From e40b86a86ca236898ae61bc0d93b196a84a537b6 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Jean-Pierre=20Andr=C3=A9?= Date: Fri, 17 Apr 2015 11:03:58 +0200 Subject: [PATCH] Upgraded the upper-case table as defined by Windows 7 Newer versions of Windows use more recent definitions of upper-case table defined by the Unicode consortium. Now using the same table as Windows 7, windows 8 and Windows 10. This only has an effect on file systems newly created by mkntfs. --- include/ntfs-3g/param.h | 8 ++ libntfs-3g/unistr.c | 162 +++++++++++++++++++++++++--------------- 2 files changed, 108 insertions(+), 62 deletions(-) diff --git a/include/ntfs-3g/param.h b/include/ntfs-3g/param.h index da794abb..f67193c4 100644 --- a/include/ntfs-3g/param.h +++ b/include/ntfs-3g/param.h @@ -76,6 +76,14 @@ enum { /* only update the final extent of a runlist when appending data */ #define PARTIAL_RUNLIST_UPDATING 1 +/* + * Parameters for upper-case table + */ + + /* Create upper-case tables as defined by Windows 6.1 (Win7) */ +#define UPCASE_MAJOR 6 +#define UPCASE_MINOR 1 + /* * Parameters for user and xattr mappings */ diff --git a/libntfs-3g/unistr.c b/libntfs-3g/unistr.c index e12d21e7..94215fff 100644 --- a/libntfs-3g/unistr.c +++ b/libntfs-3g/unistr.c @@ -3,7 +3,7 @@ * * Copyright (c) 2000-2004 Anton Altaparmakov * Copyright (c) 2002-2009 Szabolcs Szakacsits - * Copyright (c) 2008-2014 Jean-Pierre Andre + * Copyright (c) 2008-2015 Jean-Pierre Andre * Copyright (c) 2008 Bernhard Kaindl * * This program/include file is free software; you can redistribute it and/or @@ -1128,66 +1128,15 @@ char *ntfs_uppercase_mbs(const char *low, */ void ntfs_upcase_table_build(ntfschar *uc, u32 uc_len) { -#if 1 /* Vista */ - /* - * This is the table as defined by Vista - */ - /* - * "Start" is inclusive and "End" is exclusive, every value has the - * value of "Add" added to it. - */ - static int uc_run_table[][3] = { /* Start, End, Add */ - {0x0061, 0x007b, -32}, {0x00e0, 0x00f7, -32}, {0x00f8, 0x00ff, -32}, - {0x0256, 0x0258, -205}, {0x028a, 0x028c, -217}, {0x037b, 0x037e, 130}, - {0x03ac, 0x03ad, -38}, {0x03ad, 0x03b0, -37}, {0x03b1, 0x03c2, -32}, - {0x03c2, 0x03c3, -31}, {0x03c3, 0x03cc, -32}, {0x03cc, 0x03cd, -64}, - {0x03cd, 0x03cf, -63}, {0x0430, 0x0450, -32}, {0x0450, 0x0460, -80}, - {0x0561, 0x0587, -48}, {0x1f00, 0x1f08, 8}, {0x1f10, 0x1f16, 8}, - {0x1f20, 0x1f28, 8}, {0x1f30, 0x1f38, 8}, {0x1f40, 0x1f46, 8}, - {0x1f51, 0x1f52, 8}, {0x1f53, 0x1f54, 8}, {0x1f55, 0x1f56, 8}, - {0x1f57, 0x1f58, 8}, {0x1f60, 0x1f68, 8}, {0x1f70, 0x1f72, 74}, - {0x1f72, 0x1f76, 86}, {0x1f76, 0x1f78, 100}, {0x1f78, 0x1f7a, 128}, - {0x1f7a, 0x1f7c, 112}, {0x1f7c, 0x1f7e, 126}, {0x1f80, 0x1f88, 8}, - {0x1f90, 0x1f98, 8}, {0x1fa0, 0x1fa8, 8}, {0x1fb0, 0x1fb2, 8}, - {0x1fb3, 0x1fb4, 9}, {0x1fcc, 0x1fcd, -9}, {0x1fd0, 0x1fd2, 8}, - {0x1fe0, 