Read and Write Lance Dataset¶

Lance dataset APIs follows the PyArrow API conventions.

Writing Lance Dataset¶

Similar to Apache Pyarrow, the simplest approach to create a Lance dataset is writing a pyarrow.Table via lance.write_dataset().

import lance
import pyarrow as pa

table = pa.Table.from_pylist([{"name": "Alice", "age": 20},
                              {"name": "Bob", "age": 30}])
lance.write_dataset(table, "./alice_and_bob.lance")

If the memory footprint of the dataset is too large to fit in memory, lance.write_dataset() also supports writing a dataset in iterator of pyarrow.RecordBatch es.

import lance
import pyarrow as pa

def producer():
    yield pa.RecordBatch.from_pylist([{"name": "Alice", "age": 20}])
    yield pa.RecordBatch.from_pylist([{"name": "Blob", "age": 30}])

schema = pa.schema([
        pa.field("name", pa.string()),
        pa.field("age", pa.int64()),
    ])

lance.write_dataset(reader, "./alice_and_bob.lance", schema)

lance.write_dataset() supports writing pyarrow.Table, pandas.DataFrame, pyarrow.Dataset, and Iterator[pyarrow.RecordBatch]. Check its doc for more details.

Adding new columns¶

New columns can be merged into an existing dataset in using lance.Dataset.merge(). This allows filling in additional columns without having to rewrite the whole dataset.

To use the merge method, provide a new table that includes the columns you want to add, and a column name to use for joining the new data to the existing dataset.

For example, imagine we have a dataset of embeddings and ids:

import lance
import pyarrow as pa
import numpy as np
table = pa.table({
   "id": pa.array([1, 2, 3]),
   "embedding": pa.array([np.array([1, 2, 3]), np.array([4, 5, 6]),
                          np.array([7, 8, 9])])
})
dataset = lance.write_dataset(table, "embeddings")

Now if we want to add a column of labels we have generated, we can do so by merging a new table:

new_data = pa.table({
   "id": pa.array([1, 2, 3]),
   "label": pa.array(["horse", "rabbit", "cat"])
})
dataset.merge(new_data, "id")
dataset.to_table().to_pandas()

   id  embedding   label
 1  [1, 2, 3]   horse
 2  [4, 5, 6]  rabbit
 3  [7, 8, 9]     cat

Deleting rows¶

Lance supports deleting rows from a dataset using a SQL filter. For example, to delete Bob’s row from the dataset above, one could use:

import lance

dataset = lance.dataset("./alice_and_bob.lance")
dataset.delete("name = 'Bob'")

lance.LanceDataset.delete() supports the same filters as described in Filter push-down.

Rows are deleted by marking them as deleted in a separate deletion index. This is faster than rewriting the files and also avoids invaliding any indices that point to those files. Any subsequent queries will not return the deleted rows.

Warning

Do not read datasets with deleted rows using Lance versions prior to 0.5.0, as they will return the deleted rows. This is fixed in 0.5.0 and later.

Committing mechanisms for S3¶

Most supported storage systems (e.g. local file system, Google Cloud Storage, Azure Blob Store) natively support atomic commits, which prevent concurrent writers from corrupting the dataset. However, S3 does not support this natively. To work around this, you may provide a locking mechanism that Lance can use to lock the table while providing a write. To do so, you should implement a context manager that acquires and releases a lock and then pass that to the commit_lock parameter of lance.write_dataset().

Note

In order for the locking mechanism to work, all writers must use the same exact mechanism. Otherwise, Lance will not be able to detect conflicts.

On entering, the context manager should acquire the lock on the table. The table version being committed is passed in as an argument, which may be used if the locking service wishes to keep track of the current version of the table, but this is not required. If the table is already locked by another transaction, it should wait until it is unlocked, since the other transaction may fail. Once unlocked, it should either lock the table or, if the lock keeps track of the current version of the table, return a CommitConflictError if the requested version has already been committed.

To prevent poisoned locks, it’s recommended to set a timeout on the locks. That way, if a process crashes while holding the lock, the lock will be released eventually. The timeout should be no less than 30 seconds.

from contextlib import contextmanager

@contextmanager
def commit_lock(version: int);
    # Acquire the lock
    my_lock.acquire()
    try:
      yield
    except:
      failed = True
    finally:
      my_lock.release()

lance.write_dataset(data, "s3://bucket/path/", commit_lock=commit_lock)

When the context manager is exited, it will raise an exception if the commit failed. This might be because of a network error or if the version has already been written. Either way, the context manager should release the lock. Use a try/finally block to ensure that the lock is released.

