organizize

crates/libsyntax2/src/lexer/classes.rs

@@ -0,0 +1,26 @@
+use unicode_xid::UnicodeXID;
+
+pub fn is_ident_start(c: char) -> bool {
+    (c >= 'a' && c <= 'z')
+        || (c >= 'A' && c <= 'Z')
+        || c == '_'
+        || (c > '\x7f' && UnicodeXID::is_xid_start(c))
+}
+
+pub fn is_ident_continue(c: char) -> bool {
+    (c >= 'a' && c <= 'z')
+        || (c >= 'A' && c <= 'Z')
+        || (c >= '0' && c <= '9')
+        || c == '_'
+        || (c > '\x7f' && UnicodeXID::is_xid_continue(c))
+}
+
+pub fn is_whitespace(c: char) -> bool {
+    //FIXME: use is_pattern_whitespace
+    //https://github.com/behnam/rust-unic/issues/192
+    c.is_whitespace()
+}
+
+pub fn is_dec_digit(c: char) -> bool {
+    '0' <= c && c <= '9'
+}

crates/libsyntax2/src/lexer/comments.rs

@@ -0,0 +1,57 @@
+use lexer::ptr::Ptr;
+
+use SyntaxKind::{self, *};
+
+pub(crate) fn scan_shebang(ptr: &mut Ptr) -> bool {
+    if ptr.next_is('!') && ptr.nnext_is('/') {
+        ptr.bump();
+        ptr.bump();
+        bump_until_eol(ptr);
+        true
+    } else {
+        false
+    }
+}
+
+fn scan_block_comment(ptr: &mut Ptr) -> Option<SyntaxKind> {
+    if ptr.next_is('*') {
+        ptr.bump();
+        let mut depth: u32 = 1;
+        while depth > 0 {
+            if ptr.next_is('*') && ptr.nnext_is('/') {
+                depth -= 1;
+                ptr.bump();
+                ptr.bump();
+            } else if ptr.next_is('/') && ptr.nnext_is('*') {
+                depth += 1;
+                ptr.bump();
+                ptr.bump();
+            } else if ptr.bump().is_none() {
+                break;
+            }
+        }
+        Some(COMMENT)
+    } else {
+        None
+    }
+}
+
+pub(crate) fn scan_comment(ptr: &mut Ptr) -> Option<SyntaxKind> {
+    if ptr.next_is('/') {
+        bump_until_eol(ptr);
+        Some(COMMENT)
+    } else {
+        scan_block_comment(ptr)
+    }
+}
+
+fn bump_until_eol(ptr: &mut Ptr) {
+    loop {
+        if ptr.next_is('\n') || ptr.next_is('\r') && ptr.nnext_is('\n') {
+            return;
+        }
+        if ptr.bump().is_none() {
+            break;
+        }
+    }
+}

