|
| 1 | +""" |
| 2 | +Test suite for Gap 2: Subquery-Based Dedup Qualify Promotion. |
| 3 | +
|
| 4 | +Tests that the common dedup pattern: |
| 5 | + SELECT ... FROM (SELECT *, ROW_NUMBER() OVER (...) AS rn FROM t) WHERE rn = 1 |
| 6 | +is promoted to qualify metadata on the outer query unit. |
| 7 | +""" |
| 8 | + |
| 9 | +import pytest |
| 10 | + |
| 11 | +from clgraph.query_parser import RecursiveQueryParser |
| 12 | + |
| 13 | +# ============================================================================ |
| 14 | +# Test Group 1: Qualify Promotion from Subquery Dedup Pattern |
| 15 | +# ============================================================================ |
| 16 | + |
| 17 | + |
| 18 | +class TestDedupQualifyPromotion: |
| 19 | + """Test promotion of subquery-based dedup WHERE to qualify_info.""" |
| 20 | + |
| 21 | + def test_qualify_promotion_eq(self): |
| 22 | + """WHERE rn = 1 with ROW_NUMBER promotes qualify_info on outer unit.""" |
| 23 | + sql = """ |
| 24 | + SELECT id, name |
| 25 | + FROM ( |
| 26 | + SELECT *, ROW_NUMBER() OVER (PARTITION BY id ORDER BY ts DESC) AS rn |
| 27 | + FROM t |
| 28 | + ) |
| 29 | + WHERE rn = 1 |
| 30 | + """ |
| 31 | + parser = RecursiveQueryParser(sql, dialect="bigquery") |
| 32 | + graph = parser.parse() |
| 33 | + |
| 34 | + main_unit = graph.units["main"] |
| 35 | + assert main_unit.qualify_info is not None |
| 36 | + assert main_unit.qualify_info["promoted_from_subquery"] is True |
| 37 | + assert "ROW_NUMBER" in main_unit.qualify_info["window_functions"] |
| 38 | + assert "id" in main_unit.qualify_info["partition_columns"] |
| 39 | + |
| 40 | + def test_qualify_promotion_lte(self): |
| 41 | + """WHERE rn <= 3 with ROW_NUMBER promotes qualify_info.""" |
| 42 | + sql = """ |
| 43 | + SELECT id, name |
| 44 | + FROM ( |
| 45 | + SELECT *, ROW_NUMBER() OVER (PARTITION BY id ORDER BY ts DESC) AS rn |
| 46 | + FROM t |
| 47 | + ) |
| 48 | + WHERE rn <= 3 |
| 49 | + """ |
| 50 | + parser = RecursiveQueryParser(sql, dialect="bigquery") |
| 51 | + graph = parser.parse() |
| 52 | + |
| 53 | + main_unit = graph.units["main"] |
| 54 | + assert main_unit.qualify_info is not None |
| 55 | + assert main_unit.qualify_info["promoted_from_subquery"] is True |
| 56 | + assert "ROW_NUMBER" in main_unit.qualify_info["window_functions"] |
| 57 | + |
| 58 | + def test_qualify_promotion_lt(self): |
| 59 | + """WHERE rn < 3 with ROW_NUMBER promotes qualify_info.""" |
| 60 | + sql = """ |
| 61 | + SELECT id, name |
| 62 | + FROM ( |
| 63 | + SELECT *, ROW_NUMBER() OVER (PARTITION BY id ORDER BY ts DESC) AS rn |
| 64 | + FROM t |
| 65 | + ) |
| 66 | + WHERE rn < 3 |
| 67 | + """ |
| 68 | + parser = RecursiveQueryParser(sql, dialect="bigquery") |
| 69 | + graph = parser.parse() |
| 70 | + |
| 71 | + main_unit = graph.units["main"] |
| 72 | + assert main_unit.qualify_info is not None |
| 73 | + assert main_unit.qualify_info["promoted_from_subquery"] is True |
| 74 | + assert "ROW_NUMBER" in main_unit.qualify_info["window_functions"] |
| 75 | + |
| 76 | + |
| 77 | +# ============================================================================ |
| 78 | +# Test Group 2: Non-Ranking Functions Should NOT Promote |
| 79 | +# ============================================================================ |
| 80 | + |
| 81 | + |
| 82 | +class TestNonRankingNotPromoted: |
| 83 | + """Test that non-ranking window functions are not promoted.""" |
| 84 | + |
| 85 | + def test_sum_window_not_promoted(self): |
| 86 | + """SUM() OVER (...) + WHERE total > 100 should NOT produce qualify_info.""" |
| 87 | + sql = """ |
| 88 | + SELECT id, total |
| 89 | + FROM ( |
| 90 | + SELECT id, SUM(amount) OVER (PARTITION BY id) AS total |
| 91 | + FROM t |
| 92 | + ) |
| 93 | + WHERE total > 100 |
| 94 | + """ |
| 95 | + parser = RecursiveQueryParser(sql, dialect="bigquery") |
| 96 | + graph = parser.parse() |
| 97 | + |
| 98 | + main_unit = graph.units["main"] |
| 99 | + assert main_unit.qualify_info is None |
