xref: /AOO41X/main/tools/bootstrp/md5.cxx (revision cdf0e10c4e3984b49a9502b011690b615761d4a3)
1*cdf0e10cSrcweir /*************************************************************************
2*cdf0e10cSrcweir  *
3*cdf0e10cSrcweir  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
4*cdf0e10cSrcweir  *
5*cdf0e10cSrcweir  * Copyright 2000, 2010 Oracle and/or its affiliates.
6*cdf0e10cSrcweir  *
7*cdf0e10cSrcweir  * OpenOffice.org - a multi-platform office productivity suite
8*cdf0e10cSrcweir  *
9*cdf0e10cSrcweir  * This file is part of OpenOffice.org.
10*cdf0e10cSrcweir  *
11*cdf0e10cSrcweir  * OpenOffice.org is free software: you can redistribute it and/or modify
12*cdf0e10cSrcweir  * it under the terms of the GNU Lesser General Public License version 3
13*cdf0e10cSrcweir  * only, as published by the Free Software Foundation.
14*cdf0e10cSrcweir  *
15*cdf0e10cSrcweir  * OpenOffice.org is distributed in the hope that it will be useful,
16*cdf0e10cSrcweir  * but WITHOUT ANY WARRANTY; without even the implied warranty of
17*cdf0e10cSrcweir  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
18*cdf0e10cSrcweir  * GNU Lesser General Public License version 3 for more details
19*cdf0e10cSrcweir  * (a copy is included in the LICENSE file that accompanied this code).
20*cdf0e10cSrcweir  *
21*cdf0e10cSrcweir  * You should have received a copy of the GNU Lesser General Public License
22*cdf0e10cSrcweir  * version 3 along with OpenOffice.org.  If not, see
23*cdf0e10cSrcweir  * <http://www.openoffice.org/license.html>
24*cdf0e10cSrcweir  * for a copy of the LGPLv3 License.
25*cdf0e10cSrcweir  *
26*cdf0e10cSrcweir  ************************************************************************/
27*cdf0e10cSrcweir 
28*cdf0e10cSrcweir // MARKER(update_precomp.py): autogen include statement, do not remove
29*cdf0e10cSrcweir #include "precompiled_tools.hxx"
30*cdf0e10cSrcweir 
31*cdf0e10cSrcweir #include "md5.hxx"
32*cdf0e10cSrcweir 
33*cdf0e10cSrcweir #include <cstddef>
34*cdf0e10cSrcweir #include <stdio.h>
35*cdf0e10cSrcweir 
36*cdf0e10cSrcweir #include <tools/string.hxx>
37*cdf0e10cSrcweir 
38*cdf0e10cSrcweir #ifdef WNT
39*cdf0e10cSrcweir #define FILE_OPEN_READ	"rb"
40*cdf0e10cSrcweir #else
41*cdf0e10cSrcweir #define FILE_OPEN_READ	"r"
42*cdf0e10cSrcweir #endif
43*cdf0e10cSrcweir 
44*cdf0e10cSrcweir // Extended calc_md5_checksum to recognize Windows executables and libraries. To
45*cdf0e10cSrcweir // create the same md5 checksum for a (code/data) identical file it ignores a different
46*cdf0e10cSrcweir // date and header checksum. Please see crashrep/source/win32/soreport.cpp
47*cdf0e10cSrcweir // where the same method is also used. The crash reporter uses the MD5
48*cdf0e10cSrcweir // checksums to transfer them to the crash database. You have to make sure that both
49*cdf0e10cSrcweir // methods use the same algorithm otherwise there could be problems with stack reports.
