feat: label-studio构建脚本 (#198)

* feat: label-studio构建脚本

* feat: label-studio构建脚本

* feat: label-studio构建脚本

* feat: label-studio安装脚本
This commit is contained in:
hhhhsc701
2025-12-25 11:44:05 +08:00
committed by GitHub
parent 1c507ac98a
commit 7e842c7cd5
10 changed files with 235 additions and 193 deletions

View File

@@ -5,6 +5,7 @@ on:
branches: [ "main" ]
paths:
- 'backend/**'
- '!backend/api-gateway/**'
- 'scripts/images/backend/**'
- '.github/workflows/docker-image-backend.yml'
- '.github/workflows/docker-images-reusable.yml'
@@ -12,6 +13,7 @@ on:
branches: [ "main" ]
paths:
- 'backend/**'
- '!backend/api-gateway/**'
- 'scripts/images/backend/**'
- '.github/workflows/docker-image-backend.yml'
- '.github/workflows/docker-images-reusable.yml'

View File

@@ -4,14 +4,14 @@ on:
push:
branches: [ "main" ]
paths:
- 'backend/**'
- 'backend/api-gateway/**'
- 'scripts/images/gateway/**'
- '.github/workflows/docker-image-gateway.yml'
- '.github/workflows/docker-images-reusable.yml'
pull_request:
branches: [ "main" ]
paths:
- 'backend/**'
- 'backend/api-gateway/**'
- 'scripts/images/gateway/**'
- '.github/workflows/docker-image-gateway.yml'
- '.github/workflows/docker-images-reusable.yml'

View File

@@ -0,0 +1,28 @@
name: Label Studio Docker Image CI
on:
push:
branches: [ "main" ]
paths:
- 'scripts/images/label-studio/**'
- '.github/workflows/docker-image-label-studio.yml'
- '.github/workflows/docker-images-reusable.yml'
pull_request:
branches: [ "main" ]
paths:
- 'scripts/images/label-studio/**'
- '.github/workflows/docker-image-label-studio.yml'
- '.github/workflows/docker-images-reusable.yml'
workflow_dispatch:
workflow_call:
jobs:
call-docker-build:
name: Build and Push Label Studio Docker Image
uses: ./.github/workflows/docker-images-reusable.yml
permissions:
contents: read
packages: write
with:
service_name: label-studio
build_dir: .

