libgo/go/regexp/all_test.go

   1 // Copyright 2009 The Go Authors. All rights reserved.
   2 // Use of this source code is governed by a BSD-style
   3 // license that can be found in the LICENSE file.
   4
   5 package regexp
   6
   7 import (
   8         "reflect"
   9         "regexp/syntax"
  10         "strings"
  11         "testing"
  12         "unicode/utf8"
  13 )
  14
  15 var goodRe = []string{
  16         ``,
  17         `.`,
  18         `^.$`,
  19         `a`,
  20         `a*`,
  21         `a+`,
  22         `a?`,
  23         `a|b`,
  24         `a*|b*`,
  25         `(a*|b)(c*|d)`,
  26         `[a-z]`,
  27         `[a-abc-c\-\]\[]`,
  28         `[a-z]+`,
  29         `[abc]`,
  30         `[^1234]`,
  31         `[^\n]`,
  32         `\!\\`,
  33 }
  34
  35 type stringError struct {
  36         re  string
  37         err string
  38 }
  39
  40 var badRe = []stringError{
  41         {`*`, "missing argument to repetition operator: `*`"},
  42         {`+`, "missing argument to repetition operator: `+`"},
  43         {`?`, "missing argument to repetition operator: `?`"},
  44         {`(abc`, "missing closing ): `(abc`"},
  45         {`abc)`, "unexpected ): `abc)`"},
  46         {`x[a-z`, "missing closing ]: `[a-z`"},
  47         {`[z-a]`, "invalid character class range: `z-a`"},
  48         {`abc\`, "trailing backslash at end of expression"},
  49         {`a**`, "invalid nested repetition operator: `**`"},
  50         {`a*+`, "invalid nested repetition operator: `*+`"},
  51         {`\x`, "invalid escape sequence: `\\x`"},
  52 }
  53
  54 func compileTest(t *testing.T, expr string, error string) *Regexp {
  55         re, err := Compile(expr)
  56         if error == "" && err != nil {
  57                 t.Error("compiling `", expr, "`; unexpected error: ", err.Error())
  58         }
  59         if error != "" && err == nil {
  60                 t.Error("compiling `", expr, "`; missing error")
  61         } else if error != "" && !strings.Contains(err.Error(), error) {
  62                 t.Error("compiling `", expr, "`; wrong error: ", err.Error(), "; want ", error)
  63         }
  64         return re
  65 }
  66
  67 func TestGoodCompile(t *testing.T) {
  68         for i := 0; i < len(goodRe); i++ {
  69                 compileTest(t, goodRe[i], "")
  70         }
  71 }
  72
  73 func TestBadCompile(t *testing.T) {
  74         for i := 0; i < len(badRe); i++ {
  75                 compileTest(t, badRe[i].re, badRe[i].err)
  76         }
  77 }
  78
  79 func matchTest(t *testing.T, test *FindTest) {
  80         re := compileTest(t, test.pat, "")
  81         if re == nil {
  82                 return
  83         }
  84         m := re.MatchString(test.text)
  85         if m != (len(test.matches) > 0) {
  86                 t.Errorf("MatchString failure on %s: %t should be %t", test, m, len(test.matches) > 0)
  87         }
  88         // now try bytes
  89         m = re.Match([]byte(test.text))
  90         if m != (len(test.matches) > 0) {
  91                 t.Errorf("Match failure on %s: %t should be %t", test, m, len(test.matches) > 0)
  92         }
  93 }
  94
  95 func TestMatch(t *testing.T) {
  96         for _, test := range findTests {
  97                 matchTest(t, &test)
  98         }
  99 }
 100
 101 func matchFunctionTest(t *testing.T, test *FindTest) {
 102         m, err := MatchString(test.pat, test.text)
 103         if err == nil {
 104                 return
 105         }
 106         if m != (len(test.matches) > 0) {
 107                 t.Errorf("Match failure on %s: %t should be %t", test, m, len(test.matches) > 0)
 108         }
 109 }
 110
 111 func TestMatchFunction(t *testing.T) {
 112         for _, test := range findTests {
 113                 matchFunctionTest(t, &test)
 114         }
 115 }
 116
 117 func copyMatchTest(t *testing.T, test *FindTest) {
 118         re := compileTest(t, test.pat, "")
 119         if re == nil {
 120                 return
 121         }
 122         m1 := re.MatchString(test.text)
 123         m2 := re.Copy().MatchString(test.text)
 124         if m1 != m2 {
 125                 t.Errorf("Copied Regexp match failure on %s: original gave %t; copy gave %t; should be %t",
 126                         test, m1, m2, len(test.matches) > 0)
 127         }
 128 }
 129
 130 func TestCopyMatch(t *testing.T) {
 131         for _, test := range findTests {
 132                 copyMatchTest(t, &test)
 133         }
 134 }
 135
 136 type ReplaceTest struct {
 137         pattern, replacement, input, output string
 138 }
 139
 140 var replaceTests = []ReplaceTest{
 141         // Test empty input and/or replacement, with pattern that matches the empty string.
