data_ai/ul/md.go

package ul

import (
	"github.com/jaytaylor/html2text"
	"regexp"
)

var md_reg1 = regexp.MustCompile("([ *]+)")
var md_reg2 = regexp.MustCompile("([\\n]+)")
var md_reg3 = regexp.MustCompile("([-]{3,})")

// 有缺陷---内存增大···不知原因
func HttpConvertToMarkdown(html string) string {
	if text := PostMarkDownText(html); text != "" {
		return text
	}
	return html
}

// 有问题，内存泄漏
func ConvertToMarkdown(html string) string {
	markdown, err := html2text.FromString(html, html2text.Options{PrettyTables: true})
	if err != nil {
		return html
	}
	markdown = md_reg1.ReplaceAllString(markdown, " ")
	markdown = md_reg2.ReplaceAllString(markdown, "\n")
	markdown = md_reg3.ReplaceAllString(markdown, "---")
	return markdown
}