Make hg verify do more thorough cross-checking.

hg

@@ -1,261 +1,279 b''
 #!/usr/bin/env python
 #
 # mercurial - a minimal scalable distributed SCM
 # v0.4d "oedipa maas"
 #
 # Copyright 2005 Matt Mackall <[e-mail]>
 #
 # This software may be used and distributed according to the terms
 # of the GNU General Public License, incorporated herein by reference.
 
 # the psyco compiler makes commits about twice as fast
 try:
     import psyco
     psyco.full()
 except:
     pass
 
 import sys, os
 from mercurial import hg, mdiff, fancyopts
 
 options = {}
 opts = [('v', 'verbose', None, 'verbose'),
         ('d', 'debug', None, 'debug')]
 
 args = fancyopts.fancyopts(sys.argv[1:], opts, options,
                            'hg [options] <command> [command options] [files]')
 
 try:
     cmd = args[0]
     args = args[1:]
 except:
     cmd = ""
 
 ui = hg.ui(options["verbose"], options["debug"])
     
 if cmd == "init":
     repo = hg.repository(ui, ".", create=1)
     sys.exit(0)
 elif cmd == "branch" or cmd == "clone":
     os.system("cp -al %s/.hg .hg" % args[0])
     sys.exit(0)
 else:
     repo = hg.repository(ui=ui)
 
 if cmd == "checkout" or cmd == "co":
     node = repo.changelog.tip()
     if len(args):
         if len(args[0]) < 40:
             rev = int(args[0])
             node = repo.changelog.node(rev)
         else:
             node = args[0]
     repo.checkout(node)
 
 elif cmd == "add":
     repo.add(args)
 
 elif cmd == "remove" or cmd == "rm" or cmd == "del" or cmd == "delete":
     repo.remove(args)
 
 elif cmd == "commit" or cmd == "checkin" or cmd == "ci":
     if 1:
         if len(args) > 0:
             repo.commit(args)
         else:
             repo.commit()
 
 elif cmd == "import" or cmd == "patch":
     ioptions = {}
     opts = [('p', 'strip', 1, 'path strip'),
             ('b', 'base', "", 'base path')]
     
     args = fancyopts.fancyopts(args, opts, ioptions,
                               'hg import [options] <patch names>')
     d = ioptions["base"]
     strip = ioptions["strip"]
 
     for patch in args:
         ui.status("applying %s\n" % patch)
         pf = d + patch
         os.system("patch -p%d < %s > /dev/null" % (strip, pf))
         f = os.popen("lsdiff --strip %d %s" % (strip, pf))
         files = f.read().splitlines()
         f.close()
         repo.commit(files)
 
 elif cmd == "status":
     (c, a, d) = repo.diffdir(repo.root)
     for f in c: print "C", f
     for f in a: print "?", f
     for f in d: print "R", f
 
 elif cmd == "diff":
     mmap = {}
     if repo.current:
         change = repo.changelog.read(repo.current)
         mmap = repo.manifest.read(change[0])
 
     (c, a, d) = repo.diffdir(repo.root)
     for f in c:
         to = repo.file(f).read(mmap[f])
         tn = file(f).read()
         sys.stdout.write(mdiff.unidiff(to, tn, f))
     for f in a:
         to = ""
         tn = file(f).read()
         sys.stdout.write(mdiff.unidiff(to, tn, f))
     for f in d:
         to = repo.file(f).read(mmap[f])
         tn = ""
         sys.stdout.write(mdiff.unidiff(to, tn, f))
 
 elif cmd == "addremove":
     (c, a, d) = repo.diffdir(repo.root)
     repo.add(a)
     repo.remove(d)
     
 elif cmd == "history":
     for i in range(repo.changelog.count()):
         n = repo.changelog.node(i)
         changes = repo.changelog.read(n)
         (p1, p2) = repo.changelog.parents(n)
         (h, h1, h2) = map(hg.hex, (n, p1, p2))
         (i1, i2) = map(repo.changelog.rev, (p1, p2))
         print "rev:      %4d:%s" % (i, h)
         print "parents:  %4d:%s" % (i1, h1)
         if i2: print "          %4d:%s" % (i2, h2)
         print "manifest: %4d:%s" % (repo.manifest.rev(changes[0]),
                                     hg.hex(changes[0]))
         print "user:", changes[1]
         print "files:", " ".join(changes[3])
         print "description:"
         print changes[4]
 
 elif cmd == "log":
     if args:
         r = repo.file(args[0])
         for i in range(r.count()):
             n = r.node(i)
             (p1, p2) = r.parents(n)
             (h, h1, h2) = map(hg.hex, (n, p1, p2))
             (i1, i2) = map(r.rev, (p1, p2))
             cr = r.linkrev(n)
             cn = hg.hex(repo.changelog.node(cr))
             print "rev:       %4d:%s" % (i, h)
             print "changeset: %4d:%s" % (cr, cn)
             print "parents:   %4d:%s" % (i1, h1)
             if i2: print "           %4d:%s" % (i2, h2)
     else:
         print "missing filename"
 
 elif cmd == "dump":
     if args:
         r = repo.file(args[0])
         n = r.tip()
         if len(args) > 1: n = hg.bin(args[1])
         sys.stdout.write(r.read(n))
     else:
         print "missing filename"
 
 elif cmd == "dumpmanifest":
     n = repo.manifest.tip()
     if len(args) > 0:
         n = hg.bin(args[0])
     m = repo.manifest.read(n)
     files = m.keys()
     files.sort()
 
     for f in files:
         print hg.hex(m[f]), f
 
 elif cmd == "merge":
     if args:
         other = hg.repository(ui, args[0])
         repo.merge(other)
     else:
         print "missing source repository"
 
 elif cmd == "verify":
     filelinkrevs = {}
     filenodes = {}
     manifestchangeset = {}
     changesets = revisions = files = 0
 
     print "checking changesets"
     for i in range(repo.changelog.count()):
         changesets += 1
         n = repo.changelog.node(i)
         changes = repo.changelog.read(n)
         manifestchangeset[changes[0]] = n
         for f in changes[3]:
             revisions += 1
             filelinkrevs.setdefault(f, []).append(i)
 
     print "checking manifests"
     for i in range(repo.manifest.count()):
         n = repo.manifest.node(i)
         ca = repo.changelog.node(repo.manifest.linkrev(n))
         cc = manifestchangeset[n]
         if ca != cc:
             print "manifest %s points to %s, not %s" % \
                   (hg.hex(n), hg.hex(ca), hg.hex(cc))
         m = repo.manifest.read(n)
         for f, fn in m.items():
             filenodes.setdefault(f, {})[fn] = 1
 
     print "crosschecking files in changesets and manifests"
     for f in filenodes:
         if f not in filelinkrevs:
             print "file %s in manifest but not in changesets"
 
     for f in filelinkrevs:
         if f not in filenodes:
             print "file %s in changeset but not in manifest"
 
     print "checking files"
     for f in filenodes:
         files += 1
         fl = repo.file(f)
         nodes = {"\0"*20: 1}
         for i in range(fl.count()):
             n = fl.node(i)
+
             if n not in filenodes[f]:
                 print "%s:%s not in manifests" % (f, hg.hex(n))
-            if fl.linkrev(n) not in filelinkrevs[f]:
+            else:
+                del filenodes[f][n]
+
+            flr = fl.linkrev(n)
+            if flr not in filelinkrevs[f]:
                 print "%s:%s points to unexpected changeset rev %d" \
                       % (f, hg.hex(n), fl.linkrev(n))
+            else:
+                filelinkrevs[f].remove(flr)
+
+            # verify contents
             t = fl.read(n)
+
+            # verify parents
             (p1, p2) = fl.parents(n)
             if p1 not in nodes:
                 print "%s:%s unknown parent 1 %s" % (f, hg.hex(n), hg.hex(p1))
             if p2 not in nodes:
                 print "file %s:%s unknown parent %s" % (f, hg.hex(n), hg.hex(p1))
+            nodes[n] = 1
 
-            nodes[n] = 1
+        # cross-check
+        for flr in filelinkrevs[f]:
+            print "changeset rev %d not in %s" % (flr, f)
+            
+        for node in filenodes[f]:
+            print "node %s in manifests not in %s" % (hg.hex(n), f)
+
 
     print "%d files, %d changesets, %d total revisions" % (files, changesets,
                                                            revisions)
     
 else:
     if cmd != "help":
         print "unknown command\n"
 
     print """\
  commands:
 
  init                  create a new repository in this directory
  branch <path>         create a branch of <path> in this directory
  merge <path>          merge changes from <path> into local repository
  checkout [changeset]  checkout the latest or given changeset
  status                show new, missing, and changed files in working dir
  add [files...]        add the given files in the next commit
  remove [files...]     remove the given files in the next commit
  addremove             add all new files, delete all missing files
  commit                commit all changes to the repository
  history               show changeset history
  log <file>            show revision history of a single file
  dump <file> [rev]     dump the latest or given revision of a file
  dumpmanifest [rev]    dump the latest or given revision of the manifest
 """
     sys.exit(1)