prolog2/parser/

tokeniser.rs

//! Lexer: converts Prolog source text into a token stream.

const DELIMINATORS: &[char] = &[
    '(', ')', ',', '.', ' ', '\r', '\n', '\t', '\\', ':', '-', '+', '/', '*', '=', '[', ']', '|',
    '>', '<', '{', '}',
];
const KNOWN_SYMBOLS: &[&str] = &[":-", "==", "=/=", "/=", "=:=", "**", "<=", ">=", "/*", "*/"];

// --------------------------------------------------------------------------------------
// Tokenise File
// --------------------------------------------------------------------------------------

fn form_empty_list_token(tokens: &mut Vec<String>) {
    let mut i = 0;
    while tokens.len() > i + 1 {
        if tokens[i] == "[" {
            let mut j = i + 1;
            let mut new_lines = 0;
            loop {
                match tokens[j].as_str() {
                    " " | "\t" => j += 1,
                    "\n" => {
                        new_lines += 1;
                        j += 1;
                    }
                    "]" => {
                        tokens.drain(i..j + 1);
                        tokens.insert(i, "[]".into());
                        for _ in 0..new_lines {
                            tokens.insert(i + 1, "\n".into());
                        }
                        break;
                    }
                    _ => break,
                }
            }
        }
        i += 1;
    }
}

fn form_empty_set_token(tokens: &mut Vec<String>) {
    let mut i = 0;
    while tokens.len() > i + 1 {
        if tokens[i] == "{" {
            let mut j = i + 1;
            let mut new_lines = 0;
            loop {
                match tokens[j].as_str() {
                    " " | "\t" => j += 1,
                    "\n" => {
                        new_lines += 1;
                        j += 1;
                    }
                    "}" => {
                        tokens.drain(i..j + 1);
                        tokens.insert(i, "{}".into());
                        for _ in 0..new_lines {
                            tokens.insert(i + 1, "\n".into());
                        }
                        break;
                    }
                    _ => break,
                }
            }
        }
        i += 1;
    }
}

fn form_empty_tuple_token(tokens: &mut Vec<String>) {
    let mut i = 0;
    while tokens.len() > i + 1 {
        if tokens[i] == "(" {
            let mut j = i + 1;
            let mut new_lines = 0;
            loop {
                match tokens[j].as_str() {
                    " " | "\t" => j += 1,
                    "\n" => {
                        new_lines += 1;
                        j += 1;
                    }
                    ")" => {
                        tokens.drain(i..j + 1);
                        tokens.insert(i, "()".into());
                        for _ in 0..new_lines {
                            tokens.insert(i + 1, "\n".into());
                        }
                        break;
                    }
                    _ => break,
                }
            }
        }
        i += 1;
    }
}

fn walk_string(
    characters: &Vec<char>,
    mut i: usize,
    mark: char,
) -> Result<(String, usize), String> {
    // TODO: more complex escape character processing
    let mut str = vec![mark];
    while let Some(&c) = characters.get(i) {
        match c {
            c if c == mark => {
                str.push(c);
                return Ok((str.iter().collect(), i + 1));
            }
            '\\' => {
                match characters.get(i + 1) {
                    Some('n') => str.push('\n'),
                    Some('t') => str.push('\t'),
                    Some('\\') => str.push('\\'),
                    Some('"') => str.push('"'),
                    Some('\'') => str.push('\''),
                    Some(_) => return Err("'\\' used without proper escape character".into()),
                    None => break,
                }
                i += 2
            }
            _ => {
                str.push(c);
                i += 1;
            }
        }
    }
    Err(format!("Unexpected end of file, missing closing {mark}"))
}

fn walk_multi_line_comment(characters: &Vec<char>, mut i: usize) -> Result<(usize, usize), String> {
    let mut newlines = 0;
    while i <= characters.len() - 2 {
        if characters[i] == '\n' {
            newlines += 1
        }
        if &characters[i..i + 2] == ['*', '/'] {
            return Ok((i + 2, newlines));
        }
        i += 1;
    }
    Err("Unclosed multi line comment".into())
}

fn walk_single_line_comment(characters: &Vec<char>, mut i: usize) -> Result<usize, ()> {
    while let Some(&c) = characters.get(i) {
        if c == '\n' {
            return Ok(i + 1);
        }
        i += 1;
    }
    Err(())
}

