reifydb-rql 0.4.12

// SPDX-License-Identifier: Apache-2.0
// Copyright (c) 2025 ReifyDB

use crate::token::{
	cursor::Cursor,
	identifier::is_identifier_char,
	token::{Literal, Token, TokenKind},
};

/// Scan for a number literal
pub fn scan_number<'b>(cursor: &mut Cursor<'b>) -> Option<Token<'b>> {
	let start_pos = cursor.pos();
	let start_line = cursor.line();
	let start_column = cursor.column();

	// Check for hex (0x...)
	if cursor.peek_str(2).eq_ignore_ascii_case("0x") {
		cursor.consume();
		cursor.consume();
		let hex_part = cursor.consume_while(|c| c.is_ascii_hexdigit() || c == '_');
		if !hex_part.is_empty()
			&& !hex_part.starts_with('_')
			&& !hex_part.ends_with('_')
			&& !hex_part.contains("__")
		{
			return Some(Token {
				kind: TokenKind::Literal(Literal::Number),
				fragment: cursor.make_fragment(start_pos, start_line, start_column),
			});
		}
		return None;
	}

	// Check for binary (0b...)
	if cursor.peek_str(2).eq_ignore_ascii_case("0b") {
		cursor.consume();
		cursor.consume();
		let bin_part = cursor.consume_while(|c| c == '0' || c == '1' || c == '_');
		if !bin_part.is_empty()
			&& !bin_part.starts_with('_')
			&& !bin_part.ends_with('_')
			&& !bin_part.contains("__")
		{
			return Some(Token {
				kind: TokenKind::Literal(Literal::Number),
				fragment: cursor.make_fragment(start_pos, start_line, start_column),
			});
		}
		return None;
	}

	// Check for octal (0o...)
	if cursor.peek_str(2).eq_ignore_ascii_case("0o") {
		cursor.consume();
		cursor.consume();
		let oct_part = cursor.consume_while(|c| ('0'..='7').contains(&c) || c == '_');
		if !oct_part.is_empty()
			&& !oct_part.starts_with('_')
			&& !oct_part.ends_with('_')
			&& !oct_part.contains("__")
		{
			return Some(Token {
				kind: TokenKind::Literal(Literal::Number),
				fragment: cursor.make_fragment(start_pos, start_line, start_column),
			});
		}
		return None;
	}

	// Decimal number (including float and scientific notation)
	let state = cursor.save_state();

	// Check for leading dot (.123)
	let has_leading_dot = cursor.peek() == Some('.');
	if has_leading_dot {
		cursor.consume();
		if !cursor.peek().is_some_and(|c| c.is_ascii_digit()) {
			cursor.restore_state(state);
			return None;
		}
	} else if !cursor.peek().is_some_and(|c| c.is_ascii_digit()) {
		return None;
	}

	// Integer part (if no leading dot)
	if !has_leading_dot {
		cursor.consume_while(|c| c.is_ascii_digit() || c == '_');
	}

	// Fractional part
	if cursor.peek() == Some('.') && !has_leading_dot {
		if cursor.peek_ahead(1).is_some_and(|c| c.is_ascii_digit()) {
			cursor.consume(); // consume '.'
			cursor.consume_while(|c| c.is_ascii_digit() || c == '_');
		}
	} else if has_leading_dot {
		// Already consumed the dot
		cursor.consume_while(|c| c.is_ascii_digit() || c == '_');
	}

	// Scientific notation (e/E)
	if let Some(e) = cursor.peek()
		&& (e == 'e' || e == 'E')
	{
		cursor.consume();
		if let Some(sign) = cursor.peek()
			&& (sign == '+' || sign == '-')
		{
			cursor.consume();
		}
		let exp_part = cursor.consume_while(|c| c.is_ascii_digit() || c == '_');
		if exp_part.is_empty() {
			// Invalid scientific notation
			cursor.restore_state(state);
			return None;
		}
	}

	// Make sure we consumed something
	if cursor.pos() == start_pos {
		return None;
	}

	// Special case: leading dot decimals followed by identifier chars should be rejected
	// This allows ".5sec" to parse as Dot + Number("5") + Identifier("sec")
	// instead of Number(".5") + Identifier("sec")
	if has_leading_dot && cursor.peek().is_some_and(is_identifier_char) {
		cursor.restore_state(state);
		return None;
	}

	Some(Token {
		kind: TokenKind::Literal(Literal::Number),
		fragment: cursor.make_fragment(start_pos, start_line, start_column),
	})
}

#[cfg(test)]
pub mod tests {
	use Literal::Number;

	use super::*;
	use crate::{
		bump::Bump,
		token::{operator::Operator, tokenize},
	};

	#[test]
	fn test_decimal_integer() {
		let bump = Bump::new();
		let tokens = tokenize(&bump, "42").unwrap();
		assert_eq!(tokens[0].kind, TokenKind::Literal(Number));
		assert_eq!(tokens[0].fragment.text(), "42");
	}

	#[test]
	fn test_decimal_float() {
		let bump = Bump::new();
		let tokens = tokenize(&bump, "3.14").unwrap();
		assert_eq!(tokens[0].kind, TokenKind::Literal(Number));
		assert_eq!(tokens[0].fragment.text(), "3.14");
	}

