Always combine line,column into Position

This commit is contained in:
Joshua Warner 2021-12-22 20:32:46 -08:00
parent f19220473a
commit 4d7070ce3b
22 changed files with 1181 additions and 1293 deletions

View file

@ -1,17 +1,18 @@
use crate::ast::CommentOrNewline;
use crate::ast::Spaceable;
use crate::parser::{self, and, backtrackable, BadInputError, Col, Parser, Progress::*, Row};
use crate::parser::{self, and, backtrackable, BadInputError, Parser, Progress::*};
use crate::state::State;
use bumpalo::collections::vec::Vec;
use bumpalo::Bump;
use roc_region::all::Loc;
use roc_region::all::Position;
pub fn space0_around_ee<'a, P, S, E>(
parser: P,
min_indent: u16,
space_problem: fn(BadInputError, Row, Col) -> E,
indent_before_problem: fn(Row, Col) -> E,
indent_after_problem: fn(Row, Col) -> E,
space_problem: fn(BadInputError, Position) -> E,
indent_before_problem: fn(Position) -> E,
indent_after_problem: fn(Position) -> E,
) -> impl Parser<'a, Loc<S>, E>
where
S: Spaceable<'a>,
@ -35,9 +36,9 @@ where
pub fn space0_before_optional_after<'a, P, S, E>(
parser: P,
min_indent: u16,
space_problem: fn(BadInputError, Row, Col) -> E,
indent_before_problem: fn(Row, Col) -> E,
indent_after_problem: fn(Row, Col) -> E,
space_problem: fn(BadInputError, Position) -> E,
indent_before_problem: fn(Position) -> E,
indent_after_problem: fn(Position) -> E,
) -> impl Parser<'a, Loc<S>, E>
where
S: Spaceable<'a>,
@ -100,8 +101,8 @@ where
pub fn space0_before_e<'a, P, S, E>(
parser: P,
min_indent: u16,
space_problem: fn(BadInputError, Row, Col) -> E,
indent_problem: fn(Row, Col) -> E,
space_problem: fn(BadInputError, Position) -> E,
indent_problem: fn(Position) -> E,
) -> impl Parser<'a, Loc<S>, E>
where
S: Spaceable<'a>,
@ -127,8 +128,8 @@ where
pub fn space0_after_e<'a, P, S, E>(
parser: P,
min_indent: u16,
space_problem: fn(BadInputError, Row, Col) -> E,
indent_problem: fn(Row, Col) -> E,
space_problem: fn(BadInputError, Position) -> E,
indent_problem: fn(Position) -> E,
) -> impl Parser<'a, Loc<S>, E>
where
S: Spaceable<'a>,
@ -153,24 +154,24 @@ where
pub fn check_indent<'a, E>(
min_indent: u16,
indent_problem: fn(Row, Col) -> E,
indent_problem: fn(Position) -> E,
) -> impl Parser<'a, (), E>
where
E: 'a,
{
move |_, state: State<'a>| {
if state.column >= min_indent {
if state.pos.column >= min_indent {
Ok((NoProgress, (), state))
} else {
Err((NoProgress, indent_problem(state.line, state.column), state))
Err((NoProgress, indent_problem(state.pos), state))
}
}
}
pub fn space0_e<'a, E>(
min_indent: u16,
space_problem: fn(BadInputError, Row, Col) -> E,
indent_problem: fn(Row, Col) -> E,
space_problem: fn(BadInputError, Position) -> E,
indent_problem: fn(Position) -> E,
) -> impl Parser<'a, &'a [CommentOrNewline<'a>], E>
where
E: 'a,
@ -181,8 +182,8 @@ where
#[inline(always)]
fn spaces_help_help<'a, E>(
min_indent: u16,
space_problem: fn(BadInputError, Row, Col) -> E,
indent_problem: fn(Row, Col) -> E,
space_problem: fn(BadInputError, Position) -> E,
indent_problem: fn(Position) -> E,
) -> impl Parser<'a, &'a [CommentOrNewline<'a>], E>
where
E: 'a,
@ -194,53 +195,49 @@ where
match eat_spaces(
state.bytes(),
state.line,
state.column,
state.pos,
comments_and_newlines,
) {
HasTab { row, col } => {
HasTab(pos) => {
// there was a tab character
let mut state = state;
state.line = row;
state.column = col;
state.pos = pos;
// TODO: it _seems_ like if we're changing the line/column, we should also be
// advancing the state by the corresponding number of bytes.
// Not doing this is likely a bug!
// state = state.advance(<something>);
Err((
MadeProgress,
space_problem(BadInputError::HasTab, row, col),
space_problem(BadInputError::HasTab, pos),
state,
))
}
Good {
row,
col,
pos,
