github_mirrors/vectorscan
3
0
Fork 1
mirror of https://github.com/VectorCamp/vectorscan.git synced 2025-09-30 03:34:25 +03:00
Code Issues Packages Projects Releases Wiki Activity

hscollider: example test cases

Browse Source
This commit is contained in:
Alex Coyte
2018-01-10 13:38:20 +11:00
committed by Xiang Wang
parent 934d6f2dad
commit 905a67375a
57 changed files with 21058 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
tools/hscollider/test_cases/pcre/accel.txt Normal file
View File

@@ -0,0 +1,3 @@
2000:/abcdef[^X].*[^X](A|B(CDE)?F)Y?foobar/s
2001:/[^abab]+\Z/si
2002:/mj(?:[pajl]|o)[tplh]thd.hq.b[frt]pk(\b)hb/i8L

174
tools/hscollider/test_cases/pcre/anchors.txt Normal file
View File

@@ -0,0 +1,174 @@
# A set of patterns that use anchors in unusual ways, also makes some use of
# capturing groups.
13000:/(^)?foo/O
13001:/(^)??foo/O
13002:/(^|)foo/O
13003:/(|^)foo/O
13004:/(^|^)?foo/O
13005:/(^||^)foo/O
13006:/foo($)?/
13007:/foo($)??/
13008:/foo($|)/
13009:/foo(|$)/
13010:/foo($||$)/
13011:/foo(\z)?/
13012:/foo(\z)??/
13013:/foo(\z|)/
13014:/foo(|\z)/
13015:/foo(\z||\z)/
13016:/(?:^|foo|)foo/O
13017:/(^)*foo/O
13018:/(^)*?foo/O
13019:/(^)+foo/O
13020:/(^)+?foo/O
13021:/(^){1}foo/O
# More anchoring tests from the rose anchoring work
13100:/a\Z/siH
13101:/a\Z/
13102:/\Aa\Z/iH
13103:/\A(\Ab\Z|a)/smi
13104:/(a|b|b$)/
13105:/(b.|cbd|[dbd]$)/si
13106:/(^h|[qj]|m|a)$/
# Various trailing .* cases
13107:/foobar.*\z/
13108:/foobar.*\z/sO
13109:/foobar.*\z/m
13110:/foobar.*\z/smO
13111:/foobar.*\Z/
13112:/foobar.*\Z/s
13113:/foobar.*\Z/m
13114:/foobar.*\Z/sm
13115:/foobar.*$/
13116:/foobar.*$/s
13117:/foobar.*$/m
13118:/foobar.*$/sm
13119:/foobar.*($|)/
13120:/foobar.*($|)/s
13121:/foobar.*($|)/m
13122:/foobar.*($|)/sm
13123:/foobar.*(\z)?/
13124:/foobar.*(\z)?/s
13125:/foobar.*(\z)?/m
13126:/foobar.*(\z)?/sm
# Some alternation variants of the above
13127:/(foo|bar).*\z/
13128:/(foo|bar).*\z/sO
13129:/(foo|bar).*\z/m
13130:/(foo|bar).*\z/smO
13131:/(foo|bar).*\Z/
13132:/(foo|bar).*\Z/s
13133:/(foo|bar).*\Z/m
13134:/(foo|bar).*\Z/sm
13135:/(foo|bar).*$/
13136:/(foo|bar).*$/s
13137:/(foo|bar).*$/m
13138:/(foo|bar).*$/sm
13139:/(foo|bar).*($|)/
13140:/(foo|bar).*($|)/s
13141:/(foo|bar).*($|)/m
13142:/(foo|bar).*($|)/sm
13143:/(foo|bar).*(\z)?/
13144:/(foo|bar).*(\z)?/s
13145:/(foo|bar).*(\z)?/m
13146:/(foo|bar).*(\z)?/sm
# Rose support for .{N,}$
13147:/abc.+$/s
13148:/abc.{2,}$/s
13149:/abc.{3,}$/s
13150:/abc.{10,}$/s
13151:/abc.{150,}$/s
# More bounded and escaped EOD fun
13152:/abc.{2,}\z/s
13153:/abc.{2,5}\z/s
13154:/abc.{2,50}\z/s
13155:/abc.{2,5}$/s
13156:/abc.{2,50}$/s
13157:/abc[^xyz]{2,}\z/s
13158:/abc[^xyz]{2,5}\z/s
13159:/abc[^xyz]{2,50}\z/s
13160:/abc[^xyz]{2,}$/s
13161:/abc[^xyz]{2,5}$/s
13162:/abc[^xyz]{2,50}$/s
# ... fun
13163:/.\z/s
13164:/.$/s
13165:/^.+\z/s
13166:/..\z/s
13167:/^.{50}foo.bar/s
# Anchored literals
13168:/^(.*\.)?trailer\z/s
13169:/^trailer\z/m
13170:/^(.*\.)?trailer$/s
13171:/^trailer$/m
13172:/\Ap\z|s|\A|n/
13173:/\Ap\z|s|\A|\Ab|n/
# Bi-anchored, non cyclic outfix
13174:/^\w{200}$/
# Leading anchored dots
13200:/^.foo/sO
13201:/^.{1}foo/sO
13202:/^.{2}foo/sO
13203:/^.{30}foo/sO
13204:/^.{1,30}foo/sO
13205:/^.{20,}foo/sO
13206:/^.{20,30}foo/sO
13207:/^.{0,20}foo/sO
13208:/^.{1,}foo/sO
13209:/^.?foo/sO
13210:/^.+foo/sO
13211:/^.*foo/sO
13212:/^.{4,20}foo|bar/sO
13213:/^(.{4,20}foo)|bar/sO
13214:/^(.{4,20}foo)|^bar/sO
13215:/^.{20,30}[^f][^o][^o]/sO
13216:/^.{0,}foo/sO
13217:/^.{20,30}a/sO
13218:/^[\x00-\xff]+/s
13219:/^..foo/sO
13220:/^..*foo/sO
13221:/^..*.foo/sO
13222:/^.*.foo/sO
13223:/^..+foo/sO
13224:/^..+.foo/sO
13225:/^.+.foo/sO
13226:/..foo/sO
13227:/..*foo/sO
13228:/..*.foo/sO
13229:/.*.foo/sO
13230:/..+foo/sO
13231:/..+.foo/sO
13232:/.+.foo/sO
13233:/(^.*.bar)?foo/sO
13234:/(^...bar)?foo/sO
13235:/((^....*bar)|(.._))foo/s
13236:/^.{0,20}foo[a-z]/sO
13237:/^.{0,20}fo[oO]/sO
13238:/^.(.{0,2}|.{0,2})foo/sO
13240:/(a|aaaa{4,}|^(a|[^a]|a)?.a[^a]*a)$/
13241:/^.{600,}/s
13242:/.{600,}/s

25
tools/hscollider/test_cases/pcre/approximate_matching.txt Normal file
View File

@@ -0,0 +1,25 @@
# patterns that should produce matches
40000:/^(012)*test$/ms{edit_distance=1}
40001:/[^k]{3}$/ms{edit_distance=2}
40002:/^test$/ms{edit_distance=1}
40003:/^test/s{edit_distance=2}
40004:/test/s{edit_distance=2}
40005:/test|^lit/s{edit_distance=2}
40006:/tta+tt/s{edit_distance=2}
40007:/^ab(..)+bc/s{edit_distance=2}
40008:/a( bc|d)e/s{edit_distance=1}
40009:/abc|def$/s{edit_distance=2}
40010:/abc|def$/ms{edit_distance=2}
40050:/^(012)*test$/ms{hamming_distance=1}
40051:/[^k]{3}$/ms{hamming_distance=2}
40052:/^test$/ms{hamming_distance=1}
40053:/^test/s{hamming_distance=2}
40054:/test/s{hamming_distance=2}
40055:/test|^lit/s{hamming_distance=2}
40056:/tta+tt/s{hamming_distance=2}
40057:/^ab(..)+bc/s{hamming_distance=2}
40058:/a( bc|d)e/s{hamming_distance=1}
40059:/abc|def$/s{hamming_distance=2}
40060:/abc|def$/ms{hamming_distance=2}
40061:/^a?/m{hamming_distance=1}

86
tools/hscollider/test_cases/pcre/asserts.txt Normal file
View File

@@ -0,0 +1,86 @@
23000:/foo.*\bbar/s
23001:/foo\b.*bar/s
23002:/\bfoo/
23003:/\Bfoo/O
23004:/(word|nonword~).*\balpha/s
23005:/\B~/
23006:/foo\b/
23007:/foo\B/
23008:/foo\b$/
23009:/foo\b\z/O
23010:/foo.*\b\bbar/s
23011:/foo.*\b\b\bbar/s
23012:/foo\b.*\bbar/s
23013:/foo\B.*\Bbar/s
23014:/foo\b.*\Bbar/s
23015:/\b\bfoo/
23016:/\B\Bfoo/O
23017:/\B\B~/
23018:/foo\b\b/
23019:/foo\B\B/
23020:/foo\b\b$/
23021:/foo\b\b\z/O
23022:/a[b~]+\b/
23023:/\b[b~]+a/
23024:/\ba/
23025:/\b/
23026:/\B/
23027:/\b\z/
23028:/\B\z/
23029:/^\b/
23030:/^\B/
23031:/(a|\A\b)/
23032:/(^a|\b)/
23033:/^\bfoo/O
23034:/^\Bfoo/O
23035:/^\b\bfoo/O
23036:/^\B\Bfoo/O
23037:/foo.*(\b|\B)bar/O
23038:/three.*\b\b\basserts/
23039:/three.*\B\B\Basserts/
23040:/can't_match\b\B/O
23041:/\b\Bcan't_match/O
23042:/\b(.*)\b/s
23043:/\b(foo|bar|baz)\b/
# More tests: repeats
23044:/((\b){2,})+/
23045:/((\b){10,})+/
23046:/((\b|a){2,})+/
23047:/(\b[a-f]\b)+/
23048:/(((\b[a-f]+\b) ?))+/
# multiline bi-anchored boundaries
23049:/^(\B)/m
23050:/^(\b)/m
23051:/^\b$/m
23052:/^\b\Z/m
23053:/^\b\z/mO
23054:/\A\b$/m
23055:/\A\b\Z/m
23056:/\A\b\z/m
23057:/^\B$/m
23058:/^\B\Z/m
23059:/^\B\z/m
23060:/\A\B$/m
23061:/\A\B\Z/m
23062:/\A\B\z/m
23063:/\b.*\b/s
23064:/\B.*\B/s
23065:/\b.*\B/s
23066:/\B.*\b/s
23067:/\b.+\b/s
23068:/\B.+\B/s
23069:/\b.+\B/s
23070:/\B.+\b/s
23071:/l(\B.)*/i
23072:/(a?.\b){4,}bbabb/
23073:/\A\B/
23074:/\A\b/
23075:/\b[a-f]+\b/
# Asserts near repeats.
23076:/godzilla\b.{0,10}mothra/s
23077:/godzilla.{0,10}\bmothra/s
23078:/godzilla\b.{0,10}\bmothra/s

34
tools/hscollider/test_cases/pcre/benefits.txt Normal file
View File

@@ -0,0 +1,34 @@
19501:/a..b/sO
19502:/a..bc/sO
19503:/ab..c/sO
19504:/ab..c../sO
19505:/ab..c../siO
19506:/.ab..c../siO
19507:/abcdefgh....abcd....efgh/sO
19508:/foo..ba/sO
19509:/foo..ba\z/sO
# Longer masks
19510:/[a-f]{3}-[a-f]{3}-[a-f]{4}/
# Mixed-sensitivity literals can use masks.
19511:/ab(?i)cdef(?-i)ghi/
# Fixed width pattern that's too long to use a benefits mask.
19512:/a{10}b{10}[Cc]{10}d{10}/
# Mixed sensitivity and dot mask
19513:/f[Yy][mPiU].W/
# Short mixed-case literals to stress literal matchers
19514:/(?i)godzill(?-i)a/
19515:/g(?i)odzilla/
19516:/G(?i)odzilla/
19517:/(?i)god(?-i)z(?i)illa/
19518:/g(?i)od(?-i)z(?i)illa/
# Masks and mixed-case
19519:/^nqt(?-i)qkf{14}bdr+k.t(?i)r[cp]q{3}\z/is
# More mixed-case
19520:/[Bb][Cc][Aa][CDc]abaaEbcd/

30
tools/hscollider/test_cases/pcre/capturing.txt Normal file
View File

@@ -0,0 +1,30 @@
25000:/((e[dcb]ce(\B)a|.*b(d.|d|c|.|ad)ad[acd]b|a{4,10}|d)[cd]|ba[bcbc].b.ebcd{0,})/m
25001:/(a[^aaaa]aaaaa..aaa{0,}|.aaa)/
25002:/(\A[aab]{1}|^a.|[aca](\B)baacccc(\b).((b|a)){1,}|aa[bc]cc.\z|\A[aa].cabc.|a(b|[bca]|c|c))/s
25003:/aaaaaa(a?|aaaa|a|.?)a+(.a.a.(\b)([aaa]|[aa]|.)aaaaa{4,}[aaa]aaa(a|[aa]))?/mi
25004:/(aaa{14,16}aa.a.a|[aaaa]aaa[aa]|(aa|a|a|[aaaa]|a)a[^aaaa]aaaaa)/smi
25005:/((aaaa[^\n]*aaaaa)|(aa)|(aa))/s
25006:/((abaaa.+a)|(aa))/s
25007:/[^p]b{3,20}/
25008:/((acbddbdcccc)|(cc))/s
25009:/(aa(a|a|a|a|a|[aa])){1,7}/s
25010:/aa(a|a+)/s
25011:/bba*/i
25012:/hg[hkjs]*/s
25013:/ke+/i
25014:/.(a|.)a?/m
25015:/(\b)../i
25016:/../
25017:/a/s
25018:/(a{5}$|(a|[aaa]|[aaaa])aaa[^aaa])/sm
25019:/(\B)..a*/i
25020:/(.bb|bbaababb|\Aba.*|b?(b|[ab]|a)[aab]b([ba]b?|bb)|\Abaaaaaa(a|a|.))/smi
25021:/foobar/
25022:/foobar/i
25023:/foo_bar/
25024:/foo_bar/i
25025:/[fg]oobar/
25026:/[fg]oobar/i
25027:/f[0o]obar/i
25028:/foo(?i)bar/
25029:/[Ff]oobar/

