refactor: Rewrite pyRevit configurations

[dosymep]

• added config abstraction
• added config overrides (need testing and design the logic of work)
• replace ini lib to ini-parser
• added json configuration
• added yaml configuration (doesn't work I need to figure out the parser library)

I think in general a start has been made, it remains to rewrite the logic of the config in python

[dosymep]

@jmcouffin @sanzoghenzo how to trigger github action?

[jmcouffin]

@jmcouffin @sanzoghenzo how to trigger github action?

[jmcouffin]

I just launched it.
15min+ to go

I won't be able to review just now, best I can do is install the wip that will be created

[sanzoghenzo]

Awesome job @dosymep !

I didn't run the code, only skimmed through, and only found some grammar to fix (ShouldThrows -> ShouldThrow).

Do you think it would be worth creating separate repo(s) and nuget package(s) for this feature (just to start the transition to a better code structure)?

[dosymep]

creating separate repo(s) and nuget package(s)

it won't work, we have dependencies on this repo, at the moment I still don't know how to do this :(

[jmcouffin]

Should we keep this one opened? @dosymep

[dosymep]

Should we keep this one opened? @dosymep

Let it be, maybe I'll continue someday, or someone else will

[jmcouffin]

@OnePowerUser88 ☝️

[OnePowerUser88]

I've been testing the branch locally with Revit 2026 (build + smoke test, then fixing runtime issues so it runs). Here's a short summary and the list of changes I have made on top of the branch to get everything working. Sharing for discussion, as I'm not shure how to contribute to a PR on GH, but can contribute my code if someone can give me guidance (I assume I can make a PR to the feature/configuration branch?).

Summary

The new C# config stack (typed sections, INI provider, GetValueOrDefault / SaveSection) is in place and the INI file is read/written correctly. Python uses the C# layer; no revert to the old Python-only config. I only fixed bugs and added a small compatibility layer so existing code and extensions that still call user_config.core.get_option() keep working.

List of what I fixed locally (for testing)

1. pyrevitlib/pyrevit/coreutils/configparser.py

• get_option: Use GetValueOrDefault(section, option, "") then json.loads(value); handle empty.
• setattr: Skip __section_name and __configuration (use object.__setattr__ for those) so __init__ can set them and the rest goes through set_option.
• Subsections: Use __configuration / __section_name; add_subsection / get_subsection return correct types and None when not found.

2. pyrevitlib/pyrevit/userconfig.py

• Use getattr(EXEC_PARAMS, 'doc_mode', False) where doc_mode is read (not present on this branch).
• User extension folders: Read from self.core.UserExtensions (and (self.core.UserExtensions or [])); write with framework.List[str](...) so C# gets a List<string>.
• _SectionCompatWrapper: Wrapper around Core/Routes/Telemetry that exposes get_option / set_option and forwards other attributes to the C# section (so extensions and Settings UI keep working).
• setattr on the wrapper: Forward property sets to the underlying C# section so Settings changes apply to the real section.
• save_changes(): Call SaveSection with config_service.Core, config_service.Routes, config_service.Telemetry (not the Python wrappers) so the C# side gets the right type and the INI is updated.

3. dev/pyRevitLabs/pyRevitLabs.Configurations.Ini/IniConfiguration.cs

• For requested type string, return the raw value (no JsonConvert.DeserializeObject), so Python can json.loads() it (handles arrays and other JSON).
• Hex integer parsing for legacy INI (e.g. 0x0).
• net48: Use Substring(1, valueToParse.Length - 2) instead of [1..^1].

4. dev/pyRevitLabs/pyRevitLabs.PyRevit/pyRevitLabs.PyRevit.csproj

• DeployDependencies: Copy INIFileParser.dll to the output so the INI provider loads.

5. pyrevitlib/pyrevit/runtime/__init__.py

• Resolve runtime assembly path: try BIN_DIR first, then BIN_DIR/engines/<EngineVersion>/ so the Runtime DLL is found when it's only under engines (e.g. after building the Runtime solution).

6. pyrevitlib/pyrevit/revit/tabs.py

• _get_tabstyle / _get_family_tabstyle: Coerce index to int(...) (config can return string).
• hex_to_brush: Defensive handling (string, default brush) so bad/malformed config doesn't crash.
• _get_tab_orderrules: Treat tab_colors as list only when isinstance(raw, list).
• _get_tab_filterrules: Treat tab_filtercolors as dict only when isinstance(tab_filtercolors, dict).

7. dev/pyRevitLabs/tests/.../ConfigurationServiceFixture.cs

• Mock: Implement GetSectionNames() and GetSectionOptionNames(string) so the solution builds.

---

After these changes, Settings persist correctly to pyRevit_config.ini based on my testing (it might have before, but pulling the clean PR and building it was not working out-of-the-box for me).
I hope this testing is aligned with the intent of the PR?

Question: What other features or fixes are planned for this PR that I could help test or look into? For example:

• Config overrides (design and testing of precedence/order)?
• YAML configuration (parser library and integration)?
• Remaining Python config logic to rewrite onto the C# layer?
• JSON configuration (same level of testing/integration as INI)?
• INI file locking / multi-process behaviour ([Bug]: pyRevit_config.ini is being used by another process #2512)?
• Other?

[ChrisCrosley]

Any objections to me taking a swing at bringing this up to date with develop?

[jmcouffin]

Any objections to me taking a swing at bringing this up to date with develop?

I don't have any, @dosymep you?

a few notes though, I'd rather have a draft PR with no code changes and a clear plan that @dosymep or others can review before deeep diving. And there is a significant amount of drift between now and the start of this PR, 1 and a half year ago.

[dosymep]

I don't mind, but maybe everything there is already very outdated