Files
HomeAgent/deploy/packaging/package-linux.sh
JianFeeeee a68d6a428c feat: 模块化内核(编译期依赖图定序)+ 密钥防火墙 + 插件三态生命周期 (#10)
* feat(kernel): 模块化内核 + 编译期依赖图定序

把单体 init 序列拆成「模块自声明依赖 + 编译期生成依赖图 + 运行期只读图」。

## 为什么

原 cmd/homed/main.go 里 17 步 init 是**手写顺序**,依赖关系只存在于
注释和人脑里。实测踩到的坑(见 docs/zh/kernel-modularization-design.md):
 · nlp.defaultParser 被 initONNXParser 写、被 distiller 读,两者相距 60 行,
   只靠 ticker 10min 才没炸(隐式依赖,无声明)
 · 加一个模块要人肉找"该插在哪",插错的表现是运行期 nil 解引用
 · 关停是 defer ks.CloseAll(),而热更新路径 os.Exit(42) **跳过 defer**
   ⇒ docStore/distiller/textMem 的 flush 全部丢失(每次热更新都发生)

## 怎么改

- internal/kernelboot/:KernelState(一个结构体收全部包级共享变量,字段
  nil 安全)、Module 接口、graphfile(依赖图编解码 + 与二进制 hash 对账)、
  scheduler(按 BFS 层并行 Init)、syncbus(层边界写通知)、
  lifecycle(Quiesce/Resume 可逆伪关机,区别于一次性 CloseAll)
- internal/kernelmod/<name>/:每个模块一个文件夹,Requires/Provides 自声明;
  all/all.go 是唯一的 import 清单(Go 循环导入约束)
- cmd/homed-depgraph:编译期生成依赖图,Makefile 注入 digest
- 运行期只**读**图,不计算顺序(顺序是编译期决定的,源码里没有中央清单)

## 判据

kernelboot 8 条 + 图失败模式 7 条(全部拒绝并给准确错误)+ 模块目录数
与注册表一致性。变异测试 15/15 咬住,含:
 · 删掉 AddCloser 的锁 → -race 报数据竞争
 · Scheduler 同批跑去重 → 203 vs 200
 · 图 hash 不匹配 → 拒绝启动

## 顺带修的缺陷

热更新路径漏调模块关闭 ⇒ 显式 stopKernelModules(),并用 AST 判据钉住
"必须在 os.Exit 之前"(字符串检查在本仓已骗过我三次)。

实测:冷启动 1630ms(5 层/17 模块,原 17 步串行);生产 22 插件全部建链。

* feat(security): 密钥防火墙 + 配置加密 + close 纪律

## 密钥防火墙

模型只见别名、不见明文;插件工具调用入口(StageHost.ExecuteTool)做入向
回填。三维能力:
 · 出向:工具输出里的密钥被替换成 [[secret:name]] 别名后才进上下文
 · 入向:模型传别名,入口处填回真值(插件完全无感知)
 · 存储:config.db 里不存明文(0644 的库可读),敏感值独立 0600 文件

模式库「宁漏勿误伤」:宁可漏掉一次替换,也不把正常文本误当密钥。

## 配置加密

- internal/security/{codec,key,keyring,store,global}.go:AES-256-GCM,
  enc:v1: 前缀,密钥轮换(OldKeys 解码 + 新钥加密)
- harden.go:数据目录权限收紧
- embedded.go:支持内嵌主密钥(可选,仍保留外部 key 文件路径)
- EncryptExistingSecrets / AuditSecrets:启动时幂等迁移 + 审计

## close 纪律(用户明确要求:只要调 close 就要有前置检查)

internal/safeclose:Safe / SafeFunc。全仓 37 处裸 close(chan) → 0。
关键教训(都写进注释和判据):
 · `select { case <-ch: default: close(ch) }` **不是**并发安全的
   (两个 goroutine 可同时命中 default)—— healthcheck 原先是这种写法
 · `sync.Once` **不可重入**:X.Do(func(){ Safe(&X, ...) }) 会自锁,
   症状是 go test 直接挂住(3 个包 600s 超时),栈里只有嵌套的 Once.Do
 · map 里按请求持有的 channel(mcp/stdio.go、clawhubadapter/sidecar.go)
   改成 *pendingRPC 包装,每通道一个 once

## 判据

security 全量 + TestNoBareCloseOutsideSafeclose(AST 级扫描,含注释剥离)
+ TestNoNestedSelfOnceDeadlock(正则修正:\w+ 不匹配点号,捕获组要含 '.')。
变异 2/2 咬住。全量 52 包 ok。

* feat(plugin+deploy): 插件三态生命周期、数据目录迁移、发行包修复

## 插件三态生命周期(用户要求:自动启动 / 启用但不自动启动 / 禁用)

- Manifest 新增 requires_facilities / auto_start(*bool,用于区分
  "未声明"与"显式 false")
- plugin-host 模块(internal/kernelmod/pluginhost):用 kernelboot.Scheduler
  按依赖时机拉起,**不再手写两阶段**
- 运行态停启 StopPlugin/StartPlugin **可逆**:不写 disabled_plugins
  (不改用户决定)、不置 shuttingDown(之后还能加载)、会摘掉注册面
- 与 StopAll 的区别写进注释 —— StopAll 置 shuttingDown 后**永久无法加载**,
  那是进程退出用的,误用于"暂停"就是关了就死

## 数据目录迁移(伪关机 → 迁移 → 唤醒)

- kernelboot/lifecycle.go:Quiesce/Resume(可逆,区别于一次性 CloseAll),
  逆序暂停、逆序恢复、Pause 失败回滚
- kernelboot/migrate.go:BeginMigration/End 编排(先停插件 → 再停内核写者
  → 等在途归零)
- distiller/textmem 增 Pause/Resume 门(不是 cancel —— cancel 不可逆,
  恢复要重建整个对象)
- 模块 migration:体检 + 生成可执行迁移脚本

★ 迁移脚本的三个实测教训(都写成了判据):
 1. 无条件改 /etc/systemd/system ⇒ 演练把**生产** -data 改成了演练目录。
    加 --dry-run(零改动)+ 强制校验 systemd 当前 -data == OLD
 2. 本机没装 rsync,脚本在"服务已停、数据未搬"处硬失败 ⇒ cp -a 回退,
    并明确警告不保硬链接/ACL(生产 plugindev 有大量 git 硬链接)

## 发行包修复(全部经实测确认)

 1. package-linux.sh **完全漏打 depgraph.bin** ⇒ 内核 exit 1
    ("生产构建(内嵌 hash 非空)但读不到依赖图")
 2. build_homed 不走 Makefile ⇒ 单独修 Makefile 无效,必须同时生成图+注入
 3. build.sh 的 LDFLAGS 用**历史模块路径 gitcode.com** ⇒ -X 静默 no-op
    ⇒ --version 显示 unknown
 4. Makefile -ldflags 用**单引号** ⇒ 注入了字面 $GRAPH_HASH

## ★ build.sh 的版本号必须校验(真实事故)

原 `VERSION="$(git describe --tags --dirty)"`。无 tag 分支上它返回**分支名**
`backup/pre-identity-rewrite-20261005-51-g03d87a4e-dirty`,被注入 meta.Version
⇒ 插件握手判据 MajorVersion() 把整串当主版本 ⇒ **22 个子进程插件全部
握手失败**("主版本不匹配")。内核活着、systemd active、日志每条都像
插件的问题 —— 实际是发行包带着非法版本号。

