Esempio n. 1
0
        /// <summary>
        /// Creates an epub from the given urls.
        /// </summary>
        /// <remarks>
        /// This method makes use of a record created for each html that is given by a url.
        /// The records contain information from the html needed to create the various files which
        /// constitute the epub format. An immutable record is created initially so the data collected can be used
        /// accross various parts of the program. For example, the record contains a mapping (src map) for each image.
        /// This maps the current src of the image to a new local one - this is then used to change the sources of the
        /// html images to a local directory and also to download the images to the relevant local directory.
        /// </remarks>
        /// <param name="urls"></param>
        /// <param name="rootDirectory">Path in which the book will be created</param>
        /// <param name="bookTitle">Name of the book</param>
        /// <param name="guid">Unique identifier for the folder</param>
        /// <returns></returns>
        public async Task FromAsync(IEnumerable <string> urls, string rootDirectory, string bookTitle, Guid guid)
        {
            var htmlDocs = _htmlInput.GetHtmlDocumentsFromAsync(urls, new HtmlWeb());

            var directoryPaths    = GetDirectoryContext(rootDirectory, guid);
            var createDirectories = _epubOutput.CreateDirectoriesAsync(directoryPaths);

            var wikiPageRecords =
                (await htmlDocs).Select(doc => _getRecords.From(doc, "image_repo")).ToList();
            var downloadImages =
                Task.WhenAll(wikiPageRecords.SelectMany(record => _epubOutput.DownloadImages(record, directoryPaths)));

            var createEpubDocuments = Task.WhenAll(_getEpubDocs.From(wikiPageRecords, bookTitle, directoryPaths));

            // Save the produced files to relevant directory and compress them
            await createDirectories;
            var   saveDocuments  = DocumentSaver.Save(await createEpubDocuments);
            var   createMimeType = _epubOutput.CreateMimeFile(directoryPaths);

            Task.WaitAll(saveDocuments, createMimeType, downloadImages);
            await _epubOutput.ZipFiles(directoryPaths, guid);
        }
Esempio n. 2
0
        public async Task FromAsync(IEnumerable <string> urls, string rootDirectory, string bookTitle, Guid folderID)
        {
            Task <HtmlDocument[]> initialDocs = _htmlInput.GetHtmlDocumentsFromAsync(urls, new HtmlWeb());

            var  directories       = GetDirectoryContext(rootDirectory, folderID);
            Task createDirectories = _epubOutput.CreateDirectoriesAsync(directories);

            List <(HtmlDocument doc, WikiPageRecord record)> htmlRecordTuple =
                (await initialDocs).Select(doc => (doc, _getRecords.From(doc, "image_repo"))).ToList();

            var  pageRecords        = htmlRecordTuple.Select(t => t.record);
            Task downloadImages     = Task.WhenAll(pageRecords.SelectMany(record => _epubOutput.DownLoadImages(record, directories)));
            var  getXmlDocs         = Task.WhenAll(_getXmlDocs.From(pageRecords, bookTitle));
            var  getParsedDocuments = Task.WhenAll(htmlRecordTuple.Select(t => (_parseHtml.ParseAsync(t.doc, t.record))));

            await createDirectories;
            Task  createMime = _epubOutput.CreateMimeFile(directories);
            Task  saveXml    = _epubOutput.SaveDocumentsAsync(directories, (await getXmlDocs));
            Task  saveHtml   = _epubOutput.SaveDocumentsAsync(directories, (await getParsedDocuments));

            Task.WaitAll(saveXml, saveHtml, createMime, downloadImages);
            await _epubOutput.ZipFiles(directories, folderID);
        }