re-organize tests

Author: sebastianMindee

tests/utils.py

@@ -7,12 +7,12 @@
     REQUEST_TIMEOUT_ENV_NAME,
 )
 
-DATA_DIR = Path("./tests/data/")
-
-EXTRAS_DIR = DATA_DIR / "extras"
-FILE_TYPES_DIR = DATA_DIR / "file_types"
-V2_DATA_DIR = DATA_DIR / "v2"
-PRODUCT_DATA_DIR = DATA_DIR / "products"
+ROOT_DATA_DIR = Path("./tests/data/")
+V1_DATA_DIR = ROOT_DATA_DIR / "v1"
+V2_DATA_DIR = ROOT_DATA_DIR / "v2"
+EXTRAS_DIR = V1_DATA_DIR / "extras"
+FILE_TYPES_DIR = ROOT_DATA_DIR / "file_types"
+V1_PRODUCT_DATA_DIR = V1_DATA_DIR / "products"
 
 
 def clear_envvars(monkeypatch) -> None:

tests/v1/api/test_async_response.py

@@ -10,8 +10,9 @@
 from mindee.parsing.common.api_request import RequestStatus
 from mindee.parsing.common.async_predict_response import AsyncPredictResponse
 from mindee.product.invoice_splitter.invoice_splitter_v1 import InvoiceSplitterV1
+from tests.utils import V1_DATA_DIR, V1_PRODUCT_DATA_DIR
 
-ASYNC_DIR = Path("./tests/data/async")
+ASYNC_DIR = V1_DATA_DIR / "async"
 
 FILE_PATH_POST_SUCCESS = ASYNC_DIR / "post_success.json"
 FILE_PATH_POST_FAIL = ASYNC_DIR / "post_fail_forbidden.json"
@@ -41,7 +42,9 @@ def content(self) -> str:
 
 @pytest.fixture
 def dummy_file_input() -> PathInput:
-    file_input = PathInput("./tests/data/products/invoice_splitter/default_sample.pdf")
+    file_input = PathInput(
+        V1_PRODUCT_DATA_DIR / "invoice_splitter" / "default_sample.pdf"
+    )
     return file_input
 
 

tests/v1/api/test_feedback_response.py

@@ -1,11 +1,12 @@
 import json
 
 from mindee.parsing.common.feedback_response import FeedbackResponse
+from tests.utils import V1_PRODUCT_DATA_DIR
 
 
 def test_empty_feedback_response():
     response = json.load(
-        open("./tests/data/products/invoices/feedback_response/empty.json")
+        open(V1_PRODUCT_DATA_DIR / "invoices" / "feedback_response" / "empty.json")
     )
     feedback_response = FeedbackResponse(response)
     assert feedback_response is not None

tests/v1/api/test_response.py

@@ -14,11 +14,12 @@
 from mindee.product.passport.passport_v1_document import PassportV1Document
 from mindee.product.receipt.receipt_v5 import ReceiptV5
 from mindee.product.receipt.receipt_v5_document import ReceiptV5Document
+from tests.utils import V1_PRODUCT_DATA_DIR
 
 
 def test_invoice_receipt_v5():
     response = json.load(
-        open("./tests/data/products/invoices/response_v4/complete.json")
+        open(V1_PRODUCT_DATA_DIR / "invoices" / "response_v4" / "complete.json")
     )
     parsed_response = PredictResponse(InvoiceV4, response)
     assert isinstance(parsed_response.document.inference, InvoiceV4)
@@ -29,7 +30,7 @@ def test_invoice_receipt_v5():
 
 def test_response_receipt_v5():
     response = json.load(
-        open("./tests/data/products/expense_receipts/response_v5/complete.json")
+        open(V1_PRODUCT_DATA_DIR / "expense_receipts" / "response_v5" / "complete.json")
     )
     parsed_response = PredictResponse(ReceiptV5, response)
     assert isinstance(parsed_response.document.inference, ReceiptV5)
@@ -41,7 +42,10 @@ def test_response_receipt_v5():
 def test_response_financial_doc_with_receipt():
     response = json.load(
         open(
-            "./tests/data/products/financial_document/response_v1/complete_receipt.json"
+            V1_PRODUCT_DATA_DIR
+            / "financial_document"
+            / "response_v1"
+            / "complete_receipt.json"
         )
     )
     parsed_response = PredictResponse(FinancialDocumentV1, response)
@@ -55,7 +59,7 @@ def test_response_financial_doc_with_receipt():
 
 def test_response_passport_v1():
     response = json.load(
-        open("./tests/data/products/passport/response_v1/complete.json")
+        open(V1_PRODUCT_DATA_DIR / "passport" / "response_v1" / "complete.json")
     )
     parsed_response = PredictResponse(PassportV1, response)
     assert isinstance(parsed_response.document.inference, PassportV1)
@@ -67,7 +71,7 @@ def test_response_passport_v1():
 
