Java Code Examples for it.unimi.dsi.logging.ProgressLogger#lightUpdate()

@Override
public Void call() throws Exception {
	final ProgressLogger pl = new ProgressLogger(LOGGER, 1, TimeUnit.MINUTES, "records");

	pl.itemsName = "records";
	pl.displayFreeMemory = true;
	pl.displayLocalSpeed = true;
	pl.start("Scanning...");

	try {
		for(;;) {
			final Result<?>[] result = queue.take();
			if (result == Result.END_OF_RESULTS) {
				pl.done();
				return null;
			}
			for (final Result<?> r : result) if (r != null) r.write();
			pl.lightUpdate();
		}
	}
	catch(final Exception e) {
		LOGGER.error("Exception in flushing thread", e);
		throw e;
	}
}
 

/**
 * Returns a list of pointers to a GZIP archive entries positions (including the end of file).
 *
 * @param in the stream from which to read the GZIP archive.
 * @param pl a progress logger.
 * @return a list of longs where the <em>i</em>-th long is the offset in the stream of the first byte of the <em>i</em>-th archive entry.
 */
public static LongBigArrayBigList index(final InputStream in, final ProgressLogger pl) throws IOException {
	final LongBigArrayBigList pointers = new LongBigArrayBigList();
	long current = 0;
	final GZIPArchiveReader gzar = new GZIPArchiveReader(in);
	GZIPArchive.ReadEntry re;
	for (;;) {
		re = gzar.skipEntry();
		if (re == null) break;
		pointers.add(current);
		current += re.compressedSkipLength;
		if (pl != null) pl.lightUpdate();
	}
	in.close();
	return pointers;
}
 

public static void main(String[] arg) throws IOException {
	if (arg.length == 0) {
		System.err.println("Usage: " + BuildRepetitionSet.class.getSimpleName() + " REPETITIONSET");
		System.exit(1);
	}

	final FastBufferedReader fastBufferedReader = new FastBufferedReader(new InputStreamReader(System.in, Charsets.US_ASCII));
	final MutableString s = new MutableString();
	final LongOpenHashSet repeatedSet = new LongOpenHashSet();
	final String outputFilename = arg[0];
	final ProgressLogger pl = new ProgressLogger();

	MutableString lastUrl = new MutableString();
	pl.itemsName = "lines";
	pl.start("Reading... ");
	while(fastBufferedReader.readLine(s) != null) {
		final int firstTab = s.indexOf('\t');
		final int secondTab = s.indexOf('\t', firstTab + 1);
		MutableString url = s.substring(secondTab + 1);
		if (url.equals(lastUrl)) {
			final int storeIndex = Integer.parseInt(new String(s.array(), 0, firstTab));
			final long storePosition = Long.parseLong(new String(s.array(), firstTab + 1, secondTab - firstTab - 1));
			repeatedSet.add((long)storeIndex << 48 | storePosition);
			System.out.print(storeIndex);
			System.out.print('\t');
			System.out.print(storePosition);
			System.out.print('\t');
			System.out.println(url);
		}

		lastUrl = url;
		pl.lightUpdate();
	}

	pl.done();

	fastBufferedReader.close();
	BinIO.storeObject(repeatedSet, outputFilename);
}
 

@SuppressWarnings("resource")
public static int[] writeRecords(final String path, final int numRecords, final WarcRecord[] randomRecords, final int parallel) throws IOException, InterruptedException {
	final ProgressLogger pl = new ProgressLogger(LOGGER, "records");
	if (parallel <= 1) pl.expectedUpdates = numRecords;
	final ProgressLogger plb = new ProgressLogger(LOGGER, "KB");
	final CountingOutputStream cos = new CountingOutputStream(new FastBufferedOutputStream(new FileOutputStream (path)));
	final WarcWriter ww;
	if (parallel == 0) {
		ww = new UncompressedWarcWriter(cos);
		pl.start("Writing records…");
	} else if (parallel == 1) {
		ww = new CompressedWarcWriter(cos);
		pl.start("Writing records (compressed)…");
	} else {
		ww = null;
		pl.start("SHOULD NOT HAPPEN");
		throw new IllegalStateException();
	}
	plb.start();
	long written = 0;
	final int[] position = new int[numRecords];
	for (int i = 0; i < numRecords; i++) {
		final int pos = RandomTestMocks.RNG.nextInt(randomRecords.length);
		position[i] = pos;
		ww.write(randomRecords[pos]);
		if (parallel <= 0) {
			pl.lightUpdate();
			plb.update((cos.getCount() - written) / 1024);
		}
		written = cos.getCount();
	}
	ww.close();
	pl.done(numRecords);
	plb.done(cos.getCount());
	return position;
}
 

public static void main( final String[] arg ) throws IOException, JSAPException, NoSuchMethodException {
	
