♻️ fix syntaxes for V2, bump test lib (#335)

Author: sebastianMindee

docs/extras/code_samples/default_v2.txt

@@ -1,10 +1,5 @@
-#
-# Install the Python client library by running:
-# pip install mindee
-#
-
 from mindee import ClientV2, InferencePredictOptions
-from mindee.parsing.v2 import InferenceResponse, PollingResponse
+from mindee.parsing.v2 import InferenceResponse
 
 input_path = "/path/to/the/file.ext"
 api_key = "MY_API_KEY"
@@ -13,12 +8,21 @@ model_id = "MY_MODEL_ID"
 # Init a new client
 mindee_client = ClientV2(api_key)
 
+# Set inference options
+options = InferencePredictOptions(
+    # ID of the model, required.
+    model_id=model_id,
+    # If set to `True`, will enable Retrieval-Augmented Generation.
+    rag=False,
+)
+
 # Load a file from disk
 input_doc = mindee_client.source_from_path(input_path)
-options = InferencePredictOptions(model_id=model_id)
 
-# Parse the file.
-response: InferenceResponse = mindee_client.enqueue_and_parse(input_doc, options)
+# Upload the file
+response: InferenceResponse = mindee_client.enqueue_and_parse(
+    input_doc, options
+)
 
 # Print a brief summary of the parsed data
-print(response.inference)
+print(response.inference)

mindee/client_v2.py

@@ -15,7 +15,7 @@
     is_valid_post_response,
 )
 from mindee.parsing.v2.inference_response import InferenceResponse
-from mindee.parsing.v2.polling_response import PollingResponse
+from mindee.parsing.v2.job_response import JobResponse
 
 
 class ClientV2(ClientMixin):
@@ -39,7 +39,7 @@ def __init__(self, api_key: Optional[str] = None) -> None:
 
     def enqueue(
         self, input_source: LocalInputSource, options: InferencePredictOptions
-    ) -> PollingResponse:
+    ) -> JobResponse:
         """
         Enqueues a document to a given model.
 
@@ -66,12 +66,12 @@ def enqueue(
         if not is_valid_post_response(response):
             handle_error_v2(dict_response)
 
-        return PollingResponse(dict_response)
+        return JobResponse(dict_response)
 
     def parse_queued(
         self,
         queue_id: str,
-    ) -> Union[InferenceResponse, PollingResponse]:
+    ) -> Union[InferenceResponse, JobResponse]:
         """
         Parses a queued document.
 
@@ -85,7 +85,7 @@ def parse_queued(
 
         dict_response = response.json()
         if "job" in dict_response:
-            return PollingResponse(dict_response)
+            return JobResponse(dict_response)
         return InferenceResponse(dict_response)
 
     def enqueue_and_parse(
@@ -118,7 +118,7 @@ def enqueue_and_parse(
             queue_result.job.id,
         )
         while retry_counter < options.polling_options.max_retries:
-            if not isinstance(poll_results, PollingResponse):
+            if not isinstance(poll_results, JobResponse):
                 break
             if poll_results.job.status == "Failed":
                 raise MindeeError(f"Parsing failed for job {poll_results.job.id}")

mindee/parsing/v2/__init__.py

@@ -13,4 +13,4 @@
 from mindee.parsing.v2.inference_response import InferenceResponse
 from mindee.parsing.v2.inference_result import InferenceResult
 from mindee.parsing.v2.job import Job
-from mindee.parsing.v2.polling_response import PollingResponse
+from mindee.parsing.v2.job_response import JobResponse

mindee/parsing/v2/inference_options.py

@@ -1,23 +1,14 @@
 from typing import List
 
 from mindee.parsing.common.string_dict import StringDict
-
-
-class RawText:
-    """Raw text extracted from the document."""
-
-    page: int
-    content: str
-
-    def __init__(self, raw_response: StringDict):
-        self.page = raw_response["page"]
-        self.content = raw_response["content"]
+from mindee.parsing.v2.raw_text import RawText
 
 
 class InferenceOptions:
     """Optional information about the document."""
 
     raw_texts: List[RawText]
+    """List of text found per page."""
 
     def __init__(self, raw_response: StringDict):
         self.raw_texts = [RawText(raw_text) for raw_text in raw_response["raw_texts"]]

mindee/parsing/v2/job.py

@@ -12,6 +12,8 @@ class Job:
     """Job ID."""
     error: Optional[ErrorResponse]
     """Error response if any."""
+    created_at: datetime
+    """Timestamp of the job creation."""
     model_id: str
     """ID of the model."""
     filename: str

mindee/parsing/v2/polling_response.py mindee/parsing/v2/job_response.pymindee/parsing/v2/polling_response.py renamed to mindee/parsing/v2/job_response.py

@@ -3,7 +3,7 @@
 from mindee.parsing.v2.job import Job
 
 
-class PollingResponse(CommonResponse):
+class JobResponse(CommonResponse):
     """Represent an inference response from Mindee V2 API."""
 
     job: Job

mindee/parsing/v2/raw_text.py

@@ -0,0 +1,14 @@
+from mindee.parsing.common.string_dict import StringDict
+
+
+class RawText:
+    """Raw text extracted from the document."""
+
+    page: int
+    """Page the raw text was found on."""
+    content: str
+    """Content of the raw text."""
+
+    def __init__(self, raw_response: StringDict):
+        self.page = raw_response["page"]
+        self.content = raw_response["content"]

tests/data

@@ -7,7 +7,7 @@
 from mindee.error.mindee_http_error_v2 import MindeeHTTPErrorV2
 from mindee.input import LocalInputSource, PathInput
 from mindee.mindee_http.base_settings import USER_AGENT
-from mindee.parsing.v2 import Job, PollingResponse
+from mindee.parsing.v2 import Job, JobResponse
 from tests.test_inputs import FILE_TYPES_DIR, V2_DATA_DIR
 from tests.utils import dummy_envvars
 
@@ -137,7 +137,7 @@ def test_enqueue(custom_base_url_client):
     response = custom_base_url_client.parse_queued(
         "12345678-1234-1234-1234-123456789ABC"
     )
-    assert isinstance(response, PollingResponse)
+    assert isinstance(response, JobResponse)
     assert isinstance(response.job, Job)
     assert response.job.id == "12345678-1234-1234-1234-123456789ABC"
     assert response.job.model_id == "87654321-4321-4321-4321-CBA987654321"