Merge pull request #3241 from Skgland/parse-error-improvements

change `ParserError` type
This commit is contained in:
Mark Thom
2026-03-09 22:07:03 -07:00
committed by GitHub
8 changed files with 173 additions and 193 deletions

View File

@@ -106,7 +106,7 @@ impl BranchStack {
let branch_num = self let branch_num = self
.last() .last()
.map(|occurrences| occurrences.current_branch_num.clone()) .map(|occurrences| occurrences.current_branch_num.clone())
.unwrap_or_else(|| BranchNumber::default()); .unwrap_or_default();
BranchDesignator { branch_num } BranchDesignator { branch_num }
} }

View File

@@ -20,7 +20,7 @@ pub type MachineStubGen = Box<dyn Fn(&mut MachineState) -> MachineStub>;
#[derive(Debug)] #[derive(Debug)]
pub(crate) struct MachineError { pub(crate) struct MachineError {
stub: MachineStub, stub: MachineStub,
location: Option<(usize, usize)>, // line_num, col_num location: Option<Location>,
} }
// from 7.12.2 b) of 13211-1:1995 // from 7.12.2 b) of 13211-1:1995
@@ -753,14 +753,15 @@ impl MachineState {
} }
pub(super) fn error_form(&mut self, err: MachineError, src: MachineStub) -> MachineStub { pub(super) fn error_form(&mut self, err: MachineError, src: MachineStub) -> MachineStub {
if let Some((line_num, _col_num)) = err.location { if let Some(location) = err.location {
let line = location.line();
functor!( functor!(
atom!("error"), atom!("error"),
[ [
functor((err.stub)), functor((err.stub)),
functor( functor(
(atom!(":")), (atom!(":")),
[functor(src), number(line_num, (&mut self.arena))] [functor(src), number(line, (&mut self.arena))]
) )
] ]
) )
@@ -853,9 +854,9 @@ impl From<ParserError> for CompilationError {
} }
impl CompilationError { impl CompilationError {
pub(crate) fn line_and_col_num(&self) -> Option<(usize, usize)> { pub(crate) fn line_and_col_num(&self) -> Option<Location> {
match self { match self {
CompilationError::ParserError(err) => err.line_and_col_num(), CompilationError::ParserError(err) => err.location(),
_ => None, _ => None,
} }
} }

View File

@@ -1952,7 +1952,7 @@ impl MachineState {
) -> Result<Stream, ParserError> { ) -> Result<Stream, ParserError> {
match stream.peek_char() { match stream.peek_char() {
None => Ok(stream), // empty stream is handled gracefully by Lexer::eof None => Ok(stream), // empty stream is handled gracefully by Lexer::eof
Some(Err(e)) => Err(ParserError::IO(e)), Some(Err(e)) => Err(ParserError::from(e)),
Some(Ok(c)) => { Some(Ok(c)) => {
if c == '\u{feff}' { if c == '\u{feff}' {
// skip UTF-8 BOM // skip UTF-8 BOM

View File

@@ -1016,7 +1016,7 @@ impl MachineState {
self.unify_fixnum(n, nx); self.unify_fixnum(n, nx);
} }
_ => { _ => {
let err = ParserError::ParseBigInt(0, 0); let err = parser.lexer.parse_big_int_error();
let err = self.syntax_error(err); let err = self.syntax_error(err);
return Err(self.error_form(err, stub_gen())); return Err(self.error_form(err, stub_gen()));
@@ -1026,9 +1026,7 @@ impl MachineState {
return Ok(()); return Ok(());
} }
Ok(c) => { Ok(c) => {
let (line_num, col_num) = (lexer.line_num, lexer.col_num); let err = lexer.unexpected_char(c);
let err = ParserError::UnexpectedChar(c, line_num, col_num);
let err = self.syntax_error(err); let err = self.syntax_error(err);
return Err(self.error_form(err, stub_gen())); return Err(self.error_form(err, stub_gen()));
@@ -9214,7 +9212,7 @@ impl Machine {
stream.add_lines_read(parser.lines_read()); stream.add_lines_read(parser.lines_read());
} }
Ok(true) => { Ok(true) => {
stream.add_lines_read(parser.lexer.line_num); stream.add_lines_read(parser.lexer.location.line());
self.machine_st.fail = true; self.machine_st.fail = true;
} }
Err(err) => { Err(err) => {

