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  //
    66  func TestRE2Search(t *testing.T) {
    67  	testRE2(t, "testdata/re2-search.txt")
    68  }
    69  
    70  func testRE2(t *testing.T, file string) {
    71  	f, err := os.Open(file)
    72  	if err != nil {
    73  		t.Fatal(err)
    74  	}
    75  	defer f.Close()
    76  	var txt io.Reader
    77  	if strings.HasSuffix(file, ".bz2") {
    78  		z := bzip2.NewReader(f)
    79  		txt = z
    80  		file = file[:len(file)-len(".bz2")] // for error messages
    81  	} else {
    82  		txt = f
    83  	}
    84  	lineno := 0
    85  	scanner := bufio.NewScanner(txt)
    86  	var (
    87  		str       []string
    88  		input     []string
    89  		inStrings bool
    90  		re        *Regexp
    91  		refull    *Regexp
    92  		nfail     int
    93  		ncase     int
    94  	)
    95  	for lineno := 1; scanner.Scan(); lineno++ {
    96  		line := scanner.Text()
    97  		switch {
    98  		case line == "":
    99  			t.Fatalf("%s:%d: unexpected blank line", file, lineno)
   100  		case line[0] == '#':
   101  			continue
   102  		case 'A' <= line[0] && line[0] <= 'Z':
   103  			// Test name.
   104  			t.Logf("%s\n", line)
   105  			continue
   106  		case line == "strings":
   107  			str = str[:0]
   108  			inStrings = true
   109  		case line == "regexps":
   110  			inStrings = false
   111  		case line[0] == '"':
   112  			q, err := strconv.Unquote(line)
   113  			if err != nil {
   114  				// Fatal because we'll get out of sync.
   115  				t.Fatalf("%s:%d: unquote %s: %v", file, lineno, line, err)
   116  			}
   117  			if inStrings {
   118  				str = append(str, q)
   119  				continue
   120  			}
   121  			// Is a regexp.
   122  			if len(input) != 0 {
   123  				t.Fatalf("%s:%d: out of sync: have %d strings left before %#q", file, lineno, len(input), q)
   124  			}
   125  			re, err = tryCompile(q)
   126  			if err != nil {
   127  				if err.Error() == "error parsing regexp: invalid escape sequence: `\\C`" {
   128  					// We don't and likely never will support \C; keep going.
   129  					continue
   130  				}
   131  				t.Errorf("%s:%d: compile %#q: %v", file, lineno, q, err)
   132  				if nfail++; nfail >= 100 {
   133  					t.Fatalf("stopping after %d errors", nfail)
   134  				}
   135  				continue
   136  			}
   137  			full := `\A(?:` + q + `)\z`
   138  			refull, err = tryCompile(full)
   139  			if err != nil {
   140  				// Fatal because q worked, so this should always work.
   141  				t.Fatalf("%s:%d: compile full %#q: %v", file, lineno, full, err)
   142  			}
   143  			input = str
   144  		case line[0] == '-' || '0' <= line[0] && line[0] <= '9':
   145  			// A sequence of match results.
   146  			ncase++
   147  			if re == nil {
   148  				// Failed to compile: skip results.
   149  				continue
   150  			}
   151  			if len(input) == 0 {
   152  				t.Fatalf("%s:%d: out of sync: no input remaining", file, lineno)
   153  			}
   154  			var text string
   155  			text, input = input[0], input[1:]
   156  			if !isSingleBytes(text) && strings.Contains(re.String(), `\B`) {
   157  				// RE2's \B considers every byte position,
   158  				// so it sees 'not word boundary' in the
   159  				// middle of UTF-8 sequences. This package
   160  				// only considers the positions between runes,
   161  				// so it disagrees. Skip those cases.