修法:git describe 只在形如 vX.Y.Z 时采用,否则回退 meta.Version(权威源),
并对最终值做形态守卫。deploy/packaging/version_guard_test.go 3 条判据,
变异 2/2 咬住。

## deb 的 systemd unit 改为 guard 模式

ExecStart=/usr/bin/homed --role=guard --data /var/lib/homeagent
(worker exit 42 → guard 重拉,不经过 systemd、不计失败)
含 RestartSteps/RestartMaxDelaySec/TimeoutStopSec + 硬化
(NoNewPrivileges/ProtectSystem/ProtectHome/DevicePolicy/ReadWritePaths)。

## 其他

- internal/plugin/proc:非法 JSON 帧的报错**可定位**(内容摘要 + 可能原因 +
  修法)。生产实测 1038 条非法帧(插件往 stdout 打日志 = 污染协议通道)
- 示例插件 4 个 14 处 fmt.Printf → log.Printf
- hmapdev 模板同步修复;scaffold 判据剥离注释(模板注释里就含 fmt.Printf)

* chore: 收尾(主密钥轮换命令、IPC/知识库加固、SDK 对齐)

- cmd/homed-key-rotate:把 security 里的 RotateKey/RotateKeyFile/RekeySecrets
  串成一条**可执行**的运维路径。没有它,E-1 就只是库里的能力而不是
  "出事时能做的事",而轮换的价值恰恰在于能执行。
- cmd/homed:kmod_runtime(模块装配入口)、plugins_scheduled(failback
  白名单)、memcandidate、bootstrap_order_test / shutdown_flush_test
- internal/ipc、internal/devicebridge、internal/knowledge:配套加固
- third_party/homeagent-sdk/sdk:PluginMgrAPI 增三态能力(mgrStop/mgrStart/
  mgrIsRuntimeStopped)+ settings 对齐

判据:bootstrap_order_test(启动顺序不依赖手写序列)、
shutdown_flush_test(停机的两条路径必须对称 —— exit 42 会跳过 defer,
热更新路径下模块关闭/flush 曾全部丢失)。

* test(security): 测试夹具改用不可误认的占位值(修 push protection 拦截)

GitHub secret scanning 拦住了推送:三处测试夹具用了"看起来像真密钥"的
字面量:

  internal/security/prefilter_test.go:39
  internal/security/store_test.go:177
  internal/security/survey_test.go:20

它们全是**假密钥**(`AKIAIOSFODNN7EXAMPLE` 是 AWS 官方文档的标准示例,
其余是 `abcdefghij...` 这类可读占位符),但扫描器无法区分 —— 它只看形态。

★ 不绕过(不用 unblock URL、不加 allowlist):直接改夹具。
原因:一旦允许"假密钥可以通过",真密钥混进来时就没有最后一道门了。

改法:保留**模式库匹配所需的结构**,把可识别内容换成 FAKE:
  glpat-abcdefghijklmnopqrst       → glpat-FAKE-TEST-TOKEN-NOT-REAL
  sk-abcdefghij...012345           → sk-FAKE-TEST-KEY-NOT-REAL-0000000
  AKIAIOSFODNN7EXAMPLE             → AKIA0000FAKE0000TEST   (保持 20 字符)
  ASIAIOSFODNN7EXAMPLE             → ASIA0000FAKE0000TEST
  xoxb-123456789012-abcdefghijklmnop → xoxb-FAKE-TEST-TOKEN-NOT-REAL

★ 中途踩到一次:AWS 键的模式是 `\b(AKIA|ASIA)[0-9A-Z]{16}\b`(共 20 字符),
我第一次替换成 19 字符的 `AKIAFAKETESTKEY0000` ⇒ 模式匹配不上 ⇒
TestSurvey_WhatIsCaught / TestRecognizesCommonShapes /
TestPrefilterNeverBlocksRealMatches 三条当场变红。
判据正确地拒绝了"为了过扫描器而削弱覆盖"。

顺带核实:全仓 grep 无任何真实密钥泄漏。

* fix(packaging): package-linux.sh 的版本解析与 build.sh 统一

上一版只修了 build.sh,漏了 package-linux.sh —— 它仍只看 git describe:

  VERSION="${VERSION:-$(git describe --tags --dirty)}"

无 tag 分支上 git describe 返回**分支名**
(backup/pre-identity-rewrite-20261005-56-g98850e1),守卫会正确拒绝,
但要求人**手动传 VERSION** 才能打包。而 build.sh 早已能自动回退 meta.Version
⇒ 同一个版本,走 build.sh 打得出来、走 package-linux.sh 打不出来。

危害不只是"要手动传参":运维为了跑通会去掉守卫,而分支名一旦进
meta.Version 就是**插件协议主版本**的来源 ⇒ 22 个子进程插件全部握手失败
(这是真实发生过的事故,见 3efa4e5)。

修法:两个脚本用同一套规则(git describe 仅在 vX.Y.Z 形态时采用,
否则回退 meta.Version),并对最终值做形态守卫。

判据:TestBothPackagingScriptsShareVersionRules —— 逐个读两个脚本,
要求都能从 meta/meta.go 取权威版本、都校验 git describe 形态。
变异(把 package-linux.sh 改回旧规则)咬住并准确指出是哪个脚本。

实测:不传 VERSION 跑 package-linux.sh,输出
  [PKG] warn: git describe 给出非版本串(...),改用 meta.Version=1.4.0
  Version: 1.4.0

* fix(test): 跨仓判据在 CI 上应 skip 而非 fail(修 PR 双红)

CI 上 "Go build / vet / test" 与 "Build linux/amd64" 双红:

  --- FAIL: TestHmapdevTemplateImplementsAllPluginMgrMethods
      bridge_sync_test.go:170: 读模板: open
      ../../../third_party/homeagent-sdk/tools/hmapdev/templates/proc_main.go.tmpl:
      no such file or directory

## 根因:判据依赖了一个**被 gitignore 的路径**

`.gitignore:33` 有 `third_party/homeagent-sdk/tools/`(SDK 仓自治,本仓不跟踪)。
于是:
  · 本机(完整 SDK 检出)有该文件 ⇒ 判据通过
  · CI(只 checkout 本仓)没有该文件 ⇒ Fatalf ⇒ **必红**

我提交时只在本机跑测试、没看 CI,所以这个问题拖到 PR 才暴露。

## 修法

文件缺失时 `t.Skipf` 而不是 `t.Fatalf`。

★ 这不是"放弃检查":
  · 跨仓一致性由 **SDK 仓自己的判据**守着
    (tools/hmapdev/scaffold_stdout_test.go 等),那里才是模板的真实归属地
  · 本判据在本机开发时**仍然生效**(已验证:有文件时 PASS,缺文件时 SKIP)

## 验证

 · 本机(有文件)  → PASS(判据真跑)
 · 模拟 CI(移走 tools/)→ SKIP(不再 fail)
 · 全量 53 包 ok / 0 fail

## 顺带

同目录还有 e2e_template_test.go 也读 tools/,但它是 `//go:build` 门控的
e2e 判据,CI 默认不跑,无需处理。

---------

Co-authored-by: JianFeeeee <JianFeeeee@users.noreply.gitcode.com>
2026-10-17 19:18:24 +08:00

