1 // ==============================================================
2 //	This file is part of Glest Shared Library (www.glest.org)
3 //
4 //	Copyright (C) 2001-2008 Martiño Figueroa
5 //
6 //	You can redistribute this code and/or modify it under
7 //	the terms of the GNU General Public License as published
8 //	by the Free Software Foundation; either version 2 of the
9 //	License, or (at your option) any later version
10 // ==============================================================
11 
12 #include "checksum.h"
13 
14 #include <cassert>
15 #include <stdexcept>
16 #include <fcntl.h> // for open()
17 
18 #ifdef WIN32
19   #include <io.h> // for open()
20 #endif
21 
22 #include <sys/stat.h> // for open()
23 
24 #include "util.h"
25 #include "platform_common.h"
26 #include "conversion.h"
27 #include "platform_util.h"
28 #include "leak_dumper.h"
29 
30 using namespace std;
31 using namespace Shared::PlatformCommon;
32 using namespace Shared::Util;
33 
34 namespace Shared{ namespace Util{
35 
36 // =====================================================
37 //	class Checksum
38 // =====================================================
39 
40 Mutex Checksum::fileListCacheSynchAccessor;
41 std::map<string,uint32> Checksum::fileListCache;
42 
43 unsigned int crc_table[256] =
44 {
45 	0x00000000, 0x77073096, 0xee0e612c, 0x990951ba, 0x076dc419, 0x706af48f, 0xe963a535, 0x9e6495a3,
46 	0x0edb8832, 0x79dcb8a4, 0xe0d5e91e, 0x97d2d988, 0x09b64c2b, 0x7eb17cbd, 0xe7b82d07, 0x90bf1d91,
47 	0x1db71064, 0x6ab020f2, 0xf3b97148, 0x84be41de, 0x1adad47d, 0x6ddde4eb, 0xf4d4b551, 0x83d385c7,
48 	0x136c9856, 0x646ba8c0, 0xfd62f97a, 0x8a65c9ec, 0x14015c4f, 0x63066cd9, 0xfa0f3d63, 0x8d080df5,
49 	0x3b6e20c8, 0x4c69105e, 0xd56041e4, 0xa2677172, 0x3c03e4d1, 0x4b04d447, 0xd20d85fd, 0xa50ab56b,
50 	0x35b5a8fa, 0x42b2986c, 0xdbbbc9d6, 0xacbcf940, 0x32d86ce3, 0x45df5c75, 0xdcd60dcf, 0xabd13d59,
51 	0x26d930ac, 0x51de003a, 0xc8d75180, 0xbfd06116, 0x21b4f4b5, 0x56b3c423, 0xcfba9599, 0xb8bda50f,
52 	0x2802b89e, 0x5f058808, 0xc60cd9b2, 0xb10be924, 0x2f6f7c87, 0x58684c11, 0xc1611dab, 0xb6662d3d,
53 	0x76dc4190, 0x01db7106, 0x98d220bc, 0xefd5102a, 0x71b18589, 0x06b6b51f, 0x9fbfe4a5, 0xe8b8d433,
54 	0x7807c9a2, 0x0f00f934, 0x9609a88e, 0xe10e9818, 0x7f6a0dbb, 0x086d3d2d, 0x91646c97, 0xe6635c01,
55 	0x6b6b51f4, 0x1c6c6162, 0x856530d8, 0xf262004e, 0x6c0695ed, 0x1b01a57b, 0x8208f4c1, 0xf50fc457,
56 	0x65b0d9c6, 0x12b7e950, 0x8bbeb8ea, 0xfcb9887c, 0x62dd1ddf, 0x15da2d49, 0x8cd37cf3, 0xfbd44c65,
57 	0x4db26158, 0x3ab551ce, 0xa3bc0074, 0xd4bb30e2, 0x4adfa541, 0x3dd895d7, 0xa4d1c46d, 0xd3d6f4fb,
58 	0x4369e96a, 0x346ed9fc, 0xad678846, 0xda60b8d0, 0x44042d73, 0x33031de5, 0xaa0a4c5f, 0xdd0d7cc9,
59 	0x5005713c, 0x270241aa, 0xbe0b1010, 0xc90c2086, 0x5768b525, 0x206f85b3, 0xb966d409, 0xce61e49f,
60 	0x5edef90e, 0x29d9c998, 0xb0d09822, 0xc7d7a8b4, 0x59b33d17, 0x2eb40d81, 0xb7bd5c3b, 0xc0ba6cad,
