private void GetPersonPageUrls(string localDir, string listPageUrl, List <Dictionary <string, string> > allPersonPageUrlInfos, string keyWords)
{
try
{
string listPageLocalPath = this.RunPage.GetFilePath(listPageUrl, localDir);
HtmlAgilityPack.HtmlDocument pageHtmlDoc = HtmlDocumentHelper.Load(listPageLocalPath);
HtmlNodeCollection allDivNodes = pageHtmlDoc.DocumentNode.SelectNodes("//div[@class=\"result c-container \"]");
foreach (HtmlNode divNode in allDivNodes)
{
string linkedinUrlPart = HtmlDocumentHelper.TryGetNodeInnerText(divNode, "./div[@class=\"f13\"]/a", true, true, null, null);
if (linkedinUrlPart == null)
{
linkedinUrlPart = HtmlDocumentHelper.TryGetNodeInnerText(divNode, "./div/div[@class=\"f13\"]/a", true, true, null, null);
}
string abstractText = HtmlDocumentHelper.TryGetNodeInnerText(divNode, true, true, null, null);
if (linkedinUrlPart != null && linkedinUrlPart.Contains(".linkedin.com/in/") && abstractText != null && abstractText.ToLower().Contains(keyWords.ToLower()))
{
try
{
string personName = HtmlDocumentHelper.TryGetNodeInnerText(divNode, "./h3/a", true, true, null, null);
string personUrl = HtmlDocumentHelper.TryGetNodeAttributeValue(divNode, "./h3/a", "href", true, true, null, null);
foreach (string postfix in this.BaiduLinkedinItemPostfix)
{
personName = personName.Replace(postfix, "").Trim();
}
Dictionary <string, string> personPageUrlInfo = new Dictionary <string, string>();
personPageUrlInfo.Add("personUrl", personUrl);
personPageUrlInfo.Add("personName", personName.Trim());
allPersonPageUrlInfos.Add(personPageUrlInfo);
}
catch (Exception ex)
{
throw new Exception("获取个人网页地址时出错", ex);
}
}
}
}
catch (Exception ex)
{
this.RunPage.InvokeAppendLogText(ex.Message, LogLevelType.Error, true);
throw new Exception("解析Baidu列表页出错, listPageUrl = +" + listPageUrl, ex);
}
}