private static PdfDocument OpenDocument(IInputBytes inputBytes, ISeekableTokenScanner scanner, ILog log, bool isLenientParsing,
IReadOnlyList <string> passwords, bool clipPaths)
{
var filterProvider = DefaultFilterProvider.Instance;
CrossReferenceTable crossReferenceTable = null;
var xrefValidator = new XrefOffsetValidator(log);
// We're ok with this since our intent is to lazily load the cross reference table.
// ReSharper disable once AccessToModifiedClosure
var locationProvider = new ObjectLocationProvider(() => crossReferenceTable, inputBytes);
var pdfScanner = new PdfTokenScanner(inputBytes, locationProvider, filterProvider, NoOpEncryptionHandler.Instance);
var crossReferenceStreamParser = new CrossReferenceStreamParser(filterProvider);
var crossReferenceParser = new CrossReferenceParser(log, xrefValidator, crossReferenceStreamParser);
var version = FileHeaderParser.Parse(scanner, isLenientParsing, log);
var crossReferenceOffset = FileTrailerParser.GetFirstCrossReferenceOffset(inputBytes, scanner,
isLenientParsing) + version.OffsetInFile;
// TODO: make this use the scanner.
var validator = new CrossReferenceOffsetValidator(xrefValidator);
crossReferenceOffset = validator.Validate(crossReferenceOffset, scanner, inputBytes, isLenientParsing);
crossReferenceTable = crossReferenceParser.Parse(inputBytes, isLenientParsing,
crossReferenceOffset,
version.OffsetInFile,
pdfScanner,
scanner);
var(rootReference, rootDictionary) = ParseTrailer(crossReferenceTable, isLenientParsing,
pdfScanner,
out var encryptionDictionary);
var encryptionHandler = encryptionDictionary != null ?
(IEncryptionHandler) new EncryptionHandler(encryptionDictionary, crossReferenceTable.Trailer, passwords)
: NoOpEncryptionHandler.Instance;
pdfScanner.UpdateEncryptionHandler(encryptionHandler);
var cidFontFactory = new CidFontFactory(pdfScanner, filterProvider);
var encodingReader = new EncodingReader(pdfScanner);
var type1Handler = new Type1FontHandler(pdfScanner, filterProvider, encodingReader);
var fontFactory = new FontFactory(log, new Type0FontHandler(cidFontFactory,
filterProvider, pdfScanner),
new TrueTypeFontHandler(log, pdfScanner, filterProvider, encodingReader, SystemFontFinder.Instance,
type1Handler),
type1Handler,
new Type3FontHandler(pdfScanner, filterProvider, encodingReader));
var resourceContainer = new ResourceStore(pdfScanner, fontFactory);
var information = DocumentInformationFactory.Create(pdfScanner, crossReferenceTable.Trailer);
var catalog = CatalogFactory.Create(rootReference, rootDictionary, pdfScanner, isLenientParsing);
var pageFactory = new PageFactory(pdfScanner, resourceContainer, filterProvider,
new PageContentParser(new ReflectionGraphicsStateOperationFactory()),
log);
var caching = new ParsingCachingProviders(resourceContainer);
var acroFormFactory = new AcroFormFactory(pdfScanner, filterProvider, crossReferenceTable);
var bookmarksProvider = new BookmarksProvider(log, pdfScanner);
return(new PdfDocument(log, inputBytes, version, crossReferenceTable, caching, pageFactory, catalog, information,
encryptionDictionary,
pdfScanner,
filterProvider,
acroFormFactory,
bookmarksProvider,
clipPaths));
}
private static PdfDocument OpenDocument(IInputBytes inputBytes, ISeekableTokenScanner scanner, IContainer container, bool isLenientParsing, string password)
{
var log = container.Get <ILog>();
var filterProvider = container.Get <IFilterProvider>();
var catalogFactory = new CatalogFactory();
var cMapCache = new CMapCache(new CMapParser());
CrossReferenceTable crossReferenceTable = null;
var bruteForceSearcher = new BruteForceSearcher(inputBytes);
var xrefValidator = new XrefOffsetValidator(log);
var objectChecker = new XrefCosOffsetChecker(log, bruteForceSearcher);
// We're ok with this since our intent is to lazily load the cross reference table.
// ReSharper disable once AccessToModifiedClosure
var locationProvider = new ObjectLocationProvider(() => crossReferenceTable, bruteForceSearcher);
var pdfScanner = new PdfTokenScanner(inputBytes, locationProvider, filterProvider, NoOpEncryptionHandler.Instance);
var crossReferenceStreamParser = new CrossReferenceStreamParser(filterProvider);
var crossReferenceParser = new CrossReferenceParser(log, xrefValidator, objectChecker, crossReferenceStreamParser, new CrossReferenceTableParser());
var version = FileHeaderParser.Parse(scanner, isLenientParsing, log);
var crossReferenceOffset = container.Get <FileTrailerParser>().GetFirstCrossReferenceOffset(inputBytes, scanner, isLenientParsing);
// TODO: make this use the scanner.
