public void WriteObject(IPdfObject obj, string trailerName = null)
{
if (gXref.ContainsKey(obj.id))
{
throw new PBException("error object already in file id {0}", obj.id);
}
gXref.Add(obj.id, new PdfXref {
objectId = obj.id, generationNumber = obj.generationNumber, filePosition = gw.Position
});
if (trailerName != null)
{
gTrailer[trailerName] = new PdfNValue {
name = trailerName, value = new PdfValueObjectRef {
valueObjectId = obj.id, valueObjectGenerationNumber = obj.generationNumber
}
}
}
;
_WriteObject(obj);
//gw.Write("{0} {1} obj\n", obj.id, obj.generationNumber);
//WriteValue(obj.value);
//gw.Write("\n");
//if (obj.stream != null)
//{
// gw.Write("stream\n");
// gw.Write(obj.stream);
// gw.Write("\n");
// gw.Write("endstream\n");
//}
//gw.Write("endobj\n");
}
public static void Test_PdfExportDeflatedStream_01(string pdfFile, int idObject, string streamFile)
{
_tr.WriteLine("export stream object {0} of pdf file \"{1}\" to \"{2}\"", idObject, pdfFile, streamFile);
PB_Pdf.PdfReader pr = null;
Writer w = null;
try
{
pr = new PB_Pdf.PdfReader(pdfFile);
IPdfObject obj = ReadPdfObject(pr, idObject, "Object");
w = new Writer(streamFile, FileMode.Create);
if (obj.deflatedStream != null)
{
w.Write(obj.deflatedStream);
}
}
finally
{
if (pr != null)
{
pr.Close();
}
if (w != null)
{
w.Close();
}
}
}
public void UpdateObject(IPdfObject obj, string trailerName = null)
{
if (!gUpdatePdf)
{
throw new PBException("error object update is not activate \"{0}\"", gw.File);
}
if (!gXref.ContainsKey(obj.id))
{
throw new PBException("error update object {0}, object dont exist", obj.id);
}
gw.Seek(gUpdatePosition);
gXref[obj.id] = new PdfXref {
objectId = obj.id, generationNumber = obj.generationNumber, filePosition = gUpdatePosition
};
if (trailerName != null)
{
gTrailer[trailerName] = new PdfNValue {
name = trailerName, value = new PdfValueObjectRef {
valueObjectId = obj.id, valueObjectGenerationNumber = obj.generationNumber
}
}
}
;
_WriteObject(obj);
gUpdatePosition = (int)gw.Position;
}
public static void Test_PdfUpdateStream_01(string pdfFile, int objectId, string streamFile)
{
_tr.WriteLine("update stream for object {0} with \"{1}\" of pdf file \"{2}\"", objectId, streamFile, pdfFile);
PB_Pdf.PdfWriter pw = null;
Reader r = null;
try
{
pw = new PB_Pdf.PdfWriter(pdfFile, updatePdf: true);
IPdfObject obj = pw.reader.ReadObject(objectId);
r = new Reader(streamFile);
obj.stream = r.ReadBytes((int)r.Length);
PdfNValue length = obj.value["Length"];
if (length == null || !length.value.isInt())
{
throw new PBException("error wrong /Length of object {0}", obj.id);
}
length.value.valueInt = obj.stream.Length;
pw.UpdateObject(obj);
}
finally
{
if (r != null)
{
r.Close();
}
if (pw != null)
{
pw.Close();
}
}
}
private void ApplyConfig(IntPtr config, IPdfObject settings, bool isGlobal)
{
if (settings == null)
{
return;
}
var properties = settings.GetType().GetProperties(BindingFlags.Instance | BindingFlags.Public | BindingFlags.NonPublic);
foreach (var propertyInfo in properties)
{
var attrs = propertyInfo.GetCustomAttributes().ToArray();
var propValue = propertyInfo.GetValue(settings);
if (propValue == null)
{
continue;
}
if (attrs.Length > 0 && attrs[0] is WkHtmlAttribute)
{
var attr = (WkHtmlAttribute)attrs[0];
Apply(config, attr.Name, propValue, isGlobal);
}
else if (propValue is IPdfObject)
{
ApplyConfig(config, propValue as IPdfObject, isGlobal);
}
}
}
public PdfContents(IPdfObject obj)
: base(PdfObjectType.Contents)
{
IsContainer = true;
if (obj is PdfIndirectReference)
{
obj = (obj as PdfIndirectReference).ReferencedObject.Object;
}
PdfArray array = obj as PdfArray;
PdfStream stream = obj as PdfStream;
if (array != null)
{
Streams = array.Items;
}
else
{
if (stream != null)
{
var list = new List<IPdfObject>();
list.Add(stream);
Streams = list.AsReadOnly();
}
else
{
