protected virtual List <KeyValuePair <string, byte[]> > FixImages(HtmlNode div)
{
var imgNodes = div.Descendants("img");// .SelectNodes("//img");
var images = new List <KeyValuePair <string, byte[]> >();
foreach (var node in imgNodes)
{
var imagePath = node.GetAttributeValue("data-original", "");
if (string.IsNullOrEmpty(imagePath))
{
imagePath = node.GetAttributeValue("src", "");
}
var imageFile = System.IO.Path.GetFileName(imagePath);
if (!FileNameSanitizer.IsBadName(imageFile))
{
var imageBytesTask = webber.DownloadFile(imagePath);
byte[] imageBytes = null;
try
{
imageBytes = imageBytesTask.Result;
if (imageBytesTask.Status != System.Threading.Tasks.TaskStatus.Faulted)
{
images.Add(new KeyValuePair <string, byte[]>(imageFile, imageBytes));
}
node.SetAttributeValue("src", imageFile); // modify the name in source
}
catch (System.AggregateException ex)
{
System.Console.WriteLine(ex);
// node.RemoveChild(node);
}
finally
{
}
}
}
return(images);
}
// include 3 files: html (content), opf (main file) and ncx (content file)
// http://www.aliciaramirez.com/2014/05/how-to-make-a-kindle-ebook-from-scratch/
public string CreateKindleFiles(string firstpageUrl, bool forNote = false)
{
// 1. get the book base on website structure
book = website.CheckBookDownloaded(firstpageUrl);
// 2. assign 3 file names
var title = VietnameseAccentsRemover.RemoveSign4VietnameseString(book.Title);
title = title.Replace(" ", "_");
title = FileNameSanitizer.GiveGoodName(title).Trim();
if (string.IsNullOrEmpty(title))
{
return(string.Empty);
}
var downloadFolder = "";
var trashFolder = "";
// get os type and set download path and kindle file
if (Environment.GetEnvironmentVariable("_system_name") == "OSX")
{
trashFolder = Path.Combine(Environment.GetEnvironmentVariable("HOME"), ".Trash");
downloadFolder = Path.Combine(Environment.GetEnvironmentVariable("HOME"), "Downloads");
KindlegenPath = "/Users/kiettran/Downloads/kindlegen";
CalibrePath = "/Applications/calibre.app/Contents/MacOS/calibredb";
}
else // windows
{
CalibrePath = @"C:\Program Files (x86)\Calibre2\calibredb.exe";
#if DEBUG
trashFolder = Path.Combine(Environment.GetEnvironmentVariable("USERPROFILE"), "Downloads");
downloadFolder = Path.Combine(Environment.GetEnvironmentVariable("USERPROFILE"), "Downloads");
// Logger logger = NLog.LogManager.GetCurrentClassLogger();
// logger.Info($"-------------DOWNLOAD FOLDER IS: {downloadFolder} --------------");
KindlegenPath = @"C:\Kiet\Kinh\kindlegen.exe";
#else
var today = System.DateTime.Today.ToString("yyyy-MM-dd-ddd");
trashFolder = Path.Combine(@"C:\Kiet\Kinh\", today);
downloadFolder = trashFolder;
if (!Directory.Exists(downloadFolder))
{
Directory.CreateDirectory(downloadFolder);
}
// Logger logger = NLog.LogManager.GetCurrentClassLogger();
// logger.Info($"-------------DOWNLOAD FOLDER IS: {downloadFolder} --------------");
KindlegenPath = @"C:\Kiet\Kinh\kindlegen.exe";
// if release
#endif
}
// in case we use Website project to reference to it
if (!string.IsNullOrEmpty(DownloadFolder))
{
downloadFolder = DownloadFolder;
}
// then download a whole book
book = website.GetOneWholeHtml(firstpageUrl);
// clean up double quote in file name
title = title.Replace("\"", "_");
htmlFilename = Path.Combine(trashFolder, title + ".html");
ncxFilename = Path.Combine(Path.GetDirectoryName(htmlFilename), Path.GetFileNameWithoutExtension(htmlFilename) + ".ncx");
opfBookFilename = Path.Combine(Path.GetDirectoryName(htmlFilename), Path.GetFileNameWithoutExtension(htmlFilename) + ".opf");
mobiBookFilename = Path.Combine(Path.GetDirectoryName(htmlFilename), Path.GetFileNameWithoutExtension(htmlFilename) + ".mobi");
var downloadMobiFileName = Path.Combine(downloadFolder, title) + ".mobi";
// check if mobi file exists, then skip it
if (File.Exists(downloadMobiFileName))
{
return(downloadMobiFileName);
}
// 3. create 3 files
if (!forNote)
{
CreateHtmlFile();
CreateNCXTableOfContent();
}
else
{
// for note
CreateHtmlFileFromNoteHtml();
CreateNCXTableOfContentForNoteHtml();
}
CreateOPFBookDetail();
// 4. Create mobile kindle file
if (!string.IsNullOrWhiteSpace(KindlegenPath))
