chore: establish project baseline

tests/test_sampling.py

@@ -0,0 +1,184 @@
+import pytest
+
+from gti_discovery.catalog import FirebirdCatalog, MAX_SAMPLE_ROWS
+from gti_discovery.identifiers import quote_firebird_identifier
+from gti_discovery.firebird_queries import _ADAPTER_REGISTRATION_TOKEN, _execute_catalog, _register_approved_adapter
+
+
+class RecordingAdapter:
+    read_only = True
+
+    def __init__(self, results):
+        self.results = results
+        self.calls = []
+
+    def execute(self, sql, params=()):
+        self.calls.append((sql, params))
+        return self.results[len(self.calls) - 1]
+
+
+_register_approved_adapter(RecordingAdapter, _ADAPTER_REGISTRATION_TOKEN)
+
+
+def make_catalog(sample_rows=None, relation='PRODUCTS', columns=('ID', 'EAN')):
+    adapter = RecordingAdapter([
+        [{'RELATION_NAME': relation, 'RELATION_ID': 1, 'SYSTEM_FLAG': 0}],
+        [{'RELATION_NAME': relation, 'FIELD_NAME': name} for name in columns],
+        sample_rows or [],
+    ])
+    return FirebirdCatalog(adapter), adapter
+
+
+@pytest.mark.parametrize('name', [
+    'PRODUCTS.ID', 'PRODUCTS; DROP TABLE X', 'PRODUCTS/* comment */',
+    'PRODUCTS -- comment', 'COUNT(*)', 'SP_RECALC(1)', '(SELECT 1)',
+    'PRODUCTS\nID',
+])
+def test_identifier_quoting_rejects_sql_shaped_input(name):
+    with pytest.raises(ValueError):
+        quote_firebird_identifier(name)
+
+
+def test_identifier_quoting_doubles_embedded_quotes():
+    assert quote_firebird_identifier('WEIRD"NAME') == '"WEIRD""NAME"'
+
+
+def test_sampling_accepts_catalog_issued_quoted_names():
+    catalog, adapter = make_catalog(
+        relation='WEIRD"TABLE', columns=('WEIRD"FIELD',)
+    )
+
+    catalog.sample_relation_rows('WEIRD"TABLE', ['WEIRD"FIELD'], 1)
+
+    assert adapter.calls[-1][0] == 'SELECT FIRST 1 "WEIRD""FIELD" FROM "WEIRD""TABLE"'
+
+
+@pytest.mark.parametrize('relation', ['UNKNOWN', 'PRODUCTS; SELECT 1', 'SP_RECALC(1)'])
+def test_sampling_rejects_unknown_or_procedure_like_relation(relation):
+    catalog, adapter = make_catalog()
+
+    with pytest.raises(ValueError):
+        catalog.sample_relation_rows(relation, ['ID'], 1)
+
+    assert not any('FIRST' in sql for sql, _ in adapter.calls)
+
+
+@pytest.mark.parametrize('column', ['UNKNOWN', 'ID, PASSWORD', 'COUNT(*)', 'ID/*x*/'])
+def test_sampling_rejects_unknown_or_expression_column(column):
+    catalog, adapter = make_catalog()
+
+    with pytest.raises(ValueError):
+        catalog.sample_relation_rows('PRODUCTS', [column], 1)
+
+    assert not any('FIRST' in sql for sql, _ in adapter.calls)
+
+
+@pytest.mark.parametrize('limit', [0, -1, MAX_SAMPLE_ROWS + 1, '1', 1.0, True])
+def test_sampling_rejects_invalid_or_oversized_limit(limit):
+    catalog, adapter = make_catalog()
+
+    with pytest.raises(ValueError):
+        catalog.sample_relation_rows('PRODUCTS', ['ID'], limit)
+
+    assert not any('FIRST' in sql for sql, _ in adapter.calls)
+
+
+def test_sampling_returns_only_sanitized_scalar_evidence():
+    catalog, _ = make_catalog([{
+        'ID': 7,
+        'DESCRIPTION': 'widget',
+        'BLOB_DATA': b'binary',
