tikhomirov@329: /*
tikhomirov@419:  * Copyright (c) 2011-2012 TMate Software Ltd
tikhomirov@329:  *  
tikhomirov@329:  * This program is free software; you can redistribute it and/or modify
tikhomirov@329:  * it under the terms of the GNU General Public License as published by
tikhomirov@329:  * the Free Software Foundation; version 2 of the License.
tikhomirov@329:  *
tikhomirov@329:  * This program is distributed in the hope that it will be useful,
tikhomirov@329:  * but WITHOUT ANY WARRANTY; without even the implied warranty of
tikhomirov@329:  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
tikhomirov@329:  * GNU General Public License for more details.
tikhomirov@329:  *
tikhomirov@329:  * For information on how to redistribute this software under
tikhomirov@329:  * the terms of a license other than GNU General Public License
tikhomirov@329:  * contact TMate Software at support@hg4j.com
tikhomirov@329:  */
tikhomirov@329: package org.tmatesoft.hg.internal;
tikhomirov@329: 
tikhomirov@329: import java.io.IOException;
tikhomirov@329: import java.util.ArrayList;
tikhomirov@330: import java.util.Formatter;
tikhomirov@329: 
tikhomirov@329: /**
tikhomirov@419:  * @see http://mercurial.selenic.com/wiki/BundleFormat
tikhomirov@419:  * in Changelog group description
tikhomirov@329:  * 
tikhomirov@329:  * range [start..end] in original source gets replaced with data of length (do not keep, use data.length instead)
tikhomirov@329:  * range [end(i)..start(i+1)] is copied from the source
tikhomirov@329:  * 
tikhomirov@329:  * @author Artem Tikhomirov
tikhomirov@329:  * @author TMate Software Ltd.
tikhomirov@329:  */
tikhomirov@329: public final class Patch {
tikhomirov@329: 	private final IntVector starts, ends;
tikhomirov@329: 	private final ArrayList<byte[]> data;
tikhomirov@330: 
tikhomirov@330: 	private static byte[] generate(int c) {
tikhomirov@330: 		byte[] rv = new byte[c];
tikhomirov@330: 		for (int i = 0; i < c; i++) {
tikhomirov@330: 			byte x = (byte) ('a' + i);
tikhomirov@330: 			rv[i] = x;
tikhomirov@330: 		}
tikhomirov@330: 		return rv;
tikhomirov@330: 	}
tikhomirov@330: 
tikhomirov@330: 	public static void main(String[] args) {
tikhomirov@330: 		Patch p1 = new Patch(), p2 = new Patch();
tikhomirov@330: 		// simple cases (one element in either patch)
tikhomirov@330: 		// III: (1,10 20) & (5,15,15) p2End from [p1End..p1AppliedEnd] (i.e. within p1 range but index is past p2 end index) 
tikhomirov@330: 		//  II: (1,10,7) & (3,15,15) insideP2 = true and no more p1 entries
tikhomirov@330: 		//  II: (1,1,10) & (3,11,15)
tikhomirov@330: 		// independent: (1,10,10) & (15,25,10);  (15, 25, 10) & (1, 10, 10) 
tikhomirov@330: 		//   I: (15, 25, 10) & (10, 20, 10). result: [10, 20, 10] [20, 25, 5]
tikhomirov@330: 		//  IV: (15, 25, 10) & (10, 30, 20)
tikhomirov@330: 		// 
tikhomirov@330: 		// cycle with insideP2
tikhomirov@330: 		//
tikhomirov@330: 		// cycle with insideP1
tikhomirov@330: 		//
tikhomirov@330: 		// multiple elements in patches (offsets)
tikhomirov@330: 		p1.add(15, 25, generate(10));
tikhomirov@330: 		p2.add(10, 30, generate(20));
tikhomirov@330: 		System.out.println("p1: " + p1);
tikhomirov@330: 		System.out.println("p2: " + p2);
