| OLD | NEW |
| (Empty) |
| 1 // Copyright (c) 2012, the Dart project authors. Please see the AUTHORS file | |
| 2 // for details. All rights reserved. Use of this source code is governed by a | |
| 3 // BSD-style license that can be found in the LICENSE file. | |
| 4 | |
| 5 // Check the validity of string literals. | |
| 6 | |
| 7 #library("stringvalidator"); | |
| 8 | |
| 9 #import("leg.dart"); | |
| 10 #import("scanner/scannerlib.dart"); | |
| 11 #import("tree/tree.dart"); | |
| 12 #import("elements/elements.dart"); | |
| 13 #import("util/characters.dart"); | |
| 14 | |
| 15 class StringValidator { | |
| 16 final DiagnosticListener listener; | |
| 17 | |
| 18 StringValidator(this.listener); | |
| 19 | |
| 20 DartString validateQuotedString(Token token) { | |
| 21 SourceString source = token.value; | |
| 22 StringQuoting quoting = quotingFromString(source); | |
| 23 int leftQuote = quoting.leftQuoteLength; | |
| 24 int rightQuote = quoting.rightQuoteLength; | |
| 25 SourceString content = source.copyWithoutQuotes(leftQuote, rightQuote); | |
| 26 return validateString(token, | |
| 27 token.charOffset + leftQuote, | |
| 28 content, | |
| 29 quoting); | |
| 30 } | |
| 31 | |
| 32 DartString validateInterpolationPart(Token token, StringQuoting quoting, | |
| 33 [bool isFirst = false, | |
| 34 bool isLast = false]) { | |
| 35 SourceString source = token.value; | |
| 36 int leftQuote = 0; | |
| 37 int rightQuote = 0; | |
| 38 if (isFirst) leftQuote = quoting.leftQuoteLength; | |
| 39 if (isLast) rightQuote = quoting.rightQuoteLength; | |
| 40 SourceString content = source.copyWithoutQuotes(leftQuote, rightQuote); | |
| 41 return validateString(token, | |
| 42 token.charOffset + leftQuote, | |
| 43 content, | |
| 44 quoting); | |
| 45 } | |
| 46 | |
| 47 static StringQuoting quotingFromString(SourceString sourceString) { | |
| 48 Iterator<int> source = sourceString.iterator(); | |
| 49 bool raw = false; | |
| 50 int quoteLength = 1; | |
| 51 int quoteChar = source.next(); | |
| 52 if (quoteChar == $AT) { | |
| 53 raw = true; | |
| 54 quoteChar = source.next(); | |
| 55 } | |
| 56 assert(quoteChar === $SQ || quoteChar === $DQ); | |
| 57 // String has at least one quote. Check it if has three. | |
| 58 // If it only have two, the string must be an empty string literal, | |
| 59 // and end after the second quote. | |
| 60 bool multiline = false; | |
| 61 if (source.hasNext() && source.next() === quoteChar && source.hasNext()) { | |
| 62 int code = source.next(); | |
| 63 assert(code === quoteChar); // If not, there is a bug in the parser. | |
| 64 quoteLength = 3; | |
| 65 // Check if a multiline string starts with a newline (CR, LF or CR+LF). | |
| 66 if (source.hasNext()) { | |
| 67 code = source.next(); | |
| 68 if (code === $CR) { | |
| 69 quoteLength += 1; | |
| 70 if (source.hasNext() && source.next() === $LF) { | |
| 71 quoteLength += 1; | |
| 72 } | |
| 73 } else if (code === $LF) { | |
| 74 quoteLength += 1; | |
| 75 } | |
| 76 } | |
| 77 } | |
| 78 return StringQuoting.getQuoting(quoteChar, raw, quoteLength); | |
| 79 } | |
| 80 | |
| 81 void stringParseError(String message, Token token, int offset) { | |
| 82 listener.cancel("$message @ $offset", token : token); | |
| 83 } | |
| 84 | |
| 85 /** | |
| 86 * Validates the escape sequences and special characters of a string literal. | |
| 87 * Returns a DartString if valid, and null if not. | |
