public static ISiteCrawler Create(SiteParameter siteParameter)
{
if (siteParameter == null)
{
throw new ArgumentNullException(nameof(siteParameter));
}
string dataServiceName = GetValueOrDefault(siteParameter.CustomProcessors, "IDataService");
string htmlReaderName = GetValueOrDefault(siteParameter.CustomProcessors, "IHtmlReader");
string pageParserName = GetValueOrDefault(siteParameter.CustomProcessors, "IPageParser");
string itemReaderName = GetValueOrDefault(siteParameter.CustomProcessors, "IItemReader");
string pageReaderName = GetValueOrDefault(siteParameter.CustomProcessors, "IPageReader");
IHtmlReader htmlReader = Container.Resolve <IHtmlReader>(htmlReaderName);
ParameterOverride htmlReaderParameter = new ParameterOverride("htmlReader", htmlReader);
ParameterOverride siteParameterParameter = new ParameterOverride("siteParameter", siteParameter);
IItemReader itemReader = Container.Resolve <IItemReader>(itemReaderName, siteParameterParameter);
ParameterOverride itemReaderParameter = new ParameterOverride("itemReader", itemReader);
IPageReader pageReader = Container.Resolve <IPageReader>(pageReaderName, siteParameterParameter, htmlReaderParameter, itemReaderParameter);
IPageParser pageParser = Container.Resolve <IPageParser>(pageParserName, siteParameterParameter, htmlReaderParameter);
return(new GeneralSiteCrawler(pageReader, pageParser));
}
public MessageFetcher(IPageReader remote, ICheckpointReader remotePos, IMemoryStreamManager streamManager, string streamName)
{
_remote = remote;
_remotePos = remotePos;
_streamManager = streamManager;
StreamName = streamName;
}
private static Image GetModifiedImage(IPageReader pageReader)
{
var rawBytes = pageReader.GetImage();
var width = pageReader.GetPageWidth();
var height = pageReader.GetPageHeight();
var characters = pageReader.GetCharacters();
using (var bmp = new Bitmap(width, height, PixelFormat.Format32bppArgb))
{
var background = new Bitmap(width, height, PixelFormat.Format32bppArgb);
bmp.AddBytes(rawBytes);
using (var g = Graphics.FromImage(background))
{
g.Clear(Color.White);
g.DrawImageUnscaled(bmp, Point.Empty);
}
//bmp.DrawRectangles(characters);
return(background);
}
}
public GeneralSiteCrawler(SiteParameter siteParameter)
{
IItemReader itemReader = new RegexItemReader(siteParameter);
IHtmlReader htmlReader = new HttpClientReader();
this.pageReader = new SequentialPageReader(siteParameter, htmlReader, itemReader);
this.pageParser = new RegexPageParser(siteParameter, htmlReader);
}
public PageReadStream(IPageReader reader, long start, long max, byte[] buffer)
{
_reader = reader;
_max = max;
_buffer = buffer;
_position = start;
_mem = new MemoryStream(buffer);
_mem.SetLength(0);
}
public PageReadStream(IPageReader reader, long start, long max, byte[] buffer)
{
_reader = reader;
_max = max;
_buffer = buffer;
_position = start;
_mem = new MemoryStream(buffer);
_mem.SetLength(0);
}
public DevourTarget(int attempts, Uri uri, IPageReader reader, int lifes = 1)
{
if (uri == null || reader == null)
{
throw new ArgumentException("Bad arguments");
}
Attempts = attempts;
Lifes = lifes;
_uri = uri;
_reader = reader;
}
public MessageCopier(
IPageReader sourceReader,
ICheckpointReader sourcePos,
IMemoryStreamManager streamManager,
IPageWriter targetWriter,
ICheckpointWriter targetPos)
{
_sourceReader = sourceReader;
SourcePos = sourcePos;
_streamManager = streamManager;
_targetWriter = targetWriter;
TargetPos = targetPos;
}
public void PageToImage(int pageIndex, int width, int height)
{
IDocReader docReader = DocLib.Instance.GetDocReader(Context.Document, new PageDimensions(width, height));
IPageReader pageReader = docReader.GetPageReader(pageIndex);
MemoryStream mem = new MemoryStream();
Image <Bgra32> image = Image.LoadPixelData <Bgra32>(pageReader.GetImage(), pageReader.GetPageWidth(), pageReader.GetPageHeight());
image.SaveAsPng(mem);
Context.Result = new ImageResult(mem.ToArray(), MimeTypes.Png);
}
public GeneralSiteCrawler(IPageReader pageReader, IPageParser pageParser, IDataService dataService)
