/// <summary>
/// 对html进行解析，得到每个信息的日期、时间、昵称、QQ号、email和内容。使用HtmlAgilityPack库解析html
/// </summary>
private void parseHtmlToMessages()
{
string date = null;
var doc = new HtmlAgilityPack.HtmlDocument();
doc.Load(tmpDir + "/index.html", new System.Text.UTF8Encoding());
foreach (HtmlAgilityPack.HtmlNode td in doc.DocumentNode.SelectNodes("//td")) {
// <tr><td style=border-bottom-width:1px;border-bottom-color:#8EC3EB;border-bottom-style:solid;color:#3568BB;font-weight:bold;height:24px;line-height:24px;padding-left:10px;margin-bottom:5px;>日期: 2012/4/11</td></tr>
var style = td.Attributes["style"];
if (style != null && style.Value != null && td.InnerText != null && td.InnerText.StartsWith("日期")) {
date = td.InnerText.SubstringAfter("日期: ");
}
// <tr><td><div style=color:#006EFE;padding-left:10px;><div style=float:left;margin-right:6px;>大魔头<[email protected]></div>12:28:18</div><div style=padding-left:20px;><font style="font-size:16pt;font-family:'华文楷体','MS Sans Serif',sans-serif;" color='000000'>看字节码干嘛。。。</font></div></td></tr>
// <tr><td><div style=color:#006EFE;padding-left:10px;><div style=float:left;margin-right:6px;>519870018(519870018)</div>12:34:43</div><div style=padding-left:20px;><IMG src="{3D48C238-CD47-4d17-9C8F-3593C6D4738B}.dat"></div></td></tr>
// <tr><td><div style=color:#42B475;padding-left:10px;><div style=float:left;margin-right:6px;>风自由(23246779)</div>20:11:16</div><div style=padding-left:20px;><font style="font-size:10pt;font-family:'微软雅黑','MS Sans Serif',sans-serif;" color='000000'>就我们两</font></div></td></tr>
else {
var firstDiv = td.FirstChild;
string msgStyle = null;
if (firstDiv != null) {
var styleAttr = firstDiv.Attributes["style"];
if (styleAttr != null) {
msgStyle = styleAttr.Value;
}
}
if (msgStyle != null && msgStyle.Contains("color") && msgStyle.Contains("padding-left:10px;")) {
QQMessage message = new QQMessage();
var name = td.FirstChild.FirstChild.InnerText.Replace("<", "<").Replace(">", ">").Replace("&", "&").Replace(" ", " ");
if (name.EndsWith(">")) {
message.nickname = name.SubstringBefore("<");
message.qqEmail = name.SubstringBetween("<", ">");
} else {
message.nickname = name.SubstringBefore("(");
message.qqNumber = name.SubstringBetween("(", ")");
}
// 日期：2012/4/11
var culture = CultureInfo.CurrentCulture;
message.date = Int32.Parse(DateTime.ParseExact(date, new string[] { "yyyy/M/d", "yyyy-M-d", "M/d/yyyy" }, culture, DateTimeStyles.None).ToString("yyyyMMdd"));
// 12:28:18
message.time = Int32.Parse(DateTime.ParseExact(td.FirstChild.LastChild.InnerText, new string[] { "H:m:s" }, culture, DateTimeStyles.None).ToString("HHmmss"));
message.content = td.LastChild.InnerHtml;
messages.Add(message);
}
}
}
}
public bool Equals(QQMessage another)
{
return (qqNumber == another.qqNumber || qqEmail == another.qqEmail) && date == another.date && time == another.time && content == another.content;
}
private bool checkMessage(List<QQMessage> messages, int fromIndex, QQMessage checkingMsg, out int index)
{
for (index = fromIndex; index < messages.Count; index++) {
var message = messages[index];
if (checkingMsg.date < message.date || (checkingMsg.date == message.date && checkingMsg.time < message.time)) {
break;
} else if (checkingMsg.Equals(message)) {
return false;
}
}
return true;
}