 142         {"", "", "", ""},
 143         {"", "x", "", "x"},
 144         {"", "", "abc", "abc"},
 145         {"", "x", "abc", "xaxbxcx"},
 146
 147         // Test empty input and/or replacement, with pattern that does not match the empty string.
 148         {"b", "", "", ""},
 149         {"b", "x", "", ""},
 150         {"b", "", "abc", "ac"},
 151         {"b", "x", "abc", "axc"},
 152         {"y", "", "", ""},
 153         {"y", "x", "", ""},
 154         {"y", "", "abc", "abc"},
 155         {"y", "x", "abc", "abc"},
 156
 157         // Multibyte characters -- verify that we don't try to match in the middle
 158         // of a character.
 159         {"[a-c]*", "x", "\u65e5", "x\u65e5x"},
 160         {"[^\u65e5]", "x", "abc\u65e5def", "xxx\u65e5xxx"},
 161
 162         // Start and end of a string.
 163         {"^[a-c]*", "x", "abcdabc", "xdabc"},
 164         {"[a-c]*$", "x", "abcdabc", "abcdx"},
 165         {"^[a-c]*$", "x", "abcdabc", "abcdabc"},
 166         {"^[a-c]*", "x", "abc", "x"},
 167         {"[a-c]*$", "x", "abc", "x"},
 168         {"^[a-c]*$", "x", "abc", "x"},
 169         {"^[a-c]*", "x", "dabce", "xdabce"},
 170         {"[a-c]*$", "x", "dabce", "dabcex"},
 171         {"^[a-c]*$", "x", "dabce", "dabce"},
 172         {"^[a-c]*", "x", "", "x"},
 173         {"[a-c]*$", "x", "", "x"},
 174         {"^[a-c]*$", "x", "", "x"},
 175
 176         {"^[a-c]+", "x", "abcdabc", "xdabc"},
 177         {"[a-c]+$", "x", "abcdabc", "abcdx"},
 178         {"^[a-c]+$", "x", "abcdabc", "abcdabc"},
 179         {"^[a-c]+", "x", "abc", "x"},
 180         {"[a-c]+$", "x", "abc", "x"},
 181         {"^[a-c]+$", "x", "abc", "x"},
 182         {"^[a-c]+", "x", "dabce", "dabce"},
 183         {"[a-c]+$", "x", "dabce", "dabce"},
 184         {"^[a-c]+$", "x", "dabce", "dabce"},
 185         {"^[a-c]+", "x", "", ""},
 186         {"[a-c]+$", "x", "", ""},
 187         {"^[a-c]+$", "x", "", ""},
 188
 189         // Other cases.
 190         {"abc", "def", "abcdefg", "defdefg"},
 191         {"bc", "BC", "abcbcdcdedef", "aBCBCdcdedef"},
 192         {"abc", "", "abcdabc", "d"},
 193         {"x", "xXx", "xxxXxxx", "xXxxXxxXxXxXxxXxxXx"},
 194         {"abc", "d", "", ""},
 195         {"abc", "d", "abc", "d"},
 196         {".+", "x", "abc", "x"},
 197         {"[a-c]*", "x", "def", "xdxexfx"},
 198         {"[a-c]+", "x", "abcbcdcdedef", "xdxdedef"},
 199         {"[a-c]*", "x", "abcbcdcdedef", "xdxdxexdxexfx"},
 200
 201         // Substitutions
 202         {"a+", "($0)", "banana", "b(a)n(a)n(a)"},
 203         {"a+", "(${0})", "banana", "b(a)n(a)n(a)"},
 204         {"a+", "(${0})$0", "banana", "b(a)an(a)an(a)a"},
 205         {"a+", "(${0})$0", "banana", "b(a)an(a)an(a)a"},
 206         {"hello, (.+)", "goodbye, ${1}", "hello, world", "goodbye, world"},
 207         {"hello, (.+)", "goodbye, $1x", "hello, world", "goodbye, "},
 208         {"hello, (.+)", "goodbye, ${1}x", "hello, world", "goodbye, worldx"},
 209         {"hello, (.+)", "<$0><$1><$2><$3>", "hello, world", "<hello, world><world><><>"},
 210         {"hello, (?P<noun>.+)", "goodbye, $noun!", "hello, world", "goodbye, world!"},
 211         {"hello, (?P<noun>.+)", "goodbye, ${noun}", "hello, world", "goodbye, world"},
 212         {"(?P<x>hi)|(?P<x>bye)", "$x$x$x", "hi", "hihihi"},
 213         {"(?P<x>hi)|(?P<x>bye)", "$x$x$x", "bye", "byebyebye"},
 214         {"(?P<x>hi)|(?P<x>bye)", "$xyz", "hi", ""},
 215         {"(?P<x>hi)|(?P<x>bye)", "${x}yz", "hi", "hiyz"},
