Fix #7: toml_utf8_to_ucs() returns incorrect results

toml.c

@@ -71,9 +71,11 @@ int toml_utf8_to_ucs(const char* orig, int len, int64_t* ret)
     if (0x6 == (i >> 5)) {
 	if (len < 2) return -1;
 	v = i & 0x1f;
-	i = *(++buf);
-	if (0x2 != (i >> 6)) return -1;
-	v = (v << 6) | (i & 0x3f);
+        for (int j = 0; j < 1; j++) {
+            i = *buf++;
+            if (0x2 != (i >> 6)) return -1;
+            v = (v << 6) | (i & 0x3f);
+        }
 	return *ret = v, (const char*) buf - orig;
     }
 
@@ -84,7 +86,7 @@ int toml_utf8_to_ucs(const char* orig, int len, int64_t* ret)
 	if (len < 3) return -1;
 	v = i & 0x0F;
 	for (int j = 0; j < 2; j++) {
-	    i = *(++buf);
+	    i = *buf++;
 	    if (0x2 != (i >> 6)) return -1;
 	    v = (v << 6) | (i & 0x3f);
 	}
@@ -98,7 +100,7 @@ int toml_utf8_to_ucs(const char* orig, int len, int64_t* ret)
 	if (len < 4) return -1;
 	v = i & 0x07;
 	for (int j = 0; j < 3; j++) {
-	    i = *(++buf);
+	    i = *buf++;
 	    if (0x2 != (i >> 6)) return -1;
 	    v = (v << 6) | (i & 0x3f);
 	}
@@ -112,7 +114,7 @@ int toml_utf8_to_ucs(const char* orig, int len, int64_t* ret)
 	if (len < 5) return -1;
 	v = i & 0x03;
 	for (int j = 0; j < 4; j++) {
-	    i = *(++buf);
+	    i = *buf++;
 	    if (0x2 != (i >> 6)) return -1;
 	    v = (v << 6) | (i & 0x3f);
 	}
@@ -126,7 +128,7 @@ int toml_utf8_to_ucs(const char* orig, int len, int64_t* ret)
 	if (len < 6) return -1;
 	v = i & 0x01;
 	for (int j = 0; j < 5; j++) {
-	    i = *(++buf);
+	    i = *buf++;
 	    if (0x2 != (i >> 6)) return -1;
 	    v = (v << 6) | (i & 0x3f);
 	}

unittest/Makefile

@@ -0,0 +1,11 @@
+CFLAGS = -g -I..
+
+TESTS = t1 
+
+all: $(TESTS)
+
+t1: t1.c ../toml.c
+
+clean:
+	rm -f $(TESTS)
+

unittest/t1.c

@@ -0,0 +1,57 @@
+#include <stdio.h>
+#include <assert.h>
+#include <stdint.h>
+#include <string.h>
+#include "../toml.h"
+
+
+
+int main(int argc, const char* argv[])
+{
+    char    xxbuf[6], buf[6];
+    int64_t xxcode, code;
+    int     xxsize;
+    
+    
+    xxsize = 2, xxcode = 0x80; memcpy(xxbuf, "\xc2\x80", xxsize);
+    assert(toml_ucs_to_utf8(xxcode, buf) == xxsize && 0 == memcmp(buf, xxbuf, xxsize));
+    assert(toml_utf8_to_ucs(buf, xxsize, &code) == xxsize && code == xxcode);
+
+    xxsize = 2, xxcode = 0x7ff; memcpy(xxbuf, "\xdf\xbf", xxsize); 
+    assert(toml_ucs_to_utf8(xxcode, buf) == xxsize && 0 == memcmp(buf, xxbuf, xxsize));
+    assert(toml_utf8_to_ucs(buf, xxsize, &code) == xxsize && code == xxcode);
+
+    xxsize = 3, xxcode = 0x800; memcpy(xxbuf, "\xe0\xa0\x80", xxsize);
+    assert(toml_ucs_to_utf8(xxcode, buf) == xxsize && 0 == memcmp(buf, xxbuf, xxsize));
+    assert(toml_utf8_to_ucs(buf, xxsize, &code) == xxsize && code == xxcode);
+    
+    xxsize = 3, xxcode = 0xfffd; memcpy(xxbuf, "\xef\xbf\xbd", xxsize);
+    assert(toml_ucs_to_utf8(xxcode, buf) == xxsize && 0 == memcmp(buf, xxbuf, xxsize));
+    assert(toml_utf8_to_ucs(buf, xxsize, &code) == xxsize && code == xxcode);
+    
+    xxsize = 4, xxcode = 0x10000; memcpy(xxbuf, "\xf0\x90\x80\x80", xxsize);
+    assert(toml_ucs_to_utf8(xxcode, buf) == xxsize && 0 == memcmp(buf, xxbuf, xxsize));
+    assert(toml_utf8_to_ucs(buf, xxsize, &code) == xxsize && code == xxcode);
+    
+    xxsize = 4, xxcode = 0x1fffff; memcpy(xxbuf, "\xf7\xbf\xbf\xbf", xxsize);
+    assert(toml_ucs_to_utf8(xxcode, buf) == xxsize && 0 == memcmp(buf, xxbuf, xxsize));
+    assert(toml_utf8_to_ucs(buf, xxsize, &code) == xxsize && code == xxcode);
+
+    xxsize = 5, xxcode = 0x200000; memcpy(xxbuf, "\xf8\x88\x80\x80\x80", xxsize);
+    assert(toml_ucs_to_utf8(xxcode, buf) == xxsize && 0 == memcmp(buf, xxbuf, xxsize));
+    assert(toml_utf8_to_ucs(buf, xxsize, &code) == xxsize && code == xxcode);
+    
+    xxsize = 5, xxcode = 0x3ffffff; memcpy(xxbuf, "\xfb\xbf\xbf\xbf\xbf", xxsize);
+    assert(toml_ucs_to_utf8(xxcode, buf) == xxsize && 0 == memcmp(buf, xxbuf, xxsize));
+    assert(toml_utf8_to_ucs(buf, xxsize, &code) == xxsize && code == xxcode);
+
+    xxsize = 6, xxcode = 0x4000000; memcpy(xxbuf, "\xfc\x84\x80\x80\x80\x80", xxsize);
+    assert(toml_ucs_to_utf8(xxcode, buf) == xxsize && 0 == memcmp(buf, xxbuf, xxsize));
+    assert(toml_utf8_to_ucs(buf, xxsize, &code) == xxsize && code == xxcode);
+
+    xxsize = 6, xxcode = 0x7fffffff; memcpy(xxbuf, "\xfd\xbf\xbf\xbf\xbf\xbf", xxsize);
+    assert(toml_ucs_to_utf8(xxcode, buf) == xxsize && 0 == memcmp(buf, xxbuf, xxsize));
+    assert(toml_utf8_to_ucs(buf, xxsize, &code) == xxsize && code == xxcode);
+
+    return 0;
+}