@Override
public BxDocument segmentDocument(BxDocument document) throws AnalysisException {
Map<BxPage, List<Component>> componentMap = new HashMap<BxPage, List<Component>>();
ExecutorService exec = Executors.newFixedThreadPool(PdfNLMContentExtractor.THREADS_NUMBER);
ArrayList<Callable<NumBxPage>> tasks = new ArrayList<Callable<NumBxPage>>();
for (BxPage page : document.getPages()) {
tasks.add(new ComponentCounter(page));
}
List<Future<NumBxPage>> results;
try {
results = exec.invokeAll(tasks);
exec.shutdown();
for (Future<NumBxPage> result : results) {
NumBxPage p = result.get();
componentMap.put(p.page, p.components);
}
} catch (ExecutionException ex) {
throw new AnalysisException("Cannot segment pages!", ex);
} catch (InterruptedException ex) {
throw new AnalysisException("Cannot segment pages!", ex);
}
this.computeDocumentOrientation(componentMap);
BxDocument output = new BxDocument();
BxPage[] pages = new BxPage[document.getPages().size()];
exec = Executors.newFixedThreadPool(PdfNLMContentExtractor.THREADS_NUMBER);
tasks = new ArrayList<Callable<NumBxPage>>();
int i = 0;
for (BxPage page : document.getPages()) {
tasks.add(new SingleSegmenter(page, i++));
}
try {
results = exec.invokeAll(tasks);
exec.shutdown();
for (Future<NumBxPage> result : results) {
NumBxPage p = result.get();
pages[p.index] = p.page;
}
for (BxPage p : pages) {
if (p.getBounds() != null) {
output.addPage(p);
}
}
return output;
} catch (ExecutionException ex) {
throw new AnalysisException("Cannot segment pages!", ex);
} catch (InterruptedException ex) {
throw new AnalysisException("Cannot segment pages!", ex);
}
}
public BxDocument setPages(Collection<BxPage> pages) {
if (pages != null) {
this.pages.clear();
curPageNumber = 0;
for (BxPage page : pages) {
addPage(page);
}
}
return this;
}