USING: accessors arrays assocs continuations http.client kernel
literals math math.parser ranges pcre pcre.ffi pcre.private
random sequences system tools.test ;
QUALIFIED: regexp
QUALIFIED: splitting
IN: pcre.tests

{ { "Bords" "words" "word" } } [
    "Bords, words, word." { ", " ", " "." } split-subseqs
] unit-test

{ { { 3 "day" } { 2 "month" } { 1 "year" } } } [
    "(?P<year>\\d{4})-(?P<month>\\d{2})-(?P<day>\\d{2})"
    <compiled-pcre> nametable>>
] unit-test

CONSTANT: iso-date "(?P<year>\\d{4})-(?P<month>\\d{2})-(?P<day>\\d{2})"

! On windows the erroffset appears to be set to 0 despite there being
! nothing wrong with the regexp.
{ t } [
    "foo" (pcre) 3array rest { { f -1 } { f 0 } } member?
] unit-test

{ { 1 2 3 } } [
    iso-date <pcre>
    { "year" "month" "day" } [ pcre_get_stringnumber ] with map
] unit-test

{ t } [
    "foo" <compiled-pcre> PCRE_UTF8 has-option?
] unit-test

! This option is not present on old PCRE versions.
{ t } [
    "foo" <compiled-pcre> version 8.10 >
    [ PCRE_UCP has-option? ] [ PCRE_UCP has-option? not ] if
] unit-test

os unix? [ [ 10 ] [ PCRE_CONFIG_NEWLINE pcre-config ] unit-test ] when

! In this day and age, not supporting utf-8 is broken.
{ 1 } [ PCRE_CONFIG_UTF8 pcre-config ] unit-test

{ 1 } [ PCRE_CONFIG_UNICODE_PROPERTIES pcre-config ] unit-test

! Ok if these options throw if the pcre library is to old to support
! these configuration parameters.
{ t } [
    [ PCRE_CONFIG_UTF16 pcre-config ] [ what>> ] recover
    { 0 $ PCRE_CONFIG_UTF16 } member?
] unit-test
{ t } [
    [ PCRE_CONFIG_UTF32 pcre-config ] [ what>> ] recover
    { 0 $ PCRE_CONFIG_UTF32 } member?
] unit-test

{ 33 }
[
    [ "foo" <pcre> f 33 pcre-fullinfo ] [ what>> ] recover
] unit-test

! Tests for findall
{
    { { f "1999-01-12" } { "year" "1999" } { "month" "01" } { "day" "12" } }
} [
    "1999-01-12" iso-date <compiled-pcre> findall first
] unit-test

{ 3 } [
    "2003-10-09 1999-09-01 1514-10-20" iso-date <compiled-pcre> findall length
] unit-test

{ 5 } [ "abcdef" "[a-e]" findall length ] unit-test

{ 3 } [ "foo bar baz" "foo|bar|baz" findall length ] unit-test

{ 3 } [ "örjan är åtta" "[åäö]" findall length ] unit-test

{ 3 } [ "ÅÄÖ" "\\p{Lu}" findall length ] unit-test

{ 3 } [ "foobar" "foo(?=bar)" findall first first second length ] unit-test

{ { { { f ", " } } { { f ", " } } { { f "." } } } } [
    "Words, words, word." "\\W+" findall
] unit-test

{ { ", " ", " "." } } [
    "Words, words, word." "\\W+" findall [ first second ] map
] unit-test

: long-string ( -- x )
    10000 [ CHAR: a CHAR: z [a..b] random ] "" replicate-as ;

! Performance
{ 0 } [ long-string ".{0,15}foobar.{0,10}" findall length ] unit-test

! Empty matches, corner case behavior is copied from pcredemo.c
{ { { { f "foo" } } { { f "" } } } }
[ "foo" ".*" findall ] unit-test

{ { { { f "" } } { { f "" } } { { f "" } } } }
[ "foo" "B*" findall ] unit-test

! Empty matches in strings with multi-byte characters are tricky.
{ { { { f "" } } { { f "" } } { { f "" } } { { f "" } } } }
[ "öööö" "x*" findall ] unit-test

! Tests for matches?
{ t } [ "örjan" "örjan" matches? ] unit-test

{ t } [ "abcö" "\\p{Ll}{4}" matches? ] unit-test

! This used to work in 8.36, but might have changed in later versions.
! See: https://bugs.exim.org/show_bug.cgi?id=1875
version 8.36 <= [
    { t t } [
        "(?s)." <compiled-pcre> PCRE_DOTALL has-option?
        "(?i)x" <compiled-pcre> PCRE_CASELESS has-option?
    ] unit-test
] when

{ f } [ "\n" "." matches? ] unit-test
{ t } [ "\n" "(?s)." matches? ] unit-test

{ f t } [
    "hello\nthere" "^.*$" <compiled-pcre> matches?
    "hello\nthere" "(?s)^.*$" <compiled-pcre> matches?
] unit-test

! Modes off by default
{ f f } [
    ! Caseless mode
    "x" <compiled-pcre> PCRE_CASELESS has-option?
    ! Dotall mode
    "." <compiled-pcre> PCRE_DOTALL has-option?
] unit-test

! Backreferences
{ { t f } } [
    { "response and responsibility" "sense and responsibility" }
    [ "(sens|respons)e and \\1ibility" matches? ] map
] unit-test

{ { t t f } } [
    { "rah rah" "RAH RAH" "RAH rah" } [ "((?i)rah)\\s+\\1" matches? ] map
] unit-test

! Splitting
{ { { "Words" "words" "word" } { "Words" "words" "word" } } } [
    "Words, words, word." { "\\W+" "[,. ]" } [ split ] with map
] unit-test

! Bigger tests
{ t } [
    "https://factorcode.org/" http-get nip
    "href=\"(?P<link>[^\"]+)\"" findall [ "link" of ] map sequence?
] unit-test

! Test that the regexp syntax works.
{ t } [ "1234abcd" regexp:R/ ^\d+\w+$/ matches? ] unit-test