feat: block slicing when filament flow ratio is 0 (with per-filament breakdown)

[iesteem]

Summary

Block slicing when any actually-used filament on a plate has filament_flow_ratio == 0, and surface a ValidateError notification listing the offending filament slots (e.g. [3] Snapmaker PLA Basic @U1) — fully aligned with the existing filament-temperature-mixing blocker. Goal: turn the "zero extrusion → empty toolpath" silent failure into a visible pre-slice error.

The PR also bundles a few related minor fixes (see "Related fixes" below).

Core changes

Flow-ratio-zero blocking core (Plater.cpp / Plater.hpp)

• New FlowRatioZeroDetail { offender_slots_1based }; slots are 1-based to match the UI.
• check_flow_ratio_zero(plate_index[, detail]): boundary checks + only when the plate has objects, collect the actually-used filament slots (same collection logic as check_filament_temp_mixing); read each preset's filament_flow_ratio, treat <= 0.0 as offender.
• is_plate_blocked_by_flow_ratio_zero(): exposed to MainFrame / GLCanvas3D for slice-button greying and plate-toolbar SLICE_FAILED state.
• sync_flow_ratio_zero_notification(): close-then-push pattern; caches flow_ratio_zero_last_error_text to match close_validate_error_notification's exact-text matching; does not touch m_apply_invalid (blocking is enforced independently via get_enable_slice_status() / find_next_sliceable_plate_for_slice_all()).

Slice-path integration

• update_background_process(): invoke sync_flow_ratio_zero_notification() after the blanket close_notification_of_type(ValidateError); fold its return value into UPDATE_BACKGROUND_PROCESS_INVALID / UPDATE_BACKGROUND_PROCESS_RESTART to plug the Preview-tab-switch / reslice path.
• start_next_slice(): new independent-dimension check; on slice_all it posts Finished to skip the current plate, on single-plate it sets process_completed_with_error. Comment emphasizes "two blockers must be parallel ifs, never else-if".
• EVT_GLVIEWTOOLBAR_PREVIEW: switched from "directly fire slice_all" to "first check can_slice() + temp_mixing + flow_ratio_zero; if blocked, sync notifications and switch to Preview view".
• guard_before_slice_plate() / guard_before_slice_all(): appended sync calls.
• find_next_sliceable_plate_for_slice_all(): added !is_plate_blocked_by_flow_ratio_zero() condition.
• EVT_FILAMENT_USAGE_CHANGED and end of load_files(): appended sync.
• MainFrame::get_enable_slice_status(), GLCanvas3D::_render_imgui_select_plate_toolbar: added blocking check; grey out slice button / show SLICE_FAILED.

Related fixes (bundled — worth a separate glance)

• PrintBase.hpp + Snapmaker_Orca.cpp: renamed STRING_EXCEPT_FILAMENTS_DIFFERENT_TEMP → STRING_EXCEPT_FILAMENTS_MIXING_TEMP (clearer semantics), added STRING_EXCEPT_FLOW_RATIO_ZERO = 6.
• GCodeReader.cpp: parse_file_raw_internal() returns false when fopen fails, preventing fread(NULL) from triggering a CRT invalid-parameter crash.
• Plater::load_gcode(): pre-check for missing file — gives an accurate error instead of the misleading "does not contain valid G-code".
• 3MF load validity-error dedup: skip the filament_flow_ratio entry (already surfaced more precisely by the flow_ratio_zero banner); don't pop an empty notification when all entries are skipped.
• History commits on this branch also include: per-filament high/low breakdown for temp-mixing notifications, on-demand checks replacing the cached flag, and unified plate-error gating.

Localization

• localization/i18n/zh_CN/Snapmaker_Orca_zh_CN.po: added flow_ratio_zero error strings (Chinese only; other languages will be synced by CI / run_gettext later).

---

Reviewer Focus

Please prioritize the following (ordered by risk):

1. Parity between check_flow_ratio_zero and check_filament_temp_mixing (most important)

Both must use exactly the same definition of "actually-used slot", otherwise you get inconsistent states ("temp mixing blocks but flow_ratio doesn't", or vice versa). Please verify item-by-item:

• collection order across plate config / object config / volume config
• uses_default_extruder flag + global extruder resolution
• boundaries of always_collect (wipe_tower / support / support_interface) vs default_keys (wall / sparse_infill / solid_infill)
• expansion of ModelVolume::get_extruders()

Forward-looking note: these two near-identical blocks should be factored into a shared helper. PR #627 (cold plate) already does this refactor — merge order between the two PRs needs to be coordinated to avoid conflicts.

2. The <= 0.0 comparison

The code deliberately uses ratio <= 0.0 instead of == 0.0, with the comment "to dodge floating-point comparison traps". But what's the smallest legal positive value of filament_flow_ratio in the UI? Could a tiny positive value (e.g. 1e-6) be mis-classified as an offender due to FP precision? Please confirm against the actual preset range (recommend either strict <= 0.0 or explicit == 0.0).

3. Correctness of close-then-push notification

• sync_flow_ratio_zero_notification() relies on the flow_ratio_zero_last_error_text cache for exact-text close. Please check:
  • The curr_plate == nullptr early-return does not clear the cache — could there be a cross-plate-switch scenario where a stale notification lingers?
  • update_background_process() requires the sync to run after the blanket close — is this ordering satisfied at every call site? Especially EVT_FILAMENT_USAGE_CHANGED and the end of load_files().
• When both flow_ratio_zero and filament_temp_mixing notifications are active, can two ValidateErrors of the same type but different text clobber each other?

4. Dual-blocker ordering in start_next_slice()

The new flow_ratio_zero check sits after temp_mixing as an independent if (not else-if). Please confirm:

• When both fire, both syncs run (user sees two notifications) — is that intended?
• On the slice_all path, either blocker posts Finished to skip the current plate; does find_next_sliceable_plate_for_slice_all() correctly advance to the next legal plate?

5. Behavior change in EVT_GLVIEWTOOLBAR_PREVIEW

Before: when has_sliceable_plate_for_slice_all() was true, the handler switched to 3D Editor and posted SLICE_ALL — auto-triggering a slice.
After: it checks blockers first; if blocked, switches to Preview view without slicing.
Please confirm this behavior change doesn't break the existing "click Preview to auto-slice_all" workflow — in particular, is the non-blocking auto-slice behavior preserved? (From the diff, the entire non-blocking path appears to have been removed — please confirm this is intentional.)

6. Enum / persistence impact

• STRING_EXCEPT_FILAMENTS_DIFFERENT_TEMP → STRING_EXCEPT_FILAMENTS_MIXING_TEMP: search the whole repo to confirm no missed call sites (CLI, 3MF serialization, Sentry reporting, etc.).
• New STRING_EXCEPT_FLOW_RATIO_ZERO = 6: does it conflict with how StringExceptionType is persisted in 3MF; is opening an old .3mf project safe?

7. Regression risk from bundled fixes

• GCodeReader::parse_file_raw_internal() NULL-FILE* defense: will Sentry crashes that previously fired on this path disappear?
• load_gcode() missing-file check: does it cover Unicode paths / network paths under wxFileExists?
• 3MF validity skipping filament_flow_ratio: confirm this is the only key the flow_ratio_zero banner takes over; are there other keys that should also be skipped?

8. i18n coverage

• This PR only updates the zh_CN .po — the other 20 languages are not synced. The flow_ratio_zero copy will fall back to English in non-Chinese locales. Is that acceptable, or should we wait for update-translation.yml to auto-sync?

---

Test Plan

• Set any filament's filament_flow_ratio to 0 → that plate's slice button greys out; ValidateError lists [n] PresetName
• Set filament_flow_ratio to a tiny positive value (e.g. 0.001) → should not be blocked (verifies the <= 0.0 boundary)
• Only one slot has ratio=0 but is unused → should not be blocked (verifies only used slots are checked)
• Only one slot has ratio=0 and is actually used by an object → blocked, and the slot is named precisely
• Both flow_ratio_zero and temp mixing triggered → two ValidateError notifications coexist without clobbering
• Slice All: plate A blocked / plate B compatible → auto-advance to plate B
• Reslice / Preview-tab switch: a blocked plate must not silently start slicing
• 3MF load: a project containing filament_flow_ratio=0 → only the flow_ratio_zero banner shows; no duplicate "Invalid values found in the 3mf" notification
• Loading a non-existent G-code file → "file does not exist" dialog, no crash
• Verify the Chinese copy renders correctly; switching to English / other locales at least doesn't crash (fallback acceptable)

---

🤖 Generated with Claude Code