tikhomirov@0: /**
tikhomirov@0:  * Copyright (c) 2010 Artem Tikhomirov 
tikhomirov@0:  */
tikhomirov@0: package com.tmate.hgkit.ll;
tikhomirov@0: 
tikhomirov@0: import java.io.DataInput;
tikhomirov@2: import java.io.EOFException;
tikhomirov@2: import java.io.IOException;
tikhomirov@2: import java.util.ArrayList;
tikhomirov@2: import java.util.Collections;
tikhomirov@2: import java.util.List;
tikhomirov@2: import java.util.zip.Inflater;
tikhomirov@0: 
tikhomirov@0: /**
tikhomirov@0:  * ? Single RevlogStream per file per repository with accessor to record access session (e.g. with back/forward operations), 
tikhomirov@0:  * or numerous RevlogStream with separate representation of the underlaying data (cached, lazy ChunkStream)?
tikhomirov@0:  * @author artem
tikhomirov@0:  * @see http://mercurial.selenic.com/wiki/Revlog
tikhomirov@0:  * @see http://mercurial.selenic.com/wiki/RevlogNG
tikhomirov@0:  */
tikhomirov@0: public class RevlogStream {
tikhomirov@2: 
tikhomirov@2: 	private List<IndexEntry> index; // indexed access highly needed
tikhomirov@2: 	private boolean inline = false;
tikhomirov@2: 
tikhomirov@0: 	private void detectVersion() {
tikhomirov@0: 		
tikhomirov@0: 	}
tikhomirov@0: 
tikhomirov@0: 	/*package*/ DataInput getIndexStream() {
tikhomirov@0: 		// TODO Auto-generated method stub
tikhomirov@0: 		return null;
tikhomirov@0: 	}
tikhomirov@0: 
tikhomirov@0: 	/*package*/ DataInput getDataStream() {
tikhomirov@0: 		// TODO Auto-generated method stub
tikhomirov@0: 		return null;
tikhomirov@0: 	}
tikhomirov@2: 	public int revisionCount() {
tikhomirov@2: 		initOutline();
tikhomirov@2: 		return index.size();
tikhomirov@2: 	}
tikhomirov@2: 
tikhomirov@2: 	// should be possible to use TIP, ALL
tikhomirov@2: 	public void iterate(int start, int end, boolean needData, Revlog.Inspector inspector) {
tikhomirov@2: 		initOutline();
tikhomirov@2: 		final int indexSize = index.size();
tikhomirov@2: 		if (start < 0 || start >= indexSize) {
tikhomirov@2: 			throw new IllegalArgumentException("Bad left range boundary " + start);
tikhomirov@2: 		}
tikhomirov@2: 		if (end < start || end >= indexSize) {
tikhomirov@2: 			throw new IllegalArgumentException("Bad right range boundary " + end);
tikhomirov@2: 		}
tikhomirov@2: 		// XXX may cache [start .. end] from index with a single read (pre-read)
tikhomirov@2: 		
tikhomirov@2: 		DataInput diIndex = null, diData = null;
tikhomirov@2: 		diIndex = getIndexStream();
tikhomirov@2: 		if (needData) {
tikhomirov@2: 			diData = getDataStream();
tikhomirov@2: 		}
tikhomirov@2: 		try {
tikhomirov@2: 			diIndex.skipBytes(inline ? (int) index.get(start).offset : start * 64);
tikhomirov@2: 			for (int i = start; i <= end && i < indexSize; i++ ) {
tikhomirov@2: 				IndexEntry ie = index.get(i);
tikhomirov@2: 				long l = diIndex.readLong();
tikhomirov@2: 				long offset = l >>> 16;
tikhomirov@2: 				int flags = (int) (l & 0X0FFFF);
tikhomirov@2: 				int compressedLen = diIndex.readInt();
tikhomirov@2: 				int actualLen = diIndex.readInt();
tikhomirov@2: 				int baseRevision = diIndex.readInt();
tikhomirov@2: 				int linkRevision = diIndex.readInt();
tikhomirov@2: 				int parent1Revision = diIndex.readInt();
tikhomirov@2: 				int parent2Revision = diIndex.readInt();
tikhomirov@2: 				byte[] buf = new byte[32];
tikhomirov@2: 				// XXX Hg keeps 12 last bytes empty, we move them into front here
tikhomirov@2: 				diIndex.readFully(buf, 12, 20);
tikhomirov@2: 				diIndex.skipBytes(12);
tikhomirov@2: 				byte[] data = null;
tikhomirov@2: 				if (needData) {
tikhomirov@2: 					byte[] dataBuf = new byte[compressedLen];
tikhomirov@2: 					if (inline) {
tikhomirov@2: 						diIndex.readFully(dataBuf);
tikhomirov@2: 					} else {
tikhomirov@2: 						diData.skipBytes((int) ie.offset); // FIXME not skip but seek!!!
