org.apache.poi.POITextExtractor Java Examples

public static POITextExtractor createExtractor(InputStream input) throws IOException {
    Class<?> cls = getOOXMLClass();
    if (cls != null) {
        // Use Reflection to get us the full OOXML-enabled version
        try {
            Method m = cls.getDeclaredMethod("createExtractor", InputStream.class);
            return (POITextExtractor)m.invoke(null, input);
        } catch (IllegalArgumentException iae) {
            throw iae;
        } catch (Exception e) {
            throw new IllegalArgumentException("Error creating Extractor for InputStream", e);
        }
    } else {
        // Best hope it's OLE2....
        return createExtractor(new NPOIFSFileSystem(input));
    }
}
 

/**
 * Create the Extractor, if possible. Generally needs the Scratchpad jar.
 * Note that this won't check for embedded OOXML resources either, use
 *  {@link org.apache.poi.extractor.ExtractorFactory} for that.
 */
public static POITextExtractor createExtractor(DirectoryNode poifsDir) throws IOException {
    // Look for certain entries in the stream, to figure it
    // out from
    for (String workbookName : WORKBOOK_DIR_ENTRY_NAMES) {
        if (poifsDir.hasEntry(workbookName)) {
            if (getPreferEventExtractor()) {
                return new EventBasedExcelExtractor(poifsDir);
            }
            return new ExcelExtractor(poifsDir);
        }
    }
    if (poifsDir.hasEntry(OLD_WORKBOOK_DIR_ENTRY_NAME)) {
        throw new OldExcelFormatException("Old Excel Spreadsheet format (1-95) "
                + "found. Please call OldExcelExtractor directly for basic text extraction");
    }
    
    // Ask Scratchpad, or fail trying
    Class<?> cls = getScratchpadClass();
    try {
        Method m = cls.getDeclaredMethod("createExtractor", DirectoryNode.class);
        POITextExtractor ext = (POITextExtractor)m.invoke(null, poifsDir);
        if (ext != null) return ext;
    } catch (IllegalArgumentException iae) {
        throw iae;
    } catch (Exception e) {
        throw new IllegalArgumentException("Error creating Scratchpad Extractor", e);
    }

    throw new IllegalArgumentException("No supported documents found in the OLE2 stream");
}
 

/**
 * Prevent recursion!
 */
public POITextExtractor getMetadataTextExtractor() {
    throw new IllegalStateException("You already have the Metadata Text Extractor, not recursing!");
}
 

private static String getTextFromExcel(String fileName)
		throws InvalidFormatException, IOException, OpenXML4JException, XmlException {
	File inputFile = new File(fileName);
	POITextExtractor extractor = ExtractorFactory.createExtractor(inputFile);
	return extractor.getText();
}