Allow symbol characters for identifier names.

othersym.patch.diff

diff -r 5bf1a8b3aeea src/cmd/gc/lex.c
--- a/src/cmd/gc/lex.c	Thu May 29 13:47:31 2014 -0400
+++ b/src/cmd/gc/lex.c	Tue Jun 03 12:11:38 2014 +0900
@@ -1309,7 +1313,7 @@
 			ungetc(c);
 			rune = getr();
 			// 0xb7 · is used for internal names
-			if(!isalpharune(rune) && !isdigitrune(rune) && (importpkg == nil || rune != 0xb7))
+			if(!isalpharune(rune) && !isdigitrune(rune) && !isothersymrune(rune) && (importpkg == nil || rune != 0xb7))
 				yyerror("invalid identifier character U+%04x", rune);
 			cp += runetochar(cp, &rune);
 		} else if(!yy_isalnum(c) && c != '_')
diff -r 5bf1a8b3aeea src/lib9/utf/mkrunetype.c
--- a/src/lib9/utf/mkrunetype.c	Thu May 29 13:47:31 2014 -0400
+++ b/src/lib9/utf/mkrunetype.c	Tue Jun 03 12:11:38 2014 +0900
@@ -76,6 +76,7 @@
 static char	myisupper[NRUNES];
 static char	myislower[NRUNES];
 static char	myistitle[NRUNES];
+static char	myisothersym[NRUNES];
 
 static int	mytoupper[NRUNES];
 static int	mytolower[NRUNES];
@@ -192,6 +193,9 @@
 			if(strcmp(p, "Nd") == 0)
 				myisdigit[code] = 1;
 
+			if(strcmp(p, "So") == 0)
+				myisothersym[code] = 1;
+
 			/*
 			 * when finding conversions, also need to mark
 			 * upper/lower case, since some chars, like
@@ -603,6 +607,7 @@
 	mkis("upper", myisupper, usepairs);
 	mkis("lower", myislower, usepairs);
 	mkis("title", myistitle, usepairs);
+	mkis("othersym", myisothersym, usepairs);
 
 	mkto("upper", mytoupper, usepairs);
 	mkto("lower", mytolower, usepairs);
diff -r 5bf1a8b3aeea src/lib9/utf/runetypebody-6.3.0.h
--- a/src/lib9/utf/runetypebody-6.3.0.h	Thu May 29 13:47:31 2014 -0400
+++ b/src/lib9/utf/runetypebody-6.3.0.h	Tue Jun 03 12:11:38 2014 +0900
@@ -1076,6 +1076,180 @@
 	return 0;
 }
 
