Kaido Kert | f309f9a | 2021-04-30 12:09:15 -0700 | [diff] [blame] | 1 | // Copyright 2019 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 | #include 'src/builtins/builtins-string-gen.h' |
| 6 | |
| 7 | namespace string { |
| 8 | |
| 9 | extern enum TrimMode extends uint31 constexpr 'String::TrimMode' { |
| 10 | kTrim, |
| 11 | kTrimStart, |
| 12 | kTrimEnd |
| 13 | } |
| 14 | |
| 15 | @export |
| 16 | macro IsWhiteSpaceOrLineTerminator(charCode: int32): bool { |
| 17 | // 0x0020 - SPACE (Intentionally out of order to fast path a commmon case) |
| 18 | if (charCode == Int32Constant(0x0020)) { |
| 19 | return true; |
| 20 | } |
| 21 | |
| 22 | // 0x0009 - HORIZONTAL TAB |
| 23 | if (charCode < Int32Constant(0x0009)) { |
| 24 | return false; |
| 25 | } |
| 26 | // 0x000A - LINE FEED OR NEW LINE |
| 27 | // 0x000B - VERTICAL TAB |
| 28 | // 0x000C - FORMFEED |
| 29 | // 0x000D - HORIZONTAL TAB |
| 30 | if (charCode <= Int32Constant(0x000D)) { |
| 31 | return true; |
| 32 | } |
| 33 | |
| 34 | // Common Non-whitespace characters |
| 35 | if (charCode < Int32Constant(0x00A0)) { |
| 36 | return false; |
| 37 | } |
| 38 | |
| 39 | // 0x00A0 - NO-BREAK SPACE |
| 40 | if (charCode == Int32Constant(0x00A0)) { |
| 41 | return true; |
| 42 | } |
| 43 | |
| 44 | // 0x1680 - Ogham Space Mark |
| 45 | if (charCode == Int32Constant(0x1680)) { |
| 46 | return true; |
| 47 | } |
| 48 | |
| 49 | // 0x2000 - EN QUAD |
| 50 | if (charCode < Int32Constant(0x2000)) { |
| 51 | return false; |
| 52 | } |
| 53 | // 0x2001 - EM QUAD |
| 54 | // 0x2002 - EN SPACE |
| 55 | // 0x2003 - EM SPACE |
| 56 | // 0x2004 - THREE-PER-EM SPACE |
| 57 | // 0x2005 - FOUR-PER-EM SPACE |
| 58 | // 0x2006 - SIX-PER-EM SPACE |
| 59 | // 0x2007 - FIGURE SPACE |
| 60 | // 0x2008 - PUNCTUATION SPACE |
| 61 | // 0x2009 - THIN SPACE |
| 62 | // 0x200A - HAIR SPACE |
| 63 | if (charCode <= Int32Constant(0x200A)) { |
| 64 | return true; |
| 65 | } |
| 66 | |
| 67 | // 0x2028 - LINE SEPARATOR |
| 68 | if (charCode == Int32Constant(0x2028)) { |
| 69 | return true; |
| 70 | } |
| 71 | // 0x2029 - PARAGRAPH SEPARATOR |
| 72 | if (charCode == Int32Constant(0x2029)) { |
| 73 | return true; |
| 74 | } |
| 75 | // 0x202F - NARROW NO-BREAK SPACE |
| 76 | if (charCode == Int32Constant(0x202F)) { |
| 77 | return true; |
| 78 | } |
| 79 | // 0x205F - MEDIUM MATHEMATICAL SPACE |
| 80 | if (charCode == Int32Constant(0x205F)) { |
| 81 | return true; |
| 82 | } |
| 83 | // 0xFEFF - BYTE ORDER MARK |
| 84 | if (charCode == Int32Constant(0xFEFF)) { |
| 85 | return true; |
| 86 | } |
| 87 | // 0x3000 - IDEOGRAPHIC SPACE |
| 88 | if (charCode == Int32Constant(0x3000)) { |
| 89 | return true; |
| 90 | } |
| 91 | |
| 92 | return false; |
| 93 | } |
| 94 | |
| 95 | transitioning macro StringTrim(implicit context: Context)( |
| 96 | receiver: JSAny, _arguments: Arguments, methodName: constexpr string, |
| 97 | variant: constexpr TrimMode): String { |
