Discard first line of multiline string if only whitespace

runtime/vm/scanner.cc

 // Copyright (c) 2012, the Dart project authors.  Please see the AUTHORS file
 // for details. All rights reserved. Use of this source code is governed by a
 // BSD-style license that can be found in the LICENSE file.
 
 #include "vm/scanner.h"
 
 #include "platform/assert.h"
 #include "vm/flags.h"
 #include "vm/object.h"
 #include "vm/object_store.h"
@@ skipping 404 matching lines @@
   SkipLine();
 }
 
 
 void Scanner::ScanLiteralString(bool is_raw) {
   ASSERT(!IsScanningString());
   ASSERT(c0_ == '"' || c0_ == '\'');
 
   // Entering string scanning mode.
   BeginStringLiteral(c0_);
-  string_is_multiline_ = (LookaheadChar(1) == c0_) &&
-      (LookaheadChar(2) == c0_);
+  ReadChar();
 
-  ReadChar();  // Skip opening delimiter.
-  if (string_is_multiline_) {
+  if ((c0_ == string_delimiter_) && (LookaheadChar(1) == string_delimiter_)) {
+    string_is_multiline_ = true;
     ReadChar();  // Skip two additional string delimiters.
     ReadChar();
-    if (c0_ == '\n') {
-      // Skip first character of multiline string if it is a newline.
-      ReadChar();
-    }
   }
-  ScanLiteralStringChars(is_raw);
+  ScanLiteralStringChars(is_raw, true);

[Ivan Posva, 2013/09/25 16:31:50]

Can't you pass string_is_multiline_ here? As this true only applies if we are
scanning a multiline string.

Correct would be "string_is_multiline_ ? true : false", but that seems a bit too
over the top.

[hausner, 2013/09/25 17:06:49]

Simplified as you suggested.

 }
 
 
 bool Scanner::ScanHexDigits(int digits, int32_t* value) {
   *value = 0;
   for (int i = 0; i < digits; ++i) {
     ReadChar();
     if (!IsHexDigit(c0_)) {
       ErrorMsg("too few hexadecimal digits");
       return false;
@@ skipping 42 matching lines @@
     }
   }
   if (is_valid &&
       ((Utf::IsOutOfRange(*code_point) ||
         (Utf16::IsSurrogate(*code_point))))) {
     ErrorMsg("invalid code point");
   }
 }
 
 
-void Scanner::ScanLiteralStringChars(bool is_raw) {
+void Scanner::ScanLiteralStringChars(bool is_raw, bool is_first_line) {
   GrowableArray<int32_t> string_chars(64);
 
   ASSERT(IsScanningString());
   // We are at the first character of a string literal piece. A string literal
   // can be broken up into multiple pieces by string interpolation.
   while (true) {
     if ((c0_ == '\0') || ((c0_ == '\n') && !string_is_multiline_)) {
       ErrorMsg("unterminated string literal");
       EndStringLiteral();
       return;
@@ skipping 77 matching lines @@
       int32_t ch1 = c0_;
       if (Utf16::IsLeadSurrogate(ch1)) {
         const int32_t ch2 = LookaheadChar(1);
         if (Utf16::IsTrailSurrogate(ch2)) {
           ch1 = Utf16::Decode(ch1, ch2);
           ReadChar();
         }
       }
       string_chars.Add(ch1);
     }
+    // The first line of a multi-line string is discarded if it only
+    // contains whitespace.
+    if (string_is_multiline_ &&
+        is_first_line &&
+        (string_chars.Last() == '\n')) {
+      bool whitespace_only = true;
+      // Last character is the newline, don't inspect it.
+      const intptr_t len = string_chars.length() - 1;
+      for (int i = 0; i < len; i++) {
+        int32_t ch = string_chars[i];
+        if ((ch != ' ') && (ch != '\t')) {

[Ivan Posva, 2013/09/25 16:31:50]

Please clarify the definition of white-space definition with Gilad.

[hausner, 2013/09/25 17:06:49]

Will do. I used the WHITESPACE production in the spec, although this production
is not used to formally define what whitespace means in a string literal.

+          // Non-whitespace character, keep the first line.
+          whitespace_only = false;
+          break;
+        }
+      }
+      if (whitespace_only) {
+        string_chars.Clear();  // Discard characters on first line.
+      }
+      is_first_line = false;
+    }
     ReadChar();
   }
 }
 
 
 void Scanner::Scan() {
   newline_seen_ = false;
 
   do {
     if (!IsScanningString()) {
@@ skipping 12 matching lines @@
           current_token_.kind = Token::kINTERPOL_VAR;
         } else if (c0_ == '{') {
           Recognize(Token::kINTERPOL_START);
           PushContext();
         } else {
           ErrorMsg("illegal character after $ in string interpolation");
           EndStringLiteral();
           break;
         }
       } else {
-        ScanLiteralStringChars(false);
+        ScanLiteralStringChars(false, false);
       }
       break;
     }
     switch (c0_) {
       case '\0':
         current_token_.kind = Token::kEOS;
         break;
 
       case '+':  // +  ++  +=
         Recognize(Token::kADD);
@@ skipping 289 matching lines @@
               "%c%#" Px "", kPrivateKeySeparator, key_value);
   const String& result = String::Handle(String::New(private_key, Heap::kOld));
   return result.raw();
 }
 
 
 void Scanner::InitOnce() {
 }
 
 }  // namespace dart