roc/compiler/parse/src/parser.rs

use crate::ast::Attempting;
use bumpalo::collections::vec::Vec;
use bumpalo::Bump;
use encode_unicode::CharExt;
use roc_region::all::{Located, Region};
use std::fmt;
use std::str::from_utf8;
use std::{char, u16};
use Progress::*;

/// A position in a source file.
#[derive(Clone, PartialEq, Eq)]
pub struct State<'a> {
    /// The raw input bytes from the file.
    pub bytes: &'a [u8],

    /// Current line of the input
    pub line: u32,
    /// Current column of the input
    pub column: u16,

    /// Current indentation level, in columns
    /// (so no indent is col 1 - this saves an arithmetic operation.)
    pub indent_col: u16,

    // true at the beginning of each line, then false after encountering
    // the first nonspace char on that line.
    pub is_indenting: bool,

    pub context_stack: &'a ContextStack<'a>,

    /// The original length of the string, before any bytes were consumed.
    /// This is used internally by the State::bytes_consumed() function.
    ///
    /// TODO make this private, in a way that doesn't break macros!
    pub original_len: usize,
}

#[derive(Debug, PartialEq, Eq)]
pub enum Either<First, Second> {
    First(First),
    Second(Second),
}

impl<'a> State<'a> {
    pub fn new_in(arena: &'a Bump, bytes: &'a [u8], _attempting: Attempting) -> State<'a> {
        State {
            bytes,
            line: 0,
            column: 0,
            indent_col: 0,
            is_indenting: true,
            context_stack: arena.alloc(ContextStack::Nil),
            original_len: bytes.len(),
        }
    }

    pub fn check_indent(self, arena: &'a Bump, min_indent: u16) -> Result<Self, (Bag<'a>, Self)> {
        if self.indent_col < min_indent {
            Err((
                Bag::from_state(arena, &self, FailReason::OutdentedTooFar),
                self,
            ))
        } else {
            Ok(self)
        }
    }

    /// Returns the total number of bytes consumed since the parser began parsing.
    ///
    /// So if the parser has consumed 8 bytes, this function will return 8.
    pub fn bytes_consumed(&self) -> usize {
        self.original_len - self.bytes.len()
    }

    /// Returns whether the parser has reached the end of the input
    pub fn has_reached_end(&self) -> bool {
        self.bytes.is_empty()
    }

    /// Increments the line, then resets column, indent_col, and is_indenting.
    /// Advances the input by 1, to consume the newline character.
    pub fn newline(&self, arena: &'a Bump) -> Result<Self, (Progress, Bag<'a>, Self)> {
        match self.line.checked_add(1) {
            Some(line) => Ok(State {
                bytes: &self.bytes[1..],
                line,
                column: 0,
                indent_col: 0,
                is_indenting: true,
                original_len: self.original_len,
                context_stack: arena.alloc(self.context_stack.clone()),
            }),
            None => Err((
                Progress::NoProgress,
                Bag::from_state(arena, &self, FailReason::TooManyLines),
                self.clone(),
            )),
        }
    }

    /// Use advance_spaces to advance with indenting.
    /// This assumes we are *not* advancing with spaces, or at least that
    /// any spaces on the line were preceded by non-spaces - which would mean
    /// they weren't eligible to indent anyway.
    pub fn advance_without_indenting(
        self,
        arena: &'a Bump,
        quantity: usize,
    ) -> Result<Self, (Progress, Bag<'a>, Self)> {
        match (self.column as usize).checked_add(quantity) {
            Some(column_usize) if column_usize <= u16::MAX as usize => {
                Ok(State {
                    bytes: &self.bytes[quantity..],
                    column: column_usize as u16,
                    // Once we hit a nonspace character, we are no longer indenting.
                    is_indenting: false,
                    ..self
                })
            }
            _ => Err(line_too_long(arena, self.clone())),
        }
    }
    /// Advance the parser while also indenting as appropriate.
    /// This assumes we are only advancing with spaces, since they can indent.
    pub fn advance_spaces(
        &self,
        arena: &'a Bump,
        spaces: usize,
    ) -> Result<Self, (Progress, Bag<'a>, Self)> {
        match (self.column as usize).checked_add(spaces) {
            Some(column_usize) if column_usize <= u16::MAX as usize => {
                // Spaces don't affect is_indenting; if we were previously indneting,
                // we still are, and if we already finished indenting, we're still done.
                let is_indenting = self.is_indenting;

                // If we're indenting, spaces indent us further.
                let indent_col = if is_indenting {
                    // This doesn't need to be checked_add because it's always true that
                    // indent_col <= col, so if this could possibly overflow, we would
                    // already have errored out from the column calculation.
                    //
                    // Leaving debug assertions in case this invariant someday disappers.
                    debug_assert!(u16::MAX - self.indent_col >= spaces as u16);
                    debug_assert!(spaces <= u16::MAX as usize);

                    self.indent_col + spaces as u16
                } else {
                    self.indent_col
                };

                Ok(State {
                    bytes: &self.bytes[spaces..],
                    line: self.line,
                    column: column_usize as u16,
                    indent_col,
                    is_indenting,
                    context_stack: arena.alloc(self.context_stack.clone()),
                    original_len: self.original_len,
                })
            }
            _ => Err(line_too_long(arena, self.clone())),
        }
    }

    /// Returns a Region corresponding to the current state, but
    /// with the end_col advanced by the given amount. This is
    /// useful when parsing something "manually" (using input.chars())
    /// and thus wanting a Region while not having access to loc().
    pub fn len_region(&self, length: u16) -> Region {
        Region {
            start_col: self.column,
            start_line: self.line,
            end_col: self
                .column
                .checked_add(length)
                .unwrap_or_else(|| panic!("len_region overflowed")),
            end_line: self.line,
        }
    }

