better char lexing

3 files changed, 27 insertions, 13 deletions

diff --git a/crates/libsyntax2/src/lexer/strings.rs b/crates/libsyntax2/src/lexer/strings.rs
index fbae767e5..795ea97b7 100644
--- a/crates/libsyntax2/src/lexer/strings.rs
+++ b/crates/libsyntax2/src/lexer/strings.rs
@@ -15,14 +15,23 @@ pub(crate) fn is_string_literal_start(c: char, c1: Option<char>, c2: Option<char
 }
 
 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
+    loop {
+        if ptr.next_is('\\') {
+            ptr.bump();
+            if ptr.next_is('\\') || ptr.next_is('\'') {
+                ptr.bump();
+            }
+            continue;
+        }
+        if ptr.next_is('\'') {
+            ptr.bump();
+            return;
+        }
+        if ptr.next_is('\n') {
+            break;
+        }
+        ptr.bump();
     }
-    ptr.bump();
 }
 
 pub(crate) fn scan_byte_char_or_string(ptr: &mut Ptr) -> SyntaxKind {
@@ -111,8 +120,3 @@ fn scan_raw_byte_string(ptr: &mut Ptr) {
         }
     }
 }
-
-fn scan_char_or_byte(ptr: &mut Ptr) {
-    //FIXME: deal with escape sequencies
-    ptr.bump();
-}
diff --git a/crates/libsyntax2/tests/data/lexer/0006_chars.rs b/crates/libsyntax2/tests/data/lexer/0006_chars.rs
index 03598d908..454ee0a5f 100644
--- a/crates/libsyntax2/tests/data/lexer/0006_chars.rs
+++ b/crates/libsyntax2/tests/data/lexer/0006_chars.rs
@@ -1 +1 @@
-'x' ' ' '0'
+'x' ' ' '0' 'hello' '\x7f' '\n' '\\' '\''
diff --git a/crates/libsyntax2/tests/data/lexer/0006_chars.txt b/crates/libsyntax2/tests/data/lexer/0006_chars.txt
index ecaf22355..950954fbc 100644
--- a/crates/libsyntax2/tests/data/lexer/0006_chars.txt
+++ b/crates/libsyntax2/tests/data/lexer/0006_chars.txt
@@ -3,4 +3,14 @@ WHITESPACE 1 " "
 CHAR 3 "\' \'"
 WHITESPACE 1 " "
 CHAR 3 "\'0\'"
+WHITESPACE 1 " "
+CHAR 7 "\'hello\'"
+WHITESPACE 1 " "
+CHAR 6 "\'\\x7f\'"
+WHITESPACE 1 " "
+CHAR 4 "\'\\n\'"
+WHITESPACE 1 " "
+CHAR 4 "\'\\\\\'"
+WHITESPACE 1 " "
+CHAR 4 "\'\\\'\'"
 WHITESPACE 1 "\n"