View File

@@ -4,7 +4,6 @@
use crate::arena::*; use crate::arena::*;
use crate::atom_table::*; use crate::atom_table::*;
use crate::offset_table::*; use crate::offset_table::*;
use crate::parser::char_reader::*;
use crate::types::HeapCellValueTag; use crate::types::HeapCellValueTag;
use std::cell::{Cell, Ref, RefCell, RefMut}; use std::cell::{Cell, Ref, RefCell, RefMut};
@@ -426,72 +425,92 @@ pub enum ArithmeticError {
UninstantiatedVar, UninstantiatedVar,
} }
#[derive(Debug, Clone)]
pub struct Location {
pub(super) line: usize,
pub(super) column: usize,
}
impl Location {
// beginning of file
pub(crate) const BOF: Self = Self { line: 0, column: 0 };
pub fn line(&self) -> usize {
self.line
}
pub fn column(&self) -> usize {
self.column
}
}
#[allow(dead_code)] #[allow(dead_code)]
#[derive(Debug)] #[derive(Debug)]
pub enum ParserError { pub struct ParserError {
BackQuotedString(usize, usize), pub(crate) location: Option<Location>,
pub(crate) kind: ParserErrorKind,
}
#[allow(dead_code)]
#[derive(Debug)]
#[non_exhaustive]
pub(crate) enum ParserErrorKind {
BackQuotedString,
IO(IOError), IO(IOError),
IncompleteReduction(usize, usize), IncompleteReduction,
InfiniteFloat(usize, usize), InfiniteFloat,
InvalidSingleQuotedCharacter(char), InvalidSingleQuotedCharacter(char),
LexicalError(lexical::Error), ParseFloat,
MissingQuote(usize, usize), MissingQuote,
NonPrologChar(usize, usize), NonPrologChar,
ParseBigInt(usize, usize), ParseBigInt,
UnexpectedChar(char, usize, usize), UnexpectedChar(char),
// UnexpectedEOF, // UnexpectedEOF,
Utf8Error(usize, usize), Utf8Error,
} }
impl ParserError { impl ParserError {
pub fn line_and_col_num(&self) -> Option<(usize, usize)> { pub(crate) fn location(&self) -> Option<Location> {
match self { self.location.as_ref().cloned()
&ParserError::BackQuotedString(line_num, col_num)
| &ParserError::IncompleteReduction(line_num, col_num)
| &ParserError::InfiniteFloat(line_num, col_num)
| &ParserError::MissingQuote(line_num, col_num)
| &ParserError::NonPrologChar(line_num, col_num)
| &ParserError::ParseBigInt(line_num, col_num)
| &ParserError::UnexpectedChar(_, line_num, col_num)
| &ParserError::Utf8Error(line_num, col_num) => Some((line_num, col_num)),
_ => None,
}
} }
pub fn as_atom(&self) -> Atom { pub(crate) fn as_atom(&self) -> Atom {
match self { match &self.kind {
ParserError::BackQuotedString(..) => atom!("back_quoted_string"), ParserErrorKind::BackQuotedString => atom!("back_quoted_string"),
ParserError::IncompleteReduction(..) => atom!("incomplete_reduction"), ParserErrorKind::IncompleteReduction => atom!("incomplete_reduction"),
ParserError::InvalidSingleQuotedCharacter(..) => { ParserErrorKind::InvalidSingleQuotedCharacter(..) => {
atom!("invalid_single_quoted_character") atom!("invalid_single_quoted_character")
} }
ParserError::InfiniteFloat(..) => { ParserErrorKind::InfiniteFloat => {
atom!("infinite_float") atom!("infinite_float")
} }
ParserError::IO(e) if e.kind() == ErrorKind::UnexpectedEof => { ParserErrorKind::IO(e) if e.kind() == ErrorKind::UnexpectedEof => {
atom!("unexpected_end_of_file") atom!("unexpected_end_of_file")
} }
ParserError::IO(e) if e.kind() == ErrorKind::InvalidData => { ParserErrorKind::IO(e) if e.kind() == ErrorKind::InvalidData => {
atom!("invalid_data") atom!("invalid_data")
} }
ParserError::IO(_) => atom!("input_output_error"), ParserErrorKind::IO(_) => atom!("input_output_error"),
ParserError::LexicalError(_) => atom!("lexical_error"), ParserErrorKind::ParseFloat => atom!("parse_float"),
ParserError::MissingQuote(..) => atom!("missing_quote"), ParserErrorKind::MissingQuote => atom!("missing_quote"),
ParserError::NonPrologChar(..) => atom!("non_prolog_character"), ParserErrorKind::NonPrologChar => atom!("non_prolog_character"),
ParserError::ParseBigInt(..) => atom!("cannot_parse_big_int"), ParserErrorKind::ParseBigInt => atom!("cannot_parse_big_int"),
ParserError::UnexpectedChar(..) => atom!("unexpected_char"), ParserErrorKind::UnexpectedChar(..) => atom!("unexpected_char"),
ParserError::Utf8Error(..) => atom!("utf8_conversion_error"), ParserErrorKind::Utf8Error => atom!("utf8_conversion_error"),
} }
} }
#[inline] #[inline]
pub fn unexpected_eof() -> Self { pub(crate) fn unexpected_eof() -> Self {
ParserError::IO(std::io::Error::from(ErrorKind::UnexpectedEof)) ParserError {
location: None,
kind: ParserErrorKind::IO(std::io::Error::from(ErrorKind::UnexpectedEof)),
}
} }
#[inline] #[inline]
pub fn is_unexpected_eof(&self) -> bool { pub(crate) fn is_unexpected_eof(&self) -> bool {
if let ParserError::IO(e) = self { if let ParserErrorKind::IO(e) = &self.kind {
e.kind() == ErrorKind::UnexpectedEof e.kind() == ErrorKind::UnexpectedEof
} else { } else {
false false
@@ -499,24 +518,11 @@ impl ParserError {
} }
} }
impl From<lexical::Error> for ParserError {
fn from(e: lexical::Error) -> ParserError {
ParserError::LexicalError(e)
}
}
impl From<IOError> for ParserError { impl From<IOError> for ParserError {
fn from(e: IOError) -> ParserError { fn from(e: IOError) -> ParserError {
ParserError::IO(e) ParserError {
} location: None,
} kind: ParserErrorKind::IO(e),
impl From<&IOError> for ParserError {
fn from(error: &IOError) -> ParserError {
if error.get_ref().filter(|e| e.is::<BadUtf8Error>()).is_some() {
ParserError::Utf8Error(0, 0)
} else {
ParserError::IO(error.kind().into())
} }
} }
} }

