hoard/src/data/ledger/parse.rs

use crate::data::{AccountName, UnitName, Span, Spanned, spec::SpecRoot, DataError, SourceFile};

use super::{LedgerEntry, Balance};

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

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

/*
#[derive(Clone, Copy, PartialEq, Eq, PartialOrd, Ord, Hash, Debug)]
struct InputWrap<'a>(SourceFile, &'a str);

impl<'a> chumsky::input::Input<'a> for InputWrap<'a> {
    type Token = char;
    type Span = Span;
    type Cache = ();
    type Cursor = usize;
    type MaybeToken = char;

    fn begin(self) -> (Self::Cursor, Self::Cache) {
        (0, ())
    }

    fn cursor_location(cursor: &Self::Cursor) -> usize {
        *cursor
    }

    unsafe fn span(_cache: &mut Self::Cache, range: std::ops::Range<&Self::Cursor>) -> Self::Span {
        let e : &[u8] = &[];
        Span::new(stringstore::StoredOsString::from(e), (*range.start..*range.end))
    }

    unsafe fn next_maybe(
        cache: &mut Self::Cache,
        cursor: &mut Self::Cursor,
    ) -> Option<Self::MaybeToken> {
        None
    }
}
*/

fn ledger_parser<'a>() -> impl Parser<
    'a,
    InputWithContext<'a>,
    Vec<Spanned<LedgerEntry>>,
    chumsky::extra::Full<
        chumsky::error::Rich<'a, char, Span>,
        chumsky::extra::SimpleState<&'a SpecRoot>,
        (),
    >,
> {
    let int = chumsky::text::digits(10)
        .to_slice()
        .map(|v: &str| v.parse::<usize>().unwrap());

    let datestamp = group((int, just('-').ignored(), int, just('-').ignored(), int))
        .map(|(y, _, m, _, d)| (y as u16, m as u8, d as u8));

    let mark = |m| just(m).padded_by(inline_whitespace());

    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(
                rust_decimal::Decimal::from_str_exact(s.trim()).map_err(|e| {
                    Rich::custom(span, format!("Failed to parse '{s}' as a decimal number"))
                })?,
                span
            ))
        });

    let balance = group((
        mark('-'),
        none_of(": \n\t").repeated().to_slice().map_with(|v, e| Spanned(stringstore::StoredString::new(v), e.span())),
        mark(':'),
        decimal,
        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, 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 {
            Some(sunit) => sunit,
            None => acc_spec.default_unit.map(|u| Spanned(u, span)).ok_or_else(||
                chumsky::error::Rich::custom(span, format!("No unit specified and no default unit specified for account '{}'", acc.as_ref())))?
        };

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

        Ok(Spanned::new(Balance {
            account: acc,
            amount,
            unit,
        }, span))
    });

    let entry = group((
        chumsky::text::whitespace(),
        datestamp,
        mark(':'),
        inline_whitespace(),
        chumsky::primitive::none_of("\n")
            .repeated()
            .collect::<String>(),
        chumsky::text::newline(),
        balance.repeated().at_least(1).collect(),
        chumsky::text::whitespace(),
    ))
    .map_with(|(_, datestamp, _, _, title, _, balances, _), e| Spanned::new(LedgerEntry {
        datestamp,
        title: (!title.is_empty()).then_some(title),
        balances,
    }, e.span()));

    entry.repeated().collect()
}

pub fn parse_ledger(
    source: SourceFile,
    spec: &SpecRoot,
    data: &str,
) -> Result<Vec<Spanned<LedgerEntry>>, 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().start()..e.span().end();

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

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