feat: better handling for synced_block feature in Notion connector

CHANGELOG.md

@@ -1,3 +1,9 @@
+## 1.0.28
+
+### Enhancements
+
+* **Added better syncblock handling in Notion connector**
+
 ## 1.0.27
 
 ### Fixes

unstructured_ingest/__version__.py

@@ -1 +1 @@
-__version__ = "1.0.27"  # pragma: no cover
+__version__ = "1.0.28"  # pragma: no cover

unstructured_ingest/processes/connectors/notion/types/blocks/synced_block.py

@@ -18,6 +18,9 @@ def can_have_children() -> bool:
 
     @classmethod
     def from_dict(cls, data: dict):
+        # Original blocks contain children content
+        if "children" not in data:
+             raise ValueError(f"OriginalSyncedBlock data missing 'children': {data}")
         return cls(children=data["children"])
 
     def get_html(self) -> Optional[HtmlTag]:
@@ -31,27 +34,56 @@ class DuplicateSyncedBlock(BlockBase):
 
     @staticmethod
     def can_have_children() -> bool:
+        # Duplicate blocks themselves don't have children directly fetched here,
+        # but they represent content that does, so Notion API might report has_children=True
+        # on the parent block object. The actual children are fetched from the original block.
         return True
 
     @classmethod
     def from_dict(cls, data: dict):
-        return cls(**data)
+        # Duplicate blocks contain a 'synced_from' reference
+        synced_from_data = data.get("synced_from")
+        if not synced_from_data or not isinstance(synced_from_data, dict):
+            raise ValueError(f"Invalid data structure for DuplicateSyncedBlock: {data}")
+        # Ensure required keys are present in the nested dictionary
+        if "type" not in synced_from_data or "block_id" not in synced_from_data:
+            raise ValueError(
+                f"Missing 'type' or 'block_id' in synced_from data: {synced_from_data}"
+            )
+        return cls(type=synced_from_data["type"], block_id=synced_from_data["block_id"])
 
     def get_html(self) -> Optional[HtmlTag]:
+        # HTML representation might need fetching the original block's content,
+        # which is outside the scope of this simple data class.
         return None
 
 
 class SyncBlock(BlockBase):
     @staticmethod
     def can_have_children() -> bool:
+        # Synced blocks (both original and duplicate) can conceptually have children.
         return True
 
     @classmethod
     def from_dict(cls, data: dict):
-        if "synced_from" in data:
+        # Determine if it's a duplicate (has 'synced_from') or original (has 'children')
+        if data.get("synced_from") is not None:
+            # It's a duplicate block containing a reference
+            return DuplicateSyncedBlock.from_dict(data)
+        elif "children" in data:
+            # It's an original block containing children
             return OriginalSyncedBlock.from_dict(data)
         else:
-            return DuplicateSyncedBlock.from_dict(data)
+            # Handle cases where neither 'synced_from' nor 'children' are present.
+            # Notion API might return this for an empty original synced block.
+            # Let's treat it as an empty OriginalSyncedBlock.
+            # If this assumption is wrong, errors might occur later.
+            # Consider logging a warning here if strictness is needed.
+            return OriginalSyncedBlock(children=[])
+
 
     def get_html(self) -> Optional[HtmlTag]:
+        # The specific instance returned by from_dict (Original or Duplicate)
+        # will handle its own get_html logic.
+        # This method on the base SyncBlock might not be directly called.
         return None