Compare commits

...

18 Commits

Author SHA1 Message Date
kurihada 2ea8e0ef87 CI: trigger social-mcp build
Build and Deploy / build-xhh (push) Failing after 15m4s
Build and Deploy / build-xhs (push) Failing after 15m6s
2026-06-02 10:56:40 +08:00
kurihada d91da9d5fd Optimize workflow with proper tool installation
Build and Deploy / build-xhs (push) Has started running
Build and Deploy / build-xhh (push) Has started running
2026-06-01 17:57:59 +08:00
kurihada 9d3a07ef97 Use direct git clone with Docker gateway
Build and Deploy / build-and-deploy (push) Failing after 11m28s
2026-06-01 17:28:47 +08:00
kurihada 5d3aa03694 CI: test with runner config
Build and Deploy / build-and-deploy (push) Has started running
2026-06-01 17:06:37 +08:00
kurihada d71a80d9b6 Retrigger CI
Build and Deploy / build-and-deploy (push) Has started running
2026-06-01 16:59:15 +08:00
kurihada fab07d4288 Simplify Gitea Actions workflow
Build and Deploy / build-and-deploy (push) Waiting to run
2026-06-01 16:58:04 +08:00
kurihada db3079d078 Trigger Gitea Actions workflow
Build and Deploy / build-and-deploy (push) Waiting to run
2026-06-01 16:47:19 +08:00
kurihada 8a63786064 Add Gitea Actions workflow (migrated from Jenkins)
Build and Deploy / build-and-deploy (push) Has started running
2026-06-01 16:35:25 +08:00
kurihada 2cbd6b28b2 重构为Monorepo:拆分xhs/xhh应用与core包并完成双服务部署改造 2026-03-03 16:06:16 +08:00
kurihada ed7fbdd5c2 增强详情与评论MCP:支持URL自动解析并为子评论引入Keyset分页 2026-03-03 13:10:14 +08:00
kurihada 68f42a9fd4 优化通知与评论MCP:评论返回ID并将通知分页升级为Keyset游标 2026-03-03 13:02:35 +08:00
kurihada ceaad6f15b 优化运营MCP:新增状态型点赞收藏、评论幂等与通知游标分页 2026-03-03 12:27:31 +08:00
kurihada a7672d0430 修复类型检查:补充DOM类型库并消除feed-detail严格模式告警 2026-03-03 12:19:16 +08:00
kurihada 892e76f7ed 优化小红书MCP:统一返回结构并增强分页、批量与幂等能力 2026-03-03 12:14:33 +08:00
kurihada 5e0543668e 新增通知任务运营MCP:手动标记、失败列表与失败重试 2026-03-03 11:57:27 +08:00
kurihada 237b528f08 实现通知自动化闭环:SQLite任务状态、定时同步、未处理查询与回复状态回写;下线MCP对外未读通知工具 2026-03-03 11:49:25 +08:00
kurihada d2d96de857 更新 README:同步最新功能并新增中文文档 2026-03-03 11:29:34 +08:00
kurihada eddc8718d0 优化通知工具描述:引导AI回复前先获取文章上下文 2026-03-02 19:26:38 +08:00
87 changed files with 9391 additions and 8252 deletions
+22 -7
View File
@@ -1,14 +1,29 @@
# Server # ============================================================================
PORT=3000 # Common runtime settings (both apps)
# ============================================================================
HOST=127.0.0.1 HOST=127.0.0.1
# Browser
HEADLESS=true HEADLESS=true
# BROWSER_BIN=/path/to/chromium # Optional: custom Chromium binary path # BROWSER_BIN=/path/to/chromium
# Allow remote access (DANGEROUS - only set if you understand the risk) # Required only when exposing HOST=0.0.0.0
# ALLOW_REMOTE=yes-i-understand-the-risk # ALLOW_REMOTE=yes-i-understand-the-risk
# Logging # Optional logging
# NODE_ENV=production # NODE_ENV=production
# LOG_LEVEL=info # LOG_LEVEL=info
# ============================================================================
# mcp-xhs (apps/xhs-mcp)
# ============================================================================
# PORT=9527
# COOKIE_DIR=~/.social-mcp-xhs
# XHS_NOTIFICATION_POLL_ENABLED=true
# XHS_NOTIFICATION_POLL_INTERVAL_SEC=60
# XHS_NOTIFICATION_POLL_MAX_COUNT=20
# ============================================================================
# mcp-xhh (apps/xhh-mcp)
# ============================================================================
# PORT=9528
# COOKIE_DIR=~/.social-mcp-xhh
+73
View File
@@ -0,0 +1,73 @@
name: Build and Deploy
on:
push:
branches: [main]
workflow_dispatch:
jobs:
build-xhs:
runs-on: ubuntu-latest
steps:
- name: Checkout
run: |
git clone http://172.17.0.1:3000/kurihada/social-mcp.git .
git checkout ${{ github.sha }}
- name: Build Docker image
run: docker build --build-arg APP_NAME=xhs-mcp -t social-mcp-xhs:latest .
- name: Deploy
run: |
docker stop mcp-xhs || true
docker rm mcp-xhs || true
mkdir -p /data/mcp-xhs && chown 1001:1001 /data/mcp-xhs
docker run -d \
--name mcp-xhs \
--network nginx \
-p 9527:9527 \
--shm-size=1g \
-v /data/mcp-xhs:/home/appuser/.social-mcp-xhs \
-e NODE_ENV=production \
-e HOST=0.0.0.0 \
-e PORT=9527 \
-e HEADLESS=true \
-e COOKIE_DIR=/home/appuser/.social-mcp-xhs \
-e APP_NAME=xhs-mcp \
-e ALLOW_REMOTE=yes-i-understand-the-risk \
--restart unless-stopped \
social-mcp-xhs:latest
build-xhh:
runs-on: ubuntu-latest
steps:
- name: Checkout
run: |
git clone http://172.17.0.1:3000/kurihada/social-mcp.git .
git checkout ${{ github.sha }}
- name: Build Docker image
run: docker build --build-arg APP_NAME=xhh-mcp -t social-mcp-xhh:latest .
- name: Deploy
run: |
docker stop mcp-xhh || true
docker rm mcp-xhh || true
mkdir -p /data/mcp-xhh && chown 1001:1001 /data/mcp-xhh
docker run -d \
--name mcp-xhh \
--network nginx \
-p 9528:9528 \
--shm-size=1g \
-v /data/mcp-xhh:/home/appuser/.social-mcp-xhh \
-e NODE_ENV=production \
-e HOST=0.0.0.0 \
-e PORT=9528 \
-e HEADLESS=true \
-e COOKIE_DIR=/home/appuser/.social-mcp-xhh \
-e APP_NAME=xhh-mcp \
-e ALLOW_REMOTE=yes-i-understand-the-risk \
--restart unless-stopped \
social-mcp-xhh:latest
+2
View File
@@ -1,8 +1,10 @@
node_modules/ node_modules/
dist/ dist/
**/dist/
web/node_modules/ web/node_modules/
web/dist/ web/dist/
*.tsbuildinfo *.tsbuildinfo
.pnpm-store/
# Environment # Environment
.env .env
+34 -41
View File
@@ -4,57 +4,54 @@
FROM node:22-slim AS builder FROM node:22-slim AS builder
# Proxy for downloading dependencies (passed via --build-arg) ARG APP_NAME=xhs-mcp
ARG HTTP_PROXY ARG HTTP_PROXY
ARG HTTPS_PROXY ARG HTTPS_PROXY
ENV HTTP_PROXY=${HTTP_PROXY} \ ENV HTTP_PROXY=${HTTP_PROXY} \
HTTPS_PROXY=${HTTPS_PROXY} HTTPS_PROXY=${HTTPS_PROXY}
# China npm mirror RUN corepack enable
RUN npm config set registry https://registry.npmmirror.com RUN pnpm config set registry https://registry.npmmirror.com
WORKDIR /app WORKDIR /app
# Copy package manifests first (layer caching for dependency install) # Copy manifests first for better caching
COPY package.json package-lock.json ./ COPY pnpm-workspace.yaml package.json pnpm-lock.yaml tsconfig.json tsconfig.base.json ./
COPY packages/core/package.json packages/core/tsconfig.json packages/core/tsup.config.ts ./packages/core/
COPY apps/xhs-mcp/package.json apps/xhs-mcp/tsconfig.json apps/xhs-mcp/tsup.config.ts ./apps/xhs-mcp/
COPY apps/xhh-mcp/package.json apps/xhh-mcp/tsconfig.json apps/xhh-mcp/tsup.config.ts ./apps/xhh-mcp/
# Install all dependencies (including devDependencies for building) RUN pnpm install --frozen-lockfile
RUN npm ci
# Install Chromium matching rebrowser-playwright version (NOT playwright) # Install Chromium matching rebrowser-playwright version (NOT playwright)
RUN npx rebrowser-playwright install chromium RUN npx rebrowser-playwright install chromium
# Copy source code # Copy source code
COPY tsconfig.json tsup.config.ts ./ COPY packages ./packages
COPY src/ src/ COPY apps ./apps
# Build the backend # Build shared core first, then selected app
RUN npm run build RUN pnpm --filter @social/core build
RUN pnpm --filter @social/${APP_NAME} build
# Build the web dashboard
COPY web/ web/
RUN cd web && npm ci && npm run build && mkdir -p ../dist/web && cp -r dist/* ../dist/web/
# Remove devDependencies to slim down node_modules for production
RUN npm prune --omit=dev
# ============================================================================= # =============================================================================
# Stage 2: Production # Stage 2: Runtime
# ============================================================================= # =============================================================================
FROM node:22-slim FROM node:22-slim
# Use China apt mirror ARG APP_NAME=xhs-mcp
RUN sed -i 's|deb.debian.org|mirrors.aliyun.com|g' /etc/apt/sources.list.d/debian.sources 2>/dev/null || \
sed -i 's|deb.debian.org|mirrors.aliyun.com|g' /etc/apt/sources.list 2>/dev/null || true
# Proxy for apt-get (passed via --build-arg)
ARG HTTP_PROXY ARG HTTP_PROXY
ARG HTTPS_PROXY ARG HTTPS_PROXY
ENV HTTP_PROXY=${HTTP_PROXY} \ ENV HTTP_PROXY=${HTTP_PROXY} \
HTTPS_PROXY=${HTTPS_PROXY} HTTPS_PROXY=${HTTPS_PROXY}
# Install Chromium dependencies required by Playwright/rebrowser-playwright RUN sed -i 's|deb.debian.org|mirrors.aliyun.com|g' /etc/apt/sources.list.d/debian.sources 2>/dev/null || \
sed -i 's|deb.debian.org|mirrors.aliyun.com|g' /etc/apt/sources.list 2>/dev/null || true
# Chromium runtime dependencies
RUN apt-get update && apt-get install -y --no-install-recommends \ RUN apt-get update && apt-get install -y --no-install-recommends \
libnss3 \ libnss3 \
libnspr4 \ libnspr4 \
@@ -77,41 +74,37 @@ RUN apt-get update && apt-get install -y --no-install-recommends \
fonts-noto-cjk \ fonts-noto-cjk \
&& rm -rf /var/lib/apt/lists/* && rm -rf /var/lib/apt/lists/*
# Create non-root user
RUN groupadd --gid 1001 appuser \ RUN groupadd --gid 1001 appuser \
&& useradd --uid 1001 --gid appuser --shell /bin/sh --create-home appuser && useradd --uid 1001 --gid appuser --shell /bin/sh --create-home appuser
WORKDIR /app WORKDIR /app
# Copy built artifacts and production dependencies from builder
COPY --from=builder --chown=appuser:appuser /app/dist ./dist
COPY --from=builder --chown=appuser:appuser /app/node_modules ./node_modules
COPY --from=builder --chown=appuser:appuser /app/package.json ./package.json COPY --from=builder --chown=appuser:appuser /app/package.json ./package.json
COPY --from=builder --chown=appuser:appuser /app/pnpm-workspace.yaml ./pnpm-workspace.yaml
COPY --from=builder --chown=appuser:appuser /app/node_modules ./node_modules
COPY --from=builder --chown=appuser:appuser /app/packages ./packages
COPY --from=builder --chown=appuser:appuser /app/apps ./apps
# Copy Playwright browsers from builder
COPY --from=builder --chown=appuser:appuser /root/.cache/ms-playwright /home/appuser/.cache/ms-playwright COPY --from=builder --chown=appuser:appuser /root/.cache/ms-playwright /home/appuser/.cache/ms-playwright
# Create data directory for cookies and API token RUN mkdir -p /home/appuser/.social-mcp-xhs /home/appuser/.social-mcp-xhh \
RUN mkdir -p /home/appuser/.social-mcp \ && chown -R appuser:appuser /home/appuser/.social-mcp-xhs /home/appuser/.social-mcp-xhh
&& chown -R appuser:appuser /home/appuser/.social-mcp
# Switch to non-root user
USER appuser USER appuser
# Environment defaults
# Clear proxy env from build stage (must not leak into runtime)
ENV HTTP_PROXY= \ ENV HTTP_PROXY= \
HTTPS_PROXY= \ HTTPS_PROXY= \
NODE_ENV=production \ NODE_ENV=production \
HOST=0.0.0.0 \ HOST=0.0.0.0 \
PORT=3000 \ PORT=9527 \
HEADLESS=true \ HEADLESS=true \
COOKIE_DIR=/home/appuser/.social-mcp \ COOKIE_DIR=/home/appuser/.social-mcp-xhs \
APP_NAME=${APP_NAME} \
ALLOW_REMOTE=yes-i-understand-the-risk ALLOW_REMOTE=yes-i-understand-the-risk
EXPOSE 3000 EXPOSE 9527
HEALTHCHECK --interval=30s --timeout=5s --retries=3 --start-period=10s \ HEALTHCHECK --interval=30s --timeout=5s --retries=3 --start-period=10s \
CMD node -e "fetch('http://localhost:3000/health').then(r => process.exit(r.ok ? 0 : 1)).catch(() => process.exit(1))" CMD node -e "fetch('http://localhost:' + (process.env.PORT || '9527') + '/health').then(r => process.exit(r.ok ? 0 : 1)).catch(() => process.exit(1))"
CMD ["node", "dist/index.js"] CMD ["sh", "-lc", "node apps/${APP_NAME}/dist/main.js"]
Vendored
+41 -14
View File
@@ -2,7 +2,10 @@ pipeline {
agent any agent any
environment { environment {
APP_NAME = 'social-mcp' IMAGE_XHS = 'social-mcp-xhs'
IMAGE_XHH = 'social-mcp-xhh'
APP_XHS = 'mcp-xhs'
APP_XHH = 'mcp-xhh'
} }
triggers { triggers {
@@ -16,33 +19,57 @@ pipeline {
} }
} }
stage('Build Docker Image') { stage('Build Docker Images') {
steps { steps {
sh "docker build --build-arg HTTP_PROXY=http://172.17.0.1:7897 --build-arg HTTPS_PROXY=http://172.17.0.1:7897 -t ${APP_NAME}:${BUILD_NUMBER} -t ${APP_NAME}:latest ." sh "docker build --build-arg APP_NAME=xhs-mcp --build-arg HTTP_PROXY=http://172.17.0.1:7897 --build-arg HTTPS_PROXY=http://172.17.0.1:7897 -t ${IMAGE_XHS}:${BUILD_NUMBER} -t ${IMAGE_XHS}:latest ."
sh "docker build --build-arg APP_NAME=xhh-mcp --build-arg HTTP_PROXY=http://172.17.0.1:7897 --build-arg HTTPS_PROXY=http://172.17.0.1:7897 -t ${IMAGE_XHH}:${BUILD_NUMBER} -t ${IMAGE_XHH}:latest ."
} }
} }
stage('Deploy') { stage('Deploy') {
steps { steps {
sh """ sh """
docker stop ${APP_NAME} || true docker stop ${APP_XHS} || true
docker rm ${APP_NAME} || true docker rm ${APP_XHS} || true
mkdir -p /data/${APP_NAME} docker stop ${APP_XHH} || true
chown 1001:1001 /data/${APP_NAME} docker rm ${APP_XHH} || true
mkdir -p /data/${APP_XHS}
mkdir -p /data/${APP_XHH}
chown 1001:1001 /data/${APP_XHS}
chown 1001:1001 /data/${APP_XHH}
docker run -d \ docker run -d \
--name ${APP_NAME} \ --name ${APP_XHS} \
--network nginx \ --network nginx \
-p 3010:3000 \ -p 9527:9527 \
--shm-size=1g \ --shm-size=1g \
-v /data/${APP_NAME}:/home/appuser/.social-mcp \ -v /data/${APP_XHS}:/home/appuser/.social-mcp-xhs \
-e NODE_ENV=production \ -e NODE_ENV=production \
-e HOST=0.0.0.0 \ -e HOST=0.0.0.0 \
-e PORT=3000 \ -e PORT=9527 \
-e HEADLESS=true \ -e HEADLESS=true \
-e COOKIE_DIR=/home/appuser/.social-mcp \ -e COOKIE_DIR=/home/appuser/.social-mcp-xhs \
-e APP_NAME=xhs-mcp \
-e ALLOW_REMOTE=yes-i-understand-the-risk \ -e ALLOW_REMOTE=yes-i-understand-the-risk \
--restart unless-stopped \ --restart unless-stopped \
${APP_NAME}:latest ${IMAGE_XHS}:latest
docker run -d \
--name ${APP_XHH} \
--network nginx \
-p 9528:9528 \
--shm-size=1g \
-v /data/${APP_XHH}:/home/appuser/.social-mcp-xhh \
-e NODE_ENV=production \
-e HOST=0.0.0.0 \
-e PORT=9528 \
-e HEADLESS=true \
-e COOKIE_DIR=/home/appuser/.social-mcp-xhh \
-e APP_NAME=xhh-mcp \
-e ALLOW_REMOTE=yes-i-understand-the-risk \
--restart unless-stopped \
${IMAGE_XHH}:latest
""" """
} }
} }
@@ -50,7 +77,7 @@ pipeline {
post { post {
success { success {
echo "Deployed ${APP_NAME} build #${BUILD_NUMBER} successfully" echo "Deployed ${APP_XHS} and ${APP_XHH} build #${BUILD_NUMBER} successfully"
} }
failure { failure {
echo "Build #${BUILD_NUMBER} failed" echo "Build #${BUILD_NUMBER} failed"
+91 -211
View File
@@ -1,247 +1,127 @@
# Social MCP # Social Auto Hub (Monorepo)
Multi-platform social media automation service that exposes browser-based actions as both MCP (Model Context Protocol) tools and a REST API. Currently supports **Xiaohongshu** (Little Red Book). [中文文档](./README.zh-CN.md)
## Features This repository is now a **workspace monorepo** with clear visual separation:
- **13 MCP tools** for Xiaohongshu: login management, content browsing, publishing, and interactions - `apps/xhs-mcp`: Xiaohongshu MCP service (`xhs_*` tools)
- **REST API** with Bearer token authentication and rate limiting - `apps/xhh-mcp`: Xiaoheihe MCP service (`xhh_*` tools)
- **Browser automation** via rebrowser-playwright with anti-detection patches - `packages/core`: shared infrastructure (`browser/config/cookie/server/utils`)
- **Cookie persistence** with file-based storage (0600 permissions, atomic writes)
- **Security**: DNS rebinding protection, Host header validation, error message sanitization, log redaction ## Workspace Layout
- **Docker support** with hardened configuration (non-root user, read-only filesystem, resource limits)
- **Plugin architecture** for adding new platforms ```text
apps/
xhs-mcp/
xhh-mcp/
packages/
core/
```
## Quick Start ## Quick Start
### Prerequisites ### Prerequisites
- Node.js >= 22.0.0 - Node.js >= 22
- pnpm - pnpm
### Install and Run ### Install
```bash ```bash
# Install dependencies
pnpm install pnpm install
npx rebrowser-playwright install chromium
# Install Playwright browsers (first time only)
npx playwright install chromium
# Build
pnpm build
# Start the server
pnpm start
``` ```
The server starts on `http://127.0.0.1:3000` by default. A REST API Bearer token is printed to the console on first startup and saved to `~/.social-mcp/.api-token`. ### Build all packages
### Development
```bash ```bash
# Watch mode (rebuilds on file changes) pnpm build
pnpm dev
# Type check without emitting
pnpm lint
# Run tests
pnpm test
``` ```
## MCP Integration ### Run XHS service
### Claude Desktop ```bash
pnpm start:xhs
```
Add the following to your Claude Desktop configuration file (`claude_desktop_config.json`): Default runtime values:
- `PORT=9527`
- `COOKIE_DIR=~/.social-mcp-xhs`
### Run XHH service
```bash
pnpm start:xhh
```
Default runtime values:
- `PORT=9528`
- `COOKIE_DIR=~/.social-mcp-xhh`
## Compatibility Root Scripts
Root scripts are kept as forwarding wrappers:
- `pnpm build`
- `pnpm lint`
- `pnpm test`
- `pnpm start:xhs`
- `pnpm start:xhh`
- `pnpm dev:xhs`
- `pnpm dev:xhh`
You can also run package-level commands directly:
```bash
pnpm --filter @social/core build
pnpm --filter @social/xhs-mcp start
pnpm --filter @social/xhh-mcp start
```
## MCP Endpoints
- XHS MCP: `http://127.0.0.1:9527/mcp`
- XHH MCP: `http://127.0.0.1:9528/mcp`
## REST Endpoints
- XHS REST: `http://127.0.0.1:9527/api/xhs/*`
- XHH REST: `http://127.0.0.1:9528/api/xhh/*`
Each service uses its own bearer token file under its `COOKIE_DIR`.
## Claude Desktop Example
```json ```json
{ {
"mcpServers": { "mcpServers": {
"social-mcp": { "mcp-xhs": {
"url": "http://127.0.0.1:3000/sse" "url": "http://127.0.0.1:9527/mcp"
},
"mcp-xhh": {
"url": "http://127.0.0.1:9528/mcp"
} }
} }
} }
``` ```
### Available MCP Tools ## Docker
| Tool | Description | Single Dockerfile supports dual app targets via build arg:
|------|-------------|
| `xhs_check_login` | Check Xiaohongshu login status |
| `xhs_get_login_qrcode` | Get login QR code for phone scanning |
| `xhs_delete_cookies` | Delete cookies and reset login session |
| `xhs_list_feeds` | Get explore page recommended feed list |
| `xhs_search` | Search notes by keyword with filters |
| `xhs_get_feed_detail` | Get note detail with content, images, stats, comments |
| `xhs_get_user_profile` | Get user profile with bio, stats, recent notes |
| `xhs_publish_image` | Publish an image note |
| `xhs_publish_video` | Publish a video note |
| `xhs_post_comment` | Post a comment on a note |
| `xhs_reply_comment` | Reply to a comment |
| `xhs_like` | Like or unlike a note |
| `xhs_favorite` | Favorite or unfavorite a note |
## REST API - `APP_NAME=xhs-mcp`
- `APP_NAME=xhh-mcp`
All REST endpoints require a `Bearer` token in the `Authorization` header. The token is generated on first startup and printed to the console. Compose files already define both services:
```bash - `docker-compose.yml`
# Example: check login status - `deploy/docker-compose.yml`
curl -H "Authorization: Bearer <token>" http://127.0.0.1:3000/api/xhs/login/status
# Example: search notes ## Migration Notes
curl -X POST -H "Authorization: Bearer <token>" \
-H "Content-Type: application/json" \
-d '{"keyword": "travel", "filters": {"sort": "popularity_descending"}}' \
http://127.0.0.1:3000/api/xhs/search
```
### Endpoints - Old root runtime entry (`src/index.ts`) is removed.
- Service code now lives inside `apps/*`.
| Method | Path | Description | Rate Limit | - Shared runtime logic now lives inside `packages/core`.
|--------|------|-------------|------------| - Lockfile strategy is pnpm-only (`pnpm-lock.yaml`).
| `GET` | `/api/xhs/login/status` | Check login status | 60/min |
| `GET` | `/api/xhs/login/qrcode` | Get login QR code | 60/min |
| `DELETE` | `/api/xhs/login/cookies` | Delete cookies | 10/min |
| `GET` | `/api/xhs/feeds` | Get recommended feeds | 60/min |
| `POST` | `/api/xhs/search` | Search notes | 60/min |
| `POST` | `/api/xhs/feeds/detail` | Get note detail | 60/min |
| `POST` | `/api/xhs/user/profile` | Get user profile | 60/min |
| `POST` | `/api/xhs/publish/image` | Publish image note | 10/min |
| `POST` | `/api/xhs/publish/video` | Publish video note | 10/min |
| `POST` | `/api/xhs/comment` | Post a comment | 10/min |
| `POST` | `/api/xhs/comment/reply` | Reply to a comment | 10/min |
| `POST` | `/api/xhs/like` | Like/unlike a note | 10/min |
| `POST` | `/api/xhs/favorite` | Favorite/unfavorite a note | 10/min |
### Response Format
All REST responses follow a consistent JSON format:
```json
// Success
{
"success": true,
"data": { ... }
}
// Error
{
"success": false,
"error": {
"code": "VALIDATION_ERROR",
"message": "keyword: Required"
}
}
```
### Other Endpoints (no auth required)
| Method | Path | Description |
|--------|------|-------------|
| `GET` | `/health` | Health check (memory, uptime, plugin status) |
| `GET` | `/sse` | MCP SSE transport |
| `POST` | `/messages` | MCP JSON-RPC messages |
## Docker Deployment
### Using Docker Compose (recommended)
```bash
cd deploy
docker compose up -d
# View logs
docker compose logs -f
# The API token is printed in the logs on first start
docker compose logs social-mcp | grep "Bearer Token"
```
### Using Docker directly
```bash
# Build the image
docker build -t social-mcp .
# Run with required settings
docker run -d \
--name social-mcp \
-p 127.0.0.1:3000:3000 \
--shm-size=1gb \
--memory=2g \
--cpus=2.0 \
--security-opt=no-new-privileges:true \
--cap-drop=ALL \
--read-only \
--tmpfs /tmp:size=512m \
-v social-mcp-data:/home/appuser/.social-mcp \
social-mcp
```
**Important**: The `--shm-size=1gb` flag is required. Chromium uses `/dev/shm` for shared memory and the default 64MB causes crashes.
## Environment Variables
| Variable | Default | Description |
|----------|---------|-------------|
| `PORT` | `3000` | HTTP server port |
| `HOST` | `127.0.0.1` | Bind address (`0.0.0.0` requires `ALLOW_REMOTE`) |
| `HEADLESS` | `true` | Run browser in headless mode |
| `BROWSER_BIN` | (auto) | Custom Chromium executable path |
| `LOG_LEVEL` | `info` | Pino log level (`debug`, `info`, `warn`, `error`) |
| `NODE_ENV` | `development` | Environment (`production` disables pretty logs) |
| `COOKIE_DIR` | `~/.social-mcp` | Directory for cookie and token storage |
| `MAX_QUEUE_DEPTH` | `10` | Max pending operations per platform queue |
| `ALLOW_REMOTE` | (unset) | Set to `yes-i-understand-the-risk` to allow `HOST=0.0.0.0` |
## Project Structure
```
social-mcp/
├── package.json
├── tsconfig.json
├── tsup.config.ts
├── Dockerfile
├── deploy/
│ └── docker-compose.yml
├── src/
│ ├── index.ts # Entry point: bootstrap, plugin registration, graceful shutdown
│ ├── server/
│ │ ├── app.ts # AppServer: Express + MCP lifecycle
│ │ └── middleware.ts # DNS rebinding guard, bearer auth, rate limiter, error handler
│ ├── browser/
│ │ └── manager.ts # BrowserManager: browser lifecycle, serial queues, backpressure
│ ├── cookie/
│ │ └── store.ts # CookieStore: per-platform cookie persistence (0600, atomic writes)
│ ├── config/
│ │ └── index.ts # Environment-based configuration
│ ├── utils/
│ │ ├── logger.ts # Pino logger with deep redaction
│ │ ├── errors.ts # Error classification, sanitization, MCP error wrapper
│ │ └── downloader.ts # Media file download and path validation
│ └── platforms/
│ └── xiaohongshu/
│ ├── index.ts # PlatformPlugin: MCP tool + REST route registration
│ ├── routes.ts # REST API route handlers
│ ├── schemas.ts # Zod schemas for tool/API parameter validation
│ ├── types.ts # Domain types (Feed, Comment, UserProfile, etc.)