tikhomirov@330: 		Patch r = p1.apply(p2);
tikhomirov@330: 		System.out.println("r: " + r);
tikhomirov@330: 	}
tikhomirov@330: 
tikhomirov@329: 	public Patch() {
tikhomirov@329: 		starts = new IntVector();
tikhomirov@329: 		ends = new IntVector();
tikhomirov@329: 		data = new ArrayList<byte[]>();
tikhomirov@329: 	}
tikhomirov@329: 	
tikhomirov@330: 	public String toString() {
tikhomirov@330: 		StringBuilder sb = new StringBuilder();
tikhomirov@330: 		Formatter f = new Formatter(sb);
tikhomirov@330: 		for (int i = 0; i < count(); i++) {
tikhomirov@330: 			f.format("[%d, %d, %d] ", starts.get(i), ends.get(i), data.get(i).length);
tikhomirov@330: 		}
tikhomirov@330: 		return sb.toString();
tikhomirov@330: 	}
tikhomirov@330: 	
tikhomirov@329: 	public int count() {
tikhomirov@329: 		return data.size();
tikhomirov@329: 	}
tikhomirov@329: 
tikhomirov@329: 	// number of bytes this patch will add (or remove, if negative) from the base revision
tikhomirov@329: 	private int patchSizeDelta() {
tikhomirov@329: 		int rv = 0;
tikhomirov@329: 		int prevEnd = 0;
tikhomirov@329: 		for (int i = 0, x = data.size(); i < x; i++) {
tikhomirov@329: 			final int start = starts.get(i);
tikhomirov@329: 			final int len = data.get(i).length;
tikhomirov@329: 			rv += start - prevEnd; // would copy from original
tikhomirov@329: 			rv += len; // and add new
tikhomirov@329: 			prevEnd = ends.get(i);
tikhomirov@329: 		}
tikhomirov@329: 		rv -= prevEnd;
tikhomirov@329: 		return rv;
tikhomirov@329: 	}
tikhomirov@329: 	
tikhomirov@329: 	public byte[] apply(DataAccess baseRevisionContent, int outcomeLen) throws IOException {
tikhomirov@329: 		if (outcomeLen == -1) {
tikhomirov@329: 			outcomeLen = baseRevisionContent.length() + patchSizeDelta();
tikhomirov@329: 		}
tikhomirov@329: 		int prevEnd = 0, destIndex = 0;
tikhomirov@329: 		byte[] rv = new byte[outcomeLen];
tikhomirov@329: 		for (int i = 0, x = data.size(); i < x; i++) {
tikhomirov@329: 			final int start = starts.get(i);
tikhomirov@329: 			baseRevisionContent.seek(prevEnd);
tikhomirov@329: 			// copy source bytes that were not modified (up to start of the record)
tikhomirov@329: 			baseRevisionContent.readBytes(rv, destIndex, start - prevEnd);
tikhomirov@329: 			destIndex += start - prevEnd;
tikhomirov@329: 			// insert new data from the patch, if any
tikhomirov@329: 			byte[] d = data.get(i);
tikhomirov@532: 			System.arraycopy(d, 0, rv, destIndex, d.length);
tikhomirov@329: 			destIndex += d.length;
tikhomirov@329: 			prevEnd = ends.get(i);
tikhomirov@329: 		}
tikhomirov@329: 		baseRevisionContent.seek(prevEnd);
tikhomirov@329: 		// copy everything in the source past last record's end
tikhomirov@420: 		baseRevisionContent.readBytes(rv, destIndex, (baseRevisionContent.length() - prevEnd));
tikhomirov@329: 		return rv;
tikhomirov@329: 	}
tikhomirov@329: 	
tikhomirov@329: 	public void clear() {
tikhomirov@329: 		starts.clear();
tikhomirov@329: 		ends.clear();
tikhomirov@329: 		data.clear();
tikhomirov@329: 	}
tikhomirov@329: 	
tikhomirov@329: 	/**
tikhomirov@329: 	 * Initialize instance from stream. Any previous patch information (i.e. if instance if reused) is cleared first.
tikhomirov@329: 	 * Read up to the end of DataAccess and interpret data as patch records.