Warning

Lance _detects_ but does not yet resolve conflicts. So if there are multiple writers, only one will succeed if they try to commit at exactly the same time. This will be fixed in future releases.

Reading Lance Dataset¶

To open a Lance dataset, use the lance.dataset() function:

import lance
ds = lance.dataset("s3://bucket/path/imagenet.lance")
# Or local path
ds = lance.dataset("./imagenet.lance")
Note

Lance supports local file system, AWS s3 and Google Cloud Storage(gs) as storage backends at the moment. See storages for more details.

The most straightforward approach for reading a Lance dataset is to utilize the lance.LanceDataset.to_table() method in order to load the entire dataset into memory.

table = ds.to_table()

Due to Lance being a high-performance columnar format, it enables efficient reading of subsets of the dataset by utilizing Column (projection) push-down and filter (predicates) push-downs.

table = ds.to_table(
    columns=["image", "label"],
    filter="label = 2 AND text IS NOT NULL",
    limit=1000,
    offset=3000)

Lance understands the cost of reading heavy columns such as image. Consequently, it employs an optimized query plan to execute the operation efficiently.

Iterative Read¶

If the dataset is too large to fit in memory, you can read it in batches using the lance.LanceDataset.to_batches() method:

for batch in ds.to_batches(columns=["image"], filter="label = 10"):
    # do something with batch
    compute_on_batch(batch)

Unsurprisingly, to_batches() takes the same parameters as to_table() function.

Filter push-down¶

Lance embraces the utilization of standard SQL expressions as predicates for dataset filtering. By pushing down the SQL predicates directly to the storage system, the overall I/O load during a scan is significantly reduced.

Currently, Lance supports a growing list of expressions.

>, >=, <, <=, =
AND, OR, NOT
IS NULL, IS NOT NULL
IS TRUE, IS NOT TRUE, IS FALSE, IS NOT FALSE
IN
LIKE, NOT LIKE
regexp_match(column, pattern)
CAST

For example, the following filter string is acceptable:

((label IN [10, 20]) AND (note.email IS NOT NULL))
    OR NOT note.created

If your column name contains special characters or is a SQL Keyword, you can use backtick (`) to escape it. For nested fields, each segment of the path must be wrapped in backticks.

`CUBE` = 10 AND `column name with space` IS NOT NULL
  AND `nested with space`.`inner with space` < 2

Warning

Field names containing periods (.) are not supported.

Literals for dates, timestamps, and decimals can be written by writing the string value after the type name. For example

date_col = date '2021-01-01'
and timestamp_col = timestamp '2021-01-01 00:00:00'
and decimal_col = decimal(8,3) '1.000'

For timestamp columns, the precision can be specified as a number in the type parameter. Microsecond precision (6) is the default.

SQL	Time unit
`timestamp(0)`	Seconds
`timestamp(3)`	Milliseconds
`timestamp(6)`	Microseconds
`timestamp(9)`	Nanoseconds

Lance internally stores data in Arrow format. The mapping from SQL types to Arrow is:

SQL type	Arrow type
`boolean`	`Boolean`
`tinyint` / `tinyint unsigned`	`Int8` / `UInt8`
`smallint` / `smallint unsigned`	`Int16` / `UInt16`
`int` or `integer` / `int unsigned` or `integer unsigned`	`Int32` / `UInt32`
`bigint` / `bigint unsigned`	`Int64` / `UInt64`
`float`	`Float32`
`double`	`Float64`
`decimal(precision, scale)`	`Decimal128`
`date`	`Date32`
`timestamp`	`Timestamp` (1)
`string`	`Utf8`
`binary`	`Binary`

See precision mapping in previous table.

Random read¶

One district feature of Lance, as columnar format, is that it allows you to read random samples quickly.

# Access the 2nd, 101th and 501th rows
data = ds.take([1, 100, 500], columns=["image", "label"])

The ability to achieve fast random access to individual rows plays a crucial role in facilitating various workflows such as random sampling and shuffling in ML training. Additionally, it empowers users to construct secondary indices, enabling swift execution of queries for enhanced performance.