│ ├── selectors.ts # CSS selector constants
│ ├── login.ts # Login management (QR code, status check)
│ ├── feeds.ts # Explore page feed extraction
│ ├── search.ts # Search with filters
│ ├── feed-detail.ts # Note detail + comment loading
│ ├── user-profile.ts # User profile extraction
│ ├── publish.ts # Image note publishing
│ ├── publish-video.ts # Video note publishing
│ ├── comment.ts # Comment and reply posting
│ └── interaction.ts # Like and favorite toggling
└── tests/
```
## License
ISC
+124
View File
@@ -0,0 +1,124 @@
# Social Auto HubMonorepo
[English](./README.md)
仓库已改造为 **workspace monorepo**,结构上清晰拆分为:
- `apps/xhs-mcp`:小红书 MCP 服务(`xhs_*`
- `apps/xhh-mcp`:小黑盒 MCP 服务(`xhh_*`
- `packages/core`:共享基础设施(`browser/config/cookie/server/utils`
## 目录结构
```text
apps/
xhs-mcp/
xhh-mcp/
packages/
core/
```
## 快速开始
### 前置要求
- Node.js >= 22
- pnpm
### 安装依赖
```bash
pnpm install
npx rebrowser-playwright install chromium
```
### 构建全部包
```bash
pnpm build
```
### 启动小红书服务
```bash
pnpm start:xhs
```
默认运行参数:
- `PORT=9527`
- `COOKIE_DIR=~/.social-mcp-xhs`
### 启动小黑盒服务
```bash
pnpm start:xhh
```
默认运行参数:
- `PORT=9528`
- `COOKIE_DIR=~/.social-mcp-xhh`
## 根目录兼容脚本
根脚本保留为转发壳:
- `pnpm build`
- `pnpm lint`
- `pnpm test`
- `pnpm start:xhs`
- `pnpm start:xhh`
- `pnpm dev:xhs`
- `pnpm dev:xhh`
也可以直接按包执行:
```bash
pnpm --filter @social/core build
pnpm --filter @social/xhs-mcp start
pnpm --filter @social/xhh-mcp start
```
## MCP 地址
- XHS MCP: `http://127.0.0.1:9527/mcp`
- XHH MCP: `http://127.0.0.1:9528/mcp`
## REST 地址
- XHS REST: `http://127.0.0.1:9527/api/xhs/*`
- XHH REST: `http://127.0.0.1:9528/api/xhh/*`
两个服务使用各自 `COOKIE_DIR` 下的独立 Bearer token。
## Claude Desktop 接入示例
```json
{
"mcpServers": {
"mcp-xhs": {
"url": "http://127.0.0.1:9527/mcp"
},
"mcp-xhh": {
"url": "http://127.0.0.1:9528/mcp"
}
}
}
```
## Docker
单一 Dockerfile 支持通过 `APP_NAME` 构建双目标:
- `APP_NAME=xhs-mcp`
- `APP_NAME=xhh-mcp`
`docker-compose.yml``deploy/docker-compose.yml` 已配置双服务。
## 迁移说明
- 旧的根入口(`src/index.ts`)已移除。
- 服务代码迁移到 `apps/*`
- 共享基础能力迁移到 `packages/core`
- 锁文件策略改为 pnpm`pnpm-lock.yaml`)。
+29
View File
@@ -0,0 +1,29 @@
{
"name": "@social/xhh-mcp",
"version": "0.1.0",
"private": true,
"type": "module",
"main": "dist/main.js",
"bin": {
"mcp-xhh": "dist/main.js"
},
"scripts": {
"build": "tsup",
"lint": "tsc --noEmit",
"test": "vitest run",
"start": "PORT=${PORT:-9528} COOKIE_DIR=${COOKIE_DIR:-$HOME/.social-mcp-xhh} node dist/main.js",
"dev": "pnpm build && pnpm start"
},
"dependencies": {
"@modelcontextprotocol/sdk": "^1.27.0",
"@social/core": "workspace:*",
"zod": "^3.25.0"
},
"devDependencies": {
"@types/node": "^22.0.0",
"tsup": "^8.0.0",
"typescript": "^5.7.0",
"vitest": "^3.0.0"
}
}
+5
View File
@@ -0,0 +1,5 @@
import { startServerWithPlugins } from '@social/core/server/bootstrap.js';
import { xiaoheihePlugin } from './platforms/xiaoheihe/index.js';
startServerWithPlugins([xiaoheihePlugin]);
@@ -0,0 +1,187 @@
import type { Page } from 'rebrowser-playwright';
import { logger } from '@social/core/utils/logger.js';
import { XHH_SELECTORS } from './selectors.js';
import { detectCaptchaText } from './extractors.js';
const log = logger.child({ module: 'xhh-comment' });
function buildDetailUrl(linkId: string): string {
return `https://www.xiaoheihe.cn/app/bbs/link/${encodeURIComponent(linkId)}`;
}
export async function postComment(
page: Page,
linkId: string,
content: string,
): Promise<{ success: boolean; comment_id?: string }> {
await page.goto(buildDetailUrl(linkId), { waitUntil: 'domcontentloaded' });
await page.waitForTimeout(1_000);
const text = await page.textContent('body').catch(() => '');
if (text && detectCaptchaText(text)) {
throw new Error('CAPTCHA_REQUIRED: captcha detected on comment page');
}
const ok = await fillCommentInput(page, content);
if (!ok) return { success: false };
const submitted = await clickFirstVisible(page, XHH_SELECTORS.detail.commentSubmit);
if (!submitted) return { success: false };
await page.waitForTimeout(1_500);
const commentId = await page.evaluate(
({ selectors, contentLike }: { selectors: typeof XHH_SELECTORS; contentLike: string }) => {
const nodes = [...document.querySelectorAll<HTMLElement>(selectors.detail.commentItem.join(', '))];
const hit = nodes.find((node) => node.textContent?.includes(contentLike));
if (!hit) return '';
return (
hit.getAttribute('data-comment-id') ||
hit.getAttribute('comment-id') ||
hit.id ||
''
);
},
{ selectors: XHH_SELECTORS, contentLike: content.slice(0, 24) },
);
return {
success: true,
...(commentId ? { comment_id: commentId as string } : {}),
};
}
export async function replyComment(
page: Page,
linkId: string,
commentId: string,
content: string,
): Promise<{ success: boolean; reply_id?: string }> {
await page.goto(buildDetailUrl(linkId), { waitUntil: 'domcontentloaded' });
await page.waitForTimeout(1_000);
const text = await page.textContent('body').catch(() => '');
if (text && detectCaptchaText(text)) {
throw new Error('CAPTCHA_REQUIRED: captcha detected on reply page');
}
await page.evaluate(
({ selectors, targetCommentId }) => {
const comments = [...document.querySelectorAll<HTMLElement>(selectors.detail.commentItem.join(', '))];
const target = comments.find((node) => {
const id =
node.getAttribute('data-comment-id') ||
node.getAttribute('comment-id') ||
node.id ||
'';
if (id === targetCommentId) return true;
return node.outerHTML.includes(targetCommentId);
});
if (!target) return;
const replyBtn = [...target.querySelectorAll<HTMLElement>('button, [role="button"], .reply-btn, .comment-reply')]
.find((node) => {
const text = (node.textContent ?? '').trim();
const cls = node.className.toString().toLowerCase();
return text.includes('回复') || cls.includes('reply');
}) ?? null;
replyBtn?.click();
},
{ selectors: XHH_SELECTORS, targetCommentId: commentId },
);
await page.waitForTimeout(500);
const ok = await fillCommentInput(page, content);
if (!ok) return { success: false };
const submitted = await clickFirstVisible(page, XHH_SELECTORS.detail.commentSubmit);
if (!submitted) return { success: false };
await page.waitForTimeout(1_500);
const replyId = await page.evaluate(
(
{
selectors,
targetCommentId,
contentLike,
}: { selectors: typeof XHH_SELECTORS; targetCommentId: string; contentLike: string },
) => {
const comments = [...document.querySelectorAll<HTMLElement>(selectors.detail.commentItem.join(', '))];
const target = comments.find((node) => {
const id =
node.getAttribute('data-comment-id') ||
node.getAttribute('comment-id') ||
node.id ||
'';
if (id === targetCommentId) return true;
return node.outerHTML.includes(targetCommentId);
});
if (!target) return '';
const replies = [...target.querySelectorAll<HTMLElement>(selectors.detail.subCommentItem.join(', '))];
const hit = replies.find((node) => node.textContent?.includes(contentLike));
if (!hit) return '';
return (
hit.getAttribute('data-comment-id') ||
hit.getAttribute('comment-id') ||
hit.id ||
''
);
},
{ selectors: XHH_SELECTORS, targetCommentId: commentId, contentLike: content.slice(0, 24) },
);
return {
success: true,
...(replyId ? { reply_id: replyId as string } : {}),
};
}
async function fillCommentInput(page: Page, content: string): Promise<boolean> {
for (const selector of XHH_SELECTORS.detail.commentInput) {
const input = await page.$(selector).catch(() => null);
if (!input) continue;
await input.click().catch(() => {});
await page.waitForTimeout(200);
const isOk = await page
.evaluate(
({ selector, content }) => {
const node = document.querySelector(selector);
if (!node) return false;
if (node instanceof HTMLTextAreaElement || node instanceof HTMLInputElement) {
node.value = content;
node.dispatchEvent(new Event('input', { bubbles: true }));
return true;
}
if (node instanceof HTMLElement && node.isContentEditable) {
node.focus();
node.textContent = content;
node.dispatchEvent(new Event('input', { bubbles: true }));
return true;
}
return false;
},
{ selector, content },
)
.catch(() => false);
if (isOk) return true;
}
return false;
}
async function clickFirstVisible(page: Page, selectors: readonly string[]): Promise<boolean> {
for (const selector of selectors) {
const clicked = await page
.locator(selector)
.first()
.click({ timeout: 2_000 })
.then(() => true)
.catch(() => false);
if (clicked) return true;
}
log.warn({ selectors }, 'no clickable submit button');
return false;
}
@@ -0,0 +1,66 @@
export interface KeysetCursorPayload {
key: string;
}
export interface KeysetPage<T> {
items: T[];
hasMore: boolean;
nextCursor?: string;
}
export function encodeKeysetCursor(payload: KeysetCursorPayload): string {
return Buffer.from(JSON.stringify(payload), 'utf8').toString('base64url');
}
export function decodeKeysetCursor(cursor?: string): KeysetCursorPayload | undefined {
if (!cursor) return undefined;
try {
const raw = JSON.parse(Buffer.from(cursor, 'base64url').toString('utf8')) as {
key?: unknown;
};
if (typeof raw.key !== 'string' || raw.key.length === 0) {
throw new Error('Invalid keyset cursor payload');
}
return { key: raw.key };
} catch {
throw new Error('Invalid cursor for keyset pagination');
}
}
export function paginateByKeyset<T>(
items: T[],
maxCount: number,
cursor: KeysetCursorPayload | undefined,
keyOf: (item: T) => string,
): KeysetPage<T> {
if (maxCount <= 0) {
return { items: [], hasMore: false };
}
const start = cursor
? Math.max(0, items.findIndex((item) => keyOf(item) === cursor.key) + 1)
: 0;
const pageItems = items.slice(start, start + maxCount);
const hasMore = start + pageItems.length < items.length;
if (!hasMore || pageItems.length === 0) {
return {
items: pageItems,
hasMore,
};
}
const nextCursor = encodeKeysetCursor({
key: keyOf(pageItems[pageItems.length - 1]!),
});
return {
items: pageItems,
hasMore,
nextCursor,
};
}
@@ -0,0 +1,110 @@
import type { Feed } from './types.js';
export function parseCountString(raw: string | number | null | undefined): number {
if (typeof raw === 'number') {
return Number.isFinite(raw) ? raw : 0;
}
const text = (raw ?? '').toString().trim().replace(/,/g, '');
if (!text) return 0;
if (text.endsWith('万')) {
const num = Number.parseFloat(text.slice(0, -1));
if (Number.isNaN(num)) return 0;
return Math.round(num * 10_000);
}
const intNum = Number.parseInt(text, 10);
return Number.isNaN(intNum) ? 0 : intNum;
}
export function detectCaptchaText(text: string): boolean {
const haystack = text.toLowerCase();
return (
haystack.includes('captcha') ||
haystack.includes('show_captcha') ||
haystack.includes('验证码') ||
haystack.includes('tencentcaptcha')
);
}
export function extractLinkIdFromUrl(rawUrl: string): string | undefined {
const trimmed = rawUrl.trim();
if (!trimmed) return undefined;
try {
const url = /^https?:\/\//i.test(trimmed)
? new URL(trimmed)
: trimmed.startsWith('/')
? new URL(`https://www.xiaoheihe.cn${trimmed}`)
: new URL(`https://${trimmed}`);
const pathMatch = url.pathname.match(/\/app\/bbs\/link\/(\d+)/);
if (pathMatch?.[1]) return pathMatch[1];
const queryLinkId = url.searchParams.get('link_id') ?? url.searchParams.get('linkid');
return queryLinkId || undefined;
} catch {
return undefined;
}
}
export function extractUserIdFromUrl(rawUrl: string): string | undefined {
const trimmed = rawUrl.trim();
if (!trimmed) return undefined;
try {
const url = /^https?:\/\//i.test(trimmed)
? new URL(trimmed)
: trimmed.startsWith('/')
? new URL(`https://www.xiaoheihe.cn${trimmed}`)
: new URL(`https://${trimmed}`);
const pathMatch = url.pathname.match(/\/app\/user\/profile\/(\d+)/);
if (pathMatch?.[1]) return pathMatch[1];
const queryUserId = url.searchParams.get('userid') ?? url.searchParams.get('user_id');
return queryUserId || undefined;
} catch {
return undefined;
}
}
export function parseFeedsFromHtmlSnapshot(html: string): Feed[] {
const matches = [...html.matchAll(/href="(\/app\/bbs\/link\/\d+)"/g)];
const seen = new Set<string>();
const feeds: Feed[] = [];
for (const m of matches) {
const href = m[1];
if (!href) continue;
const id = extractLinkIdFromUrl(href);
if (!id || seen.has(id)) continue;
seen.add(id);
feeds.push({
id,
title: '',
description: '',
coverUrl: '',
likeCount: 0,
commentCount: 0,
user: {
id: '',
nickname: '',
avatar: '',
},
linkUrl: `https://www.xiaoheihe.cn${href}`,
});
}
return feeds;
}
export function firstNonEmpty(...values: Array<string | null | undefined>): string {
for (const value of values) {
const trimmed = value?.trim();
if (trimmed) return trimmed;
}
return '';
}
@@ -0,0 +1,266 @@
import type { Page } from 'rebrowser-playwright';
import { logger } from '@social/core/utils/logger.js';
import { XHH_SELECTORS } from './selectors.js';
import type { Comment, FeedDetail } from './types.js';
import { detectCaptchaText, firstNonEmpty, parseCountString } from './extractors.js';
const log = logger.child({ module: 'xhh-feed-detail' });
function buildDetailUrl(linkId: string): string {
return `https://www.xiaoheihe.cn/app/bbs/link/${encodeURIComponent(linkId)}`;
}
interface RawComment {
id: string;
parentId?: string;
userId: string;
nickname: string;
avatar: string;
content: string;
likeCount: string | number;
createTime: string;
subComments: RawComment[];
}
interface RawDetail {
title: string;
description: string;
images: string[];
likeCount: string | number;
favoriteCount: string | number;
commentCount: string | number;
isLiked: boolean;
isFavorited: boolean;
userId: string;
nickname: string;
avatar: string;
comments: RawComment[];
}
export async function getFeedDetail(page: Page, linkId: string): Promise<FeedDetail> {
const url = buildDetailUrl(linkId);
await page.goto(url, { waitUntil: 'domcontentloaded' });
await page.waitForTimeout(1_200);
const bodyText = await page.textContent('body').catch(() => '');
if (bodyText && detectCaptchaText(bodyText)) {
throw new Error('CAPTCHA_REQUIRED: captcha detected on feed detail');
}
const raw = await page.evaluate((selectors: typeof XHH_SELECTORS) => {
const pickText = (selector: string): string =>
(document.querySelector(selector)?.textContent ?? '').trim();
const pickFrom = (selectorList: readonly string[]): string => {
for (const selector of selectorList) {
const text = pickText(selector);
if (text) return text;
}
return '';
};
const pickAttr = (selector: string, attr: string): string =>
(document.querySelector(selector)?.getAttribute(attr) ?? '').trim();
const title = pickFrom(selectors.detail.title);
const description = pickFrom(selectors.detail.description);
const images = [...document.querySelectorAll<HTMLImageElement>(selectors.detail.image)]
.map((img) => img.src)
.filter(Boolean);
const likeCount = pickText(selectors.detail.likeCount);
const favoriteCount = pickText(selectors.detail.favoriteCount);
const commentCount = pickText(selectors.detail.commentCount);
const likeBtn = selectors.detail.likeButton
.map((sel: string) => document.querySelector(sel))
.find(Boolean) as Element | undefined;
const favBtn = selectors.detail.favoriteButton
.map((sel: string) => document.querySelector(sel))
.find(Boolean) as Element | undefined;
const isLiked =
Boolean(likeBtn?.getAttribute('aria-pressed') === 'true') ||
Boolean(likeBtn?.className.toString().toLowerCase().includes('active')) ||
Boolean(likeBtn?.innerHTML.toLowerCase().includes('filled'));
const isFavorited =
Boolean(favBtn?.getAttribute('aria-pressed') === 'true') ||
Boolean(favBtn?.className.toString().toLowerCase().includes('active')) ||
Boolean(favBtn?.innerHTML.toLowerCase().includes('filled'));
const userLink = pickAttr(selectors.detail.userLink, 'href');
const nickname = pickFrom(selectors.detail.userName);
const avatar = pickAttr(selectors.detail.userAvatar, 'src');
const commentSelector = selectors.detail.commentItem.join(', ');
const subSelector = selectors.detail.subCommentItem.join(', ');
const comments: RawComment[] = [];
for (const node of document.querySelectorAll<HTMLElement>(commentSelector)) {
const id =
node.getAttribute('data-comment-id') ||
node.getAttribute('comment-id') ||
node.id ||
'';
const authorNode = node.querySelector(selectors.detail.commentAuthor);
const authorLink = authorNode?.getAttribute('href') ?? '';
const userId = authorLink.match(/\/app\/user\/profile\/(\d+)/)?.[1] ?? '';
const nickname = (authorNode?.textContent ?? '').trim();
const avatar = (node.querySelector(selectors.detail.commentAvatar) as HTMLImageElement | null)?.src ?? '';
const content = (node.querySelector(selectors.detail.commentContent)?.textContent ?? '').trim();
const createTime = (node.querySelector(selectors.detail.commentTime)?.textContent ?? '').trim();
const likeCount = (node.querySelector(selectors.detail.commentLikeCount)?.textContent ?? '').trim();
const subComments: RawComment[] = [];
for (const subNode of node.querySelectorAll<HTMLElement>(subSelector)) {
const subId =
subNode.getAttribute('data-comment-id') ||
subNode.getAttribute('comment-id') ||
subNode.id ||
'';
const subAuthorNode = subNode.querySelector(selectors.detail.commentAuthor);
const subAuthorLink = subAuthorNode?.getAttribute('href') ?? '';
const subUserId = subAuthorLink.match(/\/app\/user\/profile\/(\d+)/)?.[1] ?? '';
subComments.push({
id: subId,
parentId: id || undefined,
userId: subUserId,
nickname: (subAuthorNode?.textContent ?? '').trim(),
avatar: (subNode.querySelector(selectors.detail.commentAvatar) as HTMLImageElement | null)?.src ?? '',
content: (subNode.querySelector(selectors.detail.commentContent)?.textContent ?? '').trim(),
createTime: (subNode.querySelector(selectors.detail.commentTime)?.textContent ?? '').trim(),
likeCount: (subNode.querySelector(selectors.detail.commentLikeCount)?.textContent ?? '').trim(),
subComments: [],
});
}
comments.push({
id,
userId,
nickname,
avatar,
content,
likeCount,
createTime,
subComments,
});
}
return {
title,
description,
images,
likeCount,
favoriteCount,
commentCount,
isLiked,
isFavorited,
userId: userLink.match(/\/app\/user\/profile\/(\d+)/)?.[1] ?? '',
nickname,
avatar,
comments,
};
}, XHH_SELECTORS) as RawDetail;
const detail: FeedDetail = {
id: linkId,
title: raw.title,
description: raw.description,
images: raw.images,
likeCount: parseCountString(raw.likeCount),
favoriteCount: parseCountString(raw.favoriteCount),
commentCount: parseCountString(raw.commentCount),
isLiked: raw.isLiked,
isFavorited: raw.isFavorited,
user: {
id: raw.userId,
nickname: raw.nickname,
avatar: raw.avatar,
},
comments: raw.comments.map(normalizeComment),
};
if (!detail.title && !detail.description) {
throw new Error('waiting for selector: xhh detail not found');
}
log.info({ linkId, commentCount: detail.comments.length }, 'xhh feed detail extracted');
return detail;
}
export async function getSubComments(
page: Page,
linkId: string,
commentId: string,
maxCount = 200,
): Promise<Comment[]> {
const url = buildDetailUrl(linkId);
await page.goto(url, { waitUntil: 'domcontentloaded' });
await page.waitForTimeout(1_000);
const text = await page.textContent('body').catch(() => '');
if (text && detectCaptchaText(text)) {
throw new Error('CAPTCHA_REQUIRED: captcha detected on sub-comments page');
}
const expandSelector = XHH_SELECTORS.detail.commentExpandReplies;
await page.locator(expandSelector).first().click().catch(() => {});
await page.waitForTimeout(500);
const subComments = await page.evaluate(
({ selectors, targetCommentId }) => {
const commentSelector = selectors.detail.commentItem.join(', ');
const subSelector = selectors.detail.subCommentItem.join(', ');
const comments = [...document.querySelectorAll<HTMLElement>(commentSelector)];
const target = comments.find((node) => {
const id =
node.getAttribute('data-comment-id') ||
node.getAttribute('comment-id') ||
node.id ||
'';
if (id === targetCommentId) return true;
return node.outerHTML.includes(targetCommentId);
});
if (!target) return [] as RawComment[];
const out: RawComment[] = [];
for (const node of target.querySelectorAll<HTMLElement>(subSelector)) {
const authorNode = node.querySelector(selectors.detail.commentAuthor);
const authorLink = authorNode?.getAttribute('href') ?? '';
out.push({
id:
node.getAttribute('data-comment-id') ||
node.getAttribute('comment-id') ||
node.id ||
'',
parentId: targetCommentId,
userId: authorLink.match(/\/app\/user\/profile\/(\d+)/)?.[1] ?? '',
nickname: (authorNode?.textContent ?? '').trim(),
avatar: (node.querySelector(selectors.detail.commentAvatar) as HTMLImageElement | null)?.src ?? '',
content: (node.querySelector(selectors.detail.commentContent)?.textContent ?? '').trim(),
createTime: (node.querySelector(selectors.detail.commentTime)?.textContent ?? '').trim(),
likeCount: (node.querySelector(selectors.detail.commentLikeCount)?.textContent ?? '').trim(),
subComments: [],
});
}
return out;
},
{ selectors: XHH_SELECTORS, targetCommentId: commentId },
);
return subComments.slice(0, maxCount).map(normalizeComment);
}
function normalizeComment(input: RawComment): Comment {
return {
id: firstNonEmpty(input.id, `${Date.now()}-${Math.random()}`),
...(input.parentId ? { parentId: input.parentId } : {}),
userId: input.userId,
nickname: input.nickname,
avatar: input.avatar,
content: input.content,
likeCount: parseCountString(input.likeCount),
createTime: input.createTime,
subComments: input.subComments.map(normalizeComment),
};
}
@@ -0,0 +1,268 @@
import type { Page } from 'rebrowser-playwright';
import { logger } from '@social/core/utils/logger.js';
import { XHH_SELECTORS } from './selectors.js';
import type { Feed } from './types.js';
import {
detectCaptchaText,
extractLinkIdFromUrl,
firstNonEmpty,
parseCountString,
} from './extractors.js';
const HOME_URL = 'https://www.xiaoheihe.cn/app/bbs/home';
const log = logger.child({ module: 'xhh-feeds' });
interface RawFeedCandidate {
id?: string;
title?: string;
description?: string;
coverUrl?: string;
likeCount?: string | number;
commentCount?: string | number;
userId?: string;
nickname?: string;
avatar?: string;
linkUrl?: string;
}
export async function listFeeds(page: Page): Promise<Feed[]> {
await page.goto(HOME_URL, { waitUntil: 'domcontentloaded' });
await page.waitForTimeout(1_500);
const text = await page.textContent('body').catch(() => '');
if (text && detectCaptchaText(text)) {
throw new Error('CAPTCHA_REQUIRED: captcha detected on feeds page');
}
const nuxtFeeds = await extractFeedsFromNuxt(page);
const domFeeds = await extractFeedsFromDom(page);
const merged = [...nuxtFeeds, ...domFeeds];
const result = dedupeAndNormalize(merged);
log.info({ count: result.length }, 'xhh feeds extracted');
return result;
}
export async function searchFeeds(page: Page, keyword: string): Promise<Feed[]> {
const targetUrl = `https://www.xiaoheihe.cn/app/bbs/search?keyword=${encodeURIComponent(keyword)}`;
await page.goto(targetUrl, { waitUntil: 'domcontentloaded' });
await page.waitForTimeout(1_200);
const text = await page.textContent('body').catch(() => '');
if (text && detectCaptchaText(text)) {
throw new Error('CAPTCHA_REQUIRED: captcha detected on search page');
}
const combined = dedupeAndNormalize([
...(await extractFeedsFromNuxt(page)),
...(await extractFeedsFromDom(page)),
]);
if (combined.length > 0) {
return combined.filter((item) => {
const haystack = `${item.title} ${item.description} ${item.user.nickname}`.toLowerCase();
return haystack.includes(keyword.toLowerCase());
});
}
// Fallback: when search route structure changes, use home feeds and filter.
const homeFeeds = await listFeeds(page);
return homeFeeds.filter((item) => {
const haystack = `${item.title} ${item.description} ${item.user.nickname}`.toLowerCase();
return haystack.includes(keyword.toLowerCase());
});
}
async function extractFeedsFromNuxt(page: Page): Promise<RawFeedCandidate[]> {
const data = await page
.evaluate(() => {
const root: unknown =
(window as { __NUXT_DATA__?: unknown }).__NUXT_DATA__ ??
(window as { __NUXT__?: { data?: unknown } }).__NUXT__?.data ??
