Source file src/regexp/exec_test.go

Documentation: regexp

     1  // Copyright 2010 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  	"bufio"
     9  	"compress/bzip2"
    10  	"fmt"
    11  	"internal/testenv"
    12  	"io"
    13  	"os"
    14  	"path/filepath"
    15  	"regexp/syntax"
    16  	"strconv"
    17  	"strings"
    18  	"testing"
    19  	"unicode/utf8"
    20  )
    21  
    22  // TestRE2 tests this package's regexp API against test cases
    23  // considered during RE2's exhaustive tests, which run all possible
    24  // regexps over a given set of atoms and operators, up to a given
    25  // complexity, over all possible strings over a given alphabet,
    26  // up to a given size. Rather than try to link with RE2, we read a
    27  // log file containing the test cases and the expected matches.
    28  // The log file, re2-exhaustive.txt, is generated by running 'make log'
    29  // in the open source RE2 distribution https://github.com/google/re2/.
    30  //
    31  // The test file format is a sequence of stanzas like:
    32  //
    33  //	strings
    34  //	"abc"
    35  //	"123x"
    36  //	regexps
    37  //	"[a-z]+"
    38  //	0-3;0-3
    39  //	-;-
    40  //	"([0-9])([0-9])([0-9])"
    41  //	-;-
    42  //	-;0-3 0-1 1-2 2-3
    43  //
    44  // The stanza begins by defining a set of strings, quoted
    45  // using Go double-quote syntax, one per line. Then the
    46  // regexps section gives a sequence of regexps to run on
    47  // the strings. In the block that follows a regexp, each line
    48  // gives the semicolon-separated match results of running
    49  // the regexp on the corresponding string.
    50  // Each match result is either a single -, meaning no match, or a
    51  // space-separated sequence of pairs giving the match and
    52  // submatch indices. An unmatched subexpression formats
    53  // its pair as a single - (not illustrated above).  For now
    54  // each regexp run produces two match results, one for a
    55  // “full match” that restricts the regexp to matching the entire
    56  // string or nothing, and one for a “partial match” that gives
    57  // the leftmost first match found in the string.
    58  //
    59  // Lines beginning with # are comments. Lines beginning with
    60  // a capital letter are test names printed during RE2's test suite
    61  // and are echoed into t but otherwise ignored.
    62  //
    63  // At time of writing, re2-exhaustive.txt is 59 MB but compresses to 385 kB,
    64  // so we store re2-exhaustive.txt.bz2 in the repository and decompress it on the fly.
    65  func TestRE2Search(t *testing.T) {
    66  	testRE2(t, "testdata/re2-search.txt")
    67  }
    68  
    69  func testRE2(t *testing.T, file string) {
    70  	f, err := os.Open(file)
    71  	if err != nil {
    72  		t.Fatal(err)
    73  	}
    74  	defer f.Close()
    75  	var txt io.Reader
    76  	if strings.HasSuffix(file, ".bz2") {
    77  		z := bzip2.NewReader(f)
    78  		txt = z
    79  		file = file[:len(file)-len(".bz2")] // for error messages
    80  	} else {
    81  		txt = f
    82  	}
    83  	lineno := 0
    84  	scanner := bufio.NewScanner(txt)
    85  	var (
    86  		str       []string
    87  		input     []string
    88  		inStrings bool
    89  		re        *Regexp
    90  		refull    *Regexp
    91  		nfail     int
    92  		ncase     int
    93  	)
    94  	for lineno := 1; scanner.Scan(); lineno++ {
    95  		line := scanner.Text()
    96  		switch {
    97  		case line == "":
    98  			t.Fatalf("%s:%d: unexpected blank line", file, lineno)
    99  		case line[0] == '#':
   100  			continue
   101  		case 'A' <= line[0] && line[0] <= 'Z':
   102  			// Test name.