897 lines
36 KiB
Bash
Raw Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

#!/usr/bin/env bash
set -euo pipefail
PROJECT_ROOT="$(cd "$(dirname "$0")/../.." && pwd)"
BUILD_DIR="${PROJECT_ROOT}/build"
DIST_DIR="${PROJECT_ROOT}/dist/linux"
# ★★ VERSION 解析必须与 build.sh 用**同一套规则**(2026-10-17 统一)。
#
# 原实现只看 git describe,而无 tag 分支上它返回**分支名**:
# backup/pre-identity-rewrite-20261005-56-g98850e1
# 下面的守卫会拒绝它(正确),但要求人**手动传 VERSION** 才能打包 ——
# 而 build.sh 早已能自动回退到 meta.Version。两个脚本行为不一致的结果是:
# 同一个版本,走 build.sh 打得出来,走 package-linux.sh 打不出来。
#
# 更要紧的是:若有人为了"让它跑通"去掉守卫,分支名就会进 meta.Version,
# 而那是**插件协议主版本**的来源 ⇒ 22 个子进程插件全部握手失败(真事故)。
#
# ⇒ 统一为:git describe 只在形如 vX.Y.Z 时采用,否则回退 meta.Version。
_meta_version() {
local mv
mv="$(sed -n 's/^[[:space:]]*Version[[:space:]]*=[[:space:]]*"\([^"]*\)".*/\1/p' \
"$PROJECT_ROOT/internal/meta/meta.go" 2>/dev/null | head -1)"
case "$mv" in
[0-9]*.[0-9]*.[0-9]*) printf '%s' "$mv"; return 0 ;;
esac
return 1
}
if [ -z "${VERSION:-}" ]; then
_gd="$(git -C "$PROJECT_ROOT" describe --tags --dirty 2>/dev/null || true)"
case "$_gd" in
v[0-9]*.[0-9]*.[0-9]*) VERSION="${_gd#v}" ;;
*)
if VERSION="$(_meta_version)"; then
[ -n "$_gd" ] && echo "[PKG] warn: git describe 给出非版本串($_gd),改用 meta.Version=$VERSION" >&2
else
echo "[PKG] ERROR: 拿不到合法版本号(git describe=$_gd)。请显式传 VERSION=x.y.z。" >&2
exit 1
fi
;;
esac
fi
# git describe 给出的是 v1.0.0-68-gba0b5a1-dirty 这类描述串,它不是合法的包版本:
# deb 的 Version 必须以数字开头,rpm 的 Version 不允许 '-'(那是版本/发布的分隔符)。
# 以前只有显式传 VERSION=1.0.3 才打得出来,默认路径一跑就死在 dpkg-deb 上——
# 而且死在 stage 之后,前面每条日志都是真的,只有最后一个产物没生成。
PKG_VERSION="${VERSION#v}"
case "$PKG_VERSION" in
[0-9]*) ;;
*) echo "ERROR: 包版本必须以数字开头(得到 '$VERSION')。请显式设置 VERSION=x.y.z 后重试。" >&2; exit 1 ;;
esac
PKG_VERSION="$(printf '%s' "$PKG_VERSION" | sed -e 's/-/+/g')"
PACKAGE_ROOT="${PROJECT_ROOT}/deploy/packaging/linux"
GO="${GO:-$(command -v go 2>/dev/null || echo "go")}"
ARCH="${1:-amd64}" # amd64 or arm64
# server/full 发行包默认带 Chinese-CLIP ONNX 产物与 ONNX Runtime。
# 二进制大资产不进 git:发布环境通过这两个目录提供已验证的产物;若缺失,
# server/full 打包必须明确失败,不能生成一个「默认启用但装完不能用」的假包。
CHINESECLIP_BUNDLE_DIR="${CHINESECLIP_BUNDLE_DIR:-$BUILD_DIR/model-assets/chinese-clip-vit-b16-onnx}"
ONNXRUNTIME_ASSET_DIR="${ONNXRUNTIME_ASSET_DIR:-$BUILD_DIR/runtime-assets/$ARCH}"
ONNXRUNTIME_LIB="${ONNXRUNTIME_LIB:-$ONNXRUNTIME_ASSET_DIR/libonnxruntime.so}"
ONNXRUNTIME_LICENSE="${ONNXRUNTIME_LICENSE:-$ONNXRUNTIME_ASSET_DIR/LICENSE}"
ONNXRUNTIME_NOTICES="${ONNXRUNTIME_NOTICES:-$ONNXRUNTIME_ASSET_DIR/ThirdPartyNotices.txt}"
# 打包 staging 会把 719MB 模型真的复制一份,临时目录必须落在构建目录所在的磁盘,
# 不能落在系统临时目录:本机 /tmp 是 9.8GB tmpfs,一次 full 包 staging 就能写满,
# 而且失败发生在 cp 进行到一半,报出来是 "No space left on device"——看上去像
# 资产/版本有问题,实际只是临时目录选错了文件系统。
STAGE_TMP="${BUILD_DIR}/.stage-tmp"
# electron 官方发布物用 x64/arm64 命名,而 Debian 用 amd64/arm64。
# 两者在 arm64 上恰好同名,amd64 上不同——此前缓存查找统一用 TAR_ARCH
# (amd64),于是 electron-v*-linux-x64.zip 永远命中不到,amd64 GUI 只能
# 靠"回退到 host node_modules"这条路组装。干净 worktree 里没有完整
# node_modules,GUI 就被静默跳过。故单独映射。
ACTION="${2:-all}" # all, build, deb, tar, rpm
DEB_ARCH="$ARCH"
RPM_ARCH="$ARCH"
TAR_ARCH="$ARCH"
case "$ARCH" in
amd64) DEB_ARCH="amd64"; RPM_ARCH="x86_64"; TAR_ARCH="amd64"; ELECTRON_ARCH="x64" ;;
arm64) DEB_ARCH="arm64"; RPM_ARCH="aarch64"; TAR_ARCH="arm64"; ELECTRON_ARCH="arm64" ;;
*) echo "Unknown arch: $ARCH (use amd64 or arm64)"; exit 1 ;;
esac
echo "=== HomeAgent Linux Packager ==="
echo "Version: $VERSION"
[ "$PKG_VERSION" = "$VERSION" ] || echo "Package: $PKG_VERSION (normalized for deb/rpm)"
echo "Arch: $ARCH"
echo ""
# ---- prepare go.mod for Linux build: point SDK replace at the in-repo SDK ----
# SDK 仓库(含工具链/示例)以嵌套 git 仓库形式维护在 third_party/homeagent-sdk,
# 打包直接用该目录,不再 clone 到 /tmp。
prepare_gomod() {
local gomod="$PROJECT_ROOT/go.mod"
local sdk_local="$PROJECT_ROOT/third_party/homeagent-sdk"
local sdk_clone="/tmp/homeagent-sdk"
local patched=0
if grep -q 'replace github.com/JianFeeeee/homeagentsdk' "$gomod"; then
echo ">>> Updating go.mod: replacing Windows SDK path with in-repo SDK..."
if [ -d "$sdk_local" ]; then
sed -i.bak "s|^replace github.com/JianFeeeee/homeagentsdk => .*|replace github.com/JianFeeeee/homeagentsdk => ${sdk_local}|" "$gomod"
patched=1
elif [ ! -d "$sdk_clone" ]; then
echo ">>> Cloning SDK to $sdk_clone (in-repo SDK missing)..."
git clone git@github.com:JianFeeeee/homeagentsdk.git "$sdk_clone" 2>/dev/null || \
git clone https://github.com/JianFeeeee/homeagentsdk.git "$sdk_clone" 2>/dev/null || true
if [ -d "$sdk_clone" ]; then
sed -i.bak "s|^replace github.com/JianFeeeee/homeagentsdk => .*|replace github.com/JianFeeeee/homeagentsdk => ${sdk_clone}|" "$gomod"
patched=1
else
echo "WARNING: Cannot clone SDK. Build may fail."
fi
else
sed -i.bak "s|^replace github.com/JianFeeeee/homeagentsdk => .*|replace github.com/JianFeeeee/homeagentsdk => ${sdk_clone}|" "$gomod"
patched=1
fi
fi
return $patched
}
restore_gomod() {
if [ -f "$PROJECT_ROOT/go.mod.bak" ]; then
echo ">>> Restoring original go.mod..."
