[do not merge] Run CI on merge commit

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

@@ -17,7 +17,7 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        python-version: ["3.8","3.9","3.10","3.11","3.12","3.13","3.14","3.14t"]
+        python-version: ["3.8","3.9","3.10","3.11","3.12","3.13","3.14","3.14t","3.15","3.15t"]
         os: [ubuntu-22.04]
     env:
       # 3.7 runs in the python:X.Y container; tell uv to use that system Python.
@@ -57,6 +57,10 @@ jobs:
         run: |
           set -x # print commands that are executed
           ./scripts/runtox.sh "py${{ matrix.python-version }}-litellm"
+      - name: Test mistral
+        run: |
+          set -x # print commands that are executed
+          ./scripts/runtox.sh "py${{ matrix.python-version }}-mistral"
       - name: Test openai-base
         run: |
           set -x # print commands that are executed

CHANGELOG.md

@@ -1,5 +1,80 @@
 # Changelog
 
+## 2.70.0
+
+### New Features ✨
+
+- **New integration:** Mistral AI (#4733) by @alexander-alderman-webb
+
+  Add the Mistral integration to your `sentry_sdk.init` call:
+
+```python
+import sentry_sdk
+from sentry_sdk.integrations.mistral import MistralIntegration
+
+sentry_sdk.init(
+    dsn="...",
+    traces_sample_rate=1.0,
+    integrations=[
+        MistralIntegration(),
+    ]
+)
+```
+
+- **New option:** A fine-grained way to configure what data should be sent from the SDK.
+
+  Offers much more flexibility than `send_default_pii`, allowing you to specify what
+  data should be collected. If `data_collection` is defined, it takes precedence over
+  `send_default_pii`.
+
+  See https://docs.sentry.io/platforms/python/configuration/options/#data_collection
+  for the available configuration options.
+
+```python
+import sentry_sdk
+
+sentry_sdk.init(
+    data_collection={
+        "user_info": False,
+        "gen_ai": {"inputs": False, "outputs": False},
+        "graphql": {"document": False, "variables": False},
+        "database_query_data": False,
+        "queues": False,
+        "http_bodies": [],
+        "cookies": {
+            "mode": "denylist",
+            "terms": ["forwarded", "-ip", "remote-", "via", "-user"],
+        },
+        "http_headers": {
+            "request": {
+                "mode": "denylist",
+                "terms": ["forwarded", "-ip", "remote-", "via", "-user"],
+            },
+        },
+        "url_query_params": {
+            "mode": "denylist",
+            "terms": ["forwarded", "-ip", "remote-", "via", "-user"],
+        },
+    }
+)
+```
+
+#### Mistral
+
+- Record `gen_ai.output.messages` by @alexander-alderman-webb in [#7549](https://github.com/getsentry/sentry-python/pull/7549)
+- Record `gen_ai.input.messages` by @alexander-alderman-webb in [#7548](https://github.com/getsentry/sentry-python/pull/7548)
+- Record `gen_ai.system_instructions` by @alexander-alderman-webb in [#7547](https://github.com/getsentry/sentry-python/pull/7547)
+- Record request parameters by @alexander-alderman-webb in [#7531](https://github.com/getsentry/sentry-python/pull/7531)
+- Record token usage by @alexander-alderman-webb in [#7529](https://github.com/getsentry/sentry-python/pull/7529)
+- Add integration with `Chat.complete` and `Chat.complete_async` patches by @alexander-alderman-webb in [#7528](https://github.com/getsentry/sentry-python/pull/7528)
+
+### Internal Changes 🔧
+
+- (data-collection) Promote `data_collection` from `_experiments` o top-level by @ericapisani in [#7560](https://github.com/getsentry/sentry-python/pull/7560)
+- Close client on teardown by @alexander-alderman-webb in [#7562](https://github.com/getsentry/sentry-python/pull/7562)
+- 🤖 Update test matrix with new releases (09/21) by @github-actions in [#7546](https://github.com/getsentry/sentry-python/pull/7546)
+- Remove `getsentry/testing-ai-sdk-integrations` by @alexander-alderman-webb in [#7526](https://github.com/getsentry/sentry-python/pull/7526)
+
 ## 2.69.2
 
 ### Bug Fixes 🐛

docs/conf.py

@@ -31,7 +31,7 @@
 copyright = "2019-{}, Sentry Team and Contributors".format(datetime.now().year)
 author = "Sentry Team and Contributors"
 
-release = "2.69.2"
+release = "2.70.0"
 version = ".".join(release.split(".")[:2])  # The short X.Y version.
 
