[yunqa.de] Re: DiSqlite - How to get all snippets if more than one occurrence?
- From: Delphi Inspiration <delphi@xxxxxxxx>
- To: yunqa@xxxxxxxxxxxxx
- Date: Tue, 16 Mar 2010 13:28:06 +0100
At 12:59 16.03.2010, Edwin Yip wrote:
>Thank you for the comments. Two derived questions then, if I could:
>
>* It seems that the offsets function returns information about the bytes,
>only, but not characters? How to handle UNICDOE then? For example, the
>description of the 3rd integer of the return value of offsets function: "The
>byte offset of the matching term within the column.".
The DISQLite3_Full_Text_Search demo needs character offsets as well and has a
converter function in DISQLite3_Full_Text_Search_Form_Main.pas:
{ Converts an Offset string to an array of TOffsetInfo. Takes care to convert
UTF-8 byte positions to WideString / UnicodeString character indexes.
Therfore, the corresponding Content string must be passed as well. }
function DecodeOffsets(
const Offsets: Utf8String;
const Content: UnicodeString): TOffsetInfoArray;
>When manually extracting the snippets from a large string (thousands of
>characters), what's the most efficient way? especially that I need to also
>insert the <b></b> around the matching keywords? Do you have a library that
>make these kind of things easier? You know, you have a bunch of libraries ;)
No, I do not have a library for everything ;-) For my custom snippets()
function, I would start off with simple string concatenation and then optimize
later.
Be aware that a good snippet extractor is not as simple as it might seem: FTS3
goes quite a way to extract both relevant and "good looking" snippets.
Ralf
_______________________________________________
Delphi Inspiration mailing list
yunqa@xxxxxxxxxxxxx
//www.freelists.org/list/yunqa
Other related posts:
