|
a/src/internfile/myhtmlparse.cpp |
|
b/src/internfile/myhtmlparse.cpp |
|
... |
|
... |
| 152 |
} else if (name == "keywords") {
|
152 |
} else if (name == "keywords") {
|
| 153 |
if (!keywords.empty()) keywords += ' ';
|
153 |
if (!keywords.empty()) keywords += ' ';
|
| 154 |
string tmp = i->second;
|
154 |
string tmp = i->second;
|
| 155 |
decode_entities(tmp);
|
155 |
decode_entities(tmp);
|
| 156 |
keywords += tmp;
|
156 |
keywords += tmp;
|
|
|
157 |
} else if (name == "author") {
|
|
|
158 |
if (!author.empty()) author += ' ';
|
|
|
159 |
string tmp = i->second;
|
|
|
160 |
decode_entities(tmp);
|
|
|
161 |
author += tmp;
|
| 157 |
} else if (name == "date") {
|
162 |
} else if (name == "date") {
|
| 158 |
// Yes this doesnt exist. It's output by filters
|
163 |
// Yes this doesnt exist. It's output by filters
|
| 159 |
// And the format isn't even standard http/html
|
164 |
// And the format isn't even standard http/html
|
| 160 |
// FIXME
|
165 |
// FIXME
|
| 161 |
string tmp = i->second;
|
166 |
string tmp = i->second;
|
|
... |
|
... |
| 166 |
char ascuxtime[100];
|
171 |
char ascuxtime[100];
|
| 167 |
sprintf(ascuxtime, "%ld", (long)mktime(&tm));
|
172 |
sprintf(ascuxtime, "%ld", (long)mktime(&tm));
|
| 168 |
dmtime = ascuxtime;
|
173 |
dmtime = ascuxtime;
|
| 169 |
}
|
174 |
}
|
| 170 |
}
|
175 |
}
|
| 171 |
#if 0 // We're not a robot, so we don't care about robots metainfo
|
|
|
| 172 |
else if (name == "robots") {
|
|
|
| 173 |
string val = i->second;
|
|
|
| 174 |
decode_entities(val);
|
|
|
| 175 |
lowercase_term(val);
|
|
|
| 176 |
if (val.find("none") != string::npos ||
|
|
|
| 177 |
val.find("noindex") != string::npos) {
|
|
|
| 178 |
indexing_allowed = false;
|
|
|
| 179 |
LOGDEB1(("myhtmlparse: robots/noindex\n"));
|
|
|
| 180 |
throw false;
|
|
|
| 181 |
}
|
|
|
| 182 |
}
|
|
|
| 183 |
#endif // 0
|
|
|
| 184 |
} else if ((j = p.find("http-equiv")) != p.end()) {
|
176 |
} else if ((j = p.find("http-equiv")) != p.end()) {
|
| 185 |
string hequiv = j->second;
|
177 |
string hequiv = j->second;
|
| 186 |
lowercase_term(hequiv);
|
178 |
lowercase_term(hequiv);
|
| 187 |
if (hequiv == "content-type") {
|
179 |
if (hequiv == "content-type") {
|
| 188 |
string value = i->second;
|
180 |
string value = i->second;
|