/* If two numerical tokens are split by a '.' char unify into one token
*/
fn join_decimal_nums(tokens: &mut Vec<String>) {
    let mut i = 0;
    while tokens.len() > i + 2 {
        i += 1;
        if tokens[i] == "."
            && tokens[i - 1].chars().all(|c| c.is_numeric())
            && tokens[i + 1].chars().all(|c| c.is_numeric())
        {
            let combined_value = [
                tokens[i - 1].as_str(),
                tokens[i].as_str(),
                tokens[i + 1].as_str(),
            ]
            .concat();
            tokens.remove(i + 1);
            tokens.remove(i);
            tokens.remove(i - 1);
            tokens.insert(i - 1, combined_value);
        }
    }
}

/* If a substract sign is at the front of number unifiy into one token
*/
fn form_negative_nums(tokens: &mut Vec<String>) {
    let mut i = 0;
    while tokens.len() > i + 1 {
        if tokens[i] == "-" && tokens[i + 1].chars().all(|c| c.is_numeric() || c == '.') {
            let combined_value = [tokens[i].as_str(), tokens[i + 1].as_str()].concat();
            tokens.remove(i + 1);
            tokens.remove(i);
            tokens.insert(i, combined_value);
        }
        i += 1;
    }
}

fn form_known_symbols(tokens: &mut Vec<String>) {
    let mut i = 0;
    while i < tokens.len() - 1 {
        if DELIMINATORS.contains(&tokens[i].chars().next().unwrap()) {
            for &symbol in KNOWN_SYMBOLS {
                if symbol.len() <= tokens.len() - i {
                    if symbol == tokens[i..i + symbol.len()].concat() {
                        tokens.drain(i..i + symbol.len());
                        tokens.insert(i, symbol.into());
                    }
                }
            }
        }
        i += 1;
    }
}

pub fn tokenise(text: String) -> Result<Vec<String>, String> {
    let mut tokens = Vec::<String>::new();
    let mut last_i = 0;
    let characters: Vec<char> = text.chars().collect();
    let mut i = 0;

    while i < characters.len() {
        // println!("{tokens:?}");
        let c = characters[i];
        match c {
            '\'' | '"' => {
                let (token, i2) = walk_string(&characters, i + 1, c)?;
                tokens.push(token);
                i = i2;
                last_i = i;
            }
            '%' => match walk_single_line_comment(&characters, i + 1) {
                Ok(i2) => {
                    i = i2;
                    last_i = i;
                    tokens.push("\n".into());
                }
                Err(_) => {
                    i = characters.len();
                    last_i = i;
                    break;
                }
            },
            '/' => {
                if characters[i + 1] == '*' {
                    let (i2, newlines) = walk_multi_line_comment(&characters, i + 1)?;
                    i = i2;
                    last_i = i;
                    for _ in 0..newlines {
                        tokens.push("\n".into());
                    }
                } else {
                    tokens.push(text[last_i..i].into());
                    tokens.push(c.into());
                    i += 1;
                    last_i = i;
                }
            }
            c if DELIMINATORS.contains(&c) => {
                tokens.push(text[last_i..i].into());
                tokens.push(c.into());
                i += 1;
                last_i = i;
            }
            _ => i += 1,
        }
    }
    tokens.push(text[last_i..].into());
    tokens.retain(|token| "" != *token);
    form_known_symbols(&mut tokens);
    join_decimal_nums(&mut tokens);
    form_negative_nums(&mut tokens);
    form_empty_list_token(&mut tokens);
    form_empty_set_token(&mut tokens);
    form_empty_tuple_token(&mut tokens);
    tokens.retain(|t1| !["\t", " ", "\r"].iter().any(|t2| t1 == t2));
    Ok(tokens)
}

