BoxedAnnotations.concat should retain audio_file and annotation_file …

…lists

concatenates the lists from joined files, handling cases where the value is None instead of list

updated tests

resolves #1100

opensoundscape/annotations.py

@@ -1176,13 +1176,22 @@ def convert_labels(self, conversion_table):
     def concat(cls, list_of_boxed_annotations):
         """concatenate a list of BoxedAnnotations objects into one"""
         dfs = [ba.df for ba in list_of_boxed_annotations]
-        return cls(pd.concat(dfs).reset_index(drop=True))
-
-    @classmethod
-    def concat(cls, list_of_boxed_annotations):
-        """concatenate a list of BoxedAnnotations objects into one"""
-        dfs = [ba.df for ba in list_of_boxed_annotations]
-        return cls(pd.concat(dfs).reset_index(drop=True))
+        audio_files = []
+        annotation_files = []
+        for ba in list_of_boxed_annotations:
+            if ba.audio_files is not None:
+                audio_files.extend(ba.audio_files)
+            if ba.annotation_files is not None:
+                annotation_files.extend(ba.annotation_files)
+        if len(audio_files) == 0:
+            audio_files = None
+        if len(annotation_files) == 0:
+            annotation_files = None
+        return cls(
+            pd.concat(dfs).reset_index(drop=True),
+            audio_files=audio_files,
+            annotation_files=annotation_files,
+        )
 
 
 def diff(base_annotations, comparison_annotations):

tests/test_annotations.py

@@ -107,8 +107,8 @@ def boxed_annotations():
     )
     return BoxedAnnotations(
         df,
-        audio_files=["audio_file.wav"] * 3,
-        annotation_files=["audio_file.annotations.txt"] * 3,
+        audio_files=["audio_file.wav"],
+        annotation_files=["audio_file.annotations.txt"],
     )
 
 
@@ -186,6 +186,16 @@ def isnan(x):
 def test_concat_boxed_annotations(boxed_annotations):
     joined = BoxedAnnotations.concat([boxed_annotations] * 3)
     assert len(joined.df) == 9
+    assert len(joined.audio_files) == 3
+    assert len(joined.annotation_files) == 3
+
+    # handles scenario where audio_files and/or annotation_files are None
+    boxed_annotations.annotation_files = None
+    boxed_annotations.audio_files = None
+    joined = BoxedAnnotations.concat([boxed_annotations] * 3)
+    assert len(joined.df) == 9
+    assert joined.audio_files is None
+    assert joined.annotation_files is None
 
 
 def test_load_raven_annotations_w_audio(raven_file):