Skip to main content

Python API

Create Catalog​

Before coming into contact with the Table, you need to create a Catalog.

from pypaimon import CatalogFactory

# Note that keys and values are all string
catalog_options = {
    'warehouse': 'file:///path/to/warehouse'
}
catalog = CatalogFactory.create(catalog_options)

For an S3 warehouse, pass S3 authentication options with the filesystem catalog options:

from pypaimon import CatalogFactory

catalog_options = {
    'warehouse': 's3://bucket/path/to/warehouse',
    's3.endpoint': 'https://s3.amazonaws.com',
    's3.access-key': 'xxx',
    's3.secret-key': 'yyy',
    # Optional. Required for temporary credentials.
    's3.session-token': 'zzz',
    # Optional. Useful for S3 compatible stores such as MinIO.
    's3.path-style-access': 'true',
}
catalog = CatalogFactory.create(catalog_options)

For an HDFS warehouse with Kerberos authentication:

from pypaimon import CatalogFactory

catalog_options = {
    'warehouse': 'hdfs://namenode:8020/path/to/warehouse',
    # Keytab mode: automatic kinit
    'security.kerberos.login.principal': 'user@YOUR.REALM',
    'security.kerberos.login.keytab': '/path/to/user.keytab',
}
catalog = CatalogFactory.create(catalog_options)

If you have already run kinit externally, you can omit principal and keytab. PyPaimon will automatically pick up the ticket from KRB5CCNAME environment variable or the default /tmp/krb5cc_<uid>:

from pypaimon import CatalogFactory

catalog_options = {
    'warehouse': 'hdfs://namenode:8020/path/to/warehouse',
    # Ticket cache mode: uses existing Kerberos ticket
}
catalog = CatalogFactory.create(catalog_options)

To disable ticket cache auto-detection and force SIMPLE authentication, set:

catalog_options = {
    'warehouse': 'hdfs://namenode:8020/path/to/warehouse',
    'security.kerberos.login.use-ticket-cache': 'false',
}

Currently, PyPaimon supports filesystem catalog, jdbc catalog and rest catalog. See Catalog.

You can use the catalog to create table for writing data.

Create Database​

Table is located in a database. If you want to create table in a new database, you should create it.

catalog.create_database(
    name='database_name',
    ignore_if_exists=True,  # To raise error if the database exists, set False
    properties={'key': 'value'}  # optional database properties
)

Create Table​

Table schema contains fields definition, partition keys, primary keys, table options and comment. The field definition is described by pyarrow.Schema. All arguments except fields definition are optional.

Generally, there are two ways to build pyarrow.Schema.

First, you can use pyarrow.schema method directly, for example:

import pyarrow as pa

from pypaimon import Schema

pa_schema = pa.schema([
    ('dt', pa.string()),
    ('hh', pa.string()),
    ('pk', pa.int64()),
    ('value', pa.string())
])

schema = Schema.from_pyarrow_schema(
    pa_schema=pa_schema,
    partition_keys=['dt', 'hh'],
    primary_keys=['dt', 'hh', 'pk'],
    options={'bucket': '2'},
    comment='my test table')

See Data Types for all supported pyarrow-to-paimon data types mapping.

Second, if you have some Pandas data, the pa_schema can be extracted from DataFrame:

import pandas as pd
import pyarrow as pa

from pypaimon import Schema

# Example DataFrame data
data = {
    'dt': ['2024-01-01', '2024-01-01', '2024-01-02'],
    'hh': ['12', '15', '20'],
    'pk': [1, 2, 3],
    'value': ['a', 'b', 'c'],
}
dataframe = pd.DataFrame(data)

# Get Paimon Schema
record_batch = pa.RecordBatch.from_pandas(dataframe)
schema = Schema.from_pyarrow_schema(
    pa_schema=record_batch.schema,
    partition_keys=['dt', 'hh'],
    primary_keys=['dt', 'hh', 'pk'],
    options={'bucket': '2'},
    comment='my test table'
)

After building table schema, you can create corresponding table:

schema = ...
catalog.create_table(
    identifier='database_name.table_name',
    schema=schema,
    ignore_if_exists=True  # To raise error if the table exists, set False
)

# Get Table
table = catalog.get_table('database_name.table_name')

Alter Table​

Alter a table with a list of schema changes. Use SchemaChange from pypaimon.schema.schema_change and types from pypaimon.schema.data_types (e.g. AtomicType).

from pypaimon.schema.schema_change import SchemaChange
from pypaimon.schema.data_types import AtomicType

# Add column(s)
catalog.alter_table(
    'database_name.table_name',
    [
        SchemaChange.add_column('new_col', AtomicType('STRING')),
        SchemaChange.add_column('score', AtomicType('DOUBLE'), comment='optional'),
    ],
    ignore_if_not_exists=False
)

# Drop column
catalog.alter_table(
    'database_name.table_name',
    [SchemaChange.drop_column('col_name')],
    ignore_if_not_exists=False
)

Other supported changes: SchemaChange.rename_column, update_column_type, update_column_comment, set_option, remove_option, update_comment.

Batch Write​

Paimon table write is Two-Phase Commit, you can write many times, but once committed, no more data can be written.

warning

Currently, the feature of writing multiple times and committing once only supports append only table.

table = catalog.get_table('database_name.table_name')

# 1. Create table write and commit
write_builder = table.new_batch_write_builder()
table_write = write_builder.new_write()
table_commit = write_builder.new_commit()

# 2. Write data. Support 4 methods:
# 2.1 Write pandas.DataFrame
dataframe = ...
table_write.write_pandas(dataframe)

# 2.2 Write pyarrow.Table
pa_table = ...
table_write.write_arrow(pa_table)

# 2.3 Write pyarrow.RecordBatch
record_batch = ...
table_write.write_arrow_batch(record_batch)

# 3. Commit data (required for write_pandas/write_arrow/write_arrow_batch only)
commit_messages = table_write.prepare_commit()
table_commit.commit(commit_messages)

# 4. Close resources
table_write.close()
table_commit.close()

By default, the data will be appended to table. If you want to overwrite table, you should use TableWrite#overwrite API:

# overwrite whole table
write_builder = table.new_batch_write_builder().overwrite()

# overwrite partition 'dt=2024-01-01'
write_builder = table.new_batch_write_builder().overwrite({'dt': '2024-01-01'})

Commit Callback​

You can register CommitCallback instances on a TableCommit to be notified after each successful snapshot commit. This is useful for post-commit actions such as syncing metadata to external systems.

Implementations must be idempotent — a callback may be invoked more than once for the same commit if a failure occurs right after the commit succeeds.

from pypaimon.write.commit_callback import CommitCallback, CommitCallbackContext

class MyCallback(CommitCallback):
    def call(self, context: CommitCallbackContext) -> None:
        print(f"Committed snapshot {context.snapshot.id}, "
              f"{len(context.commit_entries)} entries, "
              f"identifier {context.identifier}")

    def close(self) -> None:
        pass  # release resources if needed

write_builder = table.new_batch_write_builder()
table_write = write_builder.new_write()
table_commit = write_builder.new_commit()
table_commit.add_commit_callback(MyCallback())

table_write.write_arrow(data)
table_commit.commit(table_write.prepare_commit())
table_write.close()
table_commit.close()

CommitCallbackContext provides:

FieldTypeDescription
snapshotSnapshotThe committed snapshot (id, commit_kind, time_millis, next_row_id, …)
commit_entriesList[ManifestEntry]Delta manifest entries in this commit (each carries file.first_row_id when row-tracking is enabled)
identifierintCommit identifier

Batch Read​

Predicate pushdown​

A ReadBuilder is used to build reading utils and perform filter and projection pushdown.

table = catalog.get_table('database_name.table_name')
read_builder = table.new_read_builder()

You can use PredicateBuilder to build filters and pushdown them by ReadBuilder:

# Example filter: ('f0' < 3 OR 'f1' > 6) AND 'f3' = 'A'

predicate_builder = read_builder.new_predicate_builder()

predicate1 = predicate_builder.less_than('f0', 3)
predicate2 = predicate_builder.greater_than('f1', 6)
predicate3 = predicate_builder.or_predicates([predicate1, predicate2])

predicate4 = predicate_builder.equal('f3', 'A')
predicate_5 = predicate_builder.and_predicates([predicate3, predicate4])

read_builder = read_builder.with_filter(predicate_5)

See Predicate for all supported filters and building methods. Filter by _ROW_ID: see Data Evolution.

You can also pushdown projection by ReadBuilder:

# select f3 and f2 columns
read_builder = read_builder.with_projection(['f3', 'f2'])

For tables with nested struct columns, you can project individual sub-fields using dotted names:

# Given a table with schema: id BIGINT, info ROW<name STRING, age INT>, val STRING

# Select a nested sub-field and a top-level field
read_builder = read_builder.with_projection(['info.name', 'val'])

# The result columns are flattened with underscore-joined names:
# info_name, val

This pushes the nested column pruning down to the file reader (Parquet/ORC use PyArrow's native nested projection; Avro falls back to Python-side extraction). Only the requested leaves are read from disk.

