rlox/src/frontend/lexer.rs

use crate::common::base_value::{BaseValue, Number};
use crate::common::lox_result::{lexical_error, LoxError, LoxResult};
use crate::frontend::source_registry::{SourceId, SourcePosition, SourceSlice};
use crate::frontend::tokens::{Token, TokenType};

pub struct Lexer {
    input: String,
    // Byte offsets into `input` (not char counts) so slicing is UTF-8 correct.
    start: usize,
    current: usize,
    start_pos: SourcePosition,
    end_pos: SourcePosition,
    source_id: SourceId,
}

fn get_keyword_token(word: &str) -> Option<TokenType> {
    match word {
        "and" => Some(TokenType::And),
        "struct" => Some(TokenType::Struct),
        "do" => Some(TokenType::StartBlock),
        "end" => Some(TokenType::EndBlock),
        "false" => Some(TokenType::False),
        "for" => Some(TokenType::For),
        "fun" => Some(TokenType::Fun),
        "if" => Some(TokenType::If),
        "then" => Some(TokenType::Then),
        "elif" => Some(TokenType::Elif),
        "else" => Some(TokenType::Else),
        "or" => Some(TokenType::Or),
        "in" => Some(TokenType::In),
        "print" => Some(TokenType::Print),
        "return" => Some(TokenType::Return),
        "break" => Some(TokenType::Break),
        "super" => Some(TokenType::Super),
        "this" => Some(TokenType::This),
        "true" => Some(TokenType::True),
        "var" => Some(TokenType::Var),
        "val" => Some(TokenType::Val),
        "while" => Some(TokenType::While),
        "True" => Some(TokenType::True),
        "False" => Some(TokenType::False),
        "Nil" => Some(TokenType::Nil),
        "fn" => Some(TokenType::Fn),
        _ => None,
    }
}

impl Lexer {
    pub fn new(input: String, source_id: SourceId) -> Lexer {
        Lexer {
            input,
            start: 0,
            current: 0,
            start_pos: SourcePosition::default(),
            end_pos: SourcePosition::default(),
            source_id,
        }
    }

    pub fn scans_tokens(&mut self) -> LoxResult<Vec<Token>> {
        let mut tokens = Vec::new();
        while !self.is_at_end() {
            self.start = self.current;
            self.start_pos = self.end_pos.clone();
            match self.scan_token() {
                Ok(Some(token)) => tokens.push(token),
                Ok(None) => {}
                Err(err) => return Err(err),
            }
        }
        tokens.push(self.make_token(TokenType::Eof));
        Ok(tokens)
    }

    fn is_at_end(&self) -> bool {
        self.current >= self.input.len()
    }

    fn advance(&mut self) -> char {
        let c = self.input[self.current..].chars().next().unwrap();
        self.current += c.len_utf8();
        if c == '\n' {
            self.end_pos.line += 1;
            self.end_pos.column = 0;
        } else {
            self.end_pos.column += 1;
        }
        c
    }

    fn peek(&self) -> char {
        self.input[self.current..].chars().next().unwrap_or('\0')
    }

    fn peek_next(&self) -> char {
        self.input[self.current..].chars().nth(1).unwrap_or('\0')
    }

    fn make_token(&self, token_type: TokenType) -> Token {
        let text = self.input[self.start..self.current].to_string();
        Token::new(
            token_type,
            text,
            SourceSlice::from_positions(
                self.source_id.clone(),
                self.start_pos.clone(),
                self.end_pos.clone(),
            ),
        )
    }
    fn make_token_with_literal(&self, token_type: TokenType, literal: BaseValue) -> Token {
        let text = self.input[self.start..self.current].to_string();
        Token::new_complete(
            token_type,
            text,
            Some(literal),
            SourceSlice::from_positions(
                self.source_id.clone(),
                self.start_pos.clone(),
                self.end_pos.clone(),
            ),
        )
    }

