private void AnalyzeEpub()
{
this.FileSizeBytes = new FileInfo(this.FullName).Length;
this.IsbnFromFileName = HCPUtils.GetIsbn13(this.FullName);
this.Isbn = this.IsbnFromFileName;
try
{
using (ZipArchive archive = ZipFile.OpenRead(this.FullName))
{
foreach (ZipArchiveEntry entry in archive.Entries)
{
// check for invalid DOCTYPE in any .xhtml file
if (entry.FullName.EndsWith(".xhtml") && !this.InvalidDocType)
{
using (var stream = entry.Open())
using (var reader = new StreamReader(stream))
{
string xhtml = reader.ReadToEnd();
if (xhtml.Contains("about:legacy-compat"))
{
this.InvalidDocType = true;
}
}
}
// check .opf - primarily for isbn
if (entry.FullName.EndsWith(".opf"))
{
this.OpfFileName = entry.FullName;
this.HasOpf = true;
// get isbn from opf
using (var stream = entry.Open())
{
XDocument xmlOpf = XDocument.Load(stream);
XNamespace dc = "http://purl.org/dc/elements/1.1/";
XNamespace opf = "http://www.idpf.org/2007/opf";
this.OpfIdentifier = xmlOpf.Descendants(opf + "metadata").Elements(dc + "identifier").FirstOrDefault().Value;
if (!String.IsNullOrEmpty(this.OpfIdentifier))
{
// sometimes isbns are formatted with dashes
string tmpIsbn = this.OpfIdentifier.Replace("-", "");
this.IsbnFromOpf = HCPUtils.GetIsbn13(tmpIsbn);
}
}
}
}
}
}
catch (System.IO.InvalidDataException exIO)
{
this.CorruptFile = true;
log.Error(String.Format("IO Exception analyzing Epub {0} - {1}", this.FullName, exIO.StackTrace));
}
catch (Exception ex)
{
log.Error(String.Format("Exception analyzing Epub {0} - {1}", this.FullName, ex.StackTrace));
}
}