#[cfg(test)]
mod tests {
    use super::tokenise;

    #[test]
    fn single_line_comments() {
        let file = "\n%simple predicate\np(x,y):-%The head\nq(x),%body 1\nr(y).%body 2".to_string();

        let tokens = tokenise(file).unwrap();

        assert_eq!(
            tokens,
            [
                "\n", "\n", "p", "(", "x", ",", "y", ")", ":-", "\n", "q", "(", "x", ")", ",",
                "\n", "r", "(", "y", ")", "."
            ]
        );
    }

    #[test]
    fn multi_line_comments() {
        let file = "/* This is a\nmutli\nline\ncomment */\np(x,y):-q(x,y).".to_string();

        let tokens = tokenise(file).unwrap();

        assert_eq!(
            tokens,
            [
                "\n", "\n", "\n", "\n", "p", "(", "x", ",", "y", ")", ":-", "q", "(", "x", ",",
                "y", ")", "."
            ]
        )
    }

    #[test]
    fn unclosed_multi_line_comments() {
        let file = "p(x,y):-q(x,y)
        /* a comment
        on two lines
        fact(a,b)."
            .to_string();

        match tokenise(file) {
            Ok(tokens) => panic!("Should have thrown error\nTokens: {tokens:?}"),
            Err(message) => assert_eq!(message, "Unclosed multi line comment"),
        }
    }

    #[test]
    fn unclosed_strings() {
        let file = "\"a string".to_string();

        match tokenise(file) {
            Ok(tokens) => panic!("Should have thrown error\nTokens: {tokens:?}"),
            Err(message) => assert_eq!(message, "Unexpected end of file, missing closing \""),
        }

        let file = "'a string".to_string();

        match tokenise(file) {
            Ok(tokens) => panic!("Should have thrown error\nTokens: {tokens:?}"),
            Err(message) => assert_eq!(message, "Unexpected end of file, missing closing '"),
        }
    }

    #[test]
    fn string_tokenisation() {
        assert_eq!(tokenise("\"a.b\'/c\"".into()).unwrap(), ["\"a.b'/c\""]);
        assert_eq!(tokenise("'a.b\"/c'".into()).unwrap(), ["'a.b\"/c'"]);

        assert_eq!(
            tokenise("p(\"a.b\'/c\").".into()).unwrap(),
            ["p", "(", "\"a.b'/c\"", ")", "."]
        );
    }

    #[test]
    fn string_escape_characters() {
        assert_eq!(tokenise("\"a\\\"b\"".into()).unwrap(), ["\"a\"b\""]);

        assert_eq!(tokenise("'a\\'b'".into()).unwrap(), ["'a'b'"]);

        assert_eq!(
            tokenise("\" \\n \\t \\\\ \"".into()).unwrap(),
            ["\" \n \t \\ \""]
        );
    }

    #[test]
    fn float_tokenisation() {
        assert_eq!(tokenise("123.123".into()).unwrap(), ["123.123"]);
        assert_eq!(
            tokenise("123.123.123".into()).unwrap(),
            ["123.123", ".", "123"]
        );
        assert_eq!(tokenise("123 . 123".into()).unwrap(), ["123", ".", "123"]);
    }

    #[test]
    fn negative_number_tokenisation() {
        assert_eq!(tokenise("-123 - 123".into()).unwrap(), ["-123", "-", "123"]);
        assert_eq!(tokenise("-123.123".into()).unwrap(), ["-123.123"]);
        assert_eq!(tokenise("- 123.123".into()).unwrap(), ["-", "123.123"]);
        assert_eq!(tokenise("-123 . 123".into()).unwrap(), ["-123", ".", "123"]);
    }

    #[test]
    fn list_tokenisation() {
        assert_eq!(
            tokenise("[123,abc, VAR]".into()).unwrap(),
            ["[", "123", ",", "abc", ",", "VAR", "]"]
        );
        assert_eq!(
            tokenise("[123,abc, VAR|T]".into()).unwrap(),
            ["[", "123", ",", "abc", ",", "VAR", "|", "T", "]"]
        );
        assert_eq!(
            tokenise("[123,abc, VAR | T]".into()).unwrap(),
            ["[", "123", ",", "abc", ",", "VAR", "|", "T", "]"]
        );
    }

