Work in progress

src/parse/formulas.rs

@@ -71,23 +71,46 @@ impl<'i> FormulaStr<'i>
 		}
 	}
 
-	fn iter_tokens(&self) -> TokenIterator<'i>
+	fn tokens(&self) -> Tokens<'i, impl FnMut(Token<'i>) -> Option<Token<'i>>>
 	{
-		TokenIterator::new(self.input)
+		Tokens::new_iter(self.input)
 	}
 
-	fn filter_logical_connective(&self, logical_connective: LogicalConnective)
-		-> std::iter::Filter<TokenIterator<'i>, impl FnMut(&Result<(usize, usize, Token<'i>), crate::parse::Error>) -> bool>
+	fn logical_connectives(&self) -> Tokens<'i, impl FnMut(Token<'i>) -> Option<LogicalConnective>>
 	{
-		let token_selector = move |token: &_| match token
+		let functor = |token| match token
 		{
-			Ok((_, _, Token::Identifier(ref identifier))) => match *identifier
+			Token::Identifier(ref identifier) => match *identifier
+			{
+				"and" => Some(LogicalConnective::And),
+				"or" => Some(LogicalConnective::Or),
+				_ => None,
+			},
+			Token::Symbol(ref symbol) => match symbol
+			{
+				Symbol::ArrowLeft => Some(LogicalConnective::ImpliesRightToLeft),
+				Symbol::ArrowLeftAndRight => Some(LogicalConnective::IfAndOnlyIf),
+				Symbol::ArrowRight => Some(LogicalConnective::ImpliesLeftToRight),
+				_ => None,
+			},
+			_ => None,
+		};
+
+		Tokens::new_filter_map(self.input, functor)
+	}
+
+	fn split_at_logical_connective(&self, logical_connective: LogicalConnective)
+		-> TokenSplit<Tokens<'i, impl FnMut(Token<'i>) -> Option<Token<'i>>>>
+	{
+		let predicate = move |token: &_| match token
+		{
+			Token::Identifier(ref identifier) => match *identifier
 			{
 				"and" => logical_connective == LogicalConnective::And,
 				"or" => logical_connective == LogicalConnective::Or,
 				_ => false,
 			},
-			Ok((_, _, Token::Symbol(ref symbol))) => match symbol
+			Token::Symbol(ref symbol) => match symbol
 			{
 				Symbol::ArrowLeft => logical_connective == LogicalConnective::ImpliesRightToLeft,
 				Symbol::ArrowLeftAndRight => logical_connective == LogicalConnective::IfAndOnlyIf,
@@ -97,7 +120,7 @@ impl<'i> FormulaStr<'i>
 			_ => false,
 		};
 
-		self.iter_tokens().filter(token_selector)
+		Tokens::new_filter(self.input, predicate).split()
 	}
 
 	pub fn top_level_logical_connective(&self)
@@ -121,16 +144,13 @@ impl<'i> FormulaStr<'i>
 			_ => None,
 		};
 
+		let logical_connectives = Tokens::new_filter_map(self.input, logical_connective);
+
 		let mut top_level_logical_connective = None;
 
-		for token in self.iter_tokens()
+		for logical_connective in logical_connectives
 		{
-			let (_, _, token) = token?;
-			let logical_connective = match logical_connective(token)
-			{
-				Some(logical_connective) => logical_connective,
-				None => continue,
-			};
+			let (_, logical_connective) = logical_connective?;
 
 			top_level_logical_connective = match top_level_logical_connective
 			{
@@ -162,39 +182,25 @@ impl<'i> FormulaStr<'i>
 		Ok(top_level_logical_connective)
 	}
 