61 	0xedb88320, 0x9abfb3b6, 0x03b6e20c, 0x74b1d29a, 0xead54739, 0x9dd277af, 0x04db2615, 0x73dc1683,
62 	0xe3630b12, 0x94643b84, 0x0d6d6a3e, 0x7a6a5aa8, 0xe40ecf0b, 0x9309ff9d, 0x0a00ae27, 0x7d079eb1,
63 	0xf00f9344, 0x8708a3d2, 0x1e01f268, 0x6906c2fe, 0xf762575d, 0x806567cb, 0x196c3671, 0x6e6b06e7,
64 	0xfed41b76, 0x89d32be0, 0x10da7a5a, 0x67dd4acc, 0xf9b9df6f, 0x8ebeeff9, 0x17b7be43, 0x60b08ed5,
65 	0xd6d6a3e8, 0xa1d1937e, 0x38d8c2c4, 0x4fdff252, 0xd1bb67f1, 0xa6bc5767, 0x3fb506dd, 0x48b2364b,
66 	0xd80d2bda, 0xaf0a1b4c, 0x36034af6, 0x41047a60, 0xdf60efc3, 0xa867df55, 0x316e8eef, 0x4669be79,
67 	0xcb61b38c, 0xbc66831a, 0x256fd2a0, 0x5268e236, 0xcc0c7795, 0xbb0b4703, 0x220216b9, 0x5505262f,
68 	0xc5ba3bbe, 0xb2bd0b28, 0x2bb45a92, 0x5cb36a04, 0xc2d7ffa7, 0xb5d0cf31, 0x2cd99e8b, 0x5bdeae1d,
69 	0x9b64c2b0, 0xec63f226, 0x756aa39c, 0x026d930a, 0x9c0906a9, 0xeb0e363f, 0x72076785, 0x05005713,
70 	0x95bf4a82, 0xe2b87a14, 0x7bb12bae, 0x0cb61b38, 0x92d28e9b, 0xe5d5be0d, 0x7cdcefb7, 0x0bdbdf21,
71 	0x86d3d2d4, 0xf1d4e242, 0x68ddb3f8, 0x1fda836e, 0x81be16cd, 0xf6b9265b, 0x6fb077e1, 0x18b74777,
72 	0x88085ae6, 0xff0f6a70, 0x66063bca, 0x11010b5c, 0x8f659eff, 0xf862ae69, 0x616bffd3, 0x166ccf45,
73 	0xa00ae278, 0xd70dd2ee, 0x4e048354, 0x3903b3c2, 0xa7672661, 0xd06016f7, 0x4969474d, 0x3e6e77db,
74 	0xaed16a4a, 0xd9d65adc, 0x40df0b66, 0x37d83bf0, 0xa9bcae53, 0xdebb9ec5, 0x47b2cf7f, 0x30b5ffe9,
75 	0xbdbdf21c, 0xcabac28a, 0x53b39330, 0x24b4a3a6, 0xbad03605, 0xcdd70693, 0x54de5729, 0x23d967bf,
76 	0xb3667a2e, 0xc4614ab8, 0x5d681b02, 0x2a6f2b94, 0xb40bbe37, 0xc30c8ea1, 0x5a05df1b, 0x2d02ef8d
77 };
78 
Checksum()79 Checksum::Checksum() {
80 	sum= 0;
81 	r= 55665;
82 	c1= 52845;
83 	c2= 22719;
84 }
85 
addByte(const char value)86 uint32 Checksum::addByte(const char value) {
87 //	int32 cipher= (value ^ (r >> 8));
88 //
89 //	r= (cipher + r) * c1 + c2;
90 //	sum += cipher;
91 //
92 //	return cipher;
93 
94 	const unsigned char *rVal = reinterpret_cast<const unsigned char *>(&value);
95 	sum = ~sum;
96 	sum = (sum >> 8) ^ crc_table[*rVal ^ (sum & 0xff)];
97 	sum = ~sum;
98 
99 	return sum;
100 }
101 
addBytes(const void * _data,size_t _size)102 uint32 Checksum::addBytes(const void *_data, size_t _size) {
103 	const unsigned char *rVal = reinterpret_cast<const unsigned char *>(_data);
104 	sum = ~sum;
105 	while (_size--) {
106 		sum = (sum >> 8) ^ crc_table[*rVal++ ^ (sum & 0xff)];
107 	}
108 	sum = ~sum;
109 
110 	return sum;
111 }
112 
113 
addSum(uint32 value)114 void Checksum::addSum(uint32 value) {
115 	sum += value;
116 }
117 
addInt(const int32 & value)118 uint32 Checksum::addInt(const int32 &value) {
119 	int8 byte 	= (value >>  0) & 0xFF;
120 	addByte(byte);
121 	byte  		= (value >>  8) & 0xFF;
122 	addByte(byte);
123 	byte 		= (value >> 16) & 0xFF;
124 	addByte(byte);
125 	byte 		= (value >> 24) & 0xFF;
126 	addByte(byte);
127 
128 	return sum;
129 }
130 
