Merge pull request #396 from OP-TED/feature/TED-1045

Updated mapping_suite files hasher

Author: kaleanych

ted_sws/mapping_suite_processor/adapters/mapping_suite_hasher.py

@@ -8,6 +8,7 @@
 """ """
 import hashlib
 import pathlib
+import re
 from typing import Tuple, List, Union
 
 from ted_sws.data_manager.adapters.mapping_suite_repository import MS_TRANSFORM_FOLDER_NAME, \
@@ -35,7 +36,10 @@ def _hash_a_file(file_path: pathlib.Path) -> Tuple[str, str]:
             """
                 Return a tuple of the relative file path and the file hash.
             """
-            hashed_line = hashlib.sha256(file_path.read_bytes()).hexdigest()
+            # remove new-lines to align content generated on different operating systems
+            new_line_pattern = re.compile(b'\r\n|\r|\n')
+            file_content = re.sub(new_line_pattern, b'', file_path.read_bytes())
+            hashed_line = hashlib.sha256(file_content).hexdigest()
             relative_path = str(file_path).replace(str(self.mapping_suite_path), "")
             return relative_path, hashed_line
 

tests/test_data/package_F03_demo/metadata.json

@@ -22,5 +22,5 @@
       ]
     }
   },
-  "mapping_suite_hash_digest": "0d02071125d6fe9d2b1e862902dae254beb253046038dfe9ddb0b5b792524b8f"
+  "mapping_suite_hash_digest": "4b30cc45f45e4134394f571ac612b48efa8738cab77b9aad0290b779a5232c8f"
 }