    /// Return a failing ParseResult for the given FailReason
    pub fn fail<T>(
        self,
        arena: &'a Bump,
        progress: Progress,
        reason: FailReason,
    ) -> Result<(Progress, T, Self), (Progress, Bag<'a>, Self)> {
        Err((progress, Bag::from_state(arena, &self, reason), self))
    }
}

impl<'a> fmt::Debug for State<'a> {
    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
        write!(f, "State {{")?;

        match from_utf8(self.bytes) {
            Ok(string) => write!(f, "\n\tbytes: [utf8] {:?}", string)?,
            Err(_) => write!(f, "\n\tbytes: [invalid utf8] {:?}", self.bytes)?,
        }

        write!(f, "\n\t(line, col): ({}, {}),", self.line, self.column)?;
        write!(f, "\n\tindent_col: {}", self.indent_col)?;
        write!(f, "\n\tis_indenting: {:?}", self.is_indenting)?;
        write!(f, "\n\toriginal_len: {}", self.original_len)?;
        write!(f, "\n\tcontext stack: {:?}", self.context_stack)?;
        write!(f, "\n}}")
    }
}

#[test]
fn state_size() {
    // State should always be under 8 machine words, so it fits in a typical
    // cache line.
    let state_size = std::mem::size_of::<State>();
    let maximum = std::mem::size_of::<usize>() * 8;
    assert!(state_size <= maximum, "{:?} <= {:?}", state_size, maximum);
}

pub type ParseResult<'a, Output> =
    Result<(Progress, Output, State<'a>), (Progress, Bag<'a>, State<'a>)>;

#[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub enum Progress {
    MadeProgress,
    NoProgress,
}

impl Progress {
    pub fn from_lengths(before: usize, after: usize) -> Self {
        Self::from_consumed(before - after)
    }
    pub fn from_consumed(chars_consumed: usize) -> Self {
        Self::progress_when(chars_consumed != 0)
    }

    pub fn progress_when(made_progress: bool) -> Self {
        if made_progress {
            Progress::MadeProgress
        } else {
            Progress::NoProgress
        }
    }

    pub fn or(&self, other: Self) -> Self {
        if (*self == MadeProgress) || (other == MadeProgress) {
            MadeProgress
        } else {
            NoProgress
        }
    }
}

#[derive(Debug, Clone, PartialEq, Eq)]
pub enum FailReason {
    Unexpected(Region),
    OutdentedTooFar,
    ConditionFailed,
    LineTooLong(u32 /* which line was too long */),
    TooManyLines,
    Eof(Region),
    InvalidPattern,
    BadUtf8,
    ReservedKeyword(Region),
    ArgumentsBeforeEquals(Region),
    NotYetImplemented(String),
    TODO,
}

#[derive(Debug, Clone, PartialEq, Eq)]
pub enum ContextStack<'a> {
    Cons(ContextItem, &'a ContextStack<'a>),
    Nil,
}

impl<'a> ContextStack<'a> {
    fn into_vec(self) -> std::vec::Vec<ContextItem> {
        let mut result = std::vec::Vec::new();
        let mut next = &self;

        while let ContextStack::Cons(item, rest) = next {
            next = rest;

            result.push(*item);
        }

        result.reverse();

        result
    }

    pub fn uncons(&'a self) -> Option<(ContextItem, &'a Self)> {
        match self {
            ContextStack::Cons(item, rest) => Some((*item, rest)),
            ContextStack::Nil => None,
        }
    }
}

#[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub struct ContextItem {
    pub line: u32,
    pub column: u16,
    pub context: Attempting,
}

#[derive(Debug, Clone, PartialEq, Eq)]
pub struct DeadEnd<'a> {
    pub line: u32,
    pub column: u16,
    pub problem: FailReason,
    pub context_stack: ContextStack<'a>,
}

#[derive(Debug, Clone, PartialEq, Eq)]
pub struct Bag<'a>(Vec<'a, DeadEnd<'a>>);

impl<'a> Bag<'a> {
    pub fn new_in(arena: &'a Bump) -> Self {
        Bag(Vec::new_in(arena))
    }

    pub fn from_state(arena: &'a Bump, state: &State<'a>, x: FailReason) -> Self {
        let mut dead_ends = Vec::with_capacity_in(1, arena);

        let dead_end = DeadEnd {
            line: state.line,
            column: state.column,
            problem: x,
            context_stack: state.context_stack.clone(),
        };
        dead_ends.push(dead_end);

        Bag(dead_ends)
    }

    fn pop(&mut self) -> Option<DeadEnd<'a>> {
        self.0.pop()
    }

    pub fn into_parse_problem(
        mut self,
        filename: std::path::PathBuf,
        bytes: &[u8],
    ) -> ParseProblem<'_> {
        match self.pop() {
            None => unreachable!("there is a parse error, but no problem"),
            Some(dead_end) => {
                let context_stack = dead_end.context_stack.into_vec();

                ParseProblem {
                    line: dead_end.line,
                    column: dead_end.column,
                    problem: dead_end.problem,
                    context_stack,
                    filename,
                    bytes,
                }
            }
        }
    }
}

/// use std vec to escape the arena's lifetime bound
/// since this is only used when there is in fact an error
/// I think this is fine
#[derive(Debug)]
pub struct ParseProblem<'a> {
    pub line: u32,
    pub column: u16,
    pub problem: FailReason,
    pub context_stack: std::vec::Vec<ContextItem>,
    pub filename: std::path::PathBuf,
    pub bytes: &'a [u8],
}

pub fn fail<'a, T>() -> impl Parser<'a, T> {
    move |arena, state: State<'a>| {
        Err((
            NoProgress,
            Bag::from_state(arena, &state, FailReason::ConditionFailed),
            state,
        ))
    }
}

pub trait Parser<'a, Output> {
    fn parse(&self, _: &'a Bump, _: State<'a>) -> ParseResult<'a, Output>;
}

impl<'a, F, Output> Parser<'a, Output> for F
where
    F: Fn(&'a Bump, State<'a>) -> ParseResult<'a, Output>,
{
    fn parse(&self, arena: &'a Bump, state: State<'a>) -> ParseResult<'a, Output> {
        self(arena, state)
    }
}

pub fn allocated<'a, P, Val>(parser: P) -> impl Parser<'a, &'a Val>
where
    P: Parser<'a, Val>,
    Val: 'a,
{
    move |arena, state: State<'a>| {
        let (progress, answer, state) = parser.parse(arena, state)?;