Limitations:

  • Nested projection is only supported on append-only tables. Primary-key tables that require a merge read will raise NotImplementedError.
  • ARRAY<ROW> and MAP nested paths are not yet supported.

Generate Splits​

Then you can step into Scan Plan stage to get splits:

table_scan = read_builder.new_scan()
splits = table_scan.plan().splits()

Finally, you can read data from the splits to various data format.

Read Apache Arrow​

This requires pyarrow to be installed.

You can read all the data into a pyarrow.Table:

table_read = read_builder.new_read()
pa_table = table_read.to_arrow(splits)
print(pa_table)

# pyarrow.Table
# f0: int32
# f1: string
# ----
# f0: [[1,2,3],[4,5,6],...]
# f1: [["a","b","c"],["d","e","f"],...]

You can also read data into a pyarrow.RecordBatchReader and iterate record batches:

table_read = read_builder.new_read()
for batch in table_read.to_arrow_batch_reader(splits):
    print(batch)

# pyarrow.RecordBatch
# f0: int32
# f1: string
# ----
# f0: [1,2,3]
# f1: ["a","b","c"]

Read Python Iterator​

You can read the data row by row into a native Python iterator. This is convenient for custom row-based processing logic.

table_read = read_builder.new_read()
for row in table_read.to_iterator(splits):
    print(row)

# [1,2,3]
# ["a","b","c"]

Read Pandas​

This requires pandas to be installed.

You can read all the data into a pandas.DataFrame:

table_read = read_builder.new_read()
df = table_read.to_pandas(splits)
print(df)

#    f0 f1
# 0   1  a
# 1   2  b
# 2   3  c
# 3   4  d
# ...

Read DuckDB​

This requires duckdb to be installed.

You can convert the splits into an in-memory DuckDB table and query it:

table_read = read_builder.new_read()
duckdb_con = table_read.to_duckdb(splits, 'duckdb_table')

print(duckdb_con.query("SELECT * FROM duckdb_table").fetchdf())
#    f0 f1
# 0   1  a
# 1   2  b
# 2   3  c
# 3   4  d
# ...

print(duckdb_con.query("SELECT * FROM duckdb_table WHERE f0 = 1").fetchdf())
#    f0 f1
# 0   1  a

Incremental Read​

This API allows reading data committed between two snapshot timestamps. The steps are as follows.

  • Set the option CoreOptions.INCREMENTAL_BETWEEN_TIMESTAMP on a copied table via table.copy({...}). The value must be a string: "startMillis,endMillis", where startMillis is exclusive and endMillis is inclusive.
  • Use SnapshotManager to obtain snapshot timestamps or you can determine them by yourself.
  • Read the data as above.

Example:

from pypaimon import CatalogFactory
from pypaimon.common.core_options import CoreOptions
from pypaimon.snapshot.snapshot_manager import SnapshotManager

# Prepare catalog and obtain a table
catalog = CatalogFactory.create({'warehouse': '/path/to/warehouse'})
table = catalog.get_table('default.your_table_name')

# Assume the table has at least two snapshots (1 and 2)
snapshot_manager = SnapshotManager(table)
t1 = snapshot_manager.get_snapshot_by_id(1).time_millis
t2 = snapshot_manager.get_snapshot_by_id(2).time_millis

# Read records committed between [t1, t2]
table_inc = table.copy({CoreOptions.INCREMENTAL_BETWEEN_TIMESTAMP: f"{t1},{t2}"})

read_builder = table_inc.new_read_builder()
table_scan = read_builder.new_scan()
table_read = read_builder.new_read()
splits = table_scan.plan().splits()

# To Arrow
arrow_table = table_read.to_arrow(splits)

# Or to pandas
pandas_df = table_read.to_pandas(splits)

Shard Read​

Shard Read allows you to read data in parallel by dividing the table into multiple shards. This is useful for distributed processing and parallel computation.

You can specify the shard index and total number of shards to read a specific portion of the data:

# Prepare read builder
table = catalog.get_table('database_name.table_name')
read_builder = table.new_read_builder()
table_read = read_builder.new_read()

# Read the second shard (index 1) out of 3 total shards
splits = read_builder.new_scan().with_shard(1, 3).plan().splits()

# Read all shards and concatenate results
splits1 = read_builder.new_scan().with_shard(0, 3).plan().splits()
splits2 = read_builder.new_scan().with_shard(1, 3).plan().splits()
splits3 = read_builder.new_scan().with_shard(2, 3).plan().splits()

# Combine results from all shards

all_splits = splits1 + splits2 + splits3
pa_table = table_read.to_arrow(all_splits)

Example with shard read:

import pyarrow as pa
from pypaimon import CatalogFactory, Schema

# Create catalog
catalog_options = {'warehouse': 'file:///path/to/warehouse'}
catalog = CatalogFactory.create(catalog_options)
catalog.create_database("default", False)
# Define schema
pa_schema = pa.schema([
    ('user_id', pa.int64()),
    ('item_id', pa.int64()),
    ('behavior', pa.string()),
    ('dt', pa.string()),
])

# Create table and write data
schema = Schema.from_pyarrow_schema(pa_schema, partition_keys=['dt'])
catalog.create_table('default.test_table', schema, False)
table = catalog.get_table('default.test_table')

# Write data in two batches
write_builder = table.new_batch_write_builder()

# First write
table_write = write_builder.new_write()
table_commit = write_builder.new_commit()
data1 = {
    'user_id': [1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14],
    'item_id': [1001, 1002, 1003, 1004, 1005, 1006, 1007, 1008, 1009, 1010, 1011, 1012, 1013, 1014],
    'behavior': ['a', 'b', 'c', None, 'd', 'e', 'f', 'g', 'h', 'i', 'j', 'k', 'l', 'm'],
    'dt': ['p1', 'p1', 'p2', 'p1', 'p2', 'p1', 'p2', 'p1', 'p2', 'p1', 'p2', 'p1', 'p2', 'p1'],
}
pa_table = pa.Table.from_pydict(data1, schema=pa_schema)
table_write.write_arrow(pa_table)
table_commit.commit(table_write.prepare_commit())
table_write.close()
table_commit.close()

# Second write
table_write = write_builder.new_write()
table_commit = write_builder.new_commit()
data2 = {
    'user_id': [5, 6, 7, 8, 18],
    'item_id': [1005, 1006, 1007, 1008, 1018],
    'behavior': ['e', 'f', 'g', 'h', 'z'],
    'dt': ['p2', 'p1', 'p2', 'p2', 'p1'],
}
pa_table = pa.Table.from_pydict(data2, schema=pa_schema)
table_write.write_arrow(pa_table)
table_commit.commit(table_write.prepare_commit())
table_write.close()
table_commit.close()

# Read specific shard
read_builder = table.new_read_builder()
table_read = read_builder.new_read()

# Read shard 2 out of 3 total shards
splits = read_builder.new_scan().with_shard(2, 3).plan().splits()
shard_data = table_read.to_arrow(splits)

# Verify shard distribution by reading all shards
splits1 = read_builder.new_scan().with_shard(0, 3).plan().splits()
splits2 = read_builder.new_scan().with_shard(1, 3).plan().splits()
splits3 = read_builder.new_scan().with_shard(2, 3).plan().splits()

# Combine all shards should equal full table read
all_shards_data = pa.concat_tables([
    table_read.to_arrow(splits1),
    table_read.to_arrow(splits2),
    table_read.to_arrow(splits3),
])
full_table_data = table_read.to_arrow(read_builder.new_scan().plan().splits())

Key points about shard read:

  • Shard Index: Zero-based index of the shard to read (0 to total_shards-1)
  • Total Shards: Total number of shards to divide the data into
  • Data Distribution: Data is distributed evenly across shards, with remainder rows going to the last shard
  • Parallel Processing: Each shard can be processed independently for better performance
  • Consistency: Combining all shards should produce the complete table data

Explain Scan Plan​

ReadBuilder.explain() returns a structured view of the scan plan without reading any data files. It is useful for understanding which splits a query will produce, how aggressively the pushdown pruned the input, and whether the resulting splits can be read on the zero-copy fast path.

table = catalog.get_table('default.events')
read_builder = table.new_read_builder()
predicate_builder = read_builder.new_predicate_builder()
read_builder = read_builder.with_filter(predicate_builder.equal('dt', '2026-05-16'))
print(read_builder.explain())

# == PyPaimon Scan Plan ==
# Table:              default.events (PK, HASH_FIXED)
# Snapshot:           1  (schema 0)
# Predicate:          dt = '2026-05-16'
# Projection:         <all columns>
# Limit:              <none>
#
# Partition pruning:  12 -> 4  (pruned 8)
# Bucket pruning:     4 -> 4  (pruned 0)
# File skipping:      4 -> 4  (pruned 0)
#
# Splits:             4
#   raw-convertible:  4 / 4
#   with DV:          0 / 4
#   all-above-L0:     0 / 4
#   files/split:      min=1  max=1  avg=1.00
#   size/split:       min=2.8 KiB  p50=2.9 KiB  p95=3.0 KiB  max=3.0 KiB
#
# Files:              4
# Total size:         11.6 KiB
# Estimated rows:     20   (merged: 20)
# Level histogram:    L0=4
# Deletion files:     0

Pass verbose=True to also list every split with its partition, bucket, file count, size, level histogram, and file paths:

print(read_builder.explain(verbose=True))

# ...
#
# Splits[]
#   [0] partition={'dt': '2026-05-16'} bucket=3 files=1 size=2.9 KiB rows=4 raw=True dv=False
#       levels: L0=1
#       file: /warehouse/default.db/events/dt=2026-05-16/bucket-3/data-...parquet
#   [1] partition={'dt': '2026-05-16'} bucket=2 files=1 size=2.8 KiB rows=2 raw=True dv=False
#       levels: L0=1
#       file: /warehouse/default.db/events/dt=2026-05-16/bucket-2/data-...parquet
#   ...

What the fields tell you:

  • Pushdown (Predicate / Projection / Limit): exactly what the reader sees after with_filter / with_projection / with_limit.
  • Pruning funnel (Partition pruning / Bucket pruning / File skipping): three before -> after counts that show at which stage the predicate paid off. n/a means the stage did not apply — for example, bucket pruning is reported for HASH_FIXED tables where every bucket key is pinned by the predicate, and for POSTPONE_BUCKET tables that skip their synthetic-bucket entries.
  • Split shape: raw-convertible counts splits that can be read zero-copy (no merge, no deletion-vector apply); with DV counts splits whose files need a deletion vector applied; all-above-L0 counts splits whose data lives entirely on L1+, i.e. the merge pipeline can skip the L0 buffer.
  • File aggregates: total file size + estimated rows (with the post-merge row estimate for primary-key tables in parentheses), plus a level histogram of where the data sits.
info

Cost: explain() reads the manifest list and manifest files but does not open any data files. It suppresses the manifest-reader's early bucket filter and forces single-threaded manifest decoding so the before/after counters are accurate. On tables where the early filter usually prunes aggressively (e.g. very wide HASH_FIXED tables with a tight predicate), this can make explain() measurably slower than a regular new_scan().plan().

ExplainResult is a plain dataclass — alongside the human-readable __str__ shown above, every field (partition_pruning, bucket_pruning, file_skipping, split_count, splits_raw_convertible, level_histogram, splits, ...) is addressable in Python for programmatic use.

CLI​

The same scan plan is available from the paimon command line — useful for previewing pruning effects of a predicate without writing any Python:

# Whole-table scan
paimon -c paimon.yaml table explain default.events

# Push down filter / projection / limit and list every split
paimon -c paimon.yaml table explain default.events \
    --where "dt = '2026-05-16' AND id = 7" \
    --select dt,id,val \
    --limit 100 \
    --verbose

# Machine-readable output (level_histogram keys are JSON strings)
paimon -c paimon.yaml table explain default.events --format json

--where accepts the same SQL-like syntax as paimon table read. With --format json, the result is a structured dump of ExplainResult suitable for piping into jq or further processing.

Rollback​

Paimon supports rolling back a table to a previous snapshot or tag. This is useful for undoing unwanted changes or restoring the table to a known good state.

Rollback to Snapshot​

You can rollback a table to a specific snapshot by its ID:

table = catalog.get_table('database_name.table_name')

# Rollback to snapshot 3
table.rollback_to(3)  # snapshot id

Rollback to Tag​

You can also rollback a table to a previously created tag:

table = catalog.get_table('database_name.table_name')

# Rollback to tag 'v3'
table.rollback_to('v3')  # tag name

The rollback_to method accepts either an int (snapshot ID) or a str (tag name) and automatically dispatches to the appropriate rollback logic.

Streaming Read​

Streaming reads allow you to continuously read new data as it arrives in a Paimon table. This is useful for building real-time data pipelines and ETL jobs.

Basic Streaming Read​

Use StreamReadBuilder to create a streaming scan that continuously polls for new snapshots:

table = catalog.get_table('database_name.table_name')

# Create streaming read builder
stream_builder = table.new_stream_read_builder()
stream_builder.with_poll_interval_ms(1000)  # Poll every 1 second

# Create streaming scan and table read
scan = stream_builder.new_streaming_scan()
table_read = stream_builder.new_read()

# Async streaming (recommended for ETL pipelines)
import asyncio

async def process_stream():
    async for plan in scan.stream():
        for split in plan.splits():
            arrow_batch = table_read.to_arrow([split])
            # Process the data
            print(f"Received {arrow_batch.num_rows} rows")

asyncio.run(process_stream())

Synchronous Streaming​

For simpler use cases, you can use the synchronous wrapper:

# Synchronous streaming
for plan in scan.stream_sync():
    arrow_table = table_read.to_arrow(plan.splits())
    process(arrow_table)

Manual Position Control​

You can directly read and set the scan position via next_snapshot_id:

# Save current position
saved_position = scan.next_snapshot_id

# Later, restore position
scan.next_snapshot_id = saved_position

# Or start from a specific snapshot
scan.next_snapshot_id = 42

Filtering Streaming Data​

You can apply predicates and projections to streaming reads:

stream_builder = table.new_stream_read_builder()

# Build predicate
predicate_builder = stream_builder.new_predicate_builder()
predicate = predicate_builder.greater_than('timestamp', 1704067200000)

# Apply filter and projection
stream_builder.with_filter(predicate)
stream_builder.with_projection(['id', 'name', 'timestamp'])

scan = stream_builder.new_streaming_scan()

Key points about streaming reads:

  • Poll Interval: Controls how often to check for new snapshots (default: 1000ms)
  • Initial Scan: First iteration returns all existing data, subsequent iterations return only new data
  • Commit Types: By default, only APPEND commits are processed; COMPACT and OVERWRITE are skipped

Parallel Consumption​

For high-throughput streaming, you can run multiple consumers in parallel, each reading a disjoint subset of buckets. This is similar to Kafka consumer groups.

Using with_buckets() for explicit bucket assignment:

# Consumer 0 reads buckets 0, 1, 2
stream_builder.with_buckets([0, 1, 2])

# Consumer 1 reads buckets 3, 4, 5
stream_builder.with_buckets([3, 4, 5])

Using with_bucket_filter() for custom filtering:

# Read only even buckets
stream_builder.with_bucket_filter(lambda b: b % 2 == 0)

Row Kind Support​

For changelog streams, you can include the row kind to distinguish between inserts, updates, and deletes:

stream_builder = table.new_stream_read_builder()
stream_builder.with_include_row_kind(True)

scan = stream_builder.new_streaming_scan()
table_read = stream_builder.new_read()

async for plan in scan.stream():
    arrow_table = table_read.to_arrow(plan.splits())
    for row in arrow_table.to_pylist():
        row_kind = row['_row_kind']  # +I, -U, +U, or -D
        if row_kind == '+I':
            handle_insert(row)
        elif row_kind == '-D':
            handle_delete(row)
        elif row_kind in ('-U', '+U'):
            handle_update(row)

Row kind values:

  • +I: Insert
  • -U: Update before (old value)
  • +U: Update after (new value)
  • -D: Delete

Data Types​

Scalar Types​

Python Native TypePyArrow TypePaimon Type
intpyarrow.int8()TINYINT
intpyarrow.int16()SMALLINT
intpyarrow.int32()INT
intpyarrow.int64()BIGINT
floatpyarrow.float32()FLOAT
floatpyarrow.float64()DOUBLE
boolpyarrow.bool_()BOOLEAN
strpyarrow.string()STRING, CHAR(n), VARCHAR(n)
bytespyarrow.binary()BYTES, VARBINARY(n)
bytespyarrow.binary(length)BINARY(length)
bytespyarrow.large_binary()BLOB
decimal.Decimalpyarrow.decimal128(precision, scale)DECIMAL(precision, scale)
datetime.datetimepyarrow.timestamp(unit, tz=None)TIMESTAMP(p) — unit: 's' p=0, 'ms' p=1–3, 'us' p=4–6, 'ns' p=7–9
datetime.datetimepyarrow.timestamp(unit, tz='UTC')TIMESTAMP_LTZ(p) — same unit/p mapping as above
datetime.datepyarrow.date32()DATE
datetime.timepyarrow.time32('ms')TIME(p)

Complex Types​

Python Native TypePyArrow TypePaimon Type
listpyarrow.list_(element_type)ARRAY<element_type>
dictpyarrow.map_(key_type, value_type)MAP<key, value>
dictpyarrow.struct([field, ...])ROW<field ...>

VARIANT Type​

VARIANT stores semi-structured, schema-flexible data (JSON objects, arrays, and primitives) in the Parquet Variant binary encoding.

pypaimon exposes VARIANT columns as Arrow struct<value: binary NOT NULL, metadata: binary NOT NULL> and provides GenericVariant for encoding and decoding.

Paimon supports two Parquet storage layouts for VARIANT:

  • Plain VARIANT — the standard two-field struct (value + metadata). Default for all writes.
  • Shredded VARIANT — typed sub-columns are stored alongside overflow bytes, enabling column-skipping inside the Parquet file. Controlled by the variant.shreddingSchema table option.

