/// <summary>
/// Clones the book instance by writing and reading it from memory.
/// </summary>
/// <param name="book"></param>
/// <returns></returns>
public static EpubBook MakeCopy(EpubBook book)
{
var stream = new MemoryStream();
var writer = new EpubWriter(book);
writer.Write(stream);
stream.Seek(0, SeekOrigin.Begin);
var epub = EpubReader.Read(stream, string.Empty);
return(epub);
}
private static IEnumerable <Tuple <string, int> > CountHtmlWords(string fileName, string password)
{
using (var zipFile = ZipFile.Read(fileName))
{
foreach (var entry in zipFile.Entries)
{
var fullFileName = !entry.FileName.StartsWith("/", StringComparison.CurrentCulture) ? "/" + entry.FileName : entry.FileName;
if (txtExtensions.Any(extention => fullFileName.EndsWith(extention, StringComparison.CurrentCulture)))
{
using (var outputStream = EpubReader.GetMemoryStream(entry, password))
{
var output = outputStream.ReadToEnd();
var text = Encoding.UTF8.GetString(output, 0, output.Length);
var parts = text.Split(' ', ';', '\r', '\n', '\t', ',', '.', '!', '?');
yield return(new Tuple <string, int>(fullFileName, parts.Length));
}
}
}
}
}
public static void Main()
{
using (var reader = new PdfReader(new FileStream(@"C:\Users\d1mne\source\repos\CSharpPlayground\BookParsing\3.pdf", FileMode.Open)))
{
var text = new StringBuilder();
for (var i = 3; i < reader.NumberOfPages - 1; i++)
{
text.Append(PdfTextExtractor.GetTextFromPage(reader, i));
}
var dumm = text.ToString();
}
var fileStream = new FileStream(@"C:\Users\d1mne\source\repos\CSharpPlayground\BookParsing\1.epub", FileMode.Open);
var dummy = ReadFully(fileStream);
var epub = EpubReader.Read(fileStream, false);
var plainText = epub.PayloadToPlainText(new List <string>
{
"cover",
"title",
"copyright",
"table-of-content",
"table_of_content",
"kolofon",
"titel"
},
new List <string>
{
"-1.xhtml",
"-2.xhtml",
"-3.xhtml",
"-4.xhtml",
"-5.xhtml",
"-6.xhtml"
});
}