|
a/src/rcldb/stemdb.cpp |
|
b/src/rcldb/stemdb.cpp |
|
... |
|
... |
| 57 |
}
|
57 |
}
|
| 58 |
|
58 |
|
| 59 |
#ifndef RCL_INDEX_STRIPCHARS
|
59 |
#ifndef RCL_INDEX_STRIPCHARS
|
| 60 |
// Expand the unaccented stem
|
60 |
// Expand the unaccented stem
|
| 61 |
if (!o_index_stripchars) {
|
61 |
if (!o_index_stripchars) {
|
|
|
62 |
string unac;
|
|
|
63 |
unacmaybefold(term, unac, "UTF-8", UNACOP_UNAC);
|
|
|
64 |
if (term != unac) {
|
| 62 |
for (vector<string>::const_iterator it = llangs.begin();
|
65 |
for (vector<string>::const_iterator it = llangs.begin();
|
| 63 |
it != llangs.end(); it++) {
|
66 |
it != llangs.end(); it++) {
|
| 64 |
SynTermTransStem stemmer(*it);
|
67 |
SynTermTransStem stemmer(*it);
|
| 65 |
XapComputableSynFamMember expander(getdb(), synFamStemUnac,
|
68 |
XapComputableSynFamMember expander(getdb(), synFamStemUnac,
|
| 66 |
*it, &stemmer);
|
69 |
*it, &stemmer);
|
| 67 |
string unac;
|
|
|
| 68 |
unacmaybefold(term, unac, "UTF-8", UNACOP_UNAC);
|
|
|
| 69 |
(void)expander.synExpand(unac, result);
|
70 |
(void)expander.synExpand(unac, result);
|
|
|
71 |
}
|
| 70 |
}
|
72 |
}
|
| 71 |
}
|
73 |
}
|
| 72 |
#endif
|
74 |
#endif
|
| 73 |
|
75 |
|
| 74 |
if (result.empty())
|
76 |
if (result.empty())
|
| 75 |
result.push_back(term);
|
77 |
result.push_back(term);
|
| 76 |
|
78 |
|
| 77 |
sort(result.begin(), result.end());
|
79 |
sort(result.begin(), result.end());
|
| 78 |
vector<string>::iterator uit = unique(result.begin(), result.end());
|
80 |
vector<string>::iterator uit = unique(result.begin(), result.end());
|
| 79 |
result.resize(uit - result.begin());
|
81 |
result.resize(uit - result.begin());
|
| 80 |
LOGDEB0(("stemExpand:%s: %s -> %s\n", langs.c_str(), term.c_str(),
|
82 |
LOGDEB1(("stemExpand:%s: %s -> %s\n", langs.c_str(), term.c_str(),
|
| 81 |
stringsToString(result).c_str()));
|
83 |
stringsToString(result).c_str()));
|
| 82 |
return true;
|
84 |
return true;
|
| 83 |
}
|
85 |
}
|
| 84 |
|
86 |
|
| 85 |
|
87 |
|