View File

@@ -1,110 +0,0 @@
name: DataMate Package
on:
workflow_dispatch:
inputs:
milvus:
type: boolean
description: 是否打包milvus
required: false
default: true
deer-flow:
type: boolean
description: 是否打包deer-flow
required: false
default: false
version:
type: string
description: 版本号
required: false
default: 1.0.0
jobs:
package-all:
runs-on: ubuntu-latest
steps:
- name: Checkout
uses: actions/checkout@v4
- name: Clone DataMate-Deploy
uses: GuillaumeFalourd/clone-github-repo-action@v2.3
with:
owner: ModelEngine-Group
repository: 'DataMate-Deploy'
access-token: ${{ secrets.ACCESS_TOKEN }}
- name: Copy tools
run: |
cp -r DataMate-Deploy/tools tools
- name: DataMate Package
run: |
mkdir helm
cp -r deployment/helm/datamate helm/datamate
sed -i "s#^\(\s*repository:\s*\).*#\1\"\"#" helm/datamate/values.yaml
sed -i "s#^\(\s*type:\s*\).*#\1ClusterIP#" helm/datamate/values.yaml
- name: DeerFlow Package
if: inputs.deer-flow == true
run: |
cp runtime/deer-flow/.env deployment/helm/deer-flow/charts/public/.env
cp runtime/deer-flow/conf.yaml deployment/helm/deer-flow/charts/public/conf.yaml
cp -r deployment/helm/deer-flow helm/deer-flow
- name: Milvus Package
if: inputs.milvus == true
run: |
cp -r deployment/helm/milvus helm/milvus
sed -i "s#^\(\s*workload:\s*\).*#\1deployment#" helm/milvus/values.yaml
- name: Download DataMate Image
run: |
mkdir -p images/datamate
LOWERCASE_REPO=$(echo "${{ github.repository_owner }}" | tr '[:upper:]' '[:lower:]')
services=("runtime" "backend" "frontend" "database" "backend-python" "gateway")
for service in "${services[@]}"; do
docker pull ghcr.io/$LOWERCASE_REPO/datamate-$service:latest --platform arm64
docker tag ghcr.io/$LOWERCASE_REPO/datamate-$service:latest datamate-$service:latest
docker save -o images/datamate/datamate-$service.tar datamate-$service:latest
docker rmi ghcr.io/$LOWERCASE_REPO/datamate-$service:latest datamate-$service:latest
done
docker pull quay.io/kuberay/operator:v1.4.2 --platform arm64
docker save -o images/datamate/kuberay-operator.tar quay.io/kuberay/operator:v1.4.2
docker rmi quay.io/kuberay/operator:v1.4.2
- name: Download DeerFlow Image
if: inputs.deer-flow == true
run: |
mkdir -p images/deer-flow
LOWERCASE_REPO=$(echo "${{ github.repository_owner }}" | tr '[:upper:]' '[:lower:]')
services=("backend" "frontend")
for service in "${services[@]}"; do
docker pull ghcr.io/$LOWERCASE_REPO/deer-flow-$service:latest --platform arm64
docker tag ghcr.io/$LOWERCASE_REPO/deer-flow-$service:latest deer-flow-$service:latest
docker save -o images/deer-flow/deer-flow-$service.tar deer-flow-$service:latest
docekr rmi ghcr.io/$LOWERCASE_REPO/deer-flow-$service:latest deer-flow-$service:latest
done
- name: Download Milvus Image
if: inputs.milvus == true
run: |
mkdir -p images/milvus
docker pull milvusdb/milvus:v2.6.5 --platform arm64
docker save -o images/milvus/milvus.tar milvusdb/milvus:v2.6.5
docker rmi milvusdb/milvus:v2.6.5
docker pull minio/minio:RELEASE.2024-12-18T13-15-44Z --platform arm64
docker save -o images/milvus/minio.tar minio/minio:RELEASE.2024-12-18T13-15-44Z
docker rmi minio/minio:RELEASE.2024-12-18T13-15-44Z
docker pull milvusdb/etcd:3.5.18-r1 --platform arm64
docker save -o images/milvus/etcd.tar milvusdb/etcd:3.5.18-r1
docker rmi milvusdb/etcd:3.5.18-r1
- name: Upload Package
uses: actions/upload-artifact@v4
with:
name: DataMate_${{ inputs.version }}_Aarch64
include-hidden-files: true
path: |
helm/
images/
tools/

View File

@@ -1,78 +0,0 @@
name: Package All
on:
workflow_dispatch:
jobs:
backend-docker-build:
name: Build and Push Backend Docker Image
uses: ./.github/workflows/docker-image-save.yml
with:
service_name: backend
frontend-docker-build:
name: Build and Push Frontend Docker Image
uses: ./.github/workflows/docker-image-save.yml
with:
service_name: frontend
database-docker-build:
name: Build and Push Database Docker Image
uses: ./.github/workflows/docker-image-save.yml
with:
service_name: database
runtime-docker-build:
name: Build and Push Runtime Docker Image
uses: ./.github/workflows/docker-image-save.yml
with:
service_name: runtime
backend-python-docker-build:
name: Build and Push Backend Python Docker Image
uses: ./.github/workflows/docker-image-save.yml
with:
service_name: backend-python
gateway-docker-build:
name: Build and Push Frontend Docker Image
uses: ./.github/workflows/docker-image-save.yml
with:
service_name: gateway
package-all:
needs:
- backend-docker-build
- frontend-docker-build
- database-docker-build
- backend-python-docker-build
- runtime-docker-build
- gateway-docker-build
runs-on: ubuntu-latest
steps:
- name: Checkout
uses: actions/checkout@v4
- name: Download Image
uses: actions/download-artifact@v5
with:
path: images
pattern: datamate-*
merge-multiple: true
- name: Package
run: |
cp runtime/deer-flow/.env deployment/docker/deer-flow/.env
cp runtime/deer-flow/conf.yaml deployment/docker/deer-flow/conf.yaml
cp runtime/deer-flow/.env deployment/helm/deer-flow/charts/public/.env
cp runtime/deer-flow/conf.yaml deployment/helm/deer-flow/charts/public/conf.yaml
tar -cvf datamate.tar deployment/ images
- name: Upload Package
uses: actions/upload-artifact@v4
with:
name: DataMate
include-hidden-files: true
path: |
deployment/
images/

