hoard/src/data/parse.rs

use crate::prelude::*;

use data::{AccountName, DataError, Datestamp, Decimal, Spanned, UnitName, spec::SpecRoot};

use super::{Change, RawLedgerEntry, Transaction};

use chumsky::{prelude::*, text::inline_whitespace};

type InputWithContext<'a> = chumsky::input::WithContext<io::Span, &'a str>;

fn ledger_parser<'a>() -> impl Parser<
    'a,
    InputWithContext<'a>,
    Vec<RawLedgerEntry>,
    chumsky::extra::Full<
        chumsky::error::Rich<'a, char, io::Span>,
        chumsky::extra::SimpleState<&'a SpecRoot>,
        (),
    >,
> {
    let mark = |m| just(m).padded_by(inline_whitespace());

    let int = chumsky::text::digits(10)
        .to_slice()
        .map(|v: &str| v.parse::<usize>().unwrap());

    let comment = mark('#')
        .ignore_then(none_of("\n").repeated())
        .to_slice()
        .padded()
        .map_with(|s: &str, e| RawLedgerEntry::Comment(Spanned::new(s.into(), e.span())));

    let datestamp =
        group((int, just('-').ignored(), int, just('-').ignored(), int)).map(|(y, _, m, _, d)| {
            Datestamp::from_ymd_opt(y as i32, m as u32, d as u32).expect("parse error: date format did not match")
        });

    let decimal_digit = one_of("0123456789.,");
    let decimal_digits = decimal_digit
        .or(just(' ').repeated().ignore_then(decimal_digit))
        .repeated();

    let decimal = choice((just('-').ignored(), just('+').ignored(), empty()))
        .then(decimal_digits)
        .to_slice()
        .try_map(|s: &str, span| {
            Ok(Spanned::new(
                Decimal::from_str_exact(s.trim()).map_err(|e| {
                    Rich::custom(
                        span,
                        format!("Failed to parse '{s}' as a decimal number: {e}"),
                    )
                })?,
                span,
            ))
        });

    let change = group((
        mark('-'),
        none_of(": \n\t").repeated().to_slice().map_with(|v, e| Spanned(AccountName::new(v), e.span())),
        mark(':'),
        decimal,
        choice((
            mark('=').ignore_then(decimal).map(Some),
            empty().map(|_| None)
        )),
        choice((
            inline_whitespace()
                .at_least(1)
                .ignore_then(chumsky::text::ident())
                .then_ignore(inline_whitespace())
                .map_with(|u, e| Some(Spanned(UnitName::new(u), e.span()))),
            inline_whitespace().map(|_| None),
        ))
        .then_ignore(chumsky::text::newline()),
    ))
    .try_map_with(|(_, acc, _, amount, balance, unit, ), e| {
        let span = e.span();
        let spec: &mut chumsky::extra::SimpleState<&SpecRoot> = e.state();

        let Some(acc_spec) = spec.accounts.get(acc.as_ref()) else {
            return Err(chumsky::error::Rich::custom(acc.span(), "no such account"));
        };

        let unit = match (unit, acc_spec.unit) {
            (None, None) => {
                return Err(chumsky::error::Rich::custom(span, "account does not have a unit specified, so all transactions must specify units"))
            }
            (Some(unit), None) => unit,
            (None, Some(unit)) => Spanned(unit, amount.span().to_end()),
            (Some(unit1), Some(unit2)) => {
                if *unit1 != unit2 {
                    return Err(chumsky::error::Rich::custom(span, "unit mismatch between account and transaction"))
                } else {
                    unit1
                }
            }
        };

        if !spec.units.contains_key(&unit) {
            return Err(chumsky::error::Rich::custom(unit.span(), format!("no such unit '{unit}' found")))
        }

        Ok(Change {
            account: acc,
            amount,
            balance,
            unit,
            source: Some(span),
        })
    });

    let annotation = mark('[')
        .ignore_then(none_of("]\n\t").repeated().to_slice())
        .then_ignore(mark(']'))
        .map(|v: &str| String::from(v.trim()));

    let transaction = group((
        chumsky::text::whitespace(),
        datestamp,
        mark(':'),
        inline_whitespace(),
        chumsky::primitive::none_of("\n")
            .repeated()
            .collect::<String>(),
        chumsky::text::newline(),
        choice((
            annotation
                .repeated()
                .collect()
                .then_ignore(chumsky::text::newline()),
            empty().map(|_| vec![]),
        )),
        change.repeated().at_least(1).collect(),
        chumsky::text::whitespace(),
    ))
    .map_with(
        |(_, datestamp, _, _, title, _, annotations, changes, _), e| {
            RawLedgerEntry::Transaction(
                Transaction {
                    datestamp,
                    title: (!title.is_empty()).then_some(title),
                    annotations,
                    changes,
                    source: Some(e.span()),
                }
            )
        },
    );

    (transaction.or(comment)).repeated().collect()
}

pub fn parse_ledger(
    source: io::SourceFile,
    spec: &SpecRoot,
    data: &str,
) -> Result<Vec<RawLedgerEntry>, DataError> {
    let parser = ledger_parser();

    let (presult, errors) = parser
        .parse_with_state(
            data.with_context(source),
            &mut chumsky::extra::SimpleState(spec),
        )
        .into_output_errors();

    if let Some(e) = errors.first() {
        let span = *e.span();

        let report = ariadne::Report::build(ariadne::ReportKind::Error, span)
            .with_label(ariadne::Label::new(span).with_message(e.reason()))
            .finish();

        Err(report.into())
    } else {
        Ok(presult.unwrap())
    }
}