services.py — Constants, Helpers, Extension Mapping, and Pipelines

[whisper67265]

Problem

The service layer is the core of the add-or-update workflow. It needs foundational infrastructure: constants (supported file extensions, default branch names, Weblate API base URLs), helper functions (slug generation, path normalization, VCS workspace management), extension-to-format mapping (.qbk -> QuickBook, .adoc -> AsciiDoc, .po -> gettext), and the construction logic that assembles a processing pipeline. This is the first half of the services module, focused on setup and scanning.

Acceptance Criteria

• src/boost_weblate/services.py defines constants: SUPPORTED_EXTENSIONS, DEFAULT_BRANCH, WORKSPACE_ROOT, format-class mapping
• Helper functions: slugify_component(path) -> deterministic slug, normalize_repo_url(url) -> canonical form, ensure_workspace(project_slug) -> local clone path
• Extension mapping: given a file path, return the correct Weblate format class (QuickBook, AsciiDoc, gettext, JSON, etc.)
• scan_repository(workspace_path, extensions) walks the repo tree and returns a list of (relative_path, format_class) tuples
• Workspace management: clone or pull the repo into a temp/workspace directory, handle branch checkout
• All functions are importable and testable in isolation (no global state, no Django ORM calls in helpers)

Implementation Notes

Keep constants at module top level. The extension mapping should be a dict, not a chain of if/elif. scan_repository should use pathlib.Path.rglob() and filter against SUPPORTED_EXTENSIONS. Workspace management can use subprocess + git or dulwich — match whatever the existing codebase uses. This module will be extended in the next services item (project/component/translation flows), so design for composability.

References

• Related files: src/boost_weblate/services.py, src/boost_weblate/formats/quickbook.py
• Depends on: Week 20 QuickBook format and utility modules