Merge #188

188: Introduce `SyntaxErrorKind` and `TextRange` to `SyntaxError` r=matklad a=aochagavia



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

1 files changed, 61 insertions, 23 deletions

diff --git a/crates/ra_syntax/src/validation.rs b/crates/ra_syntax/src/validation.rs
index 03d98eff4..009f5052f 100644
--- a/crates/ra_syntax/src/validation.rs
+++ b/crates/ra_syntax/src/validation.rs
@@ -1,40 +1,78 @@
 use crate::{
+    algo::visit::{visitor_ctx, VisitorCtx},
     ast::{self, AstNode},
     File,
-    string_lexing,
+    string_lexing::{self, CharComponentKind},
     yellow::{
         SyntaxError,
+        SyntaxErrorKind::*,
     },
 };
 
 pub(crate) fn validate(file: &File) -> Vec<SyntaxError> {
     let mut errors = Vec::new();
-    for d in file.root.borrowed().descendants() {
-        if let Some(c) = ast::Char::cast(d) {
-            let components = &mut string_lexing::parse_char_literal(c.text());
-            let len = components.count();
+    for node in file.root.borrowed().descendants() {
+        let _ = visitor_ctx(&mut errors)
+            .visit::<ast::Char, _>(validate_char)
+            .accept(node);
+    }
+    errors
+}
 
-            if !components.has_closing_quote {
-                errors.push(SyntaxError {
-                    msg: "Unclosed char literal".to_string(),
-                    offset: d.range().start(),
-                });
-            }
+fn validate_char(node: ast::Char, errors: &mut Vec<SyntaxError>) {
+    let mut components = string_lexing::parse_char_literal(node.text());
+    let mut len = 0;
+    for component in &mut components {
+        len += 1;
 
-            if len == 0 {
-                errors.push(SyntaxError {
-                    msg: "Empty char literal".to_string(),
-                    offset: d.range().start(),
-                });
+        // Validate escapes
+        let text = &node.text()[component.range];
+        let range = component.range + node.syntax().range().start();
+        use self::CharComponentKind::*;
+        match component.kind {
+            AsciiEscape => {
+                if text.len() == 1 {
+                    // Escape sequence consists only of leading `\`
+                    errors.push(SyntaxError::new(EmptyAsciiEscape, range));
+                } else {
+                    let escape_code = text.chars().skip(1).next().unwrap();
+                    if !is_ascii_escape(escape_code) {
+                        errors.push(SyntaxError::new(InvalidAsciiEscape, range));
+                    }
+                }
             }
-
-            if len > 1 {
-                errors.push(SyntaxError {
-                    msg: "Character literal should be only one character long".to_string(),
-                    offset: d.range().start(),
-                });
+            AsciiCodeEscape => {
+                // TODO:
+                // * First digit is octal
+                // * Second digit is hex
+            }
+            UnicodeEscape => {
+                // TODO:
+                // * Only hex digits or underscores allowed
+                // * Max 6 chars
+                // * Within allowed range (must be at most 10FFFF)
             }
+            // Code points are always valid
+            CodePoint => (),
         }
     }
-    errors
+
+    if !components.has_closing_quote {
+        errors.push(SyntaxError::new(UnclosedChar, node.syntax().range()));
+    }
+
+    if len == 0 {
+        errors.push(SyntaxError::new(EmptyChar, node.syntax().range()));
+    }
+
+    if len > 1 {
+        errors.push(SyntaxError::new(LongChar, node.syntax().range()));
+    }
+}
+
+fn is_ascii_escape(code: char) -> bool {
+    match code {
+        '\'' | '"' | 'n' | 'r' | 't' | '0' => true,
+        _ => false,
+    }
 }