|
a/src/common/unacpp.cpp |
|
b/src/common/unacpp.cpp |
|
... |
|
... |
| 23 |
#include <cstdlib>
|
23 |
#include <cstdlib>
|
| 24 |
#include <errno.h>
|
24 |
#include <errno.h>
|
| 25 |
|
25 |
|
| 26 |
#include <string>
|
26 |
#include <string>
|
| 27 |
|
27 |
|
| 28 |
#ifndef NO_NAMESPACES
|
|
|
| 29 |
using std::string;
|
|
|
| 30 |
#endif /* NO_NAMESPACES */
|
|
|
| 31 |
|
|
|
| 32 |
#include "unacpp.h"
|
28 |
#include "unacpp.h"
|
| 33 |
#include "unac.h"
|
29 |
#include "unac.h"
|
|
|
30 |
#include "debuglog.h"
|
|
|
31 |
#include "utf8iter.h"
|
| 34 |
|
32 |
|
| 35 |
|
|
|
| 36 |
bool unacmaybefold(const std::string &in, std::string &out,
|
33 |
bool unacmaybefold(const string &in, string &out,
|
| 37 |
const char *encoding, bool dofold)
|
34 |
const char *encoding, bool dofold)
|
| 38 |
{
|
35 |
{
|
| 39 |
char *cout = 0;
|
36 |
char *cout = 0;
|
| 40 |
size_t out_len;
|
37 |
size_t out_len;
|
| 41 |
int status;
|
38 |
int status;
|
|
... |
|
... |
| 52 |
}
|
49 |
}
|
| 53 |
out.assign(cout, out_len);
|
50 |
out.assign(cout, out_len);
|
| 54 |
if (cout)
|
51 |
if (cout)
|
| 55 |
free(cout);
|
52 |
free(cout);
|
| 56 |
return true;
|
53 |
return true;
|
|
|
54 |
}
|
|
|
55 |
|
|
|
56 |
bool unaciscapital(const string& in)
|
|
|
57 |
{
|
|
|
58 |
if (in.empty())
|
|
|
59 |
return false;
|
|
|
60 |
Utf8Iter it(in);
|
|
|
61 |
string shorter;
|
|
|
62 |
it.appendchartostring(shorter);
|
|
|
63 |
|
|
|
64 |
string noacterm, noaclowterm;
|
|
|
65 |
if (!unacmaybefold(shorter, noacterm, "UTF-8", false)) {
|
|
|
66 |
LOGINFO(("unaciscapital: unac failed for [%s]\n", in.c_str()));
|
|
|
67 |
return false;
|
|
|
68 |
}
|
|
|
69 |
if (!unacmaybefold(noacterm, noaclowterm, "UTF-8", true)) {
|
|
|
70 |
LOGINFO(("unaciscapital: unacfold failed for [%s]\n", in.c_str()));
|
|
|
71 |
return false;
|
|
|
72 |
}
|
|
|
73 |
Utf8Iter it1(noacterm);
|
|
|
74 |
Utf8Iter it2(noaclowterm);
|
|
|
75 |
if (*it1 != *it2)
|
|
|
76 |
return true;
|
|
|
77 |
else
|
|
|
78 |
return false;
|
| 57 |
}
|
79 |
}
|
| 58 |
|
80 |
|
| 59 |
#else // not testing
|
81 |
#else // not testing
|
| 60 |
|
82 |
|
| 61 |
#include <unistd.h>
|
83 |
#include <unistd.h>
|