View File

@@ -155,7 +155,7 @@ endef
# ========== Build Targets ==========
# Valid build targets
VALID_BUILD_TARGETS := backend database frontend runtime backend-python deer-flow mineru mineru-npu gateway
VALID_BUILD_TARGETS := backend database frontend runtime backend-python deer-flow mineru mineru-npu gateway label-studio
# Generic docker build target with service name as parameter
# Automatically prefixes image names with "datamate-" unless it's deer-flow
@@ -231,7 +231,6 @@ else
fi
@$(MAKE) label-studio-$(INSTALLER)-uninstall DELETE_VOLUMES_CHOICE=$$DELETE_VOLUMES_CHOICE; \
$(MAKE) milvus-$(INSTALLER)-uninstall DELETE_VOLUMES_CHOICE=$$DELETE_VOLUMES_CHOICE; \
$(MAKE) datamate-$(INSTALLER)-uninstall DELETE_VOLUMES_CHOICE=$$DELETE_VOLUMES_CHOICE; \
$(MAKE) deer-flow-$(INSTALLER)-uninstall DELETE_VOLUMES_CHOICE=$$DELETE_VOLUMES_CHOICE; \
$(MAKE) datamate-$(INSTALLER)-uninstall DELETE_VOLUMES_CHOICE=$$DELETE_VOLUMES_CHOICE
endif
@@ -321,7 +320,9 @@ VALID_K8S_TARGETS := mineru datamate deer-flow milvus label-studio
done; \
exit 1; \
fi
@if [ "$*" = "mineru" ]; then \
@if [ "$*" = "label-studio" ]; then \
helm upgrade label-studio deployment/helm/label-studio/ -n $(NAMESPACE) --install; \
elif [ "$*" = "mineru" ]; then \
kubectl apply -f deployment/kubernetes/mineru/deploy.yaml -n $(NAMESPACE); \
elif [ "$*" = "datamate" ]; then \
helm upgrade datamate deployment/helm/datamate/ -n $(NAMESPACE) --install --set global.image.repository=$(REGISTRY); \

View File

@@ -21,7 +21,11 @@ spec:
spec:
containers:
- name: label-studio
{{- if .Values.global.imageRegistry }}
image: "{{ .Values.global.imageRegistry | trimSuffix "/" }}/{{ .Values.image.repository }}:{{ .Values.image.tag }}"
{{- else }}
image: "{{ .Values.image.repository }}:{{ .Values.image.tag }}"
{{- end }}
imagePullPolicy: {{ .Values.image.pullPolicy }}
args: ["label-studio-uwsgi"]
env:

View File

@@ -31,7 +31,11 @@ spec:
spec:
containers:
- name: postgres
{{- if .Values.global.imageRegistry }}
image: "{{ .Values.global.imageRegistry | trimSuffix "/" }}/{{ .Values.postgres.image.repository }}:{{ .Values.postgres.image.tag }}"
{{- else }}
image: "{{ .Values.postgres.image.repository }}:{{ .Values.postgres.image.tag }}"
{{- end }}
imagePullPolicy: {{ .Values.postgres.image.pullPolicy }}
env:
- name: POSTGRES_HOST_AUTH_METHOD

View File

@@ -4,6 +4,9 @@ fullnameOverride: label-studio
replicaCount: 1
global:
imageRegistry: ""
image:
repository: heartexlabs/label-studio
tag: "latest"

