Tagging constraints (#1305)

* update tagging constraints

* openml python tests

* small fix

Author: janvanrijn

tests/test_datasets/test_dataset.py

@@ -316,7 +316,9 @@ def setUp(self):
         self.dataset = openml.datasets.get_dataset(125, download_data=False)
 
     def test_tagging(self):
-        tag = f"test_tag_OpenMLDatasetTestOnTestServer_{time()}"
+        # tags can be at most 64 alphanumeric (+ underscore) chars
+        unique_indicator = str(time()).replace('.', '')
+        tag = f"test_tag_OpenMLDatasetTestOnTestServer_{unique_indicator}"
         datasets = openml.datasets.list_datasets(tag=tag, output_format="dataframe")
         assert datasets.empty
         self.dataset.push_tag(tag)
@@ -327,7 +329,6 @@ def test_tagging(self):
         datasets = openml.datasets.list_datasets(tag=tag, output_format="dataframe")
         assert datasets.empty
 
-
 class OpenMLDatasetTestSparse(TestBase):
     _multiprocess_can_split_ = True
 

tests/test_datasets/test_dataset_functions.py

@@ -150,6 +150,24 @@ def test_check_datasets_active(self):
         )
         openml.config.server = self.test_server
 
+    def test_illegal_character_tag(self):
+        dataset = openml.datasets.get_dataset(1)
+        tag = "illegal_tag&"
+        try:
+            dataset.push_tag(tag)
+            assert False
+        except openml.exceptions.OpenMLServerException as e:
+            assert e.code == 477
+
+    def test_illegal_length_tag(self):
+        dataset = openml.datasets.get_dataset(1)
+        tag = "a" * 65
+        try:
+            dataset.push_tag(tag)
+            assert False
+        except openml.exceptions.OpenMLServerException as e:
+            assert e.code == 477
+
     def _datasets_retrieved_successfully(self, dids, metadata_only=True):
         """Checks that all files for the given dids have been downloaded.
 

tests/test_flows/test_flow.py

@@ -102,7 +102,9 @@ def test_tagging(self):
         flows = openml.flows.list_flows(size=1, output_format="dataframe")
         flow_id = flows["id"].iloc[0]
         flow = openml.flows.get_flow(flow_id)
-        tag = f"test_tag_TestFlow_{time.time()}"
+        # tags can be at most 64 alphanumeric (+ underscore) chars
+        unique_indicator = str(time()).replace('.', '')
+        tag = f"test_tag_TestFlow_{unique_indicator}"
         flows = openml.flows.list_flows(tag=tag, output_format="dataframe")
         assert len(flows) == 0
         flow.push_tag(tag)

tests/test_runs/test_run.py

@@ -30,7 +30,9 @@ def test_tagging(self):
         assert not runs.empty, "Test server state is incorrect"
         run_id = runs["run_id"].iloc[0]
         run = openml.runs.get_run(run_id)
-        tag = f"test_tag_TestRun_{time()}"
+        # tags can be at most 64 alphanumeric (+ underscore) chars
+        unique_indicator = str(time()).replace('.', '')
+        tag = f"test_tag_TestRun_{unique_indicator}"
         runs = openml.runs.list_runs(tag=tag, output_format="dataframe")
         assert len(runs) == 0
         run.push_tag(tag)

tests/test_tasks/test_task_methods.py

@@ -17,7 +17,9 @@ def tearDown(self):
 
     def test_tagging(self):
         task = openml.tasks.get_task(1)  # anneal; crossvalidation
-        tag = f"test_tag_OpenMLTaskMethodsTest_{time()}"
+        # tags can be at most 64 alphanumeric (+ underscore) chars
+        unique_indicator = str(time()).replace('.', '')
+        tag = f"test_tag_OpenMLTaskMethodsTest_{unique_indicator}"
         tasks = openml.tasks.list_tasks(tag=tag, output_format="dataframe")
         assert len(tasks) == 0
         task.push_tag(tag)