null;
const out: Array<Record<string, unknown>> = [];
const visited = new Set<unknown>();
function walk(value: unknown): void {
if (!value || typeof value !== 'object') return;
if (visited.has(value)) return;
visited.add(value);
if (Array.isArray(value)) {
for (const item of value) walk(item);
return;
}
const obj = value as Record<string, unknown>;
const id =
(typeof obj['link_id'] === 'string' && obj['link_id']) ||
(typeof obj['linkid'] === 'string' && obj['linkid']) ||
(typeof obj['id'] === 'string' && obj['id']) ||
(typeof obj['post_id'] === 'string' && obj['post_id']) ||
'';
const url =
(typeof obj['link_url'] === 'string' && obj['link_url']) ||
(typeof obj['url'] === 'string' && obj['url']) ||
'';
const title =
(typeof obj['title'] === 'string' && obj['title']) ||
(typeof obj['subject'] === 'string' && obj['subject']) ||
'';
const hasLink = (typeof url === 'string' && url.includes('/app/bbs/link/'));
if (id || hasLink || title) {
out.push(obj);
}
for (const next of Object.values(obj)) {
walk(next);
}
}
walk(root);
return out.slice(0, 500);
})
.catch(() => []);
return (data as Array<Record<string, unknown>>).map((item) => {
const linkUrl = firstNonEmpty(
valueString(item['link_url']),
valueString(item['url']),
valueString(item['jump_url']),
);
const user = (item['user'] ?? item['author']) as Record<string, unknown> | undefined;
return {
id: firstNonEmpty(
valueString(item['link_id']),
valueString(item['linkid']),
valueString(item['post_id']),
valueString(item['id']),
),
title: firstNonEmpty(valueString(item['title']), valueString(item['subject'])),
description: firstNonEmpty(
valueString(item['description']),
valueString(item['content']),
valueString(item['desc']),
),
coverUrl: firstNonEmpty(
valueString(item['cover']),
valueString(item['cover_url']),
valueString(item['image']),
),
likeCount: valueString(item['like_count']) || valueString(item['likes']),
commentCount: valueString(item['comment_count']) || valueString(item['comments']),
userId: firstNonEmpty(
valueString(user?.['userid']),
valueString(user?.['user_id']),
valueString(item['userid']),
),
nickname: firstNonEmpty(
valueString(user?.['nickname']),
valueString(user?.['name']),
valueString(item['nickname']),
),
avatar: firstNonEmpty(
valueString(user?.['avatar']),
valueString(user?.['avatar_url']),
),
linkUrl,
};
});
}
async function extractFeedsFromDom(page: Page): Promise<RawFeedCandidate[]> {
return page
.evaluate((selectors) => {
const anchors = [...document.querySelectorAll<HTMLAnchorElement>(selectors.feed.link)]
.filter((a) => Boolean(a.getAttribute('href')));
const feeds: RawFeedCandidate[] = [];
const cardSelector = selectors.feed.card.join(', ');
const titleSelector = selectors.feed.title.join(', ');
const descSelector = selectors.feed.description.join(', ');
const userNameSelector = selectors.feed.userName.join(', ');
const likeSelector = selectors.feed.likeCount.join(', ');
const commentSelector = selectors.feed.commentCount.join(', ');
for (const link of anchors) {
const href = link.getAttribute('href') ?? '';
const card = link.closest(cardSelector) ?? link.parentElement;
const title = (card?.querySelector(titleSelector)?.textContent ?? '').trim();
const description = (card?.querySelector(descSelector)?.textContent ?? '').trim();
const cover = (card?.querySelector(selectors.feed.cover) as HTMLImageElement | null)?.src ?? '';
const userNode = card?.querySelector(selectors.feed.userLink) as HTMLAnchorElement | null;
const username = (card?.querySelector(userNameSelector)?.textContent ?? '').trim();
const likeCount = (card?.querySelector(likeSelector)?.textContent ?? '').trim();
const commentCount = (card?.querySelector(commentSelector)?.textContent ?? '').trim();
feeds.push({
linkUrl: href,
title,
description,
coverUrl: cover,
userId: userNode?.getAttribute('href') ?? '',
nickname: username,
avatar: (card?.querySelector('img') as HTMLImageElement | null)?.src ?? '',
likeCount,
commentCount,
});
}
return feeds;
}, XHH_SELECTORS)
.catch(() => []);
}
function dedupeAndNormalize(items: RawFeedCandidate[]): Feed[] {
const output: Feed[] = [];
const seen = new Set<string>();
for (const item of items) {
const linkId = firstNonEmpty(item.id, item.linkUrl ? extractLinkIdFromUrl(item.linkUrl) ?? '' : '');
if (!linkId || seen.has(linkId)) continue;
seen.add(linkId);
const linkUrl = normalizeLinkUrl(item.linkUrl, linkId);
const userIdFromHref = item.userId ? extractUserIdFromMaybeHref(item.userId) : '';
output.push({
id: linkId,
title: item.title?.trim() ?? '',
description: item.description?.trim() ?? '',
coverUrl: item.coverUrl?.trim() ?? '',
likeCount: parseCountString(item.likeCount),
commentCount: parseCountString(item.commentCount),
user: {
id: userIdFromHref,
nickname: item.nickname?.trim() ?? '',
avatar: item.avatar?.trim() ?? '',
},
linkUrl,
});
}
return output;
}
function normalizeLinkUrl(rawUrl: string | undefined, linkId: string): string {
const trimmed = rawUrl?.trim() ?? '';
if (!trimmed) return `https://www.xiaoheihe.cn/app/bbs/link/${linkId}`;
if (/^https?:\/\//i.test(trimmed)) return trimmed;
if (trimmed.startsWith('/')) return `https://www.xiaoheihe.cn${trimmed}`;
return `https://${trimmed}`;
}
function extractUserIdFromMaybeHref(raw: string): string {
const normalized = raw.startsWith('/') ? `https://www.xiaoheihe.cn${raw}` : raw;
return normalized.match(/\/app\/user\/profile\/(\d+)/)?.[1] ?? '';
}
function valueString(value: unknown): string {
if (typeof value === 'string') return value;
if (typeof value === 'number') return String(value);
return '';
}
@@ -0,0 +1,389 @@
import type { McpServer } from '@modelcontextprotocol/sdk/server/mcp.js';
import type { Router } from 'express';
import type { BrowserManager } from '@social/core/browser/manager.js';
import { config } from '@social/core/config/index.js';
import type { PlatformPlugin } from '@social/core/server/app.js';
import { withErrorHandling, type McpToolResult } from '@social/core/utils/errors.js';
import { computeIdempotencyHash, getIdempotencyStore } from '@social/core/utils/idempotency.js';
import { deleteCookies, checkLoginStatus, getLoginQRCode } from './login.js';
import { listFeeds } from './feeds.js';
import { searchFeeds } from './search.js';
import { getFeedDetail, getSubComments } from './feed-detail.js';
import { getUserProfile } from './user-profile.js';
import { listMyPosts } from './my-posts.js';
import { postComment, replyComment } from './comment.js';
import { setFavoriteState, setLikeState } from './interaction.js';
import { resolveFeedTarget, resolveUserTarget } from './target-resolver.js';
import {
CheckLoginSchema,
DeleteCookiesSchema,
GetFeedDetailSchema,
GetLoginQRCodeSchema,
GetSubCommentsSchema,
GetUserProfileSchema,
ListFeedsSchema,
ListMyPostsSchema,
PostCommentSchema,
ReplyCommentSchema,
SearchSchema,
SetFavoriteStateSchema,
SetLikeStateSchema,
} from './schemas.js';
import { createXhhRoutes } from './routes.js';
import { decodeKeysetCursor, paginateByKeyset } from './cursor.js';
import type { Comment } from './types.js';
const PLATFORM = 'xiaoheihe';
const DEFAULT_PAGE_SIZE = 20;
const MAX_PAGE_SIZE = 200;
type McpMeta = Record<string, unknown>;
function ok(data: unknown, meta?: McpMeta): McpToolResult {
return {
content: [{
type: 'text',
text: JSON.stringify({
success: true,
data,
meta: meta ?? {},
}),
}],
};
}
function clampPageSize(maxCount?: number): number {
return Math.min(MAX_PAGE_SIZE, Math.max(1, maxCount ?? DEFAULT_PAGE_SIZE));
}
function paginationMeta(
cursor: string | undefined,
maxCount: number,
returned: number,
nextCursor?: string,
): McpMeta {
return {
pagination: {
mode: 'keyset',
cursor: cursor ?? '',
max_count: maxCount,
returned,
...(nextCursor ? { next_cursor: nextCursor } : {}),
},
};
}
async function runWithIdempotency<T>(
toolName: string,
requestId: string | undefined,
inputForHash: unknown,
execute: () => Promise<T>,
): Promise<{ data: T; meta?: McpMeta }> {
if (!requestId) {
return { data: await execute() };
}
const store = getIdempotencyStore();
const inputHash = computeIdempotencyHash(inputForHash);
const existing = store.get(toolName, requestId);
if (existing) {
if (existing.inputHash !== inputHash) {
throw new Error('request_id already used with different parameters');
}
return {
data: existing.responseData as T,
meta: {
request_id: requestId,
idempotent_replay: true,
first_processed_at: existing.createdAt,
},
};
}
const data = await execute();
store.put(toolName, requestId, inputHash, data);
return {
data,
meta: {
request_id: requestId,
idempotent_replay: false,
},
};
}
function compareCommentKey(a: Comment, b: Comment): number {
const timeCmp = a.createTime.localeCompare(b.createTime);
if (timeCmp !== 0) return timeCmp;
return a.id.localeCompare(b.id);
}
export const xiaoheihePlugin: PlatformPlugin = {
name: PLATFORM,
apiNamespace: 'xhh',
registerRoutes(router: Router, browser: BrowserManager): void {
const xhhRouter = createXhhRoutes(browser);
router.use('/', xhhRouter);
},
registerTools(server: McpServer, browser: BrowserManager): void {
server.tool(
'xhh_check_login',
'Check Xiaoheihe login status',
CheckLoginSchema,
async () => withErrorHandling('xhh_check_login', async () => {
const timeoutMs = config.operationTimeouts['login'] ?? config.operationTimeouts['default'] ?? 60_000;
const status = await browser.withPage(PLATFORM, async (page) => checkLoginStatus(page), timeoutMs);
return ok({
logged_in: status.loggedIn,
...(status.username ? { username: status.username } : {}),
...(status.avatar ? { avatar: status.avatar } : {}),
...(status.userId ? { user_id: status.userId } : {}),
});
}),
);
server.tool(
'xhh_get_login_qrcode',
'Get Xiaoheihe login QR code',
GetLoginQRCodeSchema,
async () => withErrorHandling('xhh_get_login_qrcode', async () => {
const qr = await getLoginQRCode(browser);
return ok({
qrcode_data: qr.qrcodeData,
already_logged_in: qr.alreadyLoggedIn,
timeout: qr.timeout,
});
}),
);
server.tool(
'xhh_delete_cookies',
'Delete Xiaoheihe cookies and reset login session',
DeleteCookiesSchema,
async () => withErrorHandling('xhh_delete_cookies', async () => {
await deleteCookies(browser);
return ok({ deleted: true });
}),
);
server.tool(
'xhh_list_feeds',
'List Xiaoheihe feed cards',
ListFeedsSchema,
async (args) => withErrorHandling('xhh_list_feeds', async () => {
const timeoutMs = config.operationTimeouts['feed_list'] ?? config.operationTimeouts['default'] ?? 60_000;
const feeds = await browser.withPage(PLATFORM, async (page) => listFeeds(page), timeoutMs);
const limit = clampPageSize(args.max_count);
const cursor = decodeKeysetCursor(args.cursor);
const paged = paginateByKeyset(feeds, limit, cursor, (item) => item.id);
return ok(
paged.items,
paginationMeta(args.cursor, limit, paged.items.length, paged.nextCursor),
);
}),
);
server.tool(
'xhh_search',
'Search Xiaoheihe posts by keyword',
SearchSchema,
async (args) => withErrorHandling('xhh_search', async () => {
const timeoutMs = config.operationTimeouts['search'] ?? config.operationTimeouts['default'] ?? 60_000;
const feeds = await browser.withPage(
PLATFORM,
async (page) => searchFeeds(page, args.keyword),
timeoutMs,
);
const limit = clampPageSize(args.max_count);
const cursor = decodeKeysetCursor(args.cursor);
const paged = paginateByKeyset(feeds, limit, cursor, (item) => item.id);
return ok(
paged.items,
paginationMeta(args.cursor, limit, paged.items.length, paged.nextCursor),
);
}),
);
server.tool(
'xhh_get_feed_detail',
'Get Xiaoheihe feed detail with first-screen comments',
GetFeedDetailSchema,
async (args) => withErrorHandling('xhh_get_feed_detail', async () => {
const timeoutMs = config.operationTimeouts['feed_detail'] ?? config.operationTimeouts['default'] ?? 60_000;
const target = resolveFeedTarget({
link_id: args.link_id,
url: args.url,
});
const detail = await browser.withPage(
PLATFORM,
async (page) => getFeedDetail(page, target.linkId),
timeoutMs,
);
const { comments, ...rest } = detail;
return ok({
detail: rest,
first_screen_comments: comments,
});
}),
);
server.tool(
'xhh_get_sub_comments',
'Get sub-comments for a Xiaoheihe comment with keyset pagination',
GetSubCommentsSchema,
async (args) => withErrorHandling('xhh_get_sub_comments', async () => {
const timeoutMs = config.operationTimeouts['feed_detail'] ?? config.operationTimeouts['default'] ?? 60_000;
const loaded = await browser.withPage(
PLATFORM,
async (page) => getSubComments(page, args.link_id, args.comment_id, MAX_PAGE_SIZE),
timeoutMs,
);
const sorted = [...loaded].sort(compareCommentKey);
const limit = clampPageSize(args.max_count);
const cursor = decodeKeysetCursor(args.cursor);
const paged = paginateByKeyset(
sorted,
limit,
cursor,
(item) => `${item.createTime}|${item.id}`,
);
return ok(
paged.items,
paginationMeta(args.cursor, limit, paged.items.length, paged.nextCursor),
);
}),
);
server.tool(
'xhh_get_user_profile',
'Get Xiaoheihe user profile',
GetUserProfileSchema,
async (args) => withErrorHandling('xhh_get_user_profile', async () => {
const timeoutMs = config.operationTimeouts['user_profile'] ?? config.operationTimeouts['default'] ?? 60_000;
const target = resolveUserTarget({
user_id: args.user_id,
url: args.url,
});
const profile = await browser.withPage(
PLATFORM,
async (page) => getUserProfile(page, target.userId),
timeoutMs,
);
return ok({
profile: {
id: profile.id,
nickname: profile.nickname,
avatar: profile.avatar,
description: profile.description,
follows: profile.follows,
fans: profile.fans,
likes: profile.likes,
},
recent_posts: profile.posts,
});
}),
);
server.tool(
'xhh_list_my_posts',
'List my Xiaoheihe posts',
ListMyPostsSchema,
async (args) => withErrorHandling('xhh_list_my_posts', async () => {
const timeoutMs = config.operationTimeouts['feed_list'] ?? config.operationTimeouts['default'] ?? 60_000;
const posts = await browser.withPage(
PLATFORM,
async (page) => listMyPosts(page, args.type),
timeoutMs,
);
const limit = clampPageSize(args.max_count);
const cursor = decodeKeysetCursor(args.cursor);
const paged = paginateByKeyset(posts, limit, cursor, (item) => `${item.modifyTime ?? item.createTime ?? ''}|${item.id}`);
return ok(
paged.items,
paginationMeta(args.cursor, limit, paged.items.length, paged.nextCursor),
);
}),
);
server.tool(
'xhh_post_comment',
'Post a comment on Xiaoheihe',
PostCommentSchema,
async (args) => withErrorHandling('xhh_post_comment', async () => {
const timeoutMs = config.operationTimeouts['comment'] ?? config.operationTimeouts['default'] ?? 60_000;
const result = await runWithIdempotency(
'xhh_post_comment',
args.request_id,
{
link_id: args.link_id,
content: args.content,
},
async () => browser.withPage(
PLATFORM,
async (page) => postComment(page, args.link_id, args.content),
timeoutMs,
),
);
return ok(result.data, result.meta);
}),
);
server.tool(
'xhh_reply_comment',
'Reply a comment on Xiaoheihe',
ReplyCommentSchema,
async (args) => withErrorHandling('xhh_reply_comment', async () => {
const timeoutMs = config.operationTimeouts['reply'] ?? config.operationTimeouts['default'] ?? 60_000;
const result = await runWithIdempotency(
'xhh_reply_comment',
args.request_id,
{
link_id: args.link_id,
comment_id: args.comment_id,
content: args.content,
},
async () => browser.withPage(
PLATFORM,
async (page) => replyComment(page, args.link_id, args.comment_id, args.content),
timeoutMs,
),
);
return ok(result.data, result.meta);
}),
);
server.tool(
'xhh_set_like_state',
'Set like state for a Xiaoheihe post',
SetLikeStateSchema,
async (args) => withErrorHandling('xhh_set_like_state', async () => {
const timeoutMs = config.operationTimeouts['like'] ?? config.operationTimeouts['default'] ?? 60_000;
const result = await browser.withPage(
PLATFORM,
async (page) => setLikeState(page, args.link_id, args.liked),
timeoutMs,
);
return ok(result);
}),
);
server.tool(
'xhh_set_favorite_state',
'Set favorite state for a Xiaoheihe post',
SetFavoriteStateSchema,
async (args) => withErrorHandling('xhh_set_favorite_state', async () => {
const timeoutMs = config.operationTimeouts['favorite'] ?? config.operationTimeouts['default'] ?? 60_000;
const result = await browser.withPage(
PLATFORM,
async (page) => setFavoriteState(page, args.link_id, args.favorited),
timeoutMs,
);
return ok(result);
}),
);
},
};
@@ -0,0 +1,98 @@
import type { Page } from 'rebrowser-playwright';
import { XHH_SELECTORS } from './selectors.js';
import { detectCaptchaText } from './extractors.js';
function buildDetailUrl(linkId: string): string {
return `https://www.xiaoheihe.cn/app/bbs/link/${encodeURIComponent(linkId)}`;
}
export async function setLikeState(
page: Page,
linkId: string,
targetState: boolean,
): Promise<{ success: boolean; state: boolean; changed: boolean }> {
await page.goto(buildDetailUrl(linkId), { waitUntil: 'domcontentloaded' });
await page.waitForTimeout(1_000);
const text = await page.textContent('body').catch(() => '');
if (text && detectCaptchaText(text)) {
throw new Error('CAPTCHA_REQUIRED: captcha detected on interaction page');
}
const current = await readButtonState(page, XHH_SELECTORS.detail.likeButton);
if (current === targetState) {
return { success: true, state: current, changed: false };
}
const clicked = await clickAny(page, XHH_SELECTORS.detail.likeButton);
if (!clicked) {
return { success: false, state: current, changed: false };
}
await page.waitForTimeout(700);
const state = await readButtonState(page, XHH_SELECTORS.detail.likeButton);
return {
success: state === targetState,
state,
changed: state !== current,
};
}
export async function setFavoriteState(
page: Page,
linkId: string,
targetState: boolean,
): Promise<{ success: boolean; state: boolean; changed: boolean }> {
await page.goto(buildDetailUrl(linkId), { waitUntil: 'domcontentloaded' });
await page.waitForTimeout(1_000);
const text = await page.textContent('body').catch(() => '');
if (text && detectCaptchaText(text)) {
throw new Error('CAPTCHA_REQUIRED: captcha detected on interaction page');
}
const current = await readButtonState(page, XHH_SELECTORS.detail.favoriteButton);
if (current === targetState) {
return { success: true, state: current, changed: false };
}
const clicked = await clickAny(page, XHH_SELECTORS.detail.favoriteButton);
if (!clicked) {
return { success: false, state: current, changed: false };
}
await page.waitForTimeout(700);
const state = await readButtonState(page, XHH_SELECTORS.detail.favoriteButton);
return {
success: state === targetState,
state,
changed: state !== current,
};
}
async function clickAny(page: Page, selectors: readonly string[]): Promise<boolean> {
for (const selector of selectors) {
const ok = await page.locator(selector).first().click({ timeout: 2_000 }).then(() => true).catch(() => false);
if (ok) return true;
}
return false;
}
async function readButtonState(page: Page, selectors: readonly string[]): Promise<boolean> {
for (const selector of selectors) {
const state = await page
.evaluate((sel) => {
const node = document.querySelector(sel) as HTMLElement | null;
if (!node) return null;
if (node.getAttribute('aria-pressed') === 'true') return true;
const cls = node.className.toString().toLowerCase();
if (cls.includes('active') || cls.includes('selected')) return true;
const html = node.innerHTML.toLowerCase();
if (html.includes('filled') || html.includes('checked')) return true;
return false;
}, selector)
.catch(() => null);
if (typeof state === 'boolean') return state;
}
return false;
}
@@ -0,0 +1,169 @@
import type { Page } from 'rebrowser-playwright';
import type { BrowserManager } from '@social/core/browser/manager.js';
import { cookieStore } from '@social/core/cookie/store.js';
import { logger } from '@social/core/utils/logger.js';
import { XHH_SELECTORS } from './selectors.js';
import type { LoginStatus, QRCodeResult } from './types.js';
import { extractUserIdFromUrl, firstNonEmpty } from './extractors.js';
const PLATFORM = 'xiaoheihe';
const HOME_URL = 'https://www.xiaoheihe.cn/app/bbs/home';
const QR_SCAN_TIMEOUT_MS = 4 * 60_000;
const LOGIN_SAFETY_TIMEOUT_MS = 5 * 60_000;
const log = logger.child({ module: 'xhh-login' });
export async function checkLoginStatus(page: Page): Promise<LoginStatus> {
await page.goto(HOME_URL, { waitUntil: 'domcontentloaded' });
await page.waitForTimeout(1_200);
const indicator = await waitFirstSelector(page, XHH_SELECTORS.login.loggedInIndicators, 4_000);
if (!indicator) {
return { loggedIn: false };
}
const username = firstNonEmpty(
await textFromSelector(page, XHH_SELECTORS.login.username),
await indicator.textContent().catch(() => ''),
);
const avatar = await attrFromSelector(page, XHH_SELECTORS.login.avatar, 'src');
const userLink = await attrFromSelector(page, XHH_SELECTORS.login.userLink, 'href');
const userId = userLink ? extractUserIdFromUrl(userLink) : undefined;
return {
loggedIn: true,
...(username ? { username } : {}),
...(avatar ? { avatar } : {}),
...(userId ? { userId } : {}),
};
}
export async function getLoginQRCode(browser: BrowserManager): Promise<QRCodeResult> {
const { page, release } = await browser.acquirePage(PLATFORM);
const releaseTimer = setTimeout(() => {
void release();
}, LOGIN_SAFETY_TIMEOUT_MS);
if (typeof releaseTimer === 'object' && 'unref' in releaseTimer) {
releaseTimer.unref();
}
try {
await page.goto(HOME_URL, { waitUntil: 'domcontentloaded' });
await page.waitForTimeout(1_200);
const status = await checkLoginStatus(page);
if (status.loggedIn) {
await release();
clearTimeout(releaseTimer);
return {
qrcodeData: '',
alreadyLoggedIn: true,
timeout: '0',
};
}
const loginBtn = await page.$(XHH_SELECTORS.login.loginButton).catch(() => null);
if (loginBtn) {
await loginBtn.click().catch(() => {});
await page.waitForTimeout(500);
}
const qrcodeData = await extractQrCodeData(page);
if (!qrcodeData) {
await release();
clearTimeout(releaseTimer);
throw new Error('waiting for selector: xhh login qrcode');
}
waitForLoginAndRelease(page, browser, release).catch((err: unknown) => {
log.warn({ err }, 'background login wait failed');
});
return {
qrcodeData,
alreadyLoggedIn: false,
timeout: '4m',
};
} catch (err) {
clearTimeout(releaseTimer);
await release();
throw err;
}
}
export async function deleteCookies(browser: BrowserManager): Promise<void> {
await cookieStore.delete(PLATFORM);
await browser.clearContext(PLATFORM);
}
async function waitForLoginAndRelease(
page: Page,
browser: BrowserManager,
release: () => Promise<void>,
): Promise<void> {
try {
await waitFirstSelector(page, XHH_SELECTORS.login.loggedInIndicators, QR_SCAN_TIMEOUT_MS);
await browser.saveCookies(PLATFORM);
await browser.clearContext(PLATFORM);
} finally {
await release();
}
}
async function waitFirstSelector(
page: Page,
selectors: readonly string[],
timeout: number,
) {
const started = Date.now();
for (const selector of selectors) {
const remaining = Math.max(1, timeout - (Date.now() - started));
const handle = await page.waitForSelector(selector, { timeout: remaining }).catch(() => null);
if (handle) return handle;
}
return null;
}
async function extractQrCodeData(page: Page): Promise<string> {
for (const selector of XHH_SELECTORS.login.qrCodeImage) {
const data = await page
.evaluate((sel) => {
const node = document.querySelector(sel);
if (!node) return '';
if (node instanceof HTMLImageElement) {
return node.src || '';
}
if (node instanceof HTMLCanvasElement) {
try {
return node.toDataURL();
} catch {
return '';
}
}
return '';
}, selector)
.catch(() => '');
if (data) return data;
}
return '';
}
async function textFromSelector(page: Page, selector: string): Promise<string> {
return page
.$eval(selector, (el) => (el.textContent ?? '').trim())
.catch(() => '');
}
async function attrFromSelector(
page: Page,
selector: string,
attr: string,
): Promise<string> {
return page
.$eval(selector, (el, attrName) => el.getAttribute(attrName) ?? '', attr)
.catch(() => '');
}
@@ -0,0 +1,88 @@
import type { Page } from 'rebrowser-playwright';
import { logger } from '@social/core/utils/logger.js';
import { XHH_SELECTORS } from './selectors.js';
import type { MyPost, MyPostType } from './types.js';
import { detectCaptchaText, extractLinkIdFromUrl, parseCountString } from './extractors.js';
const URL = 'https://www.xiaoheihe.cn/creator/content_management/home';
const log = logger.child({ module: 'xhh-my-posts' });
const TAB_KEYWORDS: Record<MyPostType, string> = {
all: '全部',
article: '文章',
image_text: '图文',
video: '视频',
};
export async function listMyPosts(
page: Page,
type: MyPostType = 'all',
): Promise<MyPost[]> {
await page.goto(URL, { waitUntil: 'domcontentloaded' });
await page.waitForTimeout(1_200);
const text = await page.textContent('body').catch(() => '');
if (text && detectCaptchaText(text)) {
throw new Error('CAPTCHA_REQUIRED: captcha detected on my-posts page');
}
if (type !== 'all') {
const keyword = TAB_KEYWORDS[type];
const tabs = page.locator(XHH_SELECTORS.myPosts.tabButton);
const count = await tabs.count().catch(() => 0);
for (let i = 0; i < count; i += 1) {
const tab = tabs.nth(i);
const tabText = (await tab.textContent().catch(() => '')) ?? '';
if (tabText.includes(keyword)) {
await tab.click().catch(() => {});
await page.waitForTimeout(500);
break;
}
}
}
const rawItems = await page.evaluate((selectors) => {
const container = [...document.querySelectorAll<HTMLElement>(selectors.myPosts.postItem)];
return container.map((node) => {
const linkNode = node.querySelector<HTMLAnchorElement>(selectors.myPosts.postLink);
const href = linkNode?.getAttribute('href') ?? '';
const title = (node.querySelector(selectors.myPosts.title)?.textContent ?? '').trim();
const description = (node.querySelector(selectors.myPosts.description)?.textContent ?? '').trim();
const time = (node.querySelector(selectors.myPosts.time)?.textContent ?? '').trim();
const likeRaw = (node.querySelector('.like-count, .content-list__like-cnt')?.textContent ?? '').trim();
const commentRaw = (node.querySelector('.comment-count, .content-list__comment-cnt')?.textContent ?? '').trim();
const cover = (node.querySelector('img') as HTMLImageElement | null)?.src ?? '';
return { href, title, description, time, likeRaw, commentRaw, cover };
});
}, XHH_SELECTORS);
const posts: MyPost[] = [];
for (const item of rawItems) {
const linkId = extractLinkIdFromUrl(item.href);
if (!linkId) continue;
const linkUrl = item.href.startsWith('http')
? item.href
: `https://www.xiaoheihe.cn${item.href}`;
posts.push({
id: linkId,
type,
title: item.title,
description: item.description,
coverUrl: item.cover,
likeCount: parseCountString(item.likeRaw),
commentCount: parseCountString(item.commentRaw),
user: {
id: '',
nickname: '',
avatar: '',
},
linkUrl,
createTime: item.time,
modifyTime: item.time,
});
}
log.info({ type, count: posts.length }, 'xhh my posts listed');
return posts;
}
@@ -0,0 +1,464 @@
import { Router } from 'express';
import { z, ZodError } from 'zod';
import type { BrowserManager } from '@social/core/browser/manager.js';
import { config } from '@social/core/config/index.js';
import { rateLimiter } from '@social/core/server/middleware.js';
import { classifyError, sanitizeErrorMessage } from '@social/core/utils/errors.js';
import { computeIdempotencyHash, getIdempotencyStore } from '@social/core/utils/idempotency.js';
import { decodeKeysetCursor, paginateByKeyset } from './cursor.js';
import { postComment, replyComment } from './comment.js';
import { getFeedDetail, getSubComments } from './feed-detail.js';
import { listFeeds } from './feeds.js';
import { setFavoriteState, setLikeState } from './interaction.js';
import { checkLoginStatus, deleteCookies, getLoginQRCode } from './login.js';
import { listMyPosts } from './my-posts.js';
import {
GetFeedDetailSchema,
GetSubCommentsSchema,
GetUserProfileSchema,
ListMyPostsSchema,
PostCommentSchema,
ReplyCommentSchema,
SearchSchema,
SetFavoriteStateSchema,
SetLikeStateSchema,
} from './schemas.js';
import { searchFeeds } from './search.js';
import { resolveFeedTarget, resolveUserTarget } from './target-resolver.js';
import { getUserProfile } from './user-profile.js';
const PLATFORM = 'xiaoheihe';
const DEFAULT_PAGE_SIZE = 20;
const MAX_PAGE_SIZE = 200;
const readRateLimiter = rateLimiter({ windowMs: 60_000, maxRequests: 60 });
const writeRateLimiter = rateLimiter({ windowMs: 60_000, maxRequests: 10 });
const SearchBodySchema = z.object({
keyword: SearchSchema.keyword,
max_count: SearchSchema.max_count,
cursor: SearchSchema.cursor,
});
const FeedDetailBodySchema = z.object({
link_id: GetFeedDetailSchema.link_id,
url: GetFeedDetailSchema.url,
});
const SubCommentsBodySchema = z.object({
link_id: GetSubCommentsSchema.link_id,
comment_id: GetSubCommentsSchema.comment_id,
max_count: GetSubCommentsSchema.max_count,
cursor: GetSubCommentsSchema.cursor,
});
const UserProfileBodySchema = z.object({
user_id: GetUserProfileSchema.user_id,
url: GetUserProfileSchema.url,
});
const PostCommentBodySchema = z.object({
request_id: PostCommentSchema.request_id,
link_id: PostCommentSchema.link_id,
content: PostCommentSchema.content,
});
const ReplyCommentBodySchema = z.object({
request_id: ReplyCommentSchema.request_id,
link_id: ReplyCommentSchema.link_id,
comment_id: ReplyCommentSchema.comment_id,
content: ReplyCommentSchema.content,
});
const LikeBodySchema = z.object({
link_id: SetLikeStateSchema.link_id,
liked: SetLikeStateSchema.liked,
});
const FavoriteBodySchema = z.object({
link_id: SetFavoriteStateSchema.link_id,
favorited: SetFavoriteStateSchema.favorited,
});
interface ApiSuccessResponse<T> {
success: true;
data: T;
}
interface ApiErrorResponse {
success: false;
error: {
code: string;
message: string;
};
}
function successResponse<T>(data: T): ApiSuccessResponse<T> {
return { success: true, data };
}
function errorResponse(code: string, message: string): ApiErrorResponse {
return { success: false, error: { code, message } };
}
function clampPageSize(maxCount?: number): number {
return Math.min(MAX_PAGE_SIZE, Math.max(1, maxCount ?? DEFAULT_PAGE_SIZE));
}
async function runWithIdempotency<T>(
toolName: string,
requestId: string | undefined,
inputForHash: unknown,
execute: () => Promise<T>,
): Promise<{ data: T; meta?: Record<string, unknown> }> {
if (!requestId) {
return { data: await execute() };
}
const store = getIdempotencyStore();
const inputHash = computeIdempotencyHash(inputForHash);
const existing = store.get(toolName, requestId);
if (existing) {
if (existing.inputHash !== inputHash) {
throw new Error('request_id already used with different parameters');
}
return {
data: existing.responseData as T,
meta: {
request_id: requestId,
idempotent_replay: true,
first_processed_at: existing.createdAt,
},
};
}
const data = await execute();
store.put(toolName, requestId, inputHash, data);
return {
data,
meta: {
request_id: requestId,
idempotent_replay: false,
},
};
}
export function createXhhRoutes(browser: BrowserManager): Router {
const router = Router();
router.get('/login/status', readRateLimiter, (_req, res) => {
void (async () => {
try {
const timeoutMs = config.operationTimeouts['login'] ?? config.operationTimeouts['default'] ?? 60_000;
const status = await browser.withPage(PLATFORM, async (page) => checkLoginStatus(page), timeoutMs);
res.json(successResponse({
logged_in: status.loggedIn,
...(status.username ? { username: status.username } : {}),
...(status.avatar ? { avatar: status.avatar } : {}),
...(status.userId ? { user_id: status.userId } : {}),
}));
} catch (err) {
handleError(res, err);
}
})();
});
router.get('/login/qrcode', readRateLimiter, (_req, res) => {
void (async () => {
try {
const qr = await getLoginQRCode(browser);
res.json(successResponse({
qrcode_data: qr.qrcodeData,
already_logged_in: qr.alreadyLoggedIn,
timeout: qr.timeout,
}));
} catch (err) {
handleError(res, err);
}
})();
});
router.delete('/login/cookies', writeRateLimiter, (_req, res) => {
void (async () => {
try {
await deleteCookies(browser);
res.json(successResponse({ deleted: true }));
} catch (err) {
handleError(res, err);
}
})();
});
router.get('/feeds', readRateLimiter, (req, res) => {
void (async () => {
try {
const query = z.object({
max_count: z.coerce.number().int().min(1).max(200).optional().default(20),
cursor: z.string().optional(),
}).parse(req.query);
const timeoutMs = config.operationTimeouts['feed_list'] ?? config.operationTimeouts['default'] ?? 60_000;
const feeds = await browser.withPage(PLATFORM, async (page) => listFeeds(page), timeoutMs);
const limit = clampPageSize(query.max_count);
const paged = paginateByKeyset(feeds, limit, decodeKeysetCursor(query.cursor), (item) => item.id);
res.json(successResponse({
items: paged.items,
pagination: {
mode: 'keyset',
cursor: query.cursor ?? '',
max_count: limit,
returned: paged.items.length,
...(paged.nextCursor ? { next_cursor: paged.nextCursor } : {}),
},
}));
} catch (err) {
handleError(res, err);
}
})();
});
router.post('/search', readRateLimiter, (req, res) => {
void (async () => {
try {
const body = SearchBodySchema.parse(req.body);
const timeoutMs = config.operationTimeouts['search'] ?? config.operationTimeouts['default'] ?? 60_000;
const items = await browser.withPage(PLATFORM, async (page) => searchFeeds(page, body.keyword), timeoutMs);
const limit = clampPageSize(body.max_count);
const paged = paginateByKeyset(items, limit, decodeKeysetCursor(body.cursor), (item) => item.id);
res.json(successResponse({
items: paged.items,
pagination: {
mode: 'keyset',
cursor: body.cursor ?? '',
max_count: limit,
returned: paged.items.length,
...(paged.nextCursor ? { next_cursor: paged.nextCursor } : {}),
},
}));
} catch (err) {
