arnaudne/Epicure
1
0
Fork 0
Code Issues 2 Pull Requests Actions Packages Projects Releases Wiki Activity

feat: split photo-import into vision recognition + text generation

Browse Source 
The photo-import flow used one vision-capable model to both read the
photo and structure the full recipe (quantities, steps, timing) in a
single call. Split it into two: a vision model recognizes what's in
the picture, then a text model reconstructs the recipe from that
description — same pattern the pantry photo-scan already uses for
recognition, and lets structuring use whichever model is actually
configured for text generation. Still counts as one AI-quota unit.

v0.37.0
This commit is contained in:
Arnaud
2026-07-17 16:12:39 +02:00
parent 5763bd3318
commit f0340859fa
12 changed files with 157 additions and 59 deletions
Download Patch File Download Diff File Expand all files Collapse all files
apps/web/app/api/v1/ai/import-photo/route.ts
+12 -8
View File
@@ -28,19 +28,23 @@ export async function POST(req: NextRequest) {
const userId = session!.user.id;
const locale = (session!.user as { locale?: string }).locale ?? "en";
const configResult = await resolveAiConfigOrError(() => getModelConfigForUseCase(userId, "vision"));
if (!configResult.ok) return configResult.response;
const aiConfig = configResult.data;
const visionConfigResult = await resolveAiConfigOrError(() => getModelConfigForUseCase(userId, "vision"));
if (!visionConfigResult.ok) return visionConfigResult.response;
const visionConfig = visionConfigResult.data;
// Fall back to vision-capable defaults if no explicit model configured
if (!aiConfig.model) {
if (aiConfig.provider === "openai") aiConfig.model = "gpt-4o";
else if (aiConfig.provider === "anthropic") aiConfig.model = "claude-sonnet-4-6";
if (!visionConfig.model) {
if (visionConfig.provider === "openai") visionConfig.model = "gpt-4o";
else if (visionConfig.provider === "anthropic") visionConfig.model = "claude-sonnet-4-6";
}
const textConfigResult = await resolveAiConfigOrError(() => getModelConfigForUseCase(userId, "text"));
if (!textConfigResult.ok) return textConfigResult.response;
const textConfig = textConfigResult.data;
const result = await withAiQuota(userId, session!.user.tier as "free" | "pro", () =>
importFromPhoto(parsed.data.imageBase64, parsed.data.mimeType, aiConfig, locale),
{ skipQuota: aiConfig.isByok }
importFromPhoto(parsed.data.imageBase64, parsed.data.mimeType, visionConfig, textConfig, locale),
{ skipQuota: visionConfig.isByok && textConfig.isByok }
);
if (!result.ok) return result.response;
const recipe = result.data;