RE: mime types, parsers, filters

Thank you.  Anyway, I guess I could at least use dd which I think is
more 
efficient.  Now on to clustering...

-----Original Message-----
From: Alexander Barkov [mailto:[email protected]] 
Sent: Tuesday, March 09, 2010 10:29 AM
To: King, Stefan
Cc: [email protected]
Subject: Re: mime types, parsers, filters

Hi Stefan,

King, Stefan wrote:
> I am trying to set up search for a few million files.  There are no
web 
> servers.
> 
> It is very important that I be able to specify the type of document 
> during search,
> 
> so I have set up in indexer.conf to treat every recognizable mime type

> distinctly,
> 
> 
> For example, mime type text/plain is restricted to just .txt files.
> 
> AddType text/plain                      *.txt
> 
> while Java source is
> 
> AddType text/x-java-source              *.java
> 
> so I added this :
> 
> Mime text/x-java-source   text/plain
> 
> This caused the search terms to be recognized only if they were in the

> name of the java
> 
> source file itself, not in the document body.
> 
> So, I did this
> 
> Mime text/x-java-source   text/plain /bin/cat
> 
>  
> 
> and reindexed the test directory, and now the search is able to see
the 
> java source code
> 
> itself.  So I have basically a trivial filter.  However, it seems like
I 
> should not have to do this.
> 
> My question is,
>  
> 
> Is there a way to tell mnogosearch to index a document as if it were 
> mime type
> 
> X even but store the attributes for the document as mime type Y?
> 
>  
> 
> In this case I want mnogosearch to store java-source as the mime type 
> but to index the
> 
> document as text/plain otherwise.  The "cat" trick does work, but I
get 
> the feeling I am
> 
> missing something.  I should not need an external parser at all.
> 
> 
> Where did I go wrong?


Using /bin/cat is the correct way.

Empty command line parameter (to do Mime type "cast", without any 
program execution) is not supported.




> 
>  
> 
> Stefan
> 
>  
> 
>  
> 
>  
>