tikhomirov@21: /*
tikhomirov@74:  * Copyright (c) 2010-2011 TMate Software Ltd
tikhomirov@74:  *  
tikhomirov@74:  * This program is free software; you can redistribute it and/or modify
tikhomirov@74:  * it under the terms of the GNU General Public License as published by
tikhomirov@74:  * the Free Software Foundation; version 2 of the License.
tikhomirov@74:  *
tikhomirov@74:  * This program is distributed in the hope that it will be useful,
tikhomirov@74:  * but WITHOUT ANY WARRANTY; without even the implied warranty of
tikhomirov@74:  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
tikhomirov@74:  * GNU General Public License for more details.
tikhomirov@74:  *
tikhomirov@74:  * For information on how to redistribute this software under
tikhomirov@74:  * the terms of a license other than GNU General Public License
tikhomirov@102:  * contact TMate Software at support@hg4j.com
tikhomirov@0:  */
tikhomirov@74: package org.tmatesoft.hg.repo;
tikhomirov@0: 
tikhomirov@157: import java.io.IOException;
tikhomirov@129: import java.io.UnsupportedEncodingException;
tikhomirov@2: import java.util.ArrayList;
tikhomirov@3: import java.util.Arrays;
tikhomirov@146: import java.util.Calendar;
tikhomirov@317: import java.util.Collection;
tikhomirov@129: import java.util.Collections;
tikhomirov@129: import java.util.Date;
tikhomirov@129: import java.util.Formatter;
tikhomirov@129: import java.util.HashMap;
tikhomirov@2: import java.util.List;
tikhomirov@129: import java.util.Locale;
tikhomirov@129: import java.util.Map;
tikhomirov@146: import java.util.TimeZone;
tikhomirov@2: 
tikhomirov@196: import org.tmatesoft.hg.core.HgBadStateException;
tikhomirov@328: import org.tmatesoft.hg.core.HgLogCommand;
tikhomirov@74: import org.tmatesoft.hg.core.Nodeid;
tikhomirov@157: import org.tmatesoft.hg.internal.DataAccess;
tikhomirov@312: import org.tmatesoft.hg.internal.IterateControlMediator;
tikhomirov@312: import org.tmatesoft.hg.internal.Lifecycle;
tikhomirov@196: import org.tmatesoft.hg.internal.Pool;
tikhomirov@77: import org.tmatesoft.hg.internal.RevlogStream;
tikhomirov@312: import org.tmatesoft.hg.util.CancelSupport;
tikhomirov@317: import org.tmatesoft.hg.util.Pair;
tikhomirov@312: import org.tmatesoft.hg.util.ProgressSupport;
tikhomirov@74: 
tikhomirov@0: /**
tikhomirov@0:  * Representation of the Mercurial changelog file (list of ChangeSets)
tikhomirov@153:  * 
tikhomirov@74:  * @author Artem Tikhomirov
tikhomirov@74:  * @author TMate Software Ltd.
tikhomirov@0:  */
tikhomirov@97: public class HgChangelog extends Revlog {
tikhomirov@0: 
tikhomirov@153: 	/* package-local */HgChangelog(HgRepository hgRepo, RevlogStream content) {
tikhomirov@21: 		super(hgRepo, content);
tikhomirov@2: 	}
tikhomirov@2: 
tikhomirov@129: 	public void all(final HgChangelog.Inspector inspector) {
tikhomirov@137: 		range(0, getLastRevision(), inspector);
tikhomirov@48: 	}
tikhomirov@48: 
tikhomirov@129: 	public void range(int start, int end, final HgChangelog.Inspector inspector) {
tikhomirov@196: 		if (inspector == null) {
tikhomirov@196: 			throw new IllegalArgumentException();
tikhomirov@196: 		}
tikhomirov@196: 		content.iterate(start, end, true, new RawCsetParser(inspector));
tikhomirov@2: 	}
tikhomirov@2: 
tikhomirov@154: 	public List<RawChangeset> range(int start, int end) {
tikhomirov@196: 		final RawCsetCollector c = new RawCsetCollector(end - start + 1);
tikhomirov@196: 		range(start, end, c);
tikhomirov@196: 		return c.result;
tikhomirov@2: 	}
tikhomirov@3: 
tikhomirov@242: 	/**
tikhomirov@242: 	 * Access individual revisions. Note, regardless of supplied revision order, inspector gets
tikhomirov@242: 	 * changesets strictly in the order they are in the changelog.
