✨ add support for extraction in crop, split, and classification

mindee/client_mixin.py

@@ -2,7 +2,11 @@
 from typing import BinaryIO, Union
 
 from mindee.error import MindeeClientError
-from mindee.input import Base64Input, BytesInput, FileInput, PathInput, UrlInputSource
+from mindee.input.sources.base_64_input import Base64Input
+from mindee.input.sources.bytes_input import BytesInput
+from mindee.input.sources.file_input import FileInput
+from mindee.input.sources.path_input import PathInput
+from mindee.input.sources.url_input_source import UrlInputSource
 
 
 class ClientMixin:

mindee/client_v2.py

@@ -16,7 +16,7 @@
     is_valid_post_response,
 )
 from mindee.parsing.v2.common_response import CommonStatus
-from mindee.v2.parsing.inference.base_response import BaseResponse
+from mindee.parsing.v2.base_response import BaseResponse
 from mindee.parsing.v2.inference_response import InferenceResponse
 from mindee.parsing.v2.job_response import JobResponse
 

mindee/parsing/v2/base_inference.py

@@ -0,0 +1,29 @@
+from abc import ABC
+from typing import TypeVar
+
+from mindee.parsing.common.string_dict import StringDict
+from mindee.parsing.v2.inference_file import InferenceFile
+from mindee.parsing.v2.inference_job import InferenceJob
+from mindee.parsing.v2.inference_model import InferenceModel
+
+
+class BaseInference(ABC):
+    """Base class for V2 inference objects."""
+
+    job: InferenceJob
+    """Job the inference belongs to."""
+    model: InferenceModel
+    """Model info for the inference."""
+    file: InferenceFile
+    """File info for the inference."""
+    id: str
+    """ID of the inference."""
+
+    def __init__(self, raw_response: StringDict):
+        self.id = raw_response["id"]
+        self.job = InferenceJob(raw_response["job"])
+        self.model = InferenceModel(raw_response["model"])
+        self.file = InferenceFile(raw_response["file"])
+
+
+TypeBaseInference = TypeVar("TypeBaseInference", bound=BaseInference)

mindee/parsing/v2/base_response.py

@@ -0,0 +1,21 @@
+from abc import ABC
+
+from mindee.parsing.v2.base_inference import BaseInference
+from mindee.parsing.v2.common_response import CommonResponse
+
+
+class BaseResponse(ABC, CommonResponse):
+    """Base class for V2 inference responses."""
+
+    inference: BaseInference
+    """The inference result for a split utility request"""
+    _slug: str
+    """Slug of the inference."""
+
+    def __str__(self) -> str:
+        return str(self.inference)
+
+    @classmethod
+    def get_result_slug(cls) -> str:
+        """Getter for the inference slug."""
+        return cls._slug

mindee/parsing/v2/inference.py

@@ -1,5 +1,5 @@
 from mindee.parsing.common.string_dict import StringDict
-from mindee.v2.parsing.inference import BaseInference
+from mindee.parsing.v2.base_inference import BaseInference
 from mindee.parsing.v2.inference_active_options import InferenceActiveOptions
 from mindee.parsing.v2.inference_result import InferenceResult
 

mindee/parsing/v2/inference_job.py

@@ -0,0 +1,14 @@
+from mindee.parsing.common.string_dict import StringDict
+
+
+class InferenceJob:
+    """Inference Job info."""
+
+    id: str
+    """UUID of the Job."""
+
+    def __init__(self, raw_response: StringDict) -> None:
+        self.id = raw_response["id"]
+
+    def __str__(self) -> str:
+        return f"Job\n===\n:ID: {self.id}"

mindee/parsing/v2/inference_response.py

@@ -1,8 +1,6 @@
 from mindee.parsing.common.string_dict import StringDict
+from mindee.parsing.v2.base_response import BaseResponse
 from mindee.parsing.v2.inference import Inference
-from mindee.v2.parsing.inference.base_response import (
-    BaseResponse,
-)
 
 
 class InferenceResponse(BaseResponse):

mindee/v2/commands/cli_parser.py

@@ -16,7 +16,7 @@
 from mindee.input import BaseParameters
 
 from mindee.input.sources import PathInput, UrlInputSource
-from mindee.v2.parsing import BaseResponse
+from mindee.parsing.v2.base_response import BaseResponse
 
 
 @dataclass

mindee/v2/parsing/__init__.py

@@ -1,5 +1,5 @@
-from mindee.v2.parsing.inference.base_inference import BaseInference
-from mindee.v2.parsing.inference.base_response import BaseResponse
+from mindee.parsing.v2.base_inference import BaseInference
+from mindee.parsing.v2.base_response import BaseResponse
 
 __all__ = [
     "BaseInference",

mindee/v2/parsing/inference/__init__.py

@@ -1,7 +1,5 @@
-from mindee.v2.parsing.inference.base_inference import BaseInference
-from mindee.v2.parsing.inference.base_response import (
-    BaseResponse,
-)
+from mindee.parsing.v2.base_inference import BaseInference
+from mindee.parsing.v2.base_response import BaseResponse
 
