pipeline/config/default.yaml

# Pipeline default configuration
#
# Priority: CLI flags > this file > .env (for secrets/endpoints)

branding:
  channelName: "知势·科技洞察"

llm:
  # baseURL: set via OPENAI_BASE_URL in .env
  # apiKey: set via OPENAI_API_KEY in .env
  model: "model-stable"

tts:
  provider: "openai-tts"
  speed: 1.0
  format: "mp3"
  # alignment:
  #   provider: "whisper"
  #   whisperModel: "base"
  #   language: "zh"

  openai-tts:
    # apiKey: set via OPENAI_TTS_API_KEY in .env
    # baseURL: set via OPENAI_TTS_BASE_URL in .env
    model: "seed-tts-1.1"
    defaultVoice: "zh_female_vv_uranus_bigtts"

  fish-audio:
    # apiKey: set via FISH_AUDIO_API_KEY in .env
    model: "s2-pro"
    defaultVoice: ""

  minimax:
    # apiKey: set via MINIMAX_API_KEY in .env
    # groupId: set via MINIMAX_GROUP_ID in .env
    model: "speech-02-hd"
    defaultVoice: ""

  elevenlabs:
    # apiKey: set via ELEVENLABS_API_KEY in .env
    model: "eleven_multilingual_v2"
    defaultVoice: "21m00Tcm4TlvDq8ikWAM"

render:
  fps: 30
  codec: "h264"

output:
  dir: "./output"
  format: "mp4"
  # Auto-clean cached outputs older than this many days (0 disables). Default 30.
  # Override with the OUTPUT_RETENTION_DAYS env var.
  retentionDays: 30

# 每日推荐的"已讲历史"去重状态（covered store，@pipeline/shared/node）。
#   coveredWindowDays — 滚动去重窗口（天）：窗口内讲过的项目不再讲，过期释放。
# 文件位置由 PIPELINE_COVERED_DIR 指定（默认系统临时目录，重启即丢——部署时
# 必须指向持久卷，推荐 jobs 卷的 covered/ 子目录，见 docs/DEPLOYMENT.md）。
state:
  coveredWindowDays: 90

# Alibaba Cloud OSS — rendered videos are uploaded here and a resource URL is
# returned. Non-sensitive values may live in this file; secrets (access keys)
# belong in .env. Every field can be overridden by an OSS_* env var.
oss:
  region: "oss-cn-hangzhou"          # OSS_REGION
  bucket: ""                         # OSS_BUCKET
  accessKeyId: ""                    # OSS_ACCESS_KEY_ID  (use .env)
  accessKeySecret: ""                # OSS_ACCESS_KEY_SECRET (use .env)
  # endpoint: ""                     # custom endpoint, e.g. https://oss-cn-hangzhou.aliyuncs.com — OSS_ENDPOINT
  # path: "videos/"                  # object key prefix — OSS_PATH
  # publicDomain: ""                 # base URL for resource links (e.g. a CDN) — OSS_PUBLIC_DOMAIN
  # secure: true                     # use HTTPS — OSS_SECURE

# Feishu (Lark) custom-bot webhook — push results / failures after each job.
feishu:
  webhook: ""                        # FEISHU_WEBHOOK_URL (use .env)
  # secret: ""                       # signing secret — FEISHU_WEBHOOK_SECRET

# Publish manifest — written as {video}.yaml next to each rendered MP4 with the
# title/description/tags (LLM-produced for github-trending) plus the
# platform-specific 分区/category defined here. Lookup: publishMeta[platform][template].
# Only fill the combos you publish; missing combos just omit those keys.
publishMeta:
  bilibili:
    github-trending:
      tid: 122                  # 分区 id (122 = 科技→科普). Adjust to your target board.
      tags: ["GitHub", "开源", "热榜"]
    github-weekly:
      tid: 122
      tags: ["GitHub", "开源", "周榜"]
    github-daily-pick:
      tid: 122
      tags: ["GitHub", "开源", "推荐"]
    github-weekly-recap:
      tid: 122
      tags: ["GitHub", "开源", "周报"]
  douyin-long:
    github-trending:
      category: "科技"
      tags: ["GitHub", "开源"]
    github-weekly:
      category: "科技"
      tags: ["GitHub", "开源", "周榜"]
    github-daily-pick:
      category: "科技"
      tags: ["GitHub", "开源", "推荐"]
    github-weekly-recap:
      category: "科技"
      tags: ["GitHub", "开源", "周报"]
  douyin-short:
    github-trending:
      category: "科技"
      tags: ["GitHub", "开源"]
    github-weekly:
      category: "科技"
      tags: ["GitHub", "开源", "周榜"]
    github-daily-pick:
      category: "科技"
      tags: ["GitHub", "开源", "推荐"]
    github-weekly-recap:
      category: "科技"
      tags: ["GitHub", "开源", "周报"]