handleError(res, err);
}
})();
});
router.post('/feeds/detail', readRateLimiter, (req, res) => {
void (async () => {
try {
const body = FeedDetailBodySchema.parse(req.body);
const target = resolveFeedTarget(body);
const timeoutMs = config.operationTimeouts['feed_detail'] ?? config.operationTimeouts['default'] ?? 60_000;
const detail = await browser.withPage(PLATFORM, async (page) => getFeedDetail(page, target.linkId), timeoutMs);
const { comments, ...rest } = detail;
res.json(successResponse({
detail: rest,
first_screen_comments: comments,
}));
} catch (err) {
handleError(res, err);
}
})();
});
router.post('/feeds/sub-comments', readRateLimiter, (req, res) => {
void (async () => {
try {
const body = SubCommentsBodySchema.parse(req.body);
const timeoutMs = config.operationTimeouts['feed_detail'] ?? config.operationTimeouts['default'] ?? 60_000;
const loaded = await browser.withPage(
PLATFORM,
async (page) => getSubComments(page, body.link_id, body.comment_id, MAX_PAGE_SIZE),
timeoutMs,
);
const sorted = [...loaded].sort((a, b) => {
const timeCmp = a.createTime.localeCompare(b.createTime);
if (timeCmp !== 0) return timeCmp;
return a.id.localeCompare(b.id);
});
const limit = clampPageSize(body.max_count);
const paged = paginateByKeyset(
sorted,
limit,
decodeKeysetCursor(body.cursor),
(item) => `${item.createTime}|${item.id}`,
);
res.json(successResponse({
items: paged.items,
pagination: {
mode: 'keyset',
cursor: body.cursor ?? '',
max_count: limit,
returned: paged.items.length,
...(paged.nextCursor ? { next_cursor: paged.nextCursor } : {}),
},
}));
} catch (err) {
handleError(res, err);
}
})();
});
router.post('/user/profile', readRateLimiter, (req, res) => {
void (async () => {
try {
const body = UserProfileBodySchema.parse(req.body);
const target = resolveUserTarget(body);
const timeoutMs = config.operationTimeouts['user_profile'] ?? config.operationTimeouts['default'] ?? 60_000;
const profile = await browser.withPage(PLATFORM, async (page) => getUserProfile(page, target.userId), timeoutMs);
res.json(successResponse({
profile: {
id: profile.id,
nickname: profile.nickname,
avatar: profile.avatar,
description: profile.description,
follows: profile.follows,
fans: profile.fans,
likes: profile.likes,
},
recent_posts: profile.posts,
}));
} catch (err) {
handleError(res, err);
}
})();
});
router.get('/my-posts', readRateLimiter, (req, res) => {
void (async () => {
try {
const query = z.object({
type: ListMyPostsSchema.type,
max_count: z.coerce.number().int().min(1).max(200).optional().default(20),
cursor: z.string().optional(),
}).parse(req.query);
const timeoutMs = config.operationTimeouts['feed_list'] ?? config.operationTimeouts['default'] ?? 60_000;
const posts = await browser.withPage(PLATFORM, async (page) => listMyPosts(page, query.type), timeoutMs);
const limit = clampPageSize(query.max_count);
const paged = paginateByKeyset(
posts,
limit,
decodeKeysetCursor(query.cursor),
(item) => `${item.modifyTime ?? item.createTime ?? ''}|${item.id}`,
);
res.json(successResponse({
items: paged.items,
pagination: {
mode: 'keyset',
cursor: query.cursor ?? '',
max_count: limit,
returned: paged.items.length,
...(paged.nextCursor ? { next_cursor: paged.nextCursor } : {}),
},
}));
} catch (err) {
handleError(res, err);
}
})();
});
router.post('/comment', writeRateLimiter, (req, res) => {
void (async () => {
try {
const body = PostCommentBodySchema.parse(req.body);
const timeoutMs = config.operationTimeouts['comment'] ?? config.operationTimeouts['default'] ?? 60_000;
const result = await runWithIdempotency(
'xhh_post_comment',
body.request_id,
{
link_id: body.link_id,
content: body.content,
},
async () => browser.withPage(
PLATFORM,
async (page) => postComment(page, body.link_id, body.content),
timeoutMs,
),
);
res.json(successResponse({
...result.data,
...(result.meta ? { meta: result.meta } : {}),
}));
} catch (err) {
handleError(res, err);
}
})();
});
router.post('/comment/reply', writeRateLimiter, (req, res) => {
void (async () => {
try {
const body = ReplyCommentBodySchema.parse(req.body);
const timeoutMs = config.operationTimeouts['reply'] ?? config.operationTimeouts['default'] ?? 60_000;
const result = await runWithIdempotency(
'xhh_reply_comment',
body.request_id,
{
link_id: body.link_id,
comment_id: body.comment_id,
content: body.content,
},
async () => browser.withPage(
PLATFORM,
async (page) => replyComment(page, body.link_id, body.comment_id, body.content),
timeoutMs,
),
);
res.json(successResponse({
...result.data,
...(result.meta ? { meta: result.meta } : {}),
}));
} catch (err) {
handleError(res, err);
}
})();
});
router.post('/like/set-state', writeRateLimiter, (req, res) => {
void (async () => {
try {
const body = LikeBodySchema.parse(req.body);
const timeoutMs = config.operationTimeouts['like'] ?? config.operationTimeouts['default'] ?? 60_000;
const result = await browser.withPage(
PLATFORM,
async (page) => setLikeState(page, body.link_id, body.liked),
timeoutMs,
);
res.json(successResponse(result));
} catch (err) {
handleError(res, err);
}
})();
});
router.post('/favorite/set-state', writeRateLimiter, (req, res) => {
void (async () => {
try {
const body = FavoriteBodySchema.parse(req.body);
const timeoutMs = config.operationTimeouts['favorite'] ?? config.operationTimeouts['default'] ?? 60_000;
const result = await browser.withPage(
PLATFORM,
async (page) => setFavoriteState(page, body.link_id, body.favorited),
timeoutMs,
);
res.json(successResponse(result));
} catch (err) {
handleError(res, err);
}
})();
});
return router;
}
function handleError(res: { status: (n: number) => { json: (body: ApiErrorResponse) => void } }, err: unknown): void {
if (err instanceof ZodError) {
const detail = err.issues.map((issue) => `${issue.path.join('.') || '<root>'}: ${issue.message}`).join('; ');
res.status(400).json(errorResponse('VALIDATION_ERROR', detail));
return;
}
const e = err instanceof Error ? err : new Error(String(err));
const category = classifyError(e);
const message = sanitizeErrorMessage(e.message);
const statusCode = category === 'AUTH_REQUIRED' ? 401 : 500;
res.status(statusCode).json(errorResponse(category, message));
}
@@ -0,0 +1,82 @@
import { z } from 'zod';
export const CheckLoginSchema = {};
export const GetLoginQRCodeSchema = {};
export const DeleteCookiesSchema = {};
export const ListFeedsSchema = {
max_count: z
.number()
.int()
.min(1)
.max(200)
.optional()
.default(20)
.describe('Maximum number of feeds to return per page (1-200, default 20)'),
cursor: z
.string()
.optional()
.describe('Keyset pagination cursor returned by previous call'),
};
export const SearchSchema = {
keyword: z.string().min(1).describe('Search keyword'),
max_count: ListFeedsSchema.max_count,
cursor: ListFeedsSchema.cursor,
};
export const GetFeedDetailSchema = {
link_id: z.string().optional().describe('Link ID (required when url is absent)'),
url: z.string().optional().describe('Detail page URL (auto-parse link_id)'),
};
export const GetSubCommentsSchema = {
link_id: z.string().describe('Link ID'),
comment_id: z.string().describe('Parent comment ID'),
max_count: ListFeedsSchema.max_count,
cursor: ListFeedsSchema.cursor,
};
export const GetUserProfileSchema = {
user_id: z.string().optional().describe('User ID (required when url is absent)'),
url: z.string().optional().describe('User profile URL (auto-parse user_id)'),
};
export const ListMyPostsSchema = {
type: z
.enum(['all', 'article', 'image_text', 'video'])
.optional()
.default('all')
.describe('Post type filter'),
max_count: ListFeedsSchema.max_count,
cursor: ListFeedsSchema.cursor,
};
export const PostCommentSchema = {
request_id: z
.string()
.min(1)
.max(128)
.optional()
.describe('Optional idempotency key'),
link_id: z.string().describe('Link ID'),
content: z.string().min(1).describe('Comment content'),
};
export const ReplyCommentSchema = {
request_id: PostCommentSchema.request_id,
link_id: z.string().describe('Link ID'),
comment_id: z.string().describe('Target comment ID'),
content: z.string().min(1).describe('Reply content'),
};
export const SetLikeStateSchema = {
link_id: z.string().describe('Link ID'),
liked: z.boolean().describe('Target like state'),
};
export const SetFavoriteStateSchema = {
link_id: z.string().describe('Link ID'),
favorited: z.boolean().describe('Target favorite state'),
};
@@ -0,0 +1,9 @@
import type { Page } from 'rebrowser-playwright';
import type { Feed } from './types.js';
import { searchFeeds as runSearch } from './feeds.js';
export async function searchFeeds(page: Page, keyword: string): Promise<Feed[]> {
return runSearch(page, keyword);
}
@@ -0,0 +1,147 @@
export const XHH_SELECTORS = {
login: {
loggedInIndicators: [
'.user-profile-user-head',
'.user-info .user-name',
'.view-header__user-box',
],
loginButton: '.user-box__login, .login-btn, button:has-text("登录")',
qrCodeImage: [
'#login-qrcode img',
'#login-qrcode canvas',
'.qr-code-wrapper img',
'.qrcode-box img',
'img[src*="qrcode"]',
],
username: '.user-profile-user-head .name, .user-info .user-name',
avatar: '.user-profile-user-head img, .user-info img.user-image',
userLink: 'a[href*="/app/user/profile/"]',
},
feed: {
card: [
'.content-management-home__content',
'.hb-cpt__moment-list-content',
'.related-recommend__link-item--content',
'.bbs-home__content-list > *',
],
link: 'a[href*="/app/bbs/link/"]',
title: [
'.link-item__title',
'.content-list__title',
'.article-title .title',
'.title',
],
description: [
'.link-item__desc',
'.content-list__desc',
'.article-desc',
'.desc',
],
cover: 'img',
userLink: 'a[href*="/app/user/profile/"]',
userName: [
'.list-content__username',
'.user-name',
'.name',
],
likeCount: [
'.content-list__like-cnt',
'.like-count',
'.link-award-num',
],
commentCount: [
'.content-list__comment-cnt',
'.comment-count',
'.comment-num',
],
},
detail: {
title: [
'.link-detail__title',
'.bbs-link__title',
'.article-title .title',
'h1',
],
description: [
'.link-detail__content',
'.bbs-link__content',
'.article-content',
'.description',
],
image: '.article-content img, .bbs-link img, .link-detail img',
userLink: 'a[href*="/app/user/profile/"]',
userName: [
'.user-profile-user-head .name',
'.user-info .user-name',
'.header .name',
],
userAvatar: '.user-profile-user-head img, .user-info img, .header img',
commentItem: [
'.comment-item',
'.bbs-comment-item',
'.link-comment-item',
'.comment__item',
'[id*="comment"]',
],
subCommentItem: [
'.sub-comment-item',
'.reply-item',
'.sub-comment',
'.child-comment',
],
commentAuthor: '.name, .nickname, a[href*="/app/user/profile/"]',
commentAvatar: 'img',
commentContent: '.content, .comment-content, p',
commentTime: '.time, .date, .create-time',
commentLikeCount: '.like-count, .like .count',
commentReplyButton: 'button:has-text("回复"), .reply-btn, .comment-reply',
commentExpandReplies: 'button:has-text("展开"), .show-more, .expand-replies',
likeButton: [
'.engage-bar-style .like-wrapper',
'.like-wrapper',
'button:has(.heybox-bbs_thumbs-up_line_24x24)',
'button:has(.heybox-bbs_thumbs-up_filled_24x24)',
],
favoriteButton: [
'.engage-bar-style .collect-wrapper',
'.collect-wrapper',
'button:has(.heybox-bbs_collect_line_24x24)',
'button:has(.heybox-bbs_collect_filled_24x24)',
],
commentCount: '.content-list__comment-cnt, .comment-count, .comment-num',
likeCount: '.content-list__like-cnt, .like-count, .link-award-num',
favoriteCount: '.favorite-count, .collect-count, .favour-count',
commentInput: [
'textarea[placeholder*="评论"]',
'textarea',
'[contenteditable="true"][placeholder*="评论"]',
'[contenteditable="true"]',
],
commentSubmit: [
'button:has-text("发送")',
'button:has-text("发布")',
'button:has-text("评论")',
'.comment-submit',
'.submit',
],
},
profile: {
nickname: '.user-profile-user-head .name, .user-info .user-name',
avatar: '.user-profile-user-head img, .user-info img.user-image',
description: '.user-profile-user-head .desc, .user-info .user-desc, .signature',
followCount: '.bbs-info-item .value, .follow-num',
postLink: 'a[href*="/app/bbs/link/"]',
},
myPosts: {
tabButton: '.creator-content-management__tabs button',
postItem: '.content-management-home__content',
postLink: 'a[href*="/app/bbs/link/"]',
title: '.link-item__title, .title',
description: '.link-item__desc, .desc',
time: '.time, .date',
},
} as const;
@@ -0,0 +1,52 @@
import { extractLinkIdFromUrl, extractUserIdFromUrl } from './extractors.js';
interface FeedTargetInput {
link_id?: string;
url?: string;
}
interface UserTargetInput {
user_id?: string;
url?: string;
}
export interface FeedTargetResolved {
linkId: string;
}
export interface UserTargetResolved {
userId: string;
}
function normalizeUrl(url: string): string {
const trimmed = url.trim();
if (!trimmed) {
throw new Error('url cannot be empty');
}
return trimmed;
}
export function resolveFeedTarget(input: FeedTargetInput): FeedTargetResolved {
const direct = input.link_id?.trim();
if (direct) return { linkId: direct };
if (input.url) {
const parsed = extractLinkIdFromUrl(normalizeUrl(input.url));
if (parsed) return { linkId: parsed };
}
throw new Error('xhh_get_feed_detail requires link_id or url containing link_id');
}
export function resolveUserTarget(input: UserTargetInput): UserTargetResolved {
const direct = input.user_id?.trim();
if (direct) return { userId: direct };
if (input.url) {
const parsed = extractUserIdFromUrl(normalizeUrl(input.url));
if (parsed) return { userId: parsed };
}
throw new Error('xhh_get_user_profile requires user_id or url containing user_id');
}
@@ -0,0 +1,74 @@
export interface LoginStatus {
loggedIn: boolean;
username?: string;
avatar?: string;
userId?: string;
}
export interface QRCodeResult {
qrcodeData: string;
alreadyLoggedIn: boolean;
timeout: string;
}
export interface FeedUser {
id: string;
nickname: string;
avatar: string;
}
export interface Feed {
id: string;
title: string;
description: string;
coverUrl: string;
likeCount: number;
commentCount: number;
user: FeedUser;
linkUrl: string;
}
export interface Comment {
id: string;
parentId?: string;
userId: string;
nickname: string;
avatar: string;
content: string;
likeCount: number;
createTime: string;
subComments: Comment[];
}
export interface FeedDetail {
id: string;
title: string;
description: string;
images: string[];
likeCount: number;
favoriteCount: number;
commentCount: number;
isLiked: boolean;
isFavorited: boolean;
user: FeedUser;
comments: Comment[];
}
export interface UserProfile {
id: string;
nickname: string;
avatar: string;
description: string;
follows: number;
fans: number;
likes: number;
posts: Feed[];
}
export type MyPostType = 'all' | 'article' | 'image_text' | 'video';
export interface MyPost extends Feed {
type: MyPostType;
createTime?: string;
modifyTime?: string;
}
@@ -0,0 +1,72 @@
import type { Page } from 'rebrowser-playwright';
import { logger } from '@social/core/utils/logger.js';
import { XHH_SELECTORS } from './selectors.js';
import type { UserProfile } from './types.js';
import { detectCaptchaText, parseCountString } from './extractors.js';
import { listFeeds } from './feeds.js';
const log = logger.child({ module: 'xhh-user-profile' });
function buildProfileUrl(userId: string): string {
return `https://www.xiaoheihe.cn/app/user/profile/${encodeURIComponent(userId)}`;
}
export async function getUserProfile(page: Page, userId: string): Promise<UserProfile> {
await page.goto(buildProfileUrl(userId), { waitUntil: 'domcontentloaded' });
await page.waitForTimeout(1_200);
const text = await page.textContent('body').catch(() => '');
if (text && detectCaptchaText(text)) {
throw new Error('CAPTCHA_REQUIRED: captcha detected on user profile page');
}
const raw = await page.evaluate((selectors) => {
const pickText = (selector: string) =>
(document.querySelector(selector)?.textContent ?? '').trim();
const pickAttr = (selector: string, attr: string) =>
(document.querySelector(selector)?.getAttribute(attr) ?? '').trim();
const counters = [...document.querySelectorAll(selectors.profile.followCount)]
.map((node) => (node.textContent ?? '').trim())
.filter(Boolean);
const postLinks = [...document.querySelectorAll<HTMLAnchorElement>(selectors.profile.postLink)]
.map((node) => node.getAttribute('href') ?? '')
.filter(Boolean);
return {
nickname: pickText(selectors.profile.nickname),
avatar: pickAttr(selectors.profile.avatar, 'src'),
description: pickText(selectors.profile.description),
counters,
postLinks,
};
}, XHH_SELECTORS);
const [followRaw, fansRaw, likesRaw] = raw.counters;
const recentPosts = await listFeeds(page).catch(() => []);
const filteredPosts = recentPosts
.filter((item) => item.user.id === userId || raw.postLinks.some((href: string) => href.includes(item.id)))
.slice(0, 20);
const profile: UserProfile = {
id: userId,
nickname: raw.nickname,
avatar: raw.avatar,
description: raw.description,
follows: parseCountString(followRaw),
fans: parseCountString(fansRaw),
likes: parseCountString(likesRaw),
posts: filteredPosts,
};
if (!profile.nickname && !profile.avatar) {
throw new Error('waiting for selector: xhh profile not found');
}
log.info({ userId, posts: profile.posts.length }, 'xhh user profile extracted');
return profile;
}
+45
View File
@@ -0,0 +1,45 @@
import { describe, expect, it } from 'vitest';
import {
decodeKeysetCursor,
encodeKeysetCursor,
paginateByKeyset,
} from '../src/platforms/xiaoheihe/cursor.js';
describe('xhh keyset cursor', () => {
it('encodes and decodes cursor payload', () => {
const encoded = encodeKeysetCursor({ key: 'abc-123' });
const decoded = decodeKeysetCursor(encoded);
expect(decoded).toEqual({ key: 'abc-123' });
});
it('throws on invalid cursor payload', () => {
expect(() => decodeKeysetCursor('not-base64')).toThrow();
});
it('paginates deterministically without duplicates', () => {
const items = [
{ id: 'a' },
{ id: 'b' },
{ id: 'c' },
{ id: 'd' },
{ id: 'e' },
];
const page1 = paginateByKeyset(items, 2, undefined, (item) => item.id);
expect(page1.items.map((i) => i.id)).toEqual(['a', 'b']);
expect(page1.nextCursor).toBeTruthy();
const page2 = paginateByKeyset(items, 2, decodeKeysetCursor(page1.nextCursor), (item) => item.id);
expect(page2.items.map((i) => i.id)).toEqual(['c', 'd']);
expect(page2.nextCursor).toBeTruthy();
const page3 = paginateByKeyset(items, 2, decodeKeysetCursor(page2.nextCursor), (item) => item.id);
expect(page3.items.map((i) => i.id)).toEqual(['e']);
expect(page3.hasMore).toBe(false);
const combined = [...page1.items, ...page2.items, ...page3.items].map((i) => i.id);
expect(combined).toEqual(['a', 'b', 'c', 'd', 'e']);
});
});
+36
View File
@@ -0,0 +1,36 @@
import { describe, expect, it } from 'vitest';
import {
detectCaptchaText,
extractLinkIdFromUrl,
extractUserIdFromUrl,
firstNonEmpty,
parseCountString,
} from '../src/platforms/xiaoheihe/extractors.js';
describe('xhh extractors', () => {
it('parses count strings', () => {
expect(parseCountString('123')).toBe(123);
expect(parseCountString('1.2万')).toBe(12000);
expect(parseCountString('')).toBe(0);
});
it('detects captcha text', () => {
expect(detectCaptchaText('show_captcha')).toBe(true);
expect(detectCaptchaText('请完成验证码')).toBe(true);
expect(detectCaptchaText('normal page')).toBe(false);
});
it('extracts link_id and user_id from url', () => {
expect(extractLinkIdFromUrl('https://www.xiaoheihe.cn/app/bbs/link/123456')).toBe('123456');
expect(extractLinkIdFromUrl('/app/bbs/link/998877')).toBe('998877');
expect(extractUserIdFromUrl('https://www.xiaoheihe.cn/app/user/profile/112233')).toBe('112233');
expect(extractUserIdFromUrl('/app/user/profile/778899')).toBe('778899');
});
it('returns first non-empty value', () => {
expect(firstNonEmpty('', ' ', 'x', 'y')).toBe('x');
expect(firstNonEmpty('', ' ')).toBe('');
});
});
+49
View File
@@ -0,0 +1,49 @@
import { describe, expect, it } from 'vitest';
import { z } from 'zod';
import {
GetFeedDetailSchema,
ListFeedsSchema,
PostCommentSchema,
ReplyCommentSchema,
SearchSchema,
SetFavoriteStateSchema,
SetLikeStateSchema,
} from '../src/platforms/xiaoheihe/schemas.js';
describe('xhh schemas', () => {
it('validates list/query boundaries', () => {
const schema = z.object(ListFeedsSchema);
expect(schema.parse({ max_count: 20 }).max_count).toBe(20);
expect(() => schema.parse({ max_count: 0 })).toThrow();
expect(() => schema.parse({ max_count: 201 })).toThrow();
});
it('validates search required keyword', () => {
const schema = z.object(SearchSchema);
expect(() => schema.parse({})).toThrow();
expect(schema.parse({ keyword: 'aaa' }).keyword).toBe('aaa');
});
it('allows feed detail by link_id or url', () => {
const schema = z.object(GetFeedDetailSchema);
expect(schema.parse({ link_id: '123' }).link_id).toBe('123');
expect(schema.parse({ url: 'https://www.xiaoheihe.cn/app/bbs/link/123' }).url).toContain('/app/bbs/link/');
});
it('validates comment payloads', () => {
const postSchema = z.object(PostCommentSchema);
const replySchema = z.object(ReplyCommentSchema);
expect(postSchema.parse({ link_id: '1', content: 'hi' }).content).toBe('hi');
expect(replySchema.parse({ link_id: '1', comment_id: '2', content: 'ok' }).comment_id).toBe('2');
expect(() => postSchema.parse({ link_id: '1', content: '' })).toThrow();
});
it('validates set-state tools', () => {
const likeSchema = z.object(SetLikeStateSchema);
const favSchema = z.object(SetFavoriteStateSchema);
expect(likeSchema.parse({ link_id: '1', liked: true }).liked).toBe(true);
expect(favSchema.parse({ link_id: '1', favorited: false }).favorited).toBe(false);
});
});
@@ -0,0 +1,30 @@
import { describe, expect, it } from 'vitest';
import { resolveFeedTarget, resolveUserTarget } from '../src/platforms/xiaoheihe/target-resolver.js';
describe('xhh target resolver', () => {
it('resolves feed target from link_id', () => {
expect(resolveFeedTarget({ link_id: '123' })).toEqual({ linkId: '123' });
});
it('resolves feed target from url', () => {
expect(resolveFeedTarget({ url: 'https://www.xiaoheihe.cn/app/bbs/link/123' })).toEqual({ linkId: '123' });
});
it('throws on invalid feed target', () => {
expect(() => resolveFeedTarget({})).toThrow();
});
it('resolves user target from user_id', () => {
expect(resolveUserTarget({ user_id: '999' })).toEqual({ userId: '999' });
});
it('resolves user target from url', () => {
expect(resolveUserTarget({ url: 'https://www.xiaoheihe.cn/app/user/profile/888' })).toEqual({ userId: '888' });
});
it('throws on invalid user target', () => {
expect(() => resolveUserTarget({})).toThrow();
});
});
+7
View File
@@ -0,0 +1,7 @@
{
"extends": "../../tsconfig.base.json",
"compilerOptions": {
"outDir": "dist"
},
"include": ["src"]
}
+24
View File
@@ -0,0 +1,24 @@
import { defineConfig } from 'tsup';
export default defineConfig({
entry: ['src/main.ts'],
noExternal: [/^@social\/core/],
external: [
'@modelcontextprotocol/sdk',
/^@modelcontextprotocol\/sdk\//,
'express',
'pino',
'pino-pretty',
'rebrowser-playwright',
'chromium-bidi/lib/cjs/bidiMapper/BidiMapper',
'chromium-bidi/lib/cjs/cdp/CdpConnection',
],
format: ['esm'],
target: 'node22',
outDir: 'dist',
clean: true,
sourcemap: true,
dts: false,
splitting: false,
shims: false,
});
+14
View File
@@ -0,0 +1,14 @@
import { defineConfig } from 'vitest/config';
import path from 'node:path';
export default defineConfig({
resolve: {
alias: {
'@social/core': path.resolve(__dirname, '../../packages/core/src'),
},
},
test: {
include: ['test/**/*.test.ts'],
environment: 'node',
},
});
+29
View File
@@ -0,0 +1,29 @@
{
"name": "@social/xhs-mcp",
"version": "0.1.0",
"private": true,
"type": "module",
"main": "dist/main.js",
"bin": {
"mcp-xhs": "dist/main.js"
},
"scripts": {
"build": "tsup",
"lint": "tsc --noEmit",
"test": "vitest run",
"start": "PORT=${PORT:-9527} COOKIE_DIR=${COOKIE_DIR:-$HOME/.social-mcp-xhs} node dist/main.js",
"dev": "pnpm build && pnpm start"
},
"dependencies": {
"@modelcontextprotocol/sdk": "^1.27.0",
"@social/core": "workspace:*",
"zod": "^3.25.0"
},
"devDependencies": {
"@types/node": "^22.0.0",
"tsup": "^8.0.0",
"typescript": "^5.7.0",
"vitest": "^3.0.0"
}
}
+5
View File
@@ -0,0 +1,5 @@
import { startServerWithPlugins } from '@social/core/server/bootstrap.js';
import { xiaohongshuPlugin } from './platforms/xiaohongshu/index.js';
startServerWithPlugins([xiaohongshuPlugin]);
@@ -1,6 +1,6 @@
import type { Page } from 'rebrowser-playwright'; import type { Page } from 'rebrowser-playwright';
import { logger } from '../../utils/logger.js'; import { logger } from '@social/core/utils/logger.js';
import { XHS_SELECTORS } from './selectors.js'; import { XHS_SELECTORS } from './selectors.js';
// --------------------------------------------------------------------------- // ---------------------------------------------------------------------------
@@ -38,7 +38,7 @@ export async function postComment(
feedId: string, feedId: string,
xsecToken: string, xsecToken: string,
content: string, content: string,
): Promise<{ success: boolean }> { ): Promise<{ success: boolean; comment_id?: string }> {
log.info({ feedId }, 'Posting comment on note'); log.info({ feedId }, 'Posting comment on note');
// ------------------------------------------------------------------------- // -------------------------------------------------------------------------
@@ -91,11 +91,16 @@ export async function postComment(
// Check for the comment text in the page to verify success. // Check for the comment text in the page to verify success.
const pageContent = await page.content(); const pageContent = await page.content();
const success = pageContent.includes(content.slice(0, 20)); const textHit = pageContent.includes(content.slice(0, 20));
const commentId = await extractTopLevelCommentId(page, content);
const success = textHit || !!commentId;
log.info({ feedId, success }, 'Comment post complete'); log.info({ feedId, success, commentId }, 'Comment post complete');
return { success }; return {
success,
...(commentId ? { comment_id: commentId } : {}),
};
} }
// --------------------------------------------------------------------------- // ---------------------------------------------------------------------------
@@ -120,7 +125,7 @@ export async function replyComment(
content: string, content: string,
commentId?: string, commentId?: string,
userId?: string, userId?: string,
): Promise<{ success: boolean }> { ): Promise<{ success: boolean; reply_id?: string }> {
log.info({ feedId, commentId, userId }, 'Replying to comment on note'); log.info({ feedId, commentId, userId }, 'Replying to comment on note');
// ------------------------------------------------------------------------- // -------------------------------------------------------------------------
@@ -212,11 +217,16 @@ export async function replyComment(
await page.waitForTimeout(SUBMIT_SETTLE_MS); await page.waitForTimeout(SUBMIT_SETTLE_MS);
const pageContent = await page.content(); const pageContent = await page.content();
const success = pageContent.includes(content.slice(0, 20)); const textHit = pageContent.includes(content.slice(0, 20));
const replyId = await extractReplyCommentId(page, content, commentId);
const success = textHit || !!replyId;
log.info({ feedId, commentId, success }, 'Reply post complete'); log.info({ feedId, commentId, success, replyId }, 'Reply post complete');
return { success }; return {
success,
...(replyId ? { reply_id: replyId } : {}),
};
} }
// --------------------------------------------------------------------------- // ---------------------------------------------------------------------------
@@ -320,3 +330,149 @@ async function submitComment(page: Page): Promise<boolean> {
await submitBtn.click(); await submitBtn.click();
return true; return true;
} }
function normalizeText(input: string): string {
return input.replace(/\s+/g, ' ').trim();
}
function parseCommentElementId(value: string | null): string {
if (!value) return '';
return value.replace(/^comment-/, '').trim();
}
function matchContent(candidate: string, target: string): boolean {
const c = normalizeText(candidate);
const t = normalizeText(target);
if (!c || !t) return false;
return c === t || c.includes(t) || t.includes(c);
}
async function extractTopLevelCommentId(
page: Page,
content: string,
): Promise<string | undefined> {
const fromStore = await page.evaluate((targetContent: string) => {
const match = (candidate: unknown): boolean => {
const c = String(candidate ?? '').replace(/\s+/g, ' ').trim();
const t = String(targetContent ?? '').replace(/\s+/g, ' ').trim();
if (!c || !t) return false;
return c === t || c.includes(t) || t.includes(c);
};
const normalizeId = (id: unknown): string =>
String(id ?? '').replace(/^comment-/, '').trim();
const state = (window as unknown as Record<string, unknown>).__INITIAL_STATE__ as
Record<string, unknown> | undefined;
const note = state?.note as Record<string, unknown> | undefined;
const map = note?.noteDetailMap as Record<string, Record<string, unknown>> | undefined;
if (!map) return null;
for (const entry of Object.values(map)) {
const comments = entry?.comments as { list?: Array<Record<string, unknown>> } | undefined;
if (!comments?.list) continue;
for (const one of comments.list) {
if (match(one.content)) {
const id = normalizeId(one.id);
if (id) return id;
}
}
}
return null;
}, content).catch(() => null);
if (typeof fromStore === 'string' && fromStore) {
return fromStore;
}
const candidates = await page.$$('.parent-comment .comment-item, [id^="comment-"]');
for (const candidate of candidates) {
const text = (await candidate.$eval('.content', (el) => el.textContent ?? '').catch(() => ''));
if (!matchContent(text, content)) continue;
const id = parseCommentElementId(
await candidate.getAttribute('id').catch(() => null),
) || parseCommentElementId(
await candidate.getAttribute('data-comment-id').catch(() => null),
) || parseCommentElementId(
await candidate.getAttribute('data-id').catch(() => null),
);
if (id) return id;
}
return undefined;
}
async function extractReplyCommentId(
page: Page,
content: string,
parentCommentId?: string,
): Promise<string | undefined> {
const fromStore = await page.evaluate(
(args: { targetContent: string; parentCommentId?: string }) => {
const match = (candidate: unknown): boolean => {
const c = String(candidate ?? '').replace(/\s+/g, ' ').trim();
const t = String(args.targetContent ?? '').replace(/\s+/g, ' ').trim();
if (!c || !t) return false;
return c === t || c.includes(t) || t.includes(c);
};
const normalizeId = (id: unknown): string =>
String(id ?? '').replace(/^comment-/, '').trim();
const state = (window as unknown as Record<string, unknown>).__INITIAL_STATE__ as
Record<string, unknown> | undefined;
const note = state?.note as Record<string, unknown> | undefined;
const map = note?.noteDetailMap as Record<string, Record<string, unknown>> | undefined;
if (!map) return null;
for (const entry of Object.values(map)) {
const comments = entry?.comments as { list?: Array<Record<string, unknown>> } | undefined;
if (!comments?.list) continue;
for (const parent of comments.list) {
const parentId = normalizeId(parent.id);
if (args.parentCommentId && parentId !== args.parentCommentId) continue;
const subs = (parent.subComments ?? parent.sub_comments ?? []) as Array<Record<string, unknown>>;
for (const sub of subs) {
if (match(sub.content)) {
const id = normalizeId(sub.id);
if (id) return id;
}
}
}
}
return null;
},
{ targetContent: content, parentCommentId },
).catch(() => null);
if (typeof fromStore === 'string' && fromStore) {
return fromStore;
}
const parentCandidates = parentCommentId
? await page.$$(
`[id="comment-${parentCommentId}"], [data-comment-id="${parentCommentId}"], [data-id="${parentCommentId}"]`,
)
: await page.$$('.parent-comment');
for (const parent of parentCandidates) {
const replyItems = await parent.$$('.sub-comment-item, [id^="comment-"]');
for (const item of replyItems) {
const text = await item
.$eval('.content', (el) => el.textContent ?? '')
.catch(() => item.textContent().catch(() => ''));
if (!matchContent(text ?? '', content)) continue;
const id = parseCommentElementId(
await item.getAttribute('id').catch(() => null),
) || parseCommentElementId(
await item.getAttribute('data-comment-id').catch(() => null),
) || parseCommentElementId(
await item.getAttribute('data-id').catch(() => null),
);
if (id) return id;
}
}
return undefined;
}
@@ -1,6 +1,6 @@
import type { Page } from 'rebrowser-playwright'; import type { Page } from 'rebrowser-playwright';
import { logger } from '../../utils/logger.js'; import { logger } from '@social/core/utils/logger.js';
import { XHS_SELECTORS } from './selectors.js'; import { XHS_SELECTORS } from './selectors.js';
import { extractInitialState, parseCountString, ensureHttps } from './feeds.js'; import { extractInitialState, parseCountString, ensureHttps } from './feeds.js';
import type { FeedDetail, Comment } from './types.js'; import type { FeedDetail, Comment } from './types.js';
@@ -297,7 +297,9 @@ export async function getSubComments(
const parentIndex = await page.evaluate((cid: string) => { const parentIndex = await page.evaluate((cid: string) => {
const parents = document.querySelectorAll('.parent-comment'); const parents = document.querySelectorAll('.parent-comment');
for (let i = 0; i < parents.length; i++) { for (let i = 0; i < parents.length; i++) {
const item = parents[i].querySelector('.comment-item'); const parent = parents.item(i);
if (!parent) continue;
const item = parent.querySelector('.comment-item');
if (!item) continue; if (!item) continue;
const id = const id =
item.getAttribute('id')?.replace(/^comment-/, '') ?? item.getAttribute('id')?.replace(/^comment-/, '') ??
