1 |
torben |
26 |
|
2 |
|
|
#include <string> |
3 |
|
|
#include <sstream> |
4 |
|
|
#include <vector> |
5 |
torben |
38 |
#include <cctype> |
6 |
|
|
|
7 |
torben |
26 |
#include <iostream> |
8 |
|
|
#include <fstream> |
9 |
|
|
#include <iconv.h> |
10 |
|
|
|
11 |
|
|
#include <time.h> |
12 |
|
|
|
13 |
|
|
#include "util.h" |
14 |
|
|
|
15 |
|
|
using namespace std; |
16 |
|
|
|
17 |
|
|
namespace Util |
18 |
|
|
{ |
19 |
|
|
void str_dump(const string& str) |
20 |
|
|
{ |
21 |
|
|
for (unsigned i=0; i<str.length(); ++i) |
22 |
|
|
{ |
23 |
|
|
cout.width(2); |
24 |
|
|
cout.fill('0'); |
25 |
|
|
cout << hex << static_cast<int>(str.at(i)) << " "; |
26 |
|
|
} |
27 |
|
|
cout.width(1); |
28 |
|
|
cout << dec << endl; |
29 |
|
|
} |
30 |
|
|
|
31 |
|
|
|
32 |
|
|
|
33 |
torben |
38 |
string str_replace(string str, string search, string replace) |
34 |
torben |
26 |
{ |
35 |
|
|
unsigned int pos = 0; |
36 |
|
|
|
37 |
torben |
38 |
while ( (pos = str.find(search,pos)) != string::npos) |
38 |
torben |
26 |
{ |
39 |
torben |
38 |
str.replace(pos, search.size(), replace); |
40 |
torben |
26 |
pos += replace.size(); |
41 |
|
|
} |
42 |
torben |
38 |
return str; |
43 |
torben |
26 |
} |
44 |
|
|
|
45 |
torben |
42 |
std::string str_replace_char(std::string str, char search, char replace) |
46 |
|
|
{ |
47 |
|
|
unsigned int pos = 0; |
48 |
|
|
|
49 |
|
|
while ( (pos = str.find(search,pos)) != string::npos) |
50 |
|
|
{ |
51 |
|
|
str.replace(pos, 1, 1, replace); |
52 |
|
|
} |
53 |
|
|
return str; |
54 |
|
|
} |
55 |
|
|
|
56 |
torben |
38 |
|
57 |
|
|
|
58 |
|
|
string str_trim(string str) |
59 |
|
|
{ |
60 |
|
|
while (str.length() > 0 && isspace(str.at(0))) |
61 |
|
|
str.erase(0,1); |
62 |
|
|
while (str.length() > 0 && isspace(str.at(str.length()-1))) |
63 |
|
|
str.erase(str.length()-1,1); |
64 |
|
|
return str; |
65 |
|
|
} |
66 |
|
|
|
67 |
|
|
string str_striptags(string str) |
68 |
|
|
{ |
69 |
|
|
unsigned int pos=0; |
70 |
|
|
|
71 |
|
|
while ( (pos=str.find("<",pos)) != string::npos) |
72 |
|
|
{ |
73 |
|
|
unsigned int endpos = str.find(">",pos); |
74 |
|
|
if (endpos == string::npos) |
75 |
|
|
break; |
76 |
|
|
str.erase(pos, (endpos-pos)+1); |
77 |
|
|
} |
78 |
|
|
|
79 |
|
|
return str; |
80 |
|
|
} |
81 |
|
|
|
82 |
torben |
26 |
vector<string> str_split(string input, string delimiter) |
83 |
|
|
{ |
84 |
|
|
vector<string> retval; |
85 |
|
|
|
86 |
|
|
while ( 1 ) |
87 |
|
|
{ |
88 |
|
|
unsigned int endpos = input.find(delimiter); |
89 |
|
|
|
90 |
|
|
string entry = input.substr(0,endpos); |
91 |
|
|
retval.push_back( entry ); |
92 |
|
|
|
93 |
|
|
|
94 |
|
|
if (endpos == string::npos) |
95 |
|
|
break; |
96 |
|
|
|
97 |
|
|
endpos += delimiter.length(); |
98 |
|
|
input = input.substr(endpos, input.length() - endpos); |
99 |
|
|
} |
100 |
|
|
|
101 |
|
|
return retval; |
102 |
|
|
} |
103 |
|
|
|
104 |
torben |
38 |
string str_characters(string str) |
105 |
|
|
{ |
106 |
|
|
string rep; |
107 |
|
|
rep.append(1,197); |
108 |
|
|
str = str_replace(str, "Å", rep); |
109 |
|
|
|
110 |
|
|
rep.at(0) = 198; |
111 |
|
|
str = str_replace(str, "Æ", rep); |
112 |
|
|
|
113 |
|
|
rep.at(0) = 216; |
114 |
|
|
str = str_replace(str, "Ø", rep); |
115 |
|
|
|
116 |
|
|
|
117 |
|
|
rep.at(0) = 229; |
118 |
|
|
str = str_replace(str, "å", rep); |
119 |
|
|
|
120 |
|
|
rep.at(0) = 230; |
121 |
|
|
str = str_replace(str, "æ", rep); |
122 |
|
|
|
123 |
|
|
rep.at(0) = 248; |
124 |
|
|
str = str_replace(str, "ø", rep); |
125 |
|
|
|
126 |
|
|
return str; |
127 |
|
|
} |
128 |
|
|
|
129 |
torben |
