Merge branch 'master' into fetchhttp-hc

Conflicts:
	modules/src/main/java/org/archive/modules/fetcher/FetchHTTP.java

(conflicts are resolved)
This commit is contained in:
Noah Levitt
2012-07-15 11:58:41 -07:00
60 changed files with 75 additions and 23 deletions
@@ -59,6 +59,7 @@ import org.springframework.context.ApplicationContextAware;
public class ScriptedProcessor extends Processor
implements ApplicationContextAware, InitializingBean {
@SuppressWarnings("unused")
private static final long serialVersionUID = 3L;
private static final Logger logger =
@@ -18,6 +18,8 @@
*/
package org.archive.modules.deciderules;
import java.util.logging.Logger;
import org.archive.modules.CrawlURI;
/**
@@ -30,6 +32,8 @@ import org.archive.modules.CrawlURI;
public class TooManyPathSegmentsDecideRule extends PredicatedDecideRule {
private static final long serialVersionUID = 3L;
@SuppressWarnings("unused")
private static final Logger logger = Logger.getLogger(TooManyPathSegmentsDecideRule.class.getName());
/** default for this class is to REJECT */
{
@@ -63,11 +67,15 @@ public class TooManyPathSegmentsDecideRule extends PredicatedDecideRule {
*/
@Override
protected boolean evaluate(CrawlURI curi) {
String uri = curi.toString();
String uriPath = curi.getUURI().getEscapedPath();
if (uriPath == null) {
// no path means no segments
return false;
}
int count = 0;
int threshold = getMaxPathDepth();
for (int i = 0; i < uri.length(); i++) {
if (uri.charAt(i) == '/') {
for (int i = 0; i < uriPath.length(); i++) {
if (uriPath.charAt(i) == '/') {
count++;
}
if (count > threshold) {
@@ -32,6 +32,7 @@ import org.archive.modules.CrawlURI;
*/
public class AggressiveExtractorHTML
extends ExtractorHTML {
@SuppressWarnings("unused")
private static final long serialVersionUID = 3L;
protected static Logger logger =
@@ -57,6 +57,7 @@ import org.archive.util.TextUtils;
public class ExtractorCSS extends ContentExtractor {
@SuppressWarnings("unused")
private static final long serialVersionUID = 2L;
private static Logger logger =
@@ -42,6 +42,7 @@ import org.archive.util.ms.Doc;
*/
public class ExtractorDOC extends ContentExtractor {
@SuppressWarnings("unused")
private static final long serialVersionUID = 3L;
private static Pattern PATTERN = Pattern.compile("HYPERLINK.*?\"(.*?)\"");
@@ -61,6 +61,7 @@ import org.springframework.beans.factory.annotation.Autowired;
*/
public class ExtractorHTML extends ContentExtractor implements InitializingBean {
@SuppressWarnings("unused")
private static final long serialVersionUID = 2L;
private static Logger logger =
@@ -32,6 +32,7 @@ import org.archive.net.UURIFactory;
*/
public class ExtractorHTTP extends Extractor {
@SuppressWarnings("unused")
private static final long serialVersionUID = 3L;
public ExtractorHTTP() {
@@ -51,6 +51,7 @@ import org.archive.net.UURIFactory;
public class ExtractorImpliedURI extends Extractor {
@SuppressWarnings("unused")
private static final long serialVersionUID = 3L;
private static Logger LOGGER =
@@ -56,6 +56,7 @@ import org.archive.util.UriUtils;
*/
public class ExtractorJS extends ContentExtractor {
@SuppressWarnings("unused")
private static final long serialVersionUID = 2L;
private static Logger LOGGER =
@@ -38,6 +38,7 @@ import org.archive.util.FileUtils;
*/
public class ExtractorPDF extends ContentExtractor {
@SuppressWarnings("unused")
private static final long serialVersionUID = 3L;
private static final Logger LOGGER =
@@ -46,6 +46,7 @@ import com.anotherbigidea.io.InStream;
*/
public class ExtractorSWF extends ContentExtractor {
@SuppressWarnings("unused")
private static final long serialVersionUID = 3L;
private static Logger logger =
@@ -48,6 +48,7 @@ import org.archive.util.TextUtils;
public class ExtractorURI extends Extractor {
@SuppressWarnings("unused")
private static final long serialVersionUID = 3L;
private static Logger LOGGER =
@@ -49,6 +49,7 @@ import org.archive.net.UURIFactory;
*/
public class ExtractorUniversal extends ContentExtractor {
@SuppressWarnings("unused")
private static final long serialVersionUID = 3L;
/**
@@ -49,6 +49,7 @@ import org.archive.util.UriUtils;
*/
public class ExtractorXML extends ContentExtractor {
@SuppressWarnings("unused")
private static final long serialVersionUID = 3L;
private static Logger logger =
@@ -65,6 +65,7 @@ import org.archive.util.TextUtils;
*/
public class HTTPContentDigest extends Processor {
@SuppressWarnings("unused")
private static final long serialVersionUID = 3L;
private static Logger logger =
@@ -76,6 +76,7 @@ import au.id.jericho.lib.html.StartTagType;
@SuppressWarnings("unchecked")
public class JerichoExtractorHTML extends ExtractorHTML {
@SuppressWarnings("unused")
private static final long serialVersionUID = 1684681316546343615L;
final private static Logger logger =
@@ -28,6 +28,7 @@ import org.archive.modules.CrawlURI;
*
*/
public class TrapSuppressExtractor extends ContentExtractor {
@SuppressWarnings("unused")
private static final long serialVersionUID = -1028783453022579530L;
/** ALIst attribute key for carrying-forward content-digest from 'via'*/
@@ -43,6 +43,7 @@ import com.sleepycat.je.DatabaseException;
* @author pjack
*/
