tacoreader

library2.4.21pypypi✓ verified 85d ago

TacoReader is a Python library for querying AI-ready datasets, supporting lazy SQL evaluation via DuckDB. It reads TACO datasets in ZIP (.tacozip), folder, or TacoCat consolidated formats. Current version: 2.4.21, requires Python >=3.10, <3.14. Release cadence: feature releases every few months, patch releases as needed.

data database ai-ml

pip install tacoreader

github ↗package ↗docs ↗changelog ↗

INSTALL

IMPORT

SIG · TACOREADER

tacoreader

datapythonv2.4.21

Install

7.8s avg

Import

1048ms

Disk

293MB

Pass rate

5/ 10

Env Coverage5 / 10

glibc

3.9–3.13

musl

3.9–3.13

Install & Compatibility

Where this runs

tested against v2.4.21 · pip install

no network on importno background threads

Install × environment matrix

Each cell = how many times install + import succeeded across repeated harness runs. Partial = flaky.
glibc = Debian/Ubuntu slim · musl = Alpine Linux

musl

py 3.10–3.95 runs

build_error

glibc

py 3.10–3.95 runs

installs and imports cleanly · install 7.8s · import 1.048s · 250MB

293MB installed

● package 293MB

Code

Verified usage

Verified import paths — ran on the pinned version, not inferred.

load

✓ from tacoreader import load

TacoDataset

✓ from tacoreader import TacoDataset

Load a TACO dataset, filter by time, and execute a SQL query.

from tacoreader import load dataset = load("path/to/dataset.tacozip") # Apply a filter and query filtered = dataset.filter("time==2024") result = filtered.sql("SELECT * FROM dataset LIMIT 5") print(result)

Debug

Known issues

breakingIn v2.4.0, `load()` accepts `pathlib.Path` objects directly. If you were converting Path to str, update to use Path directly.

fix

Replace `str(path)` with just `path` when calling `load()`.

affects: <2.4.0

deprecatedThe `base_path` parameter in `load()` is deprecated as of v2.4.0. Use `load()` with absolute paths or ensure paths are relative to the dataset root.

fix

Remove `base_path` argument and provide paths as absolute or relative to dataset.

affects: >=2.4.0

gotchaWhen using `Dataset.concat()` with pre-filtered datasets, filters are preserved. If you expect filters to be cleared, you may get empty results.

fix

Ensure you understand filter propagation; if needed, apply the same filter after concat.

affects: >=2.4.0

Errors

Common errors & fixes

ModuleNotFoundError: No module named 'tacoreader'

tacoreader is not installed or Python environment is wrong.

fix

Install with `pip install tacoreader`. Verify with `python -m pip list`. Use correct Python version (>=3.10, <3.14).

AttributeError: module 'tacoreader' has no attribute 'load'

Importing wrong symbol or using outdated version before 'load' was added.

fix

Use `from tacoreader import load` and ensure tacoreader >=2.3.0. Check version: `python -c 'import tacoreader; print(tacoreader.__version__)'`

duckdb.InvalidInputException: Invalid Input Error: No files found that match the pattern

Dataset path is incorrect or the file format is not recognized.

fix

Verify the dataset path exists and has .tacozip extension or is a folder with correct structure. Use absolute paths if ambiguous.

Upgrade

Version history

2.4.21latest on PyPI · released Mar 27, 2026

Audit

Dependencies

duckdbrequiredUsed for lazy SQL query execution.

Agent activity

18 hits · last 30 days

node

OpenAI (training)

Resources

githubhttps://github.com/tacofoundation/tacoreader ↗

packagehttps://pypi.org/project/tacoreader/ ↗

docshttps://tacofoundation.github.io/tacoreader/ ↗

changeloghttps://github.com/tacofoundation/tacoreader/releases ↗