The data contained in this repository can be downloaded to your computer using one of several clients.
Please see the documentation of your version control software client for more information.
Please select the desired protocol below to get the URL.
This URL has Read-Only access.
main_repo / deps / v8 / test / cctest / test-regexp.cc @ 40c0f755
History | View | Annotate | Download (49 KB)
| 1 |
// Copyright 2008 the V8 project authors. All rights reserved.
|
|---|---|
| 2 |
// Redistribution and use in source and binary forms, with or without
|
| 3 |
// modification, are permitted provided that the following conditions are
|
| 4 |
// met:
|
| 5 |
//
|
| 6 |
// * Redistributions of source code must retain the above copyright
|
| 7 |
// notice, this list of conditions and the following disclaimer.
|
| 8 |
// * Redistributions in binary form must reproduce the above
|
| 9 |
// copyright notice, this list of conditions and the following
|
| 10 |
// disclaimer in the documentation and/or other materials provided
|
| 11 |
// with the distribution.
|
| 12 |
// * Neither the name of Google Inc. nor the names of its
|
| 13 |
// contributors may be used to endorse or promote products derived
|
| 14 |
// from this software without specific prior written permission.
|
| 15 |
//
|
| 16 |
// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
|
| 17 |
// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
|
| 18 |
// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
|
| 19 |
// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
|
| 20 |
// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
|
| 21 |
// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
|
| 22 |
// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
|
| 23 |
// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
|
| 24 |
// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
|
| 25 |
// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
|
| 26 |
// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
|
| 27 |
|
| 28 |
|
| 29 |
#include <stdlib.h> |
| 30 |
#include <set> |
| 31 |
|
| 32 |
#include "v8.h" |
| 33 |
|
| 34 |
#include "string-stream.h" |
| 35 |
#include "cctest.h" |
| 36 |
#include "zone-inl.h" |
| 37 |
#include "parser.h" |
| 38 |
#include "ast.h" |
| 39 |
#include "jsregexp-inl.h" |
| 40 |
#include "regexp-macro-assembler.h" |
| 41 |
#include "regexp-macro-assembler-irregexp.h" |
| 42 |
#ifdef ARM
|
| 43 |
#include "regexp-macro-assembler-arm.h" |
| 44 |
#else // IA32 |
| 45 |
#include "macro-assembler-ia32.h" |
| 46 |
#include "regexp-macro-assembler-ia32.h" |
| 47 |
#endif
|
| 48 |
#include "interpreter-irregexp.h" |
| 49 |
|
| 50 |
|
| 51 |
using namespace v8::internal; |
| 52 |
|
| 53 |
|
| 54 |
static SmartPointer<const char> Parse(const char* input) { |
| 55 |
V8::Initialize(NULL);
|
| 56 |
v8::HandleScope scope; |
| 57 |
ZoneScope zone_scope(DELETE_ON_EXIT); |
| 58 |
FlatStringReader reader(CStrVector(input)); |
| 59 |
RegExpCompileData result; |
| 60 |
CHECK(v8::internal::ParseRegExp(&reader, false, &result));
|
| 61 |
CHECK(result.tree != NULL);
|
| 62 |
CHECK(result.error.is_null()); |
| 63 |
SmartPointer<const char> output = result.tree->ToString(); |
| 64 |
return output;
|
| 65 |
} |
| 66 |
|
| 67 |
static bool CheckSimple(const char* input) { |
| 68 |
V8::Initialize(NULL);
|
| 69 |
v8::HandleScope scope; |
| 70 |
unibrow::Utf8InputBuffer<> buffer(input, strlen(input)); |
| 71 |
ZoneScope zone_scope(DELETE_ON_EXIT); |
| 72 |
FlatStringReader reader(CStrVector(input)); |
| 73 |
RegExpCompileData result; |
| 74 |
CHECK(v8::internal::ParseRegExp(&reader, false, &result));
|
| 75 |
CHECK(result.tree != NULL);
|
| 76 |
CHECK(result.error.is_null()); |
| 77 |
return result.simple;
|
| 78 |
} |
| 79 |
|
| 80 |
struct MinMaxPair {
|
| 81 |
int min_match;
|
| 82 |
int max_match;
|
| 83 |
}; |
| 84 |
|
| 85 |
static MinMaxPair CheckMinMaxMatch(const char* input) { |
| 86 |
V8::Initialize(NULL);
|
| 87 |
v8::HandleScope scope; |
| 88 |
unibrow::Utf8InputBuffer<> buffer(input, strlen(input)); |
| 89 |
ZoneScope zone_scope(DELETE_ON_EXIT); |
| 90 |
FlatStringReader reader(CStrVector(input)); |
| 91 |
RegExpCompileData result; |
| 92 |
CHECK(v8::internal::ParseRegExp(&reader, false, &result));
|
| 93 |
CHECK(result.tree != NULL);
|
| 94 |
CHECK(result.error.is_null()); |
| 95 |
int min_match = result.tree->min_match();
|
| 96 |
int max_match = result.tree->max_match();
|
| 97 |
MinMaxPair pair = { min_match, max_match };
|
| 98 |
return pair;
|
| 99 |
} |
| 100 |
|
| 101 |
|
| 102 |
|
| 103 |
#define CHECK_PARSE_EQ(input, expected) CHECK_EQ(expected, *Parse(input))
|
| 104 |
#define CHECK_SIMPLE(input, simple) CHECK_EQ(simple, CheckSimple(input));
|
| 105 |
#define CHECK_MIN_MAX(input, min, max) \
|
| 106 |
{ MinMaxPair min_max = CheckMinMaxMatch(input); \
|
| 107 |
CHECK_EQ(min, min_max.min_match); \ |
| 108 |
CHECK_EQ(max, min_max.max_match); \ |
| 109 |
} |
| 110 |
|
| 111 |
TEST(Parser) {
|
| 112 |
V8::Initialize(NULL);
|
| 113 |
CHECK_PARSE_EQ("abc", "'abc'"); |
| 114 |
CHECK_PARSE_EQ("", "%"); |
| 115 |
CHECK_PARSE_EQ("abc|def", "(| 'abc' 'def')"); |
| 116 |
CHECK_PARSE_EQ("abc|def|ghi", "(| 'abc' 'def' 'ghi')"); |
| 117 |
CHECK_PARSE_EQ("^xxx$", "(: @^i 'xxx' @$i)"); |
| 118 |
CHECK_PARSE_EQ("ab\\b\\d\\bcd", "(: 'ab' @b [0-9] @b 'cd')"); |
| 119 |
CHECK_PARSE_EQ("\\w|\\d", "(| [0-9 A-Z _ a-z] [0-9])"); |
| 120 |
CHECK_PARSE_EQ("a*", "(# 0 - g 'a')"); |
| 121 |
CHECK_PARSE_EQ("a*?", "(# 0 - n 'a')"); |
| 122 |
CHECK_PARSE_EQ("abc+", "(: 'ab' (# 1 - g 'c'))"); |
| 123 |
CHECK_PARSE_EQ("abc+?", "(: 'ab' (# 1 - n 'c'))"); |
| 124 |
CHECK_PARSE_EQ("xyz?", "(: 'xy' (# 0 1 g 'z'))"); |
| 125 |
CHECK_PARSE_EQ("xyz??", "(: 'xy' (# 0 1 n 'z'))"); |
| 126 |
CHECK_PARSE_EQ("xyz{0,1}", "(: 'xy' (# 0 1 g 'z'))"); |
| 127 |
CHECK_PARSE_EQ("xyz{0,1}?", "(: 'xy' (# 0 1 n 'z'))"); |
| 128 |
CHECK_PARSE_EQ("xyz{93}", "(: 'xy' (# 93 93 g 'z'))"); |
| 129 |
CHECK_PARSE_EQ("xyz{93}?", "(: 'xy' (# 93 93 n 'z'))"); |
| 130 |
CHECK_PARSE_EQ("xyz{1,32}", "(: 'xy' (# 1 32 g 'z'))"); |
| 131 |
CHECK_PARSE_EQ("xyz{1,32}?", "(: 'xy' (# 1 32 n 'z'))"); |
| 132 |
CHECK_PARSE_EQ("xyz{1,}", "(: 'xy' (# 1 - g 'z'))"); |
| 133 |
CHECK_PARSE_EQ("xyz{1,}?", "(: 'xy' (# 1 - n 'z'))"); |
| 134 |
CHECK_PARSE_EQ("a\\fb\\nc\\rd\\te\\vf", "'a\\x0cb\\x0ac\\x0dd\\x09e\\x0bf'"); |
| 135 |
CHECK_PARSE_EQ("a\\nb\\bc", "(: 'a\\x0ab' @b 'c')"); |
| 136 |
CHECK_PARSE_EQ("(?:foo)", "'foo'"); |
| 137 |
CHECK_PARSE_EQ("(?: foo )", "' foo '"); |
| 138 |
CHECK_PARSE_EQ("(foo|bar|baz)", "(^ (| 'foo' 'bar' 'baz'))"); |
| 139 |
CHECK_PARSE_EQ("foo|(bar|baz)|quux", "(| 'foo' (^ (| 'bar' 'baz')) 'quux')"); |
| 140 |
CHECK_PARSE_EQ("foo(?=bar)baz", "(: 'foo' (-> + 'bar') 'baz')"); |
| 141 |
CHECK_PARSE_EQ("foo(?!bar)baz", "(: 'foo' (-> - 'bar') 'baz')"); |
| 142 |
CHECK_PARSE_EQ("()", "(^ %)"); |
| 143 |
CHECK_PARSE_EQ("(?=)", "(-> + %)"); |
| 144 |
CHECK_PARSE_EQ("[]", "^[\\x00-\\uffff]"); // Doesn't compile on windows |
| 145 |
CHECK_PARSE_EQ("[^]", "[\\x00-\\uffff]"); // \uffff isn't in codepage 1252 |
| 146 |
CHECK_PARSE_EQ("[x]", "[x]"); |
| 147 |
CHECK_PARSE_EQ("[xyz]", "[x y z]"); |
| 148 |
CHECK_PARSE_EQ("[a-zA-Z0-9]", "[a-z A-Z 0-9]"); |
| 149 |
CHECK_PARSE_EQ("[-123]", "[- 1 2 3]"); |
| 150 |
CHECK_PARSE_EQ("[^123]", "^[1 2 3]"); |
| 151 |
CHECK_PARSE_EQ("]", "']'"); |
| 152 |
CHECK_PARSE_EQ("}", "'}'"); |
| 153 |
CHECK_PARSE_EQ("[a-b-c]", "[a-b - c]"); |
| 154 |
CHECK_PARSE_EQ("[\\d]", "[0-9]"); |
