unicode: update to 15.0.0

[factor.git] / basis / unicode / collation / collation.factor

diff --git a/basis/unicode/collation/collation.factor b/basis/unicode/collation/collation.factor
index b6c4941b9d7fcb4315d24f338795a6271f83e631..f29adc50eb05ece45a99a415676f636aec0f481d 100644 (file)
--- a/basis/unicode/collation/collation.factor
+++ b/basis/unicode/collation/collation.factor
@@ -1,7 +1,7 @@
 ! Copyright (C) 2008 Daniel Ehrenberg.
 ! See http://factorcode.org/license.txt for BSD license.
 USING: accessors arrays assocs combinators
-combinators.short-circuit combinators.smart fry kernel locals make
+combinators.short-circuit combinators.smart fry kernel make
 math math.order math.parser namespaces sequences
 simple-flat-file splitting strings unicode.data ;
 IN: unicode.collation
@@ -25,7 +25,7 @@ TUPLE: weight-levels primary secondary tertiary ignorable? ;
     ] map ;
 
 : parse-keys ( string -- chars )
-    " " split [ hex> ] "" map-as ;
+    split-words [ hex> ] "" map-as ;
 
 : parse-ducet ( file -- ducet )
     load-data-file [ [ parse-keys ] [ parse-weight ] bi* ] H{ } assoc-map-as ;
@@ -203,23 +203,32 @@ TUPLE: weight-levels primary secondary tertiary ignorable? ;
 fixup-ducet-for-tibetan
 
 : tangut-block? ( char -- ? )
-    ! Tangut Block, Tangut Components Block
-    { [ 0x17000 0x187FF between? ] [ 0x18800 0x18AFF between? ] } 1|| ; inline
+    {
+        [ 0x17000 0x18AFF between? ] ! Tangut and Tangut Components
+        [ 0x18D00 0x18D8F between? ] ! Tangut Supplement
+    } 1|| ; inline
 
 : nushu-block? ( char -- ? )
-    0x1b170 0x1B2FB between? ; inline
+    0x1b170 0x1B2FF between? ; inline
+
+: khitan-block? ( char -- ? )
+    0x18b00 0x18cff between? ; inline
 
 ! https://wiki.computercraft.cc/Module:Unicode_data
 ! Unicode TR10 - Computing Implicit Weights
 : base ( char -- base )
     {
-        { [ dup 0x03400 0x04DB5 between? ] [ drop 0xFB80 ] } ! Extension A
-        { [ dup 0x20000 0x2A6D6 between? ] [ drop 0xFB80 ] } ! Extension B
-        { [ dup 0x2A700 0x2B734 between? ] [ drop 0xFB80 ] } ! Extension C
+        { [ dup 0x03400 0x04DBF between? ] [ drop 0xFB80 ] } ! Extension A
+        { [ dup 0x20000 0x2A6DF between? ] [ drop 0xFB80 ] } ! Extension B
+        { [ dup 0x2A700 0x2B739 between? ] [ drop 0xFB80 ] } ! Extension C
         { [ dup 0x2B740 0x2B81D between? ] [ drop 0xFB80 ] } ! Extension D
         { [ dup 0x2B820 0x2CEA1 between? ] [ drop 0xFB80 ] } ! Extension E
         { [ dup 0x2CEB0 0x2EBE0 between? ] [ drop 0xFB80 ] } ! Extension F
-        { [ dup 0x04E00 0x09FEF between? ] [ drop 0xFB40 ] } ! CJK
+        { [ dup 0x30000 0x3134A between? ] [ drop 0xFB80 ] } ! Extension G
+        { [ dup 0x31350 0x323AF between? ] [ drop 0xFB80 ] } ! Extension H
+        { [ dup 0x2F800 0x2FA1D between? ] [ drop 0xFB80 ] } ! CJK Compatibility
+        { [ dup 0x04E00 0x09FFF between? ] [ drop 0xFB40 ] } ! CJK
+        { [ dup 0x0F900 0x0FAD9 between? ] [ drop 0xFB40 ] } ! CJK
         [ drop 0xFBC0 ] ! Other
     } cond ;
 
@@ -235,6 +244,12 @@ fixup-ducet-for-tibetan
 : nushu-BBBB ( char -- weight-levels )
     0x1B170 - 0x8000 bitor 0 0 <weight-levels> ; inline
 
+: khitan-AAAA ( char -- weight-levels )
+    drop 0xfb02 0x0020 0x0002 <weight-levels> ; inline
+
+: khitan-BBBB ( char -- weight-levels )
+    0x18b00 - 0x8000 bitor 0 0 <weight-levels> ; inline
+
 : AAAA ( char -- weight-levels )
     [ base ] [ -15 shift ] bi + 0x0020 0x0002 <weight-levels> ; inline
 
@@ -246,6 +261,7 @@ fixup-ducet-for-tibetan
     {
         { [ dup tangut-block? ] [ [ tangut-AAAA ] [ tangut-BBBB ] bi 2array ] }
         { [ dup nushu-block? ] [ [ nushu-AAAA ] [ nushu-BBBB ] bi 2array ] }
+        { [ dup khitan-block? ] [ [ khitan-AAAA ] [ khitan-BBBB ] bi 2array ] }
         [ [ AAAA ] [ BBBB ] bi 2array ]
     } cond ;
 
@@ -300,7 +316,7 @@ fixup-ducet-for-tibetan
             [ [ primary>> ] append-weights { 0 } ]
             [ [ secondary>> ] append-weights { 0 } ]
             [ [ tertiary>> ] append-weights { 0 } ]
-            [ [ [ secondary>> ] [ tertiary>> ] bi [ zero? ] bi@ and not ] filter [ variable-weight ] map ]
+            [ [ [ secondary>> ] [ tertiary>> ] bi [ zero? ] both? ] reject [ variable-weight ] map ]
         } cleave
     ] { } append-outputs-as ;