public_git/bambuddy
1
0
Fork 0
mirror of https://github.com/maziggy/bambuddy.git synced 2026-10-17 11:25:48 +02:00
Code Issues Packages Projects Releases Wiki Activity

Match slicer presets on what they declare, not what they are named (issue #2982)

Browse Source 
The internal slicer picked PETG for a PLA plate and an A1 process for a
P1S. Both come from the sidecar's bundled-profile listing, fixed in the
sidecar repo; this is the consuming half plus the hardening that keeps an
older sidecar degrading rather than breaking.

Standard-tier presets now carry the compatible_printers the sidecar
reports. That list is the only truthful account of which printer a preset
belongs to, because the bundle ships no process preset named after a P1S,
an X1, an X1E or an H2D Pro -- all ten of the P1S's are named "@BBL X1C"
and name the P1S only in that list. Reading the printer out of the preset
NAME therefore made a P1S look like it had no compatible process at all:
all 198 hid behind "Show all" and the auto-pick fell through to an
alphabetically-first 0.06mm Fine @BBL A1 0.2 nozzle the CLI refused. A
P1S now gets 0.20mm Standard @BBL X1C and 73 filaments instead of 4. An
older sidecar reports nothing here, which leaves the name matcher in
place -- degraded as before, not broken.

Material is now a hard partition in the filament pre-pick rather than a
+10 bonus. A preset stating a different material than the plate asks for
is the wrong preset, not a worse one: wrong nozzle temperature, wrong bed
temperature, wrong flow. A preset stating NO material stays eligible --
unknown is not wrong, and 32 shipped profiles genuinely have none. The
same rule reaches the retain path, which held a slot on
printer-compatibility alone and so cemented a wrong-material pick through
every re-pick. A preset the user chose themselves is exempt: printing
PETG on a plate a designer labelled PLA is a legitimate thing to do, and
this rule exists to correct the auto-pick, not to overrule the user.

Two more, both found while tracing this and neither reported:

Among process presets equally valid for the selected printer, the one
nearest a 0.2mm layer height now wins. Within a tier the list is
alphabetical and Bambu's naming puts the finest height first, so every
slice that did not name its own process silently got 0.08mm Extra Fine on
an X1 Carbon and 0.06mm Fine on an A1 mini -- correct presets, nobody's
default. Ties break toward the coarser, faster height; a name with no
readable height is still pickable when it is the only candidate; a
process the 3MF named still wins outright.

H2DP is aliased to H2D Pro, the same shape as the A1M rename in #1649 --
the bundle spells the model one way in preset names and another in the
printer preset, so an H2D Pro classified all 198 processes as another
printer's. Deliberately narrow: H2DP and a plain H2D are different
machines and must not collapse.

A dropdown the printer filter would empty now shows the unfiltered list
instead. That state was reachable for four printer models and told the
user nothing; a visible preset for the wrong printer can be changed, an
empty dropdown cannot.

