private String toArticleName(String name) {
name = StringUtils.trim(name);
name = RegexUtils.trim(name, "_");
name = StringUtils.trim(name);
name = StringUtils.ucfirst(name);
return name;
}
@Override
public List<Word> modifyWords(List<Word> words, XmlList xfact) {
for (Word word : words) {
if (word.hasTags() && word.getTags().isCategory("n")) {
word.setWord(StringUtils.ucfirst(word.getWord()));
}
}
return words;
}
private String getShortDescription(String name) {
LogUtils.i("Searching a short description for: \"" + name + "\"");
final Iterable<String> html = FreehalFiles.createInstance("wikipedia", name).readLines();
List<String> desc = new ArrayList<String>();
final int limitLines = 3;
for (String line : html) {
// skip empty lines
line = StringUtils.trim(line);
if (line.length() == 0) continue;
// skip disambiguation links
if (langUtils.lineContainsDisambiguation(line)) continue;
// skip images
if (langUtils.lineContainsFile(line)) continue;
// skip templates
if (line.contains("[[Wikipedia:")) continue;
// skip headings
if (line.startsWith("==")) continue;
// skip infoboxes
if (line.contains("{{")
|| line.contains("}}")
|| (line.startsWith("|") && line.contains("="))) continue;
// add sentences to description
// is it the beginning of a list?
if (line.endsWith(":") && !line.contains(".")) desc.add(line + "\n");
// or a list element
else if (line.startsWith("*")) desc.add(line + "\n");
// or a normal paragraph...
else if (desc.size() < limitLines) {
for (String sentence : line.split("\\.( |$)")) {
sentence = sentence.trim();
if (sentence.length() > 0 && desc.size() < limitLines) desc.add(sentence + ". ");
}
} else break;
}
if (desc.size() > 0) return toPlainText(StringUtils.join("", desc));
else return null;
}