public void testWarcIteration() throws Exception {
warcJob =
new ExtractCDXFromWarcJob() {
public void processRecord(ArchiveRecord sar, OutputStream os) {
super.processRecord(sar, new ByteArrayOutputStream());
processed++;
}
};
OutputStream os = new ByteArrayOutputStream();
assertFalse("The to-be-generated file should not exist aforehand", TestInfo.CDX_FILE.exists());
os = new FileOutputStream(TestInfo.CDX_FILE);
warcBlaf.run(warcJob, os);
os.close();
}
public void testExtractCDXJobWithWarcfilesIncludeChecksum() throws Exception {
warcJob = new ExtractCDXFromWarcJob(true);
OutputStream os = new ByteArrayOutputStream();
assertFalse("The to-be-generated file should not exist aforehand", TestInfo.CDX_FILE.exists());
os = new FileOutputStream(TestInfo.CDX_FILE);
warcBlaf.run(warcJob, os);
os.close();
List<ExceptionOccurrence> exceptions = warcJob.getExceptions();
for (ExceptionOccurrence eo : exceptions) {
System.out.println("Exception: " + eo.getException());
}
// assertFalse(warcJob.getExceptions().isEmpty());
System.out.println(FileUtils.readFile(TestInfo.CDX_FILE));
}
/** Verify that the job runs without problems and visits all relevant records. */
public void testRun() throws IOException {
warcJob =
new ExtractCDXFromWarcJob() {
public void processRecord(ArchiveRecord sar, OutputStream os) {
super.processRecord(sar, new ByteArrayOutputStream());
processed++;
}
};
OutputStream os = new FileOutputStream(TestInfo.TEMP_FILE);
warcBlaf.run(warcJob, os);
os.close();
Exception[] es = warcJob.getExceptionArray();
printExceptions(es);
assertEquals("No exceptions should be thrown", 0, es.length);
assertEquals(
"The correct number of records should be processed", TestInfo.NUM_RECORDS, processed);
}
/** Test whether the class is really Serializable. */
public void testSerializability() throws IOException, ClassNotFoundException {
// Take two jobs: one for study and one for reference.
ExtractCDXFromWarcJob job1 = new StubbornJob();
ExtractCDXFromWarcJob job2 = new StubbornJob();
// Now serialize and deserialize the studied job (but NOT the reference):
ByteArrayOutputStream baos = new ByteArrayOutputStream();
ObjectOutputStream ous = new ObjectOutputStream(baos);
ous.writeObject(job1);
ous.close();
baos.close();
ObjectInputStream ois = new ObjectInputStream(new ByteArrayInputStream(baos.toByteArray()));
job1 = (ExtractCDXFromWarcJob) ois.readObject();
// Finally, compare their outputs:
ByteArrayOutputStream baos1 = new ByteArrayOutputStream();
ByteArrayOutputStream baos2 = new ByteArrayOutputStream();
// Run both jobs ordinarily:
warcBlaf.run(job1, baos1);
warcBlaf.run(job2, baos2);
baos1.close();
baos2.close();
byte[] b1 = baos1.toByteArray();
byte[] b2 = baos2.toByteArray();
for (int i = 0; i < b1.length; ++i) {
if (b1[i] != b2[i]) {
System.out.println("b1=" + b1[i]);
System.out.println("b2=" + b2[i]);
fail("\nDifference at position " + i);
} else {
// System.out.println(b1[i]);
}
}
assertTrue(
"Output from cdx jobs should be the same",
Arrays.equals(baos1.toByteArray(), baos2.toByteArray()));
}
