public HtmlParser(IHtmlDocumentFactory htmlDocumentFactory, IMetaKeywordsTagParser metaKeywordsTagParser,
IBodyParser bodyParser)
{
_htmlDocumentFactory = htmlDocumentFactory;
_metaKeywordsTagParser = metaKeywordsTagParser;
_bodyParser = bodyParser;
}
private static void ConvertContentTo(IHtmlNode node, TextWriter outText, IHtmlDocumentFactory documentFactory)
{
foreach (var subnode in node.ChildNodes)
{
ConvertHtmlToText(subnode, outText, documentFactory);
}
}
//CONSTRUCTOR
public WordReferenceScraper(IHtmlDocumentFactory htmlDocumentFactory, IWordReferenceURLBuilder urlBuilder, ICaretAnalyser caretAnalyser)
{
this.htmlDocumentFactory = htmlDocumentFactory;
this.wordReferenceURLBuilder = urlBuilder;
this.caretAnalyser = caretAnalyser;
}
protected BasicHtmlParser(IHtmlDocumentFactory htmlDocumentFactory)
{
if (htmlDocumentFactory == null)
{
throw new ArgumentNullException(nameof(htmlDocumentFactory));
}
this.HtmlDocumentFactory = htmlDocumentFactory;
}
public HtmlParser(IHtmlDocumentFactory htmlDocumentFactory, IHttpEncoder httpEncoder) : base(htmlDocumentFactory)
{
if (httpEncoder == null)
{
throw new ArgumentNullException(nameof(httpEncoder));
}
this.HttpEncoder = httpEncoder;
}
/// <summary>
/// Действие по инициализации.
/// </summary>
/// <param name="moduleProvider">Провайдер модулей.</param>
protected override async ValueTask <Nothing> OnInitialize(IModuleProvider moduleProvider)
{
await base.OnInitialize(moduleProvider);
_htmlDocumentFactory = await moduleProvider.QueryModuleAsync <IHtmlDocumentFactory>() ?? throw new ModuleNotFoundException(typeof(IHtmlDocumentFactory));
_youtubeIdService = await moduleProvider.QueryModuleAsync <IYoutubeIdService>() ?? throw new ModuleNotFoundException(typeof(IYoutubeIdService));
_linkParser = await moduleProvider.QueryEngineCapabilityAsync <IEngineLinkParser>(MakabaConstants.MakabaEngineId) ?? throw new ModuleNotFoundException(typeof(IEngineLinkParser));
return(Nothing.Value);
}
/// <summary>
/// Конвертировать HTML в текст.
/// </summary>
/// <param name="node"></param>
/// <param name="outText"></param>
/// <param name="documentFactory">Фабрика документов.</param>
public static void ConvertHtmlToText(IHtmlNode node, TextWriter outText, IHtmlDocumentFactory documentFactory)
{
if (documentFactory == null)
{
throw new ArgumentNullException(nameof(documentFactory));
}
switch (node)
{
case IHtmlCommentNode _:
// don't output comments
break;
case IHtmlTextNode tn:
// script and style must not be output
string parentName = tn.ParentNode.Name;
if ((parentName == "script") || (parentName == "style"))
{
break;
}
// get text
var html = tn.Text;
// is it in fact a special closing node output as text?
if (documentFactory.IsOverlappedClosingElement(html))
{
break;
}
// check the text is meaningful and not a bunch of whitespaces
if (html.Trim().Length > 0)
{
outText.Write(documentFactory.DeEntitize(html));
}
break;
default:
switch (node.Name)
{
case "p":
// treat paragraphs as crlf
outText.WriteLine();
break;
}
if (node.HasChildNodes)
{
ConvertContentTo(node, outText, documentFactory);
}
break;
}
}
/// <summary>
/// Действие по инициализации.
/// </summary>
/// <param name="moduleProvider">Провайдер модулей.</param>
protected override async ValueTask <Nothing> OnInitialize(IModuleProvider moduleProvider)
{
await base.OnInitialize(moduleProvider);
_htmlParser = await moduleProvider.QueryEngineCapabilityAsync <IHtmlParser>(MakabaConstants.MakabaEngineId) ?? throw new ModuleNotFoundException(typeof(IHtmlParser));
_htmlDocumentFactory = await moduleProvider.QueryModuleAsync <IHtmlDocumentFactory>() ?? throw new ModuleNotFoundException(typeof(IHtmlDocumentFactory));
_postsParser = await moduleProvider.FindNetworkDtoParserAsync <BoardPost2WithParentLink, IBoardPost>() ?? throw new ModuleNotFoundException(typeof(INetworkDtoParser <BoardPost2WithParentLink, IBoardPost>));
_threadPreviewParser = await moduleProvider.FindNetworkDtoParserAsync <ThreadPreviewData, IThreadPreviewPostCollection>() ?? throw new ModuleNotFoundException(typeof(INetworkDtoParser <ThreadPreviewData, IThreadPreviewPostCollection>));
return(Nothing.Value);
}
public DefaultHtmlTransformingInitializer(IHtmlInvestigator htmlInvestigator, IHtmlDocumentFactory htmlDocumentFactory, IHtmlTransformingContext htmlTransformingContext)
{
if(htmlInvestigator == null)
throw new ArgumentNullException("htmlInvestigator");
if(htmlDocumentFactory == null)
throw new ArgumentNullException("htmlDocumentFactory");
if(htmlTransformingContext == null)
throw new ArgumentNullException("htmlTransformingContext");
this._htmlDocumentFactory = htmlDocumentFactory;
this._htmlInvestigator = htmlInvestigator;
this._htmlTransformingContext = htmlTransformingContext;
}
public TagBuilderParser(IHtmlDocumentFactory htmlDocumentFactory) : base(htmlDocumentFactory)
{
}
public ResultParser(IHtmlDocumentFactory htmlDocumentFactory)
{
_htmlDocumentFactory = htmlDocumentFactory;
}
