Andreas/siyuan
1
0
Fork 0
mirror of https://github.com/siyuan-note/siyuan.git synced 2025-12-23 18:10:12 +01:00
Code Issues Projects Releases Packages Wiki Activity Actions 1

🎨 Improve HTML clipping https://github.com/siyuan-note/siyuan/issues/13125#issuecomment-2475461284

Browse source
This commit is contained in:
Daniel 2024-11-15 13:13:43 +08:00
parent ab147598e1
commit e408546be5
No known key found for this signature in database
GPG key ID: 86211BA83DF03017
3 changed files with 40 additions and 22 deletions
Download patch file Download diff file Expand all files Collapse all files

18
kernel/api/extension.go
View file

@ -123,7 +123,12 @@ func extensionCopy(c *gin.Context) {
} }
luteEngine := util.NewLute() luteEngine := util.NewLute()
luteEngine.SetHTMLTag2TextMark(true) luteEngine.SetSup(true)
luteEngine.SetSub(true)
luteEngine.SetMark(true)
luteEngine.SetGFMStrikethrough(true)
luteEngine.SetInlineAsterisk(true)
luteEngine.SetInlineUnderscore(true)
var md string var md string
var withMath bool var withMath bool
if nil != form.Value["href"] { if nil != form.Value["href"] {
@ -175,16 +180,19 @@ func extensionCopy(c *gin.Context) {
} }
} }
var tree *parse.Tree
if "" == md { if "" == md {
tree, withMath = model.HTML2Tree(dom, luteEngine)
if nil == tree {
md, withMath, _ = model.HTML2Markdown(dom, luteEngine) md, withMath, _ = model.HTML2Markdown(dom, luteEngine)
}
md = strings.TrimSpace(md)
if withMath { if withMath {
luteEngine.SetInlineMath(true) luteEngine.SetInlineMath(true)
} }
tree = parse.Parse("", []byte(md), luteEngine.ParseOptions)
}
}
var unlinks []*ast.Node var unlinks []*ast.Node
tree := parse.Parse("", []byte(md), luteEngine.ParseOptions)
ast.Walk(tree.Root, func(n *ast.Node, entering bool) ast.WalkStatus { ast.Walk(tree.Root, func(n *ast.Node, entering bool) ast.WalkStatus {
if !entering { if !entering {
return ast.WalkContinue return ast.WalkContinue

7
kernel/api/lute.go
View file

@ -58,7 +58,12 @@ func html2BlockDOM(c *gin.Context) {
dom := arg["dom"].(string) dom := arg["dom"].(string)
luteEngine := util.NewLute() luteEngine := util.NewLute()
luteEngine.SetHTMLTag2TextMark(true) luteEngine.SetSup(true)
luteEngine.SetSub(true)
luteEngine.SetMark(true)
luteEngine.SetGFMStrikethrough(true)
luteEngine.SetInlineAsterisk(true)
luteEngine.SetInlineUnderscore(true)
markdown, withMath, err := model.HTML2Markdown(dom, luteEngine) markdown, withMath, err := model.HTML2Markdown(dom, luteEngine)
if err != nil { if err != nil {
ret.Data = "Failed to convert" ret.Data = "Failed to convert"

