graphique 1.2

GraphQL service for arrow tables and parquet data sets.

GraphQL service for arrow tables and parquet data sets. The schema for a query API is derived automatically.

Usage

% env PARQUET_PATH=... uvicorn graphique.service:app

Open http://localhost:8000/ to try out the API in GraphiQL. There is a test fixture at ./tests/fixtures/zipcodes.parquet.

% env PARQUET_PATH=... strawberry export-schema graphique.service:app.schema

outputs the graphql schema for a parquet data set.

Configuration

Graphique uses Starlette's config: in environment variables or a .env file. Config variables are used as input to a parquet dataset.

• PARQUET_PATH: path to the parquet directory or file
• FEDERATED = '': field name to extend type Query with a federated Table
• DEBUG = False: run service in debug mode, which includes timing
• COLUMNS = None: list of names, or mapping of aliases, of columns to select
• FILTERS = None: json filter query for which rows to read at startup

For more options create a custom ASGI app. Call graphique's GraphQL on an arrow Dataset, Scanner, or Table. The GraphQL Table type will be the root Query type.

Supply a mapping of names to datasets for multiple roots, and to enable federation.

import pyarrow.dataset as ds
from graphique import GraphQL

app = GraphQL(ds.dataset(...))  # Table is root query type
app = GraphQL.federated({<name>: ds.dataset(...), ...}, keys={...})  # Tables on federated fields

Start like any ASGI app.

uvicorn <module>:app

Configuration options exist to provide a convenient no-code solution, but are subject to change in the future. Using a custom app is recommended for production usage.

API

types

• Dataset: interface for an arrow dataset, scanner, or table.
• Table: implements the Dataset interface. Adds typed row, columns, and filter fields from introspecting the schema.
• Column: interface for an arrow column (a.k.a. ChunkedArray). Each arrow data type has a corresponding column implementation: Boolean, Int, Long, Float, Decimal, Date, Datetime, Time, Duration, Base64, String, List, Struct. All columns have a values field for their list of scalars. Additional fields vary by type.
• Row: scalar fields. Arrow tables are column-oriented, and graphique encourages that usage for performance. A single row field is provided for convenience, but a field for a list of rows is not. Requesting parallel columns is far more efficient.

selection

• slice: contiguous selection of rows
• filter: select rows with simple predicates
• scan: select rows and project columns with expressions

projection

• columns: provides a field for every Column in the schema
• column: access a column of any type by name
• row: provides a field for each scalar of a single row
• apply: transform columns by applying a function
• join: join tables by key columns

aggregation

• group: group by given columns, transforming the others into list columns
• partition: partition on adjacent values in given columns, transforming the others into list columns
• aggregate: apply reduce functions to list columns
• tables: return a list of tables by splitting on the scalars in list columns

ordering

• sort: sort table by given columns
• min: select rows with smallest values
• max: select rows with largest values

Performance

Graphique relies on native PyArrow routines wherever possible. Otherwise it falls back to using NumPy or custom optimizations.

By default, datasets are read on-demand, with only the necessary rows and columns scanned. Although graphique is a running service, parquet is performant at reading a subset of data. Optionally specify FILTERS in the json filter format to read a subset of rows at startup, trading-off memory for latency. An empty filter ({}) will read the whole table.

Specifying COLUMNS will limit memory usage when reading at startup (FILTERS). There is little speed difference as unused columns are inherently ignored. Optional aliasing can also be used for camel casing.

If index columns are detected in the schema metadata, then an initial filter will also attempt a binary search on tables.

Installation

% pip install graphique[server]

Dependencies

• pyarrow >=12
• strawberry-graphql[asgi,cli]
• uvicorn (or other ASGI server)

Tests

100% branch coverage.

% pytest [--cov]

Changes

1.2

• Pyarrow >=12 required
• Grouping fragments optimized
• Group by empty columns
• Batch sorting and grouping into lists

1.1

• Pyarrow >=11 required
• Python >=3.8 required
• Scannable functions added
• List aggregations deprecated
• Group by fragments
• Month day nano interval array
• min and max fields memory optimized

1.0

• Pyarrow >=10 required
• Dataset schema introspection
• Dataset scanning with selection and projection
• Binary search on sorted columns
• List aggregation, filtering, and sorting optimizations
• Compute functions generalized
• Multiple datasets and federation
• Provisional dataset join and take

0.9

• Pyarrow >=9 required
• Multi-directional sorting
• Removed unnecessary interfaces
• Filtering has stricter typing

0.8

• Pyarrow >=8 required
• Grouping and aggregation integrated
• AbstractTable interface renamed to Dataset
• Binary scalar renamed to Base64

0.7

• Pyarrow >=7 required
• FILTERS use query syntax and trigger reading the dataset
• FEDERATED field configuration
• List columns support sorting and filtering
• Group by and aggregate optimizations
• Dataset scanning

0.6

• Pyarrow >=6 required
• Group by optimized and replaced unique field
• Dictionary related optimizations
• Null consistency with arrow count functions

0.5

• Pyarrow >=5 required
• Stricter validation of inputs
• Columns can be cast to another arrow data type
• Grouping uses large list arrays with 64-bit counts
• Datasets are read on-demand or optionally at startup

0.4

• Pyarrow >=4 required
• sort updated to use new native routines
• partition tables by adjacent values and differences
• filter supports unknown column types using tagged union pattern
• Groups replaced with Table.tables and Table.aggregate fields
• Tagged unions used for filter, apply, and partition functions

0.3

• Pyarrow >=3 required
• any and all fields
• String column split field

0.2

• Pyarrow >= 2 required
• ListColumn and StructColumn types
• Groups type with aggregate field
• group and unique optimized
• Statistical fields: mode, stddev, variance
• is_in, min, and max optimized