View File

@@ -16,7 +16,10 @@ macro_rules! consume_chars_with {
match $e { match $e {
Ok(Some(c)) => $token.push(c), Ok(Some(c)) => $token.push(c),
Ok(None) => continue, Ok(None) => continue,
Err($crate::parser::ast::ParserError::UnexpectedChar(..)) => break, Err($crate::parser::ast::ParserError {
kind: $crate::parser::ast::ParserErrorKind::UnexpectedChar(..),
..
}) => break,
Err(e) => return Err(e), Err(e) => return Err(e),
} }
} }
@@ -91,27 +94,45 @@ macro_rules! try_nt {
pub(crate) struct Lexer<'a, R> { pub(crate) struct Lexer<'a, R> {
pub(crate) reader: R, pub(crate) reader: R,
pub(crate) machine_st: &'a mut MachineState, pub(crate) machine_st: &'a mut MachineState,
pub(crate) line_num: usize, pub(crate) location: Location,
pub(crate) col_num: usize,
} }
impl<'a, R: fmt::Debug> fmt::Debug for Lexer<'a, R> { impl<'a, R: fmt::Debug> fmt::Debug for Lexer<'a, R> {
fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
f.debug_struct("LexerParser") f.debug_struct("LexerParser")
.field("reader", &"&'a mut R") // Hacky solution. .field("reader", &"&'a mut R") // Hacky solution.
.field("line_num", &self.line_num) .field("location", &self.location)
.field("col_num", &self.col_num)
.finish() .finish()
} }
} }
impl<R> Lexer<'_, R> {
pub(crate) fn located_error(&self, kind: ParserErrorKind) -> ParserError {
ParserError {
location: Some(self.location.clone()),
kind,
}
}
pub(crate) fn parse_big_int_error(&self) -> ParserError {
self.located_error(ParserErrorKind::ParseBigInt)
}
pub(crate) fn incomplete_reduction(&self) -> ParserError {
self.located_error(ParserErrorKind::IncompleteReduction)
}
pub(crate) fn unexpected_char(&self, c: char) -> ParserError {
self.located_error(ParserErrorKind::UnexpectedChar(c))
}
}
impl<'a, R: CharRead> Lexer<'a, R> { impl<'a, R: CharRead> Lexer<'a, R> {
pub fn new(src: R, machine_st: &'a mut MachineState) -> Self { pub fn new(src: R, machine_st: &'a mut MachineState) -> Self {
Self { Self {
reader: src, reader: src,
machine_st, machine_st,
line_num: 0, location: Location::BOF,
col_num: 0,
} }
} }
@@ -138,10 +159,10 @@ impl<'a, R: CharRead> Lexer<'a, R> {
self.reader.consume(c.len_utf8()); self.reader.consume(c.len_utf8());
if new_line_char!(c) { if new_line_char!(c) {
self.line_num += 1; self.location.line += 1;
self.col_num = 0; self.location.column = 0;
} else { } else {
self.col_num += 1; self.location.column += 1;
} }
} }
@@ -200,10 +221,7 @@ impl<'a, R: CharRead> Lexer<'a, R> {
match comment_loop() { match comment_loop() {
Err(e) if e.is_unexpected_eof() => { Err(e) if e.is_unexpected_eof() => {
return Err(ParserError::IncompleteReduction( return Err(self.incomplete_reduction());
self.line_num,
self.col_num,
));
} }
Err(e) => { Err(e) => {
return Err(e); return Err(e);
@@ -215,7 +233,7 @@ impl<'a, R: CharRead> Lexer<'a, R> {
self.skip_char(c); self.skip_char(c);
Ok(true) Ok(true)
} else { } else {
Err(ParserError::NonPrologChar(self.line_num, self.col_num)) Err(self.located_error(ParserErrorKind::NonPrologChar))
} }
} else { } else {
self.return_char('/'); self.return_char('/');
@@ -232,7 +250,7 @@ impl<'a, R: CharRead> Lexer<'a, R> {
if !back_quote_char!(c2) { if !back_quote_char!(c2) {
self.return_char(c); self.return_char(c);
Err(ParserError::UnexpectedChar(c, self.line_num, self.col_num)) Err(self.unexpected_char(c))
} else { } else {
