@Override
protected void parseContent(StreamLimiter streamLimiter, LanguageEnum lang) throws IOException {
HSSFWorkbook workbook = new HSSFWorkbook(streamLimiter.getNewInputStream());
ExcelExtractor excel = null;
try {
excel = new ExcelExtractor(workbook);
ParserResultItem result = getNewParserResultItem();
SummaryInformation info = excel.getSummaryInformation();
if (info != null) {
result.addField(ParserFieldEnum.title, info.getTitle());
result.addField(ParserFieldEnum.author, info.getAuthor());
result.addField(ParserFieldEnum.subject, info.getSubject());
}
String content = excel.getText();
result.addField(ParserFieldEnum.content, StringUtils.replaceConsecutiveSpaces(content, " "));
result.langDetection(10000, ParserFieldEnum.content);
} finally {
IOUtils.close(excel);
}
}
/** Command line extractor. */
public static void main(String[] args) {
CommandArgs cmdArgs;
try {
cmdArgs = new CommandArgs(args);
} catch (CommandParseException e) {
System.err.println(e.getMessage());
printUsageMessage(System.err);
System.exit(1);
return; // suppress compiler error
}
if (cmdArgs.isRequestHelp()) {
printUsageMessage(System.out);
return;
}
try {
InputStream is;
if (cmdArgs.getInputFile() == null) {
is = System.in;
} else {
is = new FileInputStream(cmdArgs.getInputFile());
}
HSSFWorkbook wb = new HSSFWorkbook(is);
ExcelExtractor extractor = new ExcelExtractor(wb);
extractor.setIncludeSheetNames(cmdArgs.shouldShowSheetNames());
extractor.setFormulasNotResults(!cmdArgs.shouldEvaluateFormulas());
extractor.setIncludeCellComments(cmdArgs.shouldShowCellComments());
extractor.setIncludeBlankCells(cmdArgs.shouldShowBlankCells());
extractor.setIncludeHeadersFooters(cmdArgs.shouldIncludeHeadersFooters());
System.out.println(extractor.getText());
} catch (Exception e) {
e.printStackTrace();
System.exit(1);
}
}