@@ -1,6 +1,6 @@
import type { Page } from 'rebrowser-playwright'; import type { Page } from 'rebrowser-playwright';
import { logger } from '../../utils/logger.js'; import { logger } from '@social/core/utils/logger.js';
import type { Feed } from './types.js'; import type { Feed } from './types.js';
// --------------------------------------------------------------------------- // ---------------------------------------------------------------------------
File diff suppressed because it is too large Load Diff
@@ -1,6 +1,6 @@
import type { Page } from 'rebrowser-playwright'; import type { Page } from 'rebrowser-playwright';
import { logger } from '../../utils/logger.js'; import { logger } from '@social/core/utils/logger.js';
import { XHS_SELECTORS } from './selectors.js'; import { XHS_SELECTORS } from './selectors.js';
// --------------------------------------------------------------------------- // ---------------------------------------------------------------------------
@@ -48,6 +48,11 @@ async function readState(page: Page, btnSelector: string, activeHref: string): P
.catch(() => false); .catch(() => false);
} }
async function openFeedOverlay(page: Page, feedId: string, xsecToken: string): Promise<void> {
await page.goto(buildFeedUrl(feedId, xsecToken), { waitUntil: 'domcontentloaded' });
await page.waitForSelector(selDetail.noteContainer, { timeout: 10_000 });
}
// --------------------------------------------------------------------------- // ---------------------------------------------------------------------------
// toggleLike — pure toggle, clicks the like button once // toggleLike — pure toggle, clicks the like button once
// --------------------------------------------------------------------------- // ---------------------------------------------------------------------------
@@ -59,8 +64,7 @@ export async function toggleLike(
): Promise<{ success: boolean; liked: boolean }> { ): Promise<{ success: boolean; liked: boolean }> {
log.info({ feedId }, 'Toggling like on note'); log.info({ feedId }, 'Toggling like on note');
await page.goto(buildFeedUrl(feedId, xsecToken), { waitUntil: 'domcontentloaded' }); await openFeedOverlay(page, feedId, xsecToken);
await page.waitForSelector(selDetail.noteContainer, { timeout: 10_000 });
const clicked = await clickLastMatch(page, '.engage-bar-style .like-wrapper'); const clicked = await clickLastMatch(page, '.engage-bar-style .like-wrapper');
if (!clicked) { if (!clicked) {
@@ -86,8 +90,7 @@ export async function toggleFavorite(
): Promise<{ success: boolean; favorited: boolean }> { ): Promise<{ success: boolean; favorited: boolean }> {
log.info({ feedId }, 'Toggling favorite on note'); log.info({ feedId }, 'Toggling favorite on note');
await page.goto(buildFeedUrl(feedId, xsecToken), { waitUntil: 'domcontentloaded' }); await openFeedOverlay(page, feedId, xsecToken);
await page.waitForSelector(selDetail.noteContainer, { timeout: 10_000 });
const clicked = await clickLastMatch(page, '.engage-bar-style .collect-wrapper'); const clicked = await clickLastMatch(page, '.engage-bar-style .collect-wrapper');
if (!clicked) { if (!clicked) {
@@ -101,3 +104,71 @@ export async function toggleFavorite(
log.info({ feedId, favorited }, 'Favorite toggle complete'); log.info({ feedId, favorited }, 'Favorite toggle complete');
return { success: true, favorited }; return { success: true, favorited };
} }
// ---------------------------------------------------------------------------
// setLikeState / setFavoriteState — idempotent state-setting operations
// ---------------------------------------------------------------------------
export async function setLikeState(
page: Page,
feedId: string,
xsecToken: string,
targetLiked: boolean,
): Promise<{ success: boolean; liked: boolean; changed: boolean }> {
log.info({ feedId, targetLiked }, 'Setting like state on note');
await openFeedOverlay(page, feedId, xsecToken);
const currentLiked = await readState(page, '.engage-bar-style .like-wrapper', '#liked');
if (currentLiked === targetLiked) {
return { success: true, liked: currentLiked, changed: false };
}
const clicked = await clickLastMatch(page, '.engage-bar-style .like-wrapper');
if (!clicked) {
log.warn('Like button not found in note detail overlay');
return { success: false, liked: currentLiked, changed: false };
}
await page.waitForTimeout(TOGGLE_SETTLE_MS);
const liked = await readState(page, '.engage-bar-style .like-wrapper', '#liked');
return {
success: liked === targetLiked,
liked,
changed: liked !== currentLiked,
};
}
export async function setFavoriteState(
page: Page,
feedId: string,
xsecToken: string,
targetFavorited: boolean,
): Promise<{ success: boolean; favorited: boolean; changed: boolean }> {
log.info({ feedId, targetFavorited }, 'Setting favorite state on note');
await openFeedOverlay(page, feedId, xsecToken);
const currentFavorited = await readState(
page,
'.engage-bar-style .collect-wrapper',
'#collected',
);
if (currentFavorited === targetFavorited) {
return { success: true, favorited: currentFavorited, changed: false };
}
const clicked = await clickLastMatch(page, '.engage-bar-style .collect-wrapper');
if (!clicked) {
log.warn('Favorite button not found in note detail overlay');
return { success: false, favorited: currentFavorited, changed: false };
}
await page.waitForTimeout(TOGGLE_SETTLE_MS);
const favorited = await readState(page, '.engage-bar-style .collect-wrapper', '#collected');
return {
success: favorited === targetFavorited,
favorited,
changed: favorited !== currentFavorited,
};
}
@@ -1,10 +1,10 @@
import { chromium } from 'rebrowser-playwright'; import { chromium } from 'rebrowser-playwright';
import type { Page, BrowserContext } from 'rebrowser-playwright'; import type { Page, BrowserContext } from 'rebrowser-playwright';
import type { BrowserManager } from '../../browser/manager.js'; import type { BrowserManager } from '@social/core/browser/manager.js';
import { config } from '../../config/index.js'; import { config } from '@social/core/config/index.js';
import { logger } from '../../utils/logger.js'; import { logger } from '@social/core/utils/logger.js';
import { cookieStore } from '../../cookie/store.js'; import { cookieStore } from '@social/core/cookie/store.js';
import { XHS_SELECTORS } from './selectors.js'; import { XHS_SELECTORS } from './selectors.js';
import type { LoginStatus, QRCodeResult } from './types.js'; import type { LoginStatus, QRCodeResult } from './types.js';
@@ -1,6 +1,6 @@
import type { Page } from 'rebrowser-playwright'; import type { Page } from 'rebrowser-playwright';
import { logger } from '../../utils/logger.js'; import { logger } from '@social/core/utils/logger.js';
// --------------------------------------------------------------------------- // ---------------------------------------------------------------------------
// Constants // Constants
@@ -0,0 +1,446 @@
import crypto from 'node:crypto';
import fs from 'node:fs';
import path from 'node:path';
import { config } from '@social/core/config/index.js';
import { logger } from '@social/core/utils/logger.js';
import { DatabaseSync } from '@social/core/utils/sqlite.js';
import type { CommentNotification } from './types.js';
export type NotificationTaskStatus =
| 'new'
| 'pending'
| 'replied'
| 'failed'
| 'ignored';
interface NotificationRow {
fingerprint: string;
user_id: string;
nickname: string;
avatar: string;
content: string;
type: string;
time: string;
feed_id: string;
xsec_token: string;
note_image: string;
status: NotificationTaskStatus;
first_seen_at: number;
last_seen_at: number;
retry_count: number;
last_attempt_at: number | null;
replied_at: number | null;
reply_content: string | null;
error_message: string | null;
}
export interface NotificationTask {
fingerprint: string;
notification: CommentNotification;
status: NotificationTaskStatus;
firstSeenAt: string;
lastSeenAt: string;
retryCount: number;
lastAttemptAt?: string;
repliedAt?: string;
replyContent?: string;
errorMessage?: string;
}
export interface NotificationUpsertResult {
fetched: number;
inserted: number;
updated: number;
}
export interface NotificationKeysetCursor {
firstSeenAt: number;
fingerprint: string;
}
const PLATFORM = 'xiaohongshu';
const DB_FILENAME = 'automation.db';
const log = logger.child({ module: 'xhs-notification-state' });
export class NotificationStateStore {
private readonly db: InstanceType<typeof DatabaseSync>;
private readonly dbPath: string;
constructor(baseDir = config.cookieDir, dbFilename = DB_FILENAME) {
const dir = path.join(baseDir, PLATFORM);
fs.mkdirSync(dir, { recursive: true, mode: 0o700 });
this.dbPath = path.join(dir, dbFilename);
this.db = new DatabaseSync(this.dbPath);
this.db.exec('PRAGMA journal_mode = WAL;');
this.db.exec('PRAGMA synchronous = NORMAL;');
this.db.exec(`
CREATE TABLE IF NOT EXISTS notification_tasks (
fingerprint TEXT PRIMARY KEY,
user_id TEXT NOT NULL,
nickname TEXT NOT NULL,
avatar TEXT NOT NULL,
content TEXT NOT NULL,
type TEXT NOT NULL,
time TEXT NOT NULL,
feed_id TEXT NOT NULL,
xsec_token TEXT NOT NULL,
note_image TEXT NOT NULL,
status TEXT NOT NULL,
first_seen_at INTEGER NOT NULL,
last_seen_at INTEGER NOT NULL,
retry_count INTEGER NOT NULL DEFAULT 0,
last_attempt_at INTEGER,
replied_at INTEGER,
reply_content TEXT,
error_message TEXT
);
CREATE INDEX IF NOT EXISTS idx_notification_tasks_status_first_seen
ON notification_tasks(status, first_seen_at);
CREATE INDEX IF NOT EXISTS idx_notification_tasks_user_content_status
ON notification_tasks(user_id, content, status);
`);
log.info({ dbPath: this.dbPath }, 'Notification state store initialized');
}
buildFingerprint(notification: CommentNotification): string {
const payload = [
notification.feedId,
notification.userId,
notification.content.trim(),
notification.time.trim(),
notification.type.trim(),
].join('|');
return crypto.createHash('sha256').update(payload).digest('hex');
}
upsertNotifications(notifications: CommentNotification[]): NotificationUpsertResult {
if (notifications.length === 0) {
return { fetched: 0, inserted: 0, updated: 0 };
}
const now = Date.now();
let inserted = 0;
let updated = 0;
const selectStmt = this.db.prepare(
'SELECT fingerprint FROM notification_tasks WHERE fingerprint = ?',
);
const insertStmt = this.db.prepare(`
INSERT INTO notification_tasks (
fingerprint, user_id, nickname, avatar, content, type, time,
feed_id, xsec_token, note_image, status, first_seen_at, last_seen_at
) VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, 'new', ?, ?)
`);
const updateStmt = this.db.prepare(`
UPDATE notification_tasks
SET
nickname = ?,
avatar = ?,
type = ?,
time = ?,
feed_id = ?,
xsec_token = ?,
note_image = ?,
last_seen_at = ?
WHERE fingerprint = ?
`);
this.db.exec('BEGIN');
try {
for (const n of notifications) {
const fp = this.buildFingerprint(n);
const exists = selectStmt.get(fp) as { fingerprint: string } | undefined;
if (!exists) {
insertStmt.run(
fp,
n.userId,
n.nickname,
n.avatar,
n.content,
n.type,
n.time,
n.feedId,
n.xsecToken,
n.noteImage,
now,
now,
);
inserted++;
} else {
updateStmt.run(
n.nickname,
n.avatar,
n.type,
n.time,
n.feedId,
n.xsecToken,
n.noteImage,
now,
fp,
);
updated++;
}
}
this.db.exec('COMMIT');
} catch (err) {
this.db.exec('ROLLBACK');
throw err;
}
return {
fetched: notifications.length,
inserted,
updated,
};
}
listByStatuses(
statuses: NotificationTaskStatus[],
maxCount: number,
offset = 0,
): NotificationTask[] {
if (statuses.length === 0) return [];
const placeholders = statuses.map(() => '?').join(', ');
const query = `
SELECT
fingerprint, user_id, nickname, avatar, content, type, time,
feed_id, xsec_token, note_image, status, first_seen_at, last_seen_at,
retry_count, last_attempt_at, replied_at, reply_content, error_message
FROM notification_tasks
WHERE status IN (${placeholders})
ORDER BY first_seen_at ASC
LIMIT ?
OFFSET ?
`;
const stmt = this.db.prepare(query);
const rows = stmt.all(...statuses, maxCount, offset) as unknown as NotificationRow[];
return rows.map((r) => this.rowToTask(r));
}
listByStatusesKeyset(
statuses: NotificationTaskStatus[],
maxCount: number,
cursor?: NotificationKeysetCursor,
): { tasks: NotificationTask[]; hasMore: boolean; nextCursor?: NotificationKeysetCursor } {
if (statuses.length === 0 || maxCount <= 0) {
return { tasks: [], hasMore: false };
}
const placeholders = statuses.map(() => '?').join(', ');
const condition = cursor
? `
AND (
first_seen_at > ?
OR (first_seen_at = ? AND fingerprint > ?)
)
`
: '';
const query = `
SELECT
fingerprint, user_id, nickname, avatar, content, type, time,
feed_id, xsec_token, note_image, status, first_seen_at, last_seen_at,
retry_count, last_attempt_at, replied_at, reply_content, error_message
FROM notification_tasks
WHERE status IN (${placeholders})
${condition}
ORDER BY first_seen_at ASC, fingerprint ASC
LIMIT ?
`;
const stmt = this.db.prepare(query);
const limitWithSentinel = maxCount + 1;
const rows = cursor
? stmt.all(
...statuses,
cursor.firstSeenAt,
cursor.firstSeenAt,
cursor.fingerprint,
limitWithSentinel,
) as unknown as NotificationRow[]
: stmt.all(...statuses, limitWithSentinel) as unknown as NotificationRow[];
const hasMore = rows.length > maxCount;
const pageRows = hasMore ? rows.slice(0, maxCount) : rows;
const tasks = pageRows.map((r) => this.rowToTask(r));
if (!hasMore || pageRows.length === 0) {
return { tasks, hasMore };
}
const last = pageRows[pageRows.length - 1]!;
return {
tasks,
hasMore,
nextCursor: {
firstSeenAt: last.first_seen_at,
fingerprint: last.fingerprint,
},
};
}
countByStatuses(statuses: NotificationTaskStatus[]): number {
if (statuses.length === 0) return 0;
const placeholders = statuses.map(() => '?').join(', ');
const query = `
SELECT COUNT(1) AS count
FROM notification_tasks
WHERE status IN (${placeholders})
`;
const stmt = this.db.prepare(query);
const row = stmt.get(...statuses) as { count?: number } | undefined;
return row?.count ?? 0;
}
getByFingerprint(fingerprint: string): NotificationTask | null {
const stmt = this.db.prepare(`
SELECT
fingerprint, user_id, nickname, avatar, content, type, time,
feed_id, xsec_token, note_image, status, first_seen_at, last_seen_at,
retry_count, last_attempt_at, replied_at, reply_content, error_message
FROM notification_tasks
WHERE fingerprint = ?
LIMIT 1
`);
const row = stmt.get(fingerprint) as NotificationRow | undefined;
return row ? this.rowToTask(row) : null;
}
findOpenFingerprint(userId: string, content: string): string | null {
const stmt = this.db.prepare(`
SELECT fingerprint
FROM notification_tasks
WHERE user_id = ?
AND content = ?
AND status IN ('new', 'failed', 'pending')
ORDER BY first_seen_at ASC
LIMIT 1
`);
const row = stmt.get(userId, content) as { fingerprint: string } | undefined;
return row?.fingerprint ?? null;
}
markPending(fingerprint: string): void {
const now = Date.now();
const stmt = this.db.prepare(`
UPDATE notification_tasks
SET status = 'pending', last_attempt_at = ?, error_message = NULL
WHERE fingerprint = ?
`);
stmt.run(now, fingerprint);
}
markReplied(fingerprint: string, replyContent: string): void {
const now = Date.now();
const stmt = this.db.prepare(`
UPDATE notification_tasks
SET
status = 'replied',
replied_at = ?,
last_attempt_at = ?,
reply_content = ?,
error_message = NULL
WHERE fingerprint = ?
`);
stmt.run(now, now, replyContent, fingerprint);
}
markFailed(fingerprint: string, errorMessage: string): void {
const now = Date.now();
const stmt = this.db.prepare(`
UPDATE notification_tasks
SET
status = 'failed',
retry_count = retry_count + 1,
last_attempt_at = ?,
error_message = ?
WHERE fingerprint = ?
`);
stmt.run(now, errorMessage, fingerprint);
}
markIgnored(fingerprint: string, reason?: string): void {
const stmt = this.db.prepare(`
UPDATE notification_tasks
SET status = 'ignored', error_message = ?
WHERE fingerprint = ?
`);
stmt.run(reason ?? 'Ignored by operator', fingerprint);
}
setStatus(
fingerprint: string,
status: NotificationTaskStatus,
note?: string,
): void {
const now = Date.now();
const stmt = this.db.prepare(`
UPDATE notification_tasks
SET
status = ?,
last_attempt_at = CASE WHEN ? IN ('pending', 'failed', 'replied') THEN ? ELSE last_attempt_at END,
replied_at = CASE WHEN ? = 'replied' THEN ? ELSE replied_at END,
error_message = CASE WHEN ? = 'failed' THEN COALESCE(?, 'Marked as failed') WHEN ? = 'ignored' THEN COALESCE(?, 'Ignored by operator') ELSE error_message END,
reply_content = CASE WHEN ? = 'replied' THEN COALESCE(?, reply_content) ELSE reply_content END
WHERE fingerprint = ?
`);
stmt.run(
status,
status,
now,
status,
now,
status,
note ?? null,
status,
note ?? null,
status,
note ?? null,
fingerprint,
);
}
private rowToTask(row: NotificationRow): NotificationTask {
return {
fingerprint: row.fingerprint,
notification: {
userId: row.user_id,
nickname: row.nickname,
avatar: row.avatar,
content: row.content,
type: row.type,
time: row.time,
feedId: row.feed_id,
xsecToken: row.xsec_token,
noteImage: row.note_image,
},
status: row.status,
firstSeenAt: new Date(row.first_seen_at).toISOString(),
lastSeenAt: new Date(row.last_seen_at).toISOString(),
retryCount: row.retry_count,
...(row.last_attempt_at ? { lastAttemptAt: new Date(row.last_attempt_at).toISOString() } : {}),
...(row.replied_at ? { repliedAt: new Date(row.replied_at).toISOString() } : {}),
...(row.reply_content ? { replyContent: row.reply_content } : {}),
...(row.error_message ? { errorMessage: row.error_message } : {}),
};
}
}
let storeSingleton: NotificationStateStore | null = null;
export function getNotificationStateStore(): NotificationStateStore {
if (!storeSingleton) {
storeSingleton = new NotificationStateStore();
}
return storeSingleton;
}
@@ -0,0 +1,90 @@
import type { BrowserManager } from '@social/core/browser/manager.js';
import { config } from '@social/core/config/index.js';
import { logger } from '@social/core/utils/logger.js';
import { getCommentNotifications } from './notification.js';
import { getNotificationStateStore, type NotificationUpsertResult } from './notification-state.js';
const PLATFORM = 'xiaohongshu';
const log = logger.child({ module: 'xhs-notification-sync' });
export async function syncCommentNotifications(
browser: BrowserManager,
maxCount = config.notificationPollMaxCount,
): Promise<NotificationUpsertResult> {
const timeoutMs =
config.operationTimeouts['feed_detail'] ??
config.operationTimeouts['default'] ??
60_000;
const notifications = await browser.withPage(
PLATFORM,
async (page) => getCommentNotifications(page, maxCount),
timeoutMs,
);
const result = getNotificationStateStore().upsertNotifications(notifications);
if (result.fetched > 0) {
log.info(result, 'Notifications synced to state store');
}
return result;
}
export class XhsNotificationPoller {
private timer: ReturnType<typeof setInterval> | null = null;
private running = false;
private started = false;
start(browser: BrowserManager): void {
if (this.started) return;
this.started = true;
if (!config.notificationPollEnabled) {
log.info('Notification poller disabled by config');
return;
}
const tick = async (): Promise<void> => {
if (this.running) return;
this.running = true;
try {
await syncCommentNotifications(browser, config.notificationPollMaxCount);
} catch (err) {
const message = err instanceof Error ? err.message : String(err);
// Login expiration and empty unread state are expected in daily usage.
log.debug({ message }, 'Notification poll tick skipped');
} finally {
this.running = false;
}
};
void tick();
this.timer = setInterval(() => {
void tick();
}, config.notificationPollIntervalMs);
if (this.timer && typeof this.timer === 'object' && 'unref' in this.timer) {
this.timer.unref();
}
log.info(
{
intervalMs: config.notificationPollIntervalMs,
maxCount: config.notificationPollMaxCount,
},
'Notification poller started',
);
}
stop(): void {
if (this.timer) {
clearInterval(this.timer);
this.timer = null;
}
this.started = false;
this.running = false;
log.info('Notification poller stopped');
}
}
export const xhsNotificationPoller = new XhsNotificationPoller();
@@ -1,6 +1,6 @@
import type { Page } from 'rebrowser-playwright'; import type { Page } from 'rebrowser-playwright';
import { logger } from '../../utils/logger.js'; import { logger } from '@social/core/utils/logger.js';
import { XHS_SELECTORS } from './selectors.js'; import { XHS_SELECTORS } from './selectors.js';
import type { CommentNotification } from './types.js'; import type { CommentNotification } from './types.js';
@@ -1,6 +1,6 @@
import type { Page } from 'rebrowser-playwright'; import type { Page } from 'rebrowser-playwright';
import { logger } from '../../utils/logger.js'; import { logger } from '@social/core/utils/logger.js';
import { XHS_SELECTORS } from './selectors.js'; import { XHS_SELECTORS } from './selectors.js';
// --------------------------------------------------------------------------- // ---------------------------------------------------------------------------
@@ -1,6 +1,6 @@
import type { Page } from 'rebrowser-playwright'; import type { Page } from 'rebrowser-playwright';
import { logger } from '../../utils/logger.js'; import { logger } from '@social/core/utils/logger.js';
import { XHS_SELECTORS } from './selectors.js'; import { XHS_SELECTORS } from './selectors.js';
// --------------------------------------------------------------------------- // ---------------------------------------------------------------------------
@@ -1,13 +1,13 @@
import { Router } from 'express'; import { Router } from 'express';
import { z, ZodError } from 'zod'; import { z, ZodError } from 'zod';
import type { BrowserManager } from '../../browser/manager.js'; import type { BrowserManager } from '@social/core/browser/manager.js';
import { config } from '../../config/index.js'; import { config } from '@social/core/config/index.js';
import { logger } from '../../utils/logger.js'; import { logger } from '@social/core/utils/logger.js';
import { classifyError, sanitizeErrorMessage } from '../../utils/errors.js'; import { classifyError, sanitizeErrorMessage } from '@social/core/utils/errors.js';
import { resolveMediaInput, cleanupFile } from '../../utils/downloader.js'; import { resolveMediaInput, cleanupFile } from '@social/core/utils/downloader.js';
import { rateLimiter } from '../../server/middleware.js'; import { rateLimiter } from '@social/core/server/middleware.js';
import { cookieStore } from '../../cookie/store.js'; import { cookieStore } from '@social/core/cookie/store.js';
import { checkLoginStatus, getLoginQRCode, deleteCookies } from './login.js'; import { checkLoginStatus, getLoginQRCode, deleteCookies } from './login.js';
import { listFeeds } from './feeds.js'; import { listFeeds } from './feeds.js';
@@ -20,6 +20,13 @@ import { listMyNotes } from './my-notes.js';
import { postComment, replyComment } from './comment.js'; import { postComment, replyComment } from './comment.js';
import { toggleLike, toggleFavorite } from './interaction.js'; import { toggleLike, toggleFavorite } from './interaction.js';
import { getCommentNotifications, replyNotification } from './notification.js'; import { getCommentNotifications, replyNotification } from './notification.js';
import { resolveFeedTarget, resolveUserTarget } from './target-resolver.js';
import {
getNotificationStateStore,
type NotificationTask,
type NotificationTaskStatus,
} from './notification-state.js';
import { syncCommentNotifications } from './notification-sync.js';
import { import {
SearchSchema, SearchSchema,
@@ -34,6 +41,7 @@ import {
FavoriteSchema, FavoriteSchema,
GetCommentNotificationsSchema, GetCommentNotificationsSchema,
ReplyNotificationSchema, ReplyNotificationSchema,
GetUnprocessedNotificationsSchema,
} from './schemas.js'; } from './schemas.js';
import type { SearchFilters } from './types.js'; import type { SearchFilters } from './types.js';
@@ -65,6 +73,7 @@ const SearchBodySchema = z.object({
}); });
const FeedDetailBodySchema = z.object({ const FeedDetailBodySchema = z.object({
url: GetFeedDetailSchema.url,
feed_id: GetFeedDetailSchema.feed_id, feed_id: GetFeedDetailSchema.feed_id,
xsec_token: GetFeedDetailSchema.xsec_token, xsec_token: GetFeedDetailSchema.xsec_token,
}); });
@@ -74,9 +83,11 @@ const SubCommentsBodySchema = z.object({
xsec_token: GetSubCommentsSchema.xsec_token, xsec_token: GetSubCommentsSchema.xsec_token,
comment_id: GetSubCommentsSchema.comment_id, comment_id: GetSubCommentsSchema.comment_id,
max_count: GetSubCommentsSchema.max_count, max_count: GetSubCommentsSchema.max_count,
cursor: GetSubCommentsSchema.cursor,
}); });
const UserProfileBodySchema = z.object({ const UserProfileBodySchema = z.object({
url: GetUserProfileSchema.url,
user_id: GetUserProfileSchema.user_id, user_id: GetUserProfileSchema.user_id,
xsec_token: GetUserProfileSchema.xsec_token, xsec_token: GetUserProfileSchema.xsec_token,
}); });
@@ -119,11 +130,19 @@ const GetCommentNotificationsQuerySchema = z.object({
}); });
const ReplyNotificationBodySchema = z.object({ const ReplyNotificationBodySchema = z.object({
fingerprint: ReplyNotificationSchema.fingerprint,
user_id: ReplyNotificationSchema.user_id, user_id: ReplyNotificationSchema.user_id,
comment_content: ReplyNotificationSchema.comment_content, comment_content: ReplyNotificationSchema.comment_content,
reply_content: ReplyNotificationSchema.reply_content, reply_content: ReplyNotificationSchema.reply_content,
}); });
const GetUnprocessedNotificationsQuerySchema = z.object({
max_count: GetUnprocessedNotificationsSchema.max_count,
cursor: GetUnprocessedNotificationsSchema.cursor,
sync: GetUnprocessedNotificationsSchema.sync,
statuses: GetUnprocessedNotificationsSchema.statuses,
});
const LikeBodySchema = z.object({ const LikeBodySchema = z.object({
feed_id: LikeSchema.feed_id, feed_id: LikeSchema.feed_id,
xsec_token: LikeSchema.xsec_token, xsec_token: LikeSchema.xsec_token,
@@ -161,6 +180,79 @@ function errorResponse(code: string, message: string): ApiErrorResponse {
return { success: false, error: { code, message } }; return { success: false, error: { code, message } };
} }
function encodeNotificationCursor(cursor: { firstSeenAt: number; fingerprint: string }): string {
return Buffer.from(
JSON.stringify({
first_seen_at: cursor.firstSeenAt,
fingerprint: cursor.fingerprint,
}),
'utf8',
).toString('base64url');
}
function decodeNotificationCursor(cursor?: string): { firstSeenAt: number; fingerprint: string } | undefined {
if (!cursor) return undefined;
try {
const raw = JSON.parse(Buffer.from(cursor, 'base64url').toString('utf8')) as {
first_seen_at?: unknown;
fingerprint?: unknown;
};
const firstSeenAt = raw.first_seen_at;
const fingerprint = raw.fingerprint;
if (
typeof firstSeenAt !== 'number' ||
!Number.isInteger(firstSeenAt) ||
firstSeenAt < 0 ||
typeof fingerprint !== 'string' ||
fingerprint.length === 0
) {
throw new Error('Invalid notification cursor payload');
}
return { firstSeenAt, fingerprint };
} catch {
throw new Error('Invalid cursor for notification keyset pagination');
}
}
function encodeSubCommentCursor(cursor: { createTime: string; replyId: string }): string {
return Buffer.from(
JSON.stringify({
create_time: cursor.createTime,
reply_id: cursor.replyId,
}),
'utf8',
).toString('base64url');
}
function decodeSubCommentCursor(cursor?: string): { createTime: string; replyId: string } | undefined {
if (!cursor) return undefined;
try {
const raw = JSON.parse(Buffer.from(cursor, 'base64url').toString('utf8')) as {
create_time?: unknown;
reply_id?: unknown;
};
const createTime = raw.create_time;
const replyId = raw.reply_id;
if (
typeof createTime !== 'string' ||
createTime.length === 0 ||
typeof replyId !== 'string' ||
replyId.length === 0
) {
throw new Error('Invalid sub-comment cursor payload');
}
return { createTime, replyId };
} catch {
throw new Error('Invalid cursor for sub-comment keyset pagination');
}
}
function compareSubCommentKey(a: { createTime: string; id: string }, b: { createTime: string; id: string }): number {
const timeCmp = a.createTime.localeCompare(b.createTime);
if (timeCmp !== 0) return timeCmp;
return a.id.localeCompare(b.id);
}
// --------------------------------------------------------------------------- // ---------------------------------------------------------------------------
// Rate limiters // Rate limiters
// --------------------------------------------------------------------------- // ---------------------------------------------------------------------------
@@ -323,6 +415,11 @@ export function createXhsRoutes(browser: BrowserManager): Router {
void (async () => { void (async () => {
try { try {
const body = FeedDetailBodySchema.parse(req.body); const body = FeedDetailBodySchema.parse(req.body);
const target = resolveFeedTarget({
url: body.url,
feed_id: body.feed_id,
xsec_token: body.xsec_token,
});
const timeoutMs = const timeoutMs =
config.operationTimeouts['feed_detail'] ?? config.operationTimeouts['feed_detail'] ??