self.skip_char(c2); self.skip_char(c2);
Ok(c2) Ok(c2)
@@ -257,7 +275,7 @@ impl<'a, R: CharRead> Lexer<'a, R> {
Ok(None) Ok(None)
} else { } else {
self.return_char(c); self.return_char(c);
Err(ParserError::UnexpectedChar(c, self.line_num, self.col_num)) Err(self.unexpected_char(c))
} }
} else { } else {
self.get_back_quoted_char().map(Some) self.get_back_quoted_char().map(Some)
@@ -279,10 +297,13 @@ impl<'a, R: CharRead> Lexer<'a, R> {
self.skip_char(c); self.skip_char(c);
Ok(token) Ok(token)
} else { } else {
Err(ParserError::MissingQuote(self.line_num, self.col_num)) Err({
let this = &self;
this.located_error(ParserErrorKind::MissingQuote)
})
} }
} else { } else {
Err(ParserError::UnexpectedChar(c, self.line_num, self.col_num)) Err(self.unexpected_char(c))
} }
} }
@@ -313,7 +334,7 @@ impl<'a, R: CharRead> Lexer<'a, R> {
if !single_quote_char!(c2) { if !single_quote_char!(c2) {
self.return_char(c); self.return_char(c);
Err(ParserError::UnexpectedChar(c, self.line_num, self.col_num)) Err(self.unexpected_char(c))
} else { } else {
self.skip_char(c2); self.skip_char(c2);
Ok(c2) Ok(c2)
@@ -354,7 +375,7 @@ impl<'a, R: CharRead> Lexer<'a, R> {
if !double_quote_char!(c2) { if !double_quote_char!(c2) {
self.return_char(c); self.return_char(c);
Err(ParserError::UnexpectedChar(c, self.line_num, self.col_num)) Err(self.unexpected_char(c))
} else { } else {
self.skip_char(c2); self.skip_char(c2);
Ok(c2) Ok(c2)
@@ -378,7 +399,7 @@ impl<'a, R: CharRead> Lexer<'a, R> {
't' => '\t', 't' => '\t',
'n' => '\n', 'n' => '\n',
'r' => '\r', 'r' => '\r',
c => return Err(ParserError::UnexpectedChar(c, self.line_num, self.col_num)), c => return Err(self.unexpected_char(c)),
}; };
self.skip_char(c); self.skip_char(c);
@@ -396,10 +417,7 @@ impl<'a, R: CharRead> Lexer<'a, R> {
if hexadecimal_digit_char!(c) { if hexadecimal_digit_char!(c) {
self.escape_sequence_to_char(|c| hexadecimal_digit_char!(c), 16) self.escape_sequence_to_char(|c| hexadecimal_digit_char!(c), 16)
} else { } else {
Err(ParserError::IncompleteReduction( Err(self.incomplete_reduction())
self.line_num,
self.col_num,
))
} }
} }
@@ -425,17 +443,11 @@ impl<'a, R: CharRead> Lexer<'a, R> {
if backslash_char!(c) { if backslash_char!(c) {
self.skip_char(c); self.skip_char(c);
u32::from_str_radix(&token, radix).map_or_else( u32::from_str_radix(&token, radix).map_or_else(
|_| Err(ParserError::ParseBigInt(self.line_num, self.col_num)), |_| Err(self.parse_big_int_error()),
|n| { |n| char::try_from(n).map_err(|_| self.located_error(ParserErrorKind::Utf8Error)),
char::try_from(n)
.map_err(|_| ParserError::Utf8Error(self.line_num, self.col_num))
},
) )
} else { } else {
Err(ParserError::IncompleteReduction( Err(self.incomplete_reduction())
self.line_num,
self.col_num,
))
} }
} }
@@ -447,7 +459,7 @@ impl<'a, R: CharRead> Lexer<'a, R> {
Ok(c) Ok(c)
} else { } else {
if !backslash_char!(c) { if !backslash_char!(c) {
return Err(ParserError::UnexpectedChar(c, self.line_num, self.col_num)); return Err(self.unexpected_char(c));
} }
self.skip_char(c); self.skip_char(c);
@@ -478,7 +490,10 @@ impl<'a, R: CharRead> Lexer<'a, R> {
self.skip_char(c); self.skip_char(c);
Ok(token) Ok(token)
} else { } else {
Err(ParserError::MissingQuote(self.line_num, self.col_num)) Err({
let this = &self;
this.located_error(ParserErrorKind::MissingQuote)
})
} }
} }
