opsxshell/tools/homepage-build/main.go

// homepage-build：把 homepage/ 源(落地页+文档 md)构建为纯静态站点 homepage/dist/。
// 产物可直接部署到任意静态托管(nginx/Pages)——官网与桌面二进制彻底解耦。
// 用法(独立 Go 模块，须在其目录内运行): cd tools/homepage-build && go run . -src ../../homepage -out ../../homepage/dist
package main

import (
	"encoding/json"
	"flag"
	"fmt"
	"os"
	"path/filepath"
	"regexp"
	"strings"

	"github.com/yuin/goldmark"
	"github.com/yuin/goldmark/extension"
	gmhtml "github.com/yuin/goldmark/renderer/html"
	"html/template"
)

type PageDef struct {
	File    string `json:"file"`
	Slug    string `json:"slug"`
	Title   string `json:"title"`
	Updated string `json:"updated,omitempty"`
}
type GroupDef struct {
	Name  string    `json:"name"`
	Pages []PageDef `json:"pages"`
}
type Manifest struct {
	ProductVersion string     `json:"productVersion"`
	Groups         []GroupDef `json:"groups"`
}
type TocItem struct {
	ID    string
	Level int
	Text  string
}
type SearchEntry struct {
	Slug, Title, Group, Text string
}
type NavLink struct {
	Slug, Title string
}
type Page struct {
	Slug, Title, Group, Updated string
	HTML                        string
	TOC                         []TocItem
	Prev, Next                  *NavLink
}

var (
	hRe     = regexp.MustCompile(`(?s)<(h[23])>(.*?)</h[23]>`)
	tagRe   = regexp.MustCompile(`<[^>]+>`)
	mdNoise = regexp.MustCompile("(?s)```.*?```|`[^`]*`|!\\[[^\\]]*\\]\\([^)]*\\)|\\[[^\\]]*\\]\\([^)]*\\)|[#>*_\\-|]+")
)

func injectHeadingIDs(html string) (string, []TocItem) {
	var toc []TocItem
	counter := 0
	out := hRe.ReplaceAllStringFunc(html, func(m string) string {
		sub := hRe.FindStringSubmatch(m)
		tag, inner := sub[1], sub[2]
		counter++
		id := fmt.Sprintf("sec-%d", counter)
		text := strings.TrimSpace(tagRe.ReplaceAllString(inner, ""))
		level := 2
		if tag == "h3" {
			level = 3
		}
		toc = append(toc, TocItem{ID: id, Level: level, Text: text})
		return fmt.Sprintf(`<%s id="%s">%s</%s>`, tag, id, inner, tag)
	})
	return out, toc
}

func plainText(src string) string {
	s := mdNoise.ReplaceAllString(src, " ")
	s = strings.Join(strings.Fields(s), " ")
	if len(s) > 6000 {
		s = s[:6000]
	}
	return s
}

func must(err error, what string) {
	if err != nil {
		fmt.Fprintf(os.Stderr, "[homepage-build] %s: %v\n", what, err)
		os.Exit(1)
	}
}

func writeFile(path, content string) {
	must(os.MkdirAll(filepath.Dir(path), 0o755), "mkdir "+path)
	must(os.WriteFile(path, []byte(content), 0o644), "write "+path)
	fmt.Println("[OK]", path)
}

// copyTree 幂等地把 src 目录整棵拷到 dst：先清空 dst 再 CopyFS。
// os.CopyFS 在目标文件已存在时会报错，故重建 dist 前必须先清空对应子目录。
func copyTree(dst, src string) {
	must(os.RemoveAll(dst), "清理 "+dst)
	must(os.CopyFS(dst, os.DirFS(src)), "拷 "+dst)
	fmt.Println("[OK]", dst)
}

func main() {
	src := flag.String("src", "homepage", "源目录")
	out := flag.String("out", "homepage/dist", "产物目录")
	flag.Parse()