@@ -332,7 +429,7 @@ export function createXhsRoutes(browser: BrowserManager): Router {
const detail = await browser.withPage( const detail = await browser.withPage(
PLATFORM, PLATFORM,
async (page) => async (page) =>
getFeedDetail(page, body.feed_id, body.xsec_token), getFeedDetail(page, target.feedId, target.xsecToken),
timeoutMs, timeoutMs,
); );
@@ -350,20 +447,66 @@ export function createXhsRoutes(browser: BrowserManager): Router {
void (async () => { void (async () => {
try { try {
const body = SubCommentsBodySchema.parse(req.body); const body = SubCommentsBodySchema.parse(req.body);
const limit = Math.min(200, Math.max(1, body.max_count));
const keysetCursor = decodeSubCommentCursor(body.cursor);
const timeoutMs = const timeoutMs =
config.operationTimeouts['feed_detail'] ?? config.operationTimeouts['feed_detail'] ??
config.operationTimeouts['default'] ?? config.operationTimeouts['default'] ??
60_000; 60_000;
const result = await browser.withPage( const allLoaded = await browser.withPage(
PLATFORM, PLATFORM,
async (page) => async (page) =>
getSubComments(page, body.feed_id, body.xsec_token, body.comment_id, body.max_count), getSubComments(page, body.feed_id, body.xsec_token, body.comment_id, 200),
timeoutMs, timeoutMs,
); );
res.json(successResponse(result) as ApiResponse<typeof result>); const sorted = [...allLoaded].sort((a, b) =>
compareSubCommentKey(
{ createTime: a.createTime, id: a.id },
{ createTime: b.createTime, id: b.id },
));
const startIndex = keysetCursor
? sorted.findIndex((item) =>
compareSubCommentKey(
{ createTime: item.createTime, id: item.id },
{ createTime: keysetCursor.createTime, id: keysetCursor.replyId },
) > 0)
: 0;
const start = startIndex < 0 ? sorted.length : startIndex;
const pageItems = sorted.slice(start, start + limit);
const hasMore = start + pageItems.length < sorted.length;
const nextCursor = hasMore && pageItems.length > 0
? encodeSubCommentCursor({
createTime: pageItems[pageItems.length - 1]!.createTime,
replyId: pageItems[pageItems.length - 1]!.id,
})
: undefined;
res.json(successResponse({
items: pageItems,
pagination: {
mode: 'keyset',
cursor: body.cursor ?? null,
max_count: limit,
returned: pageItems.length,
has_more: hasMore,
...(nextCursor ? { next_cursor: nextCursor } : {}),
},
}) as ApiResponse<{
items: typeof pageItems;
pagination: {
mode: 'keyset';
cursor: string | null;
max_count: number;
returned: number;
has_more: boolean;
next_cursor?: string;
};
}>);
} catch (err) { } catch (err) {
handleError(res, err); handleError(res, err);
} }
@@ -377,6 +520,11 @@ export function createXhsRoutes(browser: BrowserManager): Router {
void (async () => { void (async () => {
try { try {
const body = UserProfileBodySchema.parse(req.body); const body = UserProfileBodySchema.parse(req.body);
const target = resolveUserTarget({
url: body.url,
user_id: body.user_id,
xsec_token: body.xsec_token,
});
const timeoutMs = const timeoutMs =
config.operationTimeouts['user_profile'] ?? config.operationTimeouts['user_profile'] ??
@@ -386,7 +534,7 @@ export function createXhsRoutes(browser: BrowserManager): Router {
const profile = await browser.withPage( const profile = await browser.withPage(
PLATFORM, PLATFORM,
async (page) => async (page) =>
getUserProfile(page, body.user_id, body.xsec_token), getUserProfile(page, target.userId, target.xsecToken),
timeoutMs, timeoutMs,
); );
@@ -654,6 +802,7 @@ export function createXhsRoutes(browser: BrowserManager): Router {
timeoutMs, timeoutMs,
); );
getNotificationStateStore().upsertNotifications(notifications);
res.json(successResponse(notifications) as ApiResponse<typeof notifications>); res.json(successResponse(notifications) as ApiResponse<typeof notifications>);
} catch (err) { } catch (err) {
handleError(res, err); handleError(res, err);
@@ -668,20 +817,143 @@ export function createXhsRoutes(browser: BrowserManager): Router {
void (async () => { void (async () => {
try { try {
const body = ReplyNotificationBodySchema.parse(req.body); const body = ReplyNotificationBodySchema.parse(req.body);
const store = getNotificationStateStore();
const target = (() => {
if (body.fingerprint) {
const task = store.getByFingerprint(body.fingerprint);
if (!task) {
throw new Error(`Notification task not found: ${body.fingerprint}`);
}
return {
fingerprint: body.fingerprint,
userId: task.notification.userId,
commentContent: task.notification.content,
};
}
if (!body.user_id || !body.comment_content) {
throw new Error('Either fingerprint or both user_id and comment_content are required');
}
const fp = store.findOpenFingerprint(body.user_id, body.comment_content) ?? undefined;
return {
...(fp ? { fingerprint: fp } : {}),
userId: body.user_id,
commentContent: body.comment_content,
};
})();
const targetFingerprint = target.fingerprint;
if (targetFingerprint) {
store.markPending(targetFingerprint);
}
const timeoutMs = const timeoutMs =
config.operationTimeouts['reply'] ?? config.operationTimeouts['reply'] ??
config.operationTimeouts['default'] ?? config.operationTimeouts['default'] ??
20_000; 20_000;
const result = await browser.withPage( try {
PLATFORM, const result = await browser.withPage(
async (page) => PLATFORM,
replyNotification(page, body.user_id, body.comment_content, body.reply_content), async (page) =>
timeoutMs, replyNotification(page, target.userId, target.commentContent, body.reply_content),
); timeoutMs,
);
res.json(successResponse(result) as ApiResponse<typeof result>); if (targetFingerprint) {
if (result.success) {
store.markReplied(targetFingerprint, body.reply_content);
} else {
store.markFailed(
targetFingerprint,
'Reply action returned success=false',
);
}
}
res.json(successResponse({
...result,
...(targetFingerprint ? { fingerprint: targetFingerprint } : {}),
}) as ApiResponse<typeof result & { fingerprint?: string }>);
} catch (err) {
if (targetFingerprint) {
const message = err instanceof Error ? err.message : String(err);
store.markFailed(targetFingerprint, message);
}
throw err;
}
} catch (err) {
handleError(res, err);
}
})();
});
// -----------------------------------------------------------------------
// GET /notifications/unprocessed
// -----------------------------------------------------------------------
router.get('/notifications/unprocessed', readRateLimiter, (req, res) => {
void (async () => {
try {
const statusesFromQuery = (() => {
const raw = req.query.statuses;
if (!raw) return undefined;
if (Array.isArray(raw)) return raw.flatMap((s) => String(s).split(','));
return String(raw).split(',');
})()
?.map((s) => s.trim())
.filter((s) => s.length > 0);
const parsed = GetUnprocessedNotificationsQuerySchema.parse({
max_count: req.query.max_count ? Number(req.query.max_count) : undefined,
cursor: req.query.cursor ? String(req.query.cursor) : undefined,
sync: req.query.sync === undefined
? undefined
: ['1', 'true', 'yes'].includes(String(req.query.sync).toLowerCase()),
statuses: statusesFromQuery,
});
let syncResult: { fetched: number; inserted: number; updated: number } | null = null;
if (parsed.sync) {
syncResult = await syncCommentNotifications(browser, parsed.max_count);
}
const statuses: NotificationTaskStatus[] =
parsed.statuses && parsed.statuses.length > 0
? parsed.statuses
: ['new', 'failed'];
const store = getNotificationStateStore();
const keysetCursor = decodeNotificationCursor(parsed.cursor);
const pageResult = store.listByStatusesKeyset(statuses, parsed.max_count, keysetCursor);
const nextCursor = pageResult.nextCursor
? encodeNotificationCursor(pageResult.nextCursor)
: undefined;
res.json(successResponse({
tasks: pageResult.tasks,
pagination: {
mode: 'keyset',
cursor: parsed.cursor ?? null,
max_count: parsed.max_count,
returned: pageResult.tasks.length,
has_more: pageResult.hasMore,
...(nextCursor ? { next_cursor: nextCursor } : {}),
},
...(syncResult ? { synced: syncResult } : {}),
}) as ApiResponse<{
tasks: NotificationTask[];
pagination: {
mode: 'keyset';
cursor: string | null;
max_count: number;
returned: number;
has_more: boolean;
next_cursor?: string;
};
synced?: { fetched: number; inserted: number; updated: number };
}>);
} catch (err) { } catch (err) {
handleError(res, err); handleError(res, err);
} }
@@ -0,0 +1,376 @@
import { z } from 'zod';
// ---------------------------------------------------------------------------
// MCP tool parameter schemas for Xiaohongshu tools.
//
// Phase 2 tools (login) have no parameters — their schemas are empty objects.
// Phase 3/4 schemas are defined here so that the full tool surface is
// established upfront and types can be inferred with z.infer<>.
// ---------------------------------------------------------------------------
// -- Phase 2: Login management (3 tools) -----------------------------------
/** xhs_check_login — no parameters. */
export const CheckLoginSchema = {};
/** xhs_get_login_qrcode — no parameters. */
export const GetLoginQRCodeSchema = {};
/** xhs_delete_cookies — no parameters. */
export const DeleteCookiesSchema = {};
// -- Phase 3: Content browsing (4 tools) -----------------------------------
/** xhs_list_feeds — no parameters. */
export const ListFeedsSchema = {
max_count: z
.number()
.int()
.min(1)
.max(200)
.optional()
.default(20)
.describe('Maximum number of feeds to return per page (1200, default 20)'),
cursor: z
.string()
.optional()
.describe('Keyset pagination cursor returned by previous call'),
};
/** xhs_search */
export const SearchSchema = {
keyword: z.string().describe('Search keyword'),
filters: z
.object({
sort: z
.enum(['general', 'time_descending', 'popularity_descending'])
.optional()
.describe('Sort order'),
type: z
.enum(['all', 'note', 'video'])
.optional()
.describe('Content type filter'),
time: z
.enum(['all', 'day', 'week', 'half_year'])
.optional()
.describe('Time range filter'),
})
.optional()
.describe('Optional search filters'),
max_count: z
.number()
.int()
.min(1)
.max(200)
.optional()
.default(20)
.describe('Maximum number of search results to return per page (1200, default 20)'),
cursor: z
.string()
.optional()
.describe('Pagination cursor returned by previous call'),
};
/** xhs_get_feed_detail */
export const GetFeedDetailSchema = {
url: z
.string()
.optional()
.describe('Optional note URL (auto-parses feed_id and xsec_token)'),
feed_id: z.string().optional().describe('Feed (note) ID (required when url not provided)'),
xsec_token: z.string().optional().describe('Security token for the feed (required when url not provided)'),
};
/** xhs_get_sub_comments */
export const GetSubCommentsSchema = {
feed_id: z.string().describe('Feed (note) ID'),
xsec_token: z.string().describe('Security token for the feed'),
comment_id: z.string().describe('Parent comment ID whose sub-comments to load'),
max_count: z
.number()
.int()
.min(1)
.max(200)
.optional()
.default(20)
.describe('Maximum number of sub-comments to return per page (1200, default 20)'),
cursor: z
.string()
.optional()
.describe('Keyset pagination cursor returned by previous call'),
};
/** xhs_get_user_profile */
export const GetUserProfileSchema = {
url: z
.string()
.optional()
.describe('Optional user profile URL (auto-parses user_id and xsec_token)'),
user_id: z.string().optional().describe('User ID (required when url not provided)'),
xsec_token: z.string().optional().describe('Security token for the user page (required when url not provided)'),
};
// -- Phase 4: Content publishing (2 tools) ---------------------------------
/** xhs_publish_image */
export const PublishImageSchema = {
request_id: z
.string()
.min(1)
.max(128)
.optional()
.describe('Optional idempotency key for publish request'),
title: z.string().min(1).max(20, 'Title must be ≤ 20 characters').describe('Note title (max 20 chars)'),
content: z.string().max(1000, 'Content must be ≤ 1000 characters').describe('Note body text (max 1000 chars)'),
images: z
.array(z.string())
.min(1)
.max(18, 'Maximum 18 images per note')
.describe('Array of local file paths or HTTP/HTTPS URLs (118 images)'),
tags: z.array(z.string()).optional().describe('Hashtags to attach'),
schedule_at: z
.string()
.optional()
.describe('ISO 8601 datetime for scheduled publishing'),
is_original: z
.boolean()
.optional()
.default(false)
.describe('Mark as original content'),
visibility: z
.enum(['public', 'private', 'friends'])
.optional()
.default('public')
.describe('Visibility setting'),
};
/** xhs_publish_video */
export const PublishVideoSchema = {
request_id: z
.string()
.min(1)
.max(128)
.optional()
.describe('Optional idempotency key for publish request'),
title: z.string().min(1).max(20, 'Title must be ≤ 20 characters').describe('Note title (max 20 chars)'),
content: z.string().max(1000, 'Content must be ≤ 1000 characters').describe('Note body text (max 1000 chars)'),
video: z.string().describe('Local file path or HTTP/HTTPS URL for the video'),
tags: z.array(z.string()).optional().describe('Hashtags to attach'),
schedule_at: z
.string()
.optional()
.describe('ISO 8601 datetime for scheduled publishing'),
visibility: z
.enum(['public', 'private', 'friends'])
.optional()
.default('public')
.describe('Visibility setting'),
};
// -- Phase 4: Interactions (4 tools) ---------------------------------------
/** xhs_post_comment */
export const PostCommentSchema = {
request_id: z
.string()
.min(1)
.max(128)
.optional()
.describe('Optional idempotency key for comment request'),
feed_id: z.string().describe('Feed ID to comment on'),
xsec_token: z.string().describe('Security token for the feed'),
content: z.string().min(1).describe('Comment text'),
};
/** xhs_reply_comment */
export const ReplyCommentSchema = {
request_id: z
.string()
.min(1)
.max(128)
.optional()
.describe('Optional idempotency key for reply request'),
feed_id: z.string().describe('Feed ID'),
xsec_token: z.string().describe('Security token for the feed'),
comment_id: z.string().optional().describe('Comment ID to reply to'),
user_id: z.string().optional().describe('User ID of the comment author'),
content: z.string().min(1).describe('Reply text'),
};
/** xhs_set_like_state */
export const SetLikeStateSchema = {
feed_id: z.string().describe('Feed ID to set like state'),
xsec_token: z.string().describe('Security token for the feed'),
liked: z.boolean().describe('Target like state'),
};
/** Legacy schema used by REST toggle endpoint. */
export const LikeSchema = {
feed_id: SetLikeStateSchema.feed_id,
xsec_token: SetLikeStateSchema.xsec_token,
};
/** xhs_list_my_notes */
export const ListMyNotesSchema = {
max_count: z
.number()
.int()
.min(1)
.max(200)
.optional()
.default(20)
.describe('Maximum number of notes to return per page (1200, default 20)'),
cursor: z
.string()
.optional()
.describe('Keyset pagination cursor returned by previous call'),
};
// -- Phase 5: Notifications & automation -----------------------------------
/** xhs_get_comment_notifications */
export const GetCommentNotificationsSchema = {
max_count: z
.number()
.int()
.min(1)
.max(50)
.optional()
.default(20)
.describe('Maximum number of notifications to return (150, default 20)'),
};
/** xhs_reply_notification */
export const ReplyNotificationSchema = {
request_id: z
.string()
.min(1)
.max(128)
.optional()
.describe('Optional idempotency key for notification reply'),
fingerprint: z
.string()
.optional()
.describe('Optional notification fingerprint from xhs_get_unprocessed_notifications'),
user_id: z.string().optional().describe('User ID of the comment author (fallback when fingerprint is absent)'),
comment_content: z.string().optional().describe('Original comment content to match the notification (fallback when fingerprint is absent)'),
reply_content: z.string().min(1).describe('Reply text to send'),
};
/** xhs_get_unprocessed_notifications */
export const GetUnprocessedNotificationsSchema = {
max_count: z
.number()
.int()
.min(1)
.max(200)
.optional()
.default(20)
.describe('Maximum number of unprocessed notifications to return (1200, default 20)'),
cursor: z
.string()
.optional()
.describe('Keyset pagination cursor returned by previous call'),
statuses: z
.array(z.enum(['new', 'pending', 'failed']))
.optional()
.describe('Statuses to include. Defaults to ["new", "failed"]'),
sync: z
.boolean()
.optional()
.default(true)
.describe('Whether to sync latest notifications from Xiaohongshu before querying local state'),
};
/** xhs_mark_notification_task */
export const MarkNotificationTaskSchema = {
fingerprint: z.string().describe('Notification task fingerprint'),
status: z
.enum(['new', 'pending', 'ignored', 'replied', 'failed'])
.describe('Target status for this notification task'),
note: z
.string()
.optional()
.describe('Optional note/reason (used as reply_content for replied, or error_message for failed/ignored)'),
};
/** xhs_list_failed_notification_tasks */
export const ListFailedNotificationTasksSchema = {
max_count: z
.number()
.int()
.min(1)
.max(200)
.optional()
.default(20)
.describe('Maximum number of failed tasks to return (1200, default 20)'),
cursor: z
.string()
.optional()
.describe('Pagination cursor returned by previous call'),
};
/** xhs_retry_notification_task */
export const RetryNotificationTaskSchema = {
request_id: z
.string()
.min(1)
.max(128)
.optional()
.describe('Optional idempotency key for retry request'),
fingerprint: z.string().describe('Notification task fingerprint to retry'),
reply_content: z
.string()
.min(1)
.optional()
.describe('Optional override reply text. If omitted, uses stored reply_content from previous attempt.'),
};
/** xhs_mark_notification_tasks */
export const MarkNotificationTasksSchema = {
tasks: z
.array(z.object({
fingerprint: z.string(),
status: z.enum(['new', 'pending', 'ignored', 'replied', 'failed']),
note: z.string().optional(),
}))
.min(1)
.max(100)
.describe('Batch of task status updates (1100 items)'),
};
/** xhs_retry_notification_tasks */
export const RetryNotificationTasksSchema = {
request_id: z
.string()
.min(1)
.max(128)
.optional()
.describe('Optional idempotency key for batch retry request'),
tasks: z
.array(z.object({
fingerprint: z.string(),
reply_content: z.string().min(1).optional(),
}))
.min(1)
.max(100)
.describe('Batch of failed tasks to retry (1100 items)'),
continue_on_error: z
.boolean()
.optional()
.default(true)
.describe('Continue processing remaining tasks after one task fails'),
};
/** xhs_set_favorite_state */
export const SetFavoriteStateSchema = {
feed_id: z.string().describe('Feed ID to set favorite state'),
xsec_token: z.string().describe('Security token for the feed'),
favorited: z.boolean().describe('Target favorite state'),
};
/** Legacy schema used by REST toggle endpoint. */
export const FavoriteSchema = {
feed_id: SetFavoriteStateSchema.feed_id,
xsec_token: SetFavoriteStateSchema.xsec_token,
};
@@ -1,6 +1,6 @@
import type { Page } from 'rebrowser-playwright'; import type { Page } from 'rebrowser-playwright';
import { logger } from '../../utils/logger.js'; import { logger } from '@social/core/utils/logger.js';
import { extractInitialState, parseCountString, ensureHttps } from './feeds.js'; import { extractInitialState, parseCountString, ensureHttps } from './feeds.js';
import type { Feed, SearchFilters } from './types.js'; import type { Feed, SearchFilters } from './types.js';
@@ -0,0 +1,95 @@
interface FeedTargetInput {
feed_id?: string;
xsec_token?: string;
url?: string;
}
interface UserTargetInput {
user_id?: string;
xsec_token?: string;
url?: string;
}
interface FeedTargetResolved {
feedId: string;
xsecToken: string;
}
interface UserTargetResolved {
userId: string;
xsecToken: string;
}
function parseXhsUrl(rawUrl: string): URL {
const trimmed = rawUrl.trim();
if (!trimmed) {
throw new Error('url cannot be empty');
}
if (/^https?:\/\//i.test(trimmed)) {
return new URL(trimmed);
}
if (trimmed.startsWith('/')) {
return new URL(`https://www.xiaohongshu.com${trimmed}`);
}
return new URL(`https://${trimmed}`);
}
function extractFeedIdFromPath(pathname: string): string | undefined {
const patterns = [
/\/explore\/([a-zA-Z0-9_-]+)/,
/\/discovery\/item\/([a-zA-Z0-9_-]+)/,
/\/note\/([a-zA-Z0-9_-]+)/,
];
for (const pattern of patterns) {
const match = pathname.match(pattern);
if (match?.[1]) return match[1];
}
return undefined;
}
function extractUserIdFromPath(pathname: string): string | undefined {
const match = pathname.match(/\/user\/profile\/([a-zA-Z0-9_-]+)/);
return match?.[1];
}
function extractXsecToken(url: URL): string | undefined {
return url.searchParams.get('xsec_token') ?? url.searchParams.get('xsecToken') ?? undefined;
}
export function resolveFeedTarget(input: FeedTargetInput): FeedTargetResolved {
let feedId = input.feed_id?.trim();
let xsecToken = input.xsec_token?.trim();
if (input.url) {
const parsed = parseXhsUrl(input.url);
feedId = feedId || extractFeedIdFromPath(parsed.pathname);
xsecToken = xsecToken || extractXsecToken(parsed);
}
if (!feedId || !xsecToken) {
throw new Error('xhs_get_feed_detail requires either url with feed_id/xsec_token, or both feed_id and xsec_token');
}
return { feedId, xsecToken };
}
export function resolveUserTarget(input: UserTargetInput): UserTargetResolved {
let userId = input.user_id?.trim();
let xsecToken = input.xsec_token?.trim();
if (input.url) {
const parsed = parseXhsUrl(input.url);
userId = userId || extractUserIdFromPath(parsed.pathname);
xsecToken = xsecToken || extractXsecToken(parsed);
}
if (!userId || !xsecToken) {
throw new Error('xhs_get_user_profile requires either url with user_id/xsec_token, or both user_id and xsec_token');
}
return { userId, xsecToken };
}
@@ -1,6 +1,6 @@
import type { Page } from 'rebrowser-playwright'; import type { Page } from 'rebrowser-playwright';
import { logger } from '../../utils/logger.js'; import { logger } from '@social/core/utils/logger.js';
import { XHS_SELECTORS } from './selectors.js'; import { XHS_SELECTORS } from './selectors.js';
import { extractInitialState, parseCountString, ensureHttps } from './feeds.js'; import { extractInitialState, parseCountString, ensureHttps } from './feeds.js';
import type { UserProfile, Feed } from './types.js'; import type { UserProfile, Feed } from './types.js';
@@ -0,0 +1,66 @@
import fs from 'node:fs/promises';
import os from 'node:os';
import path from 'node:path';
import { describe, expect, it } from 'vitest';
import { NotificationStateStore } from '../src/platforms/xiaohongshu/notification-state.js';
import type { CommentNotification } from '../src/platforms/xiaohongshu/types.js';
function makeNotification(overrides?: Partial<CommentNotification>): CommentNotification {
return {
userId: 'u1',
nickname: 'tester',
avatar: 'https://example.com/a.png',
content: '你好',
type: '评论了你的笔记',
time: '1分钟前',
feedId: 'feed123',
xsecToken: 'token123',
noteImage: 'https://example.com/note.png',
...overrides,
};
}
describe('notification-state store', () => {
it('upserts notifications and tracks status transitions', async () => {
const tempDir = await fs.mkdtemp(path.join(os.tmpdir(), 'social-mcp-notif-'));
try {
const store = new NotificationStateStore(tempDir, 'test.db');
const n1 = makeNotification();
const first = store.upsertNotifications([n1]);
expect(first).toEqual({ fetched: 1, inserted: 1, updated: 0 });
const second = store.upsertNotifications([n1]);
expect(second).toEqual({ fetched: 1, inserted: 0, updated: 1 });
const openTasks = store.listByStatuses(['new'], 10);
expect(openTasks).toHaveLength(1);
expect(openTasks[0]?.notification.userId).toBe('u1');
const fp = store.findOpenFingerprint('u1', '你好');
expect(fp).toBeTypeOf('string');
if (!fp) {
throw new Error('fingerprint should not be null');
}
store.markPending(fp);
expect(store.listByStatuses(['pending'], 10)).toHaveLength(1);
store.markFailed(fp, 'network error');
const failed = store.listByStatuses(['failed'], 10);
expect(failed).toHaveLength(1);
expect(failed[0]?.retryCount).toBe(1);
expect(failed[0]?.errorMessage).toBe('network error');
store.markReplied(fp, '收到,感谢反馈');
const replied = store.listByStatuses(['replied'], 10);
expect(replied).toHaveLength(1);
expect(replied[0]?.replyContent).toBe('收到,感谢反馈');
} finally {
await fs.rm(tempDir, { recursive: true, force: true });
}
});
});
+7
View File
@@ -0,0 +1,7 @@
{
"extends": "../../tsconfig.base.json",
"compilerOptions": {
"outDir": "dist"
},
"include": ["src"]
}
+24
View File
@@ -0,0 +1,24 @@
import { defineConfig } from 'tsup';
export default defineConfig({
entry: ['src/main.ts'],
noExternal: [/^@social\/core/],
external: [
'@modelcontextprotocol/sdk',
/^@modelcontextprotocol\/sdk\//,
'express',
'pino',
'pino-pretty',
'rebrowser-playwright',
'chromium-bidi/lib/cjs/bidiMapper/BidiMapper',
'chromium-bidi/lib/cjs/cdp/CdpConnection',
],
format: ['esm'],
target: 'node22',
outDir: 'dist',
clean: true,
sourcemap: true,
dts: false,
splitting: false,
shims: false,
});
+14
View File
@@ -0,0 +1,14 @@
import { defineConfig } from 'vitest/config';
import path from 'node:path';
export default defineConfig({
resolve: {
alias: {
'@social/core': path.resolve(__dirname, '../../packages/core/src'),
},
},
test: {
include: ['test/**/*.test.ts'],
environment: 'node',
},
});
+46 -5
View File
@@ -1,10 +1,12 @@
services: services:
social-mcp: mcp-xhs:
build: build:
context: .. context: ..
