add is_gs_url(), split_gs_url() static methods to gs_utils.py

py/utils/gs_utils_manualtest.py

 #!/usr/bin/python
 
 """Tests for gs_utils.py.
 
 TODO(epoger): How should we exercise these self-tests? See http://skbug.com/2751
 """
 
 # System-level imports.
 import os
 import posixpath
@@ skipping 21 matching lines @@
 into gs://%s?
 """ % TEST_BUCKET
     raise
 
 
 def _get_unique_posix_dir():
   """Returns a unique directory name suitable for use in Google Storage."""
   return 'gs_utils_manualtest/%d' % random.randint(0, sys.maxint)
 
 
+def _test_static_methods():
+  """Test all static methods."""
+  gs = gs_utils.GSUtils
+
+  # (input url, output bucket, output path) for each test case
+  testcases = [
+      (None,             None, None),
+      (5,                None, None),
+      ('',               None, None),
+      ('/one/two',       None, None),
+      ('http://one/two', None, None),
+      ('gs:',            None, None),
+      ('gs://',          None, None),
+      ('gs:///',         None, None),
+      ('gs://???',       None, None),
+      ('gs:///bucket',   None, None),
+      ('gs://bucket',             'bucket', ''),
+      ('gs://bucket/',            'bucket', ''),
+      ('gs://bucket/path1',       'bucket', 'path1'),

[rmistry, 2014/07/29 20:36:20]

Could you also please add gs://bucket/path1/ and maybe gs://bucket/path1//

[epoger, 2014/07/30 01:10:49]

Done.

+      ('gs://bucket/path1/path2', 'bucket', 'path1/path2'),

[rmistry, 2014/07/29 20:36:20]

Optional Nit: Instead of aligning some of these lets not align any of them. I
always prefer not aligning because if we need to add any new test cases that are
longer than the longest one then we will have to change all of them instead of
it being a one-line change.

[epoger, 2014/07/30 01:10:49]

Yeah, that makes sense, and the style guide agrees with you, so you have THAT
going for you, which is nice...

Done.

+  ]
+  for (url, bucket, path) in testcases:
+    is_legal_url = (bucket != None)
+    assert gs.is_gs_url(url) == is_legal_url, 'gs.is_gs_url("%s") == %s' % (
+        url, is_legal_url)
+    if is_legal_url:
+      assert gs.split_gs_url(url) == (bucket, path), (
+          'gs.split_gs_url("%s") == ("%s", "%s")' % (url, bucket, path))
+
+
 def _test_public_read():
   """Make sure we can read from public files without .boto file credentials."""
   gs = gs_utils.GSUtils()
   gs.list_bucket_contents(bucket=TEST_BUCKET, subdir=None)
 
 
 def _test_upload_if_one_file():
   """Test upload_if param within upload_file()."""
   gs = _get_authenticated_gs_handle()
   filename = 'filename'
@@ skipping 296 matching lines @@
       assert file_contents == 'contents of %s\n' % filename, (
           '%s == "contents of %s\n"' % (file_contents, filename))
   finally:
     shutil.rmtree(local_dest_dir)
     for filename in filenames:
       gs.delete_file(bucket=TEST_BUCKET,
                      path=posixpath.join(remote_dir, subdir, filename))
 
 
 if __name__ == '__main__':
+  _test_static_methods()
   _test_upload_if_multiple_files()
   _test_upload_if_one_file()
   _test_public_read()
   _test_authenticated_round_trip()
   _test_dir_upload_and_download()
   # TODO(epoger): Add _test_unauthenticated_access() to make sure we raise
   # an exception when we try to access without needed credentials.