public static EpubSchema ReadSchema(ZipArchive epubArchive)
{
EpubSchema result = new EpubSchema();
string rootFilePath = RootFilePathReader.GetRootFilePath(epubArchive);
string contentDirectoryPath = ZipPathUtils.GetDirectoryPath(rootFilePath);
result.ContentDirectoryPath = contentDirectoryPath;
EpubPackage package = PackageReader.ReadPackage(epubArchive, rootFilePath);
result.Package = package;
EpubNavigation navigation = NavigationReader.ReadNavigation(epubArchive, contentDirectoryPath, package);
result.Navigation = navigation;
return(result);
}
public static async Task <EpubSchema> ReadSchemaAsync(ZipArchive epubArchive)
{
EpubSchema result = new EpubSchema();
string rootFilePath = await RootFilePathReader.GetRootFilePathAsync(epubArchive).ConfigureAwait(false);
string contentDirectoryPath = ZipPathUtils.GetDirectoryPath(rootFilePath);
result.ContentDirectoryPath = contentDirectoryPath;
EpubPackage package = await PackageReader.ReadPackageAsync(epubArchive, rootFilePath).ConfigureAwait(false);
result.Package = package;
EpubNavigation navigation = await NavigationReader.ReadNavigationAsync(epubArchive, contentDirectoryPath, package).ConfigureAwait(false);
result.Navigation = navigation;
return(result);
}
public static async Task <EpubSchema> ReadSchemaAsync(ZipArchive epubArchive)
{
EpubSchema result = new EpubSchema();
// Reading META-INF/container.xml
string rootFilePath = await RootFilePathReader.GetRootFilePathAsync(epubArchive);
// Getting directory path - usually it's: META-INF/
string contentDirectoryPath = ZipPathUtils.GetDirectoryPath(rootFilePath);
result.ContentDirectoryPath = contentDirectoryPath;
//Reading the file content.opf
EpubPackage package = await PackageReader.ReadPackageAsync(epubArchive, rootFilePath);
result.Package = package;
EpubNavigation navigation = await NavigationReader.ReadNavigationAsync(epubArchive, contentDirectoryPath, package);
result.Navigation = navigation;
return(result);
}
public static async Task <EpubNavigation> ReadNavigationAsync(ZipArchive epubArchive, string contentDirectoryPath, EpubPackage package)
{
EpubNavigation result = new EpubNavigation();
string tocId = package.Spine.Toc;
if (String.IsNullOrEmpty(tocId))
{
throw new Exception("EPUB parsing error: TOC ID is empty.");
}
EpubManifestItem tocManifestItem = package.Manifest.FirstOrDefault(item => String.Compare(item.Id, tocId, StringComparison.OrdinalIgnoreCase) == 0);
if (tocManifestItem == null)
{
throw new Exception(String.Format("EPUB parsing error: TOC item {0} not found in EPUB manifest.", tocId));
}
string tocFileEntryPath = ZipPathUtils.Combine(contentDirectoryPath, tocManifestItem.Href);
ZipArchiveEntry tocFileEntry = epubArchive.GetEntry(tocFileEntryPath);
if (tocFileEntry == null)
{
throw new Exception(String.Format("EPUB parsing error: TOC file {0} not found in archive.", tocFileEntryPath));
}
if (tocFileEntry.Length > Int32.MaxValue)
{
throw new Exception(String.Format("EPUB parsing error: TOC file {0} is larger than 2 Gb.", tocFileEntryPath));
}
XDocument containerDocument;
using (Stream containerStream = tocFileEntry.Open())
{
containerDocument = await XmlUtils.LoadDocumentAsync(containerStream).ConfigureAwait(false);
}
XNamespace ncxNamespace = "http://www.daisy.org/z3986/2005/ncx/";
XElement ncxNode = containerDocument.Element(ncxNamespace + "ncx");
if (ncxNode == null)
{
throw new Exception("EPUB parsing error: TOC file does not contain ncx element.");
}
XElement headNode = ncxNode.Element(ncxNamespace + "head");
if (headNode == null)
{
throw new Exception("EPUB parsing error: TOC file does not contain head element.");
}
EpubNavigationHead navigationHead = ReadNavigationHead(headNode);
result.Head = navigationHead;
XElement docTitleNode = ncxNode.Element(ncxNamespace + "docTitle");
if (docTitleNode == null)
{
throw new Exception("EPUB parsing error: TOC file does not contain docTitle element.");
}
EpubNavigationDocTitle navigationDocTitle = ReadNavigationDocTitle(docTitleNode);
result.DocTitle = navigationDocTitle;
result.DocAuthors = new List <EpubNavigationDocAuthor>();
foreach (XElement docAuthorNode in ncxNode.Elements(ncxNamespace + "docAuthor"))
{
EpubNavigationDocAuthor navigationDocAuthor = ReadNavigationDocAuthor(docAuthorNode);
result.DocAuthors.Add(navigationDocAuthor);
}
XElement navMapNode = ncxNode.Element(ncxNamespace + "navMap");
if (navMapNode == null)
{
throw new Exception("EPUB parsing error: TOC file does not contain navMap element.");
}
EpubNavigationMap navMap = ReadNavigationMap(navMapNode);
result.NavMap = navMap;
XElement pageListNode = ncxNode.Element(ncxNamespace + "pageList");
if (pageListNode != null)
{
EpubNavigationPageList pageList = ReadNavigationPageList(pageListNode);
result.PageList = pageList;
}
result.NavLists = new List <EpubNavigationList>();
foreach (XElement navigationListNode in ncxNode.Elements(ncxNamespace + "navList"))
{
EpubNavigationList navigationList = ReadNavigationList(navigationListNode);
result.NavLists.Add(navigationList);
}
return(result);
}
public static async Task <EpubNavigation> ReadNavigationAsync(ZipArchive epubArchive, string contentDirectoryPath, EpubPackage package)
{
EpubNavigation result = new EpubNavigation();
string tocId = package.Spine.Toc;
XmlReaderSettings xmlReaderSettings = new XmlReaderSettings
{
// XmlResolver = null,
Async = true,
DtdProcessing = DtdProcessing.Ignore
};
if (String.IsNullOrEmpty(tocId))
{
throw new Exception("EPUB parsing error: TOC ID is empty.");
}
//Cheking if toc id referenced in spine exist in manifest
EpubManifestItem tocManifestItem = package.Manifest.FirstOrDefault(item => String.Compare(item.Id, tocId, StringComparison.OrdinalIgnoreCase) == 0);
if (tocManifestItem == null)
{
throw new Exception(String.Format("EPUB parsing error: TOC item {0} not found in EPUB manifest.", tocId));
}
//Opening .toc file in archive using href-reference from manifest
string tocFileEntryPath = ZipPathUtils.Combine(contentDirectoryPath, tocManifestItem.Href);
ZipArchiveEntry tocFileEntry = epubArchive.GetEntry(tocFileEntryPath);
if (tocFileEntry == null)
{
throw new Exception(String.Format("EPUB parsing error: TOC file {0} not found in archive.", tocFileEntryPath));
}
if (tocFileEntry.Length > Int32.MaxValue)
{
throw new Exception(String.Format("EPUB parsing error: TOC file {0} is bigger than 2 Gb.", tocFileEntryPath));
}
// ------------------ Actual Parsing starts here: -------------------------
using (Stream containerStream = tocFileEntry.Open())
{
using (XmlReader xmlReader = XmlReader.Create(containerStream, xmlReaderSettings))
{
result.Head = await ReadNavigationHeadAsync(xmlReader);
result.DocTitle = await ReadNavigationDocTitleAsync(xmlReader);
result.DocAuthors = await ReadNavigationAuthorsAsync(xmlReader);
result.NavMap = await ReadNavigationMapAsync(xmlReader);
result.NavLists = new List <EpubNavigationList>(); //Empty, because not implemented
result.PageList = new EpubNavigationPageList(); //Empty, because not implemented
}
}
return(result);
//-------------------------------------------Boring old style Silverlight code...-----------------------------------------------------------------
//------------------------------------------------------------------------------------------------------------------------------------------------
//XmlDocument containerDocument;
//containerDocument = XmlDocument.Load(containerStream);
//XmlNamespaceManager xmlNamespaceManager = new XmlNamespaceManager(containerDocument.NameTable);
//xmlNamespaceManager.AddNamespace("ncx", "http://www.daisy.org/z3986/2005/ncx/");
////Parsing head section
//XmlNode headNode = containerDocument.DocumentElement.SelectSingleNode("ncx:head", xmlNamespaceManager);
//if (headNode == null)
// throw new Exception("EPUB parsing error: TOC file does not contain head element");
//EpubNavigationHead navigationHead = ReadNavigationHead(headNode);
//result.Head = navigationHead;
////Parsing title
//XmlNode docTitleNode = containerDocument.DocumentElement.SelectSingleNode("ncx:docTitle", xmlNamespaceManager);
//if (docTitleNode == null)
// throw new Exception("EPUB parsing error: TOC file does not contain docTitle element");
//EpubNavigationDocTitle navigationDocTitle = ReadNavigationDocTitle(docTitleNode);
//result.DocTitle = navigationDocTitle;
////Parsing authors section...
