reformat

examples/attack/attack_keras_parallel.py

@@ -70,7 +70,6 @@ def __init__(self, model):
         self.model = model
 
     def __call__(self, text_input_list):
-
         x_transform = []
         for i, review in enumerate(text_input_list):
             tokens = [x.strip(",") for x in review.split()]

textattack/attack_recipes/morpheus_tan_2020.py

@@ -27,7 +27,6 @@ class MorpheusTan2020(AttackRecipe):
 
     @staticmethod
     def build(model_wrapper):
-
         #
         # Goal is to minimize BLEU score between the model output given for the
         # perturbed input sequence and the reference translation

textattack/attack_recipes/seq2sick_cheng_2018_blackbox.py

@@ -31,7 +31,6 @@ class Seq2SickCheng2018BlackBox(AttackRecipe):
 
     @staticmethod
     def build(model_wrapper, goal_function="non_overlapping"):
-
         #
         # Goal is non-overlapping output.
         #

textattack/commands/augment_command.py

@@ -32,7 +32,6 @@ def run(self, args):
 
         args = textattack.AugmenterArgs(**vars(args))
         if args.interactive:
-
             print("\nRunning in interactive mode...\n")
             augmenter = eval(AUGMENTATION_RECIPE_NAMES[args.recipe])(
                 pct_words_to_swap=args.pct_words_to_swap,

textattack/commands/eval_model_command.py

@@ -56,7 +56,7 @@ def test_model_on_dataset(self, args):
         while i < min(args.num_examples, len(dataset)):
             dataset_batch = dataset[i : min(args.num_examples, i + args.batch_size)]
             batch_inputs = []
-            for (text_input, ground_truth_output) in dataset_batch:
+            for text_input, ground_truth_output in dataset_batch:
                 attacked_text = textattack.shared.AttackedText(text_input)
                 batch_inputs.append(attacked_text.tokenizer_input)
                 ground_truth_outputs.append(ground_truth_output)

textattack/constraints/overlap/max_words_perturbed.py

@@ -38,7 +38,6 @@ def __init__(
         self.max_percent = max_percent
 
     def _check_constraint(self, transformed_text, reference_text):
-
         num_words_diff = len(transformed_text.all_words_diff(reference_text))
         if self.max_percent:
             min_num_words = min(len(transformed_text.words), len(reference_text.words))

textattack/constraints/semantics/sentence_encoders/universal_sentence_encoder/multilingual_universal_sentence_encoder.py

@@ -31,7 +31,7 @@ def __init__(self, threshold=0.8, large=False, metric="angular", **kwargs):
         try:
             self.model = hub.load(self._tfhub_url)
         except Exception as e:
-            print('Error loading model from tfhub, trying mirror url')
+            print("Error loading model from tfhub, trying mirror url")
             self.model = hub.load(self.mirror_tfhub_url)
 
     def encode(self, sentences):
@@ -47,5 +47,5 @@ def __setstate__(self, state):
         try:
             self.model = hub.load(self._tfhub_url)
         except Exception as e:
-            print('Error loading model from tfhub, trying mirror url')
+            print("Error loading model from tfhub, trying mirror url")
             self.model = hub.load(self.mirror_tfhub_url)

textattack/constraints/semantics/sentence_encoders/universal_sentence_encoder/universal_sentence_encoder.py

@@ -35,7 +35,7 @@ def encode(self, sentences):
             try:
                 self.model = hub.load(self._tfhub_url)
             except Exception as e:
-                print('Error loading model from tfhub, trying mirror url')
+                print("Error loading model from tfhub, trying mirror url")
                 self.model = hub.load(self.mirror_tfhub_url)
         return self.model(sentences).numpy()
 
@@ -49,5 +49,5 @@ def __setstate__(self, state):
         try:
             self.model = hub.load(self._tfhub_url)
         except Exception as e:
-            print('Error loading model from tfhub, trying mirror url')
+            print("Error loading model from tfhub, trying mirror url")
             self.model = hub.load(self.mirror_tfhub_url)

textattack/goal_function_results/classification_goal_function_result.py

@@ -26,7 +26,6 @@ def __init__(
         num_queries,
         ground_truth_output,
     ):
-
         super().__init__(
             attacked_text,
             raw_output,

textattack/goal_function_results/text_to_text_goal_function_result.py

@@ -23,7 +23,6 @@ def __init__(
         num_queries,
         ground_truth_output,
     ):
-
         super().__init__(
             attacked_text,
             raw_output,

textattack/loggers/weights_and_biases_logger.py

@@ -13,7 +13,6 @@ class WeightsAndBiasesLogger(Logger):
     """Logs attack results to Weights & Biases."""
 
     def __init__(self, **kwargs):
-
         global wandb
         wandb = LazyLoader("wandb", globals(), "wandb")
 

textattack/metrics/quality_metrics/perplexity.py

@@ -94,7 +94,6 @@ def calculate(self, results):
         return self.all_metrics
 
     def calc_ppl(self, texts):
-
         with torch.no_grad():
             text = " ".join(texts)
             eval_loss = []

textattack/search_methods/greedy_word_swap_wir.py

@@ -65,7 +65,6 @@ def _get_index_order(self, initial_text):
             # compute the largest change in score we can find by swapping each word
             delta_ps = []
             for idx in indices_to_order:
-
                 # Exit Loop when search_over is True - but we need to make sure delta_ps
                 # is the same size as softmax_saliency_scores
                 if search_over:

textattack/shared/attacked_text.py

@@ -259,6 +259,7 @@ def ith_word_diff(self, other_attacked_text: AttackedText, i: int) -> bool:
 
     def words_diff_num(self, other_attacked_text: AttackedText) -> int:
         """The number of words different between two AttackedText objects."""
+
         # using edit distance to calculate words diff num
         def generate_tokens(words):
             result = {}

textattack/shared/validators.py

@@ -24,7 +24,10 @@
         r"^textattack.models.helpers.word_cnn_for_classification.*",
         r"^transformers.modeling_\w*\.\w*ForSequenceClassification$",
     ],
-    (NonOverlappingOutput, MinimizeBleu,): [
+    (
+        NonOverlappingOutput,
+        MinimizeBleu,
+    ): [
         r"^textattack.models.helpers.t5_for_text_to_text.*",
     ],
 }

textattack/trainer.py

@@ -407,7 +407,6 @@ def collate_fn(data):
             is_adv_sample = []
             for item in data:
                 if "_example_type" in item[0].keys():
-
                     # Get example type value from OrderedDict and remove it
 
                     adv = item[0].pop("_example_type")

textattack/training_args.py

@@ -547,7 +547,6 @@ def _create_dataset_from_args(cls, args):
             train_dataset.output_column == "label"
             and eval_dataset.output_column == "label"
         ):
-
             train_dataset_labels = train_dataset._dataset["label"]
 
             eval_dataset_labels = eval_dataset._dataset["label"]

textattack/transformations/word_swaps/word_swap_change_name.py

@@ -64,7 +64,6 @@ def _get_transformations(self, current_text, indices_to_modify):
         return transformed_texts
 
     def _get_replacement_words(self, word, word_part_of_speech):
-
         replacement_words = []
         tag = word_part_of_speech
         if (

textattack/transformations/word_swaps/word_swap_change_number.py

@@ -70,7 +70,7 @@ def _get_transformations(self, current_text, indices_to_modify):
 
         # replace original numbers with new numbers
         transformed_texts = []
-        for (idx, word) in num_words:
+        for idx, word in num_words:
             replacement_words = self._get_new_number(word)
             for r in replacement_words:
                 if r == word: