issue352: disallow '\n' and '\r' in filenames (dirstate and manifest)

tests/test-issue352

@@ -0,0 +1,21
+#!/bin/bash
+# http://www.selenic.com/mercurial/bts/issue352
+
+hg init foo
+cd foo
+
+A=`echo -e -n 'he\rllo'`
+
+echo foo > "hell
+o"
+echo foo > "$A"
+hg add
+hg ci -A -m m
+rm "$A"
+ls
+hg add
+# BUG ? we don't walk on filenames with '\n' (regexp related) ?
+hg debugwalk
+hg ci -A -m m
+
+exit 0

tests/test-issue352.out

@@ -0,0 +1,7
+adding he
llo
+abort: '\n' and '\r' disallowed in filenames
+adding he
llo
+abort: '\n' and '\r' disallowed in filenames
+hell
+o
+nothing changed

mercurial/dirstate.py

@@ -1,525 +1,529
 """
 dirstate.py - working directory tracking for mercurial
 
 Copyright 2005, 2006 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(globals(), "struct os time bisect stat strutil util re errno")
 
 class dirstate(object):
     format = ">cllll"
 
     def __init__(self, opener, ui, root):
         self.opener = opener
         self.root = root
         self.dirty = 0
         self.ui = ui
         self.map = None
         self.pl = None
         self.dirs = None
         self.copymap = {}
         self.ignorefunc = None
 
     def wjoin(self, f):
         return os.path.join(self.root, f)
 
     def getcwd(self):
         cwd = os.getcwd()
         if cwd == self.root: return ''
         return cwd[len(self.root) + 1:]
 
     def hgignore(self):
         '''return the contents of .hgignore files as a list of patterns.
 
         the files parsed for patterns include:
         .hgignore in the repository root
         any additional files specified in the [ui] section of ~/.hgrc
 
         trailing white space is dropped.
         the escape character is backslash.
         comments start with #.
         empty lines are skipped.
 
         lines can be of the following formats:
 
         syntax: regexp # defaults following lines to non-rooted regexps
         syntax: glob   # defaults following lines to non-rooted globs
         re:pattern     # non-rooted regular expression
         glob:pattern   # non-rooted glob
         pattern        # pattern of the current default type'''
         syntaxes = {'re': 'relre:', 'regexp': 'relre:', 'glob': 'relglob:'}
         def parselines(fp):
             for line in fp:
                 escape = False
                 for i in xrange(len(line)):
                     if escape: escape = False
                     elif line[i] == '\\': escape = True
                     elif line[i] == '#': break
                 line = line[:i].rstrip()
                 if line: yield line
         repoignore = self.wjoin('.hgignore')
         files = [repoignore]
         files.extend(self.ui.hgignorefiles())
         pats = {}
         for f in files:
             try:
                 pats[f] = []
                 fp = open(f)
                 syntax = 'relre:'
                 for line in parselines(fp):
                     if line.startswith('syntax:'):
                         s = line[7:].strip()
                         try:
                             syntax = syntaxes[s]
                         except KeyError:
                             self.ui.warn(_("%s: ignoring invalid "
                                            "syntax '%s'\n") % (f, s))
                         continue
                     pat = syntax + line
                     for s in syntaxes.values():
                         if line.startswith(s):
                             pat = line
                             break
                     pats[f].append(pat)
             except IOError, inst:
                 if f != repoignore:
                     self.ui.warn(_("skipping unreadable ignore file"
                                    " '%s': %s\n") % (f, inst.strerror))
         return pats
 
     def ignore(self, fn):
         '''default match function used by dirstate and
         localrepository.  this honours the repository .hgignore file
         and any other files specified in the [ui] section of .hgrc.'''
         if not self.ignorefunc:
             ignore = self.hgignore()
             allpats = []
             [allpats.extend(patlist) for patlist in ignore.values()]
             if allpats:
                 try:
                     files, self.ignorefunc, anypats = (
                         util.matcher(self.root, inc=allpats, src='.hgignore'))
                 except util.Abort:
                     # Re-raise an exception where the src is the right file
                     for f, patlist in ignore.items():
                         files, self.ignorefunc, anypats = (
                             util.matcher(self.root, inc=patlist, src=f))
             else:
                 self.ignorefunc = util.never
         return self.ignorefunc(fn)
 
     def __del__(self):
         if self.dirty:
             self.write()
 
     def __getitem__(self, key):
         try:
             return self.map[key]
         except TypeError:
             self.lazyread()
             return self[key]
 
     def __contains__(self, key):
         self.lazyread()
         return key in self.map
 
     def parents(self):
         self.lazyread()
         return self.pl
 
     def markdirty(self):
         if not self.dirty:
             self.dirty = 1
 
     def setparents(self, p1, p2=nullid):
         self.lazyread()
         self.markdirty()
         self.pl = p1, p2
 
     def state(self, key):
         try:
             return self[key][0]
         except KeyError:
             return "?"
 
     def lazyread(self):
         if self.map is None:
             self.read()
 
     def parse(self, st):
         self.pl = [st[:20], st[20: 40]]
 
         # deref fields so they will be local in loop
         map = self.map
         copymap = self.copymap
         format = self.format
         unpack = struct.unpack
 
         pos = 40
         e_size = struct.calcsize(format)
 
         while pos < len(st):
             newpos = pos + e_size
             e = unpack(format, st[pos:newpos])
             l = e[4]
             pos = newpos
             newpos = pos + l
             f = st[pos:newpos]
             if '\0' in f:
                 f, c = f.split('\0')
                 copymap[f] = c
             map[f] = e[:4]
             pos = newpos
 
     def read(self):
         self.map = {}
         self.pl = [nullid, nullid]
         try:
             st = self.opener("dirstate").read()
             if st:
                 self.parse(st)
         except IOError, err:
             if err.errno != errno.ENOENT: raise
 
     def copy(self, source, dest):
         self.lazyread()
         self.markdirty()
         self.copymap[dest] = source
 
     def copied(self, file):
         return self.copymap.get(file, None)
 
     def copies(self):
         return self.copymap
 
     def initdirs(self):
         if self.dirs is None:
             self.dirs = {}
             for f in self.map:
                 self.updatedirs(f, 1)
 
     def updatedirs(self, path, delta):
         if self.dirs is not None:
             for c in strutil.findall(path, '/'):
                 pc = path[:c]
                 self.dirs.setdefault(pc, 0)
                 self.dirs[pc] += delta
 
-    def checkshadows(self, files):
+    def checkinterfering(self, files):
         def prefixes(f):
             for c in strutil.rfindall(f, '/'):
                 yield f[:c]
         self.lazyread()
         self.initdirs()
         seendirs = {}
         for f in files:
+            # shadows
             if self.dirs.get(f):
                 raise util.Abort(_('directory named %r already in dirstate') %
                                  f)
             for d in prefixes(f):
                 if d in seendirs:
                     break
                 if d in self.map:
                     raise util.Abort(_('file named %r already in dirstate') %
                                      d)
                 seendirs[d] = True
+            # disallowed
+            if '\r' in f or '\n' in f:
+                raise util.Abort(_("'\\n' and '\\r' disallowed in filenames"))
 
     def update(self, files, state, **kw):
         ''' current states:
         n  normal
         m  needs merging
         r  marked for removal
         a  marked for addition'''
 
         if not files: return
         self.lazyread()
         self.markdirty()
         if state == "a":
             self.initdirs()
-            self.checkshadows(files)
+            self.checkinterfering(files)
         for f in files:
             if state == "r":
                 self.map[f] = ('r', 0, 0, 0)
                 self.updatedirs(f, -1)
             else:
                 if state == "a":
                     self.updatedirs(f, 1)
                 s = os.lstat(self.wjoin(f))
                 st_size = kw.get('st_size', s.st_size)
                 st_mtime = kw.get('st_mtime', s.st_mtime)
                 self.map[f] = (state, s.st_mode, st_size, st_mtime)
             if self.copymap.has_key(f):
                 del self.copymap[f]
 
     def forget(self, files):
         if not files: return
         self.lazyread()
         self.markdirty()
         self.initdirs()
         for f in files:
             try:
                 del self.map[f]
                 self.updatedirs(f, -1)
             except KeyError:
                 self.ui.warn(_("not in dirstate: %s!\n") % f)
                 pass
 
     def clear(self):
         self.map = {}
         self.copymap = {}
         self.dirs = None
         self.markdirty()
 
     def rebuild(self, parent, files):
         self.clear()
         umask = os.umask(0)
         os.umask(umask)
         for f in files:
             if files.execf(f):
                 self.map[f] = ('n', ~umask, -1, 0)
             else:
                 self.map[f] = ('n', ~umask & 0666, -1, 0)
         self.pl = (parent, nullid)
         self.markdirty()
 
     def write(self):
         if not self.dirty:
             return
         st = self.opener("dirstate", "w", atomic=True)
         st.write("".join(self.pl))
         for f, e in self.map.items():
             c = self.copied(f)
             if c:
                 f = f + "\0" + c
             e = struct.pack(self.format, e[0], e[1], e[2], e[3], len(f))
             st.write(e + f)
         self.dirty = 0
 
     def filterfiles(self, files):
         ret = {}
         unknown = []
 
         for x in files:
             if x == '.':
                 return self.map.copy()
             if x not in self.map:
                 unknown.append(x)
             else:
                 ret[x] = self.map[x]
 
         if not unknown:
             return ret
 
         b = self.map.keys()
         b.sort()
         blen = len(b)
 
         for x in unknown:
             bs = bisect.bisect(b, "%s%s" % (x, '/'))
             while bs < blen:
                 s = b[bs]
                 if len(s) > len(x) and s.startswith(x):
                     ret[s] = self.map[s]
                 else:
                     break
                 bs += 1
         return ret
 
     def supported_type(self, f, st, verbose=False):
         if stat.S_ISREG(st.st_mode):
             return True
         if verbose:
             kind = 'unknown'
             if stat.S_ISCHR(st.st_mode): kind = _('character device')
             elif stat.S_ISBLK(st.st_mode): kind = _('block device')
             elif stat.S_ISFIFO(st.st_mode): kind = _('fifo')
             elif stat.S_ISLNK(st.st_mode): kind = _('symbolic link')
             elif stat.S_ISSOCK(st.st_mode): kind = _('socket')
             elif stat.S_ISDIR(st.st_mode): kind = _('directory')
             self.ui.warn(_('%s: unsupported file type (type is %s)\n') % (
                 util.pathto(self.getcwd(), f),
                 kind))
         return False
 
     def walk(self, files=None, match=util.always, badmatch=None):
         # filter out the stat
         for src, f, st in self.statwalk(files, match, badmatch=badmatch):
             yield src, f
 
     def statwalk(self, files=None, match=util.always, ignored=False,
                  badmatch=None):
         '''
         walk recursively through the directory tree, finding all files
         matched by the match function
 
         results are yielded in a tuple (src, filename, st), where src
         is one of:
         'f' the file was found in the directory tree
         'm' the file was only in the dirstate and not in the tree
         'b' file was not found and matched badmatch
 
         and st is the stat result if the file was found in the directory.
         '''
         self.lazyread()
 
         # walk all files by default
         if not files:
             files = [self.root]
             dc = self.map.copy()
         else:
             files = util.unique(files)
             dc = self.filterfiles(files)
 
         def imatch(file_):
             if file_ not in dc and self.ignore(file_):
                 return False
             return match(file_)
 
         if ignored: imatch = match
 
         # self.root may end with a path separator when self.root == '/'
         common_prefix_len = len(self.root)
         if not self.root.endswith('/'):
             common_prefix_len += 1
         # recursion free walker, faster than os.walk.
         def findfiles(s):
             work = [s]
             while work:
                 top = work.pop()
                 names = os.listdir(top)
                 names.sort()
                 # nd is the top of the repository dir tree
                 nd = util.normpath(top[common_prefix_len:])
                 if nd == '.':
                     nd = ''
                 else:
                     # do not recurse into a repo contained in this
                     # one. use bisect to find .hg directory so speed
                     # is good on big directory.
                     hg = bisect.bisect_left(names, '.hg')
                     if hg < len(names) and names[hg] == '.hg':
                         if os.path.isdir(os.path.join(top, '.hg')):
                             continue
                 for f in names:
                     np = util.pconvert(os.path.join(nd, f))
                     if seen(np):
                         continue
                     p = os.path.join(top, f)
                     # don't trip over symlinks
                     st = os.lstat(p)
                     if stat.S_ISDIR(st.st_mode):
                         ds = util.pconvert(os.path.join(nd, f +'/'))
                         if imatch(ds):
                             work.append(p)
                         if imatch(np) and np in dc:
                             yield 'm', np, st
                     elif imatch(np):
                         if self.supported_type(np, st):
                             yield 'f', np, st
                         elif np in dc:
                             yield 'm', np, st
 
         known = {'.hg': 1}
         def seen(fn):
             if fn in known: return True
             known[fn] = 1
 
         # step one, find all files that match our criteria
         files.sort()
         for ff in files:
             nf = util.normpath(ff)
             f = self.wjoin(ff)
             try:
                 st = os.lstat(f)
             except OSError, inst:
                 found = False
                 for fn in dc:
                     if nf == fn or (fn.startswith(nf) and fn[len(nf)] == '/'):
                         found = True
                         break
                 if not found:
                     if inst.errno != errno.ENOENT or not badmatch:
                         self.ui.warn('%s: %s\n' % (
                             util.pathto(self.getcwd(), ff),
                             inst.strerror))
                     elif badmatch and badmatch(ff) and imatch(nf):
                         yield 'b', ff, None
                 continue
             if stat.S_ISDIR(st.st_mode):
                 cmp1 = (lambda x, y: cmp(x[1], y[1]))
                 sorted_ = [ x for x in findfiles(f) ]
                 sorted_.sort(cmp1)
                 for e in sorted_:
                     yield e
             else:
                 if not seen(nf) and match(nf):
                     if self.supported_type(ff, st, verbose=True):
                         yield 'f', nf, st
                     elif ff in dc:
                         yield 'm', nf, st
 
         # step two run through anything left in the dc hash and yield
         # if we haven't already seen it
         ks = dc.keys()
         ks.sort()
         for k in ks:
             if not seen(k) and imatch(k):
                 yield 'm', k, None
 
     def status(self, files=None, match=util.always, list_ignored=False,
                list_clean=False):
         lookup, modified, added, unknown, ignored = [], [], [], [], []
         removed, deleted, clean = [], [], []
 
         for src, fn, st in self.statwalk(files, match, ignored=list_ignored):
             try:
                 type_, mode, size, time = self[fn]
             except KeyError:
                 if list_ignored and self.ignore(fn):
                     ignored.append(fn)
                 else:
                     unknown.append(fn)
                 continue
             if src == 'm':
                 nonexistent = True
                 if not st:
                     try:
                         st = os.lstat(self.wjoin(fn))
                     except OSError, inst:
                         if inst.errno != errno.ENOENT:
                             raise
                         st = None
                     # We need to re-check that it is a valid file
                     if st and self.supported_type(fn, st):
                         nonexistent = False
                 # XXX: what to do with file no longer present in the fs
                 # who are not removed in the dirstate ?
                 if nonexistent and type_ in "nm":
                     deleted.append(fn)
                     continue
             # check the common case first
             if type_ == 'n':
                 if not st:
                     st = os.lstat(self.wjoin(fn))
                 if size >= 0 and (size != st.st_size
                                   or (mode ^ st.st_mode) & 0100):
                     modified.append(fn)
                 elif time != int(st.st_mtime):
                     lookup.append(fn)
                 elif list_clean:
                     clean.append(fn)
             elif type_ == 'm':
                 modified.append(fn)
             elif type_ == 'a':
                 added.append(fn)
             elif type_ == 'r':
                 removed.append(fn)
 
         return (lookup, modified, added, removed, deleted, unknown, ignored,
                 clean)

mercurial/manifest.py

@@ -1,209 +1,218
 # manifest.py - manifest revision class for mercurial
 #
 # Copyright 2005, 2006 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 revlog import *
 from i18n import gettext as _
 from demandload import *
 demandload(globals(), "array bisect struct")
 demandload(globals(), "mdiff")
 
 class manifestdict(dict):
     def __init__(self, mapping=None, flags=None):
         if mapping is None: mapping = {}
         if flags is None: flags = {}
         dict.__init__(self, mapping)
         self._flags = flags
     def flags(self, f):
         return self._flags.get(f, "")
     def execf(self, f):
         "test for executable in manifest flags"
         return "x" in self.flags(f)
     def linkf(self, f):
         "test for symlink in manifest flags"
         return "l" in self.flags(f)
     def rawset(self, f, entry):
         self[f] = bin(entry[:40])
         fl = entry[40:-1]
         if fl: self._flags[f] = fl
     def set(self, f, execf=False, linkf=False):
         if linkf: self._flags[f] = "l"
         elif execf: self._flags[f] = "x"
         else: self._flags[f] = ""
     def copy(self):
         return manifestdict(dict.copy(self), dict.copy(self._flags))
 
 class manifest(revlog):
     def __init__(self, opener, defversion=REVLOGV0):
         self.mapcache = None
         self.listcache = None
         revlog.__init__(self, opener, "00manifest.i", "00manifest.d",
                         defversion)
 
     def parselines(self, lines):
         for l in lines.splitlines(1):
             yield l.split('\0')
 
     def readdelta(self, node):
         delta = mdiff.patchtext(self.delta(node))
         deltamap = manifestdict()
         for f, n in self.parselines(delta):
             deltamap.rawset(f, n)
         return deltamap
 
     def read(self, node):
         if node == nullid: return manifestdict() # don't upset local cache
         if self.mapcache and self.mapcache[0] == node:
             return self.mapcache[1]
         text = self.revision(node)
         self.listcache = array.array('c', text)
         mapping = manifestdict()
         for f, n in self.parselines(text):
             mapping.rawset(f, n)
         self.mapcache = (node, mapping)
         return mapping
 
     def _search(self, m, s, lo=0, hi=None):
         '''return a tuple (start, end) that says where to find s within m.
 
         If the string is found m[start:end] are the line containing
         that string.  If start == end the string was not found and
         they indicate the proper sorted insertion point.  This was
         taken from bisect_left, and modified to find line start/end as
         it goes along.
 
         m should be a buffer or a string
         s is a string'''
         def advance(i, c):
             while i < lenm and m[i] != c:
                 i += 1
             return i
         lenm = len(m)
         if not hi:
             hi = lenm
         while lo < hi:
             mid = (lo + hi) // 2
             start = mid
             while start > 0 and m[start-1] != '\n':
                 start -= 1
             end = advance(start, '\0')
             if m[start:end] < s:
                 # we know that after the null there are 40 bytes of sha1
                 # this translates to the bisect lo = mid + 1
                 lo = advance(end + 40, '\n') + 1
             else:
                 # this translates to the bisect hi = mid
                 hi = start
         end = advance(lo, '\0')
         found = m[lo:end]
         if cmp(s, found) == 0:
             # we know that after the null there are 40 bytes of sha1
             end = advance(end + 40, '\n')
             return (lo, end+1)
         else:
             return (lo, lo)
 
     def find(self, node, f):
         '''look up entry for a single file efficiently.
         return (node, flag) pair if found, (None, None) if not.'''
         if self.mapcache and node == self.mapcache[0]:
             return self.mapcache[1].get(f), self.mapcache[1].flags(f)
         text = self.revision(node)
         start, end = self._search(text, f)
         if start == end:
             return None, None
         l = text[start:end]
         f, n = l.split('\0')
         return bin(n[:40]), n[40:-1] == 'x'
 
     def add(self, map, transaction, link, p1=None, p2=None,
             changed=None):
         # apply the changes collected during the bisect loop to our addlist
         # return a delta suitable for addrevision
         def addlistdelta(addlist, x):
             # start from the bottom up
             # so changes to the offsets don't mess things up.
             i = len(x)
             while i > 0:
                 i -= 1
                 start = x[i][0]
                 end = x[i][1]
                 if x[i][2]:
                     addlist[start:end] = array.array('c', x[i][2])
                 else:
                     del addlist[start:end]
             return "".join([struct.pack(">lll", d[0], d[1], len(d[2])) + d[2] \
                             for d in x ])
 
+        def checkforbidden(f):
+            if '\n' in f or '\r' in f:
+                raise RevlogError(_("'\\n' and '\\r' disallowed in filenames"))
+
         # if we're using the listcache, make sure it is valid and
         # parented by the same node we're diffing against
         if not changed or not self.listcache or not p1 or \
                self.mapcache[0] != p1:
             files = map.keys()
             files.sort()
 
+            for f in files:
+                checkforbidden(f)
+
             # if this is changed to support newlines in filenames,
             # be sure to check the templates/ dir again (especially *-raw.tmpl)
             text = ["%s\000%s%s\n" % (f, hex(map[f]), map.flags(f)) for f in files]
             self.listcache = array.array('c', "".join(text))
             cachedelta = None
         else:
             addlist = self.listcache
 
+            for f in changed[0]:
+                checkforbidden(f)
             # combine the changed lists into one list for sorting
             work = [[x, 0] for x in changed[0]]
             work[len(work):] = [[x, 1] for x in changed[1]]
             work.sort()
 
             delta = []
             dstart = None
             dend = None
             dline = [""]
             start = 0
             # zero copy representation of addlist as a buffer
             addbuf = buffer(addlist)
 
             # start with a readonly loop that finds the offset of
             # each line and creates the deltas
             for w in work:
                 f = w[0]
                 # bs will either be the index of the item or the insert point
                 start, end = self._search(addbuf, f, start)
                 if w[1] == 0:
                     l = "%s\000%s%s\n" % (f, hex(map[f]), map.flags(f))
                 else:
                     l = ""
                 if start == end and w[1] == 1:
                     # item we want to delete was not found, error out
                     raise AssertionError(
                             _("failed to remove %s from manifest") % f)
                 if dstart != None and dstart <= start and dend >= start:
                     if dend < end:
                         dend = end
                     if l:
                         dline.append(l)
                 else:
                     if dstart != None:
                         delta.append([dstart, dend, "".join(dline)])
                     dstart = start
                     dend = end
                     dline = [l]
 
             if dstart != None:
                 delta.append([dstart, dend, "".join(dline)])
             # apply the delta to the addlist, and get a delta for addrevision
             cachedelta = addlistdelta(addlist, delta)
 
             # the delta is only valid if we've been processing the tip revision
             if self.mapcache[0] != self.tip():
                 cachedelta = None
             self.listcache = addlist
 
         n = self.addrevision(buffer(self.listcache), transaction, link, p1,  \
                              p2, cachedelta)
         self.mapcache = (n, map)
 
         return n