⬆️ update test data (#148)

Author: ianardee

docs/predictions/standard/documents/passport_v1.rst

@@ -7,4 +7,4 @@ Passport V1
     :language: Python
 
 .. autoclass:: mindee.documents.PassportV1
-    :members:
+    :members:

docs/predictions/standard/international.rst

@@ -11,4 +11,4 @@ International
 .. include:: ./documents/invoice_splitter_v1.rst
 .. include:: ./documents/proof_of_address_v1.rst
 .. include:: ./documents/cropper_v1.rst
-.. include:: ./documents/material_certificate.rst
+.. include:: ./documents/material_certificate_v1.rst

mindee/documents/fr/id_card/id_card_v1.py

@@ -1,35 +1,36 @@
 from typing import List, Optional, TypeVar
 
 from mindee.documents.base import Document, TypeApiPrediction, clean_out_string
+from mindee.fields.classification import ClassificationField
 from mindee.fields.date import DateField
 from mindee.fields.text import TextField
 
 
 class IdCardV1(Document):
     """Carte Nationale d'Identité v1 prediction results."""
 
-    document_side: TextField
-    """The side of the document which is visible."""
-    id_number: TextField
-    """The identification card number."""
-    given_names: List[TextField]
-    """The given name(s) of the card holder."""
-    surname: TextField
-    """The surname of the card holder."""
+    authority: TextField
+    """The name of the issuing authority."""
     birth_date: DateField
     """The date of birth of the card holder."""
     birth_place: TextField
     """The place of birth of the card holder."""
+    document_side: ClassificationField
+    """The side of the document which is visible."""
     expiry_date: DateField
     """The expiry date of the identification card."""
-    authority: TextField
-    """The name of the issuing authority."""
     gender: TextField
     """The gender of the card holder."""
+    given_names: List[TextField]
+    """The given name(s) of the card holder."""
+    id_number: TextField
+    """The identification card number."""
     mrz1: TextField
     """Machine Readable Zone, first line"""
     mrz2: TextField
     """Machine Readable Zone, second line"""
+    surname: TextField
+    """The surname of the card holder."""
 
     def __init__(
         self,
@@ -61,20 +62,8 @@ def _build_from_api_prediction(
         :param api_prediction: Raw prediction from HTTP response
         :param page_n: Page number
         """
-        self.document_side = TextField(
-            api_prediction.get("document_side", {}),
-            page_id=page_n,
-        )
-        self.id_number = TextField(
-            api_prediction["id_number"],
-            page_id=page_n,
-        )
-        self.given_names = [
-            TextField(prediction, page_id=page_n)
-            for prediction in api_prediction["given_names"]
-        ]
-        self.surname = TextField(
-            api_prediction["surname"],
+        self.authority = TextField(
+            api_prediction["authority"],
             page_id=page_n,
         )
         self.birth_date = DateField(
@@ -85,18 +74,26 @@ def _build_from_api_prediction(
             api_prediction["birth_place"],
             page_id=page_n,
         )
-        self.expiry_date = DateField(
-            api_prediction["expiry_date"],
+        self.document_side = ClassificationField(
+            api_prediction.get("document_side", {}),
             page_id=page_n,
         )
-        self.authority = TextField(
-            api_prediction["authority"],
+        self.expiry_date = DateField(
+            api_prediction["expiry_date"],
             page_id=page_n,
         )
         self.gender = TextField(
             api_prediction["gender"],
             page_id=page_n,
         )
+        self.given_names = [
+            TextField(prediction, page_id=page_n)
+            for prediction in api_prediction["given_names"]
+        ]
+        self.id_number = TextField(
+            api_prediction["id_number"],
+            page_id=page_n,
+        )
         self.mrz1 = TextField(
             api_prediction["mrz1"],
             page_id=page_n,
@@ -105,6 +102,10 @@ def _build_from_api_prediction(
             api_prediction["mrz2"],
             page_id=page_n,
         )
+        self.surname = TextField(
+            api_prediction["surname"],
+            page_id=page_n,
+        )
 
     def __str__(self) -> str:
         given_names = "\n".join([str(item) for item in self.given_names])

mindee/documents/passport/passport_v1.py

@@ -8,37 +8,37 @@
 
 
 class PassportV1(Document):
-    """Cropper v1 prediction results."""
+    """Passport v1 prediction results."""
 