throw new Exception("Contents must be either a stream or an array of streams");
}
}
}
public static PdfXRefSection Parse(Lexer lexer)
{
int firstId = int.Parse(lexer.ReadToken(), CultureInfo.InvariantCulture);
int size = int.Parse(lexer.ReadToken(), CultureInfo.InvariantCulture);
var entries = new IPdfObject[size];
for (int i = 0; i < size; i++)
{
var entry = PdfXRefEntry.Parse(firstId + i, lexer);
// first entry must be free and have a gen 65535
// head of the linked list of free objects
if (i == 0)
{
if (entry.GenerationNumber != 65535)
{
throw new ParsingException($"The first xref entry must have generation number 65535. Was {entry.GenerationNumber}");
}
if (entry.InUse)
{
throw new ParsingException($"The first xref entry must be free");
}
}
entries[i] = entry;
}
return(new PdfXRefSection(firstId, size, entries));
}
private PdfIndirectObject(int objectNumber, int generationNumber, IPdfObject obj)
: base(PdfObjectType.IndirectObject)
{
IsContainer = true;
ObjectNumber = objectNumber;
GenerationNumber = generationNumber;
Object = obj;
}
internal void Register(IPdfObject pdfObject)
{
pdfObject.ObjectId = ++m_CurrentId;
m_Objects[m_CurrentId] = pdfObject;
if (pdfObject is IPdfXObject)
{
((IPdfXObject)pdfObject).XObjectId = ++m_CurrentId;
m_Objects[m_CurrentId] = pdfObject;
}
}
internal void Register(IPdfObject pdfObject)
{
pdfObject.ObjectId = ++m_CurrentId;
m_Objects[m_CurrentId] = pdfObject;
if (pdfObject is IPdfXObject)
{
((IPdfXObject)pdfObject).XObjectId = ++m_CurrentId;
m_Objects[m_CurrentId] = pdfObject;
}
}
public static void Test_GetPdfText_01(String file)
{
_tr.WriteLine("read pdf file \"{0}\"", file);
PB_Pdf.PdfReader pr = new PB_Pdf.PdfReader(file);
try
{
IPdfObject obj = ReadPdfObject(pr, 1138, "Object");
}
finally
{
pr.Close();
}
}
private void WriteObject(IPdfObject obj)
{
if (obj is IPdfStream)
{
WriteStreamObject(obj as IPdfStream);
}
else if (obj is IPdfTypedObject)
{
WriteDictionary((obj as IPdfTypedObject).Dict);
}
else if (obj is PdfDictionary)
{
WriteDictionary(obj as PdfDictionary);
}
else if (obj is PdfArray)
{
WriteArray(obj as PdfArray);
}
else if (obj is PdfReference)
{
WriteReference(obj as PdfReference);
}
else if (obj is PdfName)
{
AsciiToOutput($"{(obj as PdfName).Name}");
}
else if (obj is PdfString)
{
WriteString(obj as PdfString);
}
else if (obj is PdfBoolean)
{
WriteBoolean(obj as PdfBoolean);
}
else if (obj is PdfNumeric)
{
AsciiToOutput(obj.ToString());
}
else if (obj is PdfHexString)
{
WriteHexString(obj as PdfHexString);
}
else if (obj is PdfNull)
{
AsciiToOutput("null");
}
else
{
throw new Exception("Unknown object while writing");
}
}
public static void PrintPdfObject(IPdfObject o)
{
_tr.WriteLine("object id {0} generation {1}", o.id, o.generationNumber);
if (o.value.isObject())
{
foreach (PdfNValue value in o.value.objectValues.Values)
{
_tr.WriteLine(" {0}", value);
}
}
else if (o.value.isArray())
{
int i = 0;
foreach (IPdfValue value in o.value.arrayValues)
{
_tr.WriteLine(" [{0}] = {1}", i++, value);
}
}
else
{
_tr.WriteLine(" {0}", o.value);
}
if (o.stream != null)
{
_tr.WriteLine(" stream {0} bytes, deflated stream {1}", o.stream.Length, o.deflatedStream.Length);
}
if (o.stream != null && logStream)
{
////PrintStream(o.stream);
//byte[] stream = o.deflatedStream;
////_tr.WriteLine("deflated stream {0} bytes", stream.Length);
////PrintStream(stream);
//StringBuilder sb = new StringBuilder();
//bool cr = false;
//foreach (byte b in stream)
//{
// if (b == 10 && !cr)
// sb.Append('\r');
// if (b == 13)
// cr = true;
// else
// cr = false;
// sb.Append((char)b);
//}
_tr.WriteLine("stream as string :");
//_tr.WriteLine(sb.ToString());
_tr.WriteLine(o.deflatedStream.zzToString());
_tr.WriteLine();
}
}
private void _WriteObject(IPdfObject obj)
{
gw.Write("{0} {1} obj\n", obj.id, obj.generationNumber);
WriteValue(obj.value);
gw.Write("\n");
if (obj.stream != null)
{
gw.Write("stream\n");
