shithub: femtolisp

Info  •  Files  •  Log  •  Branches

print_string: fix mixed ascii+non-ascii runes clobbered printout

Fixes: https://todo.sr.ht/~ft/femtolisp/34

--- a/print.c

+++ b/print.c

@@ -506,14 +506,14 @@

 static void

 print_string(ios_t *f, const char *str, size_t sz)

 {

-	char buf[512];

-	size_t i = 0;

+	char buf[64];

 	uint8_t c;

-	static char hexdig[] = "0123456789abcdef";

+	static const char hexdig[] = "0123456789abcdef";

+	size_t i = 0;

 	if(!u8_isvalid(str, sz)){

 		// alternate print algorithm that preserves data if it's not UTF-8

-		for(i = 0; i < sz; i++){

+		for(; i < sz; i++){

 			c = str[i];

 			if(c == '\\')

 				outsn(f, "\\\\", 2);

@@ -529,7 +529,7 @@

 		}

 	}else{

 		while(i < sz){

-			size_t n = u8_escape(buf, sizeof(buf), str, &i, sz, 1, 0);

+			size_t n = u8_escape(buf, sizeof(buf), str, &i, sz, true, false);

 			outsn(f, buf, n-1);

 		}

 	}

@@ -759,11 +759,12 @@

 				}

 				return;

 			}else if(eltype == FL_runesym){

-				char buf[UTFmax];

+				char buf[UTFmax+1];

 				if(!FL(print_princ))

 					outc(f, '"');

 				for(i = 0; i < cnt; i++, data = (char*)data + elsize){

 					int n = runetochar(buf, (Rune*)data);

+					buf[n] = 0;

 					if(FL(print_princ))

 						ios_write(f, buf, n);

 					else

--- a/utf8.c

+++ b/utf8.c

@@ -16,8 +16,8 @@

 #include "flisp.h"

 static const uint32_t offsetsFromUTF8[6] = {

-	0x00000000UL, 0x00003080UL, 0x000E2080UL,

-	0x03C82080UL, 0xFA082080UL, 0x82082080UL

+	0x00000000U, 0x00003080U, 0x000E2080U,

+	0x03C82080U, 0xFA082080U, 0x82082080U

 };

 static const char trailingBytesForUTF8[256] = {

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

 size_t

 u8_seqlen(const char *s)

 {

-	return trailingBytesForUTF8[(unsigned int)(uint8_t)s[0]] + 1;

+	return trailingBytesForUTF8[(uint8_t)s[0]] + 1;

 }

 /* byte offset => charnum */

@@ -80,14 +80,12 @@

 Rune

 u8_nextmemchar(const char *s, size_t *i)

 {

+	size_t sz = u8_seqlen(&s[*i]);

 	Rune ch = 0;

-	size_t sz = 0;

-

-	do{

+	for(size_t j = 0; j < sz; j++){

 		ch <<= 6;

 		ch += (uint8_t)s[(*i)++];

-		sz++;

-	}while(!isutf(s[*i]));

+	};

 	return ch - offsetsFromUTF8[sz-1];

 }

@@ -94,7 +92,7 @@

 int

 octal_digit(char c)

 {

-	return (c >= '0' && c <= '7');

+	return c >= '0' && c <= '7';

 }

 int

@@ -156,7 +154,7 @@

 }

 size_t

-u8_escape(char *buf, size_t sz, const char *src, size_t *pi, size_t end, int escape_quotes, int ascii)

+u8_escape(char *buf, size_t sz, const char *src, size_t *pi, size_t end, bool escape_quotes, bool ascii)

 {

 	size_t i = *pi, i0;

 	Rune ch;

@@ -187,7 +185,7 @@

 	}

 	*buf++ = '\0';

 	*pi = i;

-	return (buf-start);

+	return buf - start;

 }

 char *

--- a/utf8.h

+++ b/utf8.h

@@ -37,7 +37,7 @@

    returns number of bytes placed in buf, including a NUL terminator.

 */

 size_t u8_escape(char *buf, size_t sz, const char *src, size_t *pi, size_t end,

-                 int escape_quotes, int ascii);

+                 bool escape_quotes, bool ascii);

 /* utility predicates used by the above */

 int octal_digit(char c);