sql_qa_pair/test_field_similarity.py

import unittest

from field_similarity import analyze_fields


class TestAnalyzeFields(unittest.TestCase):
    def test_full_match(self):
        result = analyze_fields("hello world", "hello@@world")
        self.assertEqual(result["field_count"], 2)
        self.assertEqual(result["non_zero_count"], 2)
        self.assertAlmostEqual(result["avg_ratio"], 1.0)
        self.assertAlmostEqual(result["max_ratio"], 1.0)

    def test_partial_match(self):
        result = analyze_fields("abc", "zab")
        self.assertEqual(result["field_count"], 1)
        self.assertEqual(result["non_zero_count"], 1)
        self.assertAlmostEqual(result["avg_ratio"], 2 / 3)
        self.assertAlmostEqual(result["max_ratio"], 2 / 3)

    def test_zero_match(self):
        result = analyze_fields("abc", "xyz")
        self.assertEqual(result["field_count"], 1)
        self.assertEqual(result["non_zero_count"], 0)
        self.assertAlmostEqual(result["avg_ratio"], 0.0)
        self.assertAlmostEqual(result["max_ratio"], 0.0)

    def test_empty_fields_are_skipped(self):
        result = analyze_fields("abc", "ab@@@@bc")
        self.assertEqual(result["field_count"], 2)
        self.assertEqual(result["non_zero_count"], 2)
        self.assertAlmostEqual(result["avg_ratio"], (1 + 1) / 2)
        self.assertAlmostEqual(result["max_ratio"], 1.0)

    def test_all_empty_fields(self):
        result = analyze_fields("abc", "@@")
        self.assertEqual(result, {})

    def test_empty_field_list_string(self):
        result = analyze_fields("abc", "")
        self.assertEqual(result, {})

    def test_empty_question(self):
        result = analyze_fields("", "a@@bb")
        self.assertEqual(result["field_count"], 2)
        self.assertEqual(result["non_zero_count"], 0)
        self.assertAlmostEqual(result["avg_ratio"], 0.0)
        self.assertAlmostEqual(result["max_ratio"], 0.0)


if __name__ == "__main__":
    unittest.main()