go開源Hugo站點構建三步曲之集結渲染
Assemble
Assemble所做的事情很純粹,那就是創建站點頁面實例 – pageState。 因為支持多站點,contentMaps有多個。 所以Assemble不僅要創建pageState,還需要管理好所有的pages,這就用到瞭PageMaps。
type pageMap struct { s *Site *contentMap } type pageMaps struct { workers *para.Workers pmaps []*pageMap }
實際上pageMap就是由contentMap組合而來的。 而contentMap中的組成樹的結點就是contentNode。
正好,每個contentNode又對應一個pageState。
type contentNode struct { p *pageState // Set if source is a file. // We will soon get other sources. fi hugofs.FileMetaInfo // The source path. Unix slashes. No leading slash. path string ... }
所以Assemble不僅要為前面Process處理過生成的contentNode創建pageState,還要補齊一些缺失的contentNode,如Section。
PageState
可以看出,Assemble的重點就是組建PageState,那她到底長啥樣:
type pageState struct { // This slice will be of same length as the number of global slice of output // formats (for all sites). pageOutputs []*pageOutput // This will be shifted out when we start to render a new output format. *pageOutput // Common for all output formats. *pageCommon ... }
從註解中可以看出普通信息將由pageCommon提供,而輸出信息則由pageOutput提供。 比較特殊的是pageOutputs,是pageOutput的數組。 在 基礎架構中,對這一點有作分析。 這要歸因於Hugo的多站點渲染策略 – 允許在不同的站點中重用其它站點的頁面。
// hugo-playground/hugolib/page__new.go // line 97 // Prepare output formats for all sites. // We do this even if this page does not get rendered on // its own. It may be referenced via .Site.GetPage and // it will then need an output format. ps.pageOutputs = make([]*pageOutput, len(ps.s.h.renderFormats))
那在Assemble中Hugo是如何組織pageState實例的呢?
從上圖中,可以看出Assemble階段主要是新建pageState。 其中pageOutput在這一階段隻是一個占位符,空的nopPageOutput。 pageCommon則是在這一階段給賦予瞭很多的信息,像meta相關的信息,及各種細節信息的providers。
動手實踐 – Show Me the Code of Create a PageState
package main import ( "fmt" "html/template" ) func main() { outputFormats := createOutputFormats() renderFormats := initRenderFormats(outputFormats) s := &site{ outputFormats: outputFormats, renderFormats: renderFormats, } ps := &pageState{ pageOutputs: nil, pageOutput: nil, pageCommon: &pageCommon{m: &pageMeta{kind: KindPage}}, } ps.init(s) // prepare ps.pageOutput = ps.pageOutputs[0] // render fmt.Println(ps.targetPaths().TargetFilename) fmt.Println(ps.Content()) fmt.Println(ps.m.kind) } type site struct { outputFormats map[string]Formats renderFormats Formats } type pageState struct { // This slice will be of same length as the number of global slice of output // formats (for all sites). pageOutputs []*pageOutput // This will be shifted out when we start to render a new output format. *pageOutput // Common for all output formats. *pageCommon } func (p *pageState) init(s *site) { pp := newPagePaths(s) p.pageOutputs = make([]*pageOutput, len(s.renderFormats)) for i, f := range s.renderFormats { ft, found := pp.targetPaths[f.Name] if !found { panic("target path not found") } providers := struct{ targetPather }{ft} po := &pageOutput{ f: f, pagePerOutputProviders: providers, ContentProvider: nil, } contentProvider := newPageContentOutput(po) po.ContentProvider = contentProvider p.pageOutputs[i] = po } } func newPageContentOutput(po *pageOutput) *pageContentOutput { cp := &pageContentOutput{ f: po.f, } initContent := func() { cp.content = template.HTML("<p>hello content</p>") } cp.initMain = func() { initContent() } return cp } func newPagePaths(s *site) pagePaths { outputFormats := s.renderFormats targets := make(map[string]targetPathsHolder) for _, f := range outputFormats { target := "/" + "blog" + "/" + f.BaseName + "." + f.MediaType.SubType paths := TargetPaths{ TargetFilename: target, } targets[f.Name] = targetPathsHolder{ paths: paths, } } return pagePaths{ targetPaths: targets, } } type pagePaths struct { targetPaths map[string]targetPathsHolder } type targetPathsHolder struct { paths TargetPaths } func (t targetPathsHolder) targetPaths() TargetPaths { return t.paths } type pageOutput struct { f Format // These interface provides the functionality that is specific for this // output format. pagePerOutputProviders ContentProvider // May be nil. cp *pageContentOutput } // pageContentOutput represents the Page content for a given output format. type pageContentOutput struct { f Format initMain func() content template.HTML } func (p *pageContentOutput) Content() any { p.initMain() return p.content } // these will be shifted out when rendering a given output format. type pagePerOutputProviders interface { targetPather } type targetPather interface { targetPaths() TargetPaths } type TargetPaths struct { // Where to store the file on disk relative to the publish dir. OS slashes. TargetFilename string } type ContentProvider interface { Content() any } type pageCommon struct { m *pageMeta } type pageMeta struct { // kind is the discriminator that identifies the different page types // in the different page collections. This can, as an example, be used // to to filter regular pages, find sections etc. // Kind will, for the pages available to the templates, be one of: // page, home, section, taxonomy and term. // It is of string type to make it easy to reason about in // the templates. kind string } func initRenderFormats( outputFormats map[string]Formats) Formats { return outputFormats[KindPage] } func createOutputFormats() map[string]Formats { m := map[string]Formats{ KindPage: {HTMLFormat}, } return m } const ( KindPage = "page" ) var HTMLType = newMediaType("text", "html") // HTMLFormat An ordered list of built-in output formats. var HTMLFormat = Format{ Name: "HTML", MediaType: HTMLType, BaseName: "index", } func newMediaType(main, sub string) Type { t := Type{ MainType: main, SubType: sub, Delimiter: "."} return t } type Type struct { MainType string `json:"mainType"` // i.e. text SubType string `json:"subType"` // i.e. html Delimiter string `json:"delimiter"` // e.g. "." } type Format struct { // The Name is used as an identifier. Internal output formats (i.e. HTML and RSS) // can be overridden by providing a new definition for those types. Name string `json:"name"` MediaType Type `json:"-"` // The base output file name used when not using "ugly URLs", defaults to "index". BaseName string `json:"baseName"` } type Formats []Format
輸出結果:
/blog/index.html <p>hello content</p> page Program exited.
