fix polling & add tests

Author: sebastianMindee

mindee/client_v2.py

@@ -91,14 +91,14 @@ def parse_queued(
 
         :param queue_id: queue_id received from the API.
         """
-        logger.debug("Fetching from queue ''%s", queue_id)
+        logger.debug("Fetching from queue '%s'.", queue_id)
 
         response = self.mindee_api.document_queue_req_get(queue_id)
         if not is_valid_get_response(response):
             handle_error_v2(response.json())
 
         dict_response = response.json()
-        if dict_response.get("job"):
+        if "job" in dict_response:
             return PollingResponse(dict_response)
         return InferenceResponse(dict_response)
 

mindee/mindee_http/mindee_api_v2.py

@@ -82,28 +82,23 @@ def predict_async_req_post(
         :return: requests response.
         """
         data = {"model_id": options.model_id}
-        params = {}
         url = f"{self.url_root}/inferences/enqueue"
 
         if options.full_text:
-            params["full_text_ocr"] = "true"
+            data["full_text_ocr"] = "true"
         if options.rag:
-            params["rag"] = "true"
+            data["rag"] = "true"
         if options.webhook_ids and len(options.webhook_ids) > 0:
-            params["webhook_ids"] = ",".join(options.webhook_ids)
+            data["webhook_ids"] = ",".join(options.webhook_ids)
         if options.alias and len(options.alias):
             data["alias"] = options.alias
 
-        files = {
-            "file": input_source.read_contents(close_file)
-            + (input_source.file_mimetype,)
-        }
+        files = {"file": input_source.read_contents(close_file)}
         response = requests.post(
             url=url,
             files=files,
             headers=self.base_headers,
             data=data,
-            params=params,
             timeout=self.request_timeout,
         )
 
@@ -116,7 +111,7 @@ def document_queue_req_get(self, queue_id: str) -> requests.Response:
         :param queue_id: queue_id received from the API
         """
         return requests.get(
-            f"{self.url_root}/inferences/{queue_id}",
+            f"{self.url_root}/jobs/{queue_id}",
             headers=self.base_headers,
             timeout=self.request_timeout,
         )

mindee/mindee_http/response_validation_v2.py

@@ -36,6 +36,6 @@ def is_valid_get_response(response: requests.Response) -> bool:
     if not is_valid_sync_response(response):
         return False
     response_json = json.loads(response.content)
-    if not "inference" in response_json:
+    if not "inference" in response_json and not "job" in response_json:
         return False
     return True

mindee/parsing/v2/inference.py

@@ -1,3 +1,5 @@
+from typing import Optional
+
 from mindee.parsing.common.string_dict import StringDict
 from mindee.parsing.v2.inference_file import InferenceFile
 from mindee.parsing.v2.inference_model import InferenceModel
@@ -13,11 +15,14 @@ class Inference:
     """File info for the inference."""
     result: InferenceResult
     """Result of the inference."""
+    id: Optional[str]
+    """ID of the inference."""
 
     def __init__(self, raw_response: StringDict):
         self.model = InferenceModel(raw_response["model"])
         self.file = InferenceFile(raw_response["file"])
         self.result = InferenceResult(raw_response["result"])
+        self.id = raw_response["id"] if "id" in raw_response else None
 
     def __str__(self) -> str:
         return (

mindee/parsing/v2/inference_response.py

@@ -7,6 +7,7 @@ class InferenceResponse(CommonResponse):
     """Represent an inference response from Mindee V2 API."""
 
     inference: Inference
+    """Inference result."""
 
     def __init__(self, raw_response: StringDict) -> None:
         super().__init__(raw_response)

mindee/parsing/v2/polling_response.py

@@ -7,6 +7,7 @@ class PollingResponse(CommonResponse):
     """Represent an inference response from Mindee V2 API."""
 
     job: Job
+    """Job for the polling."""
 
     def __init__(self, raw_response: StringDict) -> None:
         super().__init__(raw_response)

tests/test_client_v2_integration.py

@@ -42,7 +42,7 @@ def test_parse_file_empty_multiple_pages_must_succeed(
     input_doc = v2_client.source_from_path(input_path)
     options = InferencePredictOptions(findoc_model_id)
 
-    response: InferenceResponse = v2_client.enqueue_and_parse(input_doc, options, False)
+    response: InferenceResponse = v2_client.enqueue_and_parse(input_doc, options)
 
     assert response is not None
     assert response.inference is not None
@@ -82,7 +82,7 @@ def test_parse_file_filled_single_page_must_succeed(
     assert response.inference.result is not None
     supplier_name = response.inference.result.fields["supplier_name"]
     assert supplier_name is not None
-    assert supplier_name.simple_field.value == "John Smith"
+    assert supplier_name.value == "John Smith"
 
 
 @pytest.mark.integration

tests/v2/test_inference_response.py

@@ -1,18 +1,23 @@
+import json
+
 import pytest
 
+from mindee import ClientV2, LocalResponse
 from mindee.parsing.common.string_dict import StringDict
 from mindee.parsing.v2 import (
     Inference,
+    InferenceFile,
+    InferenceModel,
     InferenceResponse,
-    InferenceResult,
     ListField,
     ObjectField,
     SimpleField,
 )
+from tests.test_inputs import V2_DATA_DIR
 
 
 @pytest.fixture
-def inference_json() -> StringDict:
+def inference_result_json() -> StringDict:
     return {
         "inference": {
             "model": {"id": "test-model-id"},
@@ -92,8 +97,8 @@ def inference_json() -> StringDict:
 
 
 @pytest.mark.v2
-def test_inference(inference_json):
-    inference_result = InferenceResponse(inference_json)
+def test_inference_response(inference_result_json):
+    inference_result = InferenceResponse(inference_result_json)
     assert isinstance(inference_result.inference, Inference)
     assert isinstance(
         inference_result.inference.result.fields.field_simple, SimpleField
@@ -158,3 +163,29 @@ def test_inference(inference_json):
         .value
         == "value_9"
     )
+
+
+@pytest.mark.v2
+def test_full_inference_response():
+    client_v2 = ClientV2("dummy")
+    load_response = client_v2.load_inference(
+        LocalResponse(V2_DATA_DIR / "products" / "financial_document" / "complete.json")
+    )
+
+    assert isinstance(load_response.inference, Inference)
+    assert load_response.inference.id == "12345678-1234-1234-1234-123456789abc"
+    assert isinstance(load_response.inference.result.fields.date, SimpleField)
+    assert load_response.inference.result.fields.date.value == "2019-11-02"
+    assert isinstance(load_response.inference.result.fields.taxes, ListField)
+    assert isinstance(load_response.inference.result.fields.taxes.items[0], ObjectField)
+    assert (
+        load_response.inference.result.fields.taxes.items[0].fields["base"].value
+        == 31.5
+    )
+
+    assert isinstance(load_response.inference.model, InferenceModel)
+    assert load_response.inference.model.id == "12345678-1234-1234-1234-123456789abc"
+
+    assert isinstance(load_response.inference.file, InferenceFile)
+    assert load_response.inference.file.name == "complete.jpg"
+    assert load_response.inference.file.alias == None