tikhomirov@17: /*
tikhomirov@388:  * Copyright (c) 2010-2012 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@2:  */
tikhomirov@74: package org.tmatesoft.hg.repo;
tikhomirov@2: 
tikhomirov@367: import static org.tmatesoft.hg.repo.HgInternals.wrongRevisionIndex;
tikhomirov@148: import static org.tmatesoft.hg.repo.HgRepository.*;
tikhomirov@74: 
tikhomirov@157: import java.io.ByteArrayOutputStream;
tikhomirov@237: import java.io.File;
tikhomirov@237: import java.io.FileInputStream;
tikhomirov@148: import java.io.IOException;
tikhomirov@115: import java.nio.ByteBuffer;
tikhomirov@237: import java.nio.channels.FileChannel;
tikhomirov@78: import java.util.ArrayList;
tikhomirov@240: import java.util.Arrays;
tikhomirov@78: import java.util.Collection;
tikhomirov@78: 
tikhomirov@157: import org.tmatesoft.hg.core.HgException;
tikhomirov@74: import org.tmatesoft.hg.core.Nodeid;
tikhomirov@157: import org.tmatesoft.hg.internal.DataAccess;
tikhomirov@121: import org.tmatesoft.hg.internal.FilterByteChannel;
tikhomirov@277: import org.tmatesoft.hg.internal.FilterDataAccess;
tikhomirov@276: import org.tmatesoft.hg.internal.IntMap;
tikhomirov@420: import org.tmatesoft.hg.internal.Internals;
tikhomirov@77: import org.tmatesoft.hg.internal.RevlogStream;
tikhomirov@115: import org.tmatesoft.hg.util.ByteChannel;
tikhomirov@237: import org.tmatesoft.hg.util.CancelSupport;
tikhomirov@148: import org.tmatesoft.hg.util.CancelledException;
tikhomirov@388: import org.tmatesoft.hg.util.LogFacility;
tikhomirov@305: import org.tmatesoft.hg.util.Pair;
tikhomirov@133: import org.tmatesoft.hg.util.Path;
tikhomirov@237: import org.tmatesoft.hg.util.ProgressSupport;
tikhomirov@74: 
tikhomirov@5: 
tikhomirov@17: 
tikhomirov@2: /**
tikhomirov@2:  * ? name:HgFileNode?
tikhomirov@74:  *
tikhomirov@74:  * @author Artem Tikhomirov
tikhomirov@74:  * @author TMate Software Ltd.
tikhomirov@2:  */
tikhomirov@2: public class HgDataFile extends Revlog {
tikhomirov@2: 
tikhomirov@3: 	// absolute from repo root?
tikhomirov@3: 	// slashes, unix-style?
tikhomirov@3: 	// repo location agnostic, just to give info to user, not to access real storage
tikhomirov@74: 	private final Path path;
tikhomirov@134: 	private Metadata metadata; // get initialized on first access to file content.
tikhomirov@2: 	
tikhomirov@115: 	/*package-local*/HgDataFile(HgRepository hgRepo, Path filePath, RevlogStream content) {
tikhomirov@21: 		super(hgRepo, content);
tikhomirov@115: 		path = filePath;
tikhomirov@3: 	}
tikhomirov@115: 
tikhomirov@115: 	/*package-local*/HgDataFile(HgRepository hgRepo, Path filePath) {
tikhomirov@115: 		super(hgRepo);
tikhomirov@115: 		path = filePath;
tikhomirov@115: 	}
tikhomirov@115: 
tikhomirov@115: 	// exists is not the best name possible. now it means no file with such name was ever known to the repo.
tikhomirov@115: 	// it might be confused with files existed before but lately removed. 
tikhomirov@3: 	public boolean exists() {
tikhomirov@3: 		return content != null; // XXX need better impl
tikhomirov@2: 	}
tikhomirov@2: 
tikhomirov@77: 	// human-readable (i.e. "COPYING", not "store/data/_c_o_p_y_i_n_g.i")
tikhomirov@74: 	public Path getPath() {
tikhomirov@157: 		return path; // hgRepo.backresolve(this) -> name? In this case, what about hashed long names?
tikhomirov@2: 	}
tikhomirov@2: 
tikhomirov@275: 	/**
tikhomirov@416: 	 * Handy shorthand for {@link #getLength(int) length(getRevisionIndex(nodeid))}
tikhomirov@354: 	 *
tikhomirov@354: 	 * @param nodeid revision of the file
tikhomirov@354: 	 * 
tikhomirov@275: 	 * @return size of the file content at the given revision
tikhomirov@396: 	 * @throws HgInvalidRevisionException if supplied nodeid doesn't identify any revision from this revlog (<em>runtime exception</em>)  
tikhomirov@354: 	 * @throws HgInvalidControlFileException if access to revlog index/data entry failed
tikhomirov@275: 	 */
tikhomirov@416: 	public int getLength(Nodeid nodeid) throws HgInvalidControlFileException, HgInvalidRevisionException {
tikhomirov@396: 		try {
tikhomirov@416: 			return getLength(getRevisionIndex(nodeid));
tikhomirov@396: 		} catch (HgInvalidControlFileException ex) {
tikhomirov@396: 			throw ex.isRevisionSet() ? ex : ex.setRevision(nodeid);
tikhomirov@396: 		} catch (HgInvalidRevisionException ex) {
tikhomirov@396: 			throw ex.isRevisionSet() ? ex : ex.setRevision(nodeid);
tikhomirov@396: 		}
tikhomirov@275: 	}
tikhomirov@275: 	
tikhomirov@275: 	/**
tikhomirov@368:  	 * @param fileRevisionIndex - revision local index, non-negative. From predefined constants, only {@link HgRepository#TIP} makes sense. 
