In QPDFWriter::writeLinearized use object table obj in call to getLinearizedParts

m-holger
1 parent 97d78c75
Showing 5 changed files with 26 additions and 36 deletions
include/qpdf/QPDF.hh
include/qpdf/QPDFWriter.hh
libqpdf/QPDFWriter.cc
libqpdf/QPDF_linearization.cc
libqpdf/qpdf/QPDFWriter_private.hh
@@ -739,14 +739,14 @@ class QPDF
         static void
         getLinearizedParts(
             QPDF& qpdf,
-            std::map<int, int> const& object_stream_data,
+            QPDFWriter::ObjTable const& obj,
             std::vector<QPDFObjectHandle>& part4,
             std::vector<QPDFObjectHandle>& part6,
             std::vector<QPDFObjectHandle>& part7,
             std::vector<QPDFObjectHandle>& part8,
             std::vector<QPDFObjectHandle>& part9)
         {
-            qpdf.getLinearizedParts(object_stream_data, part4, part6, part7, part8, part9);
+            qpdf.getLinearizedParts(obj, part4, part6, part7, part8, part9);
         }
         static void
@@ -1117,7 +1117,7 @@ class QPDF
     // Get lists of all objects in order according to the part of a linearized file that they belong
     // to.
     void getLinearizedParts(
-        std::map<int, int> const& object_stream_data,
+        QPDFWriter::ObjTable const& obj,
         std::vector<QPDFObjectHandle>& part4,
         std::vector<QPDFObjectHandle>& part6,
         std::vector<QPDFObjectHandle>& part7,
@@ -1382,6 +1382,7 @@ class QPDF
     qpdf_offset_t getLinearizationOffset(QPDFObjGen const&);
     QPDFObjectHandle
     getUncompressedObject(QPDFObjectHandle&, std::map<int, int> const& object_stream_data);
+    QPDFObjectHandle getUncompressedObject(QPDFObjectHandle&, QPDFWriter::ObjTable const& obj);
     int lengthNextN(int first_object, int n);
     void
     checkHPageOffset(std::vector<QPDFObjectHandle> const& pages, std::map<int, int>& idx_to_obj);
@@ -1392,11 +1393,13 @@ class QPDF
     void dumpHSharedObject();
     void dumpHGeneric(HGeneric&);
     qpdf_offset_t adjusted_offset(qpdf_offset_t offset);
-    void calculateLinearizationData(std::map<int, int> const& object_stream_data);
+    template <typename T>
+    void calculateLinearizationData(T const& object_stream_data);
+    template <typename T>
     void pushOutlinesToPart(
         std::vector<QPDFObjectHandle>& part,
         std::set<QPDFObjGen>& lc_outlines,
-        std::map<int, int> const& object_stream_data);
+        T const& object_stream_data);
     int outputLengthNextN(
         int in_object,
         int n,
@@ -611,8 +611,6 @@ class QPDFWriter
     void pushMD5Pipeline(PipelinePopper&);
     void computeDeterministicIDData();
-    void discardGeneration(std::map<int, int>& out);
-
     class Members;
     // Keep all member variables inside the Members object, which we dynamically allocate. This
@@ -2540,32 +2540,10 @@ QPDFWriter::calculateXrefStreamPadding(qpdf_offset_t xref_bytes)
 }
 void
-QPDFWriter::discardGeneration(std::map<int, int>& out)
-{
-    // There are deep assumptions in the linearization code in QPDF that there is only one object
-    // with each object number; i.e., you can't have two objects with the same object number and
-    // different generations.  This is a pretty safe assumption because Adobe Reader and Acrobat
-    // can't actually handle this case.  There is not much if any code in QPDF outside linearization
-    // that assumes this, but the linearization code as currently implemented would do weird things
-    // if we found such a case.  In order to avoid breaking ABI changes in QPDF, we will first
-    // assert that this condition holds.  Then we can create new maps for QPDF that throw away
-    // generation numbers.
-
-    out.clear();
-    m->obj.forEach([&out](auto id, auto const& item) -> void {
-        if (item.object_stream > 0) {
-            out[id] = item.object_stream;
-        }
-    });
-}
-
-void
 QPDFWriter::writeLinearized()
 {
     // Optimize file and enqueue objects in order
-    discardGeneration(m->object_to_object_stream_no_gen);
-
     auto skip_stream_parameters = [this](QPDFObjectHandle& stream) {
         bool compress_stream;
         bool is_metadata;
@@ -2583,8 +2561,7 @@ QPDFWriter::writeLinearized()
     std::vector<QPDFObjectHandle> part7;
     std::vector<QPDFObjectHandle> part8;
     std::vector<QPDFObjectHandle> part9;
-    QPDF::Writer::getLinearizedParts(
-        m->pdf, m->object_to_object_stream_no_gen, part4, part6, part7, part8, part9);
+    QPDF::Writer::getLinearizedParts(m->pdf, m->obj, part4, part6, part7, part8, part9);
     // Object number sequence:
     //
@@ -586,6 +586,17 @@ QPDF::getUncompressedObject(QPDFObjectHandle&amp; obj, std::map&lt;int, int&gt; const&amp; obj
     }
 }
+QPDFObjectHandle
+QPDF::getUncompressedObject(QPDFObjectHandle& oh, QPDFWriter::ObjTable const& obj)
+{
+    if (obj.contains(oh)) {
+        if (auto id = obj[oh].object_stream; id > 0) {
+            return oh.isNull() ? oh : getObject(id, 0);
+        }
+    }
+    return oh;
+}
+
 int
 QPDF::lengthNextN(int first_object, int n)
 {
@@ -960,8 +971,9 @@ QPDF::dumpHGeneric(HGeneric&amp; t)
                        << "group_length: " << t.group_length << "\n";
 }
+template <typename T>
 void
-QPDF::calculateLinearizationData(std::map<int, int> const& object_stream_data)
+QPDF::calculateLinearizationData(T const& object_stream_data)
 {
     // This function calculates the ordering of objects, divides them into the appropriate parts,
     // and computes some values for the linearization parameter dictionary and hint tables.  The
@@ -1403,11 +1415,12 @@ QPDF::calculateLinearizationData(std::map&lt;int, int&gt; const&amp; object_stream_data)
     }
 }
+template <typename T>
 void
 QPDF::pushOutlinesToPart(
     std::vector<QPDFObjectHandle>& part,
     std::set<QPDFObjGen>& lc_outlines,
-    std::map<int, int> const& object_stream_data)
+    T const& object_stream_data)
 {
     QPDFObjectHandle root = getRoot();
     QPDFObjectHandle outlines = root.getKey("/Outlines");
@@ -1434,14 +1447,14 @@ QPDF::pushOutlinesToPart(
 void
 QPDF::getLinearizedParts(
-    std::map<int, int> const& object_stream_data,
+    QPDFWriter::ObjTable const& obj,
     std::vector<QPDFObjectHandle>& part4,
     std::vector<QPDFObjectHandle>& part6,
     std::vector<QPDFObjectHandle>& part7,
     std::vector<QPDFObjectHandle>& part8,
     std::vector<QPDFObjectHandle>& part9)
 {
-    calculateLinearizationData(object_stream_data);
+    calculateLinearizationData(obj);
     part4 = m->part4;
     part6 = m->part6;
     part7 = m->part7;
@@ -121,7 +121,6 @@ class QPDFWriter::Members
     // For linearization only
     std::string lin_pass1_filename;
-    std::map<int, int> object_to_object_stream_no_gen;
     // For progress reporting
     std::shared_ptr<QPDFWriter::ProgressReporter> progress_reporter;