private SerpModuleRequestStats FindRequestPosition(SerpModuleRequest request, IRuntimeModule <string, List <SerpWebPage> > dataProvider, string targetPage)
{
var requestStats = new SerpModuleRequestStats();
Context.LogInform(Context.Localizer[$"Analysis position of '{request.Text}' request..."]);
var result = RuntimeTask.Run(this.Context, dataProvider, request.Text);
if (result.IsSuccessfully)
{
var target = result.Data.Where(page => page.Url.Contains(targetPage.ToLower())).SingleOrDefault();
if (target != null)
{
requestStats.Request = request;
requestStats.Position = target.Position;
requestStats.DetectionTime = DateTime.Now;
Context.LogInform(Context.Localizer[$"Request '{request.Text}' found. Position : {target.Position}"]);
}
else
{
Context.LogInform(Context.Localizer[$"Request '{request.Text}' not found"]);
}
Context.LogInform(Context.Localizer[$"Analysis position of '{request.Text}' request was completed"]);
}
throw new UserException(this.Context.Localizer[RuntimeTask.RuntimeException, dataProvider.Name]);
}
private Stream FetchData(IRuntimeModule <HttpProviderRuntimeConfig, HttpProviderData> httpModule, string request, int?page)
{
if (httpModule == null)
{
throw new NullReferenceException("Http module");
}
this.Context.LogInform(this.Context.Localizer[$"Request execution '{request}'"]);
StringBuilder queryBuilder = new StringBuilder();
queryBuilder.Append($"?user={Configuration.User}&key={Configuration.Key}&query={request}");
if (!string.IsNullOrEmpty(Configuration.Region))
{
queryBuilder.Append($"&lr={Configuration.Region}");
}
if (!string.IsNullOrEmpty(Configuration.Filter))
{
queryBuilder.Append($"&filter={Configuration.Filter}");
}
if (page.HasValue && page.Value > 0)
{
queryBuilder.Append($"&page={page.Value}");
}
UriBuilder uri = new UriBuilder();
uri.Host = "yandex.ru/search/xml";
uri.Scheme = "http";
uri.Query = queryBuilder.ToString();
var result = RuntimeTask.Run(Context, httpModule, new HttpProviderRuntimeConfig()
{
Query = uri.ToString()
});
if (result.IsSuccessfully)
{
return(result.Data.GetContent());
}
throw new UserException(this.Context.Localizer[RuntimeTask.RuntimeException, httpModule.Name]);
}
public List <SerpWebPage> Run(string config)
{
List <SerpWebPage> pages = new List <SerpWebPage>();
int numberOfRequest = Configuration.PageLimit.HasValue ? Configuration.PageLimit.Value : 1;
var registry = this.Context.Provider.GetService <IProjectModuleRegistry>();
XmlYandexSerpParser yandexXMLParser = new XmlYandexSerpParser();
IRuntimeModule <HttpProviderRuntimeConfig, HttpProviderData> httpModule = null;
if (Configuration.HttpModuleId.HasValue)
{
var targetModule = Context.Repository.GetModule(Configuration.HttpModuleId.Value);
if (targetModule == null)
{
throw new NullReferenceException("Http Module module");
}
httpModule = registry.OpenModule(Context, targetModule.Code, targetModule.Id) as IRuntimeModule <HttpProviderRuntimeConfig, HttpProviderData>;
}
for (int i = 0; i < numberOfRequest; i++)
{
using (var stream = FetchData(httpModule, config, i))
{
var result = RuntimeTask.Run(Context, yandexXMLParser, stream);
if (result.IsSuccessfully)
{
pages.AddRange(result.Data);
}
else
{
throw new UserException(this.Context.Localizer[RuntimeTask.RuntimeException, yandexXMLParser.Name]);
}
}
}
return(pages);
}
public LexModuleStats Run(MedianFrequencyAnalyserConfig config)
{
List <LexicalTextAnalyzerStatistics> lexTextStatistics = new List <LexicalTextAnalyzerStatistics>();
