- scripts/taxonomy.py: shared taxonomy with 14 categories, keyword scorer and classify_text() function - scripts/classify_server.py: FastAPI service — forwards to kreuzberg /extract, applies taxonomy, returns category/subcategory/confidence alongside full kreuzberg response - Dockerfile.classify: lightweight Python image for classify service - classify.nomad: Nomad job → classify.i80.dk - .gitea/workflows/classify.yml: CI/CD pipeline (build + deploy) - analyse_familie.py: refactored to import from taxonomy.py (no duplication) - .gitignore: exclude dokumenter_keywords.* and extract_all.log
53 lines
1.2 KiB
YAML
53 lines
1.2 KiB
YAML
name: Deploy classify service
|
|
|
|
on:
|
|
push:
|
|
branches:
|
|
- main
|
|
paths:
|
|
- "scripts/classify_server.py"
|
|
- "scripts/taxonomy.py"
|
|
- "Dockerfile.classify"
|
|
- "classify.nomad"
|
|
workflow_dispatch:
|
|
|
|
env:
|
|
REGISTRY: ghcr.io
|
|
IMAGE_NAME: hjess/kreuzberg-classify
|
|
|
|
jobs:
|
|
build-and-deploy:
|
|
runs-on: debian-host
|
|
|
|
env:
|
|
PATH: /usr/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/sbin:/bin:/snap/bin
|
|
|
|
steps:
|
|
- name: Checkout code
|
|
uses: actions/checkout@v4
|
|
|
|
- name: Build Docker image
|
|
run: |
|
|
docker build -f Dockerfile.classify -t ${{ env.REGISTRY }}/${{ env.IMAGE_NAME }}:latest .
|
|
|
|
- name: Push to registry
|
|
run: |
|
|
docker push ${{ env.REGISTRY }}/${{ env.IMAGE_NAME }}:latest
|
|
|
|
- name: Validate Nomad job
|
|
run: nomad job validate classify.nomad
|
|
env:
|
|
NOMAD_ADDR: "https://nomad.i80.dk:4646"
|
|
|
|
- name: Deploy to Nomad
|
|
run: nomad job run classify.nomad
|
|
env:
|
|
NOMAD_ADDR: "https://nomad.i80.dk:4646"
|
|
|
|
- name: Check deployment status
|
|
run: |
|
|
sleep 10
|
|
nomad job status classify
|
|
env:
|
|
NOMAD_ADDR: "https://nomad.i80.dk:4646"
|