20070906-1636

7 files changed, 178 insertions, 89 deletions

diff --git a/CHANGES b/CHANGES
index 5c1513b5..ddaa09f9 100644
--- a/CHANGES
+++ b/CHANGES
@@ -1,3 +1,5 @@
+20070906
+	add toreal to string(2)
 20070905
 	/appl/cmd/ed.b (caerwyn's fix for g/.../d)
 	also /appl/lib/ecmascript/builtin.b (toupper fix)
diff --git a/appl/lib/string.b b/appl/lib/string.b
index 16bc1940..fb123f7e 100644
--- a/appl/lib/string.b
+++ b/appl/lib/string.b
@@ -1,4 +1,5 @@
 implement String;
+
 include "string.m";
 
 splitl(s: string, cl: string): (string, string)
@@ -112,61 +113,56 @@ splitstrr(s: string, t: string): (string, string)
 
 prefix(pre: string, s: string): int
 {
-	ns := len s;
-	n := len pre;
-	if(ns < n)
+	if(len s < len pre)
 		return 0;
-	for(k := 0; k < n; k++) {
+	for(k := 0; k < len pre; k++)
 		if(pre[k] != s[k])
 			return 0;
-	}
 	return 1;
 }
 
 tolower(s: string): string
 {
-	r := s;
-	for(i := 0; i < len r; i++) {
-		c := r[i];
-		if(c >= int 'A' && c <= int 'Z')
-			r[i] = r[i] + (int 'a' - int 'A');
+	for(i := 0; i < len s; i++) {
+		c := s[i];
+		if(c >= 'A' && c <= 'Z')
+			s[i] += 'a' - 'A';
 	}
-	return r;
+	return s;
 }
 
 toupper(s: string): string
 {
-	r := s;
-	for(i := 0; i < len r; i++) {
-		c := r[i];
-		if(c >= int 'a' && c <= int 'z')
-			r[i] = r[i] - (int 'a' - int 'A');
+	for(i := 0; i < len s; i++) {
+		c := s[i];
+		if(c >= 'a' && c <= 'z')
+			s[i] += 'A' - 'a';
 	}
-	return r;
+	return s;
 }
 
-tobig(s: string, base: int): (big, string)
+startnum(s: string, base: int): (int, int, int)
 {
-	if (s == nil || base < 0 || base > 36 || base == 1)
-		return (big 0, s);
+	if(s == nil || base != 0 && (base < 2 || base > 36))
+		return (0, 0, 0);
 
 	# skip possible leading white space
-	c: int;
+	c := ' ';
 	for (i := 0; i < len s; i++) {
 		c = s[i];
 		if(c != ' ' && c != '\t' && c != '\n')
 			break;
 	}
 
-	# skip possible sign character
+	# optional sign
 	neg := 0;
-	if (c == '-' || c == '+') {
+	if(c == '-' || c == '+') {
 		if(c == '-')
 			neg = 1;
 		i++;
 	}
 
-	if (base == 0) {
+	if(base == 0) {
 		# parse possible leading base designator
 		start := i;
 		base = -1;
@@ -183,26 +179,30 @@ tobig(s: string, base: int): (big, string)
 			i = start;
 			base = 10;
 		} else if (base == 0 || base > 36)
-			return (big 0, s);
+			return (0, 0, i);
 	}
+	if(i >= len s)
+		return (0, 0, 0);
+
+	return (base, neg, i);
+}
+
+tobig(s: string, base: int): (big, string)
+{
+	neg, i: int;
+
+	(base, neg, i) = startnum(s, base);
+	if(base == 0)
+		return (big 0, s);
 
 	# parse number itself.
-	# perhaps this should check for overflow, and max out, as limbo op does?
+	# probably this should check for overflow, and max out, as limbo op does?
 	start := i;
-	dig := '9';
-	if (base < 10)
-		dig = '0' + base - 1;
 	n := big 0;
 	for (; i < len s; i++) {
-		c = s[i];
-		if ('0' <= c && c <= dig)
-			n = (n * big base) + big(c - '0');
-		else if ('a' <= c && c < 'a' + base - 10)
-			n = (n * big base) + big(c - 'a' + 10);
-		else if ('A' <= c && c  < 'A' + base - 10)
-			n = (n * big base) + big(c - 'A' + 10);
-		else
+		if((d := digit(s[i], base)) < 0)
 			break;
+		n = n*big base + big d;
 	}
 	if (i == start)
 		return (big 0, s);
@@ -213,62 +213,20 @@ tobig(s: string, base: int): (big, string)
 
 toint(s: string, base: int): (int, string)
 {
-	if (s == nil || base < 0 || base > 36 || base == 1)
-		return (0, s);
-
-	# skip possible leading white space
-	c: int;
-	for (i := 0; i < len s; i++) {
-		c = s[i];
-		if(c != ' ' && c != '\t' && c != '\n')
-			break;
-	}
+	neg, i: int;
 
