public void SplitFile(ConversionOptions options)
{
_CurrentOptions = options;
_WorkingDir = "";
_HtmlFooter = _BodyEndTag;
ProcessFile(_CurrentOptions.InputFile);
}
public static ConversionOptions FindMetaData(string html)
{
ConversionOptions returnable = new ConversionOptions();
returnable.Title = ParseTag(html, "Title");
returnable.Author = ParseMeta(html, "author");
DateTime publish = DateTime.MinValue;
DateTime.TryParse(ParseMeta(html, "date"), out publish);
returnable.PublishDate = publish;
return returnable;
}
public static byte[] GetTOCFile(ConversionOptions options)
{
StringBuilder tocBuilder = new StringBuilder();
tocBuilder.AppendLine("<?xml version=\"1.0\" encoding=\"UTF-8\"?>");
//ncx
tocBuilder.AppendLine("<ncx xmlns=\"http://www.daisy.org/z3986/2005/ncx/\" version=\"2005-1\">");
//ncx/head
tocBuilder.AppendLine("<head>");
//ncx/head/meta(s)
tocBuilder.AppendFormat("<meta name=\"dtb:uid\" content=\"{0}\"/>", options.ID);
tocBuilder.AppendFormat("<meta name=\"dtb:depth\" content=\"{0}\"/>", 1);
tocBuilder.AppendFormat("<meta name=\"dtb:totalPageCount\" content=\"{0}\"/>", 0);
tocBuilder.AppendFormat("<meta name=\"dtb:maxPageNumber\" content=\"{0}\"/>", 0);
//end /ncx/head/meta(s)
//end /ncx/head
tocBuilder.AppendLine("</head>");
//ncx/docTitle
tocBuilder.AppendLine("<docTitle>");
//ncx/docTitle/text & end
tocBuilder.AppendFormat("<text>{0}</text>", options.Title);
//end /ncx/docTitle/text
tocBuilder.AppendLine("</docTitle>");
//ncx/navMap
tocBuilder.AppendLine("<navMap>");
//ncx/navMap/navPoint(s)
//title page
tocBuilder.AppendLine(GenerateNavPoint("Title Page", 1, "title_page.xhtml"));
foreach (Chapter chapter in options.Chapters)
{
//chapters
//need the +1 because the title page isn't included in the chapter list
tocBuilder.AppendLine(GenerateNavPoint(chapter.ChapterTitle, chapter.ChapterNumber + 1, chapter.EmbeddedTitle));
}
//end /ncx/navMap/navPoint(s)
//end /ncx/navMap
tocBuilder.AppendLine("</navMap>");
//end /ncx
tocBuilder.AppendLine("</ncx>");
//return the xml in ASCII encoding
return(Encoding.ASCII.GetBytes(tocBuilder.ToString()));
}
public static void GenerateEPUB(string inputFile, string outputFile, ConversionOptions options)
{
ZipFile file = new ZipFile(outputFile);
//generate mimetype
using (Stream mime = new MemoryStream())
{
byte[] mimeBytes = System.Text.Encoding.ASCII.GetBytes(_mimetype);
mime.Write(mimeBytes, 0, mimeBytes.Length);
file.AddEntry("mimetype", mime);
}
//generate META-INF
file.AddDirectoryByName("META-INF");
{
//container.xml
using (Stream container = new MemoryStream())
{
byte[] containerBytes = System.Text.Encoding.ASCII.GetBytes(Crabwise.PDFtoEPUB.Properties.Resources.container);
container.Write(containerBytes, 0, containerBytes.Length);
file.AddEntry("META-INF\\container.xml", container);
}
}
//generate OEBPS
file.AddDirectoryByName("OEBPS");
{
//images
file.AddDirectoryByName("OEBPS\\IMAGES");
{
//add each image to the directory
}
//xhtml chapters
//page-template.xpgt
byte[] pagetemplateBytes = System.Text.Encoding.ASCII.GetBytes(Properties.Resources.page_template);
file.AddEntry("OEBPS\\page-template.xpgt", pagetemplateBytes);
//stylesheet.css
byte[] stylesheetBytes = System.Text.Encoding.ASCII.GetBytes(Properties.Resources.stylesheet);
file.AddEntry("OEBPS\\stylesheet.css", stylesheetBytes);
//title_page.xhtml
file.AddEntry("OEBPS\\title_page.xhtml", TitlePageGenerator.GetTitlePage(options));
//Content.opf
file.AddEntry("OEBPS\\content.opf", ContentFileGenerator.GetContentFile(options));
//toc.ncx
file.AddEntry("OEBPS\\toc.ncx", TOCFileGenerator.GetTOCFile(options));
}
}
