cs, crystal, es6: Detect more unterminated strings.

Specifically, correctly handle invalid strings that nonetheless end with a '"' character, by using the full string-matching regexp to check that a string token is valid.
mrmikardo · May 15, 2019 · 320807c · 320807c
1 parent 4ad84d2
commit 320807c
Show file tree

Hide file tree

Showing 3 changed files with 10 additions and 12 deletions.
diff --git a/crystal/reader.cr b/crystal/reader.cr
@@ -81,11 +81,11 @@ class Reader
     when token == "true"    then true
     when token == "false"   then false
     when token == "nil"     then nil
-    when token[0] == '"'
-      parse_error "expected '\"', got EOF" if token[-1] != '"'
+    when token =~ /^"(?:\\.|[^\\"])*"$/
       token[1..-2].gsub(/\\(.)/, {"\\\"" => "\"",
                                   "\\n"  => "\n",
                                   "\\\\" => "\\"})
+    when token[0] == '"' then parse_error "expected '\"', got EOF"
     when token[0] == ':' then "\u029e#{token[1..-1]}"
     else                      Mal::Symbol.new token
     end

diff --git a/cs/reader.cs b/cs/reader.cs
@@ -53,7 +53,7 @@ public static List<string> tokenize(string str) {
 
         public static MalVal read_atom(Reader rdr) {
             string token = rdr.next();
-            string pattern = @"(^-?[0-9]+$)|(^-?[0-9][0-9.]*$)|(^nil$)|(^true$)|(^false$)|^("".*)|:(.*)|(^[^""]*$)";
+            string pattern = @"(^-?[0-9]+$)|(^-?[0-9][0-9.]*$)|(^nil$)|(^true$)|(^false$)|(^""(?:[\\].|[^\\""])*""$)|(^"".*$)|:(.*)|(^[^""]*$)";
             Regex regex = new Regex(pattern);
             Match match = regex.Match(token);
             //Console.WriteLine("token: ^" + token + "$");
@@ -70,19 +70,18 @@ public static MalVal read_atom(Reader rdr) {
                 return Mal.types.False;
             } else if (match.Groups[6].Value != String.Empty) {
                 string str = match.Groups[6].Value;
-                if (str[str.Length-1] != '"') {
-                    throw new ParseError("expected '\"', got EOF");
-                }
                 str = str.Substring(1, str.Length-2)
                     .Replace("\\\\",   "\u029e")
                     .Replace("\\\"",   "\"")
                     .Replace("\\n",    "\n")
                     .Replace("\u029e", "\\");
                 return new Mal.types.MalString(str);
             } else if (match.Groups[7].Value != String.Empty) {
-                return new Mal.types.MalString("\u029e" + match.Groups[7].Value);
+                throw new ParseError("expected '\"', got EOF");
             } else if (match.Groups[8].Value != String.Empty) {
-                return new Mal.types.MalSymbol(match.Groups[8].Value);
+                return new Mal.types.MalString("\u029e" + match.Groups[8].Value);
+            } else if (match.Groups[9].Value != String.Empty) {
+                return new Mal.types.MalSymbol(match.Groups[9].Value);
             } else {
                 throw new ParseError("unrecognized '" + match.Groups[0] + "'");
             }

diff --git a/es6/reader.mjs b/es6/reader.mjs
@@ -29,12 +29,11 @@ function read_atom (reader) {
         return parseInt(token,10)        // integer
     } else if (token.match(/^-?[0-9][0-9.]*$/)) {
         return parseFloat(token,10)     // float
-    } else if (token[0] === "\"") {
-        if (token.slice(-1) !== "\"") {
-            throw new Error("expected '\"', got EOF");
-        }
+    } else if (token.match(/^"(?:\\.|[^\\"])*"$/)) {
         return token.slice(1,token.length-1)
             .replace(/\\(.)/g, (_, c) => c === "n" ? "\n" : c)
+    } else if (token[0] === "\"") {
+        throw new Error("expected '\"', got EOF");
     } else if (token[0] === ":") {
         return _keyword(token.slice(1))
     } else if (token === "nil") {