Dataset handle mismatches

elpis/datasets/dataset.py

@@ -153,6 +153,14 @@ def from_dict(cls, data: Dict[str, Any]) -> Dataset:
             elan_options=elan_options,
         )
 
+    @property
+    def valid_transcriptions(self):
+        return (
+            self._transcript_files()
+            .difference(self.mismatched_files())
+            .difference(self.colliding_files())
+        )
+
     def to_batches(self) -> List[ProcessingBatch]:
         """Converts a valid dataset to a list of processing jobs, matching
         transcript and audio files.
@@ -164,7 +172,7 @@ def to_batches(self) -> List[ProcessingBatch]:
                 cleaning_options=self.cleaning_options,
                 elan_options=self.elan_options,
             )
-            for transcription_file in self._transcript_files()
+            for transcription_file in self.valid_transcriptions
         ]
 
     def to_dict(self) -> Dict[str, Any]:

tests/datasets/test_dataset.py

@@ -47,13 +47,18 @@ def test_serialize_dataset_options():
 FILES_WITHOUT_ELAN = ["1.txt", "1.wav"]
 MISMATCHED_FILES = ["1.eaf", "1.wav", "2.wav", "3.txt"]
 COLLIDING_FILES = ["1.eaf", "1.wav", "1.txt"]
-
+MESSY_FILES = ["1.eaf", "1.wav", "2.eaf", "2.txt", "2.wav", "3.eaf", "4.wav"]
 
 DATASET_DICT = {
     "name": "dataset",
     "files": FILES_WITH_ELAN,
     "cleaning_options": CLEANING_OPTIONS_DICT,
 }
+MESSY_DATASET_DICT = {
+    "name": "dataset",
+    "files": MESSY_FILES,
+    "cleaning_options": CLEANING_OPTIONS_DICT,
+}
 
 DATASET_DICT_ELAN = DATASET_DICT | {"elan_options": ELAN_OPTIONS_DICT}
 
@@ -124,6 +129,11 @@ def test_duplicate_files():
     assert set(dataset.colliding_files()) == {Path("1.eaf"), Path("1.txt")}
 
 
+def test_valid_transcriptions():
+    dataset = Dataset.from_dict(MESSY_DATASET_DICT)
+    assert len(dataset.valid_transcriptions) == 1
+
+
 def test_dataset_batching():
     dataset = Dataset.from_dict(DATASET_DICT)
     batch = dataset.to_batches()