020b(gui): output-file Browse + scenario Save/Load buttons #33

Merged

ykp merged 1 commit from feat/020b-gui-file-buttons into main 2026-09-29 08:55:44 +02:00

Conversation 0 Commits 1 Files changed 4 +392 -1

ykp commented 2026-09-29 08:01:52 +02:00

Owner

Copy link

Implements backlog step 020b (docs/backlog/020b-gui-file-buttons.md). Depends on 020a (merged in #32) — 020c (Generate-scenario) is unblocked by this.

What was missing

The GUI had no way to persist or restore a scenario: closing the window discarded everything, and sidecar *.scenario.json files written by the CLI could not be reopened from the GUI (reported and confirmed by inspection — the only buttons were Set/Remove offset and Generate).

Changes

src/gengnss/gui/app.py (Settings tab):

• Browse… next to the Output-file entry — asksaveasfilename, default extension follows the fmt combobox (.i8/.i16); writes output_path through the same _on_setting path as the entry
• Save scenario… — scenario_from_model(model).to_json() to a .scenario.json sidecar; refuses an invalid model (validate_settings, e.g. NaN spans) with a status-line message, no file written
• Load scenario… — accepts both a bare to_json() dict and the full sidecar artifact written by generate_to_file ({"scenario": ...} wrapper) via the new scenario_from_sidecar helper; corrupt/wrong-shape JSON surfaces an error on the status line and leaves the model untouched
• Load repopulates the UI: settings entries, fmt/placeholder combos, doppler/delay spans, and selections grouped by signal with specs rebuilt via sat_spec_for_system (checkbox checked first, then the spec var set, so the existing _on_sat_spec trace registers them)
• resolved_seed is carried along for bookkeeping; output_path is deliberately not loaded (a sidecar travels with its IQ file) — the status line says output path unchanged
• generation lock: all three buttons disable while a job runs

src/gengnss/gui/controllers.py: scenario_to_model(scenario, model) (inverse of scenario_from_model) and scenario_from_sidecar(data) (both JSON shapes).

Tests

• tests/test_gui_load_save.py (headless, 5 tests): save→load round trip incl. GLONASS :chan + power offset; loading a generate_to_file sidecar in both shapes; corrupt-JSON rejection leaves the model unchanged; resolved_seed carried / output_path untouched; sat_spec_for_system specs re-parse to identical selections for all four systems
• tests/test_gui_app.py: gui-marked smoke — buttons exist, lock behavior, Browse/Save/Load end-to-end with monkeypatched dialogs (checkbox + spec entry restored)
• ruff clean; pytest: 197 passed, 1 skipped (gui-marked tests skip headless; run in CI with a display)
• Headless shim smoke of the real handlers: 24/24 checks (020a + 020b) — Browse/Save/Load flows, save guard, corrupt-JSON guard

Note: CI legs 3.12/3.14 have been failing since run 119 on docs-only commits (known unreliable runner workspace per the CI header note); the 3.9 leg is green and local runs (Python 3.14.4) are green.

