From 86043ed931be0787dd1ad9a7f50e185087e5a56e Mon Sep 17 00:00:00 2001
From: Abdessamad Derraz <3028866+Abdess@users.noreply.github.com>
Date: Sat, 17 Oct 2026 10:31:48 +0200
Subject: [PATCH] fix: count drift since the pin in profile triage

---
 scripts/profile_sync.py |  3 ++-
 scripts/upstream.py     | 49 ++++++++++++++++++++++++++++++++++++-----
 tests/test_upstream.py  | 25 ++++++++++++++++++---
 3 files changed, 67 insertions(+), 10 deletions(-)

diff --git a/scripts/profile_sync.py b/scripts/profile_sync.py
index f3ae63cb..363c79ef 100644
--- a/scripts/profile_sync.py
+++ b/scripts/profile_sync.py
@@ -3367,7 +3367,8 @@ def _print_triage(args, selected: dict, reports: list[ProfileReport]) -> None:
             paths = _cited_paths(report)
             for path in paths[:TRIAGE_PATH_SAMPLE]:
                 commits += upstream.commits_touching(
-                    repo, path, report.pin, args.cache_dir, args.offline
+                    repo, path, report.pin, args.cache_dir, args.offline,
+                    head=report.head,
                 )
             if len(paths) > TRIAGE_PATH_SAMPLE:
                 print(
diff --git a/scripts/upstream.py b/scripts/upstream.py
index 87a2eb5a..ed855fb4 100644
--- a/scripts/upstream.py
+++ b/scripts/upstream.py
@@ -667,22 +667,59 @@ def list_tree(
     return paths, bool(payload.get("truncated"))
 
 
+def commit_date(repo: Repo, sha: str, cache_dir: str, offline: bool = False) -> str | None:
+    """When a revision was committed, as the forge reports it (ISO 8601)."""
+    if repo.family == "gitlab":
+        payload = _api(
+            f"{repo.api_base}/projects/{_project(repo)}/repository/commits/{sha}",
+            cache_dir, offline,
+        )
+        value = payload.get("committed_date") if isinstance(payload, dict) else None
+        return value if isinstance(value, str) else None
+    if repo.family == "github":
+        url = f"{repo.api_base}/repos/{repo.slug}/commits/{sha}"
+    else:
+        url = f"{repo.api_base}/repos/{repo.slug}/git/commits/{sha}"
+    stamp = _gitea_commit_date(_api(url, cache_dir, offline))
+    return stamp.isoformat() if stamp is not None else None
+
+
 def commits_touching(
-    repo: Repo, path: str, base: str, cache_dir: str, offline: bool = False
+    repo: Repo, path: str, base: str, cache_dir: str, offline: bool = False,
+    head: str | None = None,
 ) -> int:
-    """Commits touching one path since a revision."""
+    """Commits touching one path since a revision, on the default branch.
+
+    Listing from the pin counted the history before it: a profile pinned on
+    HEAD scored the age of its files as drift. The count is of the commits
+    after the pin's date that touch the path, the pin itself excluded.
+    """
+    if head is not None and head == base:
+        return 0
+    since = commit_date(repo, base, cache_dir, offline)
+    if since is None:
+        return 0
     quoted = urllib.parse.quote(path)
+    stamp = urllib.parse.quote(since)
     if repo.family == "gitlab":
         url = (
             f"{repo.api_base}/projects/{_project(repo)}"
-            f"/repository/commits?path={quoted}&per_page=100"
+            f"/repository/commits?path={quoted}&since={stamp}&per_page=100"
         )
     elif repo.family == "github":
         url = (
             f"{repo.api_base}/repos/{repo.slug}/commits"
-            f"?path={quoted}&sha={base}&per_page=100"
+            f"?path={quoted}&since={stamp}&per_page=100"
         )
     else:
-        url = f"{repo.api_base}/repos/{repo.slug}/commits?path={quoted}&limit=100"
+        url = (
+            f"{repo.api_base}/repos/{repo.slug}/commits"
+            f"?path={quoted}&since={stamp}&limit=100"
+        )
     payload = _api(url, cache_dir, offline)
-    return len(payload) if isinstance(payload, list) else 0
+    if not isinstance(payload, list):
+        return 0
+    return sum(
+        1 for entry in payload
+        if isinstance(entry, dict) and (entry.get("sha") or entry.get("id")) != base
+    )
diff --git a/tests/test_upstream.py b/tests/test_upstream.py
index 0fec7a16..ded955f3 100644
--- a/tests/test_upstream.py
+++ b/tests/test_upstream.py
@@ -602,10 +602,29 @@ class TestCompare(unittest.TestCase):
         result = upstream.compare(repo, "a", "b", self.dir)
         self.assertEqual(upstream.find_renamed(result, "old.c"), "new.c")
 
-    def test_commits_touching_counts_entries(self):
-        self.payload = [{"sha": "1"}, {"sha": "2"}, {"sha": "3"}]
+    def test_commits_touching_counts_the_commits_after_the_pin(self):
+        """Listing from the pin counted its history: freeintv's pin on HEAD
+        scored 26 commits of drift."""
+        served = {
+            "commits/base": {"commit": {"committer": {"date": "2026-01-01T00:00:00Z"}}},
+        }
+        listing = [{"sha": "1"}, {"sha": "2"}, {"sha": "base"}]
+        seen: list[str] = []
+
+        def answer(url):
+            seen.append(url)
+            for key, value in served.items():
+                if url.endswith(key):
+                    return value
+            return listing
+
+        upstream._http_json = answer
         self.assertEqual(
-            upstream.commits_touching(self.repo, "a.c", "base", self.dir), 3
+            upstream.commits_touching(self.repo, "a.c", "base", self.dir, head="tip"), 2
+        )
+        self.assertTrue(any("since=2026-01-01" in url for url in seen))
+        self.assertEqual(
+            upstream.commits_touching(self.repo, "a.c", "tip", self.dir, head="tip"), 0
         )