gw.Write(obj.stream);
gw.Write("\n");
gw.Write("endstream\n");
}
gw.Write("endobj\n");
}
public static IPdfObject ReadPdfObject(PB_Pdf.PdfReader pr, int id, string name)
{
_tr.WriteLine();
//int id = ((PdfValueObject)(pr.Trailer.values["Root"].value)).id;
_tr.WriteLine("read {0} object {1}", name, id);
IPdfObject o = pr.ReadObject(id);
_tr.WriteLine("{0} object {1}", name, id);
PrintPdfObject(o);
if (o.source != null && logSource)
{
_tr.WriteLine("Source {0} object {1}", name, id);
_tr.WriteLine(o.source);
}
return(o);
}
public static PdfStream Parse(PdfDictionary dictionary, Lexer lexer)
{
if (dictionary is null)
{
throw new ArgumentNullException(nameof(dictionary));
}
lexer.Expects("stream");
char eol = lexer.ReadChar();
if (eol == '\r')
{
eol = lexer.ReadChar();
}
if (eol != '\n')
{
throw new ParsingException($@"Stream must end with either \r\n or \n. Was '{eol}'");
}
IPdfObject lengthObject = dictionary["Length"];
if (lengthObject is null)
{
throw new ParsingException("Stream dictionary is missing 'Length' entry");
}
int length = 0;
if (lengthObject is PdfIndirectReference reference)
{
PdfIndirectObject lenobj = lexer.IndirectReferenceResolver
.GetObject(reference.ObjectNumber, reference.GenerationNumber);
length = ((PdfNumeric)lenobj.Object).ToInt32();
}
else
{
length = int.Parse(lengthObject.ToString(), CultureInfo.InvariantCulture);
}
var data = PdfData.Parse(lexer, length);
lexer.Expects("endstream");
return(new PdfStream(dictionary, data));
}
public SerializationResult SerializeTree(PdfWriter writer, IPdfObject root)
{
var references = new List <Guid>();
var indRefs = new IndirectObjectDictionary();
// Add empty GUID to represent the "free" object
references.Add(Guid.Empty);
var offsets = new List <int>();
// need to visit all nodes to reserve references for each node
var nodesToSerialize = new ConcurrentQueue <IPdfObject>();
var nodesToVisit = new Stack <IPdfObject>();
nodesToVisit.Push(root);
while (nodesToVisit.Count > 0)
{
var node = nodesToVisit.Pop();
if (node == null || indRefs.Contains(node))
{
continue;
}
nodesToVisit.PushRange(node.GetChildren());
indRefs.Add(node);
nodesToSerialize.Enqueue(node);
}
while (nodesToSerialize.TryDequeue(out IPdfObject node))
{
offsets.Add((int)writer.Position);
writer.WriteLine($"{indRefs[node].Identifier} 0 {PdfTokens.StartObject}");
SerdesFactory.GetFor(node).Serialize(writer, node, indRefs);
writer.WriteLine(PdfTokens.EndObject);
}
return(new SerializationResult()
{
Offsets = offsets,
References = indRefs
});
}
private PdfArray ParseArray()
{
var ary = new PdfArray();
Token peek = _lexer.Peek();
while (peek.Type != TokenType.EndArray)
{
IPdfObject obj = ParseObject();
ary.Add(obj);
peek = _lexer.Peek();
}
_lexer.Next();
Expect(TokenType.EndArray);
return(ary);
}
public static void Test_PdfExportDeflatedStream_02(String pdfFile, int objectId, string streamFile)
{
Writer w = null;
_tr.WriteLine("export stream object {0} pdf file \"{1}\" to \"{2}\"", objectId, pdfFile, streamFile);
PB_Pdf.PdfReader pr = new PB_Pdf.PdfReader(pdfFile);
try
{
IPdfObject obj = ReadPdfObject(pr, objectId, "Object");
if (obj.deflatedStream == null)
{
_tr.WriteLine("no stream");
return;
}
_tr.WriteLine("read data");
//IPdfInstruction[] instructions = PdfDataReader.PdfReadAll(obj.deflatedStream);
PdfDataReader pdr = new PdfDataReader(obj.deflatedStream);
//pdr.Trace += new TraceDelegate(_tr.WriteLine);
IPdfInstruction[] instructions = pdr.ReadAll();
_tr.WriteLine("{0} instructions", instructions.Length);
//var q = from instruction in instructions select instruction.opeString;
//foreach (string opeString in q.Distinct().OrderBy(opeString => opeString))
//{
// _tr.WriteLine(opeString);
//}
w = new Writer(streamFile, FileMode.Create);
foreach (IPdfInstruction instruction in instructions)
{
//_tr.WriteLine(instruction);
//w.WriteLine(instruction.ToString());
