shithub: mc

--- a/bld.sh

+++ b/bld.sh

@@ -46,6 +46,7 @@

     die.myr \

     alloc.myr\

     str.myr \

+    fmt.myr \

     chartype.myr"

 OBJ="$(echo $ASM | sed 's/\.s/.o /g') $(echo $MYR | sed 's/\.myr/.o /g')"

@@ -65,4 +66,8 @@

 echo $COMP

 $COMP

+build f.myr

+COMP="$CC -m32 -o f f.o -L. -lstd"

+echo $COMP

+$COMP

--- a/chartype.myr

+++ b/chartype.myr

@@ -1036,11 +1036,12 @@

 	0x1ffc, 491	/* ῼ ῳ */

-const bsearch = {c, t, sz, nelt, ret

+const findc = {c, t, sz, nelt, ret

 	var l

 	var m

-	while l.len > 1

+	l = t

+	while l.len > nelt

 		m = l.len/2

 		l = t[m+nelt, t.len]

 		if c >= l[0]

@@ -1059,18 +1060,16 @@

-const isalpha = {chr

+const isalpha = {c

 	var l

-	var c

-	c = chr castto(int)

-	if isupper(chr) || islower(chr)

+	if isupper(c) || islower(c)

 		-> true

-	elif bsearch(c, tabalpha2[0, tabalpha2.len], tabalpha2.len, 2, &l)

+	elif findc(c, tabalpha2[0, tabalpha2.len], tabalpha2.len, 2, &l)

 		if (c >= l[0] && c <= l[1])

 			-> true

;;

-	elif bsearch(c, tabalpha1[0, tabalpha1.len], tabalpha1.len, 1, &l)

+	elif findc(c, tabalpha1[0, tabalpha1.len], tabalpha1.len, 1, &l)

 		if (c == l[0])

 			-> true

;;

@@ -1078,11 +1077,11 @@

 	-> false

-const isnum = {chr

+const isnum = {c

 	var l

 	var c

-	if bsearch(c, tabisdigitr[0, tabisdigitr.len], tabisdigitr.len/2, 2, &l)

+	if findc(c, tabisdigitr[0, tabisdigitr.len], tabisdigitr.len/2, 2, &l)

 		if(c >= l[0] && c <= l[1])

 			-> true

;;

@@ -1089,14 +1088,20 @@

;;

 	-> false

-const isalnum = {chr

-	-> isalpha(chr) || isnum(chr)

+const isalnum = {c

+	-> isalpha(c) || isnum(c)

-const isspace = {chr

+const isspace = {c

 	var l

 	var c

+	var sl

+	var len

-	if bsearch(c, tabspace2[0,tabspace2.len], tabspace2.len/2, 2, &l)

+	sl = tabspace2[0,tabspace2.len]

+	len = tabspace2.len/2

+	if findc(c, sl, len, 2, &l)

 		if(c >= l[0] && c <= l[1])

 			-> true

;;

@@ -1104,16 +1109,15 @@

 	-> false

-const islower = {chr

+const islower = {c

 	var l

 	var c

-	c = chr castto(int)

-	if bsearch(c, tabtoupper2[0, tabtoupper2.len], tabtoupper2.len, 2, &l)

+	if findc(c, tabtoupper2[0, tabtoupper2.len], tabtoupper2.len, 2, &l)

 		if (c >= l[0] && c <= l[1])

 			-> true

;;

-	elif bsearch(c, tabtoupper1[0, tabtoupper1.len], tabtoupper1.len, 1, &l)

+	elif findc(c, tabtoupper1[0, tabtoupper1.len], tabtoupper1.len, 1, &l)

 		if (c == l[0])

 			-> true

;;

@@ -1121,15 +1125,15 @@

 	-> false

-const isupper = {chr

+const isupper = {c

 	var l

 	var c

-	if bsearch(c, tabtolower2[0, tabtolower2.len], tabtolower2.len, 2, &l)

