void openNextFile() throws NoMoreDataException, IOException {
close();
currPathType = null;
while (true) {
if (nextFile >= inputFiles.size()) {
// exhausted files, start a new round, unless forever set to false.
if (!forever) {
throw new NoMoreDataException();
}
nextFile = 0;
iteration++;
}
File f = inputFiles.get(nextFile++);
if (verbose) {
System.out.println("opening: " + f + " length: " + f.length());
}
try {
InputStream inputStream =
StreamUtils.inputStream(
f); // support either gzip, bzip2, or regular text file, by extension
reader =
new BufferedReader(
new InputStreamReader(inputStream, encoding), StreamUtils.BUFFER_SIZE);
currPathType = TrecDocParser.pathType(f);
return;
} catch (Exception e) {
if (verbose) {
System.out.println(
"Skipping 'bad' file " + f.getAbsolutePath() + " due to " + e.getMessage());
continue;
}
throw new NoMoreDataException();
}
}
}
@Override
public DocData getNextDocData(DocData docData) throws NoMoreDataException, IOException {
String name = null;
StringBuilder docBuf = getDocBuffer();
ParsePathType parsedPathType;
// protect reading from the TREC files by multiple threads. The rest of the
// method, i.e., parsing the content and returning the DocData can run unprotected.
synchronized (lock) {
if (reader == null) {
openNextFile();
}
// 1. skip until doc start - required for all TREC formats
docBuf.setLength(0);
read(docBuf, DOC, false, false);
// save parsedFile for passing trecDataParser after the sync block, in
// case another thread will open another file in between.
parsedPathType = currPathType;
// 2. name - required for all TREC formats
docBuf.setLength(0);
read(docBuf, DOCNO, true, false);
name =
docBuf
.substring(DOCNO.length(), docBuf.indexOf(TERMINATING_DOCNO, DOCNO.length()))
.trim();
if (!excludeDocnameIteration) {
name = name + "_" + iteration;
}
// 3. read all until end of doc
docBuf.setLength(0);
read(docBuf, TERMINATING_DOC, false, true);
}
// count char length of text to be parsed (may be larger than the resulted plain doc body text).
addBytes(docBuf.length());
// This code segment relies on HtmlParser being thread safe. When we get
// here, everything else is already private to that thread, so we're safe.
try {
docData = trecDocParser.parse(docData, name, this, docBuf, parsedPathType);
addDoc();
} catch (InterruptedException ie) {
throw new ThreadInterruptedException(ie);
}
return docData;
}