        Ok((progress, &*arena.alloc(answer), state))
    }
}

pub fn not_followed_by<'a, P, ByParser, By, Val>(parser: P, by: ByParser) -> impl Parser<'a, Val>
where
    ByParser: Parser<'a, By>,
    P: Parser<'a, Val>,
{
    move |arena, state: State<'a>| {
        let original_state = state.clone();

        parser
            .parse(arena, state)
            .and_then(|(progress, answer, state)| {
                let after_parse = state.clone();

                match by.parse(arena, state) {
                    Ok((_, _, state)) => Err((
                        NoProgress,
                        Bag::from_state(arena, &state, FailReason::ConditionFailed),
                        original_state,
                    )),
                    Err(_) => Ok((progress, answer, after_parse)),
                }
            })
    }
}

pub fn not<'a, P, Val>(parser: P) -> impl Parser<'a, ()>
where
    P: Parser<'a, Val>,
{
    move |arena, state: State<'a>| {
        let original_state = state.clone();

        match parser.parse(arena, state) {
            Ok((_, _, _)) => Err((
                NoProgress,
                Bag::from_state(arena, &original_state, FailReason::ConditionFailed),
                original_state,
            )),
            Err((_, _, _)) => Ok((NoProgress, (), original_state)),
        }
    }
}

pub fn lookahead<'a, Peek, P, PeekVal, Val>(peek: Peek, parser: P) -> impl Parser<'a, Val>
where
    Peek: Parser<'a, PeekVal>,
    P: Parser<'a, Val>,
{
    move |arena, state: State<'a>| {
        let original_state = state.clone();

        peek.parse(arena, state)
            .and_then(|_| parser.parse(arena, original_state))
    }
}

pub fn and_then<'a, P1, P2, F, Before, After>(parser: P1, transform: F) -> impl Parser<'a, After>
where
    P1: Parser<'a, Before>,
    P2: Parser<'a, After>,
    F: Fn(Progress, Before) -> P2,
{
    move |arena, state| {
        parser
            .parse(arena, state)
            .and_then(|(progress, output, next_state)| {
                transform(progress, output).parse(arena, next_state)
            })
    }
}

pub fn and_then_with_indent_level<'a, P1, P2, F, Before, After>(
    parser: P1,
    transform: F,
) -> impl Parser<'a, After>
where
    P1: Parser<'a, Before>,
    P2: Parser<'a, After>,
    F: Fn(Progress, Before, u16) -> P2,
{
    move |arena, state| {
        parser
            .parse(arena, state)
            .and_then(|(progress, output, next_state)| {
                transform(progress, output, next_state.indent_col).parse(arena, next_state)
            })
    }
}

pub fn then<'a, P1, F, Before, After>(parser: P1, transform: F) -> impl Parser<'a, After>
where
    P1: Parser<'a, Before>,
    After: 'a,
    F: Fn(&'a Bump, State<'a>, Progress, Before) -> ParseResult<'a, After>,
{
    move |arena, state| {
        parser
            .parse(arena, state)
            .and_then(|(progress, output, next_state)| {
                transform(arena, next_state, progress, output)
            })
    }
}

pub fn unexpected_eof<'a>(
    arena: &'a Bump,
    state: State<'a>,
    chars_consumed: usize,
) -> (Progress, Bag<'a>, State<'a>) {
    checked_unexpected(arena, state, chars_consumed, |region| {
        FailReason::Eof(region)
    })
}

pub fn unexpected<'a>(
    arena: &'a Bump,
    chars_consumed: usize,
    _attempting: Attempting,
    state: State<'a>,
) -> (Progress, Bag<'a>, State<'a>) {
    // NOTE state is the last argument because chars_consumed often depends on the state's fields
    // having state be the final argument prevents borrowing issues
    checked_unexpected(arena, state, chars_consumed, |region| {
        FailReason::Unexpected(region)
    })
}

/// Check for line overflow, then compute a new Region based on chars_consumed
/// and provide it as a way to construct a Problem.
/// If maximum line length was exceeded, return a Problem indicating as much.
#[inline(always)]
fn checked_unexpected<'a, F>(
    arena: &'a Bump,
    state: State<'a>,
    chars_consumed: usize,
    problem_from_region: F,
) -> (Progress, Bag<'a>, State<'a>)
where
    F: FnOnce(Region) -> FailReason,
{
    match (state.column as usize).checked_add(chars_consumed) {
        // Crucially, this is < u16::MAX and not <= u16::MAX. This means if
        // column ever gets set to u16::MAX, we will automatically bail out
        // with LineTooLong - which is exactly what we want! Once a line has
        // been discovered to be too long, we don't want to parse anything else
        // until that's fixed.
        Some(end_col) if end_col < u16::MAX as usize => {
            let region = Region {
                start_col: state.column,
                end_col: end_col as u16,
                start_line: state.line,
                end_line: state.line,
            };

            let problem = problem_from_region(region);

            (
                Progress::NoProgress,
                Bag::from_state(arena, &state, problem),
                state,
            )
        }
        _ => {
            let (_progress, fail, state) = line_too_long(arena, state);
            (Progress::NoProgress, fail, state)
        }
    }
}

fn line_too_long<'a>(arena: &'a Bump, state: State<'a>) -> (Progress, Bag<'a>, State<'a>) {
    let problem = FailReason::LineTooLong(state.line);
    // Set column to MAX and advance the parser to end of input.
    // This way, all future parsers will fail on EOF, and then
    // unexpected_eof will take them back here - thus propagating
    // the initial LineTooLong error all the way to the end, even if
    // (for example) the LineTooLong initially occurs in the middle of
    // a one_of chain, which would otherwise prevent it from propagating.
    let column = u16::MAX;
    let bytes = state.bytes.get(0..state.bytes.len()).unwrap();
    let state = State {
        bytes,
        line: state.line,
        column,
        ..state
    };