tikhomirov@329: 	 */
tikhomirov@329: 	public void read(DataAccess da) throws IOException {
tikhomirov@329: 		clear();
tikhomirov@329: 		while (!da.isEmpty()) {
tikhomirov@329: 			readOne(da);
tikhomirov@329: 		}
tikhomirov@329: 	}
tikhomirov@329: 
tikhomirov@329: 	/**
tikhomirov@329: 	 * Caller is responsible to ensure stream got some data to read
tikhomirov@329: 	 */
tikhomirov@329: 	public void readOne(DataAccess da) throws IOException {
tikhomirov@329: 		int s = da.readInt();
tikhomirov@329: 		int e = da.readInt();
tikhomirov@329: 		int len = da.readInt();
tikhomirov@329: 		byte[] src = new byte[len];
tikhomirov@329: 		da.readBytes(src, 0, len);
tikhomirov@329: 		starts.add(s);
tikhomirov@329: 		ends.add(e);
tikhomirov@329: 		data.add(src);
tikhomirov@329: 	}
tikhomirov@329: 
tikhomirov@330: 	private void add(Patch p, int i) {
tikhomirov@330: 		add(p.starts.get(i), p.ends.get(i), p.data.get(i));
tikhomirov@330: 	}
tikhomirov@330: 
tikhomirov@330: 	private void add(int start, int end, byte[] d) {
tikhomirov@330: 		starts.add(start);
tikhomirov@330: 		ends.add(end);
tikhomirov@330: 		data.add(d);
tikhomirov@330: 	}
tikhomirov@330: 	
tikhomirov@330: 	private static byte[] subarray(byte[] d, int start, int end) {
tikhomirov@330: 		byte[] r = new byte[end-start+1];
tikhomirov@330: 		System.arraycopy(d, start, r, 0, r.length);
tikhomirov@330: 		return r;
tikhomirov@329: 	}
tikhomirov@329: 
tikhomirov@329: 	/**
tikhomirov@329: 	 * Modify this patch with subsequent patch 
tikhomirov@330: 	 */
tikhomirov@330: 	private /*SHALL BE PUBLIC ONCE TESTING ENDS*/ Patch apply(Patch another) {
tikhomirov@329: 		Patch r = new Patch();
tikhomirov@330: 		int p1TotalAppliedDelta = 0; // value to add to start and end indexes of the older patch to get their values as if
tikhomirov@330: 		// in the patched text, iow, directly comparable with respective indexes from the newer patch.
tikhomirov@330: 		int p1EntryStart = 0, p1EntryEnd = 0, p1EntryLen = 0;
tikhomirov@330: 		byte[] p1Data = null;
tikhomirov@330: 		boolean insideP1entry = false;
tikhomirov@330: 		int p2 = 0, p1 = 0;
tikhomirov@330: 		final int p2Max = another.count(), p1Max = this.count();
tikhomirov@330: L0:		for (; p2 < p2Max; p2++) {
tikhomirov@330: 			int p2EntryStart = another.starts.get(p2);
tikhomirov@330: 			int p2EntryEnd = another.ends.get(p2);
tikhomirov@330: 			final int p2EntryRange = p2EntryEnd - p2EntryStart;
tikhomirov@330: 			final byte[] p2Data = another.data.get(p2);
tikhomirov@330: 			boolean insideP2entry = false;
tikhomirov@330: 			int p2EntryStartOffset = -1;
tikhomirov@330: 			///
tikhomirov@330: 			p1EntryStart = p1EntryEnd = p1EntryLen = 0;
tikhomirov@330: 			p1Data = null;
tikhomirov@329: 			
tikhomirov@330: L1:			while (p1 < p1Max) {
tikhomirov@330: 				if (!insideP1entry) {
tikhomirov@330: 					p1EntryStart = starts.get(p1);
tikhomirov@330: 					p1EntryEnd = ends.get(p1);
tikhomirov@330: 					p1Data = data.get(p1);
tikhomirov@330: 					p1EntryLen = p1Data.length;
tikhomirov@330: 				}// else keep values
tikhomirov@330: 
tikhomirov@330: 				final int p1EntryDelta = p1EntryLen - (p1EntryEnd - p1EntryStart); // number of actually inserted(+) or deleted(-) chars
