public override string ToString()
{
StringBuilder tag = new StringBuilder(GetStartTag());
tag.Append(GetAttributeTag());
tag.Append(_Element.ToString());
tag.Append(GetEndTag());
return(tag.ToString());
}
private ExamItem BuildEntity(int moduleId, IHtmlElement item)
{
string selector = @"td[width]";
if (false == item.Exists(selector))
{
return(null);
}
var id_element = item.FindFirst(selector);
if (id_element == null)
{
return(null);
}
try
{
var match = Regex.Match(id_element.InnerText(), @"\d+");
int id = Convert.ToInt32(match.Value.ToString().Trim());
string title = id_element.Parent().FindFirst(@".MsoNormal>span").InnerText().Trim().RemoveHtml().RemoveHtmlEncode();
string answer = item.FindFirst("#answer").InnerText().Trim();
string strExamType = item.FindFirst(".st_title").InnerText();
string examType = Regex.Match(strExamType, @"(?<=、).*").Value.Trim();
if (answer.StartsWith("答案："))
{
answer = answer.Substring(3).Trim();
}
var itemType = ItemTypeService.GetByText(examType);
if (itemType == null)
{
logger.Info(string.Format("未匹配的题目类别[{0}],ExamId=[{1}]", examType, id));
}
var model = new ExamItem()
{
Id = id,
Title = title,
Answer = answer,
OriginalHtml = item.InnerHtml(),
Module_Id = moduleId,
ItemType = itemType != null ? itemType.Id : 0,
CreateTime = DateTime.Now,
DelFlag = false
};
return(model);
}
catch (Exception ex)
{
WriteLog(item.ToString(), ex.Message);
throw;
}
}
public override string ToString()
{
StringBuilder tag = new StringBuilder("<caption");
if (_Align != null)
{
tag.Append(" align=\"");
tag.Append(_Align);
tag.Append("\"");
}
tag.Append(GetAttributeString());
tag.Append(">");
tag.Append(_Element.ToString());
tag.Append("</caption>\n");
return(tag.ToString());
}
private ExamItem BuildEntity(int moduleId, IHtmlElement item)
{
try
{
var id_element = item.FindFirst(@"tr[valign]>td");
var match = Regex.Match(id_element.InnerText(), @"\d+");
int id = Convert.ToInt32(match.Value.Trim());
string title = id_element.NextElement().InnerText().Trim().RemoveHtml().RemoveHtmlEncode();
string answer = item.FindFirst("div[id]").InnerText().Trim().RemoveHtmlEncode();
string examType = item.PreviousElement().Descendants("td")
.ElementAt(1)
.InnerText().Trim().RemoveHtml().RemoveHtmlEncode();
if (answer.StartsWith("答案："))
{
answer = answer.Substring(3).Trim();
}
var itemType = ItemTypeService.GetByText(examType);
if (itemType == null)
{
logger.Info(string.Format("未匹配的题目类别[{0}],ExamId=[{1}]", examType, id));
}
var model = new ExamItem()
{
Id = id,
Title = title,
Answer = answer,
OriginalHtml = item.InnerHtml(),
Module_Id = moduleId,
ItemType = itemType != null ? itemType.Id : 0,
CreateTime = DateTime.Now,
DelFlag = false
};
return(model);
}
catch (Exception ex)
{
WriteLog(item.ToString(), ex.Message);
throw;
}
}
private void GetUrlText_2(IHtmlDocument document, List <Class1> L_Class)
{
IEnumerable <IHtmlElement> result1 = document.Find(".list-items");
foreach (var item in result1)
{
#region MyRegion
Class1 _class = new Class1();
IHtmlElement item_a = item.FindFirst("a");
string img_str = item.Exists("img") ? item.FindFirst("img").Attribute("src").Value() : "";
_class.Image_Count = img_str.Length > 0 ? img_str.IndexOf("default.jpg") > 0?0:1:0;
IEnumerable <IHtmlElement> div = item.Find("div");
List <string> ls = new List <string>();
foreach (var d in div)
{
ls.Add(d.InnerText());
}
_class.TextName = ls[2];
_class.SumMoney = ls[4];
_class.Quyu = "赶集";
_class.Allpm = ls[1];
_class.Address = ls[0];
_class.href = item_a.Attribute("href").Value().Trim();
IEnumerable <IHtmlElement> ssa = item_a.Find("span");
string item_aa = item_a.ToString().Replace("<!--", "stu1").Replace("-->", "stp2");
item_aa = item_aa.Substring(item_aa.IndexOf("stu1") + 4);
item_aa = item_aa.Substring(0, item_aa.IndexOf("stp2"));
item_aa = item_aa.Substring(item_aa.IndexOf(">") + 1);
item_aa = item_aa.Substring(0, item_aa.IndexOf("<"));
_class.datetime = item_aa;