-	# skip possible sign character
-	neg := 0;
-	if (c == '-' || c == '+') {
-		if(c == '-')
-			neg = 1;
-		i++;
-	}
-
-	if (base == 0) {
-		# parse possible leading base designator
-		start := i;
-		base = -1;
-		for (; i < start+3 && i < len s; i++) {
-			c = s[i];
-			if (c == 'r' && i > start) {
-				base = int s[start:i];
-				i++;
-				break;
-			} else if (c < '0' || c > '9')
-				break;
-		}
-		if (base == -1) {
-			i = start;
-			base = 10;
-		} else if (base == 0 || base > 36)
-			return (0, s);
-	}
+	(base, neg, i) = startnum(s, base);
+	if(base == 0)
+		return (0, s);
 
 	# parse number itself.
-	# perhaps this should check for overflow, and max out, as limbo op does?
+	# probably this should check for overflow, and max out, as limbo op does?
 	start := i;
-	dig := '9';
-	if (base < 10)
-		dig = '0' + base - 1;
 	n := 0;
-	for (; i < len s; i++) {
-		c = s[i];
-		if ('0' <= c && c <= dig)
-			n = (n * base) + (c - '0');
-		else if ('a' <= c && c < 'a' + base - 10)
-			n = (n * base) + (c - 'a' + 10);
-		else if ('A' <= c && c  < 'A' + base - 10)
-			n = (n * base) + (c - 'A' + 10);
-		else
+	for (; i < len s; i++){
+		if((d := digit(s[i], base)) < 0)
 			break;
+		n = n*base + d;
 	}
 	if (i == start)
 		return (0, s);
@@ -277,6 +235,117 @@ toint(s: string, base: int): (int, string)
 	return (n, s[i:]);
 }
 
