LUCENE-1714: fix WriteLineDocTask to also replace \r, \n (in addition to \t) with space so those chars don't create mal-formed lines

git-svn-id: https://svn.apache.org/repos/asf/lucene/java/trunk@787750 13f79535-47bb-0310-9956-ffa450edef68
This commit is contained in:
Michael McCandless 2009-06-23 16:46:17 +00:00
parent 24d827fd70
commit 5f72065d0f
3 changed files with 53 additions and 25 deletions

View File

@ -3,6 +3,11 @@ Lucene Benchmark Contrib Change Log
The Benchmark contrib package contains code for benchmarking Lucene in a variety of ways.
$Id:$
6/23/09
LUCENE-1714: WriteLineDocTask incorrectly normalized text, by replacing only
occurrences of "\t" with a space. It now replaces "\r\n" in addition to that,
so that LineDocMaker won't fail. (Shai Erera via Michael McCandless)
6/17/09
LUCENE-1595: This issue breaks previous external algorithms. DocMaker has been
replaced with a concrete class which accepts a ContentSource for iterating over

View File

@ -22,6 +22,8 @@ import java.io.BufferedWriter;
import java.io.FileOutputStream;
import java.io.OutputStream;
import java.io.OutputStreamWriter;
import java.util.regex.Matcher;
import java.util.regex.Pattern;
import org.apache.commons.compress.compressors.CompressorStreamFactory;
import org.apache.lucene.benchmark.byTask.PerfRunData;
@ -36,7 +38,6 @@ import org.apache.lucene.document.Field;
* taske can be consumed by
* {@link org.apache.lucene.benchmark.byTask.feeds.LineDocMaker} and is intended
* to save the IO overhead of opening a file per doument to be indexed.<br>
*
* Supports the following parameters:
* <ul>
* <li>line.file.out - the name of the file to write the output to. That
@ -45,10 +46,14 @@ import org.apache.lucene.document.Field;
* recommended when the output file is expected to be large. (optional, default:
* false).
* </ul>
* <b>NOTE:</b> this class is not thread-safe and if used by multiple threads the
* output is unspecified (as all will write to the same ouput file in a
* non-synchronized way).
*/
public class WriteLineDocTask extends PerfTask {
public final static char SEP = '\t';
private static final Matcher NORMALIZER = Pattern.compile("[\t\r\n]+").matcher("");
private int docSize = 0;
private BufferedWriter lineFileOut = null;
@ -92,14 +97,14 @@ public class WriteLineDocTask extends PerfTask {
Document doc = docSize > 0 ? docMaker.makeDocument(docSize) : docMaker.makeDocument();
Field f = doc.getField(DocMaker.BODY_FIELD);
String body = f != null ? f.stringValue().replace('\t', ' ') : null;
String body = f != null ? NORMALIZER.reset(f.stringValue()).replaceAll(" ") : null;
if (body != null) {
f = doc.getField(DocMaker.TITLE_FIELD);
String title = f != null ? f.stringValue().replace('\t', ' ') : "";
String title = f != null ? NORMALIZER.reset(f.stringValue()).replaceAll(" ") : "";
f = doc.getField(DocMaker.DATE_FIELD);
String date = f != null ? f.stringValue().replace('\t', ' ') : "";
String date = f != null ? NORMALIZER.reset(f.stringValue()).replaceAll(" ") : "";
lineFileOut.write(title, 0, title.length());
lineFileOut.write(SEP);

