! See http://factorcode.org/license.txt for BSD license.
USING: combinators.short-circuit assocs math kernel sequences
io.files hashtables quotations splitting grouping arrays io
-math.parser hash2 math.order byte-arrays namespaces
+math.parser math.order byte-arrays namespaces math.bitwise
compiler.units parser io.encodings.ascii values interval-maps
ascii sets combinators locals math.ranges sorting make
strings.parser io.encodings.utf8 memoize simple-flat-file ;
+FROM: namespaces => set ;
IN: unicode.data
<PRIVATE
VALUE: special-casing
VALUE: properties
+: >2ch ( a b -- c ) [ 21 shift ] dip + ;
+: 2ch> ( c -- a b ) [ -21 shift ] [ 21 on-bits mask ] bi ;
+
PRIVATE>
VALUE: name-map
: canonical-entry ( char -- seq ) canonical-map at ; inline
-: combine-chars ( a b -- char/f ) combine-map hash2 ; inline
+: combine-chars ( a b -- char/f ) >2ch combine-map at ; inline
: compatibility-entry ( char -- seq ) compatibility-map at ; inline
: combining-class ( char -- n ) class-map at ; inline
: non-starter? ( char -- ? ) combining-class { 0 f } member? not ; inline
: name>char ( name -- char ) name-map at ; inline
: char>name ( char -- name ) name-map value-at ; inline
: property? ( char property -- ? ) properties at interval-key? ; inline
-: ch>lower ( ch -- lower ) simple-lower at-default ; inline
-: ch>upper ( ch -- upper ) simple-upper at-default ; inline
-: ch>title ( ch -- title ) simple-title at-default ; inline
+: ch>lower ( ch -- lower ) simple-lower ?at drop ; inline
+: ch>upper ( ch -- upper ) simple-upper ?at drop ; inline
+: ch>title ( ch -- title ) simple-title ?at drop ; inline
: special-case ( ch -- casing-tuple ) special-casing at ; inline
! For non-existent characters, use Cn
MEMO: categories-map ( -- hashtable )
categories <enum> [ swap ] H{ } assoc-map-as ;
-CONSTANT: num-chars HEX: 2FA1E
+CONSTANT: num-chars 0x2FA1E
PRIVATE>
! Cf = 26; Mn = 5; Cn = 29
! Use a compressed array instead?
dup category-map ?nth [ ] [
- dup HEX: E0001 HEX: E007F between?
+ dup 0xE0001 0xE007F between?
[ drop 26 ] [
- HEX: E0100 HEX: E01EF between? 5 29 ?
+ 0xE0100 0xE01EF between? 5 29 ?
] if
] ?if ;
[ [ hex> ] dip ] assoc-map ;
: process-data ( index data -- hash )
- (process-data) [ hex> ] assoc-map [ nip ] assoc-filter >hashtable ;
+ (process-data) [ hex> ] assoc-map [ nip ] H{ } assoc-filter-as ;
: (chain-decomposed) ( hash value -- newvalue )
[
: exclusions ( -- set )
exclusions-file utf8 file-lines
- [ "#" split1 drop [ blank? ] trim-tail hex> ] map harvest ;
+ [ "#" split1 drop [ blank? ] trim-tail hex> ] map
+ [ 0 = not ] filter ;
: remove-exclusions ( alist -- alist )
- exclusions [ dup ] H{ } map>assoc assoc-diff ;
+ exclusions unique assoc-diff ;
-: process-canonical ( data -- hash2 hash )
+: process-canonical ( data -- hash hash )
(process-decomposed) [ first* ] filter
[
[ second length 2 = ] filter remove-exclusions
- ! using 1009 as the size, the maximum load is 4
- [ first2 first2 rot 3array ] map 1009 alist>hash2
+ [ first2 >2ch swap ] H{ } assoc-map-as
] [ >hashtable chain-decomposed ] bi ;
: process-compatibility ( data -- hash )
2dup bounds-check? [ set-nth ] [ 3drop ] if ;
:: fill-ranges ( table -- table )
- name-map >alist sort-values keys
+ name-map sort-values keys
[ { [ "first>" tail? ] [ "last>" tail? ] } 1|| ] filter
2 group [
[ name>char ] bi@ [ [a,b] ] [ table ?nth ] bi
] assoc-map ;
: properties>intervals ( properties -- assoc[str,interval] )
- dup values prune [ f ] H{ } map>assoc
+ dup values members [ f ] H{ } map>assoc
[ [ push-at ] curry assoc-each ] keep
[ <interval-set> ] assoc-map ;
[ [ set-code-point ] each ] H{ } make-assoc ;
load-data {
- [ process-names to: name-map ]
- [ 13 swap process-data to: simple-lower ]
- [ 12 swap process-data to: simple-upper ]
- [ 14 swap process-data simple-upper assoc-union to: simple-title ]
- [ process-combining to: class-map ]
- [ process-canonical to: canonical-map to: combine-map ]
- [ process-compatibility to: compatibility-map ]
- [ process-category to: category-map ]
+ [ process-names \ name-map set-value ]
+ [ 13 swap process-data \ simple-lower set-value ]
+ [ 12 swap process-data \ simple-upper set-value ]
+ [ 14 swap process-data simple-upper assoc-union \ simple-title set-value ]
+ [ process-combining \ class-map set-value ]
+ [ process-canonical \ canonical-map set-value \ combine-map set-value ]
+ [ process-compatibility \ compatibility-map set-value ]
+ [ process-category \ category-map set-value ]
} cleave
: postprocess-class ( -- )
- combine-map [ [ second ] map ] map concat
+ combine-map keys [ 2ch> nip ] map
[ combining-class not ] filter
[ 0 swap class-map set-at ] each ;
postprocess-class
-load-special-casing to: special-casing
+load-special-casing \ special-casing set-value
-load-properties to: properties
+load-properties \ properties set-value
[ name>char [ "Invalid character" throw ] unless* ]
name>char-hook set-global