Merge branch '17800-arvput-no-follow-links'

Closes #17800

Arvados-DCO-1.1-Signed-off-by: Lucas Di Pentima <lucas.dipentima@curii.com>

diff --git a/sdk/python/arvados/commands/put.py b/sdk/python/arvados/commands/put.py
index 0920369646c6b7556babdef38fced98fee2e18b8..4f0ac1fffaf9c4ee6875c0092ba1090d6e3eb9a6 100644 (file)
--- a/sdk/python/arvados/commands/put.py
+++ b/sdk/python/arvados/commands/put.py
@@ -173,7 +173,8 @@ Follow file and directory symlinks (default).
 """)
 _group.add_argument('--no-follow-links', action='store_false', dest='follow_links',
                     help="""
-Do not follow file and directory symlinks.
+Ignore file and directory symlinks. Even paths given explicitly on the
+command line will be skipped if they are symlinks.
 """)
 
 
@@ -259,9 +260,8 @@ def parse_arguments(arguments):
 
     args.paths = ["-" if x == "/dev/stdin" else x for x in args.paths]
 
-    if len(args.paths) != 1 or os.path.isdir(args.paths[0]):
-        if args.filename:
-            arg_parser.error("""
+    if args.filename and (len(args.paths) != 1 or os.path.isdir(args.paths[0])):
+        arg_parser.error("""
     --filename argument cannot be used when storing a directory or
     multiple files.
     """)
@@ -525,6 +525,9 @@ class ArvPutUploadJob(object):
                 self._write_stdin(self.filename or 'stdin')
             elif not os.path.exists(path):
                  raise PathDoesNotExistError(u"file or directory '{}' does not exist.".format(path))
+            elif (not self.follow_links) and os.path.islink(path):
+                self.logger.warning("Skipping symlink '{}'".format(path))
+                continue
             elif os.path.isdir(path):
                 # Use absolute paths on cache index so CWD doesn't interfere
                 # with the caching logic.
@@ -659,6 +662,9 @@ class ArvPutUploadJob(object):
             self._remote_collection.save(num_retries=self.num_retries,
                                          trash_at=self._collection_trash_at())
         else:
+            if len(self._local_collection) == 0:
+                self.logger.warning("No files were uploaded, skipping collection creation.")
+                return
             self._local_collection.save_new(
                 name=self.name, owner_uuid=self.owner_uuid,
                 ensure_unique_name=self.ensure_unique_name,
@@ -1310,7 +1316,7 @@ def main(arguments=None, stdout=sys.stdout, stderr=sys.stderr,
             output = writer.manifest_text()
     elif args.raw:
         output = ','.join(writer.data_locators())
-    else:
+    elif writer.manifest_locator() is not None:
         try:
             expiration_notice = ""
             if writer.collection_trash_at() is not None:
@@ -1336,6 +1342,8 @@ def main(arguments=None, stdout=sys.stdout, stderr=sys.stderr,
                 "arv-put: Error creating Collection on project: {}.".format(
                     error))
             status = 1
+    else:
+        status = 1
 
     # Print the locator (uuid) of the new collection.
     if output is None:

diff --git a/sdk/python/tests/test_arv_put.py b/sdk/python/tests/test_arv_put.py
index caa03a3e03c02692204f3e707fe31df7a97961ad..fac970c95a6fdb13ecbef709ff850798c67840a5 100644 (file)
--- a/sdk/python/tests/test_arv_put.py
+++ b/sdk/python/tests/test_arv_put.py
@@ -295,6 +295,8 @@ class ArvPutUploadJobTest(run_test_server.TestCaseWithServers,
         shutil.rmtree(self.tempdir_with_symlink)
 
     def test_symlinks_are_followed_by_default(self):
+        self.assertTrue(os.path.islink(os.path.join(self.tempdir_with_symlink, 'linkeddir')))
+        self.assertTrue(os.path.islink(os.path.join(self.tempdir_with_symlink, 'linkedfile')))
         cwriter = arv_put.ArvPutUploadJob([self.tempdir_with_symlink])
         cwriter.start(save_collection=False)
         self.assertIn('linkeddir', cwriter.manifest_text())
@@ -302,12 +304,29 @@ class ArvPutUploadJobTest(run_test_server.TestCaseWithServers,
         cwriter.destroy_cache()
 
     def test_symlinks_are_not_followed_when_requested(self):
+        self.assertTrue(os.path.islink(os.path.join(self.tempdir_with_symlink, 'linkeddir')))
+        self.assertTrue(os.path.islink(os.path.join(self.tempdir_with_symlink, 'linkedfile')))
         cwriter = arv_put.ArvPutUploadJob([self.tempdir_with_symlink],
                                           follow_links=False)
         cwriter.start(save_collection=False)
         self.assertNotIn('linkeddir', cwriter.manifest_text())
         self.assertNotIn('linkedfile', cwriter.manifest_text())
         cwriter.destroy_cache()
+        # Check for bug #17800: passed symlinks should also be ignored.
+        linked_dir = os.path.join(self.tempdir_with_symlink, 'linkeddir')
+        cwriter = arv_put.ArvPutUploadJob([linked_dir], follow_links=False)
+        cwriter.start(save_collection=False)
+        self.assertNotIn('linkeddir', cwriter.manifest_text())
+        cwriter.destroy_cache()
+
+    def test_no_empty_collection_saved(self):
+        self.assertTrue(os.path.islink(os.path.join(self.tempdir_with_symlink, 'linkeddir')))
+        linked_dir = os.path.join(self.tempdir_with_symlink, 'linkeddir')
+        cwriter = arv_put.ArvPutUploadJob([linked_dir], follow_links=False)
+        cwriter.start(save_collection=True)
+        self.assertIsNone(cwriter.manifest_locator())
+        self.assertEqual('', cwriter.manifest_text())
+        cwriter.destroy_cache()
 
     def test_passing_nonexistant_path_raise_exception(self):
         uuid_str = str(uuid.uuid4())