tikhomirov@330: 				final int p1EntryAppliedStart = p1TotalAppliedDelta + p1EntryStart;
tikhomirov@330: 				final int p1EntryAppliedEnd = p1EntryAppliedStart + p1EntryLen; // end of j'th patch entry in the text which is source for p2
tikhomirov@330: 				
tikhomirov@330: 				if (insideP2entry) {
tikhomirov@330: 					if (p2EntryEnd < p1EntryAppliedStart) {
tikhomirov@330: 						r.add(p2EntryStart - p2EntryStartOffset, p2EntryEnd - p1TotalAppliedDelta, p2Data);
tikhomirov@330: 						insideP2entry = false;
tikhomirov@330: 						continue L0; 
tikhomirov@330: 					}
tikhomirov@330: 					if (p2EntryEnd >= p1EntryAppliedEnd) {
tikhomirov@330: 						// when p2EntryEnd == p1EntryAppliedEnd, I assume p1TotalAppliedDelta can't be used for p2EntryEnd to get it to p1 range, but rather shall be
tikhomirov@330: 						// augmented with current p1 entry and at the next p1 entry (likely to hit p1EntryAppliedStart > p2EntryEnd above) would do the rest 
tikhomirov@330: 						insideP1entry = false;
tikhomirov@330: 						p1++;
tikhomirov@330: 						p1TotalAppliedDelta += p1EntryDelta;
tikhomirov@330: 						continue L1;
tikhomirov@330: 					}
tikhomirov@330: 					// p1EntryAppliedStart <= p2EntryEnd < p1EntryAppliedEnd
tikhomirov@330: 					r.add(p2EntryStart - p2EntryStartOffset, p2EntryEnd - p1TotalAppliedDelta, p2Data);
tikhomirov@330: 					p1EntryStart = p2EntryEnd - p1TotalAppliedDelta;
tikhomirov@330: 					final int p1DataPartShift = p2EntryEnd - p1EntryAppliedStart + 1;
tikhomirov@330: 					if (p1DataPartShift >= p1EntryLen) {
tikhomirov@330: 						p1EntryLen = 0;
tikhomirov@330: 						p1Data = new byte[0];
tikhomirov@329: 					} else {
tikhomirov@330: 						p1EntryLen -= p1DataPartShift;
tikhomirov@330: 						p1Data = subarray(p1Data, p1DataPartShift, p1Data.length);
tikhomirov@330: 					}
tikhomirov@330: 					insideP1entry = true;
tikhomirov@330: 					insideP2entry = false;
tikhomirov@330: 					continue L0;
tikhomirov@330: 				}
tikhomirov@330: 
tikhomirov@330: 				if (p1EntryAppliedStart < p2EntryStart) {
tikhomirov@330: 					if (p1EntryAppliedEnd <= p2EntryStart) { // p1EntryAppliedEnd in fact index of the first char *after* patch
tikhomirov@330: 						// completely independent, copy and continue
tikhomirov@330: 						r.add(p1EntryStart, p1EntryEnd, p1Data);
tikhomirov@330: 						insideP1entry = false;
tikhomirov@330: 						p1++;
tikhomirov@330: 						// fall-through to get p1TotalAppliedDelta incremented
tikhomirov@330: 					} else { // SKETCH: II or III
tikhomirov@330: 						// remember, p1EntryDelta may be negative
tikhomirov@330: 						// shall break j'th entry into few 
tikhomirov@330: 						// fix p1's end/length
tikhomirov@330: 						// p1EntryAppliedStart < p2EntryStart < p1EntryAppliedEnd
tikhomirov@330: 						int s = p2EntryStart - p1TotalAppliedDelta; // p2EntryStart in p1 scale. Is within p1 range
tikhomirov@330: 						if (s > p1EntryEnd) {
tikhomirov@330: 							s = p1EntryEnd;
tikhomirov@330: 						}
tikhomirov@330: 						int p1DataPartEnd = p2EntryStart - p1EntryAppliedStart; // index, not count. <= (p1EntryEnd-p1EntryStart).
