Updated the attrs['source'] hash strings:

* Now includes the git commit ID in the source string when the repository has uncommitted changes for reference.
* Removed assumptions about relative paths and repository roots in the input file paths. The previous approach assumed the data issues path was already the root, causing the function to fail.

src/pypromice/process/aws.py

@@ -55,7 +55,15 @@ def __init__(
         """
         assert os.path.isfile(config_file), "cannot find " + config_file
         assert os.path.isdir(inpath), "cannot find " + inpath
-        logger.info("AWS object initialising...")
+        logger.info(
+            "AWS("
+            f"config_file={config_file},"
+            f" inpath={inpath},"
+            f" data_issues_repository={data_issues_repository},"
+            f" var_file={var_file},"
+            f" meta_file={meta_file}"
+            ")"
+        )
 
         # Load config, variables CSF standards, and L0 files
         self.config = self.loadConfig(config_file, inpath)
@@ -73,6 +81,7 @@ def __init__(
             l0_data_root=inpath_hash,
             data_issues=data_issues_hash,
         )
+        logger.debug('Source information: %s', source_dict)
         self.meta["source"] = json.dumps(source_dict)
 
         # Load config file

src/pypromice/utilities/git.py

@@ -7,12 +7,16 @@
 logger = logging.getLogger(__name__)
 
 
-def get_commit_hash_and_check_dirty(file_path) -> str:
-    repo_path = Path(file_path).parent
+def get_commit_hash_and_check_dirty(file_path: str | Path) -> str:
+    if isinstance(file_path, str):
+        file_path = Path(file_path)
+    if file_path.is_dir():
+        repo_path = file_path
+    else:
+        repo_path = file_path.parent
 
     try:
         # Ensure the file path is relative to the repository
-        relative_file_path = os.path.relpath(file_path, repo_path)
 
         # Get the latest commit hash for the file
         commit_hash = (
@@ -25,8 +29,6 @@ def get_commit_hash_and_check_dirty(file_path) -> str:
                     "-n",
                     "1",
                     "--pretty=format:%H",
-                    #"--",
-                    #relative_file_path,
                 ],
                 stderr=subprocess.STDOUT,
             )
@@ -49,7 +51,7 @@ def get_commit_hash_and_check_dirty(file_path) -> str:
 
         if is_dirty:
             logger.warning(f"Warning: The file {file_path} is dirty compared to the last commit. {commit_hash}")
-            return 'unknown'
+            return f'{commit_hash} (dirty)'
         if commit_hash == "":
             logger.warning(f"Warning: The file {file_path} is not under version control.")
             return 'unknown'

tests/e2e/test_get_l2.py

@@ -129,6 +129,6 @@ def test_get_l2_raw(self):
                 )
                 data_root_hash = source_decoded["l0_data_root"]
                 data_issues_hash = source_decoded["data_issues"]
-                self.assertNotEquals(config_hash, 'unknown', 'This test will fail while the commit is dirty')
-                self.assertNotEquals(data_root_hash, 'unknown', 'This test will fail while the commit is dirty')
-                self.assertNotEquals(data_issues_hash, 'unknown', 'This test will fail while the commit is dirty')
+                self.assertFalse(config_hash.endswith(" (dirty)"), 'This test will fail while the commit is dirty')
+                self.assertFalse(data_root_hash.endswith(" (dirty)"), 'This test will fail while the commit is dirty')
+                self.assertFalse(data_issues_hash.endswith(" (dirty)"), 'This test will fail while the commit is dirty')