public XPathDetectorModel(IEnumerable <HtmlResult> htmlResults, ScriptWorkMode workmode, Window theView,
TextBox textbox)
{
HtmlDoc = new HtmlDocument();
var xpathHelper = new Dictionary <string, string>
{
{ "all_image", "//img[@src]" },
{ "all_item_with_id", @"//*[@id=""YOUR_ID""]" },
{ "all_item_with_class", @"//*[@class=""YOUR_CLASS""]" }
};
HtmlResults = htmlResults.ToList();
view = theView;
htmlTextBox = textbox;
XPath = new TextEditSelector();
XPath.SetSource(xpathHelper.Select(d => d.Value));
if (workmode == ScriptWorkMode.List)
{
ChildCount = 5;
}
else
{
ChildCount = 1;
}
CrawlItems = new ObservableCollection <CrawlItem>();
ChildItems = new ObservableCollection <CrawlItem>();
SelectedResult = HtmlResults.FirstOrDefault();
}
public MappingPair(List <string> source_keys, List <string> target_keys)
{
Source = new TextEditSelector();
Target = new TextEditSelector();
Source.SetSource(source_keys);
Target.SetSource(target_keys);
}
public ETLBase()
{
processManager = MainDescription.MainFrm.PluginDictionary["模块管理"] as IProcessManager;
ETLSelector = new TextEditSelector {
GetItems = this.GetAllETLNames()
};
ETLRange = "";
Column = "column";
Enabled = true;
}
public DbEX()
{
dataManager = MainDescription.MainFrm.PluginDictionary["DataManager"] as IDataManager;
ConnectorSelector = new ExtendSelector <IDataBaseConnector>();
ConnectorSelector.GetItems = () => dataManager.CurrentConnectors.ToList();
TableNames = new TextEditSelector();
ConnectorSelector.SelectChanged +=
(s, e) => TableNames.SetSource(ConnectorSelector.SelectItem.RefreshTableNames().Select(d => d.Name));
TableNames.SelectChanged += (s, e) => { InformPropertyChanged("TableNames"); };
}
public SaveFileEX()
{
CrawlerSelector = new TextEditSelector
{
GetItems = () =>
{
return
(processManager.CurrentProcessCollections.Where(d => d is SmartCrawler)
.Select(d => d.Name)
.ToList());
}
};
}
public SmartCrawler()
{
Http = new HttpItem();
CrawlItems = new ObservableCollection <CrawlItem>();
helper = new HttpHelper();
URL = "";
HtmlDoc = new HtmlDocument();
SelectText = "";
IsMultiData = ScriptWorkMode.List;
IsAttribute = true;
URL = "www.cnblogs.com";
ShareCookie = new TextEditSelector();
ShareCookie.GetItems = AppHelper.GetAllCrawlerNames(null);
Commands2 = CommandBuilder.GetCommands(
this,
new[]
{
new Command(GlobalHelper.Get("key_302"), obj => AddNewItem(),
obj =>
string.IsNullOrEmpty(SelectName) == false && string.IsNullOrEmpty(SelectXPath) == false,
"add"),
new Command(GlobalHelper.Get("search"), obj => GetXPathAsync(),
obj =>
currentXPaths != null, "magnify"),
new Command(GlobalHelper.Get("feellucky"),
obj => FeelLucky(),
obj => IsMultiData != ScriptWorkMode.NoTransform && isBusy == false, "smiley_happy"
),
new Command(GlobalHelper.Get("key_624"), obj =>
{
if (!(CrawlItems.Count > 0).SafeCheck(GlobalHelper.Get("key_625")))
{
return;
}
if (IsMultiData == ScriptWorkMode.List && CrawlItems.Count < 2)
{
MessageBox.Show(GlobalHelper.Get("key_626"), GlobalHelper.Get("key_99"));
return;
}
if (string.IsNullOrEmpty(this.URLHTML))
{
this.VisitUrlAsync();
}
var datas =
HtmlDoc.DocumentNode.GetDataFromXPath(CrawlItems, IsMultiData, RootXPath, RootFormat).Take(20)
.ToList();
