Fix and test support for files >= 4 GB

9 files changed, 52 insertions, 45 deletions

diff --git a/libqpdf/BitStream.cc b/libqpdf/BitStream.cc
index 703ce8f2..eb511f72 100644
--- a/libqpdf/BitStream.cc
+++ b/libqpdf/BitStream.cc
@@ -19,7 +19,7 @@ BitStream::reset()
     bits_available = 8 * nbytes;
 }
 
-unsigned long
+unsigned long long
 BitStream::getBits(int nbits)
 {
     return read_bits(this->p, this->bit_offset,
diff --git a/libqpdf/BitWriter.cc b/libqpdf/BitWriter.cc
index 441501cb..4fb375cb 100644
--- a/libqpdf/BitWriter.cc
+++ b/libqpdf/BitWriter.cc
@@ -12,7 +12,7 @@ BitWriter::BitWriter(Pipeline* pl) :
 }
 
 void
-BitWriter::writeBits(unsigned long val, unsigned int bits)
+BitWriter::writeBits(unsigned long long val, unsigned int bits)
 {
     write_bits(this->ch, this->bit_offset, val, bits, this->pl);
 }
diff --git a/libqpdf/QPDF.cc b/libqpdf/QPDF.cc
index 6b275d28..743ba93b 100644
--- a/libqpdf/QPDF.cc
+++ b/libqpdf/QPDF.cc
@@ -571,7 +571,7 @@ QPDF::reconstruct_xref(QPDFExc& e)
 		in_obj = true;
 		int obj = atoi(m.getMatch(1).c_str());
 		int gen = atoi(m.getMatch(2).c_str());
-		int offset = this->file->getLastOffset();
+		qpdf_offset_t offset = this->file->getLastOffset();
 		insertXrefEntry(obj, 1, offset, gen, true);
 	    }
 	    else if ((! this->trailer.isInitialized()) &&
@@ -634,6 +634,11 @@ QPDF::read_xref(qpdf_offset_t xref_offset)
 	}
     }
 
+    if (! this->trailer.isInitialized())
+    {
+        throw QPDFExc(qpdf_e_damaged_pdf, this->file->getName(), "", 0,
+                      "unable to find trailer while reading xref");
+    }
     int size = this->trailer.getKey("/Size").getIntValue();
     int max_obj = 0;
     if (! xref_table.empty())
@@ -704,7 +709,8 @@ QPDF::read_xrefTable(qpdf_offset_t xref_offset)
 		    QUtil::int_to_string(i) + ")");
 	    }
 
-	    int f1 = atoi(m2.getMatch(1).c_str());
+            // For xref_table, these will always be small enough to be ints 
+	    qpdf_offset_t f1 = QUtil::string_to_ll(m2.getMatch(1).c_str());
 	    int f2 = atoi(m2.getMatch(2).c_str());
 	    char type = m2.getMatch(3)[0];
 	    if (type == 'f')
@@ -855,7 +861,7 @@ QPDF::read_xrefStream(qpdf_offset_t xref_offset)
     return xref_offset;
 }
 
-int
+qpdf_offset_t
 QPDF::processXRefStream(qpdf_offset_t xref_offset, QPDFObjectHandle& xref_obj)
 {
     QPDFObjectHandle dict = xref_obj.getDict();
@@ -957,7 +963,7 @@ QPDF::processXRefStream(qpdf_offset_t xref_offset, QPDFObjectHandle& xref_obj)
     {
 	// Read this entry
 	unsigned char const* entry = data + (entry_size * i);
-	int fields[3];
+	qpdf_offset_t fields[3];
 	unsigned char const* p = entry;
 	for (int j = 0; j < 3; ++j)
 	{
@@ -1002,7 +1008,7 @@ QPDF::processXRefStream(qpdf_offset_t xref_offset, QPDFObjectHandle& xref_obj)
 	    // This is needed by checkLinearization()
 	    this->first_xref_item_offset = xref_offset;
 	}
-	insertXrefEntry(obj, fields[0], fields[1], fields[2]);
+	insertXrefEntry(obj, (int)fields[0], fields[1], (int)fields[2]);
     }
 
     if (! this->trailer.isInitialized())
@@ -1031,7 +1037,7 @@ QPDF::processXRefStream(qpdf_offset_t xref_offset, QPDFObjectHandle& xref_obj)
 }
 