Read

A VARIANT column arrives as struct<value: binary, metadata: binary> in every Arrow batch. Use GenericVariant.from_arrow_struct to decode each row:

from pypaimon.data.generic_variant import GenericVariant

read_builder = table.new_read_builder()
result = read_builder.new_read().to_arrow(read_builder.new_scan().plan().splits())

for record in result.to_pylist():
    if (payload := record["payload"]) is not None:
        gv = GenericVariant.from_arrow_struct(payload)
        print(gv.to_python())   # decode to Python dict / list / scalar

from_arrow_struct is a lightweight operation — it only wraps the two raw byte arrays without parsing them. Actual variant binary decoding is deferred to to_python().

Write

Build GenericVariant values and convert them to an Arrow column with to_arrow_array:

import pyarrow as pa
from pypaimon.data.generic_variant import GenericVariant

gv1 = GenericVariant.from_python({'city': 'Beijing', 'age': 30})
gv2 = GenericVariant.from_python({'tags': [1, 2, 3], 'active': True})
# None represents SQL NULL

data = pa.table({
    'id':      pa.array([1, 2, 3], type=pa.int32()),
    'payload': GenericVariant.to_arrow_array([gv1, gv2, None]),
})

write_builder = table.new_batch_write_builder()
table_write = write_builder.new_write()
table_commit = write_builder.new_commit()
table_write.write_arrow(data)
table_commit.commit(table_write.prepare_commit())
table_write.close()
table_commit.close()

GenericVariant API:

MethodDescription
GenericVariant.from_python(obj)Build from a Python object (dict, list, int, str, …)
GenericVariant.from_arrow_struct({"value": b"...", "metadata": b"..."})Wrap raw bytes from an Arrow VARIANT struct row (read path)
GenericVariant.to_arrow_array([gv1, gv2, None, ...])Convert a list of GenericVariant (or None) to a pa.StructArray for writing
gv.to_python()Decode to native Python (dict, list, int, str, None, …)
gv.value()Return raw value bytes
gv.metadata()Return raw metadata bytes

Limitations:

  • VARIANT is only supported with Parquet file format. ORC and Avro are not supported.
  • VARIANT cannot be used as a primary key or partition key.

Predicate​

Predicate kindPredicate method
p1 and p2PredicateBuilder.and_predicates([p1, p2])
p1 or p2PredicateBuilder.or_predicates([p1, p2])
f = literalPredicateBuilder.equal(f, literal)
f != literalPredicateBuilder.not_equal(f, literal)
f < literalPredicateBuilder.less_than(f, literal)
f <= literalPredicateBuilder.less_or_equal(f, literal)
f > literalPredicateBuilder.greater_than(f, literal)
f >= literalPredicateBuilder.greater_or_equal(f, literal)
f is nullPredicateBuilder.is_null(f)
f is not nullPredicateBuilder.is_not_null(f)
f.startswith(literal)PredicateBuilder.startswith(f, literal)
f.endswith(literal)PredicateBuilder.endswith(f, literal)
f.contains(literal)PredicateBuilder.contains(f, literal)
f is in [l1, l2]PredicateBuilder.is_in(f, [l1, l2])
f is not in [l1, l2]PredicateBuilder.is_not_in(f, [l1, l2])
lower <= f <= upperPredicateBuilder.between(f, lower, upper)

Consumer Management​

Consumer management allows you to track consumption progress, prevent snapshot expiration, and resume from breakpoints.

Create ConsumerManager​

from pypaimon import CatalogFactory

# Get table and file_io
catalog = CatalogFactory.create({'warehouse': 'file:///path/to/warehouse'})
table = catalog.get_table('database_name.table_name')
file_io = table.file_io()

# Create consumer manager
manager = table.consumer_manager()

Get Consumer​

Retrieve a consumer by its ID:

from pypaimon.consumer.consumer import Consumer

consumer = manager.consumer('consumer_id')
if consumer:
    print(f"Next snapshot: {consumer.next_snapshot}")
else:
    print("Consumer not found")

Reset Consumer​

Create or reset a consumer with a new snapshot ID:

# Reset consumer to snapshot 10
manager.reset_consumer('consumer_id', Consumer(next_snapshot=10))

Delete Consumer​

Delete a consumer by its ID:

manager.delete_consumer('consumer_id')

List Consumers​

Get all consumers with their next snapshot IDs:

consumers = manager.consumers()
for consumer_id, next_snapshot in consumers.items():
    print(f"Consumer {consumer_id}: next snapshot {next_snapshot}")

List All Consumer IDs​

List all consumer IDs:

consumer_ids = manager.list_all_ids()
for consumer_id in consumer_ids:
    print(consumer_id)