-	fn filter_comparison_operators(&self)
-		-> std::iter::FilterMap<TokenIterator<'i>, impl FnMut(Result<(usize, usize, Token<'i>), crate::parse::Error>)
-			-> Option<Result<(usize, usize, crate::ComparisonOperator), crate::parse::Error>>>
+	fn comparison_operators(&self) -> Tokens<'i, impl FnMut(Token<'i>)
+		-> Option<crate::ComparisonOperator>>
 	{
-		let token_functor = |token| match token
+		let functor = |token| match token
 		{
-			Ok((input_left, remaining_input, Token::Symbol(symbol))) => match symbol
+			Token::Symbol(symbol) => match symbol
 			{
-				Symbol::Greater =>
-					return Some(Ok((input_left, remaining_input,
-						crate::ComparisonOperator::Greater))),
-				Symbol::GreaterOrEqual =>
-					return Some(Ok((input_left, remaining_input,
-						crate::ComparisonOperator::GreaterOrEqual))),
-				Symbol::Less =>
-					return Some(Ok((input_left, remaining_input,
-						crate::ComparisonOperator::Less))),
-				Symbol::LessOrEqual =>
-					return Some(Ok((input_left, remaining_input,
-						crate::ComparisonOperator::LessOrEqual))),
-				Symbol::Equal =>
-					return Some(Ok((input_left, remaining_input,
-						crate::ComparisonOperator::Equal))),
-				Symbol::NotEqual =>
-					return Some(Ok((input_left, remaining_input,
-						crate::ComparisonOperator::NotEqual))),
+				Symbol::Greater => Some(crate::ComparisonOperator::Greater),
+				Symbol::GreaterOrEqual => Some(crate::ComparisonOperator::GreaterOrEqual),
+				Symbol::Less => Some(crate::ComparisonOperator::Less),
+				Symbol::LessOrEqual => Some(crate::ComparisonOperator::LessOrEqual),
+				Symbol::Equal => Some(crate::ComparisonOperator::Equal),
+				Symbol::NotEqual => Some(crate::ComparisonOperator::NotEqual),
 				_ => None,
 			},
-			Err(error) => Some(Err(error)),
 			_ => None,
 		};
 
-		self.iter_tokens().filter_map(token_functor)
+		Tokens::new_filter_map(self.input, functor)
 	}
 
 	pub fn parse(&self, level: usize) -> Result<crate::Formula, crate::parse::Error>
@@ -222,8 +228,7 @@ impl<'i> FormulaStr<'i>
 			let arguments_n_ary = ||
 			{
 				// TODO: improve error handling if the formulas between the operators are invalid
-				TokenSplit::new(self.filter_logical_connective(top_level_logical_connective),
-					self.input)
+				self.split_at_logical_connective(top_level_logical_connective)
 					.map(|subformula| FormulaStr::new(subformula?).parse(level + 1))
 					.collect::<Result<Vec<_>, _>>()
 			};
@@ -236,10 +241,7 @@ impl<'i> FormulaStr<'i>
 					return Ok(crate::Formula::if_and_only_if(arguments_n_ary()?)),
 				LogicalConnective::ImpliesLeftToRight =>
 					return implication_left_to_right(
-						TokenSplit::new(
-							self.filter_logical_connective(top_level_logical_connective),
-							self.input),
-						level + 1),
+						self.split_at_logical_connective(top_level_logical_connective), level + 1),
 				/*LogicalConnective::ImpliesRightToLeft => unimplemented!(),*/
 				_ =>
 				{
@@ -270,17 +272,17 @@ impl<'i> FormulaStr<'i>
 			}
 		}
 
-		let mut comparison_operators = self.filter_comparison_operators();
+		let mut comparison_operators = self.comparison_operators();
 
 		// Parse comparisons
 		if let Some(comparison_operator) = comparison_operators.next()
 		{
-			let (_, _, comparison_operator) = comparison_operator?;
+			let (_, comparison_operator) = comparison_operator?;
 
 			// Comparisons with more than one comparison operator aren’t supported
 			if let Some(next_comparison_operator) = comparison_operators.next()
 			{
-				let (_, _, next_comparison_operator) = next_comparison_operator?;
+				let (_, next_comparison_operator) = next_comparison_operator?;
 
 				return Err(crate::parse::Error::new_multiple_comparison_operators(
 					comparison_operator, next_comparison_operator,
@@ -289,8 +291,7 @@ impl<'i> FormulaStr<'i>
 
 			println!("{}  parsing “{:?}” comparison: {}", indentation, comparison_operator, input);
 
-			let mut comparison_operator_split =
-				TokenSplit::new(self.filter_comparison_operators(), self.input);
+			let mut comparison_operator_split = self.comparison_operators().split();
 
 			// There’s exactly one comparison operator in this formula, as we have verified above.
 			// Hence, the split is guaranteed to generate exactly these two elements
@@ -371,19 +372,18 @@ impl std::fmt::Debug for Quantifier
 }
 
