EVE Technology Lab

I've pasted a kill mail parser along with an example of using it to iterate through the parsed results here:
Kill Mail Parser
Example Usage

It's implemented with declarative set of grammar rules and a few functions that get triggered when certain tokens are parsed to provide some additional semantic checks.

The result is something that's both sanitized and simple to work with for further processing such as converting strings to ids for storing in your database. The parser leverages the pyparsing module that performs a recursive descent parse given the grammar you define.

Something I found quite useful while spending a couple of hours learning about this parsing package is that each grammar rule can itself parse a string. So for example to verify that the datetime rule is working as expected you can test it directly:

g = KillMailGrammar()
r = g.datetime.parseString("2011.02.29 12:12:12") # 2011 wasn't a leap year...

>>> pyparsing.ParseException: day is out of range for month

r = g.datetime.parseString("2011.02.28 12:12") # Valid day and leave off seconds as EDK does
print r
['2011.02.28', '12:12:00']

So it adds in the missing :00 for you too. I tested it out a bit with a few ship kills, a capsule, a POS, a customs station etc. If you try it and find a glitch let me know.

Added support for Russian and German text. Wasn't too hard actually. The static text on the right-hand side of the Kill mail gets translated to English now as well - except for Faction names - I'll deal with fixing that tomorrow.

Check out a sample input file in Russian: russian.txt and the resulting print out of the parsed data including things like (laid the final blow) and (cargo) translated as well: Killmail Parse printout

Can't say enough about how fantastic the pyparsing module is for a task like this. Really helps to avoid having to write of bunch hacky code.

Kill Mail Parser
Example Usage

What's left to do is get the mapping of faction names for English to Russian and German. It now converts, the Russian and German equivalents with various spelling for Unknown, None, Victim, Faction In Container and Drone Bay properly along with switching the decimal separator from comma to dot and allowing Customs Office Kill Mails with no victim.

Also strangely for Russian Kill mails the destroyed items sometimes appear before the dropped items and an extraneous decimal point (comma) with zeroes appears for integer values like Damage Taken and Done so that's all dealt with too - along with the weird (No Data) instead of None or Unknown in Russian.