 216         {"(?P<x>hi)|(?P<x>bye)", "hello $$x", "hi", "hello $x"},
 217         {"a+", "${oops", "aaa", "${oops"},
 218         {"a+", "$$", "aaa", "$"},
 219         {"a+", "$", "aaa", "$"},
 220
 221         // Substitution when subexpression isn't found
 222         {"(x)?", "$1", "123", "123"},
 223         {"abc", "$1", "123", "123"},
 224
 225         // Substitutions involving a (x){0}
 226         {"(a)(b){0}(c)", ".$1|$3.", "xacxacx", "x.a|c.x.a|c.x"},
 227         {"(a)(((b))){0}c", ".$1.", "xacxacx", "x.a.x.a.x"},
 228         {"((a(b){0}){3}){5}(h)", "y caramb$2", "say aaaaaaaaaaaaaaaah", "say ay caramba"},
 229         {"((a(b){0}){3}){5}h", "y caramb$2", "say aaaaaaaaaaaaaaaah", "say ay caramba"},
 230 }
 231
 232 var replaceLiteralTests = []ReplaceTest{
 233         // Substitutions
 234         {"a+", "($0)", "banana", "b($0)n($0)n($0)"},
 235         {"a+", "(${0})", "banana", "b(${0})n(${0})n(${0})"},
 236         {"a+", "(${0})$0", "banana", "b(${0})$0n(${0})$0n(${0})$0"},
 237         {"a+", "(${0})$0", "banana", "b(${0})$0n(${0})$0n(${0})$0"},
 238         {"hello, (.+)", "goodbye, ${1}", "hello, world", "goodbye, ${1}"},
 239         {"hello, (?P<noun>.+)", "goodbye, $noun!", "hello, world", "goodbye, $noun!"},
 240         {"hello, (?P<noun>.+)", "goodbye, ${noun}", "hello, world", "goodbye, ${noun}"},
 241         {"(?P<x>hi)|(?P<x>bye)", "$x$x$x", "hi", "$x$x$x"},
 242         {"(?P<x>hi)|(?P<x>bye)", "$x$x$x", "bye", "$x$x$x"},
 243         {"(?P<x>hi)|(?P<x>bye)", "$xyz", "hi", "$xyz"},
 244         {"(?P<x>hi)|(?P<x>bye)", "${x}yz", "hi", "${x}yz"},
 245         {"(?P<x>hi)|(?P<x>bye)", "hello $$x", "hi", "hello $$x"},
 246         {"a+", "${oops", "aaa", "${oops"},
 247         {"a+", "$$", "aaa", "$$"},
 248         {"a+", "$", "aaa", "$"},
 249 }
 250
 251 type ReplaceFuncTest struct {
 252         pattern       string
 253         replacement   func(string) string
 254         input, output string
 255 }
 256
 257 var replaceFuncTests = []ReplaceFuncTest{
 258         {"[a-c]", func(s string) string { return "x" + s + "y" }, "defabcdef", "defxayxbyxcydef"},
 259         {"[a-c]+", func(s string) string { return "x" + s + "y" }, "defabcdef", "defxabcydef"},
 260         {"[a-c]*", func(s string) string { return "x" + s + "y" }, "defabcdef", "xydxyexyfxabcydxyexyfxy"},
 261 }
 262
 263 func TestReplaceAll(t *testing.T) {
 264         for _, tc := range replaceTests {
 265                 re, err := Compile(tc.pattern)
 266                 if err != nil {
 267                         t.Errorf("Unexpected error compiling %q: %v", tc.pattern, err)
 268                         continue
 269                 }
 270                 actual := re.ReplaceAllString(tc.input, tc.replacement)
 271                 if actual != tc.output {
 272                         t.Errorf("%q.ReplaceAllString(%q,%q) = %q; want %q",
 273                                 tc.pattern, tc.input, tc.replacement, actual, tc.output)
 274                 }
 275                 // now try bytes
 276                 actual = string(re.ReplaceAll([]byte(tc.input), []byte(tc.replacement)))
 277                 if actual != tc.output {
 278                         t.Errorf("%q.ReplaceAll(%q,%q) = %q; want %q",
 279                                 tc.pattern, tc.input, tc.replacement, actual, tc.output)
 280                 }
 281         }
 282 }
 283
 284 func TestReplaceAllLiteral(t *testing.T) {
 285         // Run ReplaceAll tests that do not have $ expansions.
