retab 0.0.77__py3-none-any.whl → 0.0.79__py3-none-any.whl

retab/resources/documents/client.py

@@ -118,19 +118,24 @@ class BaseDocumentsMixin:
 
     def _prepare_edit(
         self,
-        document: Path | str | IOBase | MIMEData | PIL.Image.Image | HttpUrl,
         filling_instructions: str,
+        document: Path | str | IOBase | MIMEData | PIL.Image.Image | HttpUrl | None = None,
         model: str = FieldUnset,
+        template_id: str | None = FieldUnset,
         **extra_body: Any,
     ) -> PreparedRequest:
-        mime_document = prepare_mime_document(document)
-        
         request_dict: dict[str, Any] = {
-            "document": mime_document,
             "filling_instructions": filling_instructions,
         }
+        
+        if document is not None:
+            mime_document = prepare_mime_document(document)
+            request_dict["document"] = mime_document
+        
         if model is not FieldUnset:
             request_dict["model"] = model
+        if template_id is not FieldUnset:
+            request_dict["template_id"] = template_id
 
         # Merge any extra fields provided by the caller
         if extra_body:
@@ -564,9 +569,10 @@ class Documents(SyncAPIResource, BaseDocumentsMixin):
 
     def edit(
         self,
-        document: Path | str | IOBase | MIMEData | PIL.Image.Image | HttpUrl,
         filling_instructions: str,
+        document: Path | str | IOBase | MIMEData | PIL.Image.Image | HttpUrl | None = None,
         model: str = FieldUnset,
+        template_id: str | None = FieldUnset,
         **extra_body: Any,
     ) -> EditResponse:
         """
@@ -578,10 +584,15 @@ class Documents(SyncAPIResource, BaseDocumentsMixin):
         3. LLM-based form filling using the provided instructions
         4. Returns the filled PDF with form field values populated
 
+        Either `document` OR `template_id` must be provided, but not both.
+
         Args:
-            document: The document to edit. Can be a file path (Path or str), file-like object, MIMEData, PIL Image, or URL.
             filling_instructions: Instructions describing how to fill the form fields.
-            model: The LLM model to use for inference. Defaults to "gemini-2.5-pro".
+            document: The document to edit. Can be a file path (Path or str), file-like object, MIMEData, PIL Image, or URL.
+                Mutually exclusive with template_id.
+            model: The LLM model to use for inference. Defaults to "retab-small".
+            template_id: Template ID to use for filling. When provided, uses the template's pre-defined form fields
+                and empty PDF. Only works for PDF documents. Mutually exclusive with document.
 
         Returns:
             EditResponse: Response containing:
@@ -592,9 +603,10 @@ class Documents(SyncAPIResource, BaseDocumentsMixin):
             HTTPException: If the request fails.
         """
         request = self._prepare_edit(
-            document=document,
             filling_instructions=filling_instructions,
+            document=document,
             model=model,
+            template_id=template_id,
             **extra_body,
         )
         response = self._client._prepared_request(request)
@@ -901,9 +913,10 @@ class AsyncDocuments(AsyncAPIResource, BaseDocumentsMixin):
 
     async def edit(
         self,
-        document: Path | str | IOBase | MIMEData | PIL.Image.Image | HttpUrl,
         filling_instructions: str,
+        document: Path | str | IOBase | MIMEData | PIL.Image.Image | HttpUrl | None = None,
         model: str = FieldUnset,
+        template_id: str | None = FieldUnset,
         **extra_body: Any,
     ) -> EditResponse:
         """
@@ -915,10 +928,15 @@ class AsyncDocuments(AsyncAPIResource, BaseDocumentsMixin):
         3. LLM-based form filling using the provided instructions
         4. Returns the filled PDF with form field values populated
 
+        Either `document` OR `template_id` must be provided, but not both.
+
         Args:
-            document: The document to edit. Can be a file path (Path or str), file-like object, MIMEData, PIL Image, or URL.
             filling_instructions: Instructions describing how to fill the form fields.
+            document: The document to edit. Can be a file path (Path or str), file-like object, MIMEData, PIL Image, or URL.
+                Mutually exclusive with template_id.
             model: The LLM model to use for inference. Defaults to "gemini-2.5-pro".
+            template_id: Template ID to use for filling. When provided, uses the template's pre-defined form fields
+                and empty PDF. Only works for PDF documents. Mutually exclusive with document.
 