33
kernel/model/import.go
View file

@ -54,8 +54,21 @@ import (
) )
func HTML2Markdown(htmlStr string, luteEngine *lute.Lute) (markdown string, withMath bool, err error) { func HTML2Markdown(htmlStr string, luteEngine *lute.Lute) (markdown string, withMath bool, err error) {
tree, withMath := HTML2Tree(htmlStr, luteEngine)
var formatted []byte
renderer := render.NewFormatRenderer(tree, luteEngine.RenderOptions)
for nodeType, rendererFunc := range luteEngine.HTML2MdRendererFuncs {
renderer.ExtRendererFuncs[nodeType] = rendererFunc
}
formatted = renderer.Render()
markdown = gulu.Str.FromBytes(formatted)
return
}
func HTML2Tree(htmlStr string, luteEngine *lute.Lute) (tree *parse.Tree, withMath bool) {
assetDirPath := filepath.Join(util.DataDir, "assets") assetDirPath := filepath.Join(util.DataDir, "assets")
tree := luteEngine.HTML2Tree(htmlStr) tree = luteEngine.HTML2Tree(htmlStr)
ast.Walk(tree.Root, func(n *ast.Node, entering bool) ast.WalkStatus { ast.Walk(tree.Root, func(n *ast.Node, entering bool) ast.WalkStatus {
if !entering { if !entering {
return ast.WalkContinue return ast.WalkContinue
@ -79,19 +92,11 @@ func HTML2Markdown(htmlStr string, luteEngine *lute.Lute) (markdown string, with
dest := n.TokensStr() dest := n.TokensStr()
if strings.HasPrefix(dest, "data:image") && strings.Contains(dest, ";base64,") { if strings.HasPrefix(dest, "data:image") && strings.Contains(dest, ";base64,") {
processBase64Img(n, dest, assetDirPath, err) processBase64Img(n, dest, assetDirPath)
return ast.WalkContinue return ast.WalkContinue
} }
return ast.WalkContinue return ast.WalkContinue
}) })
var formatted []byte
renderer := render.NewFormatRenderer(tree, luteEngine.RenderOptions)
for nodeType, rendererFunc := range luteEngine.HTML2MdRendererFuncs {
renderer.ExtRendererFuncs[nodeType] = rendererFunc
}
formatted = renderer.Render()
markdown = gulu.Str.FromBytes(formatted)
return return
} }
@ -836,7 +841,7 @@ func ImportFromLocalPath(boxID, localPath string, toPath string) (err error) {
} }
if strings.HasPrefix(dest, "data:image") && strings.Contains(dest, ";base64,") { if strings.HasPrefix(dest, "data:image") && strings.Contains(dest, ";base64,") {
processBase64Img(n, dest, assetDirPath, err) processBase64Img(n, dest, assetDirPath)
return ast.WalkContinue return ast.WalkContinue
} }
@ -953,7 +958,7 @@ func ImportFromLocalPath(boxID, localPath string, toPath string) (err error) {
} }
if strings.HasPrefix(dest, "data:image") && strings.Contains(dest, ";base64,") { if strings.HasPrefix(dest, "data:image") && strings.Contains(dest, ";base64,") {
processBase64Img(n, dest, assetDirPath, err) processBase64Img(n, dest, assetDirPath)
return ast.WalkContinue return ast.WalkContinue
} }
@ -1046,7 +1051,7 @@ func parseStdMd(markdown []byte) (ret *parse.Tree, yfmRootID, yfmTitle, yfmUpdat
return return
} }
func processBase64Img(n *ast.Node, dest string, assetDirPath string, err error) { func processBase64Img(n *ast.Node, dest string, assetDirPath string) {
base64TmpDir := filepath.Join(util.TempDir, "base64") base64TmpDir := filepath.Join(util.TempDir, "base64")
os.MkdirAll(base64TmpDir, 0755) os.MkdirAll(base64TmpDir, 0755)
@ -1110,7 +1115,7 @@ func processBase64Img(n *ast.Node, dest string, assetDirPath string, err error)
tmpFile.Close() tmpFile.Close()
assetTargetPath := filepath.Join(assetDirPath, name) assetTargetPath := filepath.Join(assetDirPath, name)
if err = filelock.Copy(tmp, assetTargetPath); err != nil { if err := filelock.Copy(tmp, assetTargetPath); err != nil {
logging.LogErrorf("copy asset from [%s] to [%s] failed: %s", tmp, assetTargetPath, err) logging.LogErrorf("copy asset from [%s] to [%s] failed: %s", tmp, assetTargetPath, err)
return return
} }