174
tools/hscollider/test_cases/pcre/charclass.txt Normal file
View File

@@ -0,0 +1,174 @@
# All the generic character types
11000:/\d\D\h\H\s\S\v\V\w\W/
# All the following should be equivalent to a dot (type plus non-type)
11001:/[\d\D]/
11002:/[\h\H]/
11003:/[\s\S]/
11004:/[\v\V]/
11005:/[\w\W]/
# POSIX classes
11006:/[[:alnum:]][[:alpha:]][[:ascii:]][[:blank:]][[:cntrl:]][[:digit:]][[:graph:]]/
11007:/[[:lower:]][[:print:]][[:punct:]][[:space:]][[:upper:]][[:word:]][[:xdigit:]]/
11008:/[01[:alpha:]%]/
# Perl extension: negation of posix classes. This matches 1, 2 or any non-digit.
11009:/[12[:^digit:]]/
# Ranges
11010:/[A-F][a-f][0-9][a-f5-7][^t-z][^a-f0-9]/
11011:/[^^][^\n]/
11012:/[\000-\037]/
11013:/[W-c]/i
11014:/[^\W_]/
# Literal 'dash' characters
11015:/[\s-\w]/
11016:/[A-]/O
11017:/[A-Z-]/
11018:/[a-f0-]/O
# \b is a backspace in a character class (hex 08)
11020:/[\b]/O
# Dollar and dot have no special meaning inside a class
11021:/[.][$]/O
# Closing square brackets can be unescaped if they are the first element in a
# class (scary!)
11022:/[]a]/O
11023:/[^]a]/
11024:/[]-_]/O
# \R and \X are the literal chars R and X (unlike outside a class)
# 11025:/[\R\X]/O
# Others
11100:/[\]]/O
11101:/[]]/O
11102:/[\[\]]/O
11103:/[\^]/O
11104:/[+--]/O
11105:/[--A]/
11106:/[A-C-E-G]/O
11107:/[A-C-E]/O
11108:/[\0]/O
11109:/[\40]/O
11110:/[\040]/O
11111:/[\0400]/O
11112:/[\0401]/O
#hint: equiv to [18\x00]
11113:/[\81]/O
11114:/[\xg]/O
11115:/[\x31]/O
11116:/[\x{31}]/O
11117:/[\x{0000000000000031}]/O
11118:/[\x{31g}]/O
11119:/[\x{foo}-~]/O
11120:/[\x00-\x{31g}]/O
11121:/[[:foo]/O
11122:/[\Q^\Ea]/O
11123:/[\Qa]\E]/O
11124:/[\Q\Q\E]/O
11125:/[\E]]/O
11126:/[\x{31]/O
11127:/[^e]/i
11128:/[(?i)a]/O
# These classes with escaped '-' chars in them aren't ranges
11200:/[\x20\x2d\x5f]/O
11201:/[A\x2dZ]/O
11202:/[A\055Z]/O
11203:/[-\x2da]/O
11204:/[\x20\x2d-]/O
# Is a range (dash-to-\x5f)
11205:/[\x2d-\x5f]/s
# void character classed
11300:/[^\x00-\xff]/O
11301:/[^\x00-\xff]foo/O
11302:/[^\x00-\xff]foo|bar/O
11303:/^[^\x00-\xff]foo/O
11304:/foo[^\x00-\xff]/O
11305:/foo[^\x00-\xff]$/O
11306:/baz.*([^\x00-\xff]foo|bar).*baz/O
# we had a buggy defn of xdigit (as '[0-9a-fA-Z]')
11307:/[[:xdigit:]]/
# test negation of all the POSIX classes
11308:/[[:^alnum:]][[:^alpha:]][[:^ascii:]][[:^blank:]][[:^cntrl:]][[:^digit:]][[:^graph:]]/
11309:/[[:^lower:]][[:^print:]][[:^punct:]][[:^space:]][[:^upper:]][[:^word:]][[:^xdigit:]]/
# test cntrl against DEL
11310:/[[:cntrl:]]/
# \C should just be the literal char C
#11311:/[\C]/O
11312:/[[:alnum:]]/
11313:/[[:^alnum:]]/
11314:/[[:alpha:]]/
11315:/[[:^alpha:]]/
11316:/[[:ascii:]]/
11317:/[[:^ascii:]]/
11318:/[[:blank:]]/
11319:/[[:^blank:]]/
11320:/[[:cntrl:]]/
11321:/[[:^cntrl:]]/
11322:/[[:digit:]]/
11323:/[[:^digit:]]/
11324:/[[:graph:]]/
11325:/[[:^graph:]]/
11326:/[[:lower:]]/
11327:/[[:^lower:]]/
11328:/[[:print:]]/
11329:/[[:^print:]]/
11330:/[[:space:]]/
11331:/[[:^space:]]/
11332:/[[:upper:]]/
11333:/[[:^upper:]]/
11334:/[[:xdigit:]]/
11335:/[[:^xdigit:]]/
11339:/[[:punct:]]/
11340:/[[:^punct:]]/
# Oddities
11336:/[\f]/
11337:/[\a]/
11338:/[\e]/
# Unterminated classes.
11341:/[ab-]/
11342:/[ab-\Q\E]/
# More negated POSIX classes and interaction with mode bits.
11343:/[[:^upper:]]/i
11344:/[[:^lower:]]/i
11345:/[^A-Z]/i
11346:/[^a-z]/i
# More unterminated classes.
11347:/[ab-]/8
11348:/[ab-\Q\E]/8
# Some classes that look a little similar to POSIX collating classes, but
# aren't.
11349:/[.abc=] [=abc.] [^=abc=] [^.abc.]/
# Bug in PCRE versions before 8.38 (see PCRE issue #1697).
11350:/[\W\p{Any}]/
# Bug in PCRE versions before 8.38 (see PCRE issue #1717).
11351:/a[[:punct:]b]/W
11352:/[[:^graph:]a]/W
# Bug in PCRE versions before 8.38 (see PCRE issue #1732).
11353:/[^[:alpha:][:^cntrl:]]/8W
# Bug in PCRE versions before 8.38 (see PCRE issue #1719).
11354:/[[:^ascii:][:alnum:]a]/8W
# Bug in PCRE < 8.38 to do with [:punct:] in UCP mode (see PCRE issue #1718).
11355:/[[:punct:]]/8W

9
tools/hscollider/test_cases/pcre/comp.txt Normal file
View File

@@ -0,0 +1,9 @@
15800:/^.*(foo.*bar|baz.*baz)/sO
15801:/^.*(foo.*bar|baz.*baz)a/s
15802:/^.{1,6}(foo.*bar|baz.*baz)/sO
15803:/.{1,6}(foo.*bar|baz.*baz)/sO
15804:/.{1,6}(foo.*bar|........baz.*baz)/sO
15805:/^.{1,}(foo.*bar|........baz.*baz)/sO
15806:/anorak.*trainspotter|ANORAK.*trainspotter/iO
15807:/anorak.*trainspotter/iO
15808:/anorak.*trainspotter\z|ANORAK.*trainspotter/iO

10
tools/hscollider/test_cases/pcre/comptree.txt Normal file
View File

@@ -0,0 +1,10 @@
1000:/(foo){2,}bar/
1001:/(foo){2,3}bar/
1002:/^(foo){2,}bar/
1003:/^(foo){2,3}bar/
1004:/x?(foo){2,}bar/
1005:/x?(foo){2,3}bar/
1006:/(foo){2,}bar/{min_length=10}
1007:/(foo){2,3}bar/{min_length=10}
1008:/(foo){2,}bar/L
1009:/(foo){2,3}bar/L

119
tools/hscollider/test_cases/pcre/extparams.txt Normal file
View File

@@ -0,0 +1,119 @@
16200:/foo.*bar/{min_offset=1}
16201:/foo.*bar/{min_offset=10}
16202:/foo.*bar/{max_offset=10}
16203:/foo.*bar/{min_offset=10,max_offset=10}
16204:/foo.*bar/{min_offset=10,max_offset=15}
16205:/a.*b/{min_length=5}
16206:/a.*b/{min_offset=10,min_length=3}
16207:/preamble.*q/{max_offset=30}
16208:/preamble.*q/{min_length=12}
16209:/aa[^a]+aa/{max_offset=10}
16210:/[0-9]{32,}/{max_offset=48}
16211:/\x04\S+/{max_offset=17,min_length=17}
16212:/[^x]+/{min_offset=10,max_offset=10}
16213:/[^x]+x/{min_length=20}
16214:/foo.*/{min_length=10}
16215:/foo.+/{min_length=10}
16216:/^hatstand.*teakettle/{min_length=25}
16217:/hatstand/{min_offset=10,max_offset=15}
16218:/[abcdef]{3}/{min_offset=5,max_offset=10}
# Test min_length with one cyclic -> bounded repeat transform.
16219:/foo.*bar/{min_length=10}
16220:/foo.+bar/{min_length=10}
16221:/.*./{min_length=4}
16222:/.+/{min_length=4}
16223:/^.*hatstand/{min_length=20}
16224:/^.+hatstand/{min_length=20}
16225:/hatstand.*/{min_length=20}
# Unnecessary min_length
16226:/long cat is lo+ng/{min_length=8}
# Be wary of assertions.
16227:/(\B|\Al)/smiV{min_offset=1,max_offset=10}
# Anchoring via max_offset.
16228:/rascal/{max_offset=6}
# Be wary of vacuous patterns.
16229:/(..v[xmdf]wn\b)*/V{max_offset=27}
16230:/abc([^a]|ab|a[^b]c)d+ef/{max_offset=30,min_length=10}
# Some alternations are disallowed by min_length. The next two also have $
# metachars, which invoke -1 offset adjustment.
16231:/(^g$|k\z|egs$|t)/{min_length=1}
16232:/(h|.ab$)/{min_length=2}
16233:/(abc|abcd|abcde|abcdef)/{min_length=5}
# Some highlander optimisations play merry hell with min_length.
16234:/kn[er]{2,10}/sH{min_length=5}
# Mixed anchored/unanchored pattern with offsets.
16235:/(\As|^c|z|[ycld]|^.)/s{min_offset=4,max_offset=22}
# More highlander shenanigans.
16236:/h|z|w|efp./H{min_offset=4}
# \b or \B offset adjustments can cause trouble.
16237:/\A(g|l|(\b)|[wfse]|^[wc])/s{min_length=1,max_offset=4}
16238:/g(\B)/sV{max_offset=2}
# min_length with virtual starts (multiline)
16239:/^p{1,}/m{min_length=5,max_offset=15}
# vacuous edges that need to to away.
16240:/c?/V{min_length=1,min_offset=4}
# min_length -> bounded repeat for a trailing cyclic.
16241:/ykmy[^kaib]g*/{min_length=8}
# Alternation that stresses our min_length/max offset transforms.
16242:/\Aq[rgm]h+|z/i{min_length=8,max_offset=20}
16243:/cj.wjn*v?/{min_length=9,min_offset=4}
# min_length transformation with an offset-adjusted report.
16244:/qye.+ys(\B)/si{min_length=7}
# More word-boundary tests.
16245:/\bfoo/{min_offset=2,max_offset=10}
16246:/foo\b/{min_offset=2,max_offset=10}
16247:/\bfoo\b/{min_offset=2,max_offset=10}
16248:/\bfoo\b.*\bbar\b/{min_offset=2,max_offset=15}
16249:/\bfoo\b.*\bbar\b/{min_length=10}
16250:/\bfoo\b.*\bbar\b/{max_offset=15}
# highlander + min_offset puff/lbr tests.
16251:/.{50}/H{min_offset=51}
16252:/.{50}/H{min_offset=52}
16253:/.{50}/H{min_offset=53}
16254:/.{50,}/H{min_offset=51}
16255:/.{50,}/H{min_offset=52}
16256:/.{50,}/H{min_offset=53}
16257:/aaa.{50}/H{min_offset=54}
16258:/aaa.{50}/H{min_offset=55}
16259:/aaa.{50}/H{min_offset=56}
16260:/aaa.{50,}/H{min_offset=54}
16261:/aaa.{50,}/H{min_offset=55}
16262:/aaa.{50,}/H{min_offset=56}
# unnecessary min_length
16263:/unambiguous/{min_length=11}
# cases with prunable paths
16264:/^a|g/m{min_offset=10,max_offset=16}
16265:/^foo|jabberwocky|apple2|foo.*bar/{max_offset=6}
16266:/^(a{8}|b{9}|c{10})|floating/{min_offset=10}
# some more cases to stress small block analyses
16267:/abcdef.{5,}/s{min_offset=20}
16268:/abcdef./s{min_offset=20}
16269:/abcdef../s{min_offset=20}
16270:/abcdef/sH{min_offset=10,max_offset=20}
16271:/abcdef/s{min_offset=10,max_offset=10}
16272:/abcdef../s{min_offset=10,max_offset=10}
# several things at once
16273:/^[^dj].b$/sH8{min_length=9,max_offset=19}

37
tools/hscollider/test_cases/pcre/highlander.txt Normal file
View File

@@ -0,0 +1,37 @@
15300:/foo/HO
15301:/foo.*bar/sHO
15302:/foo.*bar/HO
15303:/foo[^X]{15}/sHO
15304:/foo[^X]{16}/sHO
15305:/foo[^X]{17}/sHO
15306:/foo.*[^X]{17}/sH
15307:/foo[^X]{17}blah/sHO
15308:/foo[^XY]{17}/sHO
15309:/foo[^X]{17}$/sH
15310:/[^X]{17}/sH
15311:/^[^X]{17}/sHO
15312:/fo.*o[^X]{15}/sH
15313:/fo.*o[^X]{16}/sH
15314:/fo.*o[^X]{17}/sH
15315:/[fb][oa][or][^X]{15}/sH
15316:/[fb][oa][or][^X]{16}/sH
15317:/[fb][oa][or][^X]{17}/sH
15318:/foo|bar/HO
# for people who are unsure of highlander semantics
15319:/foo.|bar./HO
15320:/^(a[^aa]..aa|a*)/mHV
# highlander pruning
15321:/(foo.*bar)|(foo.*bar.*baz)/H
15322:/foo.*bar(.*baz)?/H
15323:/foo.*bar+/H
15324:/(foo.*bar)|(foo.*bar.*baz)|(foo.*bar.*eod\z)/H
15325:/foo.*eod+\z/H
15326:/foo.*eod+$/H
# SEP patterns
15327:/[a-f]/iH
15328:/a|b|c/H
# Longer, run-prone literals
15329:/(p{100})|(q{100})/H

7
tools/hscollider/test_cases/pcre/longlits.txt Normal file
View File

@@ -0,0 +1,7 @@
# Alternation of ten long literals.
50000:/ukdnsybecoqlszhxiwfcpvmnoqobdfuoovmotdeefiwdoxukyxldjxthcmnxqebsiyvwwtadafmibpxnwuxqtpcndwzaiwurnvbkgzpfutisauyagfwacajrcmjlgmomzdakzjgpgnlepnjcynzhptporgjcrjkrnhvnucgvjjgfboisxjfaywypljihrstqvmwsqdvq|ihcntajbgquaruyfimiabusvmmqcpaxpowhhucnzlpfxzmmbcqahdmposiymqscqugtmictrnomnccfcdxzlksyuqkbjvgekaebwmcmzydebtltpcfbmckvwoqtinlplzopauzkcyiinbcjrfjkncggcowuifvwvoavxrkuaxuwjhnnyotkgbrkggwkafzvzmkgijnsr|vtymzjxeeyazemvcwsvcacdzihfbgiaqwjxmcncgdzafmhtvbvnmjrpfudnflcvfbkwcfsmdfaqtawqqcbigfrnjzwrdvndstesayfgjsiofshzvtabtgblrgbksqechctlngykpladacvwiffqwjktuosdjvdonjixekrlvvxeeqenylwjgqicdpgjhojsyyhuhtphc|oxqvlanpjsnxnoodzpjhsnmgkwjfyqxmqlbqmteabqdbnwtvnpeodvvkukcxqfczyrftogophbmkeuzcxmpyqfigkynftdstdudfynmfxhssbrpdhebywvnpltqxtvdthpppdllyofyktnrwzhbiklpfrclizidkkqbgimzmdrznwkmbledtmsazljcvmfzdlpzgwymm|jtaanwnxkvwesndylwqcnsqlccnikybengkimrxauuvytagyasdomvupykaagpcthrrnkzjnysqywwyqpfbyqclabrcftwgvfcxdrhkgxlngnsxmkrvguvfugbgqruspkzojxrhqkrgjrybyoqktjrexxodbcdlfiyhclsvhsaysihhoycnpksoivyxxbglvyzhkbajh|ntafhsdlngeeeqphxqspxswcmubcwycbzusxunxrmqgmdktwblavcrdgjhecpfnqsyckxuxaboljmrvetofhgbeydypoeyydrxhordcgqafbnoylylqnvxynfcoygtiwiniwlctwmwornplgfpjbretneadneemlzzodtkkdmcyqrggrmjzlkzzjxoirfilosenpjexy|ckalndydcrodvvmyyuqbihprzzgnqympoeinwewgfqpzuhyygivfdhdxnnatccuaghjrddogabtgmcvpspptpicpftxdfdfsiilngteqvqjjsoevnqfiztgcvolmpqkemqeizzmlingcuyxyidvrlczmiifutjljifxiramtoxvtbkwzsrczyzdgbtkboudipjonydtt|fbbeioibbkbiiupjzcqrwjuvqjmbavnkvtogebhltedefahasbnvvvspugdtecfpstxsbtfluycxzfxgcvzhfyhgbgzyfcwltvyyoofolnolasemxqqywlrikjocwvhpqofufqyuhcisckvoveaeectwodmmcodisfwynzcctloqyheedjfpwcuwrixkdznnefgizrap|eosktuskzdokmshljlcazpmahliwzzmhmpzmsiymtvpctaqwdpmffcnkmkypkcrclmlcxmnysqhslegqetflncttxqiprjddowzkhyjlzytudxqnvcctpebufelzmxnzsfwqbahrgwbjrpbobfdwjfsbfjrhjsbqdlsurllezccluashcrywxhnbqqclikrnefkyutdo|gvoiwjevplfxkeempnspkgljnqdckunshelsuogizffvbplhbyhxnjfabmjiigideullxtxbnjxczvaoveafcechrilvdkyzehhuhlohtjxiocfvjzdrjosuawxqmlbcwsnfnpxusoqldoumsedxbbwummwtbqrwkcjxkvyukcxekpjacjlezesaihhpqdatiosxgbbb/
# Long flood literals.
50001:/a{300}/
50002:/A{300}/
50003:/a{300}/i

8
tools/hscollider/test_cases/pcre/lookaround.txt Normal file
View File

@@ -0,0 +1,8 @@
27000:/a[A-L]?[B-M]?[C-N]?[D-O]?foobar/
27001:/ar[A-Z]{1,9}foobar/
27002:/a{5}\w{1,8}foobar/
27003:/(ab|bc|cd|de|ef|fg|gh|hi)[\w]{12}foobar/
27004:/(abcdef|bcdefg|cdefgh|defghi|efghij|fghijk|ghijkl|hijklm)[\w]{8}foobar/
27005:/[\x20\x31\x42\x53\x64][\x20\x25\x31\x35\x42\x45\x53\x55\x64\x65][\w]{1,7}foobar/
27006:/[\dABcd][\dABCd][\dAbcd][\dabcd][\dabCD][\dEFGH][\dEFGh][\dEfgh][0-5B-Iw-z][\dA-Z]{1,6}foobar/
27007:/ec.((c|C|d){3}|.BEe|...|dE.[bdC]A.Ac[Aa]|ca)CC.cdCA.D{1,}[Dd]..b(dbe{16,18}|E|b)aaC(c|d*|e|[Da]|C|a)/s

132
tools/hscollider/test_cases/pcre/mangle.txt Normal file
View File

@@ -0,0 +1,132 @@
18000:/^foo.{64}bar/sO
18001:/^foo.{32}bar/sO
18002:/^foo.{600}bar/sO
18003:/^hoo.{64}bar|^foo.{64}bar.{178}bar/sO
18004:/foo.{64}bar/sO
18005:/^foo.{64}b(a?)r/sO
18006:/^f(o?)o.{64}bar/sO
18007:/^foo.{64}bar\z/sO
18008:/^foo.{64}bar(|t\z)/s
18009:/^foo.{63,64}bar/sO
18010:/^foo.{1,64}bar/sO
18011:/^foo.{599,600}bar/sO
18012:/^foo.{1,600}bar/sO
18013:/^(foo|fab).{400,600}(bar|baz)/sO
18014:/^foo.{63,64}.bar/sO
18015:/^foo.{1,600}.bar/sO
18016:/^foo.{63}(|.?bar)/s
18017:/^foo.{63}(|.{0,4}bar)/s
18018:/^foo.{32}.?.{32}bar/sO
18019:/^foo(..){1,32}bar/s
18020:/^(foo.{32}|).{32}bar/s
18021:/^(foo.{32}|).{0,32}bar/s
18022:/^(foo.{32}|).{1,32}bar/s
18023:/^foo.{64}bar/sHO
18024:/^foo.{1,64}bar/sHO
18025:/^foo.{63,64}bar/sHO
18026:/^foo(.{64})?bar/sH
18027:/^foo(.{1,64})?bar/sH
18028:/^foo(.{63,64})?bar/sH
18029:/^foo.{64}bar/O
18030:/^foo.{32}bar/O
18031:/^foo.{600}bar/O
18032:/^hoo.{64}bar|^foo.{64}bar.{178}bar/O
18033:/foo.{64}bar/O
18034:/^foo.{64}b(a?)r/O
18035:/^f(o?)o.{64}bar/O
18036:/^foo.{64}bar\z/O
18037:/^foo.{64}bar(|t\z)/
18038:/^foo.{63,64}bar/O
18039:/^foo.{1,64}bar/O
18040:/^foo.{599,600}bar/O
18041:/^foo.{1,600}bar/O
18042:/^(foo|fab).{400,600}(bar|baz)/O
18043:/^foo.{63,64}.bar/O
18044:/^foo.{1,1000}.bar/O
18045:/^foo.{63}(|.?bar)/
18046:/^foo.{63}(|.{0,4}bar)/
18047:/^foo.{32}.?.{32}bar/O
18048:/^foo(..){1,32}bar/
18049:/^(foo.{32}|).{32}bar/
18050:/^(foo.{32}|).{0,32}bar/
18051:/^(foo.{32}|).{1,32}bar/
18052:/^foo.{64}bar/HO
18053:/^foo.{1,64}bar/HO
18054:/^foo.{63,64}bar/HO
18055:/^foo(.{64})?bar/H
18056:/^foo(.{1,64})?bar/H
18057:/^foo(.{63,64})?bar/H
18058:/^fooa{599,600}bar/O
18059:/^fooa{1,600}bar/
18060:/^fooa{600}bar/O
18061:/^foo.{0,64}bar/sO
18062:/^foo.{0,600}bar/sO
18063:/^.{70}(aaaa|.{12})/s
18100:/^[^X]foo.{64}bar/sO
18101:/^[^X]foo.{32}bar/sO
18102:/^[^X]foo.{600}bar/s
18103:/^[^X]hoo.{64}bar|^[^X]foo.{130}bar.{178}bar/s
18105:/^[^X]foo.{64}b(a?)r/s
18106:/^[^X]f(o?)o.{64}bar/s
18107:/^[^X]foo.{64}bar\z/s
18108:/^[^X]foo.{64}bar(|t\z)/s
18109:/^[^X]foo.{63,64}bar/s
18110:/^[^X]foo.{1,64}bar/s
18111:/^[^X]foo.{599,600}bar/s
18112:/^[^X]foo.{1,600}bar/s
18113:/^[^X](foo|fab).{400,600}(bar|baz)/s
18114:/^[^X]foo.{63,64}.bar/s
18115:/^[^X]foo.{1,600}.bar/s
18116:/^[^X]foo.{63}(|.?bar)/s
18117:/^[^X]foo.{63}(|.{0,4}bar)/s
18118:/^[^X]foo.{32}.?.{32}bar/s
18119:/^[^X]foo(..){1,32}bar/s
18120:/^[^X](foo.{32}|).{32}bar/s
18121:/^[^X](foo.{32}|).{0,32}bar/s
18122:/^[^X](foo.{32}|).{1,32}bar/s
18123:/^[^X]foo.{64}bar/sHO
18124:/^[^X]foo.{1,64}bar/sH
18125:/^[^X]foo.{63,64}bar/sH
18126:/^[^X]foo(.{64})?bar/sH
18127:/^[^X]foo(.{1,64})?bar/sH
18128:/^[^X]foo(.{63,64})?bar/sH
18129:/^[^X]foo.{64}bar/O
18130:/^[^X]foo.{32}bar/O
18131:/^[^X]foo.{600}bar/
18132:/^[^X]hoo.{64}bar|^[^X]foo.{64}bar.{178}bar/O
18134:/^[^X]foo.{64}b(a?)r/
18135:/^[^X]f(o?)o.{64}bar/
18136:/^[^X]foo.{64}bar\z/
18137:/^[^X]foo.{64}bar(|t\z)/
18138:/^[^X]foo.{63,64}bar/
18139:/^[^X]foo.{1,64}bar/
18140:/^[^X]foo.{599,600}bar/
18141:/^[^X]foo.{1,600}bar/
18142:/^[^X](foo|fab).{400,600}(bar|baz)/
18143:/^[^X]foo.{63,64}.bar/
18144:/^[^X]foo.{1,1000}.bar/
18145:/^[^X]foo.{63}(|.?bar)/
18146:/^[^X]foo.{63}(|.{0,4}bar)/
18147:/^[^X]foo.{32}.?.{32}bar/
18148:/^[^X]foo(..){1,32}bar/
18149:/^[^X](foo.{32}|).{32}bar/
18150:/^[^X](foo.{32}|).{0,32}bar/
18151:/^[^X](foo.{32}|).{1,32}bar/
18152:/^[^X]foo.{64}bar/HO
18153:/^[^X]foo.{1,64}bar/H
18154:/^[^X]foo.{63,64}bar/H
18155:/^[^X]foo(.{64})?bar/H
18156:/^[^X]foo(.{1,64})?bar/H
18157:/^[^X]foo(.{63,64})?bar/H
18158:/^[^X]fooa{599,600}bar/
18159:/^[^X]fooa{1,600}bar/
18160:/^[^X]fooa{600}bar/
18161:/^[^X]foo.{0,64}bar/s
18162:/^[^X]foo.{0,600}bar/s
18163:/^[^X].{70}(aaaa|.{12})/s
18164:/^[^X].{70}(aaaa|.{0,11}.b)/s
18165:/^[^X].{70}(aaaa|.{1,11}.b)/s
18166:/^[^X]foo.{0,100}.bar/s
18167:/^[^X]foo.{100}.?bar/s
18168:/^[^X]foo.{50}.?.{50}bar/s
18169:/^[^X]foo.{50}(..)?.{50}bar/s

