diff options
| author | Roberto Ierusalimschy <roberto@inf.puc-rio.br> | 2018-07-09 12:33:01 -0300 |
|---|---|---|
| committer | Roberto Ierusalimschy <roberto@inf.puc-rio.br> | 2018-07-09 12:33:01 -0300 |
| commit | 7c519dfbd0c68b952f0849e01deaa3750e1f8153 (patch) | |
| tree | dde3ddbba310877db725df37a0d9f2cbe4e2a8f9 /testes/pm.lua | |
| parent | f59e6a93c0ad38a27a420e51abf8f13d962446b5 (diff) | |
| download | lua-7c519dfbd0c68b952f0849e01deaa3750e1f8153.tar.gz lua-7c519dfbd0c68b952f0849e01deaa3750e1f8153.tar.bz2 lua-7c519dfbd0c68b952f0849e01deaa3750e1f8153.zip | |
Added manual and tests for version 5.4-w2
Diffstat (limited to 'testes/pm.lua')
| -rw-r--r-- | testes/pm.lua | 374 |
1 files changed, 374 insertions, 0 deletions
diff --git a/testes/pm.lua b/testes/pm.lua new file mode 100644 index 00000000..e517c8b6 --- /dev/null +++ b/testes/pm.lua | |||
| @@ -0,0 +1,374 @@ | |||
| 1 | -- $Id: pm.lua,v 1.50 2018/03/12 14:19:36 roberto Exp $ | ||
| 2 | -- See Copyright Notice in file all.lua | ||
| 3 | |||
| 4 | print('testing pattern matching') | ||
| 5 | |||
| 6 | local function checkerror (msg, f, ...) | ||
| 7 | local s, err = pcall(f, ...) | ||
| 8 | assert(not s and string.find(err, msg)) | ||
| 9 | end | ||
| 10 | |||
| 11 | |||
| 12 | function f(s, p) | ||
| 13 | local i,e = string.find(s, p) | ||
| 14 | if i then return string.sub(s, i, e) end | ||
| 15 | end | ||
| 16 | |||
| 17 | a,b = string.find('', '') -- empty patterns are tricky | ||
| 18 | assert(a == 1 and b == 0); | ||
| 19 | a,b = string.find('alo', '') | ||
| 20 | assert(a == 1 and b == 0) | ||
| 21 | a,b = string.find('a\0o a\0o a\0o', 'a', 1) -- first position | ||
| 22 | assert(a == 1 and b == 1) | ||
| 23 | a,b = string.find('a\0o a\0o a\0o', 'a\0o', 2) -- starts in the midle | ||
| 24 | assert(a == 5 and b == 7) | ||
| 25 | a,b = string.find('a\0o a\0o a\0o', 'a\0o', 9) -- starts in the midle | ||
| 26 | assert(a == 9 and b == 11) | ||
| 27 | a,b = string.find('a\0a\0a\0a\0\0ab', '\0ab', 2); -- finds at the end | ||
| 28 | assert(a == 9 and b == 11); | ||
| 29 | a,b = string.find('a\0a\0a\0a\0\0ab', 'b') -- last position | ||
| 30 | assert(a == 11 and b == 11) | ||
| 31 | assert(string.find('a\0a\0a\0a\0\0ab', 'b\0') == nil) -- check ending | ||
| 32 | assert(string.find('', '\0') == nil) | ||
| 33 | assert(string.find('alo123alo', '12') == 4) | ||
| 34 | assert(string.find('alo123alo', '^12') == nil) | ||
| 35 | |||
| 36 | assert(string.match("aaab", ".*b") == "aaab") | ||
| 37 | assert(string.match("aaa", ".*a") == "aaa") | ||
| 38 | assert(string.match("b", ".*b") == "b") | ||
| 39 | |||
| 40 | assert(string.match("aaab", ".+b") == "aaab") | ||
| 41 | assert(string.match("aaa", ".+a") == "aaa") | ||
| 42 | assert(not string.match("b", ".+b")) | ||
| 43 | |||
| 44 | assert(string.match("aaab", ".?b") == "ab") | ||
| 45 | assert(string.match("aaa", ".?a") == "aa") | ||
