hugolib: markdown_title: add title from markdown

Getting title from the page itself, as first encountered header in
markdown file.

Currently it's inefficient as we parsing page twice, the first in some
ealier point when hugo renders markdown, so it should be embedded
somewhere in that place.

But currently i don't care about if a PoC implementation.

Signed-off-by: Nikita Shubin <nikita.shubin@maquefel.me>

diff --git a/hugolib/content_map_page.go b/hugolib/content_map_page.go
index 21a4e8f2abe162e88e0711dbfdb24c05991ba91d..7954c606f72187dc75c52e945dac8d401198fd41 100644 (file)
--- a/hugolib/content_map_page.go
+++ b/hugolib/content_map_page.go
@@ -141,6 +141,12 @@ func (m *pageMap) newPageFromContentNode(n *contentNode, parentBucket *pagesMapB
        }
        ps.codeowners = owners
 
+       mi, err := s.h.markdownTitleInfoForPage(ps)
+       if err != nil {
+               return nil, errors.Wrap(err, "failed to load MarkdownTitle data")
+       }
+       ps.markdownTitleInfo = mi
+
        r, err := content()
        if err != nil {
                return nil, err

diff --git a/hugolib/hugo_sites.go b/hugolib/hugo_sites.go
index d67652dab7e9bf35b6fb1649a0b37b7287ed91b4..b69cd61c32c31b17508a77adb3a0bb1ad78b7cad 100644 (file)
--- a/hugolib/hugo_sites.go
+++ b/hugolib/hugo_sites.go
@@ -81,6 +81,8 @@ type HugoSites struct {
        gitInfo       *gitInfo
        codeownerInfo *codeownerInfo
 
+       markdownTitleInfo *markdownTitleInfo
+
        // As loaded from the /data dirs
        data map[string]any
 
@@ -181,6 +183,8 @@ type hugoSitesInit struct {
        // Loads the Git info and CODEOWNERS for all the pages if enabled.
        gitInfo *lazy.Init
 
+       markdownTitleInfo *lazy.Init
+
        // Maps page translations.
        translations *lazy.Init
 }
@@ -189,6 +193,7 @@ func (h *hugoSitesInit) Reset() {
        h.data.Reset()
        h.layouts.Reset()
        h.gitInfo.Reset()
+       h.markdownTitleInfo.Reset()
        h.translations.Reset()
 }
 
@@ -224,6 +229,19 @@ func (h *HugoSites) codeownersForPage(p page.Page) ([]string, error) {
        return h.codeownerInfo.forPage(p), nil
 }
 
+
+func (h *HugoSites) markdownTitleInfoForPage(p page.Page) (*MarkdownTitleInfo, error) {
+       if _, err := h.init.markdownTitleInfo.Do(); err != nil {
+               return nil, err
+       }
+
+       if h.markdownTitleInfo == nil {
+               return nil, nil
+       }
+
+       return h.markdownTitleInfo.forPage(p), nil
+}
+
 func (h *HugoSites) siteInfos() page.Sites {
        infos := make(page.Sites, len(h.Sites))
        for i, site := range h.Sites {
@@ -353,6 +371,7 @@ func newHugoSites(cfg deps.DepsCfg, sites ...*Site) (*HugoSites, error) {
                        data:         lazy.New(),
                        layouts:      lazy.New(),
                        gitInfo:      lazy.New(),
+                       markdownTitleInfo: lazy.New(),
                        translations: lazy.New(),
                },
        }
@@ -396,6 +415,14 @@ func newHugoSites(cfg deps.DepsCfg, sites ...*Site) (*HugoSites, error) {
                return nil, nil
        })
 
+       h.init.markdownTitleInfo.Add(func() (any, error) {
+               err := h.loadMarkdownTitleInfo()
+               if err != nil {
+                       return nil, errors.Wrap(err, "failed to load MarkdownTitle info")
+               }
+               return nil, nil
+       })
+
        for _, s := range sites {
                s.h = h
        }
@@ -444,6 +471,18 @@ func (h *HugoSites) loadGitInfo() error {
        return nil
 }
 