| 98 | const receiverString: String = ToThisString(receiver, methodName); |
| 99 | const stringLength: intptr = receiverString.length_intptr; |
| 100 | |
| 101 | const directString = Cast<DirectString>(receiverString) |
| 102 | otherwise return runtime::StringTrim( |
| 103 | receiverString, SmiTag<TrimMode>(variant)); |
| 104 | |
| 105 | let startIndex: intptr = 0; |
| 106 | let endIndex: intptr = stringLength - 1; |
| 107 | |
| 108 | // TODO(duongn): It would probably be more efficient to turn StringTrim into a |
| 109 | // tempalate for the different string types and specialize the loop for them. |
| 110 | if (variant == TrimMode::kTrim || variant == TrimMode::kTrimStart) { |
| 111 | while (true) { |
| 112 | if (startIndex == stringLength) { |
| 113 | return EmptyStringConstant(); |
| 114 | } |
| 115 | if (!IsWhiteSpaceOrLineTerminator( |
| 116 | StringCharCodeAt(directString, Unsigned(startIndex)))) { |
| 117 | break; |
| 118 | } |
| 119 | startIndex++; |
| 120 | } |
| 121 | } |
| 122 | |
| 123 | if (variant == TrimMode::kTrim || variant == TrimMode::kTrimEnd) { |
| 124 | while (true) { |
| 125 | if (endIndex == -1) { |
| 126 | return EmptyStringConstant(); |
| 127 | } |
| 128 | if (!IsWhiteSpaceOrLineTerminator( |
| 129 | StringCharCodeAt(directString, Unsigned(endIndex)))) { |
| 130 | break; |
| 131 | } |
| 132 | endIndex--; |
| 133 | } |
| 134 | } |
| 135 | |
| 136 | return SubString( |
| 137 | receiverString, Unsigned(startIndex), Unsigned(endIndex + 1)); |
| 138 | } |
| 139 | |
| 140 | // ES6 #sec-string.prototype.trim |
| 141 | transitioning javascript builtin |
| 142 | StringPrototypeTrim( |
| 143 | js-implicit context: NativeContext, receiver: JSAny)(...arguments): String { |
| 144 | const methodName: constexpr string = 'String.prototype.trim'; |
| 145 | return StringTrim(receiver, arguments, methodName, TrimMode::kTrim); |
| 146 | } |
| 147 | |
| 148 | // https://github.com/tc39/proposal-string-left-right-trim |
| 149 | transitioning javascript builtin |
| 150 | StringPrototypeTrimStart( |
| 151 | js-implicit context: NativeContext, receiver: JSAny)(...arguments): String { |
| 152 | const methodName: constexpr string = 'String.prototype.trimLeft'; |
| 153 | return StringTrim(receiver, arguments, methodName, TrimMode::kTrimStart); |
| 154 | } |
| 155 | |
| 156 | // https://github.com/tc39/proposal-string-left-right-trim |
| 157 | transitioning javascript builtin |
| 158 | StringPrototypeTrimEnd( |
| 159 | js-implicit context: NativeContext, receiver: JSAny)(...arguments): String { |
| 160 | const methodName: constexpr string = 'String.prototype.trimRight'; |
| 161 | return StringTrim(receiver, arguments, methodName, TrimMode::kTrimEnd); |
| 162 | } |
| 163 | } |
| 164 | |
| 165 | namespace runtime { |
| 166 | extern runtime StringTrim(implicit context: Context)( |
| 167 | String, SmiTagged<string::TrimMode>): String; |
| 168 | } |