| 155 |
CHECK_PARSE_EQ("[x\\dz]", "[x 0-9 z]"); |
| 156 |
CHECK_PARSE_EQ("[\\d-z]", "[0-9 - z]"); |
| 157 |
CHECK_PARSE_EQ("[\\d-\\d]", "[0-9 - 0-9]"); |
| 158 |
CHECK_PARSE_EQ("[z-\\d]", "[z - 0-9]"); |
| 159 |
CHECK_PARSE_EQ("\\cj\\cJ\\ci\\cI\\ck\\cK",
|
| 160 |
"'\\x0a\\x0a\\x09\\x09\\x0b\\x0b'");
|
| 161 |
CHECK_PARSE_EQ("\\c!", "'c!'"); |
| 162 |
CHECK_PARSE_EQ("\\c_", "'c_'"); |
| 163 |
CHECK_PARSE_EQ("\\c~", "'c~'"); |
| 164 |
CHECK_PARSE_EQ("[a\\]c]", "[a ] c]"); |
| 165 |
CHECK_PARSE_EQ("\\[\\]\\{\\}\\(\\)\\%\\^\\#\\ ", "'[]{}()%^# '"); |
| 166 |
CHECK_PARSE_EQ("[\\[\\]\\{\\}\\(\\)\\%\\^\\#\\ ]", "[[ ] { } ( ) % ^ # ]"); |
| 167 |
CHECK_PARSE_EQ("\\0", "'\\x00'"); |
| 168 |
CHECK_PARSE_EQ("\\8", "'8'"); |
| 169 |
CHECK_PARSE_EQ("\\9", "'9'"); |
| 170 |
CHECK_PARSE_EQ("\\11", "'\\x09'"); |
| 171 |
CHECK_PARSE_EQ("\\11a", "'\\x09a'"); |
| 172 |
CHECK_PARSE_EQ("\\011", "'\\x09'"); |
| 173 |
CHECK_PARSE_EQ("\\00011", "'\\x0011'"); |
| 174 |
CHECK_PARSE_EQ("\\118", "'\\x098'"); |
| 175 |
CHECK_PARSE_EQ("\\111", "'I'"); |
| 176 |
CHECK_PARSE_EQ("\\1111", "'I1'"); |
| 177 |
CHECK_PARSE_EQ("(x)(x)(x)\\1", "(: (^ 'x') (^ 'x') (^ 'x') (<- 1))"); |
| 178 |
CHECK_PARSE_EQ("(x)(x)(x)\\2", "(: (^ 'x') (^ 'x') (^ 'x') (<- 2))"); |
| 179 |
CHECK_PARSE_EQ("(x)(x)(x)\\3", "(: (^ 'x') (^ 'x') (^ 'x') (<- 3))"); |
| 180 |
CHECK_PARSE_EQ("(x)(x)(x)\\4", "(: (^ 'x') (^ 'x') (^ 'x') '\\x04')"); |
| 181 |
CHECK_PARSE_EQ("(x)(x)(x)\\1*", "(: (^ 'x') (^ 'x') (^ 'x')" |
| 182 |
" (# 0 - g (<- 1)))");
|
| 183 |
CHECK_PARSE_EQ("(x)(x)(x)\\2*", "(: (^ 'x') (^ 'x') (^ 'x')" |
| 184 |
" (# 0 - g (<- 2)))");
|
| 185 |
CHECK_PARSE_EQ("(x)(x)(x)\\3*", "(: (^ 'x') (^ 'x') (^ 'x')" |
| 186 |
" (# 0 - g (<- 3)))");
|
| 187 |
CHECK_PARSE_EQ("(x)(x)(x)\\4*", "(: (^ 'x') (^ 'x') (^ 'x')" |
| 188 |
" (# 0 - g '\\x04'))");
|
| 189 |
CHECK_PARSE_EQ("(x)(x)(x)(x)(x)(x)(x)(x)(x)(x)\\10",
|
| 190 |
"(: (^ 'x') (^ 'x') (^ 'x') (^ 'x') (^ 'x') (^ 'x')"
|
| 191 |
" (^ 'x') (^ 'x') (^ 'x') (^ 'x') (<- 10))");
|
| 192 |
CHECK_PARSE_EQ("(x)(x)(x)(x)(x)(x)(x)(x)(x)(x)\\11",
|
| 193 |
"(: (^ 'x') (^ 'x') (^ 'x') (^ 'x') (^ 'x') (^ 'x')"
|
| 194 |
" (^ 'x') (^ 'x') (^ 'x') (^ 'x') '\\x09')");
|
| 195 |
CHECK_PARSE_EQ("(a)\\1", "(: (^ 'a') (<- 1))"); |
| 196 |
CHECK_PARSE_EQ("(a\\1)", "(^ 'a')"); |
| 197 |
CHECK_PARSE_EQ("(\\1a)", "(^ 'a')"); |
| 198 |
CHECK_PARSE_EQ("(?=a)?a", "'a'"); |
| 199 |
CHECK_PARSE_EQ("(?=a){0,10}a", "'a'"); |
| 200 |
CHECK_PARSE_EQ("(?=a){1,10}a", "(: (-> + 'a') 'a')"); |
| 201 |
CHECK_PARSE_EQ("(?=a){9,10}a", "(: (-> + 'a') 'a')"); |
| 202 |
CHECK_PARSE_EQ("(?!a)?a", "'a'"); |
| 203 |
CHECK_PARSE_EQ("\\1(a)", "(^ 'a')"); |
| 204 |
CHECK_PARSE_EQ("(?!(a))\\1", "(-> - (^ 'a'))"); |
| 205 |
CHECK_PARSE_EQ("(?!\\1(a\\1)\\1)\\1", "(-> - (: (^ 'a') (<- 1)))"); |
| 206 |
CHECK_PARSE_EQ("[\\0]", "[\\x00]"); |
| 207 |
CHECK_PARSE_EQ("[\\11]", "[\\x09]"); |
| 208 |
CHECK_PARSE_EQ("[\\11a]", "[\\x09 a]"); |
| 209 |
CHECK_PARSE_EQ("[\\011]", "[\\x09]"); |
| 210 |
CHECK_PARSE_EQ("[\\00011]", "[\\x00 1 1]"); |
| 211 |
CHECK_PARSE_EQ("[\\118]", "[\\x09 8]"); |
| 212 |
CHECK_PARSE_EQ("[\\111]", "[I]"); |
| 213 |
CHECK_PARSE_EQ("[\\1111]", "[I 1]"); |
| 214 |
CHECK_PARSE_EQ("\\x34", "'\x34'"); |
| 215 |
CHECK_PARSE_EQ("\\x60", "'\x60'"); |
| 216 |
CHECK_PARSE_EQ("\\x3z", "'x3z'"); |
| 217 |
CHECK_PARSE_EQ("\\c", "'c'"); |
| 218 |
CHECK_PARSE_EQ("\\u0034", "'\x34'"); |
| 219 |
CHECK_PARSE_EQ("\\u003z", "'u003z'"); |
| 220 |
CHECK_PARSE_EQ("foo[z]*", "(: 'foo' (# 0 - g [z]))"); |
| 221 |
|
| 222 |
CHECK_SIMPLE("a", true); |
| 223 |
CHECK_SIMPLE("a|b", false); |
| 224 |
CHECK_SIMPLE("a\\n", false); |
| 225 |
CHECK_SIMPLE("^a", false); |
| 226 |
CHECK_SIMPLE("a$", false); |
| 227 |
CHECK_SIMPLE("a\\b!", false); |
| 228 |
CHECK_SIMPLE("a\\Bb", false); |
| 229 |
CHECK_SIMPLE("a*", false); |
| 230 |
CHECK_SIMPLE("a*?", false); |
| 231 |
CHECK_SIMPLE("a?", false); |
| 232 |
CHECK_SIMPLE("a??", false); |
| 233 |
CHECK_SIMPLE("a{0,1}?", false); |
| 234 |
CHECK_SIMPLE("a{1,1}?", false); |
| 235 |
CHECK_SIMPLE("a{1,2}?", false); |
| 236 |
CHECK_SIMPLE("a+?", false); |
| 237 |
CHECK_SIMPLE("(a)", false); |
| 238 |
CHECK_SIMPLE("(a)\\1", false); |
| 239 |
CHECK_SIMPLE("(\\1a)", false); |
| 240 |
CHECK_SIMPLE("\\1(a)", false); |
| 241 |
CHECK_SIMPLE("a\\s", false); |
| 242 |
CHECK_SIMPLE("a\\S", false); |
| 243 |
CHECK_SIMPLE("a\\d", false); |
| 244 |
CHECK_SIMPLE("a\\D", false); |
| 245 |
CHECK_SIMPLE("a\\w", false); |
| 246 |
CHECK_SIMPLE("a\\W", false); |
| 247 |
CHECK_SIMPLE("a.", false); |
| 248 |
CHECK_SIMPLE("a\\q", false); |
| 249 |
CHECK_SIMPLE("a[a]", false); |
| 250 |
CHECK_SIMPLE("a[^a]", false); |
| 251 |
CHECK_SIMPLE("a[a-z]", false); |
| 252 |
CHECK_SIMPLE("a[\\q]", false); |
| 253 |
CHECK_SIMPLE("a(?:b)", false); |
| 254 |
CHECK_SIMPLE("a(?=b)", false); |
| 255 |
CHECK_SIMPLE("a(?!b)", false); |
| 256 |
CHECK_SIMPLE("\\x60", false); |
| 257 |
CHECK_SIMPLE("\\u0060", false); |
| 258 |
CHECK_SIMPLE("\\cA", false); |
| 259 |
CHECK_SIMPLE("\\q", false); |
| 260 |
CHECK_SIMPLE("\\1112", false); |
| 261 |
CHECK_SIMPLE("\\0", false); |
| 262 |
CHECK_SIMPLE("(a)\\1", false); |
| 263 |
CHECK_SIMPLE("(?=a)?a", false); |
| 264 |
CHECK_SIMPLE("(?!a)?a\\1", false); |
| 265 |
CHECK_SIMPLE("(?:(?=a))a\\1", false); |
| 266 |
|
| 267 |
CHECK_PARSE_EQ("a{}", "'a{}'"); |
| 268 |
CHECK_PARSE_EQ("a{,}", "'a{,}'"); |
| 269 |
CHECK_PARSE_EQ("a{", "'a{'"); |
| 270 |
CHECK_PARSE_EQ("a{z}", "'a{z}'"); |
| 271 |
CHECK_PARSE_EQ("a{1z}", "'a{1z}'"); |
| 272 |
CHECK_PARSE_EQ("a{12z}", "'a{12z}'"); |
| 273 |
CHECK_PARSE_EQ("a{12,", "'a{12,'"); |
| 274 |
CHECK_PARSE_EQ("a{12,3b", "'a{12,3b'"); |
| 275 |
CHECK_PARSE_EQ("{}", "'{}'"); |
| 276 |
CHECK_PARSE_EQ("{,}", "'{,}'"); |
| 277 |
CHECK_PARSE_EQ("{", "'{'"); |
| 278 |
CHECK_PARSE_EQ("{z}", "'{z}'"); |
| 279 |
CHECK_PARSE_EQ("{1z}", "'{1z}'"); |
| 280 |
CHECK_PARSE_EQ("{12z}", "'{12z}'"); |
| 281 |
CHECK_PARSE_EQ("{12,", "'{12,'"); |
| 282 |
CHECK_PARSE_EQ("{12,3b", "'{12,3b'"); |
| 283 |
|
| 284 |
CHECK_MIN_MAX("a", 1, 1); |
| 285 |
CHECK_MIN_MAX("abc", 3, 3); |
| 286 |
CHECK_MIN_MAX("a[bc]d", 3, 3); |
| 287 |
CHECK_MIN_MAX("a|bc", 1, 2); |
| 288 |
CHECK_MIN_MAX("ab|c", 1, 2); |
| 289 |
CHECK_MIN_MAX("a||bc", 0, 2); |
| 290 |
CHECK_MIN_MAX("|", 0, 0); |
| 291 |
CHECK_MIN_MAX("(?:ab)", 2, 2); |
| 292 |
CHECK_MIN_MAX("(?:ab|cde)", 2, 3); |
| 293 |
CHECK_MIN_MAX("(?:ab)|cde", 2, 3); |
| 294 |
CHECK_MIN_MAX("(ab)", 2, 2); |
| 295 |
CHECK_MIN_MAX("(ab|cde)", 2, 3); |
| 296 |
CHECK_MIN_MAX("(ab)\\1", 2, 4); |
| 297 |
CHECK_MIN_MAX("(ab|cde)\\1", 2, 6); |
| 298 |
CHECK_MIN_MAX("(?:ab)?", 0, 2); |
| 299 |
CHECK_MIN_MAX("(?:ab)*", 0, RegExpTree::kInfinity); |
| 300 |
CHECK_MIN_MAX("(?:ab)+", 2, RegExpTree::kInfinity); |
| 301 |
CHECK_MIN_MAX("a?", 0, 1); |
| 302 |
CHECK_MIN_MAX("a*", 0, RegExpTree::kInfinity); |
| 303 |
CHECK_MIN_MAX("a+", 1, RegExpTree::kInfinity); |
| 304 |
CHECK_MIN_MAX("a??", 0, 1); |
| 305 |
CHECK_MIN_MAX("a*?", 0, RegExpTree::kInfinity); |
| 306 |
CHECK_MIN_MAX("a+?", 1, RegExpTree::kInfinity); |
| 307 |
CHECK_MIN_MAX("(?:a?)?", 0, 1); |
| 308 |
CHECK_MIN_MAX("(?:a*)?", 0, RegExpTree::kInfinity); |
| 309 |
CHECK_MIN_MAX("(?:a+)?", 0, RegExpTree::kInfinity); |
| 310 |
CHECK_MIN_MAX("(?:a?)+", 0, RegExpTree::kInfinity); |
| 311 |
CHECK_MIN_MAX("(?:a*)+", 0, RegExpTree::kInfinity); |
| 312 |
CHECK_MIN_MAX("(?:a+)+", 1, RegExpTree::kInfinity); |
| 313 |
CHECK_MIN_MAX("(?:a?)*", 0, RegExpTree::kInfinity); |
| 314 |
CHECK_MIN_MAX("(?:a*)*", 0, RegExpTree::kInfinity); |
