!3303 fix validator that checks the vocab_size is positive in vocab.from_file

Merge pull request !3303 from ZiruiWu/fix_vocab_validator

mindspore/dataset/text/validators.py

@@ -67,7 +67,7 @@ def check_from_file(method):
             check_unique_list_of_words(special_tokens, "special_tokens")
         type_check_list([file_path, delimiter], (str,), ["file_path", "delimiter"])
         if vocab_size is not None:
-            check_value(vocab_size, (-1, INT32_MAX), "vocab_size")
+            check_positive(vocab_size, "vocab_size")
         type_check(special_first, (bool,), special_first)
 
         return method(self, *args, **kwargs)

tests/ut/python/dataset/test_vocab.py

@@ -133,7 +133,9 @@ def test_from_file():
     assert test_config("w1 w2 w3 s1 s2 s3", 3, ["s1", "s2", "s3"], False) == [0, 1, 2, 3, 4, 5]
     # text exception special_words contains duplicate words
     assert "special_tokens contains duplicate" in test_config("w1", None, ["s1", "s1"], True)
-
+    # test exception when vocab_size is negative
+    assert "Input vocab_size must be greater than 0" in test_config("w1 w2", 0, [], True)
+    assert "Input vocab_size must be greater than 0" in test_config("w1 w2", -1, [], True)
 
 if __name__ == '__main__':
     test_from_dict_exception()