public void testBaseUrlPath() throws Exception {
sau1 = setupSimAu(simAuConfig(tempDirPath));
createContent(sau1);
crawlContent(sau1);
CachedUrlSet cus1 = sau1.getAuCachedUrlSet();
tempDirPath2 = getTempDir().getAbsolutePath() + File.separator;
Configuration config2 = simAuConfig(tempDirPath2);
config2.put("base_url", "http://anotherhost.org/some/path/");
SimulatedArchivalUnit sau2 = setupSimAu(config2);
createContent(sau2);
crawlContent(sau2);
CachedUrlSet cus2 = sau1.getAuCachedUrlSet();
List urls1 = auUrls(sau1);
List urls2 = auUrls(sau2);
Pattern pat1 = Pattern.compile("http://www\\.example\\.com(/.*)$");
Pattern pat2 = Pattern.compile("http://anotherhost\\.org/some/path(/.*)$");
List<String> l1 = auUrls(sau1);
List<String> l2 = auUrls(sau2);
assertEquals(l1.size(), l2.size());
for (int ix = 0; ix < l1.size(); ix++) {
Matcher m1 = pat1.matcher(l1.get(ix));
assertTrue(m1.matches());
Matcher m2 = pat2.matcher(l2.get(ix));
assertTrue(m2.matches());
assertEquals(m1.group(1), m2.group(1));
}
}
protected void checkRoot(SimulatedArchivalUnit sau) {
log.debug("checkRoot()");
CachedUrlSet set = sau.getAuCachedUrlSet();
Iterator setIt = set.flatSetIterator();
ArrayList childL = new ArrayList(1);
CachedUrlSet cus = null;
while (setIt.hasNext()) {
cus = (CachedUrlSet) setIt.next();
childL.add(cus.getUrl());
}
String urlRoot = sau.getUrlRoot();
String[] expectedA = new String[1];
expectedA[0] = urlRoot;
assertIsomorphic(expectedA, childL);
setIt = cus.flatSetIterator();
childL = new ArrayList(7);
while (setIt.hasNext()) {
childL.add(((CachedUrlSetNode) setIt.next()).getUrl());
}
expectedA =
new String[] {
urlRoot + "/001file.html",
urlRoot + "/001file.txt",
urlRoot + "/002file.html",
urlRoot + "/002file.txt",
urlRoot + "/branch1",
urlRoot + "/branch2",
urlRoot + "/index.html"
};
assertIsomorphic(expectedA, childL);
}
public void testDualContentHash() throws Exception {
sau1 = setupSimAu(simAuConfig(tempDirPath));
createContent(sau1);
crawlContent(sau1);
CachedUrlSet set = sau1.getAuCachedUrlSet();
byte[] nameH = getHash(set, true);
byte[] contentH = getHash(set, false);
tempDirPath2 = getTempDir().getAbsolutePath() + File.separator;
SimulatedArchivalUnit sau2 = setupSimAu(simAuConfig(tempDirPath2));
createContent(sau2);
crawlContent(sau2);
set = sau2.getAuCachedUrlSet();
byte[] nameH2 = getHash(set, true);
byte[] contentH2 = getHash(set, false);
assertEquals(nameH, nameH2);
assertEquals(contentH, contentH2);
}
private void checkHashSet(
SimulatedArchivalUnit sau, boolean namesOnly, boolean filter, byte[] expected)
throws Exception {
enableFilter(sau, filter);
CachedUrlSet set = sau.getAuCachedUrlSet();
byte[] hash = getHash(set, namesOnly);
assertEquals(expected, hash);
String parent = sau.getUrlRoot() + "/branch1";
CachedUrlSetSpec spec = new RangeCachedUrlSetSpec(parent);
set = sau.makeCachedUrlSet(spec);
byte[] hash2 = getHash(set, namesOnly);
assertFalse(Arrays.equals(hash, hash2));
}
private void measureHashSpeed(SimulatedArchivalUnit sau) throws Exception {
MessageDigest dig = null;
try {
dig = MessageDigest.getInstance("SHA-1");
} catch (NoSuchAlgorithmException ex) {
fail("No algorithm.");
}
CachedUrlSet set = sau.getAuCachedUrlSet();
CachedUrlSetHasher hasher = set.getContentHasher(dig);
SystemMetrics metrics = theDaemon.getSystemMetrics();
int estimate = metrics.getBytesPerMsHashEstimate(hasher, dig);
// should be protected against this being zero by MyMockSystemMetrics,
// but otherwise use the proper calculation. This avoids test failure
// due to really slow machines
assertTrue(estimate > 0);
long estimatedTime = set.estimatedHashDuration();
long size = ((Long) PrivilegedAccessor.getValue(set, "totalNodeSize")).longValue();
assertTrue(size > 0);
System.out.println("b/ms: " + estimate);
System.out.println("size: " + size);
System.out.println("estimate: " + estimatedTime);
assertEquals(estimatedTime, theDaemon.getHashService().padHashEstimate(size / estimate));
}