10 |
#include <stdexcept> |
#include <stdexcept> |
11 |
#include <time.h> |
#include <time.h> |
12 |
#include <sys/time.h> |
#include <sys/time.h> |
13 |
|
#include <stdlib.h> |
14 |
|
|
15 |
#include "util.h" |
#include <cstring> |
16 |
|
#include "Util.h" |
17 |
|
|
18 |
using namespace std; |
using namespace std; |
19 |
|
|
20 |
namespace Util |
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) |
void str_dump(const string& str) |
84 |
{ |
{ |
85 |
for (unsigned i=0; i<str.length(); ++i) |
for (unsigned i=0; i<str.length(); ++i) |
86 |
{ |
{ |
87 |
cout.width(2); |
cout.width(2); |
88 |
cout.fill('0'); |
cout.fill('0'); |
89 |
cout << hex << static_cast<int>(str.at(i)) << " "; |
cout << hex << static_cast<int>(str.at(i)) << " "; |
97 |
string str_replace(string str, string search, string replace) |
string str_replace(string str, string search, string replace) |
98 |
{ |
{ |
99 |
unsigned int pos = 0; |
unsigned int pos = 0; |
100 |
|
|
101 |
while ( (pos = str.find(search,pos)) != string::npos) |
while ( (pos = str.find(search,pos)) != string::npos) |
102 |
{ |
{ |
103 |
str.replace(pos, search.size(), replace); |
str.replace(pos, search.size(), replace); |
106 |
return str; |
return str; |
107 |
} |
} |
108 |
|
|
109 |
std::string str_replace_char(std::string str, char search, char replace) |
std::string str_replace_char(std::string str, char search, char replace) |
110 |
{ |
{ |
111 |
unsigned int pos = 0; |
unsigned int pos = 0; |
112 |
|
|
146 |
return str; |
return str; |
147 |
} |
} |
148 |
|
|
149 |
|
|
150 |
|
vector<string> str_split(string input) |
151 |
|
{ |
152 |
|
string buf; |
153 |
|
stringstream ss(input); |
154 |
|
vector<string> tokens; |
155 |
|
while (ss >> buf) |
156 |
|
tokens.push_back(buf); |
157 |
|
return tokens; |
158 |
|
} |
159 |
|
|
160 |
|
|
161 |
vector<string> str_split(string input, string delimiter) |
vector<string> str_split(string input, string delimiter) |
162 |
{ |
{ |
163 |
vector<string> retval; |
vector<string> retval; |
164 |
|
|
165 |
while ( 1 ) |
while ( 1 ) |
166 |
{ |
{ |
180 |
return retval; |
return retval; |
181 |
} |
} |
182 |
|
|
183 |
|
char danish_map[3][2] = { {198,230}, {216,248}, {197,229} }; // aelig, oslash, aring |
184 |
|
|
185 |
string str_toupper(string str) |
unsigned char my_toupper(unsigned char ch) |
186 |
|
{ |
187 |
|
if (ch == 230) |
188 |
|
return 198; |
189 |
|
if (ch == 248) |
190 |
|
return 216; |
191 |
|
if (ch == 229) |
192 |
|
return 197; |
193 |
|
|
194 |
|
return ::toupper(ch); |
195 |
|
} |
196 |
|
|
197 |
|
unsigned char my_tolower(unsigned char ch) |
198 |
|
{ |
199 |
|
if (ch == 198) |
200 |
|
return 230; |
201 |
|
if (ch == 216) |
202 |
|
return 248; |
203 |
|
if (ch == 197) |
204 |
|
return 229; |
205 |
|
|
206 |
|
return ::tolower(ch); |
207 |
|
} |
208 |
|
string str_toupper(string str) |
209 |
{ |
{ |
210 |
for (unsigned i=0; i<str.length(); ++i) |
for (unsigned i=0; i<str.length(); ++i) |
211 |
{ |
{ |
212 |
str.replace(i, 1 ,1, ::toupper(str.at(i))); |
str.replace(i, 1 ,1, my_toupper(str.at(i))); |
213 |
} |
} |
214 |
return str; |
return str; |
215 |
} |
} |
218 |
{ |
{ |
219 |
for (unsigned i=0; i<str.length(); ++i) |
for (unsigned i=0; i<str.length(); ++i) |
220 |
{ |
{ |
221 |
str.replace(i, 1 ,1, ::tolower(str.at(i))); |
str.replace(i, 1 ,1, my_tolower(str.at(i))); |
222 |
} |
} |
223 |
return str; |
return str; |
224 |
} |
} |
237 |
string rep; |
string rep; |