mv "$PROJECT_ROOT/go.mod.bak" "$PROJECT_ROOT/go.mod"
fi
}
# ---- temporarily move .syso files (Windows COFF objects) for Linux builds ----
hide_syso() {
echo ">>> Hiding .syso files (Windows-only object files)..."
for dir in homed waiter; do
if [ -f "$PROJECT_ROOT/cmd/$dir/$dir.syso" ]; then
mv "$PROJECT_ROOT/cmd/$dir/$dir.syso" "$PROJECT_ROOT/cmd/$dir/$dir.syso.bak"
fi
done
}
restore_syso() {
for dir in homed waiter; do
if [ -f "$PROJECT_ROOT/cmd/$dir/$dir.syso.bak" ]; then
mv "$PROJECT_ROOT/cmd/$dir/$dir.syso.bak" "$PROJECT_ROOT/cmd/$dir/$dir.syso"
fi
done
}
# ensure both are always restored on exit
restore_all() { restore_gomod; restore_syso; rmdir "$STAGE_TMP" 2>/dev/null || true; }
trap restore_all EXIT
# ---- build Go binaries via existing build.sh ----
build_go() {
echo ">>> Building Go binaries for linux/$ARCH..."
prepare_gomod || true
hide_syso
local suffix="linux_${ARCH}"
local homed_bin="$BUILD_DIR/homed_$suffix"
local waiter_bin="$BUILD_DIR/waiter_$suffix"
local initconfig_bin="$BUILD_DIR/initconfig_$suffix"
# 先删旧产物:否则本次构建失败后,残留文件会让「产物存在」判据假绿。
rm -f "$homed_bin" "$waiter_bin" "$initconfig_bin"
bash "$PROJECT_ROOT/deploy/packaging/build.sh" "linux/$ARCH" "homed"
test -x "$homed_bin"
if ! go version -m "$homed_bin" | grep -Eq 'build[[:space:]]+-tags=.*onnxruntime'; then
echo "ERROR: homed 不是 onnxruntime 构建,拒绝打 server/full 包:$homed_bin" >&2
return 1
fi
bash "$PROJECT_ROOT/deploy/packaging/build.sh" "linux/$ARCH" "waiter"
test -x "$waiter_bin"
bash "$PROJECT_ROOT/deploy/packaging/build.sh" "linux/$ARCH" "initconfig"
test -x "$initconfig_bin"
echo " homed: $homed_bin ($(du -h "$homed_bin" | cut -f1), onnxruntime)"
echo " waiter: $waiter_bin ($(du -h "$waiter_bin" | cut -f1))"
echo " initconfig: $initconfig_bin ($(du -h "$initconfig_bin" | cut -f1))"
echo ""
}
# ---- build GUI (manual directory assembly, avoids electron-packager network issues) ----
#
# electron 运行时必须按**目标架构**取,不能用 host 的
# node_modules/electron/dist——那里永远是 host 架构(本机 x64)。
# v1.0.0 / v1.0.1 的 arm64 full/client 包都踩了这个坑:目录名带
# -arm64、homed/waiter 确实是 aarch64,但里面的 electron 是 x86-64,
# 在 arm64 机器上一启动就是 Exec format error(从未被交叉验证过)。
#
# 现在改为优先从 electron 缓存里取对应架构的 zip,并在最后做
# 一道强制校验:架构不符就删掉目录并跳过 GUI,宁可不发也不发坏包。
build_gui() {
local gui_dir="$PROJECT_ROOT/cmd/gui"
local gui_out="$BUILD_DIR/homeagent-gui-linux-${TAR_ARCH}"
if [ -d "$gui_out" ]; then
echo ">>> GUI already built at $gui_out (delete to rebuild)"
return
fi
echo ">>> Building GUI directory for linux/$ARCH..."
# 判据是 electron 包本身在不在,而不是 node_modules 目录在不在。
#
# npm install 失败(离线、网络受限)会留下一个只有一两个条目的空壳
# node_modules,目录存在但 electron 缺失。只看目录会以为"已安装",
# 于是 ever 读不到版本、缓存匹配退化、最后走到"host dist 也没有"而
# 静默跳过 GUI——包名和目录名全都正确,只是没有 GUI,没有任何一步报错。
if [ ! -f "$gui_dir/node_modules/electron/package.json" ]; then
if [ -d "$gui_dir/node_modules" ]; then
echo " node_modules 存在但 electron 缺失(疑似上次 npm install 未完成)"
fi
echo " npm install..."
if ! (cd "$gui_dir" && npm install --production); then
echo " WARNING: npm install 失败——离线环境下这是预期的。"
echo " GUI 需要 cmd/gui/node_modules/electron 或 ~/.cache/electron 缓存。"
fi
fi
# electron 版本优先从已安装的包里读,保证运行时与 app 依赖一致。
# 读不到时退而从 package.json 的依赖声明里取数字部分(它可能写成
# "^33.0.0" 这类范围,只用于给缓存匹配一个提示,匹配不上仍会走通配)。
local ever
ever=$(python3 -c "import json;print(json.load(open('$gui_dir/node_modules/electron/package.json'))['version'])" 2>/dev/null || true)
if [ -z "$ever" ]; then
ever=$(python3 -c "
import json, re
d = json.load(open('$gui_dir/package.json'))
spec = (d.get('devDependencies', {}) or {}).get('electron') or (d.get('dependencies', {}) or {}).get('electron') or ''
m = re.search(r'(\\d+(?:\\.\\d+)*)', spec)
print(m.group(1) if m else '')
" 2>/dev/null || true)
if [ -n "$ever" ]; then
echo " electron 版本取自 package.json 依赖声明: $ever(非精确)"
fi
fi
mkdir -p "$gui_out"
# 优先:缓存里的目标架构 zip(~/.cache/electron/<hash>/electron-v<ver>-linux-<arch>.zip)
#
# ★ 这里必须 `|| true`:`find` 对**不存在的目录**返回退出码 1,而本脚本是
# `set -euo pipefail`,命令替换里的失败会让整个脚本当场退出。
# 后果:任何**没有 ~/.cache/electron 的机器**(全新克隆、CI runner、
# 其他开发机)跑到这里就死,且只留下一行「electron 版本取自 package.json」
# 作为最后的输出,看不出真因。实测(2026-09-29,GitHub runner 与本地
# 移走缓存后均复现):build_go 全部成功,然后卡在这里静默退出。
# 本机历史上之所以一直「能打包」,只是因为碰巧有那份缓存。
local zip=""
if [ -n "$ever" ]; then
zip=$(find "$HOME/.cache/electron" -name "electron-v${ever}-linux-${ELECTRON_ARCH}.zip" 2>/dev/null | head -1 || true)
fi
if [ -z "$zip" ]; then
zip=$(find "$HOME/.cache/electron" -name "electron-v*-linux-${ELECTRON_ARCH}.zip" 2>/dev/null | head -1 || true)
fi
if [ -n "$zip" ]; then
echo " electron runtime: $(basename "$zip")"
unzip -q -o "$zip" -d "$gui_out"
else
# 回退:仅当目标架构 == host 架构时才能用 host 的 dist
local host_arch
case "$(uname -m)" in
x86_64) host_arch=amd64 ;;
aarch64|arm64) host_arch=arm64 ;;
*) host_arch=unknown ;;
esac
if [ "$TAR_ARCH" != "$host_arch" ]; then
echo " WARNING: 缺 electron-v*-linux-${ELECTRON_ARCH}.zip 缓存,且目标架构与 host"
echo " ($host_arch) 不同——不能用 host 的 electron 冒充。跳过 GUI。"