addUInt(const uint32 & value)131 uint32 Checksum::addUInt(const uint32 &value) {
132 	int8 byte 	= (value >>  0) & 0xFF;
133 	addByte(byte);
134 	byte  		= (value >>  8) & 0xFF;
135 	addByte(byte);
136 	byte 		= (value >> 16) & 0xFF;
137 	addByte(byte);
138 	byte 		= (value >> 24) & 0xFF;
139 	addByte(byte);
140 
141 	return sum;
142 }
143 
addInt64(const int64 & value)144 uint32 Checksum::addInt64(const int64 &value) {
145 	int8 byte 	= (value >>  0) & 0xFF;
146 	addByte(byte);
147 	byte  		= (value >>  8) & 0xFF;
148 	addByte(byte);
149 	byte 		= (value >> 16) & 0xFF;
150 	addByte(byte);
151 	byte 		= (value >> 24) & 0xFF;
152 	addByte(byte);
153 	byte 		= (value >> 32) & 0xFF;
154 	addByte(byte);
155 	byte 		= (value >> 40) & 0xFF;
156 	addByte(byte);
157 	byte 		= (value >> 48) & 0xFF;
158 	addByte(byte);
159 	byte 		= (value >> 56) & 0xFF;
160 	addByte(byte);
161 
162 	return sum;
163 }
164 
addString(const string & value)165 void Checksum::addString(const string &value) {
166 	for(unsigned int i = 0; i < value.size(); ++i) {
167 		addByte(value[i]);
168 	}
169 }
170 
addFile(const string & path)171 void Checksum::addFile(const string &path) {
172 	if(path != "") {
173 		fileList[path] = 0;
174 	}
175 }
176 
addFileToSum(const string & path)177 bool Checksum::addFileToSum(const string &path) {
178 
179 // OLD SLOW FILE I/O
180 /*
181 	FILE* file= fopen(path.c_str(), "rb");
182 	if(file!=NULL){
183 
184 		addString(lastFile(path));
185 
186 		while(!feof(file)){
187 			int8 byte= 0;
188 
189 			size_t readBytes = fread(&byte, 1, 1, file);
190 			addByte(byte);
191 		}
192 	}
193 	else
194 	{
195 		throw megaglest_runtime_error("Can not open file: " + path);
196 	}
197 	fclose(file);
198 */
199 
200 
201 
202 /*
203    const double MAX_CRC_FILESIZE = 100000000;
204    int fd=0;
205    size_t bytes_read, bytes_expected = MAX_CRC_FILESIZE * sizeof(int8);
206    int8 *data=0;
207    const char *infile = path.c_str();
208 
209    if ((fd = open(infile,O_RDONLY)) < 0)
210 	   throw megaglest_runtime_error("Can not open file: " + path);
211 
212    if ((data = (int8 *)malloc(bytes_expected)) == NULL)
213 	   throw megaglest_runtime_error("malloc failed, Can not open file: " + path);
214 
215    bytes_read = read(fd, data, bytes_expected);
216 
217    //if (bytes_read != bytes_expected)
218    //   throw megaglest_runtime_error("read failed, Can not open file: " + path);
219 
220    for(int i = 0; i < bytes_read; i++) {
221 		addByte(data[i]);
222    }
223    free(data);
224 */
225 
226     bool fileExists = false;
227 
228 /*
229 #ifdef WIN32
230 	FILE* file= _wfopen(utf8_decode(path).c_str(), L"rb");
231 #else
232 	FILE* file= fopen(path.c_str(), "rb");
233 #endif
234 	if(file != NULL) {
235         fileExists = true;
236 		addString(lastFile(path));
237 
238 		bool isXMLFile = (EndsWith(path, ".xml") == true);
239 		bool inCommentTag=false;
240 		char buf[4096]="";  // Should be large enough.
