test_database_migrations.py 5.7 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179
  1. from __future__ import annotations
  2. import os
  3. import subprocess
  4. import uuid
  5. from pathlib import Path
  6. import psycopg2
  7. import pytest
  8. from sqlalchemy import create_engine, inspect
  9. from sqlalchemy.engine import make_url
  10. ROOT = Path(__file__).resolve().parents[1]
  11. EXPECTED_BASELINE_TABLES = {
  12. "data_orders",
  13. "data_products",
  14. "metadata_review_records",
  15. "metadata_version_history",
  16. "task_list",
  17. "users",
  18. }
  19. EXPECTED_UPGRADED_TABLES = {
  20. "datasource_credentials",
  21. "datasource_credential_audit_events",
  22. "workflow_schedules",
  23. "workflow_runs",
  24. "workflow_task_runs",
  25. "workflow_engine_bindings",
  26. "workflow_plan_audits",
  27. "runner_task_executions",
  28. "workflow_canary_evidence",
  29. "workflow_gateway_operations",
  30. "workflow_migration_states",
  31. "workflow_dual_runs",
  32. "workflow_reconciliation_reports",
  33. "workflow_cutover_operations",
  34. "ingestion_sources",
  35. "source_artifacts",
  36. "ingestion_jobs",
  37. "evidence_fragments",
  38. "extraction_candidates",
  39. "data_elements",
  40. "data_element_versions",
  41. "candidate_decisions",
  42. }
  43. def test_data_research_ingestion_migration_is_additive_and_constrained():
  44. migration = (
  45. ROOT
  46. / "migrations"
  47. / "versions"
  48. / "20260722_100_data_research_ingestion.py"
  49. ).read_text(encoding="utf-8")
  50. assert 'revision = "20260722_100"' in migration
  51. assert 'down_revision = "20260719_90"' in migration
  52. for table in (
  53. "ingestion_sources",
  54. "source_artifacts",
  55. "ingestion_jobs",
  56. "evidence_fragments",
  57. "extraction_candidates",
  58. ):
  59. assert f"CREATE TABLE IF NOT EXISTS public.{table}" in migration
  60. assert "idempotency_key" in migration
  61. assert "parser_version" in migration
  62. assert "locator JSONB" in migration
  63. assert "DROP TABLE" not in migration.upper()
  64. def test_data_element_migration_adds_versioned_governance_tables():
  65. migration = (
  66. ROOT
  67. / "migrations"
  68. / "versions"
  69. / "20260722_105_data_research_elements.py"
  70. ).read_text(encoding="utf-8")
  71. assert 'revision = "20260722_105"' in migration
  72. assert 'down_revision = "20260722_100"' in migration
  73. for table in ("data_elements", "data_element_versions", "candidate_decisions"):
  74. assert f"CREATE TABLE IF NOT EXISTS public.{table}" in migration
  75. assert "UNIQUE (data_element_uid, version)" in migration
  76. assert "DROP TABLE" not in migration.upper()
  77. def test_ontology_migration_adds_versioned_control_plane_tables():
  78. migration = (
  79. ROOT / "migrations" / "versions" / "20260722_110_data_research_ontology.py"
  80. ).read_text(encoding="utf-8")
  81. assert 'revision = "20260722_110"' in migration
  82. assert 'down_revision = "20260722_105"' in migration
  83. for table in (
  84. "ontologies",
  85. "ontology_versions",
  86. "ontology_domain_links",
  87. "ontology_change_sets",
  88. "ontology_publish_runs",
  89. ):
  90. assert f"CREATE TABLE IF NOT EXISTS public.{table}" in migration
  91. assert "UNIQUE (ontology_uid, version)" in migration
  92. assert "DROP TABLE" not in migration.upper()
  93. def test_alembic_configuration_is_environment_only():
  94. ini = (ROOT / "alembic.ini").read_text(encoding="utf-8")
  95. env = (ROOT / "migrations" / "env.py").read_text(encoding="utf-8")
  96. assert "sqlalchemy.url" not in ini
  97. assert "SQLALCHEMY_DATABASE_URI" in env
  98. assert "DATABASE_URL" in env
  99. assert "password" not in env.lower()
  100. def test_baseline_migration_is_non_destructive():
  101. baseline = (
  102. ROOT / "migrations" / "versions" / "20260716_01_baseline.py"
  103. ).read_text(encoding="utf-8")
  104. assert "def upgrade" in baseline
  105. assert "def downgrade" in baseline
  106. assert "drop_table" not in baseline
  107. for table in EXPECTED_BASELINE_TABLES:
  108. assert table in baseline
  109. @pytest.mark.integration
  110. def test_alembic_upgrade_is_repeatable_and_downgrade_preserves_tables():
  111. admin_url = os.environ.get("TEST_POSTGRES_ADMIN_URL")
  112. if not admin_url:
  113. pytest.skip("TEST_POSTGRES_ADMIN_URL is not configured")
  114. parsed = make_url(admin_url)
  115. database_name = f"dataops_migration_{uuid.uuid4().hex[:12]}"
  116. target_url = parsed.set(database=database_name).render_as_string(
  117. hide_password=False
  118. )
  119. connection = psycopg2.connect(admin_url)
  120. connection.autocommit = True
  121. try:
  122. with connection.cursor() as cursor:
  123. cursor.execute(f'CREATE DATABASE "{database_name}"')
  124. env = os.environ.copy()
  125. env["SQLALCHEMY_DATABASE_URI"] = target_url
  126. command = [str(ROOT / ".venv" / "bin" / "alembic"), "-c", "alembic.ini"]
  127. subprocess.run(command + ["upgrade", "head"], cwd=ROOT, env=env, check=True)
  128. subprocess.run(command + ["upgrade", "head"], cwd=ROOT, env=env, check=True)
  129. engine = create_engine(target_url)
  130. try:
  131. tables = set(inspect(engine).get_table_names(schema="public"))
  132. assert EXPECTED_BASELINE_TABLES | {"alembic_version"} <= tables
  133. assert EXPECTED_UPGRADED_TABLES <= tables
  134. finally:
  135. engine.dispose()
  136. subprocess.run(command + ["downgrade", "-1"], cwd=ROOT, env=env, check=True)
  137. engine = create_engine(target_url)
  138. try:
  139. tables = set(inspect(engine).get_table_names(schema="public"))
  140. assert EXPECTED_BASELINE_TABLES <= tables
  141. finally:
  142. engine.dispose()
  143. finally:
  144. with connection.cursor() as cursor:
  145. cursor.execute(
  146. "SELECT pg_terminate_backend(pid) FROM pg_stat_activity "
  147. "WHERE datname = %s AND pid <> pg_backend_pid()",
  148. (database_name,),
  149. )
  150. cursor.execute(f'DROP DATABASE IF EXISTS "{database_name}"')
  151. connection.close()