    fn scan_token(&mut self) -> LoxResult<Option<Token>> {
        let c = self.advance();
        match (c, self.peek()) {
            ('(', _) => Ok(Some(self.make_token(TokenType::LeftParen))),
            (')', _) => Ok(Some(self.make_token(TokenType::RightParen))),
            ('{', _) => Ok(Some(self.make_token(TokenType::LeftBrace))),
            ('}', _) => Ok(Some(self.make_token(TokenType::RightBrace))),
            ('[', _) => Ok(Some(self.make_token(TokenType::LeftBracket))),
            (']', _) => Ok(Some(self.make_token(TokenType::RightBracket))),
            (',', _) => Ok(Some(self.make_token(TokenType::Comma))),
            ('.', _) => Ok(Some(self.make_token(TokenType::Dot))),
            ('-', _) => Ok(Some(self.make_token(TokenType::Minus))),
            ('+', _) => Ok(Some(self.make_token(TokenType::Plus))),
            (';', _) => Ok(Some(self.make_token(TokenType::Semicolon))),
            (':', _) => Ok(Some(self.make_token(TokenType::Colon))),
            ('*', _) => Ok(Some(self.make_token(TokenType::Star))),
            ('%', _) => Ok(Some(self.make_token(TokenType::Percent))),
            ('!', '=') => {
                self.advance(); // consuma il '='
                Ok(Some(self.make_token(TokenType::BangEqual)))
            }
            ('!', _) => Ok(Some(self.make_token(TokenType::Bang))),
            ('=', '=') => {
                self.advance(); // consuma il '='
                Ok(Some(self.make_token(TokenType::EqualEqual)))
            }
            ('=', _) => Ok(Some(self.make_token(TokenType::Equal))),
            ('<', '=') => {
                self.advance(); // consuma il '='
                Ok(Some(self.make_token(TokenType::LessEqual)))
            }
            ('<', _) => Ok(Some(self.make_token(TokenType::Less))),
            ('>', '=') => {
                self.advance(); // consuma il '='
                Ok(Some(self.make_token(TokenType::GreaterEqual)))
            }
            ('>', _) => Ok(Some(self.make_token(TokenType::Greater))),
            ('/', '/') => {
                // Commento single-line
                while self.peek() != '\n' && !self.is_at_end() {
                    self.advance();
                }
                Ok(None)
            }
            ('/', '*') => {
                // Commento multi-line
                while (self.peek() != '*' || self.peek_next() != '/') && !self.is_at_end() {
                    self.advance();
                }
                if self.is_at_end() {
                    return lexical_error(
                        SourceSlice::from_positions(
                            self.source_id.clone(),
                            self.start_pos.clone(),
                            self.end_pos.clone(),
                        ),
                        "Unterminated comment".to_string(),
                    );
                } else {
                    self.advance(); // consuma '*'
                    self.advance(); // consuma '/'
                }
                Ok(None)
            }
            ('/', _) => Ok(Some(self.make_token(TokenType::Slash))),
            (' ', _) | ('\r', _) | ('\t', _) => Ok(None),
            ('\n', _) => Ok(None),
            ('"', _) => self.string(),
            (c, _) if c.is_digit(10) => self.number(),
            (c, _) if c.is_alphanumeric() || c == '_' => self.identifier(),
            _ => Err(LoxError::LexicalError {
                source_slice: SourceSlice::from_positions(
                    self.source_id.clone(),
                    self.start_pos.clone(),
                    self.end_pos.clone(),
                ),
                message: format!("Unexpected character: {}", c),
            }),
        }
    }

    fn string(&mut self) -> LoxResult<Option<Token>> {
        while self.peek() != '"' && !self.is_at_end() {
            self.advance();
        }
        if self.is_at_end() {
            return Err(LoxError::LexicalError {
                source_slice: SourceSlice::from_positions(
                    self.source_id.clone(),
                    self.start_pos.clone(),
                    self.end_pos.clone(),
                ),
                message: "Unterminated string".to_string(),
            });
        }
        self.advance();
        Ok(Some(self.make_token_with_literal(
            TokenType::String,
            BaseValue::String(self.input[self.start..self.current].to_string()),
        )))
    }

    fn number(&mut self) -> LoxResult<Option<Token>> {
        // Leggi la parte intera
        while self.peek().is_digit(10) {
            self.advance();
        }

        // Controlla se c'è una parte decimale
        let has_decimal = if self.peek() == '.' && self.peek_next().is_digit(10) {
            self.advance(); // consuma il '.'
            while self.peek().is_digit(10) {
                self.advance();
            }
            true
        } else {
            false
        };

        // Controlla se c'è un suffisso (f o u)
        let suffix = if self.peek() == 'f' || self.peek() == 'u' {
            let s = self.peek();
            self.advance();
            Some(s)
        } else {
            None
        };

        let num_str = &self.input[self.start..self.current];
        let num_str_without_suffix = if suffix.is_some() {
            &num_str[..num_str.len() - 1]
        } else {
            num_str
        };