 __all__ = [
     "BaseInference",

mindee/v2/parsing/inference/base_inference.py

@@ -1,29 +1,3 @@
-from abc import ABC
-from typing import TypeVar
+from mindee.parsing.v2.base_inference import BaseInference, TypeBaseInference
 
-from mindee.parsing.common.string_dict import StringDict
-from mindee.parsing.v2.inference_file import InferenceFile
-from mindee.parsing.v2.inference_model import InferenceModel
-from mindee.v2.parsing.inference.inference_job import InferenceJob
-
-
-class BaseInference(ABC):
-    """Base class for V2 inference objects."""
-
-    job: InferenceJob
-    """Job the inference belongs to."""
-    model: InferenceModel
-    """Model info for the inference."""
-    file: InferenceFile
-    """File info for the inference."""
-    id: str
-    """ID of the inference."""
-
-    def __init__(self, raw_response: StringDict):
-        self.id = raw_response["id"]
-        self.job = InferenceJob(raw_response["job"])
-        self.model = InferenceModel(raw_response["model"])
-        self.file = InferenceFile(raw_response["file"])
-
-
-TypeBaseInference = TypeVar("TypeBaseInference", bound=BaseInference)
+__all__ = ["BaseInference", "TypeBaseInference"]

mindee/v2/parsing/inference/base_response.py

@@ -1,22 +1,3 @@
-from abc import ABC
+from mindee.parsing.v2.base_response import BaseResponse
 
-from mindee.v2.parsing.inference.base_inference import BaseInference
-
-from mindee.parsing.v2.common_response import CommonResponse
-
-
-class BaseResponse(ABC, CommonResponse):
-    """Base class for V2 inference responses."""
-
-    inference: BaseInference
-    """The inference result for a split utility request"""
-    _slug: str
-    """Slug of the inference."""
-
-    def __str__(self) -> str:
-        return str(self.inference)
-
-    @classmethod
-    def get_result_slug(cls) -> str:
-        """Getter for the inference slug."""
-        return cls._slug
+__all__ = ["BaseResponse"]

mindee/v2/parsing/inference/inference_job.py

@@ -1,14 +1,3 @@
-from mindee.parsing.common.string_dict import StringDict
+from mindee.parsing.v2.inference_job import InferenceJob
 
-
-class InferenceJob:
-    """Inference Job info."""
-
-    id: str
-    """UUID of the Job."""
-
-    def __init__(self, raw_response: StringDict) -> None:
-        self.id = raw_response["id"]
-
-    def __str__(self) -> str:
-        return f"Job\n===\n:ID: {self.id}"
+__all__ = ["InferenceJob"]

mindee/v2/product/classification/classification_classifier.py

@@ -1,4 +1,5 @@
 from mindee.parsing.common.string_dict import StringDict
+from mindee.parsing.v2.inference_response import InferenceResponse
 
 
 class ClassificationClassifier:
@@ -7,8 +8,15 @@ class ClassificationClassifier:
     document_type: str
     """The document type, as identified on given classification values."""
 
+    extraction_response: InferenceResponse
+    """The extraction response associated with the classification."""
+
     def __init__(self, server_response: StringDict):
         self.document_type = server_response["document_type"]
+        if server_response.get("extraction_response") is not None:
+            self.extraction_response = InferenceResponse(
+                server_response["extraction_response"]
+            )
 
     def __str__(self) -> str:
         return f":Document Type: {self.document_type}"

mindee/v2/product/classification/classification_inference.py

@@ -1,5 +1,5 @@
 from mindee.parsing.common.string_dict import StringDict
-from mindee.v2.parsing.inference.base_inference import BaseInference
+from mindee.parsing.v2.base_inference import BaseInference
 from mindee.v2.product.classification.classification_result import ClassificationResult
 
 

mindee/v2/product/classification/classification_response.py

@@ -1,5 +1,5 @@
 from mindee.parsing.common.string_dict import StringDict
-from mindee.v2.parsing.inference import BaseResponse
+from mindee.parsing.v2.base_response import BaseResponse
 from mindee.v2.product.classification.classification_inference import (
     ClassificationInference,
 )

mindee/v2/product/crop/crop_box.py

@@ -1,3 +1,4 @@
+from mindee.parsing.v2.inference_response import InferenceResponse
 from mindee.extraction import ExtractedImage, extract_multiple_images_from_source
 from mindee.input.sources.local_input_source import LocalInputSource
 from mindee.parsing.common.string_dict import StringDict
@@ -11,10 +12,16 @@ class CropBox:
     """Location which includes cropping coordinates for the detected object, within the source document."""
     object_type: str
     """Type or classification of the detected object."""
+    extraction_response: InferenceResponse
+    """The extraction response associated with the crop."""
 
     def __init__(self, server_response: StringDict):
         self.location = FieldLocation(server_response["location"])
         self.object_type = server_response["object_type"]
+        if server_response.get("extraction_response") is not None:
+            self.extraction_response = InferenceResponse(
+                server_response["extraction_response"]
+            )
 
     def __str__(self) -> str:
         return f"* :Location: {self.location}\n  :Object Type: {self.object_type}"

mindee/v2/product/crop/crop_inference.py

@@ -1,5 +1,5 @@
 from mindee.parsing.common.string_dict import StringDict
-from mindee.v2.parsing.inference.base_inference import BaseInference
+from mindee.parsing.v2.base_inference import BaseInference
 from mindee.v2.product.crop.crop_result import CropResult
 
