Inserted string validation as separate task in compiler.

frog/leg/ssa/nodes.dart

 // Copyright (c) 2011, 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.
 
 interface HVisitor<R> {
   R visitAdd(HAdd node);
   R visitBailoutTarget(HBailoutTarget node);
   R visitBitAnd(HBitAnd node);
   R visitBitNot(HBitNot node);
   R visitBitOr(HBitOr node);
@@ skipping 1538 matching lines @@
       assert(block.successors[0].id < block.id);
       assert(block.dominatedBlocks[0] === block.successors[1]);
     } else {
       assert(block.dominatedBlocks[0] === block.successors[0]);
       assert(block.dominatedBlocks[1] === block.successors[1]);;
     }
     return result;
   }
 }
 
-/**
- * A wrapper around a SourceString that stores extra information about
- * the (potentially implicit) quoting style of the original string.
- * For most strings, the quotes are included in the [wrappedString], but
- * parts of strings from a string interpolation might be missing one or
- * both quotes.
- */
-class QuotedString {
-  // Bits of flag values.
-  static final int NO_FLAGS = 0;
-  static final int RAW = 1 << 0;
-  static final int MULTI_LINE = 1 << 1;
-  static final int HAS_LEFT_QUOTE = 1 << 2;
-  static final int HAS_RIGHT_QUOTE = 1 << 3;
-  static final int HAS_BOTH_QUOTES = HAS_LEFT_QUOTE | HAS_RIGHT_QUOTE;
-  // Whether the string uses single quotes ('). Default is double quotes (").
-  static final int SINGLE_QUOTED = 1 << 4;
-
-  final SourceString wrappedString;
-  final int flags;
-
-  /**
-   * Finds the quote type for a string given a part of it containing the
-   * starting quote. Returns flags, but doesn't include HAS_LEFT_QUOTE
-   * or HAS_RIGHT_QUOTE.
-   */
-  static int flagsFromLeftQuote(SourceString sourceString) {
-    Iterator<int> source = sourceString.iterator();
-    int flags = 0;
-    int start = 0;
-    int quoteChar = source.next();
-    if (quoteChar == $AT) {
-      flags |= RAW;
-      start = 1;
-      quoteChar = source.next();
-    }
-    if (quoteChar == $SQ) {
-      flags |= SINGLE_QUOTED;
-    } else {
-      assert(quoteChar == $DQ);
-    }
-    // String has one quote. Check it if has three.
-    // If it only have two, the string must be an empty string literal,
-    // and end after the second quote.
-    if (source.hasNext() && source.next() == quoteChar && source.hasNext()) {
-      assert(source.next() == quoteChar);
-      flags |= MULTI_LINE;
-    }
-    return flags;
-  }
-
-  const QuotedString(this.wrappedString, this.flags);
-
-  // TODO(lrn): Make flags RAW for a literal string when we support it.
-  QuotedString.literal(String string) :
-      this(new SourceString(string), NO_FLAGS);
-
-  /**
-   * Crates a [QuotedString] from a [SourceString] that contains
-   * both its quotes.
-   */
-  factory QuotedString.explicit(SourceString string) {
-    int flags = flagsFromLeftQuote(string);
-    int leftQuotes = leftQuoteLengthFromFlags(flags | HAS_LEFT_QUOTE);
-    int rightQuotes = rightQuoteLengthFromFlags(flags | HAS_RIGHT_QUOTE);
-    SourceString unquoted = string.copyWithoutQuotes(leftQuotes, rightQuotes);
-    return new QuotedString(unquoted, flags);
-  }
-
-  SourceString unquotedSource() {
-    if ((flags & HAS_BOTH_QUOTES) == 0) return wrappedString;
-    return wrappedString.copyWithoutQuotes(leftQuoteLength,
-                                           rightQuoteLength);
-  }
-
-  bool get hasLeftQuote() => (flags & HAS_LEFT_QUOTE) != 0;
-  bool get hasRightQuote() => (flags & HAS_RIGHT_QUOTE) != 0;
-  bool get isMultiLine() => (flags & MULTI_LINE) != 0;
-  bool get isRaw() => (flags & RAW) != 0;
-  String get quoteChar() => ((flags & SINGLE_QUOTED) != 0) ? "'" : '"';
-  int get quoteCharCode() => ((flags & SINGLE_QUOTED) != 0) ? $SQ : $DQ;
-
-  int get leftQuoteLength() =>
-     hasLeftQuote ? (isRaw ? 1 : 0) + (isMultiLine ? 3 : 1) : 0;
-  int get rightQuoteLength() =>
