Menu
▾
▴
archive-access-cvs
|
From: <bra...@us...> - 2008-07-01 23:44:57
|
Revision: 2374
http://archive-access.svn.sourceforge.net/archive-access/?rev=2374&view=rev
Author: bradtofel
Date: 2008-07-01 16:45:04 -0700 (Tue, 01 Jul 2008)
Log Message:
-----------
REFACTOR: SearchResult => (Url|Capture)SearchResult
this includes use of accessor methods
FEATURE: IdentityUrlCanonicalizer is now default!
Modified Paths:
--------------
trunk/archive-access/projects/wayback/wayback-core/src/main/java/org/archive/wayback/resourcestore/indexer/ARCRecordToSearchResultAdapter.java
trunk/archive-access/projects/wayback/wayback-core/src/main/java/org/archive/wayback/resourcestore/indexer/ArcIndexer.java
trunk/archive-access/projects/wayback/wayback-core/src/main/java/org/archive/wayback/resourcestore/indexer/WARCRecordToSearchResultAdapter.java
trunk/archive-access/projects/wayback/wayback-core/src/main/java/org/archive/wayback/resourcestore/indexer/WarcIndexer.java
Modified: trunk/archive-access/projects/wayback/wayback-core/src/main/java/org/archive/wayback/resourcestore/indexer/ARCRecordToSearchResultAdapter.java
===================================================================
--- trunk/archive-access/projects/wayback/wayback-core/src/main/java/org/archive/wayback/resourcestore/indexer/ARCRecordToSearchResultAdapter.java 2008-07-01 23:44:18 UTC (rev 2373)
+++ trunk/archive-access/projects/wayback/wayback-core/src/main/java/org/archive/wayback/resourcestore/indexer/ARCRecordToSearchResultAdapter.java 2008-07-01 23:45:04 UTC (rev 2374)
@@ -26,19 +26,17 @@
import java.io.File;
import java.io.IOException;
-import java.util.logging.Logger;
+//import java.util.logging.Logger;
import org.apache.commons.httpclient.Header;
-import org.apache.commons.httpclient.URIException;
import org.archive.io.arc.ARCRecord;
import org.archive.io.arc.ARCRecordMetaData;
-import org.archive.net.UURI;
-import org.archive.net.UURIFactory;
import org.archive.wayback.UrlCanonicalizer;
import org.archive.wayback.WaybackConstants;
-import org.archive.wayback.core.SearchResult;
+import org.archive.wayback.core.CaptureSearchResult;
import org.archive.wayback.util.Adapter;
-import org.archive.wayback.util.url.AggressiveUrlCanonicalizer;
+import org.archive.wayback.util.url.IdentityUrlCanonicalizer;
+import org.archive.wayback.util.url.UrlOperations;
/**
*
@@ -47,22 +45,22 @@
* @version $Date$, $Revision$
*/
public class ARCRecordToSearchResultAdapter
-implements Adapter<ARCRecord,SearchResult>{
+implements Adapter<ARCRecord,CaptureSearchResult>{
- private static final Logger LOGGER = Logger.getLogger(
- ARCRecordToSearchResultAdapter.class.getName());
+// private static final Logger LOGGER = Logger.getLogger(
+// ARCRecordToSearchResultAdapter.class.getName());
private UrlCanonicalizer canonicalizer = null;
public ARCRecordToSearchResultAdapter() {
- canonicalizer = new AggressiveUrlCanonicalizer();
+ canonicalizer = new IdentityUrlCanonicalizer();
}
// public static SearchResult arcRecordToSearchResult(final ARCRecord rec)
// throws IOException, ParseException {
/* (non-Javadoc)
* @see org.archive.wayback.util.Adapter#adapt(java.lang.Object)
*/
- public SearchResult adapt(ARCRecord rec) {
+ public CaptureSearchResult adapt(ARCRecord rec) {
try {
return adaptInner(rec);
} catch (IOException e) {
@@ -71,26 +69,25 @@
}
}
- private SearchResult adaptInner(ARCRecord rec) throws IOException {
+ private CaptureSearchResult adaptInner(ARCRecord rec) throws IOException {
rec.close();
ARCRecordMetaData meta = rec.getMetaData();
- SearchResult result = new SearchResult();
+ CaptureSearchResult result = new CaptureSearchResult();
String arcName = meta.getArc();
int index = arcName.lastIndexOf(File.separator);
if (index > 0 && (index + 1) < arcName.length()) {
arcName = arcName.substring(index + 1);
}
- result.put(WaybackConstants.RESULT_ARC_FILE, arcName);
- result.put(WaybackConstants.RESULT_OFFSET, String.valueOf(meta
- .getOffset()));
+ result.setFile(arcName);
+ result.setOffset(meta.getOffset());
// initialize with default HTTP code...
