| 1 | from pathlib import Path |
| 2 | |
| 3 | WORKFLOW = Path(".github/workflows/sync-publish-to-main.yml") |
| 4 | RESTORE_WORKFLOW = Path(".github/workflows/restore-publish-backup.yml") |
| 5 | CRAWL_WORKFLOW = Path(".github/workflows/crawl-and-publish.yml") |
| 6 | |
| 7 | |
| 8 | def test_publish_sync_only_checks_out_generated_content_paths() -> None: |
| 9 | workflow = WORKFLOW.read_text(encoding="utf-8") |
| 10 | |
| 11 | for path in ( |
| 12 | "data/raw/", |
| 13 | "data/analyzed/", |
| 14 | "data/metrics/", |
| 15 | "content/weekly/", |
| 16 | "content/monthly/", |
| 17 | "content/yearly/", |
| 18 | ): |
| 19 | assert path in workflow |
| 20 | |
| 21 | assert "git checkout origin/publish -- .squad" not in workflow |
| 22 | assert "git ls-tree -r --name-only origin/publish -- .squad" not in workflow |
| 23 | assert "squad learnings" not in workflow.lower() |
| 24 | assert "python3 scripts/generate_rollups.py" in workflow |
| 25 | assert "data/raw-store/" not in workflow |
| 26 | |
| 27 | |
| 28 | def test_publish_sync_refuses_staged_squad_changes() -> None: |
| 29 | workflow = WORKFLOW.read_text(encoding="utf-8") |
| 30 | |
| 31 | assert "Refusing to sync .squad state from publish to main" in workflow |
| 32 | assert "git diff --cached --name-only | grep -E '^\\.squad/'" in workflow |
| 33 | assert "data/raw/" in workflow |
| 34 | assert ".squad/**" in workflow |
| 35 | |
| 36 | |
| 37 | def test_restore_publish_backup_workflow_uses_immutable_backup_manifest() -> None: |
| 38 | workflow = RESTORE_WORKFLOW.read_text(encoding="utf-8") |
| 39 | |
| 40 | assert "backup_manifest" in workflow |
| 41 | assert "python3 ../workflow-source/scripts/publish_safety.py restore-backup" in workflow |
| 42 | assert "--force-with-lease" in workflow |
| 43 | assert "ref: publish" in workflow |
| 44 | |
| 45 | |
| 46 | def test_restore_publish_backup_workflow_keeps_helper_available_after_publish_checkout() -> None: |
| 47 | workflow = RESTORE_WORKFLOW.read_text(encoding="utf-8") |
| 48 | |
| 49 | assert "path: workflow-source" in workflow |
| 50 | assert "ref: ${{ github.sha }}" in workflow |
| 51 | assert "path: publish" in workflow |
| 52 | assert "cd publish" in workflow |
| 53 | assert "python3 scripts/publish_safety.py restore-backup" not in workflow |
| 54 | assert "python3 ../workflow-source/scripts/publish_safety.py restore-backup" in workflow |
| 55 | |
| 56 | |
| 57 | def test_publish_sync_stages_before_cached_diff_evaluation() -> None: |
| 58 | workflow = WORKFLOW.read_text(encoding="utf-8") |
| 59 | |
| 60 | assert workflow.index("git add -A") < workflow.index("if git diff --cached --quiet; then") |
| 61 | |
| 62 | |
| 63 | def test_crawl_workflow_stores_and_restores_source_bound_raw_evidence() -> None: |
| 64 | workflow = CRAWL_WORKFLOW.read_text(encoding="utf-8") |
| 65 | |
| 66 | assert "source_run_id:" in workflow |
| 67 | assert "run_mode=restore and source_run_id" in workflow |
| 68 | assert workflow.count("python3 scripts/publish_safety.py restore-raw") == 2 |
| 69 | assert "python3 publish-safety-tool.py store-raw" in workflow |
| 70 | assert 'RAW_STORE_DIR="data/raw-store/${REBUILD_WEEK}/${SOURCE_RUN_ID}"' in workflow |
| 71 | assert ( |
| 72 | '--raw-store-manifest "data/raw-store/${WEEK}/${SOURCE_RUN_ID}/manifest.json"' in workflow |
| 73 | ) |
| 74 | assert 'source-artifact-id "$RAW_ARTIFACT_ID"' in workflow |
| 75 | assert "retention-days: 90" in workflow |
| 76 | |
| 77 | |
| 78 | def test_crawl_workflow_stages_raw_store_before_cached_diff_evaluation() -> None: |
| 79 | workflow = CRAWL_WORKFLOW.read_text(encoding="utf-8") |
| 80 | |
| 81 | stage = workflow.index("git add data/raw/ data/snapshots/ data/raw-store/") |
| 82 | cached_diff = workflow.index("if git diff --cached --quiet; then", stage) |
| 83 | assert stage < cached_diff |