238 |
rep.append(1,197); |
rep.append(1,197); |
239 |
str = str_replace(str, "Å", rep); |
str = str_replace(str, "Å", rep); |
240 |
|
str = str_replace(str, "Å", rep); |
241 |
|
|
242 |
rep.at(0) = 198; |
rep.at(0) = 198; |
243 |
str = str_replace(str, "Æ", rep); |
str = str_replace(str, "Æ", rep); |
244 |
|
str = str_replace(str, "Æ", rep); |
245 |
|
|
246 |
rep.at(0) = 216; |
rep.at(0) = 216; |
247 |
str = str_replace(str, "Ø", rep); |
str = str_replace(str, "Ø", rep); |
248 |
|
str = str_replace(str, "Ø", rep); |
249 |
|
|
250 |
|
|
251 |
rep.at(0) = 229; |
rep.at(0) = 229; |
252 |
str = str_replace(str, "å", rep); |
str = str_replace(str, "å", rep); |
253 |
|
str = str_replace(str, "å", rep); |
254 |
|
|
255 |
rep.at(0) = 230; |
rep.at(0) = 230; |
256 |
str = str_replace(str, "æ", rep); |
str = str_replace(str, "æ", rep); |
257 |
|
str = str_replace(str, "æ", rep); |
258 |
|
|
259 |
rep.at(0) = 248; |
rep.at(0) = 248; |
260 |
str = str_replace(str, "ø", rep); |
str = str_replace(str, "ø", rep); |
261 |
|
str = str_replace(str, "ø", rep); |
262 |
|
|
263 |
return str; |
return str; |
264 |
} |
} |
266 |
|
|
267 |
string str_gsm2latin(string str) |
string str_gsm2latin(string str) |
268 |
{ |
{ |
269 |
str = str_replace_char(str, 0x1c, 198); //AE |
for (unsigned i=0; i<str.size(); i++) { |
270 |
str = str_replace_char(str, 0x0b, 216); //OE |
str.at(i) = char_def_alphabet_decode(str.at(i)); |
271 |
str = str_replace_char(str, 0x0e, 197); //AA |
} |
|
|
|
|
str = str_replace_char(str, 0x1d, 230); //ae |
|
|
str = str_replace_char(str, 0x0c, 248); //oe |
|
|
str = str_replace_char(str, 0x0f, 229); //aa |
|
272 |
return str; |
return str; |
273 |
} |
} |
274 |
|
|
275 |
string str_latin2gsm(string str) |
string str_latin2gsm(string str) |
276 |
{ |
{ |
277 |
str = str_replace_char(str, 198, 0x1c); //AE |
for (unsigned i=0; i<str.size(); i++) { |
278 |
str = str_replace_char(str, 216, 0x0b); //OE |
str.at(i) = char_def_alphabet_encode(str.at(i)); |
279 |
str = str_replace_char(str, 197, 0x0e); //AA |
} |
|
|
|
|
str = str_replace_char(str, 230, 0x1d); //ae |
|
|
str = str_replace_char(str, 248, 0x0c); //oe |
|
|
str = str_replace_char(str, 229, 0x0f); //aa |
|
280 |
return str; |
return str; |
281 |
} |
} |
282 |
|
|
284 |
{ |
{ |
285 |
char buf[128000]; |
char buf[128000]; |
286 |
string document; |
string document; |
287 |
|
|
288 |
ostringstream command; |
ostringstream command; |
289 |
command << "wget -O " << tempfile << " --tries=1 --timeout=15 -o /dev/null \"" << url << "\""; |
command << "wget -O " << tempfile << " --tries=1 --timeout=15 -o /dev/null \"" << url << "\""; |
290 |
int res = my_system( command.str().c_str() ); |
int res = my_system( command.str().c_str() ); |
294 |
throw( std::runtime_error("Error retrieving document")); |
throw( std::runtime_error("Error retrieving document")); |
295 |
} |
} |
296 |
|
|
297 |
if (res>0) |
if (res>0) |
298 |
{ |
{ |
299 |
throw std::runtime_error("Command time out or document not found"); |
throw std::runtime_error("Command time out or document not found"); |
300 |
} |
} |
312 |
} |
} |
313 |
in.close(); |
in.close(); |
314 |
unlink(tempfile.c_str()); |
unlink(tempfile.c_str()); |
315 |
|
|
316 |
return document; |
return document; |
317 |
} |
} |
318 |
|
|
319 |
|
|
320 |
|
|
321 |
string iconv_wrapper(string _input, string to_format, string from_format) |
string iconv_wrapper(string _input, string to_format, string from_format) |
