Merge branch 'fix/handle-null-max-chars' into 'enterprise'

fix: harden profiling/test-execution against NULL connection.max_query_chars (TG-1114)

See merge request dkinternal/testgen/dataops-testgen!536

Author: ci bot

testgen/commands/queries/execute_tests_query.py

@@ -21,7 +21,7 @@
     is_excluded_day,
     resolve_holiday_dates,
 )
-from testgen.common.models.connection import Connection
+from testgen.common.models.connection import DEFAULT_MAX_QUERY_CHARS, Connection
 from testgen.common.models.scheduler import JobSchedule
 from testgen.common.models.table_group import TableGroup
 from testgen.common.models.test_definition import TestRunType, TestScope
@@ -534,7 +534,7 @@ def aggregate_cat_tests(
                     null_value=self.null_value,
                 )
 
-        max_query_chars = self.connection.max_query_chars - 400
+        max_query_chars = (self.connection.max_query_chars or DEFAULT_MAX_QUERY_CHARS) - 400
         groups = group_cat_tests(test_defs, max_query_chars, concat_operator, single)
 
         aggregate_queries: list[tuple[str, None]] = []

testgen/commands/queries/refresh_data_chars_query.py

@@ -5,7 +5,7 @@
 from testgen.common import read_template_sql_file
 from testgen.common.database.column_chars import ColumnChars
 from testgen.common.database.database_service import get_flavor_service, replace_params
-from testgen.common.models.connection import Connection
+from testgen.common.models.connection import DEFAULT_MAX_QUERY_CHARS, Connection
 from testgen.common.models.table_group import TableGroup
 from testgen.utils import chunk_queries, to_sql_timestamp
 
@@ -133,7 +133,8 @@ def get_row_counts(self, table_names: Iterable[str]) -> list[tuple[str, None]]:
             f"SELECT '{table}' AS table_name, COUNT(*) AS row_count FROM {self.flavor_service.get_table_ref(schema, table)}"
             for table in table_names
         ]
-        chunked_queries = chunk_queries(count_queries, " UNION ALL ", self.connection.max_query_chars)
+        max_query_chars = self.connection.max_query_chars or DEFAULT_MAX_QUERY_CHARS
+        chunked_queries = chunk_queries(count_queries, " UNION ALL ", max_query_chars)
         return [ (query, None) for query in chunked_queries ]
 
     def verify_access(self, table_name: str) -> tuple[str, None]:

testgen/common/database/connection_service.py

@@ -26,7 +26,7 @@
 from sqlalchemy.exc import DatabaseError, DBAPIError
 
 from testgen.common.database.database_service import empty_cache, get_flavor_service
-from testgen.common.models.connection import Connection
+from testgen.common.models.connection import DEFAULT_MAX_QUERY_CHARS, Connection
 from testgen.ui.services.database_service import fetch_from_target_db
 
 try:
@@ -142,4 +142,4 @@ def apply_connection_defaults(connection: Connection) -> None:
     """Fill flavor-dependent defaults for fields the caller didn't supply."""
     if connection.max_query_chars is None:
         # Salesforce Data 360's Hyper engine has a lower expression-depth limit
-        connection.max_query_chars = 15000 if connection.sql_flavor_code == "salesforce_data360" else 20000
+        connection.max_query_chars = 15000 if connection.sql_flavor_code == "salesforce_data360" else DEFAULT_MAX_QUERY_CHARS

testgen/common/models/connection.py

@@ -28,6 +28,11 @@
 
 SQLFlavorCode = Literal["redshift", "redshift_spectrum", "snowflake", "mssql", "azure_mssql", "synapse_mssql", "postgresql", "databricks", "bigquery", "oracle", "sap_hana", "salesforce_data360"]
 
+# Fallback when a connection row has a NULL max_query_chars (no DB default; older
+# rows / non-UI insert paths may not seed it). Matches the UI's non-Salesforce
+# default and the 0160 migration.
+DEFAULT_MAX_QUERY_CHARS = 20000
+
 
 @dataclass
 class ConnectionMinimal(EntityMinimal):

tests/unit/commands/queries/test_execute_tests_query.py

@@ -11,6 +11,8 @@
     group_cat_tests,
     parse_cat_results,
 )
+from testgen.common.database.database_service import get_flavor_service
+from testgen.common.models.connection import Connection
 
 pytestmark = pytest.mark.unit
 
@@ -478,3 +480,18 @@ def test_resolve_cat_no_freshness_result_uses_band_check(_mock_changed):
     assert operator == "NOT BETWEEN"
 
 
+def test_aggregate_cat_tests_handles_null_max_query_chars():
+    """A connection with NULL max_query_chars must not crash CAT batching — the
+    `- 400` headroom subtraction falls back to DEFAULT_MAX_QUERY_CHARS."""
+    instance = _make_execution_sql()
+    instance.connection = Connection(sql_flavor="postgresql", max_query_chars=None)
+    instance.flavor = "postgresql"
+    instance.flavor_service = get_flavor_service("postgresql")
+
+    td = _make_td(measure_expression="m_expr", condition_expression="c_expr")
+    queries, grouped_defs = instance.aggregate_cat_tests([td], single=True)
+
+    assert len(queries) == 1
+    assert grouped_defs == [[td]]
+
+

tests/unit/commands/queries/test_refresh_data_chars_query.py

@@ -217,3 +217,16 @@ def test_filter_schema_columns_no_filters_returns_all():
     filtered = sql_generator.filter_schema_columns(columns)
 
     assert {c.table_name for c in filtered} == {"users", "orders"}
+
+
+def test_get_row_counts_handles_null_max_query_chars():
+    """A connection with NULL max_query_chars must not crash chunking — it falls
+    back to DEFAULT_MAX_QUERY_CHARS so the UNION ALL count queries still build."""
+    connection = Connection(sql_flavor="postgresql", max_query_chars=None)
+    table_group = TableGroup(table_group_schema="test_schema")
+    sql_generator = RefreshDataCharsSQL(connection, table_group)
+
+    result = sql_generator.get_row_counts(["orders", "customers"])
+
+    assert result
+    assert all(isinstance(query, str) and query for query, _ in result)