private static int SeparateStructure(PdfDocument document, int startPage, int beforePage, int startPageStructTopIndex
) {
if (!document.IsTagged() || 1 > startPage || startPage > beforePage || beforePage > document.GetNumberOfPages
() + 1) {
return -1;
}
else {
if (beforePage == startPage) {
return startPageStructTopIndex;
}
else {
if (beforePage == document.GetNumberOfPages() + 1) {
return document.GetStructTreeRoot().GetKidsObject().Size();
}
}
}
// Here we separate the structure tree in two parts: struct elems that belong to the pages which indexes are
// less then separateBeforePage and those struct elems that belong to other pages. Some elems might belong
// to both parts and actually these are the ones that we are looking for.
ICollection<PdfObject> firstPartElems = new HashSet<PdfObject>();
for (int i = startPage; i < beforePage; ++i) {
PdfPage pageOfFirstHalf = document.GetPage(i);
ICollection<PdfMcr> pageMcrs = document.GetStructTreeRoot().GetPageMarkedContentReferences(pageOfFirstHalf
);
if (pageMcrs != null) {
foreach (PdfMcr mcr in pageMcrs) {
firstPartElems.Add(mcr.GetPdfObject());
PdfDictionary top = AddAllParentsToSet(mcr, firstPartElems);
if (top != null && top.IsFlushed()) {
throw new PdfException(PdfException.TagFromTheExistingTagStructureIsFlushedCannotAddCopiedPageTags);
}
}
}
}
IList<PdfDictionary> clonedTops = new List<PdfDictionary>();
PdfArray tops = document.GetStructTreeRoot().GetKidsObject();
// Now we "walk" through all the elems which belong to the first part, and look for the ones that contain both
// kids from first and second part. We clone found elements and move kids from the second part to cloned elems.
int lastTopBefore = startPageStructTopIndex - 1;
for (int i = 0; i < tops.Size(); ++i) {
PdfDictionary top = tops.GetAsDictionary(i);
if (firstPartElems.Contains(top)) {
lastTopBefore = i;
StructureTreeCopier.LastClonedAncestor lastCloned = new StructureTreeCopier.LastClonedAncestor();
lastCloned.ancestor = top;
PdfDictionary topClone = top.Clone(ignoreKeysForClone);
topClone.Put(PdfName.P, document.GetStructTreeRoot().GetPdfObject());
lastCloned.clone = topClone;
SeparateKids(top, firstPartElems, lastCloned, document);
if (topClone.ContainsKey(PdfName.K)) {
topClone.MakeIndirect(document);
clonedTops.Add(topClone);
}
}
}
for (int i = 0; i < clonedTops.Count; ++i) {
document.GetStructTreeRoot().AddKidObject(lastTopBefore + 1 + i, clonedTops[i]);
}
return lastTopBefore + 1;
}
private static PdfDictionary ProcessFilters(PdfDictionary dictionary)
{
PdfDictionary result;
// Create a dictionary mapping (i.e. switch statement) to process the expected filters.
var map = new Dictionary <string, Func <byte[], byte[]> >()
{
{ "/FlateDecode", (d) => {
var decoder = new FlateDecode();
return(decoder.Decode(d));
} }
};
// Get all of the filters.
var filters = ((PdfArray)dictionary.Elements["/Filter"])
.Elements.Where(e => e.IsName())
.Select(e => ((PdfName)e).Value)
.ToList();
// If only one filter in array. Just rewrite the /Filter
if (filters.Count == 1)
{
result = dictionary.Clone();
result.Elements["/Filter"] = new PdfName(filters[0]);
return(result);
}
// Process each filter in order. The last filter should be the actual encoded image.
byte[] data = dictionary.Stream.Value;
for (int index = 0; index < (filters.Count - 1); index++)
{
if (!map.ContainsKey(filters[index]))
{
throw new NotSupportedException(String.Format("Encountered embedded image with multiple filters: \"{0}\". Unable to process the filter: \"{1}\".",
String.Join(",", filters), filters[index]));
}
data = map[filters[index]].Invoke(data);
}
result = new PdfDictionary();
result.Elements.Add("/Filter", new PdfName(filters.Last()));
foreach (var element in dictionary.Elements.Where(e => !String.Equals(e.Key, "/Filter", StringComparison.OrdinalIgnoreCase)))
{
result.Elements.Add(element.Key, element.Value);
}
result.CreateStream(data);
return(result);
}
private static void CloneParents(PdfDictionary structElem, StructureTreeCopier.LastClonedAncestor lastCloned
, PdfDocument document) {
if (lastCloned.ancestor != structElem) {
PdfDictionary structElemClone = (PdfDictionary)structElem.Clone(ignoreKeysForClone).MakeIndirect(document);
PdfDictionary currClone = structElemClone;
PdfDictionary currElem = structElem;
while (currElem.Get(PdfName.P) != lastCloned.ancestor) {
PdfDictionary parent = currElem.GetAsDictionary(PdfName.P);
PdfDictionary parentClone = (PdfDictionary)parent.Clone(ignoreKeysForClone).MakeIndirect(document);
currClone.Put(PdfName.P, parentClone);
parentClone.Put(PdfName.K, currClone);
currClone = parentClone;
currElem = parent;
}
PdfStructElem.AddKidObject(lastCloned.clone, -1, currClone);
lastCloned.clone = structElemClone;
lastCloned.ancestor = structElem;
}
}
private static PdfDictionary CopyObject(PdfDictionary source, ICollection <PdfObject> objectsToCopy, PdfDocument
toDocument, IDictionary <PdfDictionary, PdfDictionary> page2page, bool copyFromDestDocument)
{
PdfDictionary copied;
if (copyFromDestDocument)
{
copied = source.Clone(ignoreKeysForCopy);
if (source.IsIndirect())
{
copied.MakeIndirect(toDocument);
}
}
else
{
copied = source.CopyTo(toDocument, ignoreKeysForCopy, true);
}
if (source.ContainsKey(PdfName.Obj))
{
PdfDictionary obj = source.GetAsDictionary(PdfName.Obj);
if (!copyFromDestDocument && obj != null)
{
// Link annotations could be not added to the toDocument, so we need to identify this case.