29
tools/hscollider/test_cases/pcre/mcclellan.txt Normal file
View File

@@ -0,0 +1,29 @@
15900:/foo.+ba[rR]/sO
15901:/foo.*ba[rR]/sO
15902:/foo.*ba[rR].*tea/sO
15903:/fo(|o.*ba[rR].*tea)/s
15904:/fo($|o.*ba[rR].*tea)/s
15905:/foo.*ba[rR]|tea.*cof[^f]ee/s
15906:/(foo|bar).*gaz/sO
15907:/foo.*(bar|gaz)/sO
15908:/foo.*(bar|gaz).*tea/sO
15909:/foo.*a(bar|gaz).*tea/sO
15910:/foo.*a(bar|gaz).*(tea|aet)/sO
15911:/foo(b[^a]r|g[^a]z).*tea/s
15912:/foo(b[^a]r|g[^a]z).*(tea|aet)/s
15913:/^p;;.*[_;]*.*:/sO
15914:/p;;.*[_;]*.*:/sO
15915:/p;;.*[_;]*.*:/O
15916:/^AA.*(a.*(Z|XX)|b.*(Z|YY))/s
15917:/^AA.*(a.*(Z|XX)|b.*Z)/s
15918:/^pppp;;.*[_;]*.*:/sO
15919:/pppp;;.*[_;]*.*:/sO
15920:/pppp;;.*[_;]*.*:/O
15921:/^AAAAA.*(a.*(Z|XX)|b.*(Z|YY))/s
15922:/^AAAAAA.*(a.*(Z|XX)|b.*Z)/s
15923:/(bdcd{26}|([kjrs]|b|b)+){2,4}/s
15924:/literal\w{0,7}\d+/s
15925:/literal\w{0,10}\d+/s
15926:/literal\w{0,10}/s
15927:/literal\w{5,}/s
15928:/[ab][^ab]a.a/s8

66
tools/hscollider/test_cases/pcre/metacharacters.txt Normal file
View File

@@ -0,0 +1,66 @@
# Some tests for specific special metacharacters
# matches any byte
24000:/\C/
# alarm, that is, the BEL character (hex 07)
24001:/\a/O
# "control-x", where x is any character
24002:/\cz/O
24003:/\c{/O
24004:/\c;/O
24005:/\ca/O
24006:/\c0/O
# escape (hex 1B)
24007:/\e/O
# formfeed (hex 0C)
24008:/\f/O
# linefeed (hex 0A)
24009:/\n/O
# carriage return (hex 0D)
24010:/\r/O
# tab (hex 09)
24011:/\t/O
# real hex escapes
24012:/\xdc\x{dc}\x00\x{ff}/O
# broken hex escape, interpreted as a null followed by some chars
24013:/\x{dc/O
# more control
24015:/\cA/O
24016:/[\cz]/O
24017:/[\c{]/O
24018:/[\c;]/O
24019:/[\ca]/O
24020:/[\c0]/O
24021:/[\cA]/O
24022:/\c\n/O
# \8 and \9 are not back-references, they are simply the literals 8 and 9.
24023:/\8 literal \9/
# Big numbers
24024:/bignum \1111111111/
24025:/bignum \2147483639/
24026:/bignum \18888/
24027:/bignum \128888/
24028:/bignum \1238888/
24029:/bignum \3778888/
# Octal escapes
24030:/\060/
24031:/\60/
24032:/\12/
24033:/\012/
24034:/\0120/
24035:/\120/
24036:/\377/
24037:/\80/

182
tools/hscollider/test_cases/pcre/notbob.txt Normal file
View File

@@ -0,0 +1,182 @@
15500:/notbob/sO
15501:/^notbob/sO
15502:/\Anotbob/sO
15503:/.notbob/sO
15504:/^.notbob/sO
15505:/\A.notbob/sO
15506:/.{6}notbob/sO
15507:/^.{6}notbob/sO
15508:/\A.{6}notbob/sO
15509:/.*notbob/sO
15510:/^.*notbob/sO
15511:/\A.*notbob/sO
15512:/.+notbob/sO
15513:/^.+notbob/sO
15514:/\A.+notbob/sO
15515:/.{4,7}notbob/sO
15516:/^.{4,7}notbob/sO
15517:/\A.{4,7}notbob/sO
15518:/(\A|)notbob/sO
15519:/(^|.)notbob/s
15520:/not.*bob/sO
15521:/^not.*bob/sO
15522:/\Anot.*bob/sO
15523:/.not.*bob/sO
15524:/^.not.*bob/sO
15525:/\A.not.*bob/sO
15526:/.{6}not.*bob/sO
15527:/^.{6}not.*bob/sO
15528:/\A.{6}not.*bob/sO
15529:/.*not.*bob/sO
15530:/^.*not.*bob/sO
15531:/\A.*not.*bob/sO
15532:/.+not.*bob/sO
15533:/^.+not.*bob/sO
15534:/\A.+not.*bob/sO
15535:/.{4,7}not.*bob/sO
15536:/^.{4,7}not.*bob/sO
15537:/\A.{4,7}not.*bob/sO
15538:/(\A|)not.*bob/sO
15539:/(^|.)not.*bob/s
15540:/no[A-Z].*bar/s
15541:/^no[A-Z].*bar/sO
15542:/\Ano[A-Z].*bar/sO
15543:/.no[A-Z].*bar/s
15544:/^.no[A-Z].*bar/s
15545:/\A.no[A-Z].*bar/s
15546:/.{6}no[A-Z].*bar/s
15547:/^.{6}no[A-Z].*bar/s
15548:/\A.{6}no[A-Z].*bar/s
15549:/.*no[A-Z].*bar/s
15550:/^.*no[A-Z].*bar/s
15551:/\A.*no[A-Z].*bar/s
15552:/.+no[A-Z].*bar/s
15553:/^.+no[A-Z].*bar/s
15554:/\A.+no[A-Z].*bar/s
15555:/.{4,7}no[A-Z].*bar/s
15556:/^.{4,7}no[A-Z].*bar/s
15557:/\A.{4,7}no[A-Z].*bar/s
15558:/(\A|)no[A-Z].*bar/s
15559:/(^|.)no[A-Z].*bar/s
15560:/notbobno[A-Z].*bar/s
15561:/^notbobno[A-Z].*bar/sO
15562:/\Anotbobno[A-Z].*bar/sO
15563:/.notbobno[A-Z].*bar/s
15564:/^.notbobno[A-Z].*bar/s
15565:/\A.notbobno[A-Z].*bar/s
15566:/.{6}notbobno[A-Z].*bar/s
15567:/^.{6}notbobno[A-Z].*bar/s
15568:/\A.{6}notbobno[A-Z].*bar/s
15569:/.*notbobno[A-Z].*bar/s
15570:/^.*notbobno[A-Z].*bar/s
15571:/\A.*notbobno[A-Z].*bar/s
15572:/.+notbobno[A-Z].*bar/s
15573:/^.+notbobno[A-Z].*bar/s
15574:/\A.+notbobno[A-Z].*bar/s
15575:/.{4,7}notbobno[A-Z].*bar/s
15576:/^.{4,7}notbobno[A-Z].*bar/s
15577:/\A.{4,7}notbobno[A-Z].*bar/s
15578:/(\A|)notbobno[A-Z].*bar/s
15579:/(^|.)notbobno[A-Z].*bar/s
15580:/^.{40}/sO
15581:/^[^X]{40}/sO
15582:/\A[^X]{40}/sO
15583:/[^X]{40}/s
15584:/.{4,}notbobno[A-Z].*bar/s
15585:/^.{4,}no[A-Z].*bar/s
15586:/\A.{4,}notbobno[A-Z].*bar/s
15587:/.{4,}notbobno[A-Z].*bar/s
15588:/^.{4,}no[A-Z].*bar/s
15589:/\A.{4,}notbobno[A-Z].*bar/s
15590:/(^.*bar)|bob/sO
15600:/notbob/smO
15601:/^notbob/smO
15602:/\Anotbob/smO
15603:/.notbob/smO
15604:/^.notbob/smO
15605:/\A.notbob/smO
15606:/.{6}notbob/smO
15607:/^.{6}notbob/smO
15608:/\A.{6}notbob/smO
15609:/.*notbob/smO
15610:/^.*notbob/smO
15611:/\A.*notbob/smO
15612:/.+notbob/smO
15613:/^.+notbob/smO
15614:/\A.+notbob/smO
15615:/.{4,7}notbob/smO
15616:/^.{4,7}notbob/smO
15617:/\A.{4,7}notbob/smO
15618:/(\A|)notbob/smO
15619:/(^|.)notbob/sm
15620:/not.*bob/smO
15621:/^not.*bob/smO
15622:/\Anot.*bob/smO
15623:/.not.*bob/smO
15624:/^.not.*bob/smO
15625:/\A.not.*bob/smO
15626:/.{6}not.*bob/smO
15627:/^.{6}not.*bob/smO
15628:/\A.{6}not.*bob/smO
15629:/.*not.*bob/smO
15630:/^.*not.*bob/smO
15631:/\A.*not.*bob/smO
15632:/.+not.*bob/smO
15633:/^.+not.*bob/smO
15634:/\A.+not.*bob/smO
15635:/.{4,7}not.*bob/smO
15636:/^.{4,7}not.*bob/smO
15637:/\A.{4,7}not.*bob/smO
15638:/(\A|)not.*bob/smO
15639:/(^|.)not.*bob/sm
15640:/no[A-Z].*bar/sm
15641:/^no[A-Z].*bar/sm
15642:/\Ano[A-Z].*bar/smO
15643:/.no[A-Z].*bar/sm
15644:/^.no[A-Z].*bar/sm
15645:/\A.no[A-Z].*bar/sm
15646:/.{6}no[A-Z].*bar/sm
15647:/^.{6}no[A-Z].*bar/sm
15648:/\A.{6}no[A-Z].*bar/sm
15649:/.*no[A-Z].*bar/sm
15650:/^.*no[A-Z].*bar/sm
15651:/\A.*no[A-Z].*bar/sm
15652:/.+no[A-Z].*bar/sm
15653:/^.+no[A-Z].*bar/sm
15654:/\A.+no[A-Z].*bar/sm
15655:/.{4,7}no[A-Z].*bar/sm
15656:/^.{4,7}no[A-Z].*bar/sm
15657:/\A.{4,7}no[A-Z].*bar/sm
15658:/(\A|)no[A-Z].*bar/sm
15659:/(^|.)no[A-Z].*bar/sm
15660:/notbobno[A-Z].*bar/sm
15661:/^notbobno[A-Z].*bar/sm
15662:/\Anotbobno[A-Z].*bar/smO
15663:/.notbobno[A-Z].*bar/sm
15664:/^.notbobno[A-Z].*bar/sm
15665:/\A.notbobno[A-Z].*bar/sm
15666:/.{6}notbobno[A-Z].*bar/sm
15667:/^.{6}notbobno[A-Z].*bar/sm
15668:/\A.{6}notbobno[A-Z].*bar/sm
15669:/.*notbobno[A-Z].*bar/sm
15670:/^.*notbobno[A-Z].*bar/sm
15671:/\A.*notbobno[A-Z].*bar/sm
15672:/.+notbobno[A-Z].*bar/sm
15673:/^.+notbobno[A-Z].*bar/sm
15674:/\A.+notbobno[A-Z].*bar/sm
15675:/.{4,7}notbobno[A-Z].*bar/sm
15676:/^.{4,7}notbobno[A-Z].*bar/sm
15677:/\A.{4,7}notbobno[A-Z].*bar/sm
15678:/(\A|)notbobno[A-Z].*bar/sm
15679:/(^|.)notbobno[A-Z].*bar/sm
15680:/^.{40}/sm
15681:/^[^X]{40}/sm
15682:/\A[^X]{40}/smO
15683:/[^X]{40}/sm
15684:/.{4,}notbobno[A-Z].*bar/sm
15685:/^.{4,}no[A-Z].*bar/sm
15686:/\A.{4,}notbobno[A-Z].*bar/sm
15687:/.{4,}notbobno[A-Z].*bar/sm
15688:/^.{4,}no[A-Z].*bar/sm
15689:/\A.{4,}notbobno[A-Z].*bar/sm
15690:/(^.*bar)|bob/smO

37
tools/hscollider/test_cases/pcre/options.txt Normal file
View File

@@ -0,0 +1,37 @@
# Patterns that set/unset various options
# DOTALL
24500:/(?s)foo.*bar/O
24501:/foo.*bar(?s).baz/O
24502:/(?-s)foo.*bar/sO
24503:/(?-s)foo.*bar/O
24504:/(?-s)foo.*bar(?s).*baz/sO
# CASELESS
24600:/(?i)foobar/O
24601:/(?i)foobar/iO
24602:/(?-i)foobar/iO
24603:/foo(?-i)bar/iO
24604:/(?-i)foo(?i)bar/iO
24605:/(?i)foo(?-i)bar/O
# MULTILINE
24700:/(?m)foobar$/
24701:/(?-m)foobar$/m
# EXTENDED
24800:/(?x)foo bar/O
24801:/(?x)foo bar(?-x) baz/O
# EPIC COMBOS
24900:/(?imsx)^ foo .* bar .* baz/O
# MORE COMPLEX CASES
24901:/^(?i:(?:abbr(?:ev(?:iation)?)))/
24902:/(?s)foo(?i).bar/
24903:/(?s)foo(?i-s).bar/
24904:/foo(?i:bar)baz/
24905:/nested(?i:caseless(?-i:caseful)caseless)literal/
24906:/(a(?i)b|c)/
24907:/(?i:hatstand|teakettle)/
24908:/foo.*(?i-s:bar.*baz).*bing/s