| 315 |
CHECK_MIN_MAX("(?:a+)*", 0, RegExpTree::kInfinity); |
| 316 |
CHECK_MIN_MAX("a{0}", 0, 0); |
| 317 |
CHECK_MIN_MAX("(?:a+){0}", 0, 0); |
| 318 |
CHECK_MIN_MAX("(?:a+){0,0}", 0, 0); |
| 319 |
CHECK_MIN_MAX("a*b", 1, RegExpTree::kInfinity); |
| 320 |
CHECK_MIN_MAX("a+b", 2, RegExpTree::kInfinity); |
| 321 |
CHECK_MIN_MAX("a*b|c", 1, RegExpTree::kInfinity); |
| 322 |
CHECK_MIN_MAX("a+b|c", 1, RegExpTree::kInfinity); |
| 323 |
CHECK_MIN_MAX("(?:a{5,1000000}){3,1000000}", 15, RegExpTree::kInfinity); |
| 324 |
CHECK_MIN_MAX("(?:ab){4,7}", 8, 14); |
| 325 |
CHECK_MIN_MAX("a\\bc", 2, 2); |
| 326 |
CHECK_MIN_MAX("a\\Bc", 2, 2); |
| 327 |
CHECK_MIN_MAX("a\\sc", 3, 3); |
| 328 |
CHECK_MIN_MAX("a\\Sc", 3, 3); |
| 329 |
CHECK_MIN_MAX("a(?=b)c", 2, 2); |
| 330 |
CHECK_MIN_MAX("a(?=bbb|bb)c", 2, 2); |
| 331 |
CHECK_MIN_MAX("a(?!bbb|bb)c", 2, 2); |
| 332 |
} |
| 333 |
|
| 334 |
TEST(ParserRegression) {
|
| 335 |
CHECK_PARSE_EQ("[A-Z$-][x]", "(! [A-Z $ -] [x])"); |
| 336 |
CHECK_PARSE_EQ("a{3,4*}", "(: 'a{3,' (# 0 - g '4') '}')"); |
| 337 |
CHECK_PARSE_EQ("{", "'{'"); |
| 338 |
CHECK_PARSE_EQ("a|", "(| 'a' %)"); |
| 339 |
} |
| 340 |
|
| 341 |
static void ExpectError(const char* input, |
| 342 |
const char* expected) { |
| 343 |
V8::Initialize(NULL);
|
| 344 |
v8::HandleScope scope; |
| 345 |
ZoneScope zone_scope(DELETE_ON_EXIT); |
| 346 |
FlatStringReader reader(CStrVector(input)); |
| 347 |
RegExpCompileData result; |
| 348 |
CHECK_EQ(false, v8::internal::ParseRegExp(&reader, false, &result)); |
| 349 |
CHECK(result.tree == NULL);
|
| 350 |
CHECK(!result.error.is_null()); |
| 351 |
SmartPointer<char> str = result.error->ToCString(ALLOW_NULLS);
|
| 352 |
CHECK_EQ(expected, *str); |
| 353 |
} |
| 354 |
|
| 355 |
|
| 356 |
TEST(Errors) {
|
| 357 |
V8::Initialize(NULL);
|
| 358 |
const char* kEndBackslash = "\\ at end of pattern"; |
| 359 |
ExpectError("\\", kEndBackslash);
|
| 360 |
const char* kUnterminatedGroup = "Unterminated group"; |
| 361 |
ExpectError("(foo", kUnterminatedGroup);
|
| 362 |
const char* kInvalidGroup = "Invalid group"; |
| 363 |
ExpectError("(?", kInvalidGroup);
|
| 364 |
const char* kUnterminatedCharacterClass = "Unterminated character class"; |
| 365 |
ExpectError("[", kUnterminatedCharacterClass);
|
| 366 |
ExpectError("[a-", kUnterminatedCharacterClass);
|
| 367 |
const char* kNothingToRepeat = "Nothing to repeat"; |
| 368 |
ExpectError("*", kNothingToRepeat);
|
| 369 |
ExpectError("?", kNothingToRepeat);
|
| 370 |
ExpectError("+", kNothingToRepeat);
|
| 371 |
ExpectError("{1}", kNothingToRepeat);
|
| 372 |
ExpectError("{1,2}", kNothingToRepeat);
|
| 373 |
ExpectError("{1,}", kNothingToRepeat);
|
| 374 |
|
| 375 |
// Check that we don't allow more than kMaxCapture captures
|
| 376 |
const int kMaxCaptures = 1 << 16; // Must match RegExpParser::kMaxCaptures. |
| 377 |
const char* kTooManyCaptures = "Too many captures"; |
| 378 |
HeapStringAllocator allocator; |
| 379 |
StringStream accumulator(&allocator); |
| 380 |
for (int i = 0; i <= kMaxCaptures; i++) { |
| 381 |
accumulator.Add("()");
|
| 382 |
} |
| 383 |
SmartPointer<const char> many_captures(accumulator.ToCString()); |
| 384 |
ExpectError(*many_captures, kTooManyCaptures); |
| 385 |
} |
| 386 |
|
| 387 |
|
| 388 |
static bool IsDigit(uc16 c) { |
| 389 |
return ('0' <= c && c <= '9'); |
| 390 |
} |
| 391 |
|
| 392 |
|
| 393 |
static bool NotDigit(uc16 c) { |
| 394 |
return !IsDigit(c);
|
| 395 |
} |
| 396 |
|
| 397 |
|
| 398 |
static bool IsWhiteSpace(uc16 c) { |
| 399 |
switch (c) {
|
| 400 |
case 0x09: |
| 401 |
case 0x0A: |
| 402 |
case 0x0B: |
| 403 |
case 0x0C: |
| 404 |
case 0x0d: |
| 405 |
case 0x20: |
| 406 |
case 0xA0: |
| 407 |
case 0x2028: |
| 408 |
case 0x2029: |
| 409 |
return true; |
| 410 |
default:
|
| 411 |
return unibrow::Space::Is(c);
|
| 412 |
} |
| 413 |
} |
| 414 |
|
| 415 |
|
| 416 |
static bool NotWhiteSpace(uc16 c) { |
| 417 |
return !IsWhiteSpace(c);
|
| 418 |
} |
| 419 |
|
| 420 |
|
| 421 |
static bool NotWord(uc16 c) { |
| 422 |
return !IsRegExpWord(c);
|
| 423 |
} |
| 424 |
|
| 425 |
|
| 426 |
static void TestCharacterClassEscapes(uc16 c, bool (pred)(uc16 c)) { |
| 427 |
ZoneScope scope(DELETE_ON_EXIT); |
| 428 |
ZoneList<CharacterRange>* ranges = new ZoneList<CharacterRange>(2); |
| 429 |
CharacterRange::AddClassEscape(c, ranges); |
| 430 |
for (unsigned i = 0; i < (1 << 16); i++) { |
| 431 |
bool in_class = false; |
| 432 |
for (int j = 0; !in_class && j < ranges->length(); j++) { |
| 433 |
CharacterRange& range = ranges->at(j); |
| 434 |
in_class = (range.from() <= i && i <= range.to()); |
| 435 |
} |
| 436 |
CHECK_EQ(pred(i), in_class); |
| 437 |
} |
| 438 |
} |
| 439 |
|
| 440 |
|
| 441 |
TEST(CharacterClassEscapes) {
|
| 442 |
TestCharacterClassEscapes('.', IsRegExpNewline);
|
| 443 |
TestCharacterClassEscapes('d', IsDigit);
|
| 444 |
TestCharacterClassEscapes('D', NotDigit);
|
| 445 |
TestCharacterClassEscapes('s', IsWhiteSpace);
|
| 446 |
TestCharacterClassEscapes('S', NotWhiteSpace);
|
| 447 |
TestCharacterClassEscapes('w', IsRegExpWord);
|
| 448 |
TestCharacterClassEscapes('W', NotWord);
|
| 449 |
} |
| 450 |
|
| 451 |
|
| 452 |
static RegExpNode* Compile(const char* input, bool multiline, bool is_ascii) { |
| 453 |
V8::Initialize(NULL);
|
| 454 |
FlatStringReader reader(CStrVector(input)); |
| 455 |
RegExpCompileData compile_data; |
| 456 |
if (!v8::internal::ParseRegExp(&reader, multiline, &compile_data))
|
| 457 |
return NULL; |
| 458 |
Handle<String> pattern = Factory::NewStringFromUtf8(CStrVector(input)); |
| 459 |
RegExpEngine::Compile(&compile_data, false, multiline, pattern, is_ascii);
|
| 460 |
return compile_data.node;
|
| 461 |
} |
| 462 |
|
| 463 |
|
| 464 |
static void Execute(const char* input, |
| 465 |
bool multiline,
|
| 466 |
bool is_ascii,
|
| 467 |
bool dot_output = false) { |
| 468 |
v8::HandleScope scope; |
| 469 |
ZoneScope zone_scope(DELETE_ON_EXIT); |
| 470 |
RegExpNode* node = Compile(input, multiline, is_ascii); |
| 471 |
USE(node); |
| 472 |
#ifdef DEBUG
|
| 473 |
if (dot_output) {
|
| 474 |
RegExpEngine::DotPrint(input, node, false);
|
| 475 |
exit(0);
|
| 476 |
} |
| 477 |
#endif // DEBUG |
| 478 |
} |
| 479 |
|
| 480 |
|
| 481 |
class TestConfig { |
| 482 |
public:
|
| 483 |
typedef int Key; |
| 484 |
typedef int Value; |
| 485 |
static const int kNoKey; |
| 486 |
static const int kNoValue; |
| 487 |
static inline int Compare(int a, int b) { |
| 488 |
if (a < b)
|
| 489 |
return -1; |
| 490 |
else if (a > b) |
| 491 |
return 1; |
| 492 |
else
|
| 493 |
return 0; |
| 494 |
} |
| 495 |
}; |
| 496 |
|
| 497 |
|
| 498 |
const int TestConfig::kNoKey = 0; |
| 499 |
const int TestConfig::kNoValue = 0; |
| 500 |
|
| 501 |
|
| 502 |
static int PseudoRandom(int i, int j) { |
| 503 |
return ~(~((i * 781) ^ (j * 329))); |
| 504 |
} |
| 505 |
|
| 506 |
|
| 507 |
TEST(SplayTreeSimple) {
|
| 508 |
static const int kLimit = 1000; |
| 509 |
ZoneScope zone_scope(DELETE_ON_EXIT); |
| 510 |
ZoneSplayTree<TestConfig> tree; |
| 511 |
std::set<int> seen;
|
| 512 |
#define CHECK_MAPS_EQUAL() do { \ |
| 513 |
for (int k = 0; k < kLimit; k++) \ |
| 514 |
CHECK_EQ(seen.find(k) != seen.end(), tree.Find(k, &loc)); \ |
| 515 |
} while (false) |
| 516 |
for (int i = 0; i < 50; i++) { |
| 517 |
for (int j = 0; j < 50; j++) { |
| 518 |
int next = PseudoRandom(i, j) % kLimit;
|
| 519 |
if (seen.find(next) != seen.end()) {
|
| 520 |
// We've already seen this one. Check the value and remove
|
| 521 |
// it.
