/**
* Loads the cached features for a given document
*
* @param document
* @param documentFile The cache file for the document.
* @return the cached features if possible. Null if a cache doesn't exist or it fails to get them.
* @throws Exception
*/
private List<EventSet> getCachedFeatures(Document document, File documentFile) {
List<EventSet> generatedEvents = null;
BufferedReader reader = null;
if (documentFile.exists() && !documentFile.isDirectory() && documentFile.canRead()) {
try {
reader = new BufferedReader(new FileReader(documentFile));
} catch (FileNotFoundException e) {
// shouldn't ever get here.. just put this here so I can keep track
// of exceptions below.
e.printStackTrace();
}
} else {
return null;
}
try {
// cachedPath is the path to the document that was used when the cache for that
// document was created. cachedLastModified is the last modified time stamp on the
// document that was cached.
String cachedPath = reader.readLine();
long cachedLastModified = Long.parseLong(reader.readLine());
String path = document.getFilePath();
File currDoc = new File(path);
long lastModified = currDoc.lastModified();
if (!(currDoc.getCanonicalPath().equals(cachedPath) && lastModified == cachedLastModified)) {
// cache is invalid
reader.close();
return null;
}
String line = null;
generatedEvents = new ArrayList<EventSet>();
while ((line = reader.readLine()) != null) {
if (line.isEmpty()) continue;
EventSet es = new EventSet();
es.setAuthor(document.getAuthor());
es.setDocumentName(document.getTitle());
es.setEventSetID(line);
String event = null;
while ((event = reader.readLine()) != null) {
if (line.isEmpty()) continue;
if (event.equals(",")) // delimiter for event sets
break;
es.addEvent(new Event(event));
}
generatedEvents.add(es);
}
reader.close();
} catch (IOException e) {
e.printStackTrace();
return null;
}
return generatedEvents;
}
/**
* Construct new Event set (with empty vector) of known author
*
* @param Author the name of the author
*/
public EventSet(String Author) {
events = new ArrayList<Event>();
setAuthor(Author);
setNewEventSetID(Author);
}
/** Creates a new list of events given a previously created list of events * */
public EventSet(List<Event> evts) {
events = new ArrayList<Event>(evts);
setAuthor("default");
setNewEventSetID("default");
}
/**
* Creates a empty list of event of the specified size
*
* @param size
*/
public EventSet(int size) {
events = new ArrayList<Event>(size);
setAuthor("default");
setNewEventSetID("default");
}
@Override
public EventSet createEventSet(Document ds) throws EventGenerationException {
String param;
HashMap<String, String> transform = new HashMap<String, String>();
boolean whitelist = false;
String line;
String[] words;
if (!(param = (getParameter("underlyingEvents"))).equals("")) {
try {
underlyingEvents = EventDriverFactory.getEventDriver(param);
} catch (Exception e) {
System.out.println("Error: cannot create EventDriver " + param);
System.out.println(" -- Using NaiveWordEventSet");
underlyingEvents = new NaiveWordEventDriver();
}
} else { // no underlyingEventsParameter, use NaiveWordEventSet
underlyingEvents = new NaiveWordEventDriver();
}
if (!(param = (getParameter("filename"))).equals("")) {
filename = param;
} else { // no underlyingfilename,
filename = null;
}
if (!(param = (getParameter("implicitWhiteList"))).equals("")) {
if (param.equalsIgnoreCase("true")) {
whitelist = true;
}
} else { // no underlyingfilename,
whitelist = false;
}
EventSet es = underlyingEvents.createEventSet(ds);
EventSet newEs = new EventSet();
newEs.setAuthor(es.getAuthor());
newEs.setNewEventSetID(es.getAuthor());
BufferedReader br = null;
if (filename != null) {
try {
FileInputStream fis = new FileInputStream(filename);
br = new BufferedReader(new InputStreamReader(fis));
while ((line = br.readLine()) != null) {
if (line.length() > 0) {
String sep = line.substring(0, 1);
words = line.substring(1).split(sep, -1);
if (words.length > 1) {
transform.put(words[0], words[1]);
System.out.println("Adding \"" + words[0] + "\" : \"" + words[1] + "\"");
}
}
}
} catch (IOException e) {
// catch io errors from FileInputStream or readLine()
System.out.println("Cannot open/read " + filename);
System.out.println("IOException error! " + e.getMessage());
transform = null;
} finally {
// if the file opened okay, make sure we close it
if (br != null) {
try {
br.close();
} catch (IOException ioe) {
}
}
}
} else {
transform = null;
}
for (Event e : es) {
String s = e.toString();
if (transform == null) {
newEs.addEvent(e);
} else if (transform.containsKey(s)) {
String newS = transform.get(s);
if (newS.length() > 0) {
newEs.addEvent(new Event(newS));
}
} else // s is not in transformation list
if (whitelist == false) {
// add only if no implicit whitelisting
newEs.addEvent(e);
} // otherwise add nothing
}
return newEs;
}