    #[test]
    fn empty_list_token() {
        assert_eq!(tokenise("[]".into()).unwrap(), ["[]"]);
        assert_eq!(tokenise("[ ]".into()).unwrap(), ["[]"]);
        assert_eq!(tokenise("[\n]".into()).unwrap(), ["[]", "\n"]);
        assert_eq!(tokenise("[\n\n]".into()).unwrap(), ["[]", "\n", "\n"]);
        assert_eq!(tokenise("[\t]".into()).unwrap(), ["[]"]);
        assert_eq!(tokenise("[  \n\t\n ]".into()).unwrap(), ["[]", "\n", "\n"]);
    }

    #[test]
    fn empty_set_token() {
        assert_eq!(tokenise("{}".into()).unwrap(), ["{}"]);
        assert_eq!(tokenise("{ }".into()).unwrap(), ["{}"]);
        assert_eq!(tokenise("{\n}".into()).unwrap(), ["{}", "\n"]);
        assert_eq!(tokenise("{\n\n}".into()).unwrap(), ["{}", "\n", "\n"]);
        assert_eq!(tokenise("{\t}".into()).unwrap(), ["{}"]);
        assert_eq!(tokenise("{  \n\t\n }".into()).unwrap(), ["{}", "\n", "\n"]);
    }

    #[test]
    fn known_symbol_formation() {
        assert_eq!(tokenise(":-".into()).unwrap(), [":-"]);
        assert_eq!(tokenise(": -".into()).unwrap(), [":", "-"]);

        assert_eq!(tokenise("==".into()).unwrap(), ["=="]);
        assert_eq!(tokenise("= =".into()).unwrap(), ["=", "="]);

        assert_eq!(tokenise("=/=".into()).unwrap(), ["=/="]);
        assert_eq!(tokenise("= / =".into()).unwrap(), ["=", "/", "="]);

        assert_eq!(tokenise("/=".into()).unwrap(), ["/="]);
        assert_eq!(tokenise("/ =".into()).unwrap(), ["/", "="]);

        assert_eq!(tokenise("=:=".into()).unwrap(), ["=:="]);
        assert_eq!(tokenise("=  : =".into()).unwrap(), ["=", ":", "="]);

        assert_eq!(tokenise("**".into()).unwrap(), ["**"]);
        assert_eq!(tokenise("* *".into()).unwrap(), ["*", "*",]);

        assert_eq!(tokenise("<=".into()).unwrap(), ["<="]);
        assert_eq!(tokenise("< =".into()).unwrap(), ["<", "=",]);

        assert_eq!(tokenise(">=".into()).unwrap(), [">="]);
        assert_eq!(tokenise("> =".into()).unwrap(), [">", "=",]);

        assert_eq!(tokenise("<=".into()).unwrap(), ["<="]);
        assert_eq!(tokenise("< =".into()).unwrap(), ["<", "=",]);
    }

    #[test]
    fn tokenise_multiple_clauses() {
        let text = " p(a,[b,c|[\t]]).
        p(X,Y):- Q(X,Y), {Q}.
        :- [\"file/name\"], test.
        "
        .to_string();

        assert_eq!(
            tokenise(text).unwrap(),
            [
                "p",
                "(",
                "a",
                ",",
                "[",
                "b",
                ",",
                "c",
                "|",
                "[]",
                "]",
                ")",
                ".",
                "\n",
                "p",
                "(",
                "X",
                ",",
                "Y",
                ")",
                ":-",
                "Q",
                "(",
                "X",
                ",",
                "Y",
                ")",
                ",",
                "{",
                "Q",
                "}",
                ".",
                "\n",
                ":-",
                "[",
                "\"file/name\"",
                "]",
                ",",
                "test",
                ".",
                "\n"
            ]
        );
    }
}