From 5e8fc42629362cad8270caf81b3dd1ab3dad9519 Mon Sep 17 00:00:00 2001
From: Jon Massey <jon.massey@thedatalab.org>
Date: Tue, 21 May 2024 21:50:55 +0100
Subject: [PATCH] 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            | 3 ++-
 metrics/github/metrics.py           | 1 +
 metrics/github/query.py             | 1 +
 metrics/timescaledb/tables.py       | 1 +
 tests/metrics/github/test_github.py | 2 ++
 5 files changed, 7 insertions(+), 1 deletion(-)

diff --git a/metrics/github/github.py b/metrics/github/github.py
index 9651206..6e3872f 100644
--- a/metrics/github/github.py
+++ b/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):
diff --git a/metrics/github/metrics.py b/metrics/github/metrics.py
index cfcfa66..7204911 100644
--- a/metrics/github/metrics.py
+++ b/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,
diff --git a/metrics/github/query.py b/metrics/github/query.py
index 3f13a11..e03cd71 100644
--- a/metrics/github/query.py
+++ b/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"],
diff --git a/metrics/timescaledb/tables.py b/metrics/timescaledb/tables.py
index 1b639a1..dad0008 100644
--- a/metrics/timescaledb/tables.py
+++ b/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),
diff --git a/tests/metrics/github/test_github.py b/tests/metrics/github/test_github.py
index e4ed5c6..f4b54e2 100644
--- a/tests/metrics/github/test_github.py
+++ b/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(),