   162  				continue
   163  			}
   164  			res := strings.Split(line, ";")
   165  			if len(res) != len(run) {
   166  				t.Fatalf("%s:%d: have %d test results, want %d", file, lineno, len(res), len(run))
   167  			}
   168  			for i := range res {
   169  				have, suffix := run[i](re, refull, text)
   170  				want := parseResult(t, file, lineno, res[i])
   171  				if !same(have, want) {
   172  					t.Errorf("%s:%d: %#q%s.FindSubmatchIndex(%#q) = %v, want %v", file, lineno, re, suffix, text, have, want)
   173  					if nfail++; nfail >= 100 {
   174  						t.Fatalf("stopping after %d errors", nfail)
   175  					}
   176  					continue
   177  				}
   178  				b, suffix := match[i](re, refull, text)
   179  				if b != (want != nil) {
   180  					t.Errorf("%s:%d: %#q%s.MatchString(%#q) = %v, want %v", file, lineno, re, suffix, text, b, !b)
   181  					if nfail++; nfail >= 100 {
   182  						t.Fatalf("stopping after %d errors", nfail)
   183  					}
   184  					continue
   185  				}
   186  			}
   187  
   188  		default:
   189  			t.Fatalf("%s:%d: out of sync: %s\n", file, lineno, line)
   190  		}
   191  	}
   192  	if err := scanner.Err(); err != nil {
   193  		t.Fatalf("%s:%d: %v", file, lineno, err)
   194  	}
   195  	if len(input) != 0 {
   196  		t.Fatalf("%s:%d: out of sync: have %d strings left at EOF", file, lineno, len(input))
   197  	}
   198  	t.Logf("%d cases tested", ncase)
   199  }
   200  
   201  var run = []func(*Regexp, *Regexp, string) ([]int, string){
   202  	runFull,
   203  	runPartial,
   204  	runFullLongest,
   205  	runPartialLongest,
   206  }
   207  
   208  func runFull(re, refull *Regexp, text string) ([]int, string) {
   209  	refull.longest = false
   210  	return refull.FindStringSubmatchIndex(text), "[full]"
   211  }
   212  
   213  func runPartial(re, refull *Regexp, text string) ([]int, string) {
   214  	re.longest = false
   215  	return re.FindStringSubmatchIndex(text), ""
   216  }
   217  
   218  func runFullLongest(re, refull *Regexp, text string) ([]int, string) {
   219  	refull.longest = true
   220  	return refull.FindStringSubmatchIndex(text), "[full,longest]"
   221  }
   222  
   223  func runPartialLongest(re, refull *Regexp, text string) ([]int, string) {
   224  	re.longest = true
   225  	return re.FindStringSubmatchIndex(text), "[longest]"
   226  }
   227  
   228  var match = []func(*Regexp, *Regexp, string) (bool, string){
   229  	matchFull,
   230  	matchPartial,
   231  	matchFullLongest,
   232  	matchPartialLongest,
   233  }
   234  
   235  func matchFull(re, refull *Regexp, text string) (bool, string) {
   236  	refull.longest = false
   237  	return refull.MatchString(text), "[full]"
   238  }
   239  
   240  func matchPartial(re, refull *Regexp, text string) (bool, string) {
   241  	re.longest = false
   242  	return re.MatchString(text), ""
   243  }
   244  
   245  func matchFullLongest(re, refull *Regexp, text string) (bool, string) {
   246  	refull.longest = true
   247  	return refull.MatchString(text), "[full,longest]"
   248  }
   249  
   250  func matchPartialLongest(re, refull *Regexp, text string) (bool, string) {
   251  	re.longest = true
   252  	return re.MatchString(text), "[longest]"
   253  }
   254  
   255  func isSingleBytes(s string) bool {
   256  	for _, c := range s {
   257  		if c >= utf8.RuneSelf {
   258  			return false
   259  		}
   260  	}
   261  	return true
   262  }
   263  
   264  func tryCompile(s string) (re *Regexp, err error) {
   265  	// Protect against panic during Compile.
