From 73dc9fe50ea486c2fa7bc9e1909077fb8ecb428d Mon Sep 17 00:00:00 2001
From: "Brian C. Lane" <bcl@redhat.com>
Date: Thu, 10 Aug 2023 15:59:05 -0700
Subject: [PATCH] tools: Add helper script to check Schutzfile and
 gitlab-ci.yml runners

This script reads the RUNNER entries from .gitlab-ci.yml and the
snapshot repo names from Schutzfile. It then prints which runners don't
have entries in Schutzfile and which Schutzfile entries don't have
runners.

This is slightly complicated by the fact that centos-stream reports
itself as centos. So when checking for missing entries they are
considered to be the same thing. eg. If centos is a RUNNER and
centos-stream is in Schutzfile it doesn't report anything missing.
---
 tools/check-runners | 117 ++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 117 insertions(+)
 create mode 100755 tools/check-runners

diff --git a/tools/check-runners b/tools/check-runners
new file mode 100755
index 000000000..113ed3208
--- /dev/null
+++ b/tools/check-runners
@@ -0,0 +1,117 @@
+#!/usr/bin/python3
+"""check-runners compares the list of runners in .gitlab-ci.yml to the list of
+snapshots in Schutzfile and prints a list of runners without Schutzfile entries,
+and a list of Schutzfile entries not used by runners."""
+
+import json
+import sys
+import yaml
+
+
+def read_gitlab_ci(path):
+    """Read the gitlab-ci.yml file and parse it into a yaml object
+    """
+    with open(path, encoding="utf8") as f:
+        data = f.read()
+
+    # Handle all !tags (like gitlab's !reference) by just returning the raw value
+    # the result isn't used, this is just to keep it from returning an error when
+    # it hits an unknown tag.
+    yaml.SafeLoader.add_multi_constructor('', lambda a, b, c: c.value)
+    return yaml.safe_load(data)
+
+
+def all_runners(gitlab):
+    """Extract all the RUNNER entries from the gitlab yaml"""
+    runners = []
+    for k in gitlab.keys():
+        if k == "RUNNER":
+            if isinstance(gitlab[k], list):
+                runners += gitlab[k]
+            else:
+                runners.append(gitlab[k])
+        elif isinstance(gitlab[k], dict):
+            runners += all_runners(gitlab[k])
+        elif isinstance(gitlab[k], list):
+            # Lists can have dicts inside them
+            for e in gitlab[k]:
+                if isinstance(e, dict):
+                    runners += all_runners(e)
+    return runners
+
+
+def runner_distros(runners):
+    """Best guess of distro versions for the runners
+    They start with SERVICE/ so that can be ignored
+    Then have DISTRO-VERSION-EXTRA-ARCH where DISTRO might be 'centos-stream' or a single word.
+
+    This should be fairly stable, but it's possible it will fail in the future.
+    """
+    distros = []
+    for r in runners:
+        try:
+            r = r.split("/", 1)[1]
+        except IndexError:
+            print("ERROR: Cannot guess {r}")
+            continue
+        if r.startswith("centos-stream"):
+            distros.append("-".join(r.split("-")[:3]))
+        else:
+            distros.append("-".join(r.split("-")[:2]))
+    return set(distros)
+
+
+def read_Schutzfile(path):
+    """Read the Schutzfile json file and return a dict"""
+    with open(path, encoding="utf8") as f:
+        return json.load(f)
+
+
+def get_distros(path):
+    """Read the Schutzfile and return a list of the distro versions it contains"""
+    sf = read_Schutzfile(path)
+    return set(k for k in sf.keys() if k != "global")
+
+
+def unused(a, b):
+    """Find the snapshots from set a that are not in set b,
+    while also checking for centos <---> centos-stream name aliasing
+    """
+    unused_snapshots = []
+    for s in a:
+        if s in b:
+            continue
+        if s.startswith("centos-stream"):
+            # centos and centos-stream both use 'centos' as the distro name
+            # so when one shows up as missing, check the other name too
+            t = s.replace("centos-stream", "centos")
+        elif s.startswith("centos"):
+            t = s.replace("centos", "centos-stream")
+        else:
+            t = None
+        if t in b:
+            continue
+        unused_snapshots.append(s)
+
+    return sorted(unused_snapshots)
+
+
+def main():
+    gitlab = read_gitlab_ci(".gitlab-ci.yml")
+    runners = all_runners(gitlab)
+    rd = runner_distros(runners)
+    sd = get_distros("Schutzfile")
+
+    missing_snapshots = unused(rd, sd)
+    if missing_snapshots:
+        print("The following RUNNERS do not have Schutzfile entries:")
+        print("\n".join(sorted(missing_snapshots)))
+
+    unused_snapshots = unused(sd, rd)
+    if unused_snapshots:
+        print("The following Schutzbot entries are not used:")
+        print("\n".join(sorted(unused_snapshots)))
+
+
+if __name__ == '__main__':
+    sys.exit(main())