//result.DocAuthors = new List<EpubNavigationDocAuthor>();
//foreach (XmlNode docAuthorNode in containerDocument.DocumentElement.SelectNodes("ncx:docAuthor", xmlNamespaceManager))
//{
// EpubNavigationDocAuthor navigationDocAuthor = ReadNavigationDocAuthor(docAuthorNode);
// result.DocAuthors.Add(navigationDocAuthor);
//}
//Parsing navMap section
//XmlNode navMapNode = containerDocument.DocumentElement.SelectSingleNode("ncx:navMap", xmlNamespaceManager);
//if (navMapNode == null)
// throw new Exception("EPUB parsing error: TOC file does not contain navMap element");
//EpubNavigationMap navMap = ReadNavigationMap(navMapNode);
//result.NavMap = navMap;
//-----------------------------------TO-DO: Implement -----------------------------------------------------------
//TO-DO: Implement pageList parsing. Needed to tide-up position inside epub to actual pages of the printed book
//--------------------------------------------------------------------------------------------------------------
//Parsing pageList node
//XmlNode pageListNode = containerDocument.DocumentElement.SelectSingleNode("ncx:pageList", xmlNamespaceManager);
//if (pageListNode != null)
//{
// EpubNavigationPageList pageList = ReadNavigationPageList(pageListNode);
// result.PageList = pageList;
//}
////TO-DO: Implement navList parsing. It is a secondary navigation system for supplied book info - schemes, fugures, diagrams, illustrations etc
////Parsing navList nodes
//result.NavLists = new List<EpubNavigationList>();
//foreach (XmlNode navigationListNode in containerDocument.DocumentElement.SelectNodes("ncx:navList", xmlNamespaceManager))
//{
// EpubNavigationList navigationList = ReadNavigationList(navigationListNode);
// result.NavLists.Add(navigationList);
//}
//--------------------------------------------------------------------------------------------------------------
}
private void Load_Click(object sender, RoutedEventArgs e)
{
var openFileDialog = new Microsoft.Win32.OpenFileDialog()
{
Filter = "Epub Files (*.epub)|*.epub"
};
var result = openFileDialog.ShowDialog();
if (result == true)
{
// Opens a book and reads all of its content into memory
epubBook = EpubReader.ReadBook(openFileDialog.FileName);
// COMMON PROPERTIES
// Book's title
string title = epubBook.Title;
// Book's authors (comma separated list)
string author = epubBook.Author;
// Book's authors (list of authors names)
List <string> authors = epubBook.AuthorList;
// Book's cover image (null if there is no cover)
byte[] coverImageContent = epubBook.CoverImage;
if (coverImageContent != null)
{
using (MemoryStream coverImageStream = new MemoryStream(coverImageContent.ToArray()))
{
// Assign the Source property of your image
try
{
File.Delete(openFileDialog.FileName + ".jpg");
File.WriteAllBytes(openFileDialog.FileName + ".jpg", coverImageStream.ToArray());
}
catch
{
File.WriteAllBytes(openFileDialog.FileName + ".jpg", coverImageStream.ToArray());
}
BitmapImage imageSource = new BitmapImage(new Uri(@"C:/Users/shish/Downloads/1.jpg", UriKind.Absolute));
Image1.Source = imageSource;
}
}
Info.Text = "Title: " + title + "\n" + "Author: " + author + "\n";
Info.FontWeight = FontWeights.Bold;
// CHAPTERS
// Enumerating chapters
foreach (EpubChapter chapter in epubBook.Chapters)
{
// Title of chapter
string chapterTitle = chapter.Title;
// HTML content of current chapter
string chapterHtmlContent = chapter.HtmlContent;
// Nested chapters
List <EpubChapter> subChapters = chapter.SubChapters;
//PrintChapter(chapter);
//Chapters.Inlines.Add(new Run(chapterTitle + "\n"));
Chapters.Items.Add(chapterTitle);
}
// CONTENT
// Book's content (HTML files, stlylesheets, images, fonts, etc.)
EpubContent bookContent = epubBook.Content;
// IMAGES
// All images in the book (file name is the key)
Dictionary <string, EpubByteContentFile> images = bookContent.Images;
EpubByteContentFile firstImage = images.Values.First();
// Content type (e.g. EpubContentType.IMAGE_JPEG, EpubContentType.IMAGE_PNG)
EpubContentType contentType = firstImage.ContentType;
// MIME type (e.g. "image/jpeg", "image/png")
string mimeContentType = firstImage.ContentMimeType;
// Creating Image class instance from the content
using (MemoryStream imageStream = new MemoryStream(firstImage.Content))
{
System.Drawing.Image image = System.Drawing.Image.FromStream(imageStream);
}
// HTML & CSS
// All XHTML files in the book (file name is the key)
Dictionary <string, EpubTextContentFile> htmlFiles = bookContent.Html;
// All CSS files in the book (file name is the key)
Dictionary <string, EpubTextContentFile> cssFiles = bookContent.Css;
// Entire HTML content of the book
foreach (EpubTextContentFile htmlFile in htmlFiles.Values)
{
string htmlContent = htmlFile.Content;
}
// All CSS content in the book
foreach (EpubTextContentFile cssFile in cssFiles.Values)
{
string cssContent = cssFile.Content;
}
// OTHER CONTENT
// All fonts in the book (file name is the key)
Dictionary <string, EpubByteContentFile> fonts = bookContent.Fonts;
// All files in the book (including HTML, CSS, images, fonts, and other types of files)
Dictionary <string, EpubContentFile> allFiles = bookContent.AllFiles;
// ACCESSING RAW SCHEMA INFORMATION
// EPUB OPF data
EpubPackage package = epubBook.Schema.Package;
// Enumerating book's contributors
foreach (EpubMetadataContributor contributor in package.Metadata.Contributors)
{
string contributorName = contributor.Contributor;
string contributorRole = contributor.Role;
}
// EPUB NCX data
EpubNavigation navigation = epubBook.Schema.Navigation;
// Enumerating NCX metadata
foreach (EpubNavigationHeadMeta meta in navigation.Head)
{
string metadataItemName = meta.Name;
string metadataItemContent = meta.Content;
}
}
}
public static EpubNavigation ReadNavigation(ZipArchive epubArchive, string contentDirectoryPath, EpubPackage package)
{
EpubNavigation result = new EpubNavigation();
string tocId = package.Spine.Toc;
if (String.IsNullOrEmpty(tocId))
{
throw new Exception("EPUB parsing error: TOC ID is empty.");
}
EpubManifestItem tocManifestItem = package.Manifest.FirstOrDefault(item => String.Compare(item.Id, tocId, StringComparison.OrdinalIgnoreCase) == 0);
if (tocManifestItem == null)
{
throw new Exception(String.Format("EPUB parsing error: TOC item {0} not found in EPUB manifest.", tocId));
}
string tocFileEntryPath = ZipPathUtils.Combine(contentDirectoryPath, tocManifestItem.Href);
ZipArchiveEntry tocFileEntry = epubArchive.GetEntry(tocFileEntryPath);
if (tocFileEntry == null)
{
throw new Exception(String.Format("EPUB parsing error: TOC file {0} not found in archive.", tocFileEntryPath));
}
if (tocFileEntry.Length > Int32.MaxValue)
{
throw new Exception(String.Format("EPUB parsing error: TOC file {0} is bigger than 2 Gb.", tocFileEntryPath));
}
XmlDocument containerDocument;
using (Stream containerStream = tocFileEntry.Open())
containerDocument = XmlUtils.LoadDocument(containerStream);
XmlNamespaceManager xmlNamespaceManager = new XmlNamespaceManager(containerDocument.NameTable);
xmlNamespaceManager.AddNamespace("ncx", "http://www.daisy.org/z3986/2005/ncx/");
XmlNode headNode = containerDocument.DocumentElement.SelectSingleNode("ncx:head", xmlNamespaceManager);
if (headNode == null)
{
throw new Exception("EPUB parsing error: TOC file does not contain head element");
}
EpubNavigationHead navigationHead = ReadNavigationHead(headNode);
result.Head = navigationHead;
XmlNode docTitleNode = containerDocument.DocumentElement.SelectSingleNode("ncx:docTitle", xmlNamespaceManager);
if (docTitleNode == null)
{
throw new Exception("EPUB parsing error: TOC file does not contain docTitle element");
}
EpubNavigationDocTitle navigationDocTitle = ReadNavigationDocTitle(docTitleNode);
result.DocTitle = navigationDocTitle;
result.DocAuthors = new List <EpubNavigationDocAuthor>();
foreach (XmlNode docAuthorNode in containerDocument.DocumentElement.SelectNodes("ncx:docAuthor", xmlNamespaceManager))
{
EpubNavigationDocAuthor navigationDocAuthor = ReadNavigationDocAuthor(docAuthorNode);
result.DocAuthors.Add(navigationDocAuthor);
}
XmlNode navMapNode = containerDocument.DocumentElement.SelectSingleNode("ncx:navMap", xmlNamespaceManager);
if (navMapNode == null)
{
throw new Exception("EPUB parsing error: TOC file does not contain navMap element");
}
EpubNavigationMap navMap = ReadNavigationMap(navMapNode);
result.NavMap = navMap;
XmlNode pageListNode = containerDocument.DocumentElement.SelectSingleNode("ncx:pageList", xmlNamespaceManager);
if (pageListNode != null)
{
EpubNavigationPageList pageList = ReadNavigationPageList(pageListNode);
result.PageList = pageList;
}
result.NavLists = new List <EpubNavigationList>();
foreach (XmlNode navigationListNode in containerDocument.DocumentElement.SelectNodes("ncx:navList", xmlNamespaceManager))
{
EpubNavigationList navigationList = ReadNavigationList(navigationListNode);
result.NavLists.Add(navigationList);
}
return(result);
}
