feat: Added a few escape sequences

alinalihassan · May 6, 2022 · 1ddba5d · 1ddba5d
1 parent caa3c25
commit 1ddba5d
Show file tree

Hide file tree

Showing 3 changed files with 31 additions and 6 deletions.
diff --git a/src/liblesma/Frontend/Lexer.cpp b/src/liblesma/Frontend/Lexer.cpp
@@ -280,21 +280,47 @@ char Lexer::Peek(int offset) {
 }
 
 Token Lexer::AddStringToken() {
+    std::string string;
+
     while (Peek() != '"' && !IsAtEnd()) {
+        // Should we allow newlines in strings? Probably not
         if (Peek() == '\n') {
             loc.Line++;
             loc.Col = 1;
         }
-        Advance();
+        // If it's not an escape sequence, proceed as usual
+        if (Peek() != '\\') {
+            string.push_back(Advance());
+            continue;
+        }
+
+        switch(Peek(1)){
+            case 'n': string.push_back('\n');  break;
+            case 'r': string.push_back('\r');  break;
+            case 't': string.push_back('\t');  break;
+            case 'b': string.push_back('\b');  break;
+            case '0': string.push_back('\0');  break;
+            case '"': string.push_back('"');   break;
+            case 'e': string.push_back(0x1B);  break;
+            case '\'': string.push_back('\''); break;
+            case '\\': string.push_back('\\'); break;
+            default:
+                Error("Unknown escape sequence.");
+        }
+
+        // Skip the backslash and the escape sequence.
+        Advance(); Advance();
     }
 
     if (IsAtEnd())
         Error("Unterminated string.");
 
-    // The closing ".
+    // Skip the closing ".
     Advance();
 
-    return AddToken(TokenType::STRING);
+    auto ret = Token(TokenType::STRING, string, Span{begin_loc, loc});
+    ResetTokenBeg();
+    return ret;
 }
 Token Lexer::AddNumToken() {
     while (IsDigit(Peek())) Advance();

diff --git a/src/liblesma/Frontend/Parser.cpp b/src/liblesma/Frontend/Parser.cpp
@@ -95,8 +95,7 @@ Expression *Parser::ParseTerm() {
         case TokenType::NIL: {
             auto token = Peek();
             Consume(token->type);
-            auto token_value = token->type == TokenType::STRING ? token->lexeme.substr(1, token->lexeme.size() - 2) : token->lexeme;
-            return new Literal(token->span, token_value, token->type);
+            return new Literal(token->span, token->lexeme, token->type);
         }
         case TokenType::IDENTIFIER: {
             if (CheckAny<TokenType::LEFT_PAREN>(1))

diff --git a/tests/lesma/print_str.les b/tests/lesma/print_str.les
@@ -2,6 +2,6 @@
 def extern exit(x: int)
 def extern printf(x: str)
 
-let x = "Hello World!"
+let x = "Hello World! 🎉\n"
 printf(x)
 exit(101)