         Returns:
             EditResponse: Response containing:
@@ -929,9 +947,10 @@ class AsyncDocuments(AsyncAPIResource, BaseDocumentsMixin):
             HTTPException: If the request fails.
         """
         request = self._prepare_edit(
-            document=document,
             filling_instructions=filling_instructions,
+            document=document,
             model=model,
+            template_id=template_id,
             **extra_body,
         )
         response = await self._client._prepared_request(request)

retab/types/documents/create_messages.py

@@ -22,7 +22,7 @@ class DocumentCreateMessageRequest(BaseModel):
     model_config = ConfigDict(extra="ignore")
     document: MIMEData = Field(description="The document to load.")
     image_resolution_dpi: int = Field(default=192, description="Resolution of the image sent to the LLM")
-    model: str = Field(default="gemini-2.5-flash", description="The model to use for the document.")
+    model: str = Field(default="retab-small", description="The model to use for the document.")
 
 class DocumentCreateInputRequest(DocumentCreateMessageRequest):
     json_schema: dict[str, Any] = Field(description="The json schema to use for the document.")

retab/types/documents/edit.py

@@ -60,6 +60,10 @@ class BaseFormField(BaseModel):
         ...,
         description="Type of field. Currently supported values: 'text' and 'checkbox'.",
     )
+    key: str = Field(
+        ...,
+        description="Key of the field. This is used to identify the field in the form data.",
+    )
     
 
 class FormField(BaseFormField):
@@ -113,10 +117,11 @@ class OCRResult(BaseModel):
 
 
 class InferFormSchemaRequest(BaseModel):
-    """Request to infer form schema from a PDF."""
+    """Request to infer form schema from a PDF or DOCX document."""
     
-    document: MIMEData = Field(..., description="Input document (PDF)")
-    model: str = Field(default="gemini-2.5-pro", description="LLM model to use for inference")
+    document: MIMEData = Field(..., description="Input document (PDF or DOCX). DOCX files will be converted to PDF.")
+    model: str = Field(default="retab-small", description="LLM model to use for inference")
+    instructions: Optional[str] = Field(default=None, description="Optional instructions to guide form field detection (e.g., which fields to focus on, specific areas to look for)")
 
 
 class InferFormSchemaResponse(BaseModel):
@@ -127,10 +132,17 @@ class InferFormSchemaResponse(BaseModel):
     form_fields_pdf: MIMEData = Field(..., description="PDF with form field bounding boxes")
 
 
-class EditRequest(InferFormSchemaRequest):
-    """Request for the infer_and_fill_schema endpoint."""
+class EditRequest(BaseModel):
+    """Request for the infer_and_fill_schema endpoint.
     
+    Either `document` OR `template_id` must be provided, but not both.
+    - When `document` is provided: OCR + LLM inference to detect and fill form fields
+    - When `template_id` is provided: Uses pre-defined form fields from the template (PDF only)
+    """
+    document: Optional[MIMEData] = Field(default=None, description="Input document (PDF or DOCX). DOCX files will be converted to PDF. Mutually exclusive with template_id.")
+    model: str = Field(default="retab-small", description="LLM model to use for inference")
     filling_instructions: str = Field(..., description="Instructions to fill the form")
+    template_id: Optional[str] = Field(default=None, description="Template ID to use for filling. When provided, uses the template's pre-defined form fields and empty PDF. Only works for PDF documents. Mutually exclusive with document.")
 
 class EditResponse(BaseModel):
     """Response from the fill_form endpoint.

retab/types/documents/extract.py

@@ -34,7 +34,7 @@ class DocumentExtractRequest(BaseModel):
     stream: bool = Field(default=False, description="If true, the extraction will be streamed to the user using the active WebSocket connection")
     seed: int | None = Field(default=None, description="Seed for the random number generator. If not provided, a random seed will be generated.", examples=[None])
     store: bool = Field(default=True, description="If true, the extraction will be stored in the database")
-    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"}])
+    chunking_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"}])
     web_search: bool = Field(default=False, description="Enable web search enrichment with Parallel AI to add external context during extraction")
     metadata: dict[str, str] = Field(default_factory=dict, description="User-defined metadata to associate with this extraction")
     extraction_id: Optional[str] = Field(default=None, description="Extraction ID to use for this extraction. If not provided, a new ID will be generated.")

