Merge #207

207: Finish implementing char validation r=aochagavia a=aochagavia

The only thing missing right now are good integration tests (and maybe more descriptive error messages)

Co-authored-by: Adolfo Ochagavía <[email protected]>

Author: bors[bot]

Cargo.lock

@@ -8,6 +8,7 @@ description = "Comment and whitespace preserving parser for the Rust langauge"
 repository = "https://github.com/rust-analyzer/rust-analyzer"
 
 [dependencies]
+arrayvec = "0.4.7"
 unicode-xid = "0.1.0"
 itertools = "0.7.8"
 drop_bomb = "0.1.4"

crates/ra_syntax/Cargo.toml

@@ -30,8 +30,7 @@ impl<'s> Ptr<'s> {
     /// Gets the nth character from the current.
     /// For example, 0 will return the current token, 1 will return the next, etc.
     pub fn nth(&self, n: u32) -> Option<char> {
-        let mut chars = self.chars().peekable();
-        chars.by_ref().nth(n as usize)
+        self.chars().nth(n as usize)
     }
 
     /// Checks whether the current character is `c`.

crates/ra_syntax/src/lexer/ptr.rs

@@ -20,6 +20,7 @@
 #![allow(missing_docs)]
 //#![warn(unreachable_pub)] // rust-lang/rust#47816
 
+extern crate arrayvec;
 extern crate drop_bomb;
 extern crate itertools;
 extern crate parking_lot;

crates/ra_syntax/src/lib.rs

@@ -219,7 +219,7 @@ mod tests {
 
     #[test]
     fn test_unicode_escapes() {
-        let unicode_escapes = &[r"{DEAD}", "{BEEF}", "{FF}", ""];
+        let unicode_escapes = &[r"{DEAD}", "{BEEF}", "{FF}", "{}", ""];
         for escape in unicode_escapes {
             let escape_sequence = format!(r"'\u{}'", escape);
             let component = closed_char_component(&escape_sequence);

crates/ra_syntax/src/string_lexing/mod.rs

@@ -1,5 +1,6 @@
 use crate::{File, SyntaxKind, SyntaxNodeRef, WalkEvent};
 use std::fmt::Write;
+use std::str;
 
 /// Parse a file and create a string representation of the resulting parse tree.
 pub fn dump_tree(syntax: SyntaxNodeRef) -> String {

