Merge pull request #1146 from NASA-IMPACT/1142-test-field-modified-unapply-logic

refactor readme for unapply logic and refactor unapply to account for overlapping patterns

Author: CarsonDavis

sde_collections/models/README_UNAPPLY_LOGIC.md

@@ -14,56 +14,105 @@
 - Delta URL exists with pattern effect
 - Pattern is removed
 ```
-Curated: None
-Delta: division=BIOLOGY (from pattern)
-[Pattern removed]
-Result: Delta remains with division=None
+Curated: None exists
+Delta: url=new.com, division=None
+```
+`[Pattern: division=BIOLOGY], created`
+```
+Curated: None exists
+Delta: url=new.com, division=BIOLOGY
+```
+`[Pattern: division=BIOLOGY], deleted`
+```
+Curated: None exists
+Delta: url=new.com, division=None
 ```
 
-### Case 2: Delta and Curated Exist
+### Case 2: Delta Created to Apply Pattern
 **Scenario:**
-- Both curated and delta URLs exist
+- A Curated with no division already exists
+- A pattern is created
+- A delta is created to  to apply a pattern
 - Pattern is removed
+- Delta should be deleted
+```
+Curated: division=None
 ```
-Curated: division=GENERAL
+`[Pattern: division=BIOLOGY], created`
+```
+Curated: division=None
 Delta: division=BIOLOGY (from pattern)
-[Pattern removed]
-Result: Delta reverts to curated value (division=GENERAL)
-If delta now matches curated exactly, delta is deleted
+```
+`[Pattern: division=BIOLOGY], deleted`
+```
+Curated: division=None
 ```
 
-### Case 3: Curated Only
-**Scenario:**
-- Only curated URL exists
+### Case 3: Pre-existing Delta
+- A Curated with no division already exists
+- A Delta with an updated scraped_title exists
+- A pattern is created to set division
+- A delta is created to apply a pattern
 - Pattern is removed
+- Delta should be maintained because of scraped_title
+
+```
+Curated: division=None
+Delta: scraped_title="Modified", division=None
+```
+`[Pattern: division=BIOLOGY], created`
+```
+Curated: division=None
+Delta: scraped_title="Modified", division=BIOLOGY (from pattern)
+```
+`[Pattern: division=BIOLOGY], deleted`
 ```
-Curated: division=GENERAL
-Delta: None
-[Pattern removed]
-Result: New delta created with division=None
+Curated: division=None
+Delta: scraped_title="Modified", division=None
 ```
 
 ### Case 4: Multiple Pattern Effects
 **Scenario:**
 - Delta has changes from multiple patterns
 - One pattern is removed
 ```
-Curated: division=GENERAL, doc_type=DOCUMENTATION
 Delta: division=BIOLOGY, doc_type=DATA (from two patterns)
-[Division pattern removed]
-Result: Delta remains with division=GENERAL, doc_type=DATA preserved
+Pattern: division=BIOLOGY
+Pattern: doc_type=DATA
+```
+`[Pattern: division=BIOLOGY], deleted`
+```
+Delta: division=None, doc_type=DATA
+Pattern: doc_type=DATA
 ```
 
-### Case 5: Pattern Removal with Manual Changes
-**Scenario:**
-- Delta has both pattern effect and manual changes
-- Pattern is removed
+### Case 5: Overlapping Patterns, Specific Deleted
 ```
-Curated: division=GENERAL, title="Original"
-Delta: division=BIOLOGY, title="Modified" (pattern + manual)
-[Pattern removed]
-Result: Delta remains with division=GENERAL, title="Modified" preserved
+Delta: division=ASTROPHYSICS (because of specific pattern)
+Specific Pattern: division=ASTROPHYSICS
+General Pattern: division=BIOLOGY
 ```
+`[Specific Pattern: division=ASTROPHYSICS], deleted`
+
+```
+Delta: division=BIOLOGY (because of general pattern)
+General Pattern: division=BIOLOGY
+```
+
+
+### Case 6: Overlapping Patterns, General Deleted
+```
+Delta: division=ASTROPHYSICS (because of specific pattern)
+Specific Pattern: division=ASTROPHYSICS
+General Pattern: division=BIOLOGY
+```
+`[General Pattern: division=BIOLOGY], deleted`
+
+```
+Delta: division=ASTROPHYSICS (because of specific pattern)
+Specific Pattern: division=ASTROPHYSICS
+```
+
 
