> I'm building a document handling system. The document files (.doc
> mostly) are stored as discrete files in a normal file folder system on
> a Windows 2003 server. Each document in the file system has a
> corresponding record in a FB 2.0.3 database. The records in the
> database contain meta-data about the documents on disk, including
> "security" information about if the currently logged in user is
> permitted access to the document. All pretty basic stuff... DB design
> class 101.
>
> I need to allow my users to do a full text search of the contents of
> the documents, a text search of each field of meta-data, and obey the
> security settings stored in the database.
>
> Several options are available to me, one of which is to use some kind
> of "text extractor" to extract the text of each document and store it
> in a BLOB field in the database. (Both keyword/list and count (for
> relevancy stats), plus all the words in the document in one continuous
> string (for "phrase" searching)). Keeping in mind the 16TB limit of
> FB 2.0.3 which I think would be adequate.
>
> Please vote:

Actually, I am not sure what is the question.
Whether to implement text search using Firebird ?
Or whether it could handle really huge volume of data ? (mentioning 16TB limit)
Or whether to index metadata extracted from discrete files ?
Or something other (the Subject mentions Security) ?

> If you vote 3, can you recommend a "text extractor"?

Perhaps you can use come standard Delphi components -
on Servers page, there is TWordDocument etc., which should be
a COM interface to Word application (I have never used it though).

Ivan