12
tools/hscollider/test_cases/pcre/passthrough.txt Normal file
View File

@@ -0,0 +1,12 @@
9000:/(abc|def)(xxx|yyyy)/O
9001:/^(abc|def)(xxx|yyyy)/O
9002:/(^anchored|floating)(XXX|YYY)/O
# Some long literals.
9005:/a{40}/
9006:/a{48}/
9007:/a{49}/
9008:/a{50}/
9009:/q{1000}/
9010:/coagulateshyperinnervationagitationreassuranceexchangeability/
9011:/nonplasticerythematoussnakebitesjubilatedworklessnesses/

72
tools/hscollider/test_cases/pcre/prefilter.txt Normal file
View File

@@ -0,0 +1,72 @@
# Prefiltering ("/P" flag) patterns.
# Assertions.
90000:/foo(?!bar)/P
90001:/foo(?!bar).*baz/sP
90002:/(?<!foo)bar/P
90003:/foo.*(?=bar)/P
# Backreferences
90200:/(sens|respons)e and \1ibility/P
90201:/((?i)rah)\s+\1/P
90202:/(\d{3}).*\1/P
90203:/this is a (["'])quoted string\1/P
90204:/<([^>+i])>.*?</\1>/sP
# Relative backreferences.
90205:/(foo|bar).*\g{-1}/P
90206:/((foo|bar).*)\g{-2}/P
# More simple back-references.
90207:/([abc])teakettle\1/P
90208:/([abc]+)teakettle\1/P
90209:/([a-f])_\1\1\1/P
# Named back-references, in a variety of syntaxes.
90210:/(?<label>[0-9])backref\k<label>/P
90211:/(?'label'[0-9])backref\k'label'/P
90212:/(?<label>[0-9])backref\g{label}/P
90213:/(?P<label>[0-9])backref(?P=label)/P
# Atomic groups
90300:/a(?>bc|b)c/P
90320:/possessive \d*+/P
90321:/possessive \d++/P
90322:/possessive \d?+/P
90323:/possessive \d{4}+/P
90324:/possessive \d{6,}+/P
90325:/possessive \d{10,20}+/P
# Conditional references
90400:/a (pine)?(?(1)apple)/P
90401:/a (water)?(?(1)melon|orange)/P
90402:/((?<animal>hamster)|guinea pig) (?(<animal>)party)/P
90403:/((?<animal>hamster)|guinea pig) (?(<animal>)party|nothing)/P
90404:/((?'animal'hamster)|guinea pig) (?('animal')party|nothing)/P
90405:/((?P<animal>hamster)|guinea pig) (?(animal)party|nothing)/P
# Empty conditional reference group
90406:/water(-buffalo)?(?(1)) field/P
# DEFINE group, which can only have one branch and is always false
90407:/jabber(?(DEFINE)wocky)/P
# Conditional references with assertions
90408:/^(?(?=hello)[a-z]+|[a-z]{3})/P
90409:/^(?(?!hello)[a-z]+|[a-z]{3})/P
90410:/[a-z]+(?(?<=hatstand)teakettle|badgerbrush)/P
90411:/[a-z]+(?(?<!hatstand)teakettle|badgerbrush)/P
# This is a case that stresses some of the rose merging code.
90413:/([Cc]C|[ab])(d|Ba[aD]De.{14})((EdeDa|d|CC|[ab]){2,8}|.bE|(\b)*|b|Cd|[ED]EE[adeD]){7}[EA]ea+/sP
# Repeats of small regions.
90414:/qq(a.|.a){300}/sP
90415:/prefix(\d{2}|\D){300}/P
# Repeats of modifiers
90416:/((?i)a)foo\1/P
# Very large case that stresses prefilter region replacement
90417:/dDcB.Ac(((\x65CA|D){0,19}){7}){1,26}/s8P

230
tools/hscollider/test_cases/pcre/priority.txt Normal file
View File

@@ -0,0 +1,230 @@
14001:/a([a-z]+)??b+/
14002:/a([a-z]{3,7}?)b+/
14003:/abc|abcd|abcde/O
14004:/a([a-z]+)?b+/
14005:/foo(abc||abcd|abcde)bar/O
14006:/a(?:fg||hi)z/O
14007:/a(foo)??b+/
14008:/ab{2,3}?c+/
14009:/ab{2,3}c+/
14010:/a(ttt)+?z/
14011:/a(ttt)+z/
14012:/a(ttt)*?z/
14013:/a[a-z]{0,4}?b+/
14014:/^a[a-z]{0,4}?b+/
14015:/a+?b/O
14016:/a*?b/O
14017:/q(a||b){3}q/O
14018:/q(a|){2}q/O
14019:/q(a|){2}?q/O
14020:/a?bb/O
14021:/a??bb/O
14022:/q(a||b)+q/
14023:/q(a||b)(a||b)(a||b)q/O
14024:/q(a??){3}/O
14025:/(foo:.*?){3}/
14026:/foo\w{0,10}bar/
14027:/foo\w{0,10}?bar/
14028:/foo\w{1,10}bar/
14029:/foo\w{1,10}?bar/
14030:/foo\w{10}bar/
14031:/foo\w{10}?bar/
14032:/foo(?:\w?){0,10}bar/
14033:/foo(?:\w??){0,10}bar/
14034:/foo(?:\w?){10}bar/
14035:/foo(?:\w??){10}bar/
14036:/foo\w{10,}bar/
14037:/foo\w{10,}?bar/
14038:/foo()+?/
14039:/foo()+/
14040:/foo()*?/
14041:/foo()*/
14042:/foo(()+?)bar/
14043:/foo(()+)bar/
14044:/foo(()*?)bar/
14045:/foo(()*)bar/
14046:/foo()+?bar/
14047:/foo(())+?bar/
14048:/foo(()+?)+?bar/
14049:/foo(()())+?bar/
14050:/^((c*?)|b){0,3}c/s
14051:/^((c*)|b){3}c/s
14052:/^((c*)|b){2,3}?c/s
14053:/^((c*)|b){0,3}?c/s
14054:/^((c*)|b){2,3}c/s
14055:/^((c*)|b){0,3}c/s
14056:/^((c*?)|b){3}c/s
14057:/^((c*?)|b){2,3}?c/s
14058:/^((c*?)|b){0,3}?c/s
14059:/^((c*?)|b){2,3}c/s
14060:/^S((a?)(\B))*T/s
14061:/^S((a??)(\B))*T/s
14062:/^S((a?)(\B))*?T/s
14063:/^S((a??)(\B))*?T/s
14064:/^[S;]((a?)(\B))*[T;]/s
14065:/^[S;]((a??)(\B))*[T;]/s
14066:/^[S;]((a?)(\B))*?[T;]/s
14067:/^[S;]((a??)(\B))*?[T;]/s
14068:/^[S;]((a?)(\b))*[T;]/s
14069:/^[S;]((a??)(\b))*[T;]/s
14070:/^[S;]((a?)(\b))*?[T;]/s
14071:/^[S;]((a??)(\b))*?[T;]/s
14072:/^S((a?)(\B))*/s
14073:/^S((a??)(\B))*/s
14074:/^S((a?)(\B))*?/s
14075:/^S((a??)(\B))*?/s
14076:/^S(((a?)(\B))*)*T/s
14077:/^S(((a??)(\B))*)*T/s
14078:/^S(((a?)(\B))*?)*T/s
14079:/^S(((a??)(\B))*?)*T/s
14080:/^[S;](((a?)(\B))*)*[T;]/s
14081:/^[S;](((a??)(\B))*)*[T;]/s
14082:/^[S;](((a?)(\B))*?)*[T;]/s
14083:/^[S;](((a??)(\B))*?)*[T;]/s
14084:/^[S;](((a?)(\b))*)*[T;]/s
14085:/^[S;](((a??)(\b))*)*[T;]/s
14086:/^[S;](((a?)(\b))*?)*[T;]/s
14087:/^[S;](((a??)(\b))*?)*[T;]/s
14088:/^S(((a?)(\B))*)*/s
14089:/^S(((a?)(\B))*)*?/s
14090:/^S(((a?)(\B))*?)*/s
14091:/^S(((a?)(\B))*?)*?/s
14092:/^[S;](((a?)(\B))*)*?[T;]/s
14093:/^[S;](((a??)(\B))*)*?[T;]/s
14094:/^[S;](((a?)(\B))*?)*?[T;]/s
14095:/^[S;](((a??)(\B))*?)*?[T;]/s
14096:/^[S;](((a?)(\b))*)*?[T;]/s
14097:/^[S;](((a??)(\b))*)*?[T;]/s
14098:/^[S;](((a?)(\b))*?)*?[T;]/s
14099:/^[S;](((a??)(\b))*?)*?[T;]/s
14100:/^(a|(c)*|b){1,3}a/s
14101:/^(c??|b){2}c/s
14102:/^((c??)|b){0,2}?c/s
14103:/^((c?)|b){0,2}?c/s
14104:/X(y||u){2,}/
14105:/X(y||u){2,}?/
14106:/X((\B)|e)+/s
14107:/X((\B)|e)*/s
14108:/X((\B)|e)+?/s
14109:/X((\B)|e)*?/s
14110:/^b(z?|c|a)*c/s
14111:/^b(z||c|a)*c/s
14112:/^b(|c|a)*c/s
14113:/^b((c|()|a)*)c/s
14114:/^b((c|)+)+X/s
14115:/^b((z??)(c|)+)+X/s
14116:/^b((c|)+(z??))+X/s
14117:/^ab((x|)(b(c)*|))*d/
14118:/^ab((x|)(bc)*())*d/
14119:/^b(()(c|)*)+X/s
14120:/^b(()(c|)+)+X/s
14121:/^X((c*|a|z)*)*Y/s
14122:/^b((c|()|a)*)*c/s
14123:/^b((c|()|a)*?)*c/s
14124:/^b((c|()|a)*)*?c/s
14125:/^b((c|()|a)*?)*?c/s
14126:/^b(c|()|a)*c/s
14127:/^b(c?|a)*c/s
14128:/^b(z|c??|a)*c/s
14129:/^b(c??|a)*c/s
14130:/^b((c|(\B\B)|a)*)*c/s
14131:/^b((c|(\B\B)|()|a)*)*c/s
14132:/^b((c|(\B\B)|(\b\b)|a)*)*c/s
14133:/^b(((c|(\B\B)|a)(z?))*)*c/s
14134:/^b(((c|(\B\B)|()|a)(z?))*)*c/s
14135:/^b(((c|(\B\B)|(\b\b)|a)(z?))*)*c/s
14136:/^b((c|(\b\b)|a)*)*c/s
14137:/^b((c|(\b\b)|()|a)*)*c/s
14138:/^b((c|(\b\b)|(\b\b)|a)*)*c/s
14139:/^b((c|(\b)|()|a)*)*c/s
14140:/^b((c|(\b)|()|a)*)*?c/s
14141:/^b((c|(\b)|()|a)*?)*c/s
14142:/^b((c|(\b)|()|a)*?)*?c/s
14143:/^b((c|(\B)|()|a)*)*c/s
14144:/^b((c|(\B)|()|a)*)*?c/s
14145:/^b((c|(\B)|()|a)*?)*c/s
14146:/^b((c|(\B)|()|a)*?)*?c/s
14147:/^b((c|()|(\B)|a)*?)*c/s
14148:/^b((c|()|(\B)|a)*)*?c/s
14149:/^b((c|()|(\B)|a)*)*c/s
14150:/^XYZ(((a*)*)*)*/
14151:/^XYZ((a*)*)*/
14152:/^XYZ(((a*)*)*)*b/
14153:/^XYZ((a*)*)*b/
14154:/^XYZ(((a?)*)*)*b/
14155:/^XYZ(((a*)?)*)*b/
14156:/^XYZ((((a*)*)*)*)*b/
14157:/^XYZ(((((a*)*)*)*)*)*b/
14158:/^XYZ(((a?)*)*)+/
14159:/^XYZ(((((a*())*())*())*())*())*/
14160:/(\b)*baa/
14161:/;(\b)*baa/
14162:/^x((a)|()|(b)){0,2}y/s
14163:/^x((a)|()|(b)){0,2}?y/s
14164:/^x((a)|()|(b)){0,}y/s
14165:/^x((a)|()|(b)){0,}?y/s
14166:/^x((a)|()|(b)){1,}y/s
14167:/^x((a)|()|(b)){1,}?y/s
14168:/^x((a)|()|(b)){5,8}y/s
14169:/^x((a)|()|(b)){5,8}?y/s
14170:/^x((a)|()|(b)){0,5}y/s
14171:/^x((a)|()|(b)){0,5}?y/s
14172:/^x((a)|()|(b)){1,5}y/s
14173:/^x((a)|()|(b)){1,5}?y/s
14174:/^X(_(X||A.)|.)+?/s
14175:/^X(_(X||A.)|.)+/s
14176:/^X(_(A.||X)|.)+?/s
14177:/^X(_(A.||X)|.)+/s
14178:/^X(_(X||A.)|.)+?/s
14179:/^X(_(X||A.)|.)+/s
14180:/^X(_(A.||X)|.){3,}?/s
14181:/^X(_(A.||X)|.){3,}/s
14182:/^X(_(A.||X)|.){3,5}?/s
14183:/^X(_(A.||X)|.){3,5}/s
14184:/^b((()|c|a)*)*c/s
14185:/^b((()|c|a)*?)*c/s
14186:/^b((()|c|a)*)*?c/s
14187:/^b((()|c|a)*?)*?c/s
14188:/^b(()|(c|()|a)*)*c/s
14189:/^b(()|(c|()|a)*?)*c/s
14190:/^b(()|(c|()|a)*)*?c/s
14191:/^b(()|(c|()|a)*?)*?c/s
14192:/^b(()|(|a)*)*c/s
14193:/^b(()|(|a)*?)*c/s
14194:/^b(()|(|a)*)*?c/s
14195:/^b(()|(|a)*?)*?c/s
14196:/^b(()|(|a)*|[cC])*c/s
14197:/^b(()|(|a)*?|[cC])*c/s
14198:/^b(()|(|a)*|[cC])*?c/s
14199:/^b(()|(|a)*?|[cC])*?c/s
14200:/^b(()|(c|()|a)*)*c/s
14201:/^b(()|(c|()|a)*?)*c/s
14202:/^b(()|(c|()|a)*)*?c/s
14203:/^b(()|(c|()|a)*?)*?c/s
14204:/^b(()|(|a)*)*c/s
14205:/^b(()|(|a)*?)*c/s
14206:/^b(()|(|a)*)*?c/s
14207:/^b(()|(|a)*?)*?c/s
14208:/^b(()|(|a)*|[cC])*c/s
14209:/^b(()|(|a)*?|[cC])*c/s
14210:/^b(()|(|a)*|[cC])*?c/s
14211:/^b(()|(|a)*?|[cC])*?c/s
14212:/^b(d|()|((a|()|b)*))+c/s
14213:/^b(d|()|((a|())*))+c/s
14214:/^b(d|()|((()|a)*))+c/s
14215:/^b(d|((a|())*)|())+c/s
14216:/^b(d|()|((a|())+))+c/s
14217:/^b(()|(|a)*|d)*c/s
14218:/^([ba]|){1,3}?a/si
14219:/^([ba]){1,3}?a/si
14220:/^(c|){2,3}?X/s
14221:/^(c|){2,4}?X/s
14222:/^b((|a)*?)*?c/s
14223:/^b((|a)*?){0,10}?c/s
14224:/^b((|a){0,10}?)*?c/s
14225:/^b((|a){0,10}?){0,10}?c/s
14226:/^b((|a)+?)+?c/s
14227:/^b((|a)*?){10,}?c/s
14228:/^b((|a){10,}?)*?c/s
14229:/^b((|a){10,}?){10,}?c/s
14230:/^x(a|()){1,3}?y/s

