From: Skgland Date: Sat, 24 Jan 2026 02:24:02 +0000 (+0100) Subject: make ParserError a struct with an enum kind field X-Git-Url: https://git.sagredo.dev/?a=commitdiff_plain;h=18b476af28e104cea2be75eeb8f3b1982d85a1a4;p=scryer-prolog.git make ParserError a struct with an enum kind field --- diff --git a/src/machine/streams.rs b/src/machine/streams.rs index 0d88ca58..760c6840 100644 --- a/src/machine/streams.rs +++ b/src/machine/streams.rs @@ -1952,7 +1952,7 @@ impl MachineState { ) -> Result { match stream.peek_char() { 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)) => { if c == '\u{feff}' { // skip UTF-8 BOM diff --git a/src/machine/system_calls.rs b/src/machine/system_calls.rs index f6763153..59ba3975 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(parser.lexer.location.clone()); + let err = parser.lexer.parse_big_int_error(); let err = self.syntax_error(err); return Err(self.error_form(err, stub_gen())); @@ -1026,7 +1026,7 @@ impl MachineState { return Ok(()); } Ok(c) => { - let err = ParserError::UnexpectedChar(c, lexer.location); + let err = lexer.unexpected_char(c); let err = self.syntax_error(err); return Err(self.error_form(err, stub_gen())); diff --git a/src/parser/ast.rs b/src/parser/ast.rs index 17caae24..18f2335d 100644 --- a/src/parser/ast.rs +++ b/src/parser/ast.rs @@ -447,71 +447,71 @@ impl Location { #[allow(dead_code)] #[derive(Debug)] -pub enum ParserError { - BackQuotedString(Location), +pub struct ParserError { + pub(crate) location: Option, + pub(crate) kind: ParserErrorKind, +} + +#[allow(dead_code)] +#[derive(Debug)] +#[non_exhaustive] +pub enum ParserErrorKind { + BackQuotedString, IO(IOError), - IncompleteReduction(Location), - InfiniteFloat(Location), - InvalidSingleQuotedCharacter(char, Location), + IncompleteReduction, + InfiniteFloat, + InvalidSingleQuotedCharacter(char), LexicalError(lexical::Error), - MissingQuote(Location), - NonPrologChar(Location), - ParseBigInt(Location), - UnexpectedChar(char, Location), + MissingQuote, + NonPrologChar, + ParseBigInt, + UnexpectedChar(char), // UnexpectedEOF, - Utf8Error(Option), + Utf8Error, } impl ParserError { pub fn location(&self) -> Option { - match self { - ParserError::BackQuotedString(location) - | ParserError::InvalidSingleQuotedCharacter(_, 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(), - ParserError::IO(_) | ParserError::LexicalError(_) => None, - } + self.location.as_ref().cloned() } pub fn as_atom(&self) -> Atom { - match self { - ParserError::BackQuotedString(..) => atom!("back_quoted_string"), - ParserError::IncompleteReduction(..) => atom!("incomplete_reduction"), - ParserError::InvalidSingleQuotedCharacter(..) => { + match &self.kind { + ParserErrorKind::BackQuotedString => atom!("back_quoted_string"), + ParserErrorKind::IncompleteReduction => atom!("incomplete_reduction"), + ParserErrorKind::InvalidSingleQuotedCharacter(..) => { atom!("invalid_single_quoted_character") } - ParserError::InfiniteFloat(..) => { + ParserErrorKind::InfiniteFloat => { atom!("infinite_float") } - ParserError::IO(e) if e.kind() == ErrorKind::UnexpectedEof => { + ParserErrorKind::IO(e) if e.kind() == ErrorKind::UnexpectedEof => { atom!("unexpected_end_of_file") } - ParserError::IO(e) if e.kind() == ErrorKind::InvalidData => { + ParserErrorKind::IO(e) if e.kind() == ErrorKind::InvalidData => { atom!("invalid_data") } - ParserError::IO(_) => atom!("input_output_error"), - ParserError::LexicalError(_) => atom!("lexical_error"), - ParserError::MissingQuote(..) => atom!("missing_quote"), - ParserError::NonPrologChar(..) => atom!("non_prolog_character"), - ParserError::ParseBigInt(..) => atom!("cannot_parse_big_int"), - ParserError::UnexpectedChar(..) => atom!("unexpected_char"), - ParserError::Utf8Error(..) => atom!