 // TODO: refactor
-fn implication_left_to_right_inner<'i, T>(
-	mut split_formula_at_logical_connective: TokenSplit<'i, T, Token<'i>>, level: usize)
+fn implication_left_to_right_inner<'i, T>(mut logical_connective_iterator: T, level: usize)
 	-> Result<Option<crate::Formula>, crate::parse::Error>
 where
-	T: std::iter::Iterator<Item = Result<(usize, usize, Token<'i>), crate::parse::Error>>
+	T: std::iter::Iterator<Item = Result<&'i str, crate::parse::Error>>
 {
-	match split_formula_at_logical_connective.next()
+	match logical_connective_iterator.next()
 	{
 		Some(argument) =>
 		{
 			// TODO: improve error handling if antecedent cannot be parsed
 			let argument = FormulaStr::new(argument?).parse(level)?;
-			match implication_left_to_right_inner(split_formula_at_logical_connective, level)?
+			match implication_left_to_right_inner(logical_connective_iterator, level)?
 			{
 				Some(next_argument) => Ok(Some(crate::Formula::implies(
 					crate::ImplicationDirection::LeftToRight, Box::new(argument),
@@ -395,19 +395,18 @@ where
 	}
 }
 
-fn implication_left_to_right<'i, T>(
-	mut split_formula_at_logical_connective: TokenSplit<'i, T, Token<'i>>, level: usize)
+fn implication_left_to_right<'i, T>(mut logical_connective_iterator: T, level: usize)
 	-> Result<crate::Formula, crate::parse::Error>
 where
-	T: std::iter::Iterator<Item = Result<(usize, usize, Token<'i>), crate::parse::Error>>
+	T: std::iter::Iterator<Item = Result<&'i str, crate::parse::Error>>
 {
-	match split_formula_at_logical_connective.next()
+	match logical_connective_iterator.next()
 	{
 		Some(argument) =>
 		{
 			// TODO: improve error handling if antecedent cannot be parsed
 			let argument = FormulaStr::new(argument?).parse(level)?;
-			match implication_left_to_right_inner(split_formula_at_logical_connective, level)?
+			match implication_left_to_right_inner(logical_connective_iterator, level)?
 			{
 				Some(next_argument) => Ok(crate::Formula::implies(
 					crate::ImplicationDirection::LeftToRight, Box::new(argument),

src/parse/tokens.rs

@@ -308,12 +308,216 @@ pub(crate) fn parenthesized_expression(input: &str)
 		crate::parse::error::Location::new(0, Some(1))))
 }
 
-pub(crate) trait OriginalInput<'i>
+pub(crate) struct Tokens<'i, F>
 {
-	fn original_input(&self) -> &'i str;
+	original_input: &'i str,
+	input: &'i str,
+	previous_index: usize,
+	reached_end_of_stream: bool,
+	functor: F,
 }
 