Get Minimum Next Snapshot​

Get the minimum next snapshot across all consumers:

min_snapshot = manager.min_next_snapshot()
if min_snapshot:
    print(f"Minimum next snapshot: {min_snapshot}")

Expire Consumers​

Expire consumers modified before a given datetime:

from datetime import datetime, timedelta

# Expire consumers older than 1 day
expire_time = datetime.now() - timedelta(days=1)
manager.expire(expire_time)

Clear Consumers​

Clear consumers matching regular expression patterns:

# Clear all consumers starting with "test_"
manager.clear_consumers('test_.*')

# Clear all consumers except those starting with "prod_"
manager.clear_consumers(
    '.*',
    'prod_.*'
)

Branch Support​

ConsumerManager supports multiple branches:

# Custom branch
branch_manager = manager.with_branch('feature_branch')

# Each branch maintains its own consumers
print(branch_manager.consumers())  # Consumers on feature branch

Branch Management​

Branch management allows you to create multiple versions of a table, enabling parallel development and experimentation. Paimon supports creating branches from the current state or from specific tags.

info

PyPaimon provides two implementations of BranchManager:

  • FileSystemBranchManager: For tables accessed directly via filesystem (default for filesystem catalog)
  • CatalogBranchManager: For tables accessed via catalog (e.g., REST catalog)

The table.branch_manager() method automatically returns the appropriate implementation based on the table's catalog environment.

Create Branch​

Create a new branch from the current table state:

from pypaimon import CatalogFactory

catalog = CatalogFactory.create({'warehouse': 'file:///path/to/warehouse'})
table = catalog.get_table('database_name.table_name')

# Create a branch from current state
table.branch_manager().create_branch('feature_branch')

Create a branch from a specific tag:

# Create a branch from tag 'v1.0'
table.branch_manager().create_branch('feature_branch', tag_name='v1.0')

Create a branch and ignore if it already exists:

# No error if branch already exists
table.branch_manager().create_branch('feature_branch', ignore_if_exists=True)

List Branches​

List all branches for a table:

# Get all branch names
branches = table.branch_manager().branches()

for branch in branches:
    print(f"Branch: {branch}")

Check Branch Exists​

Check if a specific branch exists:

if table.branch_manager().branch_exists('feature_branch'):
    print("Branch exists")
else:
    print("Branch does not exist")

Drop Branch​

Delete an existing branch:

# Drop a branch
table.branch_manager().drop_branch('feature_branch')

Rename Branch​

Rename an existing branch to a new name:

# Rename a branch
table.branch_manager().rename_branch('old_branch_name', 'new_branch_name')
warning

The source branch must exist and cannot be the main branch. The target branch name must be valid and not already exist.

Fast Forward​

Fast forward the main branch to a specific branch:

# Fast forward main to feature branch
# This is useful when you want to merge changes from a feature branch back to main
table.branch_manager().fast_forward('feature_branch')
warning

Fast forward operation is irreversible and will replace the current state of the main branch with the target branch's state.

Branch Path Structure​

Paimon organizes branches in the file system as follows:

  • Main branch: Stored directly in the table directory (e.g., /path/to/table/)
  • Feature branches: Stored in a branch subdirectory (e.g., /path/to/table/branch/branch-feature_branch/)

Branch Name Validation​

Branch names have the following constraints:

  • Cannot be "main" (the default branch)
  • Cannot be blank or whitespace only
  • Cannot be a pure numeric string
  • Valid examples: feature, develop, feature-123, my-branch

Supported Features​

The following shows the supported features of Python Paimon compared to Java Paimon:

Catalog Level

  • FileSystemCatalog
  • RestCatalog

Filesystem & Security

  • Local filesystem (file://)
  • HDFS (hdfs://) with SIMPLE authentication
  • HDFS (hdfs://) with Kerberos authentication (keytab or ticket cache)
  • Aliyun OSS (oss://)
  • S3 (s3://)

Table Level

  • Append Tables
    • bucket = -1 (unaware)
    • bucket > 0 (fixed)
  • Primary Key Tables
    • only support deduplicate
    • bucket = -2 (postpone)
    • bucket > 0 (fixed)
    • read with deletion vectors enabled
  • Format Tables
    • PARQUET
    • CSV
    • JSON
    • ORC
    • TEXT
  • Read/Write Operations
    • Batch read and write for append tables and primary key tables
    • Predicate filtering
    • Overwrite semantics
    • Incremental reading of Delta data
    • Reading and writing blob data
    • with_shard feature
    • Rollback feature
    • Streaming reads
    • Parallel consumption with bucket filtering
    • Row kind support for changelog streams