| 46 | assert(string.match("b", ".?b") == "b") | ||
| 47 | |||
| 48 | assert(f('aloALO', '%l*') == 'alo') | ||
| 49 | assert(f('aLo_ALO', '%a*') == 'aLo') | ||
| 50 | |||
| 51 | assert(f(" \n\r*&\n\r xuxu \n\n", "%g%g%g+") == "xuxu") | ||
| 52 | |||
| 53 | assert(f('aaab', 'a*') == 'aaa'); | ||
| 54 | assert(f('aaa', '^.*$') == 'aaa'); | ||
| 55 | assert(f('aaa', 'b*') == ''); | ||
| 56 | assert(f('aaa', 'ab*a') == 'aa') | ||
| 57 | assert(f('aba', 'ab*a') == 'aba') | ||
| 58 | assert(f('aaab', 'a+') == 'aaa') | ||
| 59 | assert(f('aaa', '^.+$') == 'aaa') | ||
| 60 | assert(f('aaa', 'b+') == nil) | ||
| 61 | assert(f('aaa', 'ab+a') == nil) | ||
| 62 | assert(f('aba', 'ab+a') == 'aba') | ||
| 63 | assert(f('a$a', '.$') == 'a') | ||
| 64 | assert(f('a$a', '.%$') == 'a$') | ||
| 65 | assert(f('a$a', '.$.') == 'a$a') | ||
| 66 | assert(f('a$a', '$$') == nil) | ||
| 67 | assert(f('a$b', 'a$') == nil) | ||
| 68 | assert(f('a$a', '$') == '') | ||
| 69 | assert(f('', 'b*') == '') | ||
| 70 | assert(f('aaa', 'bb*') == nil) | ||
| 71 | assert(f('aaab', 'a-') == '') | ||
| 72 | assert(f('aaa', '^.-$') == 'aaa') | ||
| 73 | assert(f('aabaaabaaabaaaba', 'b.*b') == 'baaabaaabaaab') | ||
| 74 | assert(f('aabaaabaaabaaaba', 'b.-b') == 'baaab') | ||
| 75 | assert(f('alo xo', '.o$') == 'xo') | ||
| 76 | assert(f(' \n isto é assim', '%S%S*') == 'isto') | ||
| 77 | assert(f(' \n isto é assim', '%S*$') == 'assim') | ||
| 78 | assert(f(' \n isto é assim', '[a-z]*$') == 'assim') | ||
| 79 | assert(f('um caracter ? extra', '[^%sa-z]') == '?') | ||
| 80 | assert(f('', 'a?') == '') | ||
| 81 | assert(f('á', 'á?') == 'á') | ||
| 82 | assert(f('ábl', 'á?b?l?') == 'ábl') | ||
| 83 | assert(f(' ábl', 'á?b?l?') == '') | ||
| 84 | assert(f('aa', '^aa?a?a') == 'aa') | ||
| 85 | assert(f(']]]áb', '[^]]') == 'á') | ||
| 86 | assert(f("0alo alo", "%x*") == "0a") | ||
| 87 | assert(f("alo alo", "%C+") == "alo alo") | ||
| 88 | print('+') | ||
| 89 | |||
| 90 | |||
| 91 | function f1(s, p) | ||
| 92 | p = string.gsub(p, "%%([0-9])", function (s) | ||
| 93 | return "%" .. (tonumber(s)+1) | ||
| 94 | end) | ||
| 95 | p = string.gsub(p, "^(^?)", "%1()", 1) | ||
| 96 | p = string.gsub(p, "($?)$", "()%1", 1) | ||
| 97 | local t = {string.match(s, p)} | ||
| 98 | return string.sub(s, t[1], t[#t] - 1) | ||
| 99 | end | ||
| 100 | |||
| 101 | assert(f1('alo alx 123 b\0o b\0o', '(..*) %1') == "b\0o b\0o") | ||
| 102 | assert(f1('axz123= 4= 4 34', '(.+)=(.*)=%2 %1') == '3= 4= 4 3') | ||
| 103 | assert(f1('=======', '^(=*)=%1$') == '=======') | ||
| 104 | assert(string.match('==========', '^([=]*)=%1$') == nil) | ||
| 105 | |||
| 106 | local function range (i, j) | ||
| 107 | if i <= j then | ||
| 108 | return i, range(i+1, j) | ||
| 109 | end | ||
| 110 | end | ||
| 111 | |||
| 112 | local abc = string.char(range(0, 127)) .. string.char(range(128, 255)); | ||