 286         for _, tc := range replaceTests {
 287                 if strings.Contains(tc.replacement, "$") {
 288                         continue
 289                 }
 290                 re, err := Compile(tc.pattern)
 291                 if err != nil {
 292                         t.Errorf("Unexpected error compiling %q: %v", tc.pattern, err)
 293                         continue
 294                 }
 295                 actual := re.ReplaceAllLiteralString(tc.input, tc.replacement)
 296                 if actual != tc.output {
 297                         t.Errorf("%q.ReplaceAllLiteralString(%q,%q) = %q; want %q",
 298                                 tc.pattern, tc.input, tc.replacement, actual, tc.output)
 299                 }
 300                 // now try bytes
 301                 actual = string(re.ReplaceAllLiteral([]byte(tc.input), []byte(tc.replacement)))
 302                 if actual != tc.output {
 303                         t.Errorf("%q.ReplaceAllLiteral(%q,%q) = %q; want %q",
 304                                 tc.pattern, tc.input, tc.replacement, actual, tc.output)
 305                 }
 306         }
 307
 308         // Run literal-specific tests.
 309         for _, tc := range replaceLiteralTests {
 310                 re, err := Compile(tc.pattern)
 311                 if err != nil {
 312                         t.Errorf("Unexpected error compiling %q: %v", tc.pattern, err)
 313                         continue
 314                 }
 315                 actual := re.ReplaceAllLiteralString(tc.input, tc.replacement)
 316                 if actual != tc.output {
 317                         t.Errorf("%q.ReplaceAllLiteralString(%q,%q) = %q; want %q",
 318                                 tc.pattern, tc.input, tc.replacement, actual, tc.output)
 319                 }
 320                 // now try bytes
 321                 actual = string(re.ReplaceAllLiteral([]byte(tc.input), []byte(tc.replacement)))
 322                 if actual != tc.output {
 323                         t.Errorf("%q.ReplaceAllLiteral(%q,%q) = %q; want %q",
 324                                 tc.pattern, tc.input, tc.replacement, actual, tc.output)
 325                 }
 326         }
 327 }
 328
 329 func TestReplaceAllFunc(t *testing.T) {
 330         for _, tc := range replaceFuncTests {
 331                 re, err := Compile(tc.pattern)
 332                 if err != nil {
 333                         t.Errorf("Unexpected error compiling %q: %v", tc.pattern, err)
 334                         continue
 335                 }
 336                 actual := re.ReplaceAllStringFunc(tc.input, tc.replacement)
 337                 if actual != tc.output {
 338                         t.Errorf("%q.ReplaceFunc(%q,fn) = %q; want %q",
 339                                 tc.pattern, tc.input, actual, tc.output)
 340                 }
 341                 // now try bytes
 342                 actual = string(re.ReplaceAllFunc([]byte(tc.input), func(s []byte) []byte { return []byte(tc.replacement(string(s))) }))
 343                 if actual != tc.output {
 344                         t.Errorf("%q.ReplaceFunc(%q,fn) = %q; want %q",
 345                                 tc.pattern, tc.input, actual, tc.output)
 346                 }
 347         }
 348 }
 349
 350 type MetaTest struct {
 351         pattern, output, literal string
 352         isLiteral                bool
 353 }
 354
 355 var metaTests = []MetaTest{
 356         {``, ``, ``, true},
 357         {`foo`, `foo`, `foo`, true},
 358         {`日本語+`, `日本語\+`, `日本語`, false},
 359         {`foo\.\$`, `foo\\\.\\\$`, `foo.$`, true}, // has meta but no operator
 360         {`foo.\$`, `foo\.\\\$`, `foo`, false},     // has escaped operators and real operators
 361         {`!@#$%^&*()_+-=[{]}\|,<.>/?~`, `!@#\$%\^&\*\(\)_\+-=\[\{\]\}\\\|,<\.>/\?~`, `!@#`, false},
 362 }
 363
 364 var literalPrefixTests = []MetaTest{
 365         // See golang.org/issue/11175.
 366         // output is unused.
 367         {`^0^0$`, ``, `0`, false},
 368         {`^0^`, ``, ``, false},
 369         {`^0$`, ``, `0`, true},
 370         {`$0^`, ``, ``, false},
 371         {`$0$`, ``, ``, false},
 372         {`^^0$$`, ``, ``, false},
 373         {`^$^$`, ``, ``, false},
 374         {`$$0^^`, ``, ``, false},
 375 }
 376
 377 func TestQuoteMeta(t *testing.T) {
 378         for _, tc := range metaTests {
 379                 // Verify that QuoteMeta returns the expected string.
 380                 quoted := QuoteMeta(tc.pattern)
 381                 if quoted != tc.output {
 382                         t.Errorf("QuoteMeta(`%s`) = `%s`; want `%s`",
 383                                 tc.pattern, quoted, tc.output)
 384                         continue
 385                 }
 386
 387                 // Verify that the quoted string is in fact treated as expected
 388                 // by Compile -- i.e. that it matches the original, unquoted string.
