🐛 fix(thirdparty): prune nested project snapshots

.gitea/ci/update_thirdparty_skills.sh

@@ -88,12 +88,38 @@ for entry in data["sources"]:
                 entry.get("upstream_ref", "main"),
                 entry["snapshot_dir"],
                 entry["sync_mode"],
+                "\x1e".join(entry.get("remove_paths", [])),
             ]
         )
     )
 PY
 }
 
+read_source_metadata_value() {
+  local key="$1"
+  local source_file="$2"
+
+  if [ ! -f "$source_file" ]; then
+    return 0
+  fi
+
+  sed -n "s/^- ${key}:[[:space:]]*//p" "$source_file" | head -n 1
+}
+
+remove_snapshot_paths() {
+  local snapshot_dir="$1"
+  local remove_paths="$2"
+
+  [ -n "$remove_paths" ] || return 0
+
+  local IFS=$'\x1e'
+  read -r -a paths <<< "$remove_paths"
+  for path in "${paths[@]}"; do
+    [ -n "$path" ] || continue
+    rm -rf "$snapshot_dir/$path"
+  done
+}
+
 cd "$REPO_DIR"
 
 git config user.name "$COMMIT_AUTHOR_NAME"
@@ -124,8 +150,9 @@ if ! emit_sources_tsv > "$sources_file"; then
 fi
 
 changed=0
-while IFS=$'\x1f' read -r source_id upstream_repo upstream_ref snapshot_dir sync_mode; do
+while IFS=$'\x1f' read -r source_id upstream_repo upstream_ref snapshot_dir sync_mode remove_paths; do
   [ -n "$source_id" ] || continue
+  remove_paths_md="${remove_paths//$'\x1e'/,}"
 
   gh_repo=""
   if gh_repo="$(github_owner_repo "$upstream_repo" 2>/dev/null)"; then
@@ -138,12 +165,10 @@ while IFS=$'\x1f' read -r source_id upstream_repo upstream_ref snapshot_dir sync
     exit 1
   fi
 
-  current_sha=""
-  if [ -f "$snapshot_dir/SOURCE.md" ]; then
-    current_sha="$(sed -n 's/^- Ref:[[:space:]]*//p' "$snapshot_dir/SOURCE.md" | head -n 1)"
-  fi
+  current_sha="$(read_source_metadata_value "Ref" "$snapshot_dir/SOURCE.md")"
+  current_remove_paths="$(read_source_metadata_value "Remove-Paths" "$snapshot_dir/SOURCE.md")"
 
-  if [ "$latest_sha" = "$current_sha" ]; then
+  if [ "$latest_sha" = "$current_sha" ] && [ "$remove_paths_md" = "$current_remove_paths" ]; then
     echo "Third-party snapshot is up to date for ${source_id}: $latest_sha"
     continue
   fi
@@ -169,12 +194,15 @@ while IFS=$'\x1f' read -r source_id upstream_repo upstream_ref snapshot_dir sync
     git -C "$upstream_dir" archive --format=tar HEAD | tar -xf - -C "$snapshot_dir"
   fi
 
+  remove_snapshot_paths "$snapshot_dir" "$remove_paths"
+
   snapshot_date="$(date -u +%Y-%m-%d)"
   cat > "$snapshot_dir/SOURCE.md" <<EOF
 # Source
 
 - Repo: ${upstream_repo%".git"}
 - Ref: $latest_sha
+- Remove-Paths: $remove_paths_md
 - Snapshot: $snapshot_date
 - Sync-Mode: $sync_mode
 - Notes: vendored into playbook branch $TARGET_BRANCH