@@ -509,7 +524,7 @@ impl<'a, R: CharRead> Lexer<'a, R> {
.map(NumberToken::Integer) .map(NumberToken::Integer)
} else { } else {
self.return_char(start); self.return_char(start);
Err(ParserError::ParseBigInt(self.line_num, self.col_num)) Err(self.parse_big_int_error())
} }
} }
@@ -540,7 +555,7 @@ impl<'a, R: CharRead> Lexer<'a, R> {
.map(NumberToken::Integer) .map(NumberToken::Integer)
} else { } else {
self.return_char(start); self.return_char(start);
Err(ParserError::ParseBigInt(self.line_num, self.col_num)) Err(self.parse_big_int_error())
} }
} }
@@ -571,7 +586,7 @@ impl<'a, R: CharRead> Lexer<'a, R> {
.map(NumberToken::Integer) .map(NumberToken::Integer)
} else { } else {
self.return_char(start); self.return_char(start);
Err(ParserError::ParseBigInt(self.line_num, self.col_num)) Err(self.parse_big_int_error())
} }
} }
@@ -644,11 +659,11 @@ impl<'a, R: CharRead> Lexer<'a, R> {
} }
} }
} else { } else {
return Err(ParserError::InvalidSingleQuotedCharacter(c)); return Err(self.located_error(ParserErrorKind::InvalidSingleQuotedCharacter(c)));
} }
} else { } else {
match self.get_back_quoted_string() { match self.get_back_quoted_string() {
Ok(_) => return Err(ParserError::BackQuotedString(self.line_num, self.col_num)), Ok(_) => return Err(self.located_error(ParserErrorKind::BackQuotedString)),
Err(e) => return Err(e), Err(e) => return Err(e),
} }
} }
@@ -669,7 +684,7 @@ impl<'a, R: CharRead> Lexer<'a, R> {
) -> Result<(F64Offset, OrderedFloat<f64>), ParserError> { ) -> Result<(F64Offset, OrderedFloat<f64>), ParserError> {
self.return_char(token.pop().unwrap()); self.return_char(token.pop().unwrap());
let n = parse_float_lossy(&token)?; let n = self.parse_float_lossy(&token)?;
let offset = float_alloc!(n, self.machine_st.arena); let offset = float_alloc!(n, self.machine_st.arena);
Ok((offset, OrderedFloat(n))) Ok((offset, OrderedFloat(n)))
@@ -686,7 +701,7 @@ impl<'a, R: CharRead> Lexer<'a, R> {
if decimal_digit_char!(c) { if decimal_digit_char!(c) {
Ok(c) Ok(c)
} else { } else {
Err(ParserError::ParseBigInt(self.line_num, self.col_num)) Err(self.parse_big_int_error())
} }
} else { } else {
Ok(c) Ok(c)
@@ -708,7 +723,7 @@ impl<'a, R: CharRead> Lexer<'a, R> {
.or_else(|_| { .or_else(|_| {
Integer::from_str_radix(token, radix) Integer::from_str_radix(token, radix)
.map(|n| GInteger::Integer(arena_alloc!(n, &mut self.machine_st.arena))) .map(|n| GInteger::Integer(arena_alloc!(n, &mut self.machine_st.arena)))
.map_err(|_| ParserError::ParseBigInt(self.line_num, self.col_num)) .map_err(|_| self.parse_big_int_error())
}) })
} }
@@ -806,7 +821,7 @@ impl<'a, R: CharRead> Lexer<'a, R> {
} }
} }
let n = parse_float_lossy(&token)?; let n = self.parse_float_lossy(&token)?;
let offset = float_alloc!(n, self.machine_st.arena); let offset = float_alloc!(n, self.machine_st.arena);
Ok(NumberToken::Float(offset, OrderedFloat(n))) Ok(NumberToken::Float(offset, OrderedFloat(n)))
@@ -815,7 +830,7 @@ impl<'a, R: CharRead> Lexer<'a, R> {
.map(|(offset, fl)| NumberToken::Float(offset, fl)) .map(|(offset, fl)| NumberToken::Float(offset, fl))
} }
} else { } else {
let n = parse_float_lossy(&token)?; let n = self.parse_float_lossy(&token)?;
let offset = float_alloc!(n, self.machine_st.arena); let offset = float_alloc!(n, self.machine_st.arena);
Ok(NumberToken::Float(offset, OrderedFloat(n))) Ok(NumberToken::Float(offset, OrderedFloat(n)))