("utf8_conversion_error"), + ParserErrorKind::IO(_) => atom!("input_output_error"), + ParserErrorKind::LexicalError(_) => atom!("lexical_error"), + ParserErrorKind::MissingQuote => atom!("missing_quote"), + ParserErrorKind::NonPrologChar => atom!("non_prolog_character"), + ParserErrorKind::ParseBigInt => atom!("cannot_parse_big_int"), + ParserErrorKind::UnexpectedChar(..) => atom!("unexpected_char"), + ParserErrorKind::Utf8Error => atom!("utf8_conversion_error"), } } #[inline] pub 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] pub fn is_unexpected_eof(&self) -> bool { - if let ParserError::IO(e) = self { + if let ParserErrorKind::IO(e) = &self.kind { e.kind() == ErrorKind::UnexpectedEof } else { false @@ -521,25 +521,18 @@ impl ParserError { impl From for ParserError { fn from(e: lexical::Error) -> ParserError { - ParserError::LexicalError(e) + ParserError { + location: None, + kind: ParserErrorKind::LexicalError(e), + } } } impl From for ParserError { fn from(e: IOError) -> ParserError { - ParserError::IO(e) - } -} - -impl From<&IOError> for ParserError { - fn from(error: &IOError) -> ParserError { - if let Some(_utf8_error) = error - .get_ref() - .and_then(|e| e.downcast_ref::()) - { - ParserError::Utf8Error(None) - } else { - ParserError::IO(error.kind().into()) + ParserError { + location: None, + kind: ParserErrorKind::IO(e), } } } diff --git a/src/parser/lexer.rs b/src/parser/lexer.rs index 7acf76b6..0d3f8941 100644 --- a/src/parser/lexer.rs +++ b/src/parser/lexer.rs @@ -16,7 +16,10 @@ macro_rules! consume_chars_with { match $e { Ok(Some(c)) => $token.push(c), Ok(None) => continue, - Err($crate::parser::ast::ParserError::UnexpectedChar(..)) => break, + Err($crate::parser::ast::ParserError { + kind: $crate::parser::ast::ParserErrorKind::UnexpectedChar(..), + .. + }) => break, Err(e) => return Err(e), } } @@ -103,6 +106,27 @@ impl<'a, R: fmt::Debug> fmt::Debug for Lexer<'a, R> { } } +impl 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> { pub fn new(src: R, machine_st: &'a mut MachineState) -> Self { Self { @@ -197,7 +221,7 @@ impl<'a, R: CharRead> Lexer<'a, R> { match comment_loop() { Err(e) if e.is_unexpected_eof() => { - return Err(ParserError::IncompleteReduction(self.location.clone())); + return Err(self.incomplete_reduction()); } Err(e) => { return Err(e); @@ -209,7 +233,7 @@ impl<'a, R: CharRead> Lexer<'a, R> { self.skip_char(c); Ok(true) } else { - Err(ParserError::NonPrologChar(self.location.clone())) + Err(self.located_error(ParserErrorKind::NonPrologChar)) } } else { self.return_char('/'); @@ -226,7 +250,7 @@ impl<'a, R: CharRead> Lexer<'a, R> { if !back_quote_char!(c2) { self.return_char(c); - Err(ParserError::UnexpectedChar(c, self.location.clone())) + Err(self.unexpected_char(c)) } else { self.skip_char(c2); Ok(c2) @@ -251,7 +275,7 @@ impl<'a, R: CharRead> Lexer<'a, R> { Ok(None) } else { self.return_char(c); - Err(ParserError::UnexpectedChar(c, self.location.clone())) + Err(self.unexpected_char(c)) } } else { self.get_back_quoted_char().map(Some) @@ -273,10 +297,13 @@ impl<'a, R: CharRead> Lexer<'a, R> { self.skip_char(c); Ok(token) } else { - Err(ParserError::MissingQuote(self.location.clone())) + Err({ + let this = &self; + this.located_error(ParserErrorKind::MissingQuote) + }) } } else { - Err(ParserError::UnexpectedChar(c, self.location.clone())) + Err(self.unexpected_char(c)) } } @@ -307,7 +334,7 @@ impl<'a, R: CharRead> Lexer<'a, R> { if !single_quote_char!(c2) { self.return_char(c); - Err(ParserError::UnexpectedChar(c, self.location.clone())) + Err(self.unexpected_char(c)) } else { self.skip_char(c2); Ok(c2) @@ -348,7 +375,7 @@ impl<'a, R: CharRead> Lexer<'a, R> { if !double_quote_char!