protected Article()
{
AuthorArticles = new List <AuthorArticle>();
SubjectItemArticles = new List <SubjectItemArticle>();
_versions = new List <Version>();
ScrapeContext = ArticleScrapeContextEnum.None;
}
public Article(string arxivId, string htmlLink, string pdfUrl, string otherFormatUrl,
string title, string abstractText, string comments
, string journalReference, string journalReferenceHtmlLink, DateTime scrapedDate)
{
ArxivId = arxivId;
HtmlLink = htmlLink;
PdfUrl = pdfUrl;
OtherFormatUrl = otherFormatUrl;
Title = title;
AbstractText = abstractText;
Comments = comments;
JournalReference = journalReference;
JournalReferenceHtmlLink = journalReferenceHtmlLink;
ScrapedDate = scrapedDate;
AuthorArticles = new List <AuthorArticle>();
SubjectItemArticles = new List <SubjectItemArticle>();
_versions = new List <Version>();
ScrapeContext = ArticleScrapeContextEnum.None;
}
public void AddScrapeContext(string pageHeader, string arxivIdLabel, string h3Text)
{
if (!string.IsNullOrEmpty(pageHeader) &&
pageHeader.ToLower().Contains("catchup"))
{
ScrapeContext = ArticleScrapeContextEnum.CatchUp;
}
if (!string.IsNullOrEmpty(arxivIdLabel))
{
if (arxivIdLabel.ToLower().Contains("replaced"))
{
ScrapeContext = ArticleScrapeContextEnum.Replacement;
}
else if (arxivIdLabel.ToLower().Contains("cross-list"))
{
ScrapeContext = ArticleScrapeContextEnum.CrossList;
}
}
if (!string.IsNullOrEmpty(h3Text))
{
if (h3Text.ToLower().Contains("submission"))
{
ScrapeContext = ArticleScrapeContextEnum.Submission;
}
else if (h3Text.ToLower().Contains("cross-list") || h3Text.ToLower().Contains("cross"))
{
ScrapeContext = ArticleScrapeContextEnum.CrossList;
}
else if (h3Text.ToLower().Contains("Replacement"))
{
ScrapeContext = ArticleScrapeContextEnum.Replacement;
}
}
}