 

mindee/v2/product/crop/crop_response.py

@@ -1,7 +1,7 @@
 from mindee.input.sources.local_input_source import LocalInputSource
 from mindee.parsing.common.string_dict import StringDict
 from mindee.v2.file_operations.crop_files import CropFiles
-from mindee.v2.parsing.inference import BaseResponse
+from mindee.parsing.v2.base_response import BaseResponse
 from mindee.v2.product.crop.crop_inference import CropInference
 
 

mindee/v2/product/ocr/ocr_inference.py

@@ -1,5 +1,5 @@
 from mindee.parsing.common.string_dict import StringDict
-from mindee.v2.parsing.inference.base_inference import BaseInference
+from mindee.parsing.v2.base_inference import BaseInference
 from mindee.v2.product.ocr.ocr_result import OCRResult
 
 

mindee/v2/product/ocr/ocr_response.py

@@ -1,5 +1,5 @@
 from mindee.parsing.common.string_dict import StringDict
-from mindee.v2.parsing.inference import BaseResponse
+from mindee.parsing.v2.base_response import BaseResponse
 from mindee.v2.product.ocr.ocr_inference import OCRInference
 
 

mindee/v2/product/split/split_inference.py

@@ -1,5 +1,5 @@
 from mindee.parsing.common.string_dict import StringDict
-from mindee.v2.parsing.inference.base_inference import BaseInference
+from mindee.parsing.v2.base_inference import BaseInference
 from mindee.v2.product.split.split_result import SplitResult
 
 

mindee/v2/product/split/split_range.py

@@ -1,5 +1,6 @@
 from typing import List
 
+from mindee.parsing.v2.inference_response import InferenceResponse
 from mindee.extraction.pdf_extractor.extracted_pdf import ExtractedPdf
 from mindee.input.sources.local_input_source import LocalInputSource
 from mindee.parsing.common.string_dict import StringDict
@@ -16,10 +17,16 @@ class SplitRange:
     """
     document_type: str
     """The document type, as identified on given classification values."""
+    extraction_response: InferenceResponse
+    """The extraction response associated with the split."""
 
     def __init__(self, server_response: StringDict):
         self.page_range = server_response["page_range"]
         self.document_type = server_response["document_type"]
+        if server_response.get("extraction_response") is not None:
+            self.extraction_response = InferenceResponse(
+                server_response["extraction_response"]
+            )
 
     def __str__(self) -> str:
         page_range = ",".join([str(page_index) for page_index in self.page_range])

mindee/v2/product/split/split_response.py

@@ -1,7 +1,7 @@
 from mindee.input.sources.local_input_source import LocalInputSource
 from mindee.parsing.common.string_dict import StringDict
+from mindee.parsing.v2.base_response import BaseResponse
 from mindee.v2.file_operations.split_files import SplitFiles
-from mindee.v2.parsing.inference import BaseResponse
 from mindee.v2.product.split.split_inference import SplitInference
 
 

tests/input/test_apply_page_options.py

@@ -3,7 +3,7 @@
 import pypdfium2 as pdfium
 import pytest
 
-from mindee.error import MindeeError
+from mindee.error.mindee_error import MindeeError
 from mindee.input.page_options import KEEP_ONLY, REMOVE, PageOptions
 from mindee.input.sources import (
     Base64Input,

tests/v2/file_operations/test_crop_operation_integration.py

@@ -51,8 +51,8 @@ def test_image_should_extract_crops():
     )
     check_findoc_return(invoice_0)
     extracted_images.save_all_to_disk(OUTPUT_DIR)
-    assert os.path.getsize(OUTPUT_DIR / "crop_001.jpg") == 198887
-    assert os.path.getsize(OUTPUT_DIR / "crop_002.jpg") == 197443
+    assert os.path.getsize(OUTPUT_DIR / "crop_001.jpg") == 187601
+    assert os.path.getsize(OUTPUT_DIR / "crop_002.jpg") == 197978
 
 
 @pytest.fixture(scope="module", autouse=True)

tests/v2/product/classification/test_classification_integration.py

@@ -24,14 +24,14 @@ def test_classification_default_sample(
     v2_client: ClientV2, classification_model_id: str
 ):
     input_source = PathInput(
-        V2_PRODUCT_DATA_DIR / "classification" / "default_invoice.jpg"
+        V2_PRODUCT_DATA_DIR / "classification" / "default_sample.jpg"
     )
     response = v2_client.enqueue_and_get_result(
         ClassificationResponse,
         input_source,
         ClassificationParameters(classification_model_id),
     )
     assert response.inference is not None
-    assert response.inference.file.name == "default_invoice.jpg"
+    assert response.inference.file.name == "default_sample.jpg"
     assert response.inference.result.classification
     assert response.inference.result.classification.document_type == "invoice"