[Hibernate Search] Pdf, word, excel search

fabreax · **Posted:** Mon Jun 25, 2007 9:45 am

Hello,

My search engine based on Hibernate Search works perfectly. Now, I want to index pdf, word and excel files.

I think I could use libraries to convert these files to text. Then, insert the text data into Hibernate database (which will index the text).

Do you have another solution ? Could I insert data in index without data in database ?

Thank you.

fabreax · **Posted:** Tue Jun 26, 2007 3:37 am

emmanuel · **Posted:** Tue Jun 26, 2007 11:45 am

Yes I have a solution
Write a Custom field bridge to index your excel file (probably delegating to a library)
and mark the proeprty as @Transient so that you don't need to store the data in the database

fabreax · **Posted:** Tue Jun 26, 2007 2:56 pm

Thank you again Emmanuel, could I put "@Transient" annotation only on a field, not a class ?

Because I just want to have a String field "word file content".

fabreax · **Posted:** Wed Jun 27, 2007 3:42 am

I made several tests.

I don't see a difference between

Code:

@Field(index=Index.TOKENIZED, store=Store.YES)
@Transient
private String titre;

and

Code:

@Field(index=Index.TOKENIZED, store=Store.YES)
private String titre;

If I don't want the data in database, I must delete the line :

Code:

in mapping file.

It works with or without @Transient annotation.

emmanuel · **Posted:** Fri Jun 29, 2007 8:23 am

There is definitely a difference unless @Id is on a getter.
If @Id is on a getter then you need to @Transient the getter

fabreax · **Posted:** Fri Jun 29, 2007 8:45 am

Ok thank you Emmanuel. My id is on getter ;).

fabreax · **Posted:** Tue Jul 10, 2007 10:17 am

I have a little problem ...

I have a field :

Code:

@Field(index=Index.TOKENIZED, store=Store.YES)
@Transient
@Boost(0.5f) // indexation moitie moins importante que celle des autres champs
private String pleinTexte;

First of all, I don't see the boost factor in lucene indexes :

(1.0 displayed).

Then, I would like to have the value in my indexes but not in database.
In the beginning, I deleted the line

Code:

in my Mapping.hbm.xml and I deleted setPleinTexte and getPleinTexte methods.

It works, but when I insert a second object, the field documents.pleinText is updated, not added.

The only solution to have 2 different values for 2 different objects is to keep the line in Mapping.hbm.xml (with or without @Transient, there are the 2 values in database). But if I keep the line ... I have values in database :S

I don't know what is the solution. I thought that to remove database values, we must remove the line in Mapping.hbm.xml and add @Transient annotation, but it changed index behaviour :S

Thank you.

emmanuel · **Posted:** Wed Jul 11, 2007 2:24 pm

if you're using hbm.xml files (ie if your entity is not marked as @Entity, then @Transient is not necessary

@Boost is tested in org.hibernate.search.test.fieldAccess.FieldAccessTest in the unit test suite

I suspect you might once again having issues with getters and attribute access

fabreax · **Posted:** Thu Jul 12, 2007 8:58 am

I don't understand, the only thing I changed is removing :

Code:

Full code removed.

emmanuel · **Posted:** Thu Jul 12, 2007 2:59 pm

Can you try and run with the debugger to see what's going on. I haven't see anything obvious.

fabreax · **Posted:** Fri Jul 13, 2007 11:00 am

I don't know where to put a breakpoint in hibernate search java files ...

fabreax · **Posted:** Mon Jul 16, 2007 11:39 am

emmanuel wrote:

@Boost is tested in org.hibernate.search.test.fieldAccess.FieldAccessTest in the unit test suite

I made a test :
http://opensource.atlassian.com/project ... HSEARCH-96

Thank you.

fabreax · **Posted:** Tue Jul 24, 2007 7:21 am

@Boost seems to work because I displayed the results score.

The problem is that boost value is always 1.0 in Luke !

emmanuel · **Posted:** Wed Jul 25, 2007 3:04 pm

Ah so you mean that Luke has an issue but the boost value is properly defined?