pm.lua 11 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344
  1. print('testing pattern matching')
  2. function f(s, p)
  3. local i,e = string.find(s, p)
  4. if i then return string.sub(s, i, e) end
  5. end
  6. function f1(s, p)
  7. p = string.gsub(p, "%%([0-9])", function (s) return "%" .. (s+1) end)
  8. p = string.gsub(p, "^(^?)", "%1()", 1)
  9. p = string.gsub(p, "($?)$", "()%1", 1)
  10. local t = {string.match(s, p)}
  11. return string.sub(s, t[1], t[#t] - 1)
  12. end
  13. a,b = string.find('', '') -- empty patterns are tricky
  14. assert(a == 1 and b == 0);
  15. a,b = string.find('alo', '')
  16. assert(a == 1 and b == 0)
  17. a,b = string.find('a\0o a\0o a\0o', 'a', 1) -- first position
  18. assert(a == 1 and b == 1)
  19. a,b = string.find('a\0o a\0o a\0o', 'a\0o', 2) -- starts in the midle
  20. assert(a == 5 and b == 7)
  21. a,b = string.find('a\0o a\0o a\0o', 'a\0o', 9) -- starts in the midle
  22. assert(a == 9 and b == 11)
  23. a,b = string.find('a\0a\0a\0a\0\0ab', '\0ab', 2); -- finds at the end
  24. assert(a == 9 and b == 11);
  25. a,b = string.find('a\0a\0a\0a\0\0ab', 'b') -- last position
  26. assert(a == 11 and b == 11)
  27. assert(string.find('a\0a\0a\0a\0\0ab', 'b\0') == nil) -- check ending
  28. assert(string.find('', '\0') == nil)
  29. assert(string.find('alo123alo', '12') == 4)
  30. assert(string.find('alo123alo', '^12') == nil)
  31. assert(string.match("aaab", ".*b") == "aaab")
  32. assert(string.match("aaa", ".*a") == "aaa")
  33. assert(string.match("b", ".*b") == "b")
  34. assert(string.match("aaab", ".+b") == "aaab")
  35. assert(string.match("aaa", ".+a") == "aaa")
  36. assert(not string.match("b", ".+b"))
  37. assert(string.match("aaab", ".?b") == "ab")
  38. assert(string.match("aaa", ".?a") == "aa")
  39. assert(string.match("b", ".?b") == "b")
  40. assert(f('aloALO', '%l*') == 'alo')
  41. assert(f('aLo_ALO', '%a*') == 'aLo')
  42. assert(f(" \n\r*&\n\r xuxu \n\n", "%g%g%g+") == "xuxu")
  43. assert(f('aaab', 'a*') == 'aaa');
  44. assert(f('aaa', '^.*$') == 'aaa');
  45. assert(f('aaa', 'b*') == '');
  46. assert(f('aaa', 'ab*a') == 'aa')
  47. assert(f('aba', 'ab*a') == 'aba')
  48. assert(f('aaab', 'a+') == 'aaa')
  49. assert(f('aaa', '^.+$') == 'aaa')
  50. assert(f('aaa', 'b+') == nil)
  51. assert(f('aaa', 'ab+a') == nil)
  52. assert(f('aba', 'ab+a') == 'aba')
  53. assert(f('a$a', '.$') == 'a')
  54. assert(f('a$a', '.%$') == 'a$')
  55. assert(f('a$a', '.$.') == 'a$a')
  56. assert(f('a$a', '$$') == nil)
  57. assert(f('a$b', 'a$') == nil)
  58. assert(f('a$a', '$') == '')
  59. assert(f('', 'b*') == '')
  60. assert(f('aaa', 'bb*') == nil)
  61. assert(f('aaab', 'a-') == '')
  62. assert(f('aaa', '^.-$') == 'aaa')
  63. assert(f('aabaaabaaabaaaba', 'b.*b') == 'baaabaaabaaab')
  64. assert(f('aabaaabaaabaaaba', 'b.-b') == 'baaab')
  65. assert(f('alo xo', '.o$') == 'xo')
  66. assert(f(' \n isto é assim', '%S%S*') == 'isto')
  67. assert(f(' \n isto é assim', '%S*$') == 'assim')
  68. assert(f(' \n isto é assim', '[a-z]*$') == 'assim')
  69. assert(f('um caracter ? extra', '[^%sa-z]') == '?')