| 113 | |||
| 114 | assert(string.len(abc) == 256) | ||
| 115 | |||
| 116 | function strset (p) | ||
| 117 | local res = {s=''} | ||
| 118 | string.gsub(abc, p, function (c) res.s = res.s .. c end) | ||
| 119 | return res.s | ||
| 120 | end; | ||
| 121 | |||
| 122 | assert(string.len(strset('[\200-\210]')) == 11) | ||
| 123 | |||
| 124 | assert(strset('[a-z]') == "abcdefghijklmnopqrstuvwxyz") | ||
| 125 | assert(strset('[a-z%d]') == strset('[%da-uu-z]')) | ||
| 126 | assert(strset('[a-]') == "-a") | ||
| 127 | assert(strset('[^%W]') == strset('[%w]')) | ||
| 128 | assert(strset('[]%%]') == '%]') | ||
| 129 | assert(strset('[a%-z]') == '-az') | ||
| 130 | assert(strset('[%^%[%-a%]%-b]') == '-[]^ab') | ||
| 131 | assert(strset('%Z') == strset('[\1-\255]')) | ||
| 132 | assert(strset('.') == strset('[\1-\255%z]')) | ||
| 133 | print('+'); | ||
| 134 | |||
| 135 | assert(string.match("alo xyzK", "(%w+)K") == "xyz") | ||
| 136 | assert(string.match("254 K", "(%d*)K") == "") | ||
| 137 | assert(string.match("alo ", "(%w*)$") == "") | ||
| 138 | assert(string.match("alo ", "(%w+)$") == nil) | ||
| 139 | assert(string.find("(álo)", "%(á") == 1) | ||
| 140 | local a, b, c, d, e = string.match("âlo alo", "^(((.).).* (%w*))$") | ||
| 141 | assert(a == 'âlo alo' and b == 'âl' and c == 'â' and d == 'alo' and e == nil) | ||
| 142 | a, b, c, d = string.match('0123456789', '(.+(.?)())') | ||
| 143 | assert(a == '0123456789' and b == '' and c == 11 and d == nil) | ||
| 144 | print('+') | ||
| 145 | |||
| 146 | assert(string.gsub('ülo ülo', 'ü', 'x') == 'xlo xlo') | ||
| 147 | assert(string.gsub('alo úlo ', ' +$', '') == 'alo úlo') -- trim | ||
| 148 | assert(string.gsub(' alo alo ', '^%s*(.-)%s*$', '%1') == 'alo alo') -- double trim | ||
| 149 | assert(string.gsub('alo alo \n 123\n ', '%s+', ' ') == 'alo alo 123 ') | ||
| 150 | t = "abç d" | ||
| 151 | a, b = string.gsub(t, '(.)', '%1@') | ||
| 152 | assert('@'..a == string.gsub(t, '', '@') and b == 5) | ||
| 153 | a, b = string.gsub('abçd', '(.)', '%0@', 2) | ||
| 154 | assert(a == 'a@b@çd' and b == 2) | ||
| 155 | assert(string.gsub('alo alo', '()[al]', '%1') == '12o 56o') | ||
| 156 | assert(string.gsub("abc=xyz", "(%w*)(%p)(%w+)", "%3%2%1-%0") == | ||
| 157 | "xyz=abc-abc=xyz") | ||
| 158 | assert(string.gsub("abc", "%w", "%1%0") == "aabbcc") | ||
| 159 | assert(string.gsub("abc", "%w+", "%0%1") == "abcabc") | ||
| 160 | assert(string.gsub('áéí', '$', '\0óú') == 'áéí\0óú') | ||
| 161 | assert(string.gsub('', '^', 'r') == 'r') | ||
| 162 | assert(string.gsub('', '$', 'r') == 'r') | ||
| 163 | print('+') | ||
| 164 | |||
| 165 | |||
| 166 | do -- new (5.3.3) semantics for empty matches | ||
| 167 | assert(string.gsub("a b cd", " *", "-") == "-a-b-c-d-") | ||
| 168 | |||
| 169 | local res = "" | ||
| 170 | local sub = "a \nbc\t\td" | ||
| 171 | local i = 1 | ||
