From 76da863a021b64acecc9532d5986f30a0d923405 Mon Sep 17 00:00:00 2001
From: Nick Sweeting <git@sweeting.me>
Date: Sun, 21 Jun 2026 02:53:28 -0700
Subject: [PATCH] fix docker image commit metadata

---
 .github/workflows/docker.yml              |  1 +
 Dockerfile                                |  5 ++++-
 archivebox/config/version.py              |  6 +++---
 archivebox/tests/test_version_metadata.py | 19 +++++++++++++++++++
 4 files changed, 27 insertions(+), 4 deletions(-)

diff --git a/.github/workflows/docker.yml b/.github/workflows/docker.yml
index ffd2b936..21e4ffef 100644
--- a/.github/workflows/docker.yml
+++ b/.github/workflows/docker.yml
@@ -187,6 +187,7 @@ jobs:
           labels: ${{ steps.docker_meta.outputs.labels }}
           build-args: |
             ABX_DL_IMAGE=${{ steps.abx_dl_image.outputs.image }}
+            ARCHIVEBOX_COMMIT_HASH=${{ github.sha }}
           cache-from: type=gha,scope=${{ matrix.cache_scope }}
           cache-to: type=gha,mode=max,scope=${{ matrix.cache_scope }}
           pull: true
diff --git a/Dockerfile b/Dockerfile
index 5ede94c7..241f8584 100644
--- a/Dockerfile
+++ b/Dockerfile
@@ -22,6 +22,7 @@ ARG TARGETPLATFORM
 ARG TARGETOS
 ARG TARGETARCH
 ARG TARGETVARIANT
+ARG ARCHIVEBOX_COMMIT_HASH=""
 
 ENV TZ=UTC \
     LANGUAGE=en_US:en \
@@ -165,7 +166,9 @@ COPY --chown=root:root --chmod=755 "." "$CODE_DIR/"
 RUN --mount=type=cache,target=/root/.cache/uv,sharing=locked,id=uv-$TARGETARCH$TARGETVARIANT \
     echo "[*] Installing ArchiveBox Python source code from $CODE_DIR..." \
     && COMMIT_HASH="$( \
-        if [[ -f "$CODE_DIR/.git/HEAD" ]]; then \
+        if [[ "$ARCHIVEBOX_COMMIT_HASH" =~ ^[0-9a-fA-F]{40}$ ]]; then \
+            echo "$ARCHIVEBOX_COMMIT_HASH"; \
+        elif [[ -f "$CODE_DIR/.git/HEAD" ]]; then \
             HEAD_REF="$(cat "$CODE_DIR/.git/HEAD")"; \
             if [[ "$HEAD_REF" =~ ^[0-9a-fA-F]{40}$ ]]; then \
                 echo "$HEAD_REF"; \
diff --git a/archivebox/config/version.py b/archivebox/config/version.py
index 81f4f721..eeb526cd 100644
--- a/archivebox/config/version.py
+++ b/archivebox/config/version.py
@@ -50,9 +50,9 @@ def get_COMMIT_HASH() -> str | None:
     if IN_DOCKER:
         try:
             version_text = Path("/VERSION.txt").read_text()
-            match = re.search(r"^COMMIT_HASH=([0-9a-fA-F]{40})$", version_text, re.MULTILINE)
-            if match:
-                return match.group(1)
+            matches = re.findall(r"^COMMIT_HASH=([0-9a-fA-F]{40})$", version_text, re.MULTILINE)
+            if matches:
+                return matches[-1]
         except Exception:
             pass
 
diff --git a/archivebox/tests/test_version_metadata.py b/archivebox/tests/test_version_metadata.py
index d112eaed..c60577f1 100644
--- a/archivebox/tests/test_version_metadata.py
+++ b/archivebox/tests/test_version_metadata.py
@@ -34,6 +34,25 @@ def test_get_commit_hash_from_docker_version_file_ignores_short_hash(monkeypatch
     assert version.get_COMMIT_HASH() == commit_hash
 
 
+def test_get_commit_hash_from_docker_version_file_prefers_last_full_hash(monkeypatch) -> None:
+    base_hash = "a" * 40
+    archivebox_hash = "b" * 40
+
+    class VersionPath:
+        def __init__(self, path: str):
+            self.path = path
+
+        def read_text(self) -> str:
+            assert self.path == "/VERSION.txt"
+            return f"COMMIT_HASH={base_hash}\nCOMMIT_HASH={archivebox_hash}\n"
+
+    monkeypatch.setattr(version, "IN_DOCKER", True)
+    monkeypatch.setattr(version, "Path", VersionPath)
+    version.get_COMMIT_HASH.cache_clear()
+
+    assert version.get_COMMIT_HASH() == archivebox_hash
+
+
 def test_get_commit_hash_from_detached_head(monkeypatch, tmp_path) -> None:
     commit_hash = "a" * 40
     package_dir = tmp_path / "archivebox"