Granular is a library for reading and writing multimodal datasets. Each dataset is a collection of linked files of the bag file format, a simple seekable container structure.
- 🚀 Performance: Minimal overhead for maximum read and write throughput.
- 🔎 Seekable: Fast random access from disk by datapoint index.
- 🎞️ Sequences: Datapoints can contain seekable ranges of modalities.
- 🤸 Flexible: User provides encoders and decoders; examples available.
- 👥 Sharding: Store datasets into shards to split processing workloads.
Granular is a single file, so you can just copy it to your project directory. Or you can install the package:
pip install granular
Writing
import granular
import msgpack
import numpy as np
spec = {
'foo': 'int', # integer
'bar': 'utf8[]', # list of strings
'baz': 'msgpack', # packed structure
}
# Or use the provided granular.encoders.
encoders = {
'int': lambda x: x.to_bytes(8, 'little'),
'utf8': lambda x: x.encode('utf-8'),
'msgpack': msgpack.packb,
}
with granular.ShardedDatasetWriter(
directory, spec, encoders, shardlen=1000) as writer:
writer.append({'foo': 42, 'bar': ['hello', 'world'], 'baz': {'a': 1})
# ...Files
$ tree directory
.
├── 000000
│ ├── spec.json
│ ├── refs.bag
│ ├── foo.bag
│ ├── bar.bag
│ └── baz.bag
├── 000001
│ ├── spec.json
│ ├── refs.bag
│ ├── foo.bag
│ ├── bar.bag
│ └── baz.bag
└── ...Reading
# Or use the provided granular.decoders.
decoders = {
'int': lambda x: int.from_bytes(x),
'utf8': lambda x: x.decode('utf-8'),
'msgpack': msgpack.unpackb,
}
with granular.ShardedDatasetReader(directory, decoders) as reader:
print(len(reader)) # Number of datapoints in the dataset.
print(reader.size) # Dataset size in bytes.
print(reader.shards) # Number of shards.
# Read data points by index. This will read only the relevant bytes from
# disk. An additional small read is used when caching index tables is
# disabled, supporting arbitrarily large datasets with minimal overhead.
assert reader[0] == {'foo': 42, 'bar': ['hello', 'world'], 'baz': {'a': 1}
# Read a subset of keys of a datapoint. For example, this allows quickly
# iterating over the metadata fields of all datapoints without accessing
# expensive image or video modalities.
assert reader[0, {'foo': True, 'baz': True}] == {'foo': 42, 'baz': {'a': 1}}
# Read only a slice of the 'bar' list. Only the requested slice will be
# fetched from disk. For example, the could be used to load a subsequence of
# a long video that is stored as list of consecutive MP4 clips.
assert reader[0, {'bar': range(1, 2)}] == {'bar': ['world']}For small datasets where sharding is not necessary, you can also use
DatasetReader and DatasetWriter.
For distributed processing using multiple processes or machines, use
ShardedDatasetReader and ShardedDatasetWriter and set shardstart to the
worker index and shardstop to the total number of workers.
Granular does not impose a serialization solution on the user. Any words can be used as types, as long as their encoder and decoder functions are provided.
Examples of common encode and decode functions are provided in
formats.py. These support Numpy arrays, JPG and PNG images, MP4
videos, and more. They can be used as granular.encoders and
granular.decoders.
If you have a question, please file an issue.
