Use global constant for description.json (#2246)

#2246
ludwig-ai · Jul 10, 2022 · ee1c3c9 · ee1c3c9
1 parent 5c9cffb
commit ee1c3c9
Show file tree

Hide file tree

Showing 4 changed files with 9 additions and 6 deletions.
diff --git a/ludwig/globals.py b/ludwig/globals.py
@@ -24,6 +24,8 @@
 
 TEST_STATISTICS_FILE_NAME = "test_statistics.json"
 
+DESCRIPTION_FILE_NAME = "description.json"
+
 PREDICTIONS_PARQUET_FILE_NAME = "predictions.parquet"
 PREDICTIONS_SHAPES_FILE_NAME = "predictions.shapes.json"
 

diff --git a/ludwig/utils/misc_utils.py b/ludwig/utils/misc_utils.py
@@ -23,6 +23,7 @@
 import torch
 
 from ludwig.constants import PROC_COLUMN
+from ludwig.globals import DESCRIPTION_FILE_NAME
 from ludwig.utils.fs_utils import find_non_existing_dir_by_adding_suffix
 
 
@@ -112,7 +113,7 @@ def get_output_directory(output_directory, experiment_name, model_name="run"):
 
 
 def get_file_names(output_directory):
-    description_fn = os.path.join(output_directory, "description.json")
+    description_fn = os.path.join(output_directory, DESCRIPTION_FILE_NAME)
     training_stats_fn = os.path.join(output_directory, "training_statistics.json")
 
     model_dir = os.path.join(output_directory, "model")

diff --git a/tests/integration_tests/test_model_training_options.py b/tests/integration_tests/test_model_training_options.py
@@ -17,7 +17,7 @@
 from ludwig.contribs.mlflow import MlflowCallback
 from ludwig.experiment import experiment_cli
 from ludwig.features.number_feature import numeric_transformation_registry
-from ludwig.globals import TRAINING_PREPROC_FILE_NAME
+from ludwig.globals import DESCRIPTION_FILE_NAME, TRAINING_PREPROC_FILE_NAME
 from ludwig.schema.optimizers import optimizer_registry
 from ludwig.utils.data_utils import load_json, replace_file_extension
 from ludwig.utils.misc_utils import get_from_registry
@@ -122,7 +122,7 @@ def test_early_stopping(early_stop, generated_data, tmp_path):
 
     # test existence of required files
     train_stats_fp = os.path.join(output_dir, "training_statistics.json")
-    metadata_fp = os.path.join(output_dir, "description.json")
+    metadata_fp = os.path.join(output_dir, DESCRIPTION_FILE_NAME)
     assert os.path.isfile(train_stats_fp)
     assert os.path.isfile(metadata_fp)
 
@@ -370,7 +370,7 @@ def test_regularization(generated_data, tmp_path):
 
         # test existence of required files
         train_stats_fp = os.path.join(output_dir, "training_statistics.json")
-        metadata_fp = os.path.join(output_dir, "description.json")
+        metadata_fp = os.path.join(output_dir, DESCRIPTION_FILE_NAME)
         assert os.path.isfile(train_stats_fp)
         assert os.path.isfile(metadata_fp)
 

diff --git a/tests/integration_tests/test_visualization.py b/tests/integration_tests/test_visualization.py
@@ -29,7 +29,7 @@
 
 from ludwig.constants import TRAINER
 from ludwig.experiment import experiment_cli
-from ludwig.globals import PREDICTIONS_PARQUET_FILE_NAME, TEST_STATISTICS_FILE_NAME
+from ludwig.globals import DESCRIPTION_FILE_NAME, PREDICTIONS_PARQUET_FILE_NAME, TEST_STATISTICS_FILE_NAME
 from ludwig.utils.data_utils import get_split_path
 from ludwig.visualize import _extract_ground_truth_values
 from tests.integration_tests.test_visualization_api import obtain_df_splits
@@ -81,7 +81,7 @@ def get_output_feature_name(experiment_dir, output_feature=0):
     :return output_feature_name: name of the first output feature name
                         from the experiment
     """
-    description_file = os.path.join(experiment_dir, "description.json")
+    description_file = os.path.join(experiment_dir, DESCRIPTION_FILE_NAME)
     with open(description_file, "rb") as f:
         content = json.load(f)
     output_feature_name = content["config"]["output_features"][output_feature]["name"]