    // TODO do we make progress in this case?
    // isn't this error fatal?
    (
        Progress::NoProgress,
        Bag::from_state(arena, &state, problem),
        state,
    )
}

/// A single ASCII char that isn't a newline.
/// (For newlines, use newline_char(), which handles line numbers)
pub fn ascii_char<'a>(expected: u8) -> impl Parser<'a, ()> {
    // Make sure this really is not a newline!
    debug_assert_ne!(expected, b'\n');

    move |arena, state: State<'a>| match state.bytes.first() {
        Some(&actual) if expected == actual => Ok((
            Progress::MadeProgress,
            (),
            state.advance_without_indenting(arena, 1)?,
        )),
        Some(_) => Err(unexpected(arena, 0, Attempting::Keyword, state)),
        _ => Err(unexpected_eof(arena, state, 0)),
    }
}

/// A single '\n' character.
/// Use this instead of ascii_char('\n') because it properly handles
/// incrementing the line number.
pub fn newline_char<'a>() -> impl Parser<'a, ()> {
    move |arena, state: State<'a>| match state.bytes.first() {
        Some(b'\n') => Ok((Progress::MadeProgress, (), state.newline(arena)?)),
        Some(_) => Err(unexpected(arena, 0, Attempting::Keyword, state)),
        _ => Err(unexpected_eof(arena, state, 0)),
    }
}

/// One or more ASCII hex digits. (Useful when parsing unicode escape codes,
/// which must consist entirely of ASCII hex digits.)
pub fn ascii_hex_digits<'a>() -> impl Parser<'a, &'a str> {
    move |arena, state: State<'a>| {
        let mut buf = bumpalo::collections::String::new_in(arena);

        for &byte in state.bytes.iter() {
            if (byte as char).is_ascii_hexdigit() {
                buf.push(byte as char);
            } else if buf.is_empty() {
                // We didn't find any hex digits!
                return Err(unexpected(arena, 0, Attempting::Keyword, state));
            } else {
                let state = state.advance_without_indenting(arena, buf.len())?;

                return Ok((Progress::MadeProgress, buf.into_bump_str(), state));
            }
        }

        Err(unexpected_eof(arena, state, 0))
    }
}

/// A single UTF-8-encoded char. This will both parse *and* validate that the
/// char is valid UTF-8, but it will *not* advance the state.
pub fn peek_utf8_char(state: &State) -> Result<(char, usize), FailReason> {
    if !state.bytes.is_empty() {
        match char::from_utf8_slice_start(state.bytes) {
            Ok((ch, len_utf8)) => Ok((ch, len_utf8)),
            Err(_) => Err(FailReason::BadUtf8),
        }
    } else {
        Err(FailReason::Eof(
            Region::zero(), /* TODO get a better region */
        ))
    }
}

/// A single UTF-8-encoded char, with an offset. This will both parse *and*
/// validate that the char is valid UTF-8, but it will *not* advance the state.
pub fn peek_utf8_char_at(state: &State, offset: usize) -> Result<(char, usize), FailReason> {
    if state.bytes.len() > offset {
        let bytes = &state.bytes[offset..];

        match char::from_utf8_slice_start(bytes) {
            Ok((ch, len_utf8)) => Ok((ch, len_utf8)),
            Err(_) => Err(FailReason::BadUtf8),
        }
    } else {
        Err(FailReason::Eof(
            Region::zero(), /* TODO get a better region */
        ))
    }
}

pub fn keyword<'a>(keyword: &'static str, min_indent: u16) -> impl Parser<'a, ()> {
    move |arena, state: State<'a>| {
        let initial_state = state.clone();
        // first parse the keyword characters
        let (_, _, after_keyword_state) = ascii_string(keyword).parse(arena, state)?;

        // then we must have at least one space character
        // TODO this is potentially wasteful if there are a lot of spaces
        match crate::blankspace::space1(min_indent).parse(arena, after_keyword_state.clone()) {
            Err((_, fail, _)) => {
                // this is not a keyword, maybe it's `whence` or `iffy`
                // anyway, make no progress and return the initial state
                // so we can try something else
                Err((NoProgress, fail, initial_state))
            }
            Ok((_, _, _)) => {
                // give back the state after parsing the keyword, but before the whitespace
                // that way we can attach the whitespace to whatever follows
                Ok((MadeProgress, (), after_keyword_state))
            }
        }
    }
}

/// A hardcoded string with no newlines, consisting only of ASCII characters
pub fn ascii_string<'a>(keyword: &'static str) -> impl Parser<'a, ()> {
    // Verify that this really is exclusively ASCII characters.
    // The `unsafe` block in this function relies upon this assumption!
    //
    // Also, this can't have newlines because we don't attempt to advance
    // the row in the state, only the column.
    debug_assert!(keyword.chars().all(|ch| ch.len_utf8() == 1 && ch != '\n'));

    move |arena, state: State<'a>| {
        let len = keyword.len();

        // TODO do this comparison in one SIMD instruction (on supported systems)
        match state.bytes.get(0..len) {
            Some(next_str) => {
                if next_str == keyword.as_bytes() {
                    Ok((
                        Progress::MadeProgress,
                        (),
                        state.advance_without_indenting(arena, len)?,
                    ))
                } else {
                    let (_, fail, state) = unexpected(arena, len, Attempting::Keyword, state);
                    Err((NoProgress, fail, state))
                }
            }
            _ => Err(unexpected_eof(arena, state, 0)),
        }
    }
}

/// Parse zero or more values separated by a delimiter (e.g. a comma) whose
/// values are discarded
pub fn sep_by0<'a, P, D, Val>(delimiter: D, parser: P) -> impl Parser<'a, Vec<'a, Val>>
where
    D: Parser<'a, ()>,
    P: Parser<'a, Val>,
{
    move |arena, state: State<'a>| {
        let start_bytes_len = state.bytes.len();

        match parser.parse(arena, state) {
            Ok((elem_progress, first_output, next_state)) => {
                // in practice, we want elements to make progress
                debug_assert_eq!(elem_progress, MadeProgress);

                let mut state = next_state;
                let mut buf = Vec::with_capacity_in(1, arena);

                buf.push(first_output);

                loop {
                    match delimiter.parse(arena, state) {
                        Ok((_, (), next_state)) => {
                            // If the delimiter passed, check the element parser.
                            match parser.parse(arena, next_state) {
                                Ok((element_progress, next_output, next_state)) => {
                                    // in practice, we want elements to make progress
                                    debug_assert_eq!(element_progress, MadeProgress);

