Merge #861

861: Move parsing to a separate module r=matklad a=matklad



Co-authored-by: Aleksey Kladov <[email protected]>

5 files changed, 426 insertions, 0 deletions

diff --git a/crates/ra_syntax/src/parsing/lexer/classes.rs b/crates/ra_syntax/src/parsing/lexer/classes.rs
new file mode 100644
index 000000000..4235d2648
--- /dev/null
+++ b/crates/ra_syntax/src/parsing/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'
+}
diff --git a/crates/ra_syntax/src/parsing/lexer/comments.rs b/crates/ra_syntax/src/parsing/lexer/comments.rs
new file mode 100644
index 000000000..8bbbe659b
--- /dev/null
+++ b/crates/ra_syntax/src/parsing/lexer/comments.rs
@@ -0,0 +1,57 @@
+use crate::parsing::lexer::ptr::Ptr;
+
+use crate::SyntaxKind::{self, *};
+
+pub(crate) fn scan_shebang(ptr: &mut Ptr) -> bool {
+    if ptr.at_str("!/") {
+        ptr.bump();
+        ptr.bump();
+        bump_until_eol(ptr);
+        true
+    } else {
+        false
+    }
+}
+
+fn scan_block_comment(ptr: &mut Ptr) -> Option<SyntaxKind> {
+    if ptr.at('*') {
+        ptr.bump();
+        let mut depth: u32 = 1;
+        while depth > 0 {
+            if ptr.at_str("*/") {
+                depth -= 1;
+                ptr.bump();
+                ptr.bump();
+            } else if ptr.at_str("/*") {
+                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.at('/') {
+        bump_until_eol(ptr);
+        Some(COMMENT)
+    } else {
+        scan_block_comment(ptr)
+    }
+}
+
+fn bump_until_eol(ptr: &mut Ptr) {
+    loop {
+        if ptr.at('\n') || ptr.at_str("\r\n") {
+            return;
+        }
+        if ptr.bump().is_none() {
+            break;
+        }
+    }
+}
diff --git a/crates/ra_syntax/src/parsing/lexer/numbers.rs b/crates/ra_syntax/src/parsing/lexer/numbers.rs
new file mode 100644
index 000000000..7f6abe1d5
--- /dev/null
+++ b/crates/ra_syntax/src/parsing/lexer/numbers.rs
@@ -0,0 +1,69 @@
+use crate::parsing::lexer::{
+    ptr::Ptr,
+    classes::*,
+};
+
+use crate::SyntaxKind::{self, *};
+
+pub(crate) fn scan_number(c: char, ptr: &mut Ptr) -> SyntaxKind {
+    if c == '0' {
+        match ptr.current().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.at('.') && !(ptr.at_str("..") || ptr.nth_is_p(1, 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.at('e') || ptr.at('E') {
+        scan_float_exponent(ptr);
+        return FLOAT_NUMBER;
+    }
+    INT_NUMBER
+}
+
+fn scan_digits(ptr: &mut Ptr, allow_hex: bool) {
+    while let Some(c) = ptr.current() {
+        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.at('e') || ptr.at('E') {
+        ptr.bump();
+        if ptr.at('-') || ptr.at('+') {
+            ptr.bump();
+        }
+        scan_digits(ptr, false);
+    }
+}
diff --git a/crates/ra_syntax/src/parsing/lexer/ptr.rs b/crates/ra_syntax/src/parsing/lexer/ptr.rs
new file mode 100644
index 000000000..c341c4176
--- /dev/null
+++ b/crates/ra_syntax/src/parsing/lexer/ptr.rs
@@ -0,0 +1,162 @@
+use crate::TextUnit;
+
+use std::str::Chars;
+
+/// A simple view into the characters of a string.
+pub(crate) struct Ptr<'s> {
+    text: &'s str,
+    len: TextUnit,
+}
+
+impl<'s> Ptr<'s> {
+    /// Creates a new `Ptr` from a string.
+    pub fn new(text: &'s str) -> Ptr<'s> {
+        Ptr { text, len: 0.into() }
+    }
+
+    /// Gets the length of the remaining string.
+    pub fn into_len(self) -> TextUnit {
+        self.len
+    }
+
+    /// Gets the current character, if one exists.
+    pub fn current(&self) -> Option<char> {
+        self.chars().next()
+    }
+
+    /// Gets the nth character from the current.
+    /// For example, 0 will return the current character, 1 will return the next, etc.
+    pub fn nth(&self, n: u32) -> Option<char> {
+        self.chars().nth(n as usize)
+    }
+
+    /// Checks whether the current character is `c`.
+    pub fn at(&self, c: char) -> bool {
+        self.current() == Some(c)
+    }
+
+    /// Checks whether the next characters match `s`.
+    pub fn at_str(&self, s: &str) -> bool {
+        let chars = self.chars();
+        chars.as_str().starts_with(s)
+    }
+
+    /// Checks whether the current character satisfies the predicate `p`.
+    pub fn at_p<P: Fn(char) -> bool>(&self, p: P) -> bool {
+        self.current().map(p) == Some(true)
+    }
+
+    /// Checks whether the nth character satisfies the predicate `p`.
+    pub fn nth_is_p<P: Fn(char) -> bool>(&self, n: u32, p: P) -> bool {
+        self.nth(n).map(p) == Some(true)
+    }
+
+    /// Moves to the next character.
+    pub fn bump(&mut self) -> Option<char> {
+        let ch = self.chars().next()?;
+        self.len += TextUnit::of_char(ch);
+        Some(ch)
+    }
+
+    /// Moves to the next character as long as `pred` is satisfied.
+    pub fn bump_while<F: Fn(char) -> bool>(&mut self, pred: F) {
+        loop {
+            match self.current() {
+                Some(c) if pred(c) => {
