tinybird-toolset 2.5.0__tar.gz → 2.5.4__tar.gz

{tinybird_toolset-2.5.0/src/tinybird_toolset.egg-info → tinybird_toolset-2.5.4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: tinybird-toolset
-Version: 2.5.0
+Version: 2.5.4
 Home-page: https://gitlab.com/tinybird/clickhouse-toolset
 Author: Tinybird.co
 Author-email: support@tinybird.co

{tinybird_toolset-2.5.0 → tinybird_toolset-2.5.4}/setup.py

@@ -1,7 +1,21 @@
 from setuptools import setup, Extension
 
 NAME = 'tinybird-toolset'
-VERSION = '2.5.0'
+VERSION = '2.5.4'
+
+# Shared metadata for both the full (extension) build and the metadata-only
+# fallback below, so the two setup() calls can't drift apart.
+common_kwargs = dict(
+    name=NAME,
+    version=VERSION,
+    url="https://gitlab.com/tinybird/clickhouse-toolset",
+    author="Tinybird.co",
+    author_email="support@tinybird.co",
+    packages=["chtoolset"],
+    package_dir={"": "src"},
+    python_requires=">=3.10, <3.15",
+    install_requires=[],
+)
 
 try:
     from conf import *
@@ -32,15 +46,7 @@ try:
         ],
     )
     setup(
-        name=NAME,
-        version=VERSION,
-        url="https://gitlab.com/tinybird/clickhouse-toolset",
-        author="Tinybird.co",
-        author_email="support@tinybird.co",
-        packages=["chtoolset"],
-        package_dir={"": "src"},
-        python_requires=">=3.10, <3.15",
-        install_requires=[],
+        **common_kwargs,
         extras_require={"test": requirements_from_file("requirements-test.txt")},
         cmdclass={
             "clickhouse": ClickHouseBuildExt,
@@ -51,14 +57,4 @@ try:
     )
 
 except ModuleNotFoundError:
-    setup(
-        name=NAME,
-        version=VERSION,
-        url="https://gitlab.com/tinybird/clickhouse-toolset",
-        author="Tinybird.co",
-        author_email="support@tinybird.co",
-        packages=["chtoolset"],
-        package_dir={"": "src"},
-        python_requires=">=3.10, <3.15",
-        install_requires=[],
-    )
+    setup(**common_kwargs)

{tinybird_toolset-2.5.0 → tinybird_toolset-2.5.4/src/tinybird_toolset.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: tinybird-toolset
-Version: 2.5.0
+Version: 2.5.4
 Home-page: https://gitlab.com/tinybird/clickhouse-toolset
 Author: Tinybird.co
 Author-email: support@tinybird.co

{tinybird_toolset-2.5.0 → tinybird_toolset-2.5.4}/tests/test_convert_to_row_binary.py