50*cdf0e10cSrcweir 
51*cdf0e10cSrcweir void normalize_pe_image(sal_uInt8* buffer, size_t nBufferSize)
52*cdf0e10cSrcweir {
53*cdf0e10cSrcweir 	const int OFFSET_PE_OFFSET = 0x3c;
54*cdf0e10cSrcweir 	const int OFFSET_COFF_TIMEDATESTAMP = 4;
55*cdf0e10cSrcweir 	const int PE_SIGNATURE_SIZE = 4;
56*cdf0e10cSrcweir 	const int COFFHEADER_SIZE = 20;
57*cdf0e10cSrcweir 	const int OFFSET_PE_OPTIONALHEADER_CHECKSUM = 64;
58*cdf0e10cSrcweir 
59*cdf0e10cSrcweir 	// Check the header part of the file buffer
60*cdf0e10cSrcweir 	if (buffer[0] == sal_uInt8('M') && buffer[1] == sal_uInt8('Z'))
61*cdf0e10cSrcweir 	{
62*cdf0e10cSrcweir 		unsigned long PEHeaderOffset = (long)buffer[OFFSET_PE_OFFSET];
63*cdf0e10cSrcweir 		if (PEHeaderOffset < nBufferSize-4)
64*cdf0e10cSrcweir 		{
65*cdf0e10cSrcweir 			if ( buffer[PEHeaderOffset+0] == sal_uInt8('P') &&
66*cdf0e10cSrcweir 				 buffer[PEHeaderOffset+1] == sal_uInt8('E') &&
67*cdf0e10cSrcweir 				 buffer[PEHeaderOffset+2] == 0 &&
68*cdf0e10cSrcweir 				 buffer[PEHeaderOffset+3] == 0 )
69*cdf0e10cSrcweir 			{
70*cdf0e10cSrcweir 				PEHeaderOffset += PE_SIGNATURE_SIZE;
71*cdf0e10cSrcweir 				if (PEHeaderOffset+OFFSET_COFF_TIMEDATESTAMP < nBufferSize-4)
72*cdf0e10cSrcweir 				{
73*cdf0e10cSrcweir 					// Set timedatestamp and checksum fields to a normalized
74*cdf0e10cSrcweir 					// value to enforce the same MD5 checksum for identical
75*cdf0e10cSrcweir 					// Windows	executables/libraries.
76*cdf0e10cSrcweir 					buffer[PEHeaderOffset+OFFSET_COFF_TIMEDATESTAMP+0] = 0;
77*cdf0e10cSrcweir 					buffer[PEHeaderOffset+OFFSET_COFF_TIMEDATESTAMP+1] = 0;
78*cdf0e10cSrcweir 					buffer[PEHeaderOffset+OFFSET_COFF_TIMEDATESTAMP+2] = 0;
79*cdf0e10cSrcweir 					buffer[PEHeaderOffset+OFFSET_COFF_TIMEDATESTAMP+3] = 0;
80*cdf0e10cSrcweir 				}
81*cdf0e10cSrcweir 
82*cdf0e10cSrcweir 				if (PEHeaderOffset+COFFHEADER_SIZE+OFFSET_PE_OPTIONALHEADER_CHECKSUM < nBufferSize-4)
83*cdf0e10cSrcweir 				{
84*cdf0e10cSrcweir 					// Set checksum to a normalized value
85*cdf0e10cSrcweir 					buffer[PEHeaderOffset+COFFHEADER_SIZE+OFFSET_PE_OPTIONALHEADER_CHECKSUM] = 0;
86*cdf0e10cSrcweir 					buffer[PEHeaderOffset+COFFHEADER_SIZE+OFFSET_PE_OPTIONALHEADER_CHECKSUM+1] = 0;
87*cdf0e10cSrcweir 					buffer[PEHeaderOffset+COFFHEADER_SIZE+OFFSET_PE_OPTIONALHEADER_CHECKSUM+2] = 0;
88*cdf0e10cSrcweir 					buffer[PEHeaderOffset+COFFHEADER_SIZE+OFFSET_PE_OPTIONALHEADER_CHECKSUM+3] = 0;
89*cdf0e10cSrcweir 				}
90*cdf0e10cSrcweir 			}
91*cdf0e10cSrcweir 		}