+                    self.bump();
+                }
+                _ => return,
+            }
+        }
+    }
+
+    /// Returns the text up to the current point.
+    pub fn current_token_text(&self) -> &str {
+        let len: u32 = self.len.into();
+        &self.text[..len as usize]
+    }
+
+    /// Returns an iterator over the remaining characters.
+    fn chars(&self) -> Chars {
+        let len: u32 = self.len.into();
+        self.text[len as usize..].chars()
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_current() {
+        let ptr = Ptr::new("test");
+        assert_eq!(ptr.current(), Some('t'));
+    }
+
+    #[test]
+    fn test_nth() {
+        let ptr = Ptr::new("test");
+        assert_eq!(ptr.nth(0), Some('t'));
+        assert_eq!(ptr.nth(1), Some('e'));
+        assert_eq!(ptr.nth(2), Some('s'));
+        assert_eq!(ptr.nth(3), Some('t'));
+        assert_eq!(ptr.nth(4), None);
+    }
+
+    #[test]
+    fn test_at() {
+        let ptr = Ptr::new("test");
+        assert!(ptr.at('t'));
+        assert!(!ptr.at('a'));
+    }
+
+    #[test]
+    fn test_at_str() {
+        let ptr = Ptr::new("test");
+        assert!(ptr.at_str("t"));
+        assert!(ptr.at_str("te"));
+        assert!(ptr.at_str("test"));
+        assert!(!ptr.at_str("tests"));
+        assert!(!ptr.at_str("rust"));
+    }
+
+    #[test]
+    fn test_at_p() {
+        let ptr = Ptr::new("test");
+        assert!(ptr.at_p(|c| c == 't'));
+        assert!(!ptr.at_p(|c| c == 'e'));
+    }
+
+    #[test]
+    fn test_nth_is_p() {
+        let ptr = Ptr::new("test");
+        assert!(ptr.nth_is_p(0, |c| c == 't'));
+        assert!(!ptr.nth_is_p(1, |c| c == 't'));
+        assert!(ptr.nth_is_p(3, |c| c == 't'));
+        assert!(!ptr.nth_is_p(150, |c| c == 't'));
+    }
+
+    #[test]
+    fn test_bump() {
+        let mut ptr = Ptr::new("test");
+        assert_eq!(ptr.current(), Some('t'));
+        ptr.bump();
+        assert_eq!(ptr.current(), Some('e'));
+        ptr.bump();
+        assert_eq!(ptr.current(), Some('s'));
+        ptr.bump();
+        assert_eq!(ptr.current(), Some('t'));
+        ptr.bump();
+        assert_eq!(ptr.current(), None);
+        ptr.bump();
+        assert_eq!(ptr.current(), None);
+    }
+
+    #[test]
+    fn test_bump_while() {
+        let mut ptr = Ptr::new("test");
+        assert_eq!(ptr.current(), Some('t'));
+        ptr.bump_while(|c| c != 's');
+        assert_eq!(ptr.current(), Some('s'));
+    }
+}
diff --git a/crates/ra_syntax/src/parsing/lexer/strings.rs b/crates/ra_syntax/src/parsing/lexer/strings.rs
new file mode 100644
index 000000000..f74acff9e
--- /dev/null
+++ b/crates/ra_syntax/src/parsing/lexer/strings.rs
@@ -0,0 +1,112 @@
+use crate::{
+    parsing::lexer::ptr::Ptr,
+    SyntaxKind::{self, *},
+};
+
+pub(crate) fn is_string_literal_start(c: char, c1: Option<char>, c2: Option<char>) -> bool {
+    match (c, c1, c2) {
+        ('r', Some('"'), _)
+        | ('r', Some('#'), Some('"'))
+        | ('r', Some('#'), Some('#'))
+        | ('b', Some('"'), _)
+        | ('b', Some('\''), _)
+        | ('b', Some('r'), Some('"'))
+        | ('b', Some('r'), Some('#')) => true,
+        _ => false,
+    }
+}
+
+pub(crate) fn scan_char(ptr: &mut Ptr) {
+    while let Some(c) = ptr.current() {
+        match c {
+            '\\' => {
+                ptr.bump();
+                if ptr.at('\\') || ptr.at('\'') {
+                    ptr.bump();
+                }
+            }
+            '\'' => {
+                ptr.bump();
+                return;
+            }
+            '\n' => return,
+            _ => {
+                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_string(ptr);
+            RAW_BYTE_STRING
+        }
+        _ => unreachable!(),
+    }
+}
+
+pub(crate) fn scan_string(ptr: &mut Ptr) {
+    while let Some(c) = ptr.current() {
+        match c {
+            '\\' => {
+                ptr.bump();
+                if ptr.at('\\') || ptr.at('"') {
+                    ptr.bump();
+                }
+            }
+            '"' => {
+                ptr.bump();
+                return;
+            }
+            _ => {
+                ptr.bump();
+            }
+        }
+    }
+}
+
+pub(crate) fn scan_raw_string(ptr: &mut Ptr) {
+    let mut hashes = 0;
+    while ptr.at('#') {
+        hashes += 1;
+        ptr.bump();
+    }
+    if !ptr.at('"') {
+        return;
+    }
+    ptr.bump();
+
+    while let Some(c) = ptr.bump() {
+        if c == '"' {
+            let mut hashes_left = hashes;
+            while ptr.at('#') && hashes_left > 0 {
+                hashes_left -= 1;
+                ptr.bump();
+            }
+            if hashes_left == 0 {
+                return;
+            }
+        }
+    }
+}
+
+fn scan_byte(ptr: &mut Ptr) {
+    scan_char(ptr)
+}
+
+fn scan_byte_string(ptr: &mut Ptr) {
+    scan_string(ptr)
+}