Examples with POITextExtractor org.apache.poi.POITextExtractor used on opensource projects

Example 16 with POITextExtractor

use of org.apache.poi.POITextExtractor in project poi by apache.

the class TestZipPackage method testZipEntityExpansionSharedStringTable.

@Test
public void testZipEntityExpansionSharedStringTable() throws Exception {
    Workbook wb = WorkbookFactory.create(XSSFTestDataSamples.openSamplePackage("poc-shared-strings.xlsx"));
    wb.close();
    POITextExtractor extractor = ExtractorFactory.createExtractor(HSSFTestDataSamples.getSampleFile("poc-shared-strings.xlsx"));
    try {
        assertNotNull(extractor);
        try {
            extractor.getText();
        } catch (IllegalStateException e) {
        // expected due to shared strings expansion
        }
    } finally {
        extractor.close();
    }
}

Example 17 with POITextExtractor

use of org.apache.poi.POITextExtractor in project poi by apache.

the class TestXSSFEventBasedExcelExtractor method testComparedToOLE2.

/**
	 * Test that we return pretty much the same as
	 *  ExcelExtractor does, when we're both passed
	 *  the same file, just saved as xls and xlsx
	 */
@Test
public void testComparedToOLE2() throws Exception {
    // A fairly simple file - ooxml
    XSSFEventBasedExcelExtractor ooxmlExtractor = getExtractor("SampleSS.xlsx");
    ExcelExtractor ole2Extractor = new ExcelExtractor(HSSFTestDataSamples.openSampleWorkbook("SampleSS.xls"));
    POITextExtractor[] extractors = new POITextExtractor[] { ooxmlExtractor, ole2Extractor };
    for (POITextExtractor extractor : extractors) {
        String text = extractor.getText().replaceAll("[\r\t]", "");
        assertStartsWith(text, "First Sheet\nTest spreadsheet\n2nd row2nd row 2nd column\n");
        Pattern pattern = Pattern.compile(".*13(\\.0+)?\\s+Sheet3.*", Pattern.DOTALL);
        Matcher m = pattern.matcher(text);
        assertTrue(m.matches());
    }
    ole2Extractor.close();
    ooxmlExtractor.close();
}

Example 18 with POITextExtractor

use of org.apache.poi.POITextExtractor in project poi by apache.

the class TestXSSFExcelExtractor method testComparedToOLE2.

/**
	 * Test that we return pretty much the same as
	 *  ExcelExtractor does, when we're both passed
	 *  the same file, just saved as xls and xlsx
	 */
public void testComparedToOLE2() throws IOException {
    // A fairly simple file - ooxml
    XSSFExcelExtractor ooxmlExtractor = getExtractor("SampleSS.xlsx");
    ExcelExtractor ole2Extractor = new ExcelExtractor(HSSFTestDataSamples.openSampleWorkbook("SampleSS.xls"));
    Map<String, POITextExtractor> extractors = new HashMap<String, POITextExtractor>();
    extractors.put("SampleSS.xlsx", ooxmlExtractor);
    extractors.put("SampleSS.xls", ole2Extractor);
    for (final Entry<String, POITextExtractor> e : extractors.entrySet()) {
        String filename = e.getKey();
        POITextExtractor extractor = e.getValue();
        String text = extractor.getText().replaceAll("[\r\t]", "");
        assertStartsWith(filename, text, "First Sheet\nTest spreadsheet\n2nd row2nd row 2nd column\n");
        Pattern pattern = Pattern.compile(".*13(\\.0+)?\\s+Sheet3.*", Pattern.DOTALL);
        Matcher m = pattern.matcher(text);
        assertTrue(filename, m.matches());
    }
    ole2Extractor.close();
    ooxmlExtractor.close();
}