public void ParseEntity()
{
var service = SpiderFactory.CreateScopeServiceProvider();
var dataContext = new DataFlowContext(new Response
{
Request = new Request("https://list.jd.com/list.html?cat=9987,653,655",
new Dictionary <string, string>
{
{ "cat", "手机" },
{ "cat3", "110" }
}),
RawText = File.ReadAllText("Jd.html")
}, service);
DataParser <Product> extractor = new DataParser <Product>();
extractor.HandleAsync(dataContext).GetAwaiter().GetResult();
var results = (ParseResult <Product>)dataContext.GetParseItem(typeof(Product).FullName);
Assert.Equal(60, results.Count);
Assert.Equal("手机", results[0].CategoryName);
Assert.Equal(110, results[0].CategoryId);
Assert.Equal("https://item.jd.com/3031737.html", results[0].Url);
Assert.Equal("3031737", results[0].Sku);
Assert.Equal("荣耀官方旗舰店", results[0].ShopName);
Assert.Equal("荣耀 NOTE 8 4GB+32GB 全网通版 冰河银", results[0].Name);
Assert.Equal("1000000904", results[0].VenderId);
Assert.Equal("1000000904", results[0].JdzyShopId);
Assert.Equal(DateTime.Now.ToString("yyyy-MM-dd"), results[0].RunId.ToString("yyyy-MM-dd"));
}
public Task <DataFlowResult> HandleAsync(DataFlowContext context)
{
int result = 0;
IEnumerator enumerator = context.GetParseItem(typeof(IndexEntity).FullName).GetEnumerator();
while (enumerator.MoveNext())
{
result = SaveDealIndex((IndexEntity)enumerator.Current);
if (result == -1)
{
break;
}
}
if (result == -1)
{
IEventBus bus = (IEventBus)context.Services.GetService(typeof(IEventBus));
bus.Publish(context.Response.Request.OwnerId, new Event()
{
Type = Framework.ExitCommand
});
return(Task.FromResult(DataFlowResult.Terminated));
}
else
{
return(Task.FromResult(DataFlowResult.Success));
}
}
public async Task <DataFlowResult> HandleAsync(DataFlowContext context)
{
IEnumerator enumerator = context.GetParseItem(typeof(InfoEntity).FullName).GetEnumerator();
while (enumerator.MoveNext())
{
var info = (InfoEntity)enumerator.Current;
var gps = await GetGPSAsync(info.Community);
info.Latitude = gps.Item1;
info.Longtitude = gps.Item2;
SaveDealInfo((InfoEntity)enumerator.Current);
Logger.LogInformation($"community: {info.Community}\tlat: {gps.Item1}\t lng: {gps.Item2}");
}
return(DataFlowResult.Success);
}
public void MultiEntitySelector()
{
var service = SpiderFactory.CreateScopeServiceProvider();
var dataContext = new DataFlowContext(new Response
{
Request = new Request("http://abcd.com"),
RawText = Html
}, service);
var parser = new DataParser <E>();
parser.HandleAsync(dataContext).GetAwaiter().GetResult();
var results = (ParseResult <E>)dataContext.GetParseItem(typeof(E).FullName);
Assert.Equal("a", results[0].title);
Assert.Equal("b", results[1].title);
}
public void SingleEntitySelector()
{
var service = SpiderProvider.Value.CreateScopeServiceProvider();
var dataContext = new DataFlowContext(new Response
{
Request = new Request("http://abcd.com"),
RawText = Html
}, service);
var parser = new DataParser <N>();
parser.HandleAsync(dataContext).GetAwaiter().GetResult();
var results = (ParseResult <N>)dataContext.GetParseItem(typeof(N).FullName);
Assert.Equal("i am title", results[0].title);
Assert.Equal("i am dotnetspider", results[0].dotnetspider);
}
protected override Task <DataFlowResult> Parse(DataFlowContext context)
{
if (!context.Contains(_model.TypeName))
{
context.Add(_model.TypeName, _tableMetadata);
}
var selectable = context.GetSelectable();
var results = new ParseResult <T>();
if (selectable.Properties == null)
{
selectable.Properties = new Dictionary <string, object>();
}
var environments = new Dictionary <string, string>();
foreach (var property in context.Response.Request.Properties)
{
environments.Add(property.Key, property.Value);
}
if (_model.ShareValueSelectors != null)
{
foreach (var selector in _model.ShareValueSelectors)
{
string name = selector.Name;
var value = selectable.Select(selector.ToSelector()).GetValue();
if (!environments.ContainsKey(name))
{
environments.Add(name, value);
}
else
{
environments[name] = value;
}
}
}
bool singleExtractor = _model.Selector == null;
if (!singleExtractor)
{
var selector = _model.Selector.ToSelector();
var list = selectable.SelectList(selector).Nodes()?.ToList();
if (list != null)
{
if (_model.Take > 0 && list.Count > _model.Take)
{
list = _model.TakeFromHead
? list.Take(_model.Take).ToList()
: list.Skip(list.Count - _model.Take).ToList();
}
for (var i = 0; i < list.Count; ++i)
{
var item = list.ElementAt(i);
var obj = ParseObject(environments, item, i);
if (obj != null)
{
results.Add(obj);
}
else
{
Logger?.LogWarning($"解析到空数据，类型: {_model.TypeName}");
}
}
}
}
else
{
var obj = ParseObject(environments, selectable, 0);
if (obj != null)
{
results.Add(obj);
}
else
{
Logger?.LogWarning($"解析到空数据，类型: {_model.TypeName}");
}
}
if (results.Count > 0)
{
var items = context.GetParseItem(_model.TypeName);
if (items == null)
{
context.AddParseItem(_model.TypeName, results);
}
else
{
((ParseResult <T>)items).AddRange(results);
}
}
return(Task.FromResult(DataFlowResult.Success));
}
