/** 初始化各种需要长久存放的变量,需要释放则调用close() */
public ReadTopKResults() {
try {
dir = new NIOFSDirectory(new File(indexPath));
indexReader = IndexReader.open(dir, true);
features = FeatureArray.getInstacne(indexReader);
queryList = QueryReader.readAll();
} catch (IOException e) {
e.printStackTrace();
}
isCorrectlyInitialed = true;
}
/*
public static void main(String[] args) throws Exception {
Queryer q = new Queryer(new Dials());
q.runQueries();
}
*/
public void runQueries() throws Exception {
FileOutputStream outFile = new FileOutputStream(dials.getResultsPath());
QRels qRels = new QRels(dials.getqRelPath());
QueryReader rdr = new QueryReader(dials.getqPath());
Collection<MedQuery> querys = rdr.readQueries();
Collection<Double> rPrecisions = new ArrayList<Double>();
Synonymify syn = new Synonymify();
for (MedQuery q : querys) {
// System.out.println(q.toString());
Collection<String> goldStandard = qRels.getResults(q.getNumb());
String expandedQuery = syn.expandQuery(q.getQueryText());
Collection<Document> hits = search(expandedQuery, dials.getIndexDir(), 100);
// Collection<Document> hits = search(q.getQueryText(), dials.getIndexDir(),
// goldStandard.size());
Collection<String> results = new ArrayList<String>();
for (Document doc : hits) {
results.add(doc.get("U"));
}
for (String a : results) {
outFile.write((q.getNumb() + " " + a + "\n").getBytes());
}
Double prec =
StatsUtil.rPrecision(goldStandard.size(), qRels.getResults(q.getNumb()), results);
rPrecisions.add(prec);
// System.out.println(results.toString());
System.out.println(
"There are " + hits.size() + " results for: " + q.getNumb() + ":" + q.getQueryText());
System.out.println("Expanded to : " + expandedQuery);
System.out.println("Out of a possible " + goldStandard.size() + " Precision: " + prec + "\n");
}
outFile.close();
System.out.println("Average R-Precision: " + StatsUtil.average(rPrecisions));
}
public void runCompareTest() throws Exception {
LOGGER.info("Diffs saved in: {}", targetDir.getAbsolutePath());
assertFalse("Dir should not exist: ls -la " + targetDir.getAbsolutePath(), targetDir.exists());
assertTrue("Unable to create: " + targetDir.getAbsolutePath(), targetDir.mkdirs());
assertTrue(new File(targetDir, "0_deltas_go_here.txt").createNewFile());
logVersion(config1);
logVersion(config2);
int nrQueries = 0;
int failedQueries = 0;
for (final String queryString : queryReader.getQueries()) {
if (StringUtils.isBlank(queryString) || queryString.startsWith("#")) {
continue;
}
if (logEntries == 1) {
LOGGER.info("Compare query: {}", queryString);
} else if (++nrQueries % logEntries == 0) {
LOGGER.info("Compared {} queries", nrQueries);
}
final Future<List<ResponseObject>> queryExecutor1Future =
executeQuery(queryExecutor1, queryString);
final Future<List<ResponseObject>> queryExecutor2Future =
executeQuery(queryExecutor2, queryString);
final List<ResponseObject> queryExecutor1Result = queryExecutor1Future.get();
final List<ResponseObject> queryExecutor2Result = queryExecutor2Future.get();
final KnownDifferencesPredicate knownDifferencesPredicate = new KnownDifferencesPredicate();
final List<ResponseObject> responseObjects1 =
Lists.newArrayList(Iterables.filter(queryExecutor1Result, knownDifferencesPredicate));
final List<ResponseObject> responseObjects2 =
Lists.newArrayList(Iterables.filter(queryExecutor2Result, knownDifferencesPredicate));
final Patch patch = DiffUtils.diff(responseObjects1, responseObjects2);
final List<Delta> deltas = patch.getDeltas();
if (!deltas.isEmpty()) {
writeDifferences(queryString, queryExecutor1Result, queryExecutor2Result, deltas);
failedQueries++;
LOGGER.error("Query '{}' has differences", queryString);
}
}
assertThat("Number of failed queries", failedQueries, Matchers.is(0));
}
/**
* Support method for main program. This support method can also be invoked from subclasses that
* support the same command line interface
*
* @param args the command-line arguments
* @param name name of the class, to be used in error messages
*/
protected void doQuery(String args[], String name) {
boolean showTime = false;
int repeat = 1;
String sourceFileName = null;
String queryFileName = null;
File sourceFile;
File outputFile;
boolean useURLs = false;
String outputFileName = null;
boolean explain = false;
boolean wrap = false;
boolean pullMode = false;
boolean schemaAware = false;
for (int i = 0; i < args.length; i++) {
if (args[i].equals("-sa")) {
schemaAware = true;
} else if (args[i].equals("-val")) {
schemaAware = true;
} else if (args[i].equals("-vlax")) {
schemaAware = true;
} else if (args[i].equals("-p")) {
schemaAware = true;
}
}
config = makeConfiguration(schemaAware);
config.setHostLanguage(Configuration.XQUERY);
StaticQueryContext staticEnv = new StaticQueryContext(config);
DynamicQueryContext dynamicEnv = new DynamicQueryContext(config);
Properties outputProps = new Properties();
// Check the command-line arguments.
try {
int i = 0;
while (i < args.length) {
if (args[i].charAt(0) == '-') {
if (args[i].equals("-cr")) {
i++;
if (args.length < i + 1) {
badUsage(name, "No output file name");
}
String crclass = args[i++];
Object resolver = config.getInstance(crclass, null);
if (!(resolver instanceof CollectionURIResolver)) {
quit(crclass + " is not a CollectionURIResolver", 2);
}
config.setCollectionURIResolver((CollectionURIResolver) resolver);
} else if (args[i].equals("-ds")) {
config.setTreeModel(Builder.LINKED_TREE);
i++;
} else if (args[i].equals("-dt")) {
config.setTreeModel(Builder.TINY_TREE);
i++;
} else if (args[i].equals("-e")) {
explain = true;
i++;
} else if (args[i].equals("-l")) {
config.setLineNumbering(true);
i++;
} else if (args[i].equals("-3")) { // undocumented option: do it thrice
i++;
repeat = 3;
} else if (args[i].equals("-9")) { // undocumented option: do it nine times
i++;
repeat = 9;
} else if (args[i].equals("-mr")) {
i++;
if (args.length < i + 1) {
badUsage(name, "No ModuleURIResolver class");
}
String r = args[i++];
config.setModuleURIResolver(r);
} else if (args[i].equals("-noext")) {
i++;
config.setAllowExternalFunctions(false);
} else if (args[i].equals("-o")) {
i++;
if (args.length < i + 1) {
badUsage(name, "No output file name");
}
outputFileName = args[i++];
} else if (args[i].equals("-p")) {
i++;
setPOption(config);
useURLs = true;
} else if (args[i].equals("-pull")) {
i++;
pullMode = true;
} else if (args[i].equals("-r")) {
i++;
if (args.length < i + 1) {
badUsage(name, "No URIResolver class");
}
String r = args[i++];
config.setURIResolver(config.makeURIResolver(r));
dynamicEnv.setURIResolver(config.makeURIResolver(r));
} else if (args[i].equals("-s")) {
i++;
if (args.length < i + 1) {
badUsage(name, "No source file name");
}
sourceFileName = args[i++];
} else if (args[i].equals("-sa")) {
// already handled
i++;
} else if (args[i].equals("-snone")) {
config.setStripsWhiteSpace(Whitespace.NONE);
i++;
} else if (args[i].equals("-sall")) {
config.setStripsWhiteSpace(Whitespace.ALL);
i++;
} else if (args[i].equals("-signorable")) {
config.setStripsWhiteSpace(Whitespace.IGNORABLE);
i++;
} else if (args[i].equals("-strip")) { // retained for compatibility
config.setStripsWhiteSpace(Whitespace.ALL);
i++;
} else if (args[i].equals("-t")) {
System.err.println(config.getProductTitle());
// System.err.println("Java version " + System.getProperty("java.version"));
System.err.println(config.getPlatform().getPlatformVersion());