+    birth_date: DateField
+    """The date of birth of the passport holder."""
+    birth_place: TextField
+    """The place of birth of the passport holder."""
     country: TextField
-    """Country of issue"""
-    id_number: TextField
-    """Passport number"""
+    """The country's 3 letter code (ISO 3166-1 alpha-3)."""
     expiry_date: DateField
-    """Date the passport expires"""
-    issuance_date: DateField
-    """Date the passport was issued"""
-    surname: TextField
-    """Holder's last name (surname)"""
-    given_names: List[TextField]
-    """Holder's list of first (given) names"""
+    """The expiry date of the passport."""
     full_name: TextField
     """
     Holder's full name.
     The combination of `given_names` and `surname` fields.
     """
-    birth_date: DateField
-    """Holder's date of birth"""
-    birth_place: TextField
-    """Holder's place of birth"""
     gender: TextField
-    """Holder's gender or sex"""
-    mrz1: TextField
-    """First line of the Machine-Readable Zone"""
-    mrz2: TextField
-    """Second line of the Machine-Readable Zone"""
+    """The gender of the passport holder."""
+    given_names: List[TextField]
+    """The given name(s) of the passport holder."""
+    id_number: TextField
+    """The passport's identification number."""
+    issuance_date: DateField
+    """The date the passport was issued."""
     mrz: TextField
     """Combination of both MRZ fields."""
+    mrz1: TextField
+    """Machine Readable Zone, first line"""
+    mrz2: TextField
+    """Machine Readable Zone, second line"""
+    surname: TextField
+    """The surname of the passport holder."""
 
     def __init__(
         self,
@@ -47,7 +47,7 @@ def __init__(
         page_n: Optional[int] = None,
     ):
         """
-        Passport document.
+        Passport v1 prediction results.
 
         :param api_prediction: Raw prediction from HTTP response
         :param input_source: Input object
@@ -67,25 +67,55 @@ def _build_from_api_prediction(
         self, api_prediction: TypeApiPrediction, page_n: Optional[int] = None
     ) -> None:
         """
-        Build the document from an API response JSON.
+        Build the object from the prediction API JSON.
 
         :param api_prediction: Raw prediction from HTTP response
         :param page_n: Page number for multi pages pdf input
         """
-        self.country = TextField(api_prediction["country"], page_id=page_n)
-        self.id_number = TextField(api_prediction["id_number"], page_id=page_n)
-        self.birth_date = DateField(api_prediction["birth_date"], page_id=page_n)
-        self.expiry_date = DateField(api_prediction["expiry_date"], page_id=page_n)
-        self.issuance_date = DateField(api_prediction["issuance_date"], page_id=page_n)
-        self.birth_place = TextField(api_prediction["birth_place"], page_id=page_n)
-        self.gender = TextField(api_prediction["gender"], page_id=page_n)
-        self.surname = TextField(api_prediction["surname"], page_id=page_n)
-        self.mrz1 = TextField(api_prediction["mrz1"], page_id=page_n)
-        self.mrz2 = TextField(api_prediction["mrz2"], page_id=page_n)
+        self.birth_date = DateField(
+            api_prediction["birth_date"],
+            page_id=page_n,
+        )
+        self.birth_place = TextField(
+            api_prediction["birth_place"],
+            page_id=page_n,
+        )
+        self.country = TextField(
+            api_prediction["country"],
+            page_id=page_n,
+        )
+        self.expiry_date = DateField(
+            api_prediction["expiry_date"],
+            page_id=page_n,
+        )
+        self.gender = TextField(
+            api_prediction["gender"],
+            page_id=page_n,
+        )
         self.given_names = [
-            TextField(given_name, page_id=page_n)
-            for given_name in api_prediction["given_names"]
+            TextField(prediction, page_id=page_n)
+            for prediction in api_prediction["given_names"]
         ]
+        self.id_number = TextField(
+            api_prediction["id_number"],
+            page_id=page_n,
+        )
+        self.issuance_date = DateField(
+            api_prediction["issuance_date"],
+            page_id=page_n,
+        )
+        self.mrz1 = TextField(
+            api_prediction["mrz1"],
+            page_id=page_n,
+        )
+        self.mrz2 = TextField(
+            api_prediction["mrz2"],
+            page_id=page_n,
+        )
+        self.surname = TextField(
+            api_prediction["surname"],
+            page_id=page_n,
+        )
         self.mrz = TextField({"value": None, "confidence": 0.0}, page_id=page_n)
         self.full_name = TextField({"value": None, "confidence": 0.0}, page_id=page_n)
 