                                    state = next_state;
                                    buf.push(next_output);
                                }
                                Err((_, fail, state)) => {
                                    // If the delimiter parsed, but the following
                                    // element did not, that's a fatal error.
                                    let progress =
                                        Progress::from_lengths(start_bytes_len, state.bytes.len());

                                    return Err((progress, fail, state));
                                }
                            }
                        }
                        Err((delim_progress, fail, old_state)) => match delim_progress {
                            MadeProgress => return Err((MadeProgress, fail, old_state)),
                            NoProgress => return Ok((NoProgress, buf, old_state)),
                        },
                    }
                }
            }
            Err((element_progress, fail, new_state)) => match element_progress {
                MadeProgress => Err((MadeProgress, fail, new_state)),
                NoProgress => Ok((NoProgress, Vec::new_in(arena), new_state)),
            },
        }
    }
}

/// Parse zero or more values separated by a delimiter (e.g. a comma)
/// with an optional trailing delimiter whose values are discarded
pub fn trailing_sep_by0<'a, P, D, Val>(delimiter: D, parser: P) -> impl Parser<'a, Vec<'a, Val>>
where
    D: Parser<'a, ()>,
    P: Parser<'a, Val>,
{
    move |arena, state: State<'a>| {
        let start_bytes_len = state.bytes.len();

        match parser.parse(arena, state) {
            Ok((progress, first_output, next_state)) => {
                // in practice, we want elements to make progress
                debug_assert_eq!(progress, MadeProgress);
                let mut state = next_state;
                let mut buf = Vec::with_capacity_in(1, arena);

                buf.push(first_output);

                loop {
                    match delimiter.parse(arena, state) {
                        Ok((_, (), next_state)) => {
                            // If the delimiter passed, check the element parser.
                            match parser.parse(arena, next_state) {
                                Ok((element_progress, next_output, next_state)) => {
                                    // in practice, we want elements to make progress
                                    debug_assert_eq!(element_progress, MadeProgress);

                                    state = next_state;
                                    buf.push(next_output);
                                }
                                Err((_, _fail, old_state)) => {
                                    // If the delimiter parsed, but the following
                                    // element did not, that means we saw a trailing comma
                                    let progress = Progress::from_lengths(
                                        start_bytes_len,
                                        old_state.bytes.len(),
                                    );
                                    return Ok((progress, buf, old_state));
                                }
                            }
                        }
                        Err((delim_progress, fail, old_state)) => match delim_progress {
                            MadeProgress => return Err((MadeProgress, fail, old_state)),
                            NoProgress => return Ok((NoProgress, buf, old_state)),
                        },
                    }
                }
            }
            Err((element_progress, fail, new_state)) => match element_progress {
                MadeProgress => Err((MadeProgress, fail, new_state)),
                NoProgress => Ok((NoProgress, Vec::new_in(arena), new_state)),
            },
        }
    }
}

/// Parse one or more values separated by a delimiter (e.g. a comma) whose
/// values are discarded
pub fn sep_by1<'a, P, D, Val>(delimiter: D, parser: P) -> impl Parser<'a, Vec<'a, Val>>
where
    D: Parser<'a, ()>,
    P: Parser<'a, Val>,
{
    move |arena, state: State<'a>| {
        let start_bytes_len = state.bytes.len();

        match parser.parse(arena, state) {
            Ok((progress, first_output, next_state)) => {
                debug_assert_eq!(progress, MadeProgress);
                let mut state = next_state;
                let mut buf = Vec::with_capacity_in(1, arena);

                buf.push(first_output);

                loop {
                    match delimiter.parse(arena, state) {
                        Ok((_, (), next_state)) => {
                            // If the delimiter passed, check the element parser.
                            match parser.parse(arena, next_state) {
                                Ok((_, next_output, next_state)) => {
                                    state = next_state;
                                    buf.push(next_output);
                                }
                                Err((element_progress, fail, state)) => {
                                    // If the delimiter parsed, but the following
                                    // element did not, that's a fatal error.
                                    return Err((element_progress, fail, state));
                                }
                            }
                        }
                        Err((delim_progress, fail, old_state)) => {
                            match delim_progress {
                                MadeProgress => {
                                    // fail if the delimiter made progress
                                    return Err((MadeProgress, fail, old_state));
                                }
                                NoProgress => {
                                    let progress = Progress::from_lengths(
                                        start_bytes_len,
                                        old_state.bytes.len(),
                                    );
                                    return Ok((progress, buf, old_state));
                                }
                            }
                        }
                    }
                }
            }
            Err((fail_progress, fail, new_state)) => Err((fail_progress, fail, new_state)),
        }
    }
}

pub fn fail_when_progress<'a, T>(
    progress: Progress,
    fail: Bag<'a>,
    value: T,
    state: State<'a>,
) -> ParseResult<'a, T> {
    match progress {
        MadeProgress => Err((MadeProgress, fail, state)),
        NoProgress => Ok((NoProgress, value, state)),
    }
}

pub fn satisfies<'a, P, A, F>(parser: P, predicate: F) -> impl Parser<'a, A>
where
    P: Parser<'a, A>,
    F: Fn(&A) -> bool,
{
    move |arena: &'a Bump, state: State<'a>| match parser.parse(arena, state.clone()) {
        Ok((progress, output, next_state)) if predicate(&output) => {
            Ok((progress, output, next_state))
        }
        Ok((progress, _, _)) | Err((progress, _, _)) => Err((
            progress,
            Bag::from_state(arena, &state, FailReason::ConditionFailed),
            state,
        )),
    }
}

pub fn optional<'a, P, T>(parser: P) -> impl Parser<'a, Option<T>>
where
    P: Parser<'a, T>,
{
    move |arena: &'a Bump, state: State<'a>| {
        // We have to clone this because if the optional parser fails,
        // we need to revert back to the original state.
        let original_state = state.clone();

        match parser.parse(arena, state) {
            Ok((progress, out1, state)) => Ok((progress, Some(out1), state)),
            Err((_, _, _)) => {
                // NOTE this will backtrack
                // TODO can we get rid of some of the potential backtracking?
                Ok((NoProgress, None, original_state))
            }
        }
    }
}

// MACRO COMBINATORS
//
// Using some combinators together results in combinatorial type explosion
// which makes things take forever to compile. Using macros instead avoids this!

