Remove line tracking in parser

This commit is contained in:
Joshua Warner 2021-12-23 21:25:46 -08:00
parent 8e1241adea
commit a13c474f6b
3 changed files with 40 additions and 28 deletions

View file

@ -1,10 +1,10 @@
use crate::ast::CommentOrNewline;
use crate::ast::Spaceable;
use crate::parser::{self, and, backtrackable, BadInputError, Parser, Progress::*};
use crate::state::JustColumn;
use crate::state::State;
use bumpalo::collections::vec::Vec;
use bumpalo::Bump;
use roc_region::all::LineColumn;
use roc_region::all::Loc;
use roc_region::all::Position;
@ -194,7 +194,7 @@ where
move |arena, mut state: State<'a>| {
let comments_and_newlines = Vec::new_in(arena);
match eat_spaces(state.bytes(), state.xyzlcol, state.pos(), comments_and_newlines) {
match eat_spaces(state.bytes(), false, state.xyzlcol, state.pos(), comments_and_newlines) {
HasTab(xyzlcol, pos) => {
// there was a tab character
let mut state = state;
@ -211,12 +211,13 @@ where
}
Good {
xyzcol: pos,
multiline,
bytes,
comments_and_newlines,
} => {
if bytes == state.bytes() {
Ok((NoProgress, &[] as &[_], state))
} else if state.xyzlcol.line != pos.line {
} else if multiline {
// we parsed at least one newline
state.indent_column = pos.column;
@ -242,16 +243,18 @@ where
enum SpaceState<'a> {
Good {
xyzcol: LineColumn,
xyzcol: JustColumn,
multiline: bool,
bytes: &'a [u8],
comments_and_newlines: Vec<'a, CommentOrNewline<'a>>,
},
HasTab(LineColumn, Position),
HasTab(JustColumn, Position),
}
fn eat_spaces<'a>(
mut bytes: &'a [u8],
mut xyzlcol: LineColumn,
mut multiline: bool,
mut xyzlcol: JustColumn,
mut pos: Position,
mut comments_and_newlines: Vec<'a, CommentOrNewline<'a>>,
) -> SpaceState<'a> {
@ -267,7 +270,7 @@ fn eat_spaces<'a>(
b'\n' => {
bytes = &bytes[1..];
pos = pos.bump_newline();
xyzlcol.line += 1;
multiline = true;
xyzlcol.column = 0;
comments_and_newlines.push(CommentOrNewline::Newline);
}
@ -281,7 +284,7 @@ fn eat_spaces<'a>(
b'#' => {
xyzlcol.column += 1;
pos = pos.bump_column(1);
return eat_line_comment(&bytes[1..], xyzlcol, pos, comments_and_newlines);
return eat_line_comment(&bytes[1..], multiline, xyzlcol, pos, comments_and_newlines);
}
_ => break,
}
@ -289,6 +292,7 @@ fn eat_spaces<'a>(
Good {
xyzcol: xyzlcol,
multiline,
bytes,
comments_and_newlines,
}
@ -296,7 +300,8 @@ fn eat_spaces<'a>(
fn eat_line_comment<'a>(
mut bytes: &'a [u8],
mut xyzlcol: LineColumn,
mut multiline: bool,
mut xyzlcol: JustColumn,
mut pos: Position,
mut comments_and_newlines: Vec<'a, CommentOrNewline<'a>>,
) -> SpaceState<'a> {
@ -318,9 +323,9 @@ fn eat_line_comment<'a>(
pos = pos.bump_newline();
comments_and_newlines.push(CommentOrNewline::DocComment(""));
xyzlcol.line += 1;
multiline = true;
xyzlcol.column = 0;
return eat_spaces(bytes, xyzlcol, pos, comments_and_newlines);
return eat_spaces(bytes, multiline, xyzlcol, pos, comments_and_newlines);
}
None => {
// consume the second #
@ -330,6 +335,7 @@ fn eat_line_comment<'a>(
return Good {
xyzcol: xyzlcol,
multiline,
bytes,
comments_and_newlines,
};
@ -357,9 +363,9 @@ fn eat_line_comment<'a>(
comments_and_newlines.push(CommentOrNewline::LineComment(comment));
}
pos = pos.bump_newline();
xyzlcol.line += 1;
multiline = true;
xyzlcol.column = 0;
return eat_spaces(&bytes[1..], xyzlcol, pos, comments_and_newlines);
return eat_spaces(&bytes[1..], multiline, xyzlcol, pos, comments_and_newlines);
}
_ => {
bytes = &bytes[1..];
@ -381,6 +387,7 @@ fn eat_line_comment<'a>(
Good {
xyzcol: xyzlcol,
multiline,
bytes,
comments_and_newlines,
}