var lexer = new HtmlLexer();
var lexTextAnalyzer = new LexicalTextAnalyzer();
foreach (var page in config.Pages.Take(10))
{
var httpResult = RuntimeTask.Run(Context, config.HttpModule, new HttpProviderRuntimeConfig()
{
Query = page.Url.ToString()
});
if (httpResult.IsSuccessfully)
{
using (var stream = httpResult.Data.GetContent())
{
var lexerResult = RuntimeTask.Run(Context, lexer, stream);
if (lexerResult.Details.Result != ActionExecutionResult.Error)
{
var lexTextAnalyzerResult = RuntimeTask.Run(Context, lexTextAnalyzer, new LexicalTextAnalyzerConfig()
{
Page = page, Tokens = lexerResult.Data, Filter = token => { return(token.Value.Count() > 3); }
});
if (lexTextAnalyzerResult.Details.Result != ActionExecutionResult.Error)
{
lexTextStatistics.Add(lexTextAnalyzerResult.Data);
}
}
}
}
}
return(Analyze(lexTextStatistics));
}
public AntigateCreateTaskResponse CreatTask(AntigateTaskConfig task, IRuntimeModule <HttpProviderRuntimeConfig, HttpProviderData> httpModule)
{
var creatTaskObject = new AntigateCreateTaskRequest(Configuration.ClientKey, Configuration.LanguagePool, Configuration.SoftId, Configuration.CallbackUrl);
creatTaskObject.Task = task;
HttpProviderRuntimeConfig httpConfig = new HttpProviderRuntimeConfig();
httpConfig.Query = "https://api.anti-captcha.com/createTask";
httpConfig.IsPost = true;
httpConfig.Content = new StringContent(SerializeObjectToJson(creatTaskObject), Encoding.UTF8, "application/json");
var result = RuntimeTask.Run(Context, httpModule, httpConfig);
if (result.Details.Result != ActionExecutionResult.Error)
{
using (var stream = result.Data.GetContent())
{
return(DeserializeJsonToObject <AntigateCreateTaskResponse>(stream));
}
}
return(null);
}
private void FillSerpPageDetails(IRuntimeModule <HttpProviderRuntimeConfig, HttpProviderData> httpModule, List <SerpWebPage> pages)
{
if (httpModule == null)
{
throw new NullReferenceException("httpModule");
}
foreach (var page in pages)
{
try
{
var serpUrl = new Uri(page.Url);
var targetBuilder = new UriBuilder(serpUrl.Scheme, serpUrl.Host, serpUrl.Port, "favicon.ico");
var result = RuntimeTask.Run(Context, httpModule, new HttpProviderRuntimeConfig()
{
Query = targetBuilder.Uri.ToString()
});
if (result.IsSuccessfully)
{
using (var strean = result.Data.GetContent())
{
using (MemoryStream ms = new MemoryStream())
{
strean.CopyTo(ms);
page.Favicon = ms.ToArray();
}
}
}
}
catch (Exception exc)
{
}
}
}
private List <YandexHtmlSerpItem> ExecuteRequest(string query, YandexHtmlSerpParser parser, IRuntimeModule <HttpProviderRuntimeConfig, HttpProviderData> httpModule, IRuntimeModule <AntigateTaskConfig, AntigateTaskResult> antigateModule)
{
List <YandexHtmlSerpItem> result = new List <YandexHtmlSerpItem>();
var httpResult = RuntimeTask.Run(Context, httpModule, new HttpProviderRuntimeConfig()
{
Query = query
});
if (httpResult.IsSuccessfully)
{
using (var stream = httpResult.Data.GetContent())
{
var parserResult = RuntimeTask.Run(Context, parser, stream);
if (!parserResult.IsSuccessfully)
{
throw parserResult.Error;
}
result.AddRange(parserResult.Data);
}
}
return(result);
}
private LexModuleStats Analyze(Request request)
{
if (Configuration.DataProvider == null)
{
throw new UserException("Data provider not specified");
}
var registry = this.Context.Provider.GetService <IProjectModuleRegistry>();
var module = registry.OpenModule(Context, Configuration.DataProvider.Code, Configuration.DataProvider.Id);