Verified against live Orca 2.4.2 and BambuStudio 02.08.02.61 sidecars
over the real 1156- and 1792-profile trees: every one of the eight
printer models tested now auto-picks a 0.20mm process for its own
printer, a PLA plate draws a PLA preset and a PETG plate a PETG one.
Each change was confirmed to fail its tests when reverted.
This commit is contained in:
maziggy
2026-08-28 10:30:32 +02:00
parent 73e0787bfd
commit e9daa2124e
11 changed files with 1036 additions and 144 deletions
Download Patch File Download Diff File Expand all files Collapse all files
frontend/src/utils/slicePresetPicker.ts
+102 -8
View File
@@ -14,10 +14,14 @@
// and the caller relies on the order, not a single merged list.
// - `pickProcessDefault` honours a 3MF's embedded process preset when
// it exists and isn't printer-incompatible; otherwise prefers a
// match-on-printer pick, then unknown-compat, then plain priority.
// match-on-printer pick, then unknown-compat, then plain priority --
// preferring a middle-of-the-road layer height within each of those
// (#2982).
// - `pickFilamentForSlot` partitions candidates into compatible/unknown
// vs mismatch buckets and only consults the mismatch bucket when
// the compatible bucket is empty (#1851).
// the compatible bucket is empty (#1851). Material type partitions the
// same way: a preset that states a different material than the plate
// asks for is never the right answer (#2982).
import type {
PresetRef,
@@ -100,16 +104,80 @@ export function pickProcessDefault(
}
for (const wanted of ['match', 'unknown'] as const) {
for (const tier of SLICE_MODAL_TIER_ORDER) {
for (const p of by[tier].process) {
if (presetCompatibility(p, 'process', printerName, compatIndex) === wanted) {
return { source: p.source, id: p.id };
}
}
const candidates = by[tier].process.filter(
(p) => presetCompatibility(p, 'process', printerName, compatIndex) === wanted,
);
const chosen = preferDefaultLayerHeight(candidates);
if (chosen) return { source: chosen.source, id: chosen.id };
}
}
return pickDefault(by, 'process');
}
// Bambu's default layer height, and the one its own presets are named
// "Standard" at. Used only to order candidates that are equally valid.
const DEFAULT_LAYER_HEIGHT_MM = 0.2;
// Leading "<n>mm" on a process preset name — "0.20mm Standard @BBL X1C".
const PROCESS_LAYER_HEIGHT = /^\s*([\d.]+)\s*mm\b/i;
/**
* Pick the best of a set of process presets that are all equally compatible.
*
* Tier order says which *source* to prefer, but within one tier the list is
* alphabetical, and alphabetical order on Bambu's naming scheme puts the
* finest layer height first. So the auto-pick landed on `0.08mm Extra Fine`
* for an X1C and `0.06mm Fine` for an A1 mini — a correct preset, but the
* slowest one the slicer ships, silently chosen for every slice that didn't
* name its own process (#2982).
*
* Preferring the height nearest 0.2mm gives the same answer a person would:
* `0.20mm Standard` where it exists, and the closest thing to it otherwise.
* Ties break toward the coarser height, then toward the earlier name, so the
* result stays deterministic. A name with no readable height sorts last but
* is still eligible — an imported preset called "My Draft" must remain
* pickable when it is the only candidate.
*/
function preferDefaultLayerHeight(candidates: UnifiedPreset[]): UnifiedPreset | null {
let best: UnifiedPreset | null = null;
let bestDistance = Number.POSITIVE_INFINITY;
let bestHeight = Number.NEGATIVE_INFINITY;
for (const p of candidates) {
const match = PROCESS_LAYER_HEIGHT.exec(p.name);
const height = match ? Number.parseFloat(match[1]) : Number.NaN;
const usable = Number.isFinite(height) && height > 0;
const distance = usable
? Math.abs(height - DEFAULT_LAYER_HEIGHT_MM)
: Number.POSITIVE_INFINITY;
if (best == null || distance < bestDistance
|| (distance === bestDistance && usable && height > bestHeight)) {
best = p;
bestDistance = distance;
bestHeight = usable ? height : Number.NEGATIVE_INFINITY;
}
}
return best;
}
/**
* True when ``preset`` states a material and it is not the one the plate slot
* asks for.
*
* Deliberately three-valued in effect: a preset with no stated material is not
* "different", it is unknown, and stays eligible. 32 profiles in the shipped
* BBL bundle inherit from a parent the bundle doesn't contain, so their
* material genuinely cannot be resolved — excluding those would leave slots
* with nothing to pick.
*/
export function statesDifferentMaterial(
preset: Pick<UnifiedPreset, 'filament_type'>,
requiredType: string,
): boolean {
const required = requiredType.trim().toUpperCase();
const stated = (preset.filament_type ?? '').trim().toUpperCase();
return Boolean(required) && Boolean(stated) && required !== stated;
}
export function pickFilamentForSlot(
by: UnifiedPresetsResponse,
required: { type: string; color: string },
@@ -136,11 +204,29 @@ export function pickFilamentForSlot(
// printer-correct alternative exists, which preserves the graceful-degrade
// behaviour for presets registries that genuinely have nothing for the
// selected printer.
//
// Material is the second hard partition (#2982). A preset that states a
// material the plate did not ask for is not a worse answer, it is the wrong
// one: printing a PLA plate with a PETG profile means the wrong nozzle
// temperature, the wrong bed temperature and the wrong flow. It used to be
// only a missed +10 bonus, which a colour hit plus a tier bonus could
// outweigh — and did, every time, once the standard tier's `filament_type`
// turned out to be null for every preset it listed: an A1 mini offered
// `Bambu PETG Basic` for a PLA plate, a P1S offered `Bambu PC`. Fixing the
// sidecar to report the material restores the signal; skipping a stated
// mismatch is what stops a wrong material from ever winning on colour again,
// whatever a future registry reports.
//
// A preset that states NO material stays eligible — that is "don't know",
// not "different", and 32 profiles in the shipped bundle inherit from a
// parent it doesn't contain, so their material is genuinely unknown. The
// same asymmetry `presetCompatibility` applies to printers.
const reqType = required.type.trim().toUpperCase();
const reqColor = normalizeColorForCompare(required.color);
let bestCompatible: { ref: PresetRef; score: number } | null = null;
let bestMismatch: { ref: PresetRef; score: number } | null = null;
let bestWrongType: { ref: PresetRef; score: number } | null = null;
for (const tier of SLICE_MODAL_TIER_ORDER) {
for (const p of by[tier].filament) {
let score = 0;
@@ -153,7 +239,11 @@ export function pickFilamentForSlot(
}
score += TIER_BONUS[tier];
const ref = { source: p.source, id: p.id };
if (presetCompatibility(p, 'filament', printerName, compatIndex) === 'mismatch') {
if (statesDifferentMaterial(p, reqType)) {
if (bestWrongType == null || score > bestWrongType.score) {
bestWrongType = { ref, score };
}
} else if (presetCompatibility(p, 'filament', printerName, compatIndex) === 'mismatch') {
if (bestMismatch == null || score > bestMismatch.score) {
bestMismatch = { ref, score };
}
@@ -164,6 +254,10 @@ export function pickFilamentForSlot(
}
if (bestCompatible != null) return bestCompatible.ref;
if (bestMismatch != null) return bestMismatch.ref;
// Nothing of the right material anywhere. Better a wrong-material preset the
// user can see and change in the dropdown than a null the modal renders as
// an empty slot, which is what shipped before the partition existed.
if (bestWrongType != null) return bestWrongType.ref;
// Final fallback when there are no filament presets at all (empty
// registry) — pickDefault returns null in that case too, but keeping the
// call mirrors the rest of the picker logic for shape consistency.
frontend/src/utils/slicerPrinterMatch.ts
+6
View File
@@ -45,6 +45,12 @@ export const EMPTY_COMPATIBILITY_INDEX: PrinterCompatibilityIndex = {
// (e.g. "X1" ⇄ "X1C") would silently group truly distinct printers.
const PRINTER_MODEL_SUFFIX_ALIASES: Record<string, readonly string[]> = {
'A1 MINI': ['A1M'],
// Same shape, spotted while tracing #2982: the bundle names every H2D Pro
// process and filament preset "@BBL H2DP" while the printer preset — and
// PRINTER_MODEL_MAP with it — spells the model "H2D Pro". Without the alias
// the H2D Pro classified all 198 bundled processes as belonging to another
// printer.
'H2D PRO': ['H2DP'],
};
/**