/**
* Public constructor.
*
* @param file the workflow file
* @throws IOException if an error occurs while opening the file
*/
public CommandWorkflowParser(final DataFile file) throws IOException {
this(file.open());
}
@Override
public void merge(final Iterator<DataFile> inFileIterator, DataFile outFile) throws IOException {
final Multiset<String> counts = HashMultiset.create();
final Set<String> emptyCounts = new HashSet<>();
while (inFileIterator.hasNext()) {
// Get input file
final DataFile inFile = inFileIterator.next();
EoulsanLogger.getLogger().info("Merge " + inFile.getName() + " to " + outFile.getName());
boolean first = true;
try (BufferedReader reader = new BufferedReader(new InputStreamReader(inFile.open()))) {
String line = null;
while ((line = reader.readLine()) != null) {
// Do no handle header
if (first) {
first = false;
continue;
}
final int tabPos = line.indexOf('\t');
// Do not handle empty or invalid lines
if (tabPos == -1) {
continue;
}
try {
final String id = line.substring(0, tabPos).trim();
final int count = Integer.parseInt(line.substring(tabPos).trim());
if (count == 0) {
emptyCounts.add(id);
}
counts.add(id, count);
} catch (NumberFormatException e) {
// Do not handle parsing errors
}
}
}
}
// Write the result file
try (Writer writer = new OutputStreamWriter(outFile.create())) {
writer.write(ExpressionSplitter.EXPRESSION_FILE_HEADER);
// Write the non empty counts
for (Multiset.Entry<String> e : counts.entrySet()) {
final String id = e.getElement();
// Remove the id from empty counts
emptyCounts.remove(id);
// Write the entry
writer.write(id + '\t' + e.getCount() + '\n');
}
// Write the empty counts
for (String id : emptyCounts) {
writer.write(id + "\t0\n");
}
}
}
