SC_FUNC int scan_utf8(FILE *fp,const char *filename)
{
static void *resetpos=NULL;
int utf8=TRUE;
int firstchar=TRUE,bom_found=FALSE;
const unsigned char *ptr;
resetpos=pc_getpossrc(fp,resetpos);
while (utf8 && pc_readsrc(fp,srcline,sLINEMAX)!=NULL) {
ptr=srcline;
// if (firstchar) {
// /* check whether the very first character on the very first line
// * starts with a byte order mark (BOM)
// */
// cell c=get_utf8_char(ptr,&ptr);
// bom_found= (c==0xfeff);
// utf8= (c>=0);
// firstchar=FALSE;
// } /* if */
while (utf8 && *ptr!='\0')
utf8= (get_utf8_char(ptr,&ptr)>=0);
} /* while */
pc_resetsrc(fp,resetpos);
if (bom_found) {
unsigned char bom[3];
pc_readsrc(fp,bom,3); /* read the BOM again to strip it from the file */
assert(bom[0]==0xef && bom[1]==0xbb && bom[2]==0xbf);
if (!utf8)
error(77,filename); /* malformed UTF-8 encoding */
} /* if */
return utf8;
}
SC_FUNC int scan_utf8(FILE *fp,const char *filename)
{
#if defined NO_UTF8
return 0;
#else
void *resetpos=pc_getpossrc(fp);
int utf8=TRUE;
int firstchar=TRUE,bom_found=FALSE;
const unsigned char *ptr;
while (utf8 && pc_readsrc(fp,pline,sLINEMAX)!=NULL) {
ptr=pline;
if (firstchar) {
/* check whether the very first character on the very first line
* starts with a BYTE order mark
*/
cell c=get_utf8_char(ptr,&ptr);
bom_found= (c==0xfeff);
utf8= (c>=0);
firstchar=FALSE;
} /* if */
while (utf8 && *ptr!='\0')
utf8= (get_utf8_char(ptr,&ptr)>=0);
} /* while */
pc_resetsrc(fp,resetpos);
if (bom_found) {
unsigned char bom[3];
if (!utf8)
error(77,filename); /* malformed UTF-8 encoding */
pc_readsrc(fp,bom,3);
assert(bom[0]==0xef && bom[1]==0xbb && bom[2]==0xbf);
} /* if */
return utf8;
#endif /* NO_UTF8 */
}
