search

zarr-developers / zarr-python

An implementation of chunked, compressed, N-dimensional arrays for Python.
https://zarr.readthedocs.io
MIT License
1.53k stars 285 forks source link

zarr-python cannot read arrays saved by tensorstore using the zstd compressor #2056

Open mkitti opened 3 months ago

mkitti commented 3 months ago

Zarr version

v2.18.2

Numcodecs version

v0.12.1

Python Version

3.12.4

Operating System

Linux

Installation

using conda

Description

I get the following error when trying to open a dataset compressed with tensorstore using the zstd compressor.

RuntimeError: Zstd decompression error: invalid input data

Steps to reproduce

In [8]: ds = ts.open({
   ...:     'driver': 'zarr',
   ...:     'kvstore': {
   ...:         'driver': 'file',
   ...:         'path': 'tmp/zarr_zstd_dataset',
   ...:     },
   ...:     'metadata': {
   ...:         'compressor': {
   ...:             'id': 'zstd',
   ...:             'level': 3,
   ...:         },
   ...:         'shape': [1024, 1024],
   ...:         'chunks': [64, 64],
   ...:         'dtype': '|u1',
   ...:         'dimension_separator': '/',
   ...:     },
   ...:     'create': True,
   ...:     'delete_existing': True,
   ...: }).result()

In [9]: ds[:,:] = 5

In [10]: import zarr

In [11]: arr = zarr.open_array("tmp/zarr_zstd_dataset")

In [12]: arr[:,:]
---------------------------------------------------------------------------
RuntimeError                              Traceback (most recent call last)
Cell In[12], line 1
----> 1 arr[:,:]

File ~/review_temp/conda/3/x86_64/envs/zarr_python/lib/python3.12/site-packages/zarr/core.py:798, in Array.__getitem__(self, selection)
    796     result = self.vindex[selection]
    797 elif is_pure_orthogonal_indexing(pure_selection, self.ndim):
--> 798     result = self.get_orthogonal_selection(pure_selection, fields=fields)
    799 else:
    800     result = self.get_basic_selection(pure_selection, fields=fields)

File ~/review_temp/conda/3/x86_64/envs/zarr_python/lib/python3.12/site-packages/zarr/core.py:1080, in Array.get_orthogonal_selection(self, selection, out, fields)
   1077 # setup indexer
   1078 indexer = OrthogonalIndexer(selection, self)
-> 1080 return self._get_selection(indexer=indexer, out=out, fields=fields)

File ~/review_temp/conda/3/x86_64/envs/zarr_python/lib/python3.12/site-packages/zarr/core.py:1343, in Array._get_selection(self, indexer, out, fields)
   1340 if math.prod(out_shape) > 0:
   1341     # allow storage to get multiple items at once
   1342     lchunk_coords, lchunk_selection, lout_selection = zip(*indexer)
-> 1343     self._chunk_getitems(
   1344         lchunk_coords,
   1345         lchunk_selection,
   1346         out,
   1347         lout_selection,
   1348         drop_axes=indexer.drop_axes,
   1349         fields=fields,
   1350     )
   1351 if out.shape:
   1352     return out

File ~/review_temp/conda/3/x86_64/envs/zarr_python/lib/python3.12/site-packages/zarr/core.py:2183, in Array._chunk_getitems(self, lchunk_coords, lchunk_selection, out, lout_selection, drop_axes, fields)
   2181 for ckey, chunk_select, out_select in zip(ckeys, lchunk_selection, lout_selection):
   2182     if ckey in cdatas:
-> 2183         self._process_chunk(
   2184             out,
   2185             cdatas[ckey],
   2186             chunk_select,
   2187             drop_axes,
   2188             out_is_ndarray,
   2189             fields,
   2190             out_select,
   2191             partial_read_decode=partial_read_decode,
   2192         )
   2193     else:
   2194         # check exception type
   2195         if self._fill_value is not None:

File ~/review_temp/conda/3/x86_64/envs/zarr_python/lib/python3.12/site-packages/zarr/core.py:2096, in Array._process_chunk(self, out, cdata, chunk_selection, drop_axes, out_is_ndarray, fields, out_selection, partial_read_decode)
   2094 except ArrayIndexError:
   2095     cdata = cdata.read_full()
-> 2096 chunk = self._decode_chunk(cdata)
   2098 # select data from chunk
   2099 if fields:

File ~/review_temp/conda/3/x86_64/envs/zarr_python/lib/python3.12/site-packages/zarr/core.py:2352, in Array._decode_chunk(self, cdata, start, nitems, expected_shape)
   2350         chunk = self._compressor.decode_partial(cdata, start, nitems)
   2351     else:
-> 2352         chunk = self._compressor.decode(cdata)
   2353 else:
   2354     chunk = cdata

File numcodecs/zstd.pyx:219, in numcodecs.zstd.Zstd.decode()

File numcodecs/zstd.pyx:153, in numcodecs.zstd.decompress()