- result.put(WaybackConstants.RESULT_HTTP_CODE, "-");
+ result.setHttpCode("-");
- result.put(WaybackConstants.RESULT_MD5_DIGEST, rec.getDigestStr());
- result.put(WaybackConstants.RESULT_MIME_TYPE, meta.getMimetype());
- result.put(WaybackConstants.RESULT_CAPTURE_DATE, meta.getDate());
+ result.setDigest(rec.getDigestStr());
+ result.setMimeType(meta.getMimetype());
+ result.setCaptureTimestamp(meta.getDate());
String uriStr = meta.getUrl();
if (uriStr.startsWith(ARCRecord.ARC_MAGIC_NUMBER)) {
@@ -100,67 +97,49 @@
if (uriStr.startsWith(WaybackConstants.DNS_URL_PREFIX)) {
// skip URL + HTTP header processing for dns records...
- String origHost = uriStr.substring(WaybackConstants.DNS_URL_PREFIX
- .length());
- result.put(WaybackConstants.RESULT_ORIG_HOST, origHost);
- result.put(WaybackConstants.RESULT_REDIRECT_URL, "-");
- result.put(WaybackConstants.RESULT_URL, uriStr);
- result.put(WaybackConstants.RESULT_URL_KEY, uriStr);
+ result.setOriginalUrl(uriStr);
+ result.setRedirectUrl("-");
+ result.setUrlKey(uriStr);
} else {
- UURI uri = UURIFactory.getInstance(uriStr);
- result.put(WaybackConstants.RESULT_URL, uriStr);
+ result.setOriginalUrl(uriStr);
- String uriHost = uri.getHost();
- if (uriHost == null) {
- LOGGER.info("No host in " + uriStr + " in " + meta.getArc());
- } else {
- result.put(WaybackConstants.RESULT_ORIG_HOST, uriHost);
- String statusCode = (meta.getStatusCode() == null) ? "-" : meta
- .getStatusCode();
- result.put(WaybackConstants.RESULT_HTTP_CODE, statusCode);
-
- String redirectUrl = "-";
- Header[] headers = rec.getHttpHeaders();
- if (headers != null) {
-
- for (int i = 0; i < headers.length; i++) {
- if (headers[i].getName().equals(
- WaybackConstants.LOCATION_HTTP_HEADER)) {
+ String statusCode = (meta.getStatusCode() == null) ? "-" : meta
+ .getStatusCode();
+ result.setHttpCode(statusCode);
+
+ String redirectUrl = "-";
+ Header[] headers = rec.getHttpHeaders();
+ if (headers != null) {
+
+ for (int i = 0; i < headers.length; i++) {
+ if (headers[i].getName().equals(
+ WaybackConstants.LOCATION_HTTP_HEADER)) {
- String locationStr = headers[i].getValue();
- // TODO: "Location" is supposed to be absolute:
- // (http://www.w3.org/Protocols/rfc2616/rfc2616-sec14.html)
- // (section 14.30) but Content-Location can be
- // relative.
- // is it correct to resolve a relative Location, as
- // we are?
- // it's also possible to have both in the HTTP
- // headers...
- // should we prefer one over the other?
- // right now, we're ignoring "Content-Location"
- try {
- UURI uriRedirect = UURIFactory.getInstance(uri,
- locationStr);
- redirectUrl = uriRedirect.getEscapedURI();
-
- } catch (URIException e) {
- LOGGER.info("Bad Location: " + locationStr
- + " for " + uriStr + " in "
- + meta.getArc() + " Skipped");
- }
- break;
- }
+ String locationStr = headers[i].getValue();
+ // TODO: "Location" is supposed to be absolute:
+ // (http://www.w3.org/Protocols/rfc2616/rfc2616-sec14.html)
+ // (section 14.30) but Content-Location can be
+ // relative.
+ // is it correct to resolve a relative Location, as
+ // we are?
+ // it's also possible to have both in the HTTP
+ // headers...
+ // should we prefer one over the other?