config.setTiming(true);
showTime = true;
i++;
} else if (args[i].equals("-T")) {
config.setTraceListener(new XQueryTraceListener());
i++;
} else if (args[i].equals("-TJ")) {
i++;
config.setTraceExternalFunctions(true);
} else if (args[i].equals("-TL")) {
if (args.length < i + 2) {
badUsage(name, "No TraceListener class specified");
}
TraceListener traceListener = config.makeTraceListener(args[++i]);
config.setTraceListener(traceListener);
config.setLineNumbering(true);
i++;
} else if (args[i].equals("-u")) {
useURLs = true;
i++;
} else if (args[i].equals("-untyped")) {
// TODO: this is an experimental undocumented option. It should be checked for
// consistency
config.setAllNodesUntyped(true);
i++;
} else if (args[i].equals("-v")) {
config.setValidation(true);
i++;
} else if (args[i].equals("-val")) {
if (schemaAware) {
config.setSchemaValidationMode(Validation.STRICT);
} else {
quit("The -val option requires a schema-aware processor", 2);
}
i++;
} else if (args[i].equals("-vlax")) {
if (schemaAware) {
config.setSchemaValidationMode(Validation.LAX);
} else {
quit("The -vlax option requires a schema-aware processor", 2);
}
i++;
} else if (args[i].equals("-vw")) {
if (schemaAware) {
config.setValidationWarnings(true);
} else {
quit("The -vw option requires a schema-aware processor", 2);
}
i++;
} else if (args[i].equals("-wrap")) {
wrap = true;
i++;
} else if (args[i].equals("-1.1")) {
config.setXMLVersion(Configuration.XML11);
i++;
} else if (args[i].equals("-?")) {
badUsage(name, "");
} else if (args[i].equals("-")) {
queryFileName = "-";
i++;
} else {
badUsage(name, "Unknown option " + args[i]);
}
} else {
break;
}
}
if (!("-".equals(queryFileName))) {
if (args.length < i + 1) {
badUsage(name, "No query file name");
}
queryFileName = args[i++];
}
for (int p = i; p < args.length; p++) {
String arg = args[p];
int eq = arg.indexOf("=");
if (eq < 1 || eq >= arg.length() - 1) {
badUsage(name, "Bad param=value pair on command line: " + arg);
}
String argname = arg.substring(0, eq);
if (argname.startsWith("!")) {
// parameters starting with "!" are taken as output properties
outputProps.setProperty(argname.substring(1), arg.substring(eq + 1));
} else if (argname.startsWith("+")) {
// parameters starting with "+" are taken as input documents
Object sources = Transform.loadDocuments(arg.substring(eq + 1), useURLs, config, true);
dynamicEnv.setParameter(argname.substring(1), sources);
} else {
dynamicEnv.setParameter(argname, new UntypedAtomicValue(arg.substring(eq + 1)));
}
}
config.displayLicenseMessage();
if (pullMode) {
config.setLazyConstructionMode(true);
}
Source sourceInput = null;
if (sourceFileName != null) {
if (useURLs || sourceFileName.startsWith("http:") || sourceFileName.startsWith("file:")) {
sourceInput = config.getURIResolver().resolve(sourceFileName, null);
if (sourceInput == null) {
sourceInput = config.getSystemURIResolver().resolve(sourceFileName, null);
}
} else if (sourceFileName.equals("-")) {
// take input from stdin
sourceInput = new StreamSource(System.in);
} else {
sourceFile = new File(sourceFileName);
if (!sourceFile.exists()) {
quit("Source file " + sourceFile + " does not exist", 2);
}
if (config.getPlatform() instanceof JavaPlatform) {
InputSource eis = new InputSource(sourceFile.toURI().toString());
sourceInput = new SAXSource(eis);
} else {
sourceInput = new StreamSource(sourceFile.toURI().toString());
}
}
}
long startTime = (new Date()).getTime();
if (showTime) {
System.err.println("Compiling query from " + queryFileName);
}
XQueryExpression exp;
try {
if (queryFileName.equals("-")) {
Reader queryReader = new InputStreamReader(System.in);
exp = staticEnv.compileQuery(queryReader);