tikhomirov@330: 						// add what left from p1
tikhomirov@330: 						if (p1DataPartEnd < p1EntryLen) {
tikhomirov@330: 							r.add(p1EntryStart, s, subarray(p1Data, 0, p1DataPartEnd)); 
tikhomirov@330: 						} else {
tikhomirov@330: 							p1DataPartEnd = p1EntryLen-1; // record factual number of p1 bytes we consumed.
tikhomirov@330: 							r.add(p1EntryStart, s, p1Data);
tikhomirov@330: 						}
tikhomirov@330: 						p1TotalAppliedDelta += p1DataPartEnd - (s - p1EntryStart); // (s2 - (s1+delta)) - (s2 - delta - s1) = s2-s1-delta-s2+delta+s1 = 0, unless p1DataPartEnd >= p1Data.length
tikhomirov@330: 						p1EntryLen -= (p1DataPartEnd+1); 
tikhomirov@330: 						if (p2EntryEnd < p1EntryAppliedEnd) {
tikhomirov@330: 							// SKETCH: III
tikhomirov@330: 							insideP1entry = true;
tikhomirov@330: 							// p2 completely fits into changes of p1
tikhomirov@330: 							int e = p2EntryEnd - p1TotalAppliedDelta; // p2EntryEnd in p1 scale
tikhomirov@330: 							if (e > p1EntryEnd) {
tikhomirov@330: 								// any index past p1 end shall be calculated with respect to p1 end, thus it's unsafe to go past p1 end (there may be more p1 entries there)   
tikhomirov@330: 								e = p1EntryEnd;
tikhomirov@330: 							}
tikhomirov@330: 							r.add(s, e, p2Data); // add p2
tikhomirov@330: 							// modify p1 leftover
tikhomirov@330: 							p1EntryStart = e;
tikhomirov@330: 							if (p2EntryRange >= p1EntryLen) {
tikhomirov@330: 								p1EntryLen = 0;
tikhomirov@330: 								p1Data = new byte[0];
tikhomirov@330: 							} else {
tikhomirov@330: 								p1Data = subarray(p1Data, p1DataPartEnd + p2EntryRange, p1Data.length-1 /*up to the last one*/);
tikhomirov@330: 								p1EntryLen -= p2EntryRange;
tikhomirov@330: 							}
tikhomirov@330: 							// p2 is handled, but there are leftovers of p1
tikhomirov@330: 							continue L0;
tikhomirov@330: 						} else { // p2EntryEnd >= p1EntryAppliedEnd
tikhomirov@330: 							// SKETCH: II
tikhomirov@330: 							insideP1entry = false;
tikhomirov@330: 							p1++;
tikhomirov@330: 							if (p1EntryAppliedStart + p1EntryDelta >= p2EntryEnd) {
tikhomirov@330: 								// here we know next p1 entry would be past p2 entry and thus can put p2 right away
tikhomirov@330: 								r.add(p2EntryStart - p1TotalAppliedDelta, p1EntryEnd, p2Data);
tikhomirov@330: 								p1TotalAppliedDelta += p1EntryDelta;
tikhomirov@330: 								continue L0;
tikhomirov@330: 							} else {
tikhomirov@330: 								// there are chances there are more p1 entries till p2 ends
tikhomirov@330: 								insideP2entry = true;
tikhomirov@330: 								p2EntryStartOffset = p1TotalAppliedDelta;
tikhomirov@330: 								// p2EntryEnd is past delta, no chances for p1Data leftovers to be in use
tikhomirov@330: 								// p2 processing is not over, need to fix end, depending on what else fits into p2 range (if nothing, can put p2.end right away)
tikhomirov@330: 								// fall-through to get p1TotalAppliedDelta incremented;
tikhomirov@330: 							}
tikhomirov@330: 						}
tikhomirov@330: 					}
tikhomirov@330: 				} else { // p1EntryAppliedStart >= p2EntryStart
tikhomirov@330: 					if (p2EntryEnd < p1EntryAppliedStart) {
tikhomirov@330: 						// newer patch completely fits between two older patches 
tikhomirov@330: 						r.add(p2EntryStart - p1TotalAppliedDelta, p2EntryEnd - p1TotalAppliedDelta, p2Data);
tikhomirov@330: 						// SHALL NOT increment p1TotalAppliedDelta as we didn't use any of p1
tikhomirov@330: 						continue L0; // next p2 
tikhomirov@330: 					} else { // p2EntryEnd >= p1EntryAppliedStart
tikhomirov@330: 						// SKETCH: I or IV
tikhomirov@330: 						// p2EntryEnd is either  < p1EntryAppliedEnd or past it
tikhomirov@330: 						if (p2EntryEnd <= p1EntryAppliedEnd) {
tikhomirov@330: 							// SKETCH: I: copy p2, strip p1 to start from p2EntryEnd, next i (p2)
tikhomirov@330: 							insideP1entry = true;
tikhomirov@330: 							int e = p2EntryEnd - p1TotalAppliedDelta;
tikhomirov@330: 							if (e > p1EntryEnd) {
tikhomirov@330: 								e = p1EntryEnd; // added by analogy with above. Is needed?