 389                 if tc.pattern != "" {
 390                         re, err := Compile(quoted)
 391                         if err != nil {
 392                                 t.Errorf("Unexpected error compiling QuoteMeta(`%s`): %v", tc.pattern, err)
 393                                 continue
 394                         }
 395                         src := "abc" + tc.pattern + "def"
 396                         repl := "xyz"
 397                         replaced := re.ReplaceAllString(src, repl)
 398                         expected := "abcxyzdef"
 399                         if replaced != expected {
 400                                 t.Errorf("QuoteMeta(`%s`).Replace(`%s`,`%s`) = `%s`; want `%s`",
 401                                         tc.pattern, src, repl, replaced, expected)
 402                         }
 403                 }
 404         }
 405 }
 406
 407 func TestLiteralPrefix(t *testing.T) {
 408         for _, tc := range append(metaTests, literalPrefixTests...) {
 409                 // Literal method needs to scan the pattern.
 410                 re := MustCompile(tc.pattern)
 411                 str, complete := re.LiteralPrefix()
 412                 if complete != tc.isLiteral {
 413                         t.Errorf("LiteralPrefix(`%s`) = %t; want %t", tc.pattern, complete, tc.isLiteral)
 414                 }
 415                 if str != tc.literal {
 416                         t.Errorf("LiteralPrefix(`%s`) = `%s`; want `%s`", tc.pattern, str, tc.literal)
 417                 }
 418         }
 419 }
 420
 421 type subexpCase struct {
 422         input string
 423         num   int
 424         names []string
 425 }
 426
 427 var subexpCases = []subexpCase{
 428         {``, 0, nil},
 429         {`.*`, 0, nil},
 430         {`abba`, 0, nil},
 431         {`ab(b)a`, 1, []string{"", ""}},
 432         {`ab(.*)a`, 1, []string{"", ""}},
 433         {`(.*)ab(.*)a`, 2, []string{"", "", ""}},
 434         {`(.*)(ab)(.*)a`, 3, []string{"", "", "", ""}},
 435         {`(.*)((a)b)(.*)a`, 4, []string{"", "", "", "", ""}},
 436         {`(.*)(\(ab)(.*)a`, 3, []string{"", "", "", ""}},
 437         {`(.*)(\(a\)b)(.*)a`, 3, []string{"", "", "", ""}},
 438         {`(?P<foo>.*)(?P<bar>(a)b)(?P<foo>.*)a`, 4, []string{"", "foo", "bar", "", "foo"}},
 439 }
 440
 441 func TestSubexp(t *testing.T) {
 442         for _, c := range subexpCases {
 443                 re := MustCompile(c.input)
 444                 n := re.NumSubexp()
 445                 if n != c.num {
 446                         t.Errorf("%q: NumSubexp = %d, want %d", c.input, n, c.num)
 447                         continue
 448                 }
 449                 names := re.SubexpNames()
 450                 if len(names) != 1+n {
 451                         t.Errorf("%q: len(SubexpNames) = %d, want %d", c.input, len(names), n)
 452                         continue
 453                 }
 454                 if c.names != nil {
 455                         for i := 0; i < 1+n; i++ {
 456                                 if names[i] != c.names[i] {
 457                                         t.Errorf("%q: SubexpNames[%d] = %q, want %q", c.input, i, names[i], c.names[i])
 458                                 }
 459                         }
 460                 }
 461         }
 462 }
 463
 464 var splitTests = []struct {
 465         s   string
 466         r   string
 467         n   int
 468         out []string
 469 }{
 470         {"foo:and:bar", ":", -1, []string{"foo", "and", "bar"}},
 471         {"foo:and:bar", ":", 1, []string{"foo:and:bar"}},
 472         {"foo:and:bar", ":", 2, []string{"foo", "and:bar"}},
 473         {"foo:and:bar", "foo", -1, []string{"", ":and:bar"}},
 474         {"foo:and:bar", "bar", -1, []string{"foo:and:", ""}},
 475         {"foo:and:bar", "baz", -1, []string{"foo:and:bar"}},
 476         {"baabaab", "a", -1, []string{"b", "", "b", "", "b"}},
 477         {"baabaab", "a*", -1, []string{"b", "b", "b"}},
 478         {"baabaab", "ba*", -1, []string{"", "", "", ""}},
 479         {"foobar", "f*b*", -1, []string{"", "o", "o", "a", "r"}},
 480         {"foobar", "f+.*b+", -1, []string{"", "ar"}},
 481         {"foobooboar", "o{2}", -1, []string{"f", "b", "boar"}},
 482         {"a,b,c,d,e,f", ",", 3, []string{"a", "b", "c,d,e,f"}},
 483         {"a,b,c,d,e,f", ",", 0, nil},
 484         {",", ",", -1, []string{"", ""}},
 485         {",,,", ",", -1, []string{"", "", "", ""}},
 486         {"", ",", -1, []string{""}},
 487         {"", ".*", -1, []string{""}},
 488         {"", ".+", -1, []string{""}},
 489         {"", "", -1, []string{}},
 490         {"foobar", "", -1, []string{"f", "o", "o", "b", "a", "r"}},
 491         {"abaabaccadaaae", "a*", 5, []string{"", "b", "b", "c", "cadaaae"}},
 492         {":x:y:z:", ":", -1, []string{"", "x", "y", "z", ""}},
 493 }
 494
 495 func TestSplit(t *testing.T) {
 496         for i, test := range splitTests {
 497                 re, err := Compile(test.r)
 498                 if err != nil {
 499                         t.Errorf("#%d: %q: compile error: %s", i, test.r, err.Error())
 500                         continue
 501                 }
 502
 503                 split := re.Split(test.s, test.n)
 504                 if !reflect.DeepEqual(split, test.out) {
 505                         t.Errorf("#%d: %q: got %q; want %q", i, test.r, split, test.out)
 506                 }
 507
 508                 if QuoteMeta(test.r) == test.r {
 509                         strsplit := strings.SplitN(test.s, test.r, test.n)
 510                         if !reflect.DeepEqual(split, strsplit) {
 511                                 t.Errorf("#%d: Split(%q, %q, %d): regexp vs strings mismatch\nregexp=%q\nstrings=%q", i, test.s, test.r, test.n, split, strsplit)
 512                         }
 513                 }
 514         }
 515 }
 516
 517 // The following sequence of Match calls used to panic. See issue #12980.