} }
@@ -826,7 +841,7 @@ impl<'a, R: CharRead> Lexer<'a, R> {
} else if token.starts_with('0') && token.len() == 1 { } else if token.starts_with('0') && token.len() == 1 {
if c == 'x' { if c == 'x' {
self.hexadecimal_constant(c).or_else(|e| { self.hexadecimal_constant(c).or_else(|e| {
if let ParserError::ParseBigInt(..) = e { if let ParserErrorKind::ParseBigInt = e.kind {
self.parse_integer(&token).map(NumberToken::Integer) self.parse_integer(&token).map(NumberToken::Integer)
} else { } else {
Err(e) Err(e)
@@ -834,7 +849,7 @@ impl<'a, R: CharRead> Lexer<'a, R> {
}) })
} else if c == 'o' { } else if c == 'o' {
self.octal_constant(c).or_else(|e| { self.octal_constant(c).or_else(|e| {
if let ParserError::ParseBigInt(..) = e { if let ParserErrorKind::ParseBigInt = e.kind {
self.parse_integer(&token).map(NumberToken::Integer) self.parse_integer(&token).map(NumberToken::Integer)
} else { } else {
Err(e) Err(e)
@@ -842,7 +857,7 @@ impl<'a, R: CharRead> Lexer<'a, R> {
}) })
} else if c == 'b' { } else if c == 'b' {
self.binary_constant(c).or_else(|e| { self.binary_constant(c).or_else(|e| {
if let ParserError::ParseBigInt(..) = e { if let ParserErrorKind::ParseBigInt = e.kind {
self.parse_integer(&token).map(NumberToken::Integer) self.parse_integer(&token).map(NumberToken::Integer)
} else { } else {
Err(e) Err(e)
@@ -871,9 +886,9 @@ impl<'a, R: CharRead> Lexer<'a, R> {
self.get_single_quoted_char() self.get_single_quoted_char()
.map(|c| NumberToken::Integer(GInteger::Fixnum(Fixnum::build_with(c)))) .map(|c| NumberToken::Integer(GInteger::Fixnum(Fixnum::build_with(c))))
.or_else(|err| { .or_else(|err| {
match err { match &err.kind {
ParserError::UnexpectedChar('\'', ..) => {} ParserErrorKind::UnexpectedChar('\'', ..) => {}
err => return Err(err), _ => return Err(err),
} }
self.return_char(c); self.return_char(c);
@@ -954,7 +969,7 @@ impl<'a, R: CharRead> Lexer<'a, R> {
Ok(NumberToken::Partial(token_string)) => match self.parse_integer(&token_string) { Ok(NumberToken::Partial(token_string)) => match self.parse_integer(&token_string) {
Ok(n) => Ok(Token::Literal(n.to_literal())), Ok(n) => Ok(Token::Literal(n.to_literal())),
Err(_) => { Err(_) => {
let n = parse_float_lossy(&token_string)?; let n = self.parse_float_lossy(&token_string)?;
let offset = float_alloc!(n, self.machine_st.arena); let offset = float_alloc!(n, self.machine_st.arena);
Ok(Token::Literal(Literal::F64(offset, OrderedFloat(n)))) Ok(Token::Literal(Literal::F64(offset, OrderedFloat(n))))
} }
@@ -1068,14 +1083,17 @@ impl<'a, R: CharRead> Lexer<'a, R> {
Err(e) => Err(e), Err(e) => Err(e),
} }
} }
}
fn parse_float_lossy(token: &str) -> Result<f64, ParserError> { fn parse_float_lossy(&self, token: &str) -> Result<f64, ParserError> {
const FORMAT: u128 = lexical::format::STANDARD; const FORMAT: u128 = lexical::format::STANDARD;
let options = lexical::ParseFloatOptions::builder() let Ok(options) = lexical::ParseFloatOptions::builder().lossy(true).build() else {
.lossy(true) return Err(self.located_error(ParserErrorKind::ParseFloat));
.build() };
.unwrap();
let n = lexical::parse_with_options::<f64, _, FORMAT>(token.as_bytes(), &options)?; let Ok(n) = lexical::parse_with_options::<f64, _, FORMAT>(token.as_bytes(), &options)
Ok(n) else {
return Err(self.located_error(ParserErrorKind::ParseFloat));
};
Ok(n)
}
} }