tikhomirov@242: 	 * @param inspector callback to get changesets
tikhomirov@242: 	 * @param revisions revisions to read, unrestricted ordering.
tikhomirov@242: 	 */
tikhomirov@129: 	public void range(final HgChangelog.Inspector inspector, final int... revisions) {
tikhomirov@242: 		Arrays.sort(revisions);
tikhomirov@242: 		rangeInternal(inspector, revisions);
tikhomirov@242: 	}
tikhomirov@242: 
tikhomirov@242: 	/**
tikhomirov@242: 	 * Friends-only version of {@link #range(Inspector, int...)}, when callers know array is sorted
tikhomirov@242: 	 */
tikhomirov@242: 	/*package-local*/ void rangeInternal(HgChangelog.Inspector inspector, int[] sortedRevisions) {
tikhomirov@242: 		if (sortedRevisions == null || sortedRevisions.length == 0) {
tikhomirov@3: 			return;
tikhomirov@3: 		}
tikhomirov@242: 		if (inspector == null) {
tikhomirov@242: 			throw new IllegalArgumentException();
tikhomirov@242: 		}
tikhomirov@242: 		content.iterate(sortedRevisions, true, new RawCsetParser(inspector));
tikhomirov@3: 	}
tikhomirov@236: 	
tikhomirov@236: 	public RawChangeset changeset(Nodeid nid) {
tikhomirov@236: 		int x = getLocalRevision(nid);
tikhomirov@236: 		return range(x, x).get(0);
tikhomirov@236: 	}
tikhomirov@129: 
tikhomirov@129: 	public interface Inspector {
tikhomirov@129: 		// TODO describe whether cset is new instance each time
tikhomirov@182: 		// describe what revisionNumber is when Inspector is used with HgBundle (BAD_REVISION or bundle's local order?) 
tikhomirov@154: 		void next(int revisionNumber, Nodeid nodeid, RawChangeset cset);
tikhomirov@129: 	}
tikhomirov@129: 
tikhomirov@129: 	/**
tikhomirov@317: 	 * Unlike regular {@link Inspector}, this one supplies changeset revision along with its parents and children according
tikhomirov@317: 	 * to parent information of the revlog this inspector visits.
tikhomirov@317: 	 * @see HgDataFile#history(TreeInspector)
tikhomirov@328: 	 * @deprecated use {@link HgChangesetTreeHandler} and HgLogCommand#execute(HgChangesetTreeHandler)}
tikhomirov@317: 	 */
tikhomirov@328: 	@Deprecated
tikhomirov@317: 	public interface TreeInspector {
tikhomirov@317: 		// the reason TreeInsector is in HgChangelog, not in Revlog, because despite the fact it can
tikhomirov@317: 		// be applied to any revlog, it's not meant to provide revisions of any revlog it's beeing applied to, 
tikhomirov@317: 		// but changeset revisions always.
tikhomirov@317: 		// TODO HgChangelog.walk(TreeInspector)
tikhomirov@317: 		void next(Nodeid changesetRevision, Pair<Nodeid, Nodeid> parentChangesets, Collection<Nodeid> childChangesets);
tikhomirov@317: 	}
tikhomirov@317: 
tikhomirov@317: 	/**
tikhomirov@129: 	 * Entry in the Changelog
tikhomirov@129: 	 */
tikhomirov@154: 	public static class RawChangeset implements Cloneable /* for those that would like to keep a copy */{
tikhomirov@129: 		// TODO immutable
tikhomirov@129: 		private/* final */Nodeid manifest;
tikhomirov@153: 		private String user;
tikhomirov@153: 		private String comment;
tikhomirov@153: 		private List<String> files; // unmodifiable collection (otherwise #files() and implicit #clone() shall be revised)
tikhomirov@153: 		private Date time;
tikhomirov@153: 		private int timezone;
tikhomirov@161: 		// http://mercurial.selenic.com/wiki/PruningDeadBranches - Closing changesets can be identified by close=1 in the changeset's extra field.