|
| 522 |
ZoneSplayTree<TestConfig>::Locator loc; |
| 523 |
CHECK(tree.Find(next, &loc)); |
| 524 |
CHECK_EQ(next, loc.key()); |
| 525 |
CHECK_EQ(3 * next, loc.value());
|
| 526 |
tree.Remove(next); |
| 527 |
seen.erase(next); |
| 528 |
CHECK_MAPS_EQUAL(); |
| 529 |
} else {
|
| 530 |
// Check that it wasn't there already and then add it.
|
| 531 |
ZoneSplayTree<TestConfig>::Locator loc; |
| 532 |
CHECK(!tree.Find(next, &loc)); |
| 533 |
CHECK(tree.Insert(next, &loc)); |
| 534 |
CHECK_EQ(next, loc.key()); |
| 535 |
loc.set_value(3 * next);
|
| 536 |
seen.insert(next); |
| 537 |
CHECK_MAPS_EQUAL(); |
| 538 |
} |
| 539 |
int val = PseudoRandom(j, i) % kLimit;
|
| 540 |
for (int k = val; k >= 0; k--) { |
| 541 |
if (seen.find(val) != seen.end()) {
|
| 542 |
ZoneSplayTree<TestConfig>::Locator loc; |
| 543 |
CHECK(tree.FindGreatestLessThan(val, &loc)); |
| 544 |
CHECK_EQ(loc.key(), val); |
| 545 |
break;
|
| 546 |
} |
| 547 |
} |
| 548 |
val = PseudoRandom(i + j, i - j) % kLimit; |
| 549 |
for (int k = val; k < kLimit; k++) { |
| 550 |
if (seen.find(val) != seen.end()) {
|
| 551 |
ZoneSplayTree<TestConfig>::Locator loc; |
| 552 |
CHECK(tree.FindLeastGreaterThan(val, &loc)); |
| 553 |
CHECK_EQ(loc.key(), val); |
| 554 |
break;
|
| 555 |
} |
| 556 |
} |
| 557 |
} |
| 558 |
} |
| 559 |
} |
| 560 |
|
| 561 |
|
| 562 |
TEST(DispatchTableConstruction) {
|
| 563 |
// Initialize test data.
|
| 564 |
static const int kLimit = 1000; |
| 565 |
static const int kRangeCount = 8; |
| 566 |
static const int kRangeSize = 16; |
| 567 |
uc16 ranges[kRangeCount][2 * kRangeSize];
|
| 568 |
for (int i = 0; i < kRangeCount; i++) { |
| 569 |
Vector<uc16> range(ranges[i], 2 * kRangeSize);
|
| 570 |
for (int j = 0; j < 2 * kRangeSize; j++) { |
| 571 |
range[j] = PseudoRandom(i + 25, j + 87) % kLimit; |
| 572 |
} |
| 573 |
range.Sort(); |
| 574 |
for (int j = 1; j < 2 * kRangeSize; j++) { |
| 575 |
CHECK(range[j-1] <= range[j]);
|
| 576 |
} |
| 577 |
} |
| 578 |
// Enter test data into dispatch table.
|
| 579 |
ZoneScope zone_scope(DELETE_ON_EXIT); |
| 580 |
DispatchTable table; |
| 581 |
for (int i = 0; i < kRangeCount; i++) { |
| 582 |
uc16* range = ranges[i]; |
| 583 |
for (int j = 0; j < 2 * kRangeSize; j += 2) |
| 584 |
table.AddRange(CharacterRange(range[j], range[j + 1]), i);
|
| 585 |
} |
| 586 |
// Check that the table looks as we would expect
|
| 587 |
for (int p = 0; p < kLimit; p++) { |
| 588 |
OutSet* outs = table.Get(p); |
| 589 |
for (int j = 0; j < kRangeCount; j++) { |
| 590 |
uc16* range = ranges[j]; |
| 591 |
bool is_on = false; |
| 592 |
for (int k = 0; !is_on && (k < 2 * kRangeSize); k += 2) |
| 593 |
is_on = (range[k] <= p && p <= range[k + 1]);
|
| 594 |
CHECK_EQ(is_on, outs->Get(j)); |
| 595 |
} |
| 596 |
} |
| 597 |
} |
| 598 |
|
| 599 |
|
| 600 |
TEST(MacroAssembler) {
|
| 601 |
V8::Initialize(NULL);
|
| 602 |
byte codes[1024];
|
| 603 |
RegExpMacroAssemblerIrregexp m(Vector<byte>(codes, 1024));
|
| 604 |
// ^f(o)o.
|
| 605 |
Label fail, fail2, start; |
| 606 |
uc16 foo_chars[3];
|
| 607 |
foo_chars[0] = 'f'; |
| 608 |
foo_chars[1] = 'o'; |
| 609 |
foo_chars[2] = 'o'; |
| 610 |
Vector<const uc16> foo(foo_chars, 3); |
| 611 |
m.SetRegister(4, 42); |
| 612 |
m.PushRegister(4, RegExpMacroAssembler::kNoStackLimitCheck);
|
| 613 |
m.AdvanceRegister(4, 42); |
| 614 |
m.GoTo(&start); |
| 615 |
m.Fail(); |
| 616 |
m.Bind(&start); |
| 617 |
m.PushBacktrack(&fail2); |
| 618 |
m.CheckCharacters(foo, 0, &fail, true); |
| 619 |
m.WriteCurrentPositionToRegister(0, 0); |
| 620 |
m.PushCurrentPosition(); |
| 621 |
m.AdvanceCurrentPosition(3);
|
| 622 |
m.WriteCurrentPositionToRegister(1, 0); |
| 623 |
m.PopCurrentPosition(); |
| 624 |
m.AdvanceCurrentPosition(1);
|
| 625 |
m.WriteCurrentPositionToRegister(2, 0); |
| 626 |
m.AdvanceCurrentPosition(1);
|
| 627 |
m.WriteCurrentPositionToRegister(3, 0); |
| 628 |
m.Succeed(); |
| 629 |
|
| 630 |
m.Bind(&fail); |
| 631 |
m.Backtrack(); |
| 632 |
m.Succeed(); |
| 633 |
|
| 634 |
m.Bind(&fail2); |
| 635 |
m.PopRegister(0);
|
| 636 |
m.Fail(); |
| 637 |
|
| 638 |
v8::HandleScope scope; |
| 639 |
|
| 640 |
Handle<String> source = Factory::NewStringFromAscii(CStrVector("^f(o)o"));
|
| 641 |
Handle<ByteArray> array = Handle<ByteArray>::cast(m.GetCode(source)); |
| 642 |
int captures[5]; |
| 643 |
|
| 644 |
const uc16 str1[] = {'f', 'o', 'o', 'b', 'a', 'r'}; |
| 645 |
Handle<String> f1_16 = |
| 646 |
Factory::NewStringFromTwoByte(Vector<const uc16>(str1, 6)); |
| 647 |
|
| 648 |
CHECK(IrregexpInterpreter::Match(array, f1_16, captures, 0));
|
| 649 |
CHECK_EQ(0, captures[0]); |
| 650 |
CHECK_EQ(3, captures[1]); |
| 651 |
CHECK_EQ(1, captures[2]); |
| 652 |
CHECK_EQ(2, captures[3]); |
| 653 |
CHECK_EQ(84, captures[4]); |
| 654 |
|
| 655 |
const uc16 str2[] = {'b', 'a', 'r', 'f', 'o', 'o'}; |
| 656 |
Handle<String> f2_16 = |
| 657 |
Factory::NewStringFromTwoByte(Vector<const uc16>(str2, 6)); |
| 658 |
|
| 659 |
CHECK(!IrregexpInterpreter::Match(array, f2_16, captures, 0));
|
| 660 |
CHECK_EQ(42, captures[0]); |
| 661 |
} |
| 662 |
|
| 663 |
|
| 664 |
#ifndef ARM // IA32 only tests. |
| 665 |
|
| 666 |
class ContextInitializer { |
| 667 |
public:
|
| 668 |
ContextInitializer() : env_(), scope_(), stack_guard_() {
|
| 669 |
env_ = v8::Context::New(); |
| 670 |
env_->Enter(); |
| 671 |
} |
| 672 |
~ContextInitializer() {
|
| 673 |
env_->Exit(); |
| 674 |
env_.Dispose(); |
| 675 |
} |
| 676 |
private:
|
| 677 |
v8::Persistent<v8::Context> env_; |
| 678 |
v8::HandleScope scope_; |
| 679 |
v8::internal::StackGuard stack_guard_; |
| 680 |
}; |
| 681 |
|
| 682 |
|
| 683 |
static RegExpMacroAssemblerIA32::Result ExecuteIA32(Code* code,
|
| 684 |
String* input, |
| 685 |
int start_offset,
|
| 686 |
const byte* input_start,
|
| 687 |
const byte* input_end,
|
| 688 |
int* captures,
|
| 689 |
bool at_start) {
|
| 690 |
return RegExpMacroAssemblerIA32::Execute(
|
| 691 |
code, |
| 692 |
input, |
| 693 |
start_offset, |
| 694 |
input_start, |
| 695 |
input_end, |
| 696 |
captures, |
| 697 |
at_start); |
| 698 |
} |
| 699 |
|
| 700 |
|
| 701 |
TEST(MacroAssemblerIA32Success) {
|
| 702 |
v8::V8::Initialize(); |
| 703 |
ContextInitializer initializer; |
| 704 |
|
| 705 |
RegExpMacroAssemblerIA32 m(RegExpMacroAssemblerIA32::ASCII, 4);
|
| 706 |
|
| 707 |
m.Succeed(); |
| 708 |
|
| 709 |
Handle<String> source = Factory::NewStringFromAscii(CStrVector(""));
|
| 710 |
Handle<Object> code_object = m.GetCode(source); |
| 711 |
Handle<Code> code = Handle<Code>::cast(code_object); |
| 712 |
|
| 713 |
int captures[4] = {42, 37, 87, 117}; |
| 714 |
Handle<String> input = Factory::NewStringFromAscii(CStrVector("foofoo"));
|
| 715 |
Handle<SeqAsciiString> seq_input = Handle<SeqAsciiString>::cast(input); |
| 716 |
const byte* start_adr =
|
| 717 |
reinterpret_cast<const byte*>(seq_input->GetCharsAddress()); |
| 718 |
|
| 719 |
RegExpMacroAssemblerIA32::Result result = |
| 720 |
ExecuteIA32(*code, |
| 721 |
*input, |
| 722 |
0,
|
| 723 |
start_adr, |
| 724 |
start_adr + seq_input->length(), |
| 725 |
captures, |
| 726 |
true);
|
| 727 |
|
| 728 |
CHECK_EQ(RegExpMacroAssemblerIA32::SUCCESS, result); |
| 729 |
CHECK_EQ(-1, captures[0]); |
| 730 |
CHECK_EQ(-1, captures[1]); |
| 731 |
CHECK_EQ(-1, captures[2]); |
| 732 |
CHECK_EQ(-1, captures[3]); |
| 733 |
} |
| 734 |
|
| 735 |
|
| 736 |
TEST(MacroAssemblerIA32Simple) {
|
| 737 |
v8::V8::Initialize(); |
| 738 |
ContextInitializer initializer; |
| 739 |
|
| 740 |
RegExpMacroAssemblerIA32 m(RegExpMacroAssemblerIA32::ASCII, 4);
|
| 741 |
|
| 742 |
uc16 foo_chars[3] = {'f', 'o', 'o'}; |
| 743 |
Vector<const uc16> foo(foo_chars, 3); |
| 744 |
|
| 745 |
Label fail; |
| 746 |
m.CheckCharacters(foo, 0, &fail, true); |
| 747 |
m.WriteCurrentPositionToRegister(0, 0); |
| 748 |
m.AdvanceCurrentPosition(3);
|
| 749 |
m.WriteCurrentPositionToRegister(1, 0); |
| 750 |
m.Succeed(); |
| 751 |
m.Bind(&fail); |
| 752 |
m.Fail(); |
| 753 |
|
| 754 |
Handle<String> source = Factory::NewStringFromAscii(CStrVector("^foo"));
|
| 755 |
Handle<Object> code_object = m.GetCode(source); |
| 756 |
Handle<Code> code = Handle<Code>::cast(code_object); |
| 757 |
|
| 758 |
int captures[4] = {42, 37, 87, 117}; |
| 759 |
Handle<String> input = Factory::NewStringFromAscii(CStrVector("foofoo"));
|
| 760 |
