numpy-如何方便地比较自定义dtype？

Question

I have a dtype that has more than 30 fields. I want to compare two objects with that dtype so that I know exactly which fields are unequal. A trivial solution would be to hard-code each field comparison in a series of if statements:

if (obj1['field1']==obj2['field1']) DO_SOMETHING
if (obj1['field2']==obj2['field2']) DO_SOMETHING
# ...  

Is there a better way to compare two objects with custom dtypes and know exactly which fields match or not?

Answer 1

You can access an object's dtype fields by OBJECT.dtype.names . So:

# obj1 and obj2 are elements in a numpy array with a custom dtype
for field in obj1.dtype.names:
    if obj1[field]==obj2[field]:
        # DO_SOMETHING