crates/libsyntax2/src/lexer/mod.rs

@@ -0,0 +1,209 @@
+mod classes;
+mod comments;
+mod numbers;
+mod ptr;
+mod strings;
+
+use {
+    SyntaxKind::{self, *},
+    TextUnit,
+};
+
+use self::{
+    classes::*,
+    comments::{scan_comment, scan_shebang},
+    numbers::scan_number,
+    ptr::Ptr,
+    strings::{
+        is_string_literal_start, scan_byte_char_or_string, scan_char, scan_raw_string, scan_string,
+    },
+};
+
+/// A token of Rust source.
+#[derive(Debug, Clone, Copy, PartialEq, Eq, PartialOrd, Ord, Hash)]
+pub struct Token {
+    /// The kind of token.
+    pub kind: SyntaxKind,
+    /// The length of the token.
+    pub len: TextUnit,
+}
+
+/// Break a string up into its component tokens
+pub fn tokenize(text: &str) -> Vec<Token> {
+    let mut text = text;
+    let mut acc = Vec::new();
+    while !text.is_empty() {
+        let token = next_token(text);
+        acc.push(token);
+        let len: u32 = token.len.into();
+        text = &text[len as usize..];
+    }
+    acc
+}
+
+/// Get the next token from a string
+pub fn next_token(text: &str) -> Token {
+    assert!(!text.is_empty());
+    let mut ptr = Ptr::new(text);
+    let c = ptr.bump().unwrap();
+    let kind = next_token_inner(c, &mut ptr);
+    let len = ptr.into_len();
+    Token { kind, len }
+}
+
+fn next_token_inner(c: char, ptr: &mut Ptr) -> SyntaxKind {
+    if is_whitespace(c) {
+        ptr.bump_while(is_whitespace);
+        return WHITESPACE;
+    }
+
+    match c {
+        '#' => if scan_shebang(ptr) {
+            return SHEBANG;
+        },
+        '/' => if let Some(kind) = scan_comment(ptr) {
+            return kind;
+        },
+        _ => (),
+    }
+
+    let ident_start = is_ident_start(c) && !is_string_literal_start(c, ptr.next(), ptr.nnext());
+    if ident_start {
+        return scan_ident(c, ptr);
+    }
+
+    if is_dec_digit(c) {
+        let kind = scan_number(c, ptr);
+        scan_literal_suffix(ptr);
+        return kind;
+    }
+
+    // One-byte tokens.
+    if let Some(kind) = SyntaxKind::from_char(c) {
+        return kind;
+    }
+
+    match c {
+        // Multi-byte tokens.
+        '.' => {
+            return match (ptr.next(), ptr.nnext()) {
+                (Some('.'), Some('.')) => {
+                    ptr.bump();
+                    ptr.bump();
+                    DOTDOTDOT
+                }
+                (Some('.'), Some('=')) => {
+                    ptr.bump();
+                    ptr.bump();
+                    DOTDOTEQ
+                }
+                (Some('.'), _) => {
+                    ptr.bump();
+                    DOTDOT
+                }
+                _ => DOT,
+            };
+        }
+        ':' => {
+            return match ptr.next() {
+                Some(':') => {
+                    ptr.bump();
+                    COLONCOLON
+                }
+                _ => COLON,
+            };
+        }
+        '=' => {
+            return match ptr.next() {
+                Some('=') => {
+                    ptr.bump();
+                    EQEQ
+                }
+                Some('>') => {
+                    ptr.bump();
+                    FAT_ARROW
+                }
+                _ => EQ,
+            };
+        }
+        '!' => {
+            return match ptr.next() {
+                Some('=') => {
+                    ptr.bump();
+                    NEQ
+                }
+                _ => EXCL,
+            };
+        }
+        '-' => {
+            return if ptr.next_is('>') {
+                ptr.bump();
+                THIN_ARROW
+            } else {
+                MINUS
+            };
+        }
+
+        // If the character is an ident start not followed by another single
+        // quote, then this is a lifetime name:
+        '\'' => {
+            return if ptr.next_is_p(is_ident_start) && !ptr.nnext_is('\'') {
+                ptr.bump();
+                while ptr.next_is_p(is_ident_continue) {
+                    ptr.bump();
+                }
+                // lifetimes shouldn't end with a single quote
+                // if we find one, then this is an invalid character literal
+                if ptr.next_is('\'') {
+                    ptr.bump();
+                    return CHAR; // TODO: error reporting
+                }
+                LIFETIME
+            } else {
+                scan_char(ptr);
+                scan_literal_suffix(ptr);
+                CHAR
+            };
+        }
+        'b' => {
+            let kind = scan_byte_char_or_string(ptr);
+            scan_literal_suffix(ptr);
+            return kind;
+        }
+        '"' => {
+            scan_string(ptr);
+            scan_literal_suffix(ptr);
+            return STRING;
+        }
+        'r' => {
+            scan_raw_string(ptr);
+            scan_literal_suffix(ptr);
+            return RAW_STRING;
+        }
+        _ => (),
+    }
+    ERROR
+}
+
+fn scan_ident(c: char, ptr: &mut Ptr) -> SyntaxKind {
+    let is_single_letter = match ptr.next() {
+        None => true,
+        Some(c) if !is_ident_continue(c) => true,
+        _ => false,
+    };
+    if is_single_letter {
+        return if c == '_' { UNDERSCORE } else { IDENT };
+    }
+    ptr.bump_while(is_ident_continue);
+    if let Some(kind) = SyntaxKind::from_keyword(ptr.current_token_text()) {
+        return kind;
+    }
+    IDENT
+}
+
+fn scan_literal_suffix(ptr: &mut Ptr) {
+    if ptr.next_is_p(is_ident_start) {
+        ptr.bump();
+    }
+    ptr.bump_while(is_ident_continue);
+}