 void
-QPDF::insertXrefEntry(int obj, int f0, int f1, int f2, bool overwrite)
+QPDF::insertXrefEntry(int obj, int f0, qpdf_offset_t f1, int f2, bool overwrite)
 {
     // Populate the xref table in such a way that the first reference
     // to an object that we see, which is the one in the latest xref
@@ -1558,7 +1564,7 @@ QPDF::recoverStreamLength(PointerHolder<InputSource> input,
 	    QPDFXRefEntry const& entry = (*iter).second;
 	    if (entry.getType() == 1)
 	    {
-		int obj_offset = entry.getOffset();
+		qpdf_offset_t obj_offset = entry.getOffset();
 		if ((obj_offset > stream_offset) &&
 		    ((this_obj_offset == 0) ||
 		     (this_obj_offset > obj_offset)))
diff --git a/libqpdf/QPDFWriter.cc b/libqpdf/QPDFWriter.cc
index 308f4c53..26713225 100644
--- a/libqpdf/QPDFWriter.cc
+++ b/libqpdf/QPDFWriter.cc
@@ -540,7 +540,7 @@ QPDFWriter::setDataKey(int objid)
 }
 
 int
-QPDFWriter::bytesNeeded(unsigned long n)
+QPDFWriter::bytesNeeded(unsigned long long n)
 {
     int bytes = 0;
     while (n)
@@ -552,10 +552,10 @@ QPDFWriter::bytesNeeded(unsigned long n)
 }
 
 void
-QPDFWriter::writeBinary(unsigned long val, unsigned int bytes)
+QPDFWriter::writeBinary(unsigned long long val, unsigned int bytes)
 {
-    assert(bytes <= sizeof(unsigned long));
-    unsigned char data[sizeof(unsigned long)];
+    assert(bytes <= sizeof(unsigned long long));
+    unsigned char data[sizeof(unsigned long long)];
     for (unsigned int i = 0; i < bytes; ++i)
     {
 	data[bytes - i - 1] = (unsigned char)(val & 0xff);
@@ -849,7 +849,8 @@ QPDFWriter::unparseChild(QPDFObjectHandle child, int level, int flags)
 }
 
 void
-QPDFWriter::writeTrailer(trailer_e which, int size, bool xref_stream, int prev)
+QPDFWriter::writeTrailer(trailer_e which, int size, bool xref_stream,
+                         qpdf_offset_t prev)
 {
     QPDFObjectHandle trailer = pdf.getTrailer();
     if (! xref_stream)
@@ -1812,15 +1813,15 @@ QPDFWriter::writeHintStream(int hint_id)
     closeObject(hint_id);
 }
 
-int
+qpdf_offset_t
 QPDFWriter::writeXRefTable(trailer_e which, int first, int last, int size)
 {
     return writeXRefTable(which, first, last, size, 0, false, 0, 0, 0);
 }
 
-int
+qpdf_offset_t
 QPDFWriter::writeXRefTable(trailer_e which, int first, int last, int size,
-			   int prev, bool suppress_offsets,
+			   qpdf_offset_t prev, bool suppress_offsets,
 			   int hint_id, qpdf_offset_t hint_offset,
                            qpdf_offset_t hint_length)
 {
@@ -1838,7 +1839,7 @@ QPDFWriter::writeXRefTable(trailer_e which, int first, int last, int size,
 	}
 	else
 	{
-	    int offset = 0;
+	    qpdf_offset_t offset = 0;
 	    if (! suppress_offsets)
 	    {
 		offset = this->xref[i].getOffset();
@@ -1858,24 +1859,24 @@ QPDFWriter::writeXRefTable(trailer_e which, int first, int last, int size,
     return space_before_zero;
 }
 
-int
-QPDFWriter::writeXRefStream(int objid, int max_id, int max_offset,
+qpdf_offset_t
+QPDFWriter::writeXRefStream(int objid, int max_id, qpdf_offset_t max_offset,
 			    trailer_e which, int first, int last, int size)
 {
     return writeXRefStream(objid, max_id, max_offset,
 			   which, first, last, size, 0, 0, 0, 0, false);
 }
 
