feat(open-pr-comments): get sentry projects and filenames (#59952)

Author: cathteng

src/sentry/tasks/integrations/github/open_pr_comment.py

@@ -1,9 +1,14 @@
 from __future__ import annotations
 
 import logging
-from typing import List
+from typing import List, Set, Tuple
+
+from django.db.models import Value
+from django.db.models.functions import StrIndex
 
 from sentry.integrations.github.client import GitHubAppsClient
+from sentry.models.integrations.repository_project_path_config import RepositoryProjectPathConfig
+from sentry.models.project import Project
 from sentry.models.pullrequest import PullRequest
 from sentry.models.repository import Repository
 from sentry.shared_integrations.exceptions.base import ApiError
@@ -44,7 +49,7 @@ def safe_for_comment(
                 OPEN_PR_METRIC_BASE.format(key="api_error"),
                 tags={"type": GithubAPIErrorType.UNKNOWN.value, "code": e.code},
             )
-            logger.exception("github.open_pr_comment.unknown_api_error")
+            logger.exception("github.open_pr_comment.unknown_api_error", extra={"error": str(e)})
         return False
 
     safe_to_comment = True
@@ -74,3 +79,26 @@ def get_pr_filenames(
     # new files will not have sentry issues associated with them
     pr_filenames: List[str] = [file["filename"] for file in pr_files if file["status"] != "added"]
     return pr_filenames
+
+
+def get_projects_and_filenames_from_source_file(
+    org_id: int,
+    pr_filename: str,
+) -> Tuple[Set[Project], Set[str]]:
+    # fetch the code mappings in which the source_root is a substring at the start of pr_filename
+    code_mappings = (
+        RepositoryProjectPathConfig.objects.filter(organization_id=org_id)
+        .annotate(substring_match=StrIndex(Value(pr_filename), "source_root"))
+        .filter(substring_match=1)
+    )
+
+    project_list: Set[Project] = set()
+    sentry_filenames = set()
+
+    if len(code_mappings):
+        for code_mapping in code_mappings:
+            project_list.add(code_mapping.project)
+            sentry_filenames.add(
+                pr_filename.replace(code_mapping.source_root, code_mapping.stack_root)
+            )
+    return project_list, sentry_filenames

tests/sentry/tasks/integrations/github/test_open_pr_comment.py

@@ -2,7 +2,11 @@
 
 import responses
 
-from sentry.tasks.integrations.github.open_pr_comment import get_pr_filenames, safe_for_comment
+from sentry.tasks.integrations.github.open_pr_comment import (
+    get_pr_filenames,
+    get_projects_and_filenames_from_source_file,
+    safe_for_comment,
+)
 from sentry.testutils.silo import region_silo_test
 from sentry.testutils.skips import requires_snuba
 from tests.sentry.tasks.integrations.github.test_pr_comment import GithubCommentTestCase
@@ -144,7 +148,7 @@ def setUp(self):
         self.gh_client = installation.get_client()
 
     @responses.activate
-    def test_simple(self):
+    def test_get_pr_filenames(self):
         responses.add(
             responses.GET,
             self.gh_path.format(pull_number=self.pr.key),
@@ -157,3 +161,58 @@ def test_simple(self):
         )
 
         assert set(get_pr_filenames(self.gh_client, self.gh_repo, self.pr)) == {"bar.py", "baz.py"}
+
+    def test_get_projects_and_filenames_from_source_file(self):
+        projects = [self.create_project() for _ in range(4)]
+
+        source_stack_pairs = [
+            ("", "./"),
+            ("src/sentry", "sentry/"),
+            ("src/", ""),
+            ("src/sentry/", "sentry/"),
+        ]
+        for i, pair in enumerate(source_stack_pairs):
+            source_root, stack_root = pair
+            self.create_code_mapping(
+                project=projects[i],
+                repo=self.gh_repo,
+                source_root=source_root,
+                stack_root=stack_root,
+                default_branch="master",
+            )
+
+        # matching code mapping from a different org
+        other_org_code_mapping = self.create_code_mapping(
+            project=self.another_org_project,
+            repo=self.another_org_repo,
+            source_root="",
+            stack_root="./",
+        )
+        other_org_code_mapping.organization_id = self.another_organization.id
+        other_org_code_mapping.save()
+
+        source_stack_nonmatches = [
+            ("/src/sentry", "sentry"),
+            ("tests/", "tests/"),
+            ("app/", "static/app"),
+        ]
+        for source_root, stack_root in source_stack_nonmatches:
+            self.create_code_mapping(
+                project=self.create_project(),
+                repo=self.gh_repo,
+                source_root=source_root,
+                stack_root=stack_root,
+                default_branch="master",
+            )
+
+        filename = "src/sentry/tasks/integrations/github/open_pr_comment.py"
+        correct_filenames = [
+            filename.replace(source_root, stack_root)
+            for source_root, stack_root in source_stack_pairs
+        ]
+
+        project_list, sentry_filenames = get_projects_and_filenames_from_source_file(
+            self.organization.id, filename
+        )
+        assert project_list == set(projects)
+        assert sentry_filenames == set(correct_filenames)