#[macro_export]
macro_rules! loc {
    ($parser:expr) => {
        move |arena, state: $crate::parser::State<'a>| {
            use roc_region::all::{Located, Region};

            let start_col = state.column;
            let start_line = state.line;

            match $parser.parse(arena, state) {
                Ok((progress, value, state)) => {
                    let end_col = state.column;
                    let end_line = state.line;
                    let region = Region {
                        start_col,
                        start_line,
                        end_col,
                        end_line,
                    };

                    Ok((progress, Located { region, value }, state))
                }
                Err(err) => Err(err),
            }
        }
    };
}

/// If the first one parses, ignore its output and move on to parse with the second one.
#[macro_export]
macro_rules! skip_first {
    ($p1:expr, $p2:expr) => {
        move |arena, state: $crate::parser::State<'a>| {
            let original_state = state.clone();

            match $p1.parse(arena, state) {
                Ok((p1, _, state)) => match $p2.parse(arena, state) {
                    Ok((p2, out2, state)) => Ok((p1.or(p2), out2, state)),
                    Err((p2, fail, _)) => Err((p1.or(p2), fail, original_state)),
                },
                Err((progress, fail, _)) => Err((progress, fail, original_state)),
            }
        }
    };
}

/// If the first one parses, parse the second one; if it also parses, use the
/// output from the first one.
#[macro_export]
macro_rules! skip_second {
    ($p1:expr, $p2:expr) => {
        move |arena, state: $crate::parser::State<'a>| {
            let original_state = state.clone();

            match $p1.parse(arena, state) {
                Ok((p1, out1, state)) => match $p2.parse(arena, state) {
                    Ok((p2, _, state)) => Ok((p1.or(p2), out1, state)),
                    Err((p2, fail, _)) => Err((p1.or(p2), fail, original_state)),
                },
                Err((progress, fail, _)) => Err((progress, fail, original_state)),
            }
        }
    };
}

/// Parse zero or more elements between two braces (e.g. square braces).
/// Elements can be optionally surrounded by spaces, and are separated by a
/// delimiter (e.g comma-separated). Braces and delimiters get discarded.
#[macro_export]
macro_rules! collection {
    ($opening_brace:expr, $elem:expr, $delimiter:expr, $closing_brace:expr, $min_indent:expr) => {
        skip_first!(
            $opening_brace,
            skip_first!(
                // We specifically allow space characters inside here, so that
                // `[  ]` can be successfully parsed as an empty list, and then
                // changed by the formatter back into `[]`.
                //
                // We don't allow newlines or comments in the middle of empty
                // roc_collections because those are normally stored in an Expr,
                // and there's no Expr in which to store them in an empty collection!
                //
                // We could change the AST to add extra storage specifically to
                // support empty literals containing newlines or comments, but this
                // does not seem worth even the tiniest regression in compiler performance.
                zero_or_more!($crate::parser::ascii_char(b' ')),
                skip_second!(
                    $crate::parser::sep_by0(
                        $delimiter,
                        $crate::blankspace::space0_around($elem, $min_indent)
                    ),
                    $closing_brace
                )
            )
        )
    };
}

/// Parse zero or more elements between two braces (e.g. square braces).
/// Elements can be optionally surrounded by spaces, and are separated by a
/// delimiter (e.g comma-separated) with optionally a trailing delimiter.
/// Braces and delimiters get discarded.
#[macro_export]
macro_rules! collection_trailing_sep {
    ($opening_brace:expr, $elem:expr, $delimiter:expr, $closing_brace:expr, $min_indent:expr) => {
        skip_first!(
            $opening_brace,
            skip_first!(
                // We specifically allow space characters inside here, so that
                // `[  ]` can be successfully parsed as an empty list, and then
                // changed by the formatter back into `[]`.
                //
                // We don't allow newlines or comments in the middle of empty
                // roc_collections because those are normally stored in an Expr,
                // and there's no Expr in which to store them in an empty collection!
                //
                // We could change the AST to add extra storage specifically to
                // support empty literals containing newlines or comments, but this
                // does not seem worth even the tiniest regression in compiler performance.
                zero_or_more!($crate::parser::ascii_char(b' ')),
                skip_second!(
                    and!(
                        $crate::parser::trailing_sep_by0(
                            $delimiter,
                            $crate::blankspace::space0_around($elem, $min_indent)
                        ),
                        $crate::blankspace::space0($min_indent)
                    ),
                    $closing_brace
                )
            )
        )
    };
}

#[macro_export]
macro_rules! and {
    ($p1:expr, $p2:expr) => {
        move |arena: &'a bumpalo::Bump, state: $crate::parser::State<'a>| {
            // We have to clone this because if the first parser passes and then
            // the second one fails, we need to revert back to the original state.
            let original_state = state.clone();

            match $p1.parse(arena, state) {
                Ok((p1, out1, state)) => match $p2.parse(arena, state) {
                    Ok((p2, out2, state)) => Ok((p1.or(p2), (out1, out2), state)),
                    Err((p2, fail, _)) => Err((p1.or(p2), fail, original_state)),
                },
                Err((progress, fail, state)) => Err((progress, fail, state)),
            }
        }
    };
}

