Ben Murdoch | 4a90d5f | 2016-03-22 12:00:34 +0000 | [diff] [blame] | 1 | // Copyright 2006-2008 the V8 project authors. All rights reserved. |
| 2 | // Use of this source code is governed by a BSD-style license that can be |
| 3 | // found in the LICENSE file. |
| 4 | |
| 5 | // This file contains support for URI manipulations written in |
| 6 | // JavaScript. |
| 7 | |
| 8 | (function(global, utils) { |
| 9 | |
| 10 | "use strict"; |
| 11 | |
| 12 | %CheckIsBootstrapping(); |
| 13 | |
| 14 | //- ------------------------------------------------------------------ |
| 15 | // Imports |
| 16 | |
| 17 | var GlobalObject = global.Object; |
Ben Murdoch | 4a90d5f | 2016-03-22 12:00:34 +0000 | [diff] [blame] | 18 | var InternalArray = utils.InternalArray; |
| 19 | var MakeURIError; |
| 20 | |
| 21 | utils.Import(function(from) { |
| 22 | MakeURIError = from.MakeURIError; |
| 23 | }); |
| 24 | |
| 25 | |
| 26 | // ------------------------------------------------------------------- |
| 27 | // Define internal helper functions. |
| 28 | |
| 29 | function HexValueOf(code) { |
| 30 | // 0-9 |
| 31 | if (code >= 48 && code <= 57) return code - 48; |
| 32 | // A-F |
| 33 | if (code >= 65 && code <= 70) return code - 55; |
| 34 | // a-f |
| 35 | if (code >= 97 && code <= 102) return code - 87; |
| 36 | |
| 37 | return -1; |
| 38 | } |
| 39 | |
Ben Murdoch | 4a90d5f | 2016-03-22 12:00:34 +0000 | [diff] [blame] | 40 | function URIHexCharsToCharCode(highChar, lowChar) { |
| 41 | var highCode = HexValueOf(highChar); |
| 42 | var lowCode = HexValueOf(lowChar); |
| 43 | if (highCode == -1 || lowCode == -1) throw MakeURIError(); |
| 44 | return (highCode << 4) | lowCode; |
| 45 | } |
| 46 | |
| 47 | // Callers must ensure that |result| is a sufficiently long sequential |
| 48 | // two-byte string! |
| 49 | function URIDecodeOctets(octets, result, index) { |
| 50 | var value; |
| 51 | var o0 = octets[0]; |
| 52 | if (o0 < 0x80) { |
| 53 | value = o0; |
| 54 | } else if (o0 < 0xc2) { |
| 55 | throw MakeURIError(); |
| 56 | } else { |
| 57 | var o1 = octets[1]; |
| 58 | if (o0 < 0xe0) { |
| 59 | var a = o0 & 0x1f; |
| 60 | if ((o1 < 0x80) || (o1 > 0xbf)) throw MakeURIError(); |
| 61 | var b = o1 & 0x3f; |
| 62 | value = (a << 6) + b; |
| 63 | if (value < 0x80 || value > 0x7ff) throw MakeURIError(); |
| 64 | } else { |
| 65 | var o2 = octets[2]; |
| 66 | if (o0 < 0xf0) { |
| 67 | var a = o0 & 0x0f; |
| 68 | if ((o1 < 0x80) || (o1 > 0xbf)) throw MakeURIError(); |
| 69 | var b = o1 & 0x3f; |
| 70 | if ((o2 < 0x80) || (o2 > 0xbf)) throw MakeURIError(); |
| 71 | var c = o2 & 0x3f; |
| 72 | value = (a << 12) + (b << 6) + c; |
| 73 | if ((value < 0x800) || (value > 0xffff)) throw MakeURIError(); |
| 74 | } else { |
| 75 | var o3 = octets[3]; |
| 76 | if (o0 < 0xf8) { |
| 77 | var a = (o0 & 0x07); |
| 78 | if ((o1 < 0x80) || (o1 > 0xbf)) throw MakeURIError(); |
| 79 | var b = (o1 & 0x3f); |
| 80 | if ((o2 < 0x80) || (o2 > 0xbf)) { |