+        'PASSWORD': 'secret',
+        'CUSTOMER_NAME': 'Customer One',
+        'NOT_SCALAR': {'nested': 'value'},
+        'OVERSIZED': 'x' * 2048,
+    }])
+
+    result = catalog.sample_relation_rows('PRODUCTS', ['ID'], 1)
+
+    assert result == [{
+        'ID': 7,
+        'DESCRIPTION': 'widget',
+        'BLOB_DATA': '<redacted>',
+        'OVERSIZED': '<redacted>',
+    }]
+
+
+def test_sampling_driver_receives_generated_internal_sql_not_operation_key():
+    catalog, adapter = make_catalog()
+
+    catalog.sample_relation_rows('PRODUCTS', ['ID'], 1)
+
+    sql, params = adapter.calls[-1]
+    assert sql == 'SELECT FIRST 1 "ID" FROM "PRODUCTS"'
+    assert params == ()
+    assert 'DROP' not in sql.upper()
+
+
+def test_catalog_identity_snapshots_are_immutable_and_detached():
+    relation_rows = [{'RELATION_NAME': 'PRODUCTS', 'RELATION_ID': 1, 'SYSTEM_FLAG': 0}]
+    field_rows = [{'RELATION_NAME': 'PRODUCTS', 'FIELD_NAME': 'ID'}]
+    adapter = RecordingAdapter([relation_rows, field_rows, []])
+    catalog = FirebirdCatalog(adapter)
+
+    relations = catalog.list_relations()
+    fields = catalog.list_relation_fields()
+    relation_rows[0]['RELATION_NAME'] = 'INJECTED'
+    field_rows[0]['FIELD_NAME'] = 'INJECTED'
+
+    with pytest.raises(TypeError):
+        relations[0]['RELATION_NAME'] = 'INJECTED'
+    with pytest.raises(AttributeError):
+        relations.append({'RELATION_NAME': 'INJECTED'})
+    with pytest.raises(TypeError):
+        fields[0]['FIELD_NAME'] = 'INJECTED'
+
+    catalog.sample_relation_rows('PRODUCTS', ['ID'], 1)
+    assert adapter.calls[-1][0] == 'SELECT FIRST 1 "ID" FROM "PRODUCTS"'
+
+
+@pytest.mark.parametrize('params', [
+    ('PRODUCTS; DROP TABLE X', ('ID',), 1),
+    ('PRODUCTS', ('ID, PASSWORD',), 1),
+    ('PRODUCTS', ('ID',), 0),
+    ('PRODUCTS', ('ID',), MAX_SAMPLE_ROWS + 1),
+    ('PRODUCTS', ('ID',), True),
+    ('PRODUCTS', ('ID',), '1'),
+])
+def test_private_sampling_executor_rejects_invalid_structured_parameters(params):
+    adapter = RecordingAdapter([])
+
+    with pytest.raises(ValueError):
+        _execute_catalog(adapter, 'sample_relation', params)
+
+    assert adapter.calls == []
+
+
+def test_private_sampling_executor_rejects_complete_sql_as_operation_input():
+    adapter = RecordingAdapter([])
+
+    with pytest.raises(ValueError):
+        _execute_catalog(adapter, 'SELECT FIRST 1 "ID" FROM "PRODUCTS"')
+
+    assert adapter.calls == []
+
+
+def test_private_sampling_executor_rejects_unknown_valid_identifiers_without_snapshot_capability():
+    adapter = RecordingAdapter([])
+
+    with pytest.raises(ValueError):
+        _execute_catalog(adapter, 'sample_relation', ('UNKNOWN', ('ID',), 1))
+
+    assert adapter.calls == []
+
+
+def test_consumer_cannot_bypass_snapshot_authorization_through_run_catalog():
+    catalog, adapter = make_catalog()
+
+    assert not hasattr(catalog, '_run_catalog')
+    with pytest.raises(AttributeError):
+        catalog._run_catalog('sample_relation', ('UNKNOWN', ('ID',), 1))
+    with pytest.raises(ValueError):
+        catalog._run_fixed_catalog('sample_relation')
+
+    assert adapter.calls == []