| 100 | + |
| 101 | + |
| 102 | +# ============================================================================ |
| 103 | +# Test Group 3: Explicit QUALIFY Not Overwritten |
| 104 | +# ============================================================================ |
| 105 | + |
| 106 | + |
| 107 | +class TestExplicitQualifyNotOverwritten: |
| 108 | + """Test that explicit QUALIFY clause is not overwritten by promotion.""" |
| 109 | + |
| 110 | + def test_explicit_qualify_preserved(self): |
| 111 | + """Explicit QUALIFY should remain; promotion should not overwrite.""" |
| 112 | + sql = """ |
| 113 | + SELECT customer_id, order_date |
| 114 | + FROM orders |
| 115 | + QUALIFY ROW_NUMBER() OVER (PARTITION BY customer_id ORDER BY order_date DESC) = 1 |
| 116 | + """ |
| 117 | + parser = RecursiveQueryParser(sql, dialect="bigquery") |
| 118 | + graph = parser.parse() |
| 119 | + |
| 120 | + main_unit = graph.units["main"] |
| 121 | + assert main_unit.qualify_info is not None |
| 122 | + # Explicit QUALIFY should NOT have promoted_from_subquery |
| 123 | + assert main_unit.qualify_info.get("promoted_from_subquery") is not True |
| 124 | + |
| 125 | + |
| 126 | +# ============================================================================ |
| 127 | +# Test Group 4: rn Not in Output Columns |
| 128 | +# ============================================================================ |
| 129 | + |
| 130 | + |
| 131 | +class TestRnNotInOutput: |
| 132 | + """Test that the ranking alias (rn) is not in the outer unit output columns.""" |
| 133 | + |
| 134 | + def test_rn_not_in_output(self): |
| 135 | + """Outer SELECT id, name should not include rn in output columns.""" |
| 136 | + sql = """ |
| 137 | + SELECT id, name |
| 138 | + FROM ( |
| 139 | + SELECT *, ROW_NUMBER() OVER (PARTITION BY id ORDER BY ts DESC) AS rn |
| 140 | + FROM t |
| 141 | + ) |
| 142 | + WHERE rn = 1 |
| 143 | + """ |
| 144 | + parser = RecursiveQueryParser(sql, dialect="bigquery") |
| 145 | + graph = parser.parse() |
| 146 | + |
| 147 | + main_unit = graph.units["main"] |
| 148 | + output_col_names = [c.get("name", "") for c in main_unit.output_columns] |
| 149 | + assert "rn" not in output_col_names |
| 150 | + |
| 151 | + |
| 152 | +# ============================================================================ |
| 153 | +# Test Group 5: ranking_window_columns Populated on Inner Unit |
| 154 | +# ============================================================================ |
| 155 | + |
| 156 | + |
| 157 | +class TestRankingWindowColumns: |
| 158 | + """Test that inner subquery unit has ranking_window_columns metadata.""" |
| 159 | + |
| 160 | + def test_ranking_window_columns_populated(self): |
| 161 | + """Inner unit should have ranking_window_columns with correct metadata.""" |
| 162 | + sql = """ |
| 163 | + SELECT id, name |
| 164 | + FROM ( |
| 165 | + SELECT *, ROW_NUMBER() OVER (PARTITION BY id ORDER BY ts DESC) AS rn |
| 166 | + FROM t |
| 167 | + ) |
| 168 | + WHERE rn = 1 |
| 169 | + """ |
| 170 | + parser = RecursiveQueryParser(sql, dialect="bigquery") |
| 171 | + graph = parser.parse() |
| 172 | + |
| 173 | + # Find the inner subquery unit (not 'main') |
| 174 | + inner_units = [u for uid, u in graph.units.items() if uid != "main"] |
| 175 | + assert len(inner_units) >= 1 |
| 176 | + |
| 177 | + # At least one inner unit should have ranking_window_columns |
| 178 | + inner_with_ranking = [u for u in inner_units if u.ranking_window_columns] |
| 179 | + assert len(inner_with_ranking) >= 1 |
| 180 | + |
| 181 | + inner_unit = inner_with_ranking[0] |
| 182 | + assert "rn" in inner_unit.ranking_window_columns |
| 183 | + rn_meta = inner_unit.ranking_window_columns["rn"] |
| 184 | + assert rn_meta["function"] == "ROW_NUMBER" |
| 185 | + assert "id" in rn_meta["partition_by"] |
| 186 | + |
| 187 | + |
| 188 | +if __name__ == "__main__": |
| 189 | + pytest.main([__file__, "-v", "--tb=short"]) |
0 commit comments