#[macro_export]
macro_rules! one_of {
    ($p1:expr, $p2:expr) => {
        move |arena: &'a bumpalo::Bump, state: $crate::parser::State<'a>| {

            match $p1.parse(arena, state) {
                valid @ Ok(_) => valid,
                Err((MadeProgress, fail, state)) => Err((MadeProgress, fail, state)),
                Err((NoProgress, _, state)) => $p2.parse( arena, state),
            }
        }
    };

    ($p1:expr, $($others:expr),+) => {
        one_of!($p1, one_of!($($others),+))
    };
}

#[macro_export]
macro_rules! map {
    ($parser:expr, $transform:expr) => {
        move |arena, state| {
            $parser
                .parse(arena, state)
                .map(|(progress, output, next_state)| (progress, $transform(output), next_state))
        }
    };
}

#[macro_export]
macro_rules! map_with_arena {
    ($parser:expr, $transform:expr) => {
        move |arena, state| {
            $parser
                .parse(arena, state)
                .map(|(progress, output, next_state)| {
                    (progress, $transform(arena, output), next_state)
                })
        }
    };
}

#[macro_export]
macro_rules! zero_or_more {
    ($parser:expr) => {
        move |arena, state: State<'a>| {
            use bumpalo::collections::Vec;

            let start_bytes_len = state.bytes.len();

            match $parser.parse(arena, state) {
                Ok((_, first_output, next_state)) => {
                    let mut state = next_state;
                    let mut buf = Vec::with_capacity_in(1, arena);

                    buf.push(first_output);

                    loop {
                        match $parser.parse(arena, state) {
                            Ok((_, next_output, next_state)) => {
                                state = next_state;
                                buf.push(next_output);
                            }
                            Err((fail_progress, fail, old_state)) => {
                                match fail_progress {
                                    MadeProgress => {
                                        // made progress on an element and then failed; that's an error
                                        return Err((MadeProgress, fail, old_state));
                                    }
                                    NoProgress => {
                                        // the next element failed with no progress
                                        // report whether we made progress before
                                        let progress = Progress::from_lengths(start_bytes_len, old_state.bytes.len());
                                        return Ok((progress, buf, old_state));
                                    }
                                }
                            }
                        }
                    }
                }
                Err((fail_progress, fail, new_state)) => {
                    match fail_progress {
                        MadeProgress => {
                            // made progress on an element and then failed; that's an error
                            Err((MadeProgress, fail, new_state))
                        }
                        NoProgress => {
                            // the first element failed (with no progress), but that's OK
                            // because we only need to parse 0 elements
                            Ok((NoProgress, Vec::new_in(arena), new_state))
                        }
                    }
                }
            }
        }
    };
}

#[macro_export]
macro_rules! one_or_more {
    ($parser:expr) => {
        move |arena, state: State<'a>| {
            use bumpalo::collections::Vec;

            match $parser.parse(arena, state) {
                Ok((_, first_output, next_state)) => {
                    let mut state = next_state;
                    let mut buf = Vec::with_capacity_in(1, arena);

                    buf.push(first_output);

                    loop {
                        match $parser.parse(arena, state) {
                            Ok((_, next_output, next_state)) => {
                                state = next_state;
                                buf.push(next_output);
                            }
                            Err((progress, fail, old_state)) => {
                                return $crate::parser::fail_when_progress(
                                    progress, fail, buf, old_state,
                                )
                            }
                        }
                    }
                }
                Err((progress, _, new_state)) => {
                    debug_assert_eq!(progress, NoProgress, "{:?}", &new_state);
                    Err($crate::parser::unexpected_eof(arena, new_state, 0))
                }
            }
        }
    };
}

#[macro_export]
macro_rules! debug {
    ($parser:expr) => {
        move |arena, state: $crate::parser::State<'a>| dbg!($parser.parse(arena, state))
    };
}

#[macro_export]
macro_rules! attempt {
    ($attempting:expr, $parser:expr) => {
        move |arena: &'a Bump, mut state: $crate::parser::State<'a>| {
            let item = $crate::parser::ContextItem {
                context: $attempting,
                line: state.line,
                column: state.column,
            };

            state.context_stack = arena.alloc($crate::parser::ContextStack::Cons(
                item,
                state.context_stack,
            ));

            $parser
                .parse(arena, state)
                .map(|(progress, answer, mut state)| {
                    // If the parser suceeded, go back to what we were originally attempting.
                    // (If it failed, that's exactly where we care what we were attempting!)
                    // debug_assert_eq!(!state.context_stack.is_empty());
                    match state.context_stack.uncons() {
                        Some((_item, rest)) => {
                            state.context_stack = rest;
                        }
                        None => unreachable!("context stack contains at least one element"),
                    }

                    (progress, answer, state)
                })
        }
    };
}

#[macro_export]
macro_rules! either {
    ($p1:expr, $p2:expr) => {
        move |arena: &'a bumpalo::Bump, state: $crate::parser::State<'a>| match $p1
            .parse(arena, state)
        {
            Ok((progress, output, state)) => {
                Ok((progress, $crate::parser::Either::First(output), state))
            }
            Err((NoProgress, _, state)) => match $p2.parse(arena, state) {
                Ok((progress, output, state)) => {
                    Ok((progress, $crate::parser::Either::Second(output), state))
                }
                Err((progress, fail, state)) => Err((progress, fail, state)),
            },
            Err((MadeProgress, fail, state)) => Err((MadeProgress, fail, state)),
        }
    };
}

/// Parse everything between two braces (e.g. parentheses), skipping both braces
/// and keeping only whatever was parsed in between them.
#[macro_export]
macro_rules! between {
    ($opening_brace:expr, $parser:expr, $closing_brace:expr) => {
        skip_first!($opening_brace, skip_second!($parser, $closing_brace))
    };
}

#[macro_export]
macro_rules! record_field {
    ($val_parser:expr, $min_indent:expr) => {
        move |arena: &'a bumpalo::Bump,
              state: $crate::parser::State<'a>|
              -> $crate::parser::ParseResult<'a, $crate::ast::AssignedField<'a, _>> {
            use $crate::ast::AssignedField::*;
            use $crate::blankspace::{space0, space0_before};
            use $crate::ident::lowercase_ident;
            use $crate::parser::ascii_char;
            use $crate::parser::Either::*;

