[Nepomuk] find [Filename] in Dolphin always matches all my accented file names

[S Page]

In Kubuntu 11.10 (KDE 4.7.2), searching in Dolphin by [Filename]
always matches a bunch of filenames with accents and Asian glyphs in
them, in addition to files actually matching what I entered.

My filenames with accents are on both ext3 and NTFS partitions.

There are some old bugs about accent handling like
https://bugs.kde.org/show_bug.cgi?id=259318 , but they claimed to be
fixed. Should I just file a new bug?

If I turn on debugging and run nepomukserver --nofork from a terminal,
I see the following in output when I search for the filename
TODO_email_cleanup.txt

[/usr/bin/nepomukservicestub] nepomukqueryservice(31059)/nepomuk
(query service) Nepomuk::Query::SearchRunnable::run: "<?xml
version="1.0"?><filequery queryFiles="true" queryFolders="true"
limit="0" offset="0" fullTextScoring="false"
fullTextScoringOrder="desc" flags=""><requestProperty
uri="http://www.semanticdesktop.org/ontologies/2007/01/19/nie#url"
optional="false"/><and><comparison
property="http://www.semanticdesktop.org/ontologies/2007/03/22/nfo#fileName"
comparator="regex" inverted="false"><literal
datatype="http://www.w3.org/2001/XMLSchema#string">TODO_email_cleanup\\.txt</literal></comparison><optional><and><comparison
property="http://www.semanticdesktop.org/ontologies/2007/01/19/nie#contentSize"
comparator=":" varname="size" inverted="false"/><comparison
property="http://www.semanticdesktop.org/ontologies/2007/01/19/nie#mimeType"
comparator=":" varname="mime" inverted="false"/><comparison
property="http://www.semanticdesktop.org/ontologies/2007/01/19/nie#lastModified"
comparator=":" varname="mtime" inverted="false"/><comparison
property="http://nepomuk.kde.org/ontologies/2010/11/29/kext#unixFileMode"
comparator=":" varname="mode" inverted="false"/><comparison
property="http://nepomuk.kde.org/ontologies/2010/11/29/kext#unixFileOwner"
comparator=":" varname="user" inverted="false"/><comparison
property="http://nepomuk.kde.org/ontologies/2010/11/29/kext#unixFileGroup"
comparator=":" varname="group"
inverted="false"/></and></optional></and></filequery>

" "select distinct ?r ?mime ?reqProp1 ?user ?mtime ?group ?mode ?size
where { { ?r <http://www.semanticdesktop.org/ontologies/2007/01/19/nie#url>
?reqProp1 . ?r <http://www.semanticdesktop.org/ontologies/2007/03/22/nfo#fileName>
?v2 . FILTER(REGEX(STR(?v2), 'TODO_email_cleanup\\.txt', 'i')) .
OPTIONAL { { ?r
<http://nepomuk.kde.org/ontologies/2010/11/29/kext#unixFileGroup>
?group . ?r <http://nepomuk.kde.org/ontologies/2010/11/29/kext#unixFileOwner>
?user . ?r <http://nepomuk.kde.org/ontologies/2010/11/29/kext#unixFileMode>
?mode . ?r <http://www.semanticdesktop.org/ontologies/2007/01/19/nie#lastModified>
?mtime . ?r <http://www.semanticdesktop.org/ontologies/2007/01/19/nie#mimeType>
?mime . ?r <http://www.semanticdesktop.org/ontologies/2007/01/19/nie#contentSize>
?size . } . } ?r a ?v3 . FILTER(?v3 in
(<http://www.semanticdesktop.org/ontologies/2007/03/22/nfo#Folder>,<http://www.semanticdesktop.org/ontologies/2007/03/22/nfo#FileDataObject>))
. } . ?r <http://www.semanticdesktop.org/ontologies/2007/08/15/nao#userVisible>
?v1 . FILTER(?v1>0) . }"

That XML looks legit if a little verbose, and the key part of it
  <comparison property="http://www.semanticdesktop.org/ontologies/2007/03/22/nfo#fileName"
comparator="regex" inverted="false">
    <literal datatype="http://www.w3.org/2001/XMLSchema#string">TODO_email_cleanup\\.txt</literal>
  </comparison>

seems like it should NOT match these files.  Running xmlindexer on the
files with accented file names produces output like, in part:
  <value name='http://www.semanticdesktop.org/ontologies/2007/03/22/nfo#fileName'>Déjà
vu.txt
so I'm stumped why these files match.

Thanks for any suggestions,
--
=S Page