Add check for config.fail_on_decode_error

audalign/__init__.py

@@ -807,6 +807,7 @@ def uniform_level_file(
     width: float = 5,
     overlap_ratio: float = 0.5,
     exclude_min_db: float = -70,
+    config: BaseConfig = FingerprintConfig(),
 ) -> None:
     """
     Levels the file using either of two methods: normalize or average.
@@ -838,6 +839,7 @@ def uniform_level_file(
         width=width,
         overlap_ratio=overlap_ratio,
         exclude_min_db=exclude_min_db,
+        base_config=config,
     )
 
 
@@ -851,6 +853,7 @@ def uniform_level_directory(
     exclude_min_db: float = -70,
     multiprocessing: bool = True,
     num_processors: int = None,
+    config: BaseConfig = FingerprintConfig(),
 ) -> None:
     """
     Levels the file using either of two methods: normalize or average.
@@ -886,6 +889,7 @@ def uniform_level_directory(
         exclude_min_db=exclude_min_db,
         use_multiprocessing=multiprocessing,
         num_processes=num_processors,
+        config=config,
     )
 
 
@@ -897,6 +901,7 @@ def remove_noise_file(
     write_extension: str = None,
     alt_noise_filepath: str = None,
     prop_decrease: float = 1,
+    config: BaseConfig = FingerprintConfig(),
     **kwargs,
 ):
     """Remove noise from audio file by specifying start and end seconds of representative sound sections. Writes file to destination
@@ -920,6 +925,7 @@ def remove_noise_file(
         write_extension=write_extension,
         alt_noise_filepath=alt_noise_filepath,
         prop_decrease=prop_decrease,
+        config=config,
         **kwargs,
     )
 
@@ -934,6 +940,7 @@ def remove_noise_directory(
     prop_decrease: float = 1,
     multiprocessing: bool = True,
     num_processors: int = None,
+    config: BaseConfig = FingerprintConfig(),
     **kwargs,
 ):
     """Remove noise from audio files in directory by specifying start and end seconds of
@@ -962,6 +969,7 @@ def remove_noise_directory(
         prop_decrease=prop_decrease,
         use_multiprocessing=multiprocessing,
         num_processes=num_processors,
+        config=config,
         **kwargs,
     )
 

audalign/filehandler.py

@@ -212,6 +212,7 @@ def noise_remove(
     write_extension: str = None,
     alt_noise_filepath=None,
     prop_decrease=1,
+    config: BaseConfig = FingerprintConfig(),
     **kwargs,
 ):
     audiofile = create_audiosegment(filepath)
@@ -238,6 +239,7 @@ def noise_remove(
         sr=BaseConfig.sample_rate,
         y_noise=noisy_part,
         prop_decrease=prop_decrease,
+        base_config=config,
         **kwargs,
     )
 
@@ -270,6 +272,7 @@ def noise_remove_directory(
     prop_decrease=1,
     use_multiprocessing=False,
     num_processes=None,
+    config: BaseConfig = FingerprintConfig(),
     **kwargs,
 ):
     noise_data = _floatify_data(create_audiosegment(noise_filepath))[
@@ -285,6 +288,7 @@ def noise_remove_directory(
         destination_directory=destination_directory,
         prop_decrease=prop_decrease,
         write_extension=write_extension,
+        base_config=config,
         **kwargs,
     )
 
@@ -395,6 +399,7 @@ def uniform_level_directory(
     exclude_min_db=-70,
     use_multiprocessing=False,
     num_processes=None,
+    config: BaseConfig = FingerprintConfig(),
 ):
     _uniform_level_ = partial(
         _uniform_level,
@@ -404,6 +409,7 @@ def uniform_level_directory(
         width=width,
         overlap_ratio=overlap_ratio,
         exclude_min_db=exclude_min_db,
+        base_config=config,
     )
 
     if use_multiprocessing == True:

audalign/recognizers/__init__.py

@@ -4,6 +4,7 @@
 
 import audalign.filehandler as filehandler
 from audalign.config import BaseConfig
+from pydub.exceptions import CouldntDecodeError
 
 
 class BaseRecognizer(ABC):
@@ -47,10 +48,18 @@ def align_get_file_names(
             file_names = [os.path.basename(x) for x in file_list]
         elif file_dir:
             file_names = filehandler.get_audio_files_directory(
-                    file_dir, False, self.config.can_read_extensions, self.config.cant_read_extensions)
+                file_dir,
+                False,
+                self.config.can_read_extensions,
+                self.config.cant_read_extensions,
+            )
         elif fine_aud_file_dict:
+            if fine_aud_file_dict == None or len(fine_aud_file_dict.keys()) == 0:
+                raise CouldntDecodeError("No files found", fine_aud_file_dict)
             file_names = [os.path.basename(x) for x in fine_aud_file_dict.keys()]
         else:
+            if file_list == None or len(file_list) == 0:
+                raise CouldntDecodeError("No files found", file_list)
             file_names = [os.path.basename(x) for x in file_list]
         return file_names
 

audalign/recognizers/correcognize/correcognize.py

@@ -336,8 +336,10 @@ def _correcognize_dir(
             **kwargs,
         )
 