94
tools/hscollider/test_cases/pcre/puff.txt Normal file
View File

@@ -0,0 +1,94 @@
15200:/foo[^X]{15}/sO
15201:/foo[^X]{16}/sO
15202:/foo[^X]{17}/sO
15203:/foo.*[^X]{17}/s
15204:/foo[^X]{17}blah/sO
15205:/foo[^XY]{17}/sO
15206:/foo[^X]{17}$/s
15207:/[^X]{17}/s
15208:/^[^X]{17}/sO
15209:/fo.*o[^X]{15}/s
15210:/fo.*o[^X]{16}/s
15211:/fo.*o[^X]{17}/s
15212:/[fb][oa][or][^X]{15}/s
15213:/[fb][oa][or][^X]{16}/s
15214:/[fb][oa][or][^X]{17}/s
15215:/f[^X]{255}/H
15216:/f[^X]{256}/H
15217:/f[^X]{257}/H
15218:/^[fb][oa][or][^X]{17}/sO
15219:/f.{255}/H
15220:/f.{256}/H
15221:/f.{257}/H
15222:/^[fb][oa][or].{17}/sO
15223:/foo.{15}/sHO
15224:/foo.{16}/sHO
15225:/foo.{17}/sHO
15226:/foo.*.{17}/sH
15227:/foo.{17}blah/sHO
15228:/foo.{17}/HO
15229:/foo.{17}$/sH
15230:/.{17}/sH
15231:/^.{17}/sH
15232:/fo.*o.{15}/sH
15233:/fo.*o.{16}/sH
15234:/fo.*o.{17}/sH
15235:/[fb][oa][or].{15}/sH
15236:/[fb][oa][or].{16}/sH
15237:/[fb][oa][or].{17}/sH
15238:/foo.{30,}/s
15239:/foo.*bar.{30}/s
15240:/foo.*bar.{30,}/s
15241:/^hhechbd(je[wn][oqc]bo|.{18}|l|.|r)/O
15242:/^ii.m.x(.{22}|p|y|n*)/
15243:/f[^XY]{255}/H
15244:/f[^XY]{256}/H
15245:/f[^XY]{257}/H
15246:/bar.{16,}/s
15247:/bar[^X]{16,}/s
15248:/bar[^r]{16,}/s
15249:/bar[^XY]{16,}/s
15250:/bar[^rY]{16,}/s
15251:/foo.*bar[^r]{16}/s
15252:/f.{254,}/s
15253:/f.{255,}/s
15254:/f.{256,}/s
15255:/a[fg]{35,}/s
15256:/f[fg]{35,}/s
15257:/[af][fg]{35,}/s
15258:/^[^X]{50}/
15259:/^[^X]{50,}/
15260:/^[^XY]{50}/
15261:/^[^XY]{50,}/
15262:/^[X]{50}/
15263:/^[X]{50,}/
15264:/^[XY]{50}/
15265:/^[XY]{50,}/
15266:/x/s
15267:/.{52}/s
15268:/.{100}/s
15269:/.{51}/s
15270:/^.{100}/s
15271:/^.{62,}/s
15272:/^.{60,}/s
15273:/^.{60,}|.{65}/s
15274:/^.{60}|.{65}/s
15275:/^.{60}|.{55}/s
15276:/^[^a]{51}/s
15277:/^[^a]{72}/s
15278:/^[^a]{100,}/s
15279:/[^X]{40}/s
15280:/^.*[^X]{20}|[^Y]{21}|.{60}|^.{50,}/s
15281:/[^X]{18}/s
15282:/[^XY]{18}/s
15283:/[^X\x00\x11\x22\x33\x44\x55\x66\x77\x88]{18}/s
15284:/^[A-Z]*\d\.\d{60}/s
15285:/[A-Z]*\d.\d{60}/sH

53
tools/hscollider/test_cases/pcre/pug.txt Normal file
View File

@@ -0,0 +1,53 @@
5000:/[^X]:.{111,}pug/s
5001:/[^X]:.{111,}[pugPUGxyz]{3}/s
5002:/aaa:.{111,}pug/s
5003:/aaa:.{111,}[pugPUGxyz]{3}/s
5004:/[^X]:.{112,}pug/s
5005:/[^X]:.{112,}[pugPUGxyz]{3}/s
5006:/aaa:.{112,}pug/s
5007:/aaa:.{112,}[pugPUGxyz]{3}/s
5008:/[^X]:.{113,}pug/s
5009:/[^X]:.{113,}[pugPUGxyz]{3}/s
5010:/aaa:.{113,}pug/s
5011:/aaa:.{113,}[pugPUGxyz]{3}/s
5012:/[^X]:.{127,}pug/s
5013:/[^X]:.{127,}[pugPUGxyz]{3}/s
5014:/aaa:.{127,}pug/s
5015:/aaa:.{127,}[pugPUGxyz]{3}/s
5016:/[^X]:.{128,}pug/s
5017:/[^X]:.{128,}[pugPUGxyz]{3}/s
5018:/aaa:.{128,}pug/s
5019:/aaa:.{128,}[pugPUGxyz]{3}/s
5020:/[^X]:.{129,}pug/s
5021:/[^X]:.{129,}[pugPUGxyz]{3}/s
5022:/aaa:.{129,}pug/s
5023:/aaa:.{129,}[pugPUGxyz]{3}/s
5024:/[^X]:[a-z]{111,}pug/s
5025:/[^X]:[a-z]{111,}[pugPUGxyz]{3}/s
5026:/aaa:[a-z]{111,}pug/s
5027:/aaa:[a-z]{111,}[pugPUGxyz]{3}/s
5028:/[^X]:[a-z]{112,}pug/s
5029:/[^X]:[a-z]{112,}[pugPUGxyz]{3}/s
5030:/aaa:[a-z]{112,}pug/s
5031:/aaa:[a-z]{112,}[pugPUGxyz]{3}/s
5032:/[^X]:[a-z]{113,}pug/s
5033:/[^X]:[a-z]{113,}[pugPUGxyz]{3}/s
5034:/aaa:[a-z]{113,}pug/s
5035:/aaa:[a-z]{113,}[pugPUGxyz]{3}/s
5036:/[^X]:[a-z]{127,}pug/s
5037:/[^X]:[a-z]{127,}[pugPUGxyz]{3}/s
5038:/aaa:[a-z]{127,}pug/s
5039:/aaa:[a-z]{127,}[pugPUGxyz]{3}/s
5040:/[^X]:[a-z]{128,}pug/s
5041:/[^X]:[a-z]{128,}[pugPUGxyz]{3}/s
5042:/aaa:[a-z]{128,}pug/s
5043:/aaa:[a-z]{128,}[pugPUGxyz]{3}/s
5044:/[^X]:[a-z]{129,}pug/s
5045:/[^X]:[a-z]{129,}[pugPUGxyz]{3}/s
5046:/aaa:[a-z]{129,}pug/s
5047:/aaa:[a-z]{129,}[pugPUGxyz]{3}/s
5048:/([^X]:|:[^Y]).{120,}[pugPUGxyz]{3}/s
5049:/([^X]:|:[^Y])[a-z]{120,}[pugPUGxyz]{3}/s
5900:/[^X][^Xx][defg][^:*][deDE][^bcC][deAE]/
5901:/(^|[x\- B])(a|[bBbB]|cc)[^A-Z0-9]{112,}[deDE][^bcC][geAE]/s
5902:/(^|[x\- B])(a|bb|cc).{113,}[deDE][^bcC][geAE]/s

47
tools/hscollider/test_cases/pcre/redundancy.txt Normal file
View File

@@ -0,0 +1,47 @@
15100:/hatstand.*a?teakettle/sO
15101:/foo(A|A)bar/O
15102:/foo(1|2|3|4)bar/O
15103:/^.*a+bc/
15104:/foo[Aa]+a+bar/
15105:/foo[Aa]+[aB]+bar/
15106:/foo[AaB]+[aB]+bar/
15107:/foo[Aa]*[aB]+bar/
15108:/foo[AaB]*[aB]+bar/
15109:/a(b|c.+)d+e/
15110:/a(b|c.*)d+e/
# cases specifically aimed at the cyclic-dom redundancy code
15120:/foo.*a*bar/O
15121:/fooa*.*bar/O
15122:/fooa?.*bar/O
15123:/foo.*a?bar/O
15124:/foo.*a+bar/O
15125:/fooa+.*bar/O
15126:/foo.+a*bar/O
15127:/fooa*.+bar/O
15128:/fooa?.+bar/O
15129:/foo.+a?bar/O
15130:/foo.+a+bar/O
15131:/fooa+.+bar/O
# edge redun
15132:/abc(a|\B){6}def/sO
15133:/AAA([A-Z]_?)+(([\x00-\x3d]|[\x3f-\xff])\x00?)+ZZZ/s
# misc opt
15134:/(Y.|X.+)a[^a]*foo/s
15135:/oof[^a]*a(.+X|.Y)/s
15136:/foo.*<[^<]*bar/s
15137:/foo[^<]*<.*bar/s
15138:/foo\b.*bar/s
15139:/foo.*\bbar/s
15140:/foo[^<]*<[^>]*bar/s
15141:/foo[^<]*>[^>]*bar/s
15142:/foo[^<=]*[<=][^>=]*bar/s
15143:/foo[^<=]*[>=][^>=]*bar/s
# rev misc opt
15144:/aa+.aaaa/si
# needs cyclic path redundancy followed by other redundancy passes.
15145:/^abc.(.*|foo)\Sa..a...\S./s