 ## Implementation Steps
 

sde_collections/models/collection.py

@@ -211,18 +211,18 @@ def promote_to_curated(self):
                         continue
 
                     delta_value = getattr(delta, field_name)
-                    if delta_value not in [None, ""] and getattr(curated, field_name) != delta_value:
+                    if getattr(curated, field_name) != delta_value:
                         updated_fields[field_name] = delta_value
 
                 if updated_fields:
                     CuratedUrl.objects.filter(pk=curated.pk).update(**updated_fields)
             else:
-                # If no matching CuratedUrl, create a new one using all non-null and non-empty fields
+                # Previously, we excluded fields with values of None and ""
+                # however, such null values are considered meaningful and should be copied over
                 new_data = {
                     field.name: getattr(delta, field.name)
                     for field in delta._meta.fields
-                    if field.name not in ["to_delete", "collection", "id"]
-                    and getattr(delta, field.name) not in [None, ""]
+                    if field.name not in ["to_delete", "collection", "id"] and getattr(delta, field.name)
                 }
                 CuratedUrl.objects.create(collection=self, **new_data)
 

sde_collections/models/delta_patterns.py

@@ -333,16 +333,28 @@ def unapply(self) -> None:
         affected_deltas = self.delta_urls.all()
         affected_curated = self.curated_urls.all()
 
+        # Get all other patterns of same type for this collection
+        pattern_class = self.__class__
+        other_patterns = pattern_class.objects.filter(collection=self.collection).exclude(id=self.id)
+
         # Process each affected delta URL
         for delta in affected_deltas:
             curated = CuratedUrl.objects.filter(collection=self.collection, url=delta.url).first()
 
-            if not curated:
-                # Scenario 1: Delta only - new URL
-                setattr(delta, field, None)
+            # Find next most specific matching pattern if any
+            matching_patterns = [p for p in other_patterns if re.search(p.get_regex_pattern(), delta.url)]
+
+            next_pattern = None
+            if matching_patterns:
+                # Sort by number of URLs matched (ascending) to find most specific
+                next_pattern = min(matching_patterns, key=lambda p: p.get_url_match_count())
+
+            if next_pattern:
+                # Apply next most specific pattern's value
+                setattr(delta, field, next_pattern.get_new_value())
                 delta.save()
-            else:
-                # Scenario 2: Both exist
+            elif curated:
+                # No other patterns match, revert to curated value
                 setattr(delta, field, getattr(curated, field))
                 delta.save()
 
@@ -354,17 +366,36 @@ def unapply(self) -> None:
                 )
                 if fields_match:
                     delta.delete()
+            else:
+                # No curated URL or other patterns, set to None
+                setattr(delta, field, None)
+                delta.save()
 
         # Handle curated URLs that don't have deltas
         for curated in affected_curated:
             if not DeltaUrl.objects.filter(url=curated.url).exists():
-                # Scenario 3: Curated only
-                # Copy all fields from curated except the one we're nulling
-                fields = {
-                    f.name: getattr(curated, f.name) for f in curated._meta.fields if f.name not in ["id", "collection"]
-                }
-                fields[field] = None  # Set the pattern's field to None
-                delta = DeltaUrl.objects.create(collection=self.collection, **fields)
+                # Find any matching patterns
+                matching_patterns = [p for p in other_patterns if re.search(p.get_regex_pattern(), curated.url)]
+
+                if matching_patterns:
+                    # Apply most specific pattern's value
+                    next_pattern = min(matching_patterns, key=lambda p: p.get_url_match_count())
+                    fields = {
+                        f.name: getattr(curated, f.name)
+                        for f in curated._meta.fields
+                        if f.name not in ["id", "collection"]
+                    }
+                    fields[field] = next_pattern.get_new_value()
+                    DeltaUrl.objects.create(collection=self.collection, **fields)
+                else:
+                    # No other patterns, create delta with None
+                    fields = {
+                        f.name: getattr(curated, f.name)
+                        for f in curated._meta.fields
+                        if f.name not in ["id", "collection"]
+                    }
+                    fields[field] = None
+                    DeltaUrl.objects.create(collection=self.collection, **fields)
 