+digit(c: int, base: int): int
+{
+	if ('0' <= c && c <= '0' + base - 1)
+		return c-'0';
+	else if ('a' <= c && c < 'a' + base - 10)
+		return (c - 'a' + 10);
+	else if ('A' <= c && c  < 'A' + base - 10)
+		return (c - 'A' + 10);
+	else
+		return -1;	
+}
+
+rpow(x: real, n: int): real
+{
+	inv := 0;
+	if(n < 0){
+		n = -n;
+		inv = 1;
+	}
+	r := 1.0;
+	for(;;){
+		if(n&1)
+			r *= x;
+		if((n >>= 1) == 0)
+			break;
+		x *= x;
+	}
+	if(inv)
+		r = 1.0/r;
+	return r;
+}
+
+match(p: string, s: string, i: int): int
+{
+	if(i+len p > len s)
+		return 0;
+	for(j := 0; j < len p; j++){
+		c := s[i++];
+		if(c >= 'A' && c <= 'Z')
+			c += 'a'-'A';
+		if(p[j] != c)
+			return 0;
+	}
+	return 1;
+}
+
+toreal(s: string, base: int): (real, string)
+{
+	neg, i: int;
+
+	(base, neg, i) = startnum(s, base);
+	if(base == 0)
+		return (0.0, s);
+
+	c := s[i];
+	if((c == 'i' || c == 'I') && match("infinity", s, i))
+		return (real s, s[i+8:]);
+	if((c == 'n' || c == 'N') && match("nan", s, i))
+		return (real s, s[i+3:]);
+
+	if(digit(c, base) < 0)
+		return (0.0, s);
+
+	num := 0.0;
+	for(; i < len s && (d := digit(s[i], base)) >= 0; i++)
+		num = num*real base + real d;
+	dig := 0;	# digits in fraction
+	if(i < len s && s[i] == '.'){
+		i++;
+		for(; i < len s && (d = digit(s[i], base)) >= 0; i++){
+			num = num*real base + real d;
+			dig++;
+		}
+	}
+	exp := 0;
+	eneg := 0;
+	if(i < len s && ((c = s[i]) == 'e' || c == 'E')){
+		start := i;	# might still be badly formed
+		i++;
+		if(i < len s && ((c = s[i]) == '-' || c == '+')){
+			i++;
+			if(c == '-'){
+				dig = -dig;
+				eneg = 1;
+			}
+		}
+		if(i < len s && s[i] >= '0' && s[i] <= '9'){	# exponents are always decimal
+			for(; i < len s && (d = digit(s[i], 10)) >= 0; i++)
+				exp = exp*base + d;
+		}else
+			i = start;
+	}
+	if(base == 10)
+		return (real s[0: i], s[i:]);	# conversion can be more accurate
+	exp -= dig;
+	if(exp < 0){
+		exp = -exp;
+		eneg = !eneg;
+	}
+	if(exp < 0 || exp > 19999)
+		exp = 19999;	# huge but otherwise arbitrary limit
+	dem := rpow(real base, exp);
+	if(eneg)
+		num /= dem;
+	else
+		num *= dem;
+	if(neg)
+		return  (-num,s[i:]);
+	return (num, s[i:]);
+}
+
 append(s: string, l: list of string): list of string
 {
 	t:	list of string;
diff --git a/dis/lib/string.dis b/dis/lib/string.dis
index f55bd914..19f63987 100644
--- a/dis/lib/string.dis
+++ b/dis/lib/string.dis
diff --git a/include/version.h b/include/version.h
index be5b53b0..db5adb2b 100644
--- a/include/version.h
+++ b/include/version.h
@@ -1 +1 @@
-#define VERSION	"Fourth Edition (20070905)"
+#define VERSION	"Fourth Edition (20070906)"
diff --git a/man/2/INDEX b/man/2/INDEX
index 61a1e4d0..2c46d942 100644
--- a/man/2/INDEX
+++ b/man/2/INDEX
@@ -372,6 +372,7 @@ take string
 tobig string
 toint string
 tolower string
+toreal string
 toupper string
 unquoted string
 stringinttab stringinttab
diff --git a/man/2/string b/man/2/string
index 7d1c4298..9ce877ee 100644
--- a/man/2/string
+++ b/man/2/string
@@ -1,7 +1,7 @@
 .TH STRING 2
 .SH NAME
 string: append, drop, in, prefix, quoted, splitl, splitr, splitstrl,
-splitstrr, take, tobig, toint, tolower, toupper, unquoted \- string operations
+splitstrr, take, tobig, toint, toreal, tolower, toupper, unquoted \- string operations
 .SH SYNOPSIS
 .EX
 include "string.m";
@@ -18,6 +18,7 @@ splitstrr: fn(s, t: string): (string, string);
 take:      fn(s, cl: string): string;
 tobig:     fn(s: string, base: int): (big, string);
 toint:     fn(s: string, base: int): (int, string);
+toreal:    fn(s: string, base: int): (real, string);
 tolower:   fn(s: string): string;
 toupper:   fn(s: string): string;
 quoted:    fn(args: list of string): string;
@@ -137,6 +138,21 @@ has the same specification as
 except that converts to 64-bit
 .BR big .
 .PP
+.B Toreal
+is similar to
+.BR toint ,
+except that it expects a floating-point number after optional leading white space:
+an optional sign, then a string of digits containing a decimal point, then an optional
+.RB ` e '
+or
+.RB ` E '
+followed by an optionally signed decimal integer exponent.
+The string of digits can optionally be preceded by a base (radix) specifier
+of the form
+.IB B r ,
+as for integers.
+Any exponent is then interpreted as a power of that base.
+.PP
 .B Tolower
 converts all upper case letters in the string
 .I s
diff --git a/module/string.m b/module/string.m
index 41ce5e4e..a5051738 100644
--- a/module/string.m
+++ b/module/string.m
@@ -30,6 +30,7 @@ String: module
 	# string to int returning value, remainder
 	toint:		fn(s: string, base: int): (int, string);
 	tobig:		fn(s: string, base: int): (big, string);
+	toreal:		fn(s: string, base: int): (real, string);
 
 	# append s to end of l
 	append:		fn(s: string, l: list of string): list of string;