use of org.apache.poi.hdgf.extractor.VisioTextExtractor in project poi by apache.
the class TestExtractorFactory method testOPOIFS.
@Test
public void testOPOIFS() throws Exception {
// Excel
assertTrue(ExtractorFactory.createExtractor(new OPOIFSFileSystem(new FileInputStream(xls))) instanceof ExcelExtractor);
assertTrue(ExtractorFactory.createExtractor(new OPOIFSFileSystem(new FileInputStream(xls))).getText().length() > 200);
// Word
assertTrue(ExtractorFactory.createExtractor(new OPOIFSFileSystem(new FileInputStream(doc))) instanceof WordExtractor);
assertTrue(ExtractorFactory.createExtractor(new OPOIFSFileSystem(new FileInputStream(doc))).getText().length() > 120);
assertTrue(ExtractorFactory.createExtractor(new OPOIFSFileSystem(new FileInputStream(doc6))) instanceof Word6Extractor);
assertTrue(ExtractorFactory.createExtractor(new OPOIFSFileSystem(new FileInputStream(doc6))).getText().length() > 20);
assertTrue(ExtractorFactory.createExtractor(new OPOIFSFileSystem(new FileInputStream(doc95))) instanceof Word6Extractor);
assertTrue(ExtractorFactory.createExtractor(new OPOIFSFileSystem(new FileInputStream(doc95))).getText().length() > 120);
// PowerPoint
assertTrue(ExtractorFactory.createExtractor(new OPOIFSFileSystem(new FileInputStream(ppt))) instanceof PowerPointExtractor);
assertTrue(ExtractorFactory.createExtractor(new OPOIFSFileSystem(new FileInputStream(ppt))).getText().length() > 120);
// Visio
assertTrue(ExtractorFactory.createExtractor(new OPOIFSFileSystem(new FileInputStream(vsd))) instanceof VisioTextExtractor);
assertTrue(ExtractorFactory.createExtractor(new OPOIFSFileSystem(new FileInputStream(vsd))).getText().length() > 50);
// Publisher
assertTrue(ExtractorFactory.createExtractor(new OPOIFSFileSystem(new FileInputStream(pub))) instanceof PublisherTextExtractor);
assertTrue(ExtractorFactory.createExtractor(new OPOIFSFileSystem(new FileInputStream(pub))).getText().length() > 50);
// Outlook msg
assertTrue(ExtractorFactory.createExtractor(new OPOIFSFileSystem(new FileInputStream(msg))) instanceof OutlookTextExtactor);
assertTrue(ExtractorFactory.createExtractor(new OPOIFSFileSystem(new FileInputStream(msg))).getText().length() > 50);
// Text
try {
ExtractorFactory.createExtractor(new OPOIFSFileSystem(new FileInputStream(txt)));
fail();
} catch (IOException e) {
// Good
}
}
use of org.apache.poi.hdgf.extractor.VisioTextExtractor in project poi by apache.
the class TestExtractorFactory method testInputStream.
@Test
public void testInputStream() throws Exception {
// Excel
POITextExtractor extractor = ExtractorFactory.createExtractor(new FileInputStream(xls));
assertTrue(extractor instanceof ExcelExtractor);
assertTrue(extractor.getText().length() > 200);
extractor.close();
extractor = ExtractorFactory.createExtractor(new FileInputStream(xlsx));
assertTrue(extractor.getClass().getName(), extractor instanceof XSSFExcelExtractor);
assertTrue(extractor.getText().length() > 200);
// TODO Support OOXML-Strict, see bug #57699
// assertTrue(
// ExtractorFactory.createExtractor(new FileInputStream(xlsxStrict))
// instanceof XSSFExcelExtractor
// );
// assertTrue(
// ExtractorFactory.createExtractor(new FileInputStream(xlsxStrict)).getText().length() > 200
// );
extractor.close();
// Word
extractor = ExtractorFactory.createExtractor(new FileInputStream(doc));
assertTrue(extractor.getClass().getName(), extractor instanceof WordExtractor);
assertTrue(extractor.getText().length() > 120);
extractor.close();
extractor = ExtractorFactory.createExtractor(new FileInputStream(doc6));
assertTrue(extractor.getClass().getName(), extractor instanceof Word6Extractor);
assertTrue(extractor.getText().length() > 20);
extractor.close();
extractor = ExtractorFactory.createExtractor(new FileInputStream(doc95));
assertTrue(extractor.getClass().getName(), extractor instanceof Word6Extractor);
assertTrue(extractor.getText().length() > 120);
extractor.close();
extractor = ExtractorFactory.createExtractor(new FileInputStream(docx));
assertTrue(extractor instanceof XWPFWordExtractor);
assertTrue(extractor.getText().length() > 120);
extractor.close();
// PowerPoint
extractor = ExtractorFactory.createExtractor(new FileInputStream(ppt));
assertTrue(extractor instanceof PowerPointExtractor);
assertTrue(extractor.getText().length() > 120);
extractor.close();