| 172 | for p, e in string.gmatch(sub, "()%s*()") do | ||
| 173 | res = res .. string.sub(sub, i, p - 1) .. "-" | ||
| 174 | i = e | ||
| 175 | end | ||
| 176 | assert(res == "-a-b-c-d-") | ||
| 177 | end | ||
| 178 | |||
| 179 | |||
| 180 | assert(string.gsub("um (dois) tres (quatro)", "(%(%w+%))", string.upper) == | ||
| 181 | "um (DOIS) tres (QUATRO)") | ||
| 182 | |||
| 183 | do | ||
| 184 | local function setglobal (n,v) rawset(_G, n, v) end | ||
| 185 | string.gsub("a=roberto,roberto=a", "(%w+)=(%w%w*)", setglobal) | ||
| 186 | assert(_G.a=="roberto" and _G.roberto=="a") | ||
| 187 | end | ||
| 188 | |||
| 189 | function f(a,b) return string.gsub(a,'.',b) end | ||
| 190 | assert(string.gsub("trocar tudo em |teste|b| é |beleza|al|", "|([^|]*)|([^|]*)|", f) == | ||
| 191 | "trocar tudo em bbbbb é alalalalalal") | ||
| 192 | |||
| 193 | local function dostring (s) return load(s, "")() or "" end | ||
| 194 | assert(string.gsub("alo $a='x'$ novamente $return a$", | ||
| 195 | "$([^$]*)%$", | ||
| 196 | dostring) == "alo novamente x") | ||
| 197 | |||
| 198 | x = string.gsub("$x=string.gsub('alo', '.', string.upper)$ assim vai para $return x$", | ||
| 199 | "$([^$]*)%$", dostring) | ||
| 200 | assert(x == ' assim vai para ALO') | ||
| 201 | |||
| 202 | t = {} | ||
| 203 | s = 'a alo jose joao' | ||
| 204 | r = string.gsub(s, '()(%w+)()', function (a,w,b) | ||
| 205 | assert(string.len(w) == b-a); | ||
| 206 | t[a] = b-a; | ||
| 207 | end) | ||
| 208 | assert(s == r and t[1] == 1 and t[3] == 3 and t[7] == 4 and t[13] == 4) | ||
| 209 | |||
| 210 | |||
| 211 | function isbalanced (s) | ||
| 212 | return string.find(string.gsub(s, "%b()", ""), "[()]") == nil | ||
| 213 | end | ||
| 214 | |||
| 215 | assert(isbalanced("(9 ((8))(\0) 7) \0\0 a b ()(c)() a")) | ||
| 216 | assert(not isbalanced("(9 ((8) 7) a b (\0 c) a")) | ||
| 217 | assert(string.gsub("alo 'oi' alo", "%b''", '"') == 'alo " alo') | ||
| 218 | |||
| 219 | |||
| 220 | local t = {"apple", "orange", "lime"; n=0} | ||
| 221 | assert(string.gsub("x and x and x", "x", function () t.n=t.n+1; return t[t.n] end) | ||
| 222 | == "apple and orange and lime") | ||
| 223 | |||
| 224 | t = {n=0} | ||
| 225 | string.gsub("first second word", "%w%w*", function (w) t.n=t.n+1; t[t.n] = w end) | ||
| 226 | assert(t[1] == "first" and t[2] == "second" and t[3] == "word" and t.n == 3) | ||
| 227 | |||
| 228 | t = {n=0} | ||
| 229 | assert(string.gsub("first second word", "%w+", | ||
| 230 | function (w) t.n=t.n+1; t[t.n] = w end, 2) == "first second word") | ||
| 231 | assert(t[1] == "first" and t[2] == "second" and t[3] == undef) | ||
| 232 | |||
| 233 | checkerror("invalid replacement value %(a table%)", | ||
| 234 | string.gsub, "alo", ".", {a = {}}) | ||
| 235 | checkerror("invalid capture index %%2", string.gsub, "alo", ".", "%2") | ||
| 236 | checkerror("invalid capture index %%0", string.gsub, "alo", "(%0)", "a") | ||