-int
-QPDFWriter::writeXRefStream(int xref_id, int max_id, int max_offset,
+qpdf_offset_t
+QPDFWriter::writeXRefStream(int xref_id, int max_id, qpdf_offset_t max_offset,
 			    trailer_e which, int first, int last, int size,
-			    int prev, int hint_id,
+			    qpdf_offset_t prev, int hint_id,
 			    qpdf_offset_t hint_offset,
                             qpdf_offset_t hint_length,
 			    bool skip_compression)
 {
     qpdf_offset_t xref_offset = this->pipeline->getCount();
-    int space_before_zero = xref_offset - 1;
+    qpdf_offset_t space_before_zero = xref_offset - 1;
 
     // field 1 contains offsets and object stream identifiers
     int f1_size = std::max(bytesNeeded(max_offset),
@@ -1921,7 +1922,7 @@ QPDFWriter::writeXRefStream(int xref_id, int max_id, int max_offset,
 
 	  case 1:
 	    {
-		int offset = e.getOffset();
+		qpdf_offset_t offset = e.getOffset();
 		if ((hint_id != 0) &&
 		    (i != hint_id) &&
 		    (offset >= hint_offset))
@@ -2309,7 +2310,7 @@ QPDFWriter::writeLinearized()
 
 	    // Save hint offset since it will be set to zero by
 	    // calling openObject.
-	    int hint_offset = this->xref[hint_id].getOffset();
+	    qpdf_offset_t hint_offset = this->xref[hint_id].getOffset();
 
 	    // Write hint stream to a buffer
 	    pushPipeline(new Pl_Buffer("hint buffer"));
diff --git a/libqpdf/QPDF_linearization.cc b/libqpdf/QPDF_linearization.cc
index 48bb4d2b..fdd0d702 100644
--- a/libqpdf/QPDF_linearization.cc
+++ b/libqpdf/QPDF_linearization.cc
@@ -18,10 +18,10 @@
 #include <math.h>
 #include <string.h>
 
-template <class T>
+template <class T, class int_type>
 static void
 load_vector_int(BitStream& bit_stream, int nitems, std::vector<T>& vec,
-		int bits_wanted, int T::*field)
+		int bits_wanted, int_type T::*field)
 {
     // nitems times, read bits_wanted from the given bit stream,
     // storing results in the ith vector entry.
@@ -144,7 +144,7 @@ QPDF::isLinearized()
     QPDFObjectHandle L = candidate.getKey("/L");
     if (L.isInteger())
     {
-	int Li = L.getIntValue();
+	qpdf_offset_t Li = L.getIntValue();
 	this->file->seek(0, SEEK_END);
 	if (Li != this->file->tell())
 	{
@@ -649,11 +649,11 @@ QPDF::maxEnd(ObjUser const& ou)
     return end;
 }
 
-int
+qpdf_offset_t
 QPDF::getLinearizationOffset(ObjGen const& og)
 {
     QPDFXRefEntry entry = this->xref_table[og];
-    int result = 0;
+    qpdf_offset_t result = 0;
     switch (entry.getType())
     {
       case 1:
@@ -1787,7 +1787,7 @@ static inline int nbits(int val)
 int
 QPDF::outputLengthNextN(
     int in_object, int n,
-    std::map<int, size_t> const& lengths,
+    std::map<int, qpdf_offset_t> const& lengths,
     std::map<int, int> const& obj_renumber)
 {
     // Figure out the length of a series of n consecutive objects in
@@ -1808,7 +1808,7 @@ QPDF::outputLengthNextN(
 void
 QPDF::calculateHPageOffset(
     std::map<int, QPDFXRefEntry> const& xref,
-    std::map<int, size_t> const& lengths,
+    std::map<int, qpdf_offset_t> const& lengths,
     std::map<int, int> const& obj_renumber)
 {
     // Page Offset Hint Table
@@ -1900,7 +1900,7 @@ QPDF::calculateHPageOffset(
 void
 QPDF::calculateHSharedObject(
     std::map<int, QPDFXRefEntry> const& xref,
-    std::map<int, size_t> const& lengths,
+    std::map<int, qpdf_offset_t> const& lengths,
     std::map<int, int> const& obj_renumber)
 {
     CHSharedObject& cso = this->c_shared_object_data;
@@ -1946,7 +1946,7 @@ QPDF::calculateHSharedObject(
 void
 QPDF::calculateHOutline(
     std::map<int, QPDFXRefEntry> const& xref,
-    std::map<int, size_t> const& lengths,
+    std::map<int, qpdf_offset_t> const& lengths,
     std::map<int, int> const& obj_renumber)
 {
     HGeneric& cho = this->c_outline_data;
@@ -1967,10 +1967,10 @@ QPDF::calculateHOutline(
 	cho.first_object, ho.nobjects, lengths, obj_renumber);
 }
 
-template <class T>
+template <class T, class int_type>
 static void
 write_vector_int(BitWriter& w, int nitems, std::vector<T>& vec,
-		 int bits, int T::*field)
+		 int bits, int_type T::*field)
 {
     // nitems times, write bits bits from the given field of the ith
     // vector to the given bit writer.
@@ -2095,7 +2095,7 @@ QPDF::writeHGeneric(BitWriter& w, HGeneric& t)
 
 void
 QPDF::generateHintStream(std::map<int, QPDFXRefEntry> const& xref,
-			 std::map<int, size_t> const& lengths,
+			 std::map<int, qpdf_offset_t> const& lengths,
 			 std::map<int, int> const& obj_renumber,
 			 PointerHolder<Buffer>& hint_buffer,
 			 int& S, int& O)
diff --git a/libqpdf/bits.icc b/libqpdf/bits.icc
index 31765986..bcd7dd85 100644
--- a/libqpdf/bits.icc
+++ b/libqpdf/bits.icc
@@ -15,7 +15,7 @@
 // this code includes with the symbol defined.
 
 #ifdef BITS_READ
-static unsigned long
+static unsigned long long
 read_bits(unsigned char const*& p, unsigned int& bit_offset,
 	  unsigned int& bits_available, unsigned int bits_wanted)
 {
@@ -95,7 +95,7 @@ read_bits(unsigned char const*& p, unsigned int& bit_offset,
 #ifdef BITS_WRITE
 static void
 write_bits(unsigned char& ch, unsigned int& bit_offset,
-	   unsigned long val, unsigned int bits, Pipeline* pipeline)
+	   unsigned long long val, unsigned int bits, Pipeline* pipeline)
 {
     if (bits > 32)
     {
diff --git a/libqpdf/qpdf-c.cc b/libqpdf/qpdf-c.cc
index 65a5de72..0312ae50 100644
--- a/libqpdf/qpdf-c.cc
+++ b/libqpdf/qpdf-c.cc
@@ -31,7 +31,7 @@ struct _qpdf_data
     // Parameters for functions we call
     char const* filename;	// or description
     char const* buffer;
-    unsigned long size;
+    unsigned long long size;
     char const* password;
     bool write_memory;
     Buffer* output_buffer;
@@ -218,7 +218,7 @@ char const* qpdf_get_error_filename(qpdf_data qpdf, qpdf_error e)
     return e->exc->getFilename().c_str();
 }
 
-unsigned long qpdf_get_error_file_position(qpdf_data qpdf, qpdf_error e)
+unsigned long long qpdf_get_error_file_position(qpdf_data qpdf, qpdf_error e)
 {
     if (e == 0)
     {
@@ -268,7 +268,7 @@ QPDF_ERROR_CODE qpdf_read(qpdf_data qpdf, char const* filename,
 QPDF_ERROR_CODE qpdf_read_memory(qpdf_data qpdf,
 				 char const* description,
 				 char const* buffer,
-				 unsigned long size,
+				 unsigned long long size,
 				 char const* password)
 {
     QPDF_ERROR_CODE status = QPDF_SUCCESS;
diff --git a/libqpdf/qpdf/BitStream.hh b/libqpdf/qpdf/BitStream.hh
index 92bbd735..e45a90ee 100644
--- a/libqpdf/qpdf/BitStream.hh
+++ b/libqpdf/qpdf/BitStream.hh
@@ -13,7 +13,7 @@ class BitStream
     QPDF_DLL
     void reset();
     QPDF_DLL
-    unsigned long getBits(int nbits);
+    unsigned long long getBits(int nbits);
     QPDF_DLL
     void skipToNextByte();
 
diff --git a/libqpdf/qpdf/BitWriter.hh b/libqpdf/qpdf/BitWriter.hh
index 5eae398f..7e3b07a9 100644
--- a/libqpdf/qpdf/BitWriter.hh
+++ b/libqpdf/qpdf/BitWriter.hh
@@ -15,7 +15,7 @@ class BitWriter
     QPDF_DLL
     BitWriter(Pipeline* pl);
     QPDF_DLL
-    void writeBits(unsigned long val, unsigned int bits);
+    void writeBits(unsigned long long val, unsigned int bits);
     // Force any partial byte to be written to the pipeline.
     QPDF_DLL
     void flush();