Merge branch 'master' into hc43

Conflicts:
	commons/pom.xml
	commons/src/main/java/org/archive/io/RecordingInputStream.java
	commons/src/main/java/org/archive/io/RecordingOutputStream.java
	commons/src/main/java/org/archive/io/ReplayInputStream.java
	dist/pom.xml
	engine/pom.xml
	modules/pom.xml
	modules/src/main/java/org/archive/modules/extractor/ExtractorHTTP.java
	modules/src/main/java/org/archive/modules/recrawl/FetchHistoryProcessor.java
	pom.xml
This commit is contained in:
Noah Levitt
2014-01-10 17:52:03 -08:00
171 changed files with 21168 additions and 22596 deletions
@@ -185,7 +185,7 @@ implements UriUniqFilter.CrawlUriReceiver {
assertEquals("Didn't forget", 0, this.filter.count());
}
public void testForgetSchemeHost() throws URIException {
public void testForgetAllSchemeAuthorityMatching() throws URIException {
long countBefore = this.filter.count();
for (String uri: new String[] {
@@ -205,7 +205,7 @@ implements UriUniqFilter.CrawlUriReceiver {
BdbUriUniqFilter bdbFilter = (BdbUriUniqFilter) filter;
assertFalse(bdbFilter.setAdd("http://forgetme.com/foo"));
bdbFilter.forgetSchemeHost("http://forgetme.com");
bdbFilter.forgetAllSchemeAuthorityMatching("http://forgetme.com");
assertEquals(countBefore + 2, this.filter.count());
assertTrue(bdbFilter.setAdd("http://forgetme.com/foo"));
@@ -213,10 +213,10 @@ implements UriUniqFilter.CrawlUriReceiver {
assertTrue(bdbFilter.setRemove("http://forgetme.com/foo"));
assertFalse(bdbFilter.setRemove("http://forgetme.com/foo"));
bdbFilter.forgetSchemeHost("https://forgetme.com/extra-stuff-ignored");
bdbFilter.forgetAllSchemeAuthorityMatching("https://forgetme.com/extra-stuff-ignored");
assertEquals(countBefore + 1, this.filter.count());
bdbFilter.forgetSchemeHost("http://forgetme.com:90/");
bdbFilter.forgetAllSchemeAuthorityMatching("http://forgetme.com:90/");
assertEquals(countBefore, this.filter.count());
}