[ENH] use httpx over requests (#2336)

.pre-commit-config.yaml

@@ -45,7 +45,6 @@ repos:
           ]
         additional_dependencies:
           [
-            "types-requests",
             "pydantic",
             "overrides",
             "hypothesis",

bin/windows_upgrade_sqlite.py

@@ -1,4 +1,4 @@
-import requests
+import httpx
 import zipfile
 import io
 import os
@@ -10,7 +10,7 @@
 
 if __name__ == "__main__":
     # Download and extract the DLL
-    r = requests.get(DLL_URL)
+    r = httpx.get(DLL_URL)
     z = zipfile.ZipFile(io.BytesIO(r.content))
     z.extractall(".")
     # Print current Python path

chromadb/api/async_fastapi.py

@@ -2,11 +2,10 @@
 from uuid import UUID
 import urllib.parse
 import orjson as json
-from typing import Any, Optional, TypeVar, cast, Tuple, Sequence, Dict
+from typing import Any, Optional, cast, Tuple, Sequence, Dict
 import logging
 import httpx
 from overrides import override
-from chromadb import errors
 from chromadb.api import AsyncServerAPI
 from chromadb.api.base_http_client import BaseHTTPClient
 from chromadb.config import DEFAULT_DATABASE, DEFAULT_TENANT, System, Settings
@@ -43,15 +42,6 @@
 )
 
 
-# requests removes None values from the built query string, but httpx includes it as an empty value
-T = TypeVar("T", bound=Dict[Any, Any])
-
-
-def clean_params(params: T) -> T:
-    """Remove None values from kwargs."""
-    return {k: v for k, v in params.items() if v is not None}  # type: ignore
-
-
 logger = logging.getLogger(__name__)
 
 
@@ -130,8 +120,8 @@ async def _make_request(
         escaped_path = urllib.parse.quote(path, safe="/", encoding=None, errors=None)
         url = self._api_url + escaped_path
 
-        response = await self._get_client().request(method, url, **kwargs)
-        await raise_chroma_error(response)
+        response = await self._get_client().request(method, url, **cast(Any, kwargs))
+        BaseHTTPClient._raise_chroma_error(response)
         return json.loads(response.text)
 
     @trace_method("AsyncFastAPI.heartbeat", OpenTelemetryGranularity.OPERATION)
@@ -202,7 +192,7 @@ async def list_collections(
         resp_json = await self._make_request(
             "get",
             "/collections",
-            params=clean_params(
+            params=BaseHTTPClient._clean_params(
                 {
                     "tenant": tenant,
                     "database": database,
@@ -605,30 +595,3 @@ async def get_max_batch_size(self) -> int:
             resp_json = await self._make_request("get", "/pre-flight-checks")
             self._max_batch_size = cast(int, resp_json["max_batch_size"])
         return self._max_batch_size
-
-
-async def raise_chroma_error(resp: httpx.Response) -> Any:
-    """Raises an error if the response is not ok, using a ChromaError if possible."""
-    try:
-        resp.raise_for_status()
-        return
-    except httpx.HTTPStatusError:
-        pass
-
-    chroma_error = None
-    try:
-        body = json.loads(resp.text)
-        if "error" in body:
-            if body["error"] in errors.error_types:
-                chroma_error = errors.error_types[body["error"]](body["message"])
-
-    except BaseException:
-        pass
-
-    if chroma_error:
-        raise chroma_error
-
-    try:
-        resp.raise_for_status()
-    except httpx.HTTPStatusError:
-        raise (Exception(resp.text))

chromadb/api/base_http_client.py

@@ -1,7 +1,10 @@
-from typing import Optional
+from typing import Any, Dict, Optional, TypeVar
 from urllib.parse import quote, urlparse, urlunparse
 import logging
+import orjson as json
+import httpx
 
+import chromadb.errors as errors
 from chromadb.config import Settings
 
 logger = logging.getLogger(__name__)
@@ -57,3 +60,38 @@ def resolve_url(
         )
 
         return full_url
+
+    # requests removes None values from the built query string, but httpx includes it as an empty value
+    T = TypeVar("T", bound=Dict[Any, Any])
+
+    @staticmethod
+    def _clean_params(params: T) -> T:
+        """Remove None values from provided dict."""
+        return {k: v for k, v in params.items() if v is not None}  # type: ignore
+
+    @staticmethod
+    def _raise_chroma_error(resp: httpx.Response) -> None:
+        """Raises an error if the response is not ok, using a ChromaError if possible."""
+        try:
+            resp.raise_for_status()
+            return
+        except httpx.HTTPStatusError:
+            pass
+
+        chroma_error = None
+        try:
+            body = json.loads(resp.text)
+            if "error" in body:
+                if body["error"] in errors.error_types:
+                    chroma_error = errors.error_types[body["error"]](body["message"])
+
+        except BaseException:
+            pass
+
+        if chroma_error:
+            raise chroma_error
+
+        try:
+            resp.raise_for_status()
+        except httpx.HTTPStatusError:
+            raise (Exception(resp.text))

chromadb/api/client.py

@@ -2,7 +2,7 @@
 from uuid import UUID
 
 from overrides import override
-import requests
+import httpx
 from chromadb.api import AdminAPI, ClientAPI, ServerAPI
 from chromadb.api.shared_system_client import SharedSystemClient
 from chromadb.api.types import (
@@ -349,7 +349,7 @@ def set_database(self, database: str) -> None:
     def _validate_tenant_database(self, tenant: str, database: str) -> None:
         try:
             self._admin_client.get_tenant(name=tenant)
-        except requests.exceptions.ConnectionError:
+        except httpx.ConnectError:
             raise ValueError(
                 "Could not connect to a Chroma server. Are you sure it is running?"
             )
@@ -363,7 +363,7 @@ def _validate_tenant_database(self, tenant: str, database: str) -> None:
 
         try:
             self._admin_client.get_database(name=database, tenant=tenant)
-        except requests.exceptions.ConnectionError:
+        except httpx.ConnectError:
             raise ValueError(
                 "Could not connect to a Chroma server. Are you sure it is running?"
             )