View File

@ -27,9 +27,7 @@ import java.util.Properties;
import org.apache.commons.compress.compressors.CompressorStreamFactory;
import org.apache.lucene.benchmark.BenchmarkTestCase;
import org.apache.lucene.benchmark.byTask.PerfRunData;
import org.apache.lucene.benchmark.byTask.feeds.DocData;
import org.apache.lucene.benchmark.byTask.feeds.DocMaker;
import org.apache.lucene.benchmark.byTask.feeds.NoMoreDataException;
import org.apache.lucene.benchmark.byTask.utils.Config;
import org.apache.lucene.document.Document;
import org.apache.lucene.document.Field;
@ -42,10 +40,6 @@ public class WriteLineDocTaskTest extends BenchmarkTestCase {
// class has to be public so that Class.forName.newInstance() will work
public static final class WriteLineDocMaker extends DocMaker {
protected DocData getNextDocData() throws NoMoreDataException, Exception {
throw new UnsupportedOperationException("not implemented");
}
public Document makeDocument() throws Exception {
Document doc = new Document();
doc.add(new Field(BODY_FIELD, "body", Store.NO, Index.NOT_ANALYZED_NO_NORMS));
@ -54,17 +48,28 @@ public class WriteLineDocTaskTest extends BenchmarkTestCase {
return doc;
}
public int numUniqueTexts() {
return 0;
}
// class has to be public so that Class.forName.newInstance() will work
public static final class NewLinesDocMaker extends DocMaker {
public Document makeDocument() throws Exception {
Document doc = new Document();
doc.add(new Field(BODY_FIELD, "body\r\ntext\ttwo", Store.NO, Index.NOT_ANALYZED_NO_NORMS));
doc.add(new Field(TITLE_FIELD, "title\r\ntext", Store.NO, Index.NOT_ANALYZED_NO_NORMS));
doc.add(new Field(DATE_FIELD, "date\r\ntext", Store.NO, Index.NOT_ANALYZED_NO_NORMS));
return doc;
}
}
private static final CompressorStreamFactory csFactory = new CompressorStreamFactory();
private PerfRunData createPerfRunData(File file, boolean setBZCompress, String bz2CompressVal) throws Exception {
private PerfRunData createPerfRunData(File file, boolean setBZCompress,
String bz2CompressVal,
String docMakerName) throws Exception {
Properties props = new Properties();
props.setProperty("doc.maker", WriteLineDocMaker.class.getName());
props.setProperty("doc.maker", docMakerName);
props.setProperty("line.file.out", file.getAbsolutePath());
if (setBZCompress) {
props.setProperty("bzip.compression", bz2CompressVal);
@ -74,7 +79,8 @@ public class WriteLineDocTaskTest extends BenchmarkTestCase {
return new PerfRunData(config);
}
private void doReadTest(File file, boolean bz2File) throws Exception {
private void doReadTest(File file, boolean bz2File, String expTitle,
String expDate, String expBody) throws Exception {
InputStream in = new FileInputStream(file);
if (bz2File) {
in = csFactory.createCompressorInputStream("bzip2", in);
@ -85,9 +91,9 @@ public class WriteLineDocTaskTest extends BenchmarkTestCase {
assertNotNull(line);
String[] parts = line.split(Character.toString(WriteLineDocTask.SEP));
assertEquals(3, parts.length);
assertEquals("title", parts[0]);
assertEquals("date", parts[1]);
assertEquals("body", parts[2]);
assertEquals(expTitle, parts[0]);
assertEquals(expDate, parts[1]);
assertEquals(expBody, parts[2]);
assertNull(br.readLine());
} finally {
br.close();
@ -99,36 +105,48 @@ public class WriteLineDocTaskTest extends BenchmarkTestCase {
// Create a document in bz2 format.
File file = new File(getWorkDir(), "one-line.bz2");
PerfRunData runData = createPerfRunData(file, true, "true");
PerfRunData runData = createPerfRunData(file, true, "true", WriteLineDocMaker.class.getName());
WriteLineDocTask wldt = new WriteLineDocTask(runData);
wldt.doLogic();
wldt.close();
doReadTest(file, true);
doReadTest(file, true, "title", "date", "body");
}
public void testBZip2AutoDetect() throws Exception {
// Create a document in bz2 format.
File file = new File(getWorkDir(), "one-line.bz2");
PerfRunData runData = createPerfRunData(file, false, null);
PerfRunData runData = createPerfRunData(file, false, null, WriteLineDocMaker.class.getName());
WriteLineDocTask wldt = new WriteLineDocTask(runData);
wldt.doLogic();
wldt.close();
doReadTest(file, true);
doReadTest(file, true, "title", "date", "body");
}
public void testRegularFile() throws Exception {
// Create a document in regular format.
File file = new File(getWorkDir(), "one-line");
PerfRunData runData = createPerfRunData(file, true, "false");
PerfRunData runData = createPerfRunData(file, true, "false", WriteLineDocMaker.class.getName());
WriteLineDocTask wldt = new WriteLineDocTask(runData);
wldt.doLogic();
wldt.close();
doReadTest(file, false);
doReadTest(file, false, "title", "date", "body");
}
public void testCharsReplace() throws Exception {
// WriteLineDocTask replaced only \t characters w/ a space, since that's its
// separator char. However, it didn't replace newline characters, which
// resulted in errors in LineDocMaker.
File file = new File(getWorkDir(), "one-line");
PerfRunData runData = createPerfRunData(file, false, null, NewLinesDocMaker.class.getName());
WriteLineDocTask wldt = new WriteLineDocTask(runData);
wldt.doLogic();
wldt.close();
doReadTest(file, false, "title text", "date text", "body text two");
}
}