DeforaNetworks/hashcat
3
0
Fork 0
Code Releases Activity

File handling: Do not abort on seeing a BOM in input files, just warn and ignore the BOM

Browse Source
This commit is contained in:
Jens Steube
2021-06-29 20:42:22 +02:00
parent 2af45cd03f
commit 254e33c473
7 changed files with 114 additions and 84 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/shared.c
+64 -55
View File
@@ -388,7 +388,7 @@ bool hc_path_has_bom (const char *path)
HCFILE fp;
if (hc_fopen (&fp, path, "rb") == false) return false;
if (hc_fopen_raw (&fp, path, "rb") == false) return false;
const size_t nread = hc_fread (buf, 1, sizeof (buf), &fp);
@@ -396,95 +396,104 @@ bool hc_path_has_bom (const char *path)
if (nread < 1) return false;
const int bom_size = hc_string_bom_size (buf);
const bool has_bom = bom_size > 0;
return has_bom;
}
int hc_string_bom_size (const u8 *s)
{
/* signatures from https://en.wikipedia.org/wiki/Byte_order_mark#Byte_order_marks_by_encoding */
// utf-8
if ((buf[0] == 0xef)
&& (buf[1] == 0xbb)
&& (buf[2] == 0xbf)) return true;
if ((s[0] == 0xef)
&& (s[1] == 0xbb)
&& (s[2] == 0xbf)) return 3;
// utf-16
if ((buf[0] == 0xfe)
&& (buf[1] == 0xff)) return true;
if ((s[0] == 0xfe)
&& (s[1] == 0xff)) return 2;
if ((buf[0] == 0xff)
&& (buf[1] == 0xfe)) return true;
if ((s[0] == 0xff)
&& (s[1] == 0xfe)) return 2;
// utf-32
if ((buf[0] == 0x00)
&& (buf[1] == 0x00)
&& (buf[2] == 0xfe)
&& (buf[3] == 0xff)) return true;
if ((s[0] == 0x00)
&& (s[1] == 0x00)
&& (s[2] == 0xfe)
&& (s[3] == 0xff)) return 4;
if ((buf[0] == 0xff)
&& (buf[1] == 0xfe)
&& (buf[2] == 0x00)
&& (buf[3] == 0x00)) return true;
if ((s[0] == 0xff)
&& (s[1] == 0xfe)
&& (s[2] == 0x00)
&& (s[3] == 0x00)) return 4;
// utf-7
if ((buf[0] == 0x2b)
&& (buf[1] == 0x2f)
&& (buf[2] == 0x76)
&& (buf[3] == 0x38)) return true;
if ((s[0] == 0x2b)
&& (s[1] == 0x2f)
&& (s[2] == 0x76)
&& (s[3] == 0x38)) return 4;
if ((buf[0] == 0x2b)
&& (buf[1] == 0x2f)
&& (buf[2] == 0x76)
&& (buf[3] == 0x39)) return true;
if ((s[0] == 0x2b)
&& (s[1] == 0x2f)
&& (s[2] == 0x76)
&& (s[3] == 0x39)) return 4;
if ((buf[0] == 0x2b)
&& (buf[1] == 0x2f)
&& (buf[2] == 0x76)
&& (buf[3] == 0x2b)) return true;
if ((s[0] == 0x2b)
&& (s[1] == 0x2f)
&& (s[2] == 0x76)
&& (s[3] == 0x2b)) return 4;
if ((buf[0] == 0x2b)
&& (buf[1] == 0x2f)
&& (buf[2] == 0x76)
&& (buf[3] == 0x2f)) return true;
if ((s[0] == 0x2b)
&& (s[1] == 0x2f)
&& (s[2] == 0x76)
&& (s[3] == 0x2f)) return 4;
if ((buf[0] == 0x2b)
&& (buf[1] == 0x2f)
&& (buf[2] == 0x76)
&& (buf[3] == 0x38)
&& (buf[4] == 0x2d)) return true;
if ((s[0] == 0x2b)
&& (s[1] == 0x2f)
&& (s[2] == 0x76)
&& (s[3] == 0x38)
&& (s[4] == 0x2d)) return 5;
// utf-1
if ((buf[0] == 0xf7)
&& (buf[1] == 0x64)
&& (buf[2] == 0x4c)) return true;
if ((s[0] == 0xf7)
&& (s[1] == 0x64)
&& (s[2] == 0x4c)) return 3;
// utf-ebcdic
if ((buf[0] == 0xdd)
&& (buf[1] == 0x73)
&& (buf[2] == 0x66)
&& (buf[3] == 0x73)) return true;
if ((s[0] == 0xdd)
&& (s[1] == 0x73)
&& (s[2] == 0x66)
&& (s[3] == 0x73)) return 4;
// scsu
if ((buf[0] == 0x0e)
&& (buf[1] == 0xfe)
&& (buf[2] == 0xff)) return true;
if ((s[0] == 0x0e)
&& (s[1] == 0xfe)
&& (s[2] == 0xff)) return 3;
// bocu-1
if ((buf[0] == 0xfb)
&& (buf[1] == 0xee)
&& (buf[2] == 0x28)) return true;
if ((s[0] == 0xfb)
&& (s[1] == 0xee)
&& (s[2] == 0x28)) return 3;
// gb-18030
if ((buf[0] == 0x84)
&& (buf[1] == 0x31)
&& (buf[2] == 0x95)
&& (buf[3] == 0x33)) return true;
if ((s[0] == 0x84)
&& (s[1] == 0x31)
&& (s[2] == 0x95)
&& (s[3] == 0x33)) return 4;
return false;
return 0;
}
bool hc_string_is_digit (const char *s)