71 lines
1.4 KiB
Go
71 lines
1.4 KiB
Go
package utilities
|
|
|
|
import (
|
|
"io"
|
|
"strconv"
|
|
"strings"
|
|
|
|
"golang.org/x/net/html"
|
|
)
|
|
|
|
const (
|
|
htmlNoList int = iota
|
|
htmlOrderedList
|
|
htmlUnorderedList
|
|
)
|
|
|
|
type htmlConvertState struct {
|
|
htmlListType int
|
|
orderedListNumber int
|
|
}
|
|
|
|
func ConvertHTMLToText(text string) string {
|
|
var builder strings.Builder
|
|
|
|
state := htmlConvertState{
|
|
htmlListType: htmlNoList,
|
|
orderedListNumber: 1,
|
|
}
|
|
|
|
token := html.NewTokenizer(strings.NewReader(text))
|
|
|
|
for {
|
|
tt := token.Next()
|
|
switch tt {
|
|
case html.ErrorToken:
|
|
return builder.String()
|
|
case html.TextToken:
|
|
text := token.Token().Data
|
|
builder.WriteString(text)
|
|
case html.StartTagToken, html.EndTagToken:
|
|
tag := token.Token().String()
|
|
processTagToken(&state, &builder, tag)
|
|
}
|
|
}
|
|
}
|
|
|
|
func processTagToken(state *htmlConvertState, writer io.StringWriter, tag string) {
|
|
switch tag {
|
|
case "<br>", "<p>", "</p>", "</li>":
|
|
_, _ = writer.WriteString("\n")
|
|
case "<ul>":
|
|
state.htmlListType = htmlUnorderedList
|
|
_, _ = writer.WriteString("\n")
|
|
case "<ol>":
|
|
state.htmlListType = htmlOrderedList
|
|
_, _ = writer.WriteString("\n")
|
|
case "</ul>":
|
|
state.htmlListType = htmlNoList
|
|
case "</ol>":
|
|
state.htmlListType = htmlNoList
|
|
state.orderedListNumber = 1
|
|
case "<li>":
|
|
switch state.htmlListType {
|
|
case htmlUnorderedList:
|
|
_, _ = writer.WriteString("• ")
|
|
case htmlOrderedList:
|
|
_, _ = writer.WriteString(strconv.Itoa(state.orderedListNumber) + ". ")
|
|
state.orderedListNumber++
|
|
}
|
|
}
|
|
}
|