Handle<SeqAsciiString> seq_input = Handle<SeqAsciiString>::cast(input); |
| 761 |
Address start_adr = seq_input->GetCharsAddress(); |
| 762 |
|
| 763 |
RegExpMacroAssemblerIA32::Result result = |
| 764 |
ExecuteIA32(*code, |
| 765 |
*input, |
| 766 |
0,
|
| 767 |
start_adr, |
| 768 |
start_adr + input->length(), |
| 769 |
captures, |
| 770 |
true);
|
| 771 |
|
| 772 |
CHECK_EQ(RegExpMacroAssemblerIA32::SUCCESS, result); |
| 773 |
CHECK_EQ(0, captures[0]); |
| 774 |
CHECK_EQ(3, captures[1]); |
| 775 |
CHECK_EQ(-1, captures[2]); |
| 776 |
CHECK_EQ(-1, captures[3]); |
| 777 |
|
| 778 |
input = Factory::NewStringFromAscii(CStrVector("barbarbar"));
|
| 779 |
seq_input = Handle<SeqAsciiString>::cast(input); |
| 780 |
start_adr = seq_input->GetCharsAddress(); |
| 781 |
|
| 782 |
result = ExecuteIA32(*code, |
| 783 |
*input, |
| 784 |
0,
|
| 785 |
start_adr, |
| 786 |
start_adr + input->length(), |
| 787 |
captures, |
| 788 |
true);
|
| 789 |
|
| 790 |
CHECK_EQ(RegExpMacroAssemblerIA32::FAILURE, result); |
| 791 |
} |
| 792 |
|
| 793 |
|
| 794 |
TEST(MacroAssemblerIA32SimpleUC16) {
|
| 795 |
v8::V8::Initialize(); |
| 796 |
ContextInitializer initializer; |
| 797 |
|
| 798 |
RegExpMacroAssemblerIA32 m(RegExpMacroAssemblerIA32::UC16, 4);
|
| 799 |
|
| 800 |
uc16 foo_chars[3] = {'f', 'o', 'o'}; |
| 801 |
Vector<const uc16> foo(foo_chars, 3); |
| 802 |
|
| 803 |
Label fail; |
| 804 |
m.CheckCharacters(foo, 0, &fail, true); |
| 805 |
m.WriteCurrentPositionToRegister(0, 0); |
| 806 |
m.AdvanceCurrentPosition(3);
|
| 807 |
m.WriteCurrentPositionToRegister(1, 0); |
| 808 |
m.Succeed(); |
| 809 |
m.Bind(&fail); |
| 810 |
m.Fail(); |
| 811 |
|
| 812 |
Handle<String> source = Factory::NewStringFromAscii(CStrVector("^foo"));
|
| 813 |
Handle<Object> code_object = m.GetCode(source); |
| 814 |
Handle<Code> code = Handle<Code>::cast(code_object); |
| 815 |
|
| 816 |
int captures[4] = {42, 37, 87, 117}; |
| 817 |
const uc16 input_data[6] = {'f', 'o', 'o', 'f', 'o', '\xa0'}; |
| 818 |
Handle<String> input = |
| 819 |
Factory::NewStringFromTwoByte(Vector<const uc16>(input_data, 6)); |
| 820 |
Handle<SeqTwoByteString> seq_input = Handle<SeqTwoByteString>::cast(input); |
| 821 |
Address start_adr = seq_input->GetCharsAddress(); |
| 822 |
|
| 823 |
RegExpMacroAssemblerIA32::Result result = |
| 824 |
ExecuteIA32(*code, |
| 825 |
*input, |
| 826 |
0,
|
| 827 |
start_adr, |
| 828 |
start_adr + input->length(), |
| 829 |
captures, |
| 830 |
true);
|
| 831 |
|
| 832 |
CHECK_EQ(RegExpMacroAssemblerIA32::SUCCESS, result); |
| 833 |
CHECK_EQ(0, captures[0]); |
| 834 |
CHECK_EQ(3, captures[1]); |
| 835 |
CHECK_EQ(-1, captures[2]); |
| 836 |
CHECK_EQ(-1, captures[3]); |
| 837 |
|
| 838 |
const uc16 input_data2[9] = {'b', 'a', 'r', 'b', 'a', 'r', 'b', 'a', '\xa0'}; |
| 839 |
input = Factory::NewStringFromTwoByte(Vector<const uc16>(input_data2, 9)); |
| 840 |
seq_input = Handle<SeqTwoByteString>::cast(input); |
| 841 |
start_adr = seq_input->GetCharsAddress(); |
| 842 |
|
| 843 |
result = ExecuteIA32(*code, |
| 844 |
*input, |
| 845 |
0,
|
| 846 |
start_adr, |
| 847 |
start_adr + input->length() * 2,
|
| 848 |
captures, |
| 849 |
true);
|
| 850 |
|
| 851 |
CHECK_EQ(RegExpMacroAssemblerIA32::FAILURE, result); |
| 852 |
} |
| 853 |
|
| 854 |
|
| 855 |
TEST(MacroAssemblerIA32Backtrack) {
|
| 856 |
v8::V8::Initialize(); |
| 857 |
ContextInitializer initializer; |
| 858 |
|
| 859 |
RegExpMacroAssemblerIA32 m(RegExpMacroAssemblerIA32::ASCII, 0);
|
| 860 |
|
| 861 |
Label fail; |
| 862 |
Label backtrack; |
| 863 |
m.LoadCurrentCharacter(10, &fail);
|
| 864 |
m.Succeed(); |
| 865 |
m.Bind(&fail); |
| 866 |
m.PushBacktrack(&backtrack); |
| 867 |
m.LoadCurrentCharacter(10, NULL); |
| 868 |
m.Succeed(); |
| 869 |
m.Bind(&backtrack); |
| 870 |
m.Fail(); |
| 871 |
|
| 872 |
Handle<String> source = Factory::NewStringFromAscii(CStrVector(".........."));
|
| 873 |
Handle<Object> code_object = m.GetCode(source); |
| 874 |
Handle<Code> code = Handle<Code>::cast(code_object); |
| 875 |
|
| 876 |
Handle<String> input = Factory::NewStringFromAscii(CStrVector("foofoo"));
|
| 877 |
Handle<SeqAsciiString> seq_input = Handle<SeqAsciiString>::cast(input); |
| 878 |
Address start_adr = seq_input->GetCharsAddress(); |
| 879 |
|
| 880 |
RegExpMacroAssemblerIA32::Result result = |
| 881 |
ExecuteIA32(*code, |
| 882 |
*input, |
| 883 |
0,
|
| 884 |
start_adr, |
| 885 |
start_adr + input->length(), |
| 886 |
NULL,
|
| 887 |
true);
|
| 888 |
|
| 889 |
CHECK_EQ(RegExpMacroAssemblerIA32::FAILURE, result); |
| 890 |
} |
| 891 |
|
| 892 |
|
| 893 |
TEST(MacroAssemblerIA32BackReferenceASCII) {
|
| 894 |
v8::V8::Initialize(); |
| 895 |
ContextInitializer initializer; |
| 896 |
|
| 897 |
RegExpMacroAssemblerIA32 m(RegExpMacroAssemblerIA32::ASCII, 3);
|
| 898 |
|
| 899 |
m.WriteCurrentPositionToRegister(0, 0); |
| 900 |
m.AdvanceCurrentPosition(2);
|
| 901 |
m.WriteCurrentPositionToRegister(1, 0); |
| 902 |
Label nomatch; |
| 903 |
m.CheckNotBackReference(0, &nomatch);
|
| 904 |
m.Fail(); |
| 905 |
m.Bind(&nomatch); |
| 906 |
m.AdvanceCurrentPosition(2);
|
| 907 |
Label missing_match; |
| 908 |
m.CheckNotBackReference(0, &missing_match);
|
| 909 |
m.WriteCurrentPositionToRegister(2, 0); |
| 910 |
m.Succeed(); |
| 911 |
m.Bind(&missing_match); |
| 912 |
m.Fail(); |
| 913 |
|
| 914 |
Handle<String> source = Factory::NewStringFromAscii(CStrVector("^(..)..\1"));
|
| 915 |
Handle<Object> code_object = m.GetCode(source); |
| 916 |
Handle<Code> code = Handle<Code>::cast(code_object); |
| 917 |
|
| 918 |
Handle<String> input = Factory::NewStringFromAscii(CStrVector("fooofo"));
|
| 919 |
Handle<SeqAsciiString> seq_input = Handle<SeqAsciiString>::cast(input); |
| 920 |
Address start_adr = seq_input->GetCharsAddress(); |
| 921 |
|
| 922 |
int output[3]; |
| 923 |
RegExpMacroAssemblerIA32::Result result = |
| 924 |
ExecuteIA32(*code, |
| 925 |
*input, |
| 926 |
0,
|
| 927 |
start_adr, |
| 928 |
start_adr + input->length(), |
| 929 |
output, |
| 930 |
true);
|
| 931 |
|
| 932 |
CHECK_EQ(RegExpMacroAssemblerIA32::SUCCESS, result); |
| 933 |
CHECK_EQ(0, output[0]); |
| 934 |
CHECK_EQ(2, output[1]); |
| 935 |
CHECK_EQ(6, output[2]); |
| 936 |
} |
| 937 |
|
| 938 |
|
| 939 |
TEST(MacroAssemblerIA32BackReferenceUC16) {
|
| 940 |
v8::V8::Initialize(); |
| 941 |
ContextInitializer initializer; |
| 942 |
|
| 943 |
RegExpMacroAssemblerIA32 m(RegExpMacroAssemblerIA32::UC16, 3);
|
| 944 |
|
| 945 |
m.WriteCurrentPositionToRegister(0, 0); |
| 946 |
m.AdvanceCurrentPosition(2);
|
| 947 |
m.WriteCurrentPositionToRegister(1, 0); |
| 948 |
Label nomatch; |
| 949 |
m.CheckNotBackReference(0, &nomatch);
|
| 950 |
m.Fail(); |
| 951 |
m.Bind(&nomatch); |
| 952 |
m.AdvanceCurrentPosition(2);
|
| 953 |
Label missing_match; |
| 954 |
m.CheckNotBackReference(0, &missing_match);
|
| 955 |
m.WriteCurrentPositionToRegister(2, 0); |
| 956 |
m.Succeed(); |
| 957 |
m.Bind(&missing_match); |
| 958 |
m.Fail(); |
| 959 |
|
| 960 |
Handle<String> source = Factory::NewStringFromAscii(CStrVector("^(..)..\1"));
|
| 961 |
Handle<Object> code_object = m.GetCode(source); |
| 962 |
Handle<Code> code = Handle<Code>::cast(code_object); |
| 963 |
|
| 964 |
const uc16 input_data[6] = {'f', 0x2028, 'o', 'o', 'f', 0x2028}; |
| 965 |
Handle<String> input = |
| 966 |
Factory::NewStringFromTwoByte(Vector<const uc16>(input_data, 6)); |
| 967 |
Handle<SeqTwoByteString> seq_input = Handle<SeqTwoByteString>::cast(input); |
| 968 |
Address start_adr = seq_input->GetCharsAddress(); |
| 969 |
|
| 970 |
int output[3]; |
| 971 |
RegExpMacroAssemblerIA32::Result result = |
| 972 |
ExecuteIA32(*code, |
| 973 |
*input, |
| 974 |
0,
|
| 975 |
start_adr, |
| 976 |
start_adr + input->length() * 2,
|
| 977 |
output, |
| 978 |
true);
|
| 979 |
|
| 980 |
CHECK_EQ(RegExpMacroAssemblerIA32::SUCCESS, result); |
| 981 |
CHECK_EQ(0, output[0]); |
| 982 |
CHECK_EQ(2, output[1]); |
| 983 |
CHECK_EQ(6, output[2]); |
| 984 |
} |
| 985 |
|
| 986 |
|
| 987 |
|
| 988 |
TEST(MacroAssemblerIA32AtStart) {
|
| 989 |
v8::V8::Initialize(); |
| 990 |
ContextInitializer initializer; |
| 991 |
|
| 992 |
RegExpMacroAssemblerIA32 m(RegExpMacroAssemblerIA32::ASCII, 0);
|
| 993 |
|
| 994 |
Label not_at_start, newline, fail; |
| 995 |
m.CheckNotAtStart(¬_at_start); |
| 996 |
// Check that prevchar = '\n' and current = 'f'.