var validator = new CrossReferenceOffsetValidator(xrefValidator);
crossReferenceOffset = validator.Validate(crossReferenceOffset, scanner, inputBytes, isLenientParsing);
crossReferenceTable = crossReferenceParser.Parse(inputBytes, isLenientParsing, crossReferenceOffset, pdfScanner, scanner);
var trueTypeFontParser = new TrueTypeFontParser();
var fontDescriptorFactory = new FontDescriptorFactory();
var compactFontFormatIndexReader = new CompactFontFormatIndexReader();
var compactFontFormatParser = new CompactFontFormatParser(new CompactFontFormatIndividualFontParser(compactFontFormatIndexReader, new CompactFontFormatTopLevelDictionaryReader(),
new CompactFontFormatPrivateDictionaryReader()), compactFontFormatIndexReader);
var rootDictionary = ParseTrailer(crossReferenceTable, isLenientParsing, pdfScanner, out var encryptionDictionary);
var encryptionHandler = encryptionDictionary != null ? (IEncryptionHandler) new EncryptionHandler(encryptionDictionary, crossReferenceTable.Trailer, password ?? string.Empty)
: NoOpEncryptionHandler.Instance;
pdfScanner.UpdateEncryptionHandler(encryptionHandler);
var cidFontFactory = new CidFontFactory(pdfScanner, fontDescriptorFactory, trueTypeFontParser, compactFontFormatParser, filterProvider);
var encodingReader = new EncodingReader(pdfScanner);
var fontFactory = new FontFactory(log, new Type0FontHandler(cidFontFactory,
cMapCache,
filterProvider, pdfScanner),
new TrueTypeFontHandler(log, pdfScanner, filterProvider, cMapCache, fontDescriptorFactory, trueTypeFontParser, encodingReader, new SystemFontFinder(new TrueTypeFontParser())),
new Type1FontHandler(pdfScanner, cMapCache, filterProvider, fontDescriptorFactory, encodingReader,
new Type1FontParser(new Type1EncryptedPortionParser()), compactFontFormatParser),
new Type3FontHandler(pdfScanner, cMapCache, filterProvider, encodingReader));
var resourceContainer = new ResourceContainer(pdfScanner, fontFactory);
var pageFactory = new PageFactory(pdfScanner, resourceContainer, filterProvider,
new PageContentParser(new ReflectionGraphicsStateOperationFactory()),
new XObjectFactory(), log);
var informationFactory = new DocumentInformationFactory();
var information = informationFactory.Create(pdfScanner, crossReferenceTable.Trailer);
var catalog = catalogFactory.Create(pdfScanner, rootDictionary);
var caching = new ParsingCachingProviders(bruteForceSearcher, resourceContainer);
var acroFormFactory = new AcroFormFactory(pdfScanner, filterProvider);
return(new PdfDocument(log, inputBytes, version, crossReferenceTable, isLenientParsing, caching, pageFactory, catalog, information,
encryptionDictionary,
pdfScanner,
filterProvider,
acroFormFactory));
}
private async void OpenButton_Click(object sender, RoutedEventArgs e)
{
// Open a text file.
Windows.Storage.Pickers.FileOpenPicker open =
new Windows.Storage.Pickers.FileOpenPicker();
open.SuggestedStartLocation =
Windows.Storage.Pickers.PickerLocationId.DocumentsLibrary;
open.FileTypeFilter.Add("*");
Windows.Storage.StorageFile file = await open.PickSingleFileAsync();
if (file != null)
{
//PlainText.Text = await FileIO.ReadTextAsync(file);
var provider = new StorageFileDataProvider();
var bytes = await provider.LoadDataAsync(file);
var reader = new EncodingReader();
reader.AddBytes(bytes);
byte[] bom = null;
try
{
ByteOrderMarks.ToList().ForEach(pair =>
{
var(key, value) = pair;
if (!bytes.AsSpan(0, value.Length).StartsWith(value.AsSpan(0, value.Length)))
{
return;
}
var encoding = key switch
{
ByteOrderMark.Utf8 => Encoding.UTF8,
ByteOrderMark.Utf16Be => Encoding.BigEndianUnicode,
ByteOrderMark.Utf16Le => Encoding.Unicode,
ByteOrderMark.Utf32Be => Encoding.UTF32,
ByteOrderMark.Utf32Le => Encoding.UTF32,
ByteOrderMark.Utf7A => Encoding.UTF7,
ByteOrderMark.Utf7B => Encoding.UTF7,
ByteOrderMark.Utf7C => Encoding.UTF7,
ByteOrderMark.Utf7D => Encoding.UTF7,
ByteOrderMark.Utf7E => Encoding.UTF7,
_ => Encoding.ASCII
};
documentViewModel.CurrentEncoding = encoding;
reader = new EncodingReader();
reader.AddBytes(bytes.AsSpan().Slice(value.Length).ToArray());
bom = value;
});
}
catch (Exception ex)
{
Logger.LogError(new EventId(), $"Error loading {file.Name}.", ex);
Settings.Status(ex.Message, TimeSpan.FromSeconds(60), Verbosity.Error);
}
var text = reader.Read(documentViewModel.CurrentEncoding);
}
}
