From 3343188756c2dd476d5ea01ecaf96a4d8699e897 Mon Sep 17 00:00:00 2001 From: Skgland Date: Sat, 24 Jan 2026 01:54:29 +0100 Subject: [PATCH] replace pairs of usize with location struct --- src/machine/machine_errors.rs | 11 +++--- src/machine/system_calls.rs | 8 ++--- src/parser/ast.rs | 60 +++++++++++++++++++++---------- src/parser/lexer.rs | 66 ++++++++++++++--------------------- src/parser/parser.rs | 44 ++++++++--------------- src/read.rs | 9 ++--- 6 files changed, 96 insertions(+), 102 deletions(-) diff --git a/src/machine/machine_errors.rs b/src/machine/machine_errors.rs index 4ed19893..2dccd1c8 100644 --- a/src/machine/machine_errors.rs +++ b/src/machine/machine_errors.rs @@ -20,7 +20,7 @@ pub type MachineStubGen = Box MachineStub>; #[derive(Debug)] pub(crate) struct MachineError { stub: MachineStub, - location: Option<(usize, usize)>, // line_num, col_num + location: Option, } // 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 { - if let Some((line_num, _col_num)) = err.location { + if let Some(location) = err.location { + let line = location.line(); functor!( atom!("error"), [ functor((err.stub)), functor( (atom!(":")), - [functor(src), number(line_num, (&mut self.arena))] + [functor(src), number(line, (&mut self.arena))] ) ] ) @@ -853,9 +854,9 @@ impl From for CompilationError { } impl CompilationError { - pub(crate) fn line_and_col_num(&self) -> Option<(usize, usize)> { + pub(crate) fn line_and_col_num(&self) -> Option { match self { - CompilationError::ParserError(err) => err.line_and_col_num(), + CompilationError::ParserError(err) => err.location(), _ => None, } } diff --git a/src/machine/system_calls.rs b/src/machine/system_calls.rs index edd08cb5..f6763153 100644 --- a/src/machine/system_calls.rs +++ b/src/machine/system_calls.rs @@ -1016,7 +1016,7 @@ impl MachineState { self.unify_fixnum(n, nx); } _ => { - let err = ParserError::ParseBigInt(0, 0); + let err = ParserError::ParseBigInt(parser.lexer.location.clone()); let err = self.syntax_error(err); return Err(self.error_form(err, stub_gen())); @@ -1026,9 +1026,7 @@ impl MachineState { return Ok(()); } Ok(c) => { - let (line_num, col_num) = (lexer.line_num, lexer.col_num); - - let err = ParserError::UnexpectedChar(c, line_num, col_num); + let err = ParserError::UnexpectedChar(c, lexer.location); let err = self.syntax_error(err); return Err(self.error_form(err, stub_gen())); @@ -9196,7 +9194,7 @@ impl Machine { stream.add_lines_read(parser.lines_read()); } Ok(true) => { - stream.add_lines_read(parser.lexer.line_num); + stream.add_lines_read(parser.lexer.location.line()); self.machine_st.fail = true; } Err(err) => { diff --git a/src/parser/ast.rs b/src/parser/ast.rs index ba18d36e..0d789c33 100644 --- a/src/parser/ast.rs +++ b/src/parser/ast.rs @@ -426,34 +426,53 @@ pub enum ArithmeticError { 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)] #[derive(Debug)] pub enum ParserError { - BackQuotedString(usize, usize), + BackQuotedString(Location), IO(IOError), - IncompleteReduction(usize, usize), - InfiniteFloat(usize, usize), + IncompleteReduction(Location), + InfiniteFloat(Location), InvalidSingleQuotedCharacter(char), LexicalError(lexical::Error), - MissingQuote(usize, usize), - NonPrologChar(usize, usize), - ParseBigInt(usize, usize), - UnexpectedChar(char, usize, usize), + MissingQuote(Location), + NonPrologChar(Location), + ParseBigInt(Location), + UnexpectedChar(char, Location), // UnexpectedEOF, - Utf8Error(usize, usize), + Utf8Error(Option), } impl ParserError { - pub fn line_and_col_num(&self) -> Option<(usize, usize)> { + pub fn location(&self) -> Option { match self { - &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)), + ParserError::BackQuotedString(location) + | ParserError::IncompleteReduction(location) + | ParserError::InfiniteFloat(location) + | ParserError::MissingQuote(location) + | ParserError::NonPrologChar(location) + | ParserError::ParseBigInt(location) + | ParserError::UnexpectedChar(_, location) => Some(location.clone()), + ParserError::Utf8Error(location) => location.as_ref().cloned(), _ => None, } } @@ -513,8 +532,11 @@ impl From for ParserError { impl From<&IOError> for ParserError { fn from(error: &IOError) -> ParserError { - if error.get_ref().filter(|e| e.is::()).is_some() { - ParserError::Utf8Error(0, 0) + if let Some(_utf8_error) = error + .get_ref() + .and_then(|e| e.downcast_ref::()) + { + ParserError::Utf8Error(None) } else { ParserError::IO(error.kind().into()) } diff --git a/src/parser/lexer.rs b/src/parser/lexer.rs index bfc3fb8f..ed39e210 100644 --- a/src/parser/lexer.rs +++ b/src/parser/lexer.rs @@ -91,16 +91,14 @@ macro_rules! try_nt { pub(crate) struct Lexer<'a, R> { pub(crate) reader: R, pub(crate) machine_st: &'a mut MachineState, - pub(crate) line_num: usize, - pub(crate) col_num: usize, + pub(crate) location: Location, } impl<'a, R: fmt::Debug> fmt::Debug for Lexer<'a, R> { fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { f.debug_struct("LexerParser") .field("reader", &"&'a mut R") // Hacky solution. - .field("line_num", &self.line_num) - .field("col_num", &self.col_num) + .field("location", &self.location) .finish() } } @@ -110,8 +108,7 @@ impl<'a, R: CharRead> Lexer<'a, R> { Self { reader: src, machine_st, - line_num: 0, - col_num: 0, + location: Location::BOF, } } @@ -138,10 +135,10 @@ impl<'a, R: CharRead> Lexer<'a, R> { self.reader.consume(c.len_utf8()); if new_line_char!(c) { - self.line_num += 1; - self.col_num = 0; + self.location.line += 1; + self.location.column = 0; } else { - self.col_num += 1; + self.location.column += 1; } } @@ -200,10 +197,7 @@ impl<'a, R: CharRead> Lexer<'a, R> { match comment_loop() { Err(e) if e.is_unexpected_eof() => { - return Err(ParserError::IncompleteReduction( - self.line_num, - self.col_num, - )); + return Err(ParserError::IncompleteReduction(self.location.clone())); } Err(e) => { return Err(e); @@ -215,7 +209,7 @@ impl<'a, R: CharRead> Lexer<'a, R> { self.skip_char(c); Ok(true) } else { - Err(ParserError::NonPrologChar(self.line_num, self.col_num)) + Err(ParserError::NonPrologChar(self.location.clone())) } } else { self.return_char('/'); @@ -232,7 +226,7 @@ impl<'a, R: CharRead> Lexer<'a, R> { if !back_quote_char!(c2) { self.return_char(c); - Err(ParserError::UnexpectedChar(c, self.line_num, self.col_num)) + Err(ParserError::UnexpectedChar(c, self.location.clone())) } else { self.skip_char(c2); Ok(c2) @@ -257,7 +251,7 @@ impl<'a, R: CharRead> Lexer<'a, R> { Ok(None) } else { self.return_char(c); - Err(ParserError::UnexpectedChar(c, self.line_num, self.col_num)) + Err(ParserError::UnexpectedChar(c, self.location.clone())) } } else { self.get_back_quoted_char().map(Some) @@ -279,10 +273,10 @@ impl<'a, R: CharRead> Lexer<'a, R> { self.skip_char(c); Ok(token) } else { - Err(ParserError::MissingQuote(self.line_num, self.col_num)) + Err(ParserError::MissingQuote(self.location.clone())) } } else { - Err(ParserError::UnexpectedChar(c, self.line_num, self.col_num)) + Err(ParserError::UnexpectedChar(c, self.location.clone())) } } @@ -313,7 +307,7 @@ impl<'a, R: CharRead> Lexer<'a, R> { if !single_quote_char!(c2) { self.return_char(c); - Err(ParserError::UnexpectedChar(c, self.line_num, self.col_num)) + Err(ParserError::UnexpectedChar(c, self.location.clone())) } else { self.skip_char(c2); Ok(c2) @@ -354,7 +348,7 @@ impl<'a, R: CharRead> Lexer<'a, R> { if !double_quote_char!