| 237 | checkerror("invalid capture index %%1", string.gsub, "alo", "(%1)", "a") | ||
| 238 | checkerror("invalid use of '%%'", string.gsub, "alo", ".", "%x") | ||
| 239 | |||
| 240 | -- bug since 2.5 (C-stack overflow) | ||
| 241 | do | ||
| 242 | local function f (size) | ||
| 243 | local s = string.rep("a", size) | ||
| 244 | local p = string.rep(".?", size) | ||
| 245 | return pcall(string.match, s, p) | ||
| 246 | end | ||
| 247 | local r, m = f(80) | ||
| 248 | assert(r and #m == 80) | ||
| 249 | r, m = f(200000) | ||
| 250 | assert(not r and string.find(m, "too complex")) | ||
| 251 | end | ||
| 252 | |||
| 253 | if not _soft then | ||
| 254 | print("big strings") | ||
| 255 | local a = string.rep('a', 300000) | ||
| 256 | assert(string.find(a, '^a*.?$')) | ||
| 257 | assert(not string.find(a, '^a*.?b$')) | ||
| 258 | assert(string.find(a, '^a-.?$')) | ||
| 259 | |||
| 260 | -- bug in 5.1.2 | ||
| 261 | a = string.rep('a', 10000) .. string.rep('b', 10000) | ||
| 262 | assert(not pcall(string.gsub, a, 'b')) | ||
| 263 | end | ||
| 264 | |||
| 265 | -- recursive nest of gsubs | ||
| 266 | function rev (s) | ||
| 267 | return string.gsub(s, "(.)(.+)", function (c,s1) return rev(s1)..c end) | ||
| 268 | end | ||
| 269 | |||
| 270 | local x = "abcdef" | ||
| 271 | assert(rev(rev(x)) == x) | ||
| 272 | |||
| 273 | |||
| 274 | -- gsub with tables | ||
| 275 | assert(string.gsub("alo alo", ".", {}) == "alo alo") | ||
| 276 | assert(string.gsub("alo alo", "(.)", {a="AA", l=""}) == "AAo AAo") | ||
| 277 | assert(string.gsub("alo alo", "(.).", {a="AA", l="K"}) == "AAo AAo") | ||
| 278 | assert(string.gsub("alo alo", "((.)(.?))", {al="AA", o=false}) == "AAo AAo") | ||
| 279 | |||
| 280 | assert(string.gsub("alo alo", "().", {'x','yy','zzz'}) == "xyyzzz alo") | ||
| 281 | |||
| 282 | t = {}; setmetatable(t, {__index = function (t,s) return string.upper(s) end}) | ||
| 283 | assert(string.gsub("a alo b hi", "%w%w+", t) == "a ALO b HI") | ||
| 284 | |||
| 285 | |||
| 286 | -- tests for gmatch | ||
| 287 | local a = 0 | ||
| 288 | for i in string.gmatch('abcde', '()') do assert(i == a+1); a=i end | ||
| 289 | assert(a==6) | ||
| 290 | |||
| 291 | t = {n=0} | ||
| 292 | for w in string.gmatch("first second word", "%w+") do | ||
| 293 | t.n=t.n+1; t[t.n] = w | ||
| 294 | end | ||
| 295 | assert(t[1] == "first" and t[2] == "second" and t[3] == "word") | ||
| 296 | |||
| 297 | t = {3, 6, 9} | ||
| 298 | for i in string.gmatch ("xuxx uu ppar r", "()(.)%2") do | ||
| 299 | assert(i == table.remove(t, 1)) | ||
| 300 | end | ||
| 301 | assert(#t == 0) | ||
| 302 | |||
| 303 | t = {} | ||
| 304 | for i,j in string.gmatch("13 14 10 = 11, 15= 16, 22=23", "(%d+)%s*=%s*(%d+)") do | ||
| 305 | t[tonumber(i)] = tonumber(j) | ||
| 306 | end | ||
| 307 | a = 0 | ||
| 308 | for k,v in pairs(t) do assert(k+1 == v+0); a=a+1 end | ||
