Github user omalley commented on a diff in the pull request:
https://github.com/apache/orc/pull/278#discussion_r202433445
--- Diff: java/core/src/java/org/apache/orc/impl/InStream.java ---
@@ -55,112 +60,137 @@ public long getStreamLength() {
@Override
public abstract void close();
+ static int getRangeNumber(DiskRangeList list, DiskRangeList current) {
+ int result = 0;
+ DiskRangeList range = list;
+ while (range != null && range != current) {
+ result += 1;
+ range = range.next;
+ }
+ return result;
+ }
+
+ /**
+ * Implements a stream over an uncompressed stream.
+ */
public static class UncompressedStream extends InStream {
- private List<DiskRange> bytes;
+ private DiskRangeList bytes;
private long length;
protected long currentOffset;
- private ByteBuffer range;
- private int currentRange;
+ protected ByteBuffer decrypted;
+ protected DiskRangeList currentRange;
+
+ /**
+ * Create the stream without calling reset on it.
+ * This is used for the subclass that needs to do more setup.
+ * @param name name of the stream
+ * @param length the number of bytes for the stream
+ */
+ public UncompressedStream(String name, long length) {
+ super(name, length);
+ }
- public UncompressedStream(String name, List<DiskRange> input, long
length) {
+ public UncompressedStream(String name,
+ DiskRangeList input,
+ long length) {
super(name, length);
reset(input, length);
}
- protected void reset(List<DiskRange> input, long length) {
+ protected void reset(DiskRangeList input, long length) {
this.bytes = input;
this.length = length;
- currentRange = 0;
- currentOffset = 0;
- range = null;
+ currentOffset = input == null ? 0 : input.getOffset();
+ setCurrent(input, true);
}
@Override
public int read() {
- if (range == null || range.remaining() == 0) {
+ if (decrypted == null || decrypted.remaining() == 0) {
if (currentOffset == length) {
return -1;
}
- seek(currentOffset);
+ setCurrent(currentRange.next, false);
}
currentOffset += 1;
- return 0xff & range.get();
+ return 0xff & decrypted.get();
+ }
+
+ protected void setCurrent(DiskRangeList newRange, boolean isJump) {
+ currentRange = newRange;
+ if (newRange != null) {
+ decrypted = newRange.getData().slice();
+ decrypted.position((int) (currentOffset - newRange.getOffset()));
+ }
}
@Override
public int read(byte[] data, int offset, int length) {
- if (range == null || range.remaining() == 0) {
+ if (decrypted == null || decrypted.remaining() == 0) {
if (currentOffset == this.length) {
return -1;
}
- seek(currentOffset);
+ setCurrent(currentRange.next, false);
}
- int actualLength = Math.min(length, range.remaining());
- range.get(data, offset, actualLength);
+ int actualLength = Math.min(length, decrypted.remaining());
+ decrypted.get(data, offset, actualLength);
currentOffset += actualLength;
return actualLength;
}
@Override
public int available() {
- if (range != null && range.remaining() > 0) {
- return range.remaining();
+ if (decrypted != null && decrypted.remaining() > 0) {
+ return decrypted.remaining();
}
return (int) (length - currentOffset);
}
@Override
public void close() {
- currentRange = bytes.size();
+ currentRange = null;
currentOffset = length;
// explicit de-ref of bytes[]
- bytes.clear();
+ decrypted = null;
+ bytes = null;
}
@Override
public void seek(PositionProvider index) throws IOException {
seek(index.getNext());
}
- public void seek(long desired) {
- if (desired == 0 && bytes.isEmpty()) {
+ public void seek(long desired) throws IOException {
+ if (desired == 0 && bytes == null) {
return;
}
- int i = 0;
- for (DiskRange curRange : bytes) {
- if (curRange.getOffset() <= desired &&
- (desired - curRange.getOffset()) < curRange.getLength()) {
- currentOffset = desired;
- currentRange = i;
- this.range = curRange.getData().duplicate();
- int pos = range.position();
- pos += (int)(desired - curRange.getOffset()); // this is why we
duplicate
- this.range.position(pos);
- return;
- }
- ++i;
- }
- // if they are seeking to the precise end, go ahead and let them go
there
- int segments = bytes.size();
- if (segments != 0 && desired == bytes.get(segments - 1).getEnd()) {
+ // If we are seeking inside of the current range, just reposition.
+ if (currentRange != null && desired >= currentRange.getOffset() &&
+ desired < currentRange.getEnd()) {
--- End diff --
No, the currentRange.getEnd() is the position after the range (offset +
length), so to see if an offset is in the currentRange, I need to use <.
---