92*cdf0e10cSrcweir 	}
93*cdf0e10cSrcweir }
94*cdf0e10cSrcweir 
95*cdf0e10cSrcweir rtlDigestError calc_md5_checksum( const char *filename, ByteString &aChecksum )
96*cdf0e10cSrcweir {
97*cdf0e10cSrcweir     const size_t BUFFER_SIZE  = 0x1000;
98*cdf0e10cSrcweir 	const size_t MINIMAL_SIZE = 512;
99*cdf0e10cSrcweir 
100*cdf0e10cSrcweir 	sal_uInt8 checksum[RTL_DIGEST_LENGTH_MD5];
101*cdf0e10cSrcweir 	rtlDigestError	error = rtl_Digest_E_None;
102*cdf0e10cSrcweir 
103*cdf0e10cSrcweir 	FILE *fp = fopen( filename, FILE_OPEN_READ );
104*cdf0e10cSrcweir 
105*cdf0e10cSrcweir 	if ( fp )
106*cdf0e10cSrcweir 	{
107*cdf0e10cSrcweir 		rtlDigest digest = rtl_digest_createMD5();
108*cdf0e10cSrcweir 
109*cdf0e10cSrcweir 		if ( digest )
110*cdf0e10cSrcweir 		{
111*cdf0e10cSrcweir 			size_t			nBytesRead;
112*cdf0e10cSrcweir 			sal_uInt8		buffer[BUFFER_SIZE];
113*cdf0e10cSrcweir 			bool			bHeader(true);
114*cdf0e10cSrcweir 
115*cdf0e10cSrcweir 			while ( rtl_Digest_E_None == error &&
116*cdf0e10cSrcweir 				0 != (nBytesRead = fread( buffer, 1, sizeof(buffer), fp )) )
117*cdf0e10cSrcweir 			{
118*cdf0e10cSrcweir 				if (bHeader)
119*cdf0e10cSrcweir 				{
120*cdf0e10cSrcweir 					bHeader = false;
121*cdf0e10cSrcweir 					if (nBytesRead >= MINIMAL_SIZE && buffer[0] == sal_uInt8('M') && buffer[1] == sal_uInt8('Z') )
122*cdf0e10cSrcweir 						normalize_pe_image(buffer, nBytesRead);
123*cdf0e10cSrcweir 				}
124*cdf0e10cSrcweir 
125*cdf0e10cSrcweir 				error = rtl_digest_updateMD5( digest, buffer, nBytesRead );
126*cdf0e10cSrcweir 			}
127*cdf0e10cSrcweir 
128*cdf0e10cSrcweir 			if ( rtl_Digest_E_None == error )
129*cdf0e10cSrcweir 			{
130*cdf0e10cSrcweir 				error = rtl_digest_getMD5( digest, checksum, sizeof(checksum) );
131*cdf0e10cSrcweir 			}
132*cdf0e10cSrcweir 
133*cdf0e10cSrcweir 			rtl_digest_destroyMD5( digest );
134*cdf0e10cSrcweir 
135*cdf0e10cSrcweir             for ( std::size_t i = 0; i < sizeof(checksum); i++ )
136*cdf0e10cSrcweir             {
137*cdf0e10cSrcweir                 if ( checksum[i] < 16 )
138*cdf0e10cSrcweir                     aChecksum.Append( "0" );
139*cdf0e10cSrcweir                 aChecksum += ByteString::CreateFromInt32( checksum[i], 16 );
140*cdf0e10cSrcweir             }
141*cdf0e10cSrcweir 		}
142*cdf0e10cSrcweir 
143*cdf0e10cSrcweir 		fclose( fp );
144*cdf0e10cSrcweir 	}
145*cdf0e10cSrcweir 	else
146*cdf0e10cSrcweir 		error = rtl_Digest_E_Unknown;
147*cdf0e10cSrcweir 
148*cdf0e10cSrcweir 	return error;
149*cdf0e10cSrcweir }
150