[Go] Support Array Diff

[hermanschaaf]

Describe the enhancement requested

Similar to the CPP implementation of Array Diff (Python docs: https://arrow.apache.org/docs/python/generated/pyarrow.Array.html#pyarrow.Array.diff) it would be good to have the same in Go. As far as I can tell, there is an array.Equal in Go right now, but not array.Diff.

Like its CPP counterpart, given two string arrays ["one", "two", "three"] and ["two", None, "two-and-a-half", "three"] it would output a string like this:

@@ -0, +0 @@
-"one"
@@ -2, +1 @@
+null
+"two-and-a-half"

Component(s)

Go

[hermanschaaf]

I am happy to submit a PR for this.

Currently I'm thinking it would go into a diff.go file inside go/arrow/array. The function signature would be array.Diff(left, right arrow.Array) (string, error). An error would be returned if two arrays of different types are passed in, for example. (This is also not handled in the CPP version right now. See

arrow/cpp/src/arrow/array/diff.cc

Lines 113 to 123 in ddd0a33

	Status Visit(const NullType&) { return Status::NotImplemented("null type"); }
	Status Visit(const ExtensionType&) { return Status::NotImplemented("extension type"); }
	Status Visit(const DictionaryType&) {
	return Status::NotImplemented("dictionary type");
	}
	Status Visit(const RunEndEncodedType&) {
	return Status::NotImplemented("run-end encoded type");
	}

We can also consider exposing an internal function that returns the differences in Struct format, rather than as a string.

[hermanschaaf]

I have broken the implementation into two parts:

• GH-34790: [Go]: Add array.Diff() #34806 for adding a Diff function that returns an edit script
• GH-34790: [Go]: Add array.Edits.UnifiedDiff #34827 for adding a DiffString function that returns the edit script formatted using Unified Diff format (this is still in progress, opened in draft for now)