hscollider: example test cases

tools/hscollider/test_cases/pcre/utf8.txt

@@ -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/