dockerfile: Dockerfile dockerfile: Dockerfile
args:
APP_NAME: xhs-mcp
ports: ports:
- "127.0.0.1:3000:3000" - "127.0.0.1:9527:9527"
shm_size: '1gb' shm_size: '1gb'
deploy: deploy:
resources: resources:
@@ -22,14 +24,53 @@ services:
tmpfs: tmpfs:
- /tmp:size=512m - /tmp:size=512m
volumes: volumes:
- cookie-data:/home/appuser/.social-mcp - cookie-data-xhs:/home/appuser/.social-mcp-xhs
environment: environment:
- NODE_ENV=production - NODE_ENV=production
- HOST=0.0.0.0 - HOST=0.0.0.0
- PORT=3000 - PORT=9527
- HEADLESS=true - HEADLESS=true
- COOKIE_DIR=/home/appuser/.social-mcp-xhs
- APP_NAME=xhs-mcp
- ALLOW_REMOTE=yes-i-understand-the-risk
restart: unless-stopped
mcp-xhh:
build:
context: ..
dockerfile: Dockerfile
args:
APP_NAME: xhh-mcp
ports:
- "127.0.0.1:9528:9528"
shm_size: '1gb'
deploy:
resources:
limits:
memory: 2g
cpus: '2.0'
reservations:
memory: 1g
cpus: '1.0'
security_opt:
- no-new-privileges:true
cap_drop:
- ALL
read_only: true
tmpfs:
- /tmp:size=512m
volumes:
- cookie-data-xhh:/home/appuser/.social-mcp-xhh
environment:
- NODE_ENV=production
- HOST=0.0.0.0
- PORT=9528
- HEADLESS=true
- COOKIE_DIR=/home/appuser/.social-mcp-xhh
- APP_NAME=xhh-mcp
- ALLOW_REMOTE=yes-i-understand-the-risk - ALLOW_REMOTE=yes-i-understand-the-risk
restart: unless-stopped restart: unless-stopped
volumes: volumes:
cookie-data: cookie-data-xhs:
cookie-data-xhh:
+38 -8
View File
@@ -1,19 +1,49 @@
services: services:
app: mcp-xhs:
build: . build:
image: social-mcp:latest context: .
container_name: social-mcp dockerfile: Dockerfile
args:
APP_NAME: xhs-mcp
image: social-mcp-xhs:latest
container_name: mcp-xhs
ports: ports:
- "3010:3000" - "9527:9527"
shm_size: '1gb' shm_size: '1gb'
volumes: volumes:
- /data/social-mcp:/home/appuser/.social-mcp - /data/mcp-xhs:/home/appuser/.social-mcp-xhs
environment: environment:
- NODE_ENV=production - NODE_ENV=production
- HOST=0.0.0.0 - HOST=0.0.0.0
- PORT=3000 - PORT=9527
- HEADLESS=true - HEADLESS=true
- COOKIE_DIR=/home/appuser/.social-mcp - COOKIE_DIR=/home/appuser/.social-mcp-xhs
- APP_NAME=xhs-mcp
- ALLOW_REMOTE=yes-i-understand-the-risk
restart: unless-stopped
networks:
- nginx
mcp-xhh:
build:
context: .
dockerfile: Dockerfile
args:
APP_NAME: xhh-mcp
image: social-mcp-xhh:latest
container_name: mcp-xhh
ports:
- "9528:9528"
shm_size: '1gb'
volumes:
- /data/mcp-xhh:/home/appuser/.social-mcp-xhh
environment:
- NODE_ENV=production
- HOST=0.0.0.0
- PORT=9528
- HEADLESS=true
- COOKIE_DIR=/home/appuser/.social-mcp-xhh
- APP_NAME=xhh-mcp
- ALLOW_REMOTE=yes-i-understand-the-risk - ALLOW_REMOTE=yes-i-understand-the-risk
restart: unless-stopped restart: unless-stopped
networks: networks:
+63
View File
@@ -0,0 +1,63 @@
# 小黑盒 MCP 实施计划(Monorepo 执行版)
## 1. 目标
在不改变对外协议的前提下,把项目迁移为可视化 monorepo:
- `apps/xhh-mcp` 承载小黑盒服务
- `apps/xhs-mcp` 承载小红书服务
- `packages/core` 承载共享基础设施
## 2. 小黑盒工具范围
已落地工具:
1. `xhh_check_login`
2. `xhh_get_login_qrcode`
3. `xhh_delete_cookies`
4. `xhh_list_feeds`
5. `xhh_search`
6. `xhh_get_feed_detail`
7. `xhh_get_sub_comments`
8. `xhh_get_user_profile`
9. `xhh_list_my_posts`
10. `xhh_post_comment`
11. `xhh_reply_comment`
12. `xhh_set_like_state`
13. `xhh_set_favorite_state`
统一 MCP 响应:`{ success, data, meta }`
## 3. 代码位置
- 小黑盒实现:`apps/xhh-mcp/src/platforms/xiaoheihe/*`
- 共享依赖:`packages/core/src/*`
- 小红书实现:`apps/xhs-mcp/src/platforms/xiaohongshu/*`
## 4. 构建与运行
- workspace`pnpm-workspace.yaml`
- 构建:`pnpm build`
- 启动 xhh`pnpm start:xhh`
- 启动 xhs`pnpm start:xhs`
## 5. 测试
- `apps/xhh-mcp/test/*`
- `apps/xhs-mcp/test/*`
- `packages/core/test/*`
验收命令:
1. `pnpm lint`
2. `pnpm test`
3. `pnpm build`
## 6. 部署
单 Dockerfile 双目标:
- `APP_NAME=xhs-mcp`
- `APP_NAME=xhh-mcp`
compose 与 Jenkins 已改为双服务部署(9527/9528)。
-4041
View File
File diff suppressed because it is too large Load Diff
+13 -42
View File
@@ -1,50 +1,21 @@
{ {
"name": "social-mcp", "name": "social-auto-hub",
"version": "0.1.0", "version": "0.1.0",
"description": "Multi-platform social media automation MCP service", "private": true,
"type": "module", "description": "Monorepo for XHS/XHH MCP services",
"main": "dist/index.js",
"bin": {
"social-mcp": "dist/index.js"
},
"scripts": { "scripts": {
"build": "tsup", "build": "pnpm -r --filter \"./packages/*\" --filter \"./apps/*\" run build",
"build:web": "cd web && pnpm build && mkdir -p ../dist/web && cp -r dist/* ../dist/web/", "lint": "pnpm -r --filter \"./packages/*\" --filter \"./apps/*\" run lint",
"build:all": "pnpm build && pnpm build:web", "test": "pnpm -r --filter \"./packages/*\" --filter \"./apps/*\" run test",
"restart": "pnpm build:all && pkill -f 'node dist/index.js' ; sleep 1 && node dist/index.js &", "start": "pnpm start:xhs",
"dev": "tsup --watch", "start:xhs": "pnpm --filter @social/xhs-mcp start",
"dev:web": "cd web && pnpm dev", "start:xhh": "pnpm --filter @social/xhh-mcp start",
"start": "node dist/index.js", "dev:xhs": "pnpm --filter @social/xhs-mcp dev",
"test": "vitest run", "dev:xhh": "pnpm --filter @social/xhh-mcp dev",
"test:watch": "vitest", "restart:xhs": "pnpm --filter @social/xhs-mcp build && (lsof -ti tcp:9527 | xargs kill >/dev/null 2>&1 || true) && pnpm --filter @social/xhs-mcp start",
"lint": "tsc --noEmit" "restart:xhh": "pnpm --filter @social/xhh-mcp build && (lsof -ti tcp:9528 | xargs kill >/dev/null 2>&1 || true) && pnpm --filter @social/xhh-mcp start"
}, },
"keywords": [
"mcp",
"social-media",
"automation",
"playwright",
"xiaohongshu"
],
"author": "",
"license": "ISC",
"engines": { "engines": {
"node": ">=22.0.0" "node": ">=22.0.0"
},
"dependencies": {
"@modelcontextprotocol/sdk": "^1.27.0",
"express": "^4.21.0",
"pino": "^9.0.0",
"rebrowser-playwright": "^1.52.0",
"zod": "^3.25.0"
},
"devDependencies": {
"@types/express": "^5.0.0",
"@types/node": "^22.0.0",
"pino-pretty": "^13.0.0",
"playwright": "^1.52.0",
"tsup": "^8.0.0",
"typescript": "^5.7.0",
"vitest": "^3.0.0"
} }
} }
+34
View File
@@ -0,0 +1,34 @@
{
"name": "@social/core",
"version": "0.1.0",
"private": true,
"type": "module",
"main": "dist/server/app.js",
"exports": {
"./browser/*": "./dist/browser/*",
"./config/*": "./dist/config/*",
"./cookie/*": "./dist/cookie/*",
"./server/*": "./dist/server/*",
"./utils/*": "./dist/utils/*"
},
"scripts": {
"build": "tsup",
"lint": "tsc --noEmit",
"test": "vitest run"
},
"dependencies": {
"@modelcontextprotocol/sdk": "^1.27.0",
"express": "^4.21.0",
"pino": "^9.0.0",
"rebrowser-playwright": "^1.52.0"
},
"devDependencies": {
"@types/express": "^5.0.0",
"@types/node": "^22.0.0",
"pino-pretty": "^13.0.0",
"tsup": "^8.0.0",
"typescript": "^5.7.0",
"vitest": "^3.0.0"
}
}
@@ -93,6 +93,12 @@ export interface AppConfig {
maxQueueDepth: number; maxQueueDepth: number;
/** Per-operation-type timeout in ms */ /** Per-operation-type timeout in ms */
operationTimeouts: Record<string, number>; operationTimeouts: Record<string, number>;
/** Whether the XHS notification poller is enabled */
notificationPollEnabled: boolean;
/** Poll interval for XHS notifications (ms) */
notificationPollIntervalMs: number;
/** Max notifications fetched per poll */
notificationPollMaxCount: number;
} }
// --------------------------------------------------------------------------- // ---------------------------------------------------------------------------
@@ -106,7 +112,10 @@ export const config: AppConfig = {
browserBin: process.env['BROWSER_BIN'] || undefined, browserBin: process.env['BROWSER_BIN'] || undefined,
logLevel: envString('LOG_LEVEL', 'info'), logLevel: envString('LOG_LEVEL', 'info'),
nodeEnv: envString('NODE_ENV', 'development'), nodeEnv: envString('NODE_ENV', 'development'),
cookieDir: envString('COOKIE_DIR', path.join(os.homedir(), '.social-mcp')), cookieDir: envString('COOKIE_DIR', path.join(os.homedir(), '.social-mcp-xhs')),
maxQueueDepth: envInt('MAX_QUEUE_DEPTH', 10), maxQueueDepth: envInt('MAX_QUEUE_DEPTH', 10),
operationTimeouts, operationTimeouts,
notificationPollEnabled: envBool('XHS_NOTIFICATION_POLL_ENABLED', true),
notificationPollIntervalMs: Math.max(15, envInt('XHS_NOTIFICATION_POLL_INTERVAL_SEC', 60)) * 1000,
notificationPollMaxCount: Math.max(1, Math.min(50, envInt('XHS_NOTIFICATION_POLL_MAX_COUNT', 20))),
}; };
@@ -1,10 +1,9 @@
import http from 'node:http'; import http from 'node:http';
import path from 'node:path'; import { randomUUID } from 'node:crypto';
import { fileURLToPath } from 'node:url';
import fs from 'node:fs';
import { McpServer } from '@modelcontextprotocol/sdk/server/mcp.js'; import { McpServer } from '@modelcontextprotocol/sdk/server/mcp.js';
import { SSEServerTransport } from '@modelcontextprotocol/sdk/server/sse.js'; import { SSEServerTransport } from '@modelcontextprotocol/sdk/server/sse.js';
import { StreamableHTTPServerTransport } from '@modelcontextprotocol/sdk/server/streamableHttp.js';
import express from 'express'; import express from 'express';
import { config } from '../config/index.js'; import { config } from '../config/index.js';
@@ -40,6 +39,8 @@ const PACKAGE_VERSION = '0.1.0';
export interface PlatformPlugin { export interface PlatformPlugin {
/** Human-readable name used in logs and health-check output. */ /** Human-readable name used in logs and health-check output. */
name: string; name: string;
/** REST namespace mounted at `/api/{apiNamespace}` when registerRoutes exists. */
apiNamespace?: string;
/** Register MCP tools on the shared McpServer instance. */ /** Register MCP tools on the shared McpServer instance. */
registerTools(server: McpServer, browser: BrowserManager): void; registerTools(server: McpServer, browser: BrowserManager): void;
@@ -76,11 +77,17 @@ export class AppServer {
private shuttingDown = false; private shuttingDown = false;
private readonly plugins: PlatformPlugin[] = []; private readonly plugins: PlatformPlugin[] = [];
/** /** SSE transports keyed by session ID for the deprecated `/sse` + `/messages` flow. */
* SSE transports keyed by session ID so that POST /messages can route private readonly sseTransports = new Map<string, SSEServerTransport>();
* incoming JSON-RPC messages to the correct transport instance.
*/ /** Per-session MCP servers backing active SSE sessions. */
private readonly transports = new Map<string, SSEServerTransport>(); private readonly sseSessionServers = new Map<string, McpServer>();
/** Streamable HTTP transports keyed by MCP session ID. */
private readonly streamableTransports = new Map<string, StreamableHTTPServerTransport>();
/** Per-session MCP servers backing active Streamable HTTP sessions. */
private readonly streamableSessionServers = new Map<string, McpServer>();
// -- Constructor ---------------------------------------------------------- // -- Constructor ----------------------------------------------------------
@@ -96,7 +103,8 @@ export class AppServer {
{ name: 'social-mcp', version: PACKAGE_VERSION }, { name: 'social-mcp', version: PACKAGE_VERSION },
); );
// 4. SSE transport endpoints (BEFORE body parsing — MCP SDK reads raw body) // 4. MCP transport endpoints (BEFORE body parsing — MCP SDK reads raw body)
this.setupStreamableHttpEndpoint();
this.setupSseEndpoints(); this.setupSseEndpoints();
// 5. Body parsing for non-MCP routes // 5. Body parsing for non-MCP routes
@@ -128,8 +136,8 @@ export class AppServer {
if (plugin.registerRoutes) { if (plugin.registerRoutes) {
const router = express.Router(); const router = express.Router();
plugin.registerRoutes(router, browserManager); plugin.registerRoutes(router, browserManager);
// Mount REST API routes under /api/xhs (for xiaohongshu) const apiNamespace = plugin.apiNamespace ?? plugin.name;
this.app.use(`/api/xhs`, router); this.app.use(`/api/${apiNamespace}`, router);
} }
this.plugins.push(plugin); this.plugins.push(plugin);
@@ -152,9 +160,6 @@ export class AppServer {
} }
} }
// Serve the web dashboard (static SPA) in production.
this.setupWebDashboard();
// Re-register the error handler so it sits after any plugin routes. // Re-register the error handler so it sits after any plugin routes.
this.app.use(errorHandler); this.app.use(errorHandler);
@@ -177,7 +182,7 @@ export class AppServer {
* Initiate graceful shutdown: * Initiate graceful shutdown:
* 1. Set the shutting-down flag (new requests get 503). * 1. Set the shutting-down flag (new requests get 503).
* 2. Shut down every plugin. * 2. Shut down every plugin.
* 3. Close all SSE transports and the MCP server. * 3. Close all active MCP transports and servers.
* 4. Close the HTTP server. * 4. Close the HTTP server.
*/ */
async close(): Promise<void> { async close(): Promise<void> {
@@ -197,15 +202,41 @@ export class AppServer {
} }
} }
// Close all SSE transports // Close Streamable HTTP transports first.
for (const [sessionId, transport] of this.transports) { for (const [sessionId, transport] of this.streamableTransports) {
try {
await transport.close();
} catch (err: unknown) {
logger.warn({ err, sessionId }, 'Error closing Streamable HTTP transport');
}
}
this.streamableTransports.clear();
for (const [sessionId, sessionServer] of this.streamableSessionServers) {
try {
await sessionServer.close();
} catch (err: unknown) {
logger.warn({ err, sessionId }, 'Error closing Streamable HTTP MCP server');
}
}
this.streamableSessionServers.clear();
// Close deprecated SSE transports.
for (const [sessionId, transport] of this.sseTransports) {
try { try {
await transport.close(); await transport.close();
} catch (err: unknown) { } catch (err: unknown) {
logger.warn({ err, sessionId }, 'Error closing SSE transport'); logger.warn({ err, sessionId }, 'Error closing SSE transport');
} }
} }
this.transports.clear(); this.sseTransports.clear();
for (const [sessionId, sessionServer] of this.sseSessionServers) {
try {
await sessionServer.close();
} catch (err: unknown) {
logger.warn({ err, sessionId }, 'Error closing SSE MCP server');
}
}
this.sseSessionServers.clear();
// Close the MCP server // Close the MCP server
try { try {
@@ -227,6 +258,108 @@ export class AppServer {
logger.info('AppServer shut down complete'); logger.info('AppServer shut down complete');
} }
// -- Private: Streamable HTTP endpoint ------------------------------------
private setupStreamableHttpEndpoint(): void {
this.app.all('/mcp', (req, res) => {
void this.handleStreamableHttpRequest(req, res);
});
}
private async handleStreamableHttpRequest(
req: express.Request,
res: express.Response,
): Promise<void> {
const headerSessionId = this.getFirstHeaderValue(req.headers['mcp-session-id']);
let transport = headerSessionId
? this.streamableTransports.get(headerSessionId)
: undefined;
if (!transport && headerSessionId) {
if (this.sseTransports.has(headerSessionId)) {
this.sendProtocolMismatch(
res,
'Bad Request: Session exists but uses the deprecated SSE transport protocol',
);
return;
}
res.status(404).json({
jsonrpc: '2.0',
error: {
code: -32001,
message: 'Session not found',
},
id: null,
});
return;
}
let createdTransport = false;
let sessionServer: McpServer | null = null;
if (!transport) {
createdTransport = true;
const newTransport = new StreamableHTTPServerTransport({
sessionIdGenerator: () => randomUUID(),
onsessioninitialized: (sessionId) => {
logger.info({ sessionId }, 'Streamable HTTP session initialized');
this.streamableTransports.set(sessionId, newTransport);
if (sessionServer) {
this.streamableSessionServers.set(sessionId, sessionServer);
}
},
});
transport = newTransport;
newTransport.onclose = () => {
const sessionId = newTransport.sessionId;
if (!sessionId) return;
logger.info({ sessionId }, 'Streamable HTTP session closed');
this.cleanupStreamableSession(sessionId);
};
sessionServer = this.createSessionMcpServer();
await sessionServer.connect(newTransport);
}
const activeTransport = transport;
try {
await activeTransport.handleRequest(req, res, req.body);
} catch (err: unknown) {
logger.error(
{ err, sessionId: headerSessionId ?? activeTransport.sessionId },
'Error handling /mcp request',
);
if (!res.headersSent) {
res.status(500).json({
jsonrpc: '2.0',
error: {
code: -32603,
message: 'Internal server error',
},
id: null,
});
}
} finally {
// If no session was initialized (bad request / handshake failure), avoid leaks.
if (createdTransport && !activeTransport.sessionId) {
try {
await activeTransport.close();
} catch (err: unknown) {
logger.warn({ err }, 'Error closing ephemeral Streamable HTTP transport');
}
if (sessionServer) {
try {
await sessionServer.close();
} catch (err: unknown) {
logger.warn({ err }, 'Error closing ephemeral Streamable HTTP MCP server');
}
}
}
}
}
// -- Private: SSE endpoints ----------------------------------------------- // -- Private: SSE endpoints -----------------------------------------------
private setupSseEndpoints(): void { private setupSseEndpoints(): void {
@@ -237,28 +370,22 @@ export class AppServer {
const transport = new SSEServerTransport('/messages', res); const transport = new SSEServerTransport('/messages', res);
const sessionId = transport.sessionId; const sessionId = transport.sessionId;
this.transports.set(sessionId, transport); this.sseTransports.set(sessionId, transport);
logger.info({ sessionId }, 'SSE transport created'); logger.info({ sessionId }, 'SSE transport created');
const perSessionMcp = this.createSessionMcpServer();
this.sseSessionServers.set(sessionId, perSessionMcp);
// Clean up when the client disconnects. // Clean up when the client disconnects.
res.on('close', () => { res.on('close', () => {
logger.info({ sessionId }, 'SSE client disconnected'); logger.info({ sessionId }, 'SSE client disconnected');
this.transports.delete(sessionId); this.cleanupSseSession(sessionId);
}); });
// Each SSE connection needs its own McpServer instance because the
// MCP SDK only allows one transport per server at a time.
const perSessionMcp = new McpServer(
{ name: 'social-mcp', version: PACKAGE_VERSION },
);
for (const plugin of this.plugins) {
plugin.registerTools(perSessionMcp, browserManager);
}
void perSessionMcp.connect(transport).catch((err: unknown) => { void perSessionMcp.connect(transport).catch((err: unknown) => {
logger.error({ err, sessionId }, 'Failed to connect SSE transport to MCP server'); logger.error({ err, sessionId }, 'Failed to connect SSE transport to MCP server');
this.transports.delete(sessionId); this.cleanupSseSession(sessionId);
}); });
}); });
@@ -271,9 +398,16 @@ export class AppServer {
return; return;
} }
const transport = this.transports.get(sessionId); const transport = this.sseTransports.get(sessionId);
if (!transport) { if (!transport) {
if (this.streamableTransports.has(sessionId)) {
this.sendProtocolMismatch(
res,
'Bad Request: Session exists but uses Streamable HTTP transport protocol',
);
return;
}
res.status(404).json({ error: 'Unknown or expired session' }); res.status(404).json({ error: 'Unknown or expired session' });
return; return;
} }
@@ -305,46 +439,6 @@ export class AppServer {
}); });
} }
// -- Private: Web Dashboard (SPA static files) ----------------------------
private setupWebDashboard(): void {
// Resolve the web dashboard dist directory relative to this file.
// tsup bundles to dist/index.js, so dist/web/ is a sibling.
const thisDir = path.dirname(fileURLToPath(import.meta.url));
const webDir = path.resolve(thisDir, 'web');
if (!fs.existsSync(webDir)) {
logger.debug({ webDir }, 'Web dashboard dist not found, skipping static mount');
return;
}
logger.info({ webDir }, 'Mounting web dashboard');
// Serve static assets
this.app.use(express.static(webDir, { index: false }));
// SPA fallback: any GET that doesn't match /api, /sse, /messages, /health
// returns index.html so client-side routing works.
this.app.get('*', (req, res, next) => {
// Skip API / MCP / health routes
if (
req.path.startsWith('/api/') ||
req.path.startsWith('/sse') ||
req.path.startsWith('/messages') ||
req.path === '/health'
) {
next();
return;
}
const indexPath = path.join(webDir, 'index.html');
if (fs.existsSync(indexPath)) {
res.sendFile(indexPath);
} else {
next();
}
});
}
private async buildHealthResponse(): Promise<Record<string, unknown>> { private async buildHealthResponse(): Promise<Record<string, unknown>> {
// Memory usage // Memory usage
const mem = process.memoryUsage(); const mem = process.memoryUsage();
@@ -355,8 +449,8 @@ export class AppServer {
external: Math.round(mem.external / 1024 / 1024), external: Math.round(mem.external / 1024 / 1024),
}; };
// Active SSE sessions // Active MCP sessions (streamable + deprecated SSE)
const activeSessions = this.transports.size; const activeSessions = this.streamableTransports.size + this.sseTransports.size;
// Plugin health checks // Plugin health checks
const pluginHealth: Record<string, { healthy: boolean; message?: string }> = {}; const pluginHealth: Record<string, { healthy: boolean; message?: string }> = {};
@@ -392,4 +486,50 @@ export class AppServer {
memory: memoryMb, memory: memoryMb,
}; };
} }
private createSessionMcpServer(): McpServer {
const sessionServer = new McpServer(
{ name: 'social-mcp', version: PACKAGE_VERSION },
);
for (const plugin of this.plugins) {
plugin.registerTools(sessionServer, browserManager);
}
return sessionServer;
}
private cleanupSseSession(sessionId: string): void {
this.sseTransports.delete(sessionId);
const sessionServer = this.sseSessionServers.get(sessionId);
this.sseSessionServers.delete(sessionId);
if (!sessionServer) return;
void sessionServer.close().catch((err: unknown) => {
logger.warn({ err, sessionId }, 'Error closing SSE MCP server');
});
}
private cleanupStreamableSession(sessionId: string): void {
this.streamableTransports.delete(sessionId);
const sessionServer = this.streamableSessionServers.get(sessionId);
this.streamableSessionServers.delete(sessionId);
if (!sessionServer) return;
void sessionServer.close().catch((err: unknown) => {
logger.warn({ err, sessionId }, 'Error closing Streamable HTTP MCP server');
});
}
private sendProtocolMismatch(res: express.Response, message: string): void {
res.status(400).json({
jsonrpc: '2.0',
error: {
code: -32000,
message,
},
id: null,
});
}
private getFirstHeaderValue(value: string | string[] | undefined): string | undefined {
if (!value) return undefined;
return Array.isArray(value) ? value[0] : value;
}
} }
+71
View File
@@ -0,0 +1,71 @@
import { browserManager } from '../browser/manager.js';
import { AppServer, type PlatformPlugin } from '../server/app.js';
import { logger } from '../utils/logger.js';
export function startServerWithPlugins(plugins: PlatformPlugin[]): void {
const appServer = new AppServer();
for (const plugin of plugins) {
appServer.registerPlugin(plugin);
}
appServer.start().catch((err: unknown) => {
logger.fatal({ err }, 'Failed to start server');
process.exit(1);
});
let shuttingDown = false;
async function gracefulShutdown(signal: string): Promise<void> {
if (shuttingDown) return;
shuttingDown = true;
logger.info({ signal }, 'Received shutdown signal — starting graceful shutdown');
const forceExitTimer = setTimeout(() => {
logger.fatal('Graceful shutdown timed out after 45s — forcing exit');
process.exit(1);
}, 45_000);
if (typeof forceExitTimer === 'object' && 'unref' in forceExitTimer) {
forceExitTimer.unref();
}
try {
logger.info('Shutdown step 1/5: draining browser queues');
await Promise.race([
browserManager.drain(),
new Promise<void>((resolve) => setTimeout(resolve, 30_000).unref()),
]);
logger.info('Shutdown step 2/5: closing browser');
await browserManager.close();
logger.info('Shutdown step 3/5: closing HTTP server');
await appServer.close();
logger.info('Shutdown step 4/5: flushing logger');
logger.flush();
logger.info('Shutdown step 5/5: exiting');
process.exit(0);
} catch (err: unknown) {
logger.fatal({ err }, 'Error during graceful shutdown');
process.exit(1);
}
}
process.on('SIGINT', () => void gracefulShutdown('SIGINT'));
process.on('SIGTERM', () => void gracefulShutdown('SIGTERM'));
process.on('unhandledRejection', (reason: unknown) => {
logger.fatal({ err: reason }, 'Unhandled promise rejection');
void gracefulShutdown('unhandledRejection');
});
process.on('uncaughtException', (err: Error) => {
logger.fatal({ err }, 'Uncaught exception');
void gracefulShutdown('uncaughtException');
});
}
@@ -7,6 +7,7 @@ import { logger } from './logger.js';
export enum ErrorCategory { export enum ErrorCategory {
TIMEOUT = 'TIMEOUT', TIMEOUT = 'TIMEOUT',
AUTH_REQUIRED = 'AUTH_REQUIRED', AUTH_REQUIRED = 'AUTH_REQUIRED',
CAPTCHA_REQUIRED = 'CAPTCHA_REQUIRED',
SELECTOR_NOT_FOUND = 'SELECTOR_NOT_FOUND', SELECTOR_NOT_FOUND = 'SELECTOR_NOT_FOUND',
NETWORK = 'NETWORK', NETWORK = 'NETWORK',
PLATFORM_ERROR = 'PLATFORM_ERROR', PLATFORM_ERROR = 'PLATFORM_ERROR',
@@ -21,6 +22,14 @@ export enum ErrorCategory {
export function classifyError(err: Error): ErrorCategory { export function classifyError(err: Error): ErrorCategory {
const haystack = `${err.name} ${err.message}`.toLowerCase(); const haystack = `${err.name} ${err.message}`.toLowerCase();
if (
haystack.includes('captcha') ||
haystack.includes('show_captcha') ||
haystack.includes('验证码')
) {
return ErrorCategory.CAPTCHA_REQUIRED;
}
// Selector check BEFORE timeout — Playwright's selector timeout message // Selector check BEFORE timeout — Playwright's selector timeout message
// is "Timeout waiting for selector ..." which contains both keywords. // is "Timeout waiting for selector ..." which contains both keywords.
// The more specific match must come first. // The more specific match must come first.
@@ -35,8 +44,12 @@ export function classifyError(err: Error): ErrorCategory {
return ErrorCategory.TIMEOUT; return ErrorCategory.TIMEOUT;
} }
if (haystack.includes('net::err_')) { if (
return ErrorCategory.NETWORK; haystack.includes('net::err_') ||
haystack.includes('platform_error') ||
haystack.includes('平台错误')
) {
return ErrorCategory.PLATFORM_ERROR;
} }
if (haystack.includes('login') || haystack.includes('登录')) { if (haystack.includes('login') || haystack.includes('登录')) {
@@ -123,9 +136,12 @@ export async function withErrorHandling(
); );
const payload = JSON.stringify({ const payload = JSON.stringify({
tool: toolName, success: false,
error: category, error: {
message: sanitized, tool: toolName,
code: category,
message: sanitized,
},
}); });
return { return {
+115
View File
@@ -0,0 +1,115 @@
import crypto from 'node:crypto';
import fs from 'node:fs';
import path from 'node:path';
import { config } from '../config/index.js';
import { DatabaseSync } from './sqlite.js';
interface IdempotencyRow {
tool_name: string;
request_id: string;
input_hash: string;
response_json: string;
created_at: number;
}
export interface IdempotencyRecord {
toolName: string;
requestId: string;
inputHash: string;
responseData: unknown;
createdAt: string;
}
const DB_FILENAME = 'idempotency.db';
function canonicalize(value: unknown): unknown {
if (Array.isArray(value)) {
return value.map(canonicalize);
}
if (value && typeof value === 'object') {
const input = value as Record<string, unknown>;
const out: Record<string, unknown> = {};
for (const key of Object.keys(input).sort()) {
out[key] = canonicalize(input[key]);
}
return out;
}
return value;
}
export function computeIdempotencyHash(input: unknown): string {
const canonical = canonicalize(input);
return crypto
.createHash('sha256')
.update(JSON.stringify(canonical))
.digest('hex');
}
export class IdempotencyStore {
private readonly db: InstanceType<typeof DatabaseSync>;
constructor(baseDir = config.cookieDir, dbFilename = DB_FILENAME) {
fs.mkdirSync(baseDir, { recursive: true, mode: 0o700 });
const dbPath = path.join(baseDir, dbFilename);
this.db = new DatabaseSync(dbPath);
this.db.exec(`
CREATE TABLE IF NOT EXISTS idempotency_requests (
tool_name TEXT NOT NULL,
request_id TEXT NOT NULL,
input_hash TEXT NOT NULL,
response_json TEXT NOT NULL,
created_at INTEGER NOT NULL,
PRIMARY KEY (tool_name, request_id)
);
`);
}
get(toolName: string, requestId: string): IdempotencyRecord | null {
const stmt = this.db.prepare(`
SELECT tool_name, request_id, input_hash, response_json, created_at
FROM idempotency_requests
WHERE tool_name = ? AND request_id = ?
