Change how "additional_special_tokens" argument in the ".from_pretrained" method of the tokenizer is taken into account (#13056)

* add test

* add change in PretrainedTokenizerBase

* change Luke

* deactivate

* add the possibility to add additional special tokens for M2M100

* format

* add special test for canine

* proposed changes for mbart

* proposed changes for mbart50

* proposed changes for byt5

* proposed changes for canine

* proposed changes for t5

* test fast and slow

* remove comment

* remove comment

* add fast version for all tests

* replace break by continue

* add more comments

* add check to avoid duplicates

* remove comment

* format

* proposed change for wave2vec2

* reverse changes mbart

* uncomment

* format

tests/test_tokenization_byt5.py

@@ -13,11 +13,13 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
+import json
+import os
 import shutil
 import tempfile
 import unittest
 
-from transformers import BatchEncoding, ByT5Tokenizer
+from transformers import AddedToken, BatchEncoding, ByT5Tokenizer
 from transformers.file_utils import cached_property, is_tf_available, is_torch_available
 
 from .test_tokenization_common import TokenizerTesterMixin
@@ -161,6 +163,72 @@ class ByT5TokenizationTest(TokenizerTesterMixin, unittest.TestCase):
 
                 shutil.rmtree(tmpdirname)
 
+    # There is a conflict between the default value of extra_ids and adding a new special token through additional_special_tokens
+    # We need to add the extra_ids in the list of the arg additional_special_tokens
+    def test_special_tokens_initialization_with_non_empty_additional_special_tokens(self):
+        tokenizer_list = []
+        if self.test_slow_tokenizer:
+            tokenizer_list.append((self.tokenizer_class, self.get_tokenizer()))
+
+        if self.test_rust_tokenizer:
+            tokenizer_list.append((self.rust_tokenizer_class, self.get_rust_tokenizer()))
+
+        for tokenizer_class, tokenizer_utils in tokenizer_list:
+            with tempfile.TemporaryDirectory() as tmp_dir:
+                tokenizer_utils.save_pretrained(tmp_dir)
+
+                with open(os.path.join(tmp_dir, "special_tokens_map.json"), encoding="utf-8") as json_file:
+                    special_tokens_map = json.load(json_file)
+
+                with open(os.path.join(tmp_dir, "tokenizer_config.json"), encoding="utf-8") as json_file:
+                    tokenizer_config = json.load(json_file)
+
+                added_tokens_extra_ids = [f"<extra_id_{i}>" for i in range(125)]
+
+                special_tokens_map["additional_special_tokens"] = added_tokens_extra_ids + [
+                    "an_additional_special_token"
+                ]
+                tokenizer_config["additional_special_tokens"] = added_tokens_extra_ids + [
+                    "an_additional_special_token"
+                ]
+
+                with open(os.path.join(tmp_dir, "special_tokens_map.json"), "w", encoding="utf-8") as outfile:
+                    json.dump(special_tokens_map, outfile)
+                with open(os.path.join(tmp_dir, "tokenizer_config.json"), "w", encoding="utf-8") as outfile:
+                    json.dump(tokenizer_config, outfile)
+
+                # the following checks allow us to verify that our test works as expected, i.e. that the tokenizer takes
+                # into account the new value of additional_special_tokens given in the "tokenizer_config.json" and
+                # "special_tokens_map.json" files
+                tokenizer_without_change_in_init = tokenizer_class.from_pretrained(
+                    tmp_dir,
+                )
+                self.assertIn(
+                    "an_additional_special_token", tokenizer_without_change_in_init.additional_special_tokens
+                )
+                # self.assertIn("an_additional_special_token",tokenizer_without_change_in_init.get_vocab()) # ByT5Tokenization no vocab
+                self.assertEqual(
+                    ["an_additional_special_token"],
+                    tokenizer_without_change_in_init.convert_ids_to_tokens(
+                        tokenizer_without_change_in_init.convert_tokens_to_ids(["an_additional_special_token"])
+                    ),
+                )
+
+                # Now we test that we can change the value of additional_special_tokens in the from_pretrained
+                new_added_tokens = added_tokens_extra_ids + [AddedToken("a_new_additional_special_token", lstrip=True)]
+                tokenizer = tokenizer_class.from_pretrained(
+                    tmp_dir,
+                    additional_special_tokens=new_added_tokens,
+                )
+
+                self.assertIn("a_new_additional_special_token", tokenizer.additional_special_tokens)
+                self.assertEqual(
+                    ["a_new_additional_special_token"],
+                    tokenizer.convert_ids_to_tokens(
+                        tokenizer.convert_tokens_to_ids(["a_new_additional_special_token"])
+                    ),
+                )
+
     # tokenizer can be instantiated without any pretrained files, so no need for pretrained tokenizer list
     def test_pretrained_model_lists(self):
         pass