| 81 | throw MakeURIError(); |
| 82 | } |
| 83 | var c = (o2 & 0x3f); |
| 84 | if ((o3 < 0x80) || (o3 > 0xbf)) throw MakeURIError(); |
| 85 | var d = (o3 & 0x3f); |
| 86 | value = (a << 18) + (b << 12) + (c << 6) + d; |
| 87 | if ((value < 0x10000) || (value > 0x10ffff)) throw MakeURIError(); |
| 88 | } else { |
| 89 | throw MakeURIError(); |
| 90 | } |
| 91 | } |
| 92 | } |
| 93 | } |
| 94 | if (0xD800 <= value && value <= 0xDFFF) throw MakeURIError(); |
| 95 | if (value < 0x10000) { |
| 96 | %_TwoByteSeqStringSetChar(index++, value, result); |
| 97 | } else { |
| 98 | %_TwoByteSeqStringSetChar(index++, (value >> 10) + 0xd7c0, result); |
| 99 | %_TwoByteSeqStringSetChar(index++, (value & 0x3ff) + 0xdc00, result); |
| 100 | } |
| 101 | return index; |
| 102 | } |
| 103 | |
| 104 | // ECMA-262, section 15.1.3 |
Ben Murdoch | 4a90d5f | 2016-03-22 12:00:34 +0000 | [diff] [blame] | 105 | function Decode(uri, reserved) { |
| 106 | uri = TO_STRING(uri); |
| 107 | var uriLength = uri.length; |
| 108 | var one_byte = %NewString(uriLength, NEW_ONE_BYTE_STRING); |
| 109 | var index = 0; |
| 110 | var k = 0; |
| 111 | |
| 112 | // Optimistically assume one-byte string. |
| 113 | for ( ; k < uriLength; k++) { |
| 114 | var code = %_StringCharCodeAt(uri, k); |
| 115 | if (code == 37) { // '%' |
| 116 | if (k + 2 >= uriLength) throw MakeURIError(); |
| 117 | var cc = URIHexCharsToCharCode(%_StringCharCodeAt(uri, k+1), |
| 118 | %_StringCharCodeAt(uri, k+2)); |
| 119 | if (cc >> 7) break; // Assumption wrong, two-byte string. |
| 120 | if (reserved(cc)) { |
| 121 | %_OneByteSeqStringSetChar(index++, 37, one_byte); // '%'. |
| 122 | %_OneByteSeqStringSetChar(index++, %_StringCharCodeAt(uri, k+1), |
| 123 | one_byte); |
| 124 | %_OneByteSeqStringSetChar(index++, %_StringCharCodeAt(uri, k+2), |
| 125 | one_byte); |
| 126 | } else { |
| 127 | %_OneByteSeqStringSetChar(index++, cc, one_byte); |
| 128 | } |
| 129 | k += 2; |
| 130 | } else { |
| 131 | if (code > 0x7f) break; // Assumption wrong, two-byte string. |
| 132 | %_OneByteSeqStringSetChar(index++, code, one_byte); |
| 133 | } |
| 134 | } |
| 135 | |
| 136 | one_byte = %TruncateString(one_byte, index); |
| 137 | if (k == uriLength) return one_byte; |
| 138 | |
| 139 | // Write into two byte string. |
| 140 | var two_byte = %NewString(uriLength - k, NEW_TWO_BYTE_STRING); |
| 141 | index = 0; |
| 142 | |
| 143 | for ( ; k < uriLength; k++) { |
| 144 | var code = %_StringCharCodeAt(uri, k); |
| 145 | if (code == 37) { // '%' |
| 146 | if (k + 2 >= uriLength) throw MakeURIError(); |
| 147 | var cc = URIHexCharsToCharCode(%_StringCharCodeAt(uri, ++k), |
| 148 | %_StringCharCodeAt(uri, ++k)); |
| 149 | if (cc >> 7) { |
| 150 | var n = 0; |
| 151 | while (((cc << ++n) & 0x80) != 0) { } |
| 152 | if (n == 1 || n > 4) throw MakeURIError(); |
Ben Murdoch | da12d29 | 2016-06-02 14:46:10 +0100 | [diff] [blame] | 153 | var octets = new InternalArray(n); |