241 		int bufSize = sizeof buf;
242 		while(!feof(file)) {
243 			//int8 byte= 0;
244 
245 			//size_t readBytes = fread(&byte, 1, 1, file);
246 			memset(buf,0,bufSize);
247 			if(fgets(buf, bufSize, file) != NULL) {
248 				//addByte(byte);
249 			    for(int i = 0; i < bufSize && buf[i] != 0; i++) {
250 			    	// Ignore Spaces in XML files as they are
251 			    	// ONLY for formatting
252 			    	if(isXMLFile == true) {
253 			    		if(inCommentTag == true) {
254 			    			if(buf[i] == '>' && i >= 3 && buf[i-1] == '-' && buf[i-2] == '-') {
255 			    				inCommentTag = false;
256 			    				//printf("TURNING OFF comment TAG, i = %d [%c]",i,buf[i]);
257 			    			}
258 			    			else {
259 			    				//printf("SKIPPING XML comment character, i = %d [%c]",i,buf[i]);
260 			    			}
261 			    			continue;
262 			    		}
263 			    		//else if(buf[i] == '-' && i >= 4 && buf[i-1] == '-' && buf[i-2] == '!' && buf[i-3] == '<') {
264 			    		else if(buf[i] == '<' && i+4 < bufSize && buf[i+1] == '!' && buf[i+2] == '-' && buf[i+3] == '-') {
265 		    				inCommentTag = true;
266 		    				//printf("TURNING ON comment TAG, i = %d [%c]",i,buf[i]);
267 		    				continue;
268 			    		}
269 			    		else if(buf[i] == ' ' || buf[i] == '\t' || buf[i] == '\n' || buf[i] == '\r') {
270 			    			//printf("SKIPPING special character, i = %d [%c]",i,buf[i]);
271 			    			continue;
272 			    	    }
273 			    	}
274 			 		addByte(buf[i]);
275 			    }
276 			}
277 		}
278 	}
279 	else {
280 		throw megaglest_runtime_error("Can not open file: " + path);
281 	}
282 	fclose(file);
283 */
284 
285 #if defined(WIN32) && !defined(__MINGW32__)
286 	wstring wstr = utf8_decode(path);
287 	FILE *fp = _wfopen(wstr.c_str(), L"rb");
288 	ifstream ifs(fp);
289 #else
290     ifstream ifs(path.c_str());
291 #endif
292 
293     if (ifs) {
294         fileExists = true;
295 		addString(lastFile(path));
296 
297 		bool isXMLFile = (EndsWith(path, ".xml") == true);
298 
299 		// Determine the file length
300 		ifs.seekg(0, ios::end);
301 		std::streamoff size=ifs.tellg();
302 		ifs.seekg(0, ios::beg);
303 
304 		unsigned int bufSize = (unsigned int)size / sizeof(char);
305 		// Create a vector to store the data
306 		std::vector<char> buf(bufSize);
307 		// Load the data
308 		ifs.read((char*)&buf[0], buf.size());
309 
310 		if(SystemFlags::getSystemSettingType(SystemFlags::debugSystem).enabled) SystemFlags::OutputDebug(SystemFlags::debugSystem,"In [%s::%s Line: %d] buf.size() = %d, path [%s], isXMLFile = %d\n",__FILE__,__FUNCTION__,__LINE__,buf.size(), path.c_str(),isXMLFile);
311 
312 		if(isXMLFile == true) {
313 			bool inCommentTag=false;
314 			for(std::size_t i = 0; i < buf.size(); ++i) {
315 				// Ignore Spaces in XML files as they are
316 				// ONLY for formatting
317 				//if(isXMLFile == true) {