-    except CouldntDecodeError:
+    except CouldntDecodeError as e:
         print(f'File "{against_file_path}" could not be decoded')
+        if config.fail_on_decode_error:
+            raise e
         return {}
 
 

audalign/recognizers/correcognizeSpectrogram/correcognize_spectrogram.py

@@ -336,8 +336,10 @@ def _correcognize_dir(
             **kwargs,
         )
 
-    except CouldntDecodeError:
+    except CouldntDecodeError as e:
         print(f'File "{against_file_path}" could not be decoded')
+        if config.fail_on_decode_error:
+            raise e
         return {}
 
 

audalign/recognizers/fingerprint/__init__.py

@@ -10,6 +10,7 @@
 import pickle
 import json
 import typing
+from pydub.exceptions import CouldntDecodeError
 
 
 class FingerprintRecognizer(BaseRecognizer):
@@ -266,6 +267,8 @@ def _fingerprint_directory(
                 print("All files in directory already fingerprinted")
             else:
                 print("Directory contains 0 files or could not be found")
+                if self.config.fail_on_decode_error:
+                    raise CouldntDecodeError("Directory contains 0 files or could not be found")
             return
 
         if _file_audsegs is not None:

audalign/recognizers/fingerprint/fingerprinter.py

@@ -52,10 +52,12 @@ def _fingerprint_worker(
         except FileNotFoundError:
             print(f'"{file_path}" not found')
             return None, None
-        except (
-            CouldntDecodeError,
-            IndexError,
-        ):  # Pydub throws IndexErrors for some files on Ubuntu (json, txt, others?)
+        except CouldntDecodeError as e:
+            print(f'File "{file_name}" could not be decoded')
+            if config.fail_on_decode_error:
+                raise e
+            return None, None
+        except IndexError:  # Pydub throws IndexErrors for some files on Ubuntu (json, txt, others?)
             print(f'File "{file_name}" could not be decoded')
             return None, None
     elif type(file_path) == tuple:

audalign/recognizers/visrecognize/visrecognize.py

@@ -176,8 +176,10 @@ def _visrecognize_directory(
                 imgB_title=os.path.basename(file_path),
             )
         return single_file_match
-    except CouldntDecodeError:
+    except CouldntDecodeError as e:
         print(f'File "{file_path}" could not be decoded')
+        if config.fail_on_decode_error:
+            raise e
         return {}
 
 

tests/test_recognize.py

@@ -1,6 +1,11 @@
+from pydub.exceptions import CouldntDecodeError
 import audalign as ad
 import os
 import pytest
+from audalign.config.correlation import CorrelationConfig
+
+from audalign.config.fingerprint import FingerprintConfig
+
 try:
     import skimage
 except ImportError:
@@ -94,12 +99,26 @@ def test_fingerprint_directory_not_there_already_done(self):
         fingerprint_recognizer.load_fingerprinted_files("tests/test_fingerprints.json")
         fingerprint_recognizer.fingerprint_directory("test_audio/test_shifts")
 
-    def test_fingerprint_bad_file(self):
+    def test_fingerprint_bad_file_should_fail(self):
         fingerprint_recognizer = ad.FingerprintRecognizer()
         assert len(fingerprint_recognizer.fingerprinted_files) == 0
         assert len(fingerprint_recognizer.file_names) == 0
         assert fingerprint_recognizer.total_fingerprints == 0
 
+        # both should print something and be just fine
+        with pytest.raises(CouldntDecodeError):
+            fingerprint_recognizer.fingerprint_file("filenot_even_there.txt")
+        with pytest.raises(CouldntDecodeError):
+            fingerprint_recognizer.fingerprint_file("requirements.txt")
+
+    def test_fingerprint_bad_file(self):
+        config = FingerprintConfig()
+        config.fail_on_decode_error = False
+        fingerprint_recognizer = ad.FingerprintRecognizer(config)
+        assert len(fingerprint_recognizer.fingerprinted_files) == 0
+        assert len(fingerprint_recognizer.file_names) == 0
+        assert fingerprint_recognizer.total_fingerprints == 0
+
         # both should print something and be just fine
         fingerprint_recognizer.fingerprint_file("filenot_even_there.txt")
         fingerprint_recognizer.fingerprint_file("requirements.txt")
@@ -326,8 +345,19 @@ def test_correcognize_locality_max_lags(self):
         assert min(offset_seconds) < _max_lags
         assert max(offset_seconds) < _max_lags
 
-    def test_correcognize_directory_no_return(self):
+    def test_correcognize_directory_fail_on_decode(self):
         recognizer = ad.CorrelationRecognizer()
+        with pytest.raises(CouldntDecodeError):
+            ad.recognize(
+                test_file,
+                "tests/",
+                recognizer=recognizer,
+            )
+
+    def test_correcognize_directory_no_return(self):
+        config = CorrelationConfig()
+        config.fail_on_decode_error = False
+        recognizer = ad.CorrelationRecognizer(config)
         results = ad.recognize(
             test_file,
             "tests/",