chore: use escape fn from oddrn library (#17)

odd_collector_aws/adapters/s3/mapper/dataset.py

@@ -1,3 +1,4 @@
+import re
 from typing import List, Dict, Any
 
 from lark import Lark
@@ -10,6 +11,7 @@
     DataEntityType,
 )
 from oddrn_generator.generators import S3Generator
+from oddrn_generator.utils import escape
 from pyarrow import Schema
 
 from .s3_field_type_transformer import S3FieldTypeTransformer
@@ -69,12 +71,12 @@ def __parse(field_type: str) -> Dict[str, Any]:
     return field_type_transformer.transform(column_tree)
 
 
-def s3_path_to_name(path: str, joiner: str = ":") -> str:
+def s3_path_to_name(path: str) -> str:
     """
     Remove the bucket name from the path and return the name of the file.
     """
-    without_bucket = path.rstrip("/").split("/")[1:]
-    return joiner.join(without_bucket)
+
+    return escape(re.sub("^[a-zA-z-.\d]*\/", "", path.strip("/"), 1))
 
 
 def map_dataset(

odd_collector_aws/adapters/sagemaker/domain/artifact.py

@@ -12,8 +12,8 @@
     DataInput,
 )
 from oddrn_generator import Generator
+from oddrn_generator.utils import escape
 
-from odd_collector_aws.const import S3_PATH_REPLACER
 from odd_collector_aws.domain.to_data_entity import ToDataEntity
 from odd_collector_aws.utils import parse_s3_url
 from .base_sagemaker_entity import BaseSagemakerEntity
@@ -124,7 +124,7 @@ def create_model(uri: str, arn: str):
 
 def create_dummy_dataset_artifact(uri: str, arn: str):
     bucket, key = parse_s3_url(uri)
-    name = key.replace("/", S3_PATH_REPLACER)
+    name = escape(key)
     return DummyDatasetArtifact(Name=name, Uri=uri, Arn=arn, ArtifactType="Dataset")
 
 

odd_collector_aws/const.py

@@ -1,3 +1 @@
 METADATA_PREFIX = "https://raw.githubusercontent.com/opendatadiscovery/opendatadiscovery-specification/main/specification/extensions"
-
-S3_PATH_REPLACER = ":"

tests/adapters/s3/test_s3_path_to_name.py

@@ -2,5 +2,5 @@
 
 
 def test_s3_path_to_name():
-    assert s3_path_to_name("bucket/path/to/file.csv", ":") == "path:to:file.csv"
-    assert s3_path_to_name("bucket/path/to/", ":") == "path:to"
+    assert s3_path_to_name("bucket/path/to/file.csv") == "path\\\\to\\\\file.csv"
+    assert s3_path_to_name("bucket/path/to/") == "path\\\\to"