318 					if(inCommentTag == true) {
319 						if(buf[i] == '>' && i >= 3 && buf[i-1] == '-' && buf[i-2] == '-') {
320 							inCommentTag = false;
321 							//printf("TURNING OFF comment TAG, i = %d [%c]",i,buf[i]);
322 							if(SystemFlags::getSystemSettingType(SystemFlags::debugSystem).enabled) SystemFlags::OutputDebug(SystemFlags::debugSystem,"In [%s::%s Line: %d] i = %d\n",__FILE__,__FUNCTION__,__LINE__,i);
323 						}
324 						else {
325 							//printf("SKIPPING XML comment character, i = %d [%c]",i,buf[i]);
326 							if(SystemFlags::getSystemSettingType(SystemFlags::debugSystem).enabled) SystemFlags::OutputDebug(SystemFlags::debugSystem,"In [%s::%s Line: %d] i = %d\n",__FILE__,__FUNCTION__,__LINE__,i);
327 						}
328 						continue;
329 					}
330 					//else if(buf[i] == '-' && i >= 4 && buf[i-1] == '-' && buf[i-2] == '!' && buf[i-3] == '<') {
331 					else if(buf[i] == '<' && i+4 < bufSize && buf[i+1] == '!' && buf[i+2] == '-' && buf[i+3] == '-') {
332 						inCommentTag = true;
333 						//printf("TURNING ON comment TAG, i = %d [%c]",i,buf[i]);
334 						if(SystemFlags::getSystemSettingType(SystemFlags::debugSystem).enabled) SystemFlags::OutputDebug(SystemFlags::debugSystem,"In [%s::%s Line: %d] i = %d\n",__FILE__,__FUNCTION__,__LINE__,i);
335 						continue;
336 					}
337 					else if(buf[i] == ' ' || buf[i] == '\t' || buf[i] == '\n' || buf[i] == '\r') {
338 						//printf("SKIPPING special character, i = %d [%c]",i,buf[i]);
339 						if(SystemFlags::getSystemSettingType(SystemFlags::debugSystem).enabled) SystemFlags::OutputDebug(SystemFlags::debugSystem,"In [%s::%s Line: %d] i = %d\n",__FILE__,__FUNCTION__,__LINE__,i);
340 						continue;
341 					}
342 				//}
343 				uint32 cipher = addByte(buf[i]);
344 
345 				if(SystemFlags::getSystemSettingType(SystemFlags::debugSystem).enabled) SystemFlags::OutputDebug(SystemFlags::debugSystem,"In [%s::%s Line: %d] %d / %d, cipher = %u\n",__FILE__,__FUNCTION__,__LINE__,i,buf.size(), cipher);
346 			}
347 		}
348 		else {
349 			uint32 cipher = addBytes(&buf[0],buf.size());
350 			if(SystemFlags::getSystemSettingType(SystemFlags::debugSystem).enabled) SystemFlags::OutputDebug(SystemFlags::debugSystem,"In [%s::%s Line: %d] %d, cipher = %u\n",__FILE__,__FUNCTION__,__LINE__,buf.size(), cipher);
351 		}
352 
353 		// Close the file
354 		ifs.close();
355     }
356 #if defined(WIN32) && !defined(__MINGW32__)
357 	if(fp) {
358 		fclose(fp);
359 	}
360 #endif
361 
362     return fileExists;
363 }
364 
getSum()365 uint32 Checksum::getSum() {
366 	//printf("Getting checksum for files [%d]\n",fileList.size());
367 	if(fileList.size() > 0) {
368 		if(SystemFlags::getSystemSettingType(SystemFlags::debugSystem).enabled) SystemFlags::OutputDebug(SystemFlags::debugSystem,"In [%s::%s Line: %d] fileList.size() = %d\n",__FILE__,__FUNCTION__,__LINE__,fileList.size());
