119 lines
5.1 KiB
Python
119 lines
5.1 KiB
Python
# #region test_sql_table_extractor [C:3] [TYPE TestModule] [SEMANTICS test, sql, table, extractor, pytest]
|
|
# @BRIEF Unit tests for SqlTableExtractor — T017. Verifies three-phase extraction of schema.table references from SQL+Jinja.
|
|
# @LAYER Tests
|
|
# @RELATION BINDS_TO -> [SqlTableExtractorModule]
|
|
# @TEST_CONTRACT: extract_tables_from_sql(str) -> set[str]
|
|
# @TEST_EDGE: empty_input -> returns empty set
|
|
# @TEST_EDGE: no_schema_qualified_tables -> returns empty set
|
|
# @TEST_EDGE: string_literal_false_positive -> filtered out by sqlparse
|
|
# @TEST_FIXTURE: simple_select -> INLINE_SQL: "SELECT * FROM raw.sales"
|
|
# @TEST_FIXTURE: jinja_expr -> INLINE_SQL: "SELECT * FROM {{ source('raw', 'sales') }}"
|
|
import pytest
|
|
|
|
from src.services.sql_table_extractor import extract_tables_from_sql
|
|
|
|
|
|
class TestExtractTablesFromSql:
|
|
"""T017: Unit tests for SqlTableExtractor.extract_tables_from_sql()."""
|
|
|
|
# #region test_simple_select [C:2] [TYPE Function]
|
|
# @BRIEF Happy path: simple SELECT with FROM schema.table.
|
|
def test_simple_select(self):
|
|
sql = "SELECT * FROM raw.sales"
|
|
result = extract_tables_from_sql(sql)
|
|
assert result == {"raw.sales"}
|
|
# #endregion test_simple_select
|
|
|
|
# #region test_multiple_tables [C:2] [TYPE Function]
|
|
# @BRIEF JOIN with multiple schema-qualified tables.
|
|
def test_multiple_tables(self):
|
|
sql = "SELECT a.id, b.name FROM raw.sales a JOIN raw.inventory b ON a.id = b.sale_id"
|
|
result = extract_tables_from_sql(sql)
|
|
assert result == {"raw.sales", "raw.inventory"}
|
|
# #endregion test_multiple_tables
|
|
|
|
# #region test_empty_input [C:2] [TYPE Function]
|
|
# @BRIEF Empty string returns empty set.
|
|
def test_empty_input(self):
|
|
assert extract_tables_from_sql("") == set()
|
|
assert extract_tables_from_sql(" ") == set()
|
|
assert extract_tables_from_sql(None) == set()
|
|
# #endregion test_empty_input
|
|
|
|
# #region test_no_schema_qualified_tables [C:2] [TYPE Function]
|
|
# @BRIEF Unqualified table names are NOT matched.
|
|
def test_no_schema_qualified_tables(self):
|
|
sql = "SELECT * FROM sales"
|
|
result = extract_tables_from_sql(sql)
|
|
assert result == set()
|
|
# #endregion test_no_schema_qualified_tables
|
|
|
|
# #region test_case_insensitive [C:2] [TYPE Function]
|
|
# @BRIEF Matching is case-insensitive; result is lowercased.
|
|
def test_case_insensitive(self):
|
|
sql = "SELECT * FROM RAW.Sales JOIN Raw.Inventory ON ..."
|
|
result = extract_tables_from_sql(sql)
|
|
assert result == {"raw.sales", "raw.inventory"}
|
|
# #endregion test_case_insensitive
|
|
|
|
# #region test_string_literal_false_positive [C:2] [TYPE Function]
|
|
# @BRIEF Date literal like '2026.04.30' must NOT be matched as schema.table.
|
|
def test_string_literal_false_positive(self):
|
|
sql = "SELECT * FROM raw.sales WHERE date = '2026.04.30'"
|
|
result = extract_tables_from_sql(sql)
|
|
assert result == {"raw.sales"}
|
|
# #endregion test_string_literal_false_positive
|
|
|
|
# #region test_jinja_string_concat [C:2] [TYPE Function]
|
|
# @BRIEF Table reference inside Jinja {{ 'raw.sales' }} as a direct string.
|
|
def test_jinja_string_concat(self):
|
|
sql = "SELECT * FROM {{ 'raw.sales' }}"
|
|
result = extract_tables_from_sql(sql)
|
|
assert result == {"raw.sales"}
|
|
# #endregion test_jinja_string_concat
|
|
|
|
# #region test_jinja_set_block [C:2] [TYPE Function]
|
|
# @BRIEF Table name in Jinja {% set %} block string value.
|
|
def test_jinja_set_block(self):
|
|
sql = """
|
|
{% set tables = ['raw.sales', 'raw.inventory'] %}
|
|
SELECT * FROM raw.sales
|
|
"""
|
|
result = extract_tables_from_sql(sql)
|
|
assert result == {"raw.sales", "raw.inventory"}
|
|
# #endregion test_jinja_set_block
|
|
|
|
# #region test_mixed_jinja_and_sql [C:2] [TYPE Function]
|
|
# @BRIEF Mixed Jinja string and plain SQL schema.table references.
|
|
def test_mixed_jinja_and_sql(self):
|
|
sql = """
|
|
{% set source_table = 'raw.orders' %}
|
|
SELECT * FROM raw.orders o
|
|
JOIN raw.products p ON o.id = p.product_id
|
|
"""
|
|
result = extract_tables_from_sql(sql)
|
|
assert result == {"raw.orders", "raw.products"}
|
|
# #endregion test_mixed_jinja_and_sql
|
|
|
|
# #region test_cte_with_schema_tables [C:2] [TYPE Function]
|
|
# @BRIEF CTE body with schema-qualified table references.
|
|
def test_cte_with_schema_tables(self):
|
|
sql = """
|
|
WITH active_sales AS (
|
|
SELECT * FROM raw.sales WHERE amount > 100
|
|
)
|
|
SELECT * FROM active_sales
|
|
"""
|
|
result = extract_tables_from_sql(sql)
|
|
assert result == {"raw.sales"}
|
|
# #endregion test_cte_with_schema_tables
|
|
|
|
# #region test_jinja_comment_excluded [C:2] [TYPE Function]
|
|
# @BRIEF Jinja comment {# ... #} content should not produce table extraction (no string patterns inside comments).
|
|
def test_jinja_comment_excluded(self):
|
|
sql = "SELECT * FROM raw.sales {# this references raw.archived #}"
|
|
result = extract_tables_from_sql(sql)
|
|
assert result == {"raw.sales"}
|
|
# #endregion test_jinja_comment_excluded
|
|
# #endregion test_sql_table_extractor
|