ibis-dune 0.1.1

Query Dune Analytics with Ibis table expressions (Trino + automatic REST fallback)

ibis-dune

Ibis backend for Dune Analytics.

Overview

ibis-dune lets you query Dune with Ibis table expressions: build filters, joins, and aggregates in Python, compile them to Dune SQL, and execute on Trino (trino.api.dune.com). When your API plan cannot use Trino, the backend automatically falls back to Dune REST /sql/execute and keeps results consistent across both paths.

The backend handles Dune-specific concerns that plain Trino backends do not:

• Schema discovery — get_schema() and schema-less sql() infer column types without hand-written schemas
• uint256 and varbinary — large integers and binary columns round-trip correctly in pandas, PyArrow, and rich previews
• Dune SQL helpers — hex_literal, raw_predicate, and raw_scalar for fragments Ibis cannot express natively
• Typed errors — DuneQueryError and DuneResultTooLargeError at the execution boundary

Install ibis-framework from PyPI, not the legacy ibis package. Both import as ibis and cannot coexist.

Install

Requires Python 3.11+.

pip install ibis-dune

This pulls in ibis-framework[trino] (12.x), dune-client (>=1.10), and sqlglot (>=26.4).

You need a Dune API key for live queries. Free-tier keys (no Trino access) work out of the box — the backend automatically falls back to REST when Trino rejects the performance tier. You do not need force_api=True.

Usage

import ibis

con = ibis.dune.connect(dune_api_key="YOUR_DUNE_API_KEY")

Or import the backend directly:

from ibis_dune import Backend

con = Backend().connect(dune_api_key="YOUR_DUNE_API_KEY")

By default queries run on Trino. If Trino rejects the performance tier, the backend switches to REST for that session. Call con.reset_to_trino() to try Trino again. Pass force_api=True to start on REST.

Connection options

Backend().connect() / ibis.dune.connect() accept:

Parameter	Default	Purpose
dune_api_key	(required)	Dune API key
force_api	False	Start on REST instead of Trino
dune_sql_performance	"medium"	REST /sql/execute performance tier
dune_api_warning_bytes	1 GiB	Log a warning when REST result exceeds this size
dune_api_max_bytes	4 GiB	Refuse to fetch REST results larger than this

After connect, con.uses_api is True when execution is routed through REST (either force_api=True or tier-triggered fallback). con.reset_to_trino() clears tier-triggered fallback but does not override force_api.

Example

import ibis

con = ibis.dune.connect(dune_api_key="YOUR_DUNE_API_KEY")

t = con.sql("SELECT CAST(42 AS BIGINT) AS n")
print(t.execute())

When querying Dune tables directly, always keep scans bounded — filter partitions and add .limit():

from ibis import _

t = (
    con.table("logs", database="ethereum")
    .filter(_.block_date == ibis.literal("2015-08-08", type="date"))
    .select("block_number", "index", "data")
    .limit(5)
)
print(t.execute())

Dune SQL helpers

Dune-specific SQL fragments compile through the backend compiler:

from ibis import _
from ibis_dune.ops import hex_literal, raw_predicate, raw_scalar

t = con.table("enterprise_evt_rented", database="iq_protocol_polygon")
t = t.filter(_.contract_address == hex_literal("0xbf9f6b1d910aa207daa400931430ef110570f8ff"))
t = t.filter(raw_predicate("evt_block_number > 1000"))
t = t.select(probe=raw_scalar("CAST(42 AS BIGINT)", "int64")).limit(5)

• hex_literal — validated hex string literal (Dune HEX syntax)
• raw_predicate — verbatim Trino SQL boolean filter
• raw_scalar — verbatim Trino SQL expression with an explicit ibis dtype

Errors

• DuneQueryError — wraps Trino and REST query failures at the execution boundary
• DuneResultTooLargeError — raised when a REST result exceeds dune_api_max_bytes

Development

git clone https://github.com/Firlej/ibis-dune.git
cd ibis-dune
python3 -m venv venv
source venv/bin/activate
pip install -e ".[dev]"

Offline tests (CI gate, no API key):

pytest -m "not integration and not trino and not parity_mock" -q

For integration tests against live Dune, copy tests/.env.example to tests/.env and set DUNE_API_KEY.

See CONTRIBUTING.md for test tiers, fixture refresh, and local workflow details.

To build and validate a release artifact locally:

python -m build
twine check dist/*

See UPSTREAM.md for upstream merge notes and publishing workflow.

License

Apache-2.0 — see LICENSE.