View File

@@ -0,0 +1,188 @@
ARG NODE_VERSION=22
ARG PYTHON_VERSION=3.13
ARG POETRY_VERSION=2.1.4
ARG VERSION_OVERRIDE
ARG BRANCH_OVERRIDE
ARG GIT_REPO=https://github.com/ModelEngine-Group/Label-Studio.git
ARG GIT_BRANCH=ls-release/1.21.0.wsi
################################ Stage: downloader (Source Code)
FROM alpine/git AS downloader
ARG GIT_REPO
ARG GIT_BRANCH
WORKDIR /label-studio
RUN git clone --branch ${GIT_BRANCH} ${GIT_REPO} .
################################ Overview
# This Dockerfile builds a Label Studio environment.
# It consists of five main stages:
# 1. "frontend-builder" - Compiles the frontend assets using Node.
# 2. "frontend-version-generator" - Generates version files for frontend sources.
# 3. "venv-builder" - Prepares the virtualenv environment.
# 4. "py-version-generator" - Generates version files for python sources.
# 5. "prod" - Creates the final production image with the Label Studio, Nginx, and other dependencies.
################################ Stage: frontend-builder (build frontend assets)
FROM --platform=${BUILDPLATFORM} node:${NODE_VERSION}-trixie AS frontend-builder
ENV BUILD_NO_SERVER=true \
BUILD_NO_HASH=true \
BUILD_NO_CHUNKS=true \
BUILD_MODULE=true \
YARN_CACHE_FOLDER=/root/web/.yarn \
NX_CACHE_DIRECTORY=/root/web/.nx \
NODE_ENV=production
WORKDIR /label-studio/web
# Fix Docker Arm64 Build
RUN yarn config set registry https://registry.npmjs.org/
RUN yarn config set network-timeout 1200000 # HTTP timeout used when downloading packages, set to 20 minutes
# [修改] 从 downloader 阶段复制 package.json 和 lock 文件
COPY --from=downloader /label-studio/web/package.json .
COPY --from=downloader /label-studio/web/yarn.lock .
COPY --from=downloader /label-studio/web/tools tools
RUN --mount=type=cache,target=/root/web/.yarn,id=yarn-cache,sharing=locked \
--mount=type=cache,target=/root/web/.nx,id=nx-cache,sharing=locked \
yarn install --prefer-offline --no-progress --pure-lockfile --frozen-lockfile --ignore-engines --non-interactive --production=false
# [修改] 从 downloader 阶段复制前端源码
COPY --from=downloader /label-studio/web/ .
# [修改] 从 downloader 阶段复制 pyproject.toml
COPY --from=downloader /label-studio/pyproject.toml ../pyproject.toml
RUN --mount=type=cache,target=/root/web/.yarn,id=yarn-cache,sharing=locked \
--mount=type=cache,target=/root/web/.nx,id=nx-cache,sharing=locked \
yarn run build
################################ Stage: frontend-version-generator
FROM frontend-builder AS frontend-version-generator
# [修改] 之前是 bind mount 本地 .git,现在需要将 downloader 的 .git 目录复制进来
# 注意:WORKDIR 目前是 /label-studio/web,所以我们将 git 放在上级目录以匹配原逻辑
COPY --from=downloader /label-studio/.git ../.git
RUN --mount=type=cache,target=/root/web/.yarn,id=yarn-cache,sharing=locked \
--mount=type=cache,target=/root/web/.nx,id=nx-cache,sharing=locked \
yarn version:libs
################################ Stage: venv-builder (prepare the virtualenv)
FROM python:${PYTHON_VERSION}-slim-trixie AS venv-builder
ARG POETRY_VERSION
ENV PYTHONUNBUFFERED=1 \
PYTHONDONTWRITEBYTECODE=1 \
PIP_NO_CACHE_DIR=off \
PIP_DISABLE_PIP_VERSION_CHECK=on \
PIP_DEFAULT_TIMEOUT=100 \
PIP_CACHE_DIR="/.cache" \
POETRY_CACHE_DIR="/.poetry-cache" \
POETRY_HOME="/opt/poetry" \
POETRY_VIRTUALENVS_IN_PROJECT=true \
POETRY_VIRTUALENVS_PREFER_ACTIVE_PYTHON=true \
PATH="/opt/poetry/bin:$PATH"
ADD https://install.python-poetry.org /tmp/install-poetry.py
RUN python /tmp/install-poetry.py
RUN --mount=type=cache,target=/var/cache/apt,sharing=locked \
--mount=type=cache,target=/var/lib/apt/lists,sharing=locked \
set -eux; \
apt-get update; \
apt-get install --no-install-recommends -y \
build-essential git; \
