Hi all,

I have a little problem, I don't find an easy configuration solution but
maybe my google search is wrong :)

- ElisionFilterFactory is enabled for searching and indexing analyzer.
- Index contains: *l'aventure*
=> when I search *l'avent** solr finds nothing

I would have a solution which doesn't look sexy: having another index
with a patternreplacecharfilterfactory wich removes all "'" in strings.

Some tips would be usefull.

Thanks,
Claire;

Unfortunately, use of a wildcard causes the normal token analysis processing
to be completely bypassed, including the elision filter.  So, when using a
wildcard you have to simulate in your head all of the analysis features,
such as manually performing the elision.

-- Jack Krupansky

-----Original Message-----
From: Claire Hernandez
Sent: Friday, May 04, 2012 5:08 AM
To: [EMAIL PROTECTED]
Cc: Jonathan Druart
Subject: search case: Elision and truncate in french

Hi all,

I have a little problem, I don't find an easy configuration solution but
maybe my google search is wrong :)

- ElisionFilterFactory is enabled for searching and indexing analyzer.
- Index contains: *l'aventure*
=> when I search *l'avent** solr finds nothing

I would have a solution which doesn't look sexy: having another index
with a patternreplacecharfilterfactory wich removes all "'" in strings.

Some tips would be usefull.

Thanks,
Claire;

Jack - that was true, until Solr 3.6+: <http://wiki.apache.org/solr/MultitermQueryAnalysis>

So, Claire, it's possible with the latest Solr release, to do this using bits and pieces of your existing analysis chain.

As Jack said, though, this is a manual chore in pre-Solr-3.6 releases.

Erik
On May 4, 2012, at 08:54 , Jack Krupansky wrote:

> Unfortunately, use of a wildcard causes the normal token analysis processing to be completely bypassed, including the elision filter.  So, when using a wildcard you have to simulate in your head all of the analysis features, such as manually performing the elision.
>
> -- Jack Krupansky
>
> -----Original Message----- From: Claire Hernandez
> Sent: Friday, May 04, 2012 5:08 AM
> To: [EMAIL PROTECTED]
> Cc: Jonathan Druart
> Subject: search case: Elision and truncate in french
>
> Hi all,
>
> I have a little problem, I don't find an easy configuration solution but
> maybe my google search is wrong :)
>
> - ElisionFilterFactory is enabled for searching and indexing analyzer.
> - Index contains: *l'aventure*
> => when I search *l'avent** solr finds nothing
>
> I would have a solution which doesn't look sexy: having another index
> with a patternreplacecharfilterfactory wich removes all "'" in strings.
>
> Some tips would be usefull.
>
> Thanks,
> Claire;

Well, if it was "fixed", then it is now broken again - in the 3.6 release!
Here’s a snippet from debugQuery showing that the generated query has the
elision intact in the analyzed term:

<str name="rawquerystring">text_fr:l'avion*</str>
<str name="querystring">text_fr:l'avion*</str>
<str name="parsedquery">+text_fr:l'avion*</str>
<str name="parsedquery_toString">+text_fr:l'avion*</str>

And for the same term without wildcard:

<str name="rawquerystring">text_fr:l'avion</str>
<str name="querystring">text_fr:l'avion</str>
<str name="parsedquery">+text_fr:avion</str>
<str name="parsedquery_toString">+text_fr:avion</str>

-- Jack Krupansky

-----Original Message-----
From: Erik Hatcher
Sent: Friday, May 04, 2012 9:06 AM
To: [EMAIL PROTECTED]
Subject: Re: search case: Elision and truncate in french

Jack - that was true, until Solr 3.6+:
<http://wiki.apache.org/solr/MultitermQueryAnalysis>

So, Claire, it's possible with the latest Solr release, to do this using
bits and pieces of your existing analysis chain.

As Jack said, though, this is a manual chore in pre-Solr-3.6 releases.

Erik
On May 4, 2012, at 08:54 , Jack Krupansky wrote:

> Unfortunately, use of a wildcard causes the normal token analysis
> processing to be completely bypassed, including the elision filter.  So,
> when using a wildcard you have to simulate in your head all of the
> analysis features, such as manually performing the elision.
>
> -- Jack Krupansky
>
> -----Original Message----- From: Claire Hernandez
> Sent: Friday, May 04, 2012 5:08 AM
> To: [EMAIL PROTECTED]
> Cc: Jonathan Druart
> Subject: search case: Elision and truncate in french
>
> Hi all,
>
> I have a little problem, I don't find an easy configuration solution but
> maybe my google search is wrong :)
>
> - ElisionFilterFactory is enabled for searching and indexing analyzer.
> - Index contains: *l'aventure*
> => when I search *l'avent** solr finds nothing
>
> I would have a solution which doesn't look sexy: having another index
> with a patternreplacecharfilterfactory wich removes all "'" in strings.
>
> Some tips would be usefull.
>
> Thanks,
> Claire;

Okay, the issue is that only *some* of the filters are "multi-term aware"
and the elision filter is one that is NOT multi-term aware.

-- Jack Krupansky

-----Original Message-----
From: Jack Krupansky
Sent: Friday, May 04, 2012 9:42 AM
To: [EMAIL PROTECTED]
Subject: Re: search case: Elision and truncate in french

Well, if it was "fixed", then it is now broken again - in the 3.6 release!
Here’s a snippet from debugQuery showing that the generated query has the
elision intact in the analyzed term:

<str name="rawquerystring">text_fr:l'avion*</str>
<str name="querystring">text_fr:l'avion*</str>
<str name="parsedquery">+text_fr:l'avion*</str>
<str name="parsedquery_toString">+text_fr:l'avion*</str>

And for the same term without wildcard:

<str name="rawquerystring">text_fr:l'avion</str>
<str name="querystring">text_fr:l'avion</str>
<str name="parsedquery">+text_fr:avion</str>
<str name="parsedquery_toString">+text_fr:avion</str>

-- Jack Krupansky

-----Original Message-----
From: Erik Hatcher
Sent: Friday, May 04, 2012 9:06 AM
To: [EMAIL PROTECTED]
Subject: Re: search case: Elision and truncate in french

Jack - that was true, until Solr 3.6+:
<http://wiki.apache.org/solr/MultitermQueryAnalysis>

So, Claire, it's possible with the latest Solr release, to do this using
bits and pieces of your existing analysis chain.

As Jack said, though, this is a manual chore in pre-Solr-3.6 releases.

Erik
On May 4, 2012, at 08:54 , Jack Krupansky wrote:

> Unfortunately, use of a wildcard causes the normal token analysis
> processing to be completely bypassed, including the elision filter.  So,
> when using a wildcard you have to simulate in your head all of the
> analysis features, such as manually performing the elision.
>
> -- Jack Krupansky
>
> -----Original Message----- From: Claire Hernandez
> Sent: Friday, May 04, 2012 5:08 AM
> To: [EMAIL PROTECTED]
> Cc: Jonathan Druart
> Subject: search case: Elision and truncate in french
>
> Hi all,
>
> I have a little problem, I don't find an easy configuration solution but
> maybe my google search is wrong :)
>
> - ElisionFilterFactory is enabled for searching and indexing analyzer.
> - Index contains: *l'aventure*
> => when I search *l'avent** solr finds nothing
>
> I would have a solution which doesn't look sexy: having another index
> with a patternreplacecharfilterfactory wich removes all "'" in strings.
>
> Some tips would be usefull.
>
> Thanks,
> Claire;