✨ add support for custom httpx client for V1 and V2 (#415)

Author: sebastianMindee

mindee/error/mindee_http_error.py

@@ -1,5 +1,5 @@
 from mindee.error.mindee_error import MindeeError
-from mindee.parsing.common import StringDict
+from mindee.parsing.common.string_dict import StringDict
 
 
 class MindeeHTTPError(RuntimeError):

mindee/input/url_input_source.py

@@ -10,6 +10,7 @@
 from mindee.error.mindee_error import MindeeSourceError
 from mindee.input.bytes_input import BytesInput
 from mindee.logger import logger
+from mindee.parsing.common.string_dict import StringDict
 
 
 class URLInputSource:
@@ -173,7 +174,14 @@ def __fill_filename(self, filename=None) -> str:
         return filename
 
     @staticmethod
-    def __make_request(url, auth, headers, redirects, max_redirects) -> bytes:
+    def __make_request(
+        url,
+        auth,
+        headers,
+        redirects,
+        max_redirects,
+        http_client: httpx.Client | None = None,
+    ) -> bytes:
         """
         Makes an HTTP request to the given URL, while following redirections.
 
@@ -185,20 +193,36 @@ def __make_request(url, auth, headers, redirects, max_redirects) -> bytes:
         :return: The content of the response.
         :raises MindeeSourceError: If max redirects are exceeded or the request fails.
         """
-        result = httpx.get(url, headers=headers, timeout=120, auth=auth)
+        get_kwargs: StringDict = {
+            "headers": headers,
+            "timeout": 120,
+            "auth": auth,
+            "follow_redirects": True,
+        }
+        if http_client is None:
+            result = httpx.get(url, **get_kwargs)
+        else:
+            result = http_client.get(url, **get_kwargs)
         if 299 < result.status_code < 400:
             if redirects == max_redirects:
                 raise MindeeSourceError(
-                    f"Can't reach URL after {redirects} out of {max_redirects} redirects, "
+                    f"Can't reach URL after {redirects} out of {max_redirects} "
+                    f"redirects, "
                     f"aborting operation."
                 )
             return URLInputSource.__make_request(
-                result.headers["Location"], auth, headers, redirects + 1, max_redirects
+                result.headers["Location"],
+                auth,
+                headers,
+                redirects + 1,
+                max_redirects,
+                http_client,
             )
 
         if result.status_code >= 400 or result.status_code < 200:
             raise MindeeSourceError(
                 f"Couldn't retrieve file from server, error code {result.status_code}."
             )
-
+        if http_client is not None and not http_client.is_closed:
+            http_client.close()
         return result.content

mindee/mindee_http/response_validation.py

@@ -2,7 +2,7 @@
 
 import httpx
 
-from mindee.parsing.common import StringDict
+from mindee.parsing.common.string_dict import StringDict
 
 
 def is_valid_sync_response(response: httpx.Response) -> bool:

mindee/mindee_http/settings_mixin.py

@@ -3,7 +3,7 @@ class SettingsMixin:
 
     base_url: str
     """Base URL for all V2 requests."""
-    request_timeout: int
+    request_timeout: float
     """Timeout for all requests."""
 
     def set_timeout(self, value: str | int) -> None:

mindee/v1/client.py

@@ -1,5 +1,7 @@
 from time import sleep
 
+import httpx
+
 from mindee.client_mixin import ClientMixin
 from mindee.error.mindee_error import MindeeClientError, MindeeError
 from mindee.error.mindee_http_error import handle_error
@@ -59,14 +61,21 @@ class Client(ClientMixin):
     """
 
     api_key: str
+    """API key for all endpoints."""
+    http_client: httpx.Client | None
+    """HTTP client for making requests."""
 
-    def __init__(self, api_key: str = "") -> None:
+    def __init__(
+        self, api_key: str = "", http_client: httpx.Client | None = None
+    ) -> None:
         """
         Mindee API Client.
 
         :param api_key: Your API key for all endpoints
+        :param http_client: HTTP client for making requests.
         """
         self.api_key = api_key
+        self.http_client = http_client
 
     def parse(
         self,
@@ -522,7 +531,8 @@ def _send_to_workflow(
             raise MindeeClientError("No input document provided")
 
         workflow_endpoint = WorkflowEndpoint(
-            WorkflowSettings(api_key=self.api_key, workflow_id=workflow_id)
+            WorkflowSettings(api_key=self.api_key, workflow_id=workflow_id),
+            self.http_client,
         )
 
         response = workflow_endpoint.workflow_execution_post(input_source, options)
@@ -555,8 +565,12 @@ def _build_endpoint(
             version=version,
         )
         if account_name and len(account_name) > 0 and account_name != "mindee":
-            return CustomEndpoint(endpoint_name, account_name, version, api_settings)
-        return Endpoint(endpoint_name, account_name, version, api_settings)
+            return CustomEndpoint(
+                endpoint_name, account_name, version, api_settings, self.http_client
+            )
+        return Endpoint(
+            endpoint_name, account_name, version, api_settings, self.http_client
+        )
 
     def create_endpoint(
         self,
@@ -583,3 +597,20 @@ def create_endpoint(
             )
             version = "1"
         return self._build_endpoint(endpoint_name, account_name, version)
+
+    def close(self):
+        """Close the HTTP client."""
+        if self.http_client and not self.http_client.is_closed:
+            self.http_client.close()
+
+    def __enter__(self):
+        return self
+
+    def __exit__(self, exc_type, exc_val, exc_tb):
+        self.close()
+
+    def __del__(self):
+        """Ensure the HTTP client is closed when the object is garbage collected."""
+        if self.http_client and self.http_client and not self.http_client.is_closed:
+            logger.info("Force-closing unclosed Mindee Client (V1) %s.", str(self))
+            self.close()

mindee/v1/mindee_http/base_endpoint.py

@@ -1,13 +1,24 @@
+import httpx
+
 from mindee.v1.mindee_http.base_settings import BaseSettings
 
 
 class BaseEndpoint:
     """Base endpoint class for the Mindee API."""
 
-    def __init__(self, settings: BaseSettings) -> None:
+    settings: BaseSettings
+    """Settings relating to all endpoints."""
+    http_client: httpx.Client | None
+    """HTTP client for making requests."""
+
+    def __init__(
+        self, settings: BaseSettings, http_client: httpx.Client | None = None
+    ) -> None:
         """
         Base API endpoint class for all endpoints.
 
         :param settings: Settings relating to all endpoints.
+        :param http_client: HTTP client for making requests.
         """
         self.settings = settings
+        self.http_client = http_client