import os.path
import urllib
import simplejson
+from collections import defaultdict
+from cStringIO import StringIO
from twisted.python.failure import Failure
-from allmydata.scripts.common import get_alias, escape_path, DefaultAliasMarker
-from allmydata.scripts.common_http import do_http
+from allmydata.scripts.common import get_alias, escape_path, \
+ DefaultAliasMarker, TahoeError
+from allmydata.scripts.common_http import do_http, HTTPError
from allmydata import uri
+from allmydata.util import fileutil
+from allmydata.util.fileutil import abspath_expanduser_unicode, precondition_abspath
+from allmydata.util.encodingutil import unicode_to_url, listdir_unicode, quote_output, \
+ quote_local_unicode_path, to_str
+from allmydata.util.assertutil import precondition, _assert
-def ascii_or_none(s):
- if s is None:
- return s
- return str(s)
-class TahoeError(Exception):
- def __init__(self, msg, resp):
- self.msg = msg
- self.status = resp.status
- self.reason = resp.reason
- self.body = resp.read()
+class MissingSourceError(TahoeError):
+ def __init__(self, name, quotefn=quote_output):
+ TahoeError.__init__(self, "No such file or directory %s" % quotefn(name))
- def display(self, err):
- print >>err, "%s: %s %s" % (self.msg, self.status, self.reason)
- print >>err, self.body
-
-class MissingSourceError(Exception):
- pass
def GET_to_file(url):
resp = do_http("GET", url)
if resp.status == 200:
return resp
- raise TahoeError("Error during GET", resp)
+ raise HTTPError("Error during GET", resp)
def GET_to_string(url):
f = GET_to_file(url)
resp = do_http("PUT", url, data)
if resp.status in (200, 201):
return resp.read()
- raise TahoeError("Error during PUT", resp)
+ raise HTTPError("Error during PUT", resp)
def POST(url, data):
resp = do_http("POST", url, data)
if resp.status in (200, 201):
return resp.read()
- raise TahoeError("Error during POST", resp)
+ raise HTTPError("Error during POST", resp)
def mkdir(targeturl):
url = targeturl + "?t=mkdir"
resp = do_http("POST", url)
if resp.status in (200, 201):
return resp.read().strip()
- raise TahoeError("Error during mkdir", resp)
+ raise HTTPError("Error during mkdir", resp)
def make_tahoe_subdirectory(nodeurl, parent_writecap, name):
url = nodeurl + "/".join(["uri",
urllib.quote(parent_writecap),
- urllib.quote(name),
+ urllib.quote(unicode_to_url(name)),
]) + "?t=mkdir"
resp = do_http("POST", url)
if resp.status in (200, 201):
return resp.read().strip()
- raise TahoeError("Error during mkdir", resp)
+ raise HTTPError("Error during mkdir", resp)
class LocalFileSource:
- def __init__(self, pathname):
+ def __init__(self, pathname, basename):
+ precondition_abspath(pathname)
self.pathname = pathname
+ self._basename = basename
+
+ def basename(self):
+ return self._basename
def need_to_copy_bytes(self):
return True
- def open(self):
+ def open(self, caps_only):
return open(self.pathname, "rb")
class LocalFileTarget:
def __init__(self, pathname):
+ precondition_abspath(pathname)
self.pathname = pathname
+
def put_file(self, inf):
- outf = open(self.pathname, "wb")
- while True:
- data = inf.read(32768)
- if not data:
- break
- outf.write(data)
- outf.close()
+ fileutil.put_file(self.pathname, inf)
class LocalMissingTarget:
def __init__(self, pathname):
+ precondition_abspath(pathname)
self.pathname = pathname
def put_file(self, inf):
- outf = open(self.pathname, "wb")
- while True:
- data = inf.read(32768)
- if not data:
- break
- outf.write(data)
- outf.close()
+ fileutil.put_file(self.pathname, inf)
class LocalDirectorySource:
- def __init__(self, progressfunc, pathname):
+ def __init__(self, progressfunc, pathname, basename):
+ precondition_abspath(pathname)
+
self.progressfunc = progressfunc
self.pathname = pathname
self.children = None
+ self._basename = basename
+
+ def basename(self):
+ return self._basename
def populate(self, recurse):
if self.children is not None:
return
self.children = {}
- children = os.listdir(self.pathname)
+ children = listdir_unicode(self.pathname)
for i,n in enumerate(children):
- self.progressfunc("examining %d of %d" % (i, len(children)))
+ self.progressfunc("examining %d of %d" % (i+1, len(children)))
pn = os.path.join(self.pathname, n)
if os.path.isdir(pn):
- child = LocalDirectorySource(self.progressfunc, pn)
+ child = LocalDirectorySource(self.progressfunc, pn, n)
self.children[n] = child
if recurse:
- child.populate(True)
+ child.populate(recurse=True)
elif os.path.isfile(pn):
- self.children[n] = LocalFileSource(pn)
+ self.children[n] = LocalFileSource(pn, n)
else:
# Could be dangling symlink; probably not copy-able.
+ # TODO: output a warning
pass
class LocalDirectoryTarget:
def __init__(self, progressfunc, pathname):
+ precondition_abspath(pathname)
+
self.progressfunc = progressfunc
self.pathname = pathname
self.children = None
if self.children is not None:
return
self.children = {}
- children = os.listdir(self.pathname)
+ children = listdir_unicode(self.pathname)
for i,n in enumerate(children):
- self.progressfunc("examining %d of %d" % (i, len(children)))
+ self.progressfunc("examining %d of %d" % (i+1, len(children)))
pn = os.path.join(self.pathname, n)
if os.path.isdir(pn):
child = LocalDirectoryTarget(self.progressfunc, pn)
self.children[n] = child
if recurse:
- child.populate(True)
+ child.populate(recurse=True)
else:
assert os.path.isfile(pn)
self.children[n] = LocalFileTarget(pn)
def get_child_target(self, name):
+ precondition(isinstance(name, unicode), name)
if self.children is None:
- self.populate(False)
+ self.populate(recurse=False)
if name in self.children:
return self.children[name]
pathname = os.path.join(self.pathname, name)
os.makedirs(pathname)
- return LocalDirectoryTarget(self.progressfunc, pathname)
+ child = LocalDirectoryTarget(self.progressfunc, pathname)
+ self.children[name] = child
+ return child
def put_file(self, name, inf):
+ precondition(isinstance(name, unicode), name)
pathname = os.path.join(self.pathname, name)
- outf = open(pathname, "wb")
- while True:
- data = inf.read(32768)
- if not data:
- break
- outf.write(data)
- outf.close()
+ fileutil.put_file(pathname, inf)
def set_children(self):
pass
+
class TahoeFileSource:
- def __init__(self, nodeurl, mutable, writecap, readcap):
+ def __init__(self, nodeurl, mutable, writecap, readcap, basename):
self.nodeurl = nodeurl
self.mutable = mutable
self.writecap = writecap
self.readcap = readcap
+ self._basename = basename # unicode, or None for raw filecaps
+
+ def basename(self):
+ return self._basename
def need_to_copy_bytes(self):
if self.mutable:
return True
return False
- def open(self):
+ def open(self, caps_only):
+ if caps_only:
+ return StringIO(self.readcap)
url = self.nodeurl + "uri/" + urllib.quote(self.readcap)
return GET_to_file(url)
PUT(self.url, inf)
# TODO: this always creates immutable files. We might want an option
# to always create mutable files, or to copy mutable files into new
- # mutable files.
+ # mutable files. ticket #835
class TahoeDirectorySource:
- def __init__(self, nodeurl, cache, progressfunc):
+ def __init__(self, nodeurl, cache, progressfunc, basename):
self.nodeurl = nodeurl
self.cache = cache
self.progressfunc = progressfunc
+ self._basename = basename # unicode, or None for raw dircaps
+
+ def basename(self):
+ return self._basename
def init_from_grid(self, writecap, readcap):
self.writecap = writecap
url = self.nodeurl + "uri/%s" % urllib.quote(bestcap)
resp = do_http("GET", url + "?t=json")
if resp.status != 200:
- raise TahoeError("Error examining source directory", resp)
+ raise HTTPError("Error examining source directory", resp)
parsed = simplejson.loads(resp.read())
nodetype, d = parsed
assert nodetype == "dirnode"
def init_from_parsed(self, parsed):
nodetype, d = parsed
- self.writecap = ascii_or_none(d.get("rw_uri"))
- self.readcap = ascii_or_none(d.get("ro_uri"))
+ self.writecap = to_str(d.get("rw_uri"))
+ self.readcap = to_str(d.get("ro_uri"))
self.mutable = d.get("mutable", False) # older nodes don't provide it
self.children_d = dict( [(unicode(name),value)
for (name,value)
return
self.children = {}
for i,(name, data) in enumerate(self.children_d.items()):
- self.progressfunc("examining %d of %d" % (i, len(self.children_d)))
+ self.progressfunc("examining %d of %d" % (i+1, len(self.children_d)))
if data[0] == "filenode":
mutable = data[1].get("mutable", False)
- writecap = ascii_or_none(data[1].get("rw_uri"))
- readcap = ascii_or_none(data[1].get("ro_uri"))
+ writecap = to_str(data[1].get("rw_uri"))
+ readcap = to_str(data[1].get("ro_uri"))
self.children[name] = TahoeFileSource(self.nodeurl, mutable,
- writecap, readcap)
- else:
- assert data[0] == "dirnode"
- writecap = ascii_or_none(data[1].get("rw_uri"))
- readcap = ascii_or_none(data[1].get("ro_uri"))
+ writecap, readcap, name)
+ elif data[0] == "dirnode":
+ writecap = to_str(data[1].get("rw_uri"))
+ readcap = to_str(data[1].get("ro_uri"))
if writecap and writecap in self.cache:
child = self.cache[writecap]
elif readcap and readcap in self.cache:
child = self.cache[readcap]
else:
child = TahoeDirectorySource(self.nodeurl, self.cache,
- self.progressfunc)
+ self.progressfunc, name)
child.init_from_grid(writecap, readcap)
if writecap:
self.cache[writecap] = child
if readcap:
self.cache[readcap] = child
if recurse:
- child.populate(True)
+ child.populate(recurse=True)
self.children[name] = child
+ else:
+ # TODO: there should be an option to skip unknown nodes.
+ raise TahoeError("Cannot copy unknown nodes (ticket #839). "
+ "You probably need to use a later version of "
+ "Tahoe-LAFS to copy this directory.")
class TahoeMissingTarget:
def __init__(self, url):
def init_from_parsed(self, parsed):
nodetype, d = parsed
- self.writecap = ascii_or_none(d.get("rw_uri"))
- self.readcap = ascii_or_none(d.get("ro_uri"))
+ self.writecap = to_str(d.get("rw_uri"))
+ self.readcap = to_str(d.get("ro_uri"))
self.mutable = d.get("mutable", False) # older nodes don't provide it
self.children_d = dict( [(unicode(name),value)
for (name,value)
url = self.nodeurl + "uri/%s" % urllib.quote(bestcap)
resp = do_http("GET", url + "?t=json")
if resp.status != 200:
- raise TahoeError("Error examining target directory", resp)
+ raise HTTPError("Error examining target directory", resp)
parsed = simplejson.loads(resp.read())
nodetype, d = parsed
assert nodetype == "dirnode"
self.children = None
def just_created(self, writecap):
+ # TODO: maybe integrate this with the constructor
self.writecap = writecap
self.readcap = uri.from_string(writecap).get_readonly().to_string()
self.mutable = True
return
self.children = {}
for i,(name, data) in enumerate(self.children_d.items()):
- self.progressfunc("examining %d of %d" % (i, len(self.children_d)))
+ self.progressfunc("examining %d of %d" % (i+1, len(self.children_d)))
if data[0] == "filenode":
mutable = data[1].get("mutable", False)
- writecap = ascii_or_none(data[1].get("rw_uri"))
- readcap = ascii_or_none(data[1].get("ro_uri"))
+ writecap = to_str(data[1].get("rw_uri"))
+ readcap = to_str(data[1].get("ro_uri"))
url = None
if self.writecap:
url = self.nodeurl + "/".join(["uri",
urllib.quote(self.writecap),
- urllib.quote(name.encode('utf-8'))])
+ urllib.quote(unicode_to_url(name))])
self.children[name] = TahoeFileTarget(self.nodeurl, mutable,
writecap, readcap, url)
- else:
- assert data[0] == "dirnode"
- writecap = ascii_or_none(data[1].get("rw_uri"))
- readcap = ascii_or_none(data[1].get("ro_uri"))
+ elif data[0] == "dirnode":
+ writecap = to_str(data[1].get("rw_uri"))
+ readcap = to_str(data[1].get("ro_uri"))
if writecap and writecap in self.cache:
child = self.cache[writecap]
elif readcap and readcap in self.cache:
if readcap:
self.cache[readcap] = child
if recurse:
- child.populate(True)
+ child.populate(recurse=True)
self.children[name] = child
+ else:
+ # TODO: there should be an option to skip unknown nodes.
+ raise TahoeError("Cannot copy unknown nodes (ticket #839). "
+ "You probably need to use a later version of "
+ "Tahoe-LAFS to copy this directory.")
def get_child_target(self, name):
# return a new target for a named subdirectory of this dir
+ precondition(isinstance(name, unicode), name)
if self.children is None:
- self.populate(False)
+ self.populate(recurse=False)
if name in self.children:
return self.children[name]
writecap = make_tahoe_subdirectory(self.nodeurl, self.writecap, name)
return child
def put_file(self, name, inf):
+ precondition(isinstance(name, unicode), name)
url = self.nodeurl + "uri"
if not hasattr(inf, "seek"):
inf = inf.read()
- filecap = PUT(url, inf)
- # TODO: this always creates immutable files. We might want an option
- # to always create mutable files, or to copy mutable files into new
- # mutable files.
- self.new_children[name] = filecap
+
+ if self.children is None:
+ self.populate(recurse=False)
+
+ # Check to see if we already have a mutable file by this name.
+ # If so, overwrite that file in place.
+ if name in self.children and self.children[name].mutable:
+ self.children[name].put_file(inf)
+ else:
+ filecap = PUT(url, inf)
+ # TODO: this always creates immutable files. We might want an option
+ # to always create mutable files, or to copy mutable files into new
+ # mutable files.
+ self.new_children[name] = filecap
def put_uri(self, name, filecap):
+ precondition(isinstance(name, unicode), name)
self.new_children[name] = filecap
def set_children(self):
set_data = {}
for (name, filecap) in self.new_children.items():
# it just so happens that ?t=set_children will accept both file
- # read-caps and write-caps as ['rw_uri'], and will handle eithe
+ # read-caps and write-caps as ['rw_uri'], and will handle either
# correctly. So don't bother trying to figure out whether the one
# we have is read-only or read-write.
+ # TODO: think about how this affects forward-compatibility for
+ # unknown caps
set_data[name] = ["filenode", {"rw_uri": filecap}]
body = simplejson.dumps(set_data)
POST(url, body)
+FileSources = (LocalFileSource, TahoeFileSource)
+DirectorySources = (LocalDirectorySource, TahoeDirectorySource)
+FileTargets = (LocalFileTarget, TahoeFileTarget)
+DirectoryTargets = (LocalDirectoryTarget, TahoeDirectoryTarget)
+MissingTargets = (LocalMissingTarget, TahoeMissingTarget)
+
class Copier:
def do_copy(self, options, progressfunc=None):
def progress(message):
print >>self.stderr, message
self.progressfunc = progress
+ self.caps_only = options["caps-only"]
self.cache = {}
try:
- self.try_copy()
+ status = self.try_copy()
+ return status
except TahoeError, te:
- Failure().printTraceback(self.stderr)
- print >>self.stderr
+ if verbosity >= 2:
+ Failure().printTraceback(self.stderr)
+ print >>self.stderr
te.display(self.stderr)
return 1
def try_copy(self):
+ """
+ All usage errors are caught here, not in a subroutine. This bottoms
+ out in copy_file_to_file() or copy_things_to_directory().
+ """
source_specs = self.options.sources
destination_spec = self.options.destination
recursive = self.options["recursive"]
target = self.get_target_info(destination_spec)
-
- try:
- sources = [] # list of (name, source object)
- for ss in source_specs:
- name, source = self.get_source_info(ss)
- sources.append( (name, source) )
- except MissingSourceError, e:
- self.to_stderr("No such file or directory %s" % e.args[0])
+ precondition(isinstance(target, FileTargets + DirectoryTargets + MissingTargets), target)
+ target_has_trailing_slash = destination_spec.endswith("/")
+
+ sources = [] # list of source objects
+ for ss in source_specs:
+ si = self.get_source_info(ss)
+ precondition(isinstance(si, FileSources + DirectorySources), si)
+ sources.append(si)
+
+ # if any source is a directory, must use -r
+ # if target is missing:
+ # if source is a single file, target will be a file
+ # else target will be a directory, so mkdir it
+ # if there are multiple sources, target must be a dir
+ # if target is a file, source must be a single file
+ # if target is directory, sources must be named or a dir
+
+ have_source_dirs = any([isinstance(s, DirectorySources)
+ for s in sources])
+ if have_source_dirs and not recursive:
+ # 'cp dir target' without -r: error
+ self.to_stderr("cannot copy directories without --recursive")
return 1
+ del recursive # -r is only used for signalling errors
+
+ if isinstance(target, FileTargets):
+ target_is_file = True
+ elif isinstance(target, DirectoryTargets):
+ target_is_file = False
+ else: # isinstance(target, MissingTargets)
+ if len(sources) == 1 and isinstance(sources[0], FileSources):
+ target_is_file = True
+ else:
+ target_is_file = False
- have_source_dirs = bool([s for (name,s) in sources
- if isinstance(s, (LocalDirectorySource,
- TahoeDirectorySource))])
+ if target_is_file and target_has_trailing_slash:
+ self.to_stderr("target is not a directory, but ends with a slash")
+ return 1
- if have_source_dirs and not recursive:
- self.to_stderr("cannot copy directories without --recursive")
+ if len(sources) > 1 and target_is_file:
+ self.to_stderr("copying multiple things requires target be a directory")
return 1
- if isinstance(target, (LocalFileTarget, TahoeFileTarget)):
- # cp STUFF foo.txt, where foo.txt already exists. This limits the
- # possibilities considerably.
- if len(sources) > 1:
- self.to_stderr("target '%s' is not a directory" % destination_spec)
- return 1
- if have_source_dirs:
+ if target_is_file:
+ _assert(len(sources) == 1, sources)
+ if not isinstance(sources[0], FileSources):
+ # 'cp -r dir existingfile': error
self.to_stderr("cannot copy directory into a file")
return 1
- name, source = sources[0]
- return self.copy_file(source, target)
-
- if isinstance(target, (LocalMissingTarget, TahoeMissingTarget)):
- if recursive:
- return self.copy_to_directory(sources, target)
- if len(sources) > 1:
- # if we have -r, we'll auto-create the target directory. Without
- # it, we'll only create a file.
- self.to_stderr("cannot copy multiple files into a file without -r")
+ return self.copy_file_to_file(sources[0], target)
+
+ # else target is a directory, so each source must be one of:
+ # * a named file (copied to a new file under the target)
+ # * a named directory (causes a new directory of the same name to be
+ # created under the target, then the contents of the source are
+ # copied into that directory)
+ # * an unnamed directory (the contents of the source are copied into
+ # the target, without a new directory being made)
+ #
+ # If any source is an unnamed file, throw an error, since we have no
+ # way to name the output file.
+ _assert(isinstance(target, DirectoryTargets + MissingTargets), target)
+
+ for source in sources:
+ if isinstance(source, FileSources) and not source.basename():
+ self.to_stderr("when copying into a directory, all source files must have names, but %s is unnamed" % quote_output(source_specs[0]))
return 1
- # cp file1 newfile
- name, source = sources[0]
- return self.copy_file(source, target)
-
- if isinstance(target, (LocalDirectoryTarget, TahoeDirectoryTarget)):
- return self.copy_to_directory(sources, target)
-
- self.to_stderr("unknown target")
- return 1
+ return self.copy_things_to_directory(sources, target)
def to_stderr(self, text):
print >>self.stderr, text
+ # FIXME reduce the amount of near-duplicate code between get_target_info
+ # and get_source_info.
+
def get_target_info(self, destination_spec):
- rootcap, path = get_alias(self.aliases, destination_spec, None)
+ precondition(isinstance(destination_spec, unicode), destination_spec)
+ rootcap, path_utf8 = get_alias(self.aliases, destination_spec, None)
+ path = path_utf8.decode("utf-8")
if rootcap == DefaultAliasMarker:
# no alias, so this is a local file
- pathname = os.path.abspath(os.path.expanduser(path))
+ pathname = abspath_expanduser_unicode(path)
if not os.path.exists(pathname):
t = LocalMissingTarget(pathname)
elif os.path.isdir(pathname):
t = LocalDirectoryTarget(self.progress, pathname)
else:
+ # TODO: should this be _assert? what happens if the target is
+ # a special file?
assert os.path.isfile(pathname), pathname
t = LocalFileTarget(pathname) # non-empty
else:
url = self.nodeurl + "uri/%s" % urllib.quote(rootcap)
if path:
url += "/" + escape_path(path)
- last_slash = path.rfind("/")
resp = do_http("GET", url + "?t=json")
if resp.status == 404:
self.progress)
t.init_from_parsed(parsed)
else:
- writecap = ascii_or_none(d.get("rw_uri"))
- readcap = ascii_or_none(d.get("ro_uri"))
+ writecap = to_str(d.get("rw_uri"))
+ readcap = to_str(d.get("ro_uri"))
mutable = d.get("mutable", False)
t = TahoeFileTarget(self.nodeurl, mutable,
writecap, readcap, url)
else:
- raise TahoeError("Error examining target '%s'"
- % destination_spec, resp)
+ raise HTTPError("Error examining target %s"
+ % quote_output(destination_spec), resp)
return t
def get_source_info(self, source_spec):
- rootcap, path = get_alias(self.aliases, source_spec, None)
+ """
+ This turns an argv string into a (Local|Tahoe)(File|Directory)Source.
+ """
+ precondition(isinstance(source_spec, unicode), source_spec)
+ rootcap, path_utf8 = get_alias(self.aliases, source_spec, None)
+ path = path_utf8.decode("utf-8")
if rootcap == DefaultAliasMarker:
# no alias, so this is a local file
- pathname = os.path.abspath(os.path.expanduser(path))
+ pathname = abspath_expanduser_unicode(path)
name = os.path.basename(pathname)
if not os.path.exists(pathname):
- raise MissingSourceError(source_spec)
+ raise MissingSourceError(source_spec, quotefn=quote_local_unicode_path)
if os.path.isdir(pathname):
- t = LocalDirectorySource(self.progress, pathname)
+ t = LocalDirectorySource(self.progress, pathname, name)
else:
assert os.path.isfile(pathname)
- t = LocalFileSource(pathname) # non-empty
+ t = LocalFileSource(pathname, name) # non-empty
else:
# this is a tahoe object
url = self.nodeurl + "uri/%s" % urllib.quote(rootcap)
name = None
if path:
url += "/" + escape_path(path)
- last_slash = path.rfind("/")
+ last_slash = path.rfind(u"/")
name = path
- if last_slash:
+ if last_slash != -1:
name = path[last_slash+1:]
resp = do_http("GET", url + "?t=json")
if resp.status == 404:
raise MissingSourceError(source_spec)
elif resp.status != 200:
- raise TahoeError("Error examining source '%s'" % source_spec,
- resp)
+ raise HTTPError("Error examining source %s" % quote_output(source_spec),
+ resp)
parsed = simplejson.loads(resp.read())
nodetype, d = parsed
if nodetype == "dirnode":
t = TahoeDirectorySource(self.nodeurl, self.cache,
- self.progress)
+ self.progress, name)
t.init_from_parsed(parsed)
else:
- writecap = ascii_or_none(d.get("rw_uri"))
- readcap = ascii_or_none(d.get("ro_uri"))
+ writecap = to_str(d.get("rw_uri"))
+ readcap = to_str(d.get("ro_uri"))
mutable = d.get("mutable", False) # older nodes don't provide it
- t = TahoeFileSource(self.nodeurl, mutable, writecap, readcap)
- return name, t
-
-
- def dump_graph(self, s, indent=" "):
- for name, child in s.children.items():
- print indent + name + ":" + str(child)
- if isinstance(child, (LocalDirectorySource, TahoeDirectorySource)):
- self.dump_graph(child, indent+" ")
-
- def copy_to_directory(self, source_infos, target):
- # step one: build a recursive graph of the source tree. This returns
- # a dictionary, with child names as keys, and values that are either
- # Directory or File instances (local or tahoe).
- source_dirs = self.build_graphs(source_infos)
- source_files = [source for source in source_infos
- if isinstance(source[1], (LocalFileSource,
- TahoeFileSource))]
-
- #print "graphs"
- #for s in source_dirs:
- # self.dump_graph(s)
-
- # step two: create the top-level target directory object
- if isinstance(target, LocalMissingTarget):
- os.makedirs(target.pathname)
- target = LocalDirectoryTarget(self.progress, target.pathname)
- elif isinstance(target, TahoeMissingTarget):
- writecap = mkdir(target.url)
- target = TahoeDirectoryTarget(self.nodeurl, self.cache,
- self.progress)
- target.just_created(writecap)
- assert isinstance(target, (LocalDirectoryTarget, TahoeDirectoryTarget))
- target.populate(False)
-
- # step three: find a target for each source node, creating
- # directories as necessary. 'targetmap' is a dictionary that uses
- # target Directory instances as keys, and has values of
- # (name->sourceobject) dicts for all the files that need to wind up
- # there.
-
- # sources are all LocalFile/LocalDirectory/TahoeFile/TahoeDirectory
- # target is LocalDirectory/TahoeDirectory
-
- self.progress("attaching sources to targets, "
- "%d files / %d dirs in root" %
- (len(source_files), len(source_dirs)))
-
- self.targetmap = {}
- self.files_to_copy = 0
-
- for (name,s) in source_files:
- self.attach_to_target(s, name, target)
- self.files_to_copy += 1
-
- for source in source_dirs:
- self.assign_targets(source, target)
-
- self.progress("targets assigned, %s dirs, %s files" %
- (len(self.targetmap), self.files_to_copy))
-
- self.progress("starting copy, %d files, %d directories" %
- (self.files_to_copy, len(self.targetmap)))
- self.files_copied = 0
- self.targets_finished = 0
-
- # step four: walk through the list of targets. For each one, copy all
- # the files. If the target is a TahoeDirectory, upload and create
- # read-caps, then do a set_children to the target directory.
-
- for target in self.targetmap:
- self.copy_files_to_target(self.targetmap[target], target)
- self.targets_finished += 1
- self.progress("%d/%d directories" %
- (self.targets_finished, len(self.targetmap)))
-
- return self.announce_success("files copied")
-
- def attach_to_target(self, source, name, target):
- if target not in self.targetmap:
- self.targetmap[target] = {}
- self.targetmap[target][name] = source
- self.files_to_copy += 1
-
- def assign_targets(self, source, target):
- # copy everything in the source into the target
- assert isinstance(source, (LocalDirectorySource, TahoeDirectorySource))
-
- for name, child in source.children.items():
- if isinstance(child, (LocalDirectorySource, TahoeDirectorySource)):
- # we will need a target directory for this one
- subtarget = target.get_child_target(name)
- self.assign_targets(child, subtarget)
- else:
- assert isinstance(child, (LocalFileSource, TahoeFileSource))
- self.attach_to_target(child, name, target)
+ last_slash = source_spec.rfind(u"/")
+ if last_slash != -1:
+ # TODO: this looks funny and redundant with the 'name'
+ # assignment above. cf #2329
+ name = source_spec[last_slash+1:]
+ t = TahoeFileSource(self.nodeurl, mutable, writecap, readcap, name)
+ return t
- def copy_files_to_target(self, targetmap, target):
- for name, source in targetmap.items():
- assert isinstance(source, (LocalFileSource, TahoeFileSource))
- self.copy_file_into(source, name, target)
- self.files_copied += 1
- self.progress("%d/%d files, %d/%d directories" %
- (self.files_copied, self.files_to_copy,
- self.targets_finished, len(self.targetmap)))
- target.set_children()
def need_to_copy_bytes(self, source, target):
if source.need_to_copy_bytes:
print >>self.stdout, "Success: %s" % msg
return 0
- def copy_file(self, source, target):
- assert isinstance(source, (LocalFileSource, TahoeFileSource))
- assert isinstance(target, (LocalFileTarget, TahoeFileTarget,
- LocalMissingTarget, TahoeMissingTarget))
+ def copy_file_to_file(self, source, target):
+ precondition(isinstance(source, FileSources), source)
+ precondition(isinstance(target, FileTargets + MissingTargets), target)
if self.need_to_copy_bytes(source, target):
# if the target is a local directory, this will just write the
# bytes to disk. If it is a tahoe directory, it will upload the
# data, and stash the new filecap for a later set_children call.
- f = source.open()
+ f = source.open(self.caps_only)
target.put_file(f)
return self.announce_success("file copied")
# otherwise we're copying tahoe to tahoe, and using immutable files,
target.put_uri(source.bestcap())
return self.announce_success("file linked")
- def copy_file_into(self, source, name, target):
- assert isinstance(source, (LocalFileSource, TahoeFileSource))
- assert isinstance(target, (LocalDirectoryTarget, TahoeDirectoryTarget))
+ def copy_things_to_directory(self, sources, target):
+ # step one: if the target is missing, we should mkdir it
+ target = self.maybe_create_target(target)
+ target.populate(recurse=False)
+
+ # step two: scan any source dirs, recursively, to find children
+ for s in sources:
+ if isinstance(s, DirectorySources):
+ s.populate(recurse=True)
+ if isinstance(s, FileSources):
+ # each source must have a name, or be a directory
+ _assert(s.basename() is not None, s)
+
+ # step three: find a target for each source node, creating
+ # directories as necessary. 'targetmap' is a dictionary that uses
+ # target Directory instances as keys, and has values of (name:
+ # sourceobject) dicts for all the files that need to wind up there.
+ targetmap = self.build_targetmap(sources, target)
+
+ # step four: walk through the list of targets. For each one, copy all
+ # the files. If the target is a TahoeDirectory, upload and create
+ # read-caps, then do a set_children to the target directory.
+ self.copy_to_targetmap(targetmap)
+
+ return self.announce_success("files copied")
+
+ def maybe_create_target(self, target):
+ if isinstance(target, LocalMissingTarget):
+ os.makedirs(target.pathname)
+ target = LocalDirectoryTarget(self.progress, target.pathname)
+ elif isinstance(target, TahoeMissingTarget):
+ writecap = mkdir(target.url)
+ target = TahoeDirectoryTarget(self.nodeurl, self.cache,
+ self.progress)
+ target.just_created(writecap)
+ # afterwards, or otherwise, it will be a directory
+ precondition(isinstance(target, DirectoryTargets), target)
+ return target
+
+ def build_targetmap(self, sources, target):
+ num_source_files = len([s for s in sources
+ if isinstance(s, FileSources)])
+ num_source_dirs = len([s for s in sources
+ if isinstance(s, DirectorySources)])
+ self.progress("attaching sources to targets, "
+ "%d files / %d dirs in root" %
+ (num_source_files, num_source_dirs))
+
+ # this maps each target directory to a list of source files that need
+ # to be copied into it. All source files have names.
+ targetmap = defaultdict(list)
+
+ for s in sources:
+ if isinstance(s, FileSources):
+ targetmap[target].append(s)
+ else:
+ _assert(isinstance(s, DirectorySources), s)
+ name = s.basename()
+ if name is not None:
+ # named sources get a new directory. see #2329
+ new_target = target.get_child_target(name)
+ else:
+ # unnamed sources have their contents copied directly
+ new_target = target
+ self.assign_targets(targetmap, s, new_target)
+
+ self.progress("targets assigned, %s dirs, %s files" %
+ (len(targetmap), self.count_files_to_copy(targetmap)))
+ return targetmap
+
+ def assign_targets(self, targetmap, source, target):
+ # copy everything in the source into the target
+ precondition(isinstance(source, DirectorySources), source)
+ for name, child in source.children.items():
+ if isinstance(child, DirectorySources):
+ # we will need a target directory for this one
+ subtarget = target.get_child_target(name)
+ self.assign_targets(targetmap, child, subtarget)
+ else:
+ _assert(isinstance(child, FileSources), child)
+ targetmap[target].append(child)
+
+ def copy_to_targetmap(self, targetmap):
+ files_to_copy = self.count_files_to_copy(targetmap)
+ self.progress("starting copy, %d files, %d directories" %
+ (files_to_copy, len(targetmap)))
+ files_copied = 0
+ targets_finished = 0
+
+ for target, sources in targetmap.items():
+ _assert(isinstance(target, DirectoryTargets), target)
+ for source in sources:
+ _assert(isinstance(source, FileSources), source)
+ self.copy_file_into_dir(source, source.basename(), target)
+ files_copied += 1
+ self.progress("%d/%d files, %d/%d directories" %
+ (files_copied, files_to_copy,
+ targets_finished, len(targetmap)))
+ target.set_children()
+ targets_finished += 1
+ self.progress("%d/%d directories" %
+ (targets_finished, len(targetmap)))
+
+ def count_files_to_copy(self, targetmap):
+ return sum([len(sources) for sources in targetmap.values()])
+
+ def copy_file_into_dir(self, source, name, target):
+ precondition(isinstance(source, FileSources), source)
+ precondition(isinstance(target, DirectoryTargets), target)
+ precondition(isinstance(name, unicode), name)
if self.need_to_copy_bytes(source, target):
# if the target is a local directory, this will just write the
# bytes to disk. If it is a tahoe directory, it will upload the
# data, and stash the new filecap for a later set_children call.
- f = source.open()
+ f = source.open(self.caps_only)
target.put_file(name, f)
return
# otherwise we're copying tahoe to tahoe, and using immutable files,
if self.progressfunc:
self.progressfunc(message)
- def build_graphs(self, source_infos):
- graphs = []
- for name,source in source_infos:
- if isinstance(source, (LocalDirectorySource, TahoeDirectorySource)):
- source.populate(True)
- graphs.append(source)
- return graphs
-
def copy(options):
return Copier().do_copy(options)
+
+# error cases that need improvement:
+# local-file-in-the-way
+# touch proposed
+# tahoe cp -r my:docs/proposed/denver.txt proposed/denver.txt
+# handling of unknown nodes
+
+# things that maybe should be errors but aren't
+# local-dir-in-the-way
+# mkdir denver.txt
+# tahoe cp -r my:docs/proposed/denver.txt denver.txt
+# (creates denver.txt/denver.txt)
+
+# error cases that look good:
+# tahoe cp -r my:docs/missing missing
+# disconnect servers
+# tahoe cp -r my:docs/missing missing -> No JSON object could be decoded
+# tahoe-file-in-the-way (when we want to make a directory)
+# tahoe put README my:docs
+# tahoe cp -r docs/proposed my:docs/proposed