retab/types/documents/parse.py

@@ -18,7 +18,7 @@ class ParseRequest(BaseModel):
     model_config = ConfigDict(extra="ignore")
 
     document: MIMEData = Field(..., description="Document to parse")
-    model: str = Field(default="gemini-2.5-flash", description="Model to use for parsing")
+    model: str = Field(default="retab-small", description="Model to use for parsing")
     table_parsing_format: TableParsingFormat = Field(default="html", description="Format for parsing tables")
     image_resolution_dpi: int = Field(default=192, description="DPI for image processing", ge=96, le=300)
 

retab/types/documents/split.py

@@ -10,7 +10,7 @@ class Category(BaseModel):
 class SplitRequest(BaseModel):
     document: MIMEData = Field(..., description="The document to split")
     categories: list[Category] = Field(..., description="The categories to split the document into")
-    model: str = Field(..., description="The model to use to split the document")
+    model: str = Field(default="retab-small", description="The model to use to split the document")
 
 
 class SplitResult(BaseModel):

retab/types/inference_settings.py

@@ -8,7 +8,7 @@ class InferenceSettings(BaseModel):
     reasoning_effort: ChatCompletionReasoningEffort = "minimal"
     image_resolution_dpi: int = Field(default=192, description="Resolution of the image sent to the LLM", ge=96, le=300)
     n_consensus: int = Field(default=1, ge=1, le=8, description="Number of consensus rounds to perform")
-    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"}])
+    chunking_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"}])
     web_search: bool = Field(default=False, description="Enable web search enrichment with Parallel AI to add external context during extraction")
     model_config = ConfigDict(extra="ignore")
 

retab/types/projects/model.py

@@ -9,7 +9,7 @@ from ..inference_settings import InferenceSettings
 from .predictions import PredictionData
 
 default_inference_settings = InferenceSettings(
-    model="auto-small",
+    model="retab-small",
     temperature=0.5,
     reasoning_effort="minimal",
     image_resolution_dpi=192,

{retab-0.0.77.dist-info → retab-0.0.79.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: retab
-Version: 0.0.77
+Version: 0.0.79
 Summary: Retab official python library
 Home-page: https://github.com/retab-dev/retab
 Author: Retab