   266  	defer func() {
   267  		if r := recover(); r != nil {
   268  			err = fmt.Errorf("panic: %v", r)
   269  		}
   270  	}()
   271  	return Compile(s)
   272  }
   273  
   274  func parseResult(t *testing.T, file string, lineno int, res string) []int {
   275  	// A single - indicates no match.
   276  	if res == "-" {
   277  		return nil
   278  	}
   279  	// Otherwise, a space-separated list of pairs.
   280  	n := 1
   281  	for j := 0; j < len(res); j++ {
   282  		if res[j] == ' ' {
   283  			n++
   284  		}
   285  	}
   286  	out := make([]int, 2*n)
   287  	i := 0
   288  	n = 0
   289  	for j := 0; j <= len(res); j++ {
   290  		if j == len(res) || res[j] == ' ' {
   291  			// Process a single pair.  - means no submatch.
   292  			pair := res[i:j]
   293  			if pair == "-" {
   294  				out[n] = -1
   295  				out[n+1] = -1
   296  			} else {
   297  				k := strings.Index(pair, "-")
   298  				if k < 0 {
   299  					t.Fatalf("%s:%d: invalid pair %s", file, lineno, pair)
   300  				}
   301  				lo, err1 := strconv.Atoi(pair[:k])
   302  				hi, err2 := strconv.Atoi(pair[k+1:])
   303  				if err1 != nil || err2 != nil || lo > hi {
   304  					t.Fatalf("%s:%d: invalid pair %s", file, lineno, pair)
   305  				}
   306  				out[n] = lo
   307  				out[n+1] = hi
   308  			}
   309  			n += 2
   310  			i = j + 1
   311  		}
   312  	}
   313  	return out
   314  }
   315  
   316  func same(x, y []int) bool {
   317  	if len(x) != len(y) {
   318  		return false
   319  	}
   320  	for i, xi := range x {
   321  		if xi != y[i] {
   322  			return false
   323  		}
   324  	}
   325  	return true
   326  }
   327  
   328  // TestFowler runs this package's regexp API against the
   329  // POSIX regular expression tests collected by Glenn Fowler
   330  // at http://www2.research.att.com/~astopen/testregex/testregex.html.
   331  func TestFowler(t *testing.T) {
   332  	files, err := filepath.Glob("testdata/*.dat")
   333  	if err != nil {
   334  		t.Fatal(err)
   335  	}
   336  	for _, file := range files {
   337  		t.Log(file)
   338  		testFowler(t, file)
   339  	}
   340  }
   341  
   342  var notab = MustCompilePOSIX(`[^\t]+`)
   343  
   344  func testFowler(t *testing.T, file string) {
   345  	f, err := os.Open(file)
   346  	if err != nil {
   347  		t.Error(err)
   348  		return
   349  	}
   350  	defer f.Close()
   351  	b := bufio.NewReader(f)
   352  	lineno := 0
   353  	lastRegexp := ""
   354  Reading:
   355  	for {
   356  		lineno++
   357  		line, err := b.ReadString('\n')
   358  		if err != nil {
   359  			if err != io.EOF {
   360  				t.Errorf("%s:%d: %v", file, lineno, err)
   361  			}
   362  			break Reading
   363  		}
   364  
   365  		// http://www2.research.att.com/~astopen/man/man1/testregex.html
   366  		//
   367  		// INPUT FORMAT
   368  		//   Input lines may be blank, a comment beginning with #, or a test
   369  		//   specification. A specification is five fields separated by one
   370  		//   or more tabs. NULL denotes the empty string and NIL denotes the
   371  		//   0 pointer.
