diff --git a/zack/git2filenames.py b/zack/git2filenames.py
new file mode 100755
index 0000000..9d03f36
--- /dev/null
+++ b/zack/git2filenames.py
@@ -0,0 +1,78 @@
+#!/usr/bin/env python3
+
+"""Crawl a local Git repository and write a map <SHA1, FILENAME> of all file
+*names* (not paths) given to all blobs found in the repository.
+
+As file names might be in general non readable, the map is actually emitted
+twice, once with hex-encoded file names (for those who really want all file
+names), another with utf8-encoded file names (skipping all non-utf8-encodable
+names).  If you want only one of the two, write the other to /dev/null.
+
+"""
+
+import dulwich
+import gzip
+import sys
+
+from dulwich.repo import Repo
+from pathlib import Path
+from tqdm import tqdm
+from typing import Set, Tuple
+
+
+GIT_BLOB_MODES = {0o100644, 0o100755}
+
+
+def collect_blob_entries(object_store) -> Set[Tuple[bytes, bytes]]:
+    """collect all <blob_sha1, filename> entries in memory, deduplicating"""
+    blob_entries: Set[Tuple[bytes, bytes]] = set()  # <sha1, filename>
+    objcount = len(list(object_store))
+
+    for sha1 in tqdm(object_store, desc='collecting filenames',
+                     unit='obj', total=objcount):
+        obj = object_store[sha1]
+        if not isinstance(obj, dulwich.objects.Tree):
+            continue
+        for entry in obj.items():
+            if entry.mode in GIT_BLOB_MODES:
+                blob_entries.add((entry.sha, entry.path))
+
+    return blob_entries
+
+
+def emit_blob_entries(blob_entries, hex_names: Path, utf8_names: Path) -> None:
+    """emit sorted <sha1, filename> entries to the given gzipped files
+
+    note that the gzip files are appended to, so you can incrementally add
+    other pairs to them (the *resulting* maps will not be sorted in that case
+    though)
+
+    """
+    with gzip.open(hex_names, 'at', encoding='ascii') as hex_f, \
+         gzip.open(utf8_names, 'at', encoding='utf8') as utf8_f:  # NoQA
+        for (sha, filename) in tqdm(sorted(blob_entries),
+                                    desc='writing filenames',
+                                    unit='name', total=len(blob_entries)):
+            target = sha.decode('ascii')
+            print(target, filename.hex(), sep='\t', file=hex_f)
+            try:
+                print(target, filename.decode('utf8'), sep='\t', file=utf8_f)
+            except UnicodeDecodeError:
+                pass
+
+
+def main(repo_dir: Path, hex_names: Path, utf8_names: Path) -> None:
+    repo = Repo(repo_dir)
+    blob_entries = collect_blob_entries(repo.object_store)
+    emit_blob_entries(blob_entries, hex_names, utf8_names)
+
+
+if __name__ == '__main__':
+    try:
+        repo_dir = Path(sys.argv[1])
+        hex_names = Path(sys.argv[2])
+        utf8_names = Path(sys.argv[3])
+        main(repo_dir, hex_names, utf8_names)
+    except IndexError:
+        print('Usage: git2objstorage REPO_DIR GZIP_HEX_OUT GZIP_UTF8_OUT')
+        sys.exit(1)