public void executeTest() throws Exception {
// Hey, we were able to install the file system connector etc.
// Now, create a local test job and run it.
IThreadContext tc = ThreadContextFactory.make();
// Create a basic file system connection, and save it.
IRepositoryConnectionManager mgr = RepositoryConnectionManagerFactory.make(tc);
IRepositoryConnection conn = mgr.create();
conn.setName("File Connection");
conn.setDescription("File Connection");
conn.setClassName("org.apache.manifoldcf.crawler.connectors.filesystem.FileConnector");
conn.setMaxConnections(100);
// Now, save
mgr.save(conn);
// Create a basic null output connection, and save it.
IOutputConnectionManager outputMgr = OutputConnectionManagerFactory.make(tc);
IOutputConnection outputConn = outputMgr.create();
outputConn.setName("Null Connection");
outputConn.setDescription("Null Connection");
outputConn.setClassName("org.apache.manifoldcf.agents.output.nullconnector.NullConnector");
outputConn.setMaxConnections(100);
// Now, save
outputMgr.save(outputConn);
// Create a job.
IJobManager jobManager = JobManagerFactory.make(tc);
IJobDescription job = jobManager.createJob();
job.setDescription("Test Job");
job.setConnectionName("File Connection");
job.setOutputConnectionName("Null Connection");
job.setType(job.TYPE_SPECIFIED);
job.setStartMethod(job.START_DISABLE);
job.setHopcountMode(job.HOPCOUNT_ACCURATE);
// Now, set up the document specification.
DocumentSpecification ds = job.getSpecification();
// Crawl everything underneath the 'testdata' area
File testDataFile = new File("testdata").getCanonicalFile();
if (!testDataFile.exists())
throw new ManifoldCFException(
"Test data area not found! Looking in " + testDataFile.toString());
if (!testDataFile.isDirectory())
throw new ManifoldCFException(
"Test data area not a directory! Looking in " + testDataFile.toString());
SpecificationNode sn = new SpecificationNode("startpoint");
sn.setAttribute("path", testDataFile.toString());
SpecificationNode n = new SpecificationNode("include");
n.setAttribute("type", "file");
n.setAttribute("match", "*");
sn.addChild(sn.getChildCount(), n);
n = new SpecificationNode("include");
n.setAttribute("type", "directory");
n.setAttribute("match", "*");
sn.addChild(sn.getChildCount(), n);
ds.addChild(ds.getChildCount(), sn);
// Set up the output specification.
OutputSpecification os = job.getOutputSpecification();
// Null output connections have no output specification, so this is a no-op.
// Save the job.
jobManager.save(job);
// Create the test data files.
FileHelper.createFile(new File("testdata/test1.txt"), "This is a test file");
FileHelper.createFile(new File("testdata/test2.txt"), "This is another test file");
FileHelper.createDirectory(new File("testdata/testdir"));
FileHelper.createFile(new File("testdata/testdir/test3.txt"), "This is yet another test file");
// Now, start the job, and wait until it completes.
jobManager.manualStart(job.getID());
instance.waitJobInactiveNative(jobManager, job.getID(), 120000L);
// Check to be sure we actually processed the right number of documents.
JobStatus status = jobManager.getStatus(job.getID());
// The test data area has 3 documents and one directory, and we have to count the root directory
// too.
if (status.getDocumentsProcessed() != 5)
throw new ManifoldCFException(
"Wrong number of documents processed - expected 5, saw "
+ new Long(status.getDocumentsProcessed()).toString());
// Add a file and recrawl using minimal crawl
FileHelper.createFile(new File("testdata/testdir/test4.txt"), "Added file");
// Now, start the job, and wait until it completes.
jobManager.manualStart(job.getID(), true);
instance.waitJobInactiveNative(jobManager, job.getID(), 120000L);
status = jobManager.getStatus(job.getID());
// The test data area has 4 documents and one directory, and we have to count the root directory
// too.
if (status.getDocumentsProcessed() != 6)
throw new ManifoldCFException(
"Wrong number of documents processed after add - expected 6, saw "
+ new Long(status.getDocumentsProcessed()).toString());
// Change a file, and recrawl, once again using minimal
FileHelper.changeFile(new File("testdata/test1.txt"), "Modified contents");
// Now, start the job, and wait until it completes.
jobManager.manualStart(job.getID(), true);
instance.waitJobInactiveNative(jobManager, job.getID(), 120000L);
status = jobManager.getStatus(job.getID());
// The test data area has 4 documents and one directory, and we have to count the root directory
// too.
if (status.getDocumentsProcessed() != 6)
throw new ManifoldCFException(
"Wrong number of documents processed after change - expected 6, saw "
+ new Long(status.getDocumentsProcessed()).toString());
// We also need to make sure the new document was indexed. Have to think about how to do this
// though.
