📝 update docs for RAG search

Author: ianardee

docs/client_mixin.rst

@@ -1,3 +1,6 @@
+from mindee.error import MindeeClientError
+
+
 class PollingOptions:
     """Options for asynchronous polling."""
 
@@ -17,3 +20,20 @@ def __init__(
         self.initial_delay_sec = initial_delay_sec
         self.delay_sec = delay_sec
         self.max_retries = max_retries
+
+    def validate_settings(self) -> None:
+        """Validates polling options against minimum accepted values."""
+
+        min_delay = 1
+        min_initial_delay = 1
+        min_retries = 1
+        if self.delay_sec < min_delay:
+            raise MindeeClientError(
+                f"Cannot set auto-parsing delay to less than {min_delay} second(s)."
+            )
+        if self.initial_delay_sec < min_initial_delay:
+            raise MindeeClientError(
+                f"Cannot set initial parsing delay to less than {min_initial_delay} second(s)."
+            )
+        if self.max_retries < min_retries:
+            raise MindeeClientError(f"Cannot set retries to less than {min_retries}.")

mindee/client_mixin.py

@@ -2,7 +2,7 @@
 
 import httpx
 
-from mindee.client_mixin import ClientMixin
+from mindee.client_options.polling_options import PollingOptions
 from mindee.error.mindee_error import MindeeClientError, MindeeError
 from mindee.error.mindee_http_error import handle_error
 from mindee.input.local_input_source import LocalInputSource
@@ -53,7 +53,7 @@ def _clean_account_name(account_name: str) -> str:
     return account_name
 
 
-class Client(ClientMixin):
+class Client:
     """
     Mindee API Client.
 
@@ -353,7 +353,11 @@ def enqueue_and_parse(  # pylint: disable=too-many-locals
         :param rag: If set, will enable Retrieval-Augmented Generation.
             Only works if a valid ``workflow_id`` is set.
         """
-        self._validate_async_params(initial_delay_sec, delay_sec, max_retries)
+        PollingOptions(
+            initial_delay_sec=initial_delay_sec,
+            delay_sec=delay_sec,
+            max_retries=max_retries,
+        ).validate_settings()
         if not endpoint:
             endpoint = self._initialize_ots_endpoint(product_class=product_class)
         queue_result = self.enqueue(

mindee/client_options/polling_options.py

@@ -4,7 +4,6 @@
 
 import httpx
 
-from mindee.client_mixin import ClientMixin
 from mindee.client_options.polling_options import PollingOptions
 from mindee.error.mindee_error import MindeeError
 from mindee.input import URLInputSource
@@ -27,7 +26,7 @@
 )
 
 
-class Client(ClientMixin):
+class Client:
     """
     Mindee API Client.
 
@@ -127,11 +126,7 @@ def enqueue_and_get_result(
         """
         if not params.polling_options:
             params.polling_options = PollingOptions()
-        self._validate_async_params(
-            params.polling_options.initial_delay_sec,
-            params.polling_options.delay_sec,
-            params.polling_options.max_retries,
-        )
+        params.polling_options.validate_settings()
         enqueue_response = self.enqueue(input_source, params)
         logger.debug(
             "Successfully enqueued document with job ID: %s", enqueue_response.job.id

mindee/v1/client.py

@@ -9,13 +9,20 @@
 
 @dataclass(kw_only=True)
 class RagDocumentSearchParameters(BaseSearchParameters[RagDocumentSearchResponse]):
-    """Search parameters for RAG Documents."""
+    """
+    Search for RAG documents within the organization linked to the API key.
+
+    The model ID is required, search filters are optional.
+    If no search filters are given, all documents linked to the model are returned.
+
+    Results are paginated.
+    """
 
     model_id: str
-    """Model identifier to search in."""
+    """The exact Model UUID the document is linked to."""
 
     filename: str | None = None
-    """Case-insensitive substring search on filename."""
+    """Filter documents by partial filename match, case-insensitive."""
 
     _slug: ClassVar[str] = "rag-documents"
     _response_class: type[RagDocumentSearchResponse] = RagDocumentSearchResponse

mindee/v2/client.py

@@ -11,7 +11,7 @@ def v2_client() -> Client:
 
 @pytest.mark.integration
 @pytest.mark.v2
-def test_must_have_results(v2_client: Client):
+def test_search_must_have_results(v2_client: Client):
     response = v2_client.search(ModelSearchParameters())
 
     assert response is not None
@@ -27,7 +27,7 @@ def test_must_have_results(v2_client: Client):
 
 @pytest.mark.integration
 @pytest.mark.v2
-def test_must_return_empty(v2_client: Client):
+def test_search_must_return_empty(v2_client: Client):
     response = v2_client.search(ModelSearchParameters(name="je n'existe pas tralala"))
 
     assert response is not None

mindee/v2/search/rag_documents/rag_document_search_parameters.py

@@ -3,25 +3,49 @@
 import pytest
 
 from mindee.v2.client import Client
-from mindee.v2.search.rag_documents.rag_document_search_parameters import (
-    RagDocumentSearchParameters,
-)
+from mindee.v2.search.rag_documents import RagDocumentSearchParameters
 
 
 @pytest.fixture(scope="session")
 def v2_client() -> Client:
     return Client()
 
 
-@pytest.mark.integration
-@pytest.mark.v2
-def test_must_have_results(v2_client: Client):
+@pytest.fixture(scope="session")
+def findoc_model_id() -> str:
     findoc_model_id = os.getenv("MINDEE_V2_SE_TESTS_FINDOC_MODEL_ID")
     assert findoc_model_id, "MINDEE_V2_SE_TESTS_FINDOC_MODEL_ID must be set"
+    return findoc_model_id
+
+
+@pytest.mark.integration
+@pytest.mark.v2
+def test_search_must_have_results(v2_client: Client, findoc_model_id: str):
     response = v2_client.search(RagDocumentSearchParameters(model_id=findoc_model_id))
 
     assert response is not None
     assert len(response.rag_documents) > 0
+    for rag_doc in response.rag_documents:
+        assert rag_doc.id
+        assert rag_doc.created_at
+        assert rag_doc.filename
+        assert rag_doc.total_matches >= 0
     assert response.pagination is not None
     assert response.pagination.total_items >= 1
     assert response.pagination.page == 1
+
+
+@pytest.mark.integration
+@pytest.mark.v2
+def test_search_must_return_empty(v2_client: Client, findoc_model_id: str):
+    response = v2_client.search(
+        RagDocumentSearchParameters(
+            model_id=findoc_model_id, filename="invoice_32GB-RAM_450k-USD.pdf"
+        )
+    )
+
+    assert response is not None
+    assert len(response.rag_documents) == 0
+    assert response.pagination is not None
+    assert response.pagination.total_items == 0
+    assert response.pagination.page == 1