templates:
  news:
    primaryColor: "#1a56db"
    accentColor: "#dc2626"
  knowledge:
    primaryColor: "#0d9488"
    accentColor: "#6b7280"
  opinion:
    primaryColor: "#ea580c"
    accentColor: "#374151"
  marketing:
    primaryColor: "#9333ea"
    accentColor: "#ec4899"
  github-trending:
    primaryColor: "#22c55e"
    accentColor: "#3b82f6"
  github-weekly:
    primaryColor: "#d97706"
    accentColor: "#1e3a5f"
  github-daily-pick:
    primaryColor: "#8b5cf6"
    accentColor: "#22d3ee"
  github-weekly-recap:
    primaryColor: "#0284c7"
    accentColor: "#b45309"

collect:
  github-trending:
    url: "https://github.crawler.corp.shuidi.tech/api/trending"
    # Repo-detail endpoint used to fetch each top repo's README (per-repo
    # description grounding). Optional — without it, only trending metadata is used.
    repoUrl: "https://github.crawler.corp.shuidi.tech/api/repos/:owner/:repo"
  # Weekly board — same endpoints; the collector appends ?since=weekly.
  github-weekly:
    url: "https://github.crawler.corp.shuidi.tech/api/trending"
    repoUrl: "https://github.crawler.corp.shuidi.tech/api/repos/:owner/:repo"
  github-repo:
    url: "https://github.crawler.corp.shuidi.tech/api/repos/:owner/:repo"
  github-daily:
    trendingUrl: "https://github.crawler.corp.shuidi.tech/api/trending"
    repoUrl: "https://github.crawler.corp.shuidi.tech/api/repos/:owner/:repo"
    maxRepos: 5
  # 每日推荐选题：日/周/月三榜并集，筛掉窗口内已讲项目后按综合热度取第一
  # 周榜。windowDays 与 state.coveredWindowDays 保持一致；readmeLimit 是深度
  # 版的 README 截断（比榜单类的 2000/3000 更长）。
  github-daily-pick:
    url: "https://github.crawler.corp.shuidi.tech/api/trending"
    repoUrl: "https://github.crawler.corp.shuidi.tech/api/repos/:owner/:repo"
    windowDays: 90
    readmeLimit: 4000
  # 每周推荐回顾：纯读 covered store，无网络依赖（留空对象占位注册）。
  github-weekly-recap: {}

# 定时生成（容器内进程内调度）。Next.js 服务启动时（instrumentation 钩子）注册
# 一个进程内调度器，到点 spawn CLI 渲染——复用与 WebUI/HTTP 完全相同的链路，
# 任务同样进 jobs 列表、走 OSS/飞书发布。改配置后需重启容器生效。
#   cron     — 5 字段 cron（分 时 日 月 周），按 TZ（默认 Asia/Shanghai=北京时间）
#   template — 渲染模板（news|knowledge|opinion|marketing|github-trending|github-weekly）
#   platform — 平台规格（bilibili|douyin-long|douyin-short|...），默认 bilibili
#   source   — 采集源（如 github-trending）；定时场景一般用 source 自动取数
#   sourceArgs — 采集器附加参数（如 github-repo 的 owner/repo），可选
#   publish  — 是否上传 OSS + 推送飞书，默认 true
#   enabled  — 是否启用该条，默认 true
# 环境变量 SCHEDULES（JSON 数组，同结构）可整体覆盖（部署期改无需重建镜像）；
# SCHEDULER_ENABLED=false 关闭整个调度器。
schedules:
  - cron: "50 7 * * *"          # 北京时间每天 07:50
    template: "github-trending"
    platform: "bilibili"
    source: "github-trending"    # 先采集当日 trending，再解析→渲染→发布
    publish: true
    enabled: true
  - cron: "50 8 * * 6"          # 北京时间每周六 08:50
    template: "github-weekly"
    platform: "bilibili"
    source: "github-weekly"      # 采集本周 trending（since=weekly），再解析→渲染→发布
    publish: true
    enabled: true
  - cron: "20 9 * * *"          # 北京时间每天 09:20（错开 07:50 榜单与周六 08:50 周榜）
    template: "github-daily-pick"
    platform: "bilibili"
    source: "github-daily-pick"  # 三榜并集选题，六幕叙事推荐一个开源项目
    publish: true
    enabled: false               # 已取消每日推荐
  - cron: "20 18 * * 0"         # 北京时间每周日 18:20（覆盖当天上午的推荐，回顾完整一周）
    template: "github-weekly-recap"
    platform: "bilibili"
    source: "github-weekly-recap"  # 读本周已讲记录生成回顾，不重新爬榜
    publish: true
    enabled: false               # 已取消每周推荐回顾