Examples with Specification org.apache.manifoldcf.core.interfaces.Specification used on opensource projects

Example 1 with Specification

use of org.apache.manifoldcf.core.interfaces.Specification in project manifoldcf by apache.

the class AlfrescoConnectorTest method whenTheClientIsCalledItShouldUseThePreviouslySentLastTransactionId.

@Test
public void whenTheClientIsCalledItShouldUseThePreviouslySentLastTransactionId() throws Exception {
    long firstTransactionId = 0;
    long lastTransactionId = 5;
    long firstAclChangesetId = 0;
    long lastAclChangesetId = 5;
    when(client.fetchNodes(anyInt(), anyInt(), Mockito.any(AlfrescoFilters.class))).thenReturn(new AlfrescoResponse(lastTransactionId, lastAclChangesetId));
    connector.addSeedDocuments(mock(SeedingActivity.class), new Specification(), "", 0, BaseRepositoryConnector.JOBMODE_ONCEONLY);
    verify(client, times(1)).fetchNodes(eq(firstTransactionId), eq(firstAclChangesetId), Mockito.any(AlfrescoFilters.class));
    verify(client, times(1)).fetchNodes(eq(lastTransactionId), eq(lastAclChangesetId), Mockito.any(AlfrescoFilters.class));
}

Example 2 with Specification

use of org.apache.manifoldcf.core.interfaces.Specification in project manifoldcf by apache.

the class AlfrescoConnectorTest method whenADocumentIsReturnedItShouldBeAddedToManifold.

@Test
public void whenADocumentIsReturnedItShouldBeAddedToManifold() throws Exception {
    TestDocument testDocument = new TestDocument();
    when(client.fetchNodes(anyInt(), anyInt(), Mockito.any(AlfrescoFilters.class))).thenReturn(new AlfrescoResponse(0, 0, "", "", Arrays.<Map<String, Object>>asList(testDocument)));
    SeedingActivity seedingActivity = mock(SeedingActivity.class);
    connector.addSeedDocuments(seedingActivity, new Specification(), "", 0, BaseRepositoryConnector.JOBMODE_ONCEONLY);
    verify(seedingActivity).addSeedDocument(eq(TestDocument.uuid));
}

Example 3 with Specification

use of org.apache.manifoldcf.core.interfaces.Specification in project manifoldcf by apache.

the class DocumentContentExclusionHSQLDBIT method addContentExclusionRule.

private void addContentExclusionRule(IJobDescription job) throws ManifoldCFException {
    Specification jobSpec = job.getSpecification();
    SpecificationNode sn;
    sn = new SpecificationNode(WebcrawlerConfig.NODE_EXCLUDESCONTENTINDEX);
    sn.setValue(".*expired.*\n");
    jobSpec.addChild(jobSpec.getChildCount(), sn);
    jobManager.save(job);
}

Example 4 with Specification

use of org.apache.manifoldcf.core.interfaces.Specification in project manifoldcf by apache.

the class DocumentContentExclusionHSQLDBIT method setupContentFilterJob.

private IJobDescription setupContentFilterJob() throws Exception {
    // Create a job.
    IJobDescription job = jobManager.createJob();
    job.setDescription("Test Job");
    job.setConnectionName(WEB_CONNECTION);
    job.addPipelineStage(-1, true, "Null Connection", "");
    job.setType(job.TYPE_SPECIFIED);
    job.setStartMethod(job.START_DISABLE);
    job.setHopcountMode(job.HOPCOUNT_NEVERDELETE);
    Specification jobSpec = job.getSpecification();
    // 3 seeds only
    SpecificationNode sn = new SpecificationNode(WebcrawlerConfig.NODE_SEEDS);
    StringBuilder sb = new StringBuilder();
    for (int i = 0; i < MAX_DOC_COUNT; i++) {
        sb.append(baseUrl + i + "\n");
    }
    sn.setValue(sb.toString());
    jobSpec.addChild(jobSpec.getChildCount(), sn);
    sn = new SpecificationNode(WebcrawlerConfig.NODE_INCLUDES);
    sn.setValue(".*\n");
    jobSpec.addChild(jobSpec.getChildCount(), sn);
    sn = new SpecificationNode(WebcrawlerConfig.NODE_INCLUDESINDEX);
    sn.setValue(".*\n");
    jobSpec.addChild(jobSpec.getChildCount(), sn);
    // Save the job.
    jobManager.save(job);
    return job;
}

Example 5 with Specification

use of org.apache.manifoldcf.core.interfaces.Specification in project manifoldcf by apache.

the class ConfluenceConnectorTest method mockSimpleIngestion.

@Test
public void mockSimpleIngestion() throws Exception {
    Page fakePage = mock(Page.class);
    Date date = new Date();
    DateFormat df = DateFormat.getDateTimeInstance(DateFormat.MEDIUM, DateFormat.MEDIUM, Locale.ROOT);
    String content = "A";
    String uri = "http://test";
    byte[] documentBytes = content.getBytes(StandardCharsets.UTF_8);
    long size = (long) documentBytes.length;
    when(fakePage.hasContent()).thenReturn(true);
    when(fakePage.getContent()).thenReturn(content);
    when(fakePage.getLength()).thenReturn(size);
    when(fakePage.getLastModifiedDate()).thenReturn(date);
    when(fakePage.getMediaType()).thenReturn("text/plain");
    when(fakePage.getCreatedDate()).thenReturn(date);
    when(fakePage.getWebUrl()).thenReturn(uri);
    Map<String, Object> metadata = new HashMap<String, Object>();
    metadata.put("x", "y");
    when(fakePage.getMetadataAsMap()).thenReturn(metadata);
    IProcessActivity activities = mock(IProcessActivity.class);
    when(activities.checkLengthIndexable(anyLong())).thenReturn(true);
    when(activities.checkMimeTypeIndexable(anyString())).thenReturn(true);
    when(activities.checkDateIndexable((Date) anyObject())).thenReturn(true);
    when(activities.checkURLIndexable(anyString())).thenReturn(true);
    when(activities.checkDocumentNeedsReindexing(anyString(), anyString())).thenReturn(true);
    IExistingVersions statuses = mock(IExistingVersions.class);
    String ID = df.format(date);
    when(statuses.getIndexedVersionString(ID)).thenReturn(null);
    when(client.getPage(Mockito.anyString())).thenReturn(fakePage);
    connector.processDocuments(new String[] { ID }, statuses, new Specification(), activities, 0, true);
    ArgumentCaptor<RepositoryDocument> rd = ArgumentCaptor.forClass(RepositoryDocument.class);
    verify(client, times(1)).getPage(ID);
    verify(activities, times(1)).ingestDocumentWithException(eq(ID), eq(df.format(date)), eq(uri), rd.capture());
    verify(activities, times(1)).recordActivity(anyLong(), eq("read document"), eq(size), eq(ID), eq("OK"), anyString(), Mockito.isNull(String[].class));
    RepositoryDocument doc = rd.getValue();
    Assert.assertEquals(size, doc.getBinaryLength());
    String[] values = doc.getFieldAsStrings("x");
    Assert.assertEquals(values.length, 1);
    Assert.assertEquals(values[0], "y");
}