Re: Beagle indexing of del.icio.us , Scuttle or other social bookmarking tags?

[Debajyoti Bera <dbera web gmail com>]

> As anyone through through this potential mash-up already?
>
> Related to "too many different search boxes in my life"...
>
> Have been using del.icio.us lately (alas also thinking that I should in
> addition, or instead, use Scuttle http://sourceforge.net/projects/scuttle/ 
> or some other Open Source alternative for 'social bookmarking') and am
> reaching for ability gain the additional power of index of my del.icio.us
> tags and comments when I do a Beagle search -- i.e. using index to increase
> relevancy of web site returns within Beagle, or perhaps even list my
> del.icio.us tag window as a return? Thus, remove the del.icio.us search box
> as one of my routine search box choices (leaving 'only' Google , Beagle and
> secure intranet boxes).

Hi
	I have read people talking about it, but there are no known implementations 
in that direction. For beagle to search some external (web-service) data 
source, there are two possible approaches:

1) Write a querydriver - See "Querying backends" in 
http://beagle-project.org/Architecture_Overview
    Beagle currently ships with GoogleDriver (disabled by default), which is 
an example implementation. In this mode, data is not stored in beagle, but on 
each query, beagle queries the data source on the fly and returns results. 
This approach will work with webservices if they provide soap APIs.

2) Store data in beagle itself - write a full-fledged backend (again see 
http://beagle-project.org/Architecture_Overview for what it means) which will 
find data to index and store it in beagle. Searching the data is faster.

2.a) Instead of writing a full backend, if the data source is simple enough, 
you can use the IndexingService backend. Using it, you send data to beagle 
for indexing. Search is fast. As usual, there are two choices ;-)

2.a.1) Write a script/program to put all the data to index and a descriptor 
file in files in ~/.beagle/ToIndex directory. From 
IndexingServiceQueryable.cs
// (2) Files: The file to be indexed is dropped into the ~/.beagle/ToIndex
// directory.  Another file with the same name prepended with a period is
// also dropped into the directory.  In that file is the metadata for the
// file being indexed.  The first line is the URI of the data being indexed.
// The second line is the hit type.  The third line is the mime type.  Then
// there are zero or more properties in the form "type:key=value", where
// "type" is either 't' for text or 'k' for keyword.  This method is a lot
// easier to use, but requires that Beagle have inotify support enabled to
// work.

2.a.2) Use the IndexingService API to send data to beagle for indexing. You 
can write a C program or python script to write a program to regularly fetch 
data off the webservices and send it to beagle for indexing. 
libbeagle/example contains several example C programs to use the API.

- dBera

-- 
-----------------------------------------------------
Debajyoti Bera @ http://dtecht.blogspot.com
beagle / KDE fan
Mandriva / Inspiron-1100 user