# HG changeset patch # User Artem Tikhomirov # Date 1295235909 -3600 # Node ID 05829a70b30baf3158333f69f9682806f802b969 # Parent fd4f2c98995beb051070630c272a9be87bef617d Status operation extracted into separate, cache-friendly class diff -r fd4f2c98995b -r 05829a70b30b design.txt --- a/design.txt Sun Jan 16 05:21:09 2011 +0100 +++ b/design.txt Mon Jan 17 04:45:09 2011 +0100 @@ -39,7 +39,8 @@ RevlogStream - Inflater. Perhaps, InflaterStream instead? Implement use of fncache (use names from it - perhaps, would help for Mac issues Alex mentioned) along with 'digest'-ing long file names - +repo.status - use same collector class twice, difference as external code. add external walker that keeps collected maps and use it + in Log operation to give files+,files- Status operation from GUI - guess, usually on a file/subfolder, hence API should allow for starting path (unlike cmdline, seems useless to implement include/exclide patterns - GUI users hardly enter them, ever) diff -r fd4f2c98995b -r 05829a70b30b src/com/tmate/hgkit/console/Status.java --- a/src/com/tmate/hgkit/console/Status.java Sun Jan 16 05:21:09 2011 +0100 +++ b/src/com/tmate/hgkit/console/Status.java Mon Jan 17 04:45:09 2011 +0100 @@ -5,11 +5,16 @@ import static com.tmate.hgkit.ll.HgRepository.TIP; +import java.util.ArrayList; +import java.util.Collections; +import java.util.List; + import com.tmate.hgkit.fs.RepositoryLookup; import com.tmate.hgkit.ll.HgDataFile; import com.tmate.hgkit.ll.HgRepository; import com.tmate.hgkit.ll.LocalHgRepo; import com.tmate.hgkit.ll.Nodeid; +import com.tmate.hgkit.ll.StatusCollector; /** * @@ -30,17 +35,34 @@ final StatusDump dump = new StatusDump(); dump.showIgnored = false; dump.showClean = false; - final int r1 = 0, r2 = 11; + StatusCollector sc = new StatusCollector(hgRepo); + final int r1 = 0, r2 = 3; System.out.printf("Status for changes between revision %d and %d:\n", r1, r2); - hgRepo.status(r1, r2, dump); - System.out.println("\nStatus against working dir:"); - ((LocalHgRepo) hgRepo).statusLocal(TIP, dump); - System.out.println(); - System.out.printf("Manifest of the revision %d:\n", r2); - hgRepo.getManifest().walk(r2, r2, new Manifest.Dump()); - System.out.println(); - System.out.printf("\nStatus of working dir against %d:\n", r2); - ((LocalHgRepo) hgRepo).statusLocal(r2, dump); + sc.walk(r1, r2, dump); + // + System.out.println("\n\nSame, but sorted in the way hg status does:"); + StatusCollector.Record r = sc.status(r1, r2); + sortAndPrint('M', r.getModified()); + sortAndPrint('A', r.getAdded()); + sortAndPrint('R', r.getRemoved()); +// System.out.println("\nStatus against working dir:"); +// ((LocalHgRepo) hgRepo).statusLocal(TIP, dump); +// System.out.println(); +// System.out.printf("Manifest of the revision %d:\n", r2); +// hgRepo.getManifest().walk(r2, r2, new Manifest.Dump()); +// System.out.println(); +// System.out.printf("\nStatus of working dir against %d:\n", r2); +// ((LocalHgRepo) hgRepo).statusLocal(r2, dump); + } + + private static void sortAndPrint(char prefix, List ul) { + ArrayList sortList = new ArrayList(ul); + Collections.sort(sortList); + for (String s : sortList) { + System.out.print(prefix); + System.out.print(' '); + System.out.println(s); + } } protected static void testStatusInternals(HgRepository hgRepo) { @@ -53,7 +75,7 @@ } } - private static class StatusDump implements HgRepository.StatusInspector { + private static class StatusDump implements StatusCollector.Inspector { public boolean hideStatusPrefix = false; // hg status -n option public boolean showCopied = true; // -C public boolean showIgnored = true; // -i diff -r fd4f2c98995b -r 05829a70b30b src/com/tmate/hgkit/ll/HgRepository.java --- a/src/com/tmate/hgkit/ll/HgRepository.java Sun Jan 16 05:21:09 2011 +0100 +++ b/src/com/tmate/hgkit/ll/HgRepository.java Mon Jan 17 04:45:09 2011 +0100 @@ -67,17 +67,4 @@ * Perhaps, should be separate interface, like ContentLookup */ protected abstract RevlogStream resolve(String repositoryPath); - - public abstract void status(int rev1, int rev2 /*WorkingDir - TIP, TIP?*/, StatusInspector inspector); - - public interface StatusInspector { - void modified(String fname); - void added(String fname); - void copied(String fnameOrigin, String fnameAdded); // if copied files of no interest, should delegate to self.added(fnameAdded); - void removed(String fname); - void clean(String fname); - void missing(String fname); // aka deleted (tracked by Hg, but not available in FS any more - void unknown(String fname); // not tracked - void ignored(String fname); - } } diff -r fd4f2c98995b -r 05829a70b30b src/com/tmate/hgkit/ll/LocalHgRepo.java --- a/src/com/tmate/hgkit/ll/LocalHgRepo.java Sun Jan 16 05:21:09 2011 +0100 +++ b/src/com/tmate/hgkit/ll/LocalHgRepo.java Mon Jan 17 04:45:09 2011 +0100 @@ -46,172 +46,133 @@ return repoLocation; } - @Override - public void status(int rev1, int rev2, final StatusInspector inspector) { - final ManifestRevisionCollector collect = new ManifestRevisionCollector(); - getManifest().walk(rev1, rev1, collect); - - HgManifest.Inspector compare = new HgManifest.Inspector() { - - public boolean begin(int revision, Nodeid nid) { - return true; - } - - public boolean next(Nodeid nid, String fname, String flags) { - Nodeid nidR1 = collect.idsMap.remove(fname); - String flagsR1 = collect.flagsMap.remove(fname); - if (nidR1 == null) { - inspector.added(fname); - } else { - if (nidR1.equals(nid) && ((flags == null && flagsR1 == null) || flags.equals(flagsR1))) { - inspector.clean(fname); - } else { - inspector.modified(fname); - } - } - return true; - } - - public boolean end(int revision) { - for (String fname : collect.idsMap.keySet()) { - inspector.removed(fname); - } - if (collect.idsMap.size() != collect.flagsMap.size()) { - throw new IllegalStateException(); - } - return false; - } - }; - getManifest().walk(rev2, rev2, compare); - } - - public void statusLocal(int baseRevision, StatusInspector inspector) { - LinkedList folders = new LinkedList(); - final File rootDir = repoDir.getParentFile(); - folders.add(rootDir); - final HgDirstate dirstate = loadDirstate(); - final HgIgnore hgignore = loadIgnore(); - TreeSet knownEntries = dirstate.all(); - final boolean isTipBase = baseRevision == TIP || baseRevision == getManifest().getRevisionCount(); - final ManifestRevisionCollector collect = isTipBase ? null : new ManifestRevisionCollector(); - if (!isTipBase) { - getManifest().walk(baseRevision, baseRevision, collect); - } - do { - File d = folders.removeFirst(); - for (File f : d.listFiles()) { - if (f.isDirectory()) { - if (!".hg".equals(f.getName())) { - folders.addLast(f); - } - } else { - // FIXME path relative to rootDir - need more robust approach - String fname = normalize(f.getPath().substring(rootDir.getPath().length() + 1)); - if (hgignore.isIgnored(fname)) { - inspector.ignored(fname); - } else { - if (knownEntries.remove(fname)) { - // modified, added, removed, clean - if (collect != null) { // need to check against base revision, not FS file - checkLocalStatusAgainstBaseRevision(collect, fname, f, dirstate, inspector); - } else { - checkLocalStatusAgainstFile(fname, f, dirstate, inspector); - } - } else { - inspector.unknown(fname); - } - } - } - } - } while (!folders.isEmpty()); - if (collect != null) { - for (String r : collect.idsMap.keySet()) { - inspector.removed(r); - } - } - for (String m : knownEntries) { - // removed from the repository and missing from working dir shall not be reported as 'deleted' - if (dirstate.checkRemoved(m) == null) { - inspector.missing(m); - } - } - } - - private static void checkLocalStatusAgainstFile(String fname, File f, HgDirstate dirstate, StatusInspector inspector) { - HgDirstate.Record r; - if ((r = dirstate.checkNormal(fname)) != null) { - // either clean or modified - if (f.lastModified() / 1000 == r.time && r.size == f.length()) { - inspector.clean(fname); - } else { - // FIXME check actual content to avoid false modified files - inspector.modified(fname); - } - } else if ((r = dirstate.checkAdded(fname)) != null) { - if (r.name2 == null) { - inspector.added(fname); - } else { - inspector.copied(fname, r.name2); - } - } else if ((r = dirstate.checkRemoved(fname)) != null) { - inspector.removed(fname); - } else if ((r = dirstate.checkMerged(fname)) != null) { - inspector.modified(fname); - } - } - - // XXX refactor checkLocalStatus methods in more OO way - private void checkLocalStatusAgainstBaseRevision(ManifestRevisionCollector collect, String fname, File f, HgDirstate dirstate, StatusInspector inspector) { - // fname is in the dirstate, either Normal, Added, Removed or Merged - Nodeid nid1 = collect.idsMap.remove(fname); - String flags = collect.flagsMap.remove(fname); - HgDirstate.Record r; - if (nid1 == null) { - // normal: added? - // added: not known at the time of baseRevision, shall report - // merged: was not known, report as added? - if ((r = dirstate.checkAdded(fname)) != null) { - if (r.name2 != null && collect.idsMap.containsKey(r.name2)) { - collect.idsMap.remove(r.name2); - collect.idsMap.remove(r.name2); - inspector.copied(r.name2, fname); - return; - } - // fall-through, report as added - } else if (dirstate.checkRemoved(fname) != null) { - // removed: removed file was not known at the time of baseRevision, and we should not report it as removed - return; - } - inspector.added(fname); - } else { - // was known; check whether clean or modified - // when added - seems to be the case of a file added once again, hence need to check if content is different - if ((r = dirstate.checkNormal(fname)) != null || (r = dirstate.checkMerged(fname)) != null || (r = dirstate.checkAdded(fname)) != null) { - // either clean or modified - HgDataFile fileNode = getFileNode(fname); - final int lengthAtRevision = fileNode.length(nid1); - if (r.size /* XXX File.length() ?! */ != lengthAtRevision || flags != todoGenerateFlags(fname /*java.io.File*/)) { - inspector.modified(fname); - } else { - // check actual content to see actual changes - // XXX consider adding HgDataDile.compare(File/byte[]/whatever) operation to optimize comparison - if (areTheSame(f, fileNode.content(nid1))) { - inspector.clean(fname); - } else { - inspector.modified(fname); - } - } - } - // only those left in idsMap after processing are reported as removed - } - - // 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 - // 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 - // 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: - // changeset nodeid + hash(actual content) => entry (Nodeid) in the next Manifest - // 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). - // The question is whether original Hg treats this case (same content, different parents and hence nodeids) as 'modified' or 'clean' - } +// public void statusLocal(int baseRevision, StatusCollector.Inspector inspector) { +// LinkedList folders = new LinkedList(); +// final File rootDir = repoDir.getParentFile(); +// folders.add(rootDir); +// final HgDirstate dirstate = loadDirstate(); +// final HgIgnore hgignore = loadIgnore(); +// TreeSet knownEntries = dirstate.all(); +// final boolean isTipBase = baseRevision == TIP || baseRevision == getManifest().getRevisionCount(); +// final StatusCollector.ManifestRevisionInspector collect = isTipBase ? null : new StatusCollector.ManifestRevisionInspector(); +// if (!isTipBase) { +// getManifest().walk(baseRevision, baseRevision, collect); +// } +// do { +// File d = folders.removeFirst(); +// for (File f : d.listFiles()) { +// if (f.isDirectory()) { +// if (!".hg".equals(f.getName())) { +// folders.addLast(f); +// } +// } else { +// // FIXME path relative to rootDir - need more robust approach +// String fname = normalize(f.getPath().substring(rootDir.getPath().length() + 1)); +// if (hgignore.isIgnored(fname)) { +// inspector.ignored(fname); +// } else { +// if (knownEntries.remove(fname)) { +// // modified, added, removed, clean +// if (collect != null) { // need to check against base revision, not FS file +// checkLocalStatusAgainstBaseRevision(collect, fname, f, dirstate, inspector); +// } else { +// checkLocalStatusAgainstFile(fname, f, dirstate, inspector); +// } +// } else { +// inspector.unknown(fname); +// } +// } +// } +// } +// } while (!folders.isEmpty()); +// if (collect != null) { +// for (String r : collect.idsMap.keySet()) { +// inspector.removed(r); +// } +// } +// for (String m : knownEntries) { +// // removed from the repository and missing from working dir shall not be reported as 'deleted' +// if (dirstate.checkRemoved(m) == null) { +// inspector.missing(m); +// } +// } +// } +// +// private static void checkLocalStatusAgainstFile(String fname, File f, HgDirstate dirstate, StatusCollector.Inspector inspector) { +// HgDirstate.Record r; +// if ((r = dirstate.checkNormal(fname)) != null) { +// // either clean or modified +// if (f.lastModified() / 1000 == r.time && r.size == f.length()) { +// inspector.clean(fname); +// } else { +// // FIXME check actual content to avoid false modified files +// inspector.modified(fname); +// } +// } else if ((r = dirstate.checkAdded(fname)) != null) { +// if (r.name2 == null) { +// inspector.added(fname); +// } else { +// inspector.copied(fname, r.name2); +// } +// } else if ((r = dirstate.checkRemoved(fname)) != null) { +// inspector.removed(fname); +// } else if ((r = dirstate.checkMerged(fname)) != null) { +// inspector.modified(fname); +// } +// } +// +// // XXX refactor checkLocalStatus methods in more OO way +// private void checkLocalStatusAgainstBaseRevision(StatusCollector.ManifestRevisionInspector collect, String fname, File f, HgDirstate dirstate, StatusCollector.Inspector inspector) { +// // fname is in the dirstate, either Normal, Added, Removed or Merged +// Nodeid nid1 = collect.idsMap.remove(fname); +// String flags = collect.flagsMap.remove(fname); +// HgDirstate.Record r; +// if (nid1 == null) { +// // normal: added? +// // added: not known at the time of baseRevision, shall report +// // merged: was not known, report as added? +// if ((r = dirstate.checkAdded(fname)) != null) { +// if (r.name2 != null && collect.idsMap.containsKey(r.name2)) { +// collect.idsMap.remove(r.name2); +// collect.idsMap.remove(r.name2); +// inspector.copied(r.name2, fname); +// return; +// } +// // fall-through, report as added +// } else if (dirstate.checkRemoved(fname) != null) { +// // removed: removed file was not known at the time of baseRevision, and we should not report it as removed +// return; +// } +// inspector.added(fname); +// } else { +// // was known; check whether clean or modified +// // when added - seems to be the case of a file added once again, hence need to check if content is different +// if ((r = dirstate.checkNormal(fname)) != null || (r = dirstate.checkMerged(fname)) != null || (r = dirstate.checkAdded(fname)) != null) { +// // either clean or modified +// HgDataFile fileNode = getFileNode(fname); +// final int lengthAtRevision = fileNode.length(nid1); +// if (r.size /* XXX File.length() ?! */ != lengthAtRevision || flags != todoGenerateFlags(fname /*java.io.File*/)) { +// inspector.modified(fname); +// } else { +// // check actual content to see actual changes +// // XXX consider adding HgDataDile.compare(File/byte[]/whatever) operation to optimize comparison +// if (areTheSame(f, fileNode.content(nid1))) { +// inspector.clean(fname); +// } else { +// inspector.modified(fname); +// } +// } +// } +// // only those left in idsMap after processing are reported as removed +// } +// +// // 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 +// // 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 +// // 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: +// // changeset nodeid + hash(actual content) => entry (Nodeid) in the next Manifest +// // 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). +// // The question is whether original Hg treats this case (same content, different parents and hence nodeids) as 'modified' or 'clean' +// } private static String todoGenerateFlags(String fname) { // FIXME implement @@ -416,24 +377,4 @@ } return path; } - - // XXX idsMap is being modified from outside. It's better to let outer (modifying) code to create these maps instead - private static final class ManifestRevisionCollector implements HgManifest.Inspector { - final HashMap idsMap = new HashMap(); - final HashMap flagsMap = new HashMap(); - - public boolean next(Nodeid nid, String fname, String flags) { - idsMap.put(fname, nid); - flagsMap.put(fname, flags); - return true; - } - - public boolean end(int revision) { - return false; - } - - public boolean begin(int revision, Nodeid nid) { - return true; - } - } } diff -r fd4f2c98995b -r 05829a70b30b src/com/tmate/hgkit/ll/StatusCollector.java --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/src/com/tmate/hgkit/ll/StatusCollector.java Mon Jan 17 04:45:09 2011 +0100 @@ -0,0 +1,258 @@ +/* + * Copyright (c) 2011 Artem Tikhomirov + */ +package com.tmate.hgkit.ll; + +import java.util.Collection; +import java.util.Collections; +import java.util.HashMap; +import java.util.LinkedHashMap; +import java.util.LinkedList; +import java.util.List; +import java.util.Map; +import java.util.TreeSet; + +/** + * RevisionWalker? + * @author artem + */ +public class StatusCollector { + + private final HgRepository repo; + private final Map cache; // sparse array, in fact + + public StatusCollector(HgRepository hgRepo) { + this.repo = hgRepo; + cache = new HashMap(); + } + + private ManifestRevisionInspector get(int rev) { + ManifestRevisionInspector i = cache.get(rev); + if (i == null) { + i = new ManifestRevisionInspector(rev, rev); + cache.put(rev, i); + repo.getManifest().walk(rev, rev, i); + } + return i; + } + + public void walk(int rev1, int rev2, Inspector inspector) { + if (rev1 == rev2) { + throw new IllegalArgumentException(); + } + // in fact, rev1 and rev2 are often next (or close) to each other, + // thus, we can optimize Manifest reads here (manifest.walk(rev1, rev2)) + + ManifestRevisionInspector r1, r2; + if (!cache.containsKey(rev1) && !cache.containsKey(rev2) && Math.abs(rev1 - rev2) < 5 /*subjective equivalent of 'close enough'*/) { + int minRev = rev1 < rev2 ? rev1 : rev2; + int maxRev = minRev == rev1 ? rev2 : rev1; + r1 = r2 = new ManifestRevisionInspector(minRev, maxRev); + for (int i = minRev; i <= maxRev; i++) { + cache.put(i, r1); + } + repo.getManifest().walk(minRev, maxRev, r1); + } else { + r1 = get(rev1); + r2 = get(rev2); + } + + TreeSet r1Files = new TreeSet(r1.files(rev1)); + for (String fname : r2.files(rev2)) { + if (r1Files.remove(fname)) { + Nodeid nidR1 = r1.nodeid(rev1, fname); + Nodeid nidR2 = r2.nodeid(rev2, fname); + String flagsR1 = r1.flags(rev1, fname); + String flagsR2 = r2.flags(rev2, fname); + if (nidR1.equals(nidR2) && ((flagsR2 == null && flagsR1 == null) || flagsR2.equals(flagsR1))) { + inspector.clean(fname); + } else { + inspector.modified(fname); + } + } else { + inspector.added(fname); + } + } + for (String left : r1Files) { + inspector.removed(left); + } + // inspector.done() if useful e.g. in UI client + } + + public Record status(int rev1, int rev2) { + Record rv = new Record(); + walk(rev1, rev2, rv); + return rv; + } + + public interface Inspector { + void modified(String fname); + void added(String fname); + void copied(String fnameOrigin, String fnameAdded); // if copied files of no interest, should delegate to self.added(fnameAdded); + void removed(String fname); + void clean(String fname); + void missing(String fname); // aka deleted (tracked by Hg, but not available in FS any more + void unknown(String fname); // not tracked + void ignored(String fname); + } + + // XXX for r1..r2 status, only modified, added, removed (and perhaps, clean) make sense + public static class Record implements Inspector { + private List modified, added, removed, clean, missing, unknown, ignored; + private Map copied; + + public List getModified() { + return proper(modified); + } + + public List getAdded() { + return proper(added); + } + + public List getRemoved() { + return proper(removed); + } + + public Map getCopied() { + if (copied == null) { + return Collections.emptyMap(); + } + return Collections.unmodifiableMap(copied); + } + + public List getClean() { + return proper(clean); + } + + public List getMissing() { + return proper(missing); + } + + public List getUnknown() { + return proper(unknown); + } + + public List getIgnored() { + return proper(ignored); + } + + private List proper(List l) { + if (l == null) { + return Collections.emptyList(); + } + return Collections.unmodifiableList(l); + } + + // + // + + public void modified(String fname) { + modified = doAdd(modified, fname); + } + + public void added(String fname) { + added = doAdd(added, fname); + } + + public void copied(String fnameOrigin, String fnameAdded) { + if (copied == null) { + copied = new LinkedHashMap(); + } + copied.put(fnameOrigin, fnameAdded); + } + + public void removed(String fname) { + removed = doAdd(removed, fname); + } + + public void clean(String fname) { + clean = doAdd(clean, fname); + } + + public void missing(String fname) { + missing = doAdd(missing, fname); + } + + public void unknown(String fname) { + unknown = doAdd(unknown, fname); + } + + public void ignored(String fname) { + ignored = doAdd(ignored, fname); + } + + private static List doAdd(List l, String s) { + if (l == null) { + l = new LinkedList(); + } + l.add(s); + return l; + } + } + + public /*XXX private, actually. Made public unless repo.statusLocal finds better place*/ static final class ManifestRevisionInspector implements HgManifest.Inspector { + private final HashMap[] idsMap; + private final HashMap[] flagsMap; + private final int baseRevision; + private int r = -1; // cursor + + /** + * [minRev, maxRev] + * @param minRev - inclusive + * @param maxRev - inclusive + */ + @SuppressWarnings("unchecked") + public ManifestRevisionInspector(int minRev, int maxRev) { + baseRevision = minRev; + int range = maxRev - minRev + 1; + idsMap = new HashMap[range]; + flagsMap = new HashMap[range]; + } + + public Collection files(int rev) { + if (rev < baseRevision || rev >= baseRevision + idsMap.length) { + throw new IllegalArgumentException(); + } + return idsMap[rev - baseRevision].keySet(); + } + + public Nodeid nodeid(int rev, String fname) { + if (rev < baseRevision || rev >= baseRevision + idsMap.length) { + throw new IllegalArgumentException(); + } + return idsMap[rev - baseRevision].get(fname); + } + + public String flags(int rev, String fname) { + if (rev < baseRevision || rev >= baseRevision + idsMap.length) { + throw new IllegalArgumentException(); + } + return flagsMap[rev - baseRevision].get(fname); + } + + // + + public boolean next(Nodeid nid, String fname, String flags) { + idsMap[r].put(fname, nid); + flagsMap[r].put(fname, flags); + return true; + } + + public boolean end(int revision) { + assert revision == r + baseRevision; + r = -1; + return revision+1 < baseRevision + idsMap.length; + } + + public boolean begin(int revision, Nodeid nid) { + if (revision < baseRevision || revision >= baseRevision + idsMap.length) { + throw new IllegalArgumentException(); + } + r = revision - baseRevision; + idsMap[r] = new HashMap(); + flagsMap[r] = new HashMap(); + return true; + } + } + +}