Arquero Test

const TERCdata = FileAttachment("data/TERC_Urzedowy_2024-06-21.csv").text()

display(TERCdata)

Arquero table from CSV

As you see below, data in the column WOJ is correctly parsed as strings.

const nullIfEmpty = s => s.length ? s : null
const TERC = aq.fromCSV(TERCdata, { 
    delimiter: ";",
    parse: {
        WOJ: String,
        POW: nullIfEmpty,
        GMI: nullIfEmpty,
        RODZ: nullIfEmpty,
        NAZWA: nullIfEmpty,
        NAZWA_DOD: nullIfEmpty,
        STAN_NA: d => new Date(Date.parse(d))
    } 
}).select(aq.not("STAN_NA"))
display(TERC.objects())
display(Inputs.table(TERC))

Arrow file from Arquero table

I don't think it follows from the Arquero reference that one needs to use type options when exporting to Arrow — unless you want specific data types, like UInt32 instead of Int64, say.

The arrow file is made by the following data loader:

import * as aq from "arquero"
import { Type } from "apache-arrow"
import {readFile} from "node:fs/promises"
import {fileURLToPath} from "node:url"

const nullIfEmpty = s => s.length ? s : null

const TERCdata = await readFile(
        fileURLToPath(
            import.meta.resolve(
                "./TERC_Urzedowy_2024-06-21.csv"
            )), "utf-8")

const TERC = aq.fromCSV(TERCdata, {
    delimiter: ";",
    parse: {
        WOJ: String,
        POW: nullIfEmpty,
        GMI: nullIfEmpty,
        RODZ: nullIfEmpty,
        NAZWA: nullIfEmpty,
        NAZWA_DOD: nullIfEmpty,
        STAN_NA: d => new Date(Date.parse(d))
    }
}).select(aq.not("STAN_NA"))
const bytes = TERC.toArrowBuffer({
    types: {
        WOJ: Type.Utf8,
        POW: Type.Utf8,
        GMI: Type.Utf8,
        RODZ: Type.Utf8,
    }
})
await process.stdout.write(bytes)

const TERCarrow = FileAttachment("data/TERC_Urzedowy_2024-06-21.arrow").arrow()

Inputs.table(TERCarrow, { layout: "auto" })

const TERCfromArrow = aq.fromArrow(TERCarrow)
display(TERCfromArrow.objects())