Ben Murdoch | 4a90d5f | 2016-03-22 12:00:34 +0000 | [diff] [blame] | 154 | octets[0] = cc; |
| 155 | if (k + 3 * (n - 1) >= uriLength) throw MakeURIError(); |
| 156 | for (var i = 1; i < n; i++) { |
| 157 | if (uri[++k] != '%') throw MakeURIError(); |
| 158 | octets[i] = URIHexCharsToCharCode(%_StringCharCodeAt(uri, ++k), |
| 159 | %_StringCharCodeAt(uri, ++k)); |
| 160 | } |
| 161 | index = URIDecodeOctets(octets, two_byte, index); |
| 162 | } else if (reserved(cc)) { |
| 163 | %_TwoByteSeqStringSetChar(index++, 37, two_byte); // '%'. |
| 164 | %_TwoByteSeqStringSetChar(index++, %_StringCharCodeAt(uri, k - 1), |
| 165 | two_byte); |
| 166 | %_TwoByteSeqStringSetChar(index++, %_StringCharCodeAt(uri, k), |
| 167 | two_byte); |
| 168 | } else { |
| 169 | %_TwoByteSeqStringSetChar(index++, cc, two_byte); |
| 170 | } |
| 171 | } else { |
| 172 | %_TwoByteSeqStringSetChar(index++, code, two_byte); |
| 173 | } |
| 174 | } |
| 175 | |
| 176 | two_byte = %TruncateString(two_byte, index); |
| 177 | return one_byte + two_byte; |
| 178 | } |
| 179 | |
| 180 | // ------------------------------------------------------------------- |
| 181 | // Define exported functions. |
| 182 | |
| 183 | // ECMA-262 - B.2.1. |
| 184 | function URIEscapeJS(s) { |
| 185 | return %URIEscape(s); |
| 186 | } |
| 187 | |
| 188 | // ECMA-262 - B.2.2. |
| 189 | function URIUnescapeJS(s) { |
| 190 | return %URIUnescape(s); |
| 191 | } |
| 192 | |
| 193 | // ECMA-262 - 15.1.3.1. |
| 194 | function URIDecode(uri) { |
| 195 | var reservedPredicate = function(cc) { |
| 196 | // #$ |
| 197 | if (35 <= cc && cc <= 36) return true; |
| 198 | // & |
| 199 | if (cc == 38) return true; |
| 200 | // +, |
| 201 | if (43 <= cc && cc <= 44) return true; |
| 202 | // / |
| 203 | if (cc == 47) return true; |
| 204 | // :; |
| 205 | if (58 <= cc && cc <= 59) return true; |
| 206 | // = |
| 207 | if (cc == 61) return true; |
| 208 | // ?@ |
| 209 | if (63 <= cc && cc <= 64) return true; |
| 210 | |
| 211 | return false; |
| 212 | }; |
| 213 | return Decode(uri, reservedPredicate); |
| 214 | } |
| 215 | |
| 216 | // ECMA-262 - 15.1.3.2. |
| 217 | function URIDecodeComponent(component) { |
| 218 | var reservedPredicate = function(cc) { return false; }; |
| 219 | return Decode(component, reservedPredicate); |
| 220 | } |
| 221 | |
Ben Murdoch | 4a90d5f | 2016-03-22 12:00:34 +0000 | [diff] [blame] | 222 | // ------------------------------------------------------------------- |
| 223 | // Install exported functions. |
| 224 | |
| 225 | // Set up non-enumerable URI functions on the global object and set |
| 226 | // their names. |
| 227 | utils.InstallFunctions(global, DONT_ENUM, [ |
| 228 | "escape", URIEscapeJS, |
| 229 | "unescape", URIUnescapeJS, |
| 230 | "decodeURI", URIDecode, |
Ben Murdoch | c561043 | 2016-08-08 18:44:38 +0100 | [diff] [blame^] | 231 | "decodeURIComponent", URIDecodeComponent |
Ben Murdoch | 4a90d5f | 2016-03-22 12:00:34 +0000 | [diff] [blame] | 232 | ]); |
| 233 | |
| 234 | }) |