+ // right now, we're ignoring "Content-Location"
+ redirectUrl = UrlOperations.resolveUrl(uriStr,
+ locationStr);
+
+ break;
}
}
- result.put(WaybackConstants.RESULT_REDIRECT_URL, redirectUrl);
+ result.setRedirectUrl(redirectUrl);
- String indexUrl = canonicalizer.urlStringToKey(meta.getUrl());
- result.put(WaybackConstants.RESULT_URL_KEY, indexUrl);
+ String urlKey = canonicalizer.urlStringToKey(meta.getUrl());
+ result.setUrlKey(urlKey);
}
-
}
return result;
}
Modified: trunk/archive-access/projects/wayback/wayback-core/src/main/java/org/archive/wayback/resourcestore/indexer/ArcIndexer.java
===================================================================
--- trunk/archive-access/projects/wayback/wayback-core/src/main/java/org/archive/wayback/resourcestore/indexer/ArcIndexer.java 2008-07-01 23:44:18 UTC (rev 2373)
+++ trunk/archive-access/projects/wayback/wayback-core/src/main/java/org/archive/wayback/resourcestore/indexer/ArcIndexer.java 2008-07-01 23:45:04 UTC (rev 2374)
@@ -34,7 +34,7 @@
import org.archive.io.arc.ARCReaderFactory;
import org.archive.io.arc.ARCRecord;
import org.archive.wayback.UrlCanonicalizer;
-import org.archive.wayback.core.SearchResult;
+import org.archive.wayback.core.CaptureSearchResult;
import org.archive.wayback.resourceindex.cdx.SearchResultToCDXLineAdapter;
import org.archive.wayback.util.AdaptedIterator;
import org.archive.wayback.util.Adapter;
@@ -43,7 +43,7 @@
import org.archive.wayback.util.url.IdentityUrlCanonicalizer;
/**
- * Transforms an ARC file into Iterator<SearchResult>.
+ * Transforms an ARC file into Iterator<CaptureSearchResult>.
*
* @author brad
* @version $Date$, $Revision$
@@ -65,7 +65,7 @@
* @return Iterator of SearchResults for input arc File
* @throws IOException
*/
- public CloseableIterator<SearchResult> iterator(File arc)
+ public CloseableIterator<CaptureSearchResult> iterator(File arc)
throws IOException {
return iterator(ARCReaderFactory.get(arc));
}
@@ -75,7 +75,7 @@
* @return Iterator of SearchResults for input pathOrUrl
* @throws IOException
*/
- public CloseableIterator<SearchResult> iterator(String pathOrUrl)
+ public CloseableIterator<CaptureSearchResult> iterator(String pathOrUrl)
throws IOException {
return iterator(ARCReaderFactory.get(pathOrUrl));
}
@@ -85,7 +85,7 @@
* @return Iterator of SearchResults for input ARCReader
* @throws IOException
*/
- public CloseableIterator<SearchResult> iterator(ARCReader arcReader)
+ public CloseableIterator<CaptureSearchResult> iterator(ARCReader arcReader)
throws IOException {
arcReader.setParseHttpHeaders(true);
@@ -102,7 +102,7 @@
CloseableIterator<ARCRecord> itr2 =
new AdaptedIterator<ArchiveRecord,ARCRecord>(itr1,adapter1);
- return new AdaptedIterator<ARCRecord,SearchResult>(itr2,adapter2);
+ return new AdaptedIterator<ARCRecord,CaptureSearchResult>(itr2,adapter2);
}
public UrlCanonicalizer getCanonicalizer() {
@@ -146,7 +146,7 @@
} else {
USAGE();
}
- Iterator<SearchResult> res = indexer.iterator(arc);
+ Iterator<CaptureSearchResult> res = indexer.iterator(arc);
Iterator<String> lines = SearchResultToCDXLineAdapter.adapt(res);
while(lines.hasNext()) {
pw.println(lines.next());
Modified: trunk/archive-access/projects/wayback/wayback-core/src/main/java/org/archive/wayback/resourcestore/indexer/WARCRecordToSearchResultAdapter.java
===================================================================
--- trunk/archive-access/projects/wayback/wayback-core/src/main/java/org/archive/wayback/resourcestore/indexer/WARCRecordToSearchResultAdapter.java 2008-07-01 23:44:18 UTC (rev 2373)
+++ trunk/archive-access/projects/wayback/wayback-core/src/main/java/org/archive/wayback/resourcestore/indexer/WARCRecordToSearchResultAdapter.java 2008-07-01 23:45:04 UTC (rev 2374)
