Read

Reading a TACO dataset

The same logical reader opens TACO folders, ZIPs, partition lists, and TACOCATs from Python, R, or Julia.

01Open and read

One reader in three languages

Every binding exposes the same dataset model. Use read for complete samples and SQL for filters or raw metadata.

import taco

dataset = taco.open_dataset("https://data.source.coop/major-tom/core-dem/")
samples = dataset.read(files="dem.rumi")
first_100 = dataset.sql('SELECT * FROM dataset WHERE "taco:sample_index" < 100')

library(taco)

dataset <- open_dataset("https://data.source.coop/major-tom/core-dem/")
samples <- read(dataset, files = "dem.rumi")
first_100 <- sql(dataset, 'SELECT * FROM dataset WHERE "taco:sample_index" < 100')

using Taco

dataset = Taco.open_dataset("https://data.source.coop/major-tom/core-dem/")
samples = Taco.read(dataset; files=["dem.rumi"])
first_100 = Taco.sql(dataset, 'SELECT * FROM dataset WHERE "taco:sample_index" < 100')

02Reader API

Read all samples or query a subset

Operation	Meaning
`dataset.read()`	Every sample, ordered by `taco:sample_index`.
`dataset.read(files=...)`	The complete sample table with only the selected structural file columns.
`dataset.sql(query)`	A query over `dataset` or a named metadata level.
`inspect(source, query)`	The contract, structure, levels, collection, profile, or core-generated SQL without reading samples.

03Bindings

The same reader model in every language

Python, R, and Julia open the same containers, expose the collection and contract, read complete samples, and run SQL over the dataset relation or any metadata level.