[Dxspider-support] Country prefixes

[Michael G. Carper]

I'm just trying to figure out if I can match the exact prefixes in this
country data in HRD so there is absolutely no way that HRD can get the
country's full name wrong.  So I needed to know what file did this so I
could match them up.

-----Original Message-----
From: Dirk Koopman [mailto:djk at tobit.co.uk] 
Sent: Wednesday, February 12, 2014 6:03 PM
To: Michael Carper
Cc: The DXSpider Support list
Subject: Re: [Dxspider-support] Country prefixes

I am rather hazy as to what it is that you are trying to achieve. 
Perhaps if you could lay it out in more detail, I can help you.

But in summary, this is what happens for DXSpider:

* I read through the wpxloc.raw file and create a prefix centric memory
database.

* I then take the information in cty.dat file and either add information to
the database or create prefix exceptions (which, in turn, may create more
records.

* I file that information for use in DXSpider in /spider/data/prefix_data.pl
which is read on startup and creates the memory database.

* I check whether the complete callsign is in the database. If so, I have a
result.

* Then I mess around with the proffered callsign and, if necessary, strip
off and store things like G1TLH/ZL or ZL/G1TLH (example picked completely at
random :-).

* With some heuristics (1) I create a priority list of parts of the
callsign, This may be a list of just one part (a normal callsign).

* Starting at the head of the priority list, I take that part (2) and try to
find the whole thing in the database. To save further typing and repetition
if, at any point, I find a match - that is the result and no further
processing occurs.

* If no match occurs, I chop off one character from the right hand end and
retry a match. I do this until no more characters exist and no prefix is
found.

* If there was more than one part then I take the next part in the priority
list and go through the whole rigmarole again.

There is a lot of caching of full and partial results that go on here.

(1) See /spider/perl/Prefix.pm sub extract
(2) See /spider/perl/Prefix.pm sub matchprefix

Now having obtained a match, then the rest of the information that is
available (that is derived from the wpxloc.raw file) is then returned. 
One of the fields is the country code (a number).

Dirk

On 12/02/14 22:52, Michael Carper wrote:
> Yep, I'm aware that it's more complex than matching prefixes (which is 
> why I'm asking).  We sure know that "K5K" can be anything from Kingman 
> Reef to a Boy Scout Special Event station in Little Rock, Arkansas.
>
> I was just hoping to see the file that we use in cluster... so that we 
> can match to what comes in the country field, rather than trying to 
> re-do the work that's already been done to match.
>
>
> On Wed, Feb 12, 2014 at 4:43 PM, Dirk Koopman <djk at tobit.co.uk 
> <mailto:djk at tobit.co.uk>> wrote:
>
>     On 12/02/14 20:20, Michael G. Carper wrote:
>
>         These are the same ones that are used in the cluster then?
>
>         So when the Country field says "Anguilla-VP2E"... that's coming
>         from the
>         AD1C list?
>
>
>     That, specifically, is coming from the wpxloc.raw file. This is a
>     sort of "master list" that originated in the AK1A cluster software.
>
>     Dirk
>
>