public async Task <Deputy> CrawlAsync(Uri url, TermOfOffice termOfOffice)
{
Require.NotNull(url, nameof(url));
Require.NotNull(termOfOffice, nameof(termOfOffice));
logger.LogInformation("Start Orka deputy {Url}.", url);
var perfWatch = Stopwatch.StartNew();
var page = await pageRequester.MakeRequestAsync(url)
.ConfigureAwait(false);
var document = page.AngleSharpHtmlDocument;
var list = document.QuerySelector("ul.dane1, ul.dane2");
var deputy = new Deputy
{
Name = GetName(list),
TermOfOffice = termOfOffice,
Url = url
};
SetBirths(list, deputy);
logger.LogInformation("Crawling Orka deputy {Url} finished. Took {Elapsed}", url, perfWatch.Elapsed);
return(deputy);
}
private async Task CrawlMainPageAsync(Uri root, IHtmlDocument document, SejmCrawlerOptions options)
{
var currentTermOfOffice = new TermOfOffice
{
Name = GetNameFromPage(root, document),
Url = root
};
if (options.SearchTermOfServices)
{
NewTermOfOfficeFound(currentTermOfOffice);
}
var crawlData = new CrawlData(this, root, options);
if (!options.SkipArchiwum)
{
await ProcessArchiwumAsync(root, document, crawlData)
.ConfigureAwait(false);
}
else
{
logger.LogInformation("Skipping archiwum.");
}
if (options.SearchDeputies)
{
await ProcessDeputies(document, currentTermOfOffice, crawlData)
.ConfigureAwait(false);
}
else
{
logger.LogInformation("Skipping current deputies.");
}
}
private async Task ProcessDeputies(IHtmlDocument document, TermOfOffice termOfOffice, CrawlData data)
{
var id = GetSejmId(document);
var url = new Uri(data.RootUrl.GetHostUri(), $"https://www.sejm.gov.pl/Sejm{id}.nsf/poslowie.xsp?type=A");
await deputiesCrawlerManager.CrawlAsync(url, termOfOffice, data)
.ConfigureAwait(false);
}
private async Task ProcessListAsync(IEnumerable <Uri> urls, TermOfOffice termOfOffice, CrawlData data)
{
foreach (var deputyUrl in urls)
{
var deputy = await deputyCrawlerManager.CrawlAsync(deputyUrl, termOfOffice)
.ConfigureAwait(false);
data.DataNotifier.NewDeputyFound(deputy);
}
}
private async Task ProcessListAsync(Uri url, TermOfOffice termOfOffice, CrawlData data)
{
var urls = await GetDeputiesUrlsAsync(url)
.ConfigureAwait(false);
foreach (var deputyUrl in urls)
{
var deputy = await deputyCrawlerManager.CrawlAsync(deputyUrl, termOfOffice)
.ConfigureAwait(false);
data.DataNotifier.NewDeputyFound(deputy);
}
}
public async Task <Deputy> CrawlAsync(Uri url, TermOfOffice termOfOffice)
{
Require.NotNull(url, nameof(url));
Require.NotNull(termOfOffice, nameof(termOfOffice));
var crawler = deputyCrawlers.FirstOrDefault(d => d.IsMatch(url));
if (crawler == null)
{
logger.LogWarning("Couldn't match correct crawler for url {Url}", url);
return(null);
}
return(await crawler.CrawlAsync(url, termOfOffice)
.ConfigureAwait(false));
}
public async Task CrawlAsync(Uri url, TermOfOffice termOfOffice, CrawlData data)
{
Require.NotNull(url, nameof(url));
Require.NotNull(termOfOffice, nameof(termOfOffice));
Require.NotNull(data, nameof(data));
logger.LogInformation("Start Orka deputies {Url}.", url);
var perfWatch = Stopwatch.StartNew();
var listUrl = await GetListUrlAsync(url)
.ConfigureAwait(false);
await ProcessListAsync(listUrl, termOfOffice, data)
.ConfigureAwait(false);
logger.LogInformation("Crawling Orka deputies {Url} finished. Took {Elapsed}", url, perfWatch.Elapsed);
}
public async Task CrawlAsync(Uri url, TermOfOffice termOfOffice, CrawlData data)
{
Require.NotNull(url, nameof(url));
Require.NotNull(termOfOffice, nameof(termOfOffice));
Require.NotNull(data, nameof(data));
var crawler = deputiesCrawlers.FirstOrDefault(d => d.IsMatch(url));
if (crawler == null)
{
logger.LogWarning("Couldn't match correct crawler for url {Url}", url);
}
else
{
await crawler.CrawlAsync(url, termOfOffice, data)
.ConfigureAwait(false);
}
}
private async Task ProcessPoliticans(IEnumerable <IHtmlAnchorElement> anchors, TermOfOffice termOfOffice, CrawlData data)
{
if (!data.Options.SearchDeputies)
{
return;
}
var politiciansAnchor = anchors.First(a => a.InnerHtml == "Posłowie");
var url = new Uri(politiciansAnchor.Href);
await politiciansCrawlerManager.CrawlAsync(url, termOfOffice, data)
.ConfigureAwait(false);
}
public void NewTermOfOfficeFound(TermOfOffice termOfOffice)
{
Require.NotNull(termOfOffice, nameof(termOfOffice));
TermOfOfficeFound?.Invoke(termOfOffice);
}