tikhomirov@330: 							}
tikhomirov@330: 							r.add(p2EntryStart - p1TotalAppliedDelta, e, p2Data);
tikhomirov@330: 							p1EntryStart = e;
tikhomirov@330: 							int p1DataShift = p2EntryEnd - p1EntryAppliedStart;
tikhomirov@330: 							if (p1DataShift >= p1EntryLen) {
tikhomirov@330: 								p1EntryLen = 0;
tikhomirov@330: 								p1Data = new byte[0];
tikhomirov@330: 							} else {
tikhomirov@330: 								p1EntryLen -= p1DataShift;
tikhomirov@330: 								p1Data = subarray(p1Data, p1DataShift, p1Data.length - 1);
tikhomirov@330: 							}
tikhomirov@330: 							// p1TotalAppliedDelta would get incremented once this modified p1 is handled
tikhomirov@330: 							continue L0; // next p2;
tikhomirov@330: 						} else {
tikhomirov@330: 							// p2EntryEnd > p1EntryAppliedEnd
tikhomirov@330: 							// SKETCH IV: skip (rest of) p1 completely, continue the same unless  found p1 with start or end past p2EntryEnd.
tikhomirov@330: 							insideP1entry = false;
tikhomirov@330: 							p1++;
tikhomirov@330: 							insideP2entry = true;
tikhomirov@330: 							p2EntryStartOffset = p1TotalAppliedDelta;
tikhomirov@330: 							// fall-through to get p1TotalAppliedDelta incremented
tikhomirov@330: 						}
tikhomirov@329: 					}
tikhomirov@329: 				}
tikhomirov@330: 				p1TotalAppliedDelta += p1EntryDelta;
tikhomirov@330: 			} // while (p1 < p1Max)
tikhomirov@330: 			{
tikhomirov@330: 				// no more p1 entries, shall close p2 (if it's handled, code above jumps directly to L0)
tikhomirov@330: 				// regardless of whether insideP2 is .t
tikhomirov@330: 				int s = p2EntryStart;
tikhomirov@330: 				// p2EntryStartOffset != -1 when we started p2 entry processing, but not completed
tikhomirov@330: 				// if we handled last p1 entry but didn't start with p2 entry processing, it's -1 and regular p1 delta shall be used
tikhomirov@330: 				s -= p2EntryStartOffset == -1 ? p1TotalAppliedDelta : p2EntryStartOffset;
tikhomirov@330: 				r.add(s, p2EntryEnd - p1TotalAppliedDelta, p2Data);
tikhomirov@329: 			}
tikhomirov@329: 		}
tikhomirov@330: 		if (p1 < p1Max && insideP1entry) {
tikhomirov@330: 			r.add(p1EntryStart, p1EntryEnd, p1Data);
tikhomirov@330: 			p1++;
tikhomirov@330: 		}
tikhomirov@330: 		while (p1 < p1Max) {
tikhomirov@330: 			r.add(this, p1);
tikhomirov@330: 			p1++;
tikhomirov@330: 		};
tikhomirov@330: 		return r;
tikhomirov@329: 	}
tikhomirov@329: }