|
| 997 |
m.CheckCharacter('\n', &newline);
|
| 998 |
m.Bind(&fail); |
| 999 |
m.Fail(); |
| 1000 |
m.Bind(&newline); |
| 1001 |
m.LoadCurrentCharacter(0, &fail);
|
| 1002 |
m.CheckNotCharacter('f', &fail);
|
| 1003 |
m.Succeed(); |
| 1004 |
|
| 1005 |
m.Bind(¬_at_start); |
| 1006 |
// Check that prevchar = 'o' and current = 'b'.
|
| 1007 |
Label prevo; |
| 1008 |
m.CheckCharacter('o', &prevo);
|
| 1009 |
m.Fail(); |
| 1010 |
m.Bind(&prevo); |
| 1011 |
m.LoadCurrentCharacter(0, &fail);
|
| 1012 |
m.CheckNotCharacter('b', &fail);
|
| 1013 |
m.Succeed(); |
| 1014 |
|
| 1015 |
Handle<String> source = Factory::NewStringFromAscii(CStrVector("(^f|ob)"));
|
| 1016 |
Handle<Object> code_object = m.GetCode(source); |
| 1017 |
Handle<Code> code = Handle<Code>::cast(code_object); |
| 1018 |
|
| 1019 |
Handle<String> input = Factory::NewStringFromAscii(CStrVector("foobar"));
|
| 1020 |
Handle<SeqAsciiString> seq_input = Handle<SeqAsciiString>::cast(input); |
| 1021 |
Address start_adr = seq_input->GetCharsAddress(); |
| 1022 |
|
| 1023 |
RegExpMacroAssemblerIA32::Result result = |
| 1024 |
ExecuteIA32(*code, |
| 1025 |
*input, |
| 1026 |
0,
|
| 1027 |
start_adr, |
| 1028 |
start_adr + input->length(), |
| 1029 |
NULL,
|
| 1030 |
true);
|
| 1031 |
|
| 1032 |
CHECK_EQ(RegExpMacroAssemblerIA32::SUCCESS, result); |
| 1033 |
|
| 1034 |
result = ExecuteIA32(*code, |
| 1035 |
*input, |
| 1036 |
3,
|
| 1037 |
start_adr + 3,
|
| 1038 |
start_adr + input->length(), |
| 1039 |
NULL,
|
| 1040 |
false);
|
| 1041 |
|
| 1042 |
CHECK_EQ(RegExpMacroAssemblerIA32::SUCCESS, result); |
| 1043 |
} |
| 1044 |
|
| 1045 |
|
| 1046 |
TEST(MacroAssemblerIA32BackRefNoCase) {
|
| 1047 |
v8::V8::Initialize(); |
| 1048 |
ContextInitializer initializer; |
| 1049 |
|
| 1050 |
RegExpMacroAssemblerIA32 m(RegExpMacroAssemblerIA32::ASCII, 4);
|
| 1051 |
|
| 1052 |
Label fail, succ; |
| 1053 |
|
| 1054 |
m.WriteCurrentPositionToRegister(0, 0); |
| 1055 |
m.WriteCurrentPositionToRegister(2, 0); |
| 1056 |
m.AdvanceCurrentPosition(3);
|
| 1057 |
m.WriteCurrentPositionToRegister(3, 0); |
| 1058 |
m.CheckNotBackReferenceIgnoreCase(2, &fail); // Match "AbC". |
| 1059 |
m.CheckNotBackReferenceIgnoreCase(2, &fail); // Match "ABC". |
| 1060 |
Label expected_fail; |
| 1061 |
m.CheckNotBackReferenceIgnoreCase(2, &expected_fail);
|
| 1062 |
m.Bind(&fail); |
| 1063 |
m.Fail(); |
| 1064 |
|
| 1065 |
m.Bind(&expected_fail); |
| 1066 |
m.AdvanceCurrentPosition(3); // Skip "xYz" |
| 1067 |
m.CheckNotBackReferenceIgnoreCase(2, &succ);
|
| 1068 |
m.Fail(); |
| 1069 |
|
| 1070 |
m.Bind(&succ); |
| 1071 |
m.WriteCurrentPositionToRegister(1, 0); |
| 1072 |
m.Succeed(); |
| 1073 |
|
| 1074 |
Handle<String> source = |
| 1075 |
Factory::NewStringFromAscii(CStrVector("^(abc)\1\1(?!\1)...(?!\1)"));
|
| 1076 |
Handle<Object> code_object = m.GetCode(source); |
| 1077 |
Handle<Code> code = Handle<Code>::cast(code_object); |
| 1078 |
|
| 1079 |
Handle<String> input = |
| 1080 |
Factory::NewStringFromAscii(CStrVector("aBcAbCABCxYzab"));
|
| 1081 |
Handle<SeqAsciiString> seq_input = Handle<SeqAsciiString>::cast(input); |
| 1082 |
Address start_adr = seq_input->GetCharsAddress(); |
| 1083 |
|
| 1084 |
int output[4]; |
| 1085 |
RegExpMacroAssemblerIA32::Result result = |
| 1086 |
ExecuteIA32(*code, |
| 1087 |
*input, |
| 1088 |
0,
|
| 1089 |
start_adr, |
| 1090 |
start_adr + input->length(), |
| 1091 |
output, |
| 1092 |
true);
|
| 1093 |
|
| 1094 |
CHECK_EQ(RegExpMacroAssemblerIA32::SUCCESS, result); |
| 1095 |
CHECK_EQ(0, output[0]); |
| 1096 |
CHECK_EQ(12, output[1]); |
| 1097 |
CHECK_EQ(0, output[2]); |
| 1098 |
CHECK_EQ(3, output[3]); |
| 1099 |
} |
| 1100 |
|
| 1101 |
|
| 1102 |
|
| 1103 |
TEST(MacroAssemblerIA32Registers) {
|
| 1104 |
v8::V8::Initialize(); |
| 1105 |
ContextInitializer initializer; |
| 1106 |
|
| 1107 |
RegExpMacroAssemblerIA32 m(RegExpMacroAssemblerIA32::ASCII, 5);
|
| 1108 |
|
| 1109 |
uc16 foo_chars[3] = {'f', 'o', 'o'}; |
| 1110 |
Vector<const uc16> foo(foo_chars, 3); |
| 1111 |
|
| 1112 |
enum registers { out1, out2, out3, out4, out5, sp, loop_cnt };
|
| 1113 |
Label fail; |
| 1114 |
Label backtrack; |
| 1115 |
m.WriteCurrentPositionToRegister(out1, 0); // Output: [0] |
| 1116 |
m.PushRegister(out1, RegExpMacroAssembler::kNoStackLimitCheck); |
| 1117 |
m.PushBacktrack(&backtrack); |
| 1118 |
m.WriteStackPointerToRegister(sp); |
| 1119 |
// Fill stack and registers
|
| 1120 |
m.AdvanceCurrentPosition(2);
|
| 1121 |
m.WriteCurrentPositionToRegister(out1, 0);
|
| 1122 |
m.PushRegister(out1, RegExpMacroAssembler::kNoStackLimitCheck); |
| 1123 |
m.PushBacktrack(&fail); |
| 1124 |
// Drop backtrack stack frames.