View file

@ -11,12 +11,12 @@ use crate::parser::{
EPattern, ERecord, EString, EType, EWhen, Either, ParseResult, Parser,
};
use crate::pattern::loc_closure_param;
use crate::state::State;
use crate::state::{State, JustColumn};
use crate::type_annotation;
use bumpalo::collections::Vec;
use bumpalo::Bump;
use roc_module::called_via::{BinOp, CalledVia, UnaryOp};
use roc_region::all::{Loc, Position, Region, LineColumn};
use roc_region::all::{Loc, Position, Region};
use crate::parser::Progress::{self, *};
@ -310,7 +310,7 @@ fn unary_negate<'a>() -> impl Parser<'a, (), EExpr<'a>> {
fn parse_expr_start<'a>(
min_indent: u16,
options: ExprParseOptions,
start: LineColumn,
start: JustColumn,
arena: &'a Bump,
state: State<'a>,
) -> ParseResult<'a, Loc<Expr<'a>>, EExpr<'a>> {
@ -331,7 +331,7 @@ fn parse_expr_start<'a>(
fn parse_expr_operator_chain<'a>(
min_indent: u16,
options: ExprParseOptions,
start: LineColumn,
start: JustColumn,
arena: &'a Bump,
state: State<'a>,
) -> ParseResult<'a, Expr<'a>, EExpr<'a>> {
@ -777,7 +777,7 @@ struct DefState<'a> {
fn parse_defs_end<'a>(
options: ExprParseOptions,
start: LineColumn,
start: JustColumn,
mut def_state: DefState<'a>,
arena: &'a Bump,
state: State<'a>,
@ -892,7 +892,7 @@ fn parse_defs_end<'a>(
fn parse_defs_expr<'a>(
options: ExprParseOptions,
start: LineColumn,
start: JustColumn,
def_state: DefState<'a>,
arena: &'a Bump,
state: State<'a>,
@ -933,7 +933,7 @@ fn parse_defs_expr<'a>(
fn parse_expr_operator<'a>(
min_indent: u16,
options: ExprParseOptions,
start: LineColumn,
start: JustColumn,
mut expr_state: ExprState<'a>,
loc_op: Loc<BinOp>,
arena: &'a Bump,
@ -1222,7 +1222,7 @@ fn parse_expr_operator<'a>(
fn parse_expr_end<'a>(
min_indent: u16,
options: ExprParseOptions,
start: LineColumn,
start: JustColumn,
mut expr_state: ExprState<'a>,
arena: &'a Bump,
state: State<'a>,

View file

@ -1,7 +1,7 @@
use crate::parser::Progress::*;
use crate::parser::{BadInputError, Progress};
use bumpalo::Bump;
use roc_region::all::{Position, Region, LineColumn};
use roc_region::all::{Position, Region};
use std::fmt;
/// A position in a source file.
@ -15,19 +15,25 @@ pub struct State<'a> {
input_len: usize,
/// Current position within the input (line/column)
pub xyzlcol: LineColumn,
pub xyzlcol: JustColumn,
/// Current indentation level, in columns
/// (so no indent is col 1 - this saves an arithmetic operation.)
pub indent_column: u16,
}
#[derive(Clone, Copy)]
pub struct JustColumn {
pub column: u16,
}
impl<'a> State<'a> {
pub fn new(bytes: &'a [u8]) -> State<'a> {
State {
bytes,
input_len: bytes.len(),
xyzlcol: LineColumn::default(),
xyzlcol: JustColumn { column: 0 },
indent_column: 0,
}
}
@ -80,8 +86,7 @@ impl<'a> State<'a> {
Some(column_usize) if column_usize <= u16::MAX as usize => {
Ok(State {
bytes: &self.bytes[quantity..],
xyzlcol: LineColumn {
line: self.xyzlcol.line,
xyzlcol: JustColumn {
column: column_usize as u16,
},
// Once we hit a nonspace character, we are no longer indenting.
@ -125,8 +130,8 @@ impl<'a> fmt::Debug for State<'a> {
write!(
f,
"\n\t(line, col): ({}, {}),",
self.xyzlcol.line, self.xyzlcol.column
"\n\t(col): {},",
self.xyzlcol.column
)?;
write!(f, "\n\tindent_column: {}", self.indent_column)?;
write!(f, "\n}}")