        // Determina il tipo di numero basandosi sul contenuto e suffisso
        let number_value = match suffix {
            Some('f') => {
                // Float esplicito con suffisso 'f'
                Number::F64(num_str_without_suffix.parse().unwrap())
            }
            Some('u') => {
                // Unsigned esplicito con suffisso 'u'
                Number::U128(num_str_without_suffix.parse().unwrap())
            }
            _ if has_decimal => {
                // Float implicito (contiene un punto decimale)
                Number::F64(num_str.parse().unwrap())
            }
            _ => {
                // Intero con segno di default
                Number::I32(num_str.parse().unwrap())
            }
        };

        Ok(Some(self.make_token_with_literal(
            TokenType::Number,
            BaseValue::Number(number_value),
        )))
    }

    fn identifier(&mut self) -> LoxResult<Option<Token>> {
        while self.peek().is_alphanumeric() || self.peek() == '_' {
            self.advance();
        }
        let text = self.input[self.start..self.current].to_string();
        match get_keyword_token(&text) {
            Some(TokenType::True) => Ok(Some(
                self.make_token_with_literal(TokenType::True, BaseValue::Boolean(true)),
            )),
            Some(TokenType::False) => Ok(Some(
                self.make_token_with_literal(TokenType::False, BaseValue::Boolean(false)),
            )),
            Some(TokenType::Nil) => Ok(Some(
                self.make_token_with_literal(TokenType::Nil, BaseValue::Nil),
            )),
            Some(token_type) => Ok(Some(self.make_token(token_type))),
            None => Ok(Some(self.make_token_with_literal(
                TokenType::Identifier,
                BaseValue::String(text.clone()),
            ))),
        }
    }
}

#[cfg(test)]
mod tests {
    use std::f64;

    use super::*;

    /// Lex `input` and return the resulting tokens, panicking on lexical errors.
    fn lex(input: &str) -> Vec<Token> {
        Lexer::new(input.to_string(), 0)
            .scans_tokens()
            .expect("expected input to lex without errors")
    }

    /// Lex `input` and collect just the token types (including the trailing EOF).
    fn token_types(input: &str) -> Vec<TokenType> {
        lex(input).into_iter().map(|t| t.token_type).collect()
    }

    #[test]
    fn scans_single_character_tokens() {
        assert_eq!(
            token_types("(){}[],.-+;:*%"),
            vec![
                TokenType::LeftParen,
                TokenType::RightParen,
                TokenType::LeftBrace,
                TokenType::RightBrace,
                TokenType::LeftBracket,
                TokenType::RightBracket,
                TokenType::Comma,
                TokenType::Dot,
                TokenType::Minus,
                TokenType::Plus,
                TokenType::Semicolon,
                TokenType::Colon,
                TokenType::Star,
                TokenType::Percent,
                TokenType::Eof,
            ]
        );
    }

    #[test]
    fn scans_one_and_two_char_operators() {
        assert_eq!(
            token_types("! != = == < <= > >= /"),
            vec![
                TokenType::Bang,
                TokenType::BangEqual,
                TokenType::Equal,
                TokenType::EqualEqual,
                TokenType::Less,
                TokenType::LessEqual,
                TokenType::Greater,
                TokenType::GreaterEqual,
                TokenType::Slash,
                TokenType::Eof,
            ]
        );
    }

    #[test]
    fn keyword_lookup_matches_known_words() {
        assert_eq!(get_keyword_token("and"), Some(TokenType::And));
        assert_eq!(get_keyword_token("if"), Some(TokenType::If));
        assert_eq!(get_keyword_token("then"), Some(TokenType::Then));
        assert_eq!(get_keyword_token("while"), Some(TokenType::While));
        assert_eq!(get_keyword_token("do"), Some(TokenType::StartBlock));
        assert_eq!(get_keyword_token("end"), Some(TokenType::EndBlock));
        assert_eq!(get_keyword_token("not_a_keyword"), None);
    }

    #[test]
    fn scans_keywords_in_a_stream() {
        assert_eq!(
            token_types("if then else while print return"),
            vec![
                TokenType::If,
                TokenType::Then,
                TokenType::Else,
                TokenType::While,
                TokenType::Print,
                TokenType::Return,
                TokenType::Eof,
            ]
        );
    }

    #[test]
    fn scans_integer_number() {
        let tokens = lex("42");
        assert_eq!(tokens[0].token_type, TokenType::Number);
        assert_eq!(tokens[0].literal, Some(BaseValue::Number(Number::I32(42))));
    }

    #[test]
    fn scans_float_number() {
        let tokens = lex("3.141592653589793");
        assert_eq!(
            tokens[0].literal,
            Some(BaseValue::Number(Number::F64(f64::consts::PI)))
        );
    }

