Merge pull request #19 from Unknwon/master

add func Html2str()

@@ -10,6 +10,7 @@ import (
 	"os"
 	"path"
 	"path/filepath"
+	"regexp"
 	"strings"
 	"time"
 )
@@ -30,6 +31,7 @@ func init() {
 	beegoTplFuncMap["date"] = Date
 	beegoTplFuncMap["compare"] = Compare
 	beegoTplFuncMap["substr"] = Substr
+	beegoTplFuncMap["html2str"] = Html2str
 }
 
 // MarkDown parses a string in MarkDown format and returns HTML. Used by the template parser as "markdown"
@@ -54,6 +56,33 @@ func Substr(s string, start, length int) string {
 	return string(bt[start:end])
 }
 
+// Html2str() returns escaping text convert from html
+func Html2str(html string) string {
+	src := string(html)
+
+	//将HTML标签全转换成小写
+	re, _ := regexp.Compile("\\<[\\S\\s]+?\\>")
+	src = re.ReplaceAllStringFunc(src, strings.ToLower)
+
+	//去除STYLE
+	re, _ = regexp.Compile("\\<style[\\S\\s]+?\\</style\\>")
+	src = re.ReplaceAllString(src, "")
+
+	//去除SCRIPT
+	re, _ = regexp.Compile("\\<script[\\S\\s]+?\\</script\\>")
+	src = re.ReplaceAllString(src, "")
+
+	//去除所有尖括号内的HTML代码，并换成换行符
+	re, _ = regexp.Compile("\\<[\\S\\s]+?\\>")
+	src = re.ReplaceAllString(src, "\n")
+
+	//去除连续的换行符
+	re, _ = regexp.Compile("\\s{2,}")
+	src = re.ReplaceAllString(src, "\n")
+
+	return strings.TrimSpace(src)
+}
+
 // DateFormat takes a time and a layout string and returns a string with the formatted date. Used by the template parser as "dateformat"
 func DateFormat(t time.Time, layout string) (datestring string) {
 	datestring = t.Format(layout)