array_intersect bug

[melin]

array_intersect 有数组越界bug，代码也复杂，简化代码：
` @OverRide
public Object evaluate(DeferredObject[] arguments) throws HiveException {
Object leftArray = arguments[0].get();
Object rightArray = arguments[1].get();

    int leftArrayLength = leftArrayOI.getListLength(leftArray);
    int rightArrayLength = rightArrayOI.getListLength(rightArray);

    // Check if array is null or empty
    if (leftArray == null || rightArray == null || leftArrayLength < 0 || rightArrayLength < 0) {
        return null;
    }

    if (leftArrayLength == 0) {
        return leftArray;
    }
    if (rightArrayLength == 0) {
        return rightArray;
    }

    List<?> leftList = leftArrayOI.getList(leftArray);
    List<?> rightList = leftArrayOI.getList(rightArray);
    HashSet<?> result_set = Sets.newHashSet(leftList);
    result_set.retainAll(rightList);

    return new ArrayList(result_set);
}`

[aaronshan]

@melin 能给下触发数组越界bug的测试用例么？谢谢

[melin]

数据量比较大，没有确定是那条数据，就换掉实现代码了

[minhoryang]

There are 2 bugs:

1. If the duplicate value is located at the end of first array and the same value is stored multiple times in second array, leftPositions[++leftCurrentPosition] raises ArrayIndexOutOfBoundsException.

select array_intersect(array(0,1,2,3,4,5), array(1,1,2,2,5,5))
select array_intersect(array(0,1,2,3,4,4), array(1,1,2,2,3,4))

java.lang.ArrayIndexOutOfBoundsException: 
      at cc.shanruifeng.functions.array.UDFArrayIntersect.evaluate(UDFArrayIntersect.java:147)

You can test this with INITIAL_SIZE = 1 (just set it smaller than input length)

2. Value of leftArrayElementTmp2 and rightArrayElementTmp2 never changed even position changes. So ignoring rest of the values without comparing.

select array_intersect(array(0,1,2,3,4,5), array(1,1,2,2,5,5))
expected: <[1, 2, 5]>, but was: <[1]>

[aaronshan]

@minhoryang Thank you very much! It had been resolved!

[cceazj]

https://github.com/aaronshan/hive-third-functions/blame/f98fef86d328882c85ea40b69b14375d90d44201/src/main/java/com/github/aaronshan/functions/array/UDFArrayIntersect.java#L160
代码compare方法逻辑错误：位置数组写死了leftPositions。参数中应增加int[] positions，传递进来leftPositions或者rightPositions，相应方法调用处也需一并修改

[aaronshan]

https://github.com/aaronshan/hive-third-functions/blame/f98fef86d328882c85ea40b69b14375d90d44201/src/main/java/com/github/aaronshan/functions/array/UDFArrayIntersect.java#L160 代码compare方法逻辑错误：位置数组写死了leftPositions。参数中应增加int[] positions，传递进来leftPositions或者rightPositions，相应方法调用处也需一并修改

@cceazj 能帮忙修复并提个PR吗？