@@ -2,25 +2,23 @@
import java.io.File;
import java.io.IOException;
-import java.util.logging.Logger;
+//import java.util.logging.Logger;
import org.apache.commons.httpclient.Header;
import org.apache.commons.httpclient.HttpParser;
import org.apache.commons.httpclient.StatusLine;
-import org.apache.commons.httpclient.URIException;
import org.apache.commons.httpclient.util.EncodingUtil;
import org.archive.io.ArchiveRecordHeader;
import org.archive.io.RecoverableIOException;
import org.archive.io.arc.ARCConstants;
import org.archive.io.warc.WARCConstants;
import org.archive.io.warc.WARCRecord;
-import org.archive.net.UURI;
-import org.archive.net.UURIFactory;
import org.archive.wayback.UrlCanonicalizer;
import org.archive.wayback.WaybackConstants;
-import org.archive.wayback.core.SearchResult;
+import org.archive.wayback.core.CaptureSearchResult;
import org.archive.wayback.util.Adapter;
import org.archive.wayback.util.url.AggressiveUrlCanonicalizer;
+import org.archive.wayback.util.url.UrlOperations;
/**
* Adapts certain WARCRecords into SearchResults. DNS and response records are
@@ -34,24 +32,12 @@
* @version $Date$, $Revision$
*/
public class WARCRecordToSearchResultAdapter
-implements Adapter<WARCRecord,SearchResult>{
+implements Adapter<WARCRecord,CaptureSearchResult>{
private final static String DEFAULT_VALUE = "-";
- private final static String SEARCH_FIELDS[] = {
- WaybackConstants.RESULT_URL,
- WaybackConstants.RESULT_URL_KEY,
- WaybackConstants.RESULT_ORIG_HOST,
- WaybackConstants.RESULT_CAPTURE_DATE,
- WaybackConstants.RESULT_MD5_DIGEST,
- WaybackConstants.RESULT_MIME_TYPE,
- WaybackConstants.RESULT_HTTP_CODE,
- WaybackConstants.RESULT_REDIRECT_URL,
- WaybackConstants.RESULT_ARC_FILE,
- WaybackConstants.RESULT_OFFSET,
- };
- private static final Logger LOGGER = Logger.getLogger(
- WARCRecordToSearchResultAdapter.class.getName());
+// private static final Logger LOGGER = Logger.getLogger(
+// WARCRecordToSearchResultAdapter.class.getName());
private UrlCanonicalizer canonicalizer = null;
@@ -62,7 +48,7 @@
/* (non-Javadoc)
* @see org.archive.wayback.util.Adapter#adapt(java.lang.Object)
*/
- public SearchResult adapt(WARCRecord rec) {
+ public CaptureSearchResult adapt(WARCRecord rec) {
try {
return adaptInner(rec);
} catch (IOException e) {
@@ -117,75 +103,58 @@
return orig;
}
- private SearchResult getBlankSearchResult() {
- SearchResult result = new SearchResult();
- for(String field : SEARCH_FIELDS) {
- result.put(field, DEFAULT_VALUE);
- }
+ private CaptureSearchResult getBlankSearchResult() {
+ CaptureSearchResult result = new CaptureSearchResult();
+
+ result.setUrlKey(DEFAULT_VALUE);
+ result.setOriginalUrl(DEFAULT_VALUE);
+ result.setCaptureTimestamp(DEFAULT_VALUE);
+ result.setDigest(DEFAULT_VALUE);
+ result.setMimeType(DEFAULT_VALUE);
+ result.setHttpCode(DEFAULT_VALUE);
+ result.setRedirectUrl(DEFAULT_VALUE);
+ result.setFile(DEFAULT_VALUE);
+ result.setOffset(0);
return result;
}
- private UURI addUrlDataToSearchResult(SearchResult result, String urlStr)
+ private void addUrlDataToSearchResult(CaptureSearchResult result, String urlStr)
throws IOException {
- result.put(WaybackConstants.RESULT_URL, urlStr);
- result.put(WaybackConstants.RESULT_URL_KEY, urlStr);
-
-
- UURI uri = UURIFactory.getInstance(urlStr);
- String uriHost = uri.getHost();
- if (uriHost == null) {
-
- LOGGER.info("No host in " + urlStr);
-
- } else {
-
- result.put(WaybackConstants.RESULT_ORIG_HOST, uriHost);
- }
-
+ result.setOriginalUrl(urlStr);
String urlKey = canonicalizer.urlStringToKey(urlStr);
- result.put(WaybackConstants.RESULT_URL_KEY, urlKey);
-
- return uri;
+ result.setUrlKey(urlKey);
}