extractor = ExtractorFactory.createExtractor(new FileInputStream(pptx));
assertTrue(extractor instanceof XSLFPowerPointExtractor);
assertTrue(extractor.getText().length() > 120);
extractor.close();
// Visio
extractor = ExtractorFactory.createExtractor(new FileInputStream(vsd));
assertTrue(extractor instanceof VisioTextExtractor);
assertTrue(extractor.getText().length() > 50);
extractor.close();
// Visio - vsdx
extractor = ExtractorFactory.createExtractor(new FileInputStream(vsdx));
assertTrue(extractor instanceof XDGFVisioExtractor);
assertTrue(extractor.getText().length() > 20);
extractor.close();
// Publisher
extractor = ExtractorFactory.createExtractor(new FileInputStream(pub));
assertTrue(extractor instanceof PublisherTextExtractor);
assertTrue(extractor.getText().length() > 50);
extractor.close();
// Outlook msg
extractor = ExtractorFactory.createExtractor(new FileInputStream(msg));
assertTrue(extractor instanceof OutlookTextExtactor);
assertTrue(extractor.getText().length() > 50);
extractor.close();
// Text
try {
FileInputStream stream = new FileInputStream(txt);
try {
ExtractorFactory.createExtractor(stream);
fail();
} finally {
IOUtils.closeQuietly(stream);
}
} catch (IllegalArgumentException e) {
// Good
}
}
use of org.apache.poi.hdgf.extractor.VisioTextExtractor in project poi by apache.
the class HDGFFileHandler method test.
// a test-case to test this locally without executing the full TestAllFiles
@Override
@Test
public void test() throws Exception {
File file = new File("test-data/diagram/44501.vsd");
InputStream stream = new FileInputStream(file);
try {
handleFile(stream, file.getPath());
} finally {
stream.close();
}
handleExtracting(file);
stream = new FileInputStream(file);
try {
VisioTextExtractor extractor = new VisioTextExtractor(stream);
try {
assertNotNull(extractor.getText());
} finally {
extractor.close();
}
} finally {
stream.close();
}
}
use of org.apache.poi.hdgf.extractor.VisioTextExtractor in project poi by apache.
the class TestExtractorFactory method testPOIFS.
@Test
public void testPOIFS() throws Exception {
// Excel
assertTrue(ExtractorFactory.createExtractor(new POIFSFileSystem(new FileInputStream(xls))) instanceof ExcelExtractor);
assertTrue(ExtractorFactory.createExtractor(new POIFSFileSystem(new FileInputStream(xls))).getText().length() > 200);
// Word
assertTrue(ExtractorFactory.createExtractor(new POIFSFileSystem(new FileInputStream(doc))) instanceof WordExtractor);
assertTrue(ExtractorFactory.createExtractor(new POIFSFileSystem(new FileInputStream(doc))).getText().length() > 120);
assertTrue(ExtractorFactory.createExtractor(new POIFSFileSystem(new FileInputStream(doc6))) instanceof Word6Extractor);
assertTrue(ExtractorFactory.createExtractor(new POIFSFileSystem(new FileInputStream(doc6))).getText().length() > 20);
assertTrue(ExtractorFactory.createExtractor(new POIFSFileSystem(new FileInputStream(doc95))) instanceof Word6Extractor);
assertTrue(ExtractorFactory.createExtractor(new POIFSFileSystem(new FileInputStream(doc95))).getText().length() > 120);
// PowerPoint
assertTrue(ExtractorFactory.createExtractor(new POIFSFileSystem(new FileInputStream(ppt))) instanceof PowerPointExtractor);
assertTrue(ExtractorFactory.createExtractor(new POIFSFileSystem(new FileInputStream(ppt))).getText().length() > 120);
// Visio
assertTrue(ExtractorFactory.createExtractor(new POIFSFileSystem(new FileInputStream(vsd))) instanceof VisioTextExtractor);
assertTrue(ExtractorFactory.createExtractor(new POIFSFileSystem(new FileInputStream(vsd))).getText().length() > 50);
// Publisher
assertTrue(ExtractorFactory.createExtractor(new POIFSFileSystem(new FileInputStream(pub))) instanceof PublisherTextExtractor);
assertTrue(ExtractorFactory.createExtractor(new POIFSFileSystem(new FileInputStream(pub))).getText().length() > 50);
// Outlook msg
assertTrue(ExtractorFactory.createExtractor(new POIFSFileSystem(new FileInputStream(msg))) instanceof OutlookTextExtactor);
assertTrue(ExtractorFactory.createExtractor(new POIFSFileSystem(new FileInputStream(msg))).getText().length() > 50);
// Text
try {
ExtractorFactory.createExtractor(new POIFSFileSystem(new FileInputStream(txt)));
fail();
} catch (IOException e) {
// Good
}
}
use of org.apache.poi.hdgf.extractor.VisioTextExtractor in project poi by apache.
the class TestExtractorFactory method testFile.
@Test
public void testFile() throws Exception {
// Excel
POITextExtractor xlsExtractor = ExtractorFactory.createExtractor(xls);
assertNotNull("Had empty extractor for " + xls, xlsExtractor);
assertTrue("Expected instanceof ExcelExtractor, but had: " + xlsExtractor.getClass(), xlsExtractor instanceof ExcelExtractor);
assertTrue(xlsExtractor.getText().length() > 200);
xlsExtractor.close();
POITextExtractor extractor = ExtractorFactory.createExtractor(xlsx);
assertTrue(extractor.getClass().getName(), extractor instanceof XSSFExcelExtractor);
extractor.close();
extractor = ExtractorFactory.createExtractor(xlsx);
assertTrue(extractor.getText().length() > 200);
extractor.close();
extractor = ExtractorFactory.createExtractor(xltx);
assertTrue(extractor.getClass().getName(), extractor instanceof XSSFExcelExtractor);
extractor.close();
extractor = ExtractorFactory.createExtractor(xlsb);
assertContains(extractor.getText(), "test");
extractor.close();
extractor = ExtractorFactory.createExtractor(xltx);
assertContains(extractor.getText(), "test");
extractor.close();
// TODO Support OOXML-Strict, see bug #57699
try {
/*extractor =*/
ExtractorFactory.createExtractor(xlsxStrict);
fail("OOXML-Strict isn't yet supported");
} catch (POIXMLException e) {
// Expected, for now
}
// extractor = ExtractorFactory.createExtractor(xlsxStrict);
// assertTrue(
// extractor
// instanceof XSSFExcelExtractor
// );
// extractor.close();
//
// extractor = ExtractorFactory.createExtractor(xlsxStrict);
// assertTrue(
// extractor.getText().contains("test")
// );
// extractor.close();
// Word
extractor = ExtractorFactory.createExtractor(doc);
assertTrue(extractor instanceof WordExtractor);
assertTrue(extractor.getText().length() > 120);
extractor.close();
extractor = ExtractorFactory.createExtractor(doc6);
assertTrue(extractor instanceof Word6Extractor);
assertTrue(extractor.getText().length() > 20);
extractor.close();
extractor = ExtractorFactory.createExtractor(doc95);
assertTrue(extractor instanceof Word6Extractor);
assertTrue(extractor.getText().length() > 120);
extractor.close();
extractor = ExtractorFactory.createExtractor(docx);
assertTrue(extractor instanceof XWPFWordExtractor);
extractor.close();
extractor = ExtractorFactory.createExtractor(docx);
assertTrue(extractor.getText().length() > 120);
extractor.close();
extractor = ExtractorFactory.createExtractor(dotx);
assertTrue(extractor instanceof XWPFWordExtractor);
extractor.close();
extractor = ExtractorFactory.createExtractor(dotx);
assertContains(extractor.getText(), "Test");
extractor.close();
// PowerPoint (PPT)
extractor = ExtractorFactory.createExtractor(ppt);
assertTrue(extractor instanceof PowerPointExtractor);
assertTrue(extractor.getText().length() > 120);
extractor.close();
// PowerPoint (PPTX)
extractor = ExtractorFactory.createExtractor(pptx);
assertTrue(extractor instanceof XSLFPowerPointExtractor);
assertTrue(extractor.getText().length() > 120);
extractor.close();
// Visio - binary
extractor = ExtractorFactory.createExtractor(vsd);
assertTrue(extractor instanceof VisioTextExtractor);
assertTrue(extractor.getText().length() > 50);
extractor.close();
// Visio - vsdx
extractor = ExtractorFactory.createExtractor(vsdx);
assertTrue(extractor instanceof XDGFVisioExtractor);
assertTrue(extractor.getText().length() > 20);
extractor.close();
// Publisher
extractor = ExtractorFactory.createExtractor(pub);
assertTrue(extractor instanceof PublisherTextExtractor);
assertTrue(extractor.getText().length() > 50);
extractor.close();
// Outlook msg
extractor = ExtractorFactory.createExtractor(msg);
assertTrue(extractor instanceof OutlookTextExtactor);
assertTrue(extractor.getText().length() > 50);
extractor.close();
// Text
try {
ExtractorFactory.createExtractor(txt);
fail();
} catch (IllegalArgumentException e) {
// Good
}
}
Aggregations