 

pyproject.toml

@@ -80,6 +80,7 @@ typing = [
     "tiktoken>=0.14.0",
     "aiohttp>=3.14.3",
     "openai-agents>=0.22.0",
+    "mistralai>=2.10.1",
 ]
 test = [
     "dataclasses ; python_full_version < '3.7'",

scripts/populate_tox/config.py

@@ -303,6 +303,12 @@
             "*": ["pytest-asyncio", "httpx"],
         },
     },
+    "mistral": {
+        "package": "mistralai",
+        "deps": {
+            "*": ["pytest-asyncio", "httpx"],
+        },
+    },
     "fastmcp": {
         "package": "fastmcp",
         "deps": {

scripts/split_tox_gh_actions/split_tox_gh_actions.py

@@ -92,6 +92,7 @@
         "google_genai",
         "huggingface_hub",
         "litellm",
+        "mistral",
         "openai-base",
         "openai-notiktoken",
     ],

sentry_sdk/_types.py

@@ -458,6 +458,14 @@ class ToolCallPart(TypedDict):
         name: NotRequired[str]
         arguments: NotRequired[Any]
 
+    class InputMessage(TypedDict):
+        role: Literal["user", "assistant", "tool"]
+        parts: list[Union[TextPart, ReasoningPart, ToolCallPart]]
+
+    class OutputMessage(TypedDict):
+        role: Literal["assistant"]
+        parts: list[Union[TextPart, ReasoningPart]]
+
     class ToolDefinition(TypedDict):
         type: str
         name: NotRequired[str]

sentry_sdk/consts.py

@@ -1167,12 +1167,12 @@
     """
     A list of names identifying enabled integrations.
     Example: ["AtexitIntegration", "StdlibIntegration"]
     """
 
 
 class OP:
     ANTHROPIC_MESSAGES_CREATE = "ai.messages.create.anthropic"
     CACHE_GET = "cache.get"
     CACHE_PUT = "cache.put"
     COHERE_CHAT_COMPLETIONS_CREATE = "ai.chat_completions.create.cohere"
     COHERE_EMBEDDINGS_CREATE = "ai.embeddings.create.cohere"
@@ -1265,6 +1265,7 @@
         transport_queue_size: int = DEFAULT_QUEUE_SIZE,
         sample_rate: float = 1.0,
         send_default_pii: "Optional[bool]" = None,
+        data_collection: "Optional[DataCollectionUserOptions]" = None,
         http_proxy: "Optional[str]" = None,
         https_proxy: "Optional[str]" = None,
         ignore_errors: "Sequence[Union[type, str]]" = [],  # noqa: B006
@@ -1275,19 +1276,19 @@
         before_breadcrumb: "Optional[BreadcrumbProcessor]" = None,
         debug: "Optional[bool]" = None,
         attach_stacktrace: bool = True,
         ca_certs: "Optional[str]" = None,
         traces_sample_rate: "Optional[float]" = None,
         traces_sampler: "Optional[TracesSampler]" = None,
         profiler_mode: "Optional[ContinuousProfilerMode]" = None,
         profile_lifecycle: 'Literal["manual", "trace"]' = "manual",
         profile_session_sample_rate: "Optional[float]" = None,
         auto_enabling_integrations: bool = True,
         disabled_integrations: "Optional[Sequence[Union[sentry_sdk.integrations.Integration, type[sentry_sdk.integrations.Integration]]]]" = None,
         auto_session_tracking: bool = True,
         send_client_reports: bool = True,
         _experiments: "Experiments" = {},  # noqa: B006
         proxy_headers: "Optional[Dict[str, str]]" = None,
         project_root: "Optional[str]" = None,
         include_local_variables: "Optional[bool]" = True,
         include_source_context: "Optional[bool]" = True,
         trace_propagation_targets: "Optional[Sequence[str]]" = [  # noqa: B006
@@ -1410,6 +1411,23 @@
             If you enable this option, be sure to manually remove what you don't want to send using our features for
             managing `Sensitive Data <https://docs.sentry.io/data-management/sensitive-data/>`_.
 