-     hasRightQuote ? (isMultiLine ? 3 : 1) : 0;
-  static int leftQuoteLengthFromFlags(int flags) {
-    if ((flags & HAS_LEFT_QUOTE) == 0) return 0;
-    return (flags & (RAW | MULTI_LINE)) + 1;
-  }
-  static int rightQuoteLengthFromFlags(int flags) {
-    if ((flags & HAS_RIGHT_QUOTE) == 0) return 0;
-    return (flags & MULTI_LINE) + 1;
-  }
-
-  bool isEmpty() => unquotedSource().isEmpty();
-
-  static int hexValue(int hexDigit) {
-    // hexDigit is one of '0'..'9', 'A'..'F' and 'a'..'f'.
-    if (hexDigit <= $9) {
-      return hexDigit - $0;
-    }
-    // Make letters lowercase.
-    hexDigit |= $a ^ $A;
-    hexDigit -= $a - 10;
-    assert(10 <= hexDigit && hexDigit <= 15);
-    return hexDigit;
-  }
-
-  static bool isHexDigit(int characterCode) {
-    if ($0 <= characterCode && characterCode <= $9) return true;
-    characterCode |= $a ^ $A;
-    return ($a <= characterCode && characterCode <= $f);
-  }
-
-  static int readUnicodeEscape(Iterator<int> iter,
-                                void cancel(String s)) {
-    if (!iter.hasNext()) cancel("Incomplete unicode escape.");
-    int code = iter.next();
-    if (code == $OPEN_CURLY_BRACKET) {
-      // In Dart, '\u{'x{0..7}'}' is a valid escape, but not in
-      // JS. Convert to a \uxxxx escape.
-      int value = 0;
-      int length = 0;
-      if (!iter.hasNext()) cancel("Incomplete unicode escape.");
-      int hexDigit = iter.next();
-      do {
-        if (!isHexDigit(hexDigit)) {
-          cancel("Invalid character in unicode escape");
-        }
-        value = value * 16 + hexValue(hexDigit);
-        length++;
-        if (length > 7) {
-          cancel("Invalid unicode escape length.");
-        }
-        if (!iter.hasNext()) cancel("Incomplete unicode escape.");
-        hexDigit = iter.next();
-      } while (hexDigit !== $CLOSE_CURLY_BRACKET);  // until '}'.
-      return value;
-    }
-    // Simple four-digit unicode escape.
-    int value = 0;
-    for (int i = 0; i < 4; i++) {
-      if (i > 0) {
-        if (!iter.hasNext()) cancel("Incomplete unicode escape.");
-        code = iter.next();
-      }
-      if (!isHexDigit(code)) cancel("Invalid character in unicode escape");
-      value = value * 16 + hexValue(code);
-    }
-    return value;
-  }
-
-  /**
-   * Write the contents of the quoted string to a [StringBuffer] in
-   * a form that is valid as JavaScript string literal content.
-   * The string is assumed quoted by [quote] characters.
-   * This method doesn't try to make the shortest string, but rather
-   * to be as close to the original string as possible.
-   */
-  void writeEscaped(StringBuffer buffer, int quote, void cancel(String s)) {
-    bool raw = this.isRaw;
-    Iterator<int> iter =
-        wrappedString.copyWithoutQuotes(leftQuoteLength,
-                                        rightQuoteLength).iterator();
-    while (iter.hasNext()) {
-      int code = iter.next();
-      if (code == quote) {
-        // We need to add a backslash before quotes, both in normal
-        // and in raw strings.
-        buffer.add(@'\');
-        buffer.add(code == $SQ ? "'" : '"');
-      } else if (code == $LF) {
-        // Newlines in strings only occour in multiline strings.
-        // They need to be written using escapes in JS..
-        assert(isMultiLine);
-        buffer.add(@'\n');
-      } else if (code == $CR) {
-        assert(isMultiLine);
-        buffer.add(@'\r');
-      } else if (code == $LS || code == $PS) {
-        // These Unicode line terminators are invalid in JS strings.
-        buffer.add(code == $LS ? @'\u2028' : @'\u2029');
-      } else if (code != $BACKSLASH) {
-        buffer.add(new String.fromCharCodes([code]));
-      } else if (raw) {
-        buffer.add(@'\\');
-      } else {
-        code = iter.next();
-        // TODO(lrn): Reading \x and \u escapes also validates the