} else if (queryFileName.startsWith("{") && queryFileName.endsWith("}")) {
// query is inline on the command line
String q = queryFileName.substring(1, queryFileName.length() - 1);
exp = staticEnv.compileQuery(q);
} else if (useURLs
|| queryFileName.startsWith("http:")
|| queryFileName.startsWith("file:")) {
ModuleURIResolver resolver = config.getModuleURIResolver();
String[] locations = {queryFileName};
Source[] sources = resolver.resolve(null, null, locations);
if (sources.length != 1 || !(sources[0] instanceof StreamSource)) {
quit("Module URI Resolver must return a single StreamSource", 2);
}
String queryText =
QueryReader.readSourceQuery((StreamSource) sources[0], config.getNameChecker());
exp = staticEnv.compileQuery(queryText);
} else {
InputStream queryStream = new FileInputStream(queryFileName);
staticEnv.setBaseURI(new File(queryFileName).toURI().toString());
exp = staticEnv.compileQuery(queryStream, null);
}
staticEnv = exp.getStaticContext(); // the original staticContext is copied
if (showTime) {
long endTime = (new Date()).getTime();
System.err.println("Compilation time: " + (endTime - startTime) + " milliseconds");
startTime = endTime;
}
} catch (XPathException err) {
int line = -1;
String module = null;
if (err.getLocator() != null) {
line = err.getLocator().getLineNumber();
module = err.getLocator().getSystemId();
}
if (err.hasBeenReported()) {
quit("Failed to compile query", 2);
} else {
if (line == -1) {
System.err.println("Failed to compile query: " + err.getMessage());
} else {
System.err.println("Static error at line " + line + " of " + module + ':');
System.err.println(err.getMessage());
}
}
exp = null;
System.exit(2);
}
if (explain) {
staticEnv.getExecutable().getKeyManager().explainKeys(config);
staticEnv.explainGlobalVariables();
staticEnv.explainGlobalFunctions();
exp.explain(staticEnv.getConfiguration());
}
OutputStream destination;
if (outputFileName != null) {
outputFile = new File(outputFileName);
if (outputFile.isDirectory()) {
quit("Output is a directory", 2);
}
destination = new FileOutputStream(outputFile);
} else {
destination = System.out;
}
for (int r = 0; r < repeat; r++) { // repeat is for internal testing/timing
if (sourceInput != null) {
if (showTime) {
System.err.println("Processing " + sourceInput.getSystemId());
}
DocumentInfo doc = staticEnv.buildDocument(sourceInput);
dynamicEnv.setContextItem(doc);
}
try {
if (wrap) {
SequenceIterator results = exp.iterator(dynamicEnv);
DocumentInfo resultDoc = QueryResult.wrap(results, config);
QueryResult.serialize(resultDoc, new StreamResult(destination), outputProps, config);
destination.close();
} else if (pullMode) {
if (wrap) {
outputProps.setProperty(SaxonOutputKeys.WRAP, "yes");
}
try {
exp.pull(dynamicEnv, new StreamResult(destination), outputProps);
} catch (XPathException err) {
config.reportFatalError(err);
throw err;
}
} else {
exp.run(dynamicEnv, new StreamResult(destination), outputProps);
}
} catch (TerminationException err) {
throw err;
} catch (XPathException err) {
if (err.hasBeenReported()) {
throw new DynamicError("Run-time errors were reported");
} else {
throw err;
}
}
if (showTime) {
long endTime = (new Date()).getTime();
System.err.println("Execution time: " + (endTime - startTime) + " milliseconds");
startTime = endTime;
}
}
} catch (TerminationException err) {
quit(err.getMessage(), 1);
} catch (XPathException err) {
quit("Query processing failed: " + err.getMessage(), 2);
} catch (TransformerFactoryConfigurationError err) {
err.printStackTrace();
quit("Query processing failed", 2);
} catch (Exception err2) {
err2.printStackTrace();
quit("Fatal error during transformation: " + err2.getMessage(), 2);
}
}