| 1 |
|
| 2 |
#include <string> |
| 3 |
#include <sstream> |
| 4 |
#include <vector> |
| 5 |
#include <cctype> |
| 6 |
|
| 7 |
#include <iostream> |
| 8 |
#include <fstream> |
| 9 |
#include <iconv.h> |
| 10 |
#include <stdexcept> |
| 11 |
#include <time.h> |
| 12 |
#include <sys/time.h> |
| 13 |
#include <stdlib.h> |
| 14 |
|
| 15 |
#include <cstring> |
| 16 |
#include "util.h" |
| 17 |
|
| 18 |
using namespace std; |
| 19 |
|
| 20 |
namespace Util |
| 21 |
{ |
| 22 |
const int GN_CHAR_ALPHABET_SIZE = 128; |
| 23 |
unsigned char gsm_default_alphabet[GN_CHAR_ALPHABET_SIZE] = |
| 24 |
{ |
| 25 |
|
| 26 |
/* ETSI GSM 03.38, version 6.0.1, section 6.2.1; Default alphabet */ |
| 27 |
/* Characters in hex position 10, [12 to 1a] and 24 are not present on |
| 28 |
* latin1 charset, so we cannot reproduce on the screen, however they are |
| 29 |
* greek symbol not present even on my Nokia */ |
| 30 |
|
| 31 |
'@', 0xa3, '$', 0xa5, 0xe8, 0xe9, 0xf9, 0xec, |
| 32 |
0xf2, 0xc7, '\n', 0xd8, 0xf8, '\r', 0xc5, 0xe5, |
| 33 |
'?', '_', '?', '?', '?', '?', '?', '?', |
| 34 |
'?', '?', '?', '?', 0xc6, 0xe6, 0xdf, 0xc9, |
| 35 |
' ', '!', '\"', '#', 0xa4, '%', '&', '\'', |
| 36 |
'(', ')', '*', '+', ',', '-', '.', '/', |
| 37 |
'0', '1', '2', '3', '4', '5', '6', '7', |
| 38 |
'8', '9', ':', ';', '<', '=', '>', '?', |
| 39 |
0xa1, 'A', 'B', 'C', 'D', 'E', 'F', 'G', |
| 40 |
'H', 'I', 'J', 'K', 'L', 'M', 'N', 'O', |
| 41 |
'P', 'Q', 'R', 'S', 'T', 'U', 'V', 'W', |
| 42 |
'X', 'Y', 'Z', 0xc4, 0xd6, 0xd1, 0xdc, 0xa7, |
| 43 |
0xbf, 'a', 'b', 'c', 'd', 'e', 'f', 'g', |
| 44 |
'h', 'i', 'j', 'k', 'l', 'm', 'n', 'o', |
| 45 |
'p', 'q', 'r', 's', 't', 'u', 'v', 'w', |
| 46 |
'x', 'y', 'z', 0xe4, 0xf6, 0xf1, 0xfc, 0xe0 |
| 47 |
}; |
| 48 |
|
| 49 |
static unsigned char gsm_reverse_default_alphabet[256]; |
| 50 |
static bool reversed = false; |
| 51 |
|
| 52 |
static void tbl_setup_reverse() |
| 53 |
{ |
| 54 |
int i; |
| 55 |
|
| 56 |
if (reversed) return; |
| 57 |
memset(gsm_reverse_default_alphabet, 0x3f, 256); |
| 58 |
for (i = GN_CHAR_ALPHABET_SIZE - 1; i >= 0; i--) |
| 59 |
gsm_reverse_default_alphabet[ gsm_default_alphabet[i] ] = i; |
| 60 |
gsm_reverse_default_alphabet['?'] = 0x3f; |
| 61 |
reversed = true; |
| 62 |
} |
| 63 |
|
| 64 |
unsigned char char_def_alphabet_encode(unsigned char value) |
| 65 |
{ |
| 66 |
tbl_setup_reverse(); |
| 67 |
return gsm_reverse_default_alphabet[value]; |
| 68 |
} |
| 69 |
|
| 70 |
unsigned char char_def_alphabet_decode(unsigned char value) |
| 71 |
{ |
| 72 |
if (value < GN_CHAR_ALPHABET_SIZE) |
| 73 |
{ |
| 74 |
return gsm_default_alphabet[value]; |
| 75 |
} |
| 76 |
else |
| 77 |
{ |
| 78 |
return '?'; |
| 79 |
} |
| 80 |
} |
| 81 |
|
| 82 |
|
| 83 |
void str_dump(const string& str) |
| 84 |
{ |
| 85 |
for (unsigned i=0; i<str.length(); ++i) |
| 86 |
{ |
| 87 |
cout.width(2); |
| 88 |
cout.fill('0'); |
| 89 |
cout << hex << static_cast<int>(str.at(i)) << " "; |