tikhomirov@275: 	 * @return size of the file content at the revision identified by local revision number.
tikhomirov@396: 	 * @throws HgInvalidRevisionException if supplied argument doesn't represent revision index in this revlog (<em>runtime exception</em>)
tikhomirov@354: 	 * @throws HgInvalidControlFileException if access to revlog index/data entry failed
tikhomirov@275: 	 */
tikhomirov@416: 	public int getLength(int fileRevisionIndex) throws HgInvalidControlFileException, HgInvalidRevisionException {
tikhomirov@401: 		if (wrongRevisionIndex(fileRevisionIndex) || fileRevisionIndex == BAD_REVISION) {
tikhomirov@401: 			throw new HgInvalidRevisionException(fileRevisionIndex);
tikhomirov@401: 		}
tikhomirov@401: 		if (fileRevisionIndex == TIP) {
tikhomirov@401: 			fileRevisionIndex = getLastRevision();
tikhomirov@401: 		} else if (fileRevisionIndex == WORKING_COPY) {
tikhomirov@401: 			File f = getRepo().getFile(this);
tikhomirov@401: 			if (f.exists()) {
tikhomirov@420: 				// single revision can't be greater than 2^32, shall be safe to cast to int
tikhomirov@420: 				return Internals.ltoi(f.length());
tikhomirov@401: 			}
tikhomirov@401: 			Nodeid fileRev = getWorkingCopyRevision();
tikhomirov@401: 			if (fileRev == null) {
tikhomirov@401: 				throw new HgInvalidRevisionException(String.format("File %s is not part of working copy", getPath()), null, fileRevisionIndex);
tikhomirov@401: 			}
tikhomirov@401: 			fileRevisionIndex = getRevisionIndex(fileRev);
tikhomirov@401: 		}
tikhomirov@367: 		if (metadata == null || !metadata.checked(fileRevisionIndex)) {
tikhomirov@367: 			checkAndRecordMetadata(fileRevisionIndex);
tikhomirov@275: 		}
tikhomirov@367: 		final int dataLen = content.dataLength(fileRevisionIndex);
tikhomirov@367: 		if (metadata.known(fileRevisionIndex)) {
tikhomirov@367: 			return dataLen - metadata.dataOffset(fileRevisionIndex);
tikhomirov@275: 		}
tikhomirov@275: 		return dataLen;
tikhomirov@22: 	}
tikhomirov@416: 	
tikhomirov@416: 	/**
tikhomirov@237: 	 * Reads content of the file from working directory. If file present in the working directory, its actual content without
tikhomirov@237: 	 * any filters is supplied through the sink. If file does not exist in the working dir, this method provides content of a file 
tikhomirov@401: 	 * as if it would be refreshed in the working copy, i.e. its corresponding revision (according to dirstate) is read from the 
tikhomirov@401: 	 * repository, and filters repo -> working copy get applied.
tikhomirov@401: 	 * 
tikhomirov@401: 	 * NOTE, if file is missing from the working directory and is not part of the dirstate (but otherwise legal repository file,
tikhomirov@401: 	 * e.g. from another branch), no content would be supplied.
tikhomirov@237: 	 *     
tikhomirov@396: 	 * @param sink content consumer
tikhomirov@396: 	 * @throws HgInvalidControlFileException if access to revlog index/data entry failed
tikhomirov@396: 	 * @throws HgInvalidFileException if access to file in working directory failed
tikhomirov@380: 	 * @throws CancelledException if execution of the operation was cancelled
tikhomirov@237: 	 */
tikhomirov@396: 	public void workingCopy(ByteChannel sink) throws HgException, CancelledException {
tikhomirov@237: 		File f = getRepo().getFile(this);
tikhomirov@237: 		if (f.exists()) {
tikhomirov@237: 			final CancelSupport cs = CancelSupport.Factory.get(sink);
tikhomirov@237: 			final ProgressSupport progress = ProgressSupport.Factory.get(sink);
tikhomirov@237: 			final long flength = f.length();
tikhomirov@237: 			final int bsize = (int) Math.min(flength, 32*1024);
tikhomirov@237: 			progress.start((int) (flength > Integer.MAX_VALUE ? flength >>> 15 /*32 kb buf size*/ : flength));
tikhomirov@237: 			ByteBuffer buf = ByteBuffer.allocate(bsize);
tikhomirov@237: 			FileChannel fc = null;
tikhomirov@237: 			try {
tikhomirov@237: 				fc = new FileInputStream(f).getChannel();
tikhomirov@237: 				while (fc.read(buf) != -1) {
tikhomirov@237: 					cs.checkCancelled();
tikhomirov@237: 					buf.flip();
tikhomirov@237: 					int consumed = sink.write(buf);
tikhomirov@237: 					progress.worked(flength > Integer.MAX_VALUE ? 1 : consumed);
tikhomirov@237: 					buf.compact();
tikhomirov@237: 				}
tikhomirov@237: 			} catch (IOException ex) {
tikhomirov@396: 				throw new HgInvalidFileException("Working copy read failed", ex, f);
tikhomirov@237: 			} finally {
tikhomirov@237: 				progress.done();
tikhomirov@237: 				if (fc != null) {
tikhomirov@237: 					try {
tikhomirov@237: 						fc.close();
tikhomirov@237: 					} catch (IOException ex) {
tikhomirov@295: 						getRepo().getContext().getLog().info(getClass(), ex, null);
tikhomirov@237: 					}
tikhomirov@237: 				}
tikhomirov@237: 			}
tikhomirov@237: 		} else {
tikhomirov@401: 			Nodeid fileRev = getWorkingCopyRevision();
tikhomirov@401: 			if (fileRev == null) {
tikhomirov@401: 				// no content for this data file in the working copy - it is not part of the actual working state.
tikhomirov@401: 				// XXX perhaps, shall report this to caller somehow, not silently pass no data?
tikhomirov@388: 				return;
tikhomirov@388: 			}
tikhomirov@401: 			final int fileRevIndex = getRevisionIndex(fileRev);
tikhomirov@401: 			contentWithFilters(fileRevIndex, sink);
tikhomirov@401: 		}
tikhomirov@401: 	}
tikhomirov@401: 	
tikhomirov@401: 	/**
tikhomirov@401: 	 * @return file revision as recorded in repository manifest for dirstate parent, or <code>null</code> if no file revision can be found 
tikhomirov@401: 	 */
tikhomirov@401: 	private Nodeid getWorkingCopyRevision() throws HgInvalidControlFileException {
tikhomirov@401: 		final Pair<Nodeid, Nodeid> wcParents = getRepo().getWorkingCopyParents();
tikhomirov@401: 		Nodeid p = wcParents.first().isNull() ? wcParents.second() : wcParents.first();
tikhomirov@401: 		final HgChangelog clog = getRepo().getChangelog();
tikhomirov@401: 		final int csetRevIndex;
tikhomirov@401: 		if (p.isNull()) {
tikhomirov@401: 			// no dirstate parents 
tikhomirov@401: 			getRepo().getContext().getLog().info(getClass(), "No dirstate parents, resort to TIP", getPath());
tikhomirov@401: 			// if it's a repository with no dirstate, use TIP then
tikhomirov@401: 			csetRevIndex = clog.getLastRevision();
tikhomirov@401: 			if (csetRevIndex == -1) {
tikhomirov@401: 				// shall not happen provided there's .i for this data file (hence at least one cset)
tikhomirov@401: 				// and perhaps exception is better here. However, null as "can't find" indication seems reasonable.
tikhomirov@401: 				return null;
tikhomirov@401: 			}
tikhomirov@401: 		} else {
tikhomirov@388: 			// common case to avoid searching complete changelog for nodeid match
tikhomirov@388: 			final Nodeid tipRev = clog.getRevision(TIP);
tikhomirov@388: 			if (tipRev.equals(p)) {
tikhomirov@388: 				csetRevIndex = clog.getLastRevision();
tikhomirov@388: 			} else {
tikhomirov@388: 				// bad luck, need to search honestly
tikhomirov@388: 				csetRevIndex = getRepo().getChangelog().getRevisionIndex(p);
tikhomirov@388: 			}
tikhomirov@237: 		}
tikhomirov@401: 		Nodeid fileRev = getRepo().getManifest().getFileRevision(csetRevIndex, getPath());
tikhomirov@401: 		// it's possible for a file to be in working dir and have store/.i but to belong e.g. to a different
tikhomirov@401: 		// branch than the one from dirstate. Thus it's possible to get null fileRev
tikhomirov@401: 		// which would serve as an indication this data file is not part of working copy
tikhomirov@401: 		return fileRev;
tikhomirov@2: 	}
tikhomirov@115: 	
tikhomirov@396: 	/**
tikhomirov@396: 	 * Access content of a file revision
tikhomirov@396: 	 * XXX not sure distinct method contentWithFilters() is the best way to do, perhaps, callers shall add filters themselves?
tikhomirov@396: 	 * 
tikhomirov@396:  	 * @param fileRevisionIndex - revision local index, non-negative. From predefined constants, {@link HgRepository#TIP} and {@link HgRepository#WORKING_COPY} make sense. 
tikhomirov@396: 	 * @param sink content consumer
tikhomirov@396: 	 * 
tikhomirov@396: 	 * @throws HgInvalidControlFileException if access to revlog index/data entry failed
tikhomirov@396: 	 * @throws HgInvalidFileException if access to file in working directory failed
tikhomirov@396: 	 * @throws CancelledException if execution of the operation was cancelled
tikhomirov@396: 	 * @throws HgInvalidRevisionException if supplied argument doesn't represent revision index in this revlog (<em>runtime exception</em>)
tikhomirov@396: 	 */
tikhomirov@396: 	public void contentWithFilters(int fileRevisionIndex, ByteChannel sink) throws HgException, CancelledException, HgInvalidRevisionException {
tikhomirov@396: 		if (fileRevisionIndex == WORKING_COPY) {
tikhomirov@237: 			workingCopy(sink); // pass un-mangled sink
tikhomirov@237: 		} else {
tikhomirov@396: 			content(fileRevisionIndex, new FilterByteChannel(sink, getRepo().getFiltersFromRepoToWorkingDir(getPath())));
tikhomirov@237: 		}
tikhomirov@115: 	}
tikhomirov@22: 
tikhomirov@367: 	/**
tikhomirov@396: 	 * Retrieve content of specific revision. Content is provided as is, without any filters (e.g. keywords, eol, etc.) applied.
tikhomirov@396: 	 * For filtered content, use {@link #contentWithFilters(int, ByteChannel)}. 
tikhomirov@367: 	 * 
tikhomirov@368:  	 * @param fileRevisionIndex - revision local index, non-negative. From predefined constants, {@link HgRepository#TIP} and {@link HgRepository#WORKING_COPY} make sense. 
tikhomirov@396: 	 * @param sink content consumer
tikhomirov@396: 	 * 
tikhomirov@380: 	 * @throws HgInvalidControlFileException if access to revlog index/data entry failed
tikhomirov@396: 	 * @throws HgInvalidFileException if access to file in working directory failed
tikhomirov@380: 	 * @throws CancelledException if execution of the operation was cancelled
tikhomirov@396: 	 * @throws HgInvalidRevisionException if supplied argument doesn't represent revision index in this revlog (<em>runtime exception</em>)
tikhomirov@367: 	 */
tikhomirov@396: 	public void content(int fileRevisionIndex, ByteChannel sink) throws HgException, CancelledException, HgInvalidRevisionException {
tikhomirov@367: 		// for data files need to check heading of the file content for possible metadata
tikhomirov@367: 		// @see http://mercurial.selenic.com/wiki/FileFormats#data.2BAC8-
tikhomirov@367: 		if (fileRevisionIndex == TIP) {
tikhomirov@367: 			fileRevisionIndex = getLastRevision();
tikhomirov@78: 		}
tikhomirov@367: 		if (fileRevisionIndex == WORKING_COPY) {
tikhomirov@237: 			// sink is supposed to come into workingCopy without filters
tikhomirov@237: 			// thus we shall not get here (into #content) from #contentWithFilters(WC)
tikhomirov@157: 			workingCopy(sink);
tikhomirov@157: 			return;
tikhomirov@157: 		}
tikhomirov@367: 		if (wrongRevisionIndex(fileRevisionIndex) || fileRevisionIndex == BAD_REVISION) {
tikhomirov@367: 			throw new HgInvalidRevisionException(fileRevisionIndex);
tikhomirov@148: 		}
tikhomirov@157: 		if (sink == null) {
tikhomirov@157: 			throw new IllegalArgumentException();
tikhomirov@157: 		}
tikhomirov@134: 		if (metadata == null) {
tikhomirov@134: 			metadata = new Metadata();
tikhomirov@134: 		}
tikhomirov@277: 		ErrorHandlingInspector insp;
tikhomirov@388: 		final LogFacility lf = getRepo().getContext().getLog();
tikhomirov@367: 		if (metadata.none(fileRevisionIndex)) {
tikhomirov@388: 			insp = new ContentPipe(sink, 0, lf);
tikhomirov@367: 		} else if (metadata.known(fileRevisionIndex)) {
tikhomirov@388: 			insp = new ContentPipe(sink, metadata.dataOffset(fileRevisionIndex), lf);
tikhomirov@157: 		} else {
tikhomirov@157: 			// do not know if there's metadata
tikhomirov@396: 			insp = new MetadataInspector(metadata, lf, new ContentPipe(sink, 0, lf));
tikhomirov@78: 		}
tikhomirov@157: 		insp.checkCancelled();
tikhomirov@367: 		super.content.iterate(fileRevisionIndex, fileRevisionIndex, true, insp);
tikhomirov@157: 		try {
tikhomirov@237: 			insp.checkFailed(); // XXX is there real need to throw IOException from ContentPipe?
tikhomirov@396: 		} catch (HgInvalidControlFileException ex) {
tikhomirov@396: 			ex = ex.setFileName(getPath());
tikhomirov@396: 			throw ex.isRevisionIndexSet() ? ex : ex.setRevisionIndex(fileRevisionIndex);
tikhomirov@237: 		} catch (IOException ex) {
tikhomirov@396: 			HgInvalidControlFileException e = new HgInvalidControlFileException("Revision content access failed", ex, null);
tikhomirov@396: 			throw content.initWithIndexFile(e).setFileName(getPath()).setRevisionIndex(fileRevisionIndex);
tikhomirov@157: 		} catch (HgException ex) {
tikhomirov@157: 			// shall not happen, unless we changed ContentPipe or its subclass
tikhomirov@396: 			HgInvalidControlFileException e = new HgInvalidControlFileException("Revision content access failed", ex, null);
tikhomirov@396: 			throw content.initWithIndexFile(e).setFileName(getPath()).setRevisionIndex(fileRevisionIndex);
tikhomirov@78: 		}
tikhomirov@78: 	}
tikhomirov@157: 	
tikhomirov@424: 	// FIXME is that still needed?
tikhomirov@366: 	public void history(HgChangelog.Inspector inspector) throws HgInvalidControlFileException {
tikhomirov@135: 		history(0, getLastRevision(), inspector);
tikhomirov@48: 	}
tikhomirov@48: 
tikhomirov@417: 	/**
tikhomirov@417: 	 * 
tikhomirov@417: 	 * @param start local revision index
tikhomirov@417: 	 * @param end local revision index
tikhomirov@417: 	 * @param inspector
tikhomirov@417: 	 * FIXME EXCEPTIONS
tikhomirov@417: 	 * @throws HgInvalidRevisionException
tikhomirov@417: 	 * @throws HgInvalidControlFileException
tikhomirov@417: 	 */
tikhomirov@366: 	public void history(int start, int end, HgChangelog.Inspector inspector) throws HgInvalidRevisionException, HgInvalidControlFileException {
tikhomirov@3: 		if (!exists()) {
tikhomirov@3: 			throw new IllegalStateException("Can't get history of invalid repository file node"); 
tikhomirov@3: 		}
tikhomirov@135: 		final int last = getLastRevision();
tikhomirov@77: 		if (end == TIP) {
tikhomirov@77: 			end = last;
tikhomirov@77: 		}
tikhomirov@300: 		if (start == TIP) {
tikhomirov@300: 			start = last;
tikhomirov@300: 		}
tikhomirov@300: 		HgInternals.checkRevlogRange(start, end, last);
tikhomirov@300: 
tikhomirov@48: 		final int[] commitRevisions = new int[end - start + 1];
tikhomirov@242: 		final boolean[] needsSorting = { false };
tikhomirov@77: 		RevlogStream.Inspector insp = new RevlogStream.Inspector() {
tikhomirov@3: 			int count = 0;
tikhomirov@51: 			public void next(int revisionNumber, int actualLen, int baseRevision, int linkRevision, int parent1Revision, int parent2Revision, byte[] nodeid, DataAccess data) {
tikhomirov@242: 				if (count > 0) {
tikhomirov@242: 					if (commitRevisions[count -1] > linkRevision) {
tikhomirov@242: 						needsSorting[0] = true;
tikhomirov@242: 					}
tikhomirov@242: 				}
tikhomirov@3: 				commitRevisions[count++] = linkRevision;
tikhomirov@3: 			}
tikhomirov@3: 		};
tikhomirov@48: 		content.iterate(start, end, false, insp);
tikhomirov@233: 		final HgChangelog changelog = getRepo().getChangelog();
tikhomirov@242: 		if (needsSorting[0]) {
tikhomirov@242: 			// automatic tools (svnmerge?) produce unnatural file history
tikhomirov@242: 			// (e.g. cpython/Lib/doctest.py, revision 164 points to cset 63509, 165 - to 38453) 
tikhomirov@242: 			Arrays.sort(commitRevisions);
tikhomirov@233: 		}
tikhomirov@245: 		changelog.rangeInternal(inspector, commitRevisions);
tikhomirov@3: 	}
tikhomirov@88: 	
tikhomirov@354: 	/**
tikhomirov@367: 	 * For a given revision of the file (identified with revision index), find out index of the corresponding changeset.
tikhomirov@354: 	 *
tikhomirov@354: 	 * @return changeset revision index
tikhomirov@354: 	 * @throws HgInvalidRevisionException if supplied argument doesn't represent revision index in this revlog
tikhomirov@354: 	 * @throws HgInvalidControlFileException if access to revlog index/data entry failed
tikhomirov@354: 	 */
tikhomirov@367: 	public int getChangesetRevisionIndex(int revision) throws HgInvalidControlFileException, HgInvalidRevisionException {
tikhomirov@367: 		return content.linkRevision(revision);
tikhomirov@367: 	}
tikhomirov@88: 
tikhomirov@354: 	/**
tikhomirov@367: 	 * Complements {@link #getChangesetRevisionIndex(int)} to get changeset revision that corresponds to supplied file revision
tikhomirov@354: 	 * 
tikhomirov@354: 	 * @param nid revision of the file
tikhomirov@354: 	 * @return changeset revision
tikhomirov@354: 	 * @throws HgInvalidRevisionException if supplied argument doesn't represent revision index in this revlog
tikhomirov@354: 	 * @throws HgInvalidControlFileException if access to revlog index/data entry failed
tikhomirov@354: 	 */
tikhomirov@354: 	public Nodeid getChangesetRevision(Nodeid nid) throws HgInvalidControlFileException, HgInvalidRevisionException {
tikhomirov@367: 		int changelogRevision = getChangesetRevisionIndex(getRevisionIndex(nid));
tikhomirov@88: 		return getRepo().getChangelog().getRevision(changelogRevision);
tikhomirov@88: 	}
tikhomirov@78: 
tikhomirov@354: 	/**
tikhomirov@396: 	 * Tells whether this file originates from another repository file
tikhomirov@396: 	 * @return <code>true</code> if this file is a copy of another from the repository
tikhomirov@396: 	 * @throws HgInvalidControlFileException if access to revlog or file metadata failed
tikhomirov@354: 	 */
tikhomirov@396: 	public boolean isCopy() throws HgInvalidControlFileException {
tikhomirov@134: 		if (metadata == null || !metadata.checked(0)) {
tikhomirov@275: 			checkAndRecordMetadata(0);
tikhomirov@78: 		}
tikhomirov@134: 		if (!metadata.known(0)) {
tikhomirov@78: 			return false;
tikhomirov@78: 		}
tikhomirov@78: 		return metadata.find(0, "copy") != null;
tikhomirov@78: 	}
tikhomirov@78: 
tikhomirov@354: 	/**
tikhomirov@354: 	 * Get name of the file this one was copied from.
tikhomirov@354: 	 * 
tikhomirov@354: 	 * @return name of the file origin
tikhomirov@396: 	 * @throws HgInvalidControlFileException if access to revlog or file metadata failed
tikhomirov@354: 	 * @throws UnsupportedOperationException if this file doesn't represent a copy ({@link #isCopy()} was false)
tikhomirov@354: 	 */
tikhomirov@396: 	public Path getCopySourceName() throws HgInvalidControlFileException {
tikhomirov@78: 		if (isCopy()) {
tikhomirov@78: 			return Path.create(metadata.find(0, "copy"));
tikhomirov@78: 		}
tikhomirov@78: 		throw new UnsupportedOperationException(); // XXX REVISIT, think over if Exception is good (clients would check isCopy() anyway, perhaps null is sufficient?)
tikhomirov@78: 	}
tikhomirov@78: 	
tikhomirov@415: 	/**
tikhomirov@415: 	 * 
tikhomirov@415: 	 * @return revision this file was copied from
tikhomirov@415: 	 * @throws HgInvalidControlFileException if access to revlog or file metadata failed
tikhomirov@415: 	 * @throws UnsupportedOperationException if this file doesn't represent a copy ({@link #isCopy()} was false)
tikhomirov@415: 	 */
tikhomirov@396: 	public Nodeid getCopySourceRevision() throws HgInvalidControlFileException {
tikhomirov@78: 		if (isCopy()) {
tikhomirov@78: 			return Nodeid.fromAscii(metadata.find(0, "copyrev")); // XXX reuse/cache Nodeid
tikhomirov@78: 		}
tikhomirov@78: 		throw new UnsupportedOperationException();
tikhomirov@78: 	}
tikhomirov@417: 
tikhomirov@415: 	/**
tikhomirov@417: 	 * Get file flags recorded in the manifest
tikhomirov@416:  	 * @param fileRevisionIndex - revision local index, non-negative, or {@link HgRepository#TIP}. 
tikhomirov@417: 	 * @see HgManifest#getFileFlags(int, Path) 
tikhomirov@417: 	 * FIXME EXCEPTIONS
tikhomirov@415: 	 * @throws HgInvalidControlFileException
tikhomirov@415: 	 * @throws HgInvalidRevisionException
tikhomirov@415: 	 */
tikhomirov@415: 	public HgManifest.Flags getFlags(int fileRevisionIndex) throws HgInvalidControlFileException, HgInvalidRevisionException {
tikhomirov@415: 		int changesetRevIndex = getChangesetRevisionIndex(fileRevisionIndex);
tikhomirov@417: 		return getRepo().getManifest().getFileFlags(changesetRevIndex, getPath());
tikhomirov@415: 	}
tikhomirov@416: 	
tikhomirov@88: 	@Override
tikhomirov@88: 	public String toString() {
tikhomirov@88: 		StringBuilder sb = new StringBuilder(getClass().getSimpleName());
tikhomirov@88: 		sb.append('(');
tikhomirov@88: 		sb.append(getPath());
tikhomirov@88: 		sb.append(')');
tikhomirov@88: 		return sb.toString();
tikhomirov@88: 	}
tikhomirov@275: 	
tikhomirov@396: 	private void checkAndRecordMetadata(int localRev) throws HgInvalidControlFileException {
tikhomirov@275: 		// content() always initializes metadata.
tikhomirov@401: 		// TODO [post-1.0] this is expensive way to find out metadata, distinct RevlogStream.Iterator would be better.
tikhomirov@275: 		// Alternatively, may parameterize MetadataContentPipe to do prepare only.
tikhomirov@275: 		// For reference, when throwing CancelledException, hg status -A --rev 3:80 takes 70 ms
tikhomirov@275: 		// however, if we just consume buffer instead (buffer.position(buffer.limit()), same command takes ~320ms
tikhomirov@275: 		// (compared to command-line counterpart of 190ms)
tikhomirov@275: 		try {
tikhomirov@275: 			content(localRev, new ByteChannel() { // No-op channel
tikhomirov@275: 				public int write(ByteBuffer buffer) throws IOException, CancelledException {
tikhomirov@275: 					throw new CancelledException();
tikhomirov@275: 				}
tikhomirov@275: 			});
tikhomirov@275: 		} catch (CancelledException ex) {
tikhomirov@275: 			// it's ok, we did that
tikhomirov@366: 		} catch (HgInvalidControlFileException ex) {
tikhomirov@396: 			throw ex.isRevisionIndexSet() ? ex : ex.setRevisionIndex(localRev);
tikhomirov@396: 		} catch (HgException ex) {
tikhomirov@396: 			// metadata comes from the content, hence initWithDataFile
tikhomirov@396: 			throw content.initWithDataFile(new HgInvalidControlFileException(null, ex, null));
tikhomirov@275: 		}
tikhomirov@275: 	}
tikhomirov@78: 
tikhomirov@88: 	private static final class MetadataEntry {
tikhomirov@78: 		private final String entry;
tikhomirov@78: 		private final int valueStart;
tikhomirov@388: 
tikhomirov@388: 		// key may be null
tikhomirov@78: 		/*package-local*/MetadataEntry(String key, String value) {
tikhomirov@388: 			if (key == null) {
tikhomirov@388: 				entry = value;
tikhomirov@388: 				valueStart = -1; // not 0 to tell between key == null and key == ""
tikhomirov@388: 			} else {
tikhomirov@388: 				entry = key + value;
tikhomirov@388: 				valueStart = key.length();
tikhomirov@388: 			}
tikhomirov@78: 		}
tikhomirov@78: 		/*package-local*/boolean matchKey(String key) {
tikhomirov@388: 			return key == null ? valueStart == -1 : key.length() == valueStart && entry.startsWith(key);
tikhomirov@78: 		}
tikhomirov@134: //		uncomment once/if needed
tikhomirov@134: //		public String key() {
tikhomirov@134: //			return entry.substring(0, valueStart);
tikhomirov@134: //		}
tikhomirov@78: 		public String value() {
tikhomirov@388: 			return valueStart == -1 ? entry : entry.substring(valueStart);
tikhomirov@78: 		}
tikhomirov@78: 	}
tikhomirov@78: 
tikhomirov@78: 	private static class Metadata {
tikhomirov@276: 		private static class Record {
tikhomirov@276: 			public final int offset;
tikhomirov@276: 			public final MetadataEntry[] entries;
tikhomirov@276: 			
tikhomirov@276: 			public Record(int off, MetadataEntry[] entr) {
tikhomirov@276: 				offset = off;
tikhomirov@276: 				entries = entr;
tikhomirov@276: 			}
tikhomirov@276: 		}
tikhomirov@78: 		// XXX sparse array needed
tikhomirov@276: 		private final IntMap<Record> entries = new IntMap<Record>(5);
tikhomirov@134: 		
tikhomirov@276: 		private final Record NONE = new Record(-1, null); // don't want statics
tikhomirov@134: 
tikhomirov@134: 		// true when there's metadata for given revision
tikhomirov@78: 		boolean known(int revision) {
tikhomirov@276: 			Record i = entries.get(revision);
tikhomirov@134: 			return i != null && NONE != i;
tikhomirov@134: 		}
tikhomirov@134: 
tikhomirov@134: 		// true when revision has been checked for metadata presence.
tikhomirov@134: 		public boolean checked(int revision) {
tikhomirov@276: 			return entries.containsKey(revision);
tikhomirov@78: 		}
tikhomirov@134: 
tikhomirov@134: 		// true when revision has been checked and found not having any metadata
tikhomirov@134: 		boolean none(int revision) {
tikhomirov@276: 			Record i = entries.get(revision);
tikhomirov@134: 			return i == NONE;
tikhomirov@134: 		}
tikhomirov@134: 
tikhomirov@134: 		// mark revision as having no metadata.
tikhomirov@134: 		void recordNone(int revision) {
tikhomirov@276: 			Record i = entries.get(revision);
tikhomirov@134: 			if (i == NONE) {
tikhomirov@134: 				return; // already there
tikhomirov@134: 			} 
tikhomirov@134: 			if (i != null) {
tikhomirov@134: 				throw new IllegalStateException(String.format("Trying to override Metadata state for revision %d (known offset: %d)", revision, i));
tikhomirov@134: 			}
tikhomirov@276: 			entries.put(revision, NONE);
tikhomirov@134: 		}
tikhomirov@134: 
tikhomirov@78: 		// since this is internal class, callers are supposed to ensure arg correctness (i.e. ask known() before)
tikhomirov@78: 		int dataOffset(int revision) {
tikhomirov@276: 			return entries.get(revision).offset;
tikhomirov@78: 		}
tikhomirov@78: 		void add(int revision, int dataOffset, Collection<MetadataEntry> e) {
tikhomirov@276: 			assert !entries.containsKey(revision);
tikhomirov@276: 			entries.put(revision, new Record(dataOffset, e.toArray(new MetadataEntry[e.size()])));
tikhomirov@78: 		}
tikhomirov@276: 
tikhomirov@78: 		String find(int revision, String key) {
tikhomirov@276: 			for (MetadataEntry me : entries.get(revision).entries) {
tikhomirov@78: 				if (me.matchKey(key)) {
tikhomirov@78: 					return me.value();
tikhomirov@78: 				}
tikhomirov@78: 			}
tikhomirov@78: 			return null;
tikhomirov@78: 		}
tikhomirov@78: 	}
tikhomirov@157: 
tikhomirov@277: 	private static class MetadataInspector extends ErrorHandlingInspector implements RevlogStream.Inspector {
tikhomirov@157: 		private final Metadata metadata;
tikhomirov@277: 		private final RevlogStream.Inspector delegate;
tikhomirov@388: 		private final LogFacility log;
tikhomirov@157: 
tikhomirov@396: 		public MetadataInspector(Metadata _metadata, LogFacility logFacility, RevlogStream.Inspector chain) {
tikhomirov@157: 			metadata = _metadata;
tikhomirov@388: 			log = logFacility;
tikhomirov@277: 			delegate = chain;
tikhomirov@277: 			setCancelSupport(CancelSupport.Factory.get(chain));
tikhomirov@157: 		}
tikhomirov@157: 
tikhomirov@423: 		public void next(int revisionNumber, int actualLen, int baseRevision, int linkRevision, int parent1Revision, int parent2Revision, byte[] nodeid, DataAccess data) {
tikhomirov@277: 			try {
tikhomirov@277: 				final int daLength = data.length();
tikhomirov@277: 				if (daLength < 4 || data.readByte() != 1 || data.readByte() != 10) {
tikhomirov@277: 					metadata.recordNone(revisionNumber);
tikhomirov@277: 					data.reset();
tikhomirov@277: 				} else {
tikhomirov@277: 					ArrayList<MetadataEntry> _metadata = new ArrayList<MetadataEntry>();
tikhomirov@277: 					int offset = parseMetadata(data, daLength, _metadata);
tikhomirov@277: 					metadata.add(revisionNumber, offset, _metadata);
tikhomirov@277: 					// da is in prepared state (i.e. we consumed all bytes up to metadata end).
tikhomirov@277: 					// However, it's not safe to assume delegate won't call da.reset() for some reason,
tikhomirov@277: 					// and we need to ensure predictable result.
tikhomirov@277: 					data.reset();
tikhomirov@277: 					data = new FilterDataAccess(data, offset, daLength - offset);
tikhomirov@277: 				}
tikhomirov@277: 				if (delegate != null) {
tikhomirov@277: 					delegate.next(revisionNumber, actualLen, baseRevision, linkRevision, parent1Revision, parent2Revision, nodeid, data);
tikhomirov@277: 				}
tikhomirov@277: 			} catch (IOException ex) {
tikhomirov@277: 				recordFailure(ex);
tikhomirov@396: 			} catch (HgInvalidControlFileException ex) {
tikhomirov@423: 				// TODO RevlogStream, where this RevlogStream.Inspector goes, shall set File (as it's the only one having access to it)
tikhomirov@396: 				recordFailure(ex.isRevisionIndexSet() ? ex : ex.setRevisionIndex(revisionNumber));
tikhomirov@157: 			}
tikhomirov@277: 		}
tikhomirov@277: 
tikhomirov@396: 		private int parseMetadata(DataAccess data, final int daLength, ArrayList<MetadataEntry> _metadata) throws IOException, HgInvalidControlFileException {
tikhomirov@157: 			int lastEntryStart = 2;
tikhomirov@157: 			int lastColon = -1;
tikhomirov@157: 			// XXX in fact, need smth like ByteArrayBuilder, similar to StringBuilder,
tikhomirov@157: 			// which can't be used here because we can't convert bytes to chars as we read them
tikhomirov@157: 			// (there might be multi-byte encoding), and we need to collect all bytes before converting to string 
tikhomirov@157: 			ByteArrayOutputStream bos = new ByteArrayOutputStream();
tikhomirov@157: 			String key = null, value = null;
tikhomirov@157: 			boolean byteOne = false;
tikhomirov@323: 			boolean metadataIsComplete = false;
tikhomirov@157: 			for (int i = 2; i < daLength; i++) {
tikhomirov@277: 				byte b = data.readByte();
tikhomirov@157: 				if (b == '\n') {
tikhomirov@157: 					if (byteOne) { // i.e. \n follows 1
tikhomirov@157: 						lastEntryStart = i+1;
tikhomirov@323: 						metadataIsComplete = true;
tikhomirov@157: 						// XXX is it possible to have here incomplete key/value (i.e. if last pair didn't end with \n)
tikhomirov@323: 						// if yes, need to set metadataIsComplete to true in that case as well
tikhomirov@157: 						break;
tikhomirov@157: 					}
tikhomirov@157: 					if (key == null || lastColon == -1 || i <= lastColon) {
tikhomirov@388: 						log.error(getClass(), "Missing key in file revision metadata at index %d", i);
tikhomirov@157: 					}
tikhomirov@157: 					value = new String(bos.toByteArray()).trim();
tikhomirov@157: 					bos.reset();
tikhomirov@157: 					_metadata.add(new MetadataEntry(key, value));
tikhomirov@157: 					key = value = null;
tikhomirov@157: 					lastColon = -1;
tikhomirov@157: 					lastEntryStart = i+1;
tikhomirov@157: 					continue;
tikhomirov@157: 				} 
tikhomirov@277: 				// byteOne has to be consumed up to this line, if not yet, consume it
tikhomirov@157: 				if (byteOne) {
tikhomirov@157: 					// insert 1 we've read on previous step into the byte builder
tikhomirov@157: 					bos.write(1);
tikhomirov@277: 					byteOne = false;
tikhomirov@157: 					// fall-through to consume current byte
tikhomirov@157: 				}
tikhomirov@157: 				if (b == (int) ':') {
tikhomirov@157: 					assert value == null;
tikhomirov@157: 					key = new String(bos.toByteArray());
tikhomirov@157: 					bos.reset();
tikhomirov@157: 					lastColon = i;
tikhomirov@157: 				} else if (b == 1) {
tikhomirov@157: 					byteOne = true;
tikhomirov@157: 				} else {
tikhomirov@157: 					bos.write(b);
tikhomirov@157: 				}
tikhomirov@157: 			}
tikhomirov@323: 			// data.isEmpty is not reliable, renamed files of size==0 keep only metadata
tikhomirov@323: 			if (!metadataIsComplete) {
tikhomirov@323: 				// XXX perhaps, worth a testcase (empty file, renamed, read or ask ifCopy
tikhomirov@396: 				throw new HgInvalidControlFileException("Metadata is not closed properly", null, null);
tikhomirov@157: 			}
tikhomirov@277: 			return lastEntryStart;
tikhomirov@17: 		}
tikhomirov@322: 
tikhomirov@322: 		@Override
tikhomirov@322: 		public void checkFailed() throws HgException, IOException, CancelledException {
tikhomirov@322: 			super.checkFailed();
tikhomirov@322: 			if (delegate instanceof ErrorHandlingInspector) {
tikhomirov@322: 				// XXX need to add ErrorDestination and pass it around (much like CancelSupport get passed)
tikhomirov@322: 				// so that delegate would be able report its failures directly to caller without this hack
tikhomirov@322: 				((ErrorHandlingInspector) delegate).checkFailed();
tikhomirov@322: 			}
tikhomirov@322: 		}
tikhomirov@17: 	}
tikhomirov@2: }