	// 1) 落地页与附加单页: 纯 HTML 直接拷 + site assets
	// index.html 必需；pricing/terms/license 为可选页，源存在即产出。
	// 附加页同时产出 <name>.html 与 <name>/index.html：兼容 nginx 的干净
	// URL(location = /license) 与纯静态服务器(python -m http.server)。
	for _, page := range []string{"index.html", "pricing.html", "terms.html", "license.html"} {
		raw, err := os.ReadFile(filepath.Join(*src, page))
		if err != nil {
			if page == "index.html" {
				must(err, "读 index.html")
			}
			continue
		}
		writeFile(filepath.Join(*out, page), string(raw))
		if page != "index.html" {
			writeFile(filepath.Join(*out, strings.TrimSuffix(page, ".html"), "index.html"), string(raw))
		}
	}
	siteAssets := filepath.Join(*src, "assets")
	if st, _ := os.Stat(siteAssets); st != nil {
		copyTree(filepath.Join(*out, "site", "assets"), siteAssets)
	}

	// 2) 文档: manifest → goldmark 渲染 → docs.html 模板
	mraw, err := os.ReadFile(filepath.Join(*src, "content", "manifest.json"))
	must(err, "读 manifest")
	var manifest Manifest
	must(json.Unmarshal(mraw, &manifest), "解析 manifest")
	tplRaw, err := os.ReadFile(filepath.Join(*src, "docs.html"))
	must(err, "读 docs.html")
	tpl, err := template.New("doc").Parse(string(tplRaw))
	must(err, "解析 docs 模板")
	md := goldmark.New(goldmark.WithExtensions(extension.Table, extension.Strikethrough, extension.TaskList), goldmark.WithRendererOptions(gmhtml.WithHardWraps()))

	var order []string
	pages := map[string]*Page{}
	var search []SearchEntry
	var prev *NavLink
	for _, g := range manifest.Groups {
		for _, def := range g.Pages {
			srcb, err := os.ReadFile(filepath.Join(*src, "content", def.File))
			must(err, "读文档 "+def.File)
			var buf strings.Builder
			must(md.Convert(srcb, &buf), "渲染 "+def.File)
			body, toc := injectHeadingIDs(buf.String())
			p := &Page{Slug: def.Slug, Title: def.Title, Group: g.Name, Updated: def.Updated, HTML: body, TOC: toc, Prev: prev}
			pages[def.Slug] = p
			order = append(order, def.Slug)
			prev = &NavLink{Slug: def.Slug, Title: def.Title}
			search = append(search, SearchEntry{Slug: def.Slug, Title: def.Title, Group: g.Name, Text: plainText(string(srcb))})
		}
	}
	for i, slug := range order {
		if i+1 < len(order) {
			n := pages[order[i+1]]
			pages[slug].Next = &NavLink{Slug: n.Slug, Title: n.Title}
		}
	}
	// /docs/ 首页 = 第一篇(静态站无重定向,直接渲染)
	render := func(p *Page) string {
		var sb strings.Builder
		must(tpl.Execute(&sb, map[string]any{"Manifest": manifest, "Page": p, "TOC": p.TOC, "Content": template.HTML(p.HTML)}), "渲染模板 "+p.Slug)
		return sb.String()
	}
	writeFile(filepath.Join(*out, "docs", "index.html"), render(pages[order[0]]))
	for _, slug := range order {
		writeFile(filepath.Join(*out, "docs", slug, "index.html"), render(pages[slug]))
	}
	sj, _ := json.Marshal(search)
	writeFile(filepath.Join(*out, "docs", "search-data.json"), string(sj))
	// docs assets(模板引用 /docs/assets/*)
	da := filepath.Join(*src, "docs-assets")
	if st, _ := os.Stat(da); st != nil {
		copyTree(filepath.Join(*out, "docs", "assets"), da)
	}
	fmt.Printf("[DONE] %d 页文档 + 落地页 → %s（纯静态,可部署任意域名）\n", len(order), *out)
}