protected override Task <DataFlowResult> Store(DataFlowContext context)
{
var items = context.GetData();
Console.WriteLine(JsonConvert.SerializeObject(items));
return(Task.FromResult(DataFlowResult.Success));
}
public override async Task HandleAsync(DataFlowContext context)
{
if (IsNullOrEmpty(context))
{
Logger.LogWarning("数据流上下文不包含实体解析结果");
return;
}
var data = context.GetData();
var result = new Dictionary <Type, ICollection <dynamic> >();
foreach (var kv in data)
{
var type = kv.Key as Type;
if (type == null || !_baseType.IsAssignableFrom(type))
{
continue;
}
if (kv.Value is IEnumerable list)
{
foreach (var obj in list)
{
AddResult(result, type, obj);
}
}
else
{
AddResult(result, type, kv.Value);
}
}
await HandleAsync(context, result);
}
protected override async Task <DataFlowResult> Store(DataFlowContext context)
{
var items = context.GetData();
foreach (var item in items)
{
var tableMetadata = (TableMetadata)context[item.Key];
if (!_cache.ContainsKey(tableMetadata.Schema.Database))
{
_cache.TryAdd(tableMetadata.Schema.Database, _client.GetDatabase(tableMetadata.Schema.Database));
}
var db = _cache[tableMetadata.Schema.Database];
var collection = db.GetCollection <BsonDocument>(tableMetadata.Schema.Table);
var bsonDocs = new List <BsonDocument>();
foreach (var data in item.Value)
{
bsonDocs.Add(BsonDocument.Create(data));
}
await collection.InsertManyAsync(bsonDocs);
}
return(DataFlowResult.Success);
}
public override async Task HandleAsync(DataFlowContext context)
{
if (context.IsEmpty)
{
Logger.LogWarning("数据流上下文不包含实体解析结果");
return;
}
var data = context.GetData();
var dict = new Dictionary <Type, List <dynamic> >();
foreach (var d in data)
{
var type = d.Key as Type;
if (type == null || !_baseType.IsAssignableFrom(type))
{
continue;
}
if (d.Value is IEnumerable list)
{
foreach (var obj in list)
{
InsertData(dict, type, obj);
}
}
else
{
InsertData(dict, type, d.Value);
}
}
await StoreAsync(context, dict);
}
protected override Task StoreAsync(DataFlowContext context)
{
var items = context.GetData().Where(x => !ReferenceEquals(x.Key, Consts.ResponseBytes));
Console.ForegroundColor = ConsoleColor.Cyan;
Console.WriteLine($"{Environment.NewLine}DATA: {JsonConvert.SerializeObject(items)}{Environment.NewLine}");
return(Task.CompletedTask);
}
protected override async Task StoreAsync(DataFlowContext context)
{
var file = Path.Combine(GetDataFolder(context.Request.Owner),
$"{context.Request.Hash}.json");
using var writer = OpenWrite(file);
var items = context.GetData();
await writer.WriteLineAsync(JsonConvert.SerializeObject(items));
}
protected override Task StoreAsync(DataFlowContext context)
{
var items = context.GetData();
Console.ForegroundColor = ConsoleColor.Cyan;
Console.WriteLine($"{Environment.NewLine}DATA: {JsonConvert.SerializeObject(items)}{Environment.NewLine}");
return(Task.CompletedTask);
}
protected override async Task StoreAsync(DataFlowContext context)
{
var file = Path.Combine(GetDataFolder(context.Request.Owner),
$"{context.Request.Hash}.json");
using var writer = OpenWrite(file);
var items = context.GetData().Where(x => !ReferenceEquals(x.Key, Consts.ResponseBytes));
await writer.WriteLineAsync(JsonConvert.SerializeObject(items));
}
protected override async Task <DataFlowResult> Store(DataFlowContext context)
{
var items = context.GetData();
var file = Path.Combine(GetDataFolder(context.Response.Request.OwnerId),
$"{context.Response.Request.Hash}.json");
CreateFile(file);
await Writer.WriteLineAsync(JsonConvert.SerializeObject(items));
return(DataFlowResult.Success);
}
protected override Task StoreAsync(DataFlowContext context)
{
var typeName = typeof(News).FullName;
var data = context.GetData(typeName);
if (data is News news)
{
Console.WriteLine($"URL: {news.Url}, TITLE: {news.Title}, VIEWS: {news.Views}");
}
return(Task.CompletedTask);
}
protected override Task <DataFlowResult> Store(DataFlowContext context)
{
var items = context.GetData();
foreach (var item in items)
{
foreach (var data in item.Value)
{
Console.WriteLine(JsonConvert.SerializeObject(data));
}
}
return(Task.FromResult(DataFlowResult.Success));
}
protected override async Task StoreAsync(DataFlowContext context)
{
var typeName = typeof(News).FullName;
var data = (News)context.GetData(typeName);
if (data != null)
{
await using var conn =
new MySqlConnection(
"Database='mysql';Data Source=localhost;password=1qazZAQ!;User ID=root;Port=3306;");
await conn.ExecuteAsync(
