Fix OOV file paths in mfa validate output and a minor logging typo

montreal_forced_aligner/dictionary/multispeaker.py

@@ -1478,7 +1478,7 @@ def save_oovs_found(self, directory: str) -> None:
         Parameters
         ----------
         directory : str
-            Path to directory to save ``oovs_found.txt``
+            Path to directory to save ``oovs_found_*.txt`` files.
         """
         with self.session() as session:
             for dict_id, base_name in self.dictionary_base_names.items():

montreal_forced_aligner/validation/corpus_validator.py

@@ -140,7 +140,15 @@ def analyze_oovs(self, output_directory: Path = None) -> None:
         if output_directory is None:
             output_directory = self.output_directory
         os.makedirs(output_directory, exist_ok=True)
-        oov_path = os.path.join(output_directory, "oovs_found.txt")
+        # Point users to the actual OOV files written by save_oovs_found()
+        if hasattr(self, "dictionary_base_names"):
+            if len(self.dictionary_base_names) == 1:
+                only_base = next(iter(self.dictionary_base_names.values()))
+                oov_path = os.path.join(output_directory, f"oovs_found_{only_base}.txt")
+            else:
+                oov_path = os.path.join(output_directory, "oovs_found_*.txt")
+        else:
+            oov_path = os.path.join(output_directory, "oovs_found.txt")
         utterance_oov_path = os.path.join(output_directory, "utterance_oovs.txt")
 
         total_instances = 0
@@ -169,7 +177,7 @@ def analyze_oovs(self, output_directory: Path = None) -> None:
         if self.oovs_found:
             self.save_oovs_found(output_directory)
             logger.warning(f"{len(self.oovs_found)} OOV word types")
-            logger.warning(f"{total_instances}total OOV tokens")
+            logger.warning(f"{total_instances} total OOV tokens")
             logger.warning(
                 f"For a full list of the word types, please see: {oov_path}. "
                 f"For a by-utterance breakdown of missing words, see: {utterance_oov_path}"

tests/test_commandline_validate.py

@@ -36,6 +36,8 @@ def test_validate_corpus(
         print(result.exc_info)
         raise result.exception
     assert not result.return_value
+    # Regression test for oovs_found_*.txt path
+    assert "oovs_found_english_us" in result.stderr
 
 
 def test_validate_training_corpus(