Update to .md-files

2026-05-01 00:28:09 +02:00
parent 32358a4e54
commit 9417d30574
3 changed files with 107 additions and 5 deletions
@@ -1,11 +1,107 @@
 # Migrering: Import-funktion → microservice-importer

-## Beslut
+## Status: ✅ GENOMFÖRD 2026-04-30

 - **Scope:** quick-import, parse-markdown, receipt-import
 - **Arkitektur:** Backend-till-backend — recipe-app NestJS-backend anropar microservice-importer internt via HTTP. Frontend ändras inte.
- **OCR:** Läggs till i microservice-importer (tesseract.js)
- **Infra:** Separata Docker Compose-filer, microservice-importer på port 3001
+- **OCR:** Implementerat i microservice-importer (tesseract.js + Alpine apk-paket)
+- **Infra:** `importer-api`-tjänst i `recipe-app/compose.yml`, port 3001 intern
+- **Driftsatt:** 2026-04-30, alla containers Healthy
+
+---
+
+## Fas 1 — Utöka microservice-importer ✅
+
+**1. OCR-stöd och multipart i quick-import** ✅
+- `QuickImportService.importFromUpload()` tillagd — hanterar PDF (pdf-parse) och bild (tesseract.js)
+- `quick-import.controller.ts` utökat med `FileInterceptor`, `@HttpCode(200)`
+- `Dockerfile` uppdaterad: `apk add tesseract-ocr tesseract-ocr-data-swe tesseract-ocr-data-eng`
+
+**2. `imageUrl` i quick-import-svaret** ✅
+- `imageUrl?: string` tillagd i `ParsedRecipe`-interfacet (`base.parser.ts`)
+- ICA-parsern extraherar nu `recipe.image` (string/array/objekt-varianter)
+- `QuickImportResult` utökad med `imageUrl?`, `imageWarning?`, source `'image'`
+- `normalizeImageUrl()` hanterar protokollrelativa URL:er (`//cdn.ica.se/...`)
+
+**3. Ny `ReceiptParsingModule`** ✅
+- `backend/src/receipt-parsing/receipt-parsing.service.ts` — Mistral AI-parsning av kvitto (bild/PDF)
+- `backend/src/receipt-parsing/receipt-parsing.controller.ts` — `POST /api/receipt-import/parse`, `@HttpCode(200)`, tillåter `application/octet-stream`
+- `backend/src/receipt-parsing/receipt-parsing.module.ts`
+- Registrerad i `app.module.ts`
+
+**4. Health-endpoint** ✅
+- `GET /api/health` → `{status: "ok"}` inline i `app.module.ts`
+
+**5. Bugfixar i document-service** ✅
+- `document-service.module.ts`: korrigerade importvägar + klassnamn (`DocumentImportModule` → `DocumentServiceModule`)
+- `services/document-import.service.ts`: parsersökväg `./parsers/` → `../parsers/`
+- Borttagna dubbletter: `services/web-scraping.module.ts`, `services/document-service.module.ts`
+
+---
+
+## Fas 2 — Anpassa recipe-app backend ✅
+
+**5. Refaktorera `QuickImportService`** ✅
+- All lokal parsning (ICA, pdf-parse, tesseract) borttagen
+- Delegerar URL-import: `POST importer-api:3001/api/quick-import` (JSON)
+- Delegerar filuploading: `POST importer-api:3001/api/quick-import` (FormData, `new Uint8Array(file.buffer)`)
+- `downloadAndOptimizeImage()` behålls lokalt (körs efter microservice returnerat `imageUrl`)
+- `IMPORTER_SERVICE_URL` env-variabel med fallback `http://importer-api:3001`
+
+**6. Refaktorera `ReceiptImportService`** ✅
+- AI-parsning (Mistral, pdf-parse) borttagen ur recipe-app
+- Delegerar till `POST importer-api:3001/api/receipt-import/parse` (FormData)
+- `matchProducts()` och `enrichWithAiCategories()` behålls (DB-krav)
+- `RECEIPT_IMPORT_MODEL`-konstanten flyttad till `ai.controller.ts` (lokal konstant)
+
+**7. Refaktorera `RecipesService.parseMarkdown()`** ✅
+- Delegerar markdown-parsning till `POST importer-api:3001/api/recipes/parse-markdown`
+- Fallback till lokal `parseRecipeMarkdown()` vid driftavbrott
+- Levenshtein-produktmatchning behålls lokalt
+
+---
+
+## Fas 3 — Infrastruktur ✅
+
+**8. `importer-api` i `recipe-app/compose.yml`** ✅
+- Build-context: `../microservice-importer`, dockerfile `backend/Dockerfile`
+- Image: `recipe-importer-api:local`, `pull_policy: never`
+- Nätverk: `recipe-internal` (ej exponerad externt)
+- Env: `MISTRAL_API_KEY`, `PORT=3001`
+- Healthcheck: `wget -qO- http://127.0.0.1:3001/api/health`
+- `recipe-api` får `depends_on: importer-api: condition: service_healthy`
+
+---
+
+## Relevanta filer som ändrades
+
+| Fil | Förändring |
+|---|---|
+| `microservice-importer/backend/src/web-scraping-service/parsers/base.parser.ts` | `imageUrl?` i `ParsedRecipe` |
+| `microservice-importer/backend/src/web-scraping-service/parsers/ica.parser.ts` | Extraherar `recipe.image` |
+| `microservice-importer/backend/src/web-scraping-service/services/quick-import.service.ts` | Omskriven: OCR, PDF, imageUrl, importFromUpload |
+| `microservice-importer/backend/src/web-scraping-service/controllers/quick-import.controller.ts` | FileInterceptor, HttpCode(200) |
+| `microservice-importer/backend/src/web-scraping-service/web-scraping.module.ts` | Fixade importvägar + klassnamn |
+| `microservice-importer/backend/src/document-service/document-service.module.ts` | Fixade importvägar + klassnamn |
+| `microservice-importer/backend/src/document-service/services/document-import.service.ts` | Fixad parsersökväg |
+| `microservice-importer/backend/src/receipt-parsing/` | Ny modul (service, controller, module) |
+| `microservice-importer/backend/src/app.module.ts` | ReceiptParsingModule + HealthController |
+| `microservice-importer/backend/Dockerfile` | apk add tesseract-ocr |
+| `recipe-app/backend/src/quick-import/quick-import.service.ts` | Delegerar till importer-api |
+| `recipe-app/backend/src/receipt-import/receipt-import.service.ts` | AI-del delegeras, matchning behålls |
+| `recipe-app/backend/src/recipes/recipes.service.ts` | parseMarkdown delegeras, matchning behålls |
+| `recipe-app/backend/src/ai/ai.controller.ts` | RECEIPT_IMPORT_MODEL lokal konstant |
+| `recipe-app/compose.yml` | importer-api-tjänst tillagd |
+
+---
+
+## Avgränsningar (oförändrade)
+
+- **Frontend ändras inte** — samma proxy-routes, samma API-kontrakt
+- **Auth stannar i recipe-app backend** — microservice-importer exponeras bara internt
+- **Bildoptimering** behålls i recipe-app (`downloadAndOptimizeImage` vid `RecipesService.create()`)
+- `matchProducts()` och `enrichWithAiCategories()` stannar i recipe-app (DB-krav)
+

 ---