| 298 | } |
| 299 | |
| 300 | func trimHTML(src string) string { |
| 301 | // 将HTML标签全转换成小写 |
| 302 | re := regexp.MustCompile(`<[\S\s]+?>`) |
| 303 | src = re.ReplaceAllStringFunc(src, strings.ToLower) |
| 304 | // 去除STYLE |
| 305 | re = regexp.MustCompile(`<style[\S\s]+?</style>`) |
| 306 | src = re.ReplaceAllString(src, "") |
| 307 | // 去除SCRIPT |
| 308 | re = regexp.MustCompile(`<script[\S\s]+?</script>`) |
| 309 | src = re.ReplaceAllString(src, "") |
| 310 | // 去除所有尖括号内的HTML代码,并换成换行符 |
| 311 | re = regexp.MustCompile(`<[\S\s]+?>`) |
| 312 | src = re.ReplaceAllString(src, "\n") |
| 313 | // 去除连续的换行符 |
| 314 | re = regexp.MustCompile(`\s{2,}`) |
| 315 | src = re.ReplaceAllString(src, "\n") |
| 316 | return strings.TrimSpace(src) |
| 317 | } |
| 318 | func getRequest(url string) (resp *resty.Response, err error) { |
| 319 | client := resty.New() |
| 320 | resp, err = client.R().Get(url) |