| 88 */ | |
| 89 DartString validateString(Token token, | |
| 90 int startOffset, | |
| 91 SourceString string, | |
| 92 StringQuoting quoting) { | |
| 93 // We only need to check for invalid x and u escapes, for line | |
| 94 // terminators in non-multiline strings, and for invalid Unicode | |
| 95 // scalar values (either directly or as u-escape values). | |
| 96 int length = 0; | |
| 97 int index = startOffset; | |
| 98 bool containsEscape = false; | |
| 99 for(Iterator<int> iter = string.iterator(); iter.hasNext(); length++) { | |
| 100 index++; | |
| 101 int code = iter.next(); | |
| 102 if (code === $BACKSLASH) { | |
| 103 if (quoting.raw) continue; | |
| 104 containsEscape = true; | |
| 105 if (!iter.hasNext()) { | |
| 106 stringParseError("Incomplete escape sequence",token, index); | |
| 107 return null; | |
| 108 } | |
| 109 index++; | |
| 110 code = iter.next(); | |
| 111 if (code === $x) { | |
| 112 for (int i = 0; i < 2; i++) { | |
| 113 if (!iter.hasNext()) { | |
| 114 stringParseError("Incomplete escape sequence", token, index); | |
| 115 return null; | |
| 116 } | |
| 117 index++; | |
| 118 code = iter.next(); | |
| 119 if (!isHexDigit(code)) { | |
| 120 stringParseError("Invalid character in escape sequence", | |
| 121 token, index); | |
| 122 return null; | |
| 123 } | |
| 124 } | |
| 125 // A two-byte hex escape can't generate an invalid value. | |
| 126 continue; | |
| 127 } else if (code === $u) { | |
| 128 int escapeStart = index - 1; | |
| 129 index++; | |
| 130 code = iter.next(); | |
| 131 int value = 0; | |
| 132 if (code == $OPEN_CURLY_BRACKET) { | |
| 133 // expect 1-6 hex digits. | |
| 134 int count = 0; | |
| 135 index++; | |
| 136 code = iter.next(); | |
| 137 do { | |
| 138 if (!isHexDigit(code)) { | |
| 139 stringParseError("Invalid character in escape sequence", | |
| 140 token, index); | |
| 141 return null; | |
| 142 } | |
| 143 count++; | |
| 144 value = value * 16 + hexDigitValue(code); | |
| 145 index++; | |
| 146 code = iter.next(); | |
| 147 } while (code != $CLOSE_CURLY_BRACKET); | |
| 148 if (count > 6) { | |
| 149 stringParseError("Invalid character in escape sequence", | |
| 150 token, index - (count - 6)); | |
| 151 return null; | |
| 152 } | |
| 153 } else { | |
| 154 // Expect four hex digits, including the one just read. | |
| 155 for (int i = 0; i < 4; i++) { | |
| 156 if (i > 0) { | |
| 157 index++; | |
| 158 code = iter.next(); | |
| 159 } | |
| 160 if (!isHexDigit(code)) { | |
| 161 stringParseError("Invalid character in escape sequence", | |
| 162 token, index); | |
| 163 return null; | |
| 164 } | |
| 165 value = value * 16 + hexDigitValue(code); | |
| 166 } | |
| 167 } | |
| 168 code = value; | |
| 169 } | |
| 170 } | |
| 171 // This handles both unescaped characters and the value of unicode | |
| 172 // escapes. | |
| 173 if (!isUnicodeScalarValue(code)) { | |
| 174 stringParseError( | |
| 175 "Invalid Unicode scalar value U+${code.toRadixString(16)}", | |
| 176 token, index); | |
| 177 return null; | |
| 178 } | |
| 179 } | |
| 180 // String literal successfully validated. | |
| 181 if (quoting.raw || !containsEscape) { | |
| 182 // A string without escapes could just as well have been raw. | |
| 183 return new DartString.rawString(string, length); | |
| 184 } | |
| 185 return new DartString.escapedString(string, length); | |
| 186 } | |
| 187 } | |
| OLD | NEW |