(c2) { self.return_char(c); - Err(ParserError::UnexpectedChar(c, self.location.clone())) + Err(self.unexpected_char(c)) } else { self.skip_char(c2); Ok(c2) @@ -372,7 +399,7 @@ impl<'a, R: CharRead> Lexer<'a, R> { 't' => '\t', 'n' => '\n', 'r' => '\r', - c => return Err(ParserError::UnexpectedChar(c, self.location.clone())), + c => return Err(self.unexpected_char(c)), }; self.skip_char(c); @@ -390,7 +417,7 @@ 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.location.clone())) + Err(self.incomplete_reduction()) } } @@ -416,14 +443,11 @@ 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.location.clone())), - |n| { - char::try_from(n) - .map_err(|_| ParserError::Utf8Error(Some(self.location.clone()))) - }, + |_| Err(self.parse_big_int_error()), + |n| char::try_from(n).map_err(|_| self.located_error(ParserErrorKind::Utf8Error)), ) } else { - Err(ParserError::IncompleteReduction(self.location.clone())) + Err(self.incomplete_reduction()) } } @@ -435,7 +459,7 @@ impl<'a, R: CharRead> Lexer<'a, R> { Ok(c) } else { if !backslash_char!(c) { - return Err(ParserError::UnexpectedChar(c, self.location.clone())); + return Err(self.unexpected_char(c)); } self.skip_char(c); @@ -466,7 +490,10 @@ impl<'a, R: CharRead> Lexer<'a, R> { self.skip_char(c); Ok(token) } else { - Err(ParserError::MissingQuote(self.location.clone())) + Err({ + let this = &self; + this.located_error(ParserErrorKind::MissingQuote) + }) } } @@ -497,7 +524,7 @@ impl<'a, R: CharRead> Lexer<'a, R> { .map(NumberToken::Integer) } else { self.return_char(start); - Err(ParserError::ParseBigInt(self.location.clone())) + Err(self.parse_big_int_error()) } } @@ -528,7 +555,7 @@ impl<'a, R: CharRead> Lexer<'a, R> { .map(NumberToken::Integer) } else { self.return_char(start); - Err(ParserError::ParseBigInt(self.location.clone())) + Err(self.parse_big_int_error()) } } @@ -559,7 +586,7 @@ impl<'a, R: CharRead> Lexer<'a, R> { .map(NumberToken::Integer) } else { self.return_char(start); - Err(ParserError::ParseBigInt(self.location.clone())) + Err(self.parse_big_int_error()) } } @@ -632,14 +659,11 @@ impl<'a, R: CharRead> Lexer<'a, R> { } } } else { - return Err(ParserError::InvalidSingleQuotedCharacter( - c, - self.location.clone(), - )); + return Err(self.located_error(ParserErrorKind::InvalidSingleQuotedCharacter(c))); } } else { match self.get_back_quoted_string() { - Ok(_) => return Err(ParserError::BackQuotedString(self.location.clone())), + Ok(_) => return Err(self.located_error(ParserErrorKind::BackQuotedString)), Err(e) => return Err(e), } } @@ -677,7 +701,7 @@ impl<'a, R: CharRead> Lexer<'a, R> { if decimal_digit_char!(c) { Ok(c) } else { - Err(ParserError::ParseBigInt(self.location.clone())) + Err(self.parse_big_int_error()) } } else { Ok(c) @@ -699,7 +723,7 @@ 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.location.clone())) + .map_err(|_| self.parse_big_int_error()) }) } @@ -817,7 +841,7 @@ impl<'a, R: CharRead> Lexer<'a, R> { } else if token.starts_with('0') && token.len() == 1 { if c == 'x' { 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) } else { Err(e) @@ -825,7 +849,7 @@ impl<'a, R: CharRead> Lexer<'a, R> { }) } else if c == 'o' { 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) } else { Err(e) @@ -833,7 +857,7 @@ impl<'a, R: CharRead> Lexer<'a, R> { }) } else if c == 'b' { 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) } else { Err(e) @@ -862,9 +886,9 @@ impl<'a, R: CharRead> Lexer<'a, R> { self.get_single_quoted_char() .map(|c| NumberToken::Integer(GInteger::Fixnum(Fixnum::build_with(c)))) .or_else(|err| { - match err { - ParserError::UnexpectedChar('\'', ..) => {} - err => return Err(err), + match &err.kind { + ParserErrorKind::UnexpectedChar('\'', ..) => {} + _ => return Err(err), } self.return_char(c); diff --git a/src/parser/parser.rs