private bool IsExpectedFromRedaction(IAbstractTextType tt)
{
if (tt == null)
return true;
int iChildCount = tt.GetChildCount();
ContentType contType = tt.GetContentType();
switch (tt.GetContentType())
{
case ContentType.Header:
case ContentType.Footer:
case ContentType.Endnote:
case ContentType.Footnote:
case ContentType.Paragraph:
case ContentType.WorkshareProperty:
case ContentType.CellText:
case ContentType.HiddenColumn:
case ContentType.HiddenRow:
return true; // we don't clean any of these
case ContentType.TextBox:
if (IsPowerPoint() || IsExcel())
return true;
break;
case ContentType.AttachedTemplate:
if (IsThereASignificantAttachedTemplate(tt))
return false;
return true ;
case ContentType.Field:
if (AreThereAnyNonQuoteFields(tt))
return false;
return true;
case ContentType.HiddenText:
if (TypeHasNonWhitespaceContent(tt))
return false;
return true;
}
if (tt.GetChildCount() > 0)
return false;
return true;
}
static private List<KeyValuePair<string, string>> _typeList2Pair(
string key, string value, IAbstractTextType textType)
{
List<KeyValuePair<string, string>> pairList = new List<KeyValuePair<string, string>>();
try
{
for (int i = 0; i < textType.GetChildCount(); ++i)
{
IAbstractTextNode textNode = textType.GetChild(i);
string keyStr = _nodeValueFromName(textNode, key);
string valStr = _nodeValueFromName(textNode, value);
pairList.Add(new KeyValuePair<string, string>(keyStr, valStr));
}
}
catch( System.Exception e )
{
Logger.LogError(e);
}
return pairList;
}
public static string DumpTextType(IAbstractTextType ttOrig)
{
StringBuilder bld = new StringBuilder();
for (int i = 0; i < ttOrig.GetChildCount(); i++)
{
bld.Append(DumpNode(ttOrig.GetChild(i)));
}
return bld.ToString();
}
public static void CheckNode(IAbstractTextType ttype, string expectedName, string expectedValue)
{
for (int i = 0; i < ttype.GetChildCount(); i++)
{
IAbstractTextNode iAbstractTextNode = ttype.GetChild(i);
Assert.IsNotNull(iAbstractTextNode);
Assert.Greater(iAbstractTextNode.GetInfo("Name").Count, 0, "missing info for the property name");
if (expectedName == iAbstractTextNode.GetInfo("Name")[0].value)
{
Assert.Greater(iAbstractTextNode.GetInfo("Value").Count, 0, "missing info for the property value");
Assert.AreEqual(expectedValue, iAbstractTextNode.GetInfo("Value")[0].value, "Built in property name incorrect");
return;
}
}
Assert.Fail("Didn't find the property type expected - " + expectedName);
}
static private List<string> _typeList2List(IAbstractTextType textType)
{
List<string> list = new List<string>();
try
{
for (int i = 0; i < textType.GetChildCount(); ++i)
{
IAbstractTextNode textNode = textType.GetChild(i);
for (int j = 0; j < textNode.GetInfoCount(); ++j)
{
NodeInfo nodeInfo = textNode.GetInfo(j);
string nodeValue = (nodeInfo != null) ? nodeInfo.value : "";
list.Add(EscapeXML(nodeValue));
}
}
}
catch (System.Exception e)
{
Logger.LogError(e);
}
return list;
}
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);
}
}
static private List<KeyValuePair<string, string>> _typeList2Pair(IAbstractTextType textType)
{
List<KeyValuePair<string, string>> pairList = new List<KeyValuePair<string, string>>();
try
{
for (int i = 0; i < textType.GetChildCount(); ++i)
{
IAbstractTextNode textNode = textType.GetChild(i);
for (int j = 0; j < textNode.GetInfoCount(); ++j)
{
NodeInfo nodeInfo = textNode.GetInfo(j);
pairList.Add(new KeyValuePair<string, string>(nodeInfo.name, EscapeXML(nodeInfo.value)));
}
}
}
catch( System.Exception e )
{
Logger.LogError(e);
}
return pairList;
}
private bool HadJustThisDocumentMacro(IAbstractTextType ttOrig)
{
if (ttOrig.GetChildCount() != 1)
return false;
return (ttOrig.GetChild(0).GetInfo("Name")[0].value == "ThisDocument");
}
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);
}
}
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);
}
}
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);
}
}
}
static private string _typeList2String(IAbstractTextType textType)
{
NodeInfo nodeInfo = null;
if (textType.GetChildCount() > 0)
{
IAbstractTextNode textNode = textType.GetChild(0);
if (textNode.GetInfoCount() > 0)
{
nodeInfo = textNode.GetInfo(0);
}
}
string strValue = (nodeInfo == null) ? "" : nodeInfo.value;
return EscapeXML(strValue);
}
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 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 bool AreAllTextBoxesPictures(IAbstractTextType ttOrig)
{
for (int i = 0; i < ttOrig.GetChildCount(); i++)
{
if (!IsEmbeddingOrEmpty(ttOrig.GetChild(i)))
return false;
}
return true;
}
private void CheckContentNode(IAbstractTextType ttype, string expectedValue)
{
for (int i = 0; i < ttype.GetChildCount(); i++)
{
IAbstractTextNode iAbstractTextNode = ttype.GetChild(i);
Assert.IsNotNull(iAbstractTextNode);
Assert.Greater(iAbstractTextNode.GetInfo("Content").Count, 0, "missing info for the property name");
string thisValue = iAbstractTextNode.GetInfo("Content")[0].value;
if (expectedValue == thisValue)
return;
}
Assert.Fail("Didn't find the content expected - " + expectedValue);
}
private bool IsThereASignificantAttachedTemplate(IAbstractTextType tt)
{
if (tt.GetChildCount() == 0)
return false;
for (int i = 0; i < tt.GetChildCount(); i++)
{
IAbstractTextNode node = tt.GetChild(i);
for (int j = 0; j < node.GetInfoCount(); j++)
{
NodeInfo ni = node.GetInfo(j);
if (ni.name == "Content")
{
if (ni.value.Trim() != "")
return true;
}
}
}
return false;
}
private bool IsAllWhiteSpace(IAbstractTextType ttOrig)
{
for (int i = 0; i < ttOrig.GetChildCount(); i++)
{
if (!IsAllWhiteSpace(ttOrig.GetChild(i)))
return false;
}
return true;
}
private bool TypeHasNonWhitespaceContent(IAbstractTextType tt)
{
for (int i = 0; i < tt.GetChildCount(); i++)
{
IAbstractTextNode node = tt.GetChild(i);
for (int j = 0; j < node.GetInfoCount(); j++)
{
NodeInfo ni = node.GetInfo(j);
if (ni.name == "Content")
{
if (!string.IsNullOrEmpty(ni.value.Trim()))
return true;
}
}
}
return false;
}
private bool AreThereAnyNonQuoteFields(IAbstractTextType tt)
{
for (int i = 0; i < tt.GetChildCount(); i++)
{
IAbstractTextNode node = tt.GetChild(i);
for (int j = 0; j < node.GetInfoCount(); j++)
{
NodeInfo ni = node.GetInfo(j);
if (ni.name == "Content")
{
if (ni.value != "QUOTE")
return true;
}
}
}
return false;
}