tikhomirov@2: 						diData.readFully(dataBuf);
tikhomirov@2: 					}
tikhomirov@2: 					if (dataBuf[0] == 0x78 /* 'x' */) {
tikhomirov@2: 						Inflater zlib = new Inflater();
tikhomirov@2: 						zlib.setInput(dataBuf, 0, compressedLen);
tikhomirov@2: 						byte[] result = new byte[actualLen*2]; // FIXME need to use zlib.finished() instead 
tikhomirov@2: 						int resultLen = zlib.inflate(result);
tikhomirov@2: 						zlib.end();
tikhomirov@2: 						data = new byte[resultLen];
tikhomirov@2: 						System.arraycopy(result, 0, data, 0, resultLen);
tikhomirov@2: 					} else if (dataBuf[0] == 0x75 /* 'u' */) {
tikhomirov@2: 						data = new byte[dataBuf.length - 1];
tikhomirov@2: 						System.arraycopy(dataBuf, 1, data, 0, data.length);
tikhomirov@2: 					} else {
tikhomirov@2: 						// XXX Python impl in fact throws exception when there's not 'x', 'u' or '0'
tikhomirov@2: 						// but I don't see reason not to just return data as is 
tikhomirov@2: 						data = dataBuf;
tikhomirov@2: 					}
tikhomirov@2: 					// FIXME if patch data (based on baseRevision), then apply patches to get true content
tikhomirov@2: 				}
tikhomirov@2: 				inspector.next(compressedLen, actualLen, baseRevision, linkRevision, parent1Revision, parent2Revision, buf, data);
tikhomirov@2: 			}
tikhomirov@2: 		} catch (EOFException ex) {
tikhomirov@2: 			// should not happen as long as we read inside known boundaries
tikhomirov@2: 			throw new IllegalStateException(ex);
tikhomirov@2: 		} catch (IOException ex) {
tikhomirov@2: 			FIXME 
tikhomirov@2: 		}
tikhomirov@2: 	}
tikhomirov@0: 	
tikhomirov@2: 	private void initOutline() {
tikhomirov@2: 		if (index != null && !index.isEmpty()) {
tikhomirov@2: 			return;
tikhomirov@2: 		}
tikhomirov@2: 		ArrayList<IndexEntry> res = new ArrayList<IndexEntry>();
tikhomirov@2: 		DataInput di = getIndexStream();
tikhomirov@2: 		try {
tikhomirov@2: 			int versionField = di.readInt();
tikhomirov@2: 			// di.undreadInt();
tikhomirov@2: 			final int INLINEDATA = 1 << 16;
tikhomirov@2: 			inline = (versionField & INLINEDATA) != 0;
tikhomirov@2: 			long offset = 0; // first offset is always 0, thus Hg uses it for other purposes
tikhomirov@2: 			while(true) { // EOFExcepiton should get us outta here. FIXME Out inputstream should has explicit no-more-data indicator
tikhomirov@2: 				int compressedLen = di.readInt();
tikhomirov@2: 				// 8+4 = 12 bytes total read
tikhomirov@2: //				int actualLen = di.readInt();
tikhomirov@2: //				int baseRevision = di.readInt();
tikhomirov@2: //				int linkRevision = di.readInt();
tikhomirov@2: //				int parent1Revision = di.readInt();
tikhomirov@2: //				int parent2Revision = di.readInt();
tikhomirov@2: //				byte[] nodeid = new byte[32];
tikhomirov@2: 				res.add(new IndexEntry(offset, compressedLen));
tikhomirov@2: 				if (inline) {
tikhomirov@2: 					di.skipBytes(6*4 + 32 + compressedLen); // Check: 56 (skip) + 12 (read) = 64 (total RevlogNG record size)
tikhomirov@2: 				} else {
tikhomirov@2: 					di.skipBytes(6*4 + 32);
tikhomirov@2: 				}
tikhomirov@2: 				long l = di.readLong();
tikhomirov@2: 				offset = l >>> 16;
tikhomirov@2: 			}
tikhomirov@2: 		} catch (EOFException ex) {
tikhomirov@2: 			// fine, done then
tikhomirov@2: 			index = res;
tikhomirov@2: 		} catch (IOException ex) {
tikhomirov@2: 			ex.printStackTrace();
tikhomirov@2: 			// too bad, no outline then
tikhomirov@2: 			index = Collections.emptyList();
tikhomirov@2: 		}
tikhomirov@2: 	}
tikhomirov@2: 
tikhomirov@2: 
tikhomirov@2: 	// perhaps, package-local or protected, if anyone else from low-level needs them
tikhomirov@2: 	private static class IndexEntry {
tikhomirov@2: 		public final long offset;
tikhomirov@2: 		public final int length; // data past fixed record (need to decide whether including header size or not), and whether length is of compressed data or not
tikhomirov@2: 
tikhomirov@2: 		public IndexEntry(long o, int l) {
tikhomirov@2: 			offset = o;
tikhomirov@2: 			length = l;
tikhomirov@2: 		}
tikhomirov@2: 	}
tikhomirov@0: }