extra/pcre/pcre-tests.factor

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