instruction.Export(w);
}
}
finally
{
if (w != null)
{
w.Close();
}
pr.Close();
}
}
private PdfDictionary ParseDictionary()
{
var dict = new PdfDictionary();
_lexer.Next();
while (!Accept(TokenType.EndDict))
{
Expect(TokenType.Name);
var key = new PdfName()
{
Name = (string)_lexer.Current.Value
};
IPdfObject obj = ParseObject();
dict.Add(key, obj);
_lexer.Next();
}
return(dict);
}
public PdfContents(IPdfObject obj)
: base(PdfObjectType.Contents)
{
IsContainer = true;
if (obj is PdfIndirectReference reference)
{
obj = reference.ReferencedObject.Object;
}
if (obj is PdfArray array)
{
Streams = array.Items;
}
else if (obj is PdfStream stream)
{
Streams = new[] { stream };
}
else
{
throw new Exception("Contents must be either a stream or an array of streams");
}
}
internal IPdfObject ResolveReference(PdfReference reference)
{
PdfCrossReference indirectObj = Xref.Get(reference.ObjectNumber, reference.Generation);
if (indirectObj == null)
{
return(null);
}
if (!indirectObj.Loaded)
{
Parser.ParseIndirectObjectDefinition(indirectObj);
}
IPdfObject obj = indirectObj.Object;
if (obj is PdfReference)
{
return(ResolveReference(obj as PdfReference));
}
return(obj);
}
public IPdfObject ReadObject(int id)
{
// trailer
// <<
// /Size 5274
// /Root 5273 0 R
// /Info 5272 0 R
// >>
// 5273 0 obj
// <<
// /Type /Catalog
// /Pages 1 0 R
// /OpenAction [3 0 R /FitH null]
// /PageLayout /OneColumn
// >>
// endobj
// 3 0 obj
// <</Type /Page
// /Parent 1 0 R
// /MediaBox [0 0 907.09 1292.59]
// /Resources 2 0 R
// /Contents 4 0 R>>
// endobj
if (!gXref.ContainsKey(id))
{
throw new PBException("error unknow object {0}", id);
}
PdfXref xref = gXref[id];
gpsr.StartReadObject(string.Format("object {0}", id));
gpsr.Seek(xref.filePosition);
// 5273 0 obj
Regex rg_object_begin = new Regex("^([0-9]+) ([0-9]+) obj$", RegexOptions.Compiled);
PdfObjectDictionary o = new PdfObjectDictionary(id, xref.generationNumber);
string s = gpsr.ReadLine();
Match match = rg_object_begin.Match(s);
if (!match.Success || int.Parse(match.Groups[1].Value) != id || int.Parse(match.Groups[2].Value) != xref.generationNumber)
{
throw new PBException("error reading {0} line {1} \"{2}\"", gpsr.ObjectName, gpsr.LineNumber, s);
}
s = gpsr.ReadLine();
o.value = ReadPdfValue(ref s);
if (s == "")
{
s = gpsr.ReadLine();
}
if (s == "stream")
{
int length = 0;
PdfNValue lengthValue = null;
if (o.value.isObject())
{
lengthValue = o.value["Length"];
}
if (lengthValue == null)
{
throw new PBException("error stream without /Length reading {0} line {1} \"{2}\"", gpsr.ObjectName, gpsr.LineNumber, s);
}
if (lengthValue.value.isInt())
{
length = lengthValue.value.valueInt;
}
else if (lengthValue.value.isObjectRef())
{
long position = gpsr.Position;
PdfObjectReader or = new PdfObjectReader(this);
IPdfObject lengthObject = or.ReadObject(lengthValue.value.valueObjectId);
gpsr.Seek(position);
if (lengthObject.value.isInt())
{
length = lengthObject.value.valueInt;
}
}
if (length != 0)
{
o.stream = gpsr.ReadStream(length);
}
else
{
o.stream = gpsr.ReadStream();
}
s = gpsr.ReadLine();
}
if (s != "endobj")
{
throw new PBException("error endobj not found reading {0} line {1} \"{2}\"", gpsr.ObjectName, gpsr.LineNumber, s);
}
o.source = gpsr.GetObjectSource();
return(o);
}
public static void PrintPdfObject(IPdfObject o)
{
_tr.WriteLine("object id {0} generation {1}", o.id, o.generationNumber);
if (o.value.isObject())
{
foreach (PdfNValue value in o.value.objectValues.Values)
{
_tr.WriteLine(" {0}", value);
}
}
else if (o.value.isArray())
{
int i = 0;
foreach (IPdfValue value in o.value.arrayValues)
{
_tr.WriteLine(" [{0}] = {1}", i++, value);
}
}
else
_tr.WriteLine(" {0}", o.value);
if (o.stream != null)
{
_tr.WriteLine(" stream {0} bytes, deflated stream {1}", o.stream.Length, o.deflatedStream.Length);
}
if (o.stream != null && logStream)
{
////PrintStream(o.stream);