PageState線上可直接運行版本
Render
基礎信息是由pageCommon提供瞭,那渲染過程中的輸出由誰提供呢?
沒錯,輪到pageOutput瞭:
可以看到,在render階段,pageState的pageOutput得到瞭最終的處理,為發佈做準備瞭。 為瞭發佈,最重的信息是發佈什麼,以及發佈到哪裡去。 這些信息都在pageOutput中,其中ContentProvider是提供發佈內容的,而targetPathsProvider則是提供發佈地址信息的。 其中地址信息主要來源於PagePath,這又和站點的RenderFormats和OutputFormats相關,哪下圖所示:
其中OutputFormats, RenderFormats及PageOutput之間的關系有在 基礎架構中有詳細提到,這裡就不再贅述。
// We create a pageOutput for every output format combination, even if this // particular page isn't configured to be rendered to that format. type pageOutput struct { ... // These interface provides the functionality that is specific for this // output format. pagePerOutputProviders page.ContentProvider page.TableOfContentsProvider page.PageRenderProvider // May be nil. cp *pageContentOutput }
其中pageContentOutput正是實現瞭ContentProvider接口的實例。 其中有包含markdown文件原始信息的workContent字段,以及包含處理過後的內容content字段。 如Hugo Shortcode特性。 就是在這裡經過contentToRender方法將原始信息進行處理,而最終實現的。
動手實踐 – Show Me the Code of Publish
package main import ( "bytes" "fmt" "io" "os" "path/filepath" ) // publisher needs to know: // 1: what to publish // 2: where to publish func main() { // 1 // src is template executed result // it is the source that we need to publish // take a look at template executor example // https://c.sunwei.xyz/template-executor.html src := &bytes.Buffer{} src.Write([]byte("template executed result")) b := &bytes.Buffer{} transformers := createTransformerChain() if err := transformers.Apply(b, src); err != nil { fmt.Println(err) return } dir, _ := os.MkdirTemp("", "hugo") defer os.RemoveAll(dir) // 2 // targetPath is from pageState // this is where we need to publish // take a look at page state example // https://c.sunwei.xyz/page-state.html targetPath := filepath.Join(dir, "index.html") if err := os.WriteFile( targetPath, bytes.TrimSuffix(b.Bytes(), []byte("\n")), os.ModePerm); err != nil { panic(err) } fmt.Println("1. what to publish: ", string(b.Bytes())) fmt.Println("2. where to publish: ", dir) } func (c *Chain) Apply(to io.Writer, from io.Reader) error { fb := &bytes.Buffer{} if _, err := fb.ReadFrom(from); err != nil { return err } tb := &bytes.Buffer{} ftb := &fromToBuffer{from: fb, to: tb} for i, tr := range *c { if i > 0 { panic("switch from/to and reset to") } if err := tr(ftb); err != nil { continue } } _, err := ftb.to.WriteTo(to) return err } func createTransformerChain() Chain { transformers := NewEmpty() transformers = append(transformers, func(ft FromTo) error { content := ft.From().Bytes() w := ft.To() tc := bytes.Replace( content, []byte("result"), []byte("transferred result"), 1) _, _ = w.Write(tc) return nil }) return transformers } // Chain is an ordered processing chain. The next transform operation will // receive the output from the previous. type Chain []Transformer // Transformer is the func that needs to be implemented by a transformation step. type Transformer func(ft FromTo) error // FromTo is sent to each transformation step in the chain. type FromTo interface { From() BytesReader To() io.Writer } // BytesReader wraps the Bytes method, usually implemented by bytes.Buffer, and an // io.Reader. type BytesReader interface { // Bytes The slice given by Bytes is valid for use only until the next buffer modification. // That is, if you want to use this value outside of the current transformer step, // you need to take a copy. Bytes() []byte io.Reader } // NewEmpty creates a new slice of transformers with a capacity of 20. func NewEmpty() Chain { return make(Chain, 0, 2) } // Implements contentTransformer // Content is read from the from-buffer and rewritten to to the to-buffer. type fromToBuffer struct { from *bytes.Buffer to *bytes.Buffer } func (ft fromToBuffer) From() BytesReader { return ft.from } func (ft fromToBuffer) To() io.Writer { return ft.to }
輸出結果:
1. what to publish: template executed transferred result
2. where to publish: /tmp/hugo2834984546
Program exited.
Publish線上可直接運行版本
以上就是go開源Hugo站點構建三步曲之集結渲染的詳細內容,更多關於go Hugo站點構建集結渲染的資料請關註WalkonNet其它相關文章!
推薦閱讀:
- Go語言設計模式之實現觀察者模式解決代碼臃腫
- 關於Go 空結構體的 3 種使用場景
- Ajv format校驗使用示例分析
- Go中string與[]byte高效互轉的方法實例
- Go結構體SliceHeader及StringHeader作用詳解