- private SearchResult adaptDNS(ArchiveRecordHeader header, WARCRecord rec)
+ private CaptureSearchResult adaptDNS(ArchiveRecordHeader header, WARCRecord rec)
throws IOException {
- SearchResult result = getBlankSearchResult();
+ CaptureSearchResult result = getBlankSearchResult();
- result.put(WaybackConstants.RESULT_CAPTURE_DATE,
- transformDate(header.getDate()));
- result.put(WaybackConstants.RESULT_ARC_FILE,
- transformWarcFilename(header.getReaderIdentifier()));
- result.put(WaybackConstants.RESULT_OFFSET,
- String.valueOf(header.getOffset()));
+ result.setCaptureTimestamp(transformDate(header.getDate()));
+ result.setFile(transformWarcFilename(header.getReaderIdentifier()));
+ result.setOffset(header.getOffset());
String uriStr = header.getUrl();
- String origHost = uriStr.substring(WaybackConstants.DNS_URL_PREFIX
- .length());
- result.put(WaybackConstants.RESULT_MIME_TYPE, header.getMimetype());
+ result.setMimeType(header.getMimetype());
- result.put(WaybackConstants.RESULT_ORIG_HOST, origHost);
- result.put(WaybackConstants.RESULT_URL, uriStr);
- result.put(WaybackConstants.RESULT_URL_KEY, uriStr);
+ result.setOriginalUrl(uriStr);
+ result.setUrlKey(uriStr);
rec.close();
- result.put(WaybackConstants.RESULT_MD5_DIGEST, rec.getDigestStr());
+ result.setDigest(rec.getDigestStr());
return result;
}
- private SearchResult adaptRevisit(ArchiveRecordHeader header, WARCRecord rec)
+ private CaptureSearchResult adaptRevisit(ArchiveRecordHeader header, WARCRecord rec)
throws IOException {
- SearchResult result = getBlankSearchResult();
+ CaptureSearchResult result = getBlankSearchResult();
- result.put(WaybackConstants.RESULT_CAPTURE_DATE,
- transformDate(header.getDate()));
- result.put(WaybackConstants.RESULT_MD5_DIGEST,
- transformDigest(header.getHeaderValue(
+ result.setCaptureTimestamp(transformDate(header.getDate()));
+ result.setDigest(transformDigest(header.getHeaderValue(
WARCRecord.HEADER_KEY_PAYLOAD_DIGEST)));
addUrlDataToSearchResult(result,header.getUrl());
@@ -211,20 +180,17 @@
return count;
}
- private SearchResult adaptResponse(ArchiveRecordHeader header, WARCRecord rec)
+ private CaptureSearchResult adaptResponse(ArchiveRecordHeader header, WARCRecord rec)
throws IOException {
- SearchResult result = getBlankSearchResult();
+ CaptureSearchResult result = getBlankSearchResult();
- result.put(WaybackConstants.RESULT_CAPTURE_DATE,
- transformDate(header.getDate()));
- result.put(WaybackConstants.RESULT_ARC_FILE,
- transformWarcFilename(header.getReaderIdentifier()));
- result.put(WaybackConstants.RESULT_OFFSET,
- String.valueOf(header.getOffset()));
+ result.setCaptureTimestamp(transformDate(header.getDate()));
+ result.setFile(transformWarcFilename(header.getReaderIdentifier()));
+ result.setOffset(header.getOffset());
String origUrl = header.getUrl();
- UURI uri = addUrlDataToSearchResult(result,origUrl);
+ addUrlDataToSearchResult(result,origUrl);
// need to parse the documents HTTP message and headers here: WARCReader
// does not implement this... yet..
@@ -242,15 +208,13 @@
throw new RecoverableIOException("Failed parse of http status line.");
}
StatusLine status = new StatusLine(statusLine);
- result.put(WaybackConstants.RESULT_HTTP_CODE,
- String.valueOf(status.getStatusCode()));
+ result.setHttpCode(String.valueOf(status.getStatusCode()));
Header[] headers = HttpParser.parseHeaders(rec,
ARCConstants.DEFAULT_ENCODING);
rec.close();
- result.put(WaybackConstants.RESULT_MD5_DIGEST,
- transformDigest(header.getHeaderValue(
+ result.setDigest(transformDigest(header.getHeaderValue(
WARCRecord.HEADER_KEY_PAYLOAD_DIGEST)));
if (headers != null) {
@@ -270,28 +234,19 @@
// headers...
// should we prefer one over the other?
