shithub: sl

Info  •  Files  •  Log  •  Branches

str-find: index using runes instead of bytes

References: https://todo.sr.ht/~ft/sl/30

--- a/src/str.c

+++ b/src/str.c

@@ -289,40 +289,45 @@

 BUILTIN("str-find", str_find)

 {

-	char cbuf[UTFmax+1];

-	usize start = 0;

+	usize startrune = 0;

 	if(nargs == 3)

-		start = tosize(args[2]);

+		startrune = tosize(args[2]);

 	else

 		argcount(nargs, 2);

 	char *s = tostr(args[0]);

-	usize len = cv_len(ptr(args[0]));

-	if(start > len)

-		bounds_error(args[0], args[2]);

-	char *needle; usize needlesz;

+	usize sbytes = cv_len(ptr(args[0]));

+	// needle

+	char *nd;

+	usize ndbytes;

 	sl_v v = args[1];

+	char rbuf[UTFmax+1];

 	sl_cprim *cp = ptr(v);

 	if(iscprim(v) && cp_class(cp) == sl_runetype){

 		Rune r = *(Rune*)cp_data(cp);

-		needlesz = runetochar(cbuf, &r);

-		needle = cbuf;

-		needle[needlesz] = 0;

+		ndbytes = runetochar(rbuf, &r);

+		nd = rbuf;

+		nd[ndbytes] = 0;

 	}else if(sl_isstr(v)){

 		csl_v *cv = ptr(v);

-		needlesz = cv_len(cv);

-		needle = (char*)cv_data(cv);

+		ndbytes = cv_len(cv);

+		nd = (char*)cv_data(cv);

 	}else{

-		type_error("str", args[1]);

+		type_error("str or rune", args[1]);

 	}

-	if(needlesz > len-start)

-		return sl_nil;

-	if(needlesz == 0)

-		return size_wrap(start);

-	usize i;

-	for(i = start; i < len-needlesz+1; i++){

-		if(s[i] == needle[0] && memcmp(&s[i+1], needle+1, needlesz-1) == 0)

-			return size_wrap(i);

+	if(ndbytes == 0)

+		return size_wrap(startrune);

+	usize i, n;

+	// first iterate to the starting rune

+	for(i = n = 0; n < startrune && i < sbytes; n++)

+		i += u8_seqlen(s+i);

+	if(n != startrune)

+		bounds_error(args[0], fixnum(startrune));

+	// now search for the needle

+	for(; i < sbytes-ndbytes+1; n++){

+		if(s[i] == nd[0] && memcmp(&s[i+1], nd+1, ndbytes-1) == 0)

+			return size_wrap(n);

+		i += u8_seqlen(s+i);

 	}

 	return sl_nil;

 }

--- a/test/unittest.lsp

+++ b/test/unittest.lsp

@@ -596,20 +596,20 @@

 (assert (rune-title-case? #\Dž))

 (def s "hello й goodbye")

-(assert (= 6 (str-find s #\й)))

-(assert (= 6 (str-find s #\й 6)))

-(assert-fail (str-find s #\o -1))

-(assert (not (str-find s #\o 16)))

-(assert-fail (str-find s #\o 17))

 (assert (= 4 (str-find s #\o)))

-(assert (= 10 (str-find s #\o 5)))

-(assert (= 11 (str-find s #\o 11)))

-(assert (not (str-find s #\o 12)))

+(assert (= 9 (str-find s #\o 5)))

+(assert (= 10 (str-find s #\o 10)))

+(assert (not (str-find s #\o 11)))

+(assert (not (str-find s #\o 15)))

 (assert (= 4 (str-find s "o")))

 (assert (= 2 (str-find s "ll")))

 (assert (not (str-find s "ll" 3)))

 (assert (= 0 (str-find s "")))

 (assert (= 7 (str-find s "" 7)))

+(assert (= 6 (str-find s #\й)))

+(assert (= 6 (str-find s #\й 6)))

+(assert-fail (str-find s #\o -1))

+(assert-fail (str-find s #\o 16))

 (assert-fail (str-find s 0))

 (assert-fail (str-find s (byte #\o)))