echo " 解法:下载 electron-v${ever:-<ver>}-linux-${ELECTRON_ARCH}.zip 到"
echo " ~/.cache/electron/<任意子目录>/ 后重跑。"
rm -rf "$gui_out"
return
fi
local electron_dir="$gui_dir/node_modules/electron/dist"
if [ ! -f "$electron_dir/electron" ]; then
echo " WARNING: electron binary not found at $electron_dir. GUI will be skipped."
rm -rf "$gui_out"
return
fi
echo " electron runtime: host node_modules (同架构 $host_arch)"
cp -r "$electron_dir"/* "$gui_out/" 2>/dev/null
fi
mkdir -p "$gui_out/resources/app/node_modules"
mkdir -p "$gui_out/resources/app/renderer"
rm -f "$gui_out/resources/default_app.asar" 2>/dev/null
# copy app source
#
# ★ 改成**按目录整体同步**,不再逐文件手列。
#
# 原实现手工 cp 七个文件,每加一个新文件就要记得补一行,漏了不会有
# 任何报错,只会做出一个"装完缺东西"的包。已实际漏掉过:
# · renderer/mascot.svg —— 源码里根本不存在(真名 mascot.webp)
# · icon.svg / icon.ico / icon-tray*.png —— 托盘与窗口图标
# · renderer/icon.svg
# · renderer/vendor/* —— three/OrbitControls/marked/purify
# (这四个是本地化后的第三方库,缺了 Linux 发行版的星图直接坏)
#
# 整体同步还顺带保证 .hmap 之外的任何新增资源都会进包。
cp "$gui_dir/main.js" "$gui_out/resources/app/"
cp "$gui_dir/preload.js" "$gui_out/resources/app/"
cp "$gui_dir/package.json" "$gui_out/resources/app/"
cp "$gui_dir"/*.svg "$gui_out/resources/app/" 2>/dev/null || true
cp "$gui_dir"/*.ico "$gui_out/resources/app/" 2>/dev/null || true
cp "$gui_dir"/icon-tray*.png "$gui_out/resources/app/" 2>/dev/null || true
# renderer 整目录同步(含 vendor/ 与全部静态资源)
cp -r "$gui_dir/renderer/." "$gui_out/resources/app/renderer/"
# ★ 必备文件校验:缺了就是"装完坏掉"的包,且不会有任何构建报错。
# renderer/vendor 是本地化后的第三方库(three/OrbitControls/marked/purify),
# 缺任何一份都会让星图或 Markdown 渲染在断网/离线环境下直接坏掉。
local missing=""
for _f in index.html app.js style.css vendor/three.min.js vendor/OrbitControls.js \
vendor/marked.min.js vendor/purify.min.js; do
[ -f "$gui_out/resources/app/renderer/$_f" ] || missing="$_f "
done
[ -f "$gui_out/resources/app/icon.svg" ] || missing="icon.svg "
if [ -n "$missing" ]; then
echo " ERROR: GUI 资源缺失:$missing"
echo " 宁可不发,也不发装了跑不起来的包。"
rm -rf "$gui_out"
return
fi
# production node_modules for app
if [ -d "$gui_dir/node_modules" ]; then
for mod in icojs; do
if [ -d "$gui_dir/node_modules/$mod" ]; then
cp -r "$gui_dir/node_modules/$mod" "$gui_out/resources/app/node_modules/"
fi
done
fi
# create desktop entry and symlink
cat > "$gui_out/resources/app/homeagent-gui.desktop" << DESKTOP
[Desktop Entry]
Name=HomeAgent
Comment=HomeAgent Desktop GUI
Exec=$gui_out/homeagent-gui
Terminal=false
Type=Application
Categories=Utility;
Icon=$gui_out/resources/app/icon.svg
DESKTOP
# create launcher script
cat > "$gui_out/homeagent-gui" << 'LAUNCHER'
#!/bin/sh
DIR="$(cd "$(dirname "$0")" && pwd)"
exec "$DIR/electron" "$DIR/resources/app" "$@"
LAUNCHER
chmod +x "$gui_out/homeagent-gui"
chmod +x "$gui_out/electron"
# 最后一道强制校验:electron 二进制的实际架构必须匹配目标架构。
# 不做这步就会重现 v1.0.0/v1.0.1 的隐形坏包:包名、目录名、
# homed/waiter 全对,只有 electron 是错架构,直到用户在 arm64 机器上
# 双击才发现 Exec format error。
local want_pat
case "$TAR_ARCH" in
amd64) want_pat="x86-64" ;;
arm64) want_pat="aarch64" ;;
*) want_pat="" ;;
esac
if [ -n "$want_pat" ]; then
local got
got=$(file -b "$gui_out/electron" 2>/dev/null || echo "")
if ! printf '%s' "$got" | grep -q "$want_pat"; then
echo " ERROR: electron 架构不符——期望 $want_pat,实际: ${got%%,*}"
echo " 删除 GUI 目录并跳过(宁可不发,也不发装了跑不起来的包)。"
rm -rf "$gui_out"
return
fi
fi
echo " GUI built: $gui_out ($(du -sh "$gui_out" | cut -f1), $(file -b "$gui_out/electron" | cut -d, -f2 | tr -d ' '))"
echo ""
}
# ---- stage files for a variant ----
stage_variant() {
local variant="$1"
local staging="$2"
local suffix="linux_${ARCH}"
echo ">>> Staging $variant..."
mkdir -p "$staging/usr/bin"
mkdir -p "$staging/etc/systemd/system"
mkdir -p "$staging/var/lib/homeagent"
local initconfig_bin="$BUILD_DIR/initconfig_$suffix"
# ★★ 启动依赖图必须与 homed **同目录**(2026-10-17 实测缺陷)。
#
# 内核启动时会比对
# 「磁盘图的 digest」 vs 「二进制内嵌的 digest」
# 读不到图 ⇒ 生产构建(内嵌 hash 非空)**直接拒绝启动**:
# kernelboot: 生产构建(内嵌 hash 非空)但读不到依赖图
# <dir>/depgraph.bin: no such file or directory
# ⇒ 退出码 1,内核完全起不来。
#
# 原先 stage_variant 只拷 homed/waiter ⇒ **每个 deb/tar 包里的 homed
# 都起不来**。实测确认:只放 homed 不放图,启动即失败退出 1。
#
# 图由 `make build` 产出到 build/depgraph.bin(与 homed 同一次构建,
# digest 一一对应;手工复制别的版本的图会 digest 不符,同样拒绝启动)。
stage_depgraph() {
local staging="$1"
if [ -f "$BUILD_DIR/depgraph.bin" ]; then
cp "$BUILD_DIR/depgraph.bin" "$staging/usr/bin/depgraph.bin"
chmod 644 "$staging/usr/bin/depgraph.bin"
else
echo "ERROR: 缺少 $BUILD_DIR/depgraph.bin —— 内核与依赖图必须来自同一次构建。" >&2
echo " 先跑 make build(它会生成图并注入 digest),再打包。" >&2
return 1
fi
}
# stage_hmapdev:插件工具链(init/build/package .hmap)
#
# 放在 /usr/bin 而不是 /usr/lib:它是使用者直接调用的命令,
# 与 homed/waiter 同级更符合直觉(hmapdev 自身按 PATH 找 SDK store)。
stage_hmapdev() {
local staging="$1"
local hv
hv="$(ls "$BUILD_DIR"/hmapdev* 2>/dev/null | grep -vE '\.(exe)$' | head -1)"
if [ -z "$hv" ]; then
# 退而求其次:从 SDK 源码现编一个(CGO 无关,快)
echo " [hmapdev] build/ 下没有 hmapdev,尝试现编..."
( cd "$PROJECT_ROOT/third_party/homeagent-sdk/tools/hmapdev" && \