	#[test]
	fn test_decimal_with_underscores() {
		let bump = Bump::new();
		let tokens = tokenize(&bump, "1_234_567").unwrap();
		assert_eq!(tokens[0].kind, TokenKind::Literal(Number));
		assert_eq!(tokens[0].fragment.text(), "1_234_567");
	}

	#[test]
	fn test_scientific_notation() {
		let bump = Bump::new();
		let tokens = tokenize(&bump, "1.23e10").unwrap();
		assert_eq!(tokens[0].kind, TokenKind::Literal(Number));
		assert_eq!(tokens[0].fragment.text(), "1.23e10");

		let tokens = tokenize(&bump, "5E-3").unwrap();
		assert_eq!(tokens[0].kind, TokenKind::Literal(Number));
		assert_eq!(tokens[0].fragment.text(), "5E-3");
	}

	#[test]
	fn test_hex_number() {
		let bump = Bump::new();
		let tokens = tokenize(&bump, "0x2A").unwrap();
		assert_eq!(tokens[0].kind, TokenKind::Literal(Number));
		assert_eq!(tokens[0].fragment.text(), "0x2A");

		let tokens = tokenize(&bump, "0xDEAD_BEEF").unwrap();
		assert_eq!(tokens[0].kind, TokenKind::Literal(Number));
		assert_eq!(tokens[0].fragment.text(), "0xDEAD_BEEF");
	}

	#[test]
	fn test_binary_number() {
		let bump = Bump::new();
		let tokens = tokenize(&bump, "0b1010").unwrap();
		assert_eq!(tokens[0].kind, TokenKind::Literal(Number));
		assert_eq!(tokens[0].fragment.text(), "0b1010");

		let tokens = tokenize(&bump, "0b1111_0000").unwrap();
		assert_eq!(tokens[0].kind, TokenKind::Literal(Number));
		assert_eq!(tokens[0].fragment.text(), "0b1111_0000");
	}

	#[test]
	fn test_octal_number() {
		let bump = Bump::new();
		let tokens = tokenize(&bump, "0o777").unwrap();
		assert_eq!(tokens[0].kind, TokenKind::Literal(Number));
		assert_eq!(tokens[0].fragment.text(), "0o777");

		let tokens = tokenize(&bump, "0o12_34").unwrap();
		assert_eq!(tokens[0].kind, TokenKind::Literal(Number));
		assert_eq!(tokens[0].fragment.text(), "0o12_34");
	}

	#[test]
	fn test_leading_dot() {
		let bump = Bump::new();
		let tokens = tokenize(&bump, ".5").unwrap();
		assert_eq!(tokens[0].kind, TokenKind::Literal(Number));
		assert_eq!(tokens[0].fragment.text(), ".5");
	}

	#[test]
	fn test_leading_dot_decimal_with_identifier() {
		let bump = Bump::new();
		// Leading dot followed by digit and then identifier tokenizes as Dot + Identifier
		let tokens = tokenize(&bump, ".5sec").unwrap();
		assert_eq!(tokens.len(), 2); // Dot, Identifier("5sec")
		assert_eq!(tokens[0].kind, TokenKind::Operator(Operator::Dot));
		assert_eq!(tokens[1].kind, TokenKind::Identifier);
		assert_eq!(tokens[1].fragment.text(), "5sec");
	}

	#[test]
	fn test_leading_dot_decimal_standalone() {
		let bump = Bump::new();
		// Leading dot decimals should work when standalone or with spacing
		let tokens = tokenize(&bump, ".5").unwrap();
		assert_eq!(tokens.len(), 1);
		assert_eq!(tokens[0].kind, TokenKind::Literal(Number));
		assert_eq!(tokens[0].fragment.text(), ".5");
	}

	#[test]
	fn test_number_with_trailing() {
		let bump = Bump::new();
		// Digit-starting tokens with alpha chars are now single Identifier tokens
		let tokens = tokenize(&bump, "42abc").unwrap();
		assert_eq!(tokens.len(), 1);
		assert_eq!(tokens[0].kind, TokenKind::Identifier);
		assert_eq!(tokens[0].fragment.text(), "42abc");

		// With proper spacing, they remain separate
		let tokens = tokenize(&bump, "42 abc").unwrap();
		assert_eq!(tokens[0].kind, TokenKind::Literal(Number));
		assert_eq!(tokens[0].fragment.text(), "42");
	}

	#[test]
	fn test_invalid_numbers() {
		let bump = Bump::new();
		// Invalid hex (starts with _)
		let result = tokenize(&bump, "0x_FF");
		assert!(result.is_err() || result.unwrap()[0].fragment.text() != "0x_FF");

		// Invalid binary (contains 2)
		let result = tokenize(&bump, "0b102");
		assert!(result.is_ok()); // Will be parsed as 0b10 followed by 2
		let tokens = result.unwrap();
		assert_eq!(tokens[0].fragment.text(), "0b10");
		assert_eq!(tokens[1].fragment.text(), "2");
	}
}