(c2) { self.return_char(c); - Err(ParserError::UnexpectedChar(c, self.line_num, self.col_num)) + Err(ParserError::UnexpectedChar(c, self.location.clone())) } else { self.skip_char(c2); Ok(c2) @@ -378,7 +372,7 @@ impl<'a, R: CharRead> Lexer<'a, R> { 't' => '\t', 'n' => '\n', 'r' => '\r', - c => return Err(ParserError::UnexpectedChar(c, self.line_num, self.col_num)), + c => return Err(ParserError::UnexpectedChar(c, self.location.clone())), }; self.skip_char(c); @@ -396,10 +390,7 @@ impl<'a, R: CharRead> Lexer<'a, R> { if hexadecimal_digit_char!(c) { self.escape_sequence_to_char(|c| hexadecimal_digit_char!(c), 16) } else { - Err(ParserError::IncompleteReduction( - self.line_num, - self.col_num, - )) + Err(ParserError::IncompleteReduction(self.location.clone())) } } @@ -425,17 +416,14 @@ impl<'a, R: CharRead> Lexer<'a, R> { if backslash_char!(c) { self.skip_char(c); u32::from_str_radix(&token, radix).map_or_else( - |_| Err(ParserError::ParseBigInt(self.line_num, self.col_num)), + |_| Err(ParserError::ParseBigInt(self.location.clone())), |n| { char::try_from(n) - .map_err(|_| ParserError::Utf8Error(self.line_num, self.col_num)) + .map_err(|_| ParserError::Utf8Error(Some(self.location.clone()))) }, ) } else { - Err(ParserError::IncompleteReduction( - self.line_num, - self.col_num, - )) + Err(ParserError::IncompleteReduction(self.location.clone())) } } @@ -447,7 +435,7 @@ impl<'a, R: CharRead> Lexer<'a, R> { Ok(c) } else { if !backslash_char!(c) { - return Err(ParserError::UnexpectedChar(c, self.line_num, self.col_num)); + return Err(ParserError::UnexpectedChar(c, self.location.clone())); } self.skip_char(c); @@ -478,7 +466,7 @@ impl<'a, R: CharRead> Lexer<'a, R> { self.skip_char(c); Ok(token) } else { - Err(ParserError::MissingQuote(self.line_num, self.col_num)) + Err(ParserError::MissingQuote(self.location.clone())) } } @@ -509,7 +497,7 @@ impl<'a, R: CharRead> Lexer<'a, R> { .map(NumberToken::Integer) } else { self.return_char(start); - Err(ParserError::ParseBigInt(self.line_num, self.col_num)) + Err(ParserError::ParseBigInt(self.location.clone())) } } @@ -540,7 +528,7 @@ impl<'a, R: CharRead> Lexer<'a, R> { .map(NumberToken::Integer) } else { self.return_char(start); - Err(ParserError::ParseBigInt(self.line_num, self.col_num)) + Err(ParserError::ParseBigInt(self.location.clone())) } } @@ -571,7 +559,7 @@ impl<'a, R: CharRead> Lexer<'a, R> { .map(NumberToken::Integer) } else { self.return_char(start); - Err(ParserError::ParseBigInt(self.line_num, self.col_num)) + Err(ParserError::ParseBigInt(self.location.clone())) } } @@ -648,7 +636,7 @@ impl<'a, R: CharRead> Lexer<'a, R> { } } else { match self.get_back_quoted_string() { - Ok(_) => return Err(ParserError::BackQuotedString(self.line_num, self.col_num)), + Ok(_) => return Err(ParserError::BackQuotedString(self.location.clone())), Err(e) => return Err(e), } } @@ -686,7 +674,7 @@ impl<'a, R: CharRead> Lexer<'a, R> { if decimal_digit_char!(c) { Ok(c) } else { - Err(ParserError::ParseBigInt(self.line_num, self.col_num)) + Err(ParserError::ParseBigInt(self.location.clone())) } } else { Ok(c) @@ -708,7 +696,7 @@ impl<'a, R: CharRead> Lexer<'a, R> { .or_else(|_| { Integer::from_str_radix(token, radix) .map(|n| GInteger::Integer(arena_alloc!