   372  		if line[0] == '#' || line[0] == '\n' {
   373  			continue Reading
   374  		}
   375  		line = line[:len(line)-1]
   376  		field := notab.FindAllString(line, -1)
   377  		for i, f := range field {
   378  			if f == "NULL" {
   379  				field[i] = ""
   380  			}
   381  			if f == "NIL" {
   382  				t.Logf("%s:%d: skip: %s", file, lineno, line)
   383  				continue Reading
   384  			}
   385  		}
   386  		if len(field) == 0 {
   387  			continue Reading
   388  		}
   389  
   390  		//   Field 1: the regex(3) flags to apply, one character per REG_feature
   391  		//   flag. The test is skipped if REG_feature is not supported by the
   392  		//   implementation. If the first character is not [BEASKLP] then the
   393  		//   specification is a global control line. One or more of [BEASKLP] may be
   394  		//   specified; the test will be repeated for each mode.
   395  		//
   396  		//     B 	basic			BRE	(grep, ed, sed)
   397  		//     E 	REG_EXTENDED		ERE	(egrep)
   398  		//     A	REG_AUGMENTED		ARE	(egrep with negation)
   399  		//     S	REG_SHELL		SRE	(sh glob)
   400  		//     K	REG_SHELL|REG_AUGMENTED	KRE	(ksh glob)
   401  		//     L	REG_LITERAL		LRE	(fgrep)
   402  		//
   403  		//     a	REG_LEFT|REG_RIGHT	implicit ^...$
   404  		//     b	REG_NOTBOL		lhs does not match ^
   405  		//     c	REG_COMMENT		ignore space and #...\n
   406  		//     d	REG_SHELL_DOT		explicit leading . match
   407  		//     e	REG_NOTEOL		rhs does not match $
   408  		//     f	REG_MULTIPLE		multiple \n separated patterns
   409  		//     g	FNM_LEADING_DIR		testfnmatch only -- match until /
   410  		//     h	REG_MULTIREF		multiple digit backref
   411  		//     i	REG_ICASE		ignore case
   412  		//     j	REG_SPAN		. matches \n
   413  		//     k	REG_ESCAPE		\ to escape [...] delimiter
   414  		//     l	REG_LEFT		implicit ^...
   415  		//     m	REG_MINIMAL		minimal match
   416  		//     n	REG_NEWLINE		explicit \n match
   417  		//     o	REG_ENCLOSED		(|&) magic inside [@|&](...)
   418  		//     p	REG_SHELL_PATH		explicit / match
   419  		//     q	REG_DELIMITED		delimited pattern
   420  		//     r	REG_RIGHT		implicit ...$
   421  		//     s	REG_SHELL_ESCAPED	\ not special
   422  		//     t	REG_MUSTDELIM		all delimiters must be specified
   423  		//     u	standard unspecified behavior -- errors not counted
   424  		//     v	REG_CLASS_ESCAPE	\ special inside [...]
   425  		//     w	REG_NOSUB		no subexpression match array
   426  		//     x	REG_LENIENT		let some errors slide
   427  		//     y	REG_LEFT		regexec() implicit ^...
   428  		//     z	REG_NULL		NULL subexpressions ok
   429  		//     $	                        expand C \c escapes in fields 2 and 3
   430  		//     /	                        field 2 is a regsubcomp() expression
   431  		//     =	                        field 3 is a regdecomp() expression
   432  		//
   433  		//   Field 1 control lines:
   434  		//
   435  		//     C		set LC_COLLATE and LC_CTYPE to locale in field 2
   436  		//
   437  		//     ?test ...	output field 5 if passed and != EXPECTED, silent otherwise
   438  		//     &test ...	output field 5 if current and previous passed
   439  		//     |test ...	output field 5 if current passed and previous failed
   440  		//     ; ...	output field 2 if previous failed
   441  		//     {test ...	skip if failed until }
   442  		//     }		end of skip
   443  		//
   444  		//     : comment		comment copied as output NOTE
   445  		//     :comment:test	:comment: ignored
   446  		//     N[OTE] comment	comment copied as output NOTE
   447  		//     T[EST] comment	comment
   448  		//
   449  		//     number		use number for nmatch (20 by default)
   450  		flag := field[0]
   451  		switch flag[0] {
   452  		case '?', '&', '|', ';', '{', '}':
   453  			// Ignore all the control operators.