 518 func TestParseAndCompile(t *testing.T) {
 519         expr := "a$"
 520         s := "a\nb"
 521
 522         for i, tc := range []struct {
 523                 reFlags  syntax.Flags
 524                 expMatch bool
 525         }{
 526                 {syntax.Perl | syntax.OneLine, false},
 527                 {syntax.Perl &^ syntax.OneLine, true},
 528         } {
 529                 parsed, err := syntax.Parse(expr, tc.reFlags)
 530                 if err != nil {
 531                         t.Fatalf("%d: parse: %v", i, err)
 532                 }
 533                 re, err := Compile(parsed.String())
 534                 if err != nil {
 535                         t.Fatalf("%d: compile: %v", i, err)
 536                 }
 537                 if match := re.MatchString(s); match != tc.expMatch {
 538                         t.Errorf("%d: %q.MatchString(%q)=%t; expected=%t", i, re, s, match, tc.expMatch)
 539                 }
 540         }
 541 }
 542
 543 // Check that one-pass cutoff does trigger.
 544 func TestOnePassCutoff(t *testing.T) {
 545         re, err := syntax.Parse(`^x{1,1000}y{1,1000}$`, syntax.Perl)
 546         if err != nil {
 547                 t.Fatalf("parse: %v", err)
 548         }
 549         p, err := syntax.Compile(re.Simplify())
 550         if err != nil {
 551                 t.Fatalf("compile: %v", err)
 552         }
 553         if compileOnePass(p) != notOnePass {
 554                 t.Fatalf("makeOnePass succeeded; wanted notOnePass")
 555         }
 556 }
 557
 558 // Check that the same machine can be used with the standard matcher
 559 // and then the backtracker when there are no captures.
 560 func TestSwitchBacktrack(t *testing.T) {
 561         re := MustCompile(`a|b`)
 562         long := make([]byte, maxBacktrackVector+1)
 563
 564         // The following sequence of Match calls used to panic. See issue #10319.
 565         re.Match(long)     // triggers standard matcher
 566         re.Match(long[:1]) // triggers backtracker
 567 }
 568
 569 func BenchmarkFind(b *testing.B) {
 570         b.StopTimer()
 571         re := MustCompile("a+b+")
 572         wantSubs := "aaabb"
 573         s := []byte("acbb" + wantSubs + "dd")
 574         b.StartTimer()
 575         b.ReportAllocs()
 576         for i := 0; i < b.N; i++ {
 577                 subs := re.Find(s)
 578                 if string(subs) != wantSubs {
 579                         b.Fatalf("Find(%q) = %q; want %q", s, subs, wantSubs)
 580                 }
 581         }
 582 }
 583
 584 func BenchmarkFindAllNoMatches(b *testing.B) {
 585         re := MustCompile("a+b+")
 586         s := []byte("acddee")
 587         b.ReportAllocs()
 588         b.ResetTimer()
 589         for i := 0; i < b.N; i++ {
 590                 all := re.FindAll(s, -1)
 591                 if all != nil {
 592                         b.Fatalf("FindAll(%q) = %q; want nil", s, all)
 593                 }
 594         }
 595 }
 596
 597 func BenchmarkFindString(b *testing.B) {
 598         b.StopTimer()
 599         re := MustCompile("a+b+")
 600         wantSubs := "aaabb"
 601         s := "acbb" + wantSubs + "dd"
 602         b.StartTimer()
 603         b.ReportAllocs()
 604         for i := 0; i < b.N; i++ {
 605                 subs := re.FindString(s)
 606                 if subs != wantSubs {
 607                         b.Fatalf("FindString(%q) = %q; want %q", s, subs, wantSubs)
 608                 }
 609         }
 610 }
 611
 612 func BenchmarkFindSubmatch(b *testing.B) {
 613         b.StopTimer()
 614         re := MustCompile("a(a+b+)b")
 615         wantSubs := "aaabb"
 616         s := []byte("acbb" + wantSubs + "dd")