369 
370 		Checksum newResult;
371 
372 		{
373 
374 		for(std::map<string,uint32>::iterator iterMap = fileList.begin();
375 			iterMap != fileList.end(); ++iterMap) {
376 
377 			MutexSafeWrapper safeMutexSocketDestructorFlag(&Checksum::fileListCacheSynchAccessor,string(__FILE__) + "_" + intToStr(__LINE__));
378 			if(Checksum::fileListCache.find(iterMap->first) == Checksum::fileListCache.end()) {
379 				Checksum fileResult;
380 				//bool fileAddedOk = fileResult.addFileToSum(iterMap->first);
381 				fileResult.addFileToSum(iterMap->first);
382 				Checksum::fileListCache[iterMap->first] = fileResult.getSum();
383 				//printf("fileAddedOk = %d for file [%s] CRC [%d]\n",fileAddedOk,iterMap->first.c_str(),Checksum::fileListCache[iterMap->first]);
384 			}
385 			else {
386 				//printf("Getting checksum from CACHE for file [%s] CRC [%d]\n",iterMap->first.c_str(),Checksum::fileListCache[iterMap->first]);
387 			}
388 			newResult.addSum(Checksum::fileListCache[iterMap->first]);
389 		}
390 		}
391 
392 		if(SystemFlags::getSystemSettingType(SystemFlags::debugSystem).enabled) SystemFlags::OutputDebug(SystemFlags::debugSystem,"In [%s::%s Line: %d] fileList.size() = %d\n",__FILE__,__FUNCTION__,__LINE__,fileList.size());
393 
394 		return newResult.getSum();
395 	}
396 	return sum;
397 }
398 
getFinalFileListSum()399 uint32 Checksum::getFinalFileListSum() {
400 	sum = 0;
401 	return getSum();
402 }
403 
404 //uint32 Checksum::getFinalFileListSum() {
405 //	if(SystemFlags::getSystemSettingType(SystemFlags::debugSystem).enabled) SystemFlags::OutputDebug(SystemFlags::debugSystem,"In [%s::%s Line: %d] sum = %u\n",extractFileFromDirectoryPath(__FILE__).c_str(),__FUNCTION__,__LINE__,sum);
406 //
407 //	sum = 0;
408 //	//return getSum();
409 //	sum = getSum();
410 //
411 //	if(SystemFlags::getSystemSettingType(SystemFlags::debugSystem).enabled) SystemFlags::OutputDebug(SystemFlags::debugSystem,"In [%s::%s Line: %d] sum = %u\n",extractFileFromDirectoryPath(__FILE__).c_str(),__FUNCTION__,__LINE__,sum);
412 //
413 //	return sum;
414 //}
415 
getFileCount()416 uint32 Checksum::getFileCount() {
417 	return (uint32)fileList.size();
418 }
419 
removeFileFromCache(const string file)420 void Checksum::removeFileFromCache(const string file) {
421 	MutexSafeWrapper safeMutexSocketDestructorFlag(&Checksum::fileListCacheSynchAccessor,string(__FILE__) + "_" + intToStr(__LINE__));
422     if(Checksum::fileListCache.find(file) != Checksum::fileListCache.end()) {
423         Checksum::fileListCache.erase(file);
424     }
425 }
426 
clearFileCache()427 void Checksum::clearFileCache() {
428 	MutexSafeWrapper safeMutexSocketDestructorFlag(&Checksum::fileListCacheSynchAccessor,string(__FILE__) + "_" + intToStr(__LINE__));
429     Checksum::fileListCache.clear();
430 }
431 
432 }}//end namespace
433