tests/regex/wxreg.test

   1 #############################################################################
   2 # Name:        wxreg.test
   3 # Purpose:     Additional tests for the regex lib and wxRegEx
   4 # Author:      Mike Wetherell
   5 # Copyright:   (c) 2004 Mike Wetherell.
   6 # Licence:     wxWindows licence
   7 #############################################################################
   8
   9 #
  10 # The layout of this file is the same as reg.test. See the comments in that
  11 # file for full details. The encoding used in here is UTF-8.
  12
  13 # wx_1 tests the character classifications over the ascii range pretty
  14 # thoroughly, since hopefully these will be similar for all platforms and
  15 # locales where wxWidgets runs.
  16
  17 # wx_2 & wx_3 do some tests involving western european and cyrillic characters.
  18 # In Unicode mode, all these tests should succeed, which verifies that the
  19 # classifications aren't limited to a single 8-bit character set.
  20
  21 # In non-unicode mode wx_2 and wx_3 are skipped since they depend on the
  22 # behaviour of the runtime library's 'is' functions (isalpha, isdigit, etc.),
  23 # which are not consistent enough between implementations to allow testing.
  24 #
  25
  26 doing wx_1 "character classification: ascii"
  27 m   1   &   {[^[:alnum:]]}  "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz!" "!"
  28 m   2   &   {[[:alnum:]]}   "\a\b\t\n\v\f\r !\"#$%&'()*+,-./:;<=>?@[\\]^_`{|}~X" "X"
  29 m   3   &   {[^[:alpha:]]}  "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz!" "!"
  30 m   4   &   {[[:alpha:]]}   "\a\b\t\n\v\f\r !\"#$%&'()*+,-./0123456789:;<=>?@[\\]^_`{|}~X" "X"
  31 m   5   &   {[^[:cntrl:]]}  "\a\b\t\n\v\f\r!" "!"
  32 m   6   &   {[[:cntrl:]]}   " !\"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\\]^_`abcdefghijklmnopqrstuvwxyz{|}~\n" "\n"
  33 m   7   &   {[^[:digit:]]}  "0123456789!" "!"
  34 m   8   &   {[[:digit:]]}   "\a\b\t\n\v\f\r !\"#$%&'()*+,-./:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\\]^_`abcdefghijklmnopqrstuvwxyz{|}~ 0" "0"
  35 m   9   &   {[^[:graph:]]}  "!\"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\\]^_`abcdefghijklmnopqrstuvwxyz{|}~\n" "\n"
  36 m   10  &   {[[:graph:]]}   "\a\b\t\n\v\f\r !" "!"
  37 m   11  &   {[^[:lower:]]}  "abcdefghijklmnopqrstuvwxyz!" "!"
  38 m   12  &   {[[:lower:]]}   "\a\b\t\n\v\f\r !\"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\\]^_`{|}~x" "x"
  39 m   13  &   {[^[:print:]]}  "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz\n" "\n"
  40 m   14  &   {[[:print:]]}   "\a\b\n\v\f\rX" "X"
  41 m   15  &   {[^[:punct:]]}  "!\"#%&'()*,-./:;?@[\\]_{}X" "X"
  42 m   16  &   {[[:punct:]]}   "\a\b\t\n\v\f\r 0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz!" "!"
  43 m   17  &   {[^[:space:]]}  "\t\n\v\f\r X" "X"
  44 m   18  &   {[[:space:]]}   "\a\b!\"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\\]^_`abcdefghijklmnopqrstuvwxyz{|}~\n" "\n"
  45 m   19  &   {[^[:upper:]]}  "ABCDEFGHIJKLMNOPQRSTUVWXYZ!" "!"
  46 m   20  &   {[[:upper:]]}   "\a\b\t\n\v\f\r !\"#$%&'()*+,-./0123456789:;<=>?@[\\]^_`abcdefghijklmnopqrstuvwxyz{|}~X" "X"
  47 m   21  &   {[^[:xdigit:]]} "0123456789ABCDEFabcdef!" "!"
  48 m   22  &   {[[:xdigit:]]}  "\a\b\t\n\v\f\r !\"#$%&'()*+,-./:;<=>?@GHIJKLMNOPQRSTUVWXYZ[\\]^_`ghijklmnopqrstuvwxyz{|}~a" "a"
  49 i   23  &i  "AbCdEfGhIjKlMnOpQrStUvWxYz" "aBcDeFgHiJkLmNoPqRsTuVwXyZ" "0 25"
  50
  51 doing wx_2 "character classification: western european"
  52 m   1   &*  {[^[:alpha:]]}  "ÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖØÙÚÛÜÝÞßàáâãäåæçèéêëìíîïðñòóôõöøùúûüýþÿ!" "!"
  53 m   2   &*  {[[:alpha:]]}   " ¡¢£¤¥¦§¨©«¬®¯°±²³´¶·¸¹»¼½¾¿×÷X" "X"
  54 m   3   &*  {[^[:lower:]]}  "ßàáâãäåæçèéêëìíîïðñòóôõöøùúûüýþÿ!" "!"
  55 m   4   &*  {[[:lower:]]}   " ¡¢£¤¥¦§¨©«¬®¯°±²³´¶·¸¹»¼½¾¿ÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖ×ØÙÚÛÜÝÞ÷x" "x"
  56 m   5   &*  {[^[:upper:]]}  "ÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖØÙÚÛÜÝÞ!" "!"
  57 m   6   &*  {[[:upper:]]}   " ¡¢£¤¥¦§¨©«¬®¯°±²³´¶·¸¹»¼½¾¿×ßàáâãäåæçèéêëìíîïðñòóôõö÷øùúûüýþÿX" "X"
  58 i   7   &i* "ÀáÂãÄåÆçÈéÊëÌíÎïÐñÒóÔõÖøÙúÛüÝþ" "àÁâÃäÅæÇèÉêËìÍîÏðÑòÓôÕöØùÚûÜýÞ" "0 29"
  59
  60 doing wx_3 "character classification: cyrillic"
  61 m   1   &*  {[^[:alpha:]]}  "ёЁюабцдефгхийклмнопярстужвьызшэщчъЮАБЦДЕФГХИЙКЛМНОПЯРСТУЖВЬЫЗШЭЩЧЪ!" "!"
  62 m   2   &*  {[^[:lower:]]}  "ёюабцдефгхийклмнопярстужвьызшэщчъ!" "!"
  63 m   3   &*  {[[:lower:]]}   "ЁЮАБЦДЕФГХИЙКЛМНОПЯРСТУЖВЬЫЗШЭЩЧЪx" "x"
  64 m   4   &*  {[^[:upper:]]}  "ЁЮАБЦДЕФГХИЙКЛМНОПЯРСТУЖВЬЫЗШЭЩЧЪ!" "!"
  65 m   5   &*  {[[:upper:]]}   "ёюабцдефгхийклмнопярстужвьызшэщчъX" "X"
  66 i   6   &i* "ЁюАбЦдЕфГхИйКлМнОпЯрСтУжВьЫзШэЩчЪ" "ёЮаБцДеФгХиЙкЛмНоПяРсТуЖвЬыЗшЭщЧъ" "0 32"
  67
  68 #doing bugs "known bugs"
  69 #m  1    -   {(\w+).*?(\d\d:\d\d)} "from 10:30 until 12:00" "from" "10:30"
  70