2024-05-29 18:33:16 +01:00
|
|
|
package utilities
|
|
|
|
|
|
|
|
import (
|
2024-06-01 01:08:43 +01:00
|
|
|
"io"
|
2024-06-01 02:57:48 +01:00
|
|
|
"strconv"
|
2024-05-29 18:33:16 +01:00
|
|
|
"strings"
|
|
|
|
|
|
|
|
"golang.org/x/net/html"
|
|
|
|
)
|
|
|
|
|
2024-06-01 02:57:48 +01:00
|
|
|
const (
|
|
|
|
htmlNoList int = iota
|
|
|
|
htmlOrderedList
|
|
|
|
htmlUnorderedList
|
|
|
|
)
|
|
|
|
|
2024-06-01 01:08:43 +01:00
|
|
|
func ConvertHTMLToText(text string) string {
|
2024-05-29 18:33:16 +01:00
|
|
|
var builder strings.Builder
|
|
|
|
|
2024-06-01 02:57:48 +01:00
|
|
|
htmlList := htmlNoList
|
|
|
|
orderedListNumber := 1
|
2024-06-01 01:08:43 +01:00
|
|
|
|
|
|
|
token := html.NewTokenizer(strings.NewReader(text))
|
|
|
|
|
2024-05-29 18:33:16 +01:00
|
|
|
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()
|
2024-06-01 02:57:48 +01:00
|
|
|
htmlList, orderedListNumber = processTagToken(&builder, tag, htmlList, orderedListNumber)
|
2024-05-29 18:33:16 +01:00
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2024-06-01 02:57:48 +01:00
|
|
|
func processTagToken(writer io.StringWriter, tag string, htmlList, orderedListNumber int) (int, int) {
|
2024-05-29 18:33:16 +01:00
|
|
|
switch tag {
|
2024-06-01 01:08:43 +01:00
|
|
|
case "<br>", "<p>", "</p>", "</li>":
|
|
|
|
_, _ = writer.WriteString("\n")
|
|
|
|
case "<ul>":
|
2024-06-01 02:57:48 +01:00
|
|
|
htmlList = htmlUnorderedList
|
|
|
|
_, _ = writer.WriteString("\n")
|
|
|
|
case "<ol>":
|
|
|
|
htmlList = htmlOrderedList
|
|
|
|
_, _ = writer.WriteString("\n")
|
2024-06-01 01:08:43 +01:00
|
|
|
case "</ul>":
|
2024-06-01 02:57:48 +01:00
|
|
|
htmlList = htmlNoList
|
|
|
|
case "</ol>":
|
|
|
|
htmlList = htmlNoList
|
|
|
|
orderedListNumber = 1
|
2024-06-01 01:08:43 +01:00
|
|
|
case "<li>":
|
2024-06-01 02:57:48 +01:00
|
|
|
switch htmlList {
|
|
|
|
case htmlUnorderedList:
|
2024-06-01 01:08:43 +01:00
|
|
|
_, _ = writer.WriteString("• ")
|
2024-06-01 02:57:48 +01:00
|
|
|
case htmlOrderedList:
|
|
|
|
_, _ = writer.WriteString(strconv.Itoa(orderedListNumber) + ". ")
|
|
|
|
orderedListNumber++
|
2024-06-01 01:08:43 +01:00
|
|
|
}
|
2024-05-29 18:33:16 +01:00
|
|
|
}
|
|
|
|
|
2024-06-01 02:57:48 +01:00
|
|
|
return htmlList, orderedListNumber
|
2024-05-29 18:33:16 +01:00
|
|
|
}
|