CGO_ENABLED=0 go build -trimpath -o "$BUILD_DIR/hmapdev" . ) || {
echo "ERROR: 无法获得 hmapdev(插件工具链),拒绝打包。" >&2
return 1; }
hv="$BUILD_DIR/hmapdev"
fi
install -m 0755 "$hv" "$staging/usr/bin/hmapdev"
echo " hmapdev: $hv ($(du -h "$hv" | cut -f1))"
}
# stage_sdk:与本次工具链同源的 SDK 源码(含 sdk/ tools/ skills/ meta/)。
#
# 为什么要随包发:hmapdev 需要按插件声明的 sdk 版本解析依赖;没有本地 SDK
# store 时它会去网上拉(内网/受限出口环境直接失败)。随包带一份 →
# 「装完即可离线构建插件」,且版本与内核严格同源(防版本漂移)。
stage_sdk() {
local staging="$1"
local sdk_src="$PROJECT_ROOT/third_party/homeagent-sdk"
local sdk_dst="$staging/usr/lib/homeagent/sdk"
[ -d "$sdk_src" ] || { echo "ERROR: 找不到 SDK 源码 $sdk_src" >&2; return 1; }
mkdir -p "$sdk_dst"
for item in go.mod go.sum LICENSE README.md sdk meta tools skills; do
[ -e "$sdk_src/$item" ] && cp -a "$sdk_src/$item" "$sdk_dst/"
done
# 不带陈旧工具链产物:那是上次构建留下的,可能与本次源码不符。
rm -rf "$sdk_dst/tools/vscode-hmapdev/node_modules" 2>/dev/null || true
local sv
sv="$(grep -oE 'Version = "[0-9.]+"' "$sdk_src/meta/meta.go" 2>/dev/null | head -1 | grep -oE '[0-9.]+')"
echo " sdk: $sdk_dst (v${sv:-?}, $(du -sh "$sdk_dst" | cut -f1))"
}
case "$variant" in
full)
cp "$BUILD_DIR/homed_$suffix" "$staging/usr/bin/homed"
cp "$BUILD_DIR/waiter_$suffix" "$staging/usr/bin/waiter"
stage_depgraph "$staging"
cp "$PROJECT_ROOT/deploy/homeagent.service" "$staging/etc/systemd/system/homeagent.service"
[ -f "$initconfig_bin" ] && cp "$initconfig_bin" "$staging/usr/bin/initconfig"
stage_setup "$staging"
stage_license "$staging"
stage_multimodal_assets "$staging"
stage_gui "$staging"
;;
server)
cp "$BUILD_DIR/homed_$suffix" "$staging/usr/bin/homed"
cp "$BUILD_DIR/waiter_$suffix" "$staging/usr/bin/waiter"
stage_depgraph "$staging"
cp "$PROJECT_ROOT/deploy/homeagent.service" "$staging/etc/systemd/system/homeagent.service"
[ -f "$initconfig_bin" ] && cp "$initconfig_bin" "$staging/usr/bin/initconfig"
stage_setup "$staging"
stage_license "$staging"
stage_multimodal_assets "$staging"
;;
toolchain)
# 内核 + CLI + 依赖图 + 插件工具链 + SDK(开发者/自建一体化)
cp "$BUILD_DIR/homed_$suffix" "$staging/usr/bin/homed"
cp "$BUILD_DIR/waiter_$suffix" "$staging/usr/bin/waiter"
stage_depgraph "$staging"
stage_hmapdev "$staging"
stage_sdk "$staging"
cp "$PROJECT_ROOT/deploy/homeagent.service" "$staging/etc/systemd/system/homeagent.service"
[ -f "$initconfig_bin" ] && cp "$initconfig_bin" "$staging/usr/bin/initconfig"
stage_setup "$staging"
stage_license "$staging"
# ★★ onnx 运行库 + 模型必须进包(用户明确要求,2026-10-17)。
# 否则装到干净机器上「默认启用但用不了」——本地向量空间会静默降级。
stage_multimodal_assets "$staging"
;;
client)
cp "$BUILD_DIR/waiter_$suffix" "$staging/usr/bin/waiter"
stage_license "$staging"
stage_gui "$staging"
;;
esac
chmod 755 "$staging/usr/bin/"* 2>/dev/null || true
echo " OK"
}
stage_gui() {
local staging="$1"
local gui_src="$BUILD_DIR/homeagent-gui-linux-${TAR_ARCH}"
if [ -d "$gui_src" ]; then
mkdir -p "$staging/usr/lib/homeagent-gui"
cp -r "$gui_src"/* "$staging/usr/lib/homeagent-gui/"
cat > "$staging/usr/bin/homeagent-gui" << 'SCRIPT'
#!/bin/sh
exec /usr/lib/homeagent-gui/homeagent-gui "$@"
SCRIPT
chmod 755 "$staging/usr/bin/homeagent-gui"
else
echo " WARNING: GUI not built, skipping GUI staging"
fi
}
stage_setup() {
local staging="$1"
local setup_src="$PROJECT_ROOT/deploy/packaging/linux/setup.sh"
if [ -f "$setup_src" ]; then
mkdir -p "$staging/usr/lib/homeagent"
cp "$setup_src" "$staging/usr/lib/homeagent/setup.sh"
chmod 755 "$staging/usr/lib/homeagent/setup.sh"
fi
}
# 项目自身的许可:**所有变体**都要带(client 也分发 waiter 与 GUI)。
#
# deb 按 Debian 惯例给 /usr/share/doc/homeagent/copyright(DEP-5 机器可读格式),
# 同时把 LICENSE 全文放进去;rpm 的许可走 fpm 的 --license 元数据。
# 与 stage_multimodal_assets 的 licenses/ 分工:那里放**第三方**(模型/运行库)的
# 许可全文,这里放本项目自己的。
stage_license() {
local staging="$1"
local docdir="$staging/usr/share/doc/homeagent"
mkdir -p "$docdir"
cp "$PROJECT_ROOT/LICENSE" "$docdir/LICENSE"
cat > "$docdir/copyright" <<'EOF'
Format: https://www.debian.org/doc/packaging-manuals/copyright-format/1.0/
Upstream-Name: HomeAgent
Source: https://github.com/JianFeeeee/HomeAgent
Files: *
Copyright: HomeAgent contributors
License: AGPL-3.0-only
This program is free software: you can redistribute it and/or modify it under
the terms of the GNU Affero General Public License as published by the Free
Software Foundation, version 3 of the License.
.
This program is distributed in the hope that it will be useful, but WITHOUT
ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
FOR A PARTICULAR PURPOSE. See the GNU Affero General Public License for more
details.
.
You should have received a copy of the GNU Affero General Public License along
with this program. If not, see <https://www.gnu.org/licenses/>.
.
The license is AGPL-3.0-only: no later version may be chosen. Note the network
clause (§13 Remote Network Interaction) — offering modified versions of this
software to users over a network also requires offering them the source.
.
Full text: /usr/share/doc/homeagent/LICENSE
Files: usr/lib/homeagent/models/chinese-clip-vit-b16-onnx/*