         # Clear pattern relationships
         self.delta_urls.clear()
@@ -523,11 +554,12 @@ def apply(self) -> None:
 
     def unapply(self) -> None:
         """
-        Remove title modifications:
-        1. Create Delta URLs for affected Curated URLs to explicitly clear titles
-        2. Remove generated titles from affected Delta URLs
-        3. Clean up Delta URLs that become identical to their Curated URL
-        4. Clear resolution tracking
+        Remove title modifications, maintaining pattern precedence:
+        1. Find any remaining patterns that match each URL
+        2. Apply most specific matching pattern's title if one exists
+        3. Otherwise revert to curated title or clear title
+        4. Update title resolution tracking
+        5. Clean up redundant deltas
         """
         DeltaUrl = apps.get_model("sde_collections", "DeltaUrl")
         CuratedUrl = apps.get_model("sde_collections", "CuratedUrl")
@@ -538,16 +570,36 @@ def unapply(self) -> None:
         affected_deltas = self.delta_urls.all()
         affected_curated = self.curated_urls.all()
 
+        # Get all other title patterns for this collection
+        other_patterns = DeltaTitlePattern.objects.filter(collection=self.collection).exclude(id=self.id)
+
         # Process each affected delta URL
         for delta in affected_deltas:
             curated = CuratedUrl.objects.filter(collection=self.collection, url=delta.url).first()
 
-            if not curated:
-                # Scenario 1: Delta only - clear generated title
-                delta.generated_title = ""
-                delta.save()
-            else:
-                # Scenario 2: Both exist - revert to curated title
+            # Find next most specific matching pattern if any
+            matching_patterns = [p for p in other_patterns if re.search(p.get_regex_pattern(), delta.url)]
+
+            next_pattern = None
+            if matching_patterns:
+                # Sort by number of URLs matched (ascending) to find most specific
+                next_pattern = min(matching_patterns, key=lambda p: p.get_url_match_count())
+
+            if next_pattern:
+                # Apply next most specific pattern's title
+                new_title, error = next_pattern.generate_title_for_url(delta)
+                if error:
+                    DeltaResolvedTitleError.objects.update_or_create(
+                        delta_url=delta, defaults={"title_pattern": next_pattern, "error_string": error}
+                    )
+                else:
+                    delta.generated_title = new_title
+                    delta.save()
+                    DeltaResolvedTitle.objects.update_or_create(
+                        delta_url=delta, defaults={"title_pattern": next_pattern, "resolved_title": new_title}
+                    )
+            elif curated:
+                # No other patterns match, revert to curated title
                 delta.generated_title = curated.generated_title
                 delta.save()
 
@@ -559,18 +611,47 @@ def unapply(self) -> None:
                 )
                 if fields_match:
                     delta.delete()
+            else:
+                # No curated URL or other patterns, clear title
+                delta.generated_title = ""
+                delta.save()
 
         # Handle curated URLs that don't have deltas
         for curated in affected_curated:
             if not DeltaUrl.objects.filter(url=curated.url).exists():
-                # Scenario 3: Curated only - create delta with cleared title
-                fields = {
-                    f.name: getattr(curated, f.name) for f in curated._meta.fields if f.name not in ["id", "collection"]
-                }
-                fields["generated_title"] = ""
-                DeltaUrl.objects.create(collection=self.collection, **fields)
-
-        # Clear resolution tracking
+                # Find any matching patterns
+                matching_patterns = [p for p in other_patterns if re.search(p.get_regex_pattern(), curated.url)]
+
+                if matching_patterns:
+                    # Apply most specific pattern's title
+                    next_pattern = min(matching_patterns, key=lambda p: p.get_url_match_count())
+
+                    # Copy all fields from curated
+                    fields = {
+                        f.name: getattr(curated, f.name)
+                        for f in curated._meta.fields
+                        if f.name not in ["id", "collection"]
+                    }
+
+                    # Generate and apply new title
+                    new_title, error = next_pattern.generate_title_for_url(curated)
+                    if not error:
+                        fields["generated_title"] = new_title
+                        delta = DeltaUrl.objects.create(collection=self.collection, **fields)
+                        DeltaResolvedTitle.objects.create(
+                            title_pattern=next_pattern, delta_url=delta, resolved_title=new_title
+                        )
+                else:
+                    # No other patterns, create delta with cleared title
+                    fields = {
+                        f.name: getattr(curated, f.name)
+                        for f in curated._meta.fields
+                        if f.name not in ["id", "collection"]
+                    }
+                    fields["generated_title"] = ""
+                    DeltaUrl.objects.create(collection=self.collection, **fields)
+
+        # Clear resolution tracking for this pattern
         DeltaResolvedTitle.objects.filter(title_pattern=self).delete()
         DeltaResolvedTitleError.objects.filter(title_pattern=self).delete()