From 3b7b084419921803b7e0d59129dbeb2e457eedaf Mon Sep 17 00:00:00 2001
From: Jon Massey <jon.massey@thedatalab.org>
Date: Fri, 17 May 2024 11:54:53 +0100
Subject: [PATCH] Add task to fetch and store codespaces metrics

As the codespaces data will change over time and deleted codespaces
will disappear from the API, we do not drop and recreate the table
each time the task is run as per the other tasks. Instead, calling
the upsert() method ensures the table exists then merges the new
data with the existing.

Conversion method added to metrics.py is not a metric in the usual
sense of the word, but instead renames some fields to match the
database schema. This is neccesary due to different naming conventions
in the domain dataclasses and database tables.
---
 metrics/github/metrics.py   | 13 +++++++++++++
 metrics/tasks/codespaces.py | 24 ++++++++++++++++++++++++
 2 files changed, 37 insertions(+)
 create mode 100644 metrics/tasks/codespaces.py

diff --git a/metrics/github/metrics.py b/metrics/github/metrics.py
index 569ab2e1..c8f5fd61 100644
--- a/metrics/github/metrics.py
+++ b/metrics/github/metrics.py
@@ -75,3 +75,16 @@ def convert_issue_counts_to_metrics(counts):
             }
         )
     return metrics
+
+
+def convert_codespaces_to_dicts(codespaces):
+    return [
+        {
+            "organisation": c.org,
+            "repo": c.repo_name,
+            "user": c.user,
+            "created_at": c.created_at,
+            "last_used_at": c.last_used_at,
+        }
+        for c in codespaces
+    ]
diff --git a/metrics/tasks/codespaces.py b/metrics/tasks/codespaces.py
new file mode 100644
index 00000000..38159fb7
--- /dev/null
+++ b/metrics/tasks/codespaces.py
@@ -0,0 +1,24 @@
+import sys
+
+import structlog
+
+import metrics.github.github as github
+from metrics.github.metrics import convert_codespaces_to_dicts
+from metrics.timescaledb import db, tables
+
+
+log = structlog.get_logger()
+
+
+def main():
+    log.info("Getting codespaces")
+    codespaces = github.codespaces(org="opensafely")
+    log.info(f"Got {len(codespaces)} codespaces")
+
+    log.info("Writing data")
+    db.upsert(tables.GitHubCodespaces, convert_codespaces_to_dicts(codespaces))
+    log.info("Written data")
+
+
+if __name__ == "__main__":
+    sys.exit(main())