   454  			// Just run everything.
   455  			flag = flag[1:]
   456  			if flag == "" {
   457  				continue Reading
   458  			}
   459  		case ':':
   460  			i := strings.Index(flag[1:], ":")
   461  			if i < 0 {
   462  				t.Logf("skip: %s", line)
   463  				continue Reading
   464  			}
   465  			flag = flag[1+i+1:]
   466  		case 'C', 'N', 'T', '0', '1', '2', '3', '4', '5', '6', '7', '8', '9':
   467  			t.Logf("skip: %s", line)
   468  			continue Reading
   469  		}
   470  
   471  		// Can check field count now that we've handled the myriad comment formats.
   472  		if len(field) < 4 {
   473  			t.Errorf("%s:%d: too few fields: %s", file, lineno, line)
   474  			continue Reading
   475  		}
   476  
   477  		// Expand C escapes (a.k.a. Go escapes).
   478  		if strings.Contains(flag, "$") {
   479  			f := `"` + field[1] + `"`
   480  			if field[1], err = strconv.Unquote(f); err != nil {
   481  				t.Errorf("%s:%d: cannot unquote %s", file, lineno, f)
   482  			}
   483  			f = `"` + field[2] + `"`
   484  			if field[2], err = strconv.Unquote(f); err != nil {
   485  				t.Errorf("%s:%d: cannot unquote %s", file, lineno, f)
   486  			}
   487  		}
   488  
   489  		//   Field 2: the regular expression pattern; SAME uses the pattern from
   490  		//     the previous specification.
   491  		//
   492  		if field[1] == "SAME" {
   493  			field[1] = lastRegexp
   494  		}
   495  		lastRegexp = field[1]
   496  
   497  		//   Field 3: the string to match.
   498  		text := field[2]
   499  
   500  		//   Field 4: the test outcome...
   501  		ok, shouldCompile, shouldMatch, pos := parseFowlerResult(field[3])
   502  		if !ok {
   503  			t.Errorf("%s:%d: cannot parse result %#q", file, lineno, field[3])
   504  			continue Reading
   505  		}
   506  
   507  		//   Field 5: optional comment appended to the report.
   508  
   509  	Testing:
   510  		// Run test once for each specified capital letter mode that we support.
   511  		for _, c := range flag {
   512  			pattern := field[1]
   513  			syn := syntax.POSIX | syntax.ClassNL
   514  			switch c {
   515  			default:
   516  				continue Testing
   517  			case 'E':
   518  				// extended regexp (what we support)
   519  			case 'L':
   520  				// literal
   521  				pattern = QuoteMeta(pattern)
   522  			}
   523  
   524  			for _, c := range flag {
   525  				switch c {
   526  				case 'i':
   527  					syn |= syntax.FoldCase
   528  				}
   529  			}
   530  
   531  			re, err := compile(pattern, syn, true)
   532  			if err != nil {
   533  				if shouldCompile {
   534  					t.Errorf("%s:%d: %#q did not compile", file, lineno, pattern)
   535  				}
   536  				continue Testing
   537  			}
   538  			if !shouldCompile {
   539  				t.Errorf("%s:%d: %#q should not compile", file, lineno, pattern)
   540  				continue Testing
   541  			}
   542  			match := re.MatchString(text)
   543  			if match != shouldMatch {
   544  				t.Errorf("%s:%d: %#q.Match(%#q) = %v, want %v", file, lineno, pattern, text, match, shouldMatch)
   545  				continue Testing
   546  			}
   547  			have := re.FindStringSubmatchIndex(text)
   548  			if (len(have) > 0) != match {
   549  				t.Errorf("%s:%d: %#q.Match(%#q) = %v, but %#q.FindSubmatchIndex(%#q) = %v", file, lineno, pattern, text, match, pattern, text, have)
   550  				continue Testing
   551  			}
   552  			if len(have) > len(pos) {
   553  				have = have[:len(pos)]
   554  			}
   555  			if !same(have, pos) {
   556  				t.Errorf("%s:%d: %#q.FindSubmatchIndex(%#q) = %v, want %v", file, lineno, pattern, text, have, pos)
   557  			}
   558  		}
   559  	}
   560  }
   561  
   562  func parseFowlerResult(s string) (ok, compiled, matched bool, pos []int) {
   563  	//   Field 4: the test outcome. This is either one of the posix error
   564  	//     codes (with REG_ omitted) or the match array, a list of (m,n)
   565  	//     entries with m and n being first and last+1 positions in the
   566  	//     field 3 string, or NULL if REG_NOSUB is in effect and success
   567  	//     is expected. BADPAT is acceptable in place of any regcomp(3)
   568  	//     error code. The match[] array is initialized to (-2,-2) before
   569  	//     each test. All array elements from 0 to nmatch-1 must be specified
   570  	//     in the outcome. Unspecified endpoints (offset -1) are denoted by ?.