var view = PluginProvider.GetObjectInstance <IDataViewer>(GlobalHelper.Get("key_230"));
var r = view.SetCurrentView(datas);
ControlExtended.DockableManager.AddDockAbleContent(
FrmState.Custom, r, GlobalHelper.Get("key_627"));
var rootPath =
XPath.GetMaxCompareXPath(CrawlItems.Select(d => d.XPath));
if (datas.Count > 0 && MessageBox.Show(GlobalHelper.Get("is_save_to_tables"), GlobalHelper.Get("key_99"), MessageBoxButton.YesNo) == MessageBoxResult.Yes)
{
SysDataManager.AddDataCollection(datas,
GlobalHelper.Get("key_624") + "_" + DateTime.Now.ToShortTimeString());
}
if (datas.Count > 1 && string.IsNullOrEmpty(RootXPath) && rootPath.Length > 0 &&
IsMultiData == ScriptWorkMode.List &&
MessageBox.Show(string.Format(GlobalHelper.Get("key_628"), rootPath), GlobalHelper.Get("key_99"),
MessageBoxButton.YesNo) == MessageBoxResult.Yes)
{
RootXPath = rootPath;
RootFormat = SelectorFormat.XPath;
HtmlDoc.CompileCrawItems(CrawlItems);
OnPropertyChanged("RootXPath");
}
}, icon: "page_search")
});
}
public SmartCrawler()
{
Http = new HttpItem();
CrawlItems = new ObservableCollection <CrawlItem>();
helper = new HttpHelper();
URL = "";
HtmlDoc = new HtmlDocument();
SelectText = "";
IsMultiData = ScriptWorkMode.List;
IsAttribute = true;
URL = "www.cnblogs.com";
ShareCookie = new TextEditSelector();
ShareCookie.GetItems = AppHelper.GetAllCrawlerNames(null);
Commands2 = CommandBuilder.GetCommands(
this,
new[]
{
new Command("添加", obj => AddNewItem(),
obj =>
string.IsNullOrEmpty(SelectName) == false && string.IsNullOrEmpty(SelectXPath) == false,
"add"),
new Command("搜索", obj => GetXPathAsync(),
obj =>
currentXPaths != null, "magnify"),
new Command("手气不错",
obj => FeelLucky(),
obj => IsMultiData != ScriptWorkMode.NoTransform && isBusy == false, "smiley_happy"
),
new Command("提取测试", obj =>
{
if (!(CrawlItems.Count > 0).SafeCheck("属性数量不能为空"))
{
return;
}
if (IsMultiData == ScriptWorkMode.List && CrawlItems.Count < 2)
{
MessageBox.Show("列表模式下，属性数量不能少于2个", "提示信息");
return;
}
if (string.IsNullOrEmpty(this.URLHTML))
{
this.VisitUrlAsync();
}
var datas =
HtmlDoc.DocumentNode.GetDataFromXPath(CrawlItems, IsMultiData, RootXPath, RootFormat).Take(20)
.ToList();
var view = PluginProvider.GetObjectInstance <IDataViewer>("可编辑列表");
var r = view.SetCurrentView(datas);
ControlExtended.DockableManager.AddDockAbleContent(
FrmState.Custom, r, "提取数据测试结果(显示前20条)");
var rootPath =
XPath.GetMaxCompareXPath(CrawlItems.Select(d => d.XPath));
if (datas.Count > 1 && string.IsNullOrEmpty(RootXPath) && rootPath.Length > 0 &&
IsMultiData == ScriptWorkMode.List &&
MessageBox.Show($"检测到列表的根节点为:{rootPath}，是否设置根节点路径？ 此操作有建议有经验用户使用，小白用户请点【否】", "提示信息",
MessageBoxButton.YesNo) == MessageBoxResult.Yes)
{
RootXPath = rootPath;
RootFormat = SelectorFormat.XPath;
HtmlDoc.CompileCrawItems(CrawlItems);
OnPropertyChanged("RootXPath");
}
}, icon: "page_search")
});
}
protected ToolBase()
{
ColumnSelector = new TextEditSelector();
ColumnSelector.SelectChanged += (s, e) => Column = ColumnSelector.SelectItem;
}
public ResponseTF()
{
CrawlerSelector = new TextEditSelector();
CrawlerSelector.GetItems = this.GetAllCrawlerNames();
}
public ResponseTF()
{
CrawlerSelector = new TextEditSelector();
CrawlerSelector.GetItems = this.GetAllCrawlerNames();
IsMultiYield = false;
}