//byte[] stream = o.deflatedStream;
////_tr.WriteLine("deflated stream {0} bytes", stream.Length);
////PrintStream(stream);
//StringBuilder sb = new StringBuilder();
//bool cr = false;
//foreach (byte b in stream)
//{
// if (b == 10 && !cr)
// sb.Append('\r');
// if (b == 13)
// cr = true;
// else
// cr = false;
// sb.Append((char)b);
//}
_tr.WriteLine("stream as string :");
//_tr.WriteLine(sb.ToString());
_tr.WriteLine(o.deflatedStream.zzToString());
_tr.WriteLine();
}
}
public void WriteObjectWithChilds(IPdfObject obj, string trailerName = null)
{
WriteObject(obj, trailerName);
WriteChildsObject(obj.value);
}
public static void Test_GetPdfText_08(String file)
{
PB_Pdf.PdfWriter pw = null;
string file2 = zpath.PathSetFileName(file, Path.GetFileNameWithoutExtension(file) + "_new");
_tr.WriteLine("read pdf file \"{0}\"", file);
_tr.WriteLine("save pdf to file \"{0}\"", file2);
PB_Pdf.PdfReader pr = new PB_Pdf.PdfReader(file);
pr.Trace += new TraceDelegate(_tr.WriteLine);
try
{
pw = new PB_Pdf.PdfWriter(file2, FileMode.Create);
pw.reader = pr;
IPdfObject info = ReadPdfObject(pr, pr.Trailer["Info"].value.valueObjectId, "Info");
pw.WriteObject(info, "Info");
IPdfObject root = ReadPdfObject(pr, pr.Trailer["Root"].value.valueObjectId, "Root");
pw.WriteObject(root, "Root");
IPdfObject pages = ReadPdfObject(pr, root.value["Pages"].value.valueObjectId, "Pages");
pages.value["Kids"].value.arrayValues = new IPdfValue[] { pages.value["Kids"].value.arrayValues[0] };
pages.value["Count"].value.valueInt = 1;
pw.WriteObject(pages);
IPdfObject page1 = ReadPdfObject(pr, pages.value["Kids"].value[0].valueObjectId, "Page 1");
pw.WriteObject(page1);
IPdfObject page1Content = ReadPdfObject(pr, page1.value["Contents"].value.valueObjectId, "Contents page 1");
page1Content.value.objectValues.Remove("Filter");
page1Content.stream = page1Content.deflatedStream;
page1Content.value.objectValues["Length"].value.valueInt = page1Content.stream.Length;
pw.WriteObject(page1Content);
IPdfObject page1Ressource = ReadPdfObject(pr, page1.value["Resources"].value.valueObjectId, "Resources page 1");
Dictionary <string, PdfNValue> objectValues = new Dictionary <string, PdfNValue>();
objectValues["TPL1"] = page1Ressource.value["XObject"].value["TPL1"];
page1Ressource.value["XObject"].value.objectValues = objectValues;
pw.WriteObject(page1Ressource);
IPdfObject page1Ressource_01 = ReadPdfObject(pr, page1Ressource.value["XObject"].value["TPL1"].value.valueObjectId, "Resources page 1 /TPL1");
page1Ressource_01.value.objectValues.Remove("Filter");
page1Ressource_01.stream = page1Ressource_01.deflatedStream;
page1Ressource_01.value.objectValues["Length"].value.valueInt = page1Ressource_01.stream.Length;
pw.WriteObjectWithChilds(page1Ressource_01);
//IPdfObject page1Ressource_ProcSet = ReadPdfObject(pr, page1Ressource_01.value["Resources"].value["ProcSet"].value.valueObjectId, "Resources page 1 /TPL1 ProcSet");
//IPdfObject page1Ressource_Font_01 = ReadPdfObject(pr, page1Ressource_01.value["Resources"].value["Font"].value["F23"].value.valueObjectId, "Resources page 1 /TPL1 Font F23");
//IPdfObject page1Ressource_Font_01_Widths = ReadPdfObject(pr, page1Ressource_Font_01.value["Widths"].value.valueObjectId, "Resources page 1 /TPL1 Font F23 Widths");
//IPdfObject page1Ressource_Font_01_Encoding = ReadPdfObject(pr, page1Ressource_Font_01.value["Encoding"].value.valueObjectId, "Resources page 1 /TPL1 Font F23 Encoding");
//IPdfObject page1Ressource_Font_01_FontDescriptor = ReadPdfObject(pr, page1Ressource_Font_01.value["FontDescriptor"].value.valueObjectId, "Resources page 1 /TPL1 Font F23 FontDescriptor");
//IPdfObject page1Ressource_Font_01_FontDescriptor_FontFile3 = ReadPdfObject(pr, page1Ressource_Font_01_FontDescriptor.value["FontFile3"].value.valueObjectId, "Resources page 1 /TPL1 Font F23 FontDescriptor FontFile3");