   571  	//     Unset endpoints (offset -2) are denoted by X. {x}(o:n) denotes a
   572  	//     matched (?{...}) expression, where x is the text enclosed by {...},
   573  	//     o is the expression ordinal counting from 1, and n is the length of
   574  	//     the unmatched portion of the subject string. If x starts with a
   575  	//     number then that is the return value of re_execf(), otherwise 0 is
   576  	//     returned.
   577  	switch {
   578  	case s == "":
   579  		// Match with no position information.
   580  		ok = true
   581  		compiled = true
   582  		matched = true
   583  		return
   584  	case s == "NOMATCH":
   585  		// Match failure.
   586  		ok = true
   587  		compiled = true
   588  		matched = false
   589  		return
   590  	case 'A' <= s[0] && s[0] <= 'Z':
   591  		// All the other error codes are compile errors.
   592  		ok = true
   593  		compiled = false
   594  		return
   595  	}
   596  	compiled = true
   597  
   598  	var x []int
   599  	for s != "" {
   600  		var end byte = ')'
   601  		if len(x)%2 == 0 {
   602  			if s[0] != '(' {
   603  				ok = false
   604  				return
   605  			}
   606  			s = s[1:]
   607  			end = ','
   608  		}
   609  		i := 0
   610  		for i < len(s) && s[i] != end {
   611  			i++
   612  		}
   613  		if i == 0 || i == len(s) {
   614  			ok = false
   615  			return
   616  		}
   617  		var v = -1
   618  		var err error
   619  		if s[:i] != "?" {
   620  			v, err = strconv.Atoi(s[:i])
   621  			if err != nil {
   622  				ok = false
   623  				return
   624  			}
   625  		}
   626  		x = append(x, v)
   627  		s = s[i+1:]
   628  	}
   629  	if len(x)%2 != 0 {
   630  		ok = false
   631  		return
   632  	}
   633  	ok = true
   634  	matched = true
   635  	pos = x
   636  	return
   637  }
   638  
   639  var text []byte
   640  
   641  func makeText(n int) []byte {
   642  	if len(text) >= n {
   643  		return text[:n]
   644  	}
   645  	text = make([]byte, n)
   646  	x := ^uint32(0)
   647  	for i := range text {
   648  		x += x
   649  		x ^= 1
   650  		if int32(x) < 0 {
   651  			x ^= 0x88888eef
   652  		}
   653  		if x%31 == 0 {
   654  			text[i] = '\n'
   655  		} else {
   656  			text[i] = byte(x%(0x7E+1-0x20) + 0x20)
   657  		}
   658  	}
   659  	return text
   660  }
   661  
   662  func BenchmarkMatch(b *testing.B) {
   663  	isRaceBuilder := strings.HasSuffix(testenv.Builder(), "-race")
   664  
   665  	for _, data := range benchData {
   666  		r := MustCompile(data.re)
   667  		for _, size := range benchSizes {
   668  			if (isRaceBuilder || testing.Short()) && size.n > 1<<10 {
   669  				continue
   670  			}
   671  			t := makeText(size.n)
   672  			b.Run(data.name+"/"+size.name, func(b *testing.B) {
   673  				b.SetBytes(int64(size.n))
   674  				for i := 0; i < b.N; i++ {
   675  					if r.Match(t) {
   676  						b.Fatal("match!")