   103  			t.Logf("%s\n", line)
   104  			continue
   105  		case line == "strings":
   106  			str = str[:0]
   107  			inStrings = true
   108  		case line == "regexps":
   109  			inStrings = false
   110  		case line[0] == '"':
   111  			q, err := strconv.Unquote(line)
   112  			if err != nil {
   113  				// Fatal because we'll get out of sync.
   114  				t.Fatalf("%s:%d: unquote %s: %v", file, lineno, line, err)
   115  			}
   116  			if inStrings {
   117  				str = append(str, q)
   118  				continue
   119  			}
   120  			// Is a regexp.
   121  			if len(input) != 0 {
   122  				t.Fatalf("%s:%d: out of sync: have %d strings left before %#q", file, lineno, len(input), q)
   123  			}
   124  			re, err = tryCompile(q)
   125  			if err != nil {
   126  				if err.Error() == "error parsing regexp: invalid escape sequence: `\\C`" {
   127  					// We don't and likely never will support \C; keep going.
   128  					continue
   129  				}
   130  				t.Errorf("%s:%d: compile %#q: %v", file, lineno, q, err)
   131  				if nfail++; nfail >= 100 {
   132  					t.Fatalf("stopping after %d errors", nfail)
   133  				}
   134  				continue
   135  			}
   136  			full := `\A(?:` + q + `)\z`
   137  			refull, err = tryCompile(full)
   138  			if err != nil {
   139  				// Fatal because q worked, so this should always work.
   140  				t.Fatalf("%s:%d: compile full %#q: %v", file, lineno, full, err)
   141  			}
   142  			input = str
   143  		case line[0] == '-' || '0' <= line[0] && line[0] <= '9':
   144  			// A sequence of match results.
   145  			ncase++
   146  			if re == nil {
   147  				// Failed to compile: skip results.
   148  				continue
   149  			}
   150  			if len(input) == 0 {
   151  				t.Fatalf("%s:%d: out of sync: no input remaining", file, lineno)
   152  			}
   153  			var text string
   154  			text, input = input[0], input[1:]
   155  			if !isSingleBytes(text) && strings.Contains(re.String(), `\B`) {
   156  				// RE2's \B considers every byte position,
   157  				// so it sees 'not word boundary' in the
   158  				// middle of UTF-8 sequences. This package
   159  				// only considers the positions between runes,
   160  				// so it disagrees. Skip those cases.
   161  				continue
   162  			}
   163  			res := strings.Split(line, ";")
   164  			if len(res) != len(run) {
   165  				t.Fatalf("%s:%d: have %d test results, want %d", file, lineno, len(res), len(run))
   166  			}
   167  			for i := range res {
   168  				have, suffix := run[i](re, refull, text)
   169  				want := parseResult(t, file, lineno, res[i])
   170  				if !same(have, want) {
   171  					t.Errorf("%s:%d: %#q%s.FindSubmatchIndex(%#q) = %v, want %v", file, lineno, re, suffix, text, have, want)
   172  					if nfail++; nfail >= 100 {
   173  						t.Fatalf("stopping after %d errors", nfail)
   174  					}
   175  					continue
   176  				}
   177  				b, suffix := match[i](re, refull, text)
   178  				if b != (want != nil) {
   179  					t.Errorf("%s:%d: %#q%s.MatchString(%#q) = %v, want %v", file, lineno, re, suffix, text, b, !b)
   180  					if nfail++; nfail >= 100 {
   181  						t.Fatalf("stopping after %d errors", nfail)
   182  					}
   183  					continue
   184  				}
   185  			}
   186  
   187  		default:
   188  			t.Fatalf("%s:%d: out of sync: %s\n", file, lineno, line)
   189  		}
   190  	}
   191  	if err := scanner.Err(); err != nil {
   192  		t.Fatalf("%s:%d: %v", file, lineno, err)
   193  	}
   194  	if len(input) != 0 {
   195  		t.Fatalf("%s:%d: out of sync: have %d strings left at EOF", file, lineno, len(input))
   196  	}
   197  	t.Logf("%d cases tested", ncase)
   198  }
   199  
   200  var run = []func(*Regexp, *Regexp, string) ([]int, string){
   201  	runFull,
   202  	runPartial,
   203  	runFullLongest,
   204  	runPartialLongest,
   205  }
   206  
   207  func runFull(re, refull *Regexp, text string) ([]int, string) {
   208  	refull.longest = false
   209  	return refull.FindStringSubmatchIndex(text), "[full]"
   210  }
   211  
   212  func runPartial(re, refull *Regexp, text string) ([]int, string) {
   213  	re.longest = false
   214  	return re.FindStringSubmatchIndex(text), ""
   215  }
   216  
   217  func runFullLongest(re, refull *Regexp, text string) ([]int, string) {
   218  	refull.longest = true
   219  	return refull.FindStringSubmatchIndex(text), "[full,longest]"
   220  }
   221  
   222  func runPartialLongest(re, refull *Regexp, text string) ([]int, string) {
   223  	re.longest = true
   224  	return re.FindStringSubmatchIndex(text), "[longest]"
   225  }
   226  
   227  var match = []func(*Regexp, *Regexp, string) (bool, string){
   228  	matchFull,
   229  	matchPartial,
   230  	matchFullLongest,
   231  	matchPartialLongest,
   232  }
   233  
   234  func matchFull(re, refull *Regexp, text string) (bool, string) {
   235  	refull.longest = false
   236  	return refull.MatchString(text), "[full]"
   237  }
   238  
   239  func matchPartial(re, refull *Regexp, text string) (bool, string) {
   240  	re.longest = false
   241  	return re.MatchString(text), ""
   242  }
   243  
   244  func matchFullLongest(re, refull *Regexp, text string) (bool, string) {
   245  	refull.longest = true
   246  	return refull.MatchString(text), "[full,longest]"
   247  }
   248  
   249  func matchPartialLongest(re, refull *Regexp, text string) (bool, string) {
   250  	re.longest = true
   251  	return re.MatchString(text), "[longest]"
   252  }
   253  
   254  func isSingleBytes(s string) bool {
   255  	for _, c := range s {
   256  		if c >= utf8.RuneSelf {
   257  			return false
   258  		}
   259  	}
   260  	return true
   261  }
   262  
   263  func tryCompile(s string) (re *Regexp, err error) {
   264  	// Protect against panic during Compile.
   265  	defer func() {
   266  		if r := recover(); r != nil {
   267  			err = fmt.Errorf("panic: %v", r)
   268  		}
   269  	}()
   270  	return Compile(s)
   271  }
   272  
   273  func parseResult(t *testing.T, file string, lineno int, res string) []int {
   274  	// A single - indicates no match.
   275  	if res == "-" {
   276  		return nil
   277  	}
   278  	// Otherwise, a space-separated list of pairs.
   279  	n := 1
   280  	for j := 0; j < len(res); j++ {
   281  		if res[j] == ' ' {
   282  			n++
   283  		}
   284  	}
   285  	out := make([]int, 2*n)
   286  	i := 0
   287  	n = 0
   288  	for j := 0; j <= len(res); j++ {
   289  		if j == len(res) || res[j] == ' ' {
   290  			// Process a single pair.  - means no submatch.