533
tools/hscollider/test_cases/pcre/ucp.txt Normal file
View File

@@ -0,0 +1,533 @@
85000:/\p{C}/8W
85001:/\p{Cc}/8W
85002:/\p{Cf}/8W
85003:/\p{Cn}/8W
85004:/\p{Co}/8W
85005:/\p{Cs}/8W
85006:/\p{L}/8W
85007:/\p{L&}/8W
85008:/\p{Ll}/8W
85009:/\p{Lm}/8W
85010:/\p{Lo}/8W
85011:/\p{Lt}/8W
85012:/\p{Lu}/8W
85013:/\p{M}/8W
85014:/\p{Mc}/8W
85015:/\p{Me}/8W
85016:/\p{Mn}/8W
85017:/\p{N}/8W
85018:/\p{Nd}/8W
85019:/\p{Nl}/8W
85020:/\p{No}/8W
85021:/\p{P}/8W
85022:/\p{Pc}/8W
85023:/\p{Pd}/8W
85024:/\p{Pe}/8W
85025:/\p{Pf}/8W
85026:/\p{Pi}/8W
85027:/\p{Po}/8W
85028:/\p{Ps}/8W
85029:/\p{S}/8W
85030:/\p{Sc}/8W
85031:/\p{Sk}/8W
85032:/\p{Sm}/8W
85033:/\p{So}/8W
85034:/\p{Xan}/8W
85035:/\p{Xps}/8W
85036:/\p{Xsp}/8W
85037:/\p{Xwd}/8W
85038:/\p{Z}/8W
85039:/\p{Zl}/8W
85040:/\p{Zp}/8W
85041:/\p{Zs}/8W
85042:/\p{Arabic}/8W
85043:/\p{Armenian}/8W
85044:/\p{Avestan}/8W
85045:/\p{Balinese}/8W
85046:/\p{Bamum}/8W
85047:/\p{Batak}/8W
85048:/\p{Bengali}/8W
85049:/\p{Bopomofo}/8W
85050:/\p{Brahmi}/8W
85051:/\p{Braille}/8W
85052:/\p{Buginese}/8W
85053:/\p{Buhid}/8W
85054:/\p{Canadian_Aboriginal}/8W
85055:/\p{Carian}/8W
85056:/\p{Cham}/8W
85057:/\p{Cherokee}/8W
85058:/\p{Common}/8W
85059:/\p{Coptic}/8W
85060:/\p{Cuneiform}/8W
85061:/\p{Cypriot}/8W
85062:/\p{Cyrillic}/8W
85063:/\p{Deseret}/8W
85064:/\p{Devanagari}/8W
85065:/\p{Egyptian_Hieroglyphs}/8W
85066:/\p{Ethiopic}/8W
85067:/\p{Georgian}/8W
85068:/\p{Glagolitic}/8W
85069:/\p{Gothic}/8W
85070:/\p{Greek}/8W
85071:/\p{Gujarati}/8W
85072:/\p{Gurmukhi}/8W
85073:/\p{Han}/8W
85074:/\p{Hangul}/8W
85075:/\p{Hanunoo}/8W
85076:/\p{Hebrew}/8W
85077:/\p{Hiragana}/8W
85078:/\p{Imperial_Aramaic}/8W
85079:/\p{Inherited}/8W
85080:/\p{Inscriptional_Pahlavi}/8W
85081:/\p{Inscriptional_Parthian}/8W
85082:/\p{Javanese}/8W
85083:/\p{Kaithi}/8W
85084:/\p{Kannada}/8W
85085:/\p{Katakana}/8W
85086:/\p{Kayah_Li}/8W
85087:/\p{Kharoshthi}/8W
85088:/\p{Khmer}/8W
85089:/\p{Lao}/8W
85090:/\p{Latin}/8W
85091:/\p{Lepcha}/8W
85092:/\p{Limbu}/8W
85093:/\p{Linear_B}/8W
85094:/\p{Lisu}/8W
85095:/\p{Lycian}/8W
85096:/\p{Lydian}/8W
85097:/\p{Malayalam}/8W
85098:/\p{Mandaic}/8W
85099:/\p{Meetei_Mayek}/8W
85100:/\p{Mongolian}/8W
85101:/\p{Myanmar}/8W
85102:/\p{New_Tai_Lue}/8W
85103:/\p{Nko}/8W
85104:/\p{Ogham}/8W
85105:/\p{Ol_Chiki}/8W
85106:/\p{Old_Italic}/8W
85107:/\p{Old_Persian}/8W
85108:/\p{Old_South_Arabian}/8W
85109:/\p{Old_Turkic}/8W
85110:/\p{Oriya}/8W
85111:/\p{Osmanya}/8W
85112:/\p{Phags_Pa}/8W
85113:/\p{Phoenician}/8W
85114:/\p{Rejang}/8W
85115:/\p{Runic}/8W
85116:/\p{Samaritan}/8W
85117:/\p{Saurashtra}/8W
85118:/\p{Shavian}/8W
85119:/\p{Sinhala}/8W
85120:/\p{Sundanese}/8W
85121:/\p{Syloti_Nagri}/8W
85122:/\p{Syriac}/8W
85123:/\p{Tagalog}/8W
85124:/\p{Tagbanwa}/8W
85125:/\p{Tai_Le}/8W
85126:/\p{Tai_Tham}/8W
85127:/\p{Tai_Viet}/8W
85128:/\p{Tamil}/8W
85129:/\p{Telugu}/8W
85130:/\p{Thaana}/8W
85131:/\p{Thai}/8W
85132:/\p{Tibetan}/8W
85133:/\p{Tifinagh}/8W
85134:/\p{Ugaritic}/8W
85135:/\p{Vai}/8W
85136:/\p{Yi}/8W
85500:/\P{C}/8W
85501:/\P{Cc}/8W
85502:/\P{Cf}/8W
85503:/\P{Cn}/8W
85504:/\P{Co}/8W
85505:/\P{Cs}/8W
85506:/\P{L}/8W
85507:/\P{L&}/8W
85508:/\P{Ll}/8W
85509:/\P{Lm}/8W
85510:/\P{Lo}/8W
85511:/\P{Lt}/8W
85512:/\P{Lu}/8W
85513:/\P{M}/8W
85514:/\P{Mc}/8W
85515:/\P{Me}/8W
85516:/\P{Mn}/8W
85517:/\P{N}/8W
85518:/\P{Nd}/8W
85519:/\P{Nl}/8W
85520:/\P{No}/8W
85521:/\P{P}/8W
85522:/\P{Pc}/8W
85523:/\P{Pd}/8W
85524:/\P{Pe}/8W
85525:/\P{Pf}/8W
85526:/\P{Pi}/8W
85527:/\P{Po}/8W
85528:/\P{Ps}/8W
85529:/\P{S}/8W
85530:/\P{Sc}/8W
85531:/\P{Sk}/8W
85532:/\P{Sm}/8W
85533:/\P{So}/8W
85534:/\P{Xan}/8W
85535:/\P{Xps}/8W
85536:/\P{Xsp}/8W
85537:/\P{Xwd}/8W
85538:/\P{Z}/8W
85539:/\P{Zl}/8W
85540:/\P{Zp}/8W
85541:/\P{Zs}/8W
85542:/\P{Arabic}/8W
85543:/\P{Armenian}/8W
85544:/\P{Avestan}/8W
85545:/\P{Balinese}/8W
85546:/\P{Bamum}/8W
85547:/\P{Batak}/8W
85548:/\P{Bengali}/8W
85549:/\P{Bopomofo}/8W
85550:/\P{Brahmi}/8W
85551:/\P{Braille}/8W
85552:/\P{Buginese}/8W
85553:/\P{Buhid}/8W
85554:/\P{Canadian_Aboriginal}/8W
85555:/\P{Carian}/8W
85556:/\P{Cham}/8W
85557:/\P{Cherokee}/8W
85558:/\P{Common}/8W
85559:/\P{Coptic}/8W
85560:/\P{Cuneiform}/8W
85561:/\P{Cypriot}/8W
85562:/\P{Cyrillic}/8W
85563:/\P{Deseret}/8W
85564:/\P{Devanagari}/8W
85565:/\P{Egyptian_Hieroglyphs}/8W
85566:/\P{Ethiopic}/8W
85567:/\P{Georgian}/8W
85568:/\P{Glagolitic}/8W
85569:/\P{Gothic}/8W
85570:/\P{Greek}/8W
85571:/\P{Gujarati}/8W
85572:/\P{Gurmukhi}/8W
85573:/\P{Han}/8W
85574:/\P{Hangul}/8W
85575:/\P{Hanunoo}/8W
85576:/\P{Hebrew}/8W
85577:/\P{Hiragana}/8W
85578:/\P{Imperial_Aramaic}/8W
85579:/\P{Inherited}/8W
85580:/\P{Inscriptional_Pahlavi}/8W
85581:/\P{Inscriptional_Parthian}/8W
85582:/\P{Javanese}/8W
85583:/\P{Kaithi}/8W
85584:/\P{Kannada}/8W
85585:/\P{Katakana}/8W
85586:/\P{Kayah_Li}/8W
85587:/\P{Kharoshthi}/8W
85588:/\P{Khmer}/8W
85589:/\P{Lao}/8W
85590:/\P{Latin}/8W
85591:/\P{Lepcha}/8W
85592:/\P{Limbu}/8W
85593:/\P{Linear_B}/8W
85594:/\P{Lisu}/8W
85595:/\P{Lycian}/8W
85596:/\P{Lydian}/8W
85597:/\P{Malayalam}/8W
85598:/\P{Mandaic}/8W
85599:/\P{Meetei_Mayek}/8W
85600:/\P{Mongolian}/8W
85601:/\P{Myanmar}/8W
85602:/\P{New_Tai_Lue}/8W
85603:/\P{Nko}/8W
85604:/\P{Ogham}/8W
85605:/\P{Ol_Chiki}/8W
85606:/\P{Old_Italic}/8W
85607:/\P{Old_Persian}/8W
85608:/\P{Old_South_Arabian}/8W
85609:/\P{Old_Turkic}/8W
85610:/\P{Oriya}/8W
85611:/\P{Osmanya}/8W
85612:/\P{Phags_Pa}/8W
85613:/\P{Phoenician}/8W
85614:/\P{Rejang}/8W
85615:/\P{Runic}/8W
85616:/\P{Samaritan}/8W
85617:/\P{Saurashtra}/8W
85618:/\P{Shavian}/8W
85619:/\P{Sinhala}/8W
85620:/\P{Sundanese}/8W
85621:/\P{Syloti_Nagri}/8W
85622:/\P{Syriac}/8W
85623:/\P{Tagalog}/8W
85624:/\P{Tagbanwa}/8W
85625:/\P{Tai_Le}/8W
85626:/\P{Tai_Tham}/8W
85627:/\P{Tai_Viet}/8W
85628:/\P{Tamil}/8W
85629:/\P{Telugu}/8W
85630:/\P{Thaana}/8W
85631:/\P{Thai}/8W
85632:/\P{Tibetan}/8W
85633:/\P{Tifinagh}/8W
85634:/\P{Ugaritic}/8W
85635:/\P{Vai}/8W
85636:/\P{Yi}/8W
# Non-UTF-8 cases (the ones that can actually match)
85200:/\p{C}/W
85201:/\p{Cc}/W
85202:/\p{Cf}/W
85206:/\p{L}/W
85207:/\p{L&}/W
85208:/\p{Ll}/W
85212:/\p{Lu}/W
85217:/\p{N}/W
85218:/\p{Nd}/W
85220:/\p{No}/W
85221:/\p{P}/W
85222:/\p{Pc}/W
85223:/\p{Pd}/W
85224:/\p{Pe}/W
85225:/\p{Pf}/W
85226:/\p{Pi}/W
85227:/\p{Po}/W
85228:/\p{Ps}/W
85229:/\p{S}/W
85230:/\p{Sc}/W
85231:/\p{Sk}/W
85232:/\p{Sm}/W
85233:/\p{So}/W
85234:/\p{Xan}/W
85235:/\p{Xps}/W
85236:/\p{Xsp}/W
85237:/\p{Xwd}/W
85238:/\p{Z}/W
85241:/\p{Zs}/W
85258:/\p{Common}/W
85290:/\p{Latin}/W
85700:/\P{C}/W
85701:/\P{Cc}/W
85702:/\P{Cf}/W
85703:/\P{Cn}/W
85704:/\P{Co}/W
85705:/\P{Cs}/W
85706:/\P{L}/W
85707:/\P{L&}/W
85708:/\P{Ll}/W
85709:/\P{Lm}/W
85710:/\P{Lo}/W
85711:/\P{Lt}/W
85712:/\P{Lu}/W
85713:/\P{M}/W
85714:/\P{Mc}/W
85715:/\P{Me}/W
85716:/\P{Mn}/W
85717:/\P{N}/W
85718:/\P{Nd}/W
85719:/\P{Nl}/W
85720:/\P{No}/W
85721:/\P{P}/W
85722:/\P{Pc}/W
85723:/\P{Pd}/W
85724:/\P{Pe}/W
85725:/\P{Pf}/W
85726:/\P{Pi}/W
85727:/\P{Po}/W
85728:/\P{Ps}/W
85729:/\P{S}/W
85730:/\P{Sc}/W
85731:/\P{Sk}/W
85732:/\P{Sm}/W
85733:/\P{So}/W
85734:/\P{Xan}/W
85735:/\P{Xps}/W
85736:/\P{Xsp}/W
85737:/\P{Xwd}/W
85738:/\P{Z}/W
85739:/\P{Zl}/W
85740:/\P{Zp}/W
85741:/\P{Zs}/W
85742:/\P{Arabic}/W
85743:/\P{Armenian}/W
85744:/\P{Avestan}/W
85745:/\P{Balinese}/W
85746:/\P{Bamum}/W
85747:/\P{Batak}/W
85748:/\P{Bengali}/W
85749:/\P{Bopomofo}/W
85750:/\P{Brahmi}/W
85751:/\P{Braille}/W
85752:/\P{Buginese}/W
85753:/\P{Buhid}/W
85754:/\P{Canadian_Aboriginal}/W
85755:/\P{Carian}/W
85756:/\P{Cham}/W
85757:/\P{Cherokee}/W
85758:/\P{Common}/W
85759:/\P{Coptic}/W
85760:/\P{Cuneiform}/W
85761:/\P{Cypriot}/W
85762:/\P{Cyrillic}/W
85763:/\P{Deseret}/W
85764:/\P{Devanagari}/W
85765:/\P{Egyptian_Hieroglyphs}/W
85766:/\P{Ethiopic}/W
85767:/\P{Georgian}/W
85768:/\P{Glagolitic}/W
85769:/\P{Gothic}/W
85770:/\P{Greek}/W
85771:/\P{Gujarati}/W
85772:/\P{Gurmukhi}/W
85773:/\P{Han}/W
85774:/\P{Hangul}/W
85775:/\P{Hanunoo}/W
85776:/\P{Hebrew}/W
85777:/\P{Hiragana}/W
85778:/\P{Imperial_Aramaic}/W
85779:/\P{Inherited}/W
85780:/\P{Inscriptional_Pahlavi}/W
85781:/\P{Inscriptional_Parthian}/W
85782:/\P{Javanese}/W
85783:/\P{Kaithi}/W
85784:/\P{Kannada}/W
85785:/\P{Katakana}/W
85786:/\P{Kayah_Li}/W
85787:/\P{Kharoshthi}/W
85788:/\P{Khmer}/W
85789:/\P{Lao}/W
85790:/\P{Latin}/W
85791:/\P{Lepcha}/W
85792:/\P{Limbu}/W
85793:/\P{Linear_B}/W
85794:/\P{Lisu}/W
85795:/\P{Lycian}/W
85796:/\P{Lydian}/W
85797:/\P{Malayalam}/W
85798:/\P{Mandaic}/W
85799:/\P{Meetei_Mayek}/W
85800:/\P{Mongolian}/W
85801:/\P{Myanmar}/W
85802:/\P{New_Tai_Lue}/W
85803:/\P{Nko}/W
85804:/\P{Ogham}/W
85805:/\P{Ol_Chiki}/W
85806:/\P{Old_Italic}/W
85807:/\P{Old_Persian}/W
85808:/\P{Old_South_Arabian}/W
85809:/\P{Old_Turkic}/W
85810:/\P{Oriya}/W
85811:/\P{Osmanya}/W
85812:/\P{Phags_Pa}/W
85813:/\P{Phoenician}/W
85814:/\P{Rejang}/W
85815:/\P{Runic}/W
85816:/\P{Samaritan}/W
85817:/\P{Saurashtra}/W
85818:/\P{Shavian}/W
85819:/\P{Sinhala}/W
85820:/\P{Sundanese}/W
85821:/\P{Syloti_Nagri}/W
85822:/\P{Syriac}/W
85823:/\P{Tagalog}/W
85824:/\P{Tagbanwa}/W
85825:/\P{Tai_Le}/W
85826:/\P{Tai_Tham}/W
85827:/\P{Tai_Viet}/W
85828:/\P{Tamil}/W
85829:/\P{Telugu}/W
85830:/\P{Thaana}/W
85831:/\P{Thai}/W
85832:/\P{Tibetan}/W
85833:/\P{Tifinagh}/W
85834:/\P{Ugaritic}/W
85835:/\P{Vai}/W
85836:/\P{Yi}/W
# Don't forget \p{Any}
86000:/^\p{Any}/W
86001:/^\p{Any}/8W
# Braceless variants for {C, L, M, N, P, S, Z}
86010:/\pC/8W
86011:/\pL/8W
86012:/\pM/8W
86013:/\pN/8W
86014:/\pP/8W
86015:/\pS/8W
86016:/\pZ/8W
86020:/\PC/8W
86021:/\PL/8W
86022:/\PM/8W
86023:/\PN/8W
86024:/\PP/8W
86025:/\PS/8W
86026:/\PZ/8W
86030:/\pC/W
86031:/\pL/W
86033:/\pN/W
86034:/\pP/W
86035:/\pS/W
86036:/\pZ/W
86040:/\PC/W
86041:/\PL/W
86042:/\PM/W
86043:/\PN/W
86044:/\PP/W
86045:/\PS/W
86046:/\PZ/W
# UCP properties are immune to the nocase flag -- they are always
# case-sensitive.
86060:/^case \p{Ll}/i
86061:/^case \p{Ll}/i8W
86062:/^case \p{Lu}/i
86063:/^case \p{Lu}/i8W
86064:/^case \P{Ll}/i
86065:/^case \P{Ll}/i8W
86066:/^case \P{Lu}/i
86067:/^case \P{Lu}/i8W
# Some POSIX classes behave differently in UCP mode.
86200:/^[[:alnum:]]/8W
86201:/^[[:alpha:]]/8W
86202:/^[[:blank:]]/8W
86203:/^[[:digit:]]/8W
86204:/^[[:lower:]]/8W
86205:/^[[:space:]]/8W
86206:/^[[:upper:]]/8W
86207:/^[[:word:]]/8W
86208:/^[[:graph:]]/8W
86209:/^[[:print:]]/8W
86210:/^[[:punct:]]/8W
86211:/^[[:alnum:]]/W
86212:/^[[:alpha:]]/W
86213:/^[[:blank:]]/W
86214:/^[[:digit:]]/W
86215:/^[[:lower:]]/W
86216:/^[[:space:]]/W
86217:/^[[:upper:]]/W
86218:/^[[:word:]]/W
86219:/^[[:graph:]]/W
86220:/^[[:print:]]/W
86221:/^[[:punct:]]/W
# Negated variants
86230:/^[[:^alnum]][[:^alpha]][[:^blank]][[:^digit]][[:^lower]][[:^space]][[:^upper]][[:^word]][[:^graph]][[:^print]][[:^punct]]/8W
86231:/^[[:^alnum]][[:^alpha]][[:^blank]][[:^digit]][[:^lower]][[:^space]][[:^upper]][[:^word]][[:^graph]][[:^print]][[:^punct]]/W
# Classes where the POSIX class component disappears (fixed in PCRE 8.38)
86240:/^a[[:punct:]bc]/W
86241:/^a[[:punct:]bc]/8W
# Check [:print:] in UCP mode against U+180E, MONGOLIAN VOWEL
# SEPARATOR.
86242:/^[[:print:]]/8W
86243:/^[^[:print:]]/8W
# PCRE bug fixed in PCRE 8.40.
86244:/[\D\P{Nd}]+/8