 def test_response_fr_idcard_v2():
     response = json.load(
-        open("./tests/data/products/idcard_fr/response_v2/complete.json")
+        open(V1_PRODUCT_DATA_DIR / "idcard_fr" / "response_v2" / "complete.json")
     )
     parsed_response = PredictResponse(IdCardV2, response)
     assert isinstance(parsed_response.document.inference, IdCardV2)

tests/v1/extraction/test_image_extractor.py

@@ -6,17 +6,17 @@
 from mindee.extraction.common.image_extractor import extract_multiple_images_from_source
 from mindee.input.sources.path_input import PathInput
 from mindee.product.barcode_reader.barcode_reader_v1 import BarcodeReaderV1
-from tests.utils import PRODUCT_DATA_DIR
+from tests.utils import V1_PRODUCT_DATA_DIR
 
 
 @pytest.fixture
 def barcode_path():
-    return PRODUCT_DATA_DIR / "barcode_reader" / "default_sample.jpg"
+    return V1_PRODUCT_DATA_DIR / "barcode_reader" / "default_sample.jpg"
 
 
 @pytest.fixture
 def barcode_json_path():
-    return PRODUCT_DATA_DIR / "barcode_reader" / "response_v1" / "complete.json"
+    return V1_PRODUCT_DATA_DIR / "barcode_reader" / "response_v1" / "complete.json"
 
 
 def test_barcode_image_extraction(barcode_path, barcode_json_path):

tests/v1/extraction/test_invoice_splitter_auto_extraction.py

@@ -8,13 +8,13 @@
 from mindee.parsing.common.document import Document
 from mindee.product.invoice.invoice_v4 import InvoiceV4
 from mindee.product.invoice_splitter.invoice_splitter_v1 import InvoiceSplitterV1
-from tests.utils import PRODUCT_DATA_DIR, levenshtein_ratio
+from tests.utils import V1_PRODUCT_DATA_DIR, levenshtein_ratio
 from tests.v1.product import get_id, get_version
 
 
 @pytest.fixture
 def invoice_splitter_5p_path():
-    return PRODUCT_DATA_DIR / "invoice_splitter" / "invoice_5p.pdf"
+    return V1_PRODUCT_DATA_DIR / "invoice_splitter" / "invoice_5p.pdf"
 
 
 def prepare_invoice_return(rst_file_path: Path, invoice_prediction: Document):
@@ -31,7 +31,7 @@ def prepare_invoice_return(rst_file_path: Path, invoice_prediction: Document):
 def test_pdf_should_extract_invoices_strict():
     client = Client()
     invoice_splitter_input = PathInput(
-        PRODUCT_DATA_DIR / "invoice_splitter" / "default_sample.pdf"
+        V1_PRODUCT_DATA_DIR / "invoice_splitter" / "default_sample.pdf"
     )
     response = client.enqueue_and_parse(
         InvoiceSplitterV1, invoice_splitter_input, close_file=False
@@ -50,7 +50,10 @@ def test_pdf_should_extract_invoices_strict():
 
     invoice_0 = client.parse(InvoiceV4, extracted_pdfs_strict[0].as_input_source())
     test_string_rst_invoice_0 = prepare_invoice_return(
-        PRODUCT_DATA_DIR / "invoices" / "response_v4" / "summary_full_invoice_p1.rst",
+        V1_PRODUCT_DATA_DIR
+        / "invoices"
+        / "response_v4"
+        / "summary_full_invoice_p1.rst",
         invoice_0.document,
     )
     assert levenshtein_ratio(test_string_rst_invoice_0, str(invoice_0.document)) >= 0.97

tests/v1/extraction/test_multi_receipts_extractor.py

@@ -10,30 +10,33 @@
 from mindee.product.multi_receipts_detector.multi_receipts_detector_v1 import (
     MultiReceiptsDetectorV1,
 )
-from tests.utils import PRODUCT_DATA_DIR
+from tests.utils import V1_PRODUCT_DATA_DIR
 
 
 @pytest.fixture
 def multi_receipts_single_page_path():
-    return PRODUCT_DATA_DIR / "multi_receipts_detector" / "default_sample.jpg"
+    return V1_PRODUCT_DATA_DIR / "multi_receipts_detector" / "default_sample.jpg"
 
 
 @pytest.fixture
 def multi_receipts_single_page_json_path():
     return (
-        PRODUCT_DATA_DIR / "multi_receipts_detector" / "response_v1" / "complete.json"
+        V1_PRODUCT_DATA_DIR
+        / "multi_receipts_detector"
+        / "response_v1"
+        / "complete.json"
     )
 