public static void GenerateEPUB(string inputFile, string outputFile, ConversionOptions options)
{
ZipFile file = new ZipFile(outputFile);
//generate mimetype
using (Stream mime = new MemoryStream())
{
byte[] mimeBytes = System.Text.Encoding.ASCII.GetBytes(_mimetype);
mime.Write(mimeBytes, 0, mimeBytes.Length);
file.AddEntry("mimetype", mime);
}
//generate META-INF
file.AddDirectoryByName("META-INF");
{
//container.xml
using (Stream container = new MemoryStream())
{
byte[] containerBytes = System.Text.Encoding.ASCII.GetBytes(Crabwise.PDFtoEPUB.Properties.Resources.container);
container.Write(containerBytes, 0, containerBytes.Length);
file.AddEntry("META-INF\\container.xml", container);
}
}
//generate OEBPS
file.AddDirectoryByName("OEBPS");
{
//images
file.AddDirectoryByName("OEBPS\\IMAGES");
{
//add each image to the directory
}
//xhtml chapters
//page-template.xpgt
byte[] pagetemplateBytes = System.Text.Encoding.ASCII.GetBytes(Properties.Resources.page_template);
file.AddEntry("OEBPS\\page-template.xpgt", pagetemplateBytes);
//stylesheet.css
byte[] stylesheetBytes = System.Text.Encoding.ASCII.GetBytes(Properties.Resources.stylesheet);
file.AddEntry("OEBPS\\stylesheet.css", stylesheetBytes);
//title_page.xhtml
file.AddEntry("OEBPS\\title_page.xhtml", TitlePageGenerator.GetTitlePage(options));
//Content.opf
file.AddEntry("OEBPS\\content.opf", ContentFileGenerator.GetContentFile(options));
//toc.ncx
file.AddEntry("OEBPS\\toc.ncx", TOCFileGenerator.GetTOCFile(options));
}
}
public static byte[] GetTOCFile(ConversionOptions options)
{
StringBuilder tocBuilder = new StringBuilder();
tocBuilder.AppendLine("<?xml version=\"1.0\" encoding=\"UTF-8\"?>");
//ncx
tocBuilder.AppendLine("<ncx xmlns=\"http://www.daisy.org/z3986/2005/ncx/\" version=\"2005-1\">");
//ncx/head
tocBuilder.AppendLine("<head>");
//ncx/head/meta(s)
tocBuilder.AppendFormat("<meta name=\"dtb:uid\" content=\"{0}\"/>", options.ID);
tocBuilder.AppendFormat("<meta name=\"dtb:depth\" content=\"{0}\"/>", 1);
tocBuilder.AppendFormat("<meta name=\"dtb:totalPageCount\" content=\"{0}\"/>", 0);
tocBuilder.AppendFormat("<meta name=\"dtb:maxPageNumber\" content=\"{0}\"/>", 0);
//end /ncx/head/meta(s)
//end /ncx/head
tocBuilder.AppendLine("</head>");
//ncx/docTitle
tocBuilder.AppendLine("<docTitle>");
//ncx/docTitle/text & end
tocBuilder.AppendFormat("<text>{0}</text>", options.Title);
//end /ncx/docTitle/text
tocBuilder.AppendLine("</docTitle>");
//ncx/navMap
tocBuilder.AppendLine("<navMap>");
//ncx/navMap/navPoint(s)
//title page
tocBuilder.AppendLine(GenerateNavPoint("Title Page", 1, "title_page.xhtml"));
foreach (Chapter chapter in options.Chapters)
{
//chapters
//need the +1 because the title page isn't included in the chapter list
tocBuilder.AppendLine(GenerateNavPoint(chapter.ChapterTitle, chapter.ChapterNumber + 1, chapter.EmbeddedTitle));
}
//end /ncx/navMap/navPoint(s)
//end /ncx/navMap
tocBuilder.AppendLine("</navMap>");
//end /ncx
tocBuilder.AppendLine("</ncx>");
//return the xml in ASCII encoding
return Encoding.ASCII.GetBytes(tocBuilder.ToString());
}
public static ConversionOptions FindMetaData(string html)
{
ConversionOptions returnable = new ConversionOptions();
returnable.Title = ParseTag(html, "Title");
returnable.Author = ParseMeta(html, "author");
DateTime publish = DateTime.MinValue;
DateTime.TryParse(ParseMeta(html, "date"), out publish);
returnable.PublishDate = publish;
return(returnable);
}
public static byte[] GetTitlePage(ConversionOptions options)
{
StringBuilder titleBuilder = new StringBuilder();
titleBuilder.AppendLine("<?xml version=\"1.0\" encoding=\"iso-8859\"?>");
titleBuilder.AppendLine("<!DOCTYPE html PUBLIC \"-//W3C//DTD XHTML 1.1//EN\" \"http://www.w3.org/TR/xhtml11/DTD/xhtml11.dtd\">");
titleBuilder.AppendLine("<html xmlns=\"http://www.w3.org/1999/xhtml\">");
titleBuilder.AppendLine("<head>");
titleBuilder.AppendLine("<title>");
titleBuilder.Append(options.Title);
titleBuilder.Append("</title>");
titleBuilder.AppendLine("<link href=\"stylesheet.css\" type=\"text/css\" rel=\"stylesheet\" />");
titleBuilder.AppendLine("<link rel=\"stylesheet\" type=\"application/vnd.adobe-page-template+xml\" href=\"page-template.xpgt\"/>");
titleBuilder.AppendLine("</head>");
titleBuilder.AppendLine("<body>");
titleBuilder.AppendLine("<div>");
titleBuilder.AppendLine("<p>");
titleBuilder.Append(" ");
titleBuilder.Append("</p>");
titleBuilder.AppendLine("<h2>");
titleBuilder.Append(options.Title);
titleBuilder.Append("</h2>");
titleBuilder.AppendLine("<p> </p>");
titleBuilder.AppendLine("<h3>");
titleBuilder.Append(options.Author);
titleBuilder.Append("</h3>");
titleBuilder.AppendLine("<p> </p>");
titleBuilder.AppendLine("<h7>Generated by PDFtoEPUB (www.crabwisestudios.com/PDFtoEPUB/Overview)</h7>");
titleBuilder.AppendLine("</div></body></html>");
byte[] titleBytes = Encoding.ASCII.GetBytes(titleBuilder.ToString());
return(titleBytes);
}
public static byte[] GetContentFile(ConversionOptions options)
{
StringBuilder contentBuilder = new StringBuilder();
contentBuilder.AppendLine("<?xml version=\"1.0\"?>");
//package
contentBuilder.AppendFormat("<package version=\"2.0\" xmlns=\"http://www.idpf.org/2007/opf\" unique-identifier=\"{0}\">", options.ID);
//package/metadata
contentBuilder.AppendLine("<metadata xmlns:dc=\"http://purl.org/dc/elements/1.1/\" xmlns:opf=\"http://www.idpf.org/2007/opf\">");
//package/metadata/dc:title & end
contentBuilder.AppendFormat("<dc:title>{0}</dc:title>", options.Title);
//package/metadata/dc:creator & end
contentBuilder.AppendFormat("<dc:creator opf:role=\"aut\">{0}</dc:creator>", options.Author);
//package/metadata/dc:language & end
contentBuilder.AppendLine("<dc:language>en-US</dc:language>"); //assuming en-US
//package/metadata/dc:rights & end
contentBuilder.AppendLine("<dc:rights>Public Domain</dc:rights>");
//package/metadata/dc:publishers & end
contentBuilder.AppendLine("<dc:publisher>Crabwise Studios</dc:publisher>");
//package/metadata/dc:identifier & end
contentBuilder.AppendFormat("<dc:identifier id=\"BookId\">urn:uuid:{0}</dc:identifier>", options.ID);
//end /package/metadata
contentBuilder.AppendLine("</metadata>");
//package/manifest
contentBuilder.AppendLine("<manifest>");
//package/manifest/item(s)
//table of contents file
contentBuilder.AppendLine(GenerateItem("ncx", "toc.ncx", "text-xml"));
//main stylesheet
contentBuilder.AppendLine(GenerateItem("style", "stylesheet.css", "text/css"));
//adobe digital editions page template
contentBuilder.AppendLine(GenerateItem("pagetemplate", "page-template.xpgt", "application/vnd.adobe-page-template+xml"));
//title page
contentBuilder.AppendLine(GenerateItem("titlepage", "title_page.xhtml", "application/xhtml+xml"));
//add the chapters to the manifest
foreach (Chapter chapter in options.Chapters)
{
//example id: chapter01, chapter19
contentBuilder.AppendLine(GenerateItem("chapter" + chapter.ChapterNumber.ToString().PadLeft(2, '0'), chapter.EmbeddedTitle, "application/xhtml+xml"));
}
//add the images to the manifest
for (int imgi = 0; imgi < options.Images.Count; imgi++)
{
//example id: img001, img019, img456
contentBuilder.AppendLine(GenerateItem("img" + imgi.ToString().PadLeft(3, '0'), options.Images[imgi], "imge/jpeg"));
}
//end /package/manifest
contentBuilder.AppendLine("</manifest>");