+func (h *HugoSites) loadMarkdownTitleInfo() error {
+       if h.Cfg.GetBool("enableMarkdownTitleInfo") {
+               mi, err := newMarkdownTitleInfo(h.Cfg)
+               if err != nil {
+                       h.Log.Errorln("Failed to read MarkdownTitle:", err)
+               } else {
+                       h.markdownTitleInfo = mi
+               }
+       }
+       return nil
+}
+
 func (l configLoader) applyDeps(cfg deps.DepsCfg, sites ...*Site) error {
        if cfg.TemplateProvider == nil {
                cfg.TemplateProvider = tplimpl.DefaultTemplateProvider

diff --git a/hugolib/markdown_title.go b/hugolib/markdown_title.go
new file mode 100644 (file)
index 0000000..67d4ea7
--- /dev/null
+++ b/hugolib/markdown_title.go
@@ -0,0 +1,66 @@
+package hugolib
+
+import (
+       "path/filepath"
+       "strings"
+       "os"
+
+       "github.com/yuin/goldmark"
+       "github.com/yuin/goldmark/text"
+       "github.com/yuin/goldmark/ast"
+
+       "github.com/gohugoio/hugo/config"
+       "github.com/gohugoio/hugo/resources/page"
+)
+
+type MarkdownTitleInfo struct {
+       Title           string    `json:"title"`            // Title
+}
+
+type MarkdownTitleMap map[string]*MarkdownTitleInfo
+
+type markdownTitleInfo struct {
+       contentDir string
+       Map MarkdownTitleMap
+}
+
+func findTitle(n ast.Node, source []byte) (string, bool) {
+       ret := ""
+       if n.Kind() == ast.KindHeading {
+               return string(n.Text(source)[:]), true
+       }
+       for c := n.FirstChild(); c != nil; c = c.NextSibling() {
+               ret, res := findTitle(c, source)
+               if res {
+                       return ret, true
+               }
+       }
+
+       return ret, false
+}
+
+func (m *markdownTitleInfo) forPage(p page.Page) *MarkdownTitleInfo {
+       markdown := goldmark.New()
+       name := strings.TrimPrefix(filepath.ToSlash(p.File().Filename()), m.contentDir)
+       // name = strings.TrimPrefix(name, "/")
+
+       source, err := os.ReadFile(name)
+       if err != nil {
+               return nil
+       }
+
+       // Parse file and return title
+       n := markdown.Parser().Parse(text.NewReader([]byte(source)))
+       title, res := findTitle(n, source)
+       if !res {
+               return &MarkdownTitleInfo{Title: ""}
+       }
+
+       return &MarkdownTitleInfo{Title: title}
+}
+
+func newMarkdownTitleInfo(cfg config.Provider) (*markdownTitleInfo, error) {
+       contentDir := cfg.GetString("contentDir")
+
+       return &markdownTitleInfo{contentDir: contentDir}, nil
+}

diff --git a/hugolib/page.go b/hugolib/page.go
index 77165c072921ad0d1b4e182671197a2e35a8f6d7..5fae15ee4abc36632ae35bfefa48f57f8583a079 100644 (file)
--- a/hugolib/page.go
+++ b/hugolib/page.go
@@ -160,6 +160,10 @@ func (p *pageState) CodeOwners() []string {
        return p.codeowners
 }
 
+func (p *pageState) MarkdownTitleInfo() *MarkdownTitleInfo {
+       return p.markdownTitleInfo
+}
+
 // GetTerms gets the terms defined on this page in the given taxonomy.
 // The pages returned will be ordered according to the front matter.
 func (p *pageState) GetTerms(taxonomy string) page.Pages {

diff --git a/hugolib/page__common.go b/hugolib/page__common.go
index e55bb7e253c3624c21d309780fa1b31f8bd28554..b0ab4489fc74f2b0a802b70e58aa68eb8a6bcc2a 100644 (file)
--- a/hugolib/page__common.go
+++ b/hugolib/page__common.go
@@ -106,6 +106,8 @@ type pageCommon struct {
        gitInfo    *gitmap.GitInfo
        codeowners []string
 
+       markdownTitleInfo    *MarkdownTitleInfo
+
        // Positional navigation
        posNextPrev        *nextPrev
        posNextPrevSection *nextPrev