static void gpt_printName(uint16_t name[], const char* msg)
{
char *namestr;
namestr = (char*)cli_utf16toascii((char*)name, 72);
cli_dbgmsg("%s: %s\n", msg, namestr);
free(namestr);
}
cli_file_t cli_filetype2(int desc, const struct cl_engine *engine)
{
unsigned char smallbuff[MAGIC_BUFFER_SIZE + 1], *decoded, *bigbuff;
int bread, sret;
cli_file_t ret = CL_TYPE_UNKNOWN_DATA;
struct cli_matcher *root;
struct cli_ac_data mdata;
memset(smallbuff, 0, sizeof(smallbuff));
if((bread = read(desc, smallbuff, MAGIC_BUFFER_SIZE)) > 0)
ret = cli_filetype(smallbuff, bread);
if(engine && ret == CL_TYPE_UNKNOWN_TEXT) {
root = engine->root[0];
if(!root)
return ret;
if(cli_ac_initdata(&mdata, root->ac_partsigs, AC_DEFAULT_TRACKLEN))
return ret;
sret = cli_ac_scanbuff(smallbuff, bread, NULL, engine->root[0], &mdata, 1, 0, 0, -1, NULL);
cli_ac_freedata(&mdata);
if(sret >= CL_TYPENO) {
ret = sret;
} else {
if(cli_ac_initdata(&mdata, root->ac_partsigs, AC_DEFAULT_TRACKLEN))
return ret;
decoded = (unsigned char *) cli_utf16toascii((char *) smallbuff, bread);
if(decoded) {
sret = cli_ac_scanbuff(decoded, strlen((char *) decoded), NULL, engine->root[0], &mdata, 1, 0, 0, -1, NULL);
free(decoded);
if(sret == CL_TYPE_HTML)
ret = CL_TYPE_HTML_UTF16;
}
cli_ac_freedata(&mdata);
if((((struct cli_dconf*) engine->dconf)->phishing & PHISHING_CONF_ENTCONV) && ret != CL_TYPE_HTML_UTF16) {
struct entity_conv conv;
const size_t conv_size = 2*bread < 256 ? 256 : 2*bread;
if(init_entity_converter(&conv,UNKNOWN,conv_size) == 0) {
int end = 0;
m_area_t area;
area.buffer = (unsigned char *) smallbuff;
area.length = bread;
area.offset = 0;
while(!end) {
if(cli_ac_initdata(&mdata, root->ac_partsigs, AC_DEFAULT_TRACKLEN))
return ret;
decoded = encoding_norm_readline(&conv, NULL, &area, bread);
if(decoded) {
sret = cli_ac_scanbuff(decoded, strlen((const char *) decoded), NULL, engine->root[0], &mdata, 1, 0, 0, -1, NULL);
free(decoded);
if(sret == CL_TYPE_HTML) {
ret = CL_TYPE_HTML;
end = 1;
}
} else
end = 1;
cli_ac_freedata(&mdata);
}
entity_norm_done(&conv);
} else {
cli_warnmsg("cli_filetype2: Error initializing entity converter\n");
}
}
}
}
if(ret == CL_TYPE_UNKNOWN_DATA || ret == CL_TYPE_UNKNOWN_TEXT) {
if(!(bigbuff = (unsigned char *) cli_calloc(37638 + 1, sizeof(unsigned char))))
return ret;
lseek(desc, 0, SEEK_SET);
if((bread = read(desc, bigbuff, 37638)) > 0) {
bigbuff[bread] = 0;
switch(is_tar(bigbuff, bread)) {
case 1:
ret = CL_TYPE_OLD_TAR;
cli_dbgmsg("Recognized old fashioned tar file\n");
break;
case 2:
ret = CL_TYPE_POSIX_TAR;
cli_dbgmsg("Recognized POSIX tar file\n");
break;
}
}
if(ret == CL_TYPE_UNKNOWN_DATA || ret == CL_TYPE_UNKNOWN_TEXT) {
if(!memcmp(bigbuff + 32769, "CD001" , 5) || !memcmp(bigbuff + 37633, "CD001" , 5)) {
cli_dbgmsg("Recognized ISO 9660 CD-ROM data\n");
ret = CL_TYPE_DATA;
} else if(!memcmp(bigbuff + 32776, "CDROM" , 5)) {
cli_dbgmsg("Recognized High Sierra CD-ROM data\n");
ret = CL_TYPE_DATA;
}
}
free(bigbuff);
}
return ret;
}
