tikhomirov@58: /*
tikhomirov@74:  * Copyright (c) 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@74:  * contact TMate Software at support@svnkit.com
tikhomirov@58:  */
tikhomirov@74: package org.tmatesoft.hg.repo;
tikhomirov@58: 
tikhomirov@74: import static org.tmatesoft.hg.repo.HgRepository.BAD_REVISION;
tikhomirov@74: import static org.tmatesoft.hg.repo.HgRepository.TIP;
tikhomirov@58: 
tikhomirov@58: import java.io.BufferedInputStream;
tikhomirov@58: import java.io.File;
tikhomirov@58: import java.io.FileInputStream;
tikhomirov@58: import java.io.IOException;
tikhomirov@58: import java.util.Collections;
tikhomirov@58: import java.util.Set;
tikhomirov@58: import java.util.TreeSet;
tikhomirov@58: 
tikhomirov@74: import org.tmatesoft.hg.core.Nodeid;
tikhomirov@89: import org.tmatesoft.hg.repo.StatusCollector.ManifestRevisionInspector;
tikhomirov@74: import org.tmatesoft.hg.util.FileWalker;
tikhomirov@58: 
tikhomirov@58: /**
tikhomirov@58:  *
tikhomirov@74:  * @author Artem Tikhomirov
tikhomirov@74:  * @author TMate Software Ltd.
tikhomirov@58:  */
tikhomirov@58: public class WorkingCopyStatusCollector {
tikhomirov@58: 
tikhomirov@58: 	private final HgRepository repo;
tikhomirov@58: 	private final FileWalker repoWalker;
tikhomirov@59: 	private HgDirstate dirstate;
tikhomirov@59: 	private StatusCollector baseRevisionCollector;
tikhomirov@58: 
tikhomirov@74: 	public WorkingCopyStatusCollector(HgRepository hgRepo) {
tikhomirov@74: 		this(hgRepo, hgRepo.createWorkingDirWalker());
tikhomirov@74: 	}
tikhomirov@74: 
tikhomirov@74: 	WorkingCopyStatusCollector(HgRepository hgRepo, FileWalker hgRepoWalker) {
tikhomirov@58: 		this.repo = hgRepo;
tikhomirov@58: 		this.repoWalker = hgRepoWalker;
tikhomirov@58: 	}
tikhomirov@59: 	
tikhomirov@59: 	/**
tikhomirov@59: 	 * Optionally, supply a collector instance that may cache (or have already cached) base revision
tikhomirov@59: 	 * @param sc may be null
tikhomirov@59: 	 */
tikhomirov@63: 	public void setBaseRevisionCollector(StatusCollector sc) {
tikhomirov@59: 		baseRevisionCollector = sc;
tikhomirov@59: 	}
tikhomirov@59: 	
tikhomirov@59: 	private HgDirstate getDirstate() {
tikhomirov@59: 		if (dirstate == null) {
tikhomirov@74: 			dirstate = repo.loadDirstate();
tikhomirov@59: 		}
tikhomirov@59: 		return dirstate;
tikhomirov@59: 	}
tikhomirov@58: 
tikhomirov@59: 	// may be invoked few times
tikhomirov@58: 	public void walk(int baseRevision, StatusCollector.Inspector inspector) {
tikhomirov@91: 		final HgIgnore hgIgnore = repo.getIgnore();
tikhomirov@59: 		TreeSet<String> knownEntries = getDirstate().all();
tikhomirov@68: 		final boolean isTipBase;
tikhomirov@68: 		if (baseRevision == TIP) {
tikhomirov@68: 			baseRevision = repo.getManifest().getRevisionCount() - 1;
tikhomirov@68: 			isTipBase = true;
tikhomirov@68: 		} else {
tikhomirov@68: 			isTipBase = baseRevision == repo.getManifest().getRevisionCount() - 1;
tikhomirov@68: 		}
tikhomirov@58: 		StatusCollector.ManifestRevisionInspector collect = null;
tikhomirov@58: 		Set<String> baseRevFiles = Collections.emptySet();
tikhomirov@58: 		if (!isTipBase) {
tikhomirov@59: 			if (baseRevisionCollector != null) {
tikhomirov@59: 				collect = baseRevisionCollector.raw(baseRevision);
tikhomirov@59: 			} else {
tikhomirov@89: 				collect = new StatusCollector.ManifestRevisionInspector();
tikhomirov@59: 				repo.getManifest().walk(baseRevision, baseRevision, collect);
tikhomirov@59: 			}
tikhomirov@89: 			baseRevFiles = new TreeSet<String>(collect.files());
tikhomirov@58: 		}
tikhomirov@68: 		if (inspector instanceof StatusCollector.Record) {
tikhomirov@68: 			StatusCollector sc = baseRevisionCollector == null ? new StatusCollector(repo) : baseRevisionCollector;
tikhomirov@68: 			((StatusCollector.Record) inspector).init(baseRevision, BAD_REVISION, sc);
tikhomirov@68: 		}
tikhomirov@58: 		repoWalker.reset();
tikhomirov@58: 		while (repoWalker.hasNext()) {
tikhomirov@58: 			repoWalker.next();
tikhomirov@58: 			String fname = repoWalker.name();
tikhomirov@58: 			File f = repoWalker.file();
tikhomirov@58: 			if (hgIgnore.isIgnored(fname)) {
tikhomirov@58: 				inspector.ignored(fname);
tikhomirov@58: 			} else if (knownEntries.remove(fname)) {
tikhomirov@58: 				// modified, added, removed, clean
tikhomirov@58: 				if (collect != null) { // need to check against base revision, not FS file
tikhomirov@90: 					checkLocalStatusAgainstBaseRevision(baseRevFiles, collect, baseRevision, fname, f, inspector);
tikhomirov@58: 					baseRevFiles.remove(fname);
tikhomirov@58: 				} else {
tikhomirov@59: 					checkLocalStatusAgainstFile(fname, f, inspector);
tikhomirov@58: 				}
tikhomirov@58: 			} else {
tikhomirov@58: 				inspector.unknown(fname);
tikhomirov@58: 			}
tikhomirov@58: 		}
tikhomirov@58: 		if (collect != null) {
tikhomirov@58: 			for (String r : baseRevFiles) {
tikhomirov@58: 				inspector.removed(r);
tikhomirov@58: 			}
tikhomirov@58: 		}
tikhomirov@58: 		for (String m : knownEntries) {
tikhomirov@74: 			// missing known file from a working dir  
tikhomirov@59: 			if (getDirstate().checkRemoved(m) == null) {
tikhomirov@74: 				// not removed from the repository = 'deleted'  
tikhomirov@58: 				inspector.missing(m);
tikhomirov@74: 			} else {
tikhomirov@74: 				// removed from the repo
tikhomirov@76: 				// if we check against non-tip revision, do not report files that were added past that revision and now removed.
tikhomirov@76: 				if (collect == null || baseRevFiles.contains(m)) {
tikhomirov@76: 					inspector.removed(m);
tikhomirov@76: 				}
tikhomirov@58: 			}
tikhomirov@58: 		}
tikhomirov@58: 	}
tikhomirov@58: 
tikhomirov@58: 	public StatusCollector.Record status(int baseRevision) {
tikhomirov@58: 		StatusCollector.Record rv = new StatusCollector.Record();
tikhomirov@58: 		walk(baseRevision, rv);
tikhomirov@58: 		return rv;
tikhomirov@58: 	}
tikhomirov@58: 
tikhomirov@58: 	//********************************************
tikhomirov@58: 
tikhomirov@58: 	
tikhomirov@59: 	private void checkLocalStatusAgainstFile(String fname, File f, StatusCollector.Inspector inspector) {
tikhomirov@58: 		HgDirstate.Record r;
tikhomirov@59: 		if ((r = getDirstate().checkNormal(fname)) != null) {
tikhomirov@58: 			// either clean or modified
tikhomirov@58: 			if (f.lastModified() / 1000 == r.time && r.size == f.length()) {
tikhomirov@58: 				inspector.clean(fname);
tikhomirov@58: 			} else {
tikhomirov@58: 				// FIXME check actual content to avoid false modified files
tikhomirov@58: 				inspector.modified(fname);
tikhomirov@58: 			}
tikhomirov@59: 		} else if ((r = getDirstate().checkAdded(fname)) != null) {
tikhomirov@58: 			if (r.name2 == null) {
tikhomirov@58: 				inspector.added(fname);
tikhomirov@58: 			} else {
tikhomirov@74: 				inspector.copied(r.name2, fname);
tikhomirov@58: 			}
tikhomirov@59: 		} else if ((r = getDirstate().checkRemoved(fname)) != null) {
tikhomirov@58: 			inspector.removed(fname);
tikhomirov@59: 		} else if ((r = getDirstate().checkMerged(fname)) != null) {
tikhomirov@58: 			inspector.modified(fname);
tikhomirov@58: 		}
tikhomirov@58: 	}
tikhomirov@58: 	
tikhomirov@58: 	// XXX refactor checkLocalStatus methods in more OO way
tikhomirov@90: 	private void checkLocalStatusAgainstBaseRevision(Set<String> baseRevNames, ManifestRevisionInspector collect, int baseRevision, String fname, File f, StatusCollector.Inspector inspector) {
tikhomirov@58: 		// fname is in the dirstate, either Normal, Added, Removed or Merged
tikhomirov@89: 		Nodeid nid1 = collect.nodeid(fname);
tikhomirov@89: 		String flags = collect.flags(fname);
tikhomirov@58: 		HgDirstate.Record r;
tikhomirov@58: 		if (nid1 == null) {
tikhomirov@58: 			// normal: added?
tikhomirov@58: 			// added: not known at the time of baseRevision, shall report
tikhomirov@58: 			// merged: was not known, report as added?
tikhomirov@90: 			if ((r = getDirstate().checkNormal(fname)) != null) {
tikhomirov@90: 				String origin = StatusCollector.getOriginIfCopy(repo, fname, baseRevNames, baseRevision);
tikhomirov@90: 				if (origin != null) {
tikhomirov@90: 					inspector.copied(origin, fname);
tikhomirov@90: 					return;
tikhomirov@90: 				}
tikhomirov@90: 			} else if ((r = getDirstate().checkAdded(fname)) != null) {
tikhomirov@58: 				if (r.name2 != null && baseRevNames.contains(r.name2)) {
tikhomirov@90: 					baseRevNames.remove(r.name2); // XXX surely I shall not report rename source as Removed?
tikhomirov@58: 					inspector.copied(r.name2, fname);
tikhomirov@58: 					return;
tikhomirov@58: 				}
tikhomirov@58: 				// fall-through, report as added
tikhomirov@59: 			} else if (getDirstate().checkRemoved(fname) != null) {
tikhomirov@58: 				// removed: removed file was not known at the time of baseRevision, and we should not report it as removed
tikhomirov@58: 				return;
tikhomirov@58: 			}
tikhomirov@58: 			inspector.added(fname);
tikhomirov@58: 		} else {
tikhomirov@58: 			// was known; check whether clean or modified
tikhomirov@58: 			// when added - seems to be the case of a file added once again, hence need to check if content is different
tikhomirov@59: 			if ((r = getDirstate().checkNormal(fname)) != null || (r = getDirstate().checkMerged(fname)) != null || (r = getDirstate().checkAdded(fname)) != null) {
tikhomirov@58: 				// either clean or modified
tikhomirov@58: 				HgDataFile fileNode = repo.getFileNode(fname);
tikhomirov@58: 				final int lengthAtRevision = fileNode.length(nid1);
tikhomirov@58: 				if (r.size /* XXX File.length() ?! */ != lengthAtRevision || flags != todoGenerateFlags(fname /*java.io.File*/)) {
tikhomirov@58: 					inspector.modified(fname);
tikhomirov@58: 				} else {
tikhomirov@58: 					// check actual content to see actual changes
tikhomirov@58: 					// XXX consider adding HgDataDile.compare(File/byte[]/whatever) operation to optimize comparison
tikhomirov@58: 					if (areTheSame(f, fileNode.content(nid1))) {
tikhomirov@58: 						inspector.clean(fname);
tikhomirov@58: 					} else {
tikhomirov@58: 						inspector.modified(fname);
tikhomirov@58: 					}
tikhomirov@58: 				}
tikhomirov@58: 			}
tikhomirov@58: 			// only those left in idsMap after processing are reported as removed 
tikhomirov@58: 		}
tikhomirov@58: 
tikhomirov@58: 		// TODO think over if content comparison may be done more effectively by e.g. calculating nodeid for a local file and comparing it with nodeid from manifest
tikhomirov@58: 		// we don't need to tell exact difference, hash should be enough to detect difference, and it doesn't involve reading historical file content, and it's relatively 
tikhomirov@58: 		// cheap to calc hash on a file (no need to keep it completely in memory). OTOH, if I'm right that the next approach is used for nodeids: 
tikhomirov@58: 		// changeset nodeid + hash(actual content) => entry (Nodeid) in the next Manifest
tikhomirov@58: 		// then it's sufficient to check parents from dirstate, and if they do not match parents from file's baseRevision (non matching parents means different nodeids).
tikhomirov@58: 		// The question is whether original Hg treats this case (same content, different parents and hence nodeids) as 'modified' or 'clean'
tikhomirov@58: 	}
tikhomirov@58: 
tikhomirov@58: 	private static String todoGenerateFlags(String fname) {
tikhomirov@58: 		// FIXME implement
tikhomirov@58: 		return null;
tikhomirov@58: 	}
tikhomirov@58: 	private static boolean areTheSame(File f, byte[] data) {
tikhomirov@58: 		try {
tikhomirov@58: 			BufferedInputStream is = new BufferedInputStream(new FileInputStream(f));
tikhomirov@58: 			int i = 0;
tikhomirov@58: 			while (i < data.length && data[i] == is.read()) {
tikhomirov@58: 				i++; // increment only for successful match, otherwise won't tell last byte in data was the same as read from the stream
tikhomirov@58: 			}
tikhomirov@58: 			return i == data.length && is.read() == -1; // although data length is expected to be the same (see caller), check that we reached EOF, no more data left.
tikhomirov@58: 		} catch (IOException ex) {
tikhomirov@58: 			ex.printStackTrace(); // log warn
tikhomirov@58: 		}
tikhomirov@58: 		return false;
tikhomirov@58: 	}
tikhomirov@58: 
tikhomirov@58: }