42 |
|
130 |
|
|
string str_gsm2latin1(string str) |
131 |
|
|
{ |
132 |
|
|
str = str_replace_char(str, 0x1c, 198); //AE |
133 |
|
|
str = str_replace_char(str, 0x0b, 216); //OE |
134 |
|
|
str = str_replace_char(str, 0x0e, 197); //AA |
135 |
|
|
|
136 |
|
|
str = str_replace_char(str, 0x1d, 230); //ae |
137 |
|
|
str = str_replace_char(str, 0x0c, 248); //oe |
138 |
|
|
str = str_replace_char(str, 0x0f, 229); //aa |
139 |
|
|
return str; |
140 |
|
|
} |
141 |
|
|
|
142 |
torben |
26 |
string readUrl(string url, string tempfile) |
143 |
|
|
{ |
144 |
|
|
char buf[128000]; |
145 |
|
|
string document; |
146 |
|
|
|
147 |
|
|
ostringstream command; |
148 |
|
|
command << "wget -O " << tempfile << " -o /dev/null \"" << url << "\""; |
149 |
|
|
system( command.str().c_str() ); |
150 |
|
|
|
151 |
|
|
ifstream in( tempfile.c_str() ); |
152 |
|
|
|
153 |
|
|
if (!in) |
154 |
|
|
return ""; |
155 |
|
|
|
156 |
|
|
while (!in.eof() ) |
157 |
|
|
{ |
158 |
|
|
in.getline(buf,128000-1); |
159 |
|
|
document += buf; |
160 |
|
|
document += "\n"; |
161 |
|
|
} |
162 |
|
|
in.close(); |
163 |
|
|
unlink(tempfile.c_str()); |
164 |
|
|
|
165 |
|
|
return document; |
166 |
|
|
} |
167 |
|
|
|
168 |
torben |
42 |
|
169 |
|
|
|
170 |
|
|
string iconv_wrapper(string _input, string to_format, string from_format) |
171 |
torben |
26 |
{ |
172 |
|
|
char* input,*output,*input_ptr, *output_ptr; |
173 |
|
|
input = input_ptr = (char*) malloc(_input.size()+1); |
174 |
|
|
strcpy(input, _input.c_str()); |
175 |
|
|
|
176 |
|
|
output = output_ptr = (char*) malloc(_input.size()*2); |
177 |
|
|
|
178 |
|
|
|
179 |
|
|
|
180 |
|
|
unsigned int realinsize,insize,outsize,realsize; |
181 |
|
|
|
182 |
torben |
42 |
iconv_t icv = iconv_open(to_format.c_str(), from_format.c_str()); |
183 |
torben |
26 |
if (icv == (iconv_t)-1) |
184 |
|
|
{ |
185 |
|
|
perror(0); |
186 |
|
|
return ""; |
187 |
|
|
} |
188 |
|
|
|
189 |
|
|
|
190 |
|
|
realsize = outsize = _input.size()*2; |
191 |
|
|
realinsize = insize = _input.size(); |
192 |
|
|
|
193 |
|
|
iconv(icv, |
194 |
|
|
&input_ptr, |
195 |
|
|
&insize, |
196 |
|
|
&output_ptr, |
197 |
|
|
&outsize); |
198 |
|
|
|
199 |
|
|
|
200 |
|
|
perror(0); |
201 |
|
|
/* cout << "len=" << len << endl; |
202 |
|
|
cout << "outsize=" << outsize << endl; |
203 |
|
|
cout << "realsize=" << realsize << endl; |
204 |
|
|
cout << "insize=" << insize << endl; |
205 |
|
|
cout << "realinsize=" << realinsize << endl;*/ |
206 |
|
|
iconv_close(icv); |
207 |
|
|
|
208 |
|
|
string returnstr; |
209 |
|
|
for (unsigned int i=0; i<realsize - outsize; i++) |
210 |
|
|
{ |
211 |
|
|
//cout << i << ":" << (unsigned short) output[i] << "|" << (unsigned short) input[i]; |
212 |
|
|
//cout << " (" << output[i] << ")" << endl; |
213 |
|
|
returnstr += output[i]; |
214 |
|
|
} |
215 |
|
|
return returnstr; |
216 |
|
|
} |
217 |
|
|
|
218 |
torben |
42 |
std::string convertToUnicode(std::string str) |
219 |
|
|
{ |
220 |
|
|
return iconv_wrapper(str, "UTF-8", "ISO8859-1"); |
221 |
|
|
} |
222 |
|
|
|
223 |
|
|
std::string convertFromUnicode(std::string str) |
224 |
|
|
{ |
225 |
|
|
return iconv_wrapper(str, "ISO8859-1", "UTF-8"); |
226 |
|
|
} |
227 |
|
|
|
228 |
torben |
26 |
int uTimeDiff(const timeval& then, const timeval& now) |
229 |
|
|
{ |
230 |
|
|
return ( (now.tv_sec - then.tv_sec)*1000000) + (now.tv_usec-then.tv_usec); |
231 |
|
|
} |
232 |
|
|
|
233 |
|
|
int mTimeDiff(const timeval& then, const timeval& now) |
234 |
|
|
{ |
235 |
|
|
return uTimeDiff(then,now) / 1000; |
236 |
|
|
} |
237 |
|
|
|
238 |
|
|
} |