	final SimpleJSAP jsap = new SimpleJSAP( BloomFilter.class.getName(), "Creates a Bloom filter reading from standard input a newline-separated list of terms.",
			new Parameter[] {
				new FlaggedOption( "bufferSize", IntSizeStringParser.getParser(), "64Ki", JSAP.NOT_REQUIRED, 'b',  "buffer-size", "The size of the I/O buffer used to read terms." ),
				new FlaggedOption( "encoding", ForNameStringParser.getParser( Charset.class ), "UTF-8", JSAP.NOT_REQUIRED, 'e', "encoding", "The term file encoding." ),
				new UnflaggedOption( "bloomFilter", JSAP.STRING_PARSER, JSAP.NO_DEFAULT, JSAP.REQUIRED, JSAP.NOT_GREEDY, "The filename for the serialised front-coded list." ),
				new UnflaggedOption( "size", JSAP.INTSIZE_PARSER, JSAP.NO_DEFAULT, JSAP.REQUIRED, JSAP.NOT_GREEDY, "The size of the filter (i.e., the expected number of elements in the filter; usually, the number of terms)." ),
				new UnflaggedOption( "precision", JSAP.INTEGER_PARSER, JSAP.NO_DEFAULT, JSAP.REQUIRED, JSAP.NOT_GREEDY, "The precision of the filter." )
	});
	
	JSAPResult jsapResult = jsap.parse( arg );
	if ( jsap.messagePrinted() ) return;
	
	final int bufferSize = jsapResult.getInt( "bufferSize" );
	final String filterName = jsapResult.getString( "bloomFilter" );
	final Charset encoding = (Charset)jsapResult.getObject( "encoding" );

	BloomFilter filter = new BloomFilter( jsapResult.getInt( "size" ), jsapResult.getInt( "precision" ) );
	final ProgressLogger pl = new ProgressLogger();
	pl.itemsName = "terms";
	pl.start( "Reading terms..." );
	MutableString s = new MutableString();
	FastBufferedReader reader = new FastBufferedReader( new InputStreamReader( System.in, encoding ), bufferSize );
	while( reader.readLine( s ) != null ) { 
		filter.add( s );
		pl.lightUpdate();
	}
	pl.done();

	BinIO.storeObject( filter, filterName );
}
 

public static Result reaches(final KnowledgeBase kb, final long startSig, final int maxRevs, final ProgressLogger pl) {
	final LongOpenHashSet result = new LongOpenHashSet();
	final Object2ObjectOpenHashMap<String, IntOpenHashSet> product2Revs = new Object2ObjectOpenHashMap<>();
	final MutableLong totRevs = new MutableLong();

	// Visit queue
	final LongArrayFIFOQueue queue = new LongArrayFIFOQueue();
	queue.enqueue(startSig);
	result.add(startSig);

	String p = kb.callGraphs.get(index(startSig)).product;
	IntOpenHashSet revs = new IntOpenHashSet();
	revs.add(index(startSig));
	product2Revs.put(p, revs);
	totRevs.increment();


	pl.itemsName = "nodes";
	pl.info = new Object() {
		@Override
		public String toString() {
			return "[nodes: " + result.size() + " products: " + product2Revs.size() + " revisions: " + totRevs.getValue() + "]";
		}
	};

	pl.start("Visiting reachable nodes...");

	while (!queue.isEmpty()) {
		final long node = queue.dequeueLong();

		for (final long s : kb.successors(node)) if (!result.contains(s)) {
			p = kb.callGraphs.get(index(s)).product;
			final long gid = gid(s);
			if (badGIDs.contains(gid)) continue;
			final String targetNameSpace = kb.new Node(gid, index(s)).toFastenURI().getRawNamespace();
			if (targetNameSpace.startsWith("java.") || targetNameSpace.startsWith("javax.") || targetNameSpace.startsWith("jdk.")) {
				badGIDs.add(gid);
				continue;
			}
			revs = product2Revs.get(p);
			if (revs == null) product2Revs.put(p, revs = new IntOpenHashSet());
			if (revs.contains(index(s)) || revs.size() < maxRevs) {
				queue.enqueue(s);
				result.add(s);
				//System.out.println(kb.new Node(gid(node), index(node)).toFastenURI() + " -> " + kb.new Node(gid(s), index(s)).toFastenURI());
				if (revs.add(index(s))) totRevs.increment();
			}
		}
		pl.lightUpdate();
	}

	pl.done();
	return new Result(result, product2Revs.size(), totRevs.getValue().longValue());
}
 

public static Result coreaches(final KnowledgeBase kb, final long startSig, final int maxRevs, final ProgressLogger pl) {
	final LongOpenHashSet result = new LongOpenHashSet();
	final Object2ObjectOpenHashMap<String, IntOpenHashSet> product2Revs = new Object2ObjectOpenHashMap<>();
	final MutableLong totRevs = new MutableLong();

	// Visit queue
	final LongArrayFIFOQueue queue = new LongArrayFIFOQueue();
	queue.enqueue(startSig);
	result.add(startSig);