Implements backlog step **020b** (`docs/backlog/020b-gui-file-buttons.md`). Depends on 020a (merged in #32) — 020c (Generate-scenario) is unblocked by this. ## What was missing The GUI had no way to persist or restore a scenario: closing the window discarded everything, and sidecar `*.scenario.json` files written by the CLI could not be reopened from the GUI (reported and confirmed by inspection — the only buttons were Set/Remove offset and Generate). ## Changes `src/gengnss/gui/app.py` (Settings tab): - **Browse…** next to the Output-file entry — `asksaveasfilename`, default extension follows the fmt combobox (`.i8`/`.i16`); writes `output_path` through the same `_on_setting` path as the entry - **Save scenario…** — `scenario_from_model(model).to_json()` to a `.scenario.json` sidecar; refuses an invalid model (validate_settings, e.g. NaN spans) with a status-line message, no file written - **Load scenario…** — accepts both a bare `to_json()` dict and the full sidecar artifact written by `generate_to_file` (`{"scenario": ...}` wrapper) via the new `scenario_from_sidecar` helper; corrupt/wrong-shape JSON surfaces an error on the status line and leaves the model untouched - Load repopulates the UI: settings entries, fmt/placeholder combos, doppler/delay spans, and selections grouped by signal with specs rebuilt via `sat_spec_for_system` (checkbox checked first, then the spec var set, so the existing `_on_sat_spec` trace registers them) - `resolved_seed` is carried along for bookkeeping; `output_path` is deliberately **not** loaded (a sidecar travels with its IQ file) — the status line says `output path unchanged` - generation lock: all three buttons disable while a job runs `src/gengnss/gui/controllers.py`: `scenario_to_model(scenario, model)` (inverse of `scenario_from_model`) and `scenario_from_sidecar(data)` (both JSON shapes). ## Tests - `tests/test_gui_load_save.py` (headless, 5 tests): save→load round trip incl. GLONASS `:chan` + power offset; loading a `generate_to_file` sidecar in both shapes; corrupt-JSON rejection leaves the model unchanged; `resolved_seed` carried / `output_path` untouched; `sat_spec_for_system` specs re-parse to identical selections for all four systems - `tests/test_gui_app.py`: gui-marked smoke — buttons exist, lock behavior, Browse/Save/Load end-to-end with monkeypatched dialogs (checkbox + spec entry restored) - `ruff` clean; `pytest`: **197 passed, 1 skipped** (gui-marked tests skip headless; run in CI with a display) - Headless shim smoke of the real handlers: 24/24 checks (020a + 020b) — Browse/Save/Load flows, save guard, corrupt-JSON guard Note: CI legs 3.12/3.14 have been failing since run 119 on docs-only commits (known unreliable runner workspace per the CI header note); the 3.9 leg is green and local runs (Python 3.14.4) are green.

ykp added 1 commit 2026-09-29 08:01:52 +02:00

020b(gui): output-file Browse + scenario Save/Load buttons ... 461fbcdfa6

Implements docs/backlog/020b-gui-file-buttons.md (depends on 020a, merged
in #32):

- Browse… next to the Output-file entry (Settings tab): asksaveasfilename
  with the extension matching the fmt combobox (.i8/.i16); writes the path
  through the same _on_setting path as the entry
- Save scenario…: serializes scenario_from_model(model).to_json() to a
  sidecar JSON (defaultextension .scenario.json); refuses an invalid model
  (validate_settings) with a status-line message instead of writing NaN
- Load scenario…: accepts both a bare to_json() dict and the full sidecar
  artifact written by generate_to_file ({"scenario": ...} wrapper) via the
  new controllers.scenario_from_sidecar helper; corrupt/wrong-shape JSON
  surfaces an error on the status line and leaves the model untouched
- controllers.scenario_to_model: inverse of scenario_from_model (maps fs,
  center, duration, fmt, full scale, seed, resolved_seed, spans,
  placeholder bits, selections, power offsets); output_path deliberately
  not loaded (a sidecar travels with its IQ file) — status line says so
- selections -> UI: grouped by signal, sat-spec strings rebuilt with
  sat_spec_for_system; checkbox checked before the spec var is set so the
  existing _on_sat_spec trace registers them (idempotent)
- generation lock: all three buttons disabled while a job runs

Tests: tests/test_gui_load_save.py (headless: round trip incl. GLONASS
:chan + power offset, generate_to_file sidecar load in both shapes,
resolved_seed carried / output_path untouched, corrupt JSON rejection,
spec re-parse round trip); gui-marked smoke in tests/test_gui_app.py
(buttons exist, lock behavior, Browse/Save/Load with monkeypatched
dialogs). ruff clean; pytest 197 passed, 1 skipped (headless GUI).

ykp merged commit 623a114660 into main 2026-09-29 08:55:44 +02:00

ykp referenced this pull request from a commit 2026-09-29 08:55:45 +02:00

Merge pull request '020b(gui): output-file Browse + scenario Save/Load buttons' (#33) from feat/020b-gui-file-buttons into main

ykp referenced this pull request 2026-09-29 14:44:34 +02:00

020c(gui): Generate scenario button — sidecar-only output + CLI parity #35

Sign in to join this conversation.

Reviewers

No reviewers

Labels

Clear labels

No items

No labels

Milestone

Clear milestone

No items

No milestone

Projects

Clear projects

No items

No project

Assignees

Clear assignees

No assignees

1 participant

Notifications

Subscribe

Due date

The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference

ykp/gengnss!33

No description provided.