crates/libsyntax2/src/lexer/numbers.rs

@@ -0,0 +1,67 @@
+use lexer::classes::*;
+use lexer::ptr::Ptr;
+
+use SyntaxKind::{self, *};
+
+pub(crate) fn scan_number(c: char, ptr: &mut Ptr) -> SyntaxKind {
+    if c == '0' {
+        match ptr.next().unwrap_or('\0') {
+            'b' | 'o' => {
+                ptr.bump();
+                scan_digits(ptr, false);
+            }
+            'x' => {
+                ptr.bump();
+                scan_digits(ptr, true);
+            }
+            '0'...'9' | '_' | '.' | 'e' | 'E' => {
+                scan_digits(ptr, true);
+            }
+            _ => return INT_NUMBER,
+        }
+    } else {
+        scan_digits(ptr, false);
+    }
+
+    // might be a float, but don't be greedy if this is actually an
+    // integer literal followed by field/method access or a range pattern
+    // (`0..2` and `12.foo()`)
+    if ptr.next_is('.') && !(ptr.nnext_is('.') || ptr.nnext_is_p(is_ident_start)) {
+        // might have stuff after the ., and if it does, it needs to start
+        // with a number
+        ptr.bump();
+        scan_digits(ptr, false);
+        scan_float_exponent(ptr);
+        return FLOAT_NUMBER;
+    }
+    // it might be a float if it has an exponent
+    if ptr.next_is('e') || ptr.next_is('E') {
+        scan_float_exponent(ptr);
+        return FLOAT_NUMBER;
+    }
+    INT_NUMBER
+}
+
+fn scan_digits(ptr: &mut Ptr, allow_hex: bool) {
+    while let Some(c) = ptr.next() {
+        match c {
+            '_' | '0'...'9' => {
+                ptr.bump();
+            }
+            'a'...'f' | 'A'...'F' if allow_hex => {
+                ptr.bump();
+            }
+            _ => return,
+        }
+    }
+}
+
+fn scan_float_exponent(ptr: &mut Ptr) {
+    if ptr.next_is('e') || ptr.next_is('E') {
+        ptr.bump();
+        if ptr.next_is('-') || ptr.next_is('+') {
+            ptr.bump();
+        }
+        scan_digits(ptr, false);
+    }
+}