  70. assert(f('', 'a?') == '')
  71. assert(f('á', 'á?') == 'á')
  72. assert(f('ábl', 'á?b?l?') == 'ábl')
  73. assert(f(' ábl', 'á?b?l?') == '')
  74. assert(f('aa', '^aa?a?a') == 'aa')
  75. assert(f(']]]áb', '[^]]') == 'á')
  76. assert(f("0alo alo", "%x*") == "0a")
  77. assert(f("alo alo", "%C+") == "alo alo")
  78. print('+')
  79. assert(f1('alo alx 123 b\0o b\0o', '(..*) %1') == "b\0o b\0o")
  80. assert(f1('axz123= 4= 4 34', '(.+)=(.*)=%2 %1') == '3= 4= 4 3')
  81. assert(f1('=======', '^(=*)=%1$') == '=======')
  82. assert(string.match('==========', '^([=]*)=%1$') == nil)
  83. local function range (i, j)
  84. if i <= j then
  85. return i, range(i+1, j)
  86. end
  87. end
  88. local abc = string.char(range(0, 255));
  89. assert(string.len(abc) == 256)
  90. function strset (p)
  91. local res = {s=''}
  92. string.gsub(abc, p, function (c) res.s = res.s .. c end)
  93. return res.s
  94. end;
  95. assert(string.len(strset('[\200-\210]')) == 11)
  96. assert(strset('[a-z]') == "abcdefghijklmnopqrstuvwxyz")
  97. assert(strset('[a-z%d]') == strset('[%da-uu-z]'))
  98. assert(strset('[a-]') == "-a")
  99. assert(strset('[^%W]') == strset('[%w]'))
  100. assert(strset('[]%%]') == '%]')
  101. assert(strset('[a%-z]') == '-az')
  102. assert(strset('[%^%[%-a%]%-b]') == '-[]^ab')
  103. assert(strset('%Z') == strset('[\1-\255]'))
  104. assert(strset('.') == strset('[\1-\255%z]'))
  105. print('+');
  106. assert(string.match("alo xyzK", "(%w+)K") == "xyz")
  107. assert(string.match("254 K", "(%d*)K") == "")
  108. assert(string.match("alo ", "(%w*)$") == "")
  109. assert(string.match("alo ", "(%w+)$") == nil)
  110. assert(string.find("(álo)", "%(á") == 1)
  111. local a, b, c, d, e = string.match("âlo alo", "^(((.).).* (%w*))$")
  112. assert(a == 'âlo alo' and b == 'âl' and c == 'â' and d == 'alo' and e == nil)
  113. a, b, c, d = string.match('0123456789', '(.+(.?)())')
  114. assert(a == '0123456789' and b == '' and c == 11 and d == nil)
  115. print('+')
  116. assert(string.gsub('ülo ülo', 'ü', 'x') == 'xlo xlo')
  117. assert(string.gsub('alo úlo ', ' +$', '') == 'alo úlo') -- trim
  118. assert(string.gsub(' alo alo ', '^%s*(.-)%s*$', '%1') == 'alo alo') -- double trim
  119. assert(string.gsub('alo alo \n 123\n ', '%s+', ' ') == 'alo alo 123 ')
  120. t = "abç d"
  121. a, b = string.gsub(t, '(.)', '%1@')
  122. assert('@'..a == string.gsub(t, '', '@') and b == 5)
  123. a, b = string.gsub('abçd', '(.)', '%0@', 2)
  124. assert(a == 'a@b@çd' and b == 2)
  125. assert(string.gsub('alo alo', '()[al]', '%1') == '12o 56o')
  126. assert(string.gsub("abc=xyz", "(%w*)(%p)(%w+)", "%3%2%1-%0") ==
  127. "xyz=abc-abc=xyz")
  128. assert(string.gsub("abc", "%w", "%1%0") == "aabbcc")
  129. assert(string.gsub("abc", "%w+", "%0%1") == "abcabc")
  130. assert(string.gsub('áéí', '$', '\0óú') == 'áéí\0óú')
  131. assert(string.gsub('', '^', 'r') == 'r')
  132. assert(string.gsub('', '$', 'r') == 'r')
  133. print('+')
  134. assert(string.gsub("um (dois) tres (quatro)", "(%(%w+%))", string.upper) ==
  135. "um (DOIS) tres (QUATRO)")
  136. do
  137. local function setglobal (n,v) rawset(_G, n, v) end
  138. string.gsub("a=roberto,roberto=a", "(%w+)=(%w%w*)", setglobal)