crates/ra_syntax/src/utils.rs

@@ -1,3 +1,7 @@
+use std::u32;
+
+use arrayvec::ArrayString;
+
 use crate::{
     algo::visit::{visitor_ctx, VisitorCtx},
     ast::{self, AstNode},
@@ -42,18 +46,90 @@ fn validate_char(node: ast::Char, errors: &mut Vec<SyntaxError>) {
                 }
             }
             AsciiCodeEscape => {
-                // TODO:
-                // * First digit is octal
-                // * Second digit is hex
+                // An AsciiCodeEscape has 4 chars, example: `\xDD`
+                if text.len() < 4 {
+                    errors.push(SyntaxError::new(TooShortAsciiCodeEscape, range));
+                } else {
+                    assert!(
+                        text.chars().count() == 4,
+                        "AsciiCodeEscape cannot be longer than 4 chars"
+                    );
+
+                    match u8::from_str_radix(&text[2..], 16) {
+                        Ok(code) if code < 128 => { /* Escape code is valid */ }
+                        Ok(_) => errors.push(SyntaxError::new(AsciiCodeEscapeOutOfRange, range)),
+                        Err(_) => errors.push(SyntaxError::new(MalformedAsciiCodeEscape, range)),
+                    }
+                }
             }
             UnicodeEscape => {
-                // TODO:
-                // * Only hex digits or underscores allowed
-                // * Max 6 chars
-                // * Within allowed range (must be at most 10FFFF)
+                assert!(&text[..2] == "\\u", "UnicodeEscape always starts with \\u");
+
+                if text.len() == 2 {
+                    // No starting `{`
+                    errors.push(SyntaxError::new(MalformedUnicodeEscape, range));
+                    return;
+                }
+
+                if text.len() == 3 {
+                    // Only starting `{`
+                    errors.push(SyntaxError::new(UnclosedUnicodeEscape, range));
+                    return;
+                }
+
+                let mut code = ArrayString::<[_; 6]>::new();
+                let mut closed = false;
+                for c in text[3..].chars() {
+                    assert!(!closed, "no characters after escape is closed");
+
+                    if c.is_digit(16) {
+                        if code.len() == 6 {
+                            errors.push(SyntaxError::new(OverlongUnicodeEscape, range));
+                            return;
+                        }
+
+                        code.push(c);
+                    } else if c == '_' {
+                        // Reject leading _
+                        if code.len() == 0 {
+                            errors.push(SyntaxError::new(MalformedUnicodeEscape, range));
+                            return;
+                        }
+                    } else if c == '}' {
+                        closed = true;
+                    } else {
+                        errors.push(SyntaxError::new(MalformedUnicodeEscape, range));
+                        return;
+                    }
+                }
+
+                if !closed {
+                    errors.push(SyntaxError::new(UnclosedUnicodeEscape, range))
+                }
+
+                if code.len() == 0 {
+                    errors.push(SyntaxError::new(EmptyUnicodeEcape, range));
+                    return;
+                }
+
+                match u32::from_str_radix(&code, 16) {
+                    Ok(code_u32) if code_u32 > 0x10FFFF => {
+                        errors.push(SyntaxError::new(UnicodeEscapeOutOfRange, range));
+                    }
+                    Ok(_) => {
+                        // Valid escape code
+                    }
+                    Err(_) => {
+                        errors.push(SyntaxError::new(MalformedUnicodeEscape, range));
+                    }
+                }
+            }
+            CodePoint => {
+                // These code points must always be escaped
+                if text == "\t" || text == "\r" {
+                    errors.push(SyntaxError::new(UnescapedCodepoint, range));
+                }
             }
-            // Code points are always valid
-            CodePoint => (),
         }
     }
 
@@ -72,7 +148,124 @@ fn validate_char(node: ast::Char, errors: &mut Vec<SyntaxError>) {
 
 fn is_ascii_escape(code: char) -> bool {
     match code {
-        '\'' | '"' | 'n' | 'r' | 't' | '0' => true,
+        '\\' | '\'' | '"' | 'n' | 'r' | 't' | '0' => true,
         _ => false,
     }
 }
+
+#[cfg(test)]
+mod test {
+    use crate::File;
+
+    fn build_file(literal: &str) -> File {
+        let src = format!("const C: char = '{}';", literal);
+        File::parse(&src)
+    }
+
+    fn assert_valid_char(literal: &str) {
+        let file = build_file(literal);
+        assert!(
+            file.errors().len() == 0,
+            "Errors for literal '{}': {:?}",
+            literal,
+            file.errors()
+        );
+    }
+
+    fn assert_invalid_char(literal: &str) {
+        let file = build_file(literal);
+        assert!(file.errors().len() > 0);
+    }
+
+    #[test]
+    fn test_ansi_codepoints() {
+        for byte in 0..=255u8 {
+            match byte {
+                b'\n' | b'\r' | b'\t' => assert_invalid_char(&(byte as char).to_string()),
+                b'\'' | b'\\' => { /* Ignore character close and backslash */ }
+                _ => assert_valid_char(&(byte as char).to_string()),
+            }
+        }
+    }
+
+    #[test]
+    fn test_unicode_codepoints() {
+        let valid = ["Ƒ", "バ", "メ", "﷽"];
+        for c in &valid {
+            assert_valid_char(c);
+        }
+    }
+
+    #[test]
+    fn test_unicode_multiple_codepoints() {
+        let invalid = ["नी", "👨‍👨‍"];
+        for c in &invalid {
+            assert_invalid_char(c);
+        }
+    }
+
+    #[test]
+    fn test_valid_ascii_escape() {
+        let valid = [
+            r"\'", "\"", "\\\\", "\\\"", r"\n", r"\r", r"\t", r"\0", "a", "b",
+        ];
+        for c in &valid {
+            assert_valid_char(c);
+        }
+    }
+
+    #[test]
+    fn test_invalid_ascii_escape() {
+        let invalid = [r"\a", r"\?", r"\"];
+        for c in &invalid {
+            assert_invalid_char(c);
+        }
+    }
+
+    #[test]
+    fn test_valid_ascii_code_escape() {
+        let valid = [r"\x00", r"\x7F", r"\x55"];
+        for c in &valid {
+            assert_valid_char(c);
+        }
+    }
+
+    #[test]
+    fn test_invalid_ascii_code_escape() {
+        let invalid = [r"\x", r"\x7", r"\xF0"];
+        for c in &invalid {
+            assert_invalid_char(c);
+        }
+    }
+
+    #[test]
+    fn test_valid_unicode_escape() {
+        let valid = [
+            r"\u{FF}",
+            r"\u{0}",
+            r"\u{F}",
+            r"\u{10FFFF}",
+            r"\u{1_0__FF___FF_____}",
+        ];
+        for c in &valid {
+            assert_valid_char(c);
+        }
+    }
+
+    #[test]
+    fn test_invalid_unicode_escape() {
+        let invalid = [
+            r"\u",
+            r"\u{}",
+            r"\u{",
+            r"\u{FF",
+            r"\u{FFFFFF}",
+            r"\u{_F}",
+            r"\u{00FFFFF}",
+            r"\u{110000}",
+        ];
+        for c in &invalid {
+            assert_invalid_char(c);
+        }
+    }
+}