+	if findc(c, tabtolower2[0, tabtolower2.len], tabtolower2.len, 2, &l)

 		if (c >= l[0] && c <= l[1])

 			-> true

;;

-	elif bsearch(c, tabtolower1[0, tabtolower1.len], tabtolower1.len, 1, &l)

+	elif findc(c, tabtolower1[0, tabtolower1.len], tabtolower1.len, 1, &l)

 		if (c == l[0])

 			-> true

;;

--- a/str.myr

+++ b/str.myr

@@ -5,10 +5,11 @@

 pkg std =

 	const Badchar	: char = -1 castto(char)

-	const encode	: (buf : byte[,], chr : char -> bool)

-	const decode	: (str : byte[,] -> char)

+	const charlen	: (chr : char -> int)

+	const encode	: (chr : char, buf : byte[,] -> bool)

+	const decode	: (buf : byte[,] -> char)

 	const striter	: (str : byte[,] -> [char, byte[,]])

 	const strjoin	: (lst : byte[,][,], delim:byte[,] -> byte[,])

 	const strsep	: (str : byte[,], delim:byte[,] -> byte[,][,])

 	const strbjoin	: (lst : byte[,][,], delim:byte[,] -> byte[,])

@@ -15,6 +16,53 @@

 	const strbsep	: (str : byte[,], delim:byte[,] -> byte[,][,])

;;

+const charlen = {c

+	if c < 0x80

+		-> 1

+	elif c < 0x800

+		-> 2

+	elif c < 0x10000

+		-> 3

+	elif c < 0x200000

+		-> 4

+	else

+		-> -1

+	;;

+}

+const encode = {c, buf

+	var len

+	var mark

+	var i

+	len = charlen(c)

+	if len < 0 || buf.len < len

+		-> false

+	;;

+	if (len == 1)

+		mark = 0

+	else

+		mark = (((1 << (8 - len)) - 1) ^ 0xff) castto(char)

+	;;

+	for i = len - 1; i > 0; i--

+		buf[i] = (c & 0x3f | 0x80) castto(byte)

+		c >>= 6

+	;;

+	buf[0] = (c | mark) castto(byte)

+	-> true

+}

+const decode = {buf

+	var c

+	var b

+	(c, b) = striter(buf)

+	-> c

+}

 const striter = {str

 	var len

 	var mask

@@ -23,7 +71,11 @@

 	var c

 	var tmp

+	if !str.len

+		-> (Badchar, str)

+	;;

 	c = str[0]

+	len = 0

 	if c & 0x80 == 0	/* 0b0xxx_xxxx */

 		len = 1

 	elif c & 0xe0 == 0xc0	/* 0b110x_xxxx */

@@ -36,6 +88,10 @@

 		/* skip one char forward so we can try

 		   resyncing the character stream */

 		-> (Badchar, str[1,str.len])

+	;;

+	if len == 0 || len > str.len

+		-> (Badchar, str)

;;

 	mask = (1 << (7 - len)) - 1

--- a/test.myr

+++ b/test.myr

@@ -2,6 +2,7 @@

 const main = {

 	var x : byte*[1024]

+	var buf : byte[1024]

 	var sz

 	var i

@@ -22,9 +23,9 @@

 	for i = 0; i < 1024; i++

 		std.free(x[i])

;;

-	chartypes()

 	std.write(1, "Hello, 世界\n")

+	chartypes()

 const chartypes = {

@@ -33,6 +34,12 @@

 	s = "世界 123\n"

 	for (c, s) = std.striter(s); s.len != 0; (c, s) = std.striter(s)

-		c = c

+		if std.isspace(c)

+			std.write(1, "Space\n")

+		elif std.isalpha(c)

+			std.write(1, "Alpha\n")

+		elif std.isnum(c)

+			std.write(1, "Num\n")

+		;;

;;