   677  					}
   678  				}
   679  			})
   680  		}
   681  	}
   682  }
   683  
   684  func BenchmarkMatch_onepass_regex(b *testing.B) {
   685  	isRaceBuilder := strings.HasSuffix(testenv.Builder(), "-race")
   686  	r := MustCompile(`(?s)\A.*\z`)
   687  	if r.onepass == nil {
   688  		b.Fatalf("want onepass regex, but %q is not onepass", r)
   689  	}
   690  	for _, size := range benchSizes {
   691  		if (isRaceBuilder || testing.Short()) && size.n > 1<<10 {
   692  			continue
   693  		}
   694  		t := makeText(size.n)
   695  		b.Run(size.name, func(b *testing.B) {
   696  			b.SetBytes(int64(size.n))
   697  			b.ReportAllocs()
   698  			for i := 0; i < b.N; i++ {
   699  				if !r.Match(t) {
   700  					b.Fatal("not match!")
   701  				}
   702  			}
   703  		})
   704  	}
   705  }
   706  
   707  var benchData = []struct{ name, re string }{
   708  	{"Easy0", "ABCDEFGHIJKLMNOPQRSTUVWXYZ$"},
   709  	{"Easy0i", "(?i)ABCDEFGHIJklmnopqrstuvwxyz$"},
   710  	{"Easy1", "A[AB]B[BC]C[CD]D[DE]E[EF]F[FG]G[GH]H[HI]I[IJ]J$"},
   711  	{"Medium", "[XYZ]ABCDEFGHIJKLMNOPQRSTUVWXYZ$"},
   712  	{"Hard", "[ -~]*ABCDEFGHIJKLMNOPQRSTUVWXYZ$"},
   713  	{"Hard1", "ABCD|CDEF|EFGH|GHIJ|IJKL|KLMN|MNOP|OPQR|QRST|STUV|UVWX|WXYZ"},
   714  }
   715  
   716  var benchSizes = []struct {
   717  	name string
   718  	n    int
   719  }{
   720  	{"16", 16},
   721  	{"32", 32},
   722  	{"1K", 1 << 10},
   723  	{"32K", 32 << 10},
   724  	{"1M", 1 << 20},
   725  	{"32M", 32 << 20},
   726  }
   727  
   728  func TestLongest(t *testing.T) {
   729  	re, err := Compile(`a(|b)`)
   730  	if err != nil {
   731  		t.Fatal(err)
   732  	}
   733  	if g, w := re.FindString("ab"), "a"; g != w {
   734  		t.Errorf("first match was %q, want %q", g, w)
   735  	}
   736  	re.Longest()
   737  	if g, w := re.FindString("ab"), "ab"; g != w {
   738  		t.Errorf("longest match was %q, want %q", g, w)
   739  	}
   740  }
   741  
   742  // TestProgramTooLongForBacktrack tests that a regex which is too long
   743  // for the backtracker still executes properly.
   744  func TestProgramTooLongForBacktrack(t *testing.T) {
   745  	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)`)
   746  	if !longRegex.MatchString("two") {
   747  		t.Errorf("longRegex.MatchString(\"two\") was false, want true")
   748  	}
   749  	if longRegex.MatchString("xxx") {
   750  		t.Errorf("longRegex.MatchString(\"xxx\") was true, want false")
   751  	}
   752  }
   753  

View as plain text