[BF] Auto extractors and merge regex expressions (#275)

* fix extract_auto_entities, it was using every auto_extractor found. + fix merge auto_extractors with custom extractors * fix dict and add docstring * add random key to show fix extractors and auto_extractors * fix typo --------- Co-authored-by: bora2502 <[email protected]>
UNFmontreal · Oct 11, 2023 · cac5a25 · cac5a25
1 parent 6572021
commit cac5a25
Show file tree

Hide file tree

Showing 3 changed files with 33 additions and 9 deletions.
diff --git a/dcm2bids/sidecar.py b/dcm2bids/sidecar.py
@@ -11,7 +11,7 @@
 
 from dcm2bids.acquisition import Acquisition
 from dcm2bids.utils.io import load_json
-from dcm2bids.utils.utils import DEFAULT, convert_dir, splitext_
+from dcm2bids.utils.utils import DEFAULT, convert_dir, combine_dict_extractors, splitext_
 
 compare_float_keys = ["lt", "gt", "le", "ge", "btw", "btwe"]
 
@@ -428,7 +428,6 @@ def searchDcmTagEntity(self, sidecar, desc):
         descWithTask = desc.copy()
         concatenated_matches = {}
         entities = []
-
         if "custom_entities" in desc.keys() or self.auto_extract_entities:
             if 'custom_entities' in desc.keys():
                 if isinstance(descWithTask["custom_entities"], str):
@@ -437,7 +436,8 @@ def searchDcmTagEntity(self, sidecar, desc):
                 descWithTask["custom_entities"] = []
 
             if self.auto_extract_entities:
-                self.extractors.update(DEFAULT.auto_extractors)
+                self.extractors = combine_dict_extractors(self.extractors, DEFAULT.auto_extractors)
+
 
             for dcmTag in self.extractors:
                 if dcmTag in sidecar.data.keys():
@@ -455,8 +455,14 @@ def searchDcmTagEntity(self, sidecar, desc):
                                       compile_regex.search(curr_dcmInfo).groupdict())
                                     break
 
+            # Keep entities asked in custom_entities
+            # If dir found in custom_entities and concatenated_matches.keys we keep it
             if "custom_entities" in desc.keys():
-                entities = set(concatenated_matches.keys()).union(set(descWithTask["custom_entities"]))
+                entities = set(concatenated_matches.keys()).intersection(set(descWithTask["custom_entities"]))
+
+                # custom_entities not a key for extractor or auto_extract_entities
+                complete_entities = [ent for ent in descWithTask["custom_entities"] if '-' in ent]
+                entities = entities.union(set(complete_entities))
 
             if self.auto_extract_entities:
                 auto_acq = '_'.join([descWithTask['datatype'], descWithTask["suffix"]])
@@ -467,10 +473,11 @@ def searchDcmTagEntity(self, sidecar, desc):
                     if left_auto_entities:
                         self.logger.warning(f"{left_auto_entities} have not been found for datatype '{descWithTask['datatype']}' "
                                             f"and suffix '{descWithTask['suffix']}'.")
-                    else:
-                        entities = list(entities) + DEFAULT.auto_entities[auto_acq]
-                        entities = list(set(entities))
-                        descWithTask["custom_entities"] = entities
+
+                    entities = list(entities) + list(auto_entities)
+                    entities = list(set(entities))
+                    descWithTask["custom_entities"] = entities
+
 
             for curr_entity in entities:
                 if curr_entity in concatenated_matches.keys():

diff --git a/dcm2bids/utils/utils.py b/dcm2bids/utils/utils.py
@@ -142,6 +142,22 @@ def convert_dir(dir):
     return DEFAULT.entity_dir[dir]
 
 
+def combine_dict_extractors(d1, d2):
+    """ combine dict 
+    Args:
+        d1 (dic): dictionary
+        d2 (dic): dictionary
+    
+    Returns:
+        dict: dictionary with combined information
+              if d1 d2 use the same keys, return dict will return a list of items.
+    """    
+    return {
+            k: [d[k][0] for d in (d1, d2) if k in d]
+            for k in set(d1.keys()) | set(d2.keys())
+        }
+
+
 class TreePrinter:
     """
     Generates and prints a tree representation of a given a directory.

diff --git a/tests/data/config_test_auto_extract.json b/tests/data/config_test_auto_extract.json
@@ -1,5 +1,6 @@
 {
     "search_method": "re",
+    "extractors": {"SeriesDescription": ["random-(?P<random>[a-zA-Z0-9]+)"]},
     "descriptions": [
         {
             "datatype": "localizer",
@@ -12,7 +13,7 @@
             "id": "func_task-rest",
             "datatype": "func",
             "suffix": "bold",
-            "custom_entities": ["acq-highres"],
+            "custom_entities": ["acq-highres", "random"],
             "criteria": {
                 "SeriesDescription": ".*bold.*"
             }