private Element getDescription(String task) {
if (task != null) {
if (task.equals("define")) return definer.getDescription();
if (task.equals("compare")) return comparer.getDescription();
if (task.equals("search")) return searcher.getDescription();
if (task.equals("wikify")) return wikifier.getDescription();
}
Element description = doc.createElement("Description");
description.appendChild(
createElement(
"Details",
"<p>This servlet provides a range of services for mining information from Wikipedia. Further details depend on what you want to do.</p>"
+ "<p>You can <a href=\""
+ context.getInitParameter("service_name")
+ "?task=search&help\">search for pages</a>, <a href=\""
+ context.getInitParameter("service_name")
+ "?task=compare&help\">measure how terms or articles related to each other</a>, <a href=\""
+ context.getInitParameter("service_name")
+ "?task=define&help\">obtain short definitions from articles</a>, and <a href=\""
+ context.getInitParameter("service_name")
+ "?task=wikify&help\">detect topics in web pages</a>.</p>"));
Element paramTask =
createElement(
"Parameter",
"Specifies what you want to do: can be <em>search</em>, <em>compare</em>, <em>define</em>, or <em>wikify</em>");
paramTask.setAttribute("name", "task");
description.appendChild(paramTask);
Element paramId = doc.createElement("Parameter");
paramId.setAttribute("name", "help");
paramId.appendChild(doc.createTextNode("Specifies that you want help about the service."));
description.appendChild(paramId);
return description;
}
public void init(ServletConfig config) throws ServletException {
super.init(config);
context = config.getServletContext();
TextProcessor tp = new CaseFolder();
try {
wikipedia =
new Wikipedia(
context.getInitParameter("mysql_server"),
context.getInitParameter("mysql_database"),
context.getInitParameter("mysql_user"),
context.getInitParameter("mysql_password"));
} catch (Exception e) {
throw new ServletException("Could not connect to wikipedia database.");
}
// Escaper escaper = new Escaper() ;
definer = new Definer(this);
comparer = new Comparer(this);
searcher = new Searcher(this);
try {
wikifier = new Wikifier(this, tp);
} catch (Exception e) {
System.err.println("Could not initialize wikifier");
}
try {
File dataDirectory = new File(context.getInitParameter("data_directory"));
if (!dataDirectory.exists() || !dataDirectory.isDirectory()) {
throw new Exception();
}
cachingThread = new CacherThread(dataDirectory, tp);
cachingThread.start();
} catch (Exception e) {
throw new ServletException("Could not locate wikipedia data directory.");
}
try {
TransformerFactory tf = TransformerFactory.newInstance();
transformersByName = new HashMap<String, Transformer>();
transformersByName.put(
"help", buildTransformer("help", new File("/research/wikipediaminer/web/xsl"), tf));
transformersByName.put(
"loading", buildTransformer("loading", new File("/research/wikipediaminer/web/xsl"), tf));
transformersByName.put(
"search", buildTransformer("search", new File("/research/wikipediaminer/web/xsl"), tf));
transformersByName.put(
"compare", buildTransformer("compare", new File("/research/wikipediaminer/web/xsl"), tf));
transformersByName.put(
"wikify", buildTransformer("wikify", new File("/research/wikipediaminer/web/xsl"), tf));
Transformer serializer = TransformerFactory.newInstance().newTransformer();
serializer.setOutputProperty(OutputKeys.INDENT, "yes");
serializer.setOutputProperty(OutputKeys.METHOD, "xml");
serializer.setOutputProperty("{http://xml.apache.org/xslt}indent-amount", "3");
transformersByName.put("serializer", serializer);
} catch (Exception e) {
throw new ServletException("Could not load xslt library.");
}
}
public void doGet(HttpServletRequest request, HttpServletResponse response)
throws IOException, ServletException {
try {
response.setHeader("Cache-Control", "no-cache");
response.setCharacterEncoding("UTF-8");
String task = request.getParameter("task");
Element data = null;
// process help request
if (request.getParameter("help") != null) data = getDescription(task);
// redirect to home page if there is no task
if (data == null && task == null) {
response.setContentType("text/html");
response
.getWriter()
.append(
"<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.0 Transitional//EN\"><html><head><meta http-equiv=\"REFRESH\" content=\"0;url="
+ context.getInitParameter("server_path")
+ "></head><body></body></html>");
return;
}
// process definition request
if (data == null && task.equals("define")) {
int id = resolveIntegerArg(request.getParameter("id"), -1);
int length = resolveIntegerArg(request.getParameter("length"), definer.getDefaultLength());
int format = resolveIntegerArg(request.getParameter("format"), definer.getDefaultFormat());
int maxImageWidth =
resolveIntegerArg(
request.getParameter("maxImageWidth"), definer.getDefaultMaxImageWidth());
int maxImageHeight =
resolveIntegerArg(
request.getParameter("maxImageHeight"), definer.getDefaultMaxImageHeight());
int linkDestination =
resolveIntegerArg(
request.getParameter("linkDestination"), definer.getDefaultLinkDestination());
boolean getImages = resolveBooleanArg(request.getParameter("getImages"), false);
data =
definer.getDefinition(
id, length, format, linkDestination, getImages, maxImageWidth, maxImageHeight);
}
// all of the remaining tasks require data to be cached, so lets make sure that is finished
// before continuing.