//package/spine
contentBuilder.AppendLine("<spine toc=\"ncx\">");
//package/spine/itemref(s)
//add the title page to the spine
contentBuilder.AppendLine(GenerateItemRef("titlepage"));
//add the chapters to the spine
foreach (Chapter chapter in options.Chapters)
{
//example same as above
contentBuilder.AppendLine(GenerateItemRef("chapter" + chapter.ChapterNumber.ToString().PadLeft(2, '0')));
}
//end /package/spine
contentBuilder.AppendLine("</spine>");
//end /package
contentBuilder.AppendLine("</package>");
//return the xml in ASCII encoding
return(Encoding.ASCII.GetBytes(contentBuilder.ToString()));
}
public static byte[] GetContentFile(ConversionOptions options)
{
StringBuilder contentBuilder = new StringBuilder();
contentBuilder.AppendLine("<?xml version=\"1.0\"?>");
//package
contentBuilder.AppendFormat("<package version=\"2.0\" xmlns=\"http://www.idpf.org/2007/opf\" unique-identifier=\"{0}\">", options.ID);
//package/metadata
contentBuilder.AppendLine("<metadata xmlns:dc=\"http://purl.org/dc/elements/1.1/\" xmlns:opf=\"http://www.idpf.org/2007/opf\">");
//package/metadata/dc:title & end
contentBuilder.AppendFormat("<dc:title>{0}</dc:title>", options.Title);
//package/metadata/dc:creator & end
contentBuilder.AppendFormat("<dc:creator opf:role=\"aut\">{0}</dc:creator>", options.Author);
//package/metadata/dc:language & end
contentBuilder.AppendLine("<dc:language>en-US</dc:language>"); //assuming en-US
//package/metadata/dc:rights & end
contentBuilder.AppendLine("<dc:rights>Public Domain</dc:rights>");
//package/metadata/dc:publishers & end
contentBuilder.AppendLine("<dc:publisher>Crabwise Studios</dc:publisher>");
//package/metadata/dc:identifier & end
contentBuilder.AppendFormat("<dc:identifier id=\"BookId\">urn:uuid:{0}</dc:identifier>", options.ID);
//end /package/metadata
contentBuilder.AppendLine("</metadata>");
//package/manifest
contentBuilder.AppendLine("<manifest>");
//package/manifest/item(s)
//table of contents file
contentBuilder.AppendLine(GenerateItem("ncx", "toc.ncx", "text-xml"));
//main stylesheet
contentBuilder.AppendLine(GenerateItem("style", "stylesheet.css", "text/css"));
//adobe digital editions page template
contentBuilder.AppendLine(GenerateItem("pagetemplate", "page-template.xpgt", "application/vnd.adobe-page-template+xml"));
//title page
contentBuilder.AppendLine(GenerateItem("titlepage", "title_page.xhtml", "application/xhtml+xml"));
//add the chapters to the manifest
foreach (Chapter chapter in options.Chapters)
{
//example id: chapter01, chapter19
contentBuilder.AppendLine(GenerateItem("chapter" + chapter.ChapterNumber.ToString().PadLeft(2, '0'), chapter.EmbeddedTitle, "application/xhtml+xml"));
}
//add the images to the manifest
for (int imgi = 0; imgi < options.Images.Count; imgi++)
{
//example id: img001, img019, img456
contentBuilder.AppendLine(GenerateItem("img" + imgi.ToString().PadLeft(3, '0'), options.Images[imgi], "imge/jpeg"));
}
//end /package/manifest
contentBuilder.AppendLine("</manifest>");
//package/spine
contentBuilder.AppendLine("<spine toc=\"ncx\">");
//package/spine/itemref(s)
//add the title page to the spine
contentBuilder.AppendLine(GenerateItemRef("titlepage"));
//add the chapters to the spine
foreach (Chapter chapter in options.Chapters)
{
//example same as above
contentBuilder.AppendLine(GenerateItemRef("chapter" + chapter.ChapterNumber.ToString().PadLeft(2, '0')));
}
//end /package/spine
contentBuilder.AppendLine("</spine>");
//end /package
contentBuilder.AppendLine("</package>");
//return the xml in ASCII encoding
return Encoding.ASCII.GetBytes(contentBuilder.ToString());
}