View File

@@ -254,10 +254,7 @@ pub fn read_tokens<R: CharRead>(lexer: &mut Lexer<'_, R>) -> Result<Vec<Token>,
} }
} }
Err(e) if e.is_unexpected_eof() && !tokens.is_empty() => { Err(e) if e.is_unexpected_eof() && !tokens.is_empty() => {
return Err(ParserError::IncompleteReduction( return Err(lexer.incomplete_reduction());
lexer.line_num,
lexer.col_num,
));
} }
Err(e) => { Err(e) => {
return Err(e); return Err(e);
@@ -700,12 +697,7 @@ impl<'a, R: CharRead> Parser<'a, R> {
} else { } else {
let term = match self.terms.pop() { let term = match self.terms.pop() {
Some(term) => term, Some(term) => term,
_ => { _ => return Err(self.lexer.incomplete_reduction()),
return Err(ParserError::IncompleteReduction(
self.lexer.line_num,
self.lexer.col_num,
))
}
}; };
if self.stack[idx].priority > 1000 { if self.stack[idx].priority > 1000 {
@@ -718,10 +710,7 @@ impl<'a, R: CharRead> Parser<'a, R> {
}; };
if arity > self.terms.len() { if arity > self.terms.len() {
return Err(ParserError::IncompleteReduction( return Err(self.lexer.incomplete_reduction());
self.lexer.line_num,
self.lexer.col_num,
));
} }
let idx = self.terms.len() - arity; let idx = self.terms.len() - arity;
@@ -789,12 +778,7 @@ impl<'a, R: CharRead> Parser<'a, R> {
let term = match self.terms.pop() { let term = match self.terms.pop() {
Some(term) => term, Some(term) => term,
_ => { _ => return Err(self.lexer.incomplete_reduction()),
return Err(ParserError::IncompleteReduction(
self.lexer.line_num,
self.lexer.col_num,
))
}
}; };
self.terms self.terms
@@ -972,10 +956,7 @@ impl<'a, R: CharRead> Parser<'a, R> {
self.negate_number(n, negate_rat_rc, Literal::Rational) self.negate_number(n, negate_rat_rc, Literal::Rational)
} }
Token::Literal(Literal::F64(_offset, n)) if n.is_infinite() => { Token::Literal(Literal::F64(_offset, n)) if n.is_infinite() => {
return Err(ParserError::InfiniteFloat( return Err(self.lexer.located_error(ParserErrorKind::InfiniteFloat));
self.lexer.line_num,
self.lexer.col_num,
));
} }
Token::Literal(Literal::F64(offset, n)) => { Token::Literal(Literal::F64(offset, n)) => {
self.negate_number((offset, n), negate_f64, |(offset, n)| { self.negate_number((offset, n), negate_f64, |(offset, n)| {
@@ -997,28 +978,19 @@ impl<'a, R: CharRead> Parser<'a, R> {
Token::OpenCT => self.shift(Token::OpenCT, 1300, DELIMITER), Token::OpenCT => self.shift(Token::OpenCT, 1300, DELIMITER),
Token::Close => { Token::Close => {
if !self.reduce_term() && !self.reduce_brackets() { if !self.reduce_term() && !self.reduce_brackets() {
return Err(ParserError::IncompleteReduction( return Err(self.lexer.incomplete_reduction());
self.lexer.line_num,
self.lexer.col_num,
));
} }
} }
Token::OpenList => self.shift(Token::OpenList, 1300, DELIMITER), Token::OpenList => self.shift(Token::OpenList, 1300, DELIMITER),
Token::CloseList => { Token::CloseList => {
if !self.reduce_list()? { if !self.reduce_list()? {
return Err(ParserError::IncompleteReduction( return Err(self.lexer.incomplete_reduction());
self.lexer.line_num,
self.lexer.col_num,
));
} }
} }