//IPdfObject page1Ressource_Font_01_ToUnicode = ReadPdfObject(pr, page1Ressource_Font_01.value["Widths"].value.valueObjectId, "Resources page 1 /TPL1 Font F23 ToUnicode");
//IPdfObject page1Ressource_Img_01 = ReadPdfObject(pr, page1Ressource_01.value["Resources"].value["XObject"].value["img8"].value.valueObjectId, "Resources page 1 /TPL1 XObject img8");
//IPdfObject page1Ressource_Img_01_Length = ReadPdfObject(pr, page1Ressource_Img_01.value["Length"].value.valueObjectId, "Resources page 1 /TPL1 XObject img8 Length");
//IPdfObject page1Ressource_OPMON = ReadPdfObject(pr, page1Ressource_01.value["Resources"].value["ExtGState"].value["OPMON"].value.valueObjectId, "Resources page 1 /TPL1 ExtGState OPMON");
}
finally
{
pr.Close();
if (pw != null)
{
pw.Close();
}
}
}
private void _WriteObject(IPdfObject obj)
{
gw.Write("{0} {1} obj\n", obj.id, obj.generationNumber);
WriteValue(obj.value);
gw.Write("\n");
if (obj.stream != null)
{
gw.Write("stream\n");
gw.Write(obj.stream);
gw.Write("\n");
gw.Write("endstream\n");
}
gw.Write("endobj\n");
}
public PdfCrossReference AddObject(IPdfObject obj)
{
return(Xref.AddEntry(obj));
}
public static PdfXRefSection Parse(PdfStream pdfStream)
{
var dictionary = pdfStream.StreamDictionary;
var type = dictionary["Type"] as PdfName;
if (type.Name != "XRef")
{
throw new ParsingException("A stream of type XRef is expected");
}
// W[1 2 1] (4 columns)
// W[1 3 1] (5 columns, larger indexes)
var w = dictionary["W"] as PdfArray;
int firstId = 0;
int size = 0;
if (dictionary.TryGetValue("Index", out IPdfObject indexObject))
{
var index = (PdfArray)indexObject;
firstId = ((PdfNumeric)index.Items[0]).ToInt32();
size = ((PdfNumeric)index.Items[1]).ToInt32();
}
else if (dictionary.TryGetValue("Size", out IPdfObject sizeObject))
{
size = ((PdfNumeric)sizeObject).ToInt32();
}
int items = w.Items.Count;
// for xref this shall always be 3
if (items != 3)
{
throw new ParsingException("The W[] parameter must contain 3 columns for an XRef");
}
int[] sizes = new int[w.Items.Count];
int bytesPerEntry = 0;
for (int i = 0; i < items; i++)
{
sizes[i] = ((PdfNumeric)w.Items[i]).ToInt32();
bytesPerEntry += sizes[i];
}
var decodedXRef = pdfStream.Decode();
// Use W[...] to build up the xref
int rowCount = decodedXRef.Length / bytesPerEntry;
if (size != rowCount)
{
throw new ParsingException("The number of refs inside the Index value must match the actual refs count present in the stream");
}
var entries = new IPdfObject[rowCount];
for (int row = 0; row < rowCount; row++)
{
var entry = PdfXRefEntry.Parse(firstId + row, decodedXRef, sizes, row, bytesPerEntry);
entries[row] = entry;
}
return(new PdfXRefSection(firstId, size, entries));
}
public void WriteObject(IPdfObject obj, string trailerName = null)
{
if (gXref.ContainsKey(obj.id)) throw new PBException("error object already in file id {0}", obj.id);
gXref.Add(obj.id, new PdfXref { objectId = obj.id, generationNumber = obj.generationNumber, filePosition = gw.Position });
if (trailerName != null)
gTrailer[trailerName] = new PdfNValue { name = trailerName, value = new PdfValueObjectRef { valueObjectId = obj.id, valueObjectGenerationNumber = obj.generationNumber } };
_WriteObject(obj);
//gw.Write("{0} {1} obj\n", obj.id, obj.generationNumber);
//WriteValue(obj.value);
//gw.Write("\n");
//if (obj.stream != null)
//{
// gw.Write("stream\n");
// gw.Write(obj.stream);
// gw.Write("\n");
// gw.Write("endstream\n");
//}
//gw.Write("endobj\n");
}
private void method_1(object sender, int index, IPdfObject item)
{
this.value["Count"] = (IPdfObject) new PdfInt(this.Kids.Count);
}
internal PdfCrossReference ParseIndirectObjectDefinition(PdfCrossReference indirectObj)
{
int offset = -1;
if (indirectObj == null)
{
offset = _lexer.Current.Offset;
}
else
{
if (indirectObj.Loaded)
{
return(indirectObj);
}
// Entry points to an indirect object stream - resolve that.