   291  			pair := res[i:j]
   292  			if pair == "-" {
   293  				out[n] = -1
   294  				out[n+1] = -1
   295  			} else {
   296  				loStr, hiStr, _ := strings.Cut(pair, "-")
   297  				lo, err1 := strconv.Atoi(loStr)
   298  				hi, err2 := strconv.Atoi(hiStr)
   299  				if err1 != nil || err2 != nil || lo > hi {
   300  					t.Fatalf("%s:%d: invalid pair %s", file, lineno, pair)
   301  				}
   302  				out[n] = lo
   303  				out[n+1] = hi
   304  			}
   305  			n += 2
   306  			i = j + 1
   307  		}
   308  	}
   309  	return out
   310  }
   311  
   312  func same(x, y []int) bool {
   313  	if len(x) != len(y) {
   314  		return false
   315  	}
   316  	for i, xi := range x {
   317  		if xi != y[i] {
   318  			return false
   319  		}
   320  	}
   321  	return true
   322  }
   323  
   324  // TestFowler runs this package's regexp API against the
   325  // POSIX regular expression tests collected by Glenn Fowler
   326  // at http://www2.research.att.com/~astopen/testregex/testregex.html.
   327  func TestFowler(t *testing.T) {
   328  	files, err := filepath.Glob("testdata/*.dat")
   329  	if err != nil {
   330  		t.Fatal(err)
   331  	}
   332  	for _, file := range files {
   333  		t.Log(file)
   334  		testFowler(t, file)
   335  	}
   336  }
   337  
   338  var notab = MustCompilePOSIX(`[^\t]+`)
   339  
   340  func testFowler(t *testing.T, file string) {
   341  	f, err := os.Open(file)
   342  	if err != nil {
   343  		t.Error(err)
   344  		return
   345  	}
   346  	defer f.Close()
   347  	b := bufio.NewReader(f)
   348  	lineno := 0
   349  	lastRegexp := ""
   350  Reading:
   351  	for {
   352  		lineno++
   353  		line, err := b.ReadString('\n')
   354  		if err != nil {
   355  			if err != io.EOF {
   356  				t.Errorf("%s:%d: %v", file, lineno, err)
   357  			}
   358  			break Reading
   359  		}
   360  
   361  		// http://www2.research.att.com/~astopen/man/man1/testregex.html
   362  		//
   363  		// INPUT FORMAT
   364  		//   Input lines may be blank, a comment beginning with #, or a test
   365  		//   specification. A specification is five fields separated by one
   366  		//   or more tabs. NULL denotes the empty string and NIL denotes the
   367  		//   0 pointer.
   368  		if line[0] == '#' || line[0] == '\n' {
   369  			continue Reading
   370  		}
   371  		line = line[:len(line)-1]
   372  		field := notab.FindAllString(line, -1)
   373  		for i, f := range field {
   374  			if f == "NULL" {
   375  				field[i] = ""
   376  			}
   377  			if f == "NIL" {
   378  				t.Logf("%s:%d: skip: %s", file, lineno, line)
   379  				continue Reading
   380  			}
   381  		}
   382  		if len(field) == 0 {
   383  			continue Reading
   384  		}
   385  
   386  		//   Field 1: the regex(3) flags to apply, one character per REG_feature
   387  		//   flag. The test is skipped if REG_feature is not supported by the
   388  		//   implementation. If the first character is not [BEASKLP] then the
   389  		//   specification is a global control line. One or more of [BEASKLP] may be
   390  		//   specified; the test will be repeated for each mode.
   391  		//
   392  		//     B 	basic			BRE	(grep, ed, sed)
   393  		//     E 	REG_EXTENDED		ERE	(egrep)
   394  		//     A	REG_AUGMENTED		ARE	(egrep with negation)
   395  		//     S	REG_SHELL		SRE	(sh glob)
   396  		//     K	REG_SHELL|REG_AUGMENTED	KRE	(ksh glob)
   397  		//     L	REG_LITERAL		LRE	(fgrep)
   398  		//
   399  		//     a	REG_LEFT|REG_RIGHT	implicit ^...$
   400  		//     b	REG_NOTBOL		lhs does not match ^
   401  		//     c	REG_COMMENT		ignore space and #...\n
   402  		//     d	REG_SHELL_DOT		explicit leading . match
   403  		//     e	REG_NOTEOL		rhs does not match $
   404  		//     f	REG_MULTIPLE		multiple \n separated patterns
   405  		//     g	FNM_LEADING_DIR		testfnmatch only -- match until /
   406  		//     h	REG_MULTIREF		multiple digit backref
   407  		//     i	REG_ICASE		ignore case
   408  		//     j	REG_SPAN		. matches \n
   409  		//     k	REG_ESCAPE		\ to escape [...] delimiter
   410  		//     l	REG_LEFT		implicit ^...