// When obj.copyTo is called, and annotation was already copied, we would get this already created copy.
// If it was already copied and added, /P key would be set. Otherwise /P won't be set.
obj = obj.CopyTo(toDocument, iText.IO.Util.JavaUtil.ArraysAsList(PdfName.P), false);
copied.Put(PdfName.Obj, obj);
}
}
PdfDictionary pg = source.GetAsDictionary(PdfName.Pg);
if (pg != null)
{
//TODO It is possible, that pg will not be present in the page2page map. Consider the situation,
// that we want to copy structElem because it has marked content dictionary reference, which belongs to the page from page2page,
// but the structElem itself has /Pg which value could be arbitrary page.
copied.Put(PdfName.Pg, page2page.Get(pg));
}
PdfObject k = source.Get(PdfName.K);
if (k != null)
{
if (k.IsArray())
{
PdfArray kArr = (PdfArray)k;
PdfArray newArr = new PdfArray();
for (int i = 0; i < kArr.Size(); i++)
{
PdfObject copiedKid = CopyObjectKid(kArr.Get(i), copied, objectsToCopy, toDocument, page2page, copyFromDestDocument
);
if (copiedKid != null)
{
newArr.Add(copiedKid);
}
}
// TODO new array may be empty or with single element
copied.Put(PdfName.K, newArr);
}
else
{
PdfObject copiedKid = CopyObjectKid(k, copied, objectsToCopy, toDocument, page2page, copyFromDestDocument);
if (copiedKid != null)
{
copied.Put(PdfName.K, copiedKid);
}
}
}
return(copied);
}
/// <summary>
/// Copies structure to a
/// <paramref name="destDocument"/>
/// and insert it in a specified position in the document.
/// <br/><br/>
/// NOTE: Works only for
/// <c>PdfStructTreeRoot</c>
/// that is read from the document opened in reading mode,
/// otherwise an exception is thrown.
/// <br/>
/// Also, to insert a tagged page into existing tag structure, existing tag structure shouldn't be flushed, otherwise
/// an exception may be raised.
/// </summary>
/// <param name="destDocument">document to copy structure to.</param>
/// <param name="insertBeforePage">indicates where the structure to be inserted.</param>
/// <param name="page2page">association between original page and copied page.</param>
public static void CopyTo(PdfDocument destDocument, int insertBeforePage, IDictionary <PdfPage, PdfPage> page2page
, PdfDocument callingDocument)
{
if (!destDocument.IsTagged())
{
return;
}
// Here we separate the structure tree in two parts: struct elems that belong to the pages which indexes are
// less then insertBeforePage and those struct elems that belong to other pages. Some elems might belong
// to both parts and actually these are the ones that we are looking for.
ICollection <PdfObject> firstPartElems = new HashSet <PdfObject>();
PdfStructTreeRoot destStructTreeRoot = destDocument.GetStructTreeRoot();
for (int i = 1; i < insertBeforePage; ++i)
{
PdfPage pageOfFirstHalf = destDocument.GetPage(i);
ICollection <PdfMcr> pageMcrs = destStructTreeRoot.GetPageMarkedContentReferences(pageOfFirstHalf);
if (pageMcrs != null)
{
foreach (PdfMcr mcr in pageMcrs)
{
firstPartElems.Add(mcr.GetPdfObject());
PdfDictionary top = AddAllParentsToSet(mcr, firstPartElems);
if (top.IsFlushed())
{
throw new PdfException(PdfException.TagFromTheExistingTagStructureIsFlushedCannotAddCopiedPageTags);
}
}
}
}
IList <PdfDictionary> clonedTops = new List <PdfDictionary>();
PdfArray tops = destStructTreeRoot.GetKidsObject();
// Now we "walk" through all the elems which belong to the first part, and look for the ones that contain both
// kids from first and second part. We clone found elements and move kids from the second part to cloned elems.
