ix #77

makcedward · Jan 8, 2020 · 64aa4c5 · 64aa4c5
1 parent 74b9477
commit 64aa4c5
Show file tree

Hide file tree

Showing 2 changed files with 21 additions and 15 deletions.
diff --git a/nlpaug/augmenter/char/random.py b/nlpaug/augmenter/char/random.py
@@ -127,7 +127,7 @@ def swap(self, data):
         results = []
         tokens = self.tokenizer(data)
         aug_word_idxes = self._get_aug_idxes(tokens, self.aug_word_min, self.aug_word_max, self.aug_word_p, Method.WORD)
-        if aug_word_idxes is None:
+        if aug_word_idxes is None or len(aug_word_idxes) < 1:
             return data
 
         for token_i, token in enumerate(tokens):
@@ -137,17 +137,17 @@ def swap(self, data):
 
             result = ''
             chars = self.token2char(token)
-            original_chars = chars.copy()
 
             aug_char_idxes = self._get_aug_idxes(chars, self.aug_char_min, self.aug_char_max, self.aug_char_p,
                                                  Method.CHAR)
-            if aug_char_idxes is None:
+            if aug_char_idxes is None or len(aug_char_idxes) < 1:
                 results.append(token)
                 continue
 
             for char_i in aug_char_idxes:
                 swap_position = self._get_swap_position(char_i, len(chars)-1, mode=self.swap_mode)
                 is_original_upper, is_swap_upper = chars[char_i].isupper(), chars[swap_position].isupper()
+                original_chars = chars.copy()
                 chars[char_i], chars[swap_position] = original_chars[swap_position], original_chars[char_i]
 
                 # Swap case
@@ -171,7 +171,7 @@ def delete(self, data):
         results = []
         tokens = self.tokenizer(data)
         aug_word_idxes = self._get_aug_idxes(tokens, self.aug_word_min, self.aug_word_max, self.aug_word_p, Method.WORD)
-        if aug_word_idxes is None:
+        if aug_word_idxes is None or len(aug_word_idxes) < 1:
             return data
 
         for token_i, token in enumerate(tokens):
@@ -182,7 +182,7 @@ def delete(self, data):
             chars = self.token2char(token)
             aug_char_idxes = self._get_aug_idxes(chars, self.aug_char_min, self.aug_char_max, self.aug_char_p,
                                                  Method.CHAR)
-            if aug_char_idxes is None:
+            if aug_char_idxes is None or len(aug_char_idxes) < 1:
                 results.append(token)
                 continue
 

diff --git a/test/augmenter/char/test_random_char.py b/test/augmenter/char/test_random_char.py
@@ -1,13 +1,12 @@
 import unittest
 
 from nlpaug.augmenter.char.random import RandomCharAug
-from nlpaug.util import Action
 
 
 class TestRandomCharReplaceAug(unittest.TestCase):
     def test_insert_single_word(self):
         texts = ['Zoology', 'roku123456']
-        aug = RandomCharAug(action=Action.INSERT, min_char=1)
+        aug = RandomCharAug(action='insert', min_char=1)
         for text in texts:
             augmented_text = aug.augment(text)
             self.assertNotEqual(text, augmented_text)
@@ -17,7 +16,7 @@ def test_insert_single_word(self):
 
     def test_insert_multi_words(self):
         texts = ['The quick brown fox jumps over the lazy dog']
-        aug = RandomCharAug(action=Action.INSERT, min_char=1)
+        aug = RandomCharAug(action='insert', min_char=1)
         for text in texts:
             augmented_cnt = 0
             augmented_text = aug.augment(text)
@@ -37,7 +36,7 @@ def test_insert_multi_words(self):
 
     def test_substitute_single_word(self):
         texts = ['Zoology', 'roku123456']
-        aug = RandomCharAug(action=Action.SUBSTITUTE, min_char=1)
+        aug = RandomCharAug(action='substitute', min_char=1)
         for text in texts:
             augmented_text = aug.augment(text)
             self.assertNotEqual(text, augmented_text)
@@ -46,7 +45,7 @@ def test_substitute_single_word(self):
 
     def test_substitute_multi_words(self):
         texts = ['The quick brown fox jumps over the lazy dog']
-        aug = RandomCharAug(action=Action.SUBSTITUTE, min_char=1)
+        aug = RandomCharAug(action='substitute', min_char=1)
         for text in texts:
             augmented_cnt = 0
             augmented_text = aug.augment(text)
@@ -64,11 +63,19 @@ def test_substitute_multi_words(self):
         self.assertTrue(len(texts) > 0)
 
     def test_swap(self):
-        texts = ['The quick brown fox jumps over the lazy dog']
-        aug = RandomCharAug(action=Action.SWAP, min_char=1)
+        texts = [
+            'The quick brown fox jumps over the lazy dog',
+            'testing'
+        ]
+        aug = RandomCharAug(action="swap", min_char=1)
         for text in texts:
             augmented_cnt = 0
-            augmented_text = aug.augment(text)
+
+            augmented_text = text
+
+            # https://github.com/makcedward/nlpaug/issues/77
+            for i in range(10):
+                augmented_text = aug.augment(augmented_text)
 
             tokens = aug.tokenizer(text)
             augmented_tokens = aug.tokenizer(augmented_text)
@@ -77,14 +84,13 @@ def test_swap(self):
                 if token != augmented_token:
                     augmented_cnt += 1
 
-            self.assertLess(augmented_cnt, len(tokens))
             self.assertNotEqual(text, augmented_text)
 
         self.assertTrue(len(texts) > 0)
 
     def test_delete(self):
         tokens = ['Zoology', 'roku123456']
-        aug = RandomCharAug(action=Action.DELETE, min_char=1)
+        aug = RandomCharAug(action='delete', min_char=1)
         for t in tokens:
             augmented_text = aug.augment(t)
             self.assertNotEqual(t, augmented_text)