mirror of
https://github.com/MarginaliaSearch/MarginaliaSearch.git
synced 2025-10-08 00:02:38 +02:00
Compare commits
3 Commits
deploy-019
...
deploy-019
Author | SHA1 | Date | |
---|---|---|---|
|
f59a9eb025 | ||
|
599534806b | ||
|
7e8253dac7 |
@@ -112,14 +112,6 @@ public class EdgeDomain implements Serializable {
|
|||||||
return topDomain;
|
return topDomain;
|
||||||
}
|
}
|
||||||
|
|
||||||
public String getDomainKey() {
|
|
||||||
int cutPoint = topDomain.indexOf('.');
|
|
||||||
if (cutPoint < 0) {
|
|
||||||
return topDomain;
|
|
||||||
}
|
|
||||||
return topDomain.substring(0, cutPoint).toLowerCase();
|
|
||||||
}
|
|
||||||
|
|
||||||
/** If possible, try to provide an alias domain,
|
/** If possible, try to provide an alias domain,
|
||||||
* i.e. a domain name that is very likely to link to this one
|
* i.e. a domain name that is very likely to link to this one
|
||||||
* */
|
* */
|
||||||
|
@@ -8,14 +8,6 @@ import static org.junit.jupiter.api.Assertions.assertEquals;
|
|||||||
|
|
||||||
class EdgeDomainTest {
|
class EdgeDomainTest {
|
||||||
|
|
||||||
@Test
|
|
||||||
public void testSkepdic() throws URISyntaxException {
|
|
||||||
var domain = new EdgeUrl("http://www.skepdic.com/astrology.html");
|
|
||||||
assertEquals("skepdic", domain.getDomain().getDomainKey());
|
|
||||||
var domain2 = new EdgeUrl("http://skepdic.com/astrology.html");
|
|
||||||
assertEquals("skepdic", domain2.getDomain().getDomainKey());
|
|
||||||
}
|
|
||||||
|
|
||||||
@Test
|
@Test
|
||||||
public void testHkDomain() throws URISyntaxException {
|
public void testHkDomain() throws URISyntaxException {
|
||||||
var domain = new EdgeUrl("http://l7072i3.l7c.net");
|
var domain = new EdgeUrl("http://l7072i3.l7c.net");
|
||||||
|
@@ -61,7 +61,7 @@ public class UrlDeduplicator {
|
|||||||
|
|
||||||
private boolean limitResultsPerDomain(DecoratedSearchResultItem details) {
|
private boolean limitResultsPerDomain(DecoratedSearchResultItem details) {
|
||||||
final var domain = details.getUrl().getDomain();
|
final var domain = details.getUrl().getDomain();
|
||||||
final String key = domain.getDomainKey();
|
final String key = domain.toString();
|
||||||
|
|
||||||
return keyCount.adjustOrPutValue(key, 1, 1) <= resultsPerKey;
|
return keyCount.adjustOrPutValue(key, 1, 1) <= resultsPerKey;
|
||||||
}
|
}
|
||||||
|
@@ -112,13 +112,6 @@ public class SearchOperator {
|
|||||||
.selectStrategy(queryResponse)
|
.selectStrategy(queryResponse)
|
||||||
.clusterResults(queryResults, 25);
|
.clusterResults(queryResults, 25);
|
||||||
|
|
||||||
if (queryParams.humanQuery().equals("slackware linux")) {
|
|
||||||
logger.info("Query response: {}", queryResponse.results().subList(0, 5));
|
|
||||||
logger.info("Query results: {}", queryResults.subList(0, 5));
|
|
||||||
logger.info("Clustered results: {}", clusteredResults.subList(0, 5));
|
|
||||||
}
|
|
||||||
|
|
||||||
|
|
||||||
// Log the query and results
|
// Log the query and results
|
||||||
|
|
||||||
logger.info(queryMarker, "Human terms: {}", Strings.join(queryResponse.searchTermsHuman(), ','));
|
logger.info(queryMarker, "Human terms: {}", Strings.join(queryResponse.searchTermsHuman(), ','));
|
||||||
|
@@ -25,28 +25,14 @@ public class UrlDeduplicator {
|
|||||||
}
|
}
|
||||||
|
|
||||||
public boolean shouldRemove(DecoratedSearchResultItem details) {
|
public boolean shouldRemove(DecoratedSearchResultItem details) {
|
||||||
if (details.url.domain.topDomain.equals("slackware.com")) {
|
|
||||||
if (!deduplicateOnSuperficialHash(details)) {
|
if (!deduplicateOnSuperficialHash(details))
|
||||||
logger.info("Rejecting on superficial hash " + details.url);
|
return true;
|
||||||
return true;
|
if (!deduplicateOnLSH(details))
|
||||||
}
|
return true;
|
||||||
if (!deduplicateOnLSH(details)) {
|
if (!limitResultsPerDomain(details))
|
||||||
logger.info("Rejecting on LSH for " + details.url);
|
return true;
|
||||||
return true;
|
|
||||||
}
|
|
||||||
if (!limitResultsPerDomain(details)) {
|
|
||||||
logger.info("Rejecting on limitResultsPerDomain for " + details.url);
|
|
||||||
return true;
|
|
||||||
}
|
|
||||||
}
|
|
||||||
else {
|
|
||||||
if (!deduplicateOnSuperficialHash(details))
|
|
||||||
return true;
|
|
||||||
if (!deduplicateOnLSH(details))
|
|
||||||
return true;
|
|
||||||
if (!limitResultsPerDomain(details))
|
|
||||||
return true;
|
|
||||||
}
|
|
||||||
return false;
|
return false;
|
||||||
}
|
}
|
||||||
|
|
||||||
@@ -76,7 +62,7 @@ public class UrlDeduplicator {
|
|||||||
|
|
||||||
private boolean limitResultsPerDomain(DecoratedSearchResultItem details) {
|
private boolean limitResultsPerDomain(DecoratedSearchResultItem details) {
|
||||||
final var domain = details.getUrl().getDomain();
|
final var domain = details.getUrl().getDomain();
|
||||||
final String key = domain.getDomainKey();
|
final String key = domain.toString();
|
||||||
|
|
||||||
return keyCount.adjustOrPutValue(key, 1, 1) <= resultsPerKey;
|
return keyCount.adjustOrPutValue(key, 1, 1) <= resultsPerKey;
|
||||||
}
|
}
|
||||||
|
Reference in New Issue
Block a user