if (indirectObj != null && indirectObj.EntryType == XrefEntryType.InStream)
{
return(ParseFromStream(indirectObj));
}
_lexer.Seek(indirectObj.Offset);
_lexer.Next();
Expect(TokenType.Integer);
}
int objectNumber = (int)_lexer.Current.Value;
_lexer.Next();
Expect(TokenType.Integer);
int generation = (int)_lexer.Current.Value;
if (indirectObj == null)
{
indirectObj = new PdfCrossReference()
{
EntryType = XrefEntryType.Used, ObjectNumber = objectNumber, Generation = generation, Offset = offset
}
}
;
else if (objectNumber != indirectObj.ObjectNumber || generation != indirectObj.Generation)
{
throw new ParserException($"Expecting object {indirectObj.ObjectNumber} {indirectObj.Generation} at offset {indirectObj.Offset} but got {objectNumber} {generation}");
}
_lexer.Next();
Expect(TokenType.Object);
IPdfObject obj = ParseObject();
_lexer.Next();
if (Accept(TokenType.Stream))
{
if (!(obj is PdfDictionary))
{
throw new ParserException("Stream objects must be preceded by a dictionary");
}
obj = SetObjectType(indirectObj, ParseStream(obj as PdfDictionary));
_lexer.Next();
}
else if (obj is PdfDictionary)
{
obj = SetObjectType(indirectObj, obj as PdfDictionary);
}
Expect(TokenType.EndObject);
indirectObj.Object = obj;
indirectObj.Loaded = true;
return(indirectObj);
}
//public static void Test_GetPdfText_01()
//{
// string file = @"c:\pib\media\print\Le monde\_quotidien\Le monde - 2012-12\Le monde - 2012-12-06 - no 21113.pdf";
// _tr.WriteLine("read pdf file \"{0}\"", file);
// string s = Test_GetPdfText_01(file);
// _tr.WriteLine(s);
// //PdfTextExtractor.GetTextFromPage();
// //ITextExtractionStrategy
// // iTextSharp.text.pdf.parser.LocationTextExtractionStrategy
//}
public static void Test_GetPdfText_07(String file)
{
_tr.WriteLine("read pdf file \"{0}\"", file);
PB_Pdf.PdfReader pr = new PB_Pdf.PdfReader(file);
pr.KeepObjectSource = true;
try
{
//_tr.WriteLine();
//_tr.WriteLine("read xref position");
//pr.ReadXrefPosition();
//_tr.WriteLine("xref table position : {0}", pr.XrefPosition.zToHex());
//_tr.WriteLine();
//_tr.WriteLine("read xref headers");
//pr.ReadXrefHeaders();
//_tr.WriteLine("trailer position : {0}", pr.TrailerPosition.zToHex());
//_tr.WriteLine("objects number : {0}", pr.ObjectsNumber);
//foreach (PdfXrefHeader header in pr.XrefHeaders)
//{
// _tr.WriteLine("xref header : pos {0} id {1} nb {2}", header.filePosition.zToHex(), header.objectId, header.objectNb);
//}
//_tr.WriteLine();
//_tr.WriteLine("read trailer");
//pr.ReadTrailer();
_tr.WriteLine("trailer");
//PrintPdfObject(pr.Trailer);
PrintPdfTrailer(pr.Trailer);
//if (pr.Trailer.source != null)
//{
// _tr.WriteLine("Source trailer");
// _tr.WriteLine(pr.Trailer.source);
//}
//_tr.WriteLine();
//_tr.WriteLine("read xref");
//pr.ReadXref();
//int i = 0;
//foreach (PdfXref xref in pr.Xref.Values)
//{
// _tr.WriteLine("xref : id {0,5} pos {1} generation {2}", xref.objectId, xref.filePosition.zToHex(), xref.generationNumber);
// if (++i == 10) break;
//}
//IPdfObject info = ReadPdfObject(pr, pr.Trailer["Info"].valueObjectId, "Info");
IPdfObject info = ReadPdfObject(pr, pr.Trailer["Info"].value.valueObjectId, "Info");
IPdfObject root = ReadPdfObject(pr, pr.Trailer["Root"].value.valueObjectId, "Root");
IPdfObject pages = ReadPdfObject(pr, root.value["Pages"].value.valueObjectId, "Pages");
IPdfObject page1 = ReadPdfObject(pr, pages.value["Kids"].value[0].valueObjectId, "Page 1");
IPdfObject page1Content = ReadPdfObject(pr, page1.value["Contents"].value.valueObjectId, "Contents page 1");
IPdfObject page1Ressource = ReadPdfObject(pr, page1.value["Resources"].value.valueObjectId, "Resources page 1");
//IPdfObject page1Ressource_01 = ReadPdfObject(pr, 59, "Contents page 1 (2)");
IPdfObject page1Ressource_01 = ReadPdfObject(pr, page1Ressource.value["XObject"].value["TPL1"].value.valueObjectId, "Resources page 1 /TPL1");
IPdfObject page1Ressource_ProcSet = ReadPdfObject(pr, page1Ressource_01.value["Resources"].value["ProcSet"].value.valueObjectId, "Resources page 1 /TPL1 ProcSet");
IPdfObject page1Ressource_Font_01 = ReadPdfObject(pr, page1Ressource_01.value["Resources"].value["Font"].value["F23"].value.valueObjectId, "Resources page 1 /TPL1 Font F23");
IPdfObject page1Ressource_Font_01_Widths = ReadPdfObject(pr, page1Ressource_Font_01.value["Widths"].value.valueObjectId, "Resources page 1 /TPL1 Font F23 Widths");