322 |
{ |
{ |
323 |
char* input,*output,*input_ptr, *output_ptr; |
char* input,*output,*input_ptr, *output_ptr; |
324 |
input = input_ptr = (char*) malloc(_input.size()+1); |
input = input_ptr = (char*) malloc(_input.size()+1); |
325 |
strcpy(input, _input.c_str()); |
strcpy(input, _input.c_str()); |
326 |
|
|
327 |
output = output_ptr = (char*) malloc(_input.size()*2); |
output = output_ptr = (char*) malloc(_input.size()*2); |
328 |
|
|
329 |
|
|
330 |
|
|
331 |
unsigned int realinsize,insize,outsize,realsize; |
unsigned int realinsize,insize,outsize,realsize; |
332 |
|
|
333 |
iconv_t icv = iconv_open(to_format.c_str(), from_format.c_str()); |
iconv_t icv = iconv_open(to_format.c_str(), from_format.c_str()); |
334 |
if (icv == (iconv_t)-1) |
if (icv == (iconv_t)-1) |
335 |
{ |
{ |
337 |
return ""; |
return ""; |
338 |
} |
} |
339 |
|
|
340 |
|
|
341 |
realsize = outsize = _input.size()*2; |
realsize = outsize = _input.size()*2; |
342 |
realinsize = insize = _input.size(); |
realinsize = insize = _input.size(); |
343 |
|
|
344 |
iconv(icv, |
iconv(icv, |
345 |
&input_ptr, |
&input_ptr, |
346 |
&insize, |
&insize, |
347 |
&output_ptr, |
&output_ptr, |
348 |
&outsize); |
&outsize); |
349 |
|
|
350 |
|
|
351 |
perror(0); |
perror(0); |
352 |
/* cout << "len=" << len << endl; |
/* cout << "len=" << len << endl; |
353 |
cout << "outsize=" << outsize << endl; |
cout << "outsize=" << outsize << endl; |
354 |
cout << "realsize=" << realsize << endl; |
cout << "realsize=" << realsize << endl; |
355 |
cout << "insize=" << insize << endl; |
cout << "insize=" << insize << endl; |
356 |
cout << "realinsize=" << realinsize << endl;*/ |
cout << "realinsize=" << realinsize << endl;*/ |
357 |
iconv_close(icv); |
iconv_close(icv); |
358 |
|
|
359 |
string returnstr; |
string returnstr; |
363 |
//cout << " (" << output[i] << ")" << endl; |
//cout << " (" << output[i] << ")" << endl; |
364 |
returnstr += output[i]; |
returnstr += output[i]; |
365 |
} |
} |
366 |
return returnstr; |
return returnstr; |
367 |
} |
} |
368 |
|
|
369 |
std::string convertToUnicode(std::string str) |
std::string convertToUnicode(std::string str) |
384 |
int mTimeDiff(const timeval& then, const timeval& now) |
int mTimeDiff(const timeval& then, const timeval& now) |
385 |
{ |
{ |
386 |
return ( (now.tv_sec - then.tv_sec)*1000) + |
return ( (now.tv_sec - then.tv_sec)*1000) + |
387 |
((now.tv_usec-then.tv_usec)/1000); |
((now.tv_usec-then.tv_usec)/1000); |
388 |
} |
} |
389 |
|
|
390 |
|
|
405 |
{ |
{ |
406 |
std::string output; |
std::string output; |
407 |
char buf[256]; |
char buf[256]; |
408 |
while(!feof(p)) |
while (!feof(p)) |
409 |
{ |
{ |
410 |
int len = fread(buf,1,255, p); |
int len = fread(buf,1,255, p); |
411 |
buf[len] = 0; |
buf[len] = 0; |
417 |
return (pclose(p)); |
return (pclose(p)); |
418 |
} |
} |
419 |
|
|
420 |
} |
string readfile(string filename) |
421 |
|
{ |
422 |
|
string str; |
423 |
|
ifstream in(filename.c_str()); |
424 |
|
if (in) |
425 |
|
{ |
426 |
|
char buffer[4096]; |
427 |
|
in.read(buffer, 4095); |
428 |
|
buffer[ in.gcount() ] = 0; |
429 |
|
str = string(buffer); |
430 |
|
in.close(); |
431 |
|
} |
432 |
|
else |
433 |
|
{ |
434 |
|
string message = "Could not open "; |
435 |
|
message += filename; |
436 |
|
throw std::runtime_error(message); |
437 |
|
} |
438 |
|
return str; |
439 |
|
} |
440 |
|
|
441 |
|
|
442 |
|
|
443 |
|
} |
444 |
|
|