+static Rune __isothersymr[] = {
+	0x060e, 0x060f,
+	0x06fd, 0x06fe,
+	0x0bf3, 0x0bf8,
+	0x0f01, 0x0f03,
+	0x0f15, 0x0f17,
+	0x0f1a, 0x0f1f,
+	0x0fbe, 0x0fc5,
+	0x0fc7, 0x0fcc,
+	0x0fce, 0x0fcf,
+	0x0fd5, 0x0fd8,
+	0x109e, 0x109f,
+	0x1390, 0x1399,
+	0x19de, 0x19ff,
+	0x1b61, 0x1b6a,
+	0x1b74, 0x1b7c,
+	0x2100, 0x2101,
+	0x2103, 0x2106,
+	0x2108, 0x2109,
+	0x2116, 0x2117,
+	0x211e, 0x2123,
+	0x213a, 0x213b,
+	0x214c, 0x214d,
+	0x2195, 0x2199,
+	0x219c, 0x219f,
+	0x21a1, 0x21a2,
+	0x21a4, 0x21a5,
+	0x21a7, 0x21ad,
+	0x21af, 0x21cd,
+	0x21d0, 0x21d1,
+	0x21d5, 0x21f3,
+	0x2300, 0x2307,
+	0x230c, 0x231f,
+	0x2322, 0x2328,
+	0x232b, 0x237b,
+	0x237d, 0x239a,
+	0x23b4, 0x23db,
+	0x23e2, 0x23f3,
+	0x2400, 0x2426,
+	0x2440, 0x244a,
+	0x249c, 0x24e9,
+	0x2500, 0x25b6,
+	0x25b8, 0x25c0,
+	0x25c2, 0x25f7,
+	0x2600, 0x266e,
+	0x2670, 0x26ff,
+	0x2701, 0x2767,
+	0x2794, 0x27bf,
+	0x2800, 0x28ff,
+	0x2b00, 0x2b2f,
+	0x2b45, 0x2b46,
+	0x2b50, 0x2b59,
+	0x2ce5, 0x2cea,
+	0x2e80, 0x2e99,
+	0x2e9b, 0x2ef3,
+	0x2f00, 0x2fd5,
+	0x2ff0, 0x2ffb,
+	0x3012, 0x3013,
+	0x3036, 0x3037,
+	0x303e, 0x303f,
+	0x3190, 0x3191,
+	0x3196, 0x319f,
+	0x31c0, 0x31e3,
+	0x3200, 0x321e,
+	0x322a, 0x3247,
+	0x3260, 0x327f,
+	0x328a, 0x32b0,
+	0x32c0, 0x32fe,
+	0x3300, 0x33ff,
+	0x4dc0, 0x4dff,
+	0xa490, 0xa4c6,
+	0xa828, 0xa82b,
+	0xa836, 0xa837,
+	0xaa77, 0xaa79,
+	0xffed, 0xffee,
+	0xfffc, 0xfffd,
+	0x10137, 0x1013f,
+	0x10179, 0x10189,
+	0x10190, 0x1019b,
+	0x101d0, 0x101fc,
+	0x1d000, 0x1d0f5,
+	0x1d100, 0x1d126,
+	0x1d129, 0x1d164,
+	0x1d16a, 0x1d16c,
+	0x1d183, 0x1d184,
+	0x1d18c, 0x1d1a9,
+	0x1d1ae, 0x1d1dd,
+	0x1d200, 0x1d241,
+	0x1d300, 0x1d356,
+	0x1f000, 0x1f02b,
+	0x1f030, 0x1f093,
+	0x1f0a0, 0x1f0ae,
+	0x1f0b1, 0x1f0be,
+	0x1f0c1, 0x1f0cf,
+	0x1f0d1, 0x1f0df,
+	0x1f110, 0x1f12e,
+	0x1f130, 0x1f16b,
+	0x1f170, 0x1f19a,
+	0x1f1e6, 0x1f202,
+	0x1f210, 0x1f23a,
+	0x1f240, 0x1f248,
+	0x1f250, 0x1f251,
+	0x1f300, 0x1f320,
+	0x1f330, 0x1f335,
+	0x1f337, 0x1f37c,
+	0x1f380, 0x1f393,
+	0x1f3a0, 0x1f3c4,
+	0x1f3c6, 0x1f3ca,
+	0x1f3e0, 0x1f3f0,
+	0x1f400, 0x1f43e,
+	0x1f442, 0x1f4f7,
+	0x1f4f9, 0x1f4fc,
+	0x1f500, 0x1f53d,
+	0x1f540, 0x1f543,
+	0x1f550, 0x1f567,
+	0x1f5fb, 0x1f640,
+	0x1f645, 0x1f64f,
+	0x1f680, 0x1f6c5,
+	0x1f700, 0x1f773,
+};
+
+static Rune __isothersymp[] = {
+	0x00ae, 0x00b0,
+	0x0f34, 0x0f38,
+	0x2125, 0x2129,
+};
+
+static Rune __isothersyms[] = {
+	0x00a6,
+	0x00a9,
+	0x0482,
+	0x06de,
+	0x06e9,
+	0x07f6,
+	0x09fa,
+	0x0b70,
+	0x0bfa,
+	0x0c7f,
+	0x0d79,
+	0x0f13,
+	0x1940,
+	0x2114,
+	0x212e,
+	0x214a,
+	0x214f,
+	0x21d3,
+	0x3004,
+	0x3020,
+	0x3250,
+	0xa839,
+	0xfdfd,
+	0xffe4,
+	0xffe8,
+	0x1d245,
+	0x1f440,
+};
+
+int
+isothersymrune(Rune c)
+{
+	Rune *p;
+
+	p = rbsearch(c, __isothersymr, nelem(__isothersymr)/2, 2);
+	if(p && c >= p[0] && c <= p[1])
+		return 1;
+	p = rbsearch(c, __isothersymp, nelem(__isothersymp)/2, 2);
+	if(p && c >= p[0] && c <= p[1] && !((c - p[0]) & 1))
+		return 1;
+	p = rbsearch(c, __isothersyms, nelem(__isothersyms), 1);
+	if(p && c == p[0])
+		return 1;
+	return 0;
+}
+
 static Rune __toupperr[] = {
 	0x0061, 0x007a, 1048544,
 	0x00e0, 0x00f6, 1048544,
diff -r 5bf1a8b3aeea src/lib9/utf/utf.h
--- a/src/lib9/utf/utf.h	Thu May 29 13:47:31 2014 -0400
+++ b/src/lib9/utf/utf.h	Tue Jun 03 12:11:38 2014 +0900
@@ -229,6 +229,11 @@
 int isspacerune(Rune r);
 
 
+// isothersymrune tests for other symbols, including "Emoji" and other
+// symbol characters.
+
+int isothersymrune(Rune r);
+
 // (The comments in this file were copied from the manpage files rune.3,
 // isalpharune.3, and runestrcat.3. Some formatting changes were also made
 // to conform to Google style. /JRM 11/11/05)