mindee/documents/receipt/receipt_v5.py

@@ -16,35 +16,35 @@ class ReceiptV5(Document):
     """Receipt v5 prediction results."""
 
     category: ClassificationField
-    """The receipt category among predefined classes."""
+    """The purchase category among predefined classes."""
     date: DateField
     """The date the purchase was made."""
     document_type: ClassificationField
-    """Whether the document is an expense receipt or a credit card receipt."""
+    """One of: 'CREDIT CARD RECEIPT', 'EXPENSE RECEIPT'."""
     line_items: List[ReceiptV5LineItem]
-    """Full extraction of lines, including: description, quantity, unit price and total."""
+    """List of line item details."""
     locale: LocaleField
-    """The locale identifier in BCP 47 (RFC 5646) format: ISO language code, '-', ISO country code."""
+    """The locale detected on the document."""
     subcategory: ClassificationField
-    """The receipt sub category among predefined classes for transport and food."""
+    """The purchase subcategory among predefined classes for transport and food."""
     supplier_address: TextField
-    """The address of the supplier or merchant returned as a single string."""
+    """The address of the supplier or merchant."""
     supplier_company_registrations: List[CompanyRegistrationField]
-    """List of supplier company registrations or identifiers."""
+    """List of company registrations associated to the supplier."""
     supplier_name: TextField
     """The name of the supplier or merchant."""
     supplier_phone_number: TextField
-    """The Phone number of the supplier or merchant returned as a single string."""
+    """The phone number of the supplier or merchant."""
     taxes: Taxes
-    """List of tax lines information including: Amount, tax rate, tax base amount and tax code."""
+    """List of tax lines information."""
     time: TextField
-    """Time of purchase with 24 hours formatting (HH:MM)."""
+    """The time the purchase was made."""
     tip: AmountField
     """The total amount of tip and gratuity."""
     total_amount: AmountField
-    """The total amount paid including taxes, discounts, fees, tips, and gratuity."""
+    """The total amount paid: includes taxes, discounts, fees, tips, and gratuity."""
     total_net: AmountField
-    """The total amount excluding taxes."""
+    """The net amount paid: does not include taxes, fees, and discounts."""
     total_tax: AmountField
     """The total amount of taxes."""
 
@@ -176,13 +176,13 @@ def __str__(self) -> str:
             "=====================\n"
             f":Filename: {self.filename or ''}\n"
             f":Expense Locale: {self.locale}\n"
-            f":Expense Category: {self.category}\n"
-            f":Expense Sub Category: {self.subcategory}\n"
+            f":Purchase Category: {self.category}\n"
+            f":Purchase Subcategory: {self.subcategory}\n"
             f":Document Type: {self.document_type}\n"
             f":Purchase Date: {self.date}\n"
             f":Purchase Time: {self.time}\n"
             f":Total Amount: {self.total_amount}\n"
-            f":Total Excluding Taxes: {self.total_net}\n"
+            f":Total Net: {self.total_net}\n"
             f":Total Tax: {self.total_tax}\n"
             f":Tip and Gratuity: {self.tip}\n"
             f":Taxes: {self.taxes}\n"

mindee/documents/receipt/receipt_v5_line_item.py

@@ -10,7 +10,7 @@
 
 
 class ReceiptV5LineItem(FieldPositionMixin, FieldConfidenceMixin):
-    """Full extraction of lines, including: description, quantity, unit price and total."""
+    """List of line item details."""
 
     description: Optional[str]
     """The item description."""
@@ -20,8 +20,6 @@ class ReceiptV5LineItem(FieldPositionMixin, FieldConfidenceMixin):
     """The item total amount."""
     unit_price: Optional[float]
     """The item unit price."""
-    confidence: float = 0.0
-    """Confidence score"""
     page_n: int
     """The document page on which the information was found."""
 

tests/__init__.py

@@ -1,9 +1,7 @@
 RECEIPT_DATA_DIR = "./tests/data/receipt"
 INVOICE_DATA_DIR = "./tests/data/invoice"
-FINANCIAL_DOC_DATA_DIR = "./tests/data/financial_document"
-PASSPORT_DATA_DIR = "./tests/data/passport"
-PROOF_OF_ADDRESS_DATA_DIR = "./tests/data/proof_of_address"
 US_BANK_CHECK_DATA_DIR = "./tests/data/us/bank_check"
+PASSPORT_DATA_DIR = "./tests/data/passport"
 FR_CARTE_GRISE_DATA_DIR = "./tests/data/fr/carte_grise"
 CUSTOM_DATA_DIR = "./tests/data/custom"
 CROPPER_DATA_DIR = "./tests/data/cropper"

tests/data

@@ -5,7 +5,6 @@
 from mindee.documents.fr import IdCardV1
 
 FR_ID_CARD_DATA_DIR = "./tests/data/fr/id_card"
-
 FILE_PATH_FR_ID_CARD_V1_COMPLETE = f"{ FR_ID_CARD_DATA_DIR }/response_v1/complete.json"
 FILE_PATH_FR_ID_CARD_V1_EMPTY = f"{ FR_ID_CARD_DATA_DIR }/response_v1/empty.json"
 
@@ -28,6 +27,19 @@ def id_card_v1_page0():
     return IdCardV1(json_data["document"]["inference"]["pages"][0], page_n=0)
 
 
+def test_empty_doc_constructor(id_card_v1_doc_empty):
+    assert id_card_v1_doc_empty.id_number.value is None
+    assert len(id_card_v1_doc_empty.given_names) == 0
+    assert id_card_v1_doc_empty.surname.value is None
+    assert id_card_v1_doc_empty.birth_date.value is None
+    assert id_card_v1_doc_empty.birth_place.value is None
+    assert id_card_v1_doc_empty.expiry_date.value is None
+    assert id_card_v1_doc_empty.authority.value is None
+    assert id_card_v1_doc_empty.gender.value is None
+    assert id_card_v1_doc_empty.mrz1.value is None
+    assert id_card_v1_doc_empty.mrz2.value is None
+
+
 def test_doc_constructor(id_card_v1_doc):
     file_path = f"{ FR_ID_CARD_DATA_DIR }/response_v1/doc_to_string.txt"
     reference_str = open(file_path, "r", encoding="utf-8").read().strip()

tests/documents/fr/test_id_card_v1.py

@@ -3,8 +3,8 @@
 import pytest
 
 from mindee.documents import FinancialDocumentV1
-from tests import FINANCIAL_DOC_DATA_DIR
 
+FINANCIAL_DOC_DATA_DIR = "./tests/data/financial_document"
 FILE_PATH_FINANCIAL_DOC_V1_INVOICE = (
     f"{FINANCIAL_DOC_DATA_DIR}/response_v1/complete_invoice.json"
 )