feat(integrations): instrument pyreqwest tracing (#5682)

### Description

Implements request tracing for the
[`pyreqwest`](https://github.com/MarkusSintonen/pyreqwest) HTTP library.
Includes tests and CI updates.

#### Issues

Resolves MarkusSintonen/pyreqwest#30.

---------

Co-authored-by: Ivana Kellyer <ivana.kellyer@sentry.io>

Author: servusdei2018

.github/workflows/test-integrations-network.yml

@@ -61,6 +61,10 @@ jobs:
         run: |
           set -x # print commands that are executed
           ./scripts/runtox.sh "py${{ matrix.python-version }}-httpx"
+      - name: Test pyreqwest
+        run: |
+          set -x # print commands that are executed
+          ./scripts/runtox.sh "py${{ matrix.python-version }}-pyreqwest"
       - name: Test requests
         run: |
           set -x # print commands that are executed

scripts/populate_tox/config.py

@@ -308,6 +308,14 @@
         "deps": {
             "*": ["pytest-asyncio"],
         },
+        "python": ">=3.10",
+    },
+    "pyreqwest": {
+        "package": "pyreqwest",
+        "deps": {
+            "*": ["pytest-asyncio"],
+        },
+        "python": ">=3.11",
     },
     "pymongo": {
         "package": "pymongo",

scripts/populate_tox/package_dependencies.jsonl

@@ -124,6 +124,7 @@
     "Network": [
         "grpc",
         "httpx",
+        "pyreqwest",
         "requests",
     ],
     "Tasks": [

scripts/populate_tox/releases.jsonl

@@ -150,6 +150,7 @@ def iter_default_integrations(
     "openfeature": (0, 7, 1),
     "pydantic_ai": (1, 0, 0),
     "pymongo": (3, 5, 0),
+    "pyreqwest": (0, 11, 6),
     "quart": (0, 16, 0),
     "ray": (2, 7, 0),
     "requests": (2, 0, 0),

scripts/split_tox_gh_actions/split_tox_gh_actions.py

@@ -0,0 +1,136 @@
+import sentry_sdk
+from sentry_sdk import start_span
+from sentry_sdk.consts import OP, SPANDATA
+from sentry_sdk.integrations import Integration, DidNotEnable
+from sentry_sdk.tracing import BAGGAGE_HEADER_NAME
+from sentry_sdk.tracing_utils import (
+    should_propagate_trace,
+    add_http_request_source,
+    add_sentry_baggage_to_headers,
+)
+from sentry_sdk.utils import (
+    SENSITIVE_DATA_SUBSTITUTE,
+    capture_internal_exceptions,
+    logger,
+    parse_url,
+)
+
+from contextlib import contextmanager
+from typing import Any, Generator
+
+try:
+    from pyreqwest.client import ClientBuilder, SyncClientBuilder  # type: ignore[import-not-found]
+    from pyreqwest.request import (  # type: ignore[import-not-found]
+        Request,
+        OneOffRequestBuilder,
+        SyncOneOffRequestBuilder,
+    )
+    from pyreqwest.middleware import Next, SyncNext  # type: ignore[import-not-found]
+    from pyreqwest.response import Response, SyncResponse  # type: ignore[import-not-found]
+except ImportError:
+    raise DidNotEnable("pyreqwest not installed or incompatible version installed")
+
+
+class PyreqwestIntegration(Integration):
+    identifier = "pyreqwest"
+    origin = f"auto.http.{identifier}"
+
+    @staticmethod
+    def setup_once() -> None:
+        _patch_pyreqwest()
+
+
+def _patch_pyreqwest() -> None:
+    # Patch Client Builders
+    _patch_builder_method(ClientBuilder, "build", sentry_async_middleware)
+    _patch_builder_method(SyncClientBuilder, "build", sentry_sync_middleware)
+
+    # Patch Request Builders
+    _patch_builder_method(OneOffRequestBuilder, "send", sentry_async_middleware)
+    _patch_builder_method(SyncOneOffRequestBuilder, "send", sentry_sync_middleware)
+
+
+def _patch_builder_method(cls: type, method_name: str, middleware: "Any") -> None:
+    if not hasattr(cls, method_name):
+        return
+
+    original_method = getattr(cls, method_name)
+
+    def sentry_patched_method(self: "Any", *args: "Any", **kwargs: "Any") -> "Any":
+        if not getattr(self, "_sentry_instrumented", False):
+            integration = sentry_sdk.get_client().get_integration(PyreqwestIntegration)
+            if integration is not None:
+                self.with_middleware(middleware)
+                try:
+                    self._sentry_instrumented = True
+                except (TypeError, AttributeError):
+                    # In case the instance itself is immutable or doesn't allow extra attributes
+                    pass
+        return original_method(self, *args, **kwargs)
+
+    setattr(cls, method_name, sentry_patched_method)
+
+
+@contextmanager
+def _sentry_pyreqwest_span(request: "Request") -> "Generator[Any, None, None]":
+    parsed_url = None
+    with capture_internal_exceptions():
+        parsed_url = parse_url(str(request.url), sanitize=False)
+
+    with start_span(
+        op=OP.HTTP_CLIENT,
+        name=f"{request.method} {parsed_url.url if parsed_url else SENSITIVE_DATA_SUBSTITUTE}",
+        origin=PyreqwestIntegration.origin,
+    ) as span:
+        span.set_data(SPANDATA.HTTP_METHOD, request.method)
+        if parsed_url is not None:
+            span.set_data("url", parsed_url.url)
+            span.set_data(SPANDATA.HTTP_QUERY, parsed_url.query)
+            span.set_data(SPANDATA.HTTP_FRAGMENT, parsed_url.fragment)
+
+        if should_propagate_trace(sentry_sdk.get_client(), str(request.url)):
+            for (
+                key,
+                value,
+            ) in sentry_sdk.get_current_scope().iter_trace_propagation_headers():
+                logger.debug(
+                    "[Tracing] Adding `{key}` header {value} to outgoing request to {url}.".format(
+                        key=key, value=value, url=request.url
+                    )
+                )
+
+                if key == BAGGAGE_HEADER_NAME:
+                    add_sentry_baggage_to_headers(request.headers, value)
+                else:
+                    request.headers[key] = value
+
+        yield span
+
+    with capture_internal_exceptions():
+        add_http_request_source(span)
+
+
+async def sentry_async_middleware(
+    request: "Request", next_handler: "Next"
+) -> "Response":
+    if sentry_sdk.get_client().get_integration(PyreqwestIntegration) is None:
+        return await next_handler.run(request)
+
+    with _sentry_pyreqwest_span(request) as span:
+        response = await next_handler.run(request)
+        span.set_http_status(response.status)
+
+    return response
+
+
+def sentry_sync_middleware(
+    request: "Request", next_handler: "SyncNext"
+) -> "SyncResponse":
+    if sentry_sdk.get_client().get_integration(PyreqwestIntegration) is None:
+        return next_handler.run(request)
+
+    with _sentry_pyreqwest_span(request) as span:
+        response = next_handler.run(request)
+        span.set_http_status(response.status)
+
+    return response

sentry_sdk/integrations/__init__.py

@@ -0,0 +1,9 @@
+import os
+import sys
+import pytest
+
+pytest.importorskip("pyreqwest")
+
+# Load `pyreqwest_helpers` into the module search path to test request source path names relative to module. See
+# `test_request_source_with_module_in_search_path`
+sys.path.insert(0, os.path.join(os.path.dirname(__file__)))

sentry_sdk/integrations/pyreqwest.py

@@ -0,0 +1,2 @@
+def get_request_with_client(client, url):
+    client.get(url).build().send()