public SortingBucketMerger(
RowDownstream rowDownstream,
int numBuckets,
int[] orderByPositions,
boolean[] reverseFlags,
Boolean[] nullsFirst,
Optional<Executor> executor) {
Preconditions.checkArgument(numBuckets > 0, "must at least get 1 bucket per merge call");
this.numBuckets = numBuckets;
this.executor = executor;
List<Comparator<Row>> comparators = new ArrayList<>(orderByPositions.length);
for (int i = 0; i < orderByPositions.length; i++) {
OrderingByPosition<Row> rowOrdering =
OrderingByPosition.rowOrdering(orderByPositions[i], reverseFlags[i], nullsFirst[i]);
comparators.add(rowOrdering.reverse());
}
ordering = Ordering.compound(comparators);
downstream = rowDownstream.registerUpstream(this);
}
public FileReadingCollector(
String fileUri,
List<Input<?>> inputs,
List<LineCollectorExpression<?>> collectorExpressions,
RowDownstream downstream,
FileFormat format,
String compression,
Map<String, FileInputFactory> additionalFileInputFactories,
Boolean shared,
int numReaders,
int readerNumber) {
if (fileUri.startsWith("/")) {
// using Paths.get().toUri instead of new URI(...) as it also encodes umlauts and other
// special characters
this.fileUri = Paths.get(fileUri).toUri();
} else {
this.fileUri = URI.create(fileUri);
if (this.fileUri.getScheme() == null) {
throw new IllegalArgumentException("relative fileURIs are not allowed");
}
if (this.fileUri.getScheme().equals("file")
&& !this.fileUri.getSchemeSpecificPart().startsWith("///")) {
throw new IllegalArgumentException("Invalid fileURI");
}
if (!this.fileUri.getScheme().equals("file") && !this.fileUri.getScheme().equals("s3")) {
throw new IllegalArgumentException("URI scheme is not supported");
}
}
this.downstream = downstream.registerUpstream(this);
this.compressed = compression != null && compression.equalsIgnoreCase("gzip");
this.row = new InputRow(inputs);
this.collectorExpressions = collectorExpressions;
this.fileInputFactoryMap =
new HashMap<>(
ImmutableMap.of(
"s3",
new FileInputFactory() {
@Override
public FileInput create() throws IOException {
return new S3FileInput();
}
},
"file",
new FileInputFactory() {
@Override
public FileInput create() throws IOException {
return new LocalFsFileInput();
}
}));
this.fileInputFactoryMap.putAll(additionalFileInputFactories);
this.shared = shared;
this.numReaders = numReaders;
this.readerNumber = readerNumber;
Matcher hasGlobMatcher = HAS_GLOBS_PATTERN.matcher(this.fileUri.toString());
if (!hasGlobMatcher.matches()) {
globPredicate = null;
} else {
this.preGlobUri = URI.create(hasGlobMatcher.group(1));
final Pattern globPattern =
Pattern.compile(Globs.toUnixRegexPattern(this.fileUri.toString()));
globPredicate =
new Predicate<URI>() {
@Override
public boolean apply(URI input) {
return globPattern.matcher(input.toString()).matches();
}
};
}
}
