[MacPerl] how to grow a hash of hashes

Guten Tag,

and thanks for the response. Let me explain the context.

I have a plain text IPN archive. It contains about 7500 records and =
keeps growing. Each record is: IPN, name1, name2, etc. Some records =
contains several hundred names. I prefer to keep this archive as plain =
text for simplicity, and portability.

Every once or twice a day, I collect new IPN and names from raw text, =
email or others. It should not exceed a few dozens, or maybe a few =
hundreds if the collection has been suspended for a while.

And a few times a week I run an update of the archive. In case there are =
too many new records, I can split them in several batches.

Updating the archive consists in reading it sequentially, and matching =
its records against the new records:

For each archive record:
1- is the IPN in the new records ?
      if so, what new names should be added to the IPN in the archive ?
2- if not, what domains (2 top levels) in the new records match those in =
the archive ? (for clustering)

It seems to me convenient to build a "new record" structure geared to =
provide answers to these questions. Hence the two hashes for names and =
domains. A hierarchical structure as you mentionned (64.207.97.183  >  =
com  >  web-mania  >  server7) could work too, although I wonder if it =
would take a few more tests for matching names. E.g.

Archive contains: 64.207.97.183 nigeria.gov.ng
A new IPN comes with 64.207.97.183 www.nigeria.gov.ng

$ip =3D '64.207.97.183';
$H{$ip}{ng}{gov}{nigeria}{www}=3D'';
exists $H{$ip}{ng}{gov}{nigeria}; # is true,

but "nigeria.gov.ng" is not a name in the new IPN. We have to check keys =
at lower level to find out.

Anyway, whichever structure is used, it should allow for growing the =
lower level elements, since there may be several instances of an IPN in =
the new records.

It is trivial to grow a hash within a hash, e.g.
$H{$ip}{ng}{gov}{nigeria}{www}=3D'';
$H{$ip}{ng}{gov}{nigeria}{oil}=3D'';

but I don't know a way to grow a hash within an array, such as:

$H{$ip}[0]{$nam4}=3D'';  or
$H{$ip}[0]=3D{$nam4,''};

Both constructions erase all previous keys in the hash. Is there another =
way ?

I could replace the array with a 2-key hash. Somewhat funny, but maybe a =
solution.
$H{$ip}{nam} =3D {$nam1,'',$nam2,'',$nam3,''};
$H{$ip}{dom} =3D {$dom1,'',$dom2,''}; # and later
$H{$ip}{nam}{$nam4}=3D'';  # this works

Cheers

- -
On Sat, 30 Jul 2005 07:48:20 +0200, Detlef Lindenthal wrote:

>Your question is not clear to me.
To me it seems that you want to have a  clear hierarchy:
      IPN  >  1st level domain  >  2nd level domain  >  3rd level 
domain  > ... nth level domain
e.g.    
      64.207.97.183  >  com  >  web-mania  >  server7

And for this I would mould a hashes landscape. What is the difficulty?

BTW, do you have some hundred entries, or thousands of entries?
(If MacPerl hashes grow too big, they can become slow and malicious; so 
one could prefer an SQL-solution.)