0x1fe2, 8}, {0x1fe5, 0x1fe6, 7}, {0x1ffc, 0x1ffd, -9}, - {0x2170, 0x2180, -16}, {0x24d0, 0x24ea, -26}, {0x2c30, 0x2c5f, -48}, - {0x2d00, 0x2d26, -7264}, {0xff41, 0xff5b, -32}, {0} - }; - /* - * "Start" is exclusive and "End" is inclusive, every second value is - * decremented by one. - */ - static int uc_dup_table[][2] = { /* Start, End */ - {0x0100, 0x012f}, {0x0132, 0x0137}, {0x0139, 0x0149}, {0x014a, 0x0178}, - {0x0179, 0x017e}, {0x01a0, 0x01a6}, {0x01b3, 0x01b7}, {0x01cd, 0x01dd}, - {0x01de, 0x01ef}, {0x01f4, 0x01f5}, {0x01f8, 0x01f9}, {0x01fa, 0x0220}, - {0x0222, 0x0234}, {0x023b, 0x023c}, {0x0241, 0x0242}, {0x0246, 0x024f}, - {0x03d8, 0x03ef}, {0x03f7, 0x03f8}, {0x03fa, 0x03fb}, {0x0460, 0x0481}, - {0x048a, 0x04bf}, {0x04c1, 0x04c4}, {0x04c5, 0x04c8}, {0x04c9, 0x04ce}, - {0x04ec, 0x04ed}, {0x04d0, 0x04eb}, {0x04ee, 0x04f5}, {0x04f6, 0x0513}, - {0x1e00, 0x1e95}, {0x1ea0, 0x1ef9}, {0x2183, 0x2184}, {0x2c60, 0x2c61}, - {0x2c67, 0x2c6c}, {0x2c75, 0x2c76}, {0x2c80, 0x2ce3}, {0} - }; - /* - * Set the Unicode character at offset "Offset" to "Value". Note, - * "Value" is host endian. - */ - static int uc_byte_table[][2] = { /* Offset, Value */ - {0x00ff, 0x0178}, {0x0180, 0x0243}, {0x0183, 0x0182}, {0x0185, 0x0184}, - {0x0188, 0x0187}, {0x018c, 0x018b}, {0x0192, 0x0191}, {0x0195, 0x01f6}, - {0x0199, 0x0198}, {0x019a, 0x023d}, {0x019e, 0x0220}, {0x01a8, 0x01a7}, - {0x01ad, 0x01ac}, {0x01b0, 0x01af}, {0x01b9, 0x01b8}, {0x01bd, 0x01bc}, - {0x01bf, 0x01f7}, {0x01c6, 0x01c4}, {0x01c9, 0x01c7}, {0x01cc, 0x01ca}, - {0x01dd, 0x018e}, {0x01f3, 0x01f1}, {0x023a, 0x2c65}, {0x023e, 0x2c66}, - {0x0253, 0x0181}, {0x0254, 0x0186}, {0x0259, 0x018f}, {0x025b, 0x0190}, - {0x0260, 0x0193}, {0x0263, 0x0194}, {0x0268, 0x0197}, {0x0269, 0x0196}, - {0x026b, 0x2c62}, {0x026f, 0x019c}, {0x0272, 0x019d}, {0x0275, 0x019f}, - {0x027d, 0x2c64}, {0x0280, 0x01a6}, {0x0283, 0x01a9}, {0x0288, 0x01ae}, - {0x0289, 0x0244}, {0x028c, 0x0245}, {0x0292, 0x01b7}, {0x03f2, 0x03f9}, - {0x04cf, 0x04c0}, {0x1d7d, 0x2c63}, {0x214e, 0x2132}, {0} - }; -#else /* Vista */ + struct NEWUPPERCASE { + unsigned short first; + unsigned short last; + short diff; + unsigned char step; + unsigned char osmajor; + unsigned char osminor; + } ; + /* * This is the table as defined by Windows XP */ @@ -1227,9 +1176,88 @@ void ntfs_upcase_table_build(ntfschar *uc, u32 uc_len) {0x01A8, 0x01A7}, {0x01DD, 0x018E}, {0x0268, 0x0197}, {0} }; -#endif /* Vista */ + +/* + * Changes which were applied to later Windows versions + * + * md5 for $UpCase from Winxp : 6fa3db2468275286210751e869d36373 + * Vista : 2f03b5a69d486ff3864cecbd07f24440 + * Win8 : 7ff498a44e45e77374cc7c962b1b92f2 + */ + static const struct NEWUPPERCASE newuppercase[] = { + /* from Windows 6.0 (Vista) */ + { 0x37b, 0x37d, 0x82, 1, 6, 0 }, + { 0x1f80, 0x1f87, 0x8, 1, 6, 0 }, + { 0x1f90, 0x1f97, 0x8, 