Re: [Tracker] firefox bookmark indexing

[jamie <jamiemcc blueyonder co uk>]

On Tue, 2007-08-14 at 20:01 +0800, Jerry Tan wrote:
> Hi, guys
> I want to write code to support firefox bookmark indexing.
> After a quick evaluation, I got some idea, but still have some problem.
>
>
> 1. There is no document tell us how to add a new service.
> I have to look into code to find how.
> here is my understanding, if I am wrong or missing sth, please correct me.
> I hope that this could be a useful doc for new comer to support a new
> service
>
>
> Add a new service need to do things below.
> (1) add key for tracker configuration tools (optional)
> (2) add this service and its metadata into tracker-search-tool
> (3) modify data/service/default.service to add section for this new service
> (4) add a new function to extract its metadata,
> we can write a new program like tracker-extract-mp3.c,
> so tracker-extract can get its metadata.
>
> (5) add a new function inside index_entity to insert the extracted data
> into database.
>
> and yes, above all,
> we need to define how many data this new service has, and how to stored
> these data in DB.
>
>
>
>
> 2. Just for mozilla/firefox bookmarks,
> it is just a plain RDF file, I don't want to do like beagle to write a
> firefox extension to store bookmarks into another file, just write a
> simple parser to get its data directly.
>
> But for firefox bookmarks, every bookmark has metadata as
> "link,title,add_date,last_visit,last_modify,last_charset".
> how to store them?
>
> one bookmark file have multiple bookmarks,
>
>
> I need to insert a new record in table ServiceType, each record into
> table Service for every bookmark entry,
> but how to manage multiple bookmark information ,they are from one
> single html file,
> Use childService?

we are waiting for XESAM to define how new services are spec'ed

atm XESAM provides an ontology and a means to define new ones by
dropping ini files in a directory

also XESAM should define a way for external indexers/crawlers to operate
as well as a format for xml output that can be read by all xesam
indexers

once above is spec'ed and implemented in tracker it will be trivial for
new services, file formats and external indexers to be extended into
tracker without changing  the source of tracker.

I would prefer such services as bookmarks to be external to tracker and
use XESAM  (which means you will write a program like tracker-extract to
output the bookmark info into a file which tracker will read and index
as needed)

we already have extractors previously written for tomboy but are not
integrated yet as we wait for xesam - I certainly dont wanna bloat about
trackerd by compiling in support for all these formats

jamie