crates/ra_syntax/src/validation.rs

@@ -34,6 +34,10 @@ impl SyntaxError {
         }
     }
 
+    pub fn kind(&self) -> SyntaxErrorKind {
+        self.kind.clone()
+    }
+
     pub fn location(&self) -> Location {
         self.location.clone()
     }
@@ -64,11 +68,20 @@ impl fmt::Display for SyntaxError {
 #[derive(Debug, Clone, PartialEq, Eq, Hash)]
 pub enum SyntaxErrorKind {
     ParseError(ParseError),
+    UnescapedCodepoint,
     EmptyChar,
     UnclosedChar,
     LongChar,
     EmptyAsciiEscape,
     InvalidAsciiEscape,
+    TooShortAsciiCodeEscape,
+    AsciiCodeEscapeOutOfRange,
+    MalformedAsciiCodeEscape,
+    UnclosedUnicodeEscape,
+    MalformedUnicodeEscape,
+    EmptyUnicodeEcape,
+    OverlongUnicodeEscape,
+    UnicodeEscapeOutOfRange,
 }
 
 #[derive(Debug, Clone, PartialEq, Eq, Hash)]
@@ -78,11 +91,24 @@ impl fmt::Display for SyntaxErrorKind {
     fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
         use self::SyntaxErrorKind::*;
         match self {
+            UnescapedCodepoint => write!(f, "This codepoint should always be escaped"),
             EmptyAsciiEscape => write!(f, "Empty escape sequence"),
             InvalidAsciiEscape => write!(f, "Invalid escape sequence"),
             EmptyChar => write!(f, "Empty char literal"),
             UnclosedChar => write!(f, "Unclosed char literal"),
             LongChar => write!(f, "Char literal should be one character long"),
+            TooShortAsciiCodeEscape => write!(f, "Escape sequence should have two digits"),
+            AsciiCodeEscapeOutOfRange => {
+                write!(f, "Escape sequence should be between \\x00 and \\x7F")
+            }
+            MalformedAsciiCodeEscape => write!(f, "Escape sequence should be a hexadecimal number"),
+            UnclosedUnicodeEscape => write!(f, "Missing `}}`"),
+            MalformedUnicodeEscape => write!(f, "Malformed unicode escape sequence"),
+            EmptyUnicodeEcape => write!(f, "Empty unicode escape sequence"),
+            OverlongUnicodeEscape => {
+                write!(f, "Unicode escape sequence should have at most 6 digits")
+            }
+            UnicodeEscapeOutOfRange => write!(f, "Unicode escape code should be at most 0x10FFFF"),
             ParseError(msg) => write!(f, "{}", msg.0),
         }
     }