crates/libsyntax2/src/lexer/ptr.rs

@@ -0,0 +1,74 @@
+use TextUnit;
+
+use std::str::Chars;
+
+pub(crate) struct Ptr<'s> {
+    text: &'s str,
+    len: TextUnit,
+}
+
+impl<'s> Ptr<'s> {
+    pub fn new(text: &'s str) -> Ptr<'s> {
+        Ptr {
+            text,
+            len: 0.into(),
+        }
+    }
+
+    pub fn into_len(self) -> TextUnit {
+        self.len
+    }
+
+    pub fn next(&self) -> Option<char> {
+        self.chars().next()
+    }
+
+    pub fn nnext(&self) -> Option<char> {
+        let mut chars = self.chars();
+        chars.next()?;
+        chars.next()
+    }
+
+    pub fn next_is(&self, c: char) -> bool {
+        self.next() == Some(c)
+    }
+
+    pub fn nnext_is(&self, c: char) -> bool {
+        self.nnext() == Some(c)
+    }
+
+    pub fn next_is_p<P: Fn(char) -> bool>(&self, p: P) -> bool {
+        self.next().map(p) == Some(true)
+    }
+
+    pub fn nnext_is_p<P: Fn(char) -> bool>(&self, p: P) -> bool {
+        self.nnext().map(p) == Some(true)
+    }
+
+    pub fn bump(&mut self) -> Option<char> {
+        let ch = self.chars().next()?;
+        self.len += TextUnit::of_char(ch);
+        Some(ch)
+    }
+
+    pub fn bump_while<F: Fn(char) -> bool>(&mut self, pred: F) {
+        loop {
+            match self.next() {
+                Some(c) if pred(c) => {
+                    self.bump();
+                }
+                _ => return,
+            }
+        }
+    }
+
+    pub fn current_token_text(&self) -> &str {
+        let len: u32 = self.len.into();
+        &self.text[..len as usize]
+    }
+
+    fn chars(&self) -> Chars {
+        let len: u32 = self.len.into();
+        self.text[len as usize..].chars()
+    }
+}

crates/libsyntax2/src/lexer/strings.rs

@@ -0,0 +1,106 @@
+use SyntaxKind::{self, *};
+
+use lexer::ptr::Ptr;
+
+pub(crate) fn is_string_literal_start(c: char, c1: Option<char>, c2: Option<char>) -> bool {
+    match (c, c1, c2) {
+        ('r', Some('"'), _)
+        | ('r', Some('#'), _)
+        | ('b', Some('"'), _)
+        | ('b', Some('\''), _)
+        | ('b', Some('r'), Some('"'))
+        | ('b', Some('r'), Some('#')) => true,
+        _ => false,
+    }
+}
+
+pub(crate) fn scan_char(ptr: &mut Ptr) {
+    if ptr.bump().is_none() {
+        return; // TODO: error reporting is upper in the stack
+    }
+    scan_char_or_byte(ptr);
+    if !ptr.next_is('\'') {
+        return; // TODO: error reporting
+    }
+    ptr.bump();
+}
+
+pub(crate) fn scan_byte_char_or_string(ptr: &mut Ptr) -> SyntaxKind {
+    // unwrapping and not-exhaustive match are ok
+    // because of string_literal_start
+    let c = ptr.bump().unwrap();
+    match c {
+        '\'' => {
+            scan_byte(ptr);
+            BYTE
+        }
+        '"' => {
+            scan_byte_string(ptr);
+            BYTE_STRING
+        }
+        'r' => {
+            scan_raw_byte_string(ptr);
+            RAW_BYTE_STRING
+        }
+        _ => unreachable!(),
+    }
+}
+
+pub(crate) fn scan_string(ptr: &mut Ptr) {
+    while let Some(c) = ptr.bump() {
+        if c == '"' {
+            return;
+        }
+    }
+}
+
+pub(crate) fn scan_raw_string(ptr: &mut Ptr) {
+    if !ptr.next_is('"') {
+        return;
+    }
+    ptr.bump();
+
+    while let Some(c) = ptr.bump() {
+        if c == '"' {
+            return;
+        }
+    }
+}
+
+fn scan_byte(ptr: &mut Ptr) {
+    if ptr.next_is('\'') {
+        ptr.bump();
+        return;
+    }
+    ptr.bump();
+    if ptr.next_is('\'') {
+        ptr.bump();
+        return;
+    }
+}
+
+fn scan_byte_string(ptr: &mut Ptr) {
+    while let Some(c) = ptr.bump() {
+        if c == '"' {
+            return;
+        }
+    }
+}
+
+fn scan_raw_byte_string(ptr: &mut Ptr) {
+    if !ptr.next_is('"') {
+        return;
+    }
+    ptr.bump();
+
+    while let Some(c) = ptr.bump() {
+        if c == '"' {
+            return;
+        }
+    }
+}
+
+fn scan_char_or_byte(ptr: &mut Ptr) {
+    //FIXME: deal with escape sequencies
+    ptr.bump();
+}