   411  		//     m	REG_MINIMAL		minimal match
   412  		//     n	REG_NEWLINE		explicit \n match
   413  		//     o	REG_ENCLOSED		(|&) magic inside [@|&](...)
   414  		//     p	REG_SHELL_PATH		explicit / match
   415  		//     q	REG_DELIMITED		delimited pattern
   416  		//     r	REG_RIGHT		implicit ...$
   417  		//     s	REG_SHELL_ESCAPED	\ not special
   418  		//     t	REG_MUSTDELIM		all delimiters must be specified
   419  		//     u	standard unspecified behavior -- errors not counted
   420  		//     v	REG_CLASS_ESCAPE	\ special inside [...]
   421  		//     w	REG_NOSUB		no subexpression match array
   422  		//     x	REG_LENIENT		let some errors slide
   423  		//     y	REG_LEFT		regexec() implicit ^...
   424  		//     z	REG_NULL		NULL subexpressions ok
   425  		//     $	                        expand C \c escapes in fields 2 and 3
   426  		//     /	                        field 2 is a regsubcomp() expression
   427  		//     =	                        field 3 is a regdecomp() expression
   428  		//
   429  		//   Field 1 control lines:
   430  		//
   431  		//     C		set LC_COLLATE and LC_CTYPE to locale in field 2
   432  		//
   433  		//     ?test ...	output field 5 if passed and != EXPECTED, silent otherwise
   434  		//     &test ...	output field 5 if current and previous passed
   435  		//     |test ...	output field 5 if current passed and previous failed
   436  		//     ; ...	output field 2 if previous failed
   437  		//     {test ...	skip if failed until }
   438  		//     }		end of skip
   439  		//
   440  		//     : comment		comment copied as output NOTE
   441  		//     :comment:test	:comment: ignored
   442  		//     N[OTE] comment	comment copied as output NOTE
   443  		//     T[EST] comment	comment
   444  		//
   445  		//     number		use number for nmatch (20 by default)
   446  		flag := field[0]
   447  		switch flag[0] {
   448  		case '?', '&', '|', ';', '{', '}':
   449  			// Ignore all the control operators.
   450  			// Just run everything.
   451  			flag = flag[1:]
   452  			if flag == "" {
   453  				continue Reading
   454  			}
   455  		case ':':
   456  			var ok bool
   457  			if _, flag, ok = strings.Cut(flag[1:], ":"); !ok {
   458  				t.Logf("skip: %s", line)
   459  				continue Reading
   460  			}
   461  		case 'C', 'N', 'T', '0', '1', '2', '3', '4', '5', '6', '7', '8', '9':
   462  			t.Logf("skip: %s", line)
   463  			continue Reading
   464  		}
   465  
   466  		// Can check field count now that we've handled the myriad comment formats.
   467  		if len(field) < 4 {
   468  			t.Errorf("%s:%d: too few fields: %s", file, lineno, line)
   469  			continue Reading
   470  		}
   471  
   472  		// Expand C escapes (a.k.a. Go escapes).
   473  		if strings.Contains(flag, "$") {
   474  			f := `"` + field[1] + `"`
   475  			if field[1], err = strconv.Unquote(f); err != nil {
   476  				t.Errorf("%s:%d: cannot unquote %s", file, lineno, f)
   477  			}
   478  			f = `"` + field[2] + `"`
   479  			if field[2], err = strconv.Unquote(f); err != nil {
   480  				t.Errorf("%s:%d: cannot unquote %s", file, lineno, f)
   481  			}
   482  		}
   483  
   484  		//   Field 2: the regular expression pattern; SAME uses the pattern from
   485  		//     the previous specification.