Token::OpenCurly => self.shift(Token::OpenCurly, 1300, DELIMITER), Token::OpenCurly => self.shift(Token::OpenCurly, 1300, DELIMITER),
Token::CloseCurly => { Token::CloseCurly => {
if !self.reduce_curly()? { if !self.reduce_curly()? {
return Err(ParserError::IncompleteReduction( return Err(self.lexer.incomplete_reduction());
self.lexer.line_num,
self.lexer.col_num,
));
} }
} }
Token::HeadTailSeparator => { Token::HeadTailSeparator => {
@@ -1051,12 +1023,7 @@ impl<'a, R: CharRead> Parser<'a, R> {
| Some(TokenType::OpenList) | Some(TokenType::OpenList)
| Some(TokenType::OpenCurly) | Some(TokenType::OpenCurly)
| Some(TokenType::HeadTailSeparator) | Some(TokenType::HeadTailSeparator)
| Some(TokenType::Comma) => { | Some(TokenType::Comma) => return Err(self.lexer.incomplete_reduction()),
return Err(ParserError::IncompleteReduction(
self.lexer.line_num,
self.lexer.col_num,
))
}
_ => {} _ => {}
}, },
} }
@@ -1066,12 +1033,12 @@ impl<'a, R: CharRead> Parser<'a, R> {
#[inline] #[inline]
pub fn add_lines_read(&mut self, lines_read: usize) { pub fn add_lines_read(&mut self, lines_read: usize) {
self.lexer.line_num += lines_read; self.lexer.location.line += lines_read;
} }
#[inline] #[inline]
pub fn lines_read(&self) -> usize { pub fn lines_read(&self) -> usize {
self.lexer.line_num self.lexer.location.line
} }
// on success, returns the parsed term and the number of lines read. // on success, returns the parsed term and the number of lines read.
@@ -1092,10 +1059,7 @@ impl<'a, R: CharRead> Parser<'a, R> {
self.reduce_op(1400); self.reduce_op(1400);
if self.terms.len() > 1 || self.stack.len() > 1 { if self.terms.len() > 1 || self.stack.len() > 1 {
return Err(ParserError::IncompleteReduction( return Err(self.lexer.incomplete_reduction());
self.lexer.line_num,
self.lexer.col_num,
));
} }
match self.terms.pop() { match self.terms.pop() {
@@ -1103,16 +1067,10 @@ impl<'a, R: CharRead> Parser<'a, R> {
if self.terms.is_empty() { if self.terms.is_empty() {
Ok(term) Ok(term)
} else { } else {
Err(ParserError::IncompleteReduction( Err(self.lexer.incomplete_reduction())
self.lexer.line_num,
self.lexer.col_num,
))
} }
} }
_ => Err(ParserError::IncompleteReduction( _ => Err(self.lexer.incomplete_reduction()),
self.lexer.line_num,
self.lexer.col_num,
)),
} }
} }
} }

View File

@@ -48,12 +48,11 @@ pub(crate) fn error_after_read_term<R>(
parser: &Parser<R>, parser: &Parser<R>,
) -> CompilationError { ) -> CompilationError {
if err.is_unexpected_eof() { if err.is_unexpected_eof() {
let line_num = parser.lexer.line_num; let location = &parser.lexer.location;
let col_num = parser.lexer.col_num;
// rough overlap with errors 8.14.1.3 k) & l) of the ISO standard here // rough overlap with errors 8.14.1.3 k) & l) of the ISO standard here
if !(line_num == prior_num_lines_read && col_num == 0) { if !(location.line() == prior_num_lines_read && location.column() == 0) {
return CompilationError::from(ParserError::IncompleteReduction(line_num, col_num)); return CompilationError::from(parser.lexer.incomplete_reduction());
} }
} }