Copyright: OFA-Sys / Chinese-CLIP authors
License: Apache-2.0
Full text: /usr/share/doc/homeagent/licenses/Chinese-CLIP-Apache-2.0.txt
Comment: pre-trained model artifacts; NOT covered by this package's AGPL grant
Files: usr/lib/homeagent/onnxruntime/*
Copyright: Microsoft Corporation
License: MIT
Full text: /usr/share/doc/homeagent/licenses/ONNX-Runtime-MIT.txt
Comment: license texts and third-party notices under licenses/ONNX-Runtime-*
EOF
chmod 644 "$docdir/LICENSE" "$docdir/copyright"
}
# server/full 的 ONNX 资产。模型与运行库是发行版能力的一部分,不是可选下载:
# 只要打 server/full 包,两者缺一就失败。client 包不运行 homed,故不携带。
stage_multimodal_assets() {
local staging="$1"
local model_dst="$staging/usr/lib/homeagent/models/chinese-clip-vit-b16-onnx"
local ort_dst="$staging/usr/lib/homeagent/onnxruntime"
local licenses="$staging/usr/share/doc/homeagent/licenses"
if [ ! -d "$CHINESECLIP_BUNDLE_DIR" ]; then
echo "ERROR: Chinese-CLIP 产物目录不存在:$CHINESECLIP_BUNDLE_DIR" >&2
echo "先运行 scripts/export_chineseclip_onnx.py,再通过 CHINESECLIP_BUNDLE_DIR 指向产物。" >&2
return 1
fi
for f in TextEncoder.onnx VisionEncoder.onnx embed_config.json vocab.txt reference.json SHA256SUMS; do
if [ ! -s "$CHINESECLIP_BUNDLE_DIR/$f" ]; then
echo "ERROR: Chinese-CLIP 产物缺少或为空:$CHINESECLIP_BUNDLE_DIR/$f" >&2
return 1
fi
done
if ! (cd "$CHINESECLIP_BUNDLE_DIR" && sha256sum -c SHA256SUMS); then
echo "ERROR: Chinese-CLIP SHA256SUMS 校验失败,拒绝打包。" >&2
return 1
fi
if [ ! -s "$ONNXRUNTIME_LIB" ]; then
echo "ERROR: ONNX Runtime 不存在:$ONNXRUNTIME_LIB" >&2
echo "通过 ONNXRUNTIME_ASSET_DIR 或 ONNXRUNTIME_LIB 指向与目标架构匹配的资产。" >&2
return 1
fi
for notice in "$ONNXRUNTIME_LICENSE" "$ONNXRUNTIME_NOTICES"; do
if [ ! -s "$notice" ]; then
echo "ERROR: ONNX Runtime 许可证资产缺失:$notice" >&2
return 1
fi
done
local runtime_desc
runtime_desc=$(file -b "$ONNXRUNTIME_LIB")
case "$ARCH" in
amd64) printf '%s' "$runtime_desc" | grep -qE 'x86-64|x86_64' || {
echo "ERROR: ONNX Runtime 架构不是 amd64:$runtime_desc" >&2; return 1; } ;;
arm64) printf '%s' "$runtime_desc" | grep -qE 'aarch64|ARM aarch64' || {
echo "ERROR: ONNX Runtime 架构不是 arm64:$runtime_desc" >&2; return 1; } ;;
esac
mkdir -p "$model_dst" "$ort_dst" "$licenses"
cp -a "$CHINESECLIP_BUNDLE_DIR/." "$model_dst/"
install -m 0755 "$ONNXRUNTIME_LIB" "$ort_dst/libonnxruntime.so"
# 许可证随二进制分发:Chinese-CLIP = Apache-2.0;ONNX Runtime = MIT,
# 同时携带其 ThirdPartyNotices(含 MKL/protobuf/zlib 等第三方条款)。
cp /usr/share/common-licenses/Apache-2.0 "$licenses/Chinese-CLIP-Apache-2.0.txt"
cp "$ONNXRUNTIME_LICENSE" "$licenses/ONNX-Runtime-MIT.txt"
cp "$ONNXRUNTIME_NOTICES" "$licenses/ONNX-Runtime-ThirdPartyNotices.txt"
cat > "$licenses/MODEL-SOURCES.txt" <<EOF
Chinese-CLIP ViT-B/16
upstream: https://huggingface.co/OFA-Sys/chinese-clip-vit-base-patch16
license: Apache-2.0
exported-by: scripts/export_chineseclip_onnx.py
dimensions: 512
modalities: text,image
ONNX Runtime
upstream: https://github.com/microsoft/onnxruntime
license: MIT (see ONNX-Runtime-MIT.txt and ONNX-Runtime-ThirdPartyNotices.txt)
EOF
echo " ONNX assets: model=$(du -sh "$model_dst" | cut -f1) runtime=$(du -h "$ort_dst/libonnxruntime.so" | cut -f1)"
}
# ---- create .deb ----
build_deb() {
local variant="$1"
local staging="$2"
local deb_dir="${DIST_DIR}/deb"
mkdir -p "$deb_dir"
local pkg_name="homeagent-${variant}_${PKG_VERSION}_${DEB_ARCH}.deb"
local deb_root
deb_root="$(mktemp -d "$STAGE_TMP/deb.XXXXXX")"
mkdir -p "$deb_root/DEBIAN"
local control_file="$PACKAGE_ROOT/deb/control-${variant}"
local installed_size_kb
installed_size_kb=$(du -sk "$staging" | cut -f1)
sed -e "s/VERSION_PLACEHOLDER/$PKG_VERSION/g" \
-e "s/ARCH_PLACEHOLDER/$DEB_ARCH/g" \
-e "s/INSTALLED_SIZE_PLACEHOLDER/$installed_size_kb/g" \
"$control_file" > "$deb_root/DEBIAN/control"
if [ -f "$PACKAGE_ROOT/deb/postinst" ]; then
cp "$PACKAGE_ROOT/deb/postinst" "$deb_root/DEBIAN/postinst"
chmod 755 "$deb_root/DEBIAN/postinst"
fi
if [ -f "$PACKAGE_ROOT/deb/prerm" ]; then
cp "$PACKAGE_ROOT/deb/prerm" "$deb_root/DEBIAN/prerm"
chmod 755 "$deb_root/DEBIAN/prerm"
fi
cp -r "$staging"/* "$deb_root/" 2>/dev/null || true
echo ">>> Building .deb: $pkg_name"
fakeroot dpkg-deb --build "$deb_root" "$deb_dir/$pkg_name" 2>/dev/null || \
dpkg-deb --build "$deb_root" "$deb_dir/$pkg_name" 2>&1
rm -rf "$deb_root"
echo " Created: $deb_dir/$pkg_name ($(du -h "$deb_dir/$pkg_name" | cut -f1))"
}
# ---- create combined .tar.gz (all binaries, no variant split) ----
build_tar() {
local tar_dir="${DIST_DIR}/tar"
mkdir -p "$tar_dir"
local archive_name="homeagent_${PKG_VERSION}_linux_${TAR_ARCH}.tar.gz"
local archive_dir="homeagent-${PKG_VERSION}-linux-${TAR_ARCH}"
# build combined staging
local staging
staging="$(mktemp -d "$STAGE_TMP/tar.XXXXXX")"
mkdir -p "$staging/usr/bin" "$staging/usr/lib/homeagent" "$staging/etc/systemd/system"
# copy all available binaries
for bin in homed waiter initconfig; do
local src="$BUILD_DIR/${bin}_linux_${TAR_ARCH}"