IPdfObject page1Ressource_Font_01_Encoding = ReadPdfObject(pr, page1Ressource_Font_01.value["Encoding"].value.valueObjectId, "Resources page 1 /TPL1 Font F23 Encoding");
IPdfObject page1Ressource_Font_01_FontDescriptor = ReadPdfObject(pr, page1Ressource_Font_01.value["FontDescriptor"].value.valueObjectId, "Resources page 1 /TPL1 Font F23 FontDescriptor");
IPdfObject page1Ressource_Font_01_FontDescriptor_FontFile3 = ReadPdfObject(pr, page1Ressource_Font_01_FontDescriptor.value["FontFile3"].value.valueObjectId, "Resources page 1 /TPL1 Font F23 FontDescriptor FontFile3");
IPdfObject page1Ressource_Font_01_ToUnicode = ReadPdfObject(pr, page1Ressource_Font_01.value["Widths"].value.valueObjectId, "Resources page 1 /TPL1 Font F23 ToUnicode");
IPdfObject page1Ressource_Img_01 = ReadPdfObject(pr, page1Ressource_01.value["Resources"].value["XObject"].value["img8"].value.valueObjectId, "Resources page 1 /TPL1 XObject img8");
IPdfObject page1Ressource_Img_01_Length = ReadPdfObject(pr, page1Ressource_Img_01.value["Length"].value.valueObjectId, "Resources page 1 /TPL1 XObject img8 Length");
IPdfObject page1Ressource_OPMON = ReadPdfObject(pr, page1Ressource_01.value["Resources"].value["ExtGState"].value["OPMON"].value.valueObjectId, "Resources page 1 /TPL1 ExtGState OPMON");
//_tr.WriteLine("stream as string :");
//_tr.WriteLine(page1Ressource_01.deflatedStream.zzToString());
//_tr.WriteLine();
//IPdfObject obj = ReadPdfObject(pr, 1043, "Object 1043");
}
finally
{
pr.Close();
}
}
get => TryGetValue(name, out IPdfObject value)
? value
public static void Test_PdfImportStream_01(string inputPdfFile, string outputPdfFile, string streamFile)
{
_tr.WriteLine("import stream for object TPL1 of pdf file \"{0}\" from \"{1}\" and save pdf to \"{2}\"", inputPdfFile, streamFile, outputPdfFile);
PB_Pdf.PdfReader pr = null;
Reader r = null;
PB_Pdf.PdfWriter pw = null;
try
{
pr = new PB_Pdf.PdfReader(inputPdfFile);
//IPdfObject obj = ReadPdfObject(pr, idObject, "Object");
r = new Reader(streamFile);
//obj.stream = r.ReadBytes((int)r.Length);
//PdfNValue length = obj.value["Length"];
//if (length == null || !length.value.isInt()) throw new PBException("error wrong /Length of object {0}", idObject);
//length.value.valueInt = obj.stream.Length;
pw = new PB_Pdf.PdfWriter(outputPdfFile, FileMode.Create);
pw.reader = pr;
IPdfObject info = ReadPdfObject(pr, pr.Trailer["Info"].value.valueObjectId, "Info");
pw.WriteObject(info, "Info");
IPdfObject root = ReadPdfObject(pr, pr.Trailer["Root"].value.valueObjectId, "Root");
pw.WriteObject(root, "Root");
IPdfObject pages = ReadPdfObject(pr, root.value["Pages"].value.valueObjectId, "Pages");
pw.WriteObject(pages);
IPdfObject page1 = ReadPdfObject(pr, pages.value["Kids"].value[0].valueObjectId, "Page 1");
pw.WriteObject(page1);
IPdfObject page1Content = ReadPdfObject(pr, page1.value["Contents"].value.valueObjectId, "Contents page 1");
pw.WriteObject(page1Content);
IPdfObject page1Ressource = ReadPdfObject(pr, page1.value["Resources"].value.valueObjectId, "Resources page 1");
pw.WriteObject(page1Ressource);
IPdfObject page1Ressource_01 = ReadPdfObject(pr, page1Ressource.value["XObject"].value["TPL1"].value.valueObjectId, "Resources page 1 /TPL1");
// import stream from file
page1Ressource_01.stream = r.ReadBytes((int)r.Length);
PdfNValue length = page1Ressource_01.value["Length"];
if (length == null || !length.value.isInt())
{
throw new PBException("error wrong /Length of object {0}", page1Ressource_01.id);
}
length.value.valueInt = page1Ressource_01.stream.Length;
pw.WriteObjectWithChilds(page1Ressource_01);
}
finally
{
if (pr != null)
{
pr.Close();
}
if (r != null)
{
r.Close();
}
if (pw != null)
{
pw.Close();
}
}
}
public void WriteObjectWithChilds(IPdfObject obj, string trailerName = null)
{
WriteObject(obj, trailerName);
WriteChildsObject(obj.value);
}
public void UpdateObject(IPdfObject obj, string trailerName = null)
{
if (!gUpdatePdf) throw new PBException("error object update is not activate \"{0}\"", gw.File);
if (!gXref.ContainsKey(obj.id)) throw new PBException("error update object {0}, object dont exist", obj.id);
gw.Seek(gUpdatePosition);
gXref[obj.id] = new PdfXref { objectId = obj.id, generationNumber = obj.generationNumber, filePosition = gUpdatePosition };
if (trailerName != null)
gTrailer[trailerName] = new PdfNValue { name = trailerName, value = new PdfValueObjectRef { valueObjectId = obj.id, valueObjectGenerationNumber = obj.generationNumber } };
_WriteObject(obj);
gUpdatePosition = (int)gw.Position;
}
