Question

I’m working with PostgreSQL.

I need to transform “http://www.xyz.com/some_uri/index1.html” in something like “scdfdsffd”(some unique key, based on the URL that is a unique key in the table).

By other words… the URL is a unique key in the table, but I need to generate a small unique key based on the URL.

What can I do with standard PostgreSQL 8.4?

Best Regards,

Answer 1

Several methods:

a) Why not use an auto-incrementing column or sequence generator to generate unique integers per insert? If you have less than 100 million URLs, your identifiers are short and easy to remember. However, if that’s not an option (e.g. because you don’t want people guessing IDs and attacking the database that way):

b) The built-in MD5() function may help:

INSERT INTO table (pkey, url) VALUES (MD5('http://...'), 'http://...');

MD5() is a hash function and will most likely give you a unique identifier per URL. I say “most likely” because you get a 128-bit hash from MD5, and the likelihood of a hash collision is on the order of 2^-128 (about 10^-55).

If you need smaller identifiers you can chop the result from MD5 down to a smaller number of characters, but you could potentially significantly increase the chance of a hash collision depending on which characters you take.

[Note: timestamp answer redacted since it in no way solves the original problem. -BobG]