|
| 1125 |
m.ReadStackPointerFromRegister(sp); |
| 1126 |
// And take the first backtrack (to &backtrack)
|
| 1127 |
m.Backtrack(); |
| 1128 |
|
| 1129 |
m.PushCurrentPosition(); |
| 1130 |
m.AdvanceCurrentPosition(2);
|
| 1131 |
m.PopCurrentPosition(); |
| 1132 |
|
| 1133 |
m.Bind(&backtrack); |
| 1134 |
m.PopRegister(out1); |
| 1135 |
m.ReadCurrentPositionFromRegister(out1); |
| 1136 |
m.AdvanceCurrentPosition(3);
|
| 1137 |
m.WriteCurrentPositionToRegister(out2, 0); // [0,3] |
| 1138 |
|
| 1139 |
Label loop; |
| 1140 |
m.SetRegister(loop_cnt, 0); // loop counter |
| 1141 |
m.Bind(&loop); |
| 1142 |
m.AdvanceRegister(loop_cnt, 1);
|
| 1143 |
m.AdvanceCurrentPosition(1);
|
| 1144 |
m.IfRegisterLT(loop_cnt, 3, &loop);
|
| 1145 |
m.WriteCurrentPositionToRegister(out3, 0); // [0,3,6] |
| 1146 |
|
| 1147 |
Label loop2; |
| 1148 |
m.SetRegister(loop_cnt, 2); // loop counter |
| 1149 |
m.Bind(&loop2); |
| 1150 |
m.AdvanceRegister(loop_cnt, -1);
|
| 1151 |
m.AdvanceCurrentPosition(1);
|
| 1152 |
m.IfRegisterGE(loop_cnt, 0, &loop2);
|
| 1153 |
m.WriteCurrentPositionToRegister(out4, 0); // [0,3,6,9] |
| 1154 |
|
| 1155 |
Label loop3; |
| 1156 |
Label exit_loop3; |
| 1157 |
m.PushRegister(out4, RegExpMacroAssembler::kNoStackLimitCheck); |
| 1158 |
m.PushRegister(out4, RegExpMacroAssembler::kNoStackLimitCheck); |
| 1159 |
m.ReadCurrentPositionFromRegister(out3); |
| 1160 |
m.Bind(&loop3); |
| 1161 |
m.AdvanceCurrentPosition(1);
|
| 1162 |
m.CheckGreedyLoop(&exit_loop3); |
| 1163 |
m.GoTo(&loop3); |
| 1164 |
m.Bind(&exit_loop3); |
| 1165 |
m.PopCurrentPosition(); |
| 1166 |
m.WriteCurrentPositionToRegister(out5, 0); // [0,3,6,9,9] |
| 1167 |
|
| 1168 |
m.Succeed(); |
| 1169 |
|
| 1170 |
m.Bind(&fail); |
| 1171 |
m.Fail(); |
| 1172 |
|
| 1173 |
Handle<String> source = |
| 1174 |
Factory::NewStringFromAscii(CStrVector("<loop test>"));
|
| 1175 |
Handle<Object> code_object = m.GetCode(source); |
| 1176 |
Handle<Code> code = Handle<Code>::cast(code_object); |
| 1177 |
|
| 1178 |
// String long enough for test (content doesn't matter).
|
| 1179 |
Handle<String> input = |
| 1180 |
Factory::NewStringFromAscii(CStrVector("foofoofoofoofoo"));
|
| 1181 |
Handle<SeqAsciiString> seq_input = Handle<SeqAsciiString>::cast(input); |
| 1182 |
Address start_adr = seq_input->GetCharsAddress(); |
| 1183 |
|
| 1184 |
int output[5]; |
| 1185 |
RegExpMacroAssemblerIA32::Result result = |
| 1186 |
ExecuteIA32(*code, |
| 1187 |
*input, |
| 1188 |
0,
|
| 1189 |
start_adr, |
| 1190 |
start_adr + input->length(), |
| 1191 |
output, |
| 1192 |
true);
|
| 1193 |
|
| 1194 |
CHECK_EQ(RegExpMacroAssemblerIA32::SUCCESS, result); |
| 1195 |
CHECK_EQ(0, output[0]); |
| 1196 |
CHECK_EQ(3, output[1]); |
| 1197 |
CHECK_EQ(6, output[2]); |
| 1198 |
CHECK_EQ(9, output[3]); |
| 1199 |
CHECK_EQ(9, output[4]); |
| 1200 |
} |
| 1201 |
|
| 1202 |
|
| 1203 |
TEST(MacroAssemblerIA32StackOverflow) {
|
| 1204 |
v8::V8::Initialize(); |
| 1205 |
ContextInitializer initializer; |
| 1206 |
|
| 1207 |
RegExpMacroAssemblerIA32 m(RegExpMacroAssemblerIA32::ASCII, 0);
|
| 1208 |
|
| 1209 |
Label loop; |
| 1210 |
m.Bind(&loop); |
| 1211 |
m.PushBacktrack(&loop); |
| 1212 |
m.GoTo(&loop); |
| 1213 |
|
| 1214 |
Handle<String> source = |
| 1215 |
Factory::NewStringFromAscii(CStrVector("<stack overflow test>"));
|
| 1216 |
Handle<Object> code_object = m.GetCode(source); |
| 1217 |
Handle<Code> code = Handle<Code>::cast(code_object); |
| 1218 |
|
| 1219 |
// String long enough for test (content doesn't matter).
|
| 1220 |
Handle<String> input = |
| 1221 |
Factory::NewStringFromAscii(CStrVector("dummy"));
|
| 1222 |
Handle<SeqAsciiString> seq_input = Handle<SeqAsciiString>::cast(input); |
| 1223 |
Address start_adr = seq_input->GetCharsAddress(); |
| 1224 |
|
| 1225 |
RegExpMacroAssemblerIA32::Result result = |
| 1226 |
ExecuteIA32(*code, |
| 1227 |
*input, |
| 1228 |
0,
|
| 1229 |
start_adr, |
| 1230 |
start_adr + input->length(), |
| 1231 |
NULL,
|
| 1232 |
true);
|
| 1233 |
|
| 1234 |
CHECK_EQ(RegExpMacroAssemblerIA32::EXCEPTION, result); |
| 1235 |
CHECK(Top::has_pending_exception()); |
| 1236 |
Top::clear_pending_exception(); |
| 1237 |
} |
| 1238 |
|
| 1239 |
|
| 1240 |
TEST(MacroAssemblerIA32LotsOfRegisters) {
|
| 1241 |
v8::V8::Initialize(); |
| 1242 |
ContextInitializer initializer; |
| 1243 |
|
| 1244 |
RegExpMacroAssemblerIA32 m(RegExpMacroAssemblerIA32::ASCII, 2);
|
| 1245 |
|
| 1246 |
// At least 2048, to ensure the allocated space for registers
|
| 1247 |
// span one full page.
|
| 1248 |
const int large_number = 8000; |
| 1249 |
m.WriteCurrentPositionToRegister(large_number, 42);
|
| 1250 |
m.WriteCurrentPositionToRegister(0, 0); |
| 1251 |
m.WriteCurrentPositionToRegister(1, 1); |
| 1252 |
Label done; |
| 1253 |
m.CheckNotBackReference(0, &done); // Performs a system-stack push. |
| 1254 |
m.Bind(&done); |
| 1255 |
m.PushRegister(large_number, RegExpMacroAssembler::kNoStackLimitCheck); |
| 1256 |
m.PopRegister(1);
|
| 1257 |
m.Succeed(); |
| 1258 |
|
| 1259 |
Handle<String> source = |
| 1260 |
Factory::NewStringFromAscii(CStrVector("<huge register space test>"));
|
| 1261 |
Handle<Object> code_object = m.GetCode(source); |
| 1262 |
Handle<Code> code = Handle<Code>::cast(code_object); |
| 1263 |
|
| 1264 |
// String long enough for test (content doesn't matter).
|
| 1265 |
Handle<String> input = |
| 1266 |
Factory::NewStringFromAscii(CStrVector("sample text"));
|
| 1267 |
Handle<SeqAsciiString> seq_input = Handle<SeqAsciiString>::cast(input); |
| 1268 |
Address start_adr = seq_input->GetCharsAddress(); |
| 1269 |
|
| 1270 |
int captures[2]; |
| 1271 |
RegExpMacroAssemblerIA32::Result result = |
| 1272 |
ExecuteIA32(*code, |
| 1273 |
*input, |
| 1274 |
0,
|
| 1275 |
start_adr, |
| 1276 |
start_adr + input->length(), |
| 1277 |
captures, |
| 1278 |
true);
|
| 1279 |
|
| 1280 |
CHECK_EQ(RegExpMacroAssemblerIA32::SUCCESS, result); |
| 1281 |
CHECK_EQ(0, captures[0]); |
| 1282 |
CHECK_EQ(42, captures[1]); |
| 1283 |
|
| 1284 |
Top::clear_pending_exception(); |
| 1285 |
} |
| 1286 |
|
| 1287 |
|
| 1288 |
|
| 1289 |
#endif // !defined ARM |
| 1290 |
|
| 1291 |
TEST(AddInverseToTable) {
|
| 1292 |
static const int kLimit = 1000; |
| 1293 |
static const int kRangeCount = 16; |
| 1294 |
for (int t = 0; t < 10; t++) { |
| 1295 |
ZoneScope zone_scope(DELETE_ON_EXIT); |
| 1296 |
ZoneList<CharacterRange>* ranges = |
| 1297 |
new ZoneList<CharacterRange>(kRangeCount);
|
| 1298 |
for (int i = 0; i < kRangeCount; i++) { |
| 1299 |
int from = PseudoRandom(t + 87, i + 25) % kLimit; |
| 1300 |
int to = from + (PseudoRandom(i + 87, t + 25) % (kLimit / 20)); |
| 1301 |
if (to > kLimit) to = kLimit;
|
| 1302 |
ranges->Add(CharacterRange(from, to)); |
| 1303 |
} |
| 1304 |
DispatchTable table; |
| 1305 |
DispatchTableConstructor cons(&table, false);
|
| 1306 |
cons.set_choice_index(0);
|
| 1307 |
cons.AddInverse(ranges); |
| 1308 |
for (int i = 0; i < kLimit; i++) { |
| 1309 |
bool is_on = false; |
| 1310 |
for (int j = 0; !is_on && j < kRangeCount; j++) |
| 1311 |
is_on = ranges->at(j).Contains(i); |
| 1312 |
OutSet* set = table.Get(i); |
| 1313 |
CHECK_EQ(is_on, set->Get(0) == false); |
| 1314 |
} |
| 1315 |
} |
| 1316 |
ZoneScope zone_scope(DELETE_ON_EXIT); |
| 1317 |
ZoneList<CharacterRange>* ranges = |
| 1318 |
new ZoneList<CharacterRange>(1); |
| 1319 |
ranges->Add(CharacterRange(0xFFF0, 0xFFFE)); |
| 1320 |
DispatchTable table; |
| 1321 |
DispatchTableConstructor cons(&table, false);
|
| 1322 |
cons.set_choice_index(0);
|
| 1323 |
cons.AddInverse(ranges); |
| 1324 |
CHECK(!table.Get(0xFFFE)->Get(0)); |
| 1325 |
CHECK(table.Get(0xFFFF)->Get(0)); |
| 1326 |
} |
| 1327 |
|
| 1328 |
|
| 1329 |
static uc32 canonicalize(uc32 c) {
|
| 1330 |
unibrow::uchar canon[unibrow::Ecma262Canonicalize::kMaxWidth]; |
| 1331 |
int count = unibrow::Ecma262Canonicalize::Convert(c, '\0', canon, NULL); |
| 1332 |
if (count == 0) { |
| 1333 |
return c;
|
| 1334 |
} else {
|
| 1335 |
CHECK_EQ(1, count);
|
| 1336 |
return canon[0]; |
| 1337 |
} |
| 1338 |
} |
| 1339 |
|
| 1340 |
|
| 1341 |
TEST(LatinCanonicalize) {
|
| 1342 |
unibrow::Mapping<unibrow::Ecma262UnCanonicalize> un_canonicalize; |
| 1343 |
for (char lower = 'a'; lower <= 'z'; lower++) { |
| 1344 |
char upper = lower + ('A' - 'a'); |
| 1345 |
CHECK_EQ(canonicalize(lower), canonicalize(upper)); |
| 1346 |
unibrow::uchar uncanon[unibrow::Ecma262UnCanonicalize::kMaxWidth]; |
| 1347 |
int length = un_canonicalize.get(lower, '\0', uncanon); |
| 1348 |
CHECK_EQ(2, length);
|
| 1349 |
CHECK_EQ(upper, uncanon[0]);
|
| 1350 |
CHECK_EQ(lower, uncanon[1]);
|
| 1351 |
} |
| 1352 |
for (uc32 c = 128; c < (1 << 21); c++) |
| 1353 |
CHECK_GE(canonicalize(c), 128);
|
| 1354 |
unibrow::Mapping<unibrow::ToUppercase> to_upper; |
| 1355 |
for (uc32 c = 0; c < (1 << 21); c++) { |
| 1356 |
unibrow::uchar upper[unibrow::ToUppercase::kMaxWidth]; |
| 1357 |
int length = to_upper.get(c, '\0', upper); |
| 1358 |
if (length == 0) { |
| 1359 |
length = 1;
|
| 1360 |
upper[0] = c;
|
| 1361 |
} |
| 1362 |
uc32 u = upper[0];
|
| 1363 |
if (length > 1 || (c >= 128 && u < 128)) |
| 1364 |
u = c; |
| 1365 |
CHECK_EQ(u, canonicalize(c)); |
| 1366 |
} |
| 1367 |
} |
| 1368 |
|
| 1369 |
|
| 1370 |
static uc32 CanonRange(uc32 c) {
|
| 1371 |
unibrow::uchar canon[unibrow::CanonicalizationRange::kMaxWidth]; |
| 1372 |
int count = unibrow::CanonicalizationRange::Convert(c, '\0', canon, NULL); |
| 1373 |
if (count == 0) { |
| 1374 |
return c;
|
| 1375 |
} else {
|
| 1376 |
CHECK_EQ(1, count);
|
| 1377 |
return canon[0]; |
| 1378 |
} |
| 1379 |
} |
| 1380 |
|
| 1381 |
|
| 1382 |
TEST(RangeCanonicalization) {
|
| 1383 |
CHECK_NE(CanonRange(0) & CharacterRange::kStartMarker, 0); |
| 1384 |
// Check that we arrive at the same result when using the basic
|
| 1385 |
// range canonicalization primitives as when using immediate
|
| 1386 |
// canonicalization.
