public TitlingResult WikipediaSummarize(TitlingRequest req, string htmlDoc)
{
WikipediaArticle article = WikipediaTools.Parse(htmlDoc);
// The paragraph to be summarized.
string p = null;
// Check if the URL has an anchor to a specific section in the article.
int anchorIndex = req.Url.IndexOf("#", StringComparison.OrdinalIgnoreCase);
if (anchorIndex >= 0 && (anchorIndex + 1) < req.Url.Length)
{
var anchorId = req.Url.Substring(anchorIndex + 1);
p = article.GetFirstParagraph(anchorId);
}
// If no anchor or if we couldn't extract a paragraph for the specific anchor,
// get first paragraph of the article.
if (p == null && article.SummaryParagraphs.Length > 0)
{
p = article.SummaryParagraphs[0];
}
if (!string.IsNullOrWhiteSpace(p))
{
string summary = Format.Shorten(p, MaxCharacters, ContinuationSymbol);
req.IrcTitle.SetFormat("[ {0} ]", summary);
}
return(req.CreateResult(true));
}
public async Task <WikipediaArticle> GetArticleAsync(string language, string name)
{
if (rateLimiter.IsRatelimited())
{
return(null);
}
var article = new WikipediaArticle
{
Url = $"https://{language}.wikipedia.org/wiki/{name}"
};
{
var pageInfo = await HttpWebClient.ReturnStringAsync(new Uri($"https://{language}.wikipedia.org/w/api.php?format=json&action=query&prop=extracts&exintro=&explaintext=&titles={name}")).ConfigureAwait(false);
var jsonresp = JObject.Parse(pageInfo);
dynamic item = jsonresp["query"]["pages"].First.First;
string desc = Convert.ToString(item["extract"]);
article.Name = item["title"].ToString();
if (desc.Length >= 1024)
{
var split = desc.Split(". ").ToList();
article.Description = string.Join(". ", split.Take(4)) + ".";
}
else
{
article.Description = desc;
}
}
{
var pageImage = await HttpWebClient.ReturnStringAsync(new Uri($"https://{language}.wikipedia.org/w/api.php?format=json&action=query&prop=pageimages&piprop=original&titles={name}")).ConfigureAwait(false);
var jsonresp = JObject.Parse(pageImage);
dynamic item = jsonresp["query"]["pages"].First.First["original"];
if (item != null)
{
article.Original = new WikipediaImage
{
Source = item["source"].ToString(),
Width = Convert.ToInt32(item["width"]),
Height = Convert.ToInt32(item["height"])
};
}
}
return(article);
}
public async Task <WikipediaArticle> GetArticleAsync(string language, string name)
{
if (rateLimiter.IsRatelimited())
{
return(null);
}
var article = new WikipediaArticle
{
Url = $"https://{language}.wikipedia.org/wiki/{name}"
};
var page = new WikiPage(wikipediaSite, name);
await page.RefreshAsync(new WikiPageQueryProvider
{
Properties =
{
new ExtractsPropertyProvider
{
MaxCharacters = 1024,
AsPlainText = true,
IntroductionOnly = true
}
}
});
var extractGroup = page.GetPropertyGroup <ExtractsPropertyGroup>();
article.Name = page.Title;
article.Url = WikiLink.Parse(wikipediaSite, name).TargetUrl;
article.Description = extractGroup.Extract;
if (article.Description.Length >= 1024)
{
var split = article.Description.Split(". ").ToList();
article.Description = string.Join(". ", split.Take(4)) + ".";
}
var response = await HttpWebClient.ReturnStringAsync(new System.Uri($"https://www.wikidata.org/w/api.php?action=wbgetentities&format=json&sites={language}wiki&props=claims&titles={name}"));
var jsonresp = JObject.Parse(response);
var container = (JObject)jsonresp["entities"].First.Value <JProperty>().Value;
var claims = container["claims"];
//P18/P154/P242/P109/P1621
JToken snak = null;
if (claims["P18"] is not null)
{
snak = claims["P18"];
}
else if (claims["P154"] is not null)
{
snak = claims["P154"];
}
else if (claims["P242"] is not null)
{
snak = claims["P242"];
}
else if (claims["P109"] is not null)
{
snak = claims["P109"];
}
else if (claims["P1621"] is not null)
{
snak = claims["P1621"];
}
if (snak is not null)
{
var val = snak.First["mainsnak"]["datavalue"]["value"].ToObject <string>();
val = val.Replace(" ", "_");
var md5 = val.CreateMD5(true);;
article.ImageUrl = $"https://upload.wikimedia.org/wikipedia/commons/{md5[0]}/{md5[0]}{md5[1]}/{val}";
}
return(article);
}
