extra/pcre/pcre-tests.factor

   1 USING:
   2     accessors
   3     arrays
   4     assocs
   5     http.client
   6     kernel
   7     math math.ranges
   8     pcre pcre.ffi pcre.info
   9     random
  10     sequences
  11     system
  12     tools.test ;
  13 QUALIFIED: regexp
  14 IN: pcre.tests
  15
  16 CONSTANT: iso-date "(?P<year>\\d{4})-(?P<month>\\d{2})-(?P<day>\\d{2})"
  17
  18 ! On windows the erroffset appears to be set to 0 despite there being
  19 ! nothing wrong with the regexp.
  20 [ t ] [
  21     "foo" (pcre) 3array 1 tail { { f -1 } { f 0 } } member?
  22 ] unit-test
  23
  24 [ { 1 2 3 } ] [
  25     iso-date <pcre>
  26     { "year" "month" "day" } [ pcre_get_stringnumber ] with map
  27 ] unit-test
  28
  29 [ t ] [ "foo" <compiled-pcre> PCRE_UTF8 has-option? ] unit-test
  30
  31 os unix? [ [ 10 ] [ PCRE_CONFIG_NEWLINE config ] unit-test ] when
  32
  33 ! In this day and age, not supporting utf-8 is broken.
  34 [ 1 ] [ PCRE_CONFIG_UTF8 config ] unit-test
  35
  36 [ 1 ] [ PCRE_CONFIG_UNICODE_PROPERTIES config ] unit-test
  37
  38 ! libpcre must not support 16 or 32 bit code points.
  39 [ 0 ] [ PCRE_CONFIG_UTF16 config ] unit-test
  40 [ 0 ] [ PCRE_CONFIG_UTF32 config ] unit-test
  41
  42 ! Tests for findall
  43 [
  44     { { f "1999-01-12" } { "year" "1999" } { "month" "01" } { "day" "12" } }
  45 ] [
  46     "1999-01-12" iso-date <compiled-pcre> findall first
  47 ] unit-test
  48
  49 [ 3 ] [
  50     "2003-10-09 1999-09-01 1514-10-20" iso-date <compiled-pcre> findall length
  51 ] unit-test
  52
  53 [ 5 ] [ "abcdef" "[a-e]" findall length ] unit-test
  54
  55 [ 3 ] [ "foo bar baz" "foo|bar|baz" findall length ] unit-test
  56
  57 [ 3 ] [ "örjan är åtta" "[åäö]" findall length ] unit-test
  58
  59 [ 3 ] [ "ÅÄÖ" "\\p{Lu}" findall length ] unit-test
  60
  61 [ 3 ] [ "foobar" "foo(?=bar)" findall first first second length ] unit-test
  62
  63 [ { { { f ", " } } { { f ", " } } { { f "." } } } ] [
  64     "Words, words, word." "\\W+" findall
  65 ] unit-test
  66
  67 [ { ", " ", " "." } ] [
  68     "Words, words, word." "\\W+" findall [ first second ] map
  69 ] unit-test
  70
  71 : long-string ( -- x )
  72     10000 [ CHAR: a CHAR: z [a,b] random ] "" replicate-as ;
  73
  74 ! Performance
  75 [ 0 ] [ long-string ".{0,15}foobar.{0,10}" findall length ] unit-test
  76
  77 ! Empty matches, corner case behaviour is copied from pcredemo.c
  78 [ { { { f "foo" } } { { f "" } } } ]
  79 [ "foo" ".*" findall ] unit-test
  80
  81 [ { { { f "" } } { { f "" } } { { f "" } } } ]
  82 [ "foo" "B*" findall ] unit-test
  83
  84 ! Empty matches in strings with multi-byte characters are tricky.
  85 [ { { { f "" } } { { f "" } } { { f "" } } { { f "" } } } ]
  86 [ "öööö" "x*" findall ] unit-test
  87
  88 ! Tests for matches?
  89 [ t ] [ "örjan" "örjan" matches? ] unit-test
  90
  91 [ t ] [ "abcö" "\\p{Ll}{4}" matches? ] unit-test
  92
  93 ! Dotall mode, off by default
  94 [ f ] [ "." <compiled-pcre> PCRE_DOTALL has-option? ] unit-test
  95 [ t ] [ "(?s)." <compiled-pcre> PCRE_DOTALL has-option? ] unit-test
  96
  97 [ f ] [ "\n" "." matches? ] unit-test
  98 [ t ] [ "\n" "(?s)." matches? ] unit-test
  99
 100 ! Caseless mode, off by default
 101 [ { f t } ] [
 102     { "x" "(?i)x" } [ <compiled-pcre> PCRE_CASELESS has-option? ] map
 103 ] unit-test
 104
 105 ! Backreferences
 106 [ { t f } ] [
 107     { "response and responsibility" "sense and responsibility" }
 108     [ "(sens|respons)e and \\1ibility" matches? ] map
 109 ] unit-test
 110
 111 [ { t t f } ] [
 112     { "rah rah" "RAH RAH" "RAH rah" } [ "((?i)rah)\\s+\\1" matches? ] map
 113 ] unit-test
 114
 115 ! Splitting
 116 [ { { "Words" "words" "word" } { "Words" "words" "word" } } ] [
 117     "Words, words, word." { "\\W+" "[,. ]" } [ split ] with map
 118 ] unit-test
 119
 120 ! Bigger tests
 121 [ t ] [
 122     "http://factorcode.org/" http-get nip
 123     "href=\"(?P<link>[^\"]+)\"" findall [ "link" of ] map sequence?
 124 ] unit-test
 125
 126 ! Test that the regexp syntax works.
 127 [ t ] [ "1234abcd" regexp:R[ ^\d+\w+$] matches? ] unit-test