[ -f "$src" ] && cp "$src" "$staging/usr/bin/$bin"
done
# setup script
local setup_src="$PROJECT_ROOT/deploy/packaging/linux/setup.sh"
[ -f "$setup_src" ] && cp "$setup_src" "$staging/usr/lib/homeagent/setup.sh"
cp "$PROJECT_ROOT/deploy/homeagent.service" "$staging/etc/systemd/system/homeagent.service"
stage_license "$staging"
stage_multimodal_assets "$staging"
# GUI if available
local gui_src="$BUILD_DIR/homeagent-gui-linux-${TAR_ARCH}"
if [ -d "$gui_src" ]; then
mkdir -p "$staging/usr/lib/homeagent-gui"
cp -r "$gui_src"/* "$staging/usr/lib/homeagent-gui/"
cat > "$staging/usr/bin/homeagent-gui" << 'SCRIPT'
#!/bin/sh
exec /usr/lib/homeagent-gui/homeagent-gui "$@"
SCRIPT
chmod 755 "$staging/usr/bin/homeagent-gui"
fi
chmod 755 "$staging/usr/bin/"* 2>/dev/null || true
echo ">>> Building .tar.gz: $archive_name"
(cd "$staging" && tar czf "$tar_dir/$archive_name" --transform "s|^\.|${archive_dir}|" .)
echo " Created: $tar_dir/$archive_name ($(du -h "$tar_dir/$archive_name" | cut -f1))"
rm -rf "$staging"
}
# ---- create .rpm (via fpm if available) ----
build_rpm() {
local variant="$1"
local staging="$2"
local rpm_dir="${DIST_DIR}/rpm"
mkdir -p "$rpm_dir"
local pkg_name="homeagent-${variant}-${PKG_VERSION}-1.${RPM_ARCH}.rpm"
# find fpm
local fpm_bin="$(command -v fpm 2>/dev/null || true)"
if [ -z "$fpm_bin" ]; then
fpm_bin="$(find /home -name "fpm" -type f -path "*/bin/*" 2>/dev/null | head -1 || true)"
fi
if [ -z "$fpm_bin" ]; then
echo " SKIP .rpm: fpm not installed. Install it with: gem install fpm"
echo " Alternatively, build RPM on Fedora/RHEL using:"
echo " rpmbuild -ba deploy/packaging/linux/homeagent.spec"
return
fi
# find or extract rpmbuild (fpm needs it)
local rpmbuild_dir="/tmp/rpmext"
if [ ! -f "$rpmbuild_dir/usr/bin/rpmbuild" ]; then
# try to extract from cached deb packages
local rpm_deb
# 同 build_gui:`find` 对不存在/无命中会返回 1,`set -euo pipefail` 下
# 会让脚本当场退出(`|| true` 是给命令替换兜底,不是忽视错误)。
rpm_deb="$(find /tmp -name "rpm_*.deb" -type f 2>/dev/null | head -1 || true)"
if [ -z "$rpm_deb" ]; then
rpm_deb="$(find "$PROJECT_ROOT" -name "rpm_*.deb" -type f 2>/dev/null | head -1 || true)"
fi
if [ -n "$rpm_deb" ]; then
mkdir -p "$rpmbuild_dir"
(cd "$rpmbuild_dir" && ar x "$rpm_deb" 2>/dev/null && \
tar --no-same-permissions -xf data.tar.zst --zstd 2>/dev/null) || true
for libdeb in /tmp/librpm*.deb; do
[ -f "$libdeb" ] && (cd "$rpmbuild_dir" && ar x "$libdeb" 2>/dev/null && \
tar --no-same-permissions -xf data.tar.zst --zstd 2>/dev/null) || true
done
fi
fi
if [ -f "$rpmbuild_dir/usr/bin/rpmbuild" ]; then
export PATH="$rpmbuild_dir/usr/bin:$PATH"
export LD_LIBRARY_PATH="$rpmbuild_dir/usr/lib/x86_64-linux-gnu${LD_LIBRARY_PATH:+:$LD_LIBRARY_PATH}"
export RPM_CONFIGDIR="$rpmbuild_dir/usr/lib/rpm"
fi
echo ">>> Building .rpm via fpm: $pkg_name"
"$fpm_bin" -s dir -t rpm \
-n "homeagent-${variant}" \
-v "$VERSION" \
--iteration 1 \
-a "$RPM_ARCH" \
--description "HomeAgent ${variant^} package" \
--url "https://github.com/trueagent/HomeAgent" \
--license "AGPL-3.0-only" \
-C "$staging" \
-p "$rpm_dir/$pkg_name" \
. 2>&1
echo " Created: $rpm_dir/$pkg_name"
}
# ---- main ----
main() {
local target_arch="$ARCH"
mkdir -p "$BUILD_DIR" "$STAGE_TMP"
case "$ACTION" in
all|build)
build_go
build_gui
;;
esac
if [ "$ACTION" = "build" ]; then
echo ""
echo "=== Build complete. Binaries in $BUILD_DIR ==="
exit 0
fi
mkdir -p "$DIST_DIR"
# 上次成功构建留下的校验和必须在本次开工前删掉:本次若中途失败,脚本直接退出、
# 不重算 SHA256SUMS,旧的它会一直躺在 dist 里,看上去像在为这一批残缺产物背书。
rm -f "$DIST_DIR/SHA256SUMS"
for variant in full server client toolchain; do
echo ""
echo "=============================================="
echo " Packaging: $variant"
echo "=============================================="
local staging
staging=$(mktemp -d "$STAGE_TMP/stage.XXXXXX")
stage_variant "$variant" "$staging"
case "$ACTION" in
all|deb) build_deb "$variant" "$staging" ;;
esac
case "$ACTION" in
all|rpm) build_rpm "$variant" "$staging" ;;
esac
rm -rf "$staging"
done
case "$ACTION" in
all|tar) build_tar ;;
esac
echo ""
echo "=== Done! Packages in: $DIST_DIR ==="
echo ""
echo "Summary:"
# 只列**本批**产物:dist/ 会跨多次构建累积,用 find 全目录会让清单/SHA256SUMS
# 带上历史版本的文件名——用户下载那种清单后 `sha256sum -c` 必然报缺失。
# (v1.2.2 构建时就出现过:清单里混进了 1.2.0/1.2.1 的包名。)按本批版本号过滤。
mapfile -t release_files < <(find "$DIST_DIR" -type f \( -name "*${PKG_VERSION}*.deb" -o -name "homeagent_${PKG_VERSION}_*.tar.gz" -o -name "*${PKG_VERSION}*.rpm" \) 2>/dev/null | sort)
for f in "${release_files[@]}"; do
echo " $(du -h "$f" | cut -f1) $f"
done
# 全部包生成之后一次计算,避免边打边算漏掉后生成的产物。
# 名字用**平铺名**(basename):下载页的附件名就是平铺的,
# 清单里若写 ./deb/xxx.deb,用户下载后 `sha256sum -c` 会找不到文件。
if [ ${#release_files[@]} -gt 0 ]; then
(
cd "$DIST_DIR"
# 哈希取**真实路径**,标签用**平铺名**:两者不能混(直接对 basename 求哈希会找不到文件)。
for f in "${release_files[@]}"; do
printf '%s ./%s\n' "$(sha256sum "$f" | awk '{print $1}')" "$(basename "$f")"
done | sort -k2 > SHA256SUMS
)
echo " SHA256SUMS: $DIST_DIR/SHA256SUMS (仅本批 ${#release_files[@]} 个产物,平铺名)"
fi
}
main