  139. assert(_G.a=="roberto" and _G.roberto=="a")
  140. end
  141. function f(a,b) return string.gsub(a,'.',b) end
  142. assert(string.gsub("trocar tudo em |teste|b| é |beleza|al|", "|([^|]*)|([^|]*)|", f) ==
  143. "trocar tudo em bbbbb é alalalalalal")
  144. local function dostring (s) return load(s)() or "" end
  145. assert(string.gsub("alo $a=1$ novamente $return a$", "$([^$]*)%$", dostring) ==
  146. "alo novamente 1")
  147. x = string.gsub("$x=string.gsub('alo', '.', string.upper)$ assim vai para $return x$",
  148. "$([^$]*)%$", dostring)
  149. assert(x == ' assim vai para ALO')
  150. t = {}
  151. s = 'a alo jose joao'
  152. r = string.gsub(s, '()(%w+)()', function (a,w,b)
  153. assert(string.len(w) == b-a);
  154. t[a] = b-a;
  155. end)
  156. assert(s == r and t[1] == 1 and t[3] == 3 and t[7] == 4 and t[13] == 4)
  157. function isbalanced (s)
  158. return string.find(string.gsub(s, "%b()", ""), "[()]") == nil
  159. end
  160. assert(isbalanced("(9 ((8))(\0) 7) \0\0 a b ()(c)() a"))
  161. assert(not isbalanced("(9 ((8) 7) a b (\0 c) a"))
  162. assert(string.gsub("alo 'oi' alo", "%b''", '"') == 'alo " alo')
  163. local t = {"apple", "orange", "lime"; n=0}
  164. assert(string.gsub("x and x and x", "x", function () t.n=t.n+1; return t[t.n] end)
  165. == "apple and orange and lime")
  166. t = {n=0}
  167. string.gsub("first second word", "%w%w*", function (w) t.n=t.n+1; t[t.n] = w end)
  168. assert(t[1] == "first" and t[2] == "second" and t[3] == "word" and t.n == 3)
  169. t = {n=0}
  170. assert(string.gsub("first second word", "%w+",
  171. function (w) t.n=t.n+1; t[t.n] = w end, 2) == "first second word")
  172. assert(t[1] == "first" and t[2] == "second" and t[3] == nil)
  173. assert(not pcall(string.gsub, "alo", "(.", print))
  174. assert(not pcall(string.gsub, "alo", ".)", print))
  175. assert(not pcall(string.gsub, "alo", "(.", {}))
  176. assert(not pcall(string.gsub, "alo", "(.)", "%2"))
  177. assert(not pcall(string.gsub, "alo", "(%1)", "a"))
  178. assert(not pcall(string.gsub, "alo", "(%0)", "a"))
  179. -- bug since 2.5 (C-stack overflow)
  180. do
  181. local function f (size)
  182. local s = string.rep("a", size)
  183. local p = string.rep(".?", size)
  184. return pcall(string.match, s, p)
  185. end
  186. local r, m = f(80)
  187. assert(r and #m == 80)
  188. r, m = f(200000)
  189. assert(not r and string.find(m, "too complex"))
  190. end
  191. if not _soft then
  192. -- big strings
  193. local a = string.rep('a', 300000)
  194. assert(string.find(a, '^a*.?$'))
  195. assert(not string.find(a, '^a*.?b$'))
  196. assert(string.find(a, '^a-.?$'))
  197. -- bug in 5.1.2
  198. a = string.rep('a', 10000) .. string.rep('b', 10000)
  199. assert(not pcall(string.gsub, a, 'b'))
  200. end
  201. -- recursive nest of gsubs
  202. function rev (s)
  203. return string.gsub(s, "(.)(.+)", function (c,s1) return rev(s1)..c end)
  204. end
  205. local x = "abcdef"
  206. assert(rev(rev(x)) == x)
  207. -- gsub with tables
  208. assert(string.gsub("alo alo", ".", {}) == "alo alo")
  209. assert(string.gsub("alo alo", "(.)", {a="AA", l=""}) == "AAo AAo")
  210. assert(string.gsub("alo alo", "(.).", {a="AA", l="K"}) == "AAo AAo")