$"INSERT IGNORE INTO cnblogs2.news2 (title, url, summary, views, content) VALUES (@Title, @Url, @Summary, @Views, @Content);",
data);
}
}
public async Task MultiEntitySelector()
{
var request = new Request("http://abcd.com");
var dataContext =
new DataFlowContext(null, new SpiderOptions(), request,
new Response {
Content = new ByteArrayContent(Encoding.UTF8.GetBytes(Html))
});
var parser = new DataParser <E>();
await parser.HandleAsync(dataContext);
var results = (List <E>)dataContext.GetData(typeof(E));
Assert.Equal("a", results[0].title);
Assert.Equal("b", results[1].title);
}
public override Task HandleAsync(DataFlowContext context)
{
if (IsNullOrEmpty(context))
{
Logger.LogWarning("数据流上下文不包含解析结果");
return(Task.CompletedTask);
}
var typeName = typeof(News).FullName;
var data = context.GetData(typeName);
if (data is News news)
{
Console.WriteLine($"URL: {news.Url}, TITLE: {news.Title}, VIEWS: {news.Views}");
}
return(Task.CompletedTask);
}
protected override async Task StoreAsync(DataFlowContext context)
{
var id = context.Request.Owner;
var table = $"dotnet_spider:response_{id}";
_tableCreatedDict.GetOrAdd(table, t =>
{
EnsureDatabaseAndTableCreated(context, t);
return(true);
});
var hash = context.Request.Hash;
var bytes = context.GetData(Consts.ResponseBytes);
var data = Convert.ToBase64String(bytes);
var httpClient = context.ServiceProvider.GetRequiredService <IHttpClientFactory>().CreateClient(_rest);
for (var i = 0; i < 3; ++i)
{
try
{
var httpRequestMessage = new HttpRequestMessage(HttpMethod.Put, $"{_rest}{table}/row");
httpRequestMessage.Headers.TryAddWithoutValidation("Accept", "application/json");
var rowKey = hash.ToBase64String();
var body =
"{\"Row\":[{\"key\":\"" + rowKey +
"\", \"Cell\": [{\"column\":\"" + _columnName + "\", \"$\":\"" + data + "\"}]}]}";
var content =
new StringContent(body,
Encoding.UTF8, "application/json");
httpRequestMessage.Content = content;
var res = await httpClient.SendAsync(httpRequestMessage);
res.EnsureSuccessStatusCode();
}
catch (Exception ex)
{
Logger.LogError($"Store {context.Request.RequestUri} response to HBase failed [{i}]: {ex}");
}
}
}
protected virtual void AddParsedResult <T>(DataFlowContext context, IEnumerable <T> results)
where T : EntityBase <T>, new()
{
if (results != null)
{
var type = typeof(T);
var items = context.GetData(type);
if (items == null)
{
var list = new List <T>();
list.AddRange(results);
context.AddData(type, list);
}
else
{
items.AddRange(results);
}
}
}
public async Task SingleEntitySelector()
{
var request = new Request("http://abcd.com");
var dataContext =
new DataFlowContext(null, new SpiderOptions(), request,
new Response {
Content = new ByteArrayContent(Encoding.UTF8.GetBytes(Html))
});
var parser = new DataParser <N>();
parser.UseHtmlSelectableBuilder();
await parser.HandleAsync(dataContext);
var results = (List <N>)dataContext.GetData(typeof(N));
Assert.Equal("i am title", results[0].title);
Assert.Equal("i am dotnetspider", results[0].dotnetspider);
}
public override async Task HandleAsync(DataFlowContext context)
{
if (IsNullOrEmpty(context))
{
Logger.LogWarning("数据流上下文不包含解析结果");
return;
}
var typeName = typeof(News).FullName;
var data = (News)context.GetData(typeName);
if (data != null)
{
await using var conn =
new MySqlConnection(
"Database='mysql';Data Source=localhost;password=1qazZAQ!;User ID=root;Port=3306;");
await conn.ExecuteAsync(
$"INSERT IGNORE INTO cnblogs2.news2 (title, url, summary, views, content) VALUES (@Title, @Url, @Summary, @Views, @Content);",
data);
}
}
public async Task ParseEntity()
{
var request = new Request("https://list.jd.com/list.html?cat=9987,653,655",
new Dictionary <string, object> {
{ "cat", "手机" }, { "cat3", "110" }
});
var dataContext = new DataFlowContext(null, new SpiderOptions(), request,
new Response {
Content = new ByteArrayContent(File.ReadAllBytes("Jd.html"))
});
var parser = new DataParser <Product>();
await parser.InitializeAsync();
parser.UseHtmlSelectableBuilder();
await parser.HandleAsync(dataContext);
var results = (List <Product>)dataContext.GetData(typeof(Product));
Assert.Equal(60, results.Count);
Assert.Contains("手机商品筛选", results[0].Title);
Assert.Contains("手机商品筛选", results[1].Title);
Assert.Contains("手机商品筛选", results[2].Title);
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(DateTimeOffset.Now.ToString("yyyy-MM-dd"), results[0].RunId.ToString("yyyy-MM-dd"));
var requests = dataContext.FollowRequests;
Assert.Equal(7, requests.Count);
}