Add resumable mirror worker

mirror_worker.py

@@ -0,0 +1,96 @@
+# /// script
+# requires-python = ">=3.10"
+# dependencies = ["huggingface_hub>=1.0.0", "modelscope-hub>=0.2.0"]
+# ///
+"""Resumable, bounded-storage ModelScope -> HF Dataset mirror worker."""
+
+from __future__ import annotations
+
+import os
+import shutil
+import time
+from pathlib import Path
+
+from huggingface_hub import HfApi
+from modelscope_hub import HubApi
+
+SOURCE_REPO = "daimonrobotics/Daimon-Infinity"
+DEST_REPO = "mrfakename/Daimon-Infinity"
+WORKER_INDEX = int(os.environ["WORKER_INDEX"])
+WORKER_COUNT = int(os.environ["WORKER_COUNT"])
+MODELSCOPE_TOKEN = os.environ["MODELSCOPE_TOKEN"]
+HF_TOKEN = os.environ["HF_TOKEN"]
+WORKDIR = Path("/tmp/daimon-infinity")
+
+
+def retry(action, label: str, attempts: int = 6):
+    for attempt in range(attempts):
+        try:
+            return action()
+        except Exception:
+            if attempt == attempts - 1:
+                raise
+            time.sleep(min(180, 2 ** attempt * 5))
+
+
+def main() -> None:
+    WORKDIR.mkdir(parents=True, exist_ok=True)
+    source = HubApi(token=MODELSCOPE_TOKEN)
+    # Authenticate explicitly so the gated dataset access applies to downloads.
+    source.login(MODELSCOPE_TOKEN)
+    target = HfApi(token=HF_TOKEN)
+    uploaded = set(target.list_repo_files(DEST_REPO, repo_type="dataset"))
+    files = source.list_repo_files(SOURCE_REPO, "dataset", recursive=True)
+    selected = [
+        item for number, item in enumerate(files)
+        if number % WORKER_COUNT == WORKER_INDEX and item.type != "tree"
+    ]
+    print(f"worker {WORKER_INDEX}/{WORKER_COUNT}: {len(selected)} files")
+
+    for number, item in enumerate(selected, start=1):
+        path = item.path
+        if path in uploaded:
+            print(f"skip {number}/{len(selected)} {path}", flush=True)
+            continue
+        local = WORKDIR / path
+        local.parent.mkdir(parents=True, exist_ok=True)
+        try:
+            retry(
+                lambda: source.download_file(
+                    SOURCE_REPO,
+                    "dataset",
+                    path,
+                    local_dir=WORKDIR,
+                    force=True,
+                    expected_sha256=item.blob_id,
+                ),
+                f"download {path}",
+            )
+            retry(
+                lambda: target.upload_file(
+                    path_or_fileobj=str(local),
+                    path_in_repo=path,
+                    repo_id=DEST_REPO,
+                    repo_type="dataset",
+                    commit_message=f"Mirror {path}",
+                ),
+                f"upload {path}",
+            )
+            uploaded.add(path)
+            print(f"done {number}/{len(selected)} {path}", flush=True)
+        finally:
+            local.unlink(missing_ok=True)
+            # Remove any empty nested directories left by this file.
+            parent = local.parent
+            while parent != WORKDIR:
+                try:
+                    parent.rmdir()
+                except OSError:
+                    break
+                parent = parent.parent
+
+    shutil.rmtree(WORKDIR, ignore_errors=True)
+
+
+if __name__ == "__main__":
+    main()