|
| 1387 |
unibrow::Mapping<unibrow::Ecma262UnCanonicalize> un_canonicalize; |
| 1388 |
for (int i = 0; i < CharacterRange::kRangeCanonicalizeMax; i++) { |
| 1389 |
int range = CanonRange(i);
|
| 1390 |
int indirect_length = 0; |
| 1391 |
unibrow::uchar indirect[unibrow::Ecma262UnCanonicalize::kMaxWidth]; |
| 1392 |
if ((range & CharacterRange::kStartMarker) == 0) { |
| 1393 |
indirect_length = un_canonicalize.get(i - range, '\0', indirect);
|
| 1394 |
for (int i = 0; i < indirect_length; i++) |
| 1395 |
indirect[i] += range; |
| 1396 |
} else {
|
| 1397 |
indirect_length = un_canonicalize.get(i, '\0', indirect);
|
| 1398 |
} |
| 1399 |
unibrow::uchar direct[unibrow::Ecma262UnCanonicalize::kMaxWidth]; |
| 1400 |
int direct_length = un_canonicalize.get(i, '\0', direct); |
| 1401 |
CHECK_EQ(direct_length, indirect_length); |
| 1402 |
} |
| 1403 |
// Check that we arrive at the same results when skipping over
|
| 1404 |
// canonicalization ranges.
|
| 1405 |
int next_block = 0; |
| 1406 |
while (next_block < CharacterRange::kRangeCanonicalizeMax) {
|
| 1407 |
uc32 start = CanonRange(next_block); |
| 1408 |
CHECK_NE((start & CharacterRange::kStartMarker), 0);
|
| 1409 |
unsigned dist = start & CharacterRange::kPayloadMask;
|
| 1410 |
unibrow::uchar first[unibrow::Ecma262UnCanonicalize::kMaxWidth]; |
| 1411 |
int first_length = un_canonicalize.get(next_block, '\0', first); |
| 1412 |
for (unsigned i = 1; i < dist; i++) { |
| 1413 |
CHECK_EQ(i, CanonRange(next_block + i)); |
| 1414 |
unibrow::uchar succ[unibrow::Ecma262UnCanonicalize::kMaxWidth]; |
| 1415 |
int succ_length = un_canonicalize.get(next_block + i, '\0', succ); |
| 1416 |
CHECK_EQ(first_length, succ_length); |
| 1417 |
for (int j = 0; j < succ_length; j++) { |
| 1418 |
int calc = first[j] + i;
|
| 1419 |
int found = succ[j];
|
| 1420 |
CHECK_EQ(calc, found); |
| 1421 |
} |
| 1422 |
} |
| 1423 |
next_block = next_block + dist; |
| 1424 |
} |
| 1425 |
} |
| 1426 |
|
| 1427 |
|
| 1428 |
TEST(UncanonicalizeEquivalence) {
|
| 1429 |
unibrow::Mapping<unibrow::Ecma262UnCanonicalize> un_canonicalize; |
| 1430 |
unibrow::uchar chars[unibrow::Ecma262UnCanonicalize::kMaxWidth]; |
| 1431 |
for (int i = 0; i < (1 << 16); i++) { |
| 1432 |
int length = un_canonicalize.get(i, '\0', chars); |
| 1433 |
for (int j = 0; j < length; j++) { |
| 1434 |
unibrow::uchar chars2[unibrow::Ecma262UnCanonicalize::kMaxWidth]; |
| 1435 |
int length2 = un_canonicalize.get(chars[j], '\0', chars2); |
| 1436 |
CHECK_EQ(length, length2); |
| 1437 |
for (int k = 0; k < length; k++) |
| 1438 |
CHECK_EQ(static_cast<int>(chars[k]), static_cast<int>(chars2[k])); |
| 1439 |
} |
| 1440 |
} |
| 1441 |
} |
| 1442 |
|
| 1443 |
|
| 1444 |
static void TestRangeCaseIndependence(CharacterRange input, |
| 1445 |
Vector<CharacterRange> expected) {
|
| 1446 |
ZoneScope zone_scope(DELETE_ON_EXIT); |
| 1447 |
int count = expected.length();
|
| 1448 |
ZoneList<CharacterRange>* list = new ZoneList<CharacterRange>(count);
|
| 1449 |
input.AddCaseEquivalents(list); |
| 1450 |
CHECK_EQ(count, list->length()); |
| 1451 |
for (int i = 0; i < list->length(); i++) { |
| 1452 |
CHECK_EQ(expected[i].from(), list->at(i).from()); |
| 1453 |
CHECK_EQ(expected[i].to(), list->at(i).to()); |
| 1454 |
} |
| 1455 |
} |
| 1456 |
|
| 1457 |
|
| 1458 |
static void TestSimpleRangeCaseIndependence(CharacterRange input, |
| 1459 |
CharacterRange expected) {
|
| 1460 |
EmbeddedVector<CharacterRange, 1> vector;
|
| 1461 |
vector[0] = expected;
|
| 1462 |
TestRangeCaseIndependence(input, vector); |
| 1463 |
} |
| 1464 |
|
| 1465 |
|
| 1466 |
TEST(CharacterRangeCaseIndependence) {
|
| 1467 |
TestSimpleRangeCaseIndependence(CharacterRange::Singleton('a'),
|
| 1468 |
CharacterRange::Singleton('A'));
|
| 1469 |
TestSimpleRangeCaseIndependence(CharacterRange::Singleton('z'),
|
| 1470 |
CharacterRange::Singleton('Z'));
|
| 1471 |
TestSimpleRangeCaseIndependence(CharacterRange('a', 'z'), |
| 1472 |
CharacterRange('A', 'Z')); |
| 1473 |
TestSimpleRangeCaseIndependence(CharacterRange('c', 'f'), |
| 1474 |
CharacterRange('C', 'F')); |
| 1475 |
TestSimpleRangeCaseIndependence(CharacterRange('a', 'b'), |
| 1476 |
CharacterRange('A', 'B')); |
| 1477 |
TestSimpleRangeCaseIndependence(CharacterRange('y', 'z'), |
| 1478 |
CharacterRange('Y', 'Z')); |
| 1479 |
TestSimpleRangeCaseIndependence(CharacterRange('a' - 1, 'z' + 1), |
| 1480 |
CharacterRange('A', 'Z')); |
| 1481 |
TestSimpleRangeCaseIndependence(CharacterRange('A', 'Z'), |
| 1482 |
CharacterRange('a', 'z')); |
| 1483 |
TestSimpleRangeCaseIndependence(CharacterRange('C', 'F'), |
| 1484 |
CharacterRange('c', 'f')); |
| 1485 |
TestSimpleRangeCaseIndependence(CharacterRange('A' - 1, 'Z' + 1), |
| 1486 |
CharacterRange('a', 'z')); |
| 1487 |
// Here we need to add [l-z] to complete the case independence of
|
| 1488 |
// [A-Za-z] but we expect [a-z] to be added since we always add a
|
| 1489 |
// whole block at a time.
|
| 1490 |
TestSimpleRangeCaseIndependence(CharacterRange('A', 'k'), |
| 1491 |
CharacterRange('a', 'z')); |
| 1492 |
} |
| 1493 |
|
| 1494 |
|
| 1495 |
static bool InClass(uc16 c, ZoneList<CharacterRange>* ranges) { |
| 1496 |
if (ranges == NULL) |
| 1497 |
return false; |
| 1498 |
for (int i = 0; i < ranges->length(); i++) { |
| 1499 |
CharacterRange range = ranges->at(i); |
| 1500 |
if (range.from() <= c && c <= range.to())
|
| 1501 |
return true; |
| 1502 |
} |
| 1503 |
return false; |
| 1504 |
} |
| 1505 |
|
| 1506 |
|
| 1507 |
TEST(CharClassDifference) {
|
| 1508 |
ZoneScope zone_scope(DELETE_ON_EXIT); |
| 1509 |
ZoneList<CharacterRange>* base = new ZoneList<CharacterRange>(1); |
| 1510 |
base->Add(CharacterRange::Everything()); |
| 1511 |
Vector<const uc16> overlay = CharacterRange::GetWordBounds();
|
| 1512 |
ZoneList<CharacterRange>* included = NULL;
|
| 1513 |
ZoneList<CharacterRange>* excluded = NULL;
|
| 1514 |
CharacterRange::Split(base, overlay, &included, &excluded); |
| 1515 |
for (int i = 0; i < (1 << 16); i++) { |
| 1516 |
bool in_base = InClass(i, base);
|
| 1517 |
if (in_base) {
|
| 1518 |
bool in_overlay = false; |
| 1519 |
for (int j = 0; !in_overlay && j < overlay.length(); j += 2) { |
| 1520 |
if (overlay[j] <= i && i <= overlay[j+1]) |
| 1521 |
in_overlay = true;
|
| 1522 |
} |
| 1523 |
CHECK_EQ(in_overlay, InClass(i, included)); |
| 1524 |
CHECK_EQ(!in_overlay, InClass(i, excluded)); |
| 1525 |
} else {
|
| 1526 |
CHECK(!InClass(i, included)); |
| 1527 |
CHECK(!InClass(i, excluded)); |
| 1528 |
} |
| 1529 |
} |
| 1530 |
} |
| 1531 |
|
| 1532 |
|
| 1533 |
TEST(Graph) {
|
| 1534 |
V8::Initialize(NULL);
|
| 1535 |
Execute("(?:(?:x(.))?\1)+$", false, true, true); |
| 1536 |
} |