private Iterator<Pair<DictionaryTerm, InvertedListPointer>> mergeInvertedLists(
InvertedListBuilder output,
Map<Long, Float> tdIdfWeightAccumulator,
long newTotalDocuments,
InvertedList... inputs)
throws IOException, InterruptedException {
AnnotatingIterator<Pair<DictionaryTerm, InvertedListPointer>, InvertedList>[] listIterators =
new AnnotatingIterator[inputs.length];
int position = 0;
for (InvertedList list : inputs) {
listIterators[position++] =
new AnnotatingIterator<Pair<DictionaryTerm, InvertedListPointer>, InvertedList>(
list.getTermIterator(), list);
}
IteratorMerger<Pair<Pair<DictionaryTerm, InvertedListPointer>, InvertedList>>
mergedTermIterators =
new IteratorMerger<Pair<Pair<DictionaryTerm, InvertedListPointer>, InvertedList>>(
TERM_MERGE_COMP, listIterators);
DuplicateCollectingIterator<Pair<Pair<DictionaryTerm, InvertedListPointer>, InvertedList>>
collectedTerms =
new DuplicateCollectingIterator<
Pair<Pair<DictionaryTerm, InvertedListPointer>, InvertedList>>(
TERM_MERGE_COMP, mergedTermIterators);
return new LazyMergedInvertedListWriterIterator(
collectedTerms, output, newTotalDocuments, tdIdfWeightAccumulator);
}
public static void main(String[] args) {
// TODO Auto-generated method stub
InvertedList list = new InvertedList();
TestDoc doc = new TestDoc(list);
FileReptile visitor = new FileReptile(doc);
Scanner sc = new Scanner(System.in);
System.out.println("Please Enter Path Name:");
Path path = Paths.get(sc.nextLine()).normalize().toAbsolutePath();
System.out.println("Scanning......\nPlease Wait......");
try {
if (!path.toFile().exists()) {
sc.close();
throw new Exception();
}
Files.walkFileTree(path, visitor);
} catch (Exception e) {
// e.printStackTrace();
System.out.println("Scanning Failed");
System.exit(-1);
}
// list.testOutput();
String s;
Pattern p = Pattern.compile("[\\w]{1,}");
Matcher m;
System.out.println("Scanning complete, you can search now!");
while (true) {
System.out.println("I search:");
s = sc.nextLine();
if (s.equals("quit")) break;
m = p.matcher(s);
if (m.find()) list.searchAndOutput(s);
}
sc.close();
}