 
 @pytest.fixture
 def multi_receipts_multi_page_path():
-    return PRODUCT_DATA_DIR / "multi_receipts_detector" / "multipage_sample.pdf"
+    return V1_PRODUCT_DATA_DIR / "multi_receipts_detector" / "multipage_sample.pdf"
 
 
 @pytest.fixture
 def multi_receipts_multi_page_json_path():
     return (
-        PRODUCT_DATA_DIR
+        V1_PRODUCT_DATA_DIR
         / "multi_receipts_detector"
         / "response_v1"
         / "multipage_sample.json"

tests/v1/extraction/test_pdf_extractor.py

@@ -8,24 +8,24 @@
 from mindee.product.invoice_splitter.invoice_splitter_v1_document import (
     InvoiceSplitterV1Document,
 )
-from tests.utils import PRODUCT_DATA_DIR
+from tests.utils import V1_PRODUCT_DATA_DIR
 
 
 @pytest.fixture
 def invoice_default_sample_path():
-    return PRODUCT_DATA_DIR / "invoices" / "default_sample.jpg"
+    return V1_PRODUCT_DATA_DIR / "invoices" / "default_sample.jpg"
 
 
 @pytest.fixture
 def invoice_splitter_5p_path():
-    return PRODUCT_DATA_DIR / "invoice_splitter" / "invoice_5p.pdf"
+    return V1_PRODUCT_DATA_DIR / "invoice_splitter" / "invoice_5p.pdf"
 
 
 @pytest.fixture
 def loaded_prediction():
     dummy_client = Client("dummy_key")
     loaded_prediction_path = (
-        PRODUCT_DATA_DIR / "invoice_splitter" / "response_v1" / "complete.json"
+        V1_PRODUCT_DATA_DIR / "invoice_splitter" / "response_v1" / "complete.json"
     )
     input_response = LocalResponse(loaded_prediction_path)
     response = dummy_client.load_prediction(InvoiceSplitterV1, input_response)

tests/v1/extras/test_extras_integration.py

@@ -3,7 +3,7 @@
 from mindee import Client
 from mindee.product.international_id.international_id_v2 import InternationalIdV2
 from mindee.product.invoice.invoice_v4 import InvoiceV4
-from tests.utils import PRODUCT_DATA_DIR
+from tests.utils import V1_PRODUCT_DATA_DIR
 
 
 @pytest.fixture
@@ -15,7 +15,7 @@ def client():
 @pytest.mark.integration
 def test_send_cropper_extra(client):
     sample = client.source_from_path(
-        PRODUCT_DATA_DIR / "invoices" / "default_sample.jpg",
+        V1_PRODUCT_DATA_DIR / "invoices" / "default_sample.jpg",
     )
     response = client.parse(InvoiceV4, sample, cropper=True)
     assert response.document.inference.pages[0].extras.cropper
@@ -24,7 +24,7 @@ def test_send_cropper_extra(client):
 @pytest.mark.integration
 def test_send_full_text_ocr_extra(client):
     sample = client.source_from_path(
-        PRODUCT_DATA_DIR / "international_id" / "default_sample.jpg",
+        V1_PRODUCT_DATA_DIR / "international_id" / "default_sample.jpg",
     )
     response = client.enqueue_and_parse(InternationalIdV2, sample, full_text=True)
     assert response.document.extras.full_text_ocr

tests/v1/input/test_apply_page_options.py

@@ -12,7 +12,7 @@
     LocalInputSource,
     PathInput,
 )
-from tests.utils import FILE_TYPES_DIR, PRODUCT_DATA_DIR
+from tests.utils import FILE_TYPES_DIR, V1_PRODUCT_DATA_DIR
 
 
 def _assert_page_options(input_source: LocalInputSource, numb_pages: int):
@@ -135,15 +135,15 @@ def test_pdf_input_from_file():
 
 
 def test_pdf_input_from_base64():
-    with open(PRODUCT_DATA_DIR / "invoices" / "invoice_10p.txt", "rt") as fp:
+    with open(V1_PRODUCT_DATA_DIR / "invoices" / "invoice_10p.txt", "rt") as fp:
         input_source = Base64Input(fp.read(), filename="invoice_10p.pdf")
     assert input_source.is_pdf() is True
     input_source.process_pdf(behavior=KEEP_ONLY, on_min_pages=2, page_indexes=[0])
     assert input_source.page_count == 1
 
 
 def test_pdf_input_from_bytes():
-    with open(PRODUCT_DATA_DIR / "invoices" / "invoice_10p.pdf", "rb") as fp:
+    with open(V1_PRODUCT_DATA_DIR / "invoices" / "invoice_10p.pdf", "rb") as fp:
         input_source = BytesInput(fp.read(), filename="invoice_10p.pdf")
     assert input_source.is_pdf() is True
     input_source.process_pdf(behavior=KEEP_ONLY, on_min_pages=2, page_indexes=[0])