LIMIT 1
`);
const row = stmt.get(toolName, requestId) as IdempotencyRow | undefined;
if (!row) return null;
return {
toolName: row.tool_name,
requestId: row.request_id,
inputHash: row.input_hash,
responseData: JSON.parse(row.response_json),
createdAt: new Date(row.created_at).toISOString(),
};
}
put(
toolName: string,
requestId: string,
inputHash: string,
responseData: unknown,
): void {
const stmt = this.db.prepare(`
INSERT OR REPLACE INTO idempotency_requests (
tool_name, request_id, input_hash, response_json, created_at
) VALUES (?, ?, ?, ?, ?)
`);
stmt.run(
toolName,
requestId,
inputHash,
JSON.stringify(responseData),
Date.now(),
);
}
}
let singleton: IdempotencyStore | null = null;
export function getIdempotencyStore(): IdempotencyStore {
if (!singleton) {
singleton = new IdempotencyStore();
}
return singleton;
}
+10
View File
@@ -0,0 +1,10 @@
import { createRequire } from 'node:module';
import type { DatabaseSync as DatabaseSyncType } from 'node:sqlite';
const nodeRequire = createRequire(import.meta.url);
const loaded = nodeRequire('node:sqlite') as {
DatabaseSync: typeof DatabaseSyncType;
};
export const DatabaseSync = loaded.DatabaseSync;
@@ -39,9 +39,14 @@ describe('classifyError', () => {
expect(classifyError(err)).toBe(ErrorCategory.TIMEOUT); expect(classifyError(err)).toBe(ErrorCategory.TIMEOUT);
}); });
it('returns NETWORK when message contains "net::err_"', () => { it('returns PLATFORM_ERROR when message contains "net::err_"', () => {
const err = new Error('net::err_connection_refused'); const err = new Error('net::err_connection_refused');
expect(classifyError(err)).toBe(ErrorCategory.NETWORK); expect(classifyError(err)).toBe(ErrorCategory.PLATFORM_ERROR);
});
it('returns CAPTCHA_REQUIRED when message contains captcha keyword', () => {
const err = new Error('show_captcha');
expect(classifyError(err)).toBe(ErrorCategory.CAPTCHA_REQUIRED);
}); });
it('returns AUTH_REQUIRED when message contains "login"', () => { it('returns AUTH_REQUIRED when message contains "login"', () => {
@@ -129,9 +134,10 @@ describe('withErrorHandling', () => {
expect(result.content).toHaveLength(1); expect(result.content).toHaveLength(1);
const payload = JSON.parse(result.content[0]!.text); const payload = JSON.parse(result.content[0]!.text);
expect(payload.tool).toBe('publish_post'); expect(payload.success).toBe(false);
expect(payload.error).toBe(ErrorCategory.TIMEOUT); expect(payload.error.tool).toBe('publish_post');
expect(typeof payload.message).toBe('string'); expect(payload.error.code).toBe(ErrorCategory.TIMEOUT);
expect(typeof payload.error.message).toBe('string');
}); });
it('wraps non-Error throws into an Error', async () => { it('wraps non-Error throws into an Error', async () => {
@@ -142,8 +148,9 @@ describe('withErrorHandling', () => {
expect(result.isError).toBe(true); expect(result.isError).toBe(true);
const payload = JSON.parse(result.content[0]!.text); const payload = JSON.parse(result.content[0]!.text);
expect(payload.tool).toBe('my_tool'); expect(payload.success).toBe(false);
expect(payload.error).toBe(ErrorCategory.INTERNAL); expect(payload.error.tool).toBe('my_tool');
expect(payload.message).toContain('raw string error'); expect(payload.error.code).toBe(ErrorCategory.INTERNAL);
expect(payload.error.message).toContain('raw string error');
}); });
}); });
+9
View File
@@ -0,0 +1,9 @@
{
"extends": "../../tsconfig.base.json",
"compilerOptions": {
"outDir": "dist",
"rootDir": "src",
"composite": true
},
"include": ["src"]
}
@@ -1,13 +1,14 @@
import { defineConfig } from 'tsup'; import { defineConfig } from 'tsup';
export default defineConfig({ export default defineConfig({
entry: ['src/index.ts'], entry: ['src/**/*.ts'],
format: ['esm'], format: ['esm'],
target: 'node22', target: 'node22',
outDir: 'dist', outDir: 'dist',
clean: false, clean: true,
sourcemap: true, sourcemap: true,
dts: false, dts: false,
splitting: false, splitting: false,
shims: false, shims: false,
}); });
@@ -6,3 +6,4 @@ export default defineConfig({
environment: 'node', environment: 'node',
}, },
}); });
+2621
View File
File diff suppressed because it is too large Load Diff
+4
View File
@@ -0,0 +1,4 @@
packages:
- 'apps/*'
- 'packages/*'
-89
View File
@@ -1,89 +0,0 @@
import { logger } from './utils/logger.js';
import { browserManager } from './browser/manager.js';
import { AppServer } from './server/app.js';
import { xiaohongshuPlugin } from './platforms/xiaohongshu/index.js';
// ---------------------------------------------------------------------------
// Bootstrap
// ---------------------------------------------------------------------------
const appServer = new AppServer();
// -- Platform plugins -------------------------------------------------------
appServer.registerPlugin(xiaohongshuPlugin);
// -- Start ------------------------------------------------------------------
appServer.start().catch((err: unknown) => {
logger.fatal({ err }, 'Failed to start server');
process.exit(1);
});
// ---------------------------------------------------------------------------
// Graceful shutdown
// ---------------------------------------------------------------------------
let shuttingDown = false;
async function gracefulShutdown(signal: string): Promise<void> {
if (shuttingDown) return;
shuttingDown = true;
logger.info({ signal }, 'Received shutdown signal — starting graceful shutdown');
// Safety net: if graceful shutdown takes too long, force exit.
const forceExitTimer = setTimeout(() => {
logger.fatal('Graceful shutdown timed out after 45s — forcing exit');
process.exit(1);
}, 45_000);
// Prevent the safety-net timer from keeping the process alive on its own.
if (typeof forceExitTimer === 'object' && 'unref' in forceExitTimer) {
forceExitTimer.unref();
}
try {
// Step 1: Drain browser queues so in-flight operations finish (max 30s).
logger.info('Shutdown step 1/5: draining browser queues');
await Promise.race([
browserManager.drain(),
new Promise<void>((resolve) => setTimeout(resolve, 30_000).unref()),
]);
// Step 2: Close the browser and all contexts.
logger.info('Shutdown step 2/5: closing browser');
await browserManager.close();
// Step 3: Close the HTTP server (stop accepting new connections).
logger.info('Shutdown step 3/5: closing HTTP server');
await appServer.close();
// Step 4: Flush structured logs so nothing is lost.
logger.info('Shutdown step 4/5: flushing logger');
logger.flush();
// Step 5: Exit cleanly.
logger.info('Shutdown step 5/5: exiting');
process.exit(0);
} catch (err: unknown) {
logger.fatal({ err }, 'Error during graceful shutdown');
process.exit(1);
}
}
process.on('SIGINT', () => void gracefulShutdown('SIGINT'));
process.on('SIGTERM', () => void gracefulShutdown('SIGTERM'));
// ---------------------------------------------------------------------------
// Global error handlers
// ---------------------------------------------------------------------------
process.on('unhandledRejection', (reason: unknown) => {
logger.fatal({ err: reason }, 'Unhandled promise rejection');
void gracefulShutdown('unhandledRejection');
});
process.on('uncaughtException', (err: Error) => {
logger.fatal({ err }, 'Uncaught exception');
void gracefulShutdown('uncaughtException');
});
-657
View File
@@ -1,657 +0,0 @@
import type { McpServer } from '@modelcontextprotocol/sdk/server/mcp.js';
import type { Router } from 'express';
import type { BrowserManager } from '../../browser/manager.js';
import { config } from '../../config/index.js';
import { withErrorHandling } from '../../utils/errors.js';
import { resolveMediaInput, cleanupFile } from '../../utils/downloader.js';
import { checkLoginStatus, getLoginQRCode, deleteCookies } from './login.js';
import { listFeeds } from './feeds.js';
import { searchFeeds } from './search.js';
import { getFeedDetail, getSubComments } from './feed-detail.js';
import { getUserProfile } from './user-profile.js';
import { publishImageNote } from './publish.js';
import { publishVideoNote } from './publish-video.js';
import { listMyNotes } from './my-notes.js';
import { postComment, replyComment } from './comment.js';
import { toggleLike, toggleFavorite } from './interaction.js';
import { getCommentNotifications, replyNotification } from './notification.js';
import { createXhsRoutes } from './routes.js';
import {
CheckLoginSchema,
GetLoginQRCodeSchema,
DeleteCookiesSchema,
ListFeedsSchema,
SearchSchema,
GetFeedDetailSchema,
GetSubCommentsSchema,
GetUserProfileSchema,
PublishImageSchema,
PublishVideoSchema,
ListMyNotesSchema,
PostCommentSchema,
ReplyCommentSchema,
LikeSchema,
FavoriteSchema,
GetCommentNotificationsSchema,
ReplyNotificationSchema,
} from './schemas.js';
import type { SearchFilters } from './types.js';
import type { PlatformPlugin } from '../../server/app.js';
// ---------------------------------------------------------------------------
// Constants
// ---------------------------------------------------------------------------
const PLATFORM = 'xiaohongshu';
/** Maximum file size for video uploads (500 MB). */
const VIDEO_MAX_SIZE_MB = 500;
/** Maximum file size for image uploads (20 MB — default in validateMediaPath). */
const IMAGE_MAX_SIZE_MB = 20;
// ---------------------------------------------------------------------------
// PlatformPlugin implementation
// ---------------------------------------------------------------------------
export const xiaohongshuPlugin: PlatformPlugin = {
name: PLATFORM,
// =========================================================================
// REST API routes (Phase 5)
// =========================================================================
registerRoutes(router: Router, browser: BrowserManager): void {
const xhsRouter = createXhsRoutes(browser);
router.use('/', xhsRouter);
},
// =========================================================================
// MCP tools
// =========================================================================
registerTools(server: McpServer, browser: BrowserManager): void {
// =====================================================================
// Phase 2: Login management (3 tools)
// =====================================================================
// -----------------------------------------------------------------------
// xhs_check_login
// -----------------------------------------------------------------------
server.tool(
'xhs_check_login',
'Check Xiaohongshu login status',
CheckLoginSchema,
async () => {
return withErrorHandling('xhs_check_login', async () => {
const timeoutMs = config.operationTimeouts['login'] ?? config.operationTimeouts['default'] ?? 60_000;
const status = await browser.withPage(
PLATFORM,
async (page) => checkLoginStatus(page),
timeoutMs,
);
return {
content: [
{
type: 'text' as const,
text: JSON.stringify(status),
},
],
};
});
},
);
// -----------------------------------------------------------------------
// xhs_get_login_qrcode
// -----------------------------------------------------------------------
server.tool(
'xhs_get_login_qrcode',
'Get Xiaohongshu login QR code (user scans with phone)',
GetLoginQRCodeSchema,
async () => {
return withErrorHandling('xhs_get_login_qrcode', async () => {
const result = await getLoginQRCode(browser);
return {
content: [
{
type: 'text' as const,
text: JSON.stringify(result),
},
],
};
});
},
);
// -----------------------------------------------------------------------
// xhs_delete_cookies
// -----------------------------------------------------------------------
server.tool(
'xhs_delete_cookies',
'Delete Xiaohongshu cookies and reset login session',
DeleteCookiesSchema,
async () => {
return withErrorHandling('xhs_delete_cookies', async () => {
await deleteCookies(browser);
return {
content: [
{
type: 'text' as const,
text: JSON.stringify({ success: true, message: 'Cookies deleted' }),
},
],
};
});
},
);
// =====================================================================
// Phase 3: Content browsing (4 tools)
// =====================================================================
// -----------------------------------------------------------------------
// xhs_list_feeds
// -----------------------------------------------------------------------
server.tool(
'xhs_list_feeds',
'Get Xiaohongshu explore page recommended feed list',
ListFeedsSchema,
async () => {
return withErrorHandling('xhs_list_feeds', async () => {
const timeoutMs = config.operationTimeouts['feed_list'] ?? config.operationTimeouts['default'] ?? 60_000;
const feeds = await browser.withPage(
PLATFORM,
async (page) => listFeeds(page),
timeoutMs,
);
return {
content: [
{
type: 'text' as const,
text: JSON.stringify(feeds),
},
],
};
});
},
);
// -----------------------------------------------------------------------
// xhs_search
// -----------------------------------------------------------------------
server.tool(
'xhs_search',
'Search Xiaohongshu notes by keyword with optional filters (sort, type, time range)',
SearchSchema,
async (args) => {
return withErrorHandling('xhs_search', async () => {
const timeoutMs = config.operationTimeouts['search'] ?? config.operationTimeouts['default'] ?? 60_000;
const filters: SearchFilters | undefined = args.filters
? {
sort: args.filters.sort,
type: args.filters.type,
time: args.filters.time,
}
: undefined;
const feeds = await browser.withPage(
PLATFORM,
async (page) => searchFeeds(page, args.keyword, filters),
timeoutMs,
);
return {
content: [
{
type: 'text' as const,
text: JSON.stringify(feeds),
},
],
};
});
},
);
// -----------------------------------------------------------------------
// xhs_get_feed_detail
// -----------------------------------------------------------------------
server.tool(
'xhs_get_feed_detail',
'Get Xiaohongshu note detail including content, images, stats, and first-screen comments (use xhs_get_sub_comments to load full replies)',
GetFeedDetailSchema,
async (args) => {
return withErrorHandling('xhs_get_feed_detail', async () => {
const timeoutMs = config.operationTimeouts['feed_detail'] ?? config.operationTimeouts['default'] ?? 60_000;
const detail = await browser.withPage(
PLATFORM,
async (page) =>
getFeedDetail(page, args.feed_id, args.xsec_token),
timeoutMs,
);
return {
content: [
{
type: 'text' as const,
text: JSON.stringify(detail),
},
],
};
});
},
);
// -----------------------------------------------------------------------
// xhs_get_sub_comments
// -----------------------------------------------------------------------
server.tool(
'xhs_get_sub_comments',
'Load all sub-comments (replies) for a specific parent comment on a Xiaohongshu note',
GetSubCommentsSchema,
async (args) => {
return withErrorHandling('xhs_get_sub_comments', async () => {
const timeoutMs =
config.operationTimeouts['feed_detail'] ??
config.operationTimeouts['default'] ??
60_000;
const result = await browser.withPage(
PLATFORM,
async (page) =>
getSubComments(page, args.feed_id, args.xsec_token, args.comment_id, args.max_count),
timeoutMs,
);
return {
content: [
{
type: 'text' as const,
text: JSON.stringify(result),
},
],
};
});
},
);
// -----------------------------------------------------------------------
// xhs_get_user_profile
// -----------------------------------------------------------------------
server.tool(
'xhs_get_user_profile',
'Get Xiaohongshu user profile information including bio, stats, and recent notes',
GetUserProfileSchema,
async (args) => {
return withErrorHandling('xhs_get_user_profile', async () => {
const timeoutMs = config.operationTimeouts['user_profile'] ?? config.operationTimeouts['default'] ?? 60_000;
const profile = await browser.withPage(
PLATFORM,
async (page) =>
getUserProfile(page, args.user_id, args.xsec_token),
timeoutMs,
);
return {
content: [
{
type: 'text' as const,
text: JSON.stringify(profile),
},
],
};
});
},
);
// =====================================================================
// My published notes (1 tool)
// =====================================================================
// -----------------------------------------------------------------------
// xhs_list_my_notes
// -----------------------------------------------------------------------
server.tool(
'xhs_list_my_notes',
'List your published notes on Xiaohongshu from the creator center',
ListMyNotesSchema,
async () => {
return withErrorHandling('xhs_list_my_notes', async () => {
const timeoutMs =
config.operationTimeouts['feed_list'] ??
config.operationTimeouts['default'] ??
60_000;
const notes = await browser.withPage(
PLATFORM,
async (page) => listMyNotes(page),
timeoutMs,
);
return {
content: [{ type: 'text' as const, text: JSON.stringify(notes) }],
};
});
},
);
// =====================================================================
// Phase 4: Content publishing (2 tools)
// =====================================================================
// -----------------------------------------------------------------------
// xhs_publish_image
// -----------------------------------------------------------------------
server.tool(
'xhs_publish_image',
'Publish an image note on Xiaohongshu. Provide local file paths for images.',
PublishImageSchema,
async (args) => {
return withErrorHandling('xhs_publish_image', async () => {
// Resolve all images (local path or URL download) before acquiring browser.
const resolved: Array<{ path: string; temporary: boolean }> = [];
for (const img of args.images) {
resolved.push(await resolveMediaInput(img, { maxSizeMB: IMAGE_MAX_SIZE_MB }));
}
const validatedPaths = resolved.map((r) => r.path);
const timeoutMs =
config.operationTimeouts['publish'] ??
config.operationTimeouts['default'] ??
300_000;
try {
const result = await browser.withPage(
PLATFORM,
async (page) =>
publishImageNote(page, args.title, args.content, validatedPaths, {
tags: args.tags,
scheduleAt: args.schedule_at,
isOriginal: args.is_original,
visibility: args.visibility,
}),
timeoutMs,
);
return {
content: [{ type: 'text' as const, text: JSON.stringify(result) }],
};
} finally {
for (const r of resolved) {
if (r.temporary) await cleanupFile(r.path).catch(() => undefined);
}
}
});
},
);
// -----------------------------------------------------------------------
// xhs_publish_video
// -----------------------------------------------------------------------
server.tool(
'xhs_publish_video',
'Publish a video note on Xiaohongshu. Provide a local file path for the video.',
PublishVideoSchema,
async (args) => {
return withErrorHandling('xhs_publish_video', async () => {
// Resolve video (local path or URL download) before acquiring browser.
const resolvedVideo = await resolveMediaInput(args.video, { maxSizeMB: VIDEO_MAX_SIZE_MB });
const timeoutMs =
config.operationTimeouts['publish'] ??
config.operationTimeouts['default'] ??
300_000;
try {
const result = await browser.withPage(
PLATFORM,
async (page) =>
publishVideoNote(page, args.title, args.content, resolvedVideo.path, {
tags: args.tags,
scheduleAt: args.schedule_at,
visibility: args.visibility,
}),
timeoutMs,
);
return {
content: [{ type: 'text' as const, text: JSON.stringify(result) }],
};
} finally {
if (resolvedVideo.temporary) await cleanupFile(resolvedVideo.path).catch(() => undefined);
}
});
},
);
// =====================================================================
// Phase 4: Interactions (4 tools)
// =====================================================================
// -----------------------------------------------------------------------
// xhs_post_comment
// -----------------------------------------------------------------------
server.tool(
'xhs_post_comment',
'Post a comment on a Xiaohongshu note',
PostCommentSchema,
async (args) => {
return withErrorHandling('xhs_post_comment', async () => {
const timeoutMs =
config.operationTimeouts['comment'] ??
config.operationTimeouts['default'] ??
20_000;
const result = await browser.withPage(
PLATFORM,
async (page) =>
postComment(page, args.feed_id, args.xsec_token, args.content),
timeoutMs,
);
return {
content: [
{
type: 'text' as const,
text: JSON.stringify(result),
},
],
};
});
},
);
// -----------------------------------------------------------------------
// xhs_reply_comment
// -----------------------------------------------------------------------
server.tool(
'xhs_reply_comment',
'Reply to a comment on a Xiaohongshu note',
ReplyCommentSchema,
async (args) => {
return withErrorHandling('xhs_reply_comment', async () => {
const timeoutMs =
config.operationTimeouts['reply'] ??
config.operationTimeouts['default'] ??
20_000;
const result = await browser.withPage(
PLATFORM,
async (page) =>
replyComment(
page,
args.feed_id,
args.xsec_token,
args.content,
args.comment_id,
args.user_id,
),
timeoutMs,
);
return {
content: [
{
type: 'text' as const,
text: JSON.stringify(result),
},
],
};
});
},
);
// -----------------------------------------------------------------------
// xhs_like
// -----------------------------------------------------------------------
server.tool(
'xhs_like',
'Toggle like on a Xiaohongshu note',
LikeSchema,
async (args) => {
return withErrorHandling('xhs_like', async () => {
const timeoutMs =
config.operationTimeouts['like'] ??
config.operationTimeouts['default'] ??
15_000;
const result = await browser.withPage(
PLATFORM,
async (page) =>
toggleLike(page, args.feed_id, args.xsec_token),
timeoutMs,
);
return {
content: [
{
type: 'text' as const,
text: JSON.stringify(result),
},
],
};
});
},
);
// -----------------------------------------------------------------------
// xhs_favorite
// -----------------------------------------------------------------------
server.tool(
'xhs_favorite',
'Toggle favorite on a Xiaohongshu note',
FavoriteSchema,
async (args) => {
return withErrorHandling('xhs_favorite', async () => {
const timeoutMs =
config.operationTimeouts['favorite'] ??
config.operationTimeouts['default'] ??
15_000;
const result = await browser.withPage(
PLATFORM,
async (page) =>
toggleFavorite(page, args.feed_id, args.xsec_token),
timeoutMs,
);
return {
content: [
{
type: 'text' as const,
text: JSON.stringify(result),
},
],
};
});
},
);
// =====================================================================
// Notifications (2 tools)
// =====================================================================
// -----------------------------------------------------------------------
// xhs_get_comment_notifications
// -----------------------------------------------------------------------
server.tool(
'xhs_get_comment_notifications',
'Get comment and @ notifications from Xiaohongshu notification page',
GetCommentNotificationsSchema,
async (args) => {
return withErrorHandling('xhs_get_comment_notifications', async () => {
const timeoutMs =
config.operationTimeouts['feed_detail'] ??
config.operationTimeouts['default'] ??
60_000;
const notifications = await browser.withPage(
PLATFORM,
async (page) =>
getCommentNotifications(page, args.max_count),
timeoutMs,
);
return {
content: [{ type: 'text' as const, text: JSON.stringify(notifications) }],
};
});
},
);
// -----------------------------------------------------------------------
// xhs_reply_notification
// -----------------------------------------------------------------------
server.tool(
'xhs_reply_notification',
'Reply to a comment notification inline on the Xiaohongshu notification page',
ReplyNotificationSchema,
async (args) => {
return withErrorHandling('xhs_reply_notification', async () => {
const timeoutMs =
config.operationTimeouts['reply'] ??
config.operationTimeouts['default'] ??
20_000;
const result = await browser.withPage(
PLATFORM,
async (page) =>
replyNotification(page, args.user_id, args.comment_content, args.reply_content),
timeoutMs,
);
return {
content: [{ type: 'text' as const, text: JSON.stringify(result) }],
};
});
},
);
},
};
-173
View File
@@ -1,173 +0,0 @@
import { z } from 'zod';
// ---------------------------------------------------------------------------
// MCP tool parameter schemas for all 13 Xiaohongshu tools.
//
// Phase 2 tools (login) have no parameters — their schemas are empty objects.
// Phase 3/4 schemas are defined here so that the full tool surface is
// established upfront and types can be inferred with z.infer<>.
// ---------------------------------------------------------------------------
// -- Phase 2: Login management (3 tools) -----------------------------------
/** xhs_check_login — no parameters. */
export const CheckLoginSchema = {};
/** xhs_get_login_qrcode — no parameters. */
export const GetLoginQRCodeSchema = {};
/** xhs_delete_cookies — no parameters. */
export const DeleteCookiesSchema = {};
// -- Phase 3: Content browsing (4 tools) -----------------------------------
/** xhs_list_feeds — no parameters. */
export const ListFeedsSchema = {};
/** xhs_search */
export const SearchSchema = {
keyword: z.string().describe('Search keyword'),
filters: z
.object({
sort: z
.enum(['general', 'time_descending', 'popularity_descending'])
.optional()
.describe('Sort order'),
type: z
.enum(['all', 'note', 'video'])
.optional()
.describe('Content type filter'),
time: z
.enum(['all', 'day', 'week', 'half_year'])
.optional()
.describe('Time range filter'),
})
.optional()
.describe('Optional search filters'),
};
/** xhs_get_feed_detail */
export const GetFeedDetailSchema = {
feed_id: z.string().describe('Feed (note) ID'),
xsec_token: z.string().describe('Security token for the feed'),
};
/** xhs_get_sub_comments */
export const GetSubCommentsSchema = {
feed_id: z.string().describe('Feed (note) ID'),
xsec_token: z.string().describe('Security token for the feed'),
comment_id: z.string().describe('Parent comment ID whose sub-comments to load'),
max_count: z
.number()
.int()
.min(1)
.max(200)
.optional()
.default(20)
.describe('Maximum number of sub-comments to load (1200, default 20)'),
};
/** xhs_get_user_profile */
export const GetUserProfileSchema = {
user_id: z.string().describe('User ID'),
xsec_token: z.string().describe('Security token for the user page'),
};
// -- Phase 4: Content publishing (2 tools) ---------------------------------
/** xhs_publish_image */
export const PublishImageSchema = {
title: z.string().min(1).max(20, 'Title must be ≤ 20 characters').describe('Note title (max 20 chars)'),
content: z.string().max(1000, 'Content must be ≤ 1000 characters').describe('Note body text (max 1000 chars)'),
images: z
.array(z.string())
.min(1)
.max(18, 'Maximum 18 images per note')
.describe('Array of local file paths or HTTP/HTTPS URLs (118 images)'),
tags: z.array(z.string()).optional().describe('Hashtags to attach'),
schedule_at: z
.string()
.optional()
.describe('ISO 8601 datetime for scheduled publishing'),
is_original: z
.boolean()
.optional()
.default(false)
.describe('Mark as original content'),
visibility: z
.enum(['public', 'private', 'friends'])
.optional()
.default('public')
.describe('Visibility setting'),
};
/** xhs_publish_video */
export const PublishVideoSchema = {
title: z.string().min(1).max(20, 'Title must be ≤ 20 characters').describe('Note title (max 20 chars)'),
content: z.string().max(1000, 'Content must be ≤ 1000 characters').describe('Note body text (max 1000 chars)'),
video: z.string().describe('Local file path or HTTP/HTTPS URL for the video'),
tags: z.array(z.string()).optional().describe('Hashtags to attach'),
schedule_at: z
.string()
.optional()
.describe('ISO 8601 datetime for scheduled publishing'),
visibility: z
.enum(['public', 'private', 'friends'])
.optional()
.default('public')
.describe('Visibility setting'),
};
// -- Phase 4: Interactions (4 tools) ---------------------------------------
/** xhs_post_comment */
export const PostCommentSchema = {
feed_id: z.string().describe('Feed ID to comment on'),
xsec_token: z.string().describe('Security token for the feed'),
content: z.string().min(1).describe('Comment text'),
};
/** xhs_reply_comment */
export const ReplyCommentSchema = {
feed_id: z.string().describe('Feed ID'),
xsec_token: z.string().describe('Security token for the feed'),
comment_id: z.string().optional().describe('Comment ID to reply to'),
user_id: z.string().optional().describe('User ID of the comment author'),
content: z.string().min(1).describe('Reply text'),
};
/** xhs_like */
export const LikeSchema = {
feed_id: z.string().describe('Feed ID to toggle like'),
xsec_token: z.string().describe('Security token for the feed'),
};
/** xhs_list_my_notes — no parameters. */
export const ListMyNotesSchema = {};
// -- Phase 5: Notifications (2 tools) --------------------------------------
/** xhs_get_comment_notifications */
export const GetCommentNotificationsSchema = {
max_count: z
.number()
.int()
.min(1)
.max(50)
.optional()
.default(20)
.describe('Maximum number of notifications to return (150, default 20)'),
};
/** xhs_reply_notification */
export const ReplyNotificationSchema = {
user_id: z.string().describe('User ID of the comment author (from notification)'),
comment_content: z.string().describe('Original comment content to match the notification'),
reply_content: z.string().min(1).describe('Reply text to send'),
};
/** xhs_favorite */
export const FavoriteSchema = {
feed_id: z.string().describe('Feed ID to toggle favorite'),
xsec_token: z.string().describe('Security token for the feed'),
};
+23
View File
@@ -0,0 +1,23 @@
{
"compilerOptions": {
"target": "ES2022",
"module": "ESNext",
"moduleResolution": "bundler",
"lib": ["ES2022", "DOM", "DOM.Iterable"],
"strict": true,
"esModuleInterop": true,
"skipLibCheck": true,
"forceConsistentCasingInFileNames": true,
"resolveJsonModule": true,
"baseUrl": ".",
"paths": {
"@social/core/*": ["packages/core/src/*"]
},
"declaration": true,
"declarationMap": true,
"sourceMap": true,
"noUncheckedIndexedAccess": true,
"noUnusedLocals": true,
"noUnusedParameters": true
}
}
+6 -21
View File
@@ -1,23 +1,8 @@
{ {
"compilerOptions": { "files": [],
"target": "ES2022", "references": [
"module": "ESNext", { "path": "./packages/core" },
"moduleResolution": "bundler", { "path": "./apps/xhs-mcp" },
"lib": ["ES2022"], { "path": "./apps/xhh-mcp" }
"outDir": "dist", ]
"rootDir": "src",
"strict": true,
"esModuleInterop": true,
"skipLibCheck": true,
"forceConsistentCasingInFileNames": true,
"resolveJsonModule": true,
"declaration": true,
"declarationMap": true,
"sourceMap": true,
"noUncheckedIndexedAccess": true,
"noUnusedLocals": true,
"noUnusedParameters": true
},
"include": ["src"],
"exclude": ["node_modules", "dist", "**/*.test.ts"]
} }
-2808
View File
File diff suppressed because it is too large Load Diff