Context
packages/diff-filter's BINARY_EXTENSIONS still misses several common binary formats across bytecode, compression, ML models, and scientific data. These complement #156 (which covers image/design/mobile/3D + safetensors/onnx/pkl) without overlap.
Proposed additions to BINARY_EXTENSIONS
Bytecode and compiled (4)
| Extension |
Description |
pyc |
Python bytecode |
beam |
Erlang/Elixir BEAM bytecode |
dex |
Android Dalvik executable |
aar |
Android Archive |
Compression (2)
| Extension |
Description |
zst |
Zstandard compressed file |
lz4 |
LZ4 compressed file |
ML models not in #156 (5)
| Extension |
Description |
pt |
PyTorch model checkpoint |
pth |
PyTorch model (alt extension) |
gguf |
GGUF model format (llama.cpp) |
tflite |
TensorFlow Lite model |
mlmodel |
Apple Core ML model |
Scientific data (4)
| Extension |
Description |
h5 |
HDF5 / Keras model |
npy |
NumPy array |
npz |
NumPy compressed archive |
pcap |
Packet capture (libpcap/tcpdump) |
Size
Small - 15 entries to BINARY_EXTENSIONS + tests.
Context
packages/diff-filter'sBINARY_EXTENSIONSstill misses several common binary formats across bytecode, compression, ML models, and scientific data. These complement #156 (which covers image/design/mobile/3D + safetensors/onnx/pkl) without overlap.Proposed additions to
BINARY_EXTENSIONSBytecode and compiled (4)
pycbeamdexaarCompression (2)
zstlz4ML models not in #156 (5)
ptpthgguftflitemlmodelScientific data (4)
h5npynpzpcapSize
Small - 15 entries to BINARY_EXTENSIONS + tests.