add additional promotion tests

Author: CarsonDavis

sde_collections/tests/test_promote_collection.py

@@ -1,12 +1,15 @@
 # docker-compose -f local.yml run --rm django pytest sde_collections/tests/test_promote_collection.py
 import pytest
 
+from sde_collections.models.collection_choice_fields import Divisions, DocumentTypes
 from sde_collections.models.delta_patterns import (
+    DeltaDivisionPattern,
+    DeltaDocumentTypePattern,
     DeltaExcludePattern,
     DeltaIncludePattern,
     DeltaTitlePattern,
 )
-from sde_collections.models.delta_url import CuratedUrl, DeltaUrl
+from sde_collections.models.delta_url import CuratedUrl, DeltaUrl, DumpUrl
 from sde_collections.tests.factories import CollectionFactory
 
 
@@ -208,3 +211,182 @@ def test_promotion_with_title_change():
 
     # This should trigger the same error we're seeing in production
     collection.promote_to_curated()
+
+
+@pytest.mark.django_db
+def test_promotion_maintains_pattern_relationships_through_updates(collection):
+    """Test that pattern relationships survive multiple promotions with updates"""
+    # Initial setup
+    DeltaUrl.objects.create(collection=collection, url="https://example.com", scraped_title="Title")
+    pattern = DeltaTitlePattern.objects.create(
+        collection=collection, match_pattern="example.com", match_pattern_type=1, title_pattern="Pattern: {title}"
+    )
+
+    collection.promote_to_curated()
+
+    # Record initial state
+    curated = CuratedUrl.objects.get(url="https://example.com")
+    initial_id = curated.id
+    initial_pattern_relations = list(pattern.curated_urls.all())
+
+    # Create new delta with changes
+    DeltaUrl.objects.create(collection=collection, url="https://example.com", scraped_title="New Title")
+    collection.promote_to_curated()
+
+    # Verify relationships maintained
+    curated.refresh_from_db()
+    assert curated.id == initial_id  # ID should not change
+    assert list(pattern.curated_urls.all()) == initial_pattern_relations
+
+
+@pytest.mark.django_db
+def test_sequential_promotions_with_multiple_patterns(collection):
+    """Test complex scenario with multiple promotions and pattern changes"""
+    # Initial setup with two URLs
+    urls = ["https://example.com/doc", "https://example.com/guide"]
+    for url in urls:
+        DeltaUrl.objects.create(collection=collection, url=url, scraped_title=f"Title for {url}")
+
+    # First pattern and promotion
+    pattern1 = DeltaTitlePattern.objects.create(
+        collection=collection, match_pattern="doc", match_pattern_type=2, title_pattern="Doc: {title}"
+    )
+
+    collection.promote_to_curated()
+
+    # Record state after first promotion
+    initial_ids = {url: CuratedUrl.objects.get(url=url).id for url in urls}
+
+    DeltaUrl.objects.create(collection=collection, url="https://example.com/guide", scraped_title="Updated guide")
+    collection.promote_to_curated()
+
+    pattern2 = DeltaTitlePattern.objects.create(
+        collection=collection, match_pattern="guide", match_pattern_type=2, title_pattern="Guide: {title}"
+    )
+    assert not pattern2.curated_urls.filter(url__contains="guide").exists()
+
+    # Verify state
+    for url in urls:
+        curated = CuratedUrl.objects.get(url=url)
+        assert curated.id == initial_ids[url]  # IDs should be preserved
+
+    collection.promote_to_curated()
+
+    # Verify pattern relationships
+    assert pattern1.curated_urls.filter(url__contains="doc").exists()
+    assert pattern2.curated_urls.filter(url__contains="guide").exists()
+
+
+@pytest.mark.django_db
+def test_promotion_with_division_changes(collection):
+    """Test that division patterns are correctly promoted and applied"""
+    # Initial setup
+    DeltaUrl.objects.create(collection=collection, url="https://example.com/astrophysics", division=Divisions.GENERAL)
+    DeltaUrl.objects.create(collection=collection, url="https://example.com/helio", division=Divisions.GENERAL)
+
+    # Create and apply division patterns
+    DeltaDivisionPattern.objects.create(
+        collection=collection,
+        match_pattern="astrophysics",
