From db9a7e9d3dbd1b595f004597a0502cce0a96135a Mon Sep 17 00:00:00 2001
From: Yih-Dar <2521628+ydshieh@users.noreply.github.com>
Date: Fri, 19 Jan 2024 09:59:14 +0000
Subject: [PATCH] Don't save `processor_config.json` if a processor has no
 extra attribute  (#28584)

* not save if empty

* fix

* fix

* fix

* fix

* fix

---------

Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>
---
 src/transformers/processing_utils.py     |  9 +++++--
 tests/models/auto/test_processor_auto.py | 32 +++++++++++++++---------
 tests/test_processing_common.py          |  9 ++++---
 3 files changed, 32 insertions(+), 18 deletions(-)

diff --git a/src/transformers/processing_utils.py b/src/transformers/processing_utils.py
index 01c824f92c..f727f308ac 100644
--- a/src/transformers/processing_utils.py
+++ b/src/transformers/processing_utils.py
@@ -234,8 +234,11 @@ class ProcessorMixin(PushToHubMixin):
         # If we save using the predefined names, we can load using `from_pretrained`
         output_processor_file = os.path.join(save_directory, PROCESSOR_NAME)
 
-        self.to_json_file(output_processor_file)
-        logger.info(f"processor saved in {output_processor_file}")
+        # For now, let's not save to `processor_config.json` if the processor doesn't have extra attributes and
+        # `auto_map` is not specified.
+        if set(self.to_dict().keys()) != {"processor_class"}:
+            self.to_json_file(output_processor_file)
+            logger.info(f"processor saved in {output_processor_file}")
 
         if push_to_hub:
             self._upload_modified_files(
@@ -246,6 +249,8 @@ class ProcessorMixin(PushToHubMixin):
                 token=kwargs.get("token"),
             )
 
+        if set(self.to_dict().keys()) == {"processor_class"}:
+            return []
         return [output_processor_file]
 
     @classmethod
diff --git a/tests/models/auto/test_processor_auto.py b/tests/models/auto/test_processor_auto.py
index c22013234f..6cab1cbe81 100644
--- a/tests/models/auto/test_processor_auto.py
+++ b/tests/models/auto/test_processor_auto.py
@@ -101,6 +101,12 @@ class AutoFeatureExtractorTest(unittest.TestCase):
             # save in new folder
             processor.save_pretrained(tmpdirname)
 
+            if not os.path.isfile(os.path.join(tmpdirname, PROCESSOR_NAME)):
+                # create one manually in order to perform this test's objective
+                config_dict = {"processor_class": "Wav2Vec2Processor"}
+                with open(os.path.join(tmpdirname, PROCESSOR_NAME), "w") as fp:
+                    json.dump(config_dict, fp)
+
             # drop `processor_class` in tokenizer config
             with open(os.path.join(tmpdirname, TOKENIZER_CONFIG_FILE), "r") as f:
                 config_dict = json.load(f)
@@ -123,13 +129,14 @@ class AutoFeatureExtractorTest(unittest.TestCase):
             # save in new folder
             processor.save_pretrained(tmpdirname)
 
-            # drop `processor_class` in processor
-            with open(os.path.join(tmpdirname, PROCESSOR_NAME), "r") as f:
-                config_dict = json.load(f)
-                config_dict.pop("processor_class")
+            if os.path.isfile(os.path.join(tmpdirname, PROCESSOR_NAME)):
+                # drop `processor_class` in processor
+                with open(os.path.join(tmpdirname, PROCESSOR_NAME), "r") as f:
+                    config_dict = json.load(f)
+                    config_dict.pop("processor_class")
 
-            with open(os.path.join(tmpdirname, PROCESSOR_NAME), "w") as f:
-                f.write(json.dumps(config_dict))
+                with open(os.path.join(tmpdirname, PROCESSOR_NAME), "w") as f:
+                    f.write(json.dumps(config_dict))
 
             # drop `processor_class` in tokenizer
             with open(os.path.join(tmpdirname, TOKENIZER_CONFIG_FILE), "r") as f:
@@ -153,13 +160,14 @@ class AutoFeatureExtractorTest(unittest.TestCase):
             # save in new folder
             processor.save_pretrained(tmpdirname)
 
-            # drop `processor_class` in processor
-            with open(os.path.join(tmpdirname, PROCESSOR_NAME), "r") as f:
-                config_dict = json.load(f)
-                config_dict.pop("processor_class")
+            if os.path.isfile(os.path.join(tmpdirname, PROCESSOR_NAME)):
+                # drop `processor_class` in processor
+                with open(os.path.join(tmpdirname, PROCESSOR_NAME), "r") as f:
+                    config_dict = json.load(f)
+                    config_dict.pop("processor_class")
 
-            with open(os.path.join(tmpdirname, PROCESSOR_NAME), "w") as f:
-                f.write(json.dumps(config_dict))
+                with open(os.path.join(tmpdirname, PROCESSOR_NAME), "w") as f:
+                    f.write(json.dumps(config_dict))
 
             # drop `processor_class` in feature extractor
             with open(os.path.join(tmpdirname, FEATURE_EXTRACTOR_NAME), "r") as f:
diff --git a/tests/test_processing_common.py b/tests/test_processing_common.py
index 1ab215e34c..402e6a7351 100644
--- a/tests/test_processing_common.py
+++ b/tests/test_processing_common.py
@@ -75,11 +75,12 @@ class ProcessorTesterMixin:
         processor_first = self.get_processor()
 
         with tempfile.TemporaryDirectory() as tmpdirname:
-            saved_file = processor_first.save_pretrained(tmpdirname)[0]
-            check_json_file_has_correct_format(saved_file)
-            processor_second = self.processor_class.from_pretrained(tmpdirname)
+            saved_files = processor_first.save_pretrained(tmpdirname)
+            if len(saved_files) > 0:
+                check_json_file_has_correct_format(saved_files[0])
+                processor_second = self.processor_class.from_pretrained(tmpdirname)
 
-        self.assertEqual(processor_second.to_dict(), processor_first.to_dict())
+                self.assertEqual(processor_second.to_dict(), processor_first.to_dict())
 
 
 class MyProcessor(ProcessorMixin):