221
tools/hscollider/test_cases/pcre/utf8.txt Normal file
View File

@@ -0,0 +1,221 @@
80000:/空/8
80001:/حرية/8
80002:/[空]/8
80003:/\x{7a7a}/8
80004:/[\x{7a7a}]/8
80005:/[空]/
80006:/^./s8
80007:/^./8
80008:/æ/8
80009:/\346/8
80010:/Ā/8
80011:/\700/8
80012:/[ἀ-ῼ]/8
80013:/[\x{2e18}-⸮]/8
80014:/^空{2}/8
80015:/^空{2}/
# 2 char all
80016:/^[\x{80}-\x{7ff}]/8
80017:/\x80/8
80018:/\x{80}/8
80019:/[\x80]/8
80020:/[\x{80}]/8
80021:/\xff/8
80022:/\x{ff}/8
80023:/[\xff]/8
80024:/[\x{ff}]/8
# 80025:/\X/
80026:/foo\Xbar/8P
80100:/foo.bar/8
80101:/foo.*bar/s8
80102:/foo[a]*bar/8
80103:/foo[^a]*bar/8
80104:/foo.+bar/s8
80105:/foo[a]+bar/8
80106:/foo[^a]_bar/8
80107:/foo.*/s8
80108:/foo.+/s8
80109:/foo.*bar/8
80110:/[]a]/8
84000:/I/8i
84001:/k/8i
84002:/S/8i
84003:/i/8i
84004:/K/8i
84005:/s/8i
84006:/µ/8i
84007:/Å/8i
84008:/ß/8i
84009:/å/8i
84010:/İ/8i
84011:/ı/8i
84012:/ſ/8i
84013:/ͅ/8i
84014:/Β/8i
84015:/Ε/8i
84016:/Θ/8i
84017:/Ι/8i
84018:/Κ/8i
84019:/Μ/8i
84020:/Π/8i
84021:/Ρ/8i
84022:/Σ/8i
84023:/Φ/8i
84024:/Ω/8i
84025:/β/8i
84026:/ε/8i
84027:/θ/8i
84028:/ι/8i
84029:/κ/8i
84030:/μ/8i
84031:/π/8i
84032:/ρ/8i
84033:/ς/8i
84034:/σ/8i
84035:/φ/8i
84036:/ω/8i
84037:/ϐ/8i
84038:/ϑ/8i
84039:/ϕ/8i
84040:/ϖ/8i
84041:/ϰ/8i
84042:/ϱ/8i
84043:/ϴ/8i
84044:/ϵ/8i
84045:/Ṡ/8i
84046:/ṡ/8i
84047:/ẛ/8i
84048:/ẞ/8i
84049:/ι/8i
84050:/Ω/8i
84051:/K/8i
84052:/Å/8i
84053:/SS/8i
84054:/SZ/8i
84055:/ss/8i
84056:/sz/8i
84057:/DŽ/8i
84058:/Dž/8i
84059:/dž/8i
84060:/LJ/8i
84061:/Lj/8i
84062:/lj/8i
84063:/NJ/8i
84064:/Nj/8i
84065:/nj/8i
84066:/Dz/8i
84067:/dz/8i
88000:/\p{Vai}/8
88001:/\p{^Vai}/8
88002:/\P{Vai}/8
88003:/\P{^Vai}/8
#test for somethings which are in multiple scripts
88004:/\p{Common}/8
88005:/\p{Mongolian}/8
88006:/\p{Phags_Pa}/8
88007:/\P{Common}/8
88008:/\P{Mongolian}/8
88009:/\P{Phags_Pa}/8
#unallocated and common with the pcre twistr
88010:/\p{Cn}/8
88011:/\P{Cn}/8
88012:/\p{Common}/8
88013:/\p{^Common}/8
#'caseless' properties
88014:/\p{Ll}/8i
88015:/\p{Lu}/8i
88016:/\p{Lt}/8i
88017:/\p{Lo}/8i
#although not letters, some elements have cased variants like COMBINING GREEK
# YPOGEGRAMMENI
88018:/\p{Mn}/8i
#check Han extensions and private areas
88019:/\p{Co}/8
88020:/\p{Lo}/8
88021:/\p{Han}/8
88022:/\P{Co}/8
88023:/\P{Lo}/8
88024:/\P{Han}/8
#ucp mode
88100:/\d/8
88101:/\d/8W
88102:/[\w]/8
88103:/[\w]/8W
88104:/\s/8
88105:/\s/8W
88106:/\S/8
88107:/\S/8W
88108:/\s/
88109:/\s/W
88110:/[\w]/
88111:/[\w]/W
88112:/[[:alnum:]]/8
88113:/[[:alnum:]]/8W
88114:/[[:alpha:]]/8
88115:/[[:alpha:]]/8W
88116:/[[:blank:]]/8
88117:/[[:blank:]]/8W
88118:/[[:digit:]]/8
88119:/[[:digit:]]/8W
88120:/[[:lower:]]/8
88121:/[[:lower:]]/8W
88122:/[[:lower:]]/8i
88123:/[[:lower:]]/8Wi
88124:/[[:upper:]]/8
88125:/[[:upper:]]/8W
88126:/[[:upper:]]/8i
88127:/[[:upper:]]/8Wi
88128:/[[:space:]]/8
88129:/[[:space:]]/8W
88130:/[[:word:]]/8
88131:/[[:word:]]/8W
88132:/[[:^word:]]/8
88133:/[[:^word:]]/8W
88134:/[[:graph:]]/8
88135:/[[:graph:]]/8W
88136:/[[:cntrl:]]/8
88137:/[[:cntrl:]]/8W
88138:/\h/8
88139:/\h/8W
88140:/\H/8
88141:/\H/8W
88142:/\v/8
88143:/\v/8W
88144:/\V/8
88145:/\V/8W
#boundaries
89000:/foo\B.*\bbar/8WP
89001:/\b/8
89002:/\B/8
89003:/\bfoo/8
89004:/\bfoo/8WP
89005:/foo\b/8
89006:/foo\b/8WP
89007:/\Bfoo/8
89008:/\Bfoo/8WP
89009:/foo\B/8
89010:/foo\B/8WP
89011:/\bô/8
89012:/\bô/8WP
89013:/ô\b/8
89014:/ô\b/8WP
89015:/\b﹎/8
89016:/\b﹎/8WP
89017:/﹎\b/8
89018:/﹎\b/8WP
89019:/K/8
# (*VERB) syntax
89100:/(*UTF8)\x{7a7a}/
89101:/(*UTF8)(*UCP)\w/
89102:/(*UCP)\w/8
# Octal encoding!
89103:/\o{75172}/8
# 4byte utf8
89104:/𒀀/8
# more (*VERB) syntax
89105:/(*UCP)(*UTF8)\w/
89106:/(*UTF)(*UCP)\w/

52
tools/hscollider/test_cases/pcre/vacuous.txt Normal file
View File

@@ -0,0 +1,52 @@
# A selection of fine vacuous patterns (patterns that match the empty string).
# Only true FIREHOSE patterns need the /V flag.
22000:/.*/V
22001:/.?/V
22002:/.{0,16}/V
22003:/.*/sV
22004:/.?/sV
22005:/.{0,16}/sV
22006:/(foo|)/V
22007:/(foo|.?)/sV
22008:/(foo|.*)/sV
22009:/(foo|.{0,16})/sV
22010:/^.*/
22011:/^.?/
22012:/^.{0,16}/
22013:/^.*/s
22014:/^.?/s
22015:/^.{0,16}/s
22016:/^(foo|)/
22017:/^(foo|.?)/s
22018:/^(foo|.*)/s
22019:/^(foo|.{0,16})/s
22020:/^$/
22021:/^\z/
22022:/(foo|$)/
22023:/(foo|\z)/
22024:/(.?|$)/sV
22025:/(.?|\z)/sV
22026:/(^|\z)/
# Tricky mix of vacuous and multiline
22027:/^a?/m
22028:/^$/m
# We handle /./s as a firehose with a min start offset of zero, now.
22029:/./s
# For completeness, some other cases (especially multiline)
22030:/^/m
22031:/$/m
22032:/\z/m
22033:/\Z/m
22034:/^\z/m
22035:/^./sm
22036:/^.?/sm
22037:/(foo|$)/m
# Messy cases
22038:/((?:d|C|B|D|B*|.)){1,4}/VP