/// <summary>
/// 在百度知道查询答案。
/// </summary>
/// <param name="sentence">要查询的句子</param>
/// <param name="num">获取的答案数</param>
/// <returns></returns>
public static string[] getBaiduZhidaoAnswers(string sentence, int num = 10)
{
string url = string.Format("https://zhidao.baidu.com/search?word={0}", WebConnectHelper.UrlEncode(sentence));
List <string> res = new List <string>();
string html = WebConnectHelper.getData(url, Encoding.GetEncoding("gb2312"));
HtmlDocument hdoc = new HtmlDocument();
hdoc.LoadHtml(html);
try
{
var urls = hdoc.DocumentNode.SelectNodes("//a[@class=\"ti\"]");
foreach (var aurl in urls)
{
string dw = ItemParser.removeBlank(aurl.GetAttributeValue("href", ""), true);
var areslist = getBaiduZhidaoAnswersByUrl(dw);
if (areslist.Length > 0)
{
res.Add(areslist[0]);
}
// 暂时只查第一个
if (res.Count > 0)
{
break;
}
}
}
catch { }
return(res.ToArray());
}
/// <summary>
/// 从html文档中找出文本部分
/// </summary>
/// <param name="html"></param>
/// <returns></returns>
public static string[] getText(string html)
{
if (string.IsNullOrWhiteSpace(html) || html[0] != '<')
{
return(new string[] { ItemParser.removeUnText(html) });
}
List <string> res = new List <string>();
HtmlDocument hdoc = new HtmlDocument();
hdoc.LoadHtml(html);
foreach (var node in hdoc.DocumentNode.ChildNodes)
{
if (node.NodeType == HtmlNodeType.Text)
{
res.Add(ItemParser.removeUnText(node.InnerText));
}
//else if (node.Name == "br")
//{
// res.Add("\r\n");
//}
else if (node.NodeType == HtmlNodeType.Element)
{
var tmp = getText(node.InnerHtml);
foreach (var t in tmp)
{
res.Add(t);
}
}
}
return(res.ToArray());
}
/// <summary>
/// 从百度知识图谱中寻找答案
/// 基本就是把关键词放入百度搜索，然后看百度有没有智能返回的结果
/// </summary>
/// <param name="words"></param>
/// <returns></returns>
public static string[] getBaiduKGResult(string words)
{
List <string> reslist = new List <string>();
string askUrl = "http://www.baidu.com/s?wd=" + WebConnectHelper.UrlEncode(words);
string html = WebConnectHelper.getData(askUrl, Encoding.UTF8);
HtmlDocument hdoc = new HtmlDocument();
hdoc.LoadHtml(html);
//统计数值
try
{
string gdp = ItemParser.removeUnText(hdoc.DocumentNode.SelectSingleNode("//*[@class=\"op_gdp_subtitle\"]").InnerText);
reslist.Add(gdp);
}
catch { }
//图谱常识
try
{
string common = ItemParser.removeUnText(hdoc.DocumentNode.SelectSingleNode("//*[@class=\"op_exactqa_s_answer\"]").InnerText);
reslist.Add(common);
}
catch { }
//股票
try
{
string gp = ItemParser.removeUnText(hdoc.DocumentNode.SelectSingleNode("//*[@class=\"op_stockweakdemand_cur_num c-gap - right - small\"]").InnerText);
reslist.Add(gp);
string gpzf = ItemParser.removeUnText(hdoc.DocumentNode.SelectSingleNode("//*[@class=\"op_stockweakdemand_cur_info c-gap-icon-right-small\"]").InnerText);
reslist.Add(gpzf);
}
catch { }
//热线电话
try
{
string rx = ItemParser.removeUnText(hdoc.DocumentNode.SelectSingleNode("//*[@class=\"op_kefupoly_td2\"]").InnerText);
reslist.Add(rx);
}
catch { }
//翻译
try
{
string trans = ItemParser.removeUnText(hdoc.DocumentNode.SelectSingleNode("//*[@class=\"op_sp_fanyi_line_two\"]").InnerText);
reslist.Add(trans);
}
catch { }
//数学运算
try
{
string trans = ItemParser.removeUnText(hdoc.DocumentNode.SelectSingleNode("//*[@style=\"font-size:1.4em;line-height:22px;padding-bottom:2px;width:474px;\"]").InnerText);
reslist.Add(trans);
}
catch { }
//汇率换算
try
{
string hl = ItemParser.removeUnText(hdoc.DocumentNode.SelectSingleNode("//*[@class=\"op_exrate_result\"]").InnerText);
reslist.Add(hl);
}
catch { }
//单位换算
try
{
string dw = ItemParser.removeUnText(hdoc.DocumentNode.SelectSingleNode("//*[@class=\"op-unit-result-single c-gap-left c-gap-right\"]").InnerText);
reslist.Add(dw);
}
catch { }
//邮编
try
{
string dw = ItemParser.removeUnText(hdoc.DocumentNode.SelectSingleNode("//*[@class=\"op_post_content \"]").InnerText);
reslist.Add(dw);
}
catch { }
// 百度知道最佳答案
// 去所跳转的页面上找答案
try
{
string dw = ItemParser.removeUnText(hdoc.DocumentNode.SelectSingleNode("//*[@class=\"op_best_answer_question_link\"]").GetAttributeValue("href", ""));
var answers = getBaiduZhidaoAnswers(dw)[0];
reslist.Add(answers);
}
catch { }
// 百度知道的推荐答案的第一个
// 去所跳转的页面上找答案
try
{
string dw = ItemParser.removeUnText(hdoc.DocumentNode.SelectSingleNode("//*[@class=\"op_generalqa_answer c-gap-bottom-small op_generalqa_answer_first\"]").ChildNodes[3].FirstChild.GetAttributeValue("href", ""));
var answers = getBaiduZhidaoAnswersByUrl(dw)[0];
reslist.Add(answers);
}
catch { }
// 百度日历上的日子
// 由于它是js生成的日历，所以需要从js里正则匹配一下
try
{
Regex reg = new Regex("\"selectday\":\"([^\"]*?)\"", RegexOptions.None);
string date = reg.Match(html).Groups[1].Value;
if (!string.IsNullOrWhiteSpace(date))
{
reslist.Add(date);
}
}
catch { }
return(reslist.ToArray());
}