From: Brian Warner <warner@allmydata.com>
Date: Wed, 21 May 2008 18:49:22 +0000 (-0700)
Subject: cli: initial implementation of 'cp -r', probably doesn't work yet
X-Git-Tag: allmydata-tahoe-1.1.0~88
X-Git-Url: https://git.rkrishnan.org/%5B/%5D%20/uri/%22file:/configuration.txt?a=commitdiff_plain;h=7afd869854bc3d5df403e63f243cf7f8f37301c7;p=tahoe-lafs%2Ftahoe-lafs.git

cli: initial implementation of 'cp -r', probably doesn't work yet
---

diff --git a/src/allmydata/scripts/tahoe_cp.py b/src/allmydata/scripts/tahoe_cp.py
index b904b179..661272af 100644
--- a/src/allmydata/scripts/tahoe_cp.py
+++ b/src/allmydata/scripts/tahoe_cp.py
@@ -4,142 +4,597 @@ import urllib
 import simplejson
 from allmydata.scripts.common import get_alias, escape_path, DefaultAliasMarker
 from allmydata.scripts.common_http import do_http
+from allmydata import uri
 
 def ascii_or_none(s):
     if s is None:
         return s
     return str(s)
 
-def get_info(nodeurl, aliases, target):
-    rootcap, path = get_alias(aliases, target, None)
-    if rootcap == DefaultAliasMarker:
-        # this is a local file
-        pathname = os.path.abspath(os.path.expanduser(path))
-        if not os.path.exists(pathname):
-            return ("empty", "local", pathname)
-        if os.path.isdir(pathname):
-            return ("directory", "local", pathname)
+class WriteError(Exception):
+    pass
+class ReadError(Exception):
+    pass
+
+def GET_to_file(url):
+    resp = do_http("GET", url)
+    if resp.status == 200:
+        return resp
+    raise ReadError("Error during GET: %s %s %s" % (resp.status,
+                                                    resp.reason,
+                                                    resp.read()))
+def GET_to_string(url):
+    f = GET_to_file(url)
+    return f.read()
+
+def PUT(url, data):
+    resp = do_http("PUT", url, data)
+    if resp.status in (200, 201):
+        return resp.read()
+    raise WriteError("Error during PUT: %s %s %s" % (resp.status, resp.reason,
+                                                     resp.read()))
+
+def mkdir(targeturl):
+    resp = do_http("POST", targeturl)
+    if resp.status in (200, 201):
+        return resp.read().strip()
+    raise WriteError("Error during mkdir: %s %s %s" % (resp.status, resp.reason,
+                                                       resp.read()))
+
+def make_tahoe_subdirectory(nodeurl, parent_writecap, name):
+    url = nodeurl + "/".join(["uri",
+                              urllib.quote(parent_writecap),
+                              urllib.quote(name),
+                              ]) + "?t=mkdir"
+    resp = do_http("POST", url)
+    if resp.status in (200, 201):
+        return resp.read().strip()
+    raise WriteError("Error during mkdir: %s %s %s" % (resp.status, resp.reason,
+                                                       resp.read()))
+
+
+class LocalFileSource:
+    def __init__(self, pathname):
+        self.pathname = pathname
+
+    def need_to_copy_bytes(self):
+        return True
+
+    def open(self):
+        return open(self.pathname, "rb")
+
+class LocalFileTarget:
+    def __init__(self, pathname):
+        self.pathname = pathname
+
+class LocalDirectorySource:
+    def __init__(self, progressfunc, pathname):
+        self.progressfunc = progressfunc
+        self.pathname = pathname
+        self.children = None
+
+    def populate(self, recurse):
+        children = os.listdir(self.pathname)
+        for i,n in enumerate(children):
+            self.progressfunc("examining %d of %d" % (i, len(children)))
+            pn = os.path.join(self.pathname, n)
+            if os.path.isdir(pn):
+                child = LocalDirectorySource(self.progressfunc, pn)
+                self.children[n] = child
+                if recurse:
+                    child.populate(True)
+            else:
+                assert os.path.isfile(pn)
+                self.children[n] = LocalFileSource(pn)
+
+class LocalDirectoryTarget:
+    def __init__(self, progressfunc, pathname):
+        self.progressfunc = progressfunc
+        self.pathname = pathname
+        self.children = None
+
+    def populate(self, recurse):
+        children = os.listdir(self.pathname)
+        for i,n in enumerate(children):
+            self.progressfunc("examining %d of %d" % (i, 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)
+            else:
+                assert os.path.isfile(pn)
+                self.children[n] = LocalFileTarget(pn)
+
+    def get_child_target(self, name):
+        if self.children is None:
+            self.populate(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)
+
+    def put_file(self, name, inf):
+        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()
+
+    def set_children(self):
+        pass
+
+class TahoeFileSource:
+    def __init__(self, nodeurl, mutable, writecap, readcap):
+        self.nodeurl = nodeurl
+        self.mutable = mutable
+        self.writecap = writecap
+        self.readcap = readcap
+
+    def need_to_copy_bytes(self):
+        if self.mutable:
+            return True
+        return False
+
+    def open(self):
+        url = self.nodeurl + "uri/" + urllib.quote(self.readcap)
+        return GET_to_file(url)
+
+    def bestcap(self):
+        return self.writecap or self.readcap
+
+class TahoeFileTarget:
+    def __init__(self, nodeurl, mutable, writecap, readcap):
+        self.nodeurl = nodeurl
+        self.mutable = mutable
+        self.writecap = writecap
+        self.readcap = readcap
+
+class TahoeDirectorySource:
+    def __init__(self, nodeurl, cache, progressfunc):
+        self.nodeurl = nodeurl
+        self.cache = cache
+        self.progressfunc = progressfunc
+
+    def init_from_grid(self, writecap, readcap):
+        self.writecap = writecap
+        self.readcap = readcap
+        bestcap = writecap or readcap
+        url = self.nodeurl + "uri/%s" % urllib.quote(bestcap)
+        resp = do_http("GET", url + "?t=json")
+        assert resp.status == 200
+        parsed = simplejson.loads(resp.read())
+        nodetype, d = parsed
+        assert nodetype == "dirnode"
+        self.mutable = d.get("mutable", False) # older nodes don't provide it
+        self.children_d = d["children"]
+        self.children = None
+
+    def populate(self, recurse):
+        self.children = {}
+        for i,(name, data) in enumerate(self.children_d):
+            self.progressfunc("examining %d of %d" % (i, 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"))
+                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"))
+                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)
+                    child.init_from_grid(writecap, readcap)
+                    if writecap:
+                        self.cache[writecap] = child
+                    if readcap:
+                        self.cache[readcap] = child
+                    if recurse:
+                        child.populate(True)
+                self.children[name] = child
+
+class TahoeDirectoryTarget:
+    def __init__(self, nodeurl, cache, progressfunc):
+        self.nodeurl = nodeurl
+        self.cache = cache
+        self.progressfunc = progressfunc
+        self.new_children = {}
+
+    def init_from_grid(self, writecap, readcap):
+        self.writecap = writecap
+        self.readcap = readcap
+        bestcap = writecap or readcap
+        url = self.nodeurl + "uri/%s" % urllib.quote(bestcap)
+        resp = do_http("GET", url + "?t=json")
+        assert resp.status == 200
+        parsed = simplejson.loads(resp.read())
+        nodetype, d = parsed
+        assert nodetype == "dirnode"
+        self.mutable = d.get("mutable", False) # older nodes don't provide it
+        self.children_d = d["children"]
+        self.children = None
+
+    def just_created(self, writecap):
+        self.writecap = writecap
+        self.readcap = uri.from_string().get_readonly().to_string()
+        self.mutable = True
+        self.children_d = {}
+        self.children = {}
+
+    def populate(self, recurse):
+        self.children = {}
+        for i,(name, data) in enumerate(self.children_d):
+            self.progressfunc("examining %d of %d" % (i, 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"))
+                self.children[name] = TahoeFileTarget(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"))
+                if writecap and writecap in self.cache:
+                    child = self.cache[writecap]
+                elif readcap and readcap in self.cache:
+                    child = self.cache[readcap]
+                else:
+                    child = TahoeDirectoryTarget(self.nodeurl, self.cache,
+                                                 self.progressfunc)
+                    child.init_from_grid(writecap, readcap)
+                    if writecap:
+                        self.cache[writecap] = child
+                    if readcap:
+                        self.cache[readcap] = child
+                    if recurse:
+                        child.populate(True)
+                self.children[name] = child
+
+    def get_child_target(self, name):
+        # return a new target for a named subdirectory of this dir
+        if self.children is None:
+            self.populate(False)
+        if name in self.children:
+            return self.children[name]
+        writecap = make_tahoe_subdirectory(self.nodeurl, self.writecap, name)
+        child = TahoeDirectoryTarget(self.nodeurl, self.cache,
+                                     self.progressfunc)
+        child.just_created(writecap)
+        self.children[name] = child
+        return child
+
+    def put_file(self, name, inf):
+        url = self.nodeurl + "uri"
+        # I'm not sure this will work: we might not have .seek, so if not:
+        #inf = inf.read()
+
+        # 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.
+        resp = do_http("PUT", url, inf)
+        filecap = check_PUT(resp)
+        self.new_children[name] = filecap
+
+    def put_uri(self, name, filecap):
+        self.new_children[name] = filecap
+
+    def set_children(self):
+        if not self.new_children:
+            return
+        # XXX TODO t=set_children
+
+class Copier:
+    def __init__(self, nodeurl, config, aliases,
+                 verbosity, stdout, stderr,
+                 progressfunc=None):
+        if nodeurl[-1] != "/":
+            nodeurl += "/"
+        self.nodeurl = nodeurl
+        self.progressfunc = progressfunc
+        self.config = config
+        self.aliases = aliases
+        self.verbosity = verbosity
+        self.stdout = stdout
+        self.stderr = stderr
+
+    def to_stderr(self, text):
+        print >>self.stderr, text
+
+    def do_copy(self, sources, destination):
+        recursive = self.config["recursive"]
+
+        #print "sources:", sources
+        #print "dest:", destination
+
+        target = self.get_info(destination)
+        #print target
+
+        source_info = dict([(self.get_info(source), source)
+                            for source in sources])
+        source_files = [s for s in source_info if s[0] == "file"]
+        source_dirs = [s for s in source_info if s[0] == "directory"]
+        empty_sources = [s for s in source_info if s[0] == "empty"]
+        if empty_sources:
+            for s in empty_sources:
+                self.to_stderr("no such file or directory %s" % source_info[s])
+            return 1
+
+        #print "source_files", " ".join([source_info[s] for s in source_files])
+        #print "source_dirs", " ".join([source_info[s] for s in source_dirs])
+
+        if source_dirs and not recursive:
+            self.to_stderr("cannot copy directories without --recursive")
+            return 1
+
+        if target[0] == "file":
+            # 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)
+                return 1
+            if source_dirs:
+                self.to_stderr("cannot copy directory into a file")
+                return 1
+            return self.copy_to_file(source_files[0], target)
+
+        if target[0] == "empty":
+            if recursive:
+                return self.copy_to_directory(source_files, source_dirs, 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 1
+            # cp file1 newfile
+            return self.copy_to_file(source_files[0], target)
+
+        if target[0] == "directory":
+            return self.copy_to_directory(source_files, source_dirs, target)
+
+        self.to_stderr("unknown target")
+        return 1
+
+    def get_info(self, target):
+        rootcap, path = get_alias(self.aliases, target, None)
+        if rootcap == DefaultAliasMarker:
+            # this is a local file
+            pathname = os.path.abspath(os.path.expanduser(path))
+            if not os.path.exists(pathname):
+                name = os.path.basename(pathname)
+                return ("empty", "local", name, pathname)
+            if os.path.isdir(pathname):
+                return ("directory", "local", pathname)
+            else:
+                assert os.path.isfile(pathname)
+                name = os.path.basename(pathname)
+                return ("file", "local", name, pathname)
         else:
-            assert os.path.isfile(pathname)
-            return ("file", "local", pathname)
-    else:
-        # this is a tahoe object
-        url = nodeurl + "uri/%s" % urllib.quote(rootcap)
-        if path:
-            url += "/" + escape_path(path)
+            # 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("/")
+                name = path
+                if last_slash:
+                    name = path[last_slash+1:]
+            return self.get_info_tahoe_dirnode(url, name)
+
+    def get_info_tahoe_dirnode(self, url, name):
         resp = do_http("GET", url + "?t=json")
         if resp.status == 404:
             # doesn't exist yet
-            return ("empty", "tahoe", False, None, None, url)
+            return ("empty", "tahoe", False, name, None, None, url)
         parsed = simplejson.loads(resp.read())
         nodetype, d = parsed
         mutable = d.get("mutable", False) # older nodes don't provide 'mutable'
         rw_uri = ascii_or_none(d.get("rw_uri"))
         ro_uri = ascii_or_none(d.get("ro_uri"))
         if nodetype == "dirnode":
-            return ("directory", "tahoe", mutable, rw_uri, ro_uri, url)
+            return ("directory", "tahoe", mutable, name, rw_uri, ro_uri,
+                    d["children"], url)
         else:
-            return ("file", "tahoe", mutable, rw_uri, ro_uri, url)
+            return ("file", "tahoe", mutable, name, rw_uri, ro_uri, url)
 
-def copy(nodeurl, config, aliases, sources, destination,
-         verbosity, stdout, stderr):
-    if nodeurl[-1] != "/":
-        nodeurl += "/"
-    recursive = config["recursive"]
-
-    #print "sources:", sources
-    #print "dest:", destination
-
-    target = get_info(nodeurl, aliases, destination)
-    #print target
-
-    source_info = dict([(get_info(nodeurl, aliases, source), source)
-                        for source in sources])
-    source_files = [s for s in source_info if s[0] == "file"]
-    source_dirs = [s for s in source_info if s[0] == "directory"]
-    empty_sources = [s for s in source_info if s[0] == "empty"]
-    if empty_sources:
-        for s in empty_sources:
-            print >>stderr, "no such file or directory %s" % source_info[s]
-        return 1
 
-    #print "source_files", " ".join([source_info[s] for s in source_files])
-    #print "source_dirs", " ".join([source_info[s] for s in source_dirs])
+    def get_file_data(self, source):
+        assert source[0] == "file"
+        if source[1] == "local":
+            (ig1, ig2, name, pathname) = source
+            return open(pathname, "rb").read()
+        (ig1, ig2, mutable, name, writecap, readcap, url) = source
+        return GET_to_string(url)
 
-    if source_dirs and not recursive:
-        print >>stderr, "cannot copy directories without --recursive"
-        return 1
+    def put_file_data(self, data, target):
+        assert target[0] in ("file", "empty")
+        if target[1] == "local":
+            (ig1, ig2, name, pathname) = target
+            open(pathname, "wb").write(data)
+            return True
+        (ig1, ig2, mutable, name, writecap, readcap, url) = target
+        return PUT(url, data)
 
-    if target[0] == "file":
-        # cp STUFF foo.txt, where foo.txt already exists. This limits the
-        # possibilities considerably.
-        if len(sources) > 1:
-            print >>stderr, "target '%s' is not a directory" % destination
-            return 1
-        if source_dirs:
-            print >>stderr, "cannot copy directory into a file"
-            return 1
-        return copy_to_file(source_files[0], target)
-
-    if target[0] == "empty":
-        if recursive:
-            return copy_to_directory(source_files, source_dirs, target)
-        if len(sources) > 1:
-            # if we have -r, we'll auto-create the target directory. Without
-            # it, we'll only create a file.
-            print >>stderr, "cannot copy multiple files into a file without -r"
-            return 1
-        # cp file1 newfile
-        return copy_to_file(source_files[0], target)
+    def put_uri(self, uri, targeturl):
+        return PUT(targeturl + "?t=uri", uri)
 
-    if target[0] == "directory":
-        return copy_to_directory(source_files, source_dirs, target)
+    def upload_data(self, data):
+        url = self.nodeurl + "uri"
+        return PUT(url, data)
 
-    print >>stderr, "unknown target"
-    return 1
+    def copy_to_file(self, source, target):
+        assert source[0] == "file"
+        # do we need to copy bytes?
+        if source[1] == "local" or source[2] == True or target[1] == "local":
+            # yes
+            data = self.get_file_data(source)
+            self.put_file_data(data, target)
+            return
+        # no, we're getting data from an immutable source, and we're copying
+        # into the tahoe grid, so we can just copy the URI.
+        uri = source[3] or source[4] # prefer rw_uri, fall back to ro_uri
+        # TODO: if the original was mutable, and we're creating the target,
+        # should be we create a mutable file to match? At the moment we always
+        # create immutable files.
+        self.put_uri(uri, target[-1])
 
+    def copy_to_directory(self, source_file_infos, source_dir_infos,
+                          target_info):
+        # step one: build a 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_dir_infos)
 
-def get_file_data(source):
-    assert source[0] == "file"
-    if source[1] == "local":
-        return open(source[2], "rb").read()
-    return do_http("GET", source[-1]).read()
+        # step two: create the top-level target directory object
+        assert target_info[0] in ("empty", "directory")
+        if target_info[1] == "local":
+            pathname = target_info[-1]
+            if not os.path.exists(pathname):
+                os.makedirs(pathname)
+            assert os.path.isdir(pathname)
+            target = LocalDirectoryTarget(self.progressfunc, target_info[-1])
+        else:
+            assert target_info[1] == "tahoe"
+            target = TahoeDirectoryTarget(self.nodeurl, self.cache,
+                                          self.progressfunc)
+            if target_info[0] == "empty":
+                writecap = mkdir(target_info[-1])
+                target.just_created(writecap)
+            else:
+                (ig1, ig2, mutable, name, writecap, readcap, url) = target_info
+                target.init_from_grid(writecap, readcap)
 
-class WriteError(Exception):
-    pass
+        # 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.
 
-def check_PUT(resp):
-    if resp.status in (200, 201):
-        return True
-    raise WriteError("Error during PUT: %s %s %s" % (resp.status, resp.reason,
-                                                     resp.read()))
+        # sources are all LocalFile/LocalDirectory/TahoeFile/TahoeDirectory
+        # target is LocalDirectory/TahoeDirectory
 
-def put_file_data(data, target):
-    if target[1] == "local":
-        open(target[2], "wb").write(data)
-        return True
-    resp = do_http("PUT", target[-1], data)
-    return check_PUT(resp)
-
-def put_uri(uri, target):
-    resp = do_http("PUT", target[-1] + "?t=uri", uri)
-    return check_PUT(resp)
-
-def copy_to_file(source, target):
-    assert source[0] == "file"
-    # do we need to copy bytes?
-    if source[1] == "local" or source[2] == True or target[1] == "local":
-        # yes
-        data = get_file_data(source)
-        put_file_data(data, target)
-        return
-    # no, we're getting data from an immutable source, and we're copying into
-    # the tahoe grid, so we can just copy the URI.
-    uri = source[3] or source[4] # prefer rw_uri, fall back to ro_uri
-    # TODO: if the original was mutable, and we're creating the target,
-    # should be we create a mutable file to match? At the moment we always
-    # create immutable files.
-    put_uri(uri, target)
-
-def copy_to_directory(source_files, source_dirs, target):
-    NotImplementedError
+        self.targetmap = {}
+        self.files_to_copy = 0
+
+        for source in source_file_infos:
+            if source[1] == "local":
+                (ig1, ig2, name, pathname) = source
+                s = LocalFileSource(pathname)
+            else:
+                assert source[1] == "tahoe"
+                (ig1, ig2, mutable, name, writecap, readcap, url) = source
+                s = TahoeFileSource(self.nodeurl, mutable,
+                                    writecap, readcap)
+            self.attach_to_target(s, name, target)
+            self.files_to_copy += 1
+
+        for source in source_dirs:
+            self.assign_targets(source, target)
+
+        self.progress("starting copy, %d files, %d directories" %
+                      (self.files_to_copy, len(self.targets)))
+        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.targets:
+            self.copy_files(self.targets[target], target)
+            self.targets_finished += 1
+            self.progress("%d/%d directories" %
+                          (self.targets_finished, len(self.targets)))
+
+    def attach_to_target(self, source, name, target):
+        if target not in self.targets:
+            self.targets[target] = {}
+        self.targets[target][name] = source
+        self.files_to_copy += 1
+
+    def assign_targets(self, source, target):
+        # copy everything in s to 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(source, subtarget)
+            else:
+                assert isinstance(child, (LocalFileSource, TahoeFileSource))
+                self.attach_to_target(source, name, target)
+
+
+
+    def copy_files(self, targetmap, target):
+        for name, source in targetmap.items():
+            assert isinstance(source, (LocalFileSource, TahoeFileSource))
+            self.copy_file(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.targets)))
+        target.set_children()
+
+    def need_to_copy_bytes(self, source, target):
+        if source.need_to_copy_bytes:
+            # mutable tahoe files, and local files
+            return True
+        if isinstance(target, LocalDirectoryTarget):
+            return True
+        return False
+
+    def copy_file(self, source, name, target):
+        assert isinstance(source, (LocalFileSource, TahoeFileSource))
+        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()
+            target.put_file(name, f)
+            return
+        # otherwise we're copying tahoe to tahoe, and using immutable files,
+        # so we can just make a link
+        target.put_uri(name, source.bestcap())
+
+
+    def progress(self, message):
+        print message
+        if self.progressfunc:
+            self.progressfunc(message)
+
+    def build_graphs(self, sources):
+        cache = {}
+        graphs = []
+        for source in sources:
+            assert source[0] == "directory"
+            if source[1] == "local":
+                root = LocalDirectorySource(self.progress, source[-1])
+                root.populate(True)
+            else:
+                assert source[1] == "tahoe"
+                (ig1, ig2, mutable, name, writecap, readcap, url) = source
+                root = TahoeDirectorySource(self.nodeurl, cache, self.progress)
+                root.init_from_grid(writecap, readcap)
+                root.populate(True)
+            graphs.append(root)
+        return graphs
+
+
+def copy(nodeurl, config, aliases, sources, destination,
+         verbosity, stdout, stderr):
+    c = Copier(nodeurl, config, aliases, verbosity, stdout, stderr)
+    return c.do_copy(sources, destination)
diff --git a/src/allmydata/test/test_system.py b/src/allmydata/test/test_system.py
index fbe34113..390e84d2 100644
--- a/src/allmydata/test/test_system.py
+++ b/src/allmydata/test/test_system.py
@@ -1603,9 +1603,12 @@ class SystemTest(testutil.SignalMixin, testutil.PollMixin, testutil.StallMixin,
             datas.append(data)
             open(fn,"wb").write(data)
 
-        def _check_stdout_against((out,err), filenum):
+        def _check_stdout_against((out,err), filenum=None, data=None):
             self.failUnlessEqual(err, "")
-            self.failUnlessEqual(out, datas[filenum])
+            if filenum is not None:
+                self.failUnlessEqual(out, datas[filenum])
+            if data is not None:
+                self.failUnlessEqual(out, data)
 
         # test all both forms of put: from a file, and from stdin
         #  tahoe put bar FOO
@@ -1778,6 +1781,28 @@ class SystemTest(testutil.SignalMixin, testutil.PollMixin, testutil.StallMixin,
         d.addCallback(run, "get", "tahoe:file3")
         d.addCallback(_check_stdout_against, 5)
 
+        # recursive copy: setup
+        dn = os.path.join(self.basedir, "dir1")
+        os.makedirs(dn)
+        open(os.path.join(dn, "file1"), "wb").write("file1")
+        open(os.path.join(dn, "file2"), "wb").write("file2")
+        open(os.path.join(dn, "file3"), "wb").write("file3")
+        sdn2 = os.path.join(dn, "subdir2")
+        os.makedirs(sdn2)
+        open(os.path.join(dn, "file4"), "wb").write("file4")
+        open(os.path.join(dn, "file5"), "wb").write("file5")
+
+        # from disk into tahoe
+        #d.addCallback(run, "cp", "-r", dn, "tahoe:dir1")
+        #d.addCallback(run, "ls")
+        #d.addCallback(_check_ls, ["dir1"])
+        #d.addCallback(run, "ls", "dir1")
+        #d.addCallback(_check_ls, ["file1", "file2", "file3", "subdir2"])
+        #d.addCallback(run, "ls", "tahoe:dir1/subdir2")
+        #d.addCallback(_check_ls, ["file4", "file5"])
+        #d.addCallback(run, "get", "dir1/subdir2/file4")
+        #d.addCallback(_check_stdout_against, data="file4")
+
         # tahoe_ls doesn't currently handle the error correctly: it tries to
         # JSON-parse a traceback.
 ##         def _ls_missing(res):