Brings in the magic-link login, dev email filtering, and the Tailwind
toolchain ignores. The only conflict was .gitignore, where both sides
ignored the toolchain; kept main's broader `bin/` over `/bin/tailwindcss`.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

johnsturgeon added 1 commit

2026-09-10 13:35:44 +02:00

Updated to fix breakage

Claude Code Review / claude-review (pull_request) Has been skipped

Details

Python FastAPI Jinja Linting / build (pull_request) Failing after 27s

Details

Tests / pytest (pull_request) Failing after 18s

Details

73a6d0e5c3

johnsturgeon added 1 commit

2026-09-10 13:39:38 +02:00

Fixed up a few obvious lint issues

Claude Code Review / claude-review (pull_request) Has been skipped

Details

Python FastAPI Jinja Linting / build (pull_request) Failing after 29s

Details

Tests / pytest (pull_request) Successful in 15s

Details

53ced4dc79

johnsturgeon added 1 commit

2026-09-11 13:31:45 +02:00

migrated the picks page logic to a central class

Python FastAPI Jinja Linting / build (pull_request) Successful in 51s

Details

Tests / pytest (pull_request) Successful in 25s

Details

Claude Code Review / claude-review (pull_request) Has been skipped

Details

005cd32615

Even when I do away with the 'old' page, this is still a good design decision

johnsturgeon changed title from ~~WIP: fix-the-redesign-so-that-i-did-it-myself-426~~ to fix-the-redesign-so-that-i-did-it-myself-426

2026-09-11 13:32:09 +02:00

johnsturgeon added 1 commit

2026-09-11 13:41:39 +02:00

Bumped the version

Claude Code Review / claude-review (pull_request) Successful in 2m10s

Details

Python FastAPI Jinja Linting / build (pull_request) Successful in 29s

Details

Tests / pytest (pull_request) Successful in 14s

Details

1927c34460

johnsturgeon commented

2026-09-11 13:43:50 +02:00

Author

Owner

The legacy /picks and /picks_form routes are thoroughly tested and now delegate to the extracted picks_page/record_picks, so the moved logic stays covered. My review is ready.

This PR extracts the picks eligibility/validation logic out of app/main.py into a shared app/picks.py and adds a parallel v3 UI (new router, Tailwind-built CSS, Jinja templates). The refactor is behavior-preserving on the legacy routes and the extraction is clean; the findings below are mostly about the new v3 surface, which is wired into the live app while still partly a stub.

Worth fixing

The v3 router is mounted in production but has no test coverage. app/main.py:139 includes v3.router, yet tests/ contains no v3 requests (grep for /v3 is empty). The shared picks_page/record_picks are exercised through the legacy routes in tests/test_picks_routes.py, but the v3 wiring itself — auth redirects, the get_open_picks_week 409 gate on app/routers/v3.py:83, display_week_id handling in app/routers/v3.py:150, error-page rendering — is untested. At minimum add a happy-path GET/POST test per v3 route so a template rename or context-key change doesn't 500 silently in production.
Stub pages and dead nav links are publicly reachable. app/templates/v3/standings.html, rules.html, profile.html, and login.html are empty ({% extends "v3/base.html" %} only), and the nav in app/templates/v3/base.html:35-40 renders Profile/Settings/Sign out/Sign in as href="#". Because the routes require get_player, they're behind auth and nothing in the legacy UI links to /v3, so risk is low — but confirm this half-built UI is meant to ship in 2.34.0 rather than sit behind a feature flag or stay unmerged.
Verify pylint still passes. app/routers/v2.py:13 gained # pylint: disable=duplicate-code, but app/routers/v3.py is a near-clone of v2's structure and did not. R0801 reports a duplication cluster unless the disable is present in every file participating in it, and scripts/test_and_lint.sh:12 runs pylint without --exit-zero. I couldn't run pylint in this sandbox to confirm; please make sure test_and_lint.sh is green before merging (v3.py may need the same disable).

