/**
* Alternative to other one with similar name
*
* @param directory directory containing SDF
* @param indexFile index file
* @param loadNames should we load names
* @param loadFullNames should we load full names
* @param region region to restrict to
* @throws IOException IO exception occurs
*/
CompressedMemorySequencesReader2(
File directory,
IndexFile indexFile,
boolean loadNames,
boolean loadFullNames,
LongRange region)
throws IOException {
mIndexFile = indexFile;
mRegion = SequencesReaderFactory.resolveRange(indexFile, region);
mStart = mRegion.getStart();
mEnd = mRegion.getEnd();
mNumberSequences = mEnd - mStart;
mData =
DataInMemory.loadDelayQuality(
directory,
indexFile,
DataFileIndex.loadSequenceDataFileIndex(indexFile.dataIndexVersion(), directory),
mStart,
mEnd);
if (mNumberSequences > Integer.MAX_VALUE) {
throw new IllegalArgumentException(
"Too many sequences in region: " + region + ", maximum is: " + Integer.MAX_VALUE);
}
mDirectory = directory;
if (loadNames && mIndexFile.hasNames()) {
loadNames();
loadNameSuffixes(loadFullNames, mIndexFile.hasSequenceNameSuffixes());
}
final StringBuilder sb = new StringBuilder("CMSR2 statistics");
sb.append(LS);
this.infoString(sb);
Diagnostic.userLog(sb.toString());
}
/**
* Load the names if they haven't already been loaded.
*
* @throws IOException if an I/O related error occurs
*/
private void loadNames() throws IOException {
mNames = new PrereadNames(mDirectory, mRegion, false);
if (mIndexFile.getVersion() >= IndexFile.SEPARATE_CHECKSUM_VERSION
&& mRegion.getStart() == 0
&& mRegion.getEnd() == mIndexFile.getNumberSequences()) {
if (mNames.calcChecksum() != mIndexFile.getNameChecksum()) {
throw new CorruptSdfException(
"Sequence names failed checksum - SDF may be corrupt: \"" + mDirectory + "\"");
} else {
Diagnostic.developerLog("Sequence names passed checksum");
}
}
}
private void loadNameSuffixes(boolean attemptLoad, boolean suffixExists) throws IOException {
mNameSuffixes =
attemptLoad && suffixExists
? new PrereadNames(mDirectory, mRegion, true)
: new EmptyStringPrereadNames(mEnd - mStart);
if (attemptLoad && suffixExists) {
if (mRegion.getStart() == 0 && mRegion.getEnd() == mIndexFile.getNumberSequences()) {
if (mNameSuffixes.calcChecksum() != mIndexFile.getNameSuffixChecksum()) {
throw new CorruptSdfException(
"Sequence name suffixes failed checksum - SDF may be corrupt: \""
+ mDirectory
+ "\"");
} else {
Diagnostic.developerLog("Sequence name suffixes passed checksum");
}
}
}
}