public class BdbCookieStorage extends AbstractCookieStorage implements Checkpointable {
@SuppressWarnings("unused")
private static final long serialVersionUID = 1L;
protected BdbModule bdb;
@@ -61,6 +61,7 @@ import org.xbill.DNS.Type;
*/
public class FetchDNS extends Processor {
@SuppressWarnings("unused")
private static final long serialVersionUID = 3L;
private static Logger logger = Logger.getLogger(FetchDNS.class.getName());
@@ -62,6 +62,7 @@ import org.archive.util.Recorder;
*
*/
public class FetchFTP extends Processor {
@SuppressWarnings("unused")
private static final long serialVersionUID = 1L;
/** Logger for this class. */
@@ -122,6 +122,7 @@ import org.springframework.context.Lifecycle;
* @version $Id$
*/
public class FetchHTTP extends AbstractFetchHTTP implements Lifecycle {
@SuppressWarnings("unused")
private static final long serialVersionUID = 1L;
private static Logger logger = Logger.getLogger(FetchHTTP.class.getName());
@@ -122,6 +122,7 @@ import com.sleepycat.je.DatabaseException;
public class FetchWhois extends Processor implements CoreAttributeConstants,
FetchStatusCodes, Lifecycle {
@SuppressWarnings("unused")
private static final long serialVersionUID = 1L;
private static Logger logger = Logger.getLogger(FetchWhois.class.getName());
@@ -26,6 +26,7 @@ import org.apache.commons.httpclient.Cookie;
public class SimpleCookieStorage extends AbstractCookieStorage {
@SuppressWarnings("unused")
private static final long serialVersionUID = 1L;
final private SortedMap<String,Cookie> map = new TreeMap<String,Cookie>();
@@ -40,6 +40,7 @@ import org.archive.modules.Processor;
* @version $Date: 2006-09-25 20:19:54 +0000 (Mon, 25 Sep 2006) $, $Revision: 4654 $
*/
public class FetchHistoryProcessor extends Processor {
@SuppressWarnings("unused")
private static final long serialVersionUID = 1L;
/** Desired history array length. */
@@ -39,6 +39,7 @@ import com.sleepycat.je.DatabaseException;
* @version $Date: 2006-09-25 20:19:54 +0000 (Mon, 25 Sep 2006) $, $Revision: 4654 $
*/
public class PersistLoadProcessor extends PersistOnlineProcessor {
@SuppressWarnings("unused")
private static final long serialVersionUID = -1917169316015093131L;
private static final Logger logger =
Logger.getLogger(PersistLoadProcessor.class.getName());
@@ -43,6 +43,7 @@ import org.springframework.context.Lifecycle;
public class PersistLogProcessor extends PersistProcessor
implements Checkpointable, Lifecycle {
@SuppressWarnings("unused")
private static final long serialVersionUID = 1678691994065439346L;
protected CrawlerJournal log;
@@ -40,6 +40,7 @@ import com.sleepycat.je.DatabaseException;
public abstract class PersistOnlineProcessor extends PersistProcessor
implements Lifecycle {
@SuppressWarnings("unused")
private static final long serialVersionUID = -666479480942267268L;
protected BdbModule bdb;
@@ -64,6 +64,7 @@ import com.sleepycat.je.EnvironmentConfig;
*/
public abstract class PersistProcessor extends AbstractPersistProcessor {
@SuppressWarnings("unused")
private static final long serialVersionUID = 1L;
private static final Logger logger =
@@ -29,6 +29,7 @@ import org.archive.modules.CrawlURI;
*/
public class PersistStoreProcessor extends PersistOnlineProcessor
{
@SuppressWarnings("unused")
private static final long serialVersionUID = -8308356194337303758L;
// class description: "PersistStoreProcessor. Stores CrawlURI attributes " +
@@ -61,6 +61,7 @@ public class ARCWriterProcessor extends WriterPoolProcessor {
final static private String METADATA_TEMPLATE = readMetadataTemplate();
@SuppressWarnings("unused")
private static final long serialVersionUID = 3L;
private static final Logger logger =
@@ -78,6 +78,7 @@ import org.springframework.beans.factory.annotation.Autowired;
public class Kw3WriterProcessor extends Processor {
@SuppressWarnings("unused")
private static final long serialVersionUID = 3L;
private static String COLON = ":";
@@ -86,6 +86,7 @@ import org.archive.util.FileUtils;
@author Howard Lee Gayle
*/
public class MirrorWriterProcessor extends Processor {
@SuppressWarnings("unused")
private static final long serialVersionUID = 3L;
private static final Logger logger =
Logger.getLogger(MirrorWriterProcessor.class.getName());
@@ -99,6 +99,7 @@ import org.archive.util.anvl.ANVLRecord;
* @contributor stack
*/
public class WARCWriterProcessor extends WriterPoolProcessor implements WARCWriterPoolSettings {
@SuppressWarnings("unused")
private static final long serialVersionUID = 6182850087635847443L;
private static final Logger logger =
Logger.getLogger(WARCWriterProcessor.class.getName());
@@ -61,6 +61,7 @@ import org.springframework.context.Lifecycle;
*/
public abstract class WriterPoolProcessor extends Processor
implements Lifecycle, Checkpointable, WriterPoolSettings {
@SuppressWarnings("unused")
private static final long serialVersionUID = 1L;
private static final Logger logger =
Logger.getLogger(WriterPoolProcessor.class.getName());
@@ -14,6 +14,7 @@ import org.archive.net.UURI;
*
*/
public class UnitTestUriLoggerModule implements UriErrorLoggerModule {
@SuppressWarnings("unused")
private static final long serialVersionUID = 1L;
final private static Logger LOGGER =