{retab-0.0.77.dist-info → retab-0.0.79.dist-info}/RECORD

@@ -7,30 +7,30 @@ retab/resources/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 retab/resources/models.py,sha256=4WidFBnTGZEA65DSn2pLP2SRnCVXkMTw7o_m8xVCFC4,2469
 retab/resources/schemas.py,sha256=rZ6OzfmoYv-mGaRVzvXjO09dD-KxP74mZhOO8sMgcDQ,4632
 retab/resources/documents/__init__.py,sha256=OjXmngFN0RKqO4SI-mJBNzr6Ex6rMxfq0DxaqzP0RQs,89
-retab/resources/documents/client.py,sha256=xiHZPvaxETqZGXanOzp1zFQBcSB7WlgiXGtiv6Ys1dQ,42496
+retab/resources/documents/client.py,sha256=XxWo9FlktrpuskAPyKWTx9UIA2VA81g0SbHjHYnigMM,43583
 retab/resources/extractions/__init__.py,sha256=2H1ezUG8hI5SmTRy6NFzXdYLOdGFFsFrI60uzkitV20,97
 retab/resources/extractions/client.py,sha256=sEoNjOgX91FTOgoJUV-I1A9A9xl1ciCdPlhYwjhEjbA,11035
 retab/resources/projects/__init__.py,sha256=tPR3_3tr7bsoYd618qmGjnYN2R23PmF5oCFd7Z5_HGY,85
 retab/resources/projects/client.py,sha256=5LPAhJt5-nqBP4VWYvo0k7cW6HLGF6K9xMiHKQzIXho,15593
 retab/types/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 retab/types/chat.py,sha256=x9VbtPMa4w6Gc0HrFC3ILl6cCnfEn5ytDnwJtZmlcys,1436
-retab/types/inference_settings.py,sha256=8oeah8r9W4qj8ojn6aXCWsNkGloc2XgQ095W40Z_YUk,1135
+retab/types/inference_settings.py,sha256=wIivYffvEE7v6lhbjbhAZGssK4uYr64Oq6cZKxzY5_M,1131
 retab/types/mime.py,sha256=ZLNCD3pvgn5cbGfJwzrdkjgB9dMHCbN67YEV9bx47zE,10063
 retab/types/modality.py,sha256=4B8LctdUBZVgIjtS2FjrJpljn2Eyse0XE1bpFsGb9O4,131
 retab/types/pagination.py,sha256=A0Fw06baPTfEaYwo3kvNs4vaupzlqylBc6tQH-2DFuY,279
 retab/types/standards.py,sha256=7aGtuvzzkKidvqY8JB2Cjfn43V80FeKwrTtp162kjKc,1477
 retab/types/documents/__init__.py,sha256=YDsvsmwkS5lfGXk5aBqSqmFh6LKX3dM6q_cUo5oIydU,277
 retab/types/documents/correct_orientation.py,sha256=e-ivsslI6L6Gl0YkcslXw_DH620xMGEYVp4tdeviXeM,261
-retab/types/documents/create_messages.py,sha256=Cox0QgIyLhTXIvw1Nzd2BCnB9-5KAYgw_gads5eTaDw,7272
-retab/types/documents/edit.py,sha256=HjDjhHlj08Kks7ABVohTrAJ9QngDgwVj32AxXitjrv0,4804
-retab/types/documents/extract.py,sha256=DhS9jm0lUgXVLObKm2CnSJQ2eqMmsBfttO0K9TndfIw,16728
-retab/types/documents/parse.py,sha256=Jd6i-1UXhAtgntRBZItEHGHeevyLdLmbTQa1-HNrico,1305
-retab/types/documents/split.py,sha256=Sjp2u7Ob6nBRQL23RlgiabgyUmoyf8aEyr7zdvUdU-M,1228
+retab/types/documents/create_messages.py,sha256=Uym0SnVUGkyt1C5AOD37BsZ3puyeu_igR6X9SboojfA,7267
+retab/types/documents/edit.py,sha256=ZY-a_Q9Y76e4oojeJJsisoCZbNSU6gqwAgb9fq9S76w,5930
+retab/types/documents/extract.py,sha256=eMaVl76K_1CeuLmdttfrf4yoQqs27f10w9rNBePb0DY,16724
+retab/types/documents/parse.py,sha256=MXe7zh3DusWQhGe0Sr95nPy6cB8DRX8MA4Hmjj_AP7E,1300
+retab/types/documents/split.py,sha256=xRdJ6IpSRAPi_ZtAG2FNqg5A-v5tzfb1QQkW5UfO2pY,1246
 retab/types/extractions/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 retab/types/extractions/types.py,sha256=mnCYSfJoEKsXN2eG-PrahnnQyR6RDjP5VO9sHC1Opmg,102
 retab/types/projects/__init__.py,sha256=I7P_dems5_LOLgYQ-4Bzt9B6P6jRlQwP-D_9GxRDhVk,155
 retab/types/projects/metrics.py,sha256=J8aZdVbqlszfxosAZyTB7l6lp9WgdL5QgLMlLrckN7k,1946
-retab/types/projects/model.py,sha256=ACvqLQ850t6-mxl_s8fU87Oto6pLCW4ABN4zLe6Bog8,4592
+retab/types/projects/model.py,sha256=keuCzlqUR_WPHW5Grl68fG4VxjTFmNDXQk0MlZ-vtXs,4593
 retab/types/projects/predictions.py,sha256=GeQX-nCmbzKL50aaSFlSyn0sgnbzKserPK6DSQT8ACk,1114
 retab/types/schemas/__init__.py,sha256=9ODWiC_4pUVKxoIKglYZjvRjRyd1ZCVxG8GBdQgHNbU,57
 retab/types/schemas/chat.py,sha256=ppTidxsNslTKE5aBva04i9IxeARMqYpXYLjtR7V6pBc,21219
@@ -44,7 +44,7 @@ retab/utils/hashing.py,sha256=_BMVUvftOcJav68QL0rLkH2dbhW9RRJPzeGC2akR0fc,757
 retab/utils/json_schema.py,sha256=F3MLNGskpfPh1IkXHPLp60ceOEFD79GyL8mVvr0OiVM,19583
 retab/utils/mime.py,sha256=mTP_lqSPttOP5DYJxopiWaeFXrUCPjhwd7y53nCVGO4,6189
 retab/utils/stream_context_managers.py,sha256=gI1gVQSj3nWz6Mvjz7Ix5AiY0g6vSL-c2tPfuP04izo,2314
-retab-0.0.77.dist-info/METADATA,sha256=F2-lc5_Am2m8rqSaVLrlsp0Uwdhe1pLZmiwcplBM9KA,4532
-retab-0.0.77.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-retab-0.0.77.dist-info/top_level.txt,sha256=waQR0EGdhLIQtztoE3AXg7ik5ONQ9q_bsKVpyFuJdq0,6
-retab-0.0.77.dist-info/RECORD,,
+retab-0.0.79.dist-info/METADATA,sha256=GAgtfkDV8Zu0Bc4dBl7vL87xLutKpGUqpwCY3RxGFP0,4532
+retab-0.0.79.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+retab-0.0.79.dist-info/top_level.txt,sha256=waQR0EGdhLIQtztoE3AXg7ik5ONQ9q_bsKVpyFuJdq0,6
+retab-0.0.79.dist-info/RECORD,,