   486  		//
   487  		if field[1] == "SAME" {
   488  			field[1] = lastRegexp
   489  		}
   490  		lastRegexp = field[1]
   491  
   492  		//   Field 3: the string to match.
   493  		text := field[2]
   494  
   495  		//   Field 4: the test outcome...
   496  		ok, shouldCompile, shouldMatch, pos := parseFowlerResult(field[3])
   497  		if !ok {
   498  			t.Errorf("%s:%d: cannot parse result %#q", file, lineno, field[3])
   499  			continue Reading
   500  		}
   501  
   502  		//   Field 5: optional comment appended to the report.
   503  
   504  	Testing:
   505  		// Run test once for each specified capital letter mode that we support.
   506  		for _, c := range flag {
   507  			pattern := field[1]
   508  			syn := syntax.POSIX | syntax.ClassNL
   509  			switch c {
   510  			default:
   511  				continue Testing
   512  			case 'E':
   513  				// extended regexp (what we support)
   514  			case 'L':
   515  				// literal
   516  				pattern = QuoteMeta(pattern)
   517  			}
   518  
   519  			for _, c := range flag {
   520  				switch c {
   521  				case 'i':
   522  					syn |= syntax.FoldCase
   523  				}
   524  			}
   525  
   526  			re, err := compile(pattern, syn, true)
   527  			if err != nil {
   528  				if shouldCompile {
   529  					t.Errorf("%s:%d: %#q did not compile", file, lineno, pattern)
   530  				}
   531  				continue Testing
   532  			}
   533  			if !shouldCompile {
   534  				t.Errorf("%s:%d: %#q should not compile", file, lineno, pattern)
   535  				continue Testing
   536  			}
   537  			match := re.MatchString(text)
   538  			if match != shouldMatch {
   539  				t.Errorf("%s:%d: %#q.Match(%#q) = %v, want %v", file, lineno, pattern, text, match, shouldMatch)
   540  				continue Testing
   541  			}
   542  			have := re.FindStringSubmatchIndex(text)
   543  			if (len(have) > 0) != match {
   544  				t.Errorf("%s:%d: %#q.Match(%#q) = %v, but %#q.FindSubmatchIndex(%#q) = %v", file, lineno, pattern, text, match, pattern, text, have)
   545  				continue Testing
   546  			}
   547  			if len(have) > len(pos) {
   548  				have = have[:len(pos)]
   549  			}
   550  			if !same(have, pos) {
   551  				t.Errorf("%s:%d: %#q.FindSubmatchIndex(%#q) = %v, want %v", file, lineno, pattern, text, have, pos)
   552  			}
   553  		}
   554  	}
   555  }
   556  
   557  func parseFowlerResult(s string) (ok, compiled, matched bool, pos []int) {
   558  	//   Field 4: the test outcome. This is either one of the posix error
   559  	//     codes (with REG_ omitted) or the match array, a list of (m,n)
   560  	//     entries with m and n being first and last+1 positions in the
   561  	//     field 3 string, or NULL if REG_NOSUB is in effect and success
   562  	//     is expected. BADPAT is acceptable in place of any regcomp(3)
   563  	//     error code. The match[] array is initialized to (-2,-2) before
   564  	//     each test. All array elements from 0 to nmatch-1 must be specified
   565  	//     in the outcome. Unspecified endpoints (offset -1) are denoted by ?.
   566  	//     Unset endpoints (offset -2) are denoted by X. {x}(o:n) denotes a
   567  	//     matched (?{...}) expression, where x is the text enclosed by {...},
   568  	//     o is the expression ordinal counting from 1, and n is the length of
   569  	//     the unmatched portion of the subject string. If x starts with a
   570  	//     number then that is the return value of re_execf(), otherwise 0 is
   571  	//     returned.
