shithub: sl

Info  •  Files  •  Log  •  Branches

string-width: fix for strings containing \0

--- a/src/print.c

+++ b/src/print.c

@@ -17,18 +17,17 @@

 		FL(hpos)++;

 }

-static void

-outs(ios_t *f, const char *s)

+static inline void

+outsn(ios_t *f, const char *s, size_t n)

 {

-	ios_puts(f, s);

-	FL(hpos) += u8_strwidth(s);

+	ios_write(f, s, n);

+	FL(hpos) += u8_strwidth(s, n);

 }

-static void

-outsn(ios_t *f, const char *s, size_t n)

+static inline void

+outs(ios_t *f, const char *s)

 {

-	ios_write(f, s, n);

-	FL(hpos) += u8_strwidth(s);

+	outsn(f, s, strlen(s));

 }

 static int

@@ -167,8 +166,10 @@

 static inline int

 tinyp(value_t v)

 {

-	if(issymbol(v))

-		return u8_strwidth(symbol_name(v)) < SMALL_STR_LEN;

+	if(issymbol(v)){

+		const char *s = symbol_name(v);

+		return u8_strwidth(s, strlen(s)) < SMALL_STR_LEN;

+	}

 	if(fl_isstring(v))

 		return cv_len(ptr(v)) < SMALL_STR_LEN;

 	return (

@@ -215,8 +216,10 @@

 lengthestimate(value_t v)

 {

 	// get the width of an expression if we can do so cheaply

-	if(issymbol(v))

-		return u8_strwidth(symbol_name(v));

+	if(issymbol(v)){

+		const char *s = symbol_name(v);

+		return u8_strwidth(s, strlen(s));

+	}

 	if(iscprim(v) && ptr(v) != nil && cp_class(ptr(v)) == FL(runetype))

 		return 4;

 	return -1;

--- a/src/string.c

+++ b/src/string.c

@@ -51,7 +51,11 @@

 			return w < 0 ? FL_f : fixnum(w);

 		}

 	}

-	return size_wrap(u8_strwidth(tostring(args[0])));

+	if(!fl_isstring(args[0]))

+		type_error("string", args[0]);

+	char *str = tostring(args[0]);

+	size_t len = cv_len(ptr(args[0]));

+	return size_wrap(u8_strwidth(str, len));

 }

 BUILTIN("string-reverse", string_reverse)

--- a/src/utf8.c

+++ b/src/utf8.c

@@ -62,12 +62,12 @@

 }

 size_t

-u8_strwidth(const char *s)

+u8_strwidth(const char *s, size_t n)

 {

 	size_t i, w;

 	Rune r;

-	for(i = w = 0; s[i];){

+	for(i = w = 0; i < n;){

 		i += chartorune(&r, s+i);

 		w += fl_wcwidth(r);

 	}

--- a/src/utf8.h

+++ b/src/utf8.h

@@ -49,7 +49,7 @@

 char *u8_memchr(char *s, Rune ch, size_t sz, size_t *charn);

 /* number of columns occupied by a string */

-size_t u8_strwidth(const char *s) fl_purefn;

+size_t u8_strwidth(const char *s, size_t n) fl_purefn;

 /* determine whether a sequence of bytes is valid UTF-8. length is in bytes */

 int u8_isvalid(const char *str, int length) fl_purefn;

--- a/test/unittest.lsp

+++ b/test/unittest.lsp

@@ -514,5 +514,8 @@

 (assert (equal? "" (string-reverse "")))

 (assert (equal? "й" (string-reverse "й")))

+(assert (eq? 10 (string-width s)))

+(assert (eq? 0 (string-width "")))

+

 (princ "all tests pass")

 (newline)