private string readFile(string filePath, Encoding encoding, bool needMarkChapterHeaders)
{
string text = File.ReadAllText(filePath, encoding);
if (filePath.EndsWith("html") || filePath.EndsWith("htm") || filePath.EndsWith("asp") || filePath.EndsWith("aspx") || filePath.EndsWith("php"))
{
text = HtmlScrapper.GetChineseContent(text, needMarkChapterHeaders);
}
else if (needMarkChapterHeaders && text.Contains("\n"))
{
string text2 = text.Substring(0, text.IndexOf('\n'));
if (text2.Contains("第") && text2.Contains("章"))
{
text = "$CHAPTER_HEADER$. " + text.TrimStart(" \u3000\t".ToCharArray());
}
}
return(text);
}
public HtmlLinkHandler(HtmlScrapper scrapper)
{
_scrapper = scrapper;
}
public ScrapTest()
{
_dateTimeProvider = new Mock <IDateTimeProvider>();
_sut = new HtmlScrapper(_dateTimeProvider.Object);
}