♻️ simplify document object

Author: ianardee

mindee/documents/base.py

@@ -19,6 +19,12 @@ def serialize_for_json(obj: Any) -> Any:
     return vars(obj)
 
 
+def clean_out_string(out_string: str) -> str:
+    """Clean up the string representation."""
+    regexp = re.compile(r" \n")
+    return regexp.sub("\n", out_string)
+
+
 class Document:
     type: str
     """Document type"""
@@ -83,11 +89,5 @@ def all_checks(self) -> bool:
         """Return status of all checks."""
         return all(self.checklist)
 
-    @staticmethod
-    def clean_out_string(out_string: str) -> str:
-        """Clean up the string representation."""
-        regexp = re.compile(r" \n")
-        return regexp.sub("\n", out_string)
-
 
 TypeDocument = TypeVar("TypeDocument", bound=Document)

mindee/documents/custom/custom_v1.py

@@ -1,6 +1,6 @@
 from typing import Dict, Optional, TypeVar
 
-from mindee.documents.base import Document, TypeApiPrediction
+from mindee.documents.base import Document, TypeApiPrediction, clean_out_string
 from mindee.fields.api_builder import ClassificationField, ListField
 
 
@@ -60,7 +60,7 @@ def __str__(self) -> str:
         for field_name, field_info in self.fields.items():
             custom_doc_str += f"{field_name}: {field_info}\n"
         custom_doc_str += "----------------------"
-        return self.clean_out_string(custom_doc_str)
+        return clean_out_string(custom_doc_str)
 
     def _checklist(self) -> None:
         pass

mindee/documents/financial/financial_v1.py

@@ -1,6 +1,6 @@
 from typing import List, Optional, TypeVar
 
-from mindee.documents.base import Document, TypeApiPrediction
+from mindee.documents.base import Document, TypeApiPrediction, clean_out_string
 from mindee.documents.invoice.invoice_v3 import InvoiceV3
 from mindee.documents.receipt.receipt_v3 import ReceiptV3
 from mindee.endpoints import Endpoint
@@ -125,7 +125,7 @@ def _build_from_api_prediction(
             self.customer_address = TextField({"value": None, "confidence": 0.0})
 
     def __str__(self) -> str:
-        return self.clean_out_string(
+        return clean_out_string(
             "-----Financial Document data-----\n"
             f"Filename: {self.filename or ''}\n"
             f"Invoice number: {self.invoice_number.value}\n"

mindee/documents/invoice/invoice_v3.py

@@ -1,6 +1,6 @@
 from typing import List, Optional, TypeVar
 
-from mindee.documents.base import Document, TypeApiPrediction
+from mindee.documents.base import Document, TypeApiPrediction, clean_out_string
 from mindee.fields.amount import AmountField
 from mindee.fields.base import field_array_confidence
 from mindee.fields.company_registration import CompanyRegistrationField
@@ -126,7 +126,7 @@ def __str__(self) -> str:
             [str(p) for p in self.payment_details]
         )
         taxes = "\n       ".join(f"{t}" for t in self.taxes)
-        return self.clean_out_string(
+        return clean_out_string(
             "-----Invoice data-----\n"
             f"Filename: {self.filename or ''}\n"
             f"Invoice number: {self.invoice_number}\n"

mindee/documents/passport/passport_v1.py

@@ -1,7 +1,7 @@
 from datetime import datetime
 from typing import List, Optional, TypeVar
 
-from mindee.documents.base import Document, TypeApiPrediction
+from mindee.documents.base import Document, TypeApiPrediction, clean_out_string
 from mindee.fields.base import field_array_confidence
 from mindee.fields.date import DateField
 from mindee.fields.text import TextField
@@ -92,7 +92,7 @@ def __str__(self) -> str:
                 for given_name in self.given_names
             ]
         )
-        return self.clean_out_string(
+        return clean_out_string(
             "-----Passport data-----\n"
             f"Filename: {self.filename or ''}\n"
             f"Full name: {self.full_name}\n"

mindee/documents/receipt/receipt_v3.py

@@ -1,6 +1,6 @@
 from typing import List, Optional, TypeVar
 
-from mindee.documents.base import Document, TypeApiPrediction
+from mindee.documents.base import Document, TypeApiPrediction, clean_out_string
 from mindee.fields.amount import AmountField
 from mindee.fields.base import field_array_confidence, field_array_sum
 from mindee.fields.date import DateField
@@ -56,7 +56,7 @@ def __init__(
 
     def __str__(self) -> str:
         taxes = "\n       ".join(f"{t}" for t in self.taxes)
-        return self.clean_out_string(
+        return clean_out_string(
             "-----Receipt data-----\n"
             f"Filename: {self.filename or ''}\n"
             f"Total amount including taxes: {self.total_incl}\n"

mindee/documents/receipt/receipt_v4.py

@@ -1,6 +1,6 @@
 from typing import List, Optional, TypeVar
 
-from mindee.documents.base import Document, TypeApiPrediction
+from mindee.documents.base import Document, TypeApiPrediction, clean_out_string
 from mindee.fields.amount import AmountField
 from mindee.fields.date import DateField
 from mindee.fields.locale import LocaleField
@@ -93,7 +93,7 @@ def _build_from_api_prediction(
 
     def __str__(self) -> str:
         taxes = "\n       ".join(f"{t}" for t in self.taxes)
-        return self.clean_out_string(
+        return clean_out_string(
             "----- Receipt V4 -----\n"
             f"Filename: {self.filename or ''}\n"
             f"Total amount: {self.total_amount}\n"

mindee/documents/us/bank_check/bank_check_v1.py

@@ -1,6 +1,6 @@
 from typing import List, Optional, TypeVar
 
-from mindee.documents.base import Document, TypeApiPrediction
+from mindee.documents.base import Document, TypeApiPrediction, clean_out_string
 from mindee.fields.amount import AmountField
 from mindee.fields.date import DateField
 from mindee.fields.orientation import OrientationField
@@ -82,7 +82,7 @@ def __str__(self) -> str:
         payees = ", ".join(
             [payee.value if payee.value is not None else "" for payee in self.payees]
         )
-        return (
+        return clean_out_string(
             "----- US Bank Check -----\n"
             f"Filename: {self.filename or ''}".rstrip() + "\n"
             f"Routing number: {self.routing_number}\n"

mindee/response.py

@@ -17,6 +17,12 @@ class PredictResponse(Generic[TypeDocument]):
     """Raw HTTP response JSON"""
     document_type: str
     """Document type"""
+    input_path: Optional[str] = None
+    """Path of the input file"""
+    input_filename: Optional[str] = None
+    """Name of the input file"""
+    input_mimetype: Optional[str] = None
+    """MIME type of the input file"""
     document: Optional[TypeDocument]
     """An instance of the ``Document`` class, according to the type given."""
     pages: List[TypeDocument]
@@ -42,6 +48,11 @@ def __init__(
         self.document_type = doc_config.document_type
         self.pages = []
 
+        if input_source:
+            self.input_path = input_source.filepath
+            self.input_filename = input_source.filename
+            self.input_mimetype = input_source.file_mimetype
+
         if not response_ok:
             self.document = None
         else: