public override void Read(IDocumentReader reader)
{
if (!CanOpenContent)
return;
using (Stream str = m_fileData.BinaryFileData.AsStream())
{
EmailFormatHandler handler = EmailFormatHandler.ReadEmail(str);
if (null == handler)
return;
bool cancel = false;
reader.OnContentData("Subject", handler.Subject, ref cancel);
if (Utils.IsStringHTML(handler.Body))
{
byte[] bytes = Encoding.Unicode.GetBytes(handler.Body);
ReadContent(reader, new HtmlReader(bytes));
}
else
reader.OnContentData("Body", handler.Body, ref cancel);
foreach (string attachment in handler)
{
reader.OnContentData("Attachments", attachment, ref cancel);
}
}
}
public static void Paragraphs(IDocumentReader reader, string typeText, IAbstractTextType textType)
{
List<string> list = _typeList2List(textType);
foreach (string item in list)
{
if (!_isCommentOrParagraphMarker(item))
{
string builtString = _putCDataSection(item);
reader.OnContentData(typeText, builtString, ref MetadataProcessor._cancel);
}
}
}
public static void TrackChanges(IDocumentReader reader, string typeText, IAbstractTextType textType)
{
int trackChangeLimit = MetadataProcessor._TrackChangeLimit;
for (int i = 0; i < textType.GetChildCount() && i < trackChangeLimit; ++i)
{
IAbstractTextNode textNode = textType.GetChild(i);
string content = _nodeValueFromName(textNode, "Content");
string author = _nodeValueFromName(textNode, "Author");
string type = _nodeValueFromName(textNode, "Type");
if (!_isCommentOrParagraphMarker(content))
{
string builtString = @"";
builtString += @"<" + typeText;
builtString += @" Author=""";
builtString += author;
builtString += @"""";
builtString += @" Type=""";
builtString += type;
builtString += @""">";
builtString += _putCDataSection(content);
builtString += @"</" + typeText + @">";
reader.OnContentData(typeText, builtString, ref MetadataProcessor._cancel);
}
}
}
public static void WhiteTexts(IDocumentReader reader, string typeText, IAbstractTextType textType)
{
List<string> list = _typeList2List(textType);
try
{
foreach (string item in list)
{
if (!_isCommentOrParagraphMarker(item) && !_isCommentMarker(item))
{
string builtString = @"";
builtString += @"<" + typeText + @">";
builtString += _putCDataSection(item);
builtString += @"</" + typeText + @">";
reader.OnContentData(typeText, builtString, ref MetadataProcessor._cancel);
}
}
}
catch (System.Exception e)
{
Logger.LogError("Exception within MetadataProcessor::WhiteTexts for " + typeText + ", " + textType);
Logger.LogError(e);
}
}
public virtual void Read(IDocumentReader reader)
{
if (!((CanOpenContent) || (IsEncryptedFileAFormatWeCanReadCustomPropertiesFrom())))
return;
switch (m_fileData.FileType)
{
case FileType.WordDocument:
ReadContent(reader, new WordDocumentReader(m_fileData.BinaryFileData));
break;
case FileType.WordDocumentX:
case FileType.WordDocumentMacroX:
case FileType.WordDocumentTemplateX:
case FileType.WordDocumentMacroTemplateX:
ReadContent(reader, new DocxDocumentReader(m_fileData.BinaryFileData));
break;
case FileType.PowerPoint:
ReadContent(reader, new PptDocumentReader(m_fileData.BinaryFileData));
break;
case FileType.PowerPointX:
case FileType.PowerPointMacroX:
case FileType.PowerPointTemplateX:
case FileType.PowerPointMacroTemplateX:
case FileType.PowerPointShowX:
case FileType.PowerPointMacroShowX:
ReadContent(reader, new PptxDocumentReader(m_fileData.BinaryFileData));
break;
case FileType.ExcelSheet:
ReadContent(reader, new ExcelDocumentReader(m_fileData.BinaryFileData));
break;
case FileType.ExcelSheetX:
case FileType.ExcelSheetMacroX:
case FileType.ExcelSheetTemplateX:
case FileType.ExcelSheetMacroTemplateX:
ReadContent(reader, new XlsxDocumentReader(m_fileData.BinaryFileData));
break;
case FileType.TextDocument:
bool cancel = false;
reader.OnContentData("Paragraph", m_fileData.BinaryFileData.AsString(m_fileData.BinaryFileData.Length, m_fileData.Encoding), ref cancel);
return;
case FileType.HTTPContent:
bool cancelHttp = false;
reader.OnContentData("HTTPContent", m_fileData.BinaryFileData.AsString(m_fileData.BinaryFileData.Length, m_fileData.Encoding), ref cancelHttp);
return;
case FileType.PDFDocument:
ReadContent(reader, new PdfDocumentReader(m_fileData.Filename, m_fileData.Password));
break;
case FileType.RTFDocument:
ReadContent(reader, new RtfReader(m_fileData.BinaryFileData));
break;
case FileType.HTMLDocument:
ReadContent(reader, new HtmlReader(m_fileData.BinaryFileData));
break;
default:
Logger.LogInfo(("Reading unknown file type " + m_fileData.FileType.ToString()));
ReadContent(reader, new FallbackReader(m_fileData.BinaryFileData));
break;
}
}
private static void HiddenSheet(IDocumentReader reader, string typeText, IAbstractTextType textType)
{
List<string> list = _typeList2List(textType);
foreach (string item in list)
{
string builtString = item;
reader.OnContentData(typeText, builtString, ref MetadataProcessor._cancel);
}
}
private static void Links(IDocumentReader reader, string typeText, IAbstractTextType textType)
{
List<KeyValuePair<string, string>> pairList = _typeList2Pair("Type", "Path", textType);
for (int i = 0; i < pairList.Count; ++i)
{
KeyValuePair<string, string> pair = pairList[i];
string builtString = @"";
builtString += @"<" + typeText;
builtString += @" Type=""";
builtString += pair.Key;
builtString += @""">";
builtString += _putCDataSection(pair.Value);
builtString += @"</" + typeText + @">";
reader.OnContentData(typeText, builtString, ref MetadataProcessor._cancel);
}
}
private static void Versions(IDocumentReader reader, string typeText, IAbstractTextType textType)
{
List<KeyValuePair<string, string>> pairList = _typeList2Pair("Saved By", "Comment", textType);
for (int i = 0; i < pairList.Count; ++i)
{
KeyValuePair<string, string> pair = pairList[i];
string builtString = @"";
builtString += @"<" + typeText;
builtString += @" SavedBy=""";
builtString += pair.Key;
builtString += @"""";
builtString += @" Comment=""";
builtString += pair.Value;
builtString += @"""";
builtString += @" />";
reader.OnContentData(typeText, builtString, ref MetadataProcessor._cancel);
}
MetadataProcessor.Authors(pairList);
}
private static void DocumentStatistics(
IDocumentReader reader, string typeText, IAbstractTextType textType)
{
List<KeyValuePair<string, string>> pairList = _typeList2Pair("Name", "Value", textType);
try
{
for (int i = 0; i < pairList.Count; ++i)
{
KeyValuePair<string, string> pair = pairList[i];
string builtString = @"";
builtString += @"<" + typeText;
builtString += @" Name=""";
builtString += pair.Key;
builtString += @"""";
builtString += @" Value=""";
builtString += pair.Value;
builtString += @"""";
builtString += @" />";
reader.OnContentData(typeText, builtString, ref MetadataProcessor._cancel);
}
}
catch (System.Exception e)
{
Logger.LogError("Exception within MetadataProcessor::DocumentStatistics for " + typeText + ", " + textType);
Logger.LogError(e);
}
}
/// -----------------------------------------------------------------------------
private static void BuildInProperties(IDocumentReader reader, string typeText, IAbstractTextType textType)
{
try
{
List<KeyValuePair<string, string>> pairList = _typeList2Pair("Name", "Value", textType);
for (int i = 0; i < pairList.Count; ++i)
{
KeyValuePair<string, string> pair = pairList[i];
if (!_isCommentOrParagraphMarker(pair.Value))
{
string builtString = @"";
builtString += @"<" + typeText;
builtString += @" Name=""";
builtString += pair.Key;
builtString += @""">";
builtString += _putCDataSection(pair.Value);
builtString += @"</" + typeText + @">";
reader.OnContentData(typeText, builtString, ref MetadataProcessor._cancel);
}
}
}
catch (System.Exception e)
{
Logger.LogError("Exception within MetadataProcessor::BuildInProperties for " + typeText + ", " + textType);
Logger.LogError(e);
}
}
private static void HiddenSlide(IDocumentReader reader, string typeText, IAbstractTextType textType)
{
// ppt we get "SlideId" and "Title"
List<KeyValuePair<string, string>> pairList = _typeList2Pair("SlideId", "Title", textType);
// pptx we get "SlideNumber" and "Content"
if (pairList.Count != 0)
CheckForPPTx(ref pairList, "SlideNumber", "Content", textType);
for (int i = 0; i < pairList.Count; ++i)
{
KeyValuePair<string, string> pair = pairList[i];
string builtString = @"";
builtString += @"<" + typeText;
builtString += @" SlideId=""";
builtString += pair.Key;
builtString += @""">";
builtString += _putCDataSection(pair.Value);
builtString += @"</" + typeText + @">";
reader.OnContentData(typeText, builtString, ref MetadataProcessor._cancel);
}
}
private static void SpeakerNote(IDocumentReader reader, string typeText, IAbstractTextType textType)
{
List<KeyValuePair<string, string>> pairList = _typeList2Pair("SlideId", "SpeakerNote", textType);
for (int i = 0; i < pairList.Count; ++i)
{
KeyValuePair<string, string> pair = pairList[i];
reader.OnContentData(typeText, pair.Value, ref MetadataProcessor._cancel);
}
}
private static void RedactedText(IDocumentReader reader, string typeText, IAbstractTextType textType)
{
/*
textNode.GetInfo(0)
name: "Content"
type: String
value: "http://www.csub.edu/"
textNode.GetInfo(1)
name: "SheetName"
type: String
value: "Sheet1"
textNode.GetInfo(2)
name: "Row"
type: Long
value: "67"
textNode.GetInfo(3)
name: "Column"
type: Long
value: "1"
*/
for (int i = 0; i < textType.GetChildCount(); ++i)
{
IAbstractTextNode textNode = textType.GetChild(i);
string content = _nodeValueFromName(textNode, "Content");
string sheet = _nodeValueFromName(textNode, "SheetName");
string row = _nodeValueFromName(textNode, "Row");
string column = _nodeValueFromName(textNode, "Column");
StringBuilder builtString = new StringBuilder(@"");
builtString.Append(@"<" + typeText);
builtString.Append(@" Row=""");
builtString.Append(row);
builtString.Append(@"""");
builtString.Append(@" Column=""");
builtString.Append(column);
builtString.Append(@""">");
builtString.Append(@"<Content>");
builtString.Append(content);
builtString.Append(@"</Content>");
builtString.Append(@"<Sheet>");
builtString.Append(sheet);
builtString.Append(@"</Sheet>");
builtString.Append(@"</" + typeText + @">");
reader.OnContentData(typeText, builtString.ToString(), ref MetadataProcessor._cancel);
}
}
private static void ProcessTextType(IDocumentReader reader, string textTypeStringified, FCS.Lite.Base.IAbstractTextType textType)
{
if (MetadataProcessor.OperMode == MetadataProcessor._OperMode.Professional)
{
Workshare.FCS.Lite.Base.ContentType type = textType.GetContentType();
if (!((type == Workshare.FCS.Lite.Base.ContentType.TextBox) ||
(type == Workshare.FCS.Lite.Base.ContentType.CellText) ||
(type == Workshare.FCS.Lite.Base.ContentType.WorkshareProperty) ||
(type == Workshare.FCS.Lite.Base.ContentType.SmartTag)))
{
MetadataProcessor.ProcessType(reader, textTypeStringified, textType);
return;
}
}
bool cancel = false;
FCS.Lite.Base.IAbstractTextType basicTextType = textType;
if (basicTextType == null)
return;
for (int i = 0; i < textType.GetChildCount(); ++i)
{
FCS.Lite.Base.IAbstractTextNode textNode = basicTextType.GetChild(i);
List<FCS.Lite.Base.NodeInfo>.Enumerator enumer = textNode.InfoEnumerator();
StringBuilder builtString = new StringBuilder();
while (enumer.MoveNext())
{
if (((enumer.Current.name == "Content") && (textTypeStringified != "TrackChange")) ||
((enumer.Current.name == "Comment") && ((textTypeStringified == "Comment") || (textTypeStringified == "CellText"))))
builtString.Append(enumer.Current.value);
else if (string.Compare(textTypeStringified, "TrackChange", true, CultureInfo.InvariantCulture) == 0)
builtString.Append(enumer.Current.value);
}
reader.OnContentData(textTypeStringified, builtString.ToString(), ref cancel);
}
}
private static void RoutingSlip(IDocumentReader reader, string typeText, IAbstractTextType textType)
{
for (int i = 0; i < textType.GetChildCount(); ++i)
{
IAbstractTextNode textNode = textType.GetChild(i);
string message = _nodeValueFromName(textNode, "Message");
string subject = _nodeValueFromName(textNode, "Subject");
string builtString = @"";
builtString += @"<" + typeText;
builtString += @" Message=""";
builtString += message;
builtString += @"""";
builtString += @" Subject=""";
builtString += subject;
builtString += @""">";
builtString += @"<Recipients>";
List<NodeInfo> nodeRecipients = textNode.GetInfo("Recipient");
List<string> recipients = new List<string>();
foreach (NodeInfo nodeRecipi in nodeRecipients)
{
builtString += @"<Recipient>";
builtString += nodeRecipi.value;
builtString += @"</Recipient>";
}
builtString += @"</Recipients>";
builtString += @"</" + typeText + @">";
reader.OnContentData(typeText, builtString, ref MetadataProcessor._cancel);
}
}
private static void Footnotes(IDocumentReader reader, string typeText, IAbstractTextType textType)
{
List<string> list = _typeList2List(textType);
foreach (string item in list)
{
if (!_isCommentOrParagraphMarker(item) && !_isCommentMarker(item))
{
string builtString = @"";
builtString += @"<" + typeText + @">";
builtString += _putCDataSection(item);
builtString += @"</" + typeText + @">";
reader.OnContentData(typeText, builtString, ref MetadataProcessor._cancel);
}
}
}
private static void Macros(IDocumentReader reader, string typeText, IAbstractTextType textType)
{
List<string> list = _typeList2List(textType);
try
{
foreach (string item in list)
{
string builtString = item;
reader.OnContentData(typeText, builtString, ref MetadataProcessor._cancel);
}
}
catch (System.Exception e)
{
Logger.LogError("Exception within MetadataProcessor::Macros for " + typeText + ", " + textType);
Logger.LogError(e);
}
}
private static void HyperLinks(IDocumentReader reader, string typeText, IAbstractTextType textType)
{
/*
DOC
textNode.GetInfo(0)
name: "Content"
type: String
value: "HYPERLINK"
XLS
textNode.GetInfo(0)
name: "Description"
type: String
value: "www.bbc.co.uk/news"
textNode.GetInfo(1)
name: "Path"
type: String
value: "http://www.bbc.co.uk/news"
PPT
textNode.GetInfo(0)
name: "Content"
type: String
value: "HL 1"
textNode.GetInfo(1)
name: "Path"
type: String
value: "http://www.btyahoo.com/welcome2"
*/
/// List<KeyValuePair<string, string>> pairList = _typeList2Pair(textType);
for (int i = 0; i < textType.GetChildCount(); ++i)
{
IAbstractTextNode textNode = textType.GetChild(i);
string keyStr = textNode.GetInfo(0).name;
string valStr = (textNode.GetInfoCount() > 1) ?
textNode.GetInfo(1).value : textNode.GetInfo(0).value;
string builtString = @"";
builtString += @"<" + typeText;
builtString += @" Type=""";
builtString += keyStr;
builtString += @""">";
builtString += _putCDataSection(valStr);
builtString += @"</" + typeText + @">";
reader.OnContentData(typeText, builtString, ref MetadataProcessor._cancel);
}
}
private static void Fields(IDocumentReader reader, string typeText, IAbstractTextType textType)
{
List<KeyValuePair<string, string>> pairList = _typeList2Pair("Instruction", "Content", textType);
try
{
for (int i = 0; i < pairList.Count; ++i)
{
KeyValuePair<string, string> pair = pairList[i];
// if (!_isCommentOrParagraphMarker(pair.Value))
if (!_isHyperlink(pair.Value)) // hyperlinks added separately
{
string builtString = @"";
builtString += @"<" + typeText;
builtString += @" Instruction=""";
builtString += pair.Key;
builtString += @""">";
builtString += pair.Value;
builtString += @"</" + typeText + @">";
reader.OnContentData(typeText, builtString, ref MetadataProcessor._cancel);
}
}
}
catch (System.Exception e)
{
Logger.LogError("Exception within MetadataProcessor::Fields for " + typeText + ", " + textType);
Logger.LogError(e);
}
}
private static void AutoVersion(IDocumentReader reader, string typeText, IAbstractTextType textType)
{
string builtString = _typeList2String(textType);
reader.OnContentData(typeText, builtString, ref MetadataProcessor._cancel);
}
private static void AttachedTemplate(IDocumentReader reader, string typeText, IAbstractTextType textType)
{
if (textType.GetChildCount() > 0)
{
IAbstractTextNode textNode = textType.GetChild(0);
string path = _nodeValueFromName(textNode, "Path");
string name = _nodeValueFromName(textNode, "Name");
string builtString = @"";
builtString += @"<" + typeText;
builtString += @" Name=""";
builtString += name;
builtString += @"""";
builtString += @" Path=""";
builtString += path;
builtString += @"""";
builtString += @" />";
reader.OnContentData(typeText, builtString, ref MetadataProcessor._cancel);
}
}
private static void HiddenColumn(IDocumentReader reader, string typeText, IAbstractTextType textType)
{
List<KeyValuePair<string, string>> pairList = _typeList2Pair("SheetName", "Column", textType);
for (int i = 0; i < pairList.Count; ++i)
{
KeyValuePair<string, string> pair = pairList[i];
string builtString = @"";
builtString += @"<" + typeText;
builtString += @" Column=""";
builtString += pair.Value;
builtString += @""">";
builtString += pair.Key;
builtString += @"</" + typeText + @">";
reader.OnContentData(typeText, builtString, ref MetadataProcessor._cancel);
}
}
public static void HiddenTexts(IDocumentReader reader, string typeText, IAbstractTextType textType)
{
try
{
for (int loopi = 0; loopi < textType.GetChildCount(); ++loopi)
{
IAbstractTextNode textNode = textType.GetChild(loopi);
string content = _nodeValueFromName(textNode, "Content");
if (!_isCommentOrParagraphMarker(content))
{
string builtString = @"";
builtString += @"<" + typeText;
for (int loopj = 0; loopj < textNode.GetInfoCount(); ++loopj)
{
NodeInfo nodeInfo = textNode.GetInfo(loopj);
string nodeValue = (nodeInfo != null) ? nodeInfo.value : "";
string nodeKey = (nodeInfo != null) ? nodeInfo.name : "";
if (!nodeKey.Equals("Content"))
{
builtString += @" ";
builtString += nodeKey;
builtString += @"=""";
builtString += nodeValue;
builtString += @"""";
}
}
builtString += @">";
builtString += _putCDataSection(content);
builtString += @"</" + typeText + @">";
reader.OnContentData(typeText, builtString, ref MetadataProcessor._cancel);
}
}
}
catch (System.Exception e)
{
Logger.LogError("Exception within MetadataProcessor::HiddenTexts for " + typeText + ", " + textType);
Logger.LogError(e);
}
}
