html.go 701 B

12345678910111213141516171819202122232425262728293031
  1. package utils
  2. import (
  3. "regexp"
  4. "strings"
  5. )
  6. func StripTags(s string) string {
  7. //将HTML标签全转换成小写
  8. re, _ := regexp.Compile("\\<[\\S\\s]+?\\>")
  9. src := re.ReplaceAllStringFunc(s, strings.ToLower)
  10. //去除STYLE
  11. re, _ = regexp.Compile("\\<style[\\S\\s]+?\\</style\\>")
  12. src = re.ReplaceAllString(src, "")
  13. //去除SCRIPT
  14. re, _ = regexp.Compile("\\<script[\\S\\s]+?\\</script\\>")
  15. src = re.ReplaceAllString(src, "")
  16. //去除所有尖括号内的HTML代码,并换成换行符
  17. re, _ = regexp.Compile("\\<[\\S\\s]+?\\>")
  18. src = re.ReplaceAllString(src, "\n")
  19. //去除连续的换行符
  20. re, _ = regexp.Compile("\\s{2,}")
  21. src = re.ReplaceAllString(src, "\n")
  22. return src
  23. }