| 90 |
} |
| 91 |
cout.width(1); |
| 92 |
cout << dec << endl; |
| 93 |
} |
| 94 |
|
| 95 |
|
| 96 |
|
| 97 |
string str_replace(string str, string search, string replace) |
| 98 |
{ |
| 99 |
unsigned int pos = 0; |
| 100 |
|
| 101 |
while ( (pos = str.find(search,pos)) != string::npos) |
| 102 |
{ |
| 103 |
str.replace(pos, search.size(), replace); |
| 104 |
pos += replace.size(); |
| 105 |
} |
| 106 |
return str; |
| 107 |
} |
| 108 |
|
| 109 |
std::string str_replace_char(std::string str, char search, char replace) |
| 110 |
{ |
| 111 |
unsigned int pos = 0; |
| 112 |
|
| 113 |
while ( (pos = str.find(search,pos)) != string::npos) |
| 114 |
{ |
| 115 |
str.replace(pos, 1, 1, replace); |
| 116 |
} |
| 117 |
return str; |
| 118 |
} |
| 119 |
|
| 120 |
bool my_isspace(char ch) |
| 121 |
{ |
| 122 |
return (isspace(ch) || ch == 0); |
| 123 |
} |
| 124 |
|
| 125 |
string str_trim(string str) |
| 126 |
{ |
| 127 |
while (str.length() > 0 && my_isspace(str.at(0))) |
| 128 |
str.erase(0,1); |
| 129 |
while (str.length() > 0 && my_isspace(str.at(str.length()-1))) |
| 130 |
str.erase(str.length()-1,1); |
| 131 |
return str; |
| 132 |
} |
| 133 |
|
| 134 |
string str_striptags(string str) |
| 135 |
{ |
| 136 |
unsigned int pos=0; |
| 137 |
|
| 138 |
while ( (pos=str.find("<",pos)) != string::npos) |
| 139 |
{ |
| 140 |
unsigned int endpos = str.find(">",pos); |
| 141 |
if (endpos == string::npos) |
| 142 |
break; |
| 143 |
str.erase(pos, (endpos-pos)+1); |
| 144 |
} |
| 145 |
|
| 146 |
return str; |
| 147 |
} |
| 148 |
|
| 149 |
vector<string> str_split(string input, string delimiter) |
| 150 |
{ |
| 151 |
vector<string> retval; |
| 152 |
|
| 153 |
while ( 1 ) |
| 154 |
{ |
| 155 |
unsigned int endpos = input.find(delimiter); |
| 156 |
|
| 157 |
string entry = input.substr(0,endpos); |
| 158 |
retval.push_back( entry ); |
| 159 |
|
| 160 |
|
| 161 |
if (endpos == string::npos) |
| 162 |
break; |
| 163 |
|
| 164 |
endpos += delimiter.length(); |
| 165 |
input = input.substr(endpos, input.length() - endpos); |
| 166 |
} |
| 167 |
|
| 168 |
return retval; |
| 169 |
} |
| 170 |
|
| 171 |
char my_toupper(char ch) |
| 172 |
{ |
| 173 |
if (ch == 'æ') |
| 174 |
return 'Æ'; |
| 175 |
if (ch == 'ø') |
| 176 |
return 'Ø'; |
| 177 |
if (ch == 'å') |
| 178 |
return 'Å'; |
| 179 |
|
| 180 |
return ::toupper(ch); |
| 181 |
} |
| 182 |
|
| 183 |
char my_tolower(char ch) |
| 184 |
{ |
| 185 |
if (ch == 'Æ') |
| 186 |
return 'æ'; |
| 187 |
if (ch == 'Ø') |
| 188 |
return 'ø'; |
| 189 |
if (ch == 'Å') |
| 190 |
return 'å'; |
| 191 |
|
| 192 |
return ::tolower(ch); |
| 193 |
} |
| 194 |
string str_toupper(string str) |
| 195 |
{ |
| 196 |
for (unsigned i=0; i<str.length(); ++i) |
| 197 |
{ |
| 198 |
str.replace(i, 1 ,1, my_toupper(str.at(i))); |
| 199 |
} |
| 200 |
return str; |
| 201 |
} |
| 202 |
|
| 203 |
string str_tolower(string str) |
| 204 |
{ |
| 205 |