 617         b.StartTimer()
 618         b.ReportAllocs()
 619         for i := 0; i < b.N; i++ {
 620                 subs := re.FindSubmatch(s)
 621                 if string(subs[0]) != wantSubs {
 622                         b.Fatalf("FindSubmatch(%q)[0] = %q; want %q", s, subs[0], wantSubs)
 623                 }
 624                 if string(subs[1]) != "aab" {
 625                         b.Fatalf("FindSubmatch(%q)[1] = %q; want %q", s, subs[1], "aab")
 626                 }
 627         }
 628 }
 629
 630 func BenchmarkFindStringSubmatch(b *testing.B) {
 631         b.StopTimer()
 632         re := MustCompile("a(a+b+)b")
 633         wantSubs := "aaabb"
 634         s := "acbb" + wantSubs + "dd"
 635         b.StartTimer()
 636         b.ReportAllocs()
 637         for i := 0; i < b.N; i++ {
 638                 subs := re.FindStringSubmatch(s)
 639                 if subs[0] != wantSubs {
 640                         b.Fatalf("FindStringSubmatch(%q)[0] = %q; want %q", s, subs[0], wantSubs)
 641                 }
 642                 if subs[1] != "aab" {
 643                         b.Fatalf("FindStringSubmatch(%q)[1] = %q; want %q", s, subs[1], "aab")
 644                 }
 645         }
 646 }
 647
 648 func BenchmarkLiteral(b *testing.B) {
 649         x := strings.Repeat("x", 50) + "y"
 650         b.StopTimer()
 651         re := MustCompile("y")
 652         b.StartTimer()
 653         for i := 0; i < b.N; i++ {
 654                 if !re.MatchString(x) {
 655                         b.Fatalf("no match!")
 656                 }
 657         }
 658 }
 659
 660 func BenchmarkNotLiteral(b *testing.B) {
 661         x := strings.Repeat("x", 50) + "y"
 662         b.StopTimer()
 663         re := MustCompile(".y")
 664         b.StartTimer()
 665         for i := 0; i < b.N; i++ {
 666                 if !re.MatchString(x) {
 667                         b.Fatalf("no match!")
 668                 }
 669         }
 670 }
 671
 672 func BenchmarkMatchClass(b *testing.B) {
 673         b.StopTimer()
 674         x := strings.Repeat("xxxx", 20) + "w"
 675         re := MustCompile("[abcdw]")
 676         b.StartTimer()
 677         for i := 0; i < b.N; i++ {
 678                 if !re.MatchString(x) {
 679                         b.Fatalf("no match!")
 680                 }
 681         }
 682 }
 683
 684 func BenchmarkMatchClass_InRange(b *testing.B) {
 685         b.StopTimer()
 686         // 'b' is between 'a' and 'c', so the charclass
 687         // range checking is no help here.
 688         x := strings.Repeat("bbbb", 20) + "c"
 689         re := MustCompile("[ac]")
 690         b.StartTimer()
 691         for i := 0; i < b.N; i++ {
 692                 if !re.MatchString(x) {
 693                         b.Fatalf("no match!")
 694                 }
 695         }
 696 }
 697
 698 func BenchmarkReplaceAll(b *testing.B) {
 699         x := "abcdefghijklmnopqrstuvwxyz"
 700         b.StopTimer()
 701         re := MustCompile("[cjrw]")
 702         b.StartTimer()
 703         for i := 0; i < b.N; i++ {
 704                 re.ReplaceAllString(x, "")
 705         }
 706 }
 707
 708 func BenchmarkAnchoredLiteralShortNonMatch(b *testing.B) {
 709         b.StopTimer()
 710         x := []byte("abcdefghijklmnopqrstuvwxyz")
 711         re := MustCompile("^zbc(d|e)")
 712         b.StartTimer()
 713         for i := 0; i < b.N; i++ {
 714                 re.Match(x)
 715         }
 716 }
 717
 718 func BenchmarkAnchoredLiteralLongNonMatch(b *testing.B) {
 719         b.StopTimer()
 720         x := []byte("abcdefghijklmnopqrstuvwxyz")
 721         for i := 0; i < 15; i++ {
 722                 x = append(x, x...)