bytes,
comments_and_newlines,
} => {
if bytes == state.bytes() {
Ok((NoProgress, &[] as &[_], state))
} else if state.line != row {
} else if state.pos.line != pos.line {
// we parsed at least one newline
state.indent_col = col;
state.indent_col = pos.column;
if col >= min_indent {
state.line = row;
state.column = col;
if pos.column >= min_indent {
state.pos = pos;
state = state.advance(state.bytes().len() - bytes.len());
Ok((MadeProgress, comments_and_newlines.into_bump_slice(), state))
} else {
Err((
MadeProgress,
indent_problem(state.line, state.column),
indent_problem(state.pos),
state,
))
}
} else {
state.column = col;
state.pos.column = pos.column;
state = state.advance(state.bytes().len() - bytes.len());
Ok((MadeProgress, comments_and_newlines.into_bump_slice(), state))
@ -252,21 +249,16 @@ where
enum SpaceState<'a> {
Good {
row: Row,
col: Col,
pos: Position,
bytes: &'a [u8],
comments_and_newlines: Vec<'a, CommentOrNewline<'a>>,
},
HasTab {
row: Row,
col: Col,
},
HasTab(Position),
}
fn eat_spaces<'a>(
mut bytes: &'a [u8],
mut row: Row,
mut col: Col,
mut pos: Position,
mut comments_and_newlines: Vec<'a, CommentOrNewline<'a>>,
) -> SpaceState<'a> {
use SpaceState::*;
@ -275,30 +267,30 @@ fn eat_spaces<'a>(
match c {
b' ' => {
bytes = &bytes[1..];
col += 1;
pos.column += 1;
}
b'\n' => {
bytes = &bytes[1..];
row += 1;
col = 0;
pos.line += 1;
pos.column = 0;
comments_and_newlines.push(CommentOrNewline::Newline);
}
b'\r' => {
bytes = &bytes[1..];
}
b'\t' => {
return HasTab { row, col };
return HasTab(pos);
}
b'#' => {
return eat_line_comment(&bytes[1..], row, col + 1, comments_and_newlines);
pos.column += 1;
return eat_line_comment(&bytes[1..], pos, comments_and_newlines);
}
_ => break,
}
}
Good {
row,
col,
pos,
bytes,
comments_and_newlines,
}
@ -306,8 +298,7 @@ fn eat_spaces<'a>(
fn eat_line_comment<'a>(
mut bytes: &'a [u8],
row: Row,
mut col: Col,
mut pos: Position,
mut comments_and_newlines: Vec<'a, CommentOrNewline<'a>>,
) -> SpaceState<'a> {
use SpaceState::*;
@ -316,7 +307,7 @@ fn eat_line_comment<'a>(
match bytes.get(1) {
Some(b' ') => {
bytes = &bytes[2..];
col += 2;
pos.column += 2;
true
}
@ -325,16 +316,17 @@ fn eat_line_comment<'a>(
bytes = &bytes[2..];
comments_and_newlines.push(CommentOrNewline::DocComment(""));
return eat_spaces(bytes, row + 1, 0, comments_and_newlines);
pos.line += 1;
pos.column = 0;
return eat_spaces(bytes, pos, comments_and_newlines);
}
None => {
// consume the second #
col += 1;
pos.column += 1;
bytes = &bytes[1..];
return Good {
row,
col,
pos,
bytes,
comments_and_newlines,
};
@ -347,13 +339,13 @@ fn eat_line_comment<'a>(
};
let initial = bytes;
let initial_col = col;
let initial_column = pos.column;
for c in bytes {
match c {
b'\t' => return HasTab { row, col },
b'\t' => return HasTab(pos),
b'\n' => {
let delta = (col - initial_col) as usize;
let delta = (pos.column - initial_column) as usize;
let comment = unsafe { std::str::from_utf8_unchecked(&initial[..delta]) };
if is_doc_comment {
@ -361,17 +353,19 @@ fn eat_line_comment<'a>(
} else {
comments_and_newlines.push(CommentOrNewline::LineComment(comment));
}
return eat_spaces(&bytes[1..], row + 1, 0, comments_and_newlines);
pos.line += 1;
pos.column = 0;
return eat_spaces(&bytes[1..], pos, comments_and_newlines);
}
_ => {
bytes = &bytes[1..];
col += 1;
pos.column += 1;
}
}
}
// We made it to the end of the bytes. This means there's a comment without a trailing newline.
let delta = (col - initial_col) as usize;
let delta = (pos.column - initial_column) as usize;
let comment = unsafe { std::str::from_utf8_unchecked(&initial[..delta]) };
if is_doc_comment {
@ -381,8 +375,7 @@ fn eat_line_comment<'a>(
}
Good {
row,
col,
pos,
bytes,
comments_and_newlines,
}