Include codespace id in primary key

codespace created_at as returned from github API is only second-precision
and so no reliable as PK alone. Documentation is unclear as to the scope of uniqueness of the id field returned by the API, so compounding with created_at is a way of hedging our bets.

metrics/github/github.py

@@ -110,6 +110,7 @@ def from_dict(cls, data, repo):
 
 @dataclass(frozen=True)
 class Codespace:
+    id: int
     repo: Repo
     user: str
     created_at: datetime.datetime
@@ -119,7 +120,7 @@ class Codespace:
     def from_dict(cls, repo, **kwargs):
         if "repo_name" in kwargs:
             del kwargs["repo_name"]
-        return cls(repo, **kwargs)
+        return cls(repo=repo, **kwargs)
 
 
 def codespaces(org):

metrics/github/metrics.py

@@ -80,6 +80,7 @@ def convert_issue_counts_to_metrics(counts):
 def get_codespaces_metrics(codespaces):
     return [
         {
+            "id": c.id,
             "created_at": c.created_at,
             "organisation": c.repo.org,
             "repo": c.repo.name,

metrics/github/query.py

@@ -166,6 +166,7 @@ def codespaces(org):
     codespaces = _client().rest_query("/orgs/{org}/codespaces", org=org)
     for codespace in codespaces:
         yield {
+            "id": codespace["id"],
             "user": codespace["owner"]["login"],
             "repo_name": codespace["repository"]["name"],
             "created_at": codespace["created_at"],

metrics/timescaledb/tables.py

@@ -7,6 +7,7 @@
 GitHubCodespaces = Table(
     "github_codespaces",
     metadata,
+    Column("id", Integer, primary_key=True),
     Column("created_at", TIMESTAMP(timezone=True), primary_key=True),
     Column("organisation", Text),
     Column("repo", Text),

tests/metrics/github/test_github.py

@@ -41,12 +41,14 @@ def test_codespaces(patch):
         {
             "opensafely": [
                 {
+                    "id": 1,
                     "user": "testuser",
                     "repo_name": "testrepo",
                     "created_at": datetime.datetime.now().isoformat(),
                     "last_used_at": datetime.datetime.now().isoformat(),
                 },
                 {
+                    "id": 2,
                     "user": "testuser",
                     "repo_name": "testrepo2",
                     "created_at": datetime.datetime.now().isoformat(),