-pub(crate) struct TokenIterator<'i>
+impl<'i> Tokens<'i, ()>
+{
+	pub fn new_iter(input: &'i str) -> Tokens<'i, impl FnMut(Token<'i>) -> Option<Token<'i>>>
+	{
+		Tokens::new_filter_map(input, |x| Some(x))
+	}
+
+	pub fn new_filter<P>(input: &'i str, mut predicate: P)
+		-> Tokens<'i, impl FnMut(Token<'i>) -> Option<Token<'i>>>
+	where
+		P: FnMut(&Token<'i>) -> bool
+	{
+		Tokens::new_filter_map(input,
+			move |x|
+			{
+				if predicate(&x)
+				{
+					Some(x)
+				}
+				else
+				{
+					None
+				}
+			})
+	}
+}
+
+impl<'i, F> Tokens<'i, F>
+{
+	pub fn new_filter_map(input: &'i str, functor: F) -> Self
+	{
+		Self
+		{
+			original_input: input,
+			input,
+			previous_index: 0,
+			reached_end_of_stream: false,
+			functor,
+		}
+	}
+
+	fn next_token(&mut self) -> Option<Result<(usize, usize, Token<'i>), crate::parse::Error>>
+	{
+		self.input = self.input.trim_start();
+		let index_left = substring_offset(self.input, self.original_input);
+
+		let first_character = match self.input.chars().next()
+		{
+			None => return None,
+			Some(first_character) => first_character,
+		};
+
+		if self.input.starts_with(")")
+		{
+			return Some(Err(crate::parse::Error::new_unmatched_parenthesis(
+				crate::parse::error::Location::new(0, Some(1)))));
+		}
+
+		match parenthesized_expression(self.input)
+		{
+			Ok(Some((parenthesized_expression, remaining_input))) =>
+			{
+				self.input = remaining_input;
+				let index_right = substring_offset(self.input, self.original_input);
+
+				return Some(Ok((index_left, index_right,
+					Token::ParenthesizedExpression(parenthesized_expression))));
+			},
+			Ok(None) => (),
+			Err(error) => return Some(Err(error)),
+		}
+
+		match number(self.input)
+		{
+			Ok(Some((number, remaining_input))) =>
+			{
+				self.input = remaining_input;
+				let index_right = substring_offset(self.input, self.original_input);
+
+				return Some(Ok((index_left, index_right, Token::Number(number))));
+			},
+			Ok(None) => (),
+			Err(error) => return Some(Err(error)),
+		}
+
+		if let Some((identifier, remaining_input)) = identifier(self.input)
+		{
+			self.input = remaining_input;
+			let index_right = substring_offset(self.input, self.original_input);
+
+			return Some(Ok((index_left, index_right, Token::Identifier(identifier))));
+		}
+
+		if let Some((symbol, remaining_input)) = symbol(self.input)
+		{
+			self.input = remaining_input;
+			let index_right = substring_offset(self.input, self.original_input);
+
+			return Some(Ok((index_left, index_right, Token::Symbol(symbol))));
+		}
+
+		return Some(Err(crate::parse::Error::new_character_not_allowed(first_character,
+			crate::parse::error::Location::new(0, Some(0)))));
+	}
+
+	pub fn remaining_input(&mut self) -> Option<&'i str>
+	{
+		if self.reached_end_of_stream
+		{
+			return None;
+		}
+
+		let remaining_input = self.original_input[self.previous_index..].trim();
+		self.reached_end_of_stream = true;
+
+		Some(remaining_input)
+	}
+
+	pub fn split(self) -> TokenSplit<Self>
+	{
+		TokenSplit::new(self)
+	}
+}
+
+impl<'i, F, G> std::iter::Iterator for Tokens<'i, F>
+where
+	F: FnMut(Token<'i>) -> Option<G>,
+{
+	type Item = Result<(&'i str, G), crate::parse::Error>;
+
+	fn next(&mut self) -> Option<Self::Item>
+	{
+		if self.previous_index == self.original_input.len()
+		{
+			return None;
+		}
+
+		loop
+		{
+			match self.next_token()
+			{
+				Some(Ok((index_left, index_right, token))) =>
+				{
+					let token = match (self.functor)(token)
+					{
+						None => continue,
+						Some(token) => token,
+					};
+
+					let input_left = self.original_input[self.previous_index..index_left].trim();
+					assert!(!input_left.is_empty());
+
+					self.previous_index = index_right;
+
+					return Some(Ok((input_left, token)));
+				},
+				Some(Err(error)) => return Some(Err(error)),
+				None => return None,
+			}
+		}
+	}
+}
+
+pub(crate) struct TokenSplit<T>
+{
+	tokens: T,
+}
+
+impl TokenSplit<()>
+{
+	pub fn new<T>(tokens: T) -> TokenSplit<T>
+	{
+		TokenSplit
+		{
+			tokens,
+		}
+	}
+}
+
+impl<'i, F, G> std::iter::Iterator for TokenSplit<Tokens<'i, F>>
+where
+	F: FnMut(Token<'i>) -> Option<G>,
+{
+	type Item = Result<&'i str, crate::parse::Error>;
+
+	fn next(&mut self) -> Option<Self::Item>
+	{
+		match self.tokens.next()
+		{
+			Some(Ok((input_before, _))) => Some(Ok(input_before)),
+			Some(Err(error)) => Some(Err(error)),
+			None => match self.tokens.remaining_input()
+			{
+				Some(remaining_input) => Some(Ok(remaining_input)),
+				None => None,
+			},
+		}
+	}
+}
+
+/*pub(crate) struct TokenIterator<'i>
 {
 	original_input: &'i str,
 	input: &'i str,
@@ -329,36 +533,8 @@ impl<'i> TokenIterator<'i>
 			input,
 		}
 	}
-
-	/*pub fn filter<P>(self, pattern: P) -> TokenFilter<'i, P>
-	where
-		P: FnMut(&Token<'i>) -> bool,
-	{
-		TokenFilter::new(self, pattern)
-	}
-
-	pub fn split(self) -> TokenSplit<'i, Self>
-	{
-		TokenSplit::new(self)
-	}*/
 }
 
-/*impl<'i> OriginalInput<'i> for TokenIterator<'i>
-{
-	fn original_input(&self) -> &'i str
-	{
-		self.original_input
-	}
-}
-
-impl<'i, P> OriginalInput<'i> for std::iter::Filter<TokenIterator<'i>, P>
-{
-	fn original_input(&self) -> &'i str
-	{
-		self.iter.original_input
-	}
-}*/
-
 impl<'i> std::iter::Iterator for TokenIterator<'i>
 {
 	type Item = Result<(usize, usize, Token<'i>), crate::parse::Error>;
@@ -491,7 +667,7 @@ where
 			}
 		}
 	}
-}
+}*/
 
 #[cfg(test)]
 mod tests