fix(codecov): Fix git blame for multi repo projects (#44827)

In some projects where there are more than one repository associated to a project, we get a commit that comes from the release. For instance, in the case of the sentry project, we get the commit from the getsentry repo, thus, we can't use that commit with the sentry repo.

Fixes:

* Even if a commit is set, run the git blame experiment
  * As mentioned above, the commit may not even be for the right repo
* Report an error if we fail to git blame

Author: armenzg

src/sentry/api/endpoints/project_stacktrace_link.py

@@ -206,7 +206,8 @@ def get_latest_commit_sha_from_blame(
             git_blame_list.sort(key=lambda blame: blame["commit"]["committedDate"])
             commit_sha = git_blame_list[-1]["commit"]["oid"]
         if not commit_sha:
-            logger.warning(
+            # Report to Sentry so we can investigate
+            logger.error(
                 "Failed to get commit from git blame.",
                 extra={
                     "git_blame_response": git_blame_list,
@@ -219,7 +220,6 @@ def get_latest_commit_sha_from_blame(
 
     def fetch_codecov_data(
         self,
-        has_error_commit: bool,
         ref: Optional[str],
         integrations: BaseQuerySet,
         org: Organization,
@@ -238,8 +238,9 @@ def fetch_codecov_data(
         )
         # Get commit sha from Git blame if valid
         gh_integrations = integrations.filter(provider="github")
-        should_get_commit_sha = fetch_commit_sha and gh_integrations and not has_error_commit
+        should_get_commit_sha = fetch_commit_sha and gh_integrations
 
+        ref_source = "from_release"
         if should_get_commit_sha:
             try:
                 integration_installation = gh_integrations[0].get_installation(
@@ -252,11 +253,15 @@ def fetch_codecov_data(
                     repo,
                     branch,
                 )
+                ref_source = "from_git_blame"
             except Exception:
                 logger.exception(
                     "Failed to get commit sha from git blame, pending investigation. Continuing execution."
                 )
 
+        with configure_scope() as scope:
+            scope.set_tag("codecov.ref_source", ref_source)
+
         # Call codecov API if codecov-commit-sha-from-git-blame flag is not enabled
         # or getting ref from git blame was successful
         codecov_data = None
@@ -267,7 +272,6 @@ def fetch_codecov_data(
                 ref=ref if ref else branch,
                 ref_type="sha" if ref else "branch",
                 path=path,
-                has_error_commit=has_error_commit,
             )
             if lineCoverage and codecovUrl:
                 codecov_data = {
@@ -374,8 +378,7 @@ def get(self, request: Request, project: Project) -> Response:
                     try:
                         line_no = ctx.get("line_no")
                         codecov_data = self.fetch_codecov_data(
-                            has_error_commit := bool(ctx.get("commit_id")),
-                            ref=ctx["commit_id"] if has_error_commit else None,
+                            ref=ctx.get("commit_id"),
                             integrations=integrations,
                             org=project.organization,
                             user=request.user,

src/sentry/integrations/utils/codecov.py

@@ -11,7 +11,7 @@
 
 
 def get_codecov_data(
-    repo: str, service: str, ref: str, ref_type: REF_TYPE, path: str, has_error_commit: bool
+    repo: str, service: str, ref: str, ref_type: REF_TYPE, path: str
 ) -> Tuple[Optional[LineCoverage], Optional[str]]:
     codecov_token = options.get("codecov.client-secret")
     line_coverage = None
@@ -34,7 +34,6 @@ def get_codecov_data(
                 "codecov.request_path": path,
                 "codecov.request_ref": ref,
                 "codecov.http_code": response.status_code,
-                "codecov.ref_source": "from_release" if has_error_commit else "from_git_blame",
             }
 
             response_json = response.json()

tests/sentry/api/endpoints/test_project_stacktrace_link.py

@@ -369,10 +369,10 @@ def test_get_latest_commit_from_blame(self, get_jwt, mock_blame):
     @mock.patch("sentry.integrations.mixins.repositories.RepositoryMixin.get_stacktrace_link")
     @mock.patch("sentry.integrations.github.client.GitHubClientMixin.get_blame_for_file")
     @mock.patch("sentry.integrations.github.client.get_jwt", return_value=b"jwt_token_1")
-    def test_get_commit_sha_from_blame_logger_warning(
+    def test_get_commit_sha_from_blame_failed_to_get_commit(
         self, get_jwt, mock_blame, mock_get_stacktrace_link
     ):
-        self._caplog.set_level(logging.WARNING, logger="sentry")
+        self._caplog.set_level(logging.ERROR, logger="sentry")
         self.organization.flags.codecov_access = True
         self.organization.save()
         self.integration.provider = "github"
@@ -389,10 +389,11 @@ def test_get_commit_sha_from_blame_logger_warning(
             qs_params={"file": self.filepath, "lineNo": 26},
         )
 
+        # Error logging means an error is sent to Sentry
         assert self._caplog.record_tuples == [
             (
                 "sentry.api.endpoints.project_stacktrace_link",
-                logging.WARNING,
+                logging.ERROR,
                 "Failed to get commit from git blame.",
             )
         ]