Nits

v3 allpicks can't browse past weeks. app/routers/v3.py:150 accepts display_week_id, but no v3 template or nav emits such a link (allpicks.html has no week selector), so the historical-week navigation the legacy page offers via week_no/season_type/season isn't reachable in v3 yet. Presumably intentional for this pass — flagging so it isn't forgotten.
Malformed unicode-range for the Figtree font. In app/styles/v3.src.css the Figtree @font-face uses U+0000 -00FF, U+0152 -0153, … (stray spaces), which minifies to the mangled U0 -0FF, U152 -153, … in app/static/v3.css. Inter and Rock Salt use the correct spaceless form. Browsers reject the invalid descriptor and fall back to the full range, so the font still loads — but it's inconsistent with the other two and worth cleaning up at the source.
PicksPage.as_context() leaks an unused games key into the legacy picks.j2 context (app/main.py:247). Harmless (the template ignores it), just slightly untidy versus the old hand-built context.

Overall this is a solid, low-risk refactor; the main asks are test coverage for the new v3 routes and confirming the v3 UI is intended to go live in this state.

The legacy `/picks` and `/picks_form` routes are thoroughly tested and now delegate to the extracted `picks_page`/`record_picks`, so the moved logic stays covered. My review is ready. --- This PR extracts the picks eligibility/validation logic out of `app/main.py` into a shared `app/picks.py` and adds a parallel v3 UI (new router, Tailwind-built CSS, Jinja templates). The refactor is behavior-preserving on the legacy routes and the extraction is clean; the findings below are mostly about the new v3 surface, which is wired into the live app while still partly a stub. ## Worth fixing - **The v3 router is mounted in production but has no test coverage.** `app/main.py:139` includes `v3.router`, yet `tests/` contains no v3 requests (grep for `/v3` is empty). The shared `picks_page`/`record_picks` are exercised through the legacy routes in `tests/test_picks_routes.py`, but the v3 wiring itself — auth redirects, the `get_open_picks_week` 409 gate on `app/routers/v3.py:83`, `display_week_id` handling in `app/routers/v3.py:150`, error-page rendering — is untested. At minimum add a happy-path GET/POST test per v3 route so a template rename or context-key change doesn't 500 silently in production. - **Stub pages and dead nav links are publicly reachable.** `app/templates/v3/standings.html`, `rules.html`, `profile.html`, and `login.html` are empty (`{% extends "v3/base.html" %}` only), and the nav in `app/templates/v3/base.html:35-40` renders `Profile`/`Settings`/`Sign out`/`Sign in` as `href="#"`. Because the routes require `get_player`, they're behind auth and nothing in the legacy UI links to `/v3`, so risk is low — but confirm this half-built UI is meant to ship in 2.34.0 rather than sit behind a feature flag or stay unmerged. - **Verify pylint still passes.** `app/routers/v2.py:13` gained `# pylint: disable=duplicate-code`, but `app/routers/v3.py` is a near-clone of v2's structure and did not. `R0801` reports a duplication cluster unless the disable is present in *every* file participating in it, and `scripts/test_and_lint.sh:12` runs pylint without `--exit-zero`. I couldn't run pylint in this sandbox to confirm; please make sure `test_and_lint.sh` is green before merging (v3.py may need the same disable). ## Nits - **v3 `allpicks` can't browse past weeks.** `app/routers/v3.py:150` accepts `display_week_id`, but no v3 template or nav emits such a link (`allpicks.html` has no week selector), so the historical-week navigation the legacy page offers via `week_no`/`season_type`/`season` isn't reachable in v3 yet. Presumably intentional for this pass — flagging so it isn't forgotten. - **Malformed `unicode-range` for the Figtree font.** In `app/styles/v3.src.css` the Figtree `@font-face` uses `U+0000 -00FF, U+0152 -0153, …` (stray spaces), which minifies to the mangled `U0 -0FF, U152 -153, …` in `app/static/v3.css`. Inter and Rock Salt use the correct spaceless form. Browsers reject the invalid descriptor and fall back to the full range, so the font still loads — but it's inconsistent with the other two and worth cleaning up at the source. - **`PicksPage.as_context()` leaks an unused `games` key** into the legacy `picks.j2` context (`app/main.py:247`). Harmless (the template ignores it), just slightly untidy versus the old hand-built context. Overall this is a solid, low-risk refactor; the main asks are test coverage for the new v3 routes and confirming the v3 UI is intended to go live in this state.

johnsturgeon added 1 commit

2026-09-11 13:52:55 +02:00

Built css (tailwind)

Claude Code Review / claude-review (pull_request) Successful in 3m45s

Details

Python FastAPI Jinja Linting / build (pull_request) Successful in 30s

Details

Tests / pytest (pull_request) Successful in 15s

Details

Release on merge / release (pull_request) Successful in 0s

Details

fd373530a3