Pre commit large files

.pre-commit-config.yaml

@@ -75,9 +75,17 @@ repos:
         name: LFS data
         always_run: true
         pass_filenames: false
-        entry: bin/lfs_check
+        entry: bin/hooks/lfs_check
         language: script
 
+      - id: largefiles-check
+        name: Large files check
+        always_run: true
+        pass_filenames: false
+        entry: python bin/hooks/largefiles_check
+        language: python
+        additional_dependencies: ['tomli']
+
       - id: doclinks
         name: Doclinks
         always_run: true

bin/hooks/largefiles_check

@@ -0,0 +1,62 @@
+#!/usr/bin/env python3
+"""Pre-commit hook to detect large files that should be in LFS."""
+
+import argparse
+import fnmatch
+import os
+import shutil
+import subprocess
+import sys
+
+import tomli
+
+parser = argparse.ArgumentParser()
+parser.add_argument("--all", action="store_true", help="Check all files in repo, not just staged")
+args = parser.parse_args()
+
+# Check git-lfs is installed
+if not shutil.which("git-lfs"):
+    print("git-lfs is not installed.")
+    print("\nInstall with:")
+    print("  Arch:   pacman -S git-lfs")
+    print("  Ubuntu: apt install git-lfs")
+    print("  macOS:  brew install git-lfs")
+    print("\nThen run: git lfs install")
+    sys.exit(1)
+
+# Load config
+with open("pyproject.toml", "rb") as f:
+    config = tomli.load(f).get("tool", {}).get("largefiles", {})
+
+max_size_kb = config.get("max_size_kb", 50)
+max_bytes = max_size_kb * 1024
+ignore_patterns = config.get("ignore", [])
+
+# Get LFS files to exclude
+result = subprocess.run(
+    ["git", "lfs", "ls-files", "-n"], capture_output=True, text=True, check=True
+)
+lfs_files = set(result.stdout.splitlines())
+
+# Get files to check
+if args.all:
+    files_cmd = ["git", "ls-files"]
+else:
+    files_cmd = ["git", "diff", "--cached", "--name-only"]
+
+violations = []
+result = subprocess.run(files_cmd, capture_output=True, text=True, check=True)
+for file in result.stdout.splitlines():
+    if file in lfs_files:
+        continue
+    if any(fnmatch.fnmatch(file, p) for p in ignore_patterns):
+        continue
+    if os.path.isfile(file) and os.path.getsize(file) > max_bytes:
+        violations.append((file, os.path.getsize(file)))
+
+if violations:
+    print(f"Large files detected (limit: {max_size_kb}KB):")
+    for f, size in sorted(violations, key=lambda x: -x[1]):
+        print(f"  {size // 1024}KB  {f}")
+    print("\nEither add to LFS or to [tool.largefiles].ignore in pyproject.toml")
+    sys.exit(1)

docs/data.md

@@ -194,7 +194,7 @@ The [`lfs_push`](/bin/lfs_push) script:
 2. Uploads to Git LFS
 3. Stages the compressed file
 
-A pre-commit hook ([`bin/lfs_check`](/bin/lfs_check#L26)) blocks commits if you have uncompressed directories in `data/` without a corresponding `.tar.gz` in `data/.lfs/`.
+A pre-commit hook ([`bin/hooks/lfs_check`](/bin/hooks/lfs_check#L26)) blocks commits if you have uncompressed directories in `data/` without a corresponding `.tar.gz` in `data/.lfs/`.
 
 ## Location Resolution
 

pyproject.toml

@@ -353,6 +353,15 @@ addopts = "-v -p no:warnings -ra --color=yes -m 'not vis and not benchmark and n
 asyncio_mode = "auto"
 asyncio_default_fixture_loop_scope = "function"
 
+[tool.largefiles]
+max_size_kb = 50
+ignore = [
+    "uv.lock",
+    "*/package-lock.json",
+    "dimos/dashboard/dimos.rbl",
+    "dimos/web/dimos_interface/themes.json",
+]
+
 [tool.uv]
 # Build dependencies for packages that don't declare them properly
 extra-build-dependencies = { detectron2 = ["torch"], contact-graspnet-pytorch = ["numpy"] }