@@ -8467,6 +8467,70 @@ TEST_CASES = [
 
 
 class TestRowBinaryEncoder:
+    @pytest.mark.parametrize("legacy_conversion_mode", [True, False])
+    def test_convert_ipv4_to_row_binary(self, legacy_conversion_mode):
+        schema = json.dumps(get_schema([('address', 'IPv4', '$.address', False)]))
+
+        with chquery.RowBinaryEncoder(schema, legacy_conversion_mode=legacy_conversion_mode) as encoder:
+            result, quarantine, nrows, nquarantine = encoder.encode('{"address": "192.0.2.1"}')
+
+        assert result == b'\x00\x01\x02\x00\xc0'
+        assert quarantine == b''
+        assert nrows == 1
+        assert nquarantine == 0
+
+    @pytest.mark.parametrize("legacy_conversion_mode", [True, False])
+    def test_convert_ipv6_to_row_binary(self, legacy_conversion_mode):
+        schema = json.dumps(get_schema([('address', 'IPv6', '$.address', False)]))
+
+        with chquery.RowBinaryEncoder(schema, legacy_conversion_mode=legacy_conversion_mode) as encoder:
+            result, quarantine, nrows, nquarantine = encoder.encode('{"address": "2001:db8::1"}')
+
+        assert result == b'\x00\x20\x01\x0d\xb8' + b'\x00' * 11 + b'\x01'
+        assert quarantine == b''
+        assert nrows == 1
+        assert nquarantine == 0
+
+    @pytest.mark.parametrize("legacy_conversion_mode", [True, False])
+    def test_convert_ipv4_formatted_address_to_ipv6_row_binary(self, legacy_conversion_mode):
+        schema = json.dumps(get_schema([('address', 'IPv6', '$.address', False)]))
+
+        with chquery.RowBinaryEncoder(schema, legacy_conversion_mode=legacy_conversion_mode) as encoder:
+            result, quarantine, nrows, nquarantine = encoder.encode('{"address": "1.1.1.1"}')
+
+        assert result == b'\x00' + b'\x00' * 10 + b'\xff\xff\x01\x01\x01\x01'
+        assert quarantine == b''
+        assert nrows == 1
+        assert nquarantine == 0
+
+    @pytest.mark.parametrize("legacy_conversion_mode", [True, False])
+    @pytest.mark.parametrize("address_type", ["IPv4", "IPv6"])
+    def test_invalid_ip_address_is_quarantined(self, legacy_conversion_mode, address_type):
+        schema = json.dumps(get_schema([('address', address_type, '$.address', False)]))
+
+        with chquery.RowBinaryEncoder(schema, legacy_conversion_mode=legacy_conversion_mode) as encoder:
+            result, quarantine, nrows, nquarantine = encoder.encode('{"address": "invalid-ip"}')
+
+        assert result == b''
+        assert quarantine != b''
+        assert b"invalid-ip" in quarantine
+        assert nrows == 0
+        assert nquarantine == 1
+
+    @pytest.mark.parametrize("legacy_conversion_mode", [True, False])
+    @pytest.mark.parametrize("address", ["1.1.1.1junk", "2001:db8::1junk"])
+    def test_ipv6_with_trailing_junk_is_quarantined(self, legacy_conversion_mode, address):
+        schema = json.dumps(get_schema([('address', 'IPv6', '$.address', False)]))
+
+        with chquery.RowBinaryEncoder(schema, legacy_conversion_mode=legacy_conversion_mode) as encoder:
+            result, quarantine, nrows, nquarantine = encoder.encode(json.dumps({"address": address}))
+
+        assert result == b''
+        assert quarantine != b''
+        assert address.encode() in quarantine
+        assert nrows == 0
+        assert nquarantine == 1
+
     @pytest.mark.parametrize("conversion_mode, schema, data, expected_result, expected_quarantine, expected_nrows, expected_nquarantine", TEST_CASES)
     def test_legacy_convert_to_row_binary(self, conversion_mode, schema, data, expected_result, expected_quarantine, expected_nrows, expected_nquarantine):
         schema_rowbinary = get_schema(schema)

{tinybird_toolset-2.5.0 → tinybird_toolset-2.5.4}/tests/test_tables.py

@@ -2033,6 +2033,7 @@ SELECT *
             optimize_aggregation_in_order = 1,
             use_skip_indexes = 1,
             use_skip_indexes_if_final = 0,
+            do_not_merge_across_partitions_select_final = 1,
             optimize_move_to_prewhere = 1,
             query_plan_optimize_lazy_materialization = 0,
             min_bytes_to_use_direct_io = 0,

{tinybird_toolset-2.5.0 → tinybird_toolset-2.5.4}/tests/test_validate_ttl.py

@@ -333,6 +333,48 @@ class TestValidateTTL(unittest.TestCase):
                     f"{name} should be a known function in TTL WHERE clauses",
                 )
 
+    # --- Hash functions in sorting / primary keys (KeyDescription::parse) ---
+    def test_valid_sorting_key_with_cityhash64(self):
+        self.assertIsNone(
+            chquery.validate_ttl(
+                ttl="timestamp + INTERVAL 365 DAY",
+                columns=[
+                    {"name": "organizationId", "type": "String"},
+                    {"name": "surfaceId", "type": "String"},
+                    {"name": "timestamp", "type": "DateTime"},
+                    {"name": "name", "type": "String"},
+                    {"name": "traceId", "type": "String"},
+                    {"name": "spanId", "type": "String"},
+                    {"name": "eventIndex", "type": "UInt32"},
+                ],
+                sorting_key="organizationId, surfaceId, toDate(timestamp), name, cityHash64(traceId), spanId, eventIndex",
+            )
+        )
+
+    # --- Arithmetic functions in TTL expressions ---
+    def test_valid_ttl_with_int_div(self):
+        # Reported failure: "Unknown function intDiv".
+        self.assertIsNone(
+            chquery.validate_ttl(
+                ttl="toDateTime(intDiv(ts_server, 1000)) + INTERVAL 13 MONTH",
+                columns=[{"name": "ts_server", "type": "Int64"}],
+                sorting_key="ts_server",
+            )
+        )
+
+    def test_valid_ttl_with_greatest(self):
+        # Reported failure: "Unknown function greatest".
+        self.assertIsNone(
+            chquery.validate_ttl(
+                ttl="greatest(departure_date, toDateTime(creation_date)) + toIntervalDay(2)",
+                columns=[
+                    {"name": "departure_date", "type": "DateTime"},
+                    {"name": "creation_date", "type": "Date"},
+                ],
+                sorting_key="departure_date",
+            )
+        )
+
     # --- Input validation ---
     def test_columns_must_be_list_of_dicts(self):
         with self.assertRaises(TypeError):