@Override
public void startElement(
String nameSpaceURI, String localName, String rawName, Attributes attrs)
throws SAXException {
if (idToFileMatches.isEmpty()) return;
if (rawName == null) {
rawName = localName;
}
String elt_name = rawName;
if (elt_name.equals("files")) {
String fileCount = attrs.getValue("", "totalFileCount");
if (fileCount != null) totalFileCount = Integer.parseInt(fileCount);
Logger.minor(this, "totalfilecount = " + totalFileCount);
}
if (elt_name.equals("file")) {
try {
String id = attrs.getValue("id");
ArrayList<FileMatch> matches = idToFileMatches.get(id);
if (matches != null) {
for (FileMatch match : matches) {
String key = attrs.getValue("key");
int l = attrs.getLength();
String title = null;
int wordCount = -1;
if (l >= 3) {
try {
title = attrs.getValue("title");
} catch (Exception e) {
Logger.error(this, "Index Format not compatible " + e.toString(), e);
}
try {
String wordCountString = attrs.getValue("wordCount");
if (wordCountString != null) {
wordCount = Integer.parseInt(attrs.getValue("wordCount"));
}
} catch (Exception e) {
// Logger.minor(this, "No wordcount found " + e.toString(), e);
}
}
for (FindRequest req : match.word.searches) {
Set<TermPageEntry> result = req.getUnfinishedResult();
float relevance = 0;
if (logDEBUG)
Logger.debug(
this,
"termcount "
+ (match.termpositions == null ? 0 : match.termpositions.size())
+ " filewordcount = "
+ wordCount);
if (match.termpositions != null
&& match.termpositions.size() > 0
&& wordCount > 0) {
relevance = (float) (match.termpositions.size() / (float) wordCount);
if (totalFileCount > 0 && match.word.inWordFileCount > 0)
relevance *=
Math.log((float) totalFileCount / (float) match.word.inWordFileCount);
if (logDEBUG)
Logger.debug(
this, "Set relevance of " + title + " to " + relevance + " - " + key);
}
TermPageEntry pageEntry =
new TermPageEntry(
req.getSubject(),
relevance,
new FreenetURI(key),
title,
match.termpositions);
result.add(pageEntry);
// Logger.minor(this, "added "+inFileURI+ " to "+ match);
}
}
}
} catch (Exception e) {
Logger.error(
this, "File id and key could not be retrieved. May be due to format clash", e);
}
}
}
