[tests] relocate core integration tests (#11146)

* relocate core integration tests

* add sys.path context manager

* cleanup

* try

* try2

* fix path

* doc

* style

* add dep

* add 2 more deps

docs/source/main_classes/trainer.rst

@@ -525,7 +525,7 @@ Here is an example of running ``run_translation.py`` under DeepSpeed deploying a
 .. code-block:: bash
 
     deepspeed examples/seq2seq/run_translation.py \
-    --deepspeed examples/tests/deepspeed/ds_config.json \
+    --deepspeed tests/deepspeed/ds_config.json \
     --model_name_or_path t5-small --per_device_train_batch_size 1   \
     --output_dir output_dir --overwrite_output_dir --fp16 \
     --do_train --max_train_samples 500 --num_train_epochs 1 \
@@ -550,7 +550,7 @@ To deploy DeepSpeed with one GPU adjust the :class:`~transformers.Trainer` comma
 .. code-block:: bash
 
     deepspeed --num_gpus=1 examples/seq2seq/run_translation.py \
-    --deepspeed examples/tests/deepspeed/ds_config.json \
+    --deepspeed tests/deepspeed/ds_config.json \
     --model_name_or_path t5-small --per_device_train_batch_size 1   \
     --output_dir output_dir --overwrite_output_dir --fp16 \
     --do_train --max_train_samples 500 --num_train_epochs 1 \

docs/source/testing.rst

-.. 
+..
     Copyright 2020 The HuggingFace Team. All rights reserved.
 
     Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
@@ -388,7 +388,7 @@ For a single or a group of tests via ``pytest`` (after ``pip install pytest-pspe
 
 .. code-block:: bash
 
-   pytest --pspec tests/test_optimization.py 
+   pytest --pspec tests/test_optimization.py
 
 
 
@@ -672,7 +672,7 @@ and it will list:
 
     test_this2.py::test_floor[integer-1-1.0]
     test_this2.py::test_floor[negative--1.5--2.0]
-    test_this2.py::test_floor[large fraction-1.6-1]       
+    test_this2.py::test_floor[large fraction-1.6-1]
 
 So now you can run just the specific test:
 
@@ -795,6 +795,23 @@ leave any data in there.
    otherwise.
 
 
+Temporary sys.path override
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+If you need to temporary override ``sys.path`` to import from another test for example, you can use the
+``ExtendSysPath`` context manager. Example:
+
+
+.. code-block:: python
+
+    import os
+    from transformers.testing_utils import ExtendSysPath
+    bindir = os.path.abspath(os.path.dirname(__file__))
+    with ExtendSysPath(f"{bindir}/.."):
+        from test_trainer import TrainerIntegrationCommon  # noqa
+
+
+
 Skipping tests
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 

setup.py

@@ -19,7 +19,7 @@ To create the package for pypi.
 
 1. Run `make pre-release` (or `make pre-patch` for a patch release) then run `make fix-copies` to fix the index of the
    documentation.
-   
+
 2. Run Tests for Amazon Sagemaker. The documentation is located in `./tests/sagemaker/README.md`, otherwise @philschmid.
 
 3. Unpin specific versions from setup.py that use a git install.
@@ -85,6 +85,7 @@ if stale_egg_info.exists():
 # 1. all dependencies should be listed here with their version requirements if any
 # 2. once modified, run: `make deps_table_update` to update src/transformers/dependency_versions_table.py
 _deps = [
+    "Pillow",
     "black>=20.8b1",
     "cookiecutter==1.7.2",
     "dataclasses",
@@ -102,13 +103,13 @@ _deps = [
     "jax>=0.2.8",
     "jaxlib>=0.1.59",
     "keras2onnx",
+    "nltk",
     "numpy>=1.17",
     "onnxconverter-common",
     "onnxruntime-tools>=1.4.2",
     "onnxruntime>=1.4.0",
     "packaging",
     "parameterized",
-    "Pillow",
     "protobuf",
     "psutil",
     "pydantic",
@@ -119,15 +120,18 @@ _deps = [
     "recommonmark",
     "regex!=2019.12.17",
     "requests",
+    "rouge-score",
+    "sacrebleu>=1.4.12",
     "sacremoses",
+    "sagemaker>=2.31.0",
     "scikit-learn",
     "sentencepiece==0.1.91",
     "soundfile",
     "sphinx-copybutton",
     "sphinx-markdown-tables",
     "sphinx-rtd-theme==0.4.3",  # sphinx-rtd-theme==0.5.0 introduced big changes in the style.
-    "sphinxext-opengraph==0.4.1",
     "sphinx==3.2.1",
+    "sphinxext-opengraph==0.4.1",
     "starlette",
     "tensorflow-cpu>=2.3",
     "tensorflow>=2.3",
@@ -139,7 +143,6 @@ _deps = [
     "unidic>=1.0.2",
     "unidic_lite>=1.0.7",
     "uvicorn",
-    "sagemaker>=2.31.0",
 ]
 