| 309 | assert(a == 3) | ||
| 310 | |||
| 311 | |||
| 312 | -- tests for `%f' (`frontiers') | ||
| 313 | |||
| 314 | assert(string.gsub("aaa aa a aaa a", "%f[%w]a", "x") == "xaa xa x xaa x") | ||
| 315 | assert(string.gsub("[[]] [][] [[[[", "%f[[].", "x") == "x[]] x]x] x[[[") | ||
| 316 | assert(string.gsub("01abc45de3", "%f[%d]", ".") == ".01abc.45de.3") | ||
| 317 | assert(string.gsub("01abc45 de3x", "%f[%D]%w", ".") == "01.bc45 de3.") | ||
| 318 | assert(string.gsub("function", "%f[\1-\255]%w", ".") == ".unction") | ||
| 319 | assert(string.gsub("function", "%f[^\1-\255]", ".") == "function.") | ||
| 320 | |||
| 321 | assert(string.find("a", "%f[a]") == 1) | ||
| 322 | assert(string.find("a", "%f[^%z]") == 1) | ||
| 323 | assert(string.find("a", "%f[^%l]") == 2) | ||
| 324 | assert(string.find("aba", "%f[a%z]") == 3) | ||
| 325 | assert(string.find("aba", "%f[%z]") == 4) | ||
| 326 | assert(not string.find("aba", "%f[%l%z]")) | ||
| 327 | assert(not string.find("aba", "%f[^%l%z]")) | ||
| 328 | |||
| 329 | local i, e = string.find(" alo aalo allo", "%f[%S].-%f[%s].-%f[%S]") | ||
| 330 | assert(i == 2 and e == 5) | ||
| 331 | local k = string.match(" alo aalo allo", "%f[%S](.-%f[%s].-%f[%S])") | ||
| 332 | assert(k == 'alo ') | ||
| 333 | |||
| 334 | local a = {1, 5, 9, 14, 17,} | ||
| 335 | for k in string.gmatch("alo alo th02 is 1hat", "()%f[%w%d]") do | ||
| 336 | assert(table.remove(a, 1) == k) | ||
| 337 | end | ||
| 338 | assert(#a == 0) | ||
| 339 | |||
| 340 | |||
| 341 | -- malformed patterns | ||
| 342 | local function malform (p, m) | ||
| 343 | m = m or "malformed" | ||
| 344 | local r, msg = pcall(string.find, "a", p) | ||
| 345 | assert(not r and string.find(msg, m)) | ||
| 346 | end | ||
| 347 | |||
| 348 | malform("(.", "unfinished capture") | ||
| 349 | malform(".)", "invalid pattern capture") | ||
| 350 | malform("[a") | ||
| 351 | malform("[]") | ||
| 352 | malform("[^]") | ||
| 353 | malform("[a%]") | ||
| 354 | malform("[a%") | ||
| 355 | malform("%b") | ||
| 356 | malform("%ba") | ||
| 357 | malform("%") | ||
| 358 | malform("%f", "missing") | ||
| 359 | |||
| 360 | -- \0 in patterns | ||
| 361 | assert(string.match("ab\0\1\2c", "[\0-\2]+") == "\0\1\2") | ||
| 362 | assert(string.match("ab\0\1\2c", "[\0-\0]+") == "\0") | ||
| 363 | assert(string.find("b$a", "$\0?") == 2) | ||
| 364 | assert(string.find("abc\0efg", "%\0") == 4) | ||
| 365 | assert(string.match("abc\0efg\0\1e\1g", "%b\0\1") == "\0efg\0\1e\1") | ||
| 366 | assert(string.match("abc\0\0\0", "%\0+") == "\0\0\0") | ||
| 367 | assert(string.match("abc\0\0\0", "%\0%\0?") == "\0\0") | ||
| 368 | |||
| 369 | -- magic char after \0 | ||
| 370 | assert(string.find("abc\0\0","\0.") == 4) | ||
| 371 | assert(string.find("abcx\0\0abc\0abc","x\0\0abc\0a.") == 4) | ||
| 372 | |||
| 373 | print('OK') | ||
| 374 | |||