(n, &mut self.machine_st.arena))) - .map_err(|_| ParserError::ParseBigInt(self.line_num, self.col_num)) + .map_err(|_| ParserError::ParseBigInt(self.location.clone())) }) } diff --git a/src/parser/parser.rs b/src/parser/parser.rs index 61e85c7e..eda980c9 100644 --- a/src/parser/parser.rs +++ b/src/parser/parser.rs @@ -254,10 +254,7 @@ pub fn read_tokens(lexer: &mut Lexer<'_, R>) -> Result, } } Err(e) if e.is_unexpected_eof() && !tokens.is_empty() => { - return Err(ParserError::IncompleteReduction( - lexer.line_num, - lexer.col_num, - )); + return Err(ParserError::IncompleteReduction(lexer.location.clone())); } Err(e) => { return Err(e); @@ -702,8 +699,7 @@ impl<'a, R: CharRead> Parser<'a, R> { Some(term) => term, _ => { return Err(ParserError::IncompleteReduction( - self.lexer.line_num, - self.lexer.col_num, + self.lexer.location.clone(), )) } }; @@ -719,8 +715,7 @@ impl<'a, R: CharRead> Parser<'a, R> { if arity > self.terms.len() { return Err(ParserError::IncompleteReduction( - self.lexer.line_num, - self.lexer.col_num, + self.lexer.location.clone(), )); } @@ -791,8 +786,7 @@ impl<'a, R: CharRead> Parser<'a, R> { Some(term) => term, _ => { return Err(ParserError::IncompleteReduction( - self.lexer.line_num, - self.lexer.col_num, + self.lexer.location.clone(), )) } }; @@ -972,10 +966,7 @@ impl<'a, R: CharRead> Parser<'a, R> { self.negate_number(n, negate_rat_rc, Literal::Rational) } Token::Literal(Literal::F64(_offset, n)) if n.is_infinite() => { - return Err(ParserError::InfiniteFloat( - self.lexer.line_num, - self.lexer.col_num, - )); + return Err(ParserError::InfiniteFloat(self.lexer.location.clone())); } Token::Literal(Literal::F64(offset, n)) => { self.negate_number((offset, n), negate_f64, |(offset, n)| { @@ -998,8 +989,7 @@ impl<'a, R: CharRead> Parser<'a, R> { Token::Close => { if !self.reduce_term() && !self.reduce_brackets() { return Err(ParserError::IncompleteReduction( - self.lexer.line_num, - self.lexer.col_num, + self.lexer.location.clone(), )); } } @@ -1007,8 +997,7 @@ impl<'a, R: CharRead> Parser<'a, R> { Token::CloseList => { if !self.reduce_list()? { return Err(ParserError::IncompleteReduction( - self.lexer.line_num, - self.lexer.col_num, + self.lexer.location.clone(), )); } } @@ -1016,8 +1005,7 @@ impl<'a, R: CharRead> Parser<'a, R> { Token::CloseCurly => { if !self.reduce_curly()? { return Err(ParserError::IncompleteReduction( - self.lexer.line_num, - self.lexer.col_num, + self.lexer.location.clone(), )); } } @@ -1053,8 +1041,7 @@ impl<'a, R: CharRead> Parser<'a, R> { | Some(TokenType::HeadTailSeparator) | Some(TokenType::Comma) => { return Err(ParserError::IncompleteReduction( - self.lexer.line_num, - self.lexer.col_num, + self.lexer.location.clone(), )) } _ => {} @@ -1066,12 +1053,12 @@ impl<'a, R: CharRead> Parser<'a, R> { #[inline] pub fn add_lines_read(&mut self, lines_read: usize) { - self.lexer.line_num += lines_read; + self.lexer.location.line += lines_read; } #[inline] 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. @@ -1093,8 +1080,7 @@ impl<'a, R: CharRead> Parser<'a, R> { if self.terms.len() > 1 || self.stack.len() > 1 { return Err(ParserError::IncompleteReduction( - self.lexer.line_num, - self.lexer.col_num, + self.lexer.location.clone(), )); } @@ -1104,14 +1090,12 @@ impl<'a, R: CharRead> Parser<'a, R> { Ok(term) } else { Err(ParserError::IncompleteReduction( - self.lexer.line_num, - self.lexer.col_num, + self.lexer.location.clone(), )) } } _ => Err(ParserError::IncompleteReduction( - self.lexer.line_num, - self.lexer.col_num, + self.lexer.location.clone(), )), } } diff --git a/src/read.rs b/src/read.rs index fe709c7e..9f578bf8 100644 --- a/src/read.rs +++ b/src/read.rs @@ -48,12 +48,13 @@ pub(crate) fn error_after_read_term( parser: &Parser, ) -> CompilationError { if err.is_unexpected_eof() { - let line_num = parser.lexer.line_num; - let col_num = parser.lexer.col_num; + let location = &parser.lexer.location; // 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) { - return CompilationError::from(ParserError::IncompleteReduction(line_num, col_num)); + if !(location.line() == prior_num_lines_read && location.column() == 0) { + return CompilationError::from(ParserError::IncompleteReduction( + parser.lexer.location.clone(), + )); } } -- 2.54.0