Exemple #1
0
        /**************************************************************************/

        public static string CleanDocumentText(MacroscopeDocument msDoc)
        {
            string CleanedText = msDoc.GetDocumentTextRaw();

            if (!string.IsNullOrEmpty(CleanedText))
            {
                try
                {
                    CleanedText = HtmlEntity.DeEntitize(CleanedText);
                }
                catch (System.Collections.Generic.KeyNotFoundException ex)
                {
                    DebugMsgStatic(string.Format("CleanDocumentText: {0}", ex.Message));
                    msDoc.AddRemark("CleanDocumentText", "Possibly contains invalid HTML Entities.");
                }
                catch (Exception ex)
                {
                    DebugMsgStatic(string.Format("CleanDocumentText: {0}", ex.Message));
                    msDoc.AddRemark("CleanDocumentText", "Possibly contains invalid HTML Entities.");
                }

                CleanedText = CleanText(Text: CleanedText);
            }

            return(CleanedText);
        }