// right now, we're ignoring "Content-Location"
- try {
- UURI uriRedirect = UURIFactory.getInstance(uri,
- locationStr);
- result.put(WaybackConstants.RESULT_REDIRECT_URL,
- uriRedirect.getEscapedURI());
- } catch (URIException e) {
- LOGGER.info("Bad Location: " + locationStr
- + " for " + origUrl + " in "
- + header.getReaderIdentifier() + " Skipped");
- }
+ result.setRedirectUrl(
+ UrlOperations.resolveUrl(origUrl, locationStr));
} else if(httpHeader.getName().toLowerCase().equals("content-type")) {
- result.put(WaybackConstants.RESULT_MIME_TYPE,
- transformHTTPMime(httpHeader.getValue()));
+ result.setMimeType(transformHTTPMime(httpHeader.getValue()));
}
}
}
return result;
}
- private SearchResult adaptInner(WARCRecord rec) throws IOException {
+ private CaptureSearchResult adaptInner(WARCRecord rec) throws IOException {
- SearchResult result = null;
+ CaptureSearchResult result = null;
ArchiveRecordHeader header = rec.getHeader();
String type = header.getHeaderValue(WARCConstants.HEADER_KEY_TYPE).toString();
if(type.equals(WARCConstants.RESPONSE)) {
Modified: trunk/archive-access/projects/wayback/wayback-core/src/main/java/org/archive/wayback/resourcestore/indexer/WarcIndexer.java
===================================================================
--- trunk/archive-access/projects/wayback/wayback-core/src/main/java/org/archive/wayback/resourcestore/indexer/WarcIndexer.java 2008-07-01 23:44:18 UTC (rev 2373)
+++ trunk/archive-access/projects/wayback/wayback-core/src/main/java/org/archive/wayback/resourcestore/indexer/WarcIndexer.java 2008-07-01 23:45:04 UTC (rev 2374)
@@ -10,7 +10,7 @@
import org.archive.io.warc.WARCReaderFactory;
import org.archive.io.warc.WARCRecord;
import org.archive.wayback.UrlCanonicalizer;
-import org.archive.wayback.core.SearchResult;
+import org.archive.wayback.core.CaptureSearchResult;
import org.archive.wayback.resourceindex.cdx.SearchResultToCDXLineAdapter;
import org.archive.wayback.util.AdaptedIterator;
import org.archive.wayback.util.Adapter;
@@ -35,7 +35,7 @@
* @return Iterator of SearchResults for input arc File
* @throws IOException
*/
- public CloseableIterator<SearchResult> iterator(File warc)
+ public CloseableIterator<CaptureSearchResult> iterator(File warc)
throws IOException {
return iterator(WARCReaderFactory.get(warc));
}
@@ -44,7 +44,7 @@
* @return Iterator of SearchResults for input pathOrUrl
* @throws IOException
*/
- public CloseableIterator<SearchResult> iterator(String pathOrUrl)
+ public CloseableIterator<CaptureSearchResult> iterator(String pathOrUrl)
throws IOException {
return iterator(WARCReaderFactory.get(pathOrUrl));
}
@@ -53,7 +53,7 @@
* @return Iterator of SearchResults for input arc File
* @throws IOException
*/
- public CloseableIterator<SearchResult> iterator(WARCReader reader)
+ public CloseableIterator<CaptureSearchResult> iterator(WARCReader reader)
throws IOException {
Adapter<ArchiveRecord, WARCRecord> adapter1 = new ArchiveRecordToWARCRecordAdapter();
@@ -68,7 +68,7 @@
CloseableIterator<WARCRecord> itr2 =
new AdaptedIterator<ArchiveRecord, WARCRecord>(itr1, adapter1);
- return new AdaptedIterator<WARCRecord, SearchResult>(itr2, adapter2);
+ return new AdaptedIterator<WARCRecord, CaptureSearchResult>(itr2, adapter2);
}
public UrlCanonicalizer getCanonicalizer() {
@@ -112,7 +112,7 @@
} else {
USAGE();
}
- Iterator<SearchResult> res = indexer.iterator(arc);
+ Iterator<CaptureSearchResult> res = indexer.iterator(arc);
Iterator<String> lines = SearchResultToCDXLineAdapter.adapt(res);
while (lines.hasNext()) {
pw.println(lines.next());
This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.
|
×
Want the latest updates on software, tech news, and AI?
Get latest updates about software, tech news, and AI from SourceForge directly in your inbox once a month.
Thanks for helping keep SourceForge clean.
X