1, 6, 0 }, + { 0x1fa0, 0x1fa7, 0x8, 1, 6, 0 }, + { 0x2c30, 0x2c5e, -0x30, 1, 6, 0 }, + { 0x2d00, 0x2d25, -0x1c60, 1, 6, 0 }, + { 0x2c68, 0x2c6c, -0x1, 2, 6, 0 }, + { 0x219, 0x21f, -0x1, 2, 6, 0 }, + { 0x223, 0x233, -0x1, 2, 6, 0 }, + { 0x247, 0x24f, -0x1, 2, 6, 0 }, + { 0x3d9, 0x3e1, -0x1, 2, 6, 0 }, + { 0x48b, 0x48f, -0x1, 2, 6, 0 }, + { 0x4fb, 0x513, -0x1, 2, 6, 0 }, + { 0x2c81, 0x2ce3, -0x1, 2, 6, 0 }, + { 0x3f8, 0x3fb, -0x1, 3, 6, 0 }, + { 0x4c6, 0x4ce, -0x1, 4, 6, 0 }, + { 0x23c, 0x242, -0x1, 6, 6, 0 }, + { 0x4ed, 0x4f7, -0x1, 10, 6, 0 }, + { 0x450, 0x45d, -0x50, 13, 6, 0 }, + { 0x2c61, 0x2c76, -0x1, 21, 6, 0 }, + { 0x1fcc, 0x1ffc, -0x9, 48, 6, 0 }, + { 0x180, 0x180, 0xc3, 1, 6, 0 }, + { 0x195, 0x195, 0x61, 1, 6, 0 }, + { 0x19a, 0x19a, 0xa3, 1, 6, 0 }, + { 0x19e, 0x19e, 0x82, 1, 6, 0 }, + { 0x1bf, 0x1bf, 0x38, 1, 6, 0 }, + { 0x1f9, 0x1f9, -0x1, 1, 6, 0 }, + { 0x23a, 0x23a, 0x2a2b, 1, 6, 0 }, + { 0x23e, 0x23e, 0x2a28, 1, 6, 0 }, + { 0x26b, 0x26b, 0x29f7, 1, 6, 0 }, + { 0x27d, 0x27d, 0x29e7, 1, 6, 0 }, + { 0x280, 0x280, -0xda, 1, 6, 0 }, + { 0x289, 0x289, -0x45, 1, 6, 0 }, + { 0x28c, 0x28c, -0x47, 1, 6, 0 }, + { 0x3f2, 0x3f2, 0x7, 1, 6, 0 }, + { 0x4cf, 0x4cf, -0xf, 1, 6, 0 }, + { 0x1d7d, 0x1d7d, 0xee6, 1, 6, 0 }, + { 0x1fb3, 0x1fb3, 0x9, 1, 6, 0 }, + { 0x214e, 0x214e, -0x1c, 1, 6, 0 }, + { 0x2184, 0x2184, -0x1, 1, 6, 0 }, + /* from Windows 6.1 (Win7) */ + { 0x23a, 0x23e, 0x0, 4, 6, 1 }, + { 0x250, 0x250, 0x2a1f, 2, 6, 1 }, + { 0x251, 0x251, 0x2a1c, 2, 6, 1 }, + { 0x271, 0x271, 0x29fd, 2, 6, 1 }, + { 0x371, 0x373, -0x1, 2, 6, 1 }, + { 0x377, 0x377, -0x1, 2, 6, 1 }, + { 0x3c2, 0x3c2, 0x0, 2, 6, 1 }, + { 0x3d7, 0x3d7, -0x8, 2, 6, 1 }, + { 0x515, 0x523, -0x1, 2, 6, 1 }, + { 0x1d79, 0x1d79, 0x8a04, 2, 6, 1 }, + { 0x1efb, 0x1eff, -0x1, 2, 6, 1 }, + { 0x1fc3, 0x1ff3, 0x9, 48, 6, 1 }, + { 0x1fcc, 0x1ffc, 0x0, 48, 6, 1 }, + { 0x2c65, 0x2c65, -0x2a2b, 2, 6, 1 }, + { 0x2c66, 0x2c66, -0x2a28, 2, 6, 1 }, + { 0x2c73, 0x2c73, -0x1, 2, 6, 1 }, + { 0xa641, 0xa65f, -0x1, 2, 6, 1 }, + { 0xa663, 0xa66d, -0x1, 2, 6, 1 }, + { 0xa681, 0xa697, -0x1, 2, 6, 1 }, + { 0xa723, 0xa72f, -0x1, 2, 6, 1 }, + { 0xa733, 0xa76f, -0x1, 2, 6, 1 }, + { 0xa77a, 0xa77c, -0x1, 2, 6, 1 }, + { 0xa77f, 0xa787, -0x1, 2, 6, 1 }, + { 0xa78c, 0xa78c, -0x1, 2, 6, 1 }, + /* end mark */ + { 0 } + } ; + int i, r; int k, off; + const struct NEWUPPERCASE *puc; memset((char*)uc, 0, uc_len); uc_len >>= 1; @@ -1249,6 +1277,16 @@ void ntfs_upcase_table_build(ntfschar *uc, u32 uc_len) k = uc_byte_table[r][1]; uc[uc_byte_table[r][0]] = cpu_to_le16(k); } + for (r=0; newuppercase[r].first; r++) { + puc = &newuppercase[r]; + if ((puc->osmajor < UPCASE_MAJOR) + || ((puc->osmajor == UPCASE_MAJOR) + && (puc->osminor <= UPCASE_MINOR))) { + off = puc->diff; + for (i = puc->first; i <= puc->last; i += puc->step) + uc[i] = cpu_to_le16(i + off); + } + } } /*