+        :param data_collection: Structured configuration controlling what data integrations collect
+            automatically, superseding `send_default_pii`. Passing a dict opts into the feature; omitted
+            fields use their defaults (most categories are collected, with the sensitive denylist
+            scrubbing values). When it is not set, the SDK derives behaviour from `send_default_pii` so
+            that upgrading changes nothing. Restrict collection per category (user identity, cookies,
+            HTTP headers/bodies, query params, generative AI inputs/outputs, stack frame variables,
+            source context). If `send_default_pii` is also set, `data_collection` takes precedence.
+
+            Example::
+
+                sentry_sdk.init(
+                    dsn="...",
+                    data_collection={"user_info": False, "http_bodies": []},
+                )
+
+            See https://docs.sentry.io/platforms/python/configuration/options/#data_collection for more details.
+
         :param event_scrubber: Scrubs the event payload for sensitive information such as cookies, sessions, and
             passwords from a `denylist`.
 
@@ -1703,24 +1721,6 @@
         :param ignore_spans: A sequence of span-matching rules. Matching spans are ignored.
 
         :param _experiments: Dictionary of experimental, opt-in features that are not yet stable.
-
-            ``data_collection`` (EXPERIMENTAL): structured configuration controlling what data integrations
-            collect automatically, superseding `send_default_pii`. Passing a dict under
-            `_experiments={"data_collection": {...}}` opts into the feature; omitted fields use their
-            defaults (most categories are collected, with the sensitive denylist scrubbing values).
-            When it is not set, the SDK derives behaviour from `send_default_pii` so that upgrading
-            changes nothing. Restrict collection per category (user identity, cookies, HTTP
-            headers/bodies, query params, generative AI inputs/outputs, stack frame variables, source
-            context). If `send_default_pii` is also set, `data_collection` takes precedence.
-
-            Example::
-
-                sentry_sdk.init(
-                    dsn="...",
-                    _experiments={"data_collection": {"user_info": False, "http_bodies": []}},
-                )
-
-            See https://docs.sentry.io/platforms/python/configuration/options/#data_collection for more details.
         """
         pass
 
@@ -1744,4 +1744,4 @@
 del _get_default_options
 
 
-VERSION = "2.69.2"
+VERSION = "2.70.0"

sentry_sdk/data_collection.py

@@ -292,10 +292,17 @@ def _resolve_data_collection(options: "Dict[str, Any]") -> "DataCollection":
     concrete values for every field.
 
     ``data_collection`` must be a plain ``dict``.
+
+    Must be called exactly once per options dict, before ``client._get_options``
+    overwrites ``options["data_collection"]`` with the resolved result. Feeding an
+    already-resolved dict back in would flip ``provided_by_user`` to ``True``.
     """
     from sentry_sdk.utils import deprecation_warning
 
-    user_dc = options.get("_experiments", {}).get("data_collection")
+    user_dc = options.get("data_collection")
+    if user_dc is None:
+        user_dc = options.get("_experiments", {}).get("data_collection")
+
     send_default_pii = options.get("send_default_pii")
 
     include_local_variables = (

sentry_sdk/integrations/__init__.py

@@ -154,6 +154,7 @@
     "litellm": (1, 77, 5),
     "loguru": (0, 7, 0),
     "mcp": (2, 0),
+    "mistral": (2, 0, 5),
     "openai": (1, 0, 0),
     "openai_agents": (0, 10, 3),
     "openfeature": (0, 7, 1),
@@ -217,9 +218,9 @@
         ):
             if integration_cls.identifier not in integrations:
                 instance = integration_cls()
                 integrations[instance.identifier] = instance
                 used_as_default_integration.add(instance.identifier)
 
     for identifier, integration in integrations.items():
         with _installer_lock:
             if identifier not in _processed_integrations: