cp: error on target-filename collisions, rather than overwrite

[tahoe-lafs/tahoe-lafs.git] / src / allmydata / scripts / tahoe_cp.py

diff --git a/src/allmydata/scripts/tahoe_cp.py b/src/allmydata/scripts/tahoe_cp.py
index a5cfe16a1a25b7b65e9ca9d9856256457b7da9a4..351b6a331717db90bda955fed187998cd8f2a4e1 100644 (file)
--- a/src/allmydata/scripts/tahoe_cp.py
+++ b/src/allmydata/scripts/tahoe_cp.py
@@ -20,6 +20,14 @@ class MissingSourceError(TahoeError):
     def __init__(self, name, quotefn=quote_output):
         TahoeError.__init__(self, "No such file or directory %s" % quotefn(name))
 
+class FilenameWithTrailingSlashError(TahoeError):
+    def __init__(self, name, quotefn=quote_output):
+        TahoeError.__init__(self, "source '%s' is not a directory, but ends with a slash" % quotefn(name))
+
+class WeirdSourceError(TahoeError):
+    def __init__(self, absname):
+        quoted = quote_local_unicode_path(absname)
+        TahoeError.__init__(self, "source '%s' is neither a file nor a directory, I can't handle it" % quoted)
 
 def GET_to_file(url):
     resp = do_http("GET", url)
@@ -488,8 +496,9 @@ class Copier:
 
     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().
+        All usage errors (except for target filename collisions) 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
@@ -501,7 +510,11 @@ class Copier:
 
         sources = [] # list of source objects
         for ss in source_specs:
-            si = self.get_source_info(ss)
+            try:
+                si = self.get_source_info(ss)
+            except FilenameWithTrailingSlashError as e:
+                self.to_stderr(str(e))
+                return 1
             precondition(isinstance(si, FileSources + DirectorySources), si)
             sources.append(si)
 
@@ -622,6 +635,9 @@ class Copier:
         precondition(isinstance(source_spec, unicode), source_spec)
         rootcap, path_utf8 = get_alias(self.aliases, source_spec, None)
         path = path_utf8.decode("utf-8")
+        # any trailing slash is removed in abspath_expanduser_unicode(), so
+        # make a note of it here, to throw an error later
+        had_trailing_slash = path.endswith("/")
         if rootcap == DefaultAliasMarker:
             # no alias, so this is a local file
             pathname = abspath_expanduser_unicode(path)
@@ -631,7 +647,11 @@ class Copier:
             if os.path.isdir(pathname):
                 t = LocalDirectorySource(self.progress, pathname, name)
             else:
-                assert os.path.isfile(pathname)
+                if had_trailing_slash:
+                    raise FilenameWithTrailingSlashError(source_spec,
+                                                         quotefn=quote_local_unicode_path)
+                if not os.path.isfile(pathname):
+                    raise WeirdSourceError(pathname)
                 t = LocalFileSource(pathname, name) # non-empty
         else:
             # this is a tahoe object
@@ -659,6 +679,8 @@ class Copier:
                                          self.progress, name)
                 t.init_from_parsed(parsed)
             else:
+                if had_trailing_slash:
+                    raise FilenameWithTrailingSlashError(source_spec)
                 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
@@ -714,6 +736,23 @@ class Copier:
         # sourceobject) dicts for all the files that need to wind up there.
         targetmap = self.build_targetmap(sources, target)
 
+        # target name collisions are an error
+        collisions = []
+        for target, sources in targetmap.items():
+            target_names = {}
+            for source in sources:
+                name = source.basename()
+                if name in target_names:
+                    collisions.append((target, source, target_names[name]))
+                else:
+                    target_names[name] = source
+        if collisions:
+            self.to_stderr("cannot copy multiple files with the same name into the same target directory")
+            # I'm not sure how to show where the collisions are coming from
+            #for (target, source1, source2) in collisions:
+            #    self.to_stderr(source1.basename())
+            return 1
+
         # 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.