tikhomirov@153: 		private Map<String, String> extras;
tikhomirov@129: 
tikhomirov@153: 		/**
tikhomirov@153: 		 * @see mercurial/changelog.py:read()
tikhomirov@153: 		 * 
tikhomirov@153: 		 *      <pre>
tikhomirov@153: 		 *         format used:
tikhomirov@153: 		 *         nodeid\n        : manifest node in ascii
tikhomirov@153: 		 *         user\n          : user, no \n or \r allowed
tikhomirov@153: 		 *         time tz extra\n : date (time is int or float, timezone is int)
tikhomirov@153: 		 *                         : extra is metadatas, encoded and separated by '\0'
tikhomirov@153: 		 *                         : older versions ignore it
tikhomirov@153: 		 *         files\n\n       : files modified by the cset, no \n or \r allowed
tikhomirov@153: 		 *         (.*)            : comment (free text, ideally utf-8)
tikhomirov@153: 		 * 
tikhomirov@153: 		 *         changelog v0 doesn't use extra
tikhomirov@153: 		 * </pre>
tikhomirov@153: 		 */
tikhomirov@154: 		private RawChangeset() {
tikhomirov@153: 		}
tikhomirov@129: 
tikhomirov@153: 		public Nodeid manifest() {
tikhomirov@153: 			return manifest;
tikhomirov@153: 		}
tikhomirov@129: 
tikhomirov@153: 		public String user() {
tikhomirov@153: 			return user;
tikhomirov@153: 		}
tikhomirov@153: 
tikhomirov@153: 		public String comment() {
tikhomirov@153: 			return comment;
tikhomirov@153: 		}
tikhomirov@153: 
tikhomirov@153: 		public List<String> files() {
tikhomirov@153: 			return files;
tikhomirov@153: 		}
tikhomirov@153: 
tikhomirov@153: 		public Date date() {
tikhomirov@153: 			return time;
tikhomirov@153: 		}
tikhomirov@211: 		
tikhomirov@211: 		/**
tikhomirov@211: 		 * @return time zone value, as is, positive for Western Hemisphere.
tikhomirov@211: 		 */
tikhomirov@211: 		public int timezone() {
tikhomirov@211: 			return timezone;
tikhomirov@211: 		}
tikhomirov@153: 
tikhomirov@153: 		public String dateString() {
tikhomirov@153: 			// XXX keep once formatted? Perhaps, there's faster way to set up calendar/time zone?
tikhomirov@153: 			StringBuilder sb = new StringBuilder(30);
tikhomirov@153: 			Formatter f = new Formatter(sb, Locale.US);
tikhomirov@211: 			TimeZone tz = TimeZone.getTimeZone(TimeZone.getAvailableIDs(timezone * 1000)[0]);
tikhomirov@153: 			// apparently timezone field records number of seconds time differs from UTC,
tikhomirov@153: 			// i.e. value to substract from time to get UTC time. Calendar seems to add
tikhomirov@153: 			// timezone offset to UTC, instead, hence sign change.
tikhomirov@211: //			tz.setRawOffset(timezone * -1000);
tikhomirov@153: 			Calendar c = Calendar.getInstance(tz, Locale.US);
tikhomirov@153: 			c.setTime(time);
tikhomirov@153: 			f.format("%ta %<tb %<td %<tH:%<tM:%<tS %<tY %<tz", c);
tikhomirov@153: 			return sb.toString();
tikhomirov@153: 		}
tikhomirov@153: 
tikhomirov@153: 		public Map<String, String> extras() {
tikhomirov@153: 			return extras;
tikhomirov@153: 		}
tikhomirov@153: 
tikhomirov@153: 		public String branch() {
tikhomirov@153: 			return extras.get("branch");
tikhomirov@153: 		}
tikhomirov@153: 
tikhomirov@153: 		@Override
tikhomirov@153: 		public String toString() {
tikhomirov@153: 			StringBuilder sb = new StringBuilder();
tikhomirov@153: 			sb.append("Changeset {");
tikhomirov@153: 			sb.append("User: ").append(user).append(", ");
tikhomirov@153: 			sb.append("Comment: ").append(comment).append(", ");
tikhomirov@153: 			sb.append("Manifest: ").append(manifest).append(", ");
tikhomirov@153: 			sb.append("Date: ").append(time).append(", ");
tikhomirov@153: 			sb.append("Files: ").append(files.size());
tikhomirov@153: 			for (String s : files) {
tikhomirov@153: 				sb.append(", ").append(s);
tikhomirov@153: 			}
tikhomirov@153: 			if (extras != null) {
tikhomirov@153: 				sb.append(", Extra: ").append(extras);
tikhomirov@153: 			}
tikhomirov@153: 			sb.append("}");
tikhomirov@153: 			return sb.toString();
tikhomirov@153: 		}
tikhomirov@153: 
tikhomirov@153: 		@Override
tikhomirov@154: 		public RawChangeset clone() {
tikhomirov@153: 			try {
tikhomirov@154: 				return (RawChangeset) super.clone();
tikhomirov@153: 			} catch (CloneNotSupportedException ex) {
tikhomirov@153: 				throw new InternalError(ex.toString());
tikhomirov@153: 			}
tikhomirov@153: 		}
tikhomirov@153: 
tikhomirov@157: 		public static RawChangeset parse(DataAccess da) {
tikhomirov@157: 			try {
tikhomirov@157: 				byte[] data = da.byteArray();
tikhomirov@157: 				RawChangeset rv = new RawChangeset();
tikhomirov@196: 				rv.init(data, 0, data.length, null);
tikhomirov@157: 				return rv;
tikhomirov@157: 			} catch (IOException ex) {
tikhomirov@196: 				throw new HgBadStateException(ex); // FIXME "Error reading changeset data"
tikhomirov@157: 			}
tikhomirov@153: 		}
tikhomirov@153: 
tikhomirov@196: 		// @param usersPool - it's likely user names get repeated again and again throughout repository. can be null
tikhomirov@196: 		/* package-local */void init(byte[] data, int offset, int length, Pool<String> usersPool) {
tikhomirov@153: 			final int bufferEndIndex = offset + length;
tikhomirov@153: 			final byte lineBreak = (byte) '\n';
tikhomirov@153: 			int breakIndex1 = indexOf(data, lineBreak, offset, bufferEndIndex);
tikhomirov@153: 			if (breakIndex1 == -1) {
tikhomirov@153: 				throw new IllegalArgumentException("Bad Changeset data");
tikhomirov@153: 			}
tikhomirov@153: 			Nodeid _nodeid = Nodeid.fromAscii(data, 0, breakIndex1);
tikhomirov@153: 			int breakIndex2 = indexOf(data, lineBreak, breakIndex1 + 1, bufferEndIndex);
tikhomirov@153: 			if (breakIndex2 == -1) {
tikhomirov@153: 				throw new IllegalArgumentException("Bad Changeset data");
tikhomirov@153: 			}
tikhomirov@153: 			String _user = new String(data, breakIndex1 + 1, breakIndex2 - breakIndex1 - 1);
tikhomirov@196: 			if (usersPool != null) {
tikhomirov@196: 				_user = usersPool.unify(_user);
tikhomirov@196: 			}
tikhomirov@153: 			int breakIndex3 = indexOf(data, lineBreak, breakIndex2 + 1, bufferEndIndex);
tikhomirov@153: 			if (breakIndex3 == -1) {
tikhomirov@153: 				throw new IllegalArgumentException("Bad Changeset data");
tikhomirov@153: 			}
tikhomirov@153: 			String _timeString = new String(data, breakIndex2 + 1, breakIndex3 - breakIndex2 - 1);
tikhomirov@153: 			int space1 = _timeString.indexOf(' ');
tikhomirov@153: 			if (space1 == -1) {
tikhomirov@153: 				throw new IllegalArgumentException("Bad Changeset data");
tikhomirov@153: 			}
tikhomirov@153: 			int space2 = _timeString.indexOf(' ', space1 + 1);
tikhomirov@153: 			if (space2 == -1) {
tikhomirov@153: 				space2 = _timeString.length();
tikhomirov@153: 			}
tikhomirov@153: 			long unixTime = Long.parseLong(_timeString.substring(0, space1)); // XXX Float, perhaps
tikhomirov@153: 			int _timezone = Integer.parseInt(_timeString.substring(space1 + 1, space2));
tikhomirov@153: 			// XXX not sure need to add timezone here - I can't figure out whether Hg keeps GMT time, and records timezone just for info, or unixTime is taken local
tikhomirov@153: 			// on commit and timezone is recorded to adjust it to UTC.
tikhomirov@153: 			Date _time = new Date(unixTime * 1000);
tikhomirov@153: 			String _extras = space2 < _timeString.length() ? _timeString.substring(space2 + 1) : null;
tikhomirov@153: 			Map<String, String> _extrasMap;
tikhomirov@252: 			final String extras_branch_key = "branch";
tikhomirov@153: 			if (_extras == null) {
tikhomirov@252: 				_extrasMap = Collections.singletonMap(extras_branch_key, HgRepository.DEFAULT_BRANCH_NAME);
tikhomirov@153: 			} else {
tikhomirov@153: 				_extrasMap = new HashMap<String, String>();
tikhomirov@153: 				for (String pair : _extras.split("\00")) {
tikhomirov@153: 					int eq = pair.indexOf(':');
tikhomirov@153: 					// FIXME need to decode key/value, @see changelog.py:decodeextra
tikhomirov@153: 					_extrasMap.put(pair.substring(0, eq), pair.substring(eq + 1));
tikhomirov@153: 				}
tikhomirov@252: 				if (!_extrasMap.containsKey(extras_branch_key)) {
tikhomirov@252: 					_extrasMap.put(extras_branch_key, HgRepository.DEFAULT_BRANCH_NAME);
tikhomirov@153: 				}
tikhomirov@153: 				_extrasMap = Collections.unmodifiableMap(_extrasMap);
tikhomirov@153: 			}
tikhomirov@153: 
tikhomirov@153: 			//
tikhomirov@153: 			int lastStart = breakIndex3 + 1;
tikhomirov@153: 			int breakIndex4 = indexOf(data, lineBreak, lastStart, bufferEndIndex);
tikhomirov@153: 			ArrayList<String> _files = null;
tikhomirov@153: 			if (breakIndex4 > lastStart) {
tikhomirov@153: 				// if breakIndex4 == lastStart, we already found \n\n and hence there are no files (e.g. merge revision)
tikhomirov@153: 				_files = new ArrayList<String>(5);
tikhomirov@153: 				while (breakIndex4 != -1 && breakIndex4 + 1 < bufferEndIndex) {
tikhomirov@153: 					_files.add(new String(data, lastStart, breakIndex4 - lastStart));
tikhomirov@153: 					lastStart = breakIndex4 + 1;
tikhomirov@153: 					if (data[breakIndex4 + 1] == lineBreak) {
tikhomirov@153: 						// found \n\n
tikhomirov@153: 						break;
tikhomirov@153: 					} else {
tikhomirov@153: 						breakIndex4 = indexOf(data, lineBreak, lastStart, bufferEndIndex);
tikhomirov@129: 					}
tikhomirov@153: 				}
tikhomirov@153: 				if (breakIndex4 == -1 || breakIndex4 >= bufferEndIndex) {
tikhomirov@153: 					throw new IllegalArgumentException("Bad Changeset data");
tikhomirov@153: 				}
tikhomirov@153: 			} else {
tikhomirov@153: 				breakIndex4--;
tikhomirov@153: 			}
tikhomirov@153: 			String _comment;
tikhomirov@153: 			try {
tikhomirov@153: 				_comment = new String(data, breakIndex4 + 2, bufferEndIndex - breakIndex4 - 2, "UTF-8");
tikhomirov@153: 				// FIXME respect ui.fallbackencoding and try to decode if set
tikhomirov@153: 			} catch (UnsupportedEncodingException ex) {
tikhomirov@153: 				_comment = "";
tikhomirov@153: 				throw new IllegalStateException("Could hardly happen");
tikhomirov@153: 			}
tikhomirov@153: 			// change this instance at once, don't leave it partially changes in case of error
tikhomirov@153: 			this.manifest = _nodeid;
tikhomirov@153: 			this.user = _user;
tikhomirov@153: 			this.time = _time;
tikhomirov@153: 			this.timezone = _timezone;
tikhomirov@153: 			this.files = _files == null ? Collections.<String> emptyList() : Collections.unmodifiableList(_files);
tikhomirov@153: 			this.comment = _comment;
tikhomirov@153: 			this.extras = _extrasMap;
tikhomirov@153: 		}
tikhomirov@129: 
tikhomirov@153: 		private static int indexOf(byte[] src, byte what, int startOffset, int endIndex) {
tikhomirov@153: 			for (int i = startOffset; i < endIndex; i++) {
tikhomirov@153: 				if (src[i] == what) {
tikhomirov@153: 					return i;
tikhomirov@153: 				}
tikhomirov@153: 			}
tikhomirov@153: 			return -1;
tikhomirov@153: 		}
tikhomirov@129: 	}
tikhomirov@129: 
tikhomirov@196: 	private static class RawCsetCollector implements Inspector {
tikhomirov@196: 		final ArrayList<RawChangeset> result;
tikhomirov@196: 		
tikhomirov@196: 		public RawCsetCollector(int count) {
tikhomirov@196: 			result = new ArrayList<RawChangeset>(count > 0 ? count : 5);
tikhomirov@196: 		}
tikhomirov@196: 
tikhomirov@196: 		public void next(int revisionNumber, Nodeid nodeid, RawChangeset cset) {
tikhomirov@196: 			result.add(cset.clone());
tikhomirov@196: 		}
tikhomirov@196: 	}
tikhomirov@196: 
tikhomirov@312: 	private static class RawCsetParser implements RevlogStream.Inspector, Lifecycle {
tikhomirov@196: 		
tikhomirov@196: 		private final Inspector inspector;
tikhomirov@196: 		private final Pool<String> usersPool;
tikhomirov@196: 		private final RawChangeset cset = new RawChangeset();
tikhomirov@312: 		private final ProgressSupport progressHelper;
tikhomirov@312: 		private IterateControlMediator iterateControl;
tikhomirov@196: 
tikhomirov@196: 		public RawCsetParser(HgChangelog.Inspector delegate) {
tikhomirov@196: 			assert delegate != null;
tikhomirov@196: 			inspector = delegate;
tikhomirov@196: 			usersPool = new Pool<String>();
tikhomirov@312: 			progressHelper = ProgressSupport.Factory.get(delegate);
tikhomirov@196: 		}
tikhomirov@196: 
tikhomirov@196: 		public void next(int revisionNumber, int actualLen, int baseRevision, int linkRevision, int parent1Revision, int parent2Revision, byte[] nodeid, DataAccess da) {
tikhomirov@196: 			try {
tikhomirov@196: 				byte[] data = da.byteArray();
tikhomirov@196: 				cset.init(data, 0, data.length, usersPool);
tikhomirov@196: 				// XXX there's no guarantee for Changeset.Callback that distinct instance comes each time, consider instance reuse
tikhomirov@196: 				inspector.next(revisionNumber, Nodeid.fromBinary(nodeid, 0), cset);
tikhomirov@312: 				progressHelper.worked(1);
tikhomirov@196: 			} catch (Exception ex) {
tikhomirov@196: 				throw new HgBadStateException(ex); // FIXME exception handling
tikhomirov@196: 			}
tikhomirov@312: 			if (iterateControl != null) {
tikhomirov@312: 				iterateControl.checkCancelled();
tikhomirov@312: 			}
tikhomirov@312: 		}
tikhomirov@312: 
tikhomirov@312: 		public void start(int count, Callback callback, Object token) {
tikhomirov@312: 			CancelSupport cs = CancelSupport.Factory.get(inspector, null);
tikhomirov@312: 			iterateControl = cs == null ? null : new IterateControlMediator(cs, callback);
tikhomirov@312: 			progressHelper.start(count);
tikhomirov@312: 		}
tikhomirov@312: 
tikhomirov@312: 		public void finish(Object token) {
tikhomirov@312: 			progressHelper.done();
tikhomirov@196: 		}
tikhomirov@196: 	}
tikhomirov@0: }