	String p = kb.callGraphs.get(index(startSig)).product;
	IntOpenHashSet revs = new IntOpenHashSet();
	revs.add(index(startSig));
	product2Revs.put(p, revs);
	totRevs.increment();


	pl.itemsName = "nodes";
	pl.info = new Object() {
		@Override
		public String toString() {
			return "[nodes: " + result.size() + " products: " + product2Revs.size() + " revisions: " + totRevs.getValue() + "]";
		}
	};
	pl.start("Visiting coreachable nodes...");
	while (!queue.isEmpty()) {
		final long node = queue.dequeueLong();

		for (final long s : kb.predecessors(node)) if (!result.contains(s)) {
			p = kb.callGraphs.get(index(s)).product;
			final String targetNameSpace = kb.new Node(gid(s), index(s)).toFastenURI().getRawNamespace();
			if (targetNameSpace.startsWith("java.") || targetNameSpace.startsWith("javax.") || targetNameSpace.startsWith("jdk.")) continue;
			revs = product2Revs.get(p);
			if (revs == null) product2Revs.put(p, revs = new IntOpenHashSet());
			if (revs.contains(index(s)) || revs.size() < maxRevs) {
				queue.enqueue(s);
				result.add(s);
				//System.out.println(kb.new Node(gid(node), index(node)).toFastenURI() + " -> " + kb.new Node(gid(s), index(s)).toFastenURI());
				if (revs.add(index(s))) totRevs.increment();
			}
		}
		pl.lightUpdate();
	}

	pl.done();
	return new Result(result, product2Revs.size(), totRevs.getValue().longValue());
}
 

public static void main(String arg[]) throws IOException, InterruptedException, JSAPException {

	SimpleJSAP jsap = new SimpleJSAP(WarcCompressor.class.getName(),
		"Given a store uncompressed, write a compressed store.",
		new Parameter[] { new FlaggedOption("output", JSAP.STRING_PARSER, JSAP.NO_DEFAULT, JSAP.REQUIRED, 'o', "output", "The output filename  (- for stdout)."),
		new UnflaggedOption("store", JSAP.STRING_PARSER, JSAP.NOT_REQUIRED, "The name of the store (if omitted, stdin)."),
	});

	JSAPResult jsapResult = jsap.parse(arg);
	if (jsap.messagePrinted()) return;

	final InputStream in = jsapResult.userSpecified("store") ? new FastBufferedInputStream(new FileInputStream(jsapResult.getString("store"))) : System.in;

	final WarcReader reader = new UncompressedWarcReader(in);
	final ProgressLogger pl = new ProgressLogger(LOGGER, 1, TimeUnit.MINUTES, "records");
	final String output = jsapResult.getString("output");

	PrintStream out = "-".equals(output) ? System.out : new PrintStream(new FastBufferedOutputStream(new FileOutputStream(output)), false, "UTF-8");
	final WarcWriter writer = new CompressedWarcWriter(out);

	pl.itemsName = "records";
	pl.displayFreeMemory = true;
	pl.displayLocalSpeed = true;
	pl.start("Scanning...");

	for (long storePosition = 0;; storePosition++) {
	    LOGGER.trace("STOREPOSITION " + storePosition);
	    WarcRecord record = null;
	    try {
		record = reader.read();
	    } catch (Exception e) {
		LOGGER.error("Exception while reading record " + storePosition + " ");
		LOGGER.error(e.getMessage());
		e.printStackTrace();
		continue;
	    }
	    if (record == null)
		break;
	    writer.write(record);
	    pl.lightUpdate();
	}
	pl.done();
	writer.close();
    }
 

@Override
public void run() {
    ProgressLogger pl = new ProgressLogger();
    final int nwords = vectors.size();
    final int d = embeddingsSize;
    try {
        pl.count = count / nbThreads;
        pl.itemsName = "entities";
        final PrintWriter pw = new PrintWriter(new BufferedWriter(new FileWriter(output+"."+rank, false)));
        /*if (rank == 0)
            pw.println(count + " " + d);*/

        float alpha = 10;
        EntityEmbeddings eb = new EntityEmbeddings();
        final BufferedReader br = new BufferedReader(new InputStreamReader(new FileInputStream(new File(input)) , "UTF-8"));
        pl.start();
        String line;
        int nb = 0;
        int nbWritten = 0;
        while((line = br.readLine()) != null) {
            if ((nb < (count/nbThreads)*rank)) {
                nb++; 
                continue;
            }

            if ((rank != nbThreads-1) && (nb > (count/nbThreads)*(rank+1)))
                break;

            String[] parts = line.split("\t");
            if(parts.length > 1) {
                TrainingExamples ex = eb.getVectors(parts[1], vectors, rho, nwords, maxWords);
                if (ex.y.length == 0) {
                    nb++;
                    continue;
                }
                float[] w = eb.trainLR2(ex.x, d, ex.y, alpha);
                pw.print(parts[0] + " ");
                for(int i = 0; i < d; i++) {
                    pw.print(w[i] + " ");
                }
                pw.println();
                pl.lightUpdate();
                nbWritten++;
                if (nbWritten == 1000) {
                    pw.flush();
                    nbWritten = 0;
                }

                for(int i = 0; i < ex.y.length; i++) {
                    if(ex.y[i] > 0) {
                        eb.scoreLR(ex.x[i], w);
                        //double v = eb.scoreLR(ex.x[i], w);
                    }
                }
            }
            nb++;
        }
        br.close();
        pw.close();
        pl.stop();
    } catch(Exception e) {
        e.printStackTrace();
    }
}