apt-get autoremove -y
WORKDIR /label-studio
ENV VENV_PATH="/label-studio/.venv"
ENV PATH="$VENV_PATH/bin:$PATH"
## Starting from this line all packages will be installed in $VENV_PATH
COPY --from=downloader /label-studio/pyproject.toml /label-studio/poetry.lock /label-studio/README.md ./
# Set a default build argument for including dev dependencies
ARG INCLUDE_DEV=false
# Install dependencies
RUN --mount=type=cache,target=/.poetry-cache,id=poetry-cache,sharing=locked \
poetry check --lock && \
if [ "$INCLUDE_DEV" = "true" ]; then \
poetry install --no-root --extras uwsgi --with test; \
else \
poetry install --no-root --without test --extras uwsgi; \
fi
# Install LS
COPY --from=downloader /label-studio/label_studio label_studio
RUN --mount=type=cache,target=/.poetry-cache,id=poetry-cache,sharing=locked \
poetry install --only-root --extras uwsgi && \
python3 label_studio/manage.py collectstatic --no-input
################################ Stage: py-version-generator
FROM venv-builder AS py-version-generator
ARG VERSION_OVERRIDE
ARG BRANCH_OVERRIDE
COPY --from=downloader /label-studio/.git ./.git
# Create version_.py and ls-version_.py
RUN VERSION_OVERRIDE=${VERSION_OVERRIDE} BRANCH_OVERRIDE=${BRANCH_OVERRIDE} poetry run python label_studio/core/version.py
################################### Stage: prod
FROM python:${PYTHON_VERSION}-slim-trixie AS production
ENV LS_DIR=/label-studio \
HOME=/label-studio \
LABEL_STUDIO_BASE_DATA_DIR=/label-studio/data \
OPT_DIR=/opt/heartex/instance-data/etc \
PATH="/label-studio/.venv/bin:$PATH" \
DJANGO_SETTINGS_MODULE=core.settings.label_studio \
PYTHONUNBUFFERED=1 \
PYTHONDONTWRITEBYTECODE=1
WORKDIR $LS_DIR
# install prerequisites for app
RUN --mount=type=cache,target=/var/cache/apt,sharing=locked \
--mount=type=cache,target=/var/lib/apt/lists,sharing=locked \
set -eux; \
apt-get update; \
apt-get upgrade -y; \
apt-get install --no-install-recommends -y libexpat1 libgl1 libglx-mesa0 libglib2.0-0t64 \
gnupg2 curl nginx \
libopenjp2-7 libnuma1 libspeex1 libvdpau1 libgomp1; \
apt-get autoremove -y
RUN set -eux; \
mkdir -p $LS_DIR $LABEL_STUDIO_BASE_DATA_DIR $OPT_DIR && \
chown -R 1001:0 $LS_DIR $LABEL_STUDIO_BASE_DATA_DIR $OPT_DIR /var/log/nginx /etc/nginx
COPY --chown=1001:0 --from=downloader /label-studio/deploy/default.conf /etc/nginx/nginx.conf
# Copy essential files for installing Label Studio and its dependencies
COPY --chown=1001:0 --from=downloader /label-studio/pyproject.toml .
COPY --chown=1001:0 --from=downloader /label-studio/poetry.lock .
COPY --chown=1001:0 --from=downloader /label-studio/README.md .
COPY --chown=1001:0 --from=downloader /label-studio/LICENSE LICENSE
COPY --chown=1001:0 --from=downloader /label-studio/licenses licenses
COPY --chown=1001:0 --from=downloader /label-studio/deploy deploy
# Copy files from build stages
COPY --chown=1001:0 --from=venv-builder $LS_DIR $LS_DIR
COPY --chown=1001:0 --from=py-version-generator $LS_DIR/label_studio/core/version_.py $LS_DIR/label_studio/core/version_.py
COPY --chown=1001:0 --from=frontend-builder $LS_DIR/web/dist $LS_DIR/web/dist
COPY --chown=1001:0 --from=frontend-version-generator $LS_DIR/web/dist/apps/labelstudio/version.json $LS_DIR/web/dist/apps/labelstudio/version.json
COPY --chown=1001:0 --from=frontend-version-generator $LS_DIR/web/dist/libs/editor/version.json $LS_DIR/web/dist/libs/editor/version.json
COPY --chown=1001:0 --from=frontend-version-generator $LS_DIR/web/dist/libs/datamanager/version.json $LS_DIR/web/dist/libs/datamanager/version.json
USER 1001
EXPOSE 8080
ENTRYPOINT ["./deploy/docker-entrypoint.sh"]
CMD ["label-studio"]