From 9818e623be3aad1b325975e0ce7affd84920a757 Mon Sep 17 00:00:00 2001
From: James Frost <james.frost@metoffice.gov.uk>
Date: Fri, 4 Sep 2026 12:55:47 +0100
Subject: [PATCH] Support SHA-256 git hashes in reference checks

Git 3.0 will be defaulting to SHA-256, so will have longer (64
character) object IDs.

This is a non-exhaustive fix, as I have not checked everywhere in the
codebase, but just grepped for some common regex patterns that would be
problematic. I have checked that you can checkout a SHA-256 format
repository though.
---
 github_scripts/get_git_sources.py  | 5 ++++-
 github_scripts/git_bdiff.py        | 4 ++--
 github_scripts/suite_report_git.py | 5 ++++-
 3 files changed, 10 insertions(+), 4 deletions(-)

diff --git a/github_scripts/get_git_sources.py b/github_scripts/get_git_sources.py
index 883fd635..90169f57 100644
--- a/github_scripts/get_git_sources.py
+++ b/github_scripts/get_git_sources.py
@@ -23,6 +23,9 @@
 
 logger = logging.getLogger(__name__)
 
+# Match hex commit IDs for both SHA-1 and SHA-256 repositories.
+_hash_pattern = re.compile(r"^\s*([0-9a-f]{40}(?:[0-9a-f]{24})?)\s*")
+
 
 class SubprocessRunError(Exception):
     def __init__(self, command, returncode, stdout, stderr):
@@ -351,7 +354,7 @@ def determine_mirror_fetch(repo_source: str, repo_ref: str) -> str:
 
     # If the ref is a hash then we don't need the fork user as part of the fetch.
     # Equally, if the user is the Upstream User, it's not needed
-    if not user or re.match(r"^\s*([0-9a-f]{40})\s*$", repo_ref):
+    if not user or re.match(_hash_pattern, repo_ref):
         fetch = repo_ref
     else:
         fetch = f"{user}/{repo_ref}"
diff --git a/github_scripts/git_bdiff.py b/github_scripts/git_bdiff.py
index 2db51ddf..0836e27d 100644
--- a/github_scripts/git_bdiff.py
+++ b/github_scripts/git_bdiff.py
@@ -111,8 +111,8 @@ class GitBDiff(GitBase):
     # Name of primary branch - default is main
     primary_branch = "main"
 
-    # Match hex commit IDs
-    _hash_pattern = re.compile(r"^\s*([0-9a-f]{40})\s*$")
+    # Match hex commit IDs for both SHA-1 and SHA-256 repositories.
+    _hash_pattern = re.compile(r"^\s*([0-9a-f]{40}(?:[0-9a-f]{24})?)\s*")
 
     def __init__(self, parent=None, repo=None):
         self.parent = parent or self.primary_branch
diff --git a/github_scripts/suite_report_git.py b/github_scripts/suite_report_git.py
index 62dc2b8e..34320f3c 100755
--- a/github_scripts/suite_report_git.py
+++ b/github_scripts/suite_report_git.py
@@ -20,6 +20,9 @@
 
 from suite_data import SuiteData
 
+# Match hex commit IDs for both SHA-1 and SHA-256 repositories.
+_hash_pattern = re.compile(r"^\s*([0-9a-f]{40}(?:[0-9a-f]{24})?)\s*")
+
 
 def create_markdown_row(*columns: str, header=False) -> List[str]:
     """
@@ -151,7 +154,7 @@ def create_dependency_table(self) -> None:
             org_repo = extract_org_repo(reference)
 
             # Check if the ref is a hash and use short form if so
-            if re.match(r"^\s*([0-9a-f]{40})\s*$", ref):
+            if re.match(_hash_pattern, ref):
                 ref = ref[:7]
             url = f"https://github.com/{org_repo}/tree/{ref}"
             reference = f"[{org_repo}@{ref}]({url})"