lib9: update to Unicode 6.0.0

R=r
CC=golang-dev
https://golang.org/cl/4121042

src/lib9/utf/Makefile

@@ -14,3 +14,19 @@ OFILES=\
 	mkrunetype.$O\
 
 include ../../Make.ccmd
+
+UnicodeData-%.txt:
+	curl http://www.unicode.org/Public/$*/ucd/UnicodeData.txt >_$@
+	mv _$@ $@
+
+runetypebody-%.c: mkrunetype UnicodeData-%.txt
+	mkrunetype -p UnicodeData-$*.txt >_$@
+	mv _$@ $@
+
+CLEANFILES+=UnicodeData.txt
+
+UNICODE_VERSION=6.0.0
+
+test: mkrunetype UnicodeData-$(UNICODE_VERSION).txt
+	mkrunetype -c UnicodeData-$(UNICODE_VERSION).txt
+

src/lib9/utf/mkrunetype.c

@@ -93,8 +93,9 @@ usage(void)
 	exit(1);
 }
 
-int
-main(int argc, char *argv[]){
+void
+main(int argc, char *argv[])
+{
 	FILE *in;
 	char buf[MAX_LINE], buf2[MAX_LINE];
 	char *fields[NFIELDS + 1], *fields2[NFIELDS + 1];
@@ -239,7 +240,7 @@ main(int argc, char *argv[]){
 	}else{
 		mktables(argv[0], usepairs);
 	}
-	return 0;
+	exit(0);
 }
 
 /*
@@ -562,7 +563,8 @@ mkto(const char* label, int* map, int usepairs)
 
 // Make only range tables and a function for is<label>rune.
 static void
-mkisronly(const char* label, char* prop) {
+mkisronly(const char* label, char* prop)
+{
 	mkisrange(label, prop, 1);
 	printf(
 		"int\n"

src/lib9/utf/runetype.c

@@ -35,49 +35,4 @@ rbsearch(Rune c, Rune *t, int n, int ne)
 	return 0;
 }
 
-/*
- * The "ideographic" property is hard to extract from UnicodeData.txt,
- * so it is hard coded here.
- *
- * It is defined in the Unicode PropList.txt file, for example
- * PropList-3.0.0.txt.  Unlike the UnicodeData.txt file, the format of
- * PropList changes between versions.  This property appears relatively static;
- * it is the same in version 4.0.1, except that version defines some >16 bit
- * chars as ideographic as well: 20000..2a6d6, and 2f800..2Fa1d.
- */
-static Rune __isideographicr[] = {
-	0x3006, 0x3007,		/* 0x3006 added in 2.0.14 */
-	0x3021, 0x3029,
-	0x3038, 0x303a,		/* added in 3.0.0 */
-	0x3400, 0x4db5,		/* added in 3.0.0 */
-
-	/* consecutive */
-	0x4e00, 0x9fa5,
-	0x9fa6, 0x9fbb,		/* added in 4.1.0 */
-	0x9fbc, 0x9fc3,		/* added in 5.1.0 */
-	0x9fc4, 0x9fcb,		/* added in 5.2.0 */
-
-	0xf900, 0xfa2d,
-
-	/* consecutive */
-	0xfa30, 0xfa6a,		/* added in 5.1.0 */
-	0xfa6b, 0xfa6d,		/* added in 5.2.0 */
-
-	0xfa70, 0xfad9,		/* added in 4.1.0 */
-	0x20000, 0x2a6d6,	/* added in 3.1.0 */
-	0x2a700, 0x2b734,	/* added in 5.2.0 */
-	0x2f800, 0x2fa1d,	/* added in 3.1.0 */
-};
-
-int
-isideographicrune(Rune c)
-{
-	Rune *p;
-
-	p = rbsearch(c, __isideographicr, nelem(__isideographicr)/2, 2);
-	if(p && c >= p[0] && c <= p[1])
-		return 1;
-	return 0;
-}
-
-#include "runetypebody-5.2.0.c"
+#include "runetypebody-6.0.0.c"

src/lib9/utf/utf.h

@@ -224,12 +224,6 @@ int isalpharune(Rune r);
 int isdigitrune(Rune r);
 
 
-// isideographicrune tests for ideographic characters and numbers, as
-// defined by the Unicode standard.
-
-int isideographicrune(Rune r);
-
-
 // isspacerune tests for whitespace characters, including "C" locale
 // whitespace, Unicode defined whitespace, and the "zero-width
 // non-break space" character.