full ingestion -> OCR -> extraction flow is now working correctly.

libs/schemas/events.py

@@ -72,22 +72,23 @@ class DocumentExtractedEventData(BaseEventData):
     """Event emitted when field extraction is complete."""
 
     doc_id: str = Field(..., description="Document identifier")
+    tenant_id: str = Field(..., description="Tenant identifier")
     extraction_id: str = Field(..., description="Unique extraction run identifier")
     strategy: Literal["llm", "rules", "hybrid"] = Field(
         ..., description="Extraction strategy used"
     )
-    fields_extracted: int = Field(..., ge=0, description="Number of fields extracted")
-    confidence_avg: float = Field(
-        ..., ge=0.0, le=1.0, description="Average extraction confidence"
+    field_count: int = Field(..., ge=0, description="Number of fields extracted")
+    confidence: float = Field(
+        ..., ge=0.0, le=1.0, description="Extraction confidence score"
     )
-    calibrated_confidence: float = Field(
-        ..., ge=0.0, le=1.0, description="Calibrated confidence score"
+    extraction_results: dict[str, Any] = Field(
+        ..., description="Full extraction results including provenance"
     )
     model_name: str | None = Field(None, description="LLM model used (if applicable)")
-    processing_time_ms: int = Field(
-        ..., ge=0, description="Processing time in milliseconds"
+    processing_time_ms: int | None = Field(
+        None, ge=0, description="Processing time in milliseconds"
     )
-    storage_path: str = Field(..., description="Path to extraction results")
+    storage_path: str | None = Field(None, description="Path to extraction results")
 
 
 # Knowledge Graph events