    #[test]
    fn scans_number_with_float_suffix() {
        let tokens = lex("5f");
        assert_eq!(tokens[0].literal, Some(BaseValue::Number(Number::F64(5.0))));
    }

    #[test]
    fn scans_number_with_unsigned_suffix() {
        let tokens = lex("7u");
        assert_eq!(tokens[0].literal, Some(BaseValue::Number(Number::U128(7))));
    }

    #[test]
    fn scans_string_literal_including_quotes() {
        // The lexer slices from the opening quote through the closing quote,
        // so the literal currently retains the surrounding quotes.
        let tokens = lex("\"hello\"");
        assert_eq!(tokens[0].token_type, TokenType::String);
        assert_eq!(
            tokens[0].literal,
            Some(BaseValue::String("\"hello\"".to_string()))
        );
    }

    #[test]
    fn scans_identifier() {
        let tokens = lex("foo_bar");
        assert_eq!(tokens[0].token_type, TokenType::Identifier);
        assert_eq!(
            tokens[0].literal,
            Some(BaseValue::String("foo_bar".to_string()))
        );
    }

    #[test]
    fn scans_boolean_and_nil_literals() {
        let tokens = lex("true false Nil");
        assert_eq!(tokens[0].literal, Some(BaseValue::Boolean(true)));
        assert_eq!(tokens[1].literal, Some(BaseValue::Boolean(false)));
        assert_eq!(tokens[2].literal, Some(BaseValue::Nil));
    }

    #[test]
    fn ignores_line_comments() {
        assert_eq!(
            token_types("1 // a comment\n2"),
            vec![TokenType::Number, TokenType::Number, TokenType::Eof]
        );
    }

    #[test]
    fn ignores_block_comments() {
        assert_eq!(
            token_types("1 /* multi\nline */ 2"),
            vec![TokenType::Number, TokenType::Number, TokenType::Eof]
        );
    }

    #[test]
    fn always_appends_eof_even_for_empty_input() {
        let tokens = lex("");
        assert_eq!(tokens.len(), 1);
        assert_eq!(tokens[0].token_type, TokenType::Eof);
    }

    #[test]
    fn unterminated_string_is_a_lexical_error() {
        let result = Lexer::new("\"oops".to_string(), 0).scans_tokens();
        assert!(matches!(result, Err(LoxError::LexicalError { .. })));
    }

    #[test]
    fn unterminated_block_comment_is_a_lexical_error() {
        let result = Lexer::new("/* never ends".to_string(), 0).scans_tokens();
        assert!(matches!(result, Err(LoxError::LexicalError { .. })));
    }

    #[test]
    fn unexpected_character_is_a_lexical_error() {
        let result = Lexer::new("@".to_string(), 0).scans_tokens();
        assert!(result.is_err());
    }

    #[test]
    fn handles_multibyte_identifier() {
        // `é` is two UTF-8 bytes; the old char-counted slicing would panic or
        // slice mid-codepoint here. Byte offsets make this correct.
        let tokens = lex("café");
        assert_eq!(tokens[0].token_type, TokenType::Identifier);
        assert_eq!(
            tokens[0].literal,
            Some(BaseValue::String("café".to_string()))
        );
    }

    #[test]
    fn tracks_line_and_column_across_newlines() {
        // "1\n22": the second token sits at the start of line 1.
        let tokens = lex("1\n22");
        assert_eq!(tokens[1].lexeme, "22");
        assert_eq!(tokens[1].source_slice.start_position.line, 1);
        assert_eq!(tokens[1].source_slice.start_position.column, 0);
    }

    #[test]
    fn handle_struct() {
        let tokens = lex("Cosa :: struct {field1: Number, field2: String}");
        let tokens_tyepe: Vec<TokenType> = tokens.iter().map(|x| x.token_type.clone()).collect();
        let tokent_for_check = vec![
            TokenType::Identifier,
            TokenType::Colon,
            TokenType::Colon,
            TokenType::Struct,
            TokenType::LeftBrace,
            TokenType::Identifier,
            TokenType::Colon,
            TokenType::Identifier,
            TokenType::Comma,
            TokenType::Identifier,
            TokenType::Colon,
            TokenType::Identifier,
            TokenType::RightBrace,
            TokenType::Eof,
        ];
        assert_eq!(tokens_tyepe, tokent_for_check);
        assert_eq!(tokens[0].lexeme, "Cosa");
        assert_eq!(tokens[5].lexeme, "field1");
        assert_eq!(tokens[7].lexeme, "Number");
        assert_eq!(tokens[9].lexeme, "field2");
        assert_eq!(tokens[11].lexeme, "String");
    }
}