{
this.pageReader = pageReader ?? throw new ArgumentNullException(nameof(pageReader));
this.pageParser = pageParser ?? throw new ArgumentNullException(nameof(pageParser));
this.dataService = dataService ?? throw new ArgumentNullException(nameof(dataService));
this.pageParser.SetErrorHandler((url, exception) =>
this.dataService.AddLog(new CrawlerLog
{
Url = url,
LogTime = DateTime.Now,
Message = exception.Message
}));
}
public GeneralSiteCrawler(SiteParameter siteParameter)
{
this.dataService = new DbDataService(CrawlerDbHelper.GetContext());
IItemReader itemReader = new RegexItemReader(siteParameter);
IHtmlReader htmlReader = new HttpClientReader();
this.pageReader = new SequentialPageReader(siteParameter, htmlReader, itemReader);
this.pageParser = new RegexPageParser(siteParameter, htmlReader);
this.pageParser.SetErrorHandler((url, exception) =>
this.dataService.AddLog(new CrawlerLog
{
Url = url,
LogTime = DateTime.Now,
Message = exception.Message
}));
}
private static byte[] GetModifiedImage(IPageReader pageReader)
{
var rawBytes = pageReader.GetImage(RenderFlags.RenderAnnotations | RenderFlags.Grayscale);
var width = pageReader.GetPageWidth();
var height = pageReader.GetPageHeight();
using (var bmp = new Bitmap(width, height, PixelFormat.Format32bppArgb))
{
bmp.AddBytes(rawBytes);
using (var stream = new MemoryStream())
{
bmp.Save(stream, ImageFormat.Png);
return(stream.ToArray());
}
}
}
/// <summary>
/// Creates a new instance of <see cref="ContainerReader"/>.
/// </summary>
/// <param name="stream">The <see cref="Stream"/> to read.</param>
/// <param name="closeOnDispose"><c>True</c> to close the stream when disposed, otherwise <c>false</c>.</param>
/// <exception cref="ArgumentException"><paramref name="stream"/>'s <see cref="Stream.CanSeek"/> is <c>False</c>.</exception>
public ContainerReader(Stream stream, bool closeOnDispose)
{
if (stream == null)
{
throw new ArgumentNullException(nameof(stream));
}
_packetProviders = new List <WeakReference <Contracts.IPacketProvider> >();
if (stream.CanSeek)
{
_reader = CreatePageReader(stream, closeOnDispose, ProcessNewStream);
CanSeek = true;
}
else
{
_reader = CreateForwardOnlyPageReader(stream, closeOnDispose, ProcessNewStream);
}
}
private static byte[] GetModifiedImage(IPageReader pageReader)
{
var rawBytes = pageReader.GetImage();
var width = pageReader.GetPageWidth();
var height = pageReader.GetPageHeight();
var characters = pageReader.GetCharacters();
using (var bmp = new Bitmap(width, height, PixelFormat.Format32bppArgb))
{
bmp.AddBytes(rawBytes);
//bmp.DrawRectangles(characters);
using (var stream = new MemoryStream())
{
bmp.Save(stream, ImageFormat.Png);
return(stream.ToArray());
}
}
}
public MessageReader(ICheckpointReader position, IPageReader messages)
{
Position = position;
Messages = messages;
_buffer = new byte[Limit];
}
public MessageReader(ICheckpointReader position, IPageReader messages) {
_position = position;
_messages = messages;
_buffer = new byte[Limit];
}
/// <summary>
/// Cleans up
/// </summary>
public void Dispose()
{
_reader?.Dispose();
_reader = null;
}
public PageReadStreamTests()
{
_mem = new MemoryStream();
_given = new BinaryWriter(_mem, Encoding.UTF8, true);
_reader = new MemoryPageReader(_mem);
}
public DevourTarget(int attempts, Uri uri, IPageReader reader, int lifes = 1)
{
if (uri == null || reader == null)
throw new ArgumentException("Bad arguments");
Attempts = attempts;
Lifes = lifes;
_uri = uri;
_reader = reader;
}
public PageReadStreamTests()
{
_mem = new MemoryStream();
_given = new BinaryWriter(_mem, Encoding.UTF8, true);
_reader = new MemoryPageReader(_mem);
}
public GeneralSiteCrawler(IPageReader pageReader, IPageParser pageParser)
{
this.pageReader = pageReader ?? throw new ArgumentNullException(nameof(pageReader));
this.pageParser = pageParser ?? throw new ArgumentNullException(nameof(pageParser));
}
public PageReader(IPageReader pageReader, IDocReader doc2)
{
this.pageReader = pageReader;
this.doc2 = doc2;
}