 
@@ -238,7 +241,7 @@ extras["vision"] = deps_list("Pillow")
 extras["sentencepiece"] = deps_list("sentencepiece", "protobuf")
 extras["testing"] = (
     deps_list(
-        "pytest", "pytest-xdist", "timeout-decorator", "parameterized", "psutil", "datasets", "pytest-sugar", "black"
+        "pytest", "pytest-xdist", "timeout-decorator", "parameterized", "psutil", "datasets", "pytest-sugar", "black", "sacrebleu", "rouge-score", "nltk"
     )
     + extras["retrieval"]
     + extras["modelcreation"]

src/transformers/dependency_versions_table.py

@@ -2,6 +2,7 @@
 # 1. modify the `_deps` dict in setup.py
 # 2. run `make deps_table_update``
 deps = {
+    "Pillow": "Pillow",
     "black": "black>=20.8b1",
     "cookiecutter": "cookiecutter==1.7.2",
     "dataclasses": "dataclasses",
@@ -19,13 +20,13 @@ deps = {
     "jax": "jax>=0.2.8",
     "jaxlib": "jaxlib>=0.1.59",
     "keras2onnx": "keras2onnx",
+    "nltk": "nltk",
     "numpy": "numpy>=1.17",
     "onnxconverter-common": "onnxconverter-common",
     "onnxruntime-tools": "onnxruntime-tools>=1.4.2",
     "onnxruntime": "onnxruntime>=1.4.0",
     "packaging": "packaging",
     "parameterized": "parameterized",
-    "Pillow": "Pillow",
     "protobuf": "protobuf",
     "psutil": "psutil",
     "pydantic": "pydantic",
@@ -36,15 +37,18 @@ deps = {
     "recommonmark": "recommonmark",
     "regex": "regex!=2019.12.17",
     "requests": "requests",
+    "rouge-score": "rouge-score",
+    "sacrebleu": "sacrebleu>=1.4.12",
     "sacremoses": "sacremoses",
+    "sagemaker": "sagemaker>=2.31.0",
     "scikit-learn": "scikit-learn",
     "sentencepiece": "sentencepiece==0.1.91",
     "soundfile": "soundfile",
     "sphinx-copybutton": "sphinx-copybutton",
     "sphinx-markdown-tables": "sphinx-markdown-tables",
     "sphinx-rtd-theme": "sphinx-rtd-theme==0.4.3",
-    "sphinxext-opengraph": "sphinxext-opengraph==0.4.1",
     "sphinx": "sphinx==3.2.1",
+    "sphinxext-opengraph": "sphinxext-opengraph==0.4.1",
     "starlette": "starlette",
     "tensorflow-cpu": "tensorflow-cpu>=2.3",
     "tensorflow": "tensorflow>=2.3",
@@ -56,5 +60,4 @@ deps = {
     "unidic": "unidic>=1.0.2",
     "unidic_lite": "unidic_lite>=1.0.7",
     "uvicorn": "uvicorn",
-    "sagemaker": "sagemaker>=2.31.0",
 }

src/transformers/testing_utils.py

@@ -24,6 +24,7 @@ import unittest
 from distutils.util import strtobool
 from io import StringIO
 from pathlib import Path
+from typing import Iterator, Union
 
 from .file_utils import (
     is_datasets_available,
@@ -621,6 +622,27 @@ class CaptureLogger:
         return f"captured: {self.out}\n"
 
 
+@contextlib.contextmanager
+# adapted from https://stackoverflow.com/a/64789046/9201239
+def ExtendSysPath(path: Union[str, os.PathLike]) -> Iterator[None]:
+    """
+    Temporary add given path to `sys.path`.
+
+    Usage ::
+
+       with ExtendSysPath('/path/to/dir'):
+           mymodule = importlib.import_module('mymodule')
+
+    """
+
+    path = os.fspath(path)
+    try:
+        sys.path.insert(0, path)
+        yield
+    finally:
+        sys.path.remove(path)
+
+
 class TestCasePlus(unittest.TestCase):
     """
     This class extends `unittest.TestCase` with additional features.

examples/tests/deepspeed/test_deepspeed.py → tests/deepspeed/test_deepspeed.py

@@ -16,16 +16,16 @@ import dataclasses
 import io
 import json
 import os
-import sys
 import unittest
 from copy import deepcopy
 
 from parameterized import parameterized
-from transformers import TrainingArguments
+from transformers import TrainingArguments, is_torch_available
 from transformers.file_utils import WEIGHTS_NAME
 from transformers.integrations import is_deepspeed_available
 from transformers.testing_utils import (
     CaptureLogger,
+    ExtendSysPath,
     TestCasePlus,
     execute_subprocess_async,
     get_gpu_count,
@@ -38,8 +38,11 @@ from transformers.trainer_utils import set_seed
 
 
 bindir = os.path.abspath(os.path.dirname(__file__))
-sys.path.append(f"{bindir}/../../../tests")
-from test_trainer import TrainerIntegrationCommon, get_regression_trainer  # noqa
+with ExtendSysPath(f"{bindir}/.."):
+    from test_trainer import TrainerIntegrationCommon  # noqa
+
+    if is_torch_available():
+        from test_trainer import get_regression_trainer  # noqa
 
 
 set_seed(42)

examples/tests/trainer/test_trainer_ext.py → tests/extended/test_trainer_ext.py

@@ -21,6 +21,7 @@ from unittest.mock import patch
 from transformers.file_utils import is_apex_available
 from transformers.integrations import is_fairscale_available
 from transformers.testing_utils import (
+    ExtendSysPath,
     TestCasePlus,
     execute_subprocess_async,
     get_gpu_count,
@@ -34,8 +35,8 @@ from transformers.trainer_utils import set_seed
 
 
 bindir = os.path.abspath(os.path.dirname(__file__))
-sys.path.append(f"{bindir}/../../seq2seq")
-from run_translation import main  # noqa
+with ExtendSysPath(f"{bindir}/../../examples/seq2seq"):
+    from run_translation import main  # noqa
 
 
 set_seed(42)