kitaev@213: /*
kitaev@213:  * Copyright (c) 2010-2011 TMate Software Ltd
kitaev@213:  *  
kitaev@213:  * This program is free software; you can redistribute it and/or modify
kitaev@213:  * it under the terms of the GNU General Public License as published by
kitaev@213:  * the Free Software Foundation; version 2 of the License.
kitaev@213:  *
kitaev@213:  * This program is distributed in the hope that it will be useful,
kitaev@213:  * but WITHOUT ANY WARRANTY; without even the implied warranty of
kitaev@213:  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
kitaev@213:  * GNU General Public License for more details.
kitaev@213:  *
kitaev@213:  * For information on how to redistribute this software under
kitaev@213:  * the terms of a license other than GNU General Public License
kitaev@213:  * contact TMate Software at support@hg4j.com
kitaev@213:  */
kitaev@213: package org.tmatesoft.hg.repo;
kitaev@213: 
kitaev@213: import java.io.IOException;
kitaev@213: import java.io.UnsupportedEncodingException;
kitaev@213: import java.util.ArrayList;
kitaev@213: import java.util.Arrays;
kitaev@213: import java.util.Calendar;
kitaev@213: import java.util.Collections;
kitaev@213: import java.util.Date;
kitaev@213: import java.util.Formatter;
kitaev@213: import java.util.HashMap;
kitaev@213: import java.util.List;
kitaev@213: import java.util.Locale;
kitaev@213: import java.util.Map;
kitaev@213: import java.util.TimeZone;
kitaev@213: 
kitaev@213: import org.tmatesoft.hg.core.HgBadStateException;
kitaev@213: import org.tmatesoft.hg.core.Nodeid;
kitaev@213: import org.tmatesoft.hg.internal.DataAccess;
kitaev@213: import org.tmatesoft.hg.internal.Pool;
kitaev@213: import org.tmatesoft.hg.internal.RevlogStream;
kitaev@213: 
kitaev@213: /**
kitaev@213:  * Representation of the Mercurial changelog file (list of ChangeSets)
kitaev@213:  * 
kitaev@213:  * @author Artem Tikhomirov
kitaev@213:  * @author TMate Software Ltd.
kitaev@213:  */
kitaev@213: public class HgChangelog extends Revlog {
kitaev@213: 
kitaev@213: 	/* package-local */HgChangelog(HgRepository hgRepo, RevlogStream content) {
kitaev@213: 		super(hgRepo, content);
kitaev@213: 	}
kitaev@213: 
kitaev@213: 	public void all(final HgChangelog.Inspector inspector) {
kitaev@213: 		range(0, getLastRevision(), inspector);
kitaev@213: 	}
kitaev@213: 
kitaev@213: 	public void range(int start, int end, final HgChangelog.Inspector inspector) {
kitaev@213: 		if (inspector == null) {
kitaev@213: 			throw new IllegalArgumentException();
kitaev@213: 		}
kitaev@213: 		content.iterate(start, end, true, new RawCsetParser(inspector));
kitaev@213: 	}
kitaev@213: 
kitaev@213: 	public List<RawChangeset> range(int start, int end) {
kitaev@213: 		final RawCsetCollector c = new RawCsetCollector(end - start + 1);
kitaev@213: 		range(start, end, c);
kitaev@213: 		return c.result;
kitaev@213: 	}
kitaev@213: 
kitaev@213: 	public void range(final HgChangelog.Inspector inspector, final int... revisions) {
kitaev@213: 		if (revisions == null || revisions.length == 0) {
kitaev@213: 			return;
kitaev@213: 		}
kitaev@213: 		RevlogStream.Inspector i = new RevlogStream.Inspector() {
kitaev@213: 			private final RawCsetParser delegate = new RawCsetParser(inspector);
kitaev@213: 
kitaev@213: 			public void next(int revisionNumber, int actualLen, int baseRevision, int linkRevision, int parent1Revision, int parent2Revision, byte[] nodeid, DataAccess da) {
kitaev@213: 				if (Arrays.binarySearch(revisions, revisionNumber) >= 0) {
kitaev@213: 					delegate.next(revisionNumber, actualLen, baseRevision, linkRevision, parent1Revision, parent2Revision, nodeid, da);
kitaev@213: 				}
kitaev@213: 			}
kitaev@213: 		};
kitaev@213: 		Arrays.sort(revisions);
kitaev@213: 		content.iterate(revisions[0], revisions[revisions.length - 1], true, i);
kitaev@213: 	}
kitaev@213: 
kitaev@213: 	public interface Inspector {
kitaev@213: 		// TODO describe whether cset is new instance each time
kitaev@213: 		// describe what revisionNumber is when Inspector is used with HgBundle (BAD_REVISION or bundle's local order?) 
