Rename prepare_alliccp to prepare_aliccp (#1145)

Co-authored-by: rnyak <[email protected]>
Co-authored-by: edknv <[email protected]>

Author: 3 people

merlin/datasets/ecommerce/__init__.py

@@ -17,14 +17,14 @@
 from merlin.datasets.ecommerce.aliccp.dataset import (
     default_aliccp_transformation,
     get_aliccp,
-    prepare_alliccp,
+    prepare_aliccp,
     transform_aliccp,
 )
 from merlin.datasets.ecommerce.booking.dataset import get_booking, transform_booking
 from merlin.datasets.ecommerce.dressipi.dataset import get_dressipi2022
 
 __all__ = [
-    "prepare_alliccp",
+    "prepare_aliccp",
     "transform_aliccp",
     "get_aliccp",
     "default_aliccp_transformation",

merlin/datasets/ecommerce/aliccp/dataset.py

@@ -56,7 +56,7 @@ def get_aliccp(
     1. Download the raw data from
     [tianchi.aliyun.com](https://tianchi.aliyun.com/dataset/dataDetail?dataId=408#1).
     2. Unzip the raw data to a directory.
-    3. Run `prepare_alliccp(data_dir)` to convert the raw data to parquet files.
+    3. Run `prepare_aliccp(data_dir)` to convert the raw data to parquet files.
 
     Downloading & preparing the data can take quite a while.
     In case you want to use this dataset to run our tutorials, you can also opt for synthetic data.
@@ -98,7 +98,7 @@ def get_aliccp(
     raw_path = p / "raw"
     if not raw_path.exists():
         raw_path.mkdir(parents=True)
-        prepare_alliccp(path, output_dir=raw_path, file_size=file_size, **kwargs)
+        prepare_aliccp(path, output_dir=raw_path, file_size=file_size, **kwargs)
 
     nvt_path = p / transformed_name
     train_path, valid_path = nvt_path / "train", nvt_path / "valid"
@@ -112,7 +112,7 @@ def get_aliccp(
     return train, valid
 
 
-def prepare_alliccp(
+def prepare_aliccp(
     data_dir: Union[str, Path],
     convert_train: bool = True,
     convert_test: bool = True,

tests/unit/datasets/test_ecommerce.py

@@ -88,7 +88,7 @@ def test_synthetic_aliccp_raw_data(tmp_path):
     MAYBE_ALICCP_DATA is None,
     reason="ALI-CCP data is not available, pass it through env variable $DATA_PATH_ALICCP",
 )
-def test_get_alliccp():
+def test_get_aliccp():
     data_path = MAYBE_ALICCP_DATA
 
     nvt_workflow = ecommerce.default_aliccp_transformation(add_target_encoding=False)
@@ -104,10 +104,10 @@ def test_get_alliccp():
     MAYBE_ALICCP_DATA is None,
     reason="ALI-CCP data is not available, pass it through env variable $DATA_PATH_ALICCP",
 )
-def test_prepare_alliccp(tmp_path):
+def test_prepare_aliccp(tmp_path):
     data_path = MAYBE_ALICCP_DATA
 
-    ecommerce.prepare_alliccp(data_path, file_size=50, max_num_rows=100, output_dir=tmp_path)
+    ecommerce.prepare_aliccp(data_path, file_size=50, max_num_rows=100, output_dir=tmp_path)
     output_files = list(tmp_path.glob("*/*"))
 
     assert len(output_files) == 2
@@ -118,7 +118,7 @@ def test_prepare_alliccp(tmp_path):
     MAYBE_ALICCP_DATA is None,
     reason="ALI-CCP data is not available, pass it through env variable $DATA_PATH_ALICCP",
 )
-def test_transform_alliccp(tmp_path):
+def test_transform_aliccp(tmp_path):
     data_path = MAYBE_ALICCP_DATA
 
     ecommerce.transform_aliccp(data_path, tmp_path)