I thought something fishy with these Dijkstra solutions,
but if max degree is 4 there's not much fine for multiple pushing
of the same node into the heap.

And if max dimension for a test case is 50, you can also get away
with running a linear search for the minimum on a list instead
of a heap :D

Instructions are not clear.
Does the monster makes a move after every move you make?
And what does:

And yes, wall is close enough to escape without meeting a monster.

Means?

Description says:

Apply the Texas Hold'em rules for ranking the cards.

Anyway, good job buddy.

You did reproduce my example. In this kata the hands should be ordered from strong to weak.

The poker hands must be sortable by rank, the highest rank first:

Returned value for:

sorted(map(PokerHand, ['QS QD QH QC 3C', 'JS JD JH JC AC'])) == ['JS JD JH JC AC', 'QS QD QH QC 3C']

Should be False.

Suits are irrelevent in this example.

Intresting tricks here, but the test cases must be really weak
if this one pass.

It does not properly distinguish between hands of the same caliber:
'QS QD QH QC 3C' < 'JS JD JH JC AC' => False

Oh brother...(sigh)

“All happy families are alike; each unhappy family is unhappy in its own way.”

It's because the Test cases provided are ridicously small.
With suffiently large string as argument this function will choke.

Actually the complexity of this solution is quadratic (very bad!).

test for "("*1000000 + ")"*1000000

I see some solutions with ridiculous complexity
O(n^2) when linear is possible.