-        // escape sequences. This should be done at an earlier step
-        // to catch errors even in dead code.
-        switch (code) {
-          case $u:
-            int value = readUnicodeEscape(iter, cancel);
-            if (value >= 0xD800 && value <= 0xDFFF || value > 0x10ffff) {
-              cancel('Invalid unicode scalar value.');
-            }
-            if (value > 0xffff) {
-              cancel('Unhandled Unicode value: $value - outside the BMP.');
-            }
-            buffer.add(@'\u');
-            for (int j = 12; j >= 0; j -= 4) {
-              int digit = (value >> j) & 0xf;
-              buffer.add("0123456789abcdef"[digit]);
-            }
-            break;
-          case $x:
-            buffer.add(@'\x');
-            List<int> codes = <int>[];
-            for (int i = 0; i < 2; i++) {
-              if (!iter.hasNext()) cancel("Incomplete hex escape");
-              code = iter.next();
-              if (!isHexDigit(code)) {
-                cancel("Invalid hex digit: " +
-                       "${new String.fromCharCodes([code])}");
-              }
-              codes.add(code);
-            }
-            buffer.add(new String.fromCharCodes(codes));
-            break;
-          // Character escapes that identical in meaning in JS.
-          case $b: buffer.add(@'\b'); break;
-          case $f: buffer.add(@'\f'); break;
-          case $n: buffer.add(@'\n'); break;
-          case $r: buffer.add(@'\r'); break;
-          case $t: buffer.add(@'\t'); break;
-          case $v: buffer.add(@'\v'); break;
-          // Identity escapes that must be escaped in JS strings.
-          case $BACKSLASH: buffer.add(@'\\'); break;
-          case $LF: buffer.add(@'\n'); break;
-          case $CR: buffer.add(@'\r'); break;
-          case $LS: buffer.add(@'\u2028'); break;
-          case $PS: buffer.add(@'\u2029'); break;
-          // Quotes may or may not need the escape.
-          case $SQ:
-          case $DQ:
-            // Only escape quotes if they match the generated string quotes.
-            if (code == quote) buffer.add(@'\');
-            buffer.add(code === $SQ ? "'" : '"');
-            break;
-          default:
-            // All other escaped characters are identity escapes,
-            // and don't need a backslash in JS.
-            buffer.add(new String.fromCharCodes([code]));
-            break;
-        }
-      }
-    }
-  }
-
-  /**
-    * Does a conservative test for equality between two quoted strings.
-    * Returns true if the two definitly have the same string.
-    * Returns false if the strings are different, or if it's not possible
-    * to (quickly) determine whether they are equal.
-    */
-  bool definitlyEquals(QuotedString other) {
-    return flags == other.flags && wrappedString == other.wrappedString;
-  }
-
-  void printOn(StringBuffer buffer) {
-    int start = leftQuoteLength;
-    int end = rightQuoteLength;
-    if (start == 1 && end == 1) {
-      wrappedString.printOn(buffer);
-    } else {
-      String quote = quoteChar;
-      buffer.add(quote);
-      wrappedString.copyWithoutQuotes(start, end).printOn(buffer);
-      buffer.add(quote);
-    }
-  }
-}
-
 class HLiteral extends HInstruction {
   final value;
   HLiteral(this.value, HType type) : super(<HInstruction>[]) {
     this.type = type;
     tryGenerateAtUseSite();  // Maybe avoid this if the literal is big?
   }
   void prepareGvn() {
     // We allow global value numbering of literals, but we still
     // prefer generating them at use sites. This allows us to do
     // better GVN'ing of instructions that use literals as input.
@@ skipping 420 matching lines @@
 
   HInstruction get expression() => inputs[0];
 
   HType computeType() => HType.BOOLEAN;
   bool hasExpectedType() => true;
 
   accept(HVisitor visitor) => visitor.visitIs(this);
 
   toString() => "$expression is $typeExpression";
 }