Make it easier to develop without a dev install (#22697)

* Make it easier to develop without a dev install

* Remove ugly hack that doesn't work anyway

CONTRIBUTING.md

@@ -162,14 +162,16 @@ You'll need **[Python 3.7]((https://github.com/huggingface/transformers/blob/mai
    it with `pip uninstall transformers` before reinstalling it in editable
    mode with the `-e` flag.
    
-   Depending on your OS, you may need to install some external libraries as well if the `pip` installation fails.
-   
-   For macOS, you will likely need [MeCab](https://taku910.github.io/mecab/) which can be installed from Homebrew:
-   
+   Depending on your OS, and since the number of optional dependencies of Transformers is growing, you might get a
+   failure with this command. If that's the case make sure to install the Deep Learning framework you are working with
+   (PyTorch, TensorFlow and/or Flax) then do:
+
    ```bash
-   brew install mecab
+   pip install -e ".[quality]"
    ```
 
+   which should be enough for most use cases.
+
 5. Develop the features on your branch.
 
    As you work on your code, you should make sure the test suite

docs/source/en/add_new_model.mdx

@@ -202,7 +202,15 @@ source .env/bin/activate
 pip install -e ".[dev]"
 ```
 
-and return to the parent directory
+Depending on your OS, and since the number of optional dependencies of Transformers is growing, you might get a
+failure with this command. If that's the case make sure to install the Deep Learning framework you are working with
+(PyTorch, TensorFlow and/or Flax) then do:
+
+```bash
+pip install -e ".[quality]"
+```
+
+which should be enough for most use cases. You can then return to the parent directory
 
 ```bash
 cd ..

docs/source/en/add_tensorflow_model.mdx

@@ -119,6 +119,13 @@ source .env/bin/activate
 pip install -e ".[dev]"
 ```
 
+Depending on your OS, and since the number of optional dependencies of Transformers is growing, you might get a
+failure with this command. If that's the case make sure to install TensorFlow then do:
+
+```bash
+pip install -e ".[quality]"
+```
+
 **Note:** You don't need to have CUDA installed. Making the new model work on CPU is sufficient.
 
 4. Create a branch with a descriptive name from your main branch

docs/source/en/pr_checks.mdx

@@ -24,7 +24,7 @@ When you open a pull request on 🤗 Transformers, a fair number of checks will
 
 In this document, we will take a stab at explaining what those various checks are and the reason behind them, as well as how to debug them locally if one of them fails on your PR.
 
-Note that they all require you to have a dev install:
+Note that, ideally, they require you to have a dev install:
 
 ```bash
 pip install transformers[dev]
@@ -36,7 +36,18 @@ or for an editable install:
 pip install -e .[dev]
 ```
 
-inside the Transformers repo.
+inside the Transformers repo. Since the number of optional dependencies of Transformers has grown a lot, it's possible you don't manage to get all of them. If the dev install fails, make sure to install the Deep Learning framework you are working with (PyTorch, TensorFlow and/or Flax) then do
+
+```bash
+pip install transformers[quality]
+```
+
+or for an editable install:
+
+```bash
+pip install -e .[quality]
+```
+
 
 ## Tests
 

templates/adding_a_new_model/README.md

@@ -34,6 +34,14 @@ cd transformers
 pip install -e ".[dev]"
 ```
 
+Depending on your OS, and since the number of optional dependencies of Transformers is growing, you might get a
+failure with this command. If that's the case make sure to install the Deep Learning framework you are working with
+(PyTorch, TensorFlow and/or Flax) then do:
+
+```bash
+pip install -e ".[quality]"
+```
+
 Once the installation is done, you can use the CLI command `add-new-model` to generate your models:
 
 ```shell script
@@ -133,6 +141,14 @@ cd transformers
 pip install -e ".[dev]"
 ```
 
+Depending on your OS, and since the number of optional dependencies of Transformers is growing, you might get a
+failure with this command. If that's the case make sure to install the Deep Learning framework you are working with
+(PyTorch, TensorFlow and/or Flax) then do:
+
+```bash
+pip install -e ".[quality]"
+```
+
 Once the installation is done, you can use the CLI command `add-new-model-like` to generate your models:
 
 ```shell script

utils/check_inits.py

@@ -277,11 +277,20 @@ def check_submodules():
 
     transformers = direct_transformers_import(PATH_TO_TRANSFORMERS)
 
+    import_structure_keys = set(transformers._import_structure.keys())
+    # This contains all the base keys of the _import_structure object defined in the init, but if the user is missing
+    # some optional dependencies, they may not have all of them. Thus we read the init to read all additions and
+    # (potentiall re-) add them.
+    with open(os.path.join(PATH_TO_TRANSFORMERS, "__init__.py"), "r") as f:
+        init_content = f.read()
+    import_structure_keys.update(set(re.findall(r"import_structure\[\"([^\"]*)\"\]", init_content)))
+
     module_not_registered = [
         module
         for module in get_transformers_submodules()
-        if module not in IGNORE_SUBMODULES and module not in transformers._import_structure.keys()
+        if module not in IGNORE_SUBMODULES and module not in import_structure_keys
     ]
+
     if len(module_not_registered) > 0:
         list_of_modules = "\n".join(f"- {module}" for module in module_not_registered)
         raise ValueError(