for (unsigned i=0; i<str.length(); ++i) |
| 206 |
{ |
| 207 |
str.replace(i, 1 ,1, my_tolower(str.at(i))); |
| 208 |
} |
| 209 |
return str; |
| 210 |
} |
| 211 |
|
| 212 |
|
| 213 |
string str_formatint(int i) |
| 214 |
{ |
| 215 |
std::ostringstream os; |
| 216 |
os << i; |
| 217 |
return os.str(); |
| 218 |
} |
| 219 |
|
| 220 |
|
| 221 |
string str_characters(string str) |
| 222 |
{ |
| 223 |
string rep; |
| 224 |
rep.append(1,197); |
| 225 |
str = str_replace(str, "Å", rep); |
| 226 |
|
| 227 |
rep.at(0) = 198; |
| 228 |
str = str_replace(str, "Æ", rep); |
| 229 |
|
| 230 |
rep.at(0) = 216; |
| 231 |
str = str_replace(str, "Ø", rep); |
| 232 |
|
| 233 |
|
| 234 |
rep.at(0) = 229; |
| 235 |
str = str_replace(str, "å", rep); |
| 236 |
|
| 237 |
rep.at(0) = 230; |
| 238 |
str = str_replace(str, "æ", rep); |
| 239 |
|
| 240 |
rep.at(0) = 248; |
| 241 |
str = str_replace(str, "ø", rep); |
| 242 |
|
| 243 |
return str; |
| 244 |
} |
| 245 |
|
| 246 |
|
| 247 |
string str_gsm2latin(string str) |
| 248 |
{ |
| 249 |
for (unsigned i=0; i<str.size(); i++) { |
| 250 |
str.at(i) = char_def_alphabet_decode(str.at(i)); |
| 251 |
} |
| 252 |
return str; |
| 253 |
} |
| 254 |
|
| 255 |
string str_latin2gsm(string str) |
| 256 |
{ |
| 257 |
for (unsigned i=0; i<str.size(); i++) { |
| 258 |
str.at(i) = char_def_alphabet_encode(str.at(i)); |
| 259 |
} |
| 260 |
return str; |
| 261 |
} |
| 262 |
|
| 263 |
string readUrl(string url, string tempfile) |
| 264 |
{ |
| 265 |
char buf[128000]; |
| 266 |
string document; |
| 267 |
|
| 268 |
ostringstream command; |
| 269 |
command << "wget -O " << tempfile << " --tries=1 --timeout=15 -o /dev/null \"" << url << "\""; |
| 270 |
int res = my_system( command.str().c_str() ); |
| 271 |
|
| 272 |
if (res<0) |
| 273 |
{ |
| 274 |
throw( std::runtime_error("Error retrieving document")); |
| 275 |
} |
| 276 |
|
| 277 |
if (res>0) |
| 278 |
{ |
| 279 |
throw std::runtime_error("Command time out or document not found"); |
| 280 |
} |
| 281 |
|
| 282 |
ifstream in( tempfile.c_str() ); |
| 283 |
|
| 284 |
if (!in) |
| 285 |
return ""; |
| 286 |
|
| 287 |
while (!in.eof() ) |
| 288 |
{ |
| 289 |
in.getline(buf,128000-1); |
| 290 |
document += buf; |
| 291 |
document += "\n"; |
| 292 |
} |
| 293 |
in.close(); |
| 294 |
unlink(tempfile.c_str()); |
| 295 |
|
| 296 |
return document; |
| 297 |
} |
| 298 |
|
| 299 |
|
| 300 |
|
| 301 |
string iconv_wrapper(string _input, string to_format, string from_format) |
| 302 |
{ |
| 303 |
char* input,*output,*input_ptr, *output_ptr; |
| 304 |
input = input_ptr = (char*) malloc(_input.size()+1); |
| 305 |
strcpy(input, _input.c_str()); |
| 306 |
|
| 307 |
output = output_ptr = (char*) malloc(_input.size()*2); |
| 308 |
|
| 309 |
|
| 310 |
|
| 311 |
unsigned int realinsize,insize,outsize,realsize; |
| 312 |
|
| 313 |
iconv_t icv = iconv_open(to_format.c_str(), from_format.c_str()); |
| 314 |
if (icv == (iconv_t)-1) |
| 315 |
{ |
| 316 |
perror(0); |
| 317 |
return ""; |
| 318 |
} |
| 319 |
|
| 320 |
|
| 321 |
realsize = outsize = _input.size()*2; |
| 322 |
realinsize = insize = _input.size(); |
| 323 |
|
| 324 |
iconv(icv, |
| 325 |
&input_ptr, |
| 326 |
&insize, |
| 327 |
&output_ptr, |
| 328 |
&outsize); |
| 329 |
|
| 330 |
|
| 331 |
perror(0); |
| 332 |
/* cout << "len=" << len << endl; |
| 333 |
cout << "outsize=" << outsize << endl; |
| 334 |
cout << "realsize=" << realsize << endl; |
| 335 |
cout << "insize=" << insize << endl; |
| 336 |
cout << "realinsize=" << realinsize << endl;*/ |
| 337 |
iconv_close(icv); |
| 338 |
|
| 339 |
string returnstr; |
| 340 |
for (unsigned int i=0; i<realsize - outsize; i++) |
| 341 |
{ |
| 342 |
//cout << i << ":" << (unsigned short) output[i] << "|" << (unsigned short) input[i]; |
| 343 |
//cout << " (" << output[i] << ")" << endl; |
| 344 |
returnstr += output[i]; |
| 345 |
} |
| 346 |
return returnstr; |
| 347 |
} |
| 348 |
|
| 349 |
std::string convertToUnicode(std::string str) |
| 350 |
{ |
| 351 |
return iconv_wrapper(str, "UTF-8", "ISO8859-1"); |
| 352 |
} |
| 353 |
|
| 354 |
std::string convertFromUnicode(std::string str) |
| 355 |
{ |
| 356 |
return iconv_wrapper(str, "ISO8859-1", "UTF-8"); |
| 357 |
} |
| 358 |
|
| 359 |
int uTimeDiff(const timeval& then, const timeval& now) |
| 360 |
{ |
| 361 |
return ( (now.tv_sec - then.tv_sec)*1000000) + (now.tv_usec-then.tv_usec); |
| 362 |
} |
| 363 |
|
| 364 |
int mTimeDiff(const timeval& then, const timeval& now) |
| 365 |
{ |
| 366 |
return ( (now.tv_sec - then.tv_sec)*1000) + |
| 367 |
((now.tv_usec-then.tv_usec)/1000); |
| 368 |
} |
| 369 |
|
| 370 |
|
| 371 |
timeval GetTimeOfDay() |
| 372 |
{ |
| 373 |
timeval now; |
| 374 |
gettimeofday(&now,0); |
| 375 |
return now; |
| 376 |
} |
| 377 |
|
| 378 |
int my_system(const char* cmd, std::string* response) |
| 379 |
{ |
| 380 |
FILE* p; |
| 381 |
if ((p = popen(cmd,"r")) == NULL) |
| 382 |
return (-1); |
| 383 |
|
| 384 |
if (response) |
| 385 |
{ |
| 386 |
std::string output; |
| 387 |
char buf[256]; |
| 388 |
while (!feof(p)) |
| 389 |
{ |
| 390 |
int len = fread(buf,1,255, p); |
| 391 |
buf[len] = 0; |
| 392 |
output += buf; |
| 393 |
} |
| 394 |
*response = output; |
| 395 |
} |
| 396 |
|
| 397 |
return (pclose(p)); |
| 398 |
} |
| 399 |
|
| 400 |
string readfile(string filename) |
| 401 |
{ |
| 402 |
string str; |
| 403 |
ifstream in(filename.c_str()); |
| 404 |
if (in) |
| 405 |
{ |
| 406 |
char buffer[4096]; |
| 407 |
in.read(buffer, 4095); |
| 408 |
buffer[ in.gcount() ] = 0; |
| 409 |
str = string(buffer); |
| 410 |
in.close(); |
| 411 |
} |
| 412 |
else |
| 413 |
{ |
| 414 |
string message = "Could not open "; |
| 415 |
message += filename; |
| 416 |
throw std::runtime_error(message); |
| 417 |
} |
| 418 |
return str; |
| 419 |
} |
| 420 |
|
| 421 |
|
| 422 |
|
| 423 |
} |
Parent Directory
| 
Revision Log