retab 0.0.64__py3-none-any.whl → 0.0.66__py3-none-any.whl

retab/types/documents/extract.py

@@ -39,8 +39,7 @@ class DocumentExtractRequest(BaseModel):
     store: bool = Field(default=True, description="If true, the extraction will be stored in the database")
     need_validation: bool = Field(default=False, description="If true, the extraction will be validated against the schema")
     modality: Modality = Field(default="native", description="The modality of the document to be analyzed")
-    hardcoded_keys: Optional[list[str]] = Field(default=None, description="hardcoded keys to be used for the extraction of long lists of data")
-
+    parallel_ocr_keys: Optional[dict[str, str]] = Field(default=None, description="If set, keys to be used for the extraction of long lists of data using Parallel OCR", examples=[{"properties": "ID", "products": "identity.id"}])
 
     # Add a model validator that rejects n_consensus > 1 if temperature is 0
     @field_validator("n_consensus")
@@ -103,12 +102,16 @@ LikelihoodsSource = Literal["consensus", "log_probs"]
 
 
 class RetabParsedChatCompletion(ParsedChatCompletion):
+    model_config = ConfigDict(arbitrary_types_allowed=True, extra="ignore")
+    
     extraction_id: str | None = None
     choices: list[RetabParsedChoice]  # type: ignore
-    # Additional metadata fields (UIForm)
+    # Additional metadata fields
     likelihoods: Optional[dict[str, Any]] = Field(
         default=None, description="Object defining the uncertainties of the fields extracted when using consensus. Follows the same structure as the extraction object."
     )
+
+    requires_human_review: bool = Field(default=False, description="If true, the extraction requires human review")
     schema_validation_error: ErrorDetail | None = None
     # Timestamps
     request_at: datetime.datetime | None = Field(default=None, description="Timestamp of the request")
@@ -330,7 +333,7 @@ class RetabParsedChatCompletionChunk(StreamingBaseModel, ChatCompletionChunk):
                 for idx in range(len(self.choices))
             ],
             likelihoods=final_likelihoods,
-            usage=None,
+            usage=self.usage,
             request_at=self.request_at,
             first_token_at=self.first_token_at,
             last_token_at=self.last_token_at,

retab/types/inference_settings.py

@@ -11,6 +11,7 @@ class InferenceSettings(BaseModel):
     browser_canvas: BrowserCanvas = "A4"
     n_consensus: int = Field(default=1, description="Number of consensus rounds to perform")
     modality: Modality = "native"
+    parallel_ocr_keys: dict[str, str] | None = Field(default=None, description="If set, keys to be used for the extraction of long lists of data using Parallel OCR", examples=[{"properties": "ID", "products": "identity.id"}])
 
     model_config = ConfigDict(extra="ignore")