 	        pflag=""
ali@0: 		;;
ali@0: 	esac
ali@0: 	./xml_attached_file $pflag -i $file_id -t $type \
ali@0: 	  $cache_dir/sf/attachments/$file_id > $tmpdir/$file_id.1.xml
ali@0: 	xsltproc attachment_gather.xsl $tmpdir/$file_id.1.xml \
ali@0: 	  > $tmpdir/$file_id.2.xml
ali@0: 	xsltproc --xinclude attachment_post.xsl $tmpdir/$file_id.2.xml \
ali@0: 	  > $cache_dir/attachments/$file_id.xml
ali@0:     fi
ali@0: done
ali@0: # We can't get hold of descriptions of deleted attachments, so we
ali@0: # keep a copy of the file in case we need it. Ideally, we'd make sure
ali@0: # that we only add descriptions, but that's for another day.
ali@0: cp $tmpdir/attachment-descriptions.xml $topdir
ali@0: xsltproc get_users.xsl $input > $tmpdir/raw_users.xml
ali@0: echo '<artifacts xmlns:xi="http://www.w3.org/2001/XInclude">' \
ali@0:   >> $tmpdir/users.1.xml
ali@0: for file in $cache_dir/artifacts/*/*.xhtml; do
ali@0:     echo "<xi:include href=\"$file\"/>" >> $tmpdir/users.1.xml
ali@0: done
ali@0: echo "<xi:include href=\"raw_users.xml\"/>" >> $tmpdir/users.1.xml
ali@0: echo '</artifacts>' >> $tmpdir/users.1.xml
ali@0: xsltproc --xinclude user_names.xsl $tmpdir/users.1.xml > $tmpdir/users.2.xml
ali@0: mkdir -p $cache_dir/sf/users $cache_dir/users
ali@0: echo '<sf_users xmlns:xi="http://www.w3.org/2001/XInclude">' \
ali@0:   >> $tmpdir/users.3.xml
ali@0: xsltproc unknown_users.xsl $tmpdir/users.2.xml | while read user; do
ali@0:     case $user in
ali@0: 	*@users.sourceforge.net)
ali@0: 	    user_id=`echo $user | sed 's/@users.sourceforge.net$//'`
ali@0: 	    if [ ! -r $cache_dir/sf/users/$user_id.html ]; then
ali@0: 		wget -N -O $cache_dir/sf/users/$user_id.html \
ali@0: 		  "http://sourceforge.net/users/$user_id/"
ali@0: 	    fi
ali@0: 	    if [ ! -r $cache_dir/users/$user_id.xhtml ]; then
ali@0: 		java -jar $tagsoup $cache_dir/sf/users/$user_id.html \
ali@0: 		  > $cache_dir/users/$user_id.xhtml
ali@0: 	    fi
ali@0: 	    echo "<xi:include href=\"$cache_dir/users/$user_id.xhtml\"/>" \
ali@0: 	      >> $tmpdir/users.3.xml
ali@0:     esac
ali@0: done
ali@0: echo "<xi:include href=\"users.2.xml\"/>" >> $tmpdir/users.3.xml
ali@0: echo '</sf_users>' >> $tmpdir/users.3.xml
ali@0: xsltproc --xinclude user_names.xsl $tmpdir/users.3.xml > users.xml
ali@0: ln -s $cache_dir/attachments $tmpdir
ali@0: xsltproc bugzilla.xsl $input > artifacts.tmp.xml
ali@0: xsltproc --xinclude bugzilla_post.xsl artifacts.tmp.xml > $output
ali@0: #rm -rf $tmpdir artifacts.tmp.xml