 723         }
 724         re := MustCompile("^zbc(d|e)")
 725         b.StartTimer()
 726         for i := 0; i < b.N; i++ {
 727                 re.Match(x)
 728         }
 729 }
 730
 731 func BenchmarkAnchoredShortMatch(b *testing.B) {
 732         b.StopTimer()
 733         x := []byte("abcdefghijklmnopqrstuvwxyz")
 734         re := MustCompile("^.bc(d|e)")
 735         b.StartTimer()
 736         for i := 0; i < b.N; i++ {
 737                 re.Match(x)
 738         }
 739 }
 740
 741 func BenchmarkAnchoredLongMatch(b *testing.B) {
 742         b.StopTimer()
 743         x := []byte("abcdefghijklmnopqrstuvwxyz")
 744         for i := 0; i < 15; i++ {
 745                 x = append(x, x...)
 746         }
 747         re := MustCompile("^.bc(d|e)")
 748         b.StartTimer()
 749         for i := 0; i < b.N; i++ {
 750                 re.Match(x)
 751         }
 752 }
 753
 754 func BenchmarkOnePassShortA(b *testing.B) {
 755         b.StopTimer()
 756         x := []byte("abcddddddeeeededd")
 757         re := MustCompile("^.bc(d|e)*$")
 758         b.StartTimer()
 759         for i := 0; i < b.N; i++ {
 760                 re.Match(x)
 761         }
 762 }
 763
 764 func BenchmarkNotOnePassShortA(b *testing.B) {
 765         b.StopTimer()
 766         x := []byte("abcddddddeeeededd")
 767         re := MustCompile(".bc(d|e)*$")
 768         b.StartTimer()
 769         for i := 0; i < b.N; i++ {
 770                 re.Match(x)
 771         }
 772 }
 773
 774 func BenchmarkOnePassShortB(b *testing.B) {
 775         b.StopTimer()
 776         x := []byte("abcddddddeeeededd")
 777         re := MustCompile("^.bc(?:d|e)*$")
 778         b.StartTimer()
 779         for i := 0; i < b.N; i++ {
 780                 re.Match(x)
 781         }
 782 }
 783
 784 func BenchmarkNotOnePassShortB(b *testing.B) {
 785         b.StopTimer()
 786         x := []byte("abcddddddeeeededd")
 787         re := MustCompile(".bc(?:d|e)*$")
 788         b.StartTimer()
 789         for i := 0; i < b.N; i++ {
 790                 re.Match(x)
 791         }
 792 }
 793
 794 func BenchmarkOnePassLongPrefix(b *testing.B) {
 795         b.StopTimer()
 796         x := []byte("abcdefghijklmnopqrstuvwxyz")
 797         re := MustCompile("^abcdefghijklmnopqrstuvwxyz.*$")
 798         b.StartTimer()
 799         for i := 0; i < b.N; i++ {
 800                 re.Match(x)
 801         }
 802 }
 803
 804 func BenchmarkOnePassLongNotPrefix(b *testing.B) {
 805         b.StopTimer()
 806         x := []byte("abcdefghijklmnopqrstuvwxyz")
 807         re := MustCompile("^.bcdefghijklmnopqrstuvwxyz.*$")
 808         b.StartTimer()
 809         for i := 0; i < b.N; i++ {
 810                 re.Match(x)
 811         }
 812 }
 813
 814 func BenchmarkMatchParallelShared(b *testing.B) {
 815         x := []byte("this is a long line that contains foo bar baz")
 816         re := MustCompile("foo (ba+r)? baz")
 817         b.ResetTimer()
 818         b.RunParallel(func(pb *testing.PB) {
 819                 for pb.Next() {
 820                         re.Match(x)
 821                 }
 822         })
 823 }
 824
 825 func BenchmarkMatchParallelCopied(b *testing.B) {
 826         x := []byte("this is a long line that contains foo bar baz")
 827         re := MustCompile("foo (ba+r)? baz")
 828         b.ResetTimer()
 829         b.RunParallel(func(pb *testing.PB) {
 830                 re := re.Copy()
 831                 for pb.Next() {
 832                         re.Match(x)
 833                 }
 834         })
 835 }
 836
 837 var sink string
 838
 839 func BenchmarkQuoteMetaAll(b *testing.B) {
 840         specials := make([]byte, 0)
 841         for i := byte(0); i < utf8.RuneSelf; i++ {
 842                 if special(i) {
 843                         specials = append(specials, i)
 844                 }
 845         }
 846         s := string(specials)
 847         b.SetBytes(int64(len(s)))
 848         b.ResetTimer()
 849         for i := 0; i < b.N; i++ {
 850                 sink = QuoteMeta(s)
 851         }
 852 }
 853
 854 func BenchmarkQuoteMetaNone(b *testing.B) {
 855         s := "abcdefghijklmnopqrstuvwxyz"
 856         b.SetBytes(int64(len(s)))
 857         b.ResetTimer()
 858         for i := 0; i < b.N; i++ {
 859                 sink = QuoteMeta(s)
 860         }
 861 }