RF: Session stringification

Author: mgxd

src/smriprep/utils/misc.py

@@ -97,18 +97,35 @@ def fs_isRunning(subjects_dir, subject_id, mtime_tol=86400, logger=None):
     return subjects_dir
 
 
-def hash_list(lst: list[str], digest_size: int = 2) -> str:
+def stringify_sessions(lst: list[str], max_length: int = 10, digest_size: int = 2) -> str:
     """
-    Hash a list of strings into a string.
+    Convert a list of session into a string identifier.
+    If the list has only one element, it returns that element.
+    If the list has more than one element, it concatenates them with '+'.
+    If the concatenated string exceeds `max_length`, it returns a
+    string starting with 'multi+' followed by a BLAKE2b hash of the concatenated string.
 
     Example
     -------
-    >>> hash_list(['a', 'b', 'c'])
-    '1279'
-    >>> hash_list(['a', 'b', 'c'], digest_size=3)
-    'fb7bf6'
+    >>> stringify_sessions(['a'])
+    'a'
+    >>> stringify_sessions(['a', 'b', 'c'])
+    'a+b+c'
+    >>> stringify_sessions(['a', 'b', 'toolong'])
+    'multi+7c22'
+    >>> stringify_sessions(['a', 'b', 'toolong'], max_length=12)
+    'a+b+toolong'
+    >>> stringify_sessions(['a', 'b', 'toolong'], digest_size=4)
+    'multi+dd8bb349'
+    >>>
 
     """
-    from hashlib import blake2b
+    if len(lst) == 1:
+        return lst[0]
 
-    return blake2b(','.join(lst).encode('utf-8'), digest_size=digest_size).hexdigest()
+    ses_str = '+'.join(lst)
+    if len(ses_str) > max_length:
+        from hashlib import blake2b
+
+        ses_str = f'multi+{blake2b(ses_str.encode(), digest_size=digest_size).hexdigest()}'
+    return ses_str

src/smriprep/workflows/base.py

@@ -184,12 +184,9 @@ def init_smriprep_wf(
         if session_ids:
             ses_str = session_ids
             if isinstance(session_ids, list):
-                if len(session_ids) == 1:
-                    ses_str = session_ids[0]
-                else:
-                    from smriprep.utils.misc import hash_list
+                from smriprep.utils.misc import stringify_sessions
 
-                    ses_str = hash_list(session_ids, digest_size=2)
+                ses_str = stringify_sessions(session_ids)
 
             name = f'sub-{subject_id}_ses-{ses_str}_wf'
 
@@ -501,12 +498,9 @@ def _prefix(subject_id, session_id=None):
     if session_id:
         ses_str = session_id
         if isinstance(session_id, list):
-            if len(session_id) == 1:
-                ses_str = session_id[0]
-            else:
-                from smriprep.utils.misc import hash_list
+            from smriprep.utils.misc import stringify_sessions
 
-                ses_str = f'multi{hash_list(session_id, digest_size=2)}'
+            ses_str = stringify_sessions(session_id)
         if not ses_str.startswith('ses-'):
             ses_str = f'ses-{ses_str}'
         subject_id += f'_{ses_str}'