            // You must have a field name, e.g. "email"
            let (progress, loc_label, state) = loc!(lowercase_ident()).parse(arena, state)?;
            debug_assert_eq!(progress, MadeProgress);

            let (_, spaces, state) = space0($min_indent).parse(arena, state)?;

            // Having a value is optional; both `{ email }` and `{ email: blah }` work.
            // (This is true in both literals and types.)
            let (_, opt_loc_val, state) = $crate::parser::optional(either!(
                skip_first!(ascii_char(b':'), space0_before($val_parser, $min_indent)),
                skip_first!(ascii_char(b'?'), space0_before($val_parser, $min_indent))
            ))
            .parse(arena, state)?;

            let answer = match opt_loc_val {
                Some(either) => match either {
                    First(loc_val) => RequiredValue(loc_label, spaces, arena.alloc(loc_val)),
                    Second(loc_val) => OptionalValue(loc_label, spaces, arena.alloc(loc_val)),
                },
                // If no value was provided, record it as a Var.
                // Canonicalize will know what to do with a Var later.
                None => {
                    if !spaces.is_empty() {
                        SpaceAfter(arena.alloc(LabelOnly(loc_label)), spaces)
                    } else {
                        LabelOnly(loc_label)
                    }
                }
            };

            Ok((MadeProgress, answer, state))
        }
    };
}

#[macro_export]
macro_rules! record_without_update {
    ($val_parser:expr, $min_indent:expr) => {
        collection_trailing_sep!(
            ascii_char(b'{'),
            loc!(record_field!($val_parser, $min_indent)),
            ascii_char(b','),
            ascii_char(b'}'),
            $min_indent
        )
    };
}

#[macro_export]
macro_rules! record {
    ($val_parser:expr, $min_indent:expr) => {
        skip_first!(
            $crate::parser::ascii_char(b'{'),
            and!(
                // You can optionally have an identifier followed by an '&' to
                // make this a record update, e.g. { Foo.user & username: "blah" }.
                $crate::parser::optional(skip_second!(
                    $crate::blankspace::space0_around(
                        // We wrap the ident in an Expr here,
                        // so that we have a Spaceable value to work with,
                        // and then in canonicalization verify that it's an Expr::Var
                        // (and not e.g. an `Expr::Access`) and extract its string.
                        loc!(map_with_arena!(
                            $crate::expr::ident(),
                            $crate::expr::ident_to_expr
                        )),
                        $min_indent
                    ),
                    $crate::parser::ascii_char(b'&')
                )),
                loc!(skip_first!(
                    // We specifically allow space characters inside here, so that
                    // `{  }` can be successfully parsed as an empty record, and then
                    // changed by the formatter back into `{}`.
                    zero_or_more!($crate::parser::ascii_char(b' ')),
                    skip_second!(
                        and!(
                            $crate::parser::trailing_sep_by0(
                                $crate::parser::ascii_char(b','),
                                $crate::blankspace::space0_around(
                                    loc!(record_field!($val_parser, $min_indent)),
                                    $min_indent
                                ),
                            ),
                            $crate::blankspace::space0($min_indent)
                        ),
                        $crate::parser::ascii_char(b'}')
                    )
                ))
            )
        )
    };
}

/// For some reason, some usages won't compile unless they use this instead of the macro version
#[inline(always)]
pub fn and<'a, P1, P2, A, B>(p1: P1, p2: P2) -> impl Parser<'a, (A, B)>
where
    P1: Parser<'a, A>,
    P2: Parser<'a, B>,
    P1: 'a,
    P2: 'a,
    A: 'a,
    B: 'a,
{
    and!(p1, p2)
}

/// For some reason, some usages won't compile unless they use this instead of the macro version
#[inline(always)]
pub fn loc<'a, P, Val>(parser: P) -> impl Parser<'a, Located<Val>>
where
    P: Parser<'a, Val>,
{
    loc!(parser)
}

/// For some reason, some usages won't compile unless they use this instead of the macro version
#[inline(always)]
pub fn map<'a, P, F, Before, After>(parser: P, transform: F) -> impl Parser<'a, After>
where
    P: Parser<'a, Before>,
    F: Fn(Before) -> After,
{
    map!(parser, transform)
}

/// For some reason, some usages won't compile unless they use this instead of the macro version
#[inline(always)]
pub fn map_with_arena<'a, P, F, Before, After>(parser: P, transform: F) -> impl Parser<'a, After>
where
    P: Parser<'a, Before>,
    P: 'a,
    F: Fn(&'a Bump, Before) -> After,
    F: 'a,
    Before: 'a,
    After: 'a,
{
    map_with_arena!(parser, transform)
}

/// For some reason, some usages won't compile unless they use this instead of the macro version
#[inline(always)]
pub fn attempt<'a, P, Val>(attempting: Attempting, parser: P) -> impl Parser<'a, Val>
where
    P: Parser<'a, Val>,
{
    attempt!(attempting, parser)
}

pub fn parse_utf8(bytes: &[u8]) -> Result<&str, FailReason> {
    match from_utf8(bytes) {
        Ok(string) => Ok(string),
        Err(_) => Err(FailReason::BadUtf8),
    }
}

pub fn end_of_file<'a>() -> impl Parser<'a, ()> {
    |arena: &'a Bump, state: State<'a>| {
        if state.has_reached_end() {
            Ok((NoProgress, (), state))
        } else {
            Err((
                NoProgress,
                Bag::from_state(arena, &state, FailReason::ConditionFailed),
                state,
            ))
        }
    }
}

pub fn backtrackable<'a, P, Val>(parser: P) -> impl Parser<'a, Val>
where
    P: Parser<'a, Val>,
{
    move |arena: &'a Bump, state: State<'a>| {
        let old_state = state.clone();

        match parser.parse(arena, state) {
            Ok((_, a, s1)) => Ok((NoProgress, a, s1)),
            Err((_, f, _)) => Err((NoProgress, f, old_state)),
        }
    }
}