if (!cachingThread.isOk()) throw new ServletException("Could not cache wikipedia data");
double progress = cachingThread.getProgress();
if (data == null && (progress < 1 || task.equals("progress"))) {
// still caching up data, not ready to return a response yet.
data = doc.createElement("loading");
data.setAttribute("progress", df.format(progress));
task = "loading";
}
// process search request
if (data == null && task.equals("search")) {
String term = request.getParameter("term");
String id = request.getParameter("id");
int linkLimit =
resolveIntegerArg(request.getParameter("linkLimit"), searcher.getDefaultMaxLinkCount());
int senseLimit =
resolveIntegerArg(
request.getParameter("senseLimit"), searcher.getDefaultMaxSenseCount());
if (id == null) data = searcher.doSearch(term, linkLimit, senseLimit);
else data = searcher.doSearch(Integer.parseInt(id), linkLimit);
}
// process compare request
if (data == null && task.equals("compare")) {
String term1 = request.getParameter("term1");
String term2 = request.getParameter("term2");
int linkLimit =
resolveIntegerArg(request.getParameter("linkLimit"), comparer.getDefaultMaxLinkCount());
boolean details =
resolveBooleanArg(request.getParameter("details"), comparer.getDefaultShowDetails());
data = comparer.getRelatedness(term1, term2, details, linkLimit);
}
// process wikify request
if (data == null && task.equals("wikify")) {
if (this.wikifier == null)
throw new ServletException(
"Wikifier is not available. You must configure the servlet so that it has access to link detection and disambiguation models.");
String source = request.getParameter("source");
int sourceMode =
resolveIntegerArg(request.getParameter("sourceMode"), Wikifier.SOURCE_AUTODETECT);
String linkColor = request.getParameter("linkColor");
String baseColor = request.getParameter("baseColor");
double minProb =
resolveDoubleArg(
request.getParameter("minProbability"), wikifier.getDefaultMinProbability());
int repeatMode =
resolveIntegerArg(request.getParameter("repeatMode"), wikifier.getDefaultRepeatMode());
boolean showTooltips =
resolveBooleanArg(
request.getParameter("showTooltips"), wikifier.getDefaultShowTooltips());
String bannedTopics = request.getParameter("bannedTopics");
boolean wrapInXml = resolveBooleanArg(request.getParameter("wrapInXml"), true);
if (wrapInXml) {
data =
wikifier.wikifyAndWrapInXML(
source,
sourceMode,
minProb,
repeatMode,
bannedTopics,
baseColor,
linkColor,
showTooltips);
} else {
response.setContentType("text/html");
response
.getWriter()
.append(
wikifier.wikify(
source,
sourceMode,
minProb,
repeatMode,
bannedTopics,
baseColor,
linkColor,
showTooltips));
return;
}
}
if (data == null) throw new Exception("Unknown Task");
// wrap data
Element wrapper = doc.createElement("WikipediaMinerResponse");
wrapper.setAttribute("server_path", context.getInitParameter("server_path"));
wrapper.setAttribute("service_name", context.getInitParameter("service_name"));
wrapper.appendChild(data);
data = wrapper;
// Transform or serialize xml data as appropriate
Transformer tf = null;
if (request.getParameter("xml") == null) {
// we need to transform the data into html
tf = transformersByName.get(task);
if (request.getParameter("help") != null) tf = transformersByName.get("help");
}
if (tf == null) {
// we need to serialize the data as xml
tf = transformersByName.get("serializer");
response.setContentType("application/xml");
} else {
// output will be transformed to html
response.setContentType("text/html");
response
.getWriter()
.append(
"<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01//EN\" \"http://www.w3.org/TR/html4/strict.dtd\">\n");
}
tf.transform(new DOMSource(data), new StreamResult(response.getWriter()));
} catch (Exception error) {
response.reset();
response.setContentType("application/xml");
response.setHeader("Cache-Control", "no-cache");
response.setCharacterEncoding("UTF8");
Element xmlError = doc.createElement("Error");
if (error.getMessage() != null) xmlError.setAttribute("message", error.getMessage());
Element xmlStackTrace = doc.createElement("StackTrace");
xmlError.appendChild(xmlStackTrace);
for (StackTraceElement ste : error.getStackTrace()) {
Element xmlSte = doc.createElement("StackTraceElement");
xmlSte.setAttribute("message", ste.toString());
xmlStackTrace.appendChild(xmlSte);
}
try {
transformersByName
.get("serializer")
.transform(new DOMSource(xmlError), new StreamResult(response.getWriter()));
} catch (Exception e) {
// TODO: something for when an error is thrown processing an error????
}
;
}
}