string[] pm = _class.Allpm.Split(' ');
double pm_int = Convert.ToDouble(pm[6].Replace('㎡', ' ').Trim().Length <= 0 ? pm[5].Replace('㎡', ' ').Trim() : pm[6].Replace('㎡', ' ').Trim());
double ss = ((Convert.ToDouble(_class.SumMoney.Replace("万元", "").Trim()) / pm_int));
_class.PingMoney = "≈" + Convert.ToInt32(ss * 10000).ToString();
_class.Image_str = _class.Image_Count > 0 ? "有" : string.Empty;
L_Class.Add(_class);
r++;
#endregion
}
}
private ExamItem BuildEntity(int moduleId, IHtmlElement item)
{
try
{
var id_element = item.FindFirst(@"tr[valign]>td");
var match = Regex.Match(id_element.InnerText(), @"\d+");
int id = Convert.ToInt32(match.Value.Trim());
string title = id_element.NextElement().InnerText().Trim().RemoveHtml().RemoveHtmlEncode();
string answer = item.FindFirst("div[id]").InnerText().Trim().RemoveHtmlEncode();
string examType = item.PreviousElement().Descendants("td")
.ElementAt(1)
.InnerText().Trim().RemoveHtml().RemoveHtmlEncode();
if (answer.StartsWith("答案："))
{
answer = answer.Substring(3).Trim();
}
var itemType = ItemTypeService.GetByText(examType);
if (itemType == null)
{
logger.Info(string.Format("未匹配的题目类别[{0}],ExamId=[{1}]", examType, id));
}
var model = new ExamItem()
{
Id = id,
Title = title,
Answer = answer,
OriginalHtml = item.InnerHtml(),
Module_Id = moduleId,
ItemType = itemType != null ? itemType.Id : 0,
CreateTime = DateTime.Now,
DelFlag = false
};
return model;
}
catch (Exception ex)
{
WriteLog(item.ToString(), ex.Message);
throw;
}
}
public override string ToString()
{
return(rootElement.ToString());
}
/// <summary>
/// Gets the addresses for .png and .jpg files from "rg_meta" class "div" tags json.
/// This is the way a top level Google Image search provides images.
/// </summary>
/// <param name="tag">The parsed HTML document tag.</param>
private void GetImagesFromDivTags(HtmlTag tag)
{
try
{
// find looks recursively through the entire DOM tree
IEnumerable <HtmlTag> inputFields = tag.Find(t => string.Equals(t.TagName, "div"));
foreach (HtmlTag inputField in inputFields.Where(inputField => inputField.ContainsKey("class")))
{
try
{
string divClass = inputField["class"];
if (divClass == "rg_meta")
{
if (inputField.Contents == null || !inputField.Contents.Any())
{
continue;
}
IHtmlElement contents = inputField.Contents.First();
Dictionary <string, string> jsonValues =
JsonConvert.DeserializeObject <Dictionary <string, string> >(contents.ToString());
if (!jsonValues.ContainsKey("ity"))
{
continue;
}
if (jsonValues["ity"] != "jpg" && jsonValues["ity"] != "png")
{
continue;
}
if (jsonValues.ContainsKey("ou"))
{
_imagesInPage.Add(jsonValues["ou"]);
}
}
}
catch (Exception)
{
// ignored
}
}
}
catch (Exception)
{
// ignored
}
}
private ExamItem BuildEntity(int moduleId, IHtmlElement item)
{
string selector = @"td[width]";
if (false == item.Exists(selector)) { return null; }
var id_element = item.FindFirst(selector);
if (id_element == null) { return null; }
try
{
var match = Regex.Match(id_element.InnerText(), @"\d+");
int id = Convert.ToInt32(match.Value.ToString().Trim());
string title = id_element.Parent().FindFirst(@".MsoNormal>span").InnerText().Trim().RemoveHtml().RemoveHtmlEncode();
string answer = item.FindFirst("#answer").InnerText().Trim();
string strExamType = item.FindFirst(".st_title").InnerText();
string examType = Regex.Match(strExamType, @"(?<=、).*").Value.Trim();
if (answer.StartsWith("答案："))
{
answer = answer.Substring(3).Trim();
}
var itemType = ItemTypeService.GetByText(examType);
if (itemType == null)
{
logger.Info(string.Format("未匹配的题目类别[{0}],ExamId=[{1}]", examType, id));
}
var model = new ExamItem()
{
Id = id,
Title = title,
Answer = answer,
OriginalHtml = item.InnerHtml(),
Module_Id = moduleId,
ItemType = itemType != null ? itemType.Id : 0,
CreateTime = DateTime.Now,
DelFlag = false
};
return model;
}
catch (Exception ex)
{
WriteLog(item.ToString(), ex.Message);
throw;
}
}