   572  	switch {
   573  	case s == "":
   574  		// Match with no position information.
   575  		ok = true
   576  		compiled = true
   577  		matched = true
   578  		return
   579  	case s == "NOMATCH":
   580  		// Match failure.
   581  		ok = true
   582  		compiled = true
   583  		matched = false
   584  		return
   585  	case 'A' <= s[0] && s[0] <= 'Z':
   586  		// All the other error codes are compile errors.
   587  		ok = true
   588  		compiled = false
   589  		return
   590  	}
   591  	compiled = true
   592  
   593  	var x []int
   594  	for s != "" {
   595  		var end byte = ')'
   596  		if len(x)%2 == 0 {
   597  			if s[0] != '(' {
   598  				ok = false
   599  				return
   600  			}
   601  			s = s[1:]
   602  			end = ','
   603  		}
   604  		i := 0
   605  		for i < len(s) && s[i] != end {
   606  			i++
   607  		}
   608  		if i == 0 || i == len(s) {
   609  			ok = false
   610  			return
   611  		}
   612  		var v = -1
   613  		var err error
   614  		if s[:i] != "?" {
   615  			v, err = strconv.Atoi(s[:i])
   616  			if err != nil {
   617  				ok = false
   618  				return
   619  			}
   620  		}
   621  		x = append(x, v)
   622  		s = s[i+1:]
   623  	}
   624  	if len(x)%2 != 0 {
   625  		ok = false
   626  		return
   627  	}
   628  	ok = true
   629  	matched = true
   630  	pos = x
   631  	return
   632  }
   633  
   634  var text []byte
   635  
   636  func makeText(n int) []byte {
   637  	if len(text) >= n {
   638  		return text[:n]
   639  	}
   640  	text = make([]byte, n)
   641  	x := ^uint32(0)
   642  	for i := range text {
   643  		x += x
   644  		x ^= 1
   645  		if int32(x) < 0 {
   646  			x ^= 0x88888eef
   647  		}
   648  		if x%31 == 0 {
   649  			text[i] = '\n'
   650  		} else {
   651  			text[i] = byte(x%(0x7E+1-0x20) + 0x20)
   652  		}
   653  	}
   654  	return text
   655  }
   656  
   657  func BenchmarkMatch(b *testing.B) {
   658  	isRaceBuilder := strings.HasSuffix(testenv.Builder(), "-race")
   659  
   660  	for _, data := range benchData {
   661  		r := MustCompile(data.re)
   662  		for _, size := range benchSizes {
   663  			if (isRaceBuilder || testing.Short()) && size.n > 1<<10 {
   664  				continue
   665  			}
   666  			t := makeText(size.n)
   667  			b.Run(data.name+"/"+size.name, func(b *testing.B) {
   668  				b.SetBytes(int64(size.n))
   669  				for i := 0; i < b.N; i++ {
   670  					if r.Match(t) {
   671  						b.Fatal("match!")
   672  					}
   673  				}
   674  			})
   675  		}
   676  	}
   677  }
   678  
   679  func BenchmarkMatch_onepass_regex(b *testing.B) {
   680  	isRaceBuilder := strings.HasSuffix(testenv.Builder(), "-race")
   681  	r := MustCompile(`(?s)\A.*\z`)
   682  	if r.onepass == nil {
   683  		b.Fatalf("want onepass regex, but %q is not onepass", r)
   684  	}
   685  	for _, size := range benchSizes {
   686  		if (isRaceBuilder || testing.Short()) && size.n > 1<<10 {
   687  			continue
   688  		}
   689  		t := makeText(size.n)
   690  		b.Run(size.name, func(b *testing.B) {
   691  			b.SetBytes(int64(size.n))
   692  			b.ReportAllocs()
   693  			for i := 0; i < b.N; i++ {
   694  				if !r.Match(t) {
   695  					b.Fatal("not match!")