kitaev@213: 		void next(int revisionNumber, Nodeid nodeid, RawChangeset cset);
kitaev@213: 	}
kitaev@213: 
kitaev@213: 	/**
kitaev@213: 	 * Entry in the Changelog
kitaev@213: 	 */
kitaev@213: 	public static class RawChangeset implements Cloneable /* for those that would like to keep a copy */{
kitaev@213: 		// TODO immutable
kitaev@213: 		private/* final */Nodeid manifest;
kitaev@213: 		private String user;
kitaev@213: 		private String comment;
kitaev@213: 		private List<String> files; // unmodifiable collection (otherwise #files() and implicit #clone() shall be revised)
kitaev@213: 		private Date time;
kitaev@213: 		private int timezone;
kitaev@213: 		// http://mercurial.selenic.com/wiki/PruningDeadBranches - Closing changesets can be identified by close=1 in the changeset's extra field.
kitaev@213: 		private Map<String, String> extras;
kitaev@213: 
kitaev@213: 		/**
kitaev@213: 		 * @see mercurial/changelog.py:read()
kitaev@213: 		 * 
kitaev@213: 		 *      <pre>
kitaev@213: 		 *         format used:
kitaev@213: 		 *         nodeid\n        : manifest node in ascii
kitaev@213: 		 *         user\n          : user, no \n or \r allowed
kitaev@213: 		 *         time tz extra\n : date (time is int or float, timezone is int)
kitaev@213: 		 *                         : extra is metadatas, encoded and separated by '\0'
kitaev@213: 		 *                         : older versions ignore it
kitaev@213: 		 *         files\n\n       : files modified by the cset, no \n or \r allowed
kitaev@213: 		 *         (.*)            : comment (free text, ideally utf-8)
kitaev@213: 		 * 
kitaev@213: 		 *         changelog v0 doesn't use extra
kitaev@213: 		 * </pre>
kitaev@213: 		 */
kitaev@213: 		private RawChangeset() {
kitaev@213: 		}
kitaev@213: 
kitaev@213: 		public Nodeid manifest() {
kitaev@213: 			return manifest;
kitaev@213: 		}
kitaev@213: 
kitaev@213: 		public String user() {
kitaev@213: 			return user;
kitaev@213: 		}
kitaev@213: 
kitaev@213: 		public String comment() {
kitaev@213: 			return comment;
kitaev@213: 		}
kitaev@213: 
kitaev@213: 		public List<String> files() {
kitaev@213: 			return files;
kitaev@213: 		}
kitaev@213: 
kitaev@213: 		public Date date() {
kitaev@213: 			return time;
kitaev@213: 		}
kitaev@213: 		
kitaev@213: 		/**
kitaev@213: 		 * @return time zone value, as is, positive for Western Hemisphere.
kitaev@213: 		 */
kitaev@213: 		public int timezone() {
kitaev@213: 			return timezone;
kitaev@213: 		}
kitaev@213: 
kitaev@213: 		public String dateString() {
kitaev@213: 			// XXX keep once formatted? Perhaps, there's faster way to set up calendar/time zone?
kitaev@213: 			StringBuilder sb = new StringBuilder(30);
kitaev@213: 			Formatter f = new Formatter(sb, Locale.US);
kitaev@213: 			TimeZone tz = TimeZone.getTimeZone(TimeZone.getAvailableIDs(timezone * 1000)[0]);
kitaev@213: 			// apparently timezone field records number of seconds time differs from UTC,
kitaev@213: 			// i.e. value to substract from time to get UTC time. Calendar seems to add
kitaev@213: 			// timezone offset to UTC, instead, hence sign change.
kitaev@213: //			tz.setRawOffset(timezone * -1000);
kitaev@213: 			Calendar c = Calendar.getInstance(tz, Locale.US);
kitaev@213: 			c.setTime(time);
kitaev@213: 			f.format("%ta %<tb %<td %<tH:%<tM:%<tS %<tY %<tz", c);
kitaev@213: 			return sb.toString();
kitaev@213: 		}
kitaev@213: 
kitaev@213: 		public Map<String, String> extras() {
kitaev@213: 			return extras;
kitaev@213: 		}
kitaev@213: 
kitaev@213: 		public String branch() {
kitaev@213: 			return extras.get("branch");
kitaev@213: 		}
kitaev@213: 
kitaev@213: 		@Override
kitaev@213: 		public String toString() {
kitaev@213: 			StringBuilder sb = new StringBuilder();
kitaev@213: 			sb.append("Changeset {");
kitaev@213: 			sb.append("User: ").append(user).append(", ");
kitaev@213: 			sb.append("Comment: ").append(comment).append(", ");
kitaev@213: 			sb.append("Manifest: ").append(manifest).append(", ");
kitaev@213: 			sb.append("Date: ").append(time).append(", ");
kitaev@213: 			sb.append("Files: ").append(files.size());
kitaev@213: 			for (String s : files) {
kitaev@213: 				sb.append(", ").append(s);
kitaev@213: 			}
kitaev@213: 			if (extras != null) {
kitaev@213: 				sb.append(", Extra: ").append(extras);
kitaev@213: 			}
kitaev@213: 			sb.append("}");
kitaev@213: 			return sb.toString();
kitaev@213: 		}
kitaev@213: 
kitaev@213: 		@Override
kitaev@213: 		public RawChangeset clone() {
kitaev@213: 			try {
kitaev@213: 				return (RawChangeset) super.clone();
kitaev@213: 			} catch (CloneNotSupportedException ex) {
kitaev@213: 				throw new InternalError(ex.toString());
kitaev@213: 			}
kitaev@213: 		}
kitaev@213: 
kitaev@213: 		public static RawChangeset parse(DataAccess da) {
kitaev@213: 			try {
kitaev@213: 				byte[] data = da.byteArray();
kitaev@213: 				RawChangeset rv = new RawChangeset();
kitaev@213: 				rv.init(data, 0, data.length, null);
kitaev@213: 				return rv;
kitaev@213: 			} catch (IOException ex) {
kitaev@213: 				throw new HgBadStateException(ex); // FIXME "Error reading changeset data"
kitaev@213: 			}
kitaev@213: 		}
kitaev@213: 
kitaev@213: 		// @param usersPool - it's likely user names get repeated again and again throughout repository. can be null
kitaev@213: 		/* package-local */void init(byte[] data, int offset, int length, Pool<String> usersPool) {
kitaev@213: 			final int bufferEndIndex = offset + length;
kitaev@213: 			final byte lineBreak = (byte) '\n';
kitaev@213: 			int breakIndex1 = indexOf(data, lineBreak, offset, bufferEndIndex);
kitaev@213: 			if (breakIndex1 == -1) {
kitaev@213: 				throw new IllegalArgumentException("Bad Changeset data");
kitaev@213: 			}
kitaev@213: 			Nodeid _nodeid = Nodeid.fromAscii(data, 0, breakIndex1);
kitaev@213: 			int breakIndex2 = indexOf(data, lineBreak, breakIndex1 + 1, bufferEndIndex);
kitaev@213: 			if (breakIndex2 == -1) {
kitaev@213: 				throw new IllegalArgumentException("Bad Changeset data");
kitaev@213: 			}
kitaev@213: 			String _user = new String(data, breakIndex1 + 1, breakIndex2 - breakIndex1 - 1);
kitaev@213: 			if (usersPool != null) {
kitaev@213: 				_user = usersPool.unify(_user);
kitaev@213: 			}
kitaev@213: 			int breakIndex3 = indexOf(data, lineBreak, breakIndex2 + 1, bufferEndIndex);
kitaev@213: 			if (breakIndex3 == -1) {
kitaev@213: 				throw new IllegalArgumentException("Bad Changeset data");
kitaev@213: 			}
kitaev@213: 			String _timeString = new String(data, breakIndex2 + 1, breakIndex3 - breakIndex2 - 1);
kitaev@213: 			int space1 = _timeString.indexOf(' ');
kitaev@213: 			if (space1 == -1) {
kitaev@213: 				throw new IllegalArgumentException("Bad Changeset data");
kitaev@213: 			}
kitaev@213: 			int space2 = _timeString.indexOf(' ', space1 + 1);
kitaev@213: 			if (space2 == -1) {
kitaev@213: 				space2 = _timeString.length();
kitaev@213: 			}
kitaev@213: 			long unixTime = Long.parseLong(_timeString.substring(0, space1)); // XXX Float, perhaps
kitaev@213: 			int _timezone = Integer.parseInt(_timeString.substring(space1 + 1, space2));
kitaev@213: 			// 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
kitaev@213: 			// on commit and timezone is recorded to adjust it to UTC.
kitaev@213: 			Date _time = new Date(unixTime * 1000);
kitaev@213: 			String _extras = space2 < _timeString.length() ? _timeString.substring(space2 + 1) : null;
kitaev@213: 			Map<String, String> _extrasMap;
kitaev@213: 			if (_extras == null) {
kitaev@213: 				_extrasMap = Collections.singletonMap("branch", "default");
kitaev@213: 			} else {
kitaev@213: 				_extrasMap = new HashMap<String, String>();
kitaev@213: 				for (String pair : _extras.split("\00")) {
kitaev@213: 					int eq = pair.indexOf(':');
kitaev@213: 					// FIXME need to decode key/value, @see changelog.py:decodeextra
kitaev@213: 					_extrasMap.put(pair.substring(0, eq), pair.substring(eq + 1));
kitaev@213: 				}
kitaev@213: 				if (!_extrasMap.containsKey("branch")) {
kitaev@213: 					_extrasMap.put("branch", "default");
kitaev@213: 				}
kitaev@213: 				_extrasMap = Collections.unmodifiableMap(_extrasMap);
kitaev@213: 			}
kitaev@213: 
kitaev@213: 			//
kitaev@213: 			int lastStart = breakIndex3 + 1;
kitaev@213: 			int breakIndex4 = indexOf(data, lineBreak, lastStart, bufferEndIndex);
kitaev@213: 			ArrayList<String> _files = null;
kitaev@213: 			if (breakIndex4 > lastStart) {
kitaev@213: 				// if breakIndex4 == lastStart, we already found \n\n and hence there are no files (e.g. merge revision)
kitaev@213: 				_files = new ArrayList<String>(5);
kitaev@213: 				while (breakIndex4 != -1 && breakIndex4 + 1 < bufferEndIndex) {
kitaev@213: 					_files.add(new String(data, lastStart, breakIndex4 - lastStart));
kitaev@213: 					lastStart = breakIndex4 + 1;
kitaev@213: 					if (data[breakIndex4 + 1] == lineBreak) {
kitaev@213: 						// found \n\n
kitaev@213: 						break;
kitaev@213: 					} else {
kitaev@213: 						breakIndex4 = indexOf(data, lineBreak, lastStart, bufferEndIndex);
kitaev@213: 					}
kitaev@213: 				}
kitaev@213: 				if (breakIndex4 == -1 || breakIndex4 >= bufferEndIndex) {
kitaev@213: 					throw new IllegalArgumentException("Bad Changeset data");
kitaev@213: 				}
kitaev@213: 			} else {
kitaev@213: 				breakIndex4--;
kitaev@213: 			}
kitaev@213: 			String _comment;
kitaev@213: 			try {
kitaev@213: 				_comment = new String(data, breakIndex4 + 2, bufferEndIndex - breakIndex4 - 2, "UTF-8");
kitaev@213: 				// FIXME respect ui.fallbackencoding and try to decode if set
kitaev@213: 			} catch (UnsupportedEncodingException ex) {
kitaev@213: 				_comment = "";
kitaev@213: 				throw new IllegalStateException("Could hardly happen");
kitaev@213: 			}
kitaev@213: 			// change this instance at once, don't leave it partially changes in case of error
kitaev@213: 			this.manifest = _nodeid;
kitaev@213: 			this.user = _user;
kitaev@213: 			this.time = _time;
kitaev@213: 			this.timezone = _timezone;
kitaev@213: 			this.files = _files == null ? Collections.<String> emptyList() : Collections.unmodifiableList(_files);
kitaev@213: 			this.comment = _comment;
kitaev@213: 			this.extras = _extrasMap;
kitaev@213: 		}
kitaev@213: 
kitaev@213: 		private static int indexOf(byte[] src, byte what, int startOffset, int endIndex) {
kitaev@213: 			for (int i = startOffset; i < endIndex; i++) {
kitaev@213: 				if (src[i] == what) {
kitaev@213: 					return i;
kitaev@213: 				}
kitaev@213: 			}
kitaev@213: 			return -1;
kitaev@213: 		}
kitaev@213: 	}
kitaev@213: 
kitaev@213: 	private static class RawCsetCollector implements Inspector {
kitaev@213: 		final ArrayList<RawChangeset> result;
kitaev@213: 		
kitaev@213: 		public RawCsetCollector(int count) {
kitaev@213: 			result = new ArrayList<RawChangeset>(count > 0 ? count : 5);
kitaev@213: 		}
kitaev@213: 
kitaev@213: 		public void next(int revisionNumber, Nodeid nodeid, RawChangeset cset) {
kitaev@213: 			result.add(cset.clone());
kitaev@213: 		}
kitaev@213: 	}
kitaev@213: 
kitaev@213: 	private static class RawCsetParser implements RevlogStream.Inspector {
kitaev@213: 		
kitaev@213: 		private final Inspector inspector;
kitaev@213: 		private final Pool<String> usersPool;
kitaev@213: 		private final RawChangeset cset = new RawChangeset();
kitaev@213: 
kitaev@213: 		public RawCsetParser(HgChangelog.Inspector delegate) {
kitaev@213: 			assert delegate != null;
kitaev@213: 			inspector = delegate;
kitaev@213: 			usersPool = new Pool<String>();
kitaev@213: 		}
kitaev@213: 
kitaev@213: 		public void next(int revisionNumber, int actualLen, int baseRevision, int linkRevision, int parent1Revision, int parent2Revision, byte[] nodeid, DataAccess da) {
kitaev@213: 			try {
kitaev@213: 				byte[] data = da.byteArray();
kitaev@213: 				cset.init(data, 0, data.length, usersPool);
kitaev@213: 				// XXX there's no guarantee for Changeset.Callback that distinct instance comes each time, consider instance reuse
kitaev@213: 				inspector.next(revisionNumber, Nodeid.fromBinary(nodeid, 0), cset);
kitaev@213: 			} catch (Exception ex) {
kitaev@213: 				throw new HgBadStateException(ex); // FIXME exception handling
kitaev@213: 			}
kitaev@213: 		}
kitaev@213: 	}
kitaev@213: }