b/src/parser/parser.rs index eda980c9..a7291908 100644 --- a/src/parser/parser.rs +++ b/src/parser/parser.rs @@ -254,7 +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.location.clone())); + return Err(lexer.incomplete_reduction()); } Err(e) => { return Err(e); @@ -697,11 +697,7 @@ impl<'a, R: CharRead> Parser<'a, R> { } else { let term = match self.terms.pop() { Some(term) => term, - _ => { - return Err(ParserError::IncompleteReduction( - self.lexer.location.clone(), - )) - } + _ => return Err(self.lexer.incomplete_reduction()), }; if self.stack[idx].priority > 1000 { @@ -714,9 +710,7 @@ impl<'a, R: CharRead> Parser<'a, R> { }; if arity > self.terms.len() { - return Err(ParserError::IncompleteReduction( - self.lexer.location.clone(), - )); + return Err(self.lexer.incomplete_reduction()); } let idx = self.terms.len() - arity; @@ -784,11 +778,7 @@ impl<'a, R: CharRead> Parser<'a, R> { let term = match self.terms.pop() { Some(term) => term, - _ => { - return Err(ParserError::IncompleteReduction( - self.lexer.location.clone(), - )) - } + _ => return Err(self.lexer.incomplete_reduction()), }; self.terms @@ -966,7 +956,7 @@ 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.location.clone())); + return Err(self.lexer.located_error(ParserErrorKind::InfiniteFloat)); } Token::Literal(Literal::F64(offset, n)) => { self.negate_number((offset, n), negate_f64, |(offset, n)| { @@ -988,25 +978,19 @@ impl<'a, R: CharRead> Parser<'a, R> { Token::OpenCT => self.shift(Token::OpenCT, 1300, DELIMITER), Token::Close => { if !self.reduce_term() && !self.reduce_brackets() { - return Err(ParserError::IncompleteReduction( - self.lexer.location.clone(), - )); + return Err(self.lexer.incomplete_reduction()); } } Token::OpenList => self.shift(Token::OpenList, 1300, DELIMITER), Token::CloseList => { if !self.reduce_list()? { - return Err(ParserError::IncompleteReduction( - self.lexer.location.clone(), - )); + return Err(self.lexer.incomplete_reduction()); } } Token::OpenCurly => self.shift(Token::OpenCurly, 1300, DELIMITER), Token::CloseCurly => { if !self.reduce_curly()? { - return Err(ParserError::IncompleteReduction( - self.lexer.location.clone(), - )); + return Err(self.lexer.incomplete_reduction()); } } Token::HeadTailSeparator => { @@ -1039,11 +1023,7 @@ impl<'a, R: CharRead> Parser<'a, R> { | Some(TokenType::OpenList) | Some(TokenType::OpenCurly) | Some(TokenType::HeadTailSeparator) - | Some(TokenType::Comma) => { - return Err(ParserError::IncompleteReduction( - self.lexer.location.clone(), - )) - } + | Some(TokenType::Comma) => return Err(self.lexer.incomplete_reduction()), _ => {} }, } @@ -1079,9 +1059,7 @@ impl<'a, R: CharRead> Parser<'a, R> { self.reduce_op(1400); if self.terms.len() > 1 || self.stack.len() > 1 { - return Err(ParserError::IncompleteReduction( - self.lexer.location.clone(), - )); + return Err(self.lexer.incomplete_reduction()); } match self.terms.pop() { @@ -1089,14 +1067,10 @@ impl<'a, R: CharRead> Parser<'a, R> { if self.terms.is_empty() { Ok(term) } else { - Err(ParserError::IncompleteReduction( - self.lexer.location.clone(), - )) + Err(self.lexer.incomplete_reduction()) } } - _ => Err(ParserError::IncompleteReduction( - self.lexer.location.clone(), - )), + _ => Err(self.lexer.incomplete_reduction()), } } } diff --git a/src/read.rs b/src/read.rs index 9f578bf8..a07f4d09 100644 --- a/src/read.rs +++ b/src/read.rs @@ -52,9 +52,7 @@ pub(crate) fn error_after_read_term( // rough overlap with errors 8.14.1.3 k) & l) of the ISO standard here if !(location.line() == prior_num_lines_read && location.column() == 0) { - return CompilationError::from(ParserError::IncompleteReduction( - parser.lexer.location.clone(), - )); + return CompilationError::from(parser.lexer.incomplete_reduction()); } }