var dataProvider = module as IRuntimeModule <string, List <SerpWebPage> >;
if (dataProvider == null)
{
throw new UserException("Data provider is incorrected");
}
IRuntimeModule <HttpProviderRuntimeConfig, HttpProviderData> httpModule = null;
if (Configuration.HttpModuleId.HasValue)
{
var targetModule = Context.Repository.GetModule(Configuration.HttpModuleId.Value);
if (targetModule == null)
{
throw new NullReferenceException("Http Module module");
}
httpModule = registry.OpenModule(Context, targetModule.Code, targetModule.Id) as IRuntimeModule <HttpProviderRuntimeConfig, HttpProviderData>;
}
ModuleTaskResult <List <SerpWebPage> > dataResult;
var iteration = 0;
do
{
iteration++;
if (iteration > 1)
{
Context.LogInform(Context.Localizer[$"The attemp {iteration} to get position of '{request.Text}'"]);
}
dataResult = RuntimeTask.Run(this.Context, dataProvider, request.Text);
if (dataResult.IsSuccessfully)
{
break;
}
} while (iteration <= Configuration.RequestAttempt);
if (!dataResult.IsSuccessfully)
{
throw new UserException(this.Context.Localizer[RuntimeTask.RuntimeException, httpModule.Name]);
}
FillSerpPageDetails(httpModule, dataResult.Data);
var analyser = new MedianFrequencyAnalyser();
var result = RuntimeTask.Run(Context, analyser, new MedianFrequencyAnalyserConfig()
{
Pages = dataResult.Data, Request = request, HttpModule = httpModule
});
if (result.IsSuccessfully)
{
return(result.Data);
}
throw new UserException(this.Context.Localizer[RuntimeTask.RuntimeException, analyser.Name]);
}
private void TryExtractDetaisl(HtmlTagPageParser tagParser, IRuntimeModule <HttpProviderRuntimeConfig, HttpProviderData> httpModule, string url, out string redirectUrl, out string title)
{
redirectUrl = string.Empty;
title = string.Empty;
var httpResult = RuntimeTask.Run(Context, httpModule, new HttpProviderRuntimeConfig()
{
Query = url
});
if (httpResult.IsSuccessfully)
{
var extractor = new HtmlTextAttributeParser()
{
Attribute = "content",
PostHandle = (value) =>
{
string urlTag = "URL=";
int startIndex = value.IndexOf(urlTag);
if (startIndex != -1)
{
value = value.Substring(startIndex + urlTag.Length, value.Length - (urlTag.Length + startIndex));
value = value.Trim('\'');
}
return(value);
}
};
var tagConfig = new HtmlTagPageConfig();
tagConfig.Tags.Add("redirecturl", new SinglHtmlNodeSelector()
{
Tag = new TagProperties()
{
Path = @"//meta[@http-equiv='refresh']", Extractor = extractor
}
});
tagConfig.Tags.Add("title", new SinglHtmlNodeSelector()
{
Tag = new TagProperties()
{
Path = @"//title", Extractor = new HtmlTagInnerText()
}
});
using (var stream = httpResult.Data.GetContent())
{
tagConfig.Stream = stream;
var parseResult = RuntimeTask.Run(Context, tagParser, tagConfig);
if (parseResult.IsSuccessfully)
{
if (parseResult.Data.Values["redirecturl"] != null)
{
redirectUrl = (string)parseResult.Data.Values["redirecturl"];
}
if (parseResult.Data.Values["title"] != null)
{
title = (string)parseResult.Data.Values["title"];
}
}
}
if (!string.IsNullOrEmpty(redirectUrl))
{
httpResult = RuntimeTask.Run(Context, httpModule, new HttpProviderRuntimeConfig()
{
Query = redirectUrl
});
if (httpResult.IsSuccessfully)
{
using (var stream = httpResult.Data.GetContent())
{
tagConfig.Stream = stream;
var tagResult = tagParser.Run(tagConfig);
if (tagResult.Values["title"] != null)
{
title = (string)tagResult.Values["title"];
}
}
}
}
}
}
