fix file handling bugs on windows.

add util.posixfile class that has posix semantics on windows.
fix util.rename so it works with stupid windows delete semantics.

mercurial/appendfile.py

@@ -1,154 +1,156
 # appendfile.py - special classes to make repo updates atomic
 #
 # Copyright 2006 Vadim Gelfer <[e-mail]>
 #
 # This software may be used and distributed according to the terms
 # of the GNU General Public License, incorporated herein by reference.
 
 from demandload import *
-demandload(globals(), "cStringIO changelog errno manifest os tempfile")
+demandload(globals(), "cStringIO changelog errno manifest os tempfile util")
 
 # writes to metadata files are ordered.  reads: changelog, manifest,
 # normal files.  writes: normal files, manifest, changelog.
 
 # manifest contains pointers to offsets in normal files.  changelog
 # contains pointers to offsets in manifest.  if reader reads old
 # changelog while manifest or normal files are written, it has no
 # pointers into new parts of those files that are maybe not consistent
 # yet, so will not read them.
 
 # localrepo.addchangegroup thinks it writes changelog first, then
 # manifest, then normal files (this is order they are available, and
 # needed for computing linkrev fields), but uses appendfile to hide
 # updates from readers.  data not written to manifest or changelog
 # until all normal files updated.  write manifest first, then
 # changelog.
 
 # with this write ordering, readers cannot see inconsistent view of
 # repo during update.
 
 class appendfile(object):
     '''implement enough of file protocol to append to revlog file.
     appended data is written to temp file.  reads and seeks span real
     file and temp file.  readers cannot see appended data until
     writedata called.'''
 
     def __init__(self, fp, tmpname):
         if tmpname:
             self.tmpname = tmpname
-            self.tmpfp = open(self.tmpname, 'ab+')
+            self.tmpfp = util.posixfile(self.tmpname, 'ab+')
         else:
             fd, self.tmpname = tempfile.mkstemp()
-            self.tmpfp = os.fdopen(fd, 'ab+')
+            os.close(fd)
+            self.tmpfp = util.posixfile(self.tmpname, 'ab+')
         self.realfp = fp
         self.offset = fp.tell()
         # real file is not written by anyone else. cache its size so
         # seek and read can be fast.
-        self.realsize = os.fstat(fp.fileno()).st_size
+        self.realsize = util.fstat(fp).st_size
+        self.name = fp.name
 
     def end(self):
         self.tmpfp.flush() # make sure the stat is correct
-        return self.realsize + os.fstat(self.tmpfp.fileno()).st_size
+        return self.realsize + util.fstat(self.tmpfp).st_size
 
     def tell(self):
         return self.offset
 
     def flush(self):
         self.tmpfp.flush()
 
     def close(self):
         self.realfp.close()
         self.tmpfp.close()
 
     def seek(self, offset, whence=0):
         '''virtual file offset spans real file and temp file.'''
         if whence == 0:
             self.offset = offset
         elif whence == 1:
             self.offset += offset
         elif whence == 2:
             self.offset = self.end() + offset
 
         if self.offset < self.realsize:
             self.realfp.seek(self.offset)
         else:
             self.tmpfp.seek(self.offset - self.realsize)
 
     def read(self, count=-1):
         '''only trick here is reads that span real file and temp file.'''
         fp = cStringIO.StringIO()
         old_offset = self.offset
         if self.offset < self.realsize:
             s = self.realfp.read(count)
             fp.write(s)
             self.offset += len(s)
             if count > 0:
                 count -= len(s)
         if count != 0:
             if old_offset != self.offset:
                 self.tmpfp.seek(self.offset - self.realsize)
             s = self.tmpfp.read(count)
             fp.write(s)
             self.offset += len(s)
         return fp.getvalue()
 
     def write(self, s):
         '''append to temp file.'''
         self.tmpfp.seek(0, 2)
         self.tmpfp.write(s)
         # all writes are appends, so offset must go to end of file.
         self.offset = self.realsize + self.tmpfp.tell()
 
 class appendopener(object):
     '''special opener for files that only read or append.'''
 
     def __init__(self, opener):
         self.realopener = opener
         # key: file name, value: appendfile name
         self.tmpnames = {}
 
     def __call__(self, name, mode='r'):
         '''open file.'''
 
         assert mode in 'ra+'
         try:
             realfp = self.realopener(name, 'r')
         except IOError, err:
             if err.errno != errno.ENOENT: raise
             realfp = self.realopener(name, 'w+')
         tmpname = self.tmpnames.get(name)
         fp = appendfile(realfp, tmpname)
         if tmpname is None:
             self.tmpnames[name] = fp.tmpname
         return fp
 
     def writedata(self):
         '''copy data from temp files to real files.'''
         # write .d file before .i file.
         tmpnames = self.tmpnames.items()
         tmpnames.sort()
         for name, tmpname in tmpnames:
             fp = open(tmpname, 'rb')
             s = fp.read()
             fp.close()
             os.unlink(tmpname)
             fp = self.realopener(name, 'a')
             fp.write(s)
             fp.close()
 
 # files for changelog and manifest are in different appendopeners, so
 # not mixed up together.
 
 class appendchangelog(changelog.changelog, appendopener):
     def __init__(self, opener, version):
         appendopener.__init__(self, opener)
         changelog.changelog.__init__(self, self, version)
     def checkinlinesize(self, fp, tr):
         return
 
 class appendmanifest(manifest.manifest, appendopener):
     def __init__(self, opener, version):
         appendopener.__init__(self, opener)
         manifest.manifest.__init__(self, self, version)
     def checkinlinesize(self, fp, tr):
         return

mercurial/bundlerepo.py

@@ -1,207 +1,207
 """
 bundlerepo.py - repository class for viewing uncompressed bundles
 
 This provides a read-only repository interface to bundles as if
 they were part of the actual repository.
 
 Copyright 2006 Benoit Boissinot <[e-mail]>
 
 This software may be used and distributed according to the terms
 of the GNU General Public License, incorporated herein by reference.
 """
 
 from node import *
 from i18n import gettext as _
 from demandload import demandload
 demandload(globals(), "changegroup util os struct")
 
 import localrepo, changelog, manifest, filelog, revlog
 
 class bundlerevlog(revlog.revlog):
     def __init__(self, opener, indexfile, datafile, bundlefile,
                  linkmapper=None):
         # How it works:
         # to retrieve a revision, we need to know the offset of
         # the revision in the bundlefile (an opened file).
         #
         # We store this offset in the index (start), to differentiate a
         # rev in the bundle and from a rev in the revlog, we check
         # len(index[r]). If the tuple is bigger than 7, it is a bundle
         # (it is bigger since we store the node to which the delta is)
         #
         revlog.revlog.__init__(self, opener, indexfile, datafile)
         self.bundlefile = bundlefile
         self.basemap = {}
         def chunkpositer():
             for chunk in changegroup.chunkiter(bundlefile):
                 pos = bundlefile.tell()
                 yield chunk, pos - len(chunk)
         n = self.count()
         prev = None
         for chunk, start in chunkpositer():
             size = len(chunk)
             if size < 80:
                 raise util.Abort("invalid changegroup")
             start += 80
             size -= 80
             node, p1, p2, cs = struct.unpack("20s20s20s20s", chunk[:80])
             if node in self.nodemap:
                 prev = node
                 continue
             for p in (p1, p2):
                 if not p in self.nodemap:
                     raise RevlogError(_("unknown parent %s") % short(p1))
             if linkmapper is None:
                 link = n
             else:
                 link = linkmapper(cs)
 
             if not prev:
                 prev = p1
             # start, size, base is not used, link, p1, p2, delta ref
             if self.version == 0:
                 e = (start, size, None, link, p1, p2, node)
             else:
                 e = (self.offset_type(start, 0), size, -1, None, link,
                      self.rev(p1), self.rev(p2), node)
             self.basemap[n] = prev
             self.index.append(e)
             self.nodemap[node] = n
             prev = node
             n += 1
 
     def bundle(self, rev):
         """is rev from the bundle"""
         if rev < 0:
             return False
         return rev in self.basemap
     def bundlebase(self, rev): return self.basemap[rev]
     def chunk(self, rev, df=None):
         # Warning: in case of bundle, the diff is against bundlebase,
         # not against rev - 1
         # XXX: could use some caching
         if not self.bundle(rev):
             return revlog.revlog.chunk(self, rev)
         self.bundlefile.seek(self.start(rev))
         return self.bundlefile.read(self.length(rev))
 
     def revdiff(self, rev1, rev2):
         """return or calculate a delta between two revisions"""
         if self.bundle(rev1) and self.bundle(rev2):
             # hot path for bundle
             revb = self.rev(self.bundlebase(rev2))
             if revb == rev1:
                 return self.chunk(rev2)
         elif not self.bundle(rev1) and not self.bundle(rev2):
             return revlog.revlog.chunk(self, rev1, rev2)
 
         return self.diff(self.revision(self.node(rev1)),
                          self.revision(self.node(rev2)))
 
     def revision(self, node):
         """return an uncompressed revision of a given"""
         if node == nullid: return ""
 
         text = None
         chain = []
         iter_node = node
         rev = self.rev(iter_node)
         # reconstruct the revision if it is from a changegroup
         while self.bundle(rev):
             if self.cache and self.cache[0] == iter_node:
                 text = self.cache[2]
                 break
             chain.append(rev)
             iter_node = self.bundlebase(rev)
             rev = self.rev(iter_node)
         if text is None:
             text = revlog.revlog.revision(self, iter_node)
 
         while chain:
             delta = self.chunk(chain.pop())
             text = self.patches(text, [delta])
 
         p1, p2 = self.parents(node)
         if node != revlog.hash(text, p1, p2):
             raise RevlogError(_("integrity check failed on %s:%d")
                           % (self.datafile, self.rev(node)))
 
         self.cache = (node, self.rev(node), text)
         return text
 
     def addrevision(self, text, transaction, link, p1=None, p2=None, d=None):
         raise NotImplementedError
     def addgroup(self, revs, linkmapper, transaction, unique=0):
         raise NotImplementedError
     def strip(self, rev, minlink):
         raise NotImplementedError
     def checksize(self):
         raise NotImplementedError
 
 class bundlechangelog(bundlerevlog, changelog.changelog):
     def __init__(self, opener, bundlefile):
         changelog.changelog.__init__(self, opener)
         bundlerevlog.__init__(self, opener, "00changelog.i", "00changelog.d",
                               bundlefile)
 
 class bundlemanifest(bundlerevlog, manifest.manifest):
     def __init__(self, opener, bundlefile, linkmapper):
         manifest.manifest.__init__(self, opener)
         bundlerevlog.__init__(self, opener, self.indexfile, self.datafile,
                               bundlefile, linkmapper)
 
 class bundlefilelog(bundlerevlog, filelog.filelog):
     def __init__(self, opener, path, bundlefile, linkmapper):
         filelog.filelog.__init__(self, opener, path)
         bundlerevlog.__init__(self, opener, self.indexfile, self.datafile,
                               bundlefile, linkmapper)
 
 class bundlerepository(localrepo.localrepository):
     def __init__(self, ui, path, bundlename):
         localrepo.localrepository.__init__(self, ui, path)
         f = open(bundlename, "rb")
-        s = os.fstat(f.fileno())
+        s = util.fstat(f)
         self.bundlefile = f
         header = self.bundlefile.read(6)
         if not header.startswith("HG"):
             raise util.Abort(_("%s: not a Mercurial bundle file") % bundlename)
         elif not header.startswith("HG10"):
             raise util.Abort(_("%s: unknown bundle version") % bundlename)
         elif header == "HG10BZ":
             raise util.Abort(_("%s: compressed bundle not supported")
                              % bundlename)
         elif header == "HG10UN":
             # uncompressed bundle supported
             pass
         else:
             raise util.Abort(_("%s: unknown bundle compression type")
                              % bundlename)
         self.changelog = bundlechangelog(self.opener, self.bundlefile)
         self.manifest = bundlemanifest(self.opener, self.bundlefile,
                                        self.changelog.rev)
         # dict with the mapping 'filename' -> position in the bundle
         self.bundlefilespos = {}
         while 1:
             f = changegroup.getchunk(self.bundlefile)
             if not f:
                 break
             self.bundlefilespos[f] = self.bundlefile.tell()
             for c in changegroup.chunkiter(self.bundlefile):
                 pass
 
     def dev(self):
         return -1
 
     def file(self, f):
         if f[0] == '/':
             f = f[1:]
         if f in self.bundlefilespos:
             self.bundlefile.seek(self.bundlefilespos[f])
             return bundlefilelog(self.opener, f, self.bundlefile,
                                  self.changelog.rev)
         else:
             return filelog.filelog(self.opener, f)
 
     def close(self):
         """Close assigned bundle file immediately."""
         self.bundlefile.close()

mercurial/revlog.py

@@ -1,1241 +1,1241
 """
 revlog.py - storage back-end for mercurial
 
 This provides efficient delta storage with O(1) retrieve and append
 and O(changes) merge between branches
 
 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.
 """
 
 from node import *
 from i18n import gettext as _
 from demandload import demandload
 demandload(globals(), "binascii changegroup errno heapq mdiff os")
-demandload(globals(), "sha struct zlib")
+demandload(globals(), "sha struct util zlib")
 
 # revlog version strings
 REVLOGV0 = 0
 REVLOGNG = 1
 
 # revlog flags
 REVLOGNGINLINEDATA = (1 << 16)
 
 def flagstr(flag):
     if flag == "inline":
         return REVLOGNGINLINEDATA
     raise RevlogError(_("unknown revlog flag %s" % flag))
 
 def hash(text, p1, p2):
     """generate a hash from the given text and its parent hashes
 
     This hash combines both the current file contents and its history
     in a manner that makes it easy to distinguish nodes with the same
     content in the revision graph.
     """
     l = [p1, p2]
     l.sort()
     s = sha.new(l[0])
     s.update(l[1])
     s.update(text)
     return s.digest()
 
 def compress(text):
     """ generate a possibly-compressed representation of text """
     if not text: return ("", text)
     if len(text) < 44:
         if text[0] == '\0': return ("", text)
         return ('u', text)
     bin = zlib.compress(text)
     if len(bin) > len(text):
         if text[0] == '\0': return ("", text)
         return ('u', text)
     return ("", bin)
 
 def decompress(bin):
     """ decompress the given input """
     if not bin: return bin
     t = bin[0]
     if t == '\0': return bin
     if t == 'x': return zlib.decompress(bin)
     if t == 'u': return bin[1:]
     raise RevlogError(_("unknown compression type %r") % t)
 
 indexformatv0 = ">4l20s20s20s"
 v0shaoffset = 56
 # index ng:
 # 6 bytes offset
 # 2 bytes flags
 # 4 bytes compressed length
 # 4 bytes uncompressed length
 # 4 bytes: base rev
 # 4 bytes link rev
 # 4 bytes parent 1 rev
 # 4 bytes parent 2 rev
 # 32 bytes: nodeid
 indexformatng = ">Qiiiiii20s12x"
 ngshaoffset = 32
 versionformat = ">i"
 
 class lazyparser(object):
     """
     this class avoids the need to parse the entirety of large indices
     """
     def __init__(self, dataf, size, indexformat, shaoffset):
         self.dataf = dataf
         self.format = indexformat
         self.s = struct.calcsize(indexformat)
         self.indexformat = indexformat
         self.datasize = size
         self.l = size/self.s
         self.index = [None] * self.l
         self.map = {nullid: -1}
         self.allmap = 0
         self.all = 0
         self.mapfind_count = 0
         self.shaoffset = shaoffset
 
     def loadmap(self):
         """
         during a commit, we need to make sure the rev being added is
         not a duplicate.  This requires loading the entire index,
         which is fairly slow.  loadmap can load up just the node map,
         which takes much less time.
         """
         if self.allmap: return
         start = 0
         end = self.datasize
         self.allmap = 1
         cur = 0
         count = 0
         blocksize = self.s * 256
         self.dataf.seek(0)
         while cur < end:
             data = self.dataf.read(blocksize)
             off = 0
             for x in xrange(256):
                 n = data[off + self.shaoffset:off + self.shaoffset + 20]
                 self.map[n] = count
                 count += 1
                 if count >= self.l:
                     break
                 off += self.s
             cur += blocksize
 
     def loadblock(self, blockstart, blocksize, data=None):
         if self.all: return
         if data is None:
             self.dataf.seek(blockstart)
             data = self.dataf.read(blocksize)
         lend = len(data) / self.s
         i = blockstart / self.s
         off = 0
         for x in xrange(lend):
             if self.index[i + x] == None:
                 b = data[off : off + self.s]
                 self.index[i + x] = b
                 n = b[self.shaoffset:self.shaoffset + 20]
                 self.map[n] = i + x
             off += self.s
 
     def findnode(self, node):
         """search backwards through the index file for a specific node"""
         if self.allmap: return None
 
         # hg log will cause many many searches for the manifest
         # nodes.  After we get called a few times, just load the whole
         # thing.
         if self.mapfind_count > 8:
             self.loadmap()
             if node in self.map:
                 return node
             return None
         self.mapfind_count += 1
         last = self.l - 1
         while self.index[last] != None:
             if last == 0:
                 self.all = 1
                 self.allmap = 1
                 return None
             last -= 1
         end = (last + 1) * self.s
         blocksize = self.s * 256
         while end >= 0:
             start = max(end - blocksize, 0)
             self.dataf.seek(start)
             data = self.dataf.read(end - start)
             findend = end - start
             while True:
                 # we're searching backwards, so weh have to make sure
                 # we don't find a changeset where this node is a parent
                 off = data.rfind(node, 0, findend)
                 findend = off
                 if off >= 0:
                     i = off / self.s
                     off = i * self.s
                     n = data[off + self.shaoffset:off + self.shaoffset + 20]
                     if n == node:
                         self.map[n] = i + start / self.s
                         return node
                 else:
                     break
             end -= blocksize
         return None
 
     def loadindex(self, i=None, end=None):
         if self.all: return
         all = False
         if i == None:
             blockstart = 0
             blocksize = (512 / self.s) * self.s
             end = self.datasize
             all = True
         else:
             if end:
                 blockstart = i * self.s
                 end = end * self.s
                 blocksize = end - blockstart
             else:
                 blockstart = (i & ~(32)) * self.s
                 blocksize = self.s * 64
                 end = blockstart + blocksize
         while blockstart < end:
             self.loadblock(blockstart, blocksize)
             blockstart += blocksize
         if all: self.all = True
 
 class lazyindex(object):
     """a lazy version of the index array"""
     def __init__(self, parser):
         self.p = parser
     def __len__(self):
         return len(self.p.index)
     def load(self, pos):
         if pos < 0:
             pos += len(self.p.index)
         self.p.loadindex(pos)
         return self.p.index[pos]
     def __getitem__(self, pos):
         ret = self.p.index[pos] or self.load(pos)
         if isinstance(ret, str):
             ret = struct.unpack(self.p.indexformat, ret)
         return ret
     def __setitem__(self, pos, item):
         self.p.index[pos] = item
     def __delitem__(self, pos):
         del self.p.index[pos]
     def append(self, e):
         self.p.index.append(e)
 
 class lazymap(object):
     """a lazy version of the node map"""
     def __init__(self, parser):
         self.p = parser
     def load(self, key):
         n = self.p.findnode(key)
         if n == None:
             raise KeyError(key)
     def __contains__(self, key):
         if key in self.p.map:
             return True
         self.p.loadmap()
         return key in self.p.map
     def __iter__(self):
         yield nullid
         for i in xrange(self.p.l):
             ret = self.p.index[i]
             if not ret:
                 self.p.loadindex(i)
                 ret = self.p.index[i]
             if isinstance(ret, str):
                 ret = struct.unpack(self.p.indexformat, ret)
             yield ret[-1]
     def __getitem__(self, key):
         try:
             return self.p.map[key]
         except KeyError:
             try:
                 self.load(key)
                 return self.p.map[key]
             except KeyError:
                 raise KeyError("node " + hex(key))
     def __setitem__(self, key, val):
         self.p.map[key] = val
     def __delitem__(self, key):
         del self.p.map[key]
 
 class RevlogError(Exception): pass
 
 class revlog(object):
     """
     the underlying revision storage object
 
     A revlog consists of two parts, an index and the revision data.
 
     The index is a file with a fixed record size containing
     information on each revision, includings its nodeid (hash), the
     nodeids of its parents, the position and offset of its data within
     the data file, and the revision it's based on. Finally, each entry
     contains a linkrev entry that can serve as a pointer to external
     data.
 
     The revision data itself is a linear collection of data chunks.
     Each chunk represents a revision and is usually represented as a
     delta against the previous chunk. To bound lookup time, runs of
     deltas are limited to about 2 times the length of the original
     version data. This makes retrieval of a version proportional to
     its size, or O(1) relative to the number of revisions.
 
     Both pieces of the revlog are written to in an append-only
     fashion, which means we never need to rewrite a file to insert or
     remove data, and can use some simple techniques to avoid the need
     for locking while reading.
     """
     def __init__(self, opener, indexfile, datafile, defversion=0):
         """
         create a revlog object
 
         opener is a function that abstracts the file opening operation
         and can be used to implement COW semantics or the like.
         """
         self.indexfile = indexfile
         self.datafile = datafile
         self.opener = opener
 
         self.indexstat = None
         self.cache = None
         self.chunkcache = None
         self.defversion = defversion
         self.load()
 
     def load(self):
         v = self.defversion
         try:
             f = self.opener(self.indexfile)
             i = f.read(4)
             f.seek(0)
         except IOError, inst:
             if inst.errno != errno.ENOENT:
                 raise
             i = ""
         else:
             try:
-                st = os.fstat(f.fileno())
+                st = util.fstat(f)
             except AttributeError, inst:
                 st = None
             else:
                 oldst = self.indexstat
                 if (oldst and st.st_dev == oldst.st_dev
                     and st.st_ino == oldst.st_ino
                     and st.st_mtime == oldst.st_mtime
                     and st.st_ctime == oldst.st_ctime):
                     return
                 self.indexstat = st
                 if len(i) > 0:
                     v = struct.unpack(versionformat, i)[0]
         flags = v & ~0xFFFF
         fmt = v & 0xFFFF
         if fmt == 0:
             if flags:
                 raise RevlogError(_("index %s invalid flags %x for format v0" %
                                    (self.indexfile, flags)))
         elif fmt == REVLOGNG:
             if flags & ~REVLOGNGINLINEDATA:
                 raise RevlogError(_("index %s invalid flags %x for revlogng" %
                                    (self.indexfile, flags)))
         else:
             raise RevlogError(_("index %s invalid format %d" %
                                (self.indexfile, fmt)))
         self.version = v
         if v == 0:
             self.indexformat = indexformatv0
             shaoffset = v0shaoffset
         else:
             self.indexformat = indexformatng
             shaoffset = ngshaoffset
 
         if i:
             if not self.inlinedata() and st and st.st_size > 10000:
                 # big index, let's parse it on demand
                 parser = lazyparser(f, st.st_size, self.indexformat, shaoffset)
                 self.index = lazyindex(parser)
                 self.nodemap = lazymap(parser)
             else:
                 i = f.read()
                 self.parseindex(i)
             if self.inlinedata():
                 # we've already got the entire data file read in, save it
                 # in the chunk data
                 self.chunkcache = (0, i)
             if self.version != 0:
                 e = list(self.index[0])
                 type = self.ngtype(e[0])
                 e[0] = self.offset_type(0, type)
                 self.index[0] = e
         else:
             self.nodemap = { nullid: -1}
             self.index = []
 
 
     def parseindex(self, data):
         s = struct.calcsize(self.indexformat)
         l = len(data)
         self.index = []
         self.nodemap =  {nullid: -1}
         inline = self.inlinedata()
         off = 0
         n = 0
         while off < l:
             e = struct.unpack(self.indexformat, data[off:off + s])
             self.index.append(e)
             self.nodemap[e[-1]] = n
             n += 1
             off += s
             if inline:
                 off += e[1]
 
     def ngoffset(self, q):
         if q & 0xFFFF:
             raise RevlogError(_('%s: incompatible revision flag %x') %
                                (self.indexfile, type))
         return long(q >> 16)
 
     def ngtype(self, q):
         return int(q & 0xFFFF)
 
     def offset_type(self, offset, type):
         return long(long(offset) << 16 | type)
 
     def loadindex(self, start, end):
         """load a block of indexes all at once from the lazy parser"""
         if isinstance(self.index, lazyindex):
             self.index.p.loadindex(start, end)
 
     def loadindexmap(self):
         """loads both the map and the index from the lazy parser"""
         if isinstance(self.index, lazyindex):
             p = self.index.p
             p.loadindex()
             self.nodemap = p.map
 
     def loadmap(self):
         """loads the map from the lazy parser"""
         if isinstance(self.nodemap, lazymap):
             self.nodemap.p.loadmap()
             self.nodemap = self.nodemap.p.map
 
     def inlinedata(self): return self.version & REVLOGNGINLINEDATA
     def tip(self): return self.node(len(self.index) - 1)
     def count(self): return len(self.index)
     def node(self, rev):
         return (rev < 0) and nullid or self.index[rev][-1]
     def rev(self, node):
         try:
             return self.nodemap[node]
         except KeyError:
             raise RevlogError(_('%s: no node %s') % (self.indexfile, hex(node)))
     def linkrev(self, node): return self.index[self.rev(node)][-4]
     def parents(self, node):
         if node == nullid: return (nullid, nullid)
         r = self.rev(node)
         d = self.index[r][-3:-1]
         if self.version == 0:
             return d
         return [ self.node(x) for x in d ]
     def start(self, rev):
         if rev < 0:
             return -1
         if self.version != 0:
             return self.ngoffset(self.index[rev][0])
         return self.index[rev][0]
 
     def end(self, rev): return self.start(rev) + self.length(rev)
 
     def size(self, rev):
         """return the length of the uncompressed text for a given revision"""
         l = -1
         if self.version != 0:
             l = self.index[rev][2]
         if l >= 0:
             return l
 
         t = self.revision(self.node(rev))
         return len(t)
 
         # alternate implementation, The advantage to this code is it
         # will be faster for a single revision.  But, the results are not
         # cached, so finding the size of every revision will be slower.
         """
         if self.cache and self.cache[1] == rev:
             return len(self.cache[2])
 
         base = self.base(rev)
         if self.cache and self.cache[1] >= base and self.cache[1] < rev:
             base = self.cache[1]
             text = self.cache[2]
         else:
             text = self.revision(self.node(base))
 
         l = len(text)
         for x in xrange(base + 1, rev + 1):
             l = mdiff.patchedsize(l, self.chunk(x))
         return l
         """
 
     def length(self, rev):
         if rev < 0:
             return 0
         else:
             return self.index[rev][1]
     def base(self, rev): return (rev < 0) and rev or self.index[rev][-5]
 
     def reachable(self, rev, stop=None):
         reachable = {}
         visit = [rev]
         reachable[rev] = 1
         if stop:
             stopn = self.rev(stop)
         else:
             stopn = 0
         while visit:
             n = visit.pop(0)
             if n == stop:
                 continue
             if n == nullid:
                 continue
             for p in self.parents(n):
                 if self.rev(p) < stopn:
                     continue
                 if p not in reachable:
                     reachable[p] = 1
                     visit.append(p)
         return reachable
 
     def nodesbetween(self, roots=None, heads=None):
         """Return a tuple containing three elements. Elements 1 and 2 contain
         a final list bases and heads after all the unreachable ones have been
         pruned.  Element 0 contains a topologically sorted list of all
 
         nodes that satisfy these constraints:
         1. All nodes must be descended from a node in roots (the nodes on
            roots are considered descended from themselves).
         2. All nodes must also be ancestors of a node in heads (the nodes in
            heads are considered to be their own ancestors).
 
         If roots is unspecified, nullid is assumed as the only root.
         If heads is unspecified, it is taken to be the output of the
         heads method (i.e. a list of all nodes in the repository that
         have no children)."""
         nonodes = ([], [], [])
         if roots is not None:
             roots = list(roots)
             if not roots:
                 return nonodes
             lowestrev = min([self.rev(n) for n in roots])
         else:
             roots = [nullid] # Everybody's a descendent of nullid
             lowestrev = -1
         if (lowestrev == -1) and (heads is None):
             # We want _all_ the nodes!
             return ([self.node(r) for r in xrange(0, self.count())],
                     [nullid], list(self.heads()))
         if heads is None:
             # All nodes are ancestors, so the latest ancestor is the last
             # node.
             highestrev = self.count() - 1
             # Set ancestors to None to signal that every node is an ancestor.
             ancestors = None
             # Set heads to an empty dictionary for later discovery of heads
             heads = {}
         else:
             heads = list(heads)
             if not heads:
                 return nonodes
             ancestors = {}
             # Start at the top and keep marking parents until we're done.
             nodestotag = heads[:]
             # Turn heads into a dictionary so we can remove 'fake' heads.
             # Also, later we will be using it to filter out the heads we can't
             # find from roots.
             heads = dict.fromkeys(heads, 0)
             # Remember where the top was so we can use it as a limit later.
             highestrev = max([self.rev(n) for n in nodestotag])
             while nodestotag:
                 # grab a node to tag
                 n = nodestotag.pop()
                 # Never tag nullid
                 if n == nullid:
                     continue
                 # A node's revision number represents its place in a
                 # topologically sorted list of nodes.
                 r = self.rev(n)
                 if r >= lowestrev:
                     if n not in ancestors:
                         # If we are possibly a descendent of one of the roots
                         # and we haven't already been marked as an ancestor
                         ancestors[n] = 1 # Mark as ancestor
                         # Add non-nullid parents to list of nodes to tag.
                         nodestotag.extend([p for p in self.parents(n) if
                                            p != nullid])
                     elif n in heads: # We've seen it before, is it a fake head?
                         # So it is, real heads should not be the ancestors of
                         # any other heads.
                         heads.pop(n)
             if not ancestors:
                 return nonodes
             # Now that we have our set of ancestors, we want to remove any
             # roots that are not ancestors.
 
             # If one of the roots was nullid, everything is included anyway.
             if lowestrev > -1:
                 # But, since we weren't, let's recompute the lowest rev to not
                 # include roots that aren't ancestors.
 
                 # Filter out roots that aren't ancestors of heads
                 roots = [n for n in roots if n in ancestors]
                 # Recompute the lowest revision
                 if roots:
                     lowestrev = min([self.rev(n) for n in roots])
                 else:
                     # No more roots?  Return empty list
                     return nonodes
             else:
                 # We are descending from nullid, and don't need to care about
                 # any other roots.
                 lowestrev = -1
                 roots = [nullid]
         # Transform our roots list into a 'set' (i.e. a dictionary where the
         # values don't matter.
         descendents = dict.fromkeys(roots, 1)
         # Also, keep the original roots so we can filter out roots that aren't
         # 'real' roots (i.e. are descended from other roots).
         roots = descendents.copy()
         # Our topologically sorted list of output nodes.
         orderedout = []
         # Don't start at nullid since we don't want nullid in our output list,
         # and if nullid shows up in descedents, empty parents will look like
         # they're descendents.
         for r in xrange(max(lowestrev, 0), highestrev + 1):
             n = self.node(r)
             isdescendent = False
             if lowestrev == -1:  # Everybody is a descendent of nullid
                 isdescendent = True
             elif n in descendents:
                 # n is already a descendent
                 isdescendent = True
                 # This check only needs to be done here because all the roots
                 # will start being marked is descendents before the loop.
                 if n in roots:
                     # If n was a root, check if it's a 'real' root.
                     p = tuple(self.parents(n))
                     # If any of its parents are descendents, it's not a root.
                     if (p[0] in descendents) or (p[1] in descendents):
                         roots.pop(n)
             else:
                 p = tuple(self.parents(n))
                 # A node is a descendent if either of its parents are
                 # descendents.  (We seeded the dependents list with the roots
                 # up there, remember?)
                 if (p[0] in descendents) or (p[1] in descendents):
                     descendents[n] = 1
                     isdescendent = True
             if isdescendent and ((ancestors is None) or (n in ancestors)):
                 # Only include nodes that are both descendents and ancestors.
                 orderedout.append(n)
                 if (ancestors is not None) and (n in heads):
                     # We're trying to figure out which heads are reachable
                     # from roots.
                     # Mark this head as having been reached
                     heads[n] = 1
                 elif ancestors is None:
                     # Otherwise, we're trying to discover the heads.
                     # Assume this is a head because if it isn't, the next step
                     # will eventually remove it.
                     heads[n] = 1
                     # But, obviously its parents aren't.
                     for p in self.parents(n):
                         heads.pop(p, None)
         heads = [n for n in heads.iterkeys() if heads[n] != 0]
         roots = roots.keys()
         assert orderedout
         assert roots
         assert heads
         return (orderedout, roots, heads)
 
     def heads(self, start=None):
         """return the list of all nodes that have no children
 
         if start is specified, only heads that are descendants of
         start will be returned
 
         """
         if start is None:
             start = nullid
         reachable = {start: 1}
         heads = {start: 1}
         startrev = self.rev(start)
 
         for r in xrange(startrev + 1, self.count()):
             n = self.node(r)
             for pn in self.parents(n):
                 if pn in reachable:
                     reachable[n] = 1
                     heads[n] = 1
                 if pn in heads:
                     del heads[pn]
         return heads.keys()
 
     def children(self, node):
         """find the children of a given node"""
         c = []
         p = self.rev(node)
         for r in range(p + 1, self.count()):
             n = self.node(r)
             for pn in self.parents(n):
                 if pn == node:
                     c.append(n)
                     continue
                 elif pn == nullid:
                     continue
         return c
 
     def lookup(self, id):
         """locate a node based on revision number or subset of hex nodeid"""
         try:
             rev = int(id)
             if str(rev) != id: raise ValueError
             if rev < 0: rev = self.count() + rev
             if rev < 0 or rev >= self.count(): raise ValueError
             return self.node(rev)
         except (ValueError, OverflowError):
             c = []
             for n in self.nodemap:
                 if hex(n).startswith(id):
                     c.append(n)
             if len(c) > 1: raise RevlogError(_("Ambiguous identifier"))
             if len(c) < 1: raise RevlogError(_("No match found"))
             return c[0]
 
         return None
 
     def diff(self, a, b):
         """return a delta between two revisions"""
         return mdiff.textdiff(a, b)
 
     def patches(self, t, pl):
         """apply a list of patches to a string"""
         return mdiff.patches(t, pl)
 
     def chunk(self, rev, df=None, cachelen=4096):
         start, length = self.start(rev), self.length(rev)
         inline = self.inlinedata()
         if inline:
             start += (rev + 1) * struct.calcsize(self.indexformat)
         end = start + length
         def loadcache(df):
             cache_length = max(cachelen, length) # 4k
             if not df:
                 if inline:
                     df = self.opener(self.indexfile)
                 else:
                     df = self.opener(self.datafile)
             df.seek(start)
             self.chunkcache = (start, df.read(cache_length))
 
         if not self.chunkcache:
             loadcache(df)
 
         cache_start = self.chunkcache[0]
         cache_end = cache_start + len(self.chunkcache[1])
         if start >= cache_start and end <= cache_end:
             # it is cached
             offset = start - cache_start
         else:
             loadcache(df)
             offset = 0
 
         #def checkchunk():
         #    df = self.opener(self.datafile)
         #    df.seek(start)
         #    return df.read(length)
         #assert s == checkchunk()
         return decompress(self.chunkcache[1][offset:offset + length])
 
     def delta(self, node):
         """return or calculate a delta between a node and its predecessor"""
         r = self.rev(node)
         return self.revdiff(r - 1, r)
 
     def revdiff(self, rev1, rev2):
         """return or calculate a delta between two revisions"""
         b1 = self.base(rev1)
         b2 = self.base(rev2)
         if b1 == b2 and rev1 + 1 == rev2:
             return self.chunk(rev2)
         else:
             return self.diff(self.revision(self.node(rev1)),
                              self.revision(self.node(rev2)))
 
     def revision(self, node):
         """return an uncompressed revision of a given"""
         if node == nullid: return ""
         if self.cache and self.cache[0] == node: return self.cache[2]
 
         # look up what we need to read
         text = None
         rev = self.rev(node)
         base = self.base(rev)
 
         if self.inlinedata():
             # we probably have the whole chunk cached
             df = None
         else:
             df = self.opener(self.datafile)
 
         # do we have useful data cached?
         if self.cache and self.cache[1] >= base and self.cache[1] < rev:
             base = self.cache[1]
             text = self.cache[2]
             self.loadindex(base, rev + 1)
         else:
             self.loadindex(base, rev + 1)
             text = self.chunk(base, df=df)
 
         bins = []
         for r in xrange(base + 1, rev + 1):
             bins.append(self.chunk(r, df=df))
 
         text = self.patches(text, bins)
 
         p1, p2 = self.parents(node)
         if node != hash(text, p1, p2):
             raise RevlogError(_("integrity check failed on %s:%d")
                           % (self.datafile, rev))
 
         self.cache = (node, rev, text)
         return text
 
     def checkinlinesize(self, tr, fp=None):
         if not self.inlinedata():
             return
         if not fp:
             fp = self.opener(self.indexfile, 'r')
             fp.seek(0, 2)
         size = fp.tell()
         if size < 131072:
             return
         trinfo = tr.find(self.indexfile)
         if trinfo == None:
             raise RevlogError(_("%s not found in the transaction"  %
                               self.indexfile))
 
         trindex = trinfo[2]
         dataoff = self.start(trindex)
 
         tr.add(self.datafile, dataoff)
         df = self.opener(self.datafile, 'w')
         calc = struct.calcsize(self.indexformat)
         for r in xrange(self.count()):
             start = self.start(r) + (r + 1) * calc
             length = self.length(r)
             fp.seek(start)
             d = fp.read(length)
             df.write(d)
         fp.close()
         df.close()
         fp = self.opener(self.indexfile, 'w', atomictemp=True)
         self.version &= ~(REVLOGNGINLINEDATA)
         if self.count():
             x = self.index[0]
             e = struct.pack(self.indexformat, *x)[4:]
             l = struct.pack(versionformat, self.version)
             fp.write(l)
             fp.write(e)
 
         for i in xrange(1, self.count()):
             x = self.index[i]
             e = struct.pack(self.indexformat, *x)
             fp.write(e)
 
         # if we don't call rename, the temp file will never replace the
         # real index
         fp.rename()
 
         tr.replace(self.indexfile, trindex * calc)
         self.chunkcache = None
 
     def addrevision(self, text, transaction, link, p1=None, p2=None, d=None):
         """add a revision to the log
 
         text - the revision data to add
         transaction - the transaction object used for rollback
         link - the linkrev data to add
         p1, p2 - the parent nodeids of the revision
         d - an optional precomputed delta
         """
         if text is None: text = ""
         if p1 is None: p1 = self.tip()
         if p2 is None: p2 = nullid
 
         node = hash(text, p1, p2)
 
         if node in self.nodemap:
             return node
 
         n = self.count()
         t = n - 1
 
         if n:
             base = self.base(t)
             start = self.start(base)
             end = self.end(t)
             if not d:
                 prev = self.revision(self.tip())
                 d = self.diff(prev, str(text))
             data = compress(d)
             l = len(data[1]) + len(data[0])
             dist = end - start + l
 
         # full versions are inserted when the needed deltas
         # become comparable to the uncompressed text
         if not n or dist > len(text) * 2:
             data = compress(text)
             l = len(data[1]) + len(data[0])
             base = n
         else:
             base = self.base(t)
 
         offset = 0
         if t >= 0:
             offset = self.end(t)
 
         if self.version == 0:
             e = (offset, l, base, link, p1, p2, node)
         else:
             e = (self.offset_type(offset, 0), l, len(text),
                  base, link, self.rev(p1), self.rev(p2), node)
 
         self.index.append(e)
         self.nodemap[node] = n
         entry = struct.pack(self.indexformat, *e)
 
         if not self.inlinedata():
             transaction.add(self.datafile, offset)
             transaction.add(self.indexfile, n * len(entry))
             f = self.opener(self.datafile, "a")
             if data[0]:
                 f.write(data[0])
             f.write(data[1])
             f.close()
             f = self.opener(self.indexfile, "a")
         else:
             f = self.opener(self.indexfile, "a+")
             f.seek(0, 2)
             transaction.add(self.indexfile, f.tell(), self.count() - 1)
 
         if len(self.index) == 1 and self.version != 0:
             l = struct.pack(versionformat, self.version)
             f.write(l)
             entry = entry[4:]
 
         f.write(entry)
 
         if self.inlinedata():
             f.write(data[0])
             f.write(data[1])
             self.checkinlinesize(transaction, f)
 
         self.cache = (node, n, text)
         return node
 
     def ancestor(self, a, b):
         """calculate the least common ancestor of nodes a and b"""
 
         # start with some short cuts for the linear cases
         if a == b:
             return a
         ra = self.rev(a)
         rb = self.rev(b)
         if ra < rb:
             last = b
             first = a
         else:
             last = a
             first = b
 
         # reachable won't include stop in the list, so we have to use a parent
         reachable = self.reachable(last, stop=self.parents(first)[0])
         if first in reachable:
             return first
 
         # calculate the distance of every node from root
         dist = {nullid: 0}
         for i in xrange(self.count()):
             n = self.node(i)
             p1, p2 = self.parents(n)
             dist[n] = max(dist[p1], dist[p2]) + 1
 
         # traverse ancestors in order of decreasing distance from root
         def ancestors(node):
             # we store negative distances because heap returns smallest member
             h = [(-dist[node], node)]
             seen = {}
             while h:
                 d, n = heapq.heappop(h)
                 if n not in seen:
                     seen[n] = 1
                     yield (-d, n)
                     for p in self.parents(n):
                         heapq.heappush(h, (-dist[p], p))
 
         def generations(node):
             sg, s = None, {}
             for g,n in ancestors(node):
                 if g != sg:
                     if sg:
                         yield sg, s
                     sg, s = g, {n:1}
                 else:
                     s[n] = 1
             yield sg, s
 
         x = generations(a)
         y = generations(b)
         gx = x.next()
         gy = y.next()
 
         # increment each ancestor list until it is closer to root than
         # the other, or they match
         while 1:
             #print "ancestor gen %s %s" % (gx[0], gy[0])
             if gx[0] == gy[0]:
                 # find the intersection
                 i = [ n for n in gx[1] if n in gy[1] ]
                 if i:
                     return i[0]
                 else:
                     #print "next"
                     gy = y.next()
                     gx = x.next()
             elif gx[0] < gy[0]:
                 #print "next y"
                 gy = y.next()
             else:
                 #print "next x"
                 gx = x.next()
 
     def group(self, nodelist, lookup, infocollect=None):
         """calculate a delta group
 
         Given a list of changeset revs, return a set of deltas and
         metadata corresponding to nodes. the first delta is
         parent(nodes[0]) -> nodes[0] the receiver is guaranteed to
         have this parent as it has all history before these
         changesets. parent is parent[0]
         """
         revs = [self.rev(n) for n in nodelist]
 
         # if we don't have any revisions touched by these changesets, bail
         if not revs:
             yield changegroup.closechunk()
             return
 
         # add the parent of the first rev
         p = self.parents(self.node(revs[0]))[0]
         revs.insert(0, self.rev(p))
 
         # build deltas
         for d in xrange(0, len(revs) - 1):
             a, b = revs[d], revs[d + 1]
             nb = self.node(b)
 
             if infocollect is not None:
                 infocollect(nb)
 
             d = self.revdiff(a, b)
             p = self.parents(nb)
             meta = nb + p[0] + p[1] + lookup(nb)
             yield changegroup.genchunk("%s%s" % (meta, d))
 
         yield changegroup.closechunk()
 
     def addgroup(self, revs, linkmapper, transaction, unique=0):
         """
         add a delta group
 
         given a set of deltas, add them to the revision log. the
         first delta is against its parent, which should be in our
         log, the rest are against the previous delta.
         """
 
         #track the base of the current delta log
         r = self.count()
         t = r - 1
         node = None
 
         base = prev = -1
         start = end = textlen = 0
         if r:
             end = self.end(t)
 
         ifh = self.opener(self.indexfile, "a+")
         ifh.seek(0, 2)
         transaction.add(self.indexfile, ifh.tell(), self.count())
         if self.inlinedata():
             dfh = None
         else:
             transaction.add(self.datafile, end)
             dfh = self.opener(self.datafile, "a")
 
         # loop through our set of deltas
         chain = None
         for chunk in revs:
             node, p1, p2, cs = struct.unpack("20s20s20s20s", chunk[:80])
             link = linkmapper(cs)
             if node in self.nodemap:
                 # this can happen if two branches make the same change
                 # if unique:
                 #    raise RevlogError(_("already have %s") % hex(node[:4]))
                 chain = node
                 continue
             delta = chunk[80:]
 
             for p in (p1, p2):
                 if not p in self.nodemap:
                     raise RevlogError(_("unknown parent %s") % short(p1))
 
             if not chain:
                 # retrieve the parent revision of the delta chain
                 chain = p1
                 if not chain in self.nodemap:
                     raise RevlogError(_("unknown base %s") % short(chain[:4]))
 
             # full versions are inserted when the needed deltas become
             # comparable to the uncompressed text or when the previous
             # version is not the one we have a delta against. We use
             # the size of the previous full rev as a proxy for the
             # current size.
 
             if chain == prev:
                 tempd = compress(delta)
                 cdelta = tempd[0] + tempd[1]
                 textlen = mdiff.patchedsize(textlen, delta)
 
             if chain != prev or (end - start + len(cdelta)) > textlen * 2:
                 # flush our writes here so we can read it in revision
                 if dfh:
                     dfh.flush()
                 ifh.flush()
                 text = self.revision(chain)
                 text = self.patches(text, [delta])
                 chk = self.addrevision(text, transaction, link, p1, p2)
                 if chk != node:
                     raise RevlogError(_("consistency error adding group"))
                 textlen = len(text)
             else:
                 if self.version == 0:
                     e = (end, len(cdelta), base, link, p1, p2, node)
                 else:
                     e = (self.offset_type(end, 0), len(cdelta), textlen, base,
                          link, self.rev(p1), self.rev(p2), node)
                 self.index.append(e)
                 self.nodemap[node] = r
                 if self.inlinedata():
                     ifh.write(struct.pack(self.indexformat, *e))
                     ifh.write(cdelta)
                     self.checkinlinesize(transaction, ifh)
                     if not self.inlinedata():
                         dfh = self.opener(self.datafile, "a")
                         ifh = self.opener(self.indexfile, "a")
                 else:
                     if not dfh:
                         # addrevision switched from inline to conventional
                         # reopen the index
                         dfh = self.opener(self.datafile, "a")
                         ifh = self.opener(self.indexfile, "a")
                     dfh.write(cdelta)
                     ifh.write(struct.pack(self.indexformat, *e))
 
             t, r, chain, prev = r, r + 1, node, node
             base = self.base(t)
             start = self.start(base)
             end = self.end(t)
 
         if node is None:
             raise RevlogError(_("group to be added is empty"))
         return node
 
     def strip(self, rev, minlink):
         if self.count() == 0 or rev >= self.count():
             return
 
         if isinstance(self.index, lazyindex):
             self.loadindexmap()
 
         # When stripping away a revision, we need to make sure it
         # does not actually belong to an older changeset.
         # The minlink parameter defines the oldest revision
         # we're allowed to strip away.
         while minlink > self.index[rev][-4]:
             rev += 1
             if rev >= self.count():
                 return
 
         # first truncate the files on disk
         end = self.start(rev)
         if not self.inlinedata():
             df = self.opener(self.datafile, "a")
             df.truncate(end)
             end = rev * struct.calcsize(self.indexformat)
         else:
             end += rev * struct.calcsize(self.indexformat)
 
         indexf = self.opener(self.indexfile, "a")
         indexf.truncate(end)
 
         # then reset internal state in memory to forget those revisions
         self.cache = None
         self.chunkcache = None
         for x in xrange(rev, self.count()):
             del self.nodemap[self.node(x)]
 
         del self.index[rev:]
 
     def checksize(self):
         expected = 0
         if self.count():
             expected = self.end(self.count() - 1)
 
         try:
             f = self.opener(self.datafile)
             f.seek(0, 2)
             actual = f.tell()
             dd = actual - expected
         except IOError, inst:
             if inst.errno != errno.ENOENT:
                 raise
             dd = 0
 
         try:
             f = self.opener(self.indexfile)
             f.seek(0, 2)
             actual = f.tell()
             s = struct.calcsize(self.indexformat)
             i = actual / s
             di = actual - (i * s)
             if self.inlinedata():
                 databytes = 0
                 for r in xrange(self.count()):
                     databytes += self.length(r)
                 dd = 0
                 di = actual - self.count() * s - databytes
         except IOError, inst:
             if inst.errno != errno.ENOENT:
                 raise
             di = 0
 
         return (dd, di)
 
 

mercurial/sshrepo.py

@@ -1,155 +1,155
 # sshrepo.py - ssh repository proxy class for mercurial
 #
 # 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.
 
 from node import *
 from remoterepo import *
 from i18n import gettext as _
 from demandload import *
 demandload(globals(), "hg os re stat")
 
 class sshrepository(remoterepository):
     def __init__(self, ui, path):
         self.url = path
         self.ui = ui
 
         m = re.match(r'ssh://(([^@]+)@)?([^:/]+)(:(\d+))?(/(.*))?', path)
         if not m:
             raise hg.RepoError(_("couldn't parse destination %s") % path)
 
         self.user = m.group(2)
         self.host = m.group(3)
         self.port = m.group(5)
         self.path = m.group(7) or "."
 
         args = self.user and ("%s@%s" % (self.user, self.host)) or self.host
         args = self.port and ("%s -p %s") % (args, self.port) or args
 
         sshcmd = self.ui.config("ui", "ssh", "ssh")
         remotecmd = self.ui.config("ui", "remotecmd", "hg")
         cmd = '%s %s "%s -R %s serve --stdio"'
         cmd = cmd % (sshcmd, args, remotecmd, self.path)
 
         ui.note('running %s\n' % cmd)
         self.pipeo, self.pipei, self.pipee = os.popen3(cmd, 'b')
 
         # skip any noise generated by remote shell
         r = self.do_cmd("between", pairs=("%s-%s" % ("0"*40, "0"*40)))
         l1 = ""
         l2 = "dummy"
         max_noise = 500
         while l2 and max_noise:
             l2 = r.readline()
             self.readerr()
             if l1 == "1\n" and l2 == "\n":
                 break
             if l1:
                 ui.debug(_("remote: "), l1)
             l1 = l2
             max_noise -= 1
         else:
             if l1:
                 ui.debug(_("remote: "), l1)
             raise hg.RepoError(_("no response from remote hg"))
 
     def readerr(self):
         while 1:
-            size = os.fstat(self.pipee.fileno())[stat.ST_SIZE]
+            size = util.fstat(self.pipee).st_size
             if size == 0: break
             l = self.pipee.readline()
             if not l: break
             self.ui.status(_("remote: "), l)
 
     def __del__(self):
         try:
             self.pipeo.close()
             self.pipei.close()
             # read the error descriptor until EOF
             for l in self.pipee:
                 self.ui.status(_("remote: "), l)
             self.pipee.close()
         except:
             pass
 
     def dev(self):
         return -1
 
     def do_cmd(self, cmd, **args):
         self.ui.debug(_("sending %s command\n") % cmd)
         self.pipeo.write("%s\n" % cmd)
         for k, v in args.items():
             self.pipeo.write("%s %d\n" % (k, len(v)))
             self.pipeo.write(v)
         self.pipeo.flush()
 
         return self.pipei
 
     def call(self, cmd, **args):
         r = self.do_cmd(cmd, **args)
         l = r.readline()
         self.readerr()
         try:
             l = int(l)
         except:
             raise hg.RepoError(_("unexpected response '%s'") % l)
         return r.read(l)
 
     def lock(self):
         self.call("lock")
         return remotelock(self)
 
     def unlock(self):
         self.call("unlock")
 
     def heads(self):
         d = self.call("heads")
         try:
             return map(bin, d[:-1].split(" "))
         except:
             raise hg.RepoError(_("unexpected response '%s'") % (d[:400] + "..."))
 
     def branches(self, nodes):
         n = " ".join(map(hex, nodes))
         d = self.call("branches", nodes=n)
         try:
             br = [ tuple(map(bin, b.split(" "))) for b in d.splitlines() ]
             return br
         except:
             raise hg.RepoError(_("unexpected response '%s'") % (d[:400] + "..."))
 
     def between(self, pairs):
         n = "\n".join(["-".join(map(hex, p)) for p in pairs])
         d = self.call("between", pairs=n)
         try:
             p = [ l and map(bin, l.split(" ")) or [] for l in d.splitlines() ]
             return p
         except:
             raise hg.RepoError(_("unexpected response '%s'") % (d[:400] + "..."))
 
     def changegroup(self, nodes, kind):
         n = " ".join(map(hex, nodes))
         f = self.do_cmd("changegroup", roots=n)
         return self.pipei
 
     def addchangegroup(self, cg):
         d = self.call("addchangegroup")
         if d:
             raise hg.RepoError(_("push refused: %s"), d)
 
         while 1:
             d = cg.read(4096)
             if not d: break
             self.pipeo.write(d)
             self.readerr()
 
         self.pipeo.flush()
 
         self.readerr()
         l = int(self.pipei.readline())
         r = self.pipei.read(l)
         if not r:
             return 1
         return int(r)

mercurial/util.py

@@ -1,845 +1,865
 """
 util.py - Mercurial utility functions and platform specfic implementations
 
  Copyright 2005 K. Thananchayan <[e-mail]>
 
 This software may be used and distributed according to the terms
 of the GNU General Public License, incorporated herein by reference.
 
 This contains helper routines that are independent of the SCM core and hide
 platform-specific details from the core.
 """
 
 import os, errno
 from i18n import gettext as _
 from demandload import *
 demandload(globals(), "cStringIO errno popen2 re shutil sys tempfile")
 demandload(globals(), "threading time")
 
 def pipefilter(s, cmd):
     '''filter string S through command CMD, returning its output'''
     (pout, pin) = popen2.popen2(cmd, -1, 'b')
     def writer():
         try:
             pin.write(s)
             pin.close()
         except IOError, inst:
             if inst.errno != errno.EPIPE:
                 raise
 
     # we should use select instead on UNIX, but this will work on most
     # systems, including Windows
     w = threading.Thread(target=writer)
     w.start()
     f = pout.read()
     pout.close()
     w.join()
     return f
 
 def tempfilter(s, cmd):
     '''filter string S through a pair of temporary files with CMD.
     CMD is used as a template to create the real command to be run,
     with the strings INFILE and OUTFILE replaced by the real names of
     the temporary files generated.'''
     inname, outname = None, None
     try:
         infd, inname = tempfile.mkstemp(prefix='hgfin')
         fp = os.fdopen(infd, 'wb')
         fp.write(s)
         fp.close()
         outfd, outname = tempfile.mkstemp(prefix='hgfout')
         os.close(outfd)
         cmd = cmd.replace('INFILE', inname)
         cmd = cmd.replace('OUTFILE', outname)
         code = os.system(cmd)
         if code: raise Abort(_("command '%s' failed: %s") %
                              (cmd, explain_exit(code)))
         return open(outname, 'rb').read()
     finally:
         try:
             if inname: os.unlink(inname)
         except: pass
         try:
             if outname: os.unlink(outname)
         except: pass
 
 filtertable = {
     'tempfile:': tempfilter,
     'pipe:': pipefilter,
     }
 
 def filter(s, cmd):
     "filter a string through a command that transforms its input to its output"
     for name, fn in filtertable.iteritems():
         if cmd.startswith(name):
             return fn(s, cmd[len(name):].lstrip())
     return pipefilter(s, cmd)
 
 def find_in_path(name, path, default=None):
     '''find name in search path. path can be string (will be split
     with os.pathsep), or iterable thing that returns strings.  if name
     found, return path to name. else return default.'''
     if isinstance(path, str):
         path = path.split(os.pathsep)
     for p in path:
         p_name = os.path.join(p, name)
         if os.path.exists(p_name):
             return p_name
     return default
 
 def patch(strip, patchname, ui):
     """apply the patch <patchname> to the working directory.
     a list of patched files is returned"""
     patcher = find_in_path('gpatch', os.environ.get('PATH', ''), 'patch')
     fp = os.popen('"%s" -p%d < "%s"' % (patcher, strip, patchname))
     files = {}
     for line in fp:
         line = line.rstrip()
         ui.status("%s\n" % line)
         if line.startswith('patching file '):
             pf = parse_patch_output(line)
             files.setdefault(pf, 1)
     code = fp.close()
     if code:
         raise Abort(_("patch command failed: %s") % explain_exit(code)[0])
     return files.keys()
 
 def binary(s):
     """return true if a string is binary data using diff's heuristic"""
     if s and '\0' in s[:4096]:
         return True
     return False
 
 def unique(g):
     """return the uniq elements of iterable g"""
     seen = {}
     for f in g:
         if f not in seen:
             seen[f] = 1
             yield f
 
 class Abort(Exception):
     """Raised if a command needs to print an error and exit."""
 
 def always(fn): return True
 def never(fn): return False
 
 def patkind(name, dflt_pat='glob'):
     """Split a string into an optional pattern kind prefix and the
     actual pattern."""
     for prefix in 're', 'glob', 'path', 'relglob', 'relpath', 'relre':
         if name.startswith(prefix + ':'): return name.split(':', 1)
     return dflt_pat, name
 
 def globre(pat, head='^', tail='$'):
     "convert a glob pattern into a regexp"
     i, n = 0, len(pat)
     res = ''
     group = False
     def peek(): return i < n and pat[i]
     while i < n:
         c = pat[i]
         i = i+1
         if c == '*':
             if peek() == '*':
                 i += 1
                 res += '.*'
             else:
                 res += '[^/]*'
         elif c == '?':
             res += '.'
         elif c == '[':
             j = i
             if j < n and pat[j] in '!]':
                 j += 1
             while j < n and pat[j] != ']':
                 j += 1
             if j >= n:
                 res += '\\['
             else:
                 stuff = pat[i:j].replace('\\','\\\\')
                 i = j + 1
                 if stuff[0] == '!':
                     stuff = '^' + stuff[1:]
                 elif stuff[0] == '^':
                     stuff = '\\' + stuff
                 res = '%s[%s]' % (res, stuff)
         elif c == '{':
             group = True
             res += '(?:'
         elif c == '}' and group:
             res += ')'
             group = False
         elif c == ',' and group:
             res += '|'
         elif c == '\\':
             p = peek()
             if p:
                 i += 1
                 res += re.escape(p)
             else:
                 res += re.escape(c)
         else:
             res += re.escape(c)
     return head + res + tail
 
 _globchars = {'[': 1, '{': 1, '*': 1, '?': 1}
 
 def pathto(n1, n2):
     '''return the relative path from one place to another.
     this returns a path in the form used by the local filesystem, not hg.'''
     if not n1: return localpath(n2)
     a, b = n1.split('/'), n2.split('/')
     a.reverse()
     b.reverse()
     while a and b and a[-1] == b[-1]:
         a.pop()
         b.pop()
     b.reverse()
     return os.sep.join((['..'] * len(a)) + b)
 
 def canonpath(root, cwd, myname):
     """return the canonical path of myname, given cwd and root"""
     if root == os.sep:
         rootsep = os.sep
     else:
         rootsep = root + os.sep
     name = myname
     if not os.path.isabs(name):
         name = os.path.join(root, cwd, name)
     name = os.path.normpath(name)
     if name.startswith(rootsep):
         name = name[len(rootsep):]
         audit_path(name)
         return pconvert(name)
     elif name == root:
         return ''
     else:
         # Determine whether `name' is in the hierarchy at or beneath `root',
         # by iterating name=dirname(name) until that causes no change (can't
         # check name == '/', because that doesn't work on windows).  For each
         # `name', compare dev/inode numbers.  If they match, the list `rel'
         # holds the reversed list of components making up the relative file
         # name we want.
         root_st = os.stat(root)
         rel = []
         while True:
             try:
                 name_st = os.stat(name)
             except OSError:
                 break
             if os.path.samestat(name_st, root_st):
                 rel.reverse()
                 name = os.path.join(*rel)
                 audit_path(name)
                 return pconvert(name)
             dirname, basename = os.path.split(name)
             rel.append(basename)
             if dirname == name:
                 break
             name = dirname
 
         raise Abort('%s not under root' % myname)
 
 def matcher(canonroot, cwd='', names=['.'], inc=[], exc=[], head='', src=None):
     return _matcher(canonroot, cwd, names, inc, exc, head, 'glob', src)
 
 def cmdmatcher(canonroot, cwd='', names=['.'], inc=[], exc=[], head='', src=None):
     if os.name == 'nt':
         dflt_pat = 'glob'
     else:
         dflt_pat = 'relpath'
     return _matcher(canonroot, cwd, names, inc, exc, head, dflt_pat, src)
 
 def _matcher(canonroot, cwd, names, inc, exc, head, dflt_pat, src):
     """build a function to match a set of file patterns
 
     arguments:
     canonroot - the canonical root of the tree you're matching against
     cwd - the current working directory, if relevant
     names - patterns to find
     inc - patterns to include
     exc - patterns to exclude
     head - a regex to prepend to patterns to control whether a match is rooted
 
     a pattern is one of:
     'glob:<rooted glob>'
     're:<rooted regexp>'
     'path:<rooted path>'
     'relglob:<relative glob>'
     'relpath:<relative path>'
     'relre:<relative regexp>'
     '<rooted path or regexp>'
 
     returns:
     a 3-tuple containing
     - list of explicit non-pattern names passed in
     - a bool match(filename) function
     - a bool indicating if any patterns were passed in
 
     todo:
     make head regex a rooted bool
     """
 
     def contains_glob(name):
         for c in name:
             if c in _globchars: return True
         return False
 
     def regex(kind, name, tail):
         '''convert a pattern into a regular expression'''
         if kind == 're':
             return name
         elif kind == 'path':
             return '^' + re.escape(name) + '(?:/|$)'
         elif kind == 'relglob':
             return head + globre(name, '(?:|.*/)', tail)
         elif kind == 'relpath':
             return head + re.escape(name) + tail
         elif kind == 'relre':
             if name.startswith('^'):
                 return name
             return '.*' + name
         return head + globre(name, '', tail)
 
     def matchfn(pats, tail):
         """build a matching function from a set of patterns"""
         if not pats:
             return
         matches = []
         for k, p in pats:
             try:
                 pat = '(?:%s)' % regex(k, p, tail)
                 matches.append(re.compile(pat).match)
             except re.error:
                 if src: raise Abort("%s: invalid pattern (%s): %s" % (src, k, p))
                 else: raise Abort("invalid pattern (%s): %s" % (k, p))
 
         def buildfn(text):
             for m in matches:
                 r = m(text)
                 if r:
                     return r
 
         return buildfn
 
     def globprefix(pat):
         '''return the non-glob prefix of a path, e.g. foo/* -> foo'''
         root = []
         for p in pat.split(os.sep):
             if contains_glob(p): break
             root.append(p)
         return '/'.join(root)
 
     pats = []
     files = []
     roots = []
     for kind, name in [patkind(p, dflt_pat) for p in names]:
         if kind in ('glob', 'relpath'):
             name = canonpath(canonroot, cwd, name)
             if name == '':
                 kind, name = 'glob', '**'
         if kind in ('glob', 'path', 're'):
             pats.append((kind, name))
         if kind == 'glob':
             root = globprefix(name)
             if root: roots.append(root)
         elif kind == 'relpath':
             files.append((kind, name))
             roots.append(name)
 
     patmatch = matchfn(pats, '$') or always
     filematch = matchfn(files, '(?:/|$)') or always
     incmatch = always
     if inc:
         incmatch = matchfn(map(patkind, inc), '(?:/|$)')
     excmatch = lambda fn: False
     if exc:
         excmatch = matchfn(map(patkind, exc), '(?:/|$)')
 
     return (roots,
             lambda fn: (incmatch(fn) and not excmatch(fn) and
                         (fn.endswith('/') or
                          (not pats and not files) or
                          (pats and patmatch(fn)) or
                          (files and filematch(fn)))),
             (inc or exc or (pats and pats != [('glob', '**')])) and True)
 
 def system(cmd, environ={}, cwd=None, onerr=None, errprefix=None):
     '''enhanced shell command execution.
     run with environment maybe modified, maybe in different dir.
 
     if command fails and onerr is None, return status.  if ui object,
     print error message and return status, else raise onerr object as
     exception.'''
     oldenv = {}
     for k in environ:
         oldenv[k] = os.environ.get(k)
     if cwd is not None:
         oldcwd = os.getcwd()
     try:
         for k, v in environ.iteritems():
             os.environ[k] = str(v)
         if cwd is not None and oldcwd != cwd:
             os.chdir(cwd)
         rc = os.system(cmd)
         if rc and onerr:
             errmsg = '%s %s' % (os.path.basename(cmd.split(None, 1)[0]),
                                 explain_exit(rc)[0])
             if errprefix:
                 errmsg = '%s: %s' % (errprefix, errmsg)
             try:
                 onerr.warn(errmsg + '\n')
             except AttributeError:
                 raise onerr(errmsg)
         return rc
     finally:
         for k, v in oldenv.iteritems():
             if v is None:
                 del os.environ[k]
             else:
                 os.environ[k] = v
         if cwd is not None and oldcwd != cwd:
             os.chdir(oldcwd)
 
 def rename(src, dst):
     """forcibly rename a file"""
     try:
         os.rename(src, dst)
-    except:
-        os.unlink(dst)
+    except OSError, err:
+        # on windows, rename to existing file is not allowed, so we
+        # must delete destination first. but if file is open, unlink
+        # schedules it for delete but does not delete it. rename
+        # happens immediately even for open files, so we create
+        # temporary file, delete it, rename destination to that name,
+        # then delete that. then rename is safe to do.
+        fd, temp = tempfile.mkstemp(dir=os.path.dirname(dst) or '.')
+        os.close(fd)
+        os.unlink(temp)
+        os.rename(dst, temp)
+        os.unlink(temp)
         os.rename(src, dst)
 
 def unlink(f):
     """unlink and remove the directory if it is empty"""
     os.unlink(f)
     # try removing directories that might now be empty
     try:
         os.removedirs(os.path.dirname(f))
     except OSError:
         pass
 
 def copyfiles(src, dst, hardlink=None):
     """Copy a directory tree using hardlinks if possible"""
 
     if hardlink is None:
         hardlink = (os.stat(src).st_dev ==
                     os.stat(os.path.dirname(dst)).st_dev)
 
     if os.path.isdir(src):
         os.mkdir(dst)
         for name in os.listdir(src):
             srcname = os.path.join(src, name)
             dstname = os.path.join(dst, name)
             copyfiles(srcname, dstname, hardlink)
     else:
         if hardlink:
             try:
                 os_link(src, dst)
             except (IOError, OSError):
                 hardlink = False
                 shutil.copy(src, dst)
         else:
             shutil.copy(src, dst)
 
 def audit_path(path):
     """Abort if path contains dangerous components"""
     parts = os.path.normcase(path).split(os.sep)
     if (os.path.splitdrive(path)[0] or parts[0] in ('.hg', '')
         or os.pardir in parts):
         raise Abort(_("path contains illegal component: %s\n") % path)
 
-def opener(base, audit=True):
-    """
-    return a function that opens files relative to base
-
-    this function is used to hide the details of COW semantics and
-    remote file access from higher level code.
-    """
-    p = base
-    audit_p = audit
-
-    def mktempcopy(name):
-        d, fn = os.path.split(name)
-        fd, temp = tempfile.mkstemp(prefix=fn, dir=d)
-        fp = os.fdopen(fd, "wb")
-        try:
-            fp.write(file(name, "rb").read())
-        except:
-            try: os.unlink(temp)
-            except: pass
-            raise
-        fp.close()
-        st = os.lstat(name)
-        os.chmod(temp, st.st_mode)
-        return temp
-
-    class atomictempfile(file):
-        """the file will only be copied when rename is called"""
-        def __init__(self, name, mode):
-            self.__name = name
-            self.temp = mktempcopy(name)
-            file.__init__(self, self.temp, mode)
-        def rename(self):
-            if not self.closed:
-                file.close(self)
-                rename(self.temp, self.__name)
-        def __del__(self):
-            if not self.closed:
-                try:
-                    os.unlink(self.temp)
-                except: pass
-                file.close(self)
-
-    class atomicfile(atomictempfile):
-        """the file will only be copied on close"""
-        def __init__(self, name, mode):
-            atomictempfile.__init__(self, name, mode)
-        def close(self):
-            self.rename()
-        def __del__(self):
-            self.rename()
-
-    def o(path, mode="r", text=False, atomic=False, atomictemp=False):
-        if audit_p:
-            audit_path(path)
-        f = os.path.join(p, path)
-
-        if not text:
-            mode += "b" # for that other OS
-
-        if mode[0] != "r":
-            try:
-                nlink = nlinks(f)
-            except OSError:
-                d = os.path.dirname(f)
-                if not os.path.isdir(d):
-                    os.makedirs(d)
-            else:
-                if atomic:
-                    return atomicfile(f, mode)
-                elif atomictemp:
-                    return atomictempfile(f, mode)
-                if nlink > 1:
-                    rename(mktempcopy(f), f)
-        return file(f, mode)
-
-    return o
-
 def _makelock_file(info, pathname):
     ld = os.open(pathname, os.O_CREAT | os.O_WRONLY | os.O_EXCL)
     os.write(ld, info)
     os.close(ld)
 
 def _readlock_file(pathname):
-    return file(pathname).read()
+    return posixfile(pathname).read()
 
 def nlinks(pathname):
     """Return number of hardlinks for the given file."""
     return os.stat(pathname).st_nlink
 
 if hasattr(os, 'link'):
     os_link = os.link
 else:
     def os_link(src, dst):
         raise OSError(0, _("Hardlinks not supported"))
 
+def fstat(fp):
+    '''stat file object that may not have fileno method.'''
+    try:
+        return os.fstat(fp.fileno())
+    except AttributeError:
+        return os.stat(fp.name)
+
+posixfile = file
+
 # Platform specific variants
 if os.name == 'nt':
     demandload(globals(), "msvcrt")
     nulldev = 'NUL:'
 
     class winstdout:
         '''stdout on windows misbehaves if sent through a pipe'''
 
         def __init__(self, fp):
             self.fp = fp
 
         def __getattr__(self, key):
             return getattr(self.fp, key)
 
         def close(self):
             try:
                 self.fp.close()
             except: pass
 
         def write(self, s):
             try:
                 return self.fp.write(s)
             except IOError, inst:
                 if inst.errno != 0: raise
                 self.close()
                 raise IOError(errno.EPIPE, 'Broken pipe')
 
     sys.stdout = winstdout(sys.stdout)
 
     def system_rcpath():
         try:
             return system_rcpath_win32()
         except:
             return [r'c:\mercurial\mercurial.ini']
 
     def os_rcpath():
         '''return default os-specific hgrc search path'''
         return system_rcpath() + [os.path.join(os.path.expanduser('~'),
                                                'mercurial.ini')]
 
     def parse_patch_output(output_line):
         """parses the output produced by patch and returns the file name"""
         pf = output_line[14:]
         if pf[0] == '`':
             pf = pf[1:-1] # Remove the quotes
         return pf
 
     def testpid(pid):
         '''return False if pid dead, True if running or not known'''
         return True
 
     def is_exec(f, last):
         return last
 
     def set_exec(f, mode):
         pass
 
     def set_binary(fd):
         msvcrt.setmode(fd.fileno(), os.O_BINARY)
 
     def pconvert(path):
         return path.replace("\\", "/")
 
     def localpath(path):
         return path.replace('/', '\\')
 
     def normpath(path):
         return pconvert(os.path.normpath(path))
 
     makelock = _makelock_file
     readlock = _readlock_file
 
     def explain_exit(code):
         return _("exited with status %d") % code, code
 
     try:
         # override functions with win32 versions if possible
         from util_win32 import *
     except ImportError:
         pass
 
 else:
     nulldev = '/dev/null'
 
     def rcfiles(path):
         rcs = [os.path.join(path, 'hgrc')]
         rcdir = os.path.join(path, 'hgrc.d')
         try:
             rcs.extend([os.path.join(rcdir, f) for f in os.listdir(rcdir)
                         if f.endswith(".rc")])
         except OSError, inst: pass
         return rcs
 
     def os_rcpath():
         '''return default os-specific hgrc search path'''
         path = []
         if len(sys.argv) > 0:
             path.extend(rcfiles(os.path.dirname(sys.argv[0]) +
                                   '/../etc/mercurial'))
         path.extend(rcfiles('/etc/mercurial'))
         path.append(os.path.expanduser('~/.hgrc'))
         path = [os.path.normpath(f) for f in path]
         return path
 
     def parse_patch_output(output_line):
         """parses the output produced by patch and returns the file name"""
         pf = output_line[14:]
         if pf.startswith("'") and pf.endswith("'") and pf.find(" ") >= 0:
             pf = pf[1:-1] # Remove the quotes
         return pf
 
     def is_exec(f, last):
         """check whether a file is executable"""
         return (os.stat(f).st_mode & 0100 != 0)
 
     def set_exec(f, mode):
         s = os.stat(f).st_mode
         if (s & 0100 != 0) == mode:
             return
         if mode:
             # Turn on +x for every +r bit when making a file executable
             # and obey umask.
             umask = os.umask(0)
             os.umask(umask)
             os.chmod(f, s | (s & 0444) >> 2 & ~umask)
         else:
             os.chmod(f, s & 0666)
 
     def set_binary(fd):
         pass
 
     def pconvert(path):
         return path
 
     def localpath(path):
         return path
 
     normpath = os.path.normpath
 
     def makelock(info, pathname):
         try:
             os.symlink(info, pathname)
         except OSError, why:
             if why.errno == errno.EEXIST:
                 raise
             else:
                 _makelock_file(info, pathname)
 
     def readlock(pathname):
         try:
             return os.readlink(pathname)
         except OSError, why:
             if why.errno == errno.EINVAL:
                 return _readlock_file(pathname)
             else:
                 raise
 
     def testpid(pid):
         '''return False if pid dead, True if running or not sure'''
         try:
             os.kill(pid, 0)
             return True
         except OSError, inst:
             return inst.errno != errno.ESRCH
 
     def explain_exit(code):
         """return a 2-tuple (desc, code) describing a process's status"""
         if os.WIFEXITED(code):
             val = os.WEXITSTATUS(code)
             return _("exited with status %d") % val, val
         elif os.WIFSIGNALED(code):
             val = os.WTERMSIG(code)
             return _("killed by signal %d") % val, val
         elif os.WIFSTOPPED(code):
             val = os.WSTOPSIG(code)
             return _("stopped by signal %d") % val, val
         raise ValueError(_("invalid exit code"))
 
+def opener(base, audit=True):
+    """
+    return a function that opens files relative to base
+
+    this function is used to hide the details of COW semantics and
+    remote file access from higher level code.
+    """
+    p = base
+    audit_p = audit
+
+    def mktempcopy(name):
+        d, fn = os.path.split(name)
+        fd, temp = tempfile.mkstemp(prefix=fn, dir=d)
+        os.close(fd)
+        fp = posixfile(temp, "wb")
+        try:
+            fp.write(posixfile(name, "rb").read())
+        except:
+            try: os.unlink(temp)
+            except: pass
+            raise
+        fp.close()
+        st = os.lstat(name)
+        os.chmod(temp, st.st_mode)
+        return temp
+
+    class atomictempfile(posixfile):
+        """the file will only be copied when rename is called"""
+        def __init__(self, name, mode):
+            self.__name = name
+            self.temp = mktempcopy(name)
+            posixfile.__init__(self, self.temp, mode)
+        def rename(self):
+            if not self.closed:
+                posixfile.close(self)
+                rename(self.temp, self.__name)
+        def __del__(self):
+            if not self.closed:
+                try:
+                    os.unlink(self.temp)
+                except: pass
+                posixfile.close(self)
+
+    class atomicfile(atomictempfile):
+        """the file will only be copied on close"""
+        def __init__(self, name, mode):
+            atomictempfile.__init__(self, name, mode)
+        def close(self):
+            self.rename()
+        def __del__(self):
+            self.rename()
+
+    def o(path, mode="r", text=False, atomic=False, atomictemp=False):
+        if audit_p:
+            audit_path(path)
+        f = os.path.join(p, path)
+
+        if not text:
+            mode += "b" # for that other OS
+
+        if mode[0] != "r":
+            try:
+                nlink = nlinks(f)
+            except OSError:
+                d = os.path.dirname(f)
+                if not os.path.isdir(d):
+                    os.makedirs(d)
+            else:
+                if atomic:
+                    return atomicfile(f, mode)
+                elif atomictemp:
+                    return atomictempfile(f, mode)
+                if nlink > 1:
+                    rename(mktempcopy(f), f)
+        return posixfile(f, mode)
+
+    return o
+
 class chunkbuffer(object):
     """Allow arbitrary sized chunks of data to be efficiently read from an
     iterator over chunks of arbitrary size."""
 
     def __init__(self, in_iter, targetsize = 2**16):
         """in_iter is the iterator that's iterating over the input chunks.
         targetsize is how big a buffer to try to maintain."""
         self.in_iter = iter(in_iter)
         self.buf = ''
         self.targetsize = int(targetsize)
         if self.targetsize <= 0:
             raise ValueError(_("targetsize must be greater than 0, was %d") %
                              targetsize)
         self.iterempty = False
 
     def fillbuf(self):
         """Ignore target size; read every chunk from iterator until empty."""
         if not self.iterempty:
             collector = cStringIO.StringIO()
             collector.write(self.buf)
             for ch in self.in_iter:
                 collector.write(ch)
             self.buf = collector.getvalue()
             self.iterempty = True
 
     def read(self, l):
         """Read L bytes of data from the iterator of chunks of data.
         Returns less than L bytes if the iterator runs dry."""
         if l > len(self.buf) and not self.iterempty:
             # Clamp to a multiple of self.targetsize
             targetsize = self.targetsize * ((l // self.targetsize) + 1)
             collector = cStringIO.StringIO()
             collector.write(self.buf)
             collected = len(self.buf)
             for chunk in self.in_iter:
                 collector.write(chunk)
                 collected += len(chunk)
                 if collected >= targetsize:
                     break
             if collected < targetsize:
                 self.iterempty = True
             self.buf = collector.getvalue()
         s, self.buf = self.buf[:l], buffer(self.buf, l)
         return s
 
 def filechunkiter(f, size = 65536):
     """Create a generator that produces all the data in the file size
     (default 65536) bytes at a time.  Chunks may be less than size
     bytes if the chunk is the last chunk in the file, or the file is a
     socket or some other type of file that sometimes reads less data
     than is requested."""
     s = f.read(size)
     while len(s) > 0:
         yield s
         s = f.read(size)
 
 def makedate():
     lt = time.localtime()
     if lt[8] == 1 and time.daylight:
         tz = time.altzone
     else:
         tz = time.timezone
     return time.mktime(lt), tz
 
 def datestr(date=None, format='%a %b %d %H:%M:%S %Y', timezone=True):
     """represent a (unixtime, offset) tuple as a localized time.
     unixtime is seconds since the epoch, and offset is the time zone's
     number of seconds away from UTC. if timezone is false, do not
     append time zone to string."""
     t, tz = date or makedate()
     s = time.strftime(format, time.gmtime(float(t) - tz))
     if timezone:
         s += " %+03d%02d" % (-tz / 3600, ((-tz % 3600) / 60))
     return s
 
 def shortuser(user):
     """Return a short representation of a user name or email address."""
     f = user.find('@')
     if f >= 0:
         user = user[:f]
     f = user.find('<')
     if f >= 0:
         user = user[f+1:]
     return user
 
 def walkrepos(path):
     '''yield every hg repository under path, recursively.'''
     def errhandler(err):
         if err.filename == path:
             raise err
 
     for root, dirs, files in os.walk(path, onerror=errhandler):
         for d in dirs:
             if d == '.hg':
                 yield root
                 dirs[:] = []
                 break
 
 _rcpath = None
 
 def rcpath():
     '''return hgrc search path. if env var HGRCPATH is set, use it.
     for each item in path, if directory, use files ending in .rc,
     else use item.
     make HGRCPATH empty to only look in .hg/hgrc of current repo.
     if no HGRCPATH, use default os-specific path.'''
     global _rcpath
     if _rcpath is None:
         if 'HGRCPATH' in os.environ:
             _rcpath = []
             for p in os.environ['HGRCPATH'].split(os.pathsep):
                 if not p: continue
                 if os.path.isdir(p):
                     for f in os.listdir(p):
                         if f.endswith('.rc'):
                             _rcpath.append(os.path.join(p, f))
                 else:
                     _rcpath.append(p)
         else:
             _rcpath = os_rcpath()
     return _rcpath

mercurial/util_win32.py

@@ -1,171 +1,280
 # util_win32.py - utility functions that use win32 API
 #
 # Copyright 2005 Matt Mackall <[e-mail]>
 # Copyright 2006 Vadim Gelfer <[e-mail]>
 #
 # This software may be used and distributed according to the terms of
 # the GNU General Public License, incorporated herein by reference.
 
 # Mark Hammond's win32all package allows better functionality on
 # Windows.  this module overrides definitions in util.py.  if not
 # available, import of this module will fail, and generic code will be
 # used.
 
 import win32api
 
 from demandload import *
 from i18n import gettext as _
 demandload(globals(), 'errno os pywintypes win32con win32file win32process')
-demandload(globals(), 'winerror')
+demandload(globals(), 'cStringIO winerror')
 
-class WinError(OSError):
+class WinError:
     winerror_map = {
         winerror.ERROR_ACCESS_DENIED: errno.EACCES,
         winerror.ERROR_ACCOUNT_DISABLED: errno.EACCES,
         winerror.ERROR_ACCOUNT_RESTRICTION: errno.EACCES,
         winerror.ERROR_ALREADY_ASSIGNED: errno.EBUSY,
         winerror.ERROR_ALREADY_EXISTS: errno.EEXIST,
         winerror.ERROR_ARITHMETIC_OVERFLOW: errno.ERANGE,
         winerror.ERROR_BAD_COMMAND: errno.EIO,
         winerror.ERROR_BAD_DEVICE: errno.ENODEV,
         winerror.ERROR_BAD_DRIVER_LEVEL: errno.ENXIO,
         winerror.ERROR_BAD_EXE_FORMAT: errno.ENOEXEC,
         winerror.ERROR_BAD_FORMAT: errno.ENOEXEC,
         winerror.ERROR_BAD_LENGTH: errno.EINVAL,
         winerror.ERROR_BAD_PATHNAME: errno.ENOENT,
         winerror.ERROR_BAD_PIPE: errno.EPIPE,
         winerror.ERROR_BAD_UNIT: errno.ENODEV,
         winerror.ERROR_BAD_USERNAME: errno.EINVAL,
         winerror.ERROR_BROKEN_PIPE: errno.EPIPE,
         winerror.ERROR_BUFFER_OVERFLOW: errno.ENAMETOOLONG,
         winerror.ERROR_BUSY: errno.EBUSY,
         winerror.ERROR_BUSY_DRIVE: errno.EBUSY,
         winerror.ERROR_CALL_NOT_IMPLEMENTED: errno.ENOSYS,
         winerror.ERROR_CANNOT_MAKE: errno.EACCES,
         winerror.ERROR_CANTOPEN: errno.EIO,
         winerror.ERROR_CANTREAD: errno.EIO,
         winerror.ERROR_CANTWRITE: errno.EIO,
         winerror.ERROR_CRC: errno.EIO,
         winerror.ERROR_CURRENT_DIRECTORY: errno.EACCES,
         winerror.ERROR_DEVICE_IN_USE: errno.EBUSY,
         winerror.ERROR_DEV_NOT_EXIST: errno.ENODEV,
         winerror.ERROR_DIRECTORY: errno.EINVAL,
         winerror.ERROR_DIR_NOT_EMPTY: errno.ENOTEMPTY,
         winerror.ERROR_DISK_CHANGE: errno.EIO,
         winerror.ERROR_DISK_FULL: errno.ENOSPC,
         winerror.ERROR_DRIVE_LOCKED: errno.EBUSY,
         winerror.ERROR_ENVVAR_NOT_FOUND: errno.EINVAL,
         winerror.ERROR_EXE_MARKED_INVALID: errno.ENOEXEC,
         winerror.ERROR_FILENAME_EXCED_RANGE: errno.ENAMETOOLONG,
         winerror.ERROR_FILE_EXISTS: errno.EEXIST,
         winerror.ERROR_FILE_INVALID: errno.ENODEV,
         winerror.ERROR_FILE_NOT_FOUND: errno.ENOENT,
         winerror.ERROR_GEN_FAILURE: errno.EIO,
         winerror.ERROR_HANDLE_DISK_FULL: errno.ENOSPC,
         winerror.ERROR_INSUFFICIENT_BUFFER: errno.ENOMEM,
         winerror.ERROR_INVALID_ACCESS: errno.EACCES,
         winerror.ERROR_INVALID_ADDRESS: errno.EFAULT,
         winerror.ERROR_INVALID_BLOCK: errno.EFAULT,
         winerror.ERROR_INVALID_DATA: errno.EINVAL,
         winerror.ERROR_INVALID_DRIVE: errno.ENODEV,
         winerror.ERROR_INVALID_EXE_SIGNATURE: errno.ENOEXEC,
         winerror.ERROR_INVALID_FLAGS: errno.EINVAL,
         winerror.ERROR_INVALID_FUNCTION: errno.ENOSYS,
         winerror.ERROR_INVALID_HANDLE: errno.EBADF,
         winerror.ERROR_INVALID_LOGON_HOURS: errno.EACCES,
         winerror.ERROR_INVALID_NAME: errno.EINVAL,
         winerror.ERROR_INVALID_OWNER: errno.EINVAL,
         winerror.ERROR_INVALID_PARAMETER: errno.EINVAL,
         winerror.ERROR_INVALID_PASSWORD: errno.EPERM,
         winerror.ERROR_INVALID_PRIMARY_GROUP: errno.EINVAL,
         winerror.ERROR_INVALID_SIGNAL_NUMBER: errno.EINVAL,
         winerror.ERROR_INVALID_TARGET_HANDLE: errno.EIO,
         winerror.ERROR_INVALID_WORKSTATION: errno.EACCES,
         winerror.ERROR_IO_DEVICE: errno.EIO,
         winerror.ERROR_IO_INCOMPLETE: errno.EINTR,
         winerror.ERROR_LOCKED: errno.EBUSY,
         winerror.ERROR_LOCK_VIOLATION: errno.EACCES,
         winerror.ERROR_LOGON_FAILURE: errno.EACCES,
         winerror.ERROR_MAPPED_ALIGNMENT: errno.EINVAL,
         winerror.ERROR_META_EXPANSION_TOO_LONG: errno.E2BIG,
         winerror.ERROR_MORE_DATA: errno.EPIPE,
         winerror.ERROR_NEGATIVE_SEEK: errno.ESPIPE,
         winerror.ERROR_NOACCESS: errno.EFAULT,
         winerror.ERROR_NONE_MAPPED: errno.EINVAL,
         winerror.ERROR_NOT_ENOUGH_MEMORY: errno.ENOMEM,
         winerror.ERROR_NOT_READY: errno.EAGAIN,
         winerror.ERROR_NOT_SAME_DEVICE: errno.EXDEV,
         winerror.ERROR_NO_DATA: errno.EPIPE,
         winerror.ERROR_NO_MORE_SEARCH_HANDLES: errno.EIO,
         winerror.ERROR_NO_PROC_SLOTS: errno.EAGAIN,
         winerror.ERROR_NO_SUCH_PRIVILEGE: errno.EACCES,
         winerror.ERROR_OPEN_FAILED: errno.EIO,
         winerror.ERROR_OPEN_FILES: errno.EBUSY,
         winerror.ERROR_OPERATION_ABORTED: errno.EINTR,
         winerror.ERROR_OUTOFMEMORY: errno.ENOMEM,
         winerror.ERROR_PASSWORD_EXPIRED: errno.EACCES,
         winerror.ERROR_PATH_BUSY: errno.EBUSY,
-        winerror.ERROR_PATH_NOT_FOUND: errno.ENOTDIR,
+        winerror.ERROR_PATH_NOT_FOUND: errno.ENOENT,
         winerror.ERROR_PIPE_BUSY: errno.EBUSY,
         winerror.ERROR_PIPE_CONNECTED: errno.EPIPE,
         winerror.ERROR_PIPE_LISTENING: errno.EPIPE,
         winerror.ERROR_PIPE_NOT_CONNECTED: errno.EPIPE,
         winerror.ERROR_PRIVILEGE_NOT_HELD: errno.EACCES,
         winerror.ERROR_READ_FAULT: errno.EIO,
         winerror.ERROR_SEEK: errno.EIO,
         winerror.ERROR_SEEK_ON_DEVICE: errno.ESPIPE,
         winerror.ERROR_SHARING_BUFFER_EXCEEDED: errno.ENFILE,
         winerror.ERROR_SHARING_VIOLATION: errno.EACCES,
         winerror.ERROR_STACK_OVERFLOW: errno.ENOMEM,
         winerror.ERROR_SWAPERROR: errno.ENOENT,
         winerror.ERROR_TOO_MANY_MODULES: errno.EMFILE,
         winerror.ERROR_TOO_MANY_OPEN_FILES: errno.EMFILE,
         winerror.ERROR_UNRECOGNIZED_MEDIA: errno.ENXIO,
         winerror.ERROR_UNRECOGNIZED_VOLUME: errno.ENODEV,
         winerror.ERROR_WAIT_NO_CHILDREN: errno.ECHILD,
         winerror.ERROR_WRITE_FAULT: errno.EIO,
         winerror.ERROR_WRITE_PROTECT: errno.EROFS,
         }
 
     def __init__(self, err):
         self.win_errno, self.win_function, self.win_strerror = err
+        if self.win_strerror.endswith('.'):
+            self.win_strerror = self.win_strerror[:-1]
+
+class WinIOError(WinError, IOError):
+    def __init__(self, err, filename=None):
+        WinError.__init__(self, err)
+        IOError.__init__(self, self.winerror_map.get(self.win_errno, 0),
+                         self.win_strerror)
+        self.filename = filename
+
+class WinOSError(WinError, OSError):
+    def __init__(self, err):
+        WinError.__init__(self, err)
         OSError.__init__(self, self.winerror_map.get(self.win_errno, 0),
                          self.win_strerror)
 
 def os_link(src, dst):
     # NB will only succeed on NTFS
     try:
         win32file.CreateHardLink(dst, src)
     except pywintypes.error, details:
-        raise WinError(details)
+        raise WinOSError(details)
 
 def nlinks(pathname):
     """Return number of hardlinks for the given file."""
     try:
         fh = win32file.CreateFile(pathname,
             win32file.GENERIC_READ, win32file.FILE_SHARE_READ,
             None, win32file.OPEN_EXISTING, 0, None)
         res = win32file.GetFileInformationByHandle(fh)
         fh.Close()
         return res[7]
     except pywintypes.error:
         return os.stat(pathname).st_nlink
 
 def testpid(pid):
     '''return True if pid is still running or unable to
     determine, False otherwise'''
     try:
         handle = win32api.OpenProcess(
             win32con.PROCESS_QUERY_INFORMATION, False, pid)
         if handle:
             status = win32process.GetExitCodeProcess(handle)
             return status == win32con.STILL_ACTIVE
     except pywintypes.error, details:
         return details[0] != winerror.ERROR_INVALID_PARAMETER
     return True
 
 def system_rcpath_win32():
     '''return default os-specific hgrc search path'''
     proc = win32api.GetCurrentProcess()
     filename = win32process.GetModuleFileNameEx(proc, 0)
     return [os.path.join(os.path.dirname(filename), 'mercurial.ini')]
+
+class posixfile(object):
+    '''file object with posix-like semantics.  on windows, normal
+    files can not be deleted or renamed if they are open. must open
+    with win32file.FILE_SHARE_DELETE. this flag does not exist on
+    windows <= nt.'''
+
+    # tried to use win32file._open_osfhandle to pass fd to os.fdopen,
+    # but does not work at all. wrap win32 file api instead.
+
+    def __init__(self, name, mode='rb'):
+        access = 0
+        if 'r' in mode or '+' in mode:
+            access |= win32file.GENERIC_READ
+        if 'w' in mode or 'a' in mode:
+            access |= win32file.GENERIC_WRITE
+        if 'r' in mode:
+            creation = win32file.OPEN_EXISTING
+        elif 'a' in mode:
+            creation = win32file.OPEN_ALWAYS
+        else:
+            creation = win32file.CREATE_ALWAYS
+        try:
+            self.handle = win32file.CreateFile(name,
+                                               access,
+                                               win32file.FILE_SHARE_READ |
+                                               win32file.FILE_SHARE_WRITE |
+                                               win32file.FILE_SHARE_DELETE,
+                                               None,
+                                               creation,
+                                               win32file.FILE_ATTRIBUTE_NORMAL,
+                                               0)
+        except pywintypes.error, err:
+            raise WinIOError(err, name)
+        self.closed = False
+        self.name = name
+        self.mode = mode
+
+    def read(self, count=-1):
+        try:
+            cs = cStringIO.StringIO()
+            while count:
+                wincount = int(count)
+                if wincount == -1:
+                    wincount = 1048576
+                val, data = win32file.ReadFile(self.handle, wincount)
+                if not data: break
+                cs.write(data)
+                if count != -1:
+                    count -= len(data)
+            return cs.getvalue()
+        except pywintypes.error, err:
+            raise WinIOError(err)
+
+    def write(self, data):
+        try:
+            if 'a' in self.mode:
+                win32file.SetFilePointer(self.handle, 0, win32file.FILE_END)
+            nwrit = 0
+            while nwrit < len(data):
+                val, nwrit = win32file.WriteFile(self.handle, data)
+                data = data[nwrit:]
+        except pywintypes.error, err:
+            raise WinIOError(err)
+
+    def seek(self, pos, whence=0):
+        try:
+            win32file.SetFilePointer(self.handle, int(pos), whence)
+        except pywintypes.error, err:
+            raise WinIOError(err)
+
+    def tell(self):
+        try:
+            return win32file.SetFilePointer(self.handle, 0,
+                                            win32file.FILE_CURRENT)
+        except pywintypes.error, err:
+            raise WinIOError(err)
+
+    def close(self):
+        if not self.closed:
+            self.handle = None
+            self.closed = True
+
+    def flush(self):
+        try:
+            win32file.FlushFileBuffers(self.handle)
+        except pywintypes.error, err:
+            raise WinIOError(err)
+
+    def truncate(self, pos=0):
+        try:
+            win32file.SetFilePointer(self.handle, int(pos),
+                                     win32file.FILE_BEGIN)
+            win32file.SetEndOfFile(self.handle)
+        except pywintypes.error, err:
+            raise WinIOError(err)