  211. assert(string.gsub("alo alo", "((.)(.?))", {al="AA", o=false}) == "AAo AAo")
  212. assert(string.gsub("alo alo", "().", {2,5,6}) == "256 alo")
  213. t = {}; setmetatable(t, {__index = function (t,s) return string.upper(s) end})
  214. assert(string.gsub("a alo b hi", "%w%w+", t) == "a ALO b HI")
  215. -- tests for gmatch
  216. local a = 0
  217. for i in string.gmatch('abcde', '()') do assert(i == a+1); a=i end
  218. assert(a==6)
  219. t = {n=0}
  220. for w in string.gmatch("first second word", "%w+") do
  221. t.n=t.n+1; t[t.n] = w
  222. end
  223. assert(t[1] == "first" and t[2] == "second" and t[3] == "word")
  224. t = {3, 6, 9}
  225. for i in string.gmatch ("xuxx uu ppar r", "()(.)%2") do
  226. assert(i == table.remove(t, 1))
  227. end
  228. assert(#t == 0)
  229. t = {}
  230. for i,j in string.gmatch("13 14 10 = 11, 15= 16, 22=23", "(%d+)%s*=%s*(%d+)") do
  231. t[i] = j
  232. end
  233. a = 0
  234. for k,v in pairs(t) do assert(k+1 == v+0); a=a+1 end
  235. assert(a == 3)
  236. -- tests for `%f' (`frontiers')
  237. assert(string.gsub("aaa aa a aaa a", "%f[%w]a", "x") == "xaa xa x xaa x")
  238. assert(string.gsub("[[]] [][] [[[[", "%f[[].", "x") == "x[]] x]x] x[[[")
  239. assert(string.gsub("01abc45de3", "%f[%d]", ".") == ".01abc.45de.3")
  240. assert(string.gsub("01abc45 de3x", "%f[%D]%w", ".") == "01.bc45 de3.")
  241. assert(string.gsub("function", "%f[\1-\255]%w", ".") == ".unction")
  242. assert(string.gsub("function", "%f[^\1-\255]", ".") == "function.")
  243. assert(string.find("a", "%f[a]") == 1)
  244. assert(string.find("a", "%f[^%z]") == 1)
  245. assert(string.find("a", "%f[^%l]") == 2)
  246. assert(string.find("aba", "%f[a%z]") == 3)
  247. assert(string.find("aba", "%f[%z]") == 4)
  248. assert(not string.find("aba", "%f[%l%z]"))
  249. assert(not string.find("aba", "%f[^%l%z]"))
  250. local i, e = string.find(" alo aalo allo", "%f[%S].-%f[%s].-%f[%S]")
  251. assert(i == 2 and e == 5)
  252. local k = string.match(" alo aalo allo", "%f[%S](.-%f[%s].-%f[%S])")
  253. assert(k == 'alo ')
  254. local a = {1, 5, 9, 14, 17,}
  255. for k in string.gmatch("alo alo th02 is 1hat", "()%f[%w%d]") do
  256. assert(table.remove(a, 1) == k)
  257. end
  258. assert(#a == 0)
  259. -- malformed patterns
  260. local function malform (p, m)
  261. m = m or "malformed"
  262. local r, msg = pcall(string.find, "a", p)
  263. assert(not r and string.find(msg, m))
  264. end
  265. malform("[a")
  266. malform("[]")
  267. malform("[^]")
  268. malform("[a%]")
  269. malform("[a%")
  270. malform("%b")
  271. malform("%ba")
  272. malform("%")
  273. malform("%f", "missing")
  274. -- \0 in patterns
  275. assert(string.match("ab\0\1\2c", "[\0-\2]+") == "\0\1\2")
  276. assert(string.match("ab\0\1\2c", "[\0-\0]+") == "\0")
  277. assert(string.find("b$a", "$\0?") == 2)
  278. assert(string.find("abc\0efg", "%\0") == 4)
  279. assert(string.match("abc\0efg\0\1e\1g", "%b\0\1") == "\0efg\0\1e\1")
  280. assert(string.match("abc\0\0\0", "%\0+") == "\0\0\0")
  281. assert(string.match("abc\0\0\0", "%\0%\0?") == "\0\0")
  282. -- magic char after \0
  283. assert(string.find("abc\0\0","\0.") == 4)
  284. assert(string.find("abcx\0\0abc\0abc","x\0\0abc\0a.") == 4)
  285. print('OK')