[FIX] data/io.py Metadata file not saved anymore when it is empty #2002
Conversation
|
|
Codecov Report@@ Coverage Diff @@
## master #2002 +/- ##
=======================================
Coverage 89.53% 89.53%
=======================================
Files 91 91
Lines 9199 9199
=======================================
Hits 8236 8236
Misses 963 963Continue to review full report at Codecov.
|
Orange/tests/test_metadata_save.py
Outdated
| table = Table("iris") | ||
| fname = path.join(tempdir, "out.tab") | ||
| TabReader.write_table_metadata(fname, table) | ||
| self.assertEqual(path.isfile(fname + ".metadata"), True) |
Orange/tests/test_metadata_save.py
Outdated
| table = Table("titanic") | ||
| fname = path.join(tempdir, "out.tab") | ||
| TabReader.write_table_metadata(fname, table) | ||
| self.assertEqual(path.isfile(fname + ".metadata"), False) |
Orange/data/io.py
Outdated
| @@ -373,7 +373,7 @@ def write(cls, filename, data): | |||
|
|
|||
| @classmethod | |||
| def write_table_metadata(cls, filename, data): | |||
| if isinstance(filename, str) and hasattr(data, 'attributes'): | |||
| if isinstance(filename, str) and hasattr(data, 'attributes') and data.attributes: | |||
There was a problem hiding this comment.
hasattr(data, 'attributes') and data.attributes
could be simplified to
getattr(data, 'attributes', None)
Orange/tests/test_metadata_save.py
Outdated
| @@ -0,0 +1,36 @@ | |||
| # Test methods with long descriptive names can omit docstrings | |||
There was a problem hiding this comment.
Is there a special reason this tests were not placed in Orange/tests/test_tab_reader.py? They only test TabReaders functionality and all other TabReader tests are in that file.
Orange/tests/test_metadata_save.py
Outdated
| DiscreteVariable._clear_cache() | ||
|
|
||
| def test_metadata(self): | ||
| tempdir = tempfile.mkdtemp() |
There was a problem hiding this comment.
This tests assume that iris has metadata and titanic has not. If this would change in the future (if we added metadata for titanic as well), the second test would fail and will need to be fixed, even though the change would have nothing to do with the saving of metadata.
I suggest that you manually set the attributes dict to whatever you need in a test and then do the testing. This way, attributes dict will be the one you expect it to be, even if the datasets change in the future.
Issue Do not save the metadata file (when saving the data) if this is empty (0 Bytes) Description of changes Change if statement and add third condition. It implies that attribute should not be empty.
Issue
Do not save the metadata file (when saving the data) if this is empty (0 Bytes)
Description of changes
Change if statement and add third condition. It implies that attribute should not be empty. Add test file to check if metadata is saved when empty
Includes