   696  				}
   697  			}
   698  		})
   699  	}
   700  }
   701  
   702  var benchData = []struct{ name, re string }{
   703  	{"Easy0", "ABCDEFGHIJKLMNOPQRSTUVWXYZ$"},
   704  	{"Easy0i", "(?i)ABCDEFGHIJklmnopqrstuvwxyz$"},
   705  	{"Easy1", "A[AB]B[BC]C[CD]D[DE]E[EF]F[FG]G[GH]H[HI]I[IJ]J$"},
   706  	{"Medium", "[XYZ]ABCDEFGHIJKLMNOPQRSTUVWXYZ$"},
   707  	{"Hard", "[ -~]*ABCDEFGHIJKLMNOPQRSTUVWXYZ$"},
   708  	{"Hard1", "ABCD|CDEF|EFGH|GHIJ|IJKL|KLMN|MNOP|OPQR|QRST|STUV|UVWX|WXYZ"},
   709  }
   710  
   711  var benchSizes = []struct {
   712  	name string
   713  	n    int
   714  }{
   715  	{"16", 16},
   716  	{"32", 32},
   717  	{"1K", 1 << 10},
   718  	{"32K", 32 << 10},
   719  	{"1M", 1 << 20},
   720  	{"32M", 32 << 20},
   721  }
   722  
   723  func TestLongest(t *testing.T) {
   724  	re, err := Compile(`a(|b)`)
   725  	if err != nil {
   726  		t.Fatal(err)
   727  	}
   728  	if g, w := re.FindString("ab"), "a"; g != w {
   729  		t.Errorf("first match was %q, want %q", g, w)
   730  	}
   731  	re.Longest()
   732  	if g, w := re.FindString("ab"), "ab"; g != w {
   733  		t.Errorf("longest match was %q, want %q", g, w)
   734  	}
   735  }
   736  
   737  // TestProgramTooLongForBacktrack tests that a regex which is too long
   738  // for the backtracker still executes properly.
   739  func TestProgramTooLongForBacktrack(t *testing.T) {
   740  	longRegex := MustCompile(`(one|two|three|four|five|six|seven|eight|nine|ten|eleven|twelve|thirteen|fourteen|fifteen|sixteen|seventeen|eighteen|nineteen|twenty|twentyone|twentytwo|twentythree|twentyfour|twentyfive|twentysix|twentyseven|twentyeight|twentynine|thirty|thirtyone|thirtytwo|thirtythree|thirtyfour|thirtyfive|thirtysix|thirtyseven|thirtyeight|thirtynine|forty|fortyone|fortytwo|fortythree|fortyfour|fortyfive|fortysix|fortyseven|fortyeight|fortynine|fifty|fiftyone|fiftytwo|fiftythree|fiftyfour|fiftyfive|fiftysix|fiftyseven|fiftyeight|fiftynine|sixty|sixtyone|sixtytwo|sixtythree|sixtyfour|sixtyfive|sixtysix|sixtyseven|sixtyeight|sixtynine|seventy|seventyone|seventytwo|seventythree|seventyfour|seventyfive|seventysix|seventyseven|seventyeight|seventynine|eighty|eightyone|eightytwo|eightythree|eightyfour|eightyfive|eightysix|eightyseven|eightyeight|eightynine|ninety|ninetyone|ninetytwo|ninetythree|ninetyfour|ninetyfive|ninetysix|ninetyseven|ninetyeight|ninetynine|onehundred)`)
   741  	if !longRegex.MatchString("two") {
   742  		t.Errorf("longRegex.MatchString(\"two\") was false, want true")
   743  	}
   744  	if longRegex.MatchString("xxx") {
   745  		t.Errorf("longRegex.MatchString(\"xxx\") was true, want false")
   746  	}
   747  }
   748  

View as plain text