Examples with ExtractedText org.apache.jackrabbit.oak.plugins.index.fulltext.ExtractedText used on opensource projects

Example 1 with ExtractedText

use of org.apache.jackrabbit.oak.plugins.index.fulltext.ExtractedText in project jackrabbit-oak by apache.

the class ExtractedTextCacheTest method preExtractionReindex.

@Test
public void preExtractionReindex() throws Exception {
    ExtractedTextCache cache = new ExtractedTextCache(10 * FileUtils.ONE_MB, 100);
    PreExtractedTextProvider provider = mock(PreExtractedTextProvider.class);
    cache.setExtractedTextProvider(provider);
    when(provider.getText(anyString(), any(Blob.class))).thenReturn(new ExtractedText(ExtractionResult.SUCCESS, "bar"));
    Blob b = new IdBlob("hello", "a");
    String text = cache.get("/a", "foo", b, true);
    assertEquals("bar", text);
}

Example 2 with ExtractedText

use of org.apache.jackrabbit.oak.plugins.index.fulltext.ExtractedText in project jackrabbit-oak by apache.

the class ExtractedTextCacheTest method cacheEnabled.

@Test
public void cacheEnabled() throws Exception {
    ExtractedTextCache cache = new ExtractedTextCache(10 * FileUtils.ONE_MB, 100);
    assertNotNull(cache.getCacheStats());
    Blob b = new IdBlob("hello", "a");
    String text = cache.get("/a", "foo", b, false);
    assertNull(text);
    cache.put(b, new ExtractedText(ExtractionResult.SUCCESS, "test hello"));
    text = cache.get("/a", "foo", b, false);
    assertEquals("test hello", text);
}

Example 3 with ExtractedText

use of org.apache.jackrabbit.oak.plugins.index.fulltext.ExtractedText in project jackrabbit-oak by apache.

the class ExtractedTextCacheTest method cacheEnabledErrorInTextExtraction.

@Test
public void cacheEnabledErrorInTextExtraction() throws Exception {
    ExtractedTextCache cache = new ExtractedTextCache(10 * FileUtils.ONE_MB, 100);
    Blob b = new IdBlob("hello", "a");
    String text = cache.get("/a", "foo", b, false);
    assertNull(text);
    cache.put(b, new ExtractedText(ExtractionResult.ERROR, "test hello"));
    text = cache.get("/a", "foo", b, false);
    assertEquals(LuceneIndexEditor.TEXT_EXTRACTION_ERROR, text);
}

Example 4 with ExtractedText

use of org.apache.jackrabbit.oak.plugins.index.fulltext.ExtractedText in project jackrabbit-oak by apache.

the class ExtractedTextCacheTest method cacheEnabledNonIdBlob.

@Test
public void cacheEnabledNonIdBlob() throws Exception {
    ExtractedTextCache cache = new ExtractedTextCache(10 * FileUtils.ONE_MB, 100);
    Blob b = new ArrayBasedBlob("hello".getBytes());
    String text = cache.get("/a", "foo", b, false);
    assertNull(text);
    cache.put(b, new ExtractedText(ExtractionResult.SUCCESS, "test hello"));
    text = cache.get("/a", "foo", b, false);
    assertNull(text);
}

Example 5 with ExtractedText

use of org.apache.jackrabbit.oak.plugins.index.fulltext.ExtractedText in project jackrabbit-oak by apache.

the class DataStoreTextWriterTest method nonExistingEntry.

@Test
public void nonExistingEntry() throws Exception {
    File fdsDir = temporaryFolder.newFolder();
    FileDataStore fds = DataStoreUtils.createFDS(fdsDir, 0);
    ByteArrayInputStream is = new ByteArrayInputStream("hello".getBytes());
    DataRecord dr = fds.addRecord(is);
    File writerDir = temporaryFolder.newFolder();
    DataStoreTextWriter w = new DataStoreTextWriter(writerDir, false);
    String id = dr.getIdentifier().toString();
    assertFalse(w.isProcessed(id));
    assertNull(w.getText("/a", new IdBlob("foo", id)));
    w.write(id, "foo");
    assertTrue(w.isProcessed(id));
    ExtractedText et = w.getText("/a", new IdBlob("foo", id));
    assertEquals("foo", et.getExtractedText());
    assertEquals(ExtractionResult.SUCCESS, et.getExtractionResult());
    w.markEmpty("a");
    assertTrue(w.isProcessed("a"));
}