Add ImportAnnotations configuration option

[teobouvard]

📒 Description

This PR adds a new element ImportAnnotations to the Output section of the configuration file. The goal of this configuration option is to allow schemas introducing collisions between a class member name and its type in the generated bindings to be imported without errors.

Resolves #646

🔗 What I've Done

As discussed in the second part of #646, name collisions originally raised a TypeError because type annotations could not be resolved. However, PEP 563 discusses a way to postpone the evaluation of type hints in such a way that name collisions are acceptable. This behavior is optional in Python 3.7.0b1, and will be mandatory in Python 3.10 (see Future statement definitions).
This configuration option is disabled by default, and enabling it will add the following import at the top of all generated modules.

from __future__ import annotations

This is done by simply prepending the import statement to the default_imports filter used in the module Jinja template.

💬 Comments

Thank you for this library and the effort you're putting towards code quality. Let me know if more tests should be added for this feature.

🛫 Checklist

• Updated docs
• Added unit-tests
• Sample tests pass
• W3C tests pass

[codecov]

Codecov Report

Merging #653 (6abed7b) into master (16b12b6) will not change coverage.
The diff coverage is 100.00%.

@@            Coverage Diff            @@
##            master      #653   +/-   ##
=========================================
  Coverage   100.00%   100.00%           
=========================================
  Files           94        94           
  Lines         8214      8224   +10     
  Branches      1583      1585    +2     
=========================================
+ Hits          8214      8224   +10     

Impacted Files	Coverage Δ
xsdata/formats/dataclass/filters.py	100.00% <100.00%> (ø)
xsdata/models/config.py	100.00% <100.00%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 16b12b6...6abed7b. Read the comment docs.

[teobouvard]

So annotations is not part of __future__ in Python 3.6. We could add an include guard checking the current Python version but this feature would be broken on this version, which is not ideal. What do you think ?

[tefra]

Awesome job, @teobouvard

Can we please change the name of the new option and avoid adding a new project settings file .xsdata.xml for the tests and maybe update the docstring because it will appear in the cli,

[tefra]

All GeneratorOutput options appear in the cli automatically, no need to add a new config file.

❯ xsdata generate --help
...
...
...

Options:
  -r, --recursive                 Search files recursively in the source
                                  directory
  -c, --config TEXT               Project configuration
  -pp, --print                    Print output
  -p, --package TEXT              Target package, default: generated
  -o, --output [dataclasses]      Output format name, default: dataclasses
  --repr / --no-repr              Generate __repr__ method, default: true
  --eq / --no-eq                  Generate __eq__ method, default: true
...
...
...
...
  --import-annotations / --no-import-annotations
                                  Import annotations from :obj:`__future__` in
                                  generated modules, default: false
  --help                          Show this message and exit.

[tefra]

And fix the help text as well, it comes fromt the docstring

[teobouvard]

Is it possible to document it in such a way that the HTML docs have the link for the __future__ module, but the CLI help is only text ?

[teobouvard]

Maybe there is no need to have implementations details in the docstring. I think the following should be sufficient.

:param postponed_annotations: Enable postponed evaluation of annotations, default: false

[teobouvard]

All GeneratorOutput options appear in the cli automatically, no need to add a new config file.

That's awesome ! However, I think I still need the configuration file to set the namecase, which should be originalCase for the purpose of the test.

[tefra]

So annotations is not part of __future__ in Python 3.6. We could add an include guard checking the current Python version but this feature would be broken on this version, which is not ideal. What do you think ?

Check the OutputFormat class in the config, maybe we could add something similar there for the GeneratorOutput

xsdata/xsdata/models/config.py

Lines 177 to 192 in bdae3a3

	def __post_init__(self):
	self.validate()
	def validate(self):
	if self.order and not self.eq:
	raise GeneratorConfigError("eq must be true if order is true")
	if self.value == "dataclasses" and sys.version_info < (3, 10):
	if self.slots:
	self.slots = False
	warnings.warn(
	"slots requires python >= 3.10, reverting...",
	CodeGenerationWarning,
	)
	if self.kw_only:

[sonarqubecloud]

Kudos, SonarCloud Quality Gate passed!   

0 Bugs
0 Vulnerabilities
0 Security Hotspots
0 Code Smells

No Coverage information
0.0% Duplication

[tefra]

Awesome job @teobouvard thanks for contributing!