{
CreateKindleFile();
AddFileToCalibre();
// we don't use these files, delete it for now
htmlFilename = Path.Combine(trashFolder, htmlFilename);
ncxFilename = Path.Combine(trashFolder, ncxFilename);
opfBookFilename = Path.Combine(trashFolder, opfBookFilename);
if (File.Exists(htmlFilename))
{
File.Delete(htmlFilename);
}
if (File.Exists(ncxFilename))
{
File.Delete(ncxFilename);
}
if (File.Exists(opfBookFilename))
{
File.Delete(opfBookFilename);
}
foreach (var image in book.Chapters.SelectMany(c => c.Images))
{
var imageFilename = Path.Combine(trashFolder, image.Key);
if (File.Exists(imageFilename))
{
File.Delete(imageFilename);
}
}
//var trashMobiFileName = Path.Combine(trashFolder, title) + ".mobi";
if (mobiBookFilename != downloadMobiFileName) // on Windows, we don't need to copy because I didn't check for trash folder
{
if (File.Exists(downloadMobiFileName))
{
File.Delete(downloadMobiFileName);
}
if (File.Exists(mobiBookFilename))
{
File.Move(mobiBookFilename, downloadMobiFileName);
}
}
}
return(Path.Combine(downloadFolder, title) + ".mobi");
}
// include 3 files: html (content), opf (main file) and ncx (content file)
// http://www.aliciaramirez.com/2014/05/how-to-make-a-kindle-ebook-from-scratch/
public string CreateKindleFiles(string firstpageUrl, bool forNote = false)
{
// 1. get the book base on website structure
book = website.GetOneWholeHtml(firstpageUrl);
// 2. assign 3 file names
var title = VietnameseAccentsRemover.RemoveSign4VietnameseString(book.Title);
title = title.Replace(" ", "_");
title = FileNameSanitizer.GiveGoodName(title).Trim();
if (string.IsNullOrEmpty(title))
{
return(string.Empty);
}
var downloadFolder = "";
var trashFolder = "";
// get os type and set download path and kindle file
if (Environment.GetEnvironmentVariable("_system_name") == "OSX")
{
trashFolder = Path.Combine(Environment.GetEnvironmentVariable("HOME"), ".Trash");
downloadFolder = Path.Combine(Environment.GetEnvironmentVariable("HOME"), "Downloads");
KindlegenPath = "/Users/kiettran/Downloads/kindlegen";
}
else // windows
{
trashFolder = Path.Combine(Environment.GetEnvironmentVariable("USERPROFILE"), "Downloads");
downloadFolder = Path.Combine(Environment.GetEnvironmentVariable("USERPROFILE"), "Downloads");
KindlegenPath = @"C:\Kiet\Kinh\kindlegen.exe";
}
// in case we use Website project to reference to it
if (!string.IsNullOrEmpty(DownloadFolder))
{
downloadFolder = DownloadFolder;
}
htmlFilename = Path.Combine(trashFolder, title) + ".html";
ncxFilename = Path.Combine(Path.GetDirectoryName(htmlFilename), Path.GetFileNameWithoutExtension(htmlFilename) + ".ncx");
opfBookFilename = Path.Combine(Path.GetDirectoryName(htmlFilename), Path.GetFileNameWithoutExtension(htmlFilename) + ".opf");
// 3. create 3 files
if (!forNote)
{
CreateHtmlFile();
CreateNCXTableOfContent();
}
else
{
// for note
CreateHtmlFileFromNoteHtml();
CreateNCXTableOfContentForNoteHtml();
}
CreateOPFBookDetail();
// 4. Create mobile kindle file
if (!string.IsNullOrWhiteSpace(KindlegenPath))
{
CreateKindleFile();
// we don't use these files, delete it for now
htmlFilename = Path.Combine(trashFolder, htmlFilename);
ncxFilename = Path.Combine(trashFolder, ncxFilename);
opfBookFilename = Path.Combine(trashFolder, opfBookFilename);
if (File.Exists(htmlFilename))
{
File.Delete(htmlFilename);
}
if (File.Exists(ncxFilename))
{
File.Delete(ncxFilename);
}
if (File.Exists(opfBookFilename))
{
File.Delete(opfBookFilename);
}
foreach (var image in book.Chapters.SelectMany(c => c.Images))
{
var imageFilename = Path.Combine(trashFolder, image.Key);
if (File.Exists(imageFilename))
{
File.Delete(imageFilename);
}
}
var trashMobiFileName = Path.Combine(trashFolder, title) + ".mobi";
var downloadMobiFileName = Path.Combine(downloadFolder, title) + ".mobi";
if (trashMobiFileName != downloadMobiFileName) // on Windows, we don't need to copy because I didn't check for trash folder
{
if (File.Exists(downloadMobiFileName))
{
File.Delete(downloadMobiFileName);
}
File.Move(trashMobiFileName, downloadMobiFileName);
}
}
return(Path.Combine(downloadFolder, title) + ".mobi");
}