MetaInfo ZefaniaLex::buildIndexFromXmlDoc(KoXmlDocument *xmldoc)
{
try {
MetaInfo info;
int couldBe = 0;//1 = RMac
Document indexdoc;
const QString index = indexPath();
QString fileTitle;
QString uid;
QString type;
QDir dir("/");
dir.mkpath(index);
RefText refText;
refText.setSettings(m_settings);
IndexWriter* writer = NULL;
const TCHAR* stop_words[] = { NULL };
standard::StandardAnalyzer an(stop_words);
if(IndexReader::indexExists(index.toStdString().c_str())) {
if(IndexReader::isLocked(index.toStdString().c_str())) {
myDebug() << "Index was locked... unlocking it.";
IndexReader::unlock(index.toStdString().c_str());
}
}
writer = new IndexWriter(index.toStdString().c_str() , &an, true);
writer->setMaxFieldLength(0x7FFFFFFFL);
writer->setUseCompoundFile(false);
KoXmlNode item = xmldoc->documentElement().firstChild();
type = xmldoc->documentElement().toElement().attribute("type", "");
for(int c = 0; !item.isNull();) {
QString key = "";
QString title = "";
QString trans = "";
QString pron = "";
QString desc = "";
KoXmlElement e = item.toElement();
if(e.tagName().compare("INFORMATION", Qt::CaseInsensitive) == 0) {
KoXmlNode title = item.namedItem("subject");
KoXmlNode identifer = item.namedItem("identifier");
fileTitle = title.toElement().text();
uid = identifer.toElement().text();
} else if(e.tagName().compare("item", Qt::CaseInsensitive) == 0) {
key = e.attribute("id");
KoXmlNode details = item.firstChild();
while(!details.isNull()) {
KoXmlElement edetails = details.toElement();
if(edetails.tagName().compare("title", Qt::CaseInsensitive) == 0) {
title = edetails.text();
} else if(edetails.tagName().compare("transliteration", Qt::CaseInsensitive) == 0) {
trans = edetails.text();
} else if(edetails.tagName().compare("pronunciation", Qt::CaseInsensitive) == 0) {
KoXmlNode em = details.firstChild();
while(!em.isNull()) {
if(em.toElement().tagName().compare("em", Qt::CaseInsensitive) == 0)
pron = "<em>" + em.toElement().text() + "</em>";
em = em.nextSibling();
}
} else if(edetails.tagName().compare("description", Qt::CaseInsensitive) == 0) {
KoXmlNode descNode = details.firstChild();
while(!descNode.isNull()) {
if(descNode.nodeType() == 2) {
desc += descNode.toText().data();
} else if(descNode.nodeType() == 1) {
KoXmlElement descElement = descNode.toElement();
if(descElement.tagName().compare("reflink", Qt::CaseInsensitive) == 0) {
if(descElement.hasAttribute("mscope")) {
const QString mscope = descElement.attribute("mscope", ";;;");
VerseUrl url;
url.fromMscope(mscope);
desc += " <a href=\"" + url.toString() + "\">" + refText.toString(url) + "</a> ";
} else if(descElement.hasAttribute("target")) {
desc += descElement.text();
}
} else if(descElement.tagName().compare("see", Qt::CaseInsensitive) == 0) {
const QString target = descElement.attribute("target", "");
//todo: currently we assume target = x-self
StrongUrl url;
bool ok = url.fromText(descElement.text());
if(ok)
desc += " <a href=\"" + url.toString() + "\">" + descElement.text() + "</a> ";
}
}
descNode = descNode.nextSibling();
}
desc += "<hr />";
}
details = details.nextSibling();
}
if(couldBe == 0) {
if(key.toUpper() == "A-APF" || key.toUpper() == "X-NSN" || key.toUpper() == "V-PAP-DPN") {
couldBe = 1;
}
}
QString content = "<h3 class='strongTitle'>" + key + " - " + title + "</h3>";
if(!trans.isEmpty()) {
content += " (" + trans + ") ";
}
if(!pron.isEmpty()) {
content += " [" + pron + "] ";
}
content += "<br />" + desc;
indexdoc.clear();
#ifdef OBV_USE_WSTRING
indexdoc.add(*_CLNEW Field(_T("key"), key.toStdWString().c_str(), Field::STORE_YES | Field::INDEX_TOKENIZED));
indexdoc.add(*_CLNEW Field(_T("content"), content.toStdWString().c_str(), Field::STORE_YES | Field::INDEX_TOKENIZED));
#else
indexdoc.add(*_CLNEW Field(_T("key"), reinterpret_cast<const wchar_t *>(key.utf16()), Field::STORE_YES | Field::INDEX_TOKENIZED));
indexdoc.add(*_CLNEW Field(_T("content"), reinterpret_cast<const wchar_t *>(content.utf16()), Field::STORE_YES | Field::INDEX_TOKENIZED));
#endif
writer->addDocument(&indexdoc);
}
item = item.nextSibling();
c++;
}
writer->setUseCompoundFile(true);
writer->optimize();
writer->close();
delete writer;
info.setName(fileTitle);
info.setUID(uid);
if(type == "x-strong") {
info.setDefaultModule(OBVCore::DefaultStrongDictModule);
info.setContent(OBVCore::StrongsContent);
} else if(type == "x-dictionary") {
if(couldBe == 1) {
info.setDefaultModule(OBVCore::DefaultRMACDictModule);
info.setContent(OBVCore::RMacContent);
} else {
info.setDefaultModule(OBVCore::DefaultDictModule);
}
}
return info;
}
catch(...) {
return MetaInfo();
}
}
MetaInfo ZefaniaLex::readInfo(const QString &name)
{
DEBUG_FUNC_NAME
QFile file(name);
int couldBe = 0;//1 = RMac
QString uid = "";
QString type;
ZefaniaXmlReader reader(name);
MetaInfo info = reader.readMetaInfo();
//open the xml file
if(!file.open(QFile::ReadOnly | QFile::Text))
return MetaInfo();
m_xml = new QXmlStreamReader(&file);
if(m_xml->readNextStartElement()) {
if(cmp(m_xml->name(), "dictionary")) {
type = m_xml->attributes().value("type").toString();
if(type != "x-dictionary") {
couldBe = -1;// do not scan the keys
}
while(m_xml->readNextStartElement()) {
if(cmp(m_xml->name(), "item")) {
const QString key = m_xml->attributes().value("id").toString();
if(couldBe == 0) {
if(key.toUpper() == "A-APF" || key.toUpper() == "X-NSN" || key.toUpper() == "V-PAP-DPN") {//todo: speed
couldBe = 1;
}
} else {
break;
}
} else {
m_xml->skipCurrentElement();
}
}
}
}
if(info.name().isEmpty() && !info.subject.isEmpty()) {
info.setName(info.subject);
}
info.setUID(uid);
if(type == "x-strong") {
info.setDefaultModule(ModuleTools::DefaultStrongDictModule);
info.setContent(ModuleTools::StrongsContent);
} else if(type == "x-dictionary") {
if(couldBe == 1) {
info.setDefaultModule(ModuleTools::DefaultRMACDictModule);
info.setContent(ModuleTools::RMACContent);
} else {
info.setDefaultModule(ModuleTools::DefaultDictModule);
info.setContent(ModuleTools::DictionaryContent);
}
}
file.close();
delete m_xml;
m_xml = nullptr;
return info;
}