RuntimeError: Zstd decompression error: invalid input data

Additional output

$ conda env export
name: zarr_python
channels:
  - conda-forge
  - defaults
dependencies:
  - _libgcc_mutex=0.1=conda_forge
  - _openmp_mutex=4.5=2_gnu
  - aiohttp=3.9.5=py312h98912ed_0
  - aiosignal=1.3.1=pyhd8ed1ab_0
  - aom=3.9.1=hac33072_0
  - asciitree=0.3.3=py_2
  - asttokens=2.4.1=pyhd8ed1ab_0
  - attrs=23.2.0=pyh71513ae_0
  - blosc=1.21.6=hef167b5_0
  - brotli-python=1.1.0=py312h30efb56_1
  - bzip2=1.0.8=h4bc722e_7
  - c-ares=1.32.3=h4bc722e_0
  - ca-certificates=2024.7.4=hbcca054_0
  - certifi=2024.7.4=pyhd8ed1ab_0
  - cffi=1.16.0=py312hf06ca03_0
  - charset-normalizer=3.3.2=pyhd8ed1ab_0
  - dav1d=1.2.1=hd590300_0
  - decorator=5.1.1=pyhd8ed1ab_0
  - exceptiongroup=1.2.2=pyhd8ed1ab_0
  - executing=2.0.1=pyhd8ed1ab_0
  - fasteners=0.17.3=pyhd8ed1ab_0
  - frozenlist=1.4.1=py312h98912ed_0
  - fsspec=2024.6.1=pyhff2d567_0
  - h2=4.1.0=pyhd8ed1ab_0
  - hpack=4.0.0=pyh9f0ad1d_0
  - hyperframe=6.0.1=pyhd8ed1ab_0
  - idna=3.7=pyhd8ed1ab_0
  - ipython=8.26.0=pyh707e725_0
  - jedi=0.19.1=pyhd8ed1ab_0
  - keyutils=1.6.1=h166bdaf_0
  - krb5=1.21.3=h659f571_0
  - ld_impl_linux-64=2.40=hf3520f5_7
  - libabseil=20240116.2=cxx17_he02047a_1
  - libavif16=1.1.0=h9b56c87_0
  - libblas=3.9.0=23_linux64_openblas
  - libcblas=3.9.0=23_linux64_openblas
  - libcurl=8.9.0=hdb1bdb2_0
  - libedit=3.1.20191231=he28a2e2_2
  - libev=4.33=hd590300_2
  - libexpat=2.6.2=h59595ed_0
  - libffi=3.4.2=h7f98852_5
  - libgcc-ng=14.1.0=h77fa898_0
  - libgfortran-ng=14.1.0=h69a702a_0
  - libgfortran5=14.1.0=hc5f4f2c_0
  - libgomp=14.1.0=h77fa898_0
  - libjpeg-turbo=3.0.0=hd590300_1
  - liblapack=3.9.0=23_linux64_openblas
  - libnghttp2=1.58.0=h47da74e_1
  - libnsl=2.0.1=hd590300_0
  - libopenblas=0.3.27=pthreads_hac2b453_1
  - libpng=1.6.43=h2797004_0
  - libprotobuf=4.25.3=h08a7969_0
  - libsqlite=3.46.0=hde9e2c9_0
  - libssh2=1.11.0=h0841786_0
  - libstdcxx-ng=14.1.0=hc0a3c3a_0
  - libuuid=2.38.1=h0b41bf4_0
  - libwebp-base=1.4.0=hd590300_0
  - libxcrypt=4.4.36=hd590300_1
  - libzlib=1.3.1=h4ab18f5_1
  - lz4-c=1.9.4=hcb278e6_0
  - matplotlib-inline=0.1.7=pyhd8ed1ab_0
  - ml_dtypes=0.4.0=py312h1d6d2e6_1
  - msgpack-python=1.0.8=py312h2492b07_0
  - multidict=6.0.5=py312h98912ed_0
  - ncurses=6.5=h59595ed_0
  - numcodecs=0.12.1=py312h7070661_1
  - numpy=1.26.4=py312heda63a1_0
  - openssl=3.3.1=h4bc722e_2
  - parso=0.8.4=pyhd8ed1ab_0
  - pexpect=4.9.0=pyhd8ed1ab_0
  - pickleshare=0.7.5=py_1003
  - pip=24.0=pyhd8ed1ab_0
  - prompt-toolkit=3.0.47=pyha770c72_0
  - ptyprocess=0.7.0=pyhd3deb0d_0
  - pure_eval=0.2.3=pyhd8ed1ab_0
  - pybind11-abi=4=hd8ed1ab_3
  - pycparser=2.22=pyhd8ed1ab_0
  - pygments=2.18.0=pyhd8ed1ab_0
  - pysocks=1.7.1=pyha2e5f31_6
  - python=3.12.4=h194c7f8_0_cpython
  - python_abi=3.12=4_cp312
  - rav1e=0.6.6=he8a937b_2
  - readline=8.2=h8228510_1
  - requests=2.32.3=pyhd8ed1ab_0
  - setuptools=71.0.4=pyhd8ed1ab_0
  - six=1.16.0=pyh6c4a22f_0
  - snappy=1.2.1=ha2e4443_0
  - stack_data=0.6.2=pyhd8ed1ab_0
  - svt-av1=2.1.2=hac33072_0
  - tensorstore=0.1.62=py312h7e2185d_0
  - tk=8.6.13=noxft_h4845f30_101
  - traitlets=5.14.3=pyhd8ed1ab_0
  - typing_extensions=4.12.2=pyha770c72_0
  - tzdata=2024a=h0c530f3_0
  - urllib3=2.2.2=pyhd8ed1ab_1
  - wcwidth=0.2.13=pyhd8ed1ab_0
  - wheel=0.43.0=pyhd8ed1ab_1
  - xz=5.2.6=h166bdaf_0
  - yarl=1.9.4=py312h98912ed_0
  - zarr=2.18.2=pyhd8ed1ab_0
  - zstandard=0.23.0=py312h3483029_0
  - zstd=1.5.6=ha6fb4c9_0
prefix: /home/mkitti/review_temp/conda/3/x86_64/envs/zarr_python

xref: https://github.com/google/tensorstore/issues/182

mkitti commented 3 months ago

I previously discussed the root cause of this here: https://github.com/zarr-developers/numcodecs/pull/519#issuecomment-2186666974