1 USING: arrays assocs combinators.short-circuit grouping
2 http.download io.encodings.utf8 io.files io.files.temp kernel
3 math math.parser sequences splitting splitting.extras strings
4 tools.test unicode unicode.normalize.private ;
5 IN: unicode.normalize.tests
7 { "ab\u000323\u000302cd" } [ "ab\u000302" "\u000323cd" string-append ] unit-test
9 { "ab\u00064b\u000347\u00034e\u00034d\u000346" }
10 [ "ab\u000346\u000347\u00064b\u00034e\u00034d" dup reorder ] unit-test
12 { "hello" "hello" } [ "hello" [ nfd ] keep nfkd ] unit-test
14 { "\u00FB012\u002075\u00017F\u000323\u000307" "fi25s\u000323\u000307" }
15 [ "\u00FB012\u002075\u001E9B\u000323" [ nfd ] keep nfkd ] unit-test
17 { "\u001E69" "s\u000323\u000307" } [ "\u001E69" [ nfc ] keep nfd ] unit-test
18 { "\u001E0D\u000307" } [ "\u001E0B\u000323" nfc ] unit-test
20 { 54620 } [ 4370 4449 4523 jamo>hangul ] unit-test
21 { 4370 4449 4523 } [ 54620 hangul>jamo first3 ] unit-test
22 { t } [ 54620 hangul? ] unit-test
23 { f } [ 0 hangul? ] unit-test
24 { "\u001112\u001161\u0011ab" } [ "\u00d55c" nfd ] unit-test
25 { "\u00d55c" } [ "\u001112\u001161\u0011ab" nfc ] unit-test
27 ! Could use simple-flat-file after some cleanup
28 : parse-normalization-tests ( -- tests )
29 "https://downloads.factorcode.org/misc/UCD/15.1.0/NormalizationTest.txt"
30 "NormalizationTest-15.1.0.txt" cache-file download-once-to
31 utf8 file-lines [ "#" head? ] reject
32 [ "@" head? ] split*-when
33 2 <groups> [ first2 [ first ] dip 2array ] map
36 "#@" split first [ CHAR: \s = ] trim-tail ";" split harvest
37 [ split-words [ hex> ] "" map-as ] map
41 :: check-normalization-test? ( test spec quot -- ? )
44 [ 1 - test nth ] bi@ quot call( str -- str' ) =
49 parse-normalization-tests [
51 [ { { 2 { 1 2 3 } } { 4 { 4 5 } } } [ nfc ] check-normalization-test? ]
52 [ { { 3 { 1 2 3 } } { 5 { 4 5 } } } [ nfd ] check-normalization-test? ]
53 [ { { 4 { 1 2 3 4 5 } } } [ nfkc ] check-normalization-test? ]
54 [ { { 5 { 1 2 3 4 5 } } } [ nfkd ] check-normalization-test? ]
59 { { 4018 820 3953 3968 } }
60 [ { 3958 3953 820 } >string nfd >array ] unit-test