int lastTopBefore = 0;
for (int i_1 = 0; i_1 < tops.Size(); ++i_1)
{
PdfDictionary top = tops.GetAsDictionary(i_1);
if (firstPartElems.Contains(top))
{
lastTopBefore = i_1;
StructureTreeCopier.LastClonedAncestor lastCloned = new StructureTreeCopier.LastClonedAncestor();
lastCloned.ancestor = top;
PdfDictionary topClone = top.Clone(ignoreKeysForClone);
topClone.Put(PdfName.P, destStructTreeRoot.GetPdfObject());
lastCloned.clone = topClone;
SeparateKids(top, firstPartElems, lastCloned);
if (topClone.ContainsKey(PdfName.K))
{
topClone.MakeIndirect(destDocument);
clonedTops.Add(topClone);
}
}
}
for (int i_2 = 0; i_2 < clonedTops.Count; ++i_2)
{
destStructTreeRoot.AddKidObject(lastTopBefore + 1 + i_2, clonedTops[i_2]);
}
CopyTo(destDocument, page2page, callingDocument, false, lastTopBefore + 1);
}
public override PdfObject Visit(
PdfDictionary obj,
object data
)
{
Filter cloneFilter = MatchFilter(obj);
PdfDictionary clone = (PdfDictionary)obj.Clone();
{
clone.entries = new Dictionary<PdfName,PdfDirectObject>();
foreach(KeyValuePair<PdfName,PdfDirectObject> entry in obj.entries)
{
PdfDirectObject sourceValue = entry.Value;
if(cloneFilter.BeforeClone(this, clone, entry.Key, sourceValue))
{
PdfDirectObject cloneValue;
clone[entry.Key] = cloneValue = (PdfDirectObject)(sourceValue != null ? sourceValue.Accept(this, null) : null);
cloneFilter.AfterClone(this, clone, entry.Key, cloneValue);
}
}
}
cloneFilter.AfterClone(this, clone, obj);
return clone;
}
private static PdfDictionary CopyObject(PdfDictionary source, PdfDictionary destPage, bool parentChangePg,
StructureTreeCopier.StructElemCopyingParams copyingParams)
{
PdfDictionary copied;
if (copyingParams.IsCopyFromDestDocument())
{
//TODO: detect wether object is needed to be cloned at all
copied = source.Clone(ignoreKeysForClone);
if (source.IsIndirect())
{
copied.MakeIndirect(copyingParams.GetToDocument());
}
PdfDictionary pg = source.GetAsDictionary(PdfName.Pg);
if (pg != null)
{
if (copyingParams.IsCopyFromDestDocument())
{
if (pg != destPage)
{
copied.Put(PdfName.Pg, destPage);
parentChangePg = true;
}
else
{
parentChangePg = false;
}
}
}
}
else
{
copied = source.CopyTo(copyingParams.GetToDocument(), ignoreKeysForCopy, true);
PdfDictionary obj = source.GetAsDictionary(PdfName.Obj);
if (obj != null)
{
// Link annotations could be not added to the toDocument, so we need to identify this case.
// When obj.copyTo is called, and annotation was already copied, we would get this already created copy.
// If it was already copied and added, /P key would be set. Otherwise /P won't be set.
obj = obj.CopyTo(copyingParams.GetToDocument(), JavaUtil.ArraysAsList(PdfName.P), false);
copied.Put(PdfName.Obj, obj);
}
PdfDictionary nsDict = source.GetAsDictionary(PdfName.NS);
if (nsDict != null)
{
PdfDictionary copiedNsDict = CopyNamespaceDict(nsDict, copyingParams);
copied.Put(PdfName.NS, copiedNsDict);
}
PdfDictionary pg = source.GetAsDictionary(PdfName.Pg);
if (pg != null)
{
PdfDictionary pageAnalog = copyingParams.GetPage2page().Get(pg);
if (pageAnalog == null)
{
pageAnalog = destPage;
parentChangePg = true;
}
else
{
parentChangePg = false;
}
copied.Put(PdfName.Pg, pageAnalog);
}
}
PdfObject k = source.Get(PdfName.K);
if (k != null)
{
if (k.IsArray())
{
PdfArray kArr = (PdfArray)k;
PdfArray newArr = new PdfArray();
for (int i = 0; i < kArr.Size(); i++)
{
PdfObject copiedKid = CopyObjectKid(kArr.Get(i), copied, destPage, parentChangePg, copyingParams);
if (copiedKid != null)
{
newArr.Add(copiedKid);
}
}
if (!newArr.IsEmpty())
{
if (newArr.Size() == 1)
{
copied.Put(PdfName.K, newArr.Get(0));
}
else
{
copied.Put(PdfName.K, newArr);
}
}
}
else
{
PdfObject copiedKid = CopyObjectKid(k, copied, destPage, parentChangePg, copyingParams);
if (copiedKid != null)
{
copied.Put(PdfName.K, copiedKid);
}
}
}
return(copied);
}
