yes not able to pass the testcases in java

reported here as an issue

Java:

• sample tests do not compile (missing List import)
• the criterion for odd numbers incorrectly classifies negative odd numbers as even numbers

I optimised from passing 30 test cases to 200 test cases in python and am now running out of ideas for optimization. Is it supposed to be more optimised or is the random test case is in infinite loop causing timeout?

In java version I've not passed tests for odd numbers less than zero. For example -3 occured twice or more in arrA and arrB, but this number are not expected in final list.
I've wrote "quickfix" for filtering odd numbers less than zero and pass all test.
In reference solution I've found line like this "i % 2 == 1 : 0 (by odd or even)". But for numbers less than zero we take -1 for odd numbers hence negative numbers are filtered in reference solution.
Maybe I've not understand this task, please check my points.

super logic

Its ok now. Thanks for quick fix!

I've found next tests in random tests suite:

Incorrect answer for board:

000|000|080
258|790|000
041|036|257
---+---+---
106|179|504
109|400|076
004|368|910
---+---+---
792|600|143
000|010|005
005|203|008

Fourth and Fifth lines are started from 1 "106......" and "109......". Thats invalid sudoku board, because we have duplicate value (1) in first column.

Incorrect answer for board:

009|088|040
002|000|630
304|625|870
---+---+---
405|168|090
080|509|120
100|702|500
---+---+---
008|203|400
006|890|307

In first line we have duplicate value (8) (...088...)

And so on.

In Java version I've found next fixed test with unbalanced parenthesis (26 left, 27 right):
(((()()(()(()()()(()()(()()(()))))))(()))()(()()()))) (note 4 closing brackets at the end).

Following error message for this test is:
2*(2*(2*(1+1+2*(1+2*(1+1+1+2*(1+1+2*(1+1+2*(1))))))+2*(1))+1+2*(1+1+1))) = 790 ... (note 3 closing brackets at the end).

It should be fixed.

Thank you for the answer!

I understand how it gets the index in string, where is really a second 'c', but how it returns -1 in other cases?

You are absolutely right, it was not working properly for cases where the arrow starts from the beginning of the string. My python solution contained the right version of the operator, but somehow I mistranslated it. Sadly there were no fix tests with an arrow starting at the beginning of the string, so thank you very much for your insight!

I updated the solution, and also added a sample test with string "-->" to cover this case.

In searchArrows method (7th line): while (j > 0 && string.charAt(j) == ...) {

Should be: while (j >= 0 && string.charAt(j) == ...) {

Mine solution pass by luck (after almost 10 attempts), but others cannot solve this kata in Java for this reason (IMO).
In any case, I appreciate your contribution for Java branch. Thanks for all your work!

Lambda Calculus translation