✨ add support for direct url GET inference

Author: sebastianMindee

mindee/v2/client.py

@@ -84,6 +84,18 @@ def get_result(
 
         return self.mindee_api.get_result(response_type, inference_id)
 
+    def get_result_from_url(
+        self, response_type: type[TypeBaseResponse], url: str
+    ) -> TypeBaseResponse:
+        """
+        Get the result of an inference that was previously enqueued by its URL.
+
+        :param response_type: Type of the response to return.
+        :param url: URL of the inference to retrieve.
+        :return: The result of the inference.
+        """
+        return self.mindee_api.get_result_by_url(response_type, url)
+
     def enqueue_and_get_result(
         self,
         response_type: type[TypeBaseResponse],

mindee/v2/mindee_http/mindee_api_v2.py

@@ -131,6 +131,20 @@ def req_get_job(self, job_id: str) -> requests.Response:
             allow_redirects=False,
         )
 
+    def req_get_inference_by_url(self, url) -> requests.Response:
+        """
+        Sends a request matching a given inference_id. Returns either a Job or a Document.
+
+        :param url: URL to use for the request.
+        :return: Response object from the request.
+        """
+        return requests.get(
+            url,
+            headers=self.base_headers,
+            timeout=self.request_timeout,
+            allow_redirects=False,
+        )
+
     def req_get_inference(self, inference_id: str, slug: str) -> requests.Response:
         """
         Sends a request matching a given queue_id. Returns either a Job or a Document.
@@ -212,6 +226,20 @@ def get_result(self, response_type, inference_id: str):
             handle_error_v2(dict_response)
         return response_type(dict_response)
 
+    def get_result_by_url(self, response_type, url: str):
+        """
+        Get the result of an inference that was previously enqueued by its URL.
+
+        :param response_type: Type of the response to return.
+        :param url: URL of the inference to retrieve.
+        :return: The result of the inference.
+        """
+        response = self.req_get_inference_by_url(url)
+        dict_response = self._response_json(response)
+        if not is_valid_get_response(response):
+            handle_error_v2(dict_response)
+        return response_type(dict_response)
+
     def get_models(self, name: str | None, model_type: str | None):
         """
         Get a list of models matching the provided name and type.

tests/conftest.py

@@ -0,0 +1,9 @@
+import gc
+
+import pytest
+
+
+@pytest.fixture(autouse=True)
+def force_gc():
+    yield
+    gc.collect()

tests/v1/input/test_url_input_source_integration.py

@@ -21,32 +21,35 @@ def output_file_path():
 
 @pytest.fixture
 def reference_file_path():
-    return "https://github.com/mindee/client-lib-test-data/blob/main/v1/products/invoice_splitter/invoice_5p.pdf?raw=true"
+    ref_path = os.getenv("MINDEE_V2_SE_TESTS_BLANK_PDF_URL")
+    if ref_path is None:
+        raise ValueError("MINDEE_V2_SE_TESTS_BLANK_PDF_URL not set")
+    return ref_path
 
 
 @pytest.mark.integration
 def test_load_local_file(client, reference_file_path):
     url_source = URLInputSource(reference_file_path)
     local_source = url_source.as_local_input_source()
     result = client.parse(InvoiceV4, local_source)
-    assert result.document.n_pages == 5
-    assert result.document.filename == "invoice_5p.pdf"
+    assert result.document.n_pages == 1
+    assert result.document.filename == "blank_1.pdf"
 
 
 @pytest.mark.integration
 def test_custom_file_name(client, reference_file_path):
     url_source = URLInputSource(reference_file_path)
     local_source = url_source.as_local_input_source("customName.pdf")
     result = client.parse(InvoiceV4, local_source)
-    assert result.document.n_pages == 5
+    assert result.document.n_pages == 1
     assert result.document.filename == "customName.pdf"
 
 
 @pytest.mark.integration
 def test_save_file(client, reference_file_path, output_file_path):
     url_source = URLInputSource(reference_file_path)
     url_source.save_to_file(output_file_path)
-    assert os.path.exists(os.path.join(output_file_path, "invoice_5p.pdf"))
+    assert os.path.exists(os.path.join(output_file_path, "blank_1.pdf"))
 
 
 @pytest.mark.integration
@@ -59,4 +62,4 @@ def test_save_file_with_filename(client, reference_file_path, output_file_path):
 @pytest.fixture(autouse=True)
 def cleanup():
     yield
-    cleanup_output_files(["invoice_5p.pdf", "customFileName.pdf"])
+    cleanup_output_files(["blank_1.pdf", "customFileName.pdf"])

tests/v2/test_client.py

@@ -184,6 +184,15 @@ def test_get_inference(custom_base_url_client):
     _assert_findoc_inference(response)
 
 
+@pytest.mark.v2
+def test_get_inference_by_url(custom_base_url_client):
+    response = custom_base_url_client.get_result(
+        ExtractionResponse,
+        "https://api-v2.mindee.net/v2/inference/12345678-1234-1234-1234-123456789ABC",
+    )
+    _assert_findoc_inference(response)
+
+
 @pytest.mark.v2
 def test_error_handling(custom_base_url_client):
     with pytest.raises(MindeeHTTPErrorV2) as e: