PTFE-1547 Add branch and event label to prometheus metrics

gh_actions_exporter/config.py

@@ -42,7 +42,7 @@ class Settings(BaseSettings):
         "ubuntu-20.04": 0.008,
     }
     default_cost: Optional[float] = 0
-
+    branches: List[str] = ["main", "master"]
     check_runs_enabled: bool = False
     github_app_id: Optional[int]
     github_app_installation_id: Optional[int]

gh_actions_exporter/metrics.py

@@ -1,10 +1,11 @@
+import fnmatch
 from typing import Dict, List
 
 from prometheus_client import Counter, Histogram
 
 from gh_actions_exporter.config import Relabel, RelabelType, Settings
 from gh_actions_exporter.cost import Cost
-from gh_actions_exporter.types import WebHook, WorkflowJob
+from gh_actions_exporter.types import WebHook, WorkflowJob, WorkflowRun
 
 
 class Metrics(object):
@@ -17,7 +18,10 @@ def __init__(self, settings: Settings):
             "workflow_name",
             "repository_visibility",
         ]
-        self.workflow_labelnames = self.common_labelnames.copy()
+        self.workflow_labelnames = self.common_labelnames.copy() + [
+            "branch",
+            "event",
+        ]
         self.job_labelnames = self.common_labelnames.copy() + [
             "job_name",
             "runner_type",
@@ -107,11 +111,42 @@ def __init__(self, settings: Settings):
             labelnames=self.job_labelnames,
         )
 
+    def retrieve_branch(self, workflow_run: WorkflowRun) -> str:
+        """
+        Add the branch label to the metrics exposed by the exporter while
+        also taking into account the cardinality limitations of prometheus:
+
+        - For workflows triggered on pull_request event
+          retrieve the base branch (target branch).
+
+        - For other events, retrieve the head branch.
+
+        Then check if the branch matches any of the patterns defined in
+        self.settings.branches like: development/*, main, feature/*, etc.
+
+        Return the branch name if it matches any of the patterns,
+        otherwise return "dev".
+        """
+        ref: str
+        if workflow_run.event == "pull_request":
+            assert workflow_run.pull_requests
+            ref = workflow_run.pull_requests[0].base.ref
+        else:
+            ref = workflow_run.head_branch
+        for branch in self.settings.branches:
+            if fnmatch.fnmatch(ref, branch):
+                return ref
+        return "dev"
+
     def workflow_labels(self, webhook: WebHook) -> dict:
+        assert webhook.workflow_run
+        branch = self.retrieve_branch(webhook.workflow_run)
         return dict(
             workflow_name=webhook.workflow_run.name,
             repository=webhook.repository.full_name,
             repository_visibility=webhook.repository.visibility,
+            branch=branch,
+            event=webhook.workflow_run.event,
         )
 
     def runner_type(self, webhook: WebHook) -> str:

gh_actions_exporter/types.py

@@ -34,6 +34,15 @@ class WorkflowJob(BaseModel):
     steps: Optional[list[Steps]] = None
 
 
+class BasePullRequest(BaseModel):
+    ref: str
+
+
+class PullRequest(BaseModel):
+    number: int
+    base: BasePullRequest
+
+
 class WorkflowRun(BaseModel):
     id: int
     name: str
@@ -44,6 +53,7 @@ class WorkflowRun(BaseModel):
     workflow_id: int
     run_number: int
     head_branch: str
+    pull_requests: Optional[list[PullRequest]] = None
     created_at: datetime
     updated_at: Optional[datetime] = None
     run_attempt: int

tests/api/conftest.py

@@ -34,6 +34,31 @@ def workflow_run():
             "created_at": "2021-11-16T17:52:47Z",
             "run_started_at": "2021-11-16T17:52:47Z",
             "updated_at": "2021-11-16T17:53:32Z",
+            "pull_requests": [
+                {
+                    "url": "https://api.github.com/repos/org/repo/pulls/555",
+                    "id": 1805247,
+                    "number": 555,
+                    "head": {
+                        "ref": "feature/my-feature",
+                        "sha": "fc98b69edf5ac3f4789c210e758b1ee4b4396b9e",
+                        "repo": {
+                            "id": 395065315,
+                            "url": "https://api.github.com/repos/org/repo",
+                            "name": "repo",
+                        },
+                    },
+                    "base": {
+                        "ref": "master",
+                        "sha": "966917b16d284edde596b0705a49f8c65ecf1c84",
+                        "repo": {
+                            "id": 395065125,
+                            "url": "https://api.github.com/repos/org/repo",
+                            "name": "repo",
+                        },
+                    },
+                }
+            ],
         },
         "repository": {
             "name": "repo",

tests/api/metrics/conftest.py

@@ -27,7 +27,9 @@ def job_relabel_config():
 
 @lru_cache()
 def default_settings():
-    return Settings()
+    return Settings(
+        branches=["main", "master", "development/*"],
+    )
 
 
 @lru_cache()

tests/api/metrics/test_workflow.py

@@ -54,3 +54,41 @@ def test_rebuild(client, workflow_run, headers):
     for line in metrics.text.split("\n"):
         if "workflow_rebuild_count_total{" in line:
             assert "1.0" in line
+
+
+def test_branch_label(client, workflow_run, headers):
+    # Ensure a feature branch is not indexed
+    upload = client.post("/webhook", json=workflow_run, headers=headers)
+    assert upload.status_code == 202
+
+    metrics = client.get("/metrics")
+    assert metrics.status_code == 200
+    assert 'branch="feature"' not in metrics.text
+    assert 'branch="dev"' in metrics.text
+
+    # Index a push event with a branch label
+    workflow_run["workflow_run"]["head_branch"] = "main"
+    upload = client.post("/webhook", json=workflow_run, headers=headers)
+    assert upload.status_code == 202
+
+    metrics = client.get("/metrics")
+    assert metrics.status_code == 200
+    assert 'branch="main"' in metrics.text
+
+    # Index a branch with a fnmatch pattern configured
+    workflow_run["workflow_run"]["head_branch"] = "development/1.2"
+    upload = client.post("/webhook", json=workflow_run, headers=headers)
+    assert upload.status_code == 202
+
+    metrics = client.get("/metrics")
+    assert metrics.status_code == 200
+    assert 'branch="development/1.2"' in metrics.text
+
+    # Index the base branch of a pull request
+    workflow_run["workflow_run"]["event"] = "pull_request"
+    upload = client.post("/webhook", json=workflow_run, headers=headers)
+    assert upload.status_code == 202
+
+    metrics = client.get("/metrics")
+    assert metrics.status_code == 200
+    assert 'branch="master"' in metrics.text