protected void checkUrlContent(
SimulatedArchivalUnit sau,
String path,
int fileNum,
int depth,
int branchNum,
boolean isAbnormal,
boolean isDamaged)
throws IOException {
String file = sau.getUrlRoot() + path;
CachedUrl url = sau.makeCachedUrl(file);
String content = getUrlContent(url);
String expectedContent;
if (path.endsWith(".html")) {
String fn = path.substring(path.lastIndexOf("/") + 1);
expectedContent = scgen.getHtmlFileContent(fn, fileNum, depth, branchNum, isAbnormal);
} else {
expectedContent = scgen.getTxtContent(fileNum, depth, branchNum, isAbnormal);
}
if (isDamaged) {
assertNotEquals(expectedContent, content);
} else {
assertEquals(expectedContent, content);
}
}
protected void checkLeaf(SimulatedArchivalUnit sau) {
log.debug("checkLeaf()");
String parent = sau.getUrlRoot() + "/branch1";
CachedUrlSetSpec spec = new RangeCachedUrlSetSpec(parent);
CachedUrlSet set = sau.makeCachedUrlSet(spec);
Iterator setIt = set.contentHashIterator();
ArrayList childL = new ArrayList(16);
while (setIt.hasNext()) {
childL.add(((CachedUrlSetNode) setIt.next()).getUrl());
}
String[] expectedA =
new String[] {
parent,
parent + "/001file.html",
parent + "/001file.txt",
parent + "/002file.html",
parent + "/002file.txt",
parent + "/branch1",
parent + "/branch1/001file.html",
parent + "/branch1/001file.txt",
parent + "/branch1/002file.html",
parent + "/branch1/002file.txt",
parent + "/branch1/index.html",
parent + "/branch2",
parent + "/branch2/001file.html",
parent + "/branch2/001file.txt",
parent + "/branch2/002file.html",
parent + "/branch2/002file.txt",
parent + "/branch2/index.html",
parent + "/index.html",
};
assertIsomorphic(expectedA, childL);
}
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);
}
protected void doDamageRemoveTest(SimulatedArchivalUnit sau) throws Exception {
/* Cache the file again; this time the damage should be gone */
String file = sau.getUrlRoot() + DAMAGED_CACHED_URL;
UrlCacher uc = sau.makeUrlCacher(file);
BitSet fetchFlags = new BitSet();
fetchFlags.set(UrlCacher.REFETCH_FLAG);
uc.setFetchFlags(fetchFlags);
uc.cache();
checkUrlContent(sau, DAMAGED_CACHED_URL, 2, 2, 2, false, false);
}
protected void checkDepth(SimulatedArchivalUnit sau) {
log.debug("checkDepth()");
String URL_ROOT = sau.getUrlRoot();
assertEquals(0, sau.getLinkDepth(URL_ROOT + "/index.html"));
assertEquals(0, sau.getLinkDepth(URL_ROOT + "/"));
assertEquals(1, sau.getLinkDepth(URL_ROOT + "/001file.html"));
assertEquals(1, sau.getLinkDepth(URL_ROOT + "/branch1/index.html"));
assertEquals(1, sau.getLinkDepth(URL_ROOT + "/branch1/"));
assertEquals(2, sau.getLinkDepth(URL_ROOT + "/branch1/001file.html"));
}
protected void checkFilter(SimulatedArchivalUnit sau) throws Exception {
log.debug("checkFilter()");
CachedUrl cu = sau.makeCachedUrl(sau.getUrlRoot() + "/001file.html");
enableFilter(sau, true);
InputStream is = cu.openForHashing();
String expected = "001file.html This is file 1, depth 0, branch 0. foobar ";
assertEquals(expected, StringUtil.fromInputStream(is));
is.close();
enableFilter(sau, false);
cu = sau.makeCachedUrl(sau.getUrlRoot() + "/001file.html");
is = cu.openForHashing();
expected =
"<HTML><HEAD><TITLE>001file.html</TITLE></HEAD><BODY>\n"
+ "This is file 1, depth 0, branch 0.<br><!-- comment --> "
+ "Citation String foobar<br><script>"
+ "(defun fact (n) (cond ((= n 0) 1) (t (fact (sub1 n)))))</script>\n"
+ "</BODY></HTML>";
assertEquals(expected, StringUtil.fromInputStream(is));
is.close();
}
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));
}