+        match_pattern_type=2,  # Multi-URL pattern
+        division=Divisions.ASTROPHYSICS,
+    )
+
+    DeltaDivisionPattern.objects.create(
+        collection=collection, match_pattern="helio", match_pattern_type=2, division=Divisions.HELIOPHYSICS
+    )
+
+    # Promote and verify divisions were set
+    collection.promote_to_curated()
+
+    assert CuratedUrl.objects.get(url__contains="astrophysics").division == Divisions.ASTROPHYSICS
+    assert CuratedUrl.objects.get(url__contains="helio").division == Divisions.HELIOPHYSICS
+
+
+@pytest.mark.django_db
+def test_promotion_with_document_type_changes(collection):
+    """Test document type patterns through promotion"""
+    # Create URLs with default doc type
+    DeltaUrl.objects.create(
+        collection=collection, url="https://example.com/data/set1", document_type=DocumentTypes.DOCUMENTATION
+    )
+    DeltaUrl.objects.create(
+        collection=collection, url="https://example.com/tools/tool1", document_type=DocumentTypes.DOCUMENTATION
+    )
+
+    # Set up patterns for different doc types
+    DeltaDocumentTypePattern.objects.create(
+        collection=collection, match_pattern="data/*", match_pattern_type=2, document_type=DocumentTypes.DATA
+    ).apply()
+
+    DeltaDocumentTypePattern.objects.create(
+        collection=collection, match_pattern="tools/*", match_pattern_type=2, document_type=DocumentTypes.SOFTWARETOOLS
+    ).apply()
+
+    collection.promote_to_curated()
+
+    # Verify document types were correctly set
+    assert CuratedUrl.objects.get(url__contains="/data/").document_type == DocumentTypes.DATA
+    assert CuratedUrl.objects.get(url__contains="/tools/").document_type == DocumentTypes.SOFTWARETOOLS
+
+
+@pytest.mark.django_db
+def test_promotion_with_multiple_metadata_changes_dump(collection):
+    """Test complex scenario with multiple metadata changes through multiple promotions"""
+    # Initial URL we'll be working with
+    url = "https://example.com/helio/data"
+
+    # Create initial DumpUrl
+    DumpUrl.objects.create(
+        collection=collection,
+        url=url,
+        division=Divisions.GENERAL,
+        document_type=DocumentTypes.DOCUMENTATION,
+        scraped_title="Raw Data Title",
+    )
+
+    # Migrate DumpUrls to DeltaUrls
+    collection.migrate_dump_to_delta()
+
+    # Create patterns that will affect this URL
+    DeltaDivisionPattern.objects.create(
+        collection=collection, match_pattern="*helio*", match_pattern_type=2, division=Divisions.HELIOPHYSICS
+    )
+
+    DeltaDocumentTypePattern.objects.create(
+        collection=collection, match_pattern="*data*", match_pattern_type=2, document_type=DocumentTypes.DATA
+    )
+
+    DeltaTitlePattern.objects.create(
+        collection=collection, match_pattern="*data*", match_pattern_type=2, title_pattern="Heliophysics Data: {title}"
+    )
+
+    # First promotion
+    collection.promote_to_curated()
+
+    # Verify initial promotion worked correctly
+    curated = CuratedUrl.objects.get(url=url)
+    assert curated.division == Divisions.HELIOPHYSICS
+    assert curated.document_type == DocumentTypes.DATA
+    assert curated.generated_title == "Heliophysics Data: Raw Data Title"
+
+    # Create new DumpUrl with updated data to simulate a new crawl
+    DumpUrl.objects.create(
+        collection=collection,
+        url=url,
+        division=Divisions.GENERAL,  # These will be overridden by patterns
+        document_type=DocumentTypes.DOCUMENTATION,  # These will be overridden by patterns
+        scraped_title="Updated Data Title",
+    )
+
+    # Migrate new dump to delta
+    collection.migrate_dump_to_delta()
+
+    # Second promotion - should maintain pattern-applied metadata while updating the title
+    collection.promote_to_curated()
+
+    # Verify final state
+    curated = CuratedUrl.objects.get(url=url)
+    assert curated.division == Divisions.HELIOPHYSICS  # Should still be preserved from pattern
+    assert curated.document_type == DocumentTypes.DATA  # Should still be preserved from pattern
+    assert curated.generated_title == "Heliophysics Data: Updated Data Title"  # Should reflect new title