// MHL
// Delete a file, and recrawl
FileHelper.removeFile(new File("testdata/test2.txt"));
// Do a minimal recrawl first; the delete should not be picked up.
jobManager.manualStart(job.getID(), true);
instance.waitJobInactiveNative(jobManager, job.getID(), 120000L);
status = jobManager.getStatus(job.getID());
// The test data area has 4 documents and one directory, and we have to count the root directory
// too.
if (status.getDocumentsProcessed() != 6)
throw new ManifoldCFException(
"Wrong number of documents processed after delete with minimal crawl - expected 6, saw "
+ new Long(status.getDocumentsProcessed()).toString());
// Now, do a complete crawl - the delete should be found now.
jobManager.manualStart(job.getID());
instance.waitJobInactiveNative(jobManager, job.getID(), 120000L);
// Check to be sure we actually processed the right number of documents.
status = jobManager.getStatus(job.getID());
// The test data area has 3 documents and one directory, and we have to count the root directory
// too.
if (status.getDocumentsProcessed() != 5)
throw new ManifoldCFException(
"Wrong number of documents processed after delete - expected 5, saw "
+ new Long(status.getDocumentsProcessed()).toString());
// Now, delete the job.
jobManager.deleteJob(job.getID());
instance.waitJobDeletedNative(jobManager, job.getID(), 120000L);
// Cleanup is automatic by the base class, so we can feel free to leave jobs and connections
// lying around.
}
public static void main(String[] args) {
if (args.length != 13) {
System.err.println(
"Usage: DefineJob <description> <connection_name> <output_name> <type> <start_method> <hopcount_method> <recrawl_interval> <expiration_interval> <reseed_interval> <job_priority> <hop_filters> <filespec_xml> <outputspec_xml>");
System.err.println("<type> is one of: continuous or specified");
System.err.println("<start_method> is one of: windowbegin, windowinside, disable");
System.err.println("<hopcount_method> is one of: accurate, nodelete, neverdelete");
System.err.println("<recrawl_interval> is the default document recrawl interval in minutes");
System.err.println(
"<expiration_interval> is the default document expiration interval in minutes");
System.err.println("<reseed_interval> is the default document reseed interval in minutes");
System.err.println("<job_priority> is the job priority (and integer between 0 and 10)");
System.err.println(
"<hop_filters> is a comma-separated list of tuples, of the form 'linktype=maxhops'");
System.err.println(
"<filespec_xml> is the document specification XML, its form dependent on the connection type");
System.err.println(
"<outputspec_xml> is the output specification XML, its form dependent on the output connection type");
System.exit(-1);
}
String description = args[0];
String connectionName = args[1];
String outputConnectionName = args[2];
String typeString = args[3];
String startString = args[4];
String hopcountString = args[5];
String recrawlInterval = args[6];
String expirationInterval = args[7];
String reseedInterval = args[8];
String jobPriority = args[9];
String hopFilters = args[10];
String filespecXML = args[11];
String outputspecXML = args[12];
try {
IThreadContext tc = ThreadContextFactory.make();
ManifoldCF.initializeEnvironment(tc);
IJobManager jobManager = JobManagerFactory.make(tc);
IJobDescription desc = jobManager.createJob();
desc.setDescription(description);
desc.setConnectionName(connectionName);
desc.setOutputConnectionName(outputConnectionName);
if (typeString.equals("continuous")) desc.setType(IJobDescription.TYPE_CONTINUOUS);
else if (typeString.equals("specified")) desc.setType(IJobDescription.TYPE_SPECIFIED);
else throw new ManifoldCFException("Unknown type: '" + typeString + "'");
if (startString.equals("windowbegin")) desc.setStartMethod(IJobDescription.START_WINDOWBEGIN);
else if (startString.equals("windowinside"))
desc.setStartMethod(IJobDescription.START_WINDOWINSIDE);
else if (startString.equals("disable")) desc.setStartMethod(IJobDescription.START_DISABLE);
else throw new ManifoldCFException("Unknown start method: '" + startString + "'");
if (hopcountString.equals("accurate"))
desc.setHopcountMode(IJobDescription.HOPCOUNT_ACCURATE);
else if (hopcountString.equals("nodelete"))
desc.setHopcountMode(IJobDescription.HOPCOUNT_NODELETE);
else if (hopcountString.equals("neverdelete"))
desc.setHopcountMode(IJobDescription.HOPCOUNT_NEVERDELETE);
else throw new ManifoldCFException("Unknown hopcount mode: '" + hopcountString + "'");
if (recrawlInterval.length() > 0) desc.setInterval(new Long(recrawlInterval));
if (expirationInterval.length() > 0) desc.setExpiration(new Long(expirationInterval));
if (reseedInterval.length() > 0) desc.setReseedInterval(new Long(reseedInterval));
desc.setPriority(Integer.parseInt(jobPriority));
String[] hopFilterSet = hopFilters.split(",");
int i = 0;
while (i < hopFilterSet.length) {
String hopFilter = hopFilterSet[i++];
if (hopFilter != null && hopFilter.length() > 0) {
String[] stuff = hopFilter.trim().split("=");
if (stuff != null && stuff.length == 2)
desc.addHopCountFilter(stuff[0], ((stuff[1].length() > 0) ? new Long(stuff[1]) : null));
}
}
desc.getSpecification().fromXML(filespecXML);
if (outputspecXML.length() > 0) desc.getOutputSpecification().fromXML(outputspecXML);
// Now, save
jobManager.save(desc);
System.out.print(desc.getID().toString());
} catch (Exception e) {
e.printStackTrace();
System.exit(-2);
}
}