Originally Posted by The Iron Colonel
Can you explain this bit more? I'm unclear what your meaning is by this statement.
|
Originally, I wrote it for myself without thinking about anyone else really. So what I did to identify the mobs for each segment was search with regular expressions like "Your \w+ hits ([\w ]+) for \d+", and likewise for crits and melee hits/crits, extracting the thing in parens as the mob name. Normally, you wouldn't think that it matters too much if a combat session is labeled, but I found that I would get some combat sessions with junk (no combat at all) if I allowed sessions that didn't have any identified mobs in them. However, if you don't actually hit/crit anything on a boss fight (like say your die on the pull or you're a pure healer), then no mobs would be identified and that heuristic would filter out all of the sessions, cause it thinks there aren't any mobs.
Now what it does is builds a list of players based on who received buffs. Then it looks for the pattern " (\w+)'s \w+ hits ([\w ]+) for \d+", where the first group must be in the list of extracted player names. It works just as well on the logs I tested, though player names were coming up more (it seems a few people were casting while spell reflect was up or dpsing MCed folks)
What I might look into doing in the future is not only tagging lines I know to be out-of-combat, but also lines that I know to be in-combat. That way I might not be so dependent on guessing the player list to decide whether or not anything interesting happened in a combat session.