[java] Contiguous Array Test
 

I'm writing a client-server poker-tournament game (which I later want to extend to playing other games) and have started on the part where the system analyses the hand a given player has.

I have a jCard object that represents a card. A player's hand is represented by a jHand object that contains a Stack, which contains a set of jCards.

jCard properties are:
String suit; (Text version of suit, i.e. "Clubs", "Diamonds" etc)
String code; (2 char code 1st Char = suit, 2nd char = value - i.e. HK = King of Hearts)
int value; 1-13 (Aces through Kings)

So, I've another object called a jHandAnalyser that has a number of methods for checking what a hand might be worth. The main property of this object is a jHand reference called myHand.

Calling the constructor of jHandAnalyser passes in the hand you want to analyse. e.g.

jHandAnalyser aPlayer = new jHandAnalyser(aHand);

This jHandAnalyser needs to look at the jHand object it's given and decide what the hand is worth.

So far I have a method called isFlush() that returns true if the cards are all of the same suit, but am wondering how to implement the isStrait (sp?) method. i.e. I want to check that all the cards are in a sequential run. Has anyone got any ideas how I can do this nice and cleanly?

you could make sure the cards are always sorted,
then check the run length from [0] to length.

OK, yes, I've got that working fine - except where Aces being hi, low or Hi+Low - I converted the Stack(I want to preserve the order of dealing) to a list sorted by value and checked that after the lowest card, that the value increments by 1 each iteration.

I would say if you sort the hand it may be easier, but here is a non-sorted method I came up with, in honor of the IOCCC -- meaning it isn't obvious what I'm doing, maybe you can figure it out.

This is assuming a 5 card hand, if you are doing other sized hands You'll need to modify the check phase steps 7+. I haven't tested this, but I think it is correct

If I recall a ace cannot be high and low in a straight. Q K A 2 3 is not a valid straight.

it can't be high and low in a straight, but it can be high or low in a hand - so A2345 and TJQKA could both be valid straights.

If an ace is given value 1 (i.e. aces are low), then i can start increment-testing my card-values from 1, it it's high, then I can assume a value of 14, and expect it at the end of any straights - but if it's simultaneously 1 and 14, I'm going to need to do some special coding for it (which I'd rather avoid, but which looks unavoidable at the mometn)

a-j is that bit-shifting you're doing there? That's not a bad idea at all - any hand could be held as a 52 bit bitmask. The only problem is squeezing in cards of the same value but different suits...

I've just gone away and had a think about this, and think that using a bitmask is genius!
so each card value is assigned a bit i.e.
It also solves the roaming Ace problem too.

What's the significance of the number 0x3C02 ?

I special cased the the A high/low problem.

In binary it is 11 1100 0000 0010
Using the value of the card as the offset in the bitmask (A=1, K=13), the above number represents K Q J T A.

I like it - thanks for the idea - again, genius!

OK, I've got it to work and put in an acesHiLow variable (-1 for low, 0 for either or 1 for high) and this should work as a test to see if all the cards in a given hand are in a run.

Where aces are high or low, the thing only works for 5-card hands, but hopefully I'll work out a way to fix it so it will work for any number of cards.

Thanks again.