现在的位置: 首页 > 综合 > 正文

截取HTML C#代码

2013年09月08日 ⁄ 综合 ⁄ 共 2830字 ⁄ 字号 评论关闭

protected string TrimHtml(string input, int length)
        {
            string inputHTML = input.Trim();
            StringBuilder outHTML = new StringBuilder();

            int maxLength = length;
            if (maxLength < inputHTML.Length)
            {
                inputHTML = inputHTML.Substring(0, 505);

                string firstMatch = "";
                string lastMatch = "";
                bool isRegexMatch = false;

                string patternMatchHTML = @"(<([a-z0-9]*)([^>]*)>(.*)</s*//s*/2>)|(<([a-z0-9]*)([^>]*)//s*>)";

                Regex regex = new Regex(patternMatchHTML, RegexOptions.IgnoreCase);
                Match m = regex.Match(inputHTML);
                while (m.Success)
                {
                    isRegexMatch = true;
                    if (firstMatch == "")
                        firstMatch = m.Value;
                    lastMatch = m.Value;
                    outHTML.Append(m.Value);

                    m = m.NextMatch();

                }
                if (isRegexMatch)
                {
                    string beforeString = "";
                    if (firstMatch != "")
                    {
                        beforeString = inputHTML.Substring(0, inputHTML.IndexOf(firstMatch));
                    }
                    bool hasMeetingTag = false;
                    for (int i = beforeString.Length - 1; i >= 0; i--)
                    {
                        if (beforeString[i] == '>')
                        {
                            hasMeetingTag = true;
                        }
                        else if (beforeString[i] == '<')
                        {
                            hasMeetingTag = false;
                        }
                        else
                        {
                            if (!hasMeetingTag)
                                outHTML.Insert(0, beforeString[i]);
                        }
                    }

                    string lastString = "";
                    if (lastMatch != "")
                    {
                        lastString = inputHTML.Substring(inputHTML.LastIndexOf(lastMatch) + lastMatch.Length);
                    }
                    hasMeetingTag = false;
                    for (int i = 0; i < lastString.Length; i++)
                    {
                        if (lastString[i] == '<')
                        {
                            hasMeetingTag = true;
                        }
                        else if (lastString[i] == '>')
                        {
                            hasMeetingTag = false;
                        }
                        else
                        {
                            if (!hasMeetingTag)
                            {
                                outHTML.Append(lastString[i]);
                            }
                            else
                                outHTML.Append(inputHTML);
                        }
                        return outHTML.ToString().Trim();

                    }
                }
            }
        }

抱歉!评论已关闭.