Refactor QPDFWriter::preserveObjectStreams

libqpdf/QPDF.cc

@@ -1365,6 +1365,7 @@ QPDF::Xref_table::insert(int obj, int f0, qpdf_offset_t f1, int f2)
 
     case 2:
         entry = {0, Compressed(toI(f1), f2)};
+        object_streams_ = true;
         break;
 
     default:

libqpdf/QPDFWriter.cc

@@ -1936,50 +1936,29 @@ void
 QPDFWriter::preserveObjectStreams()
 {
     auto const& xref = QPDF::Writer::getXRefTable(m->pdf);
-    // Our object_to_object_stream map has to map ObjGen -> ObjGen since we may be generating object
+    m->obj.streams_empty = !xref.object_streams();
-    // streams out of old objects that have generation numbers greater than zero. However in an
+    if (m->obj.streams_empty) {
-    // existing PDF, all object stream objects and all objects in them must have generation 0
+        return;
-    // because the PDF spec does not provide any way to do otherwise. This code filters out objects
-    // that are not allowed to be in object streams. In addition to removing objects that were
-    // erroneously included in object streams in the source PDF, it also prevents unreferenced
-    // objects from being included.
-    auto end = xref.cend();
-    m->obj.streams_empty = true;
-    if (m->preserve_unreferenced_objects) {
-        for (auto iter = xref.cbegin(); iter != end; ++iter) {
-            if (iter->second.getType() == 2) {
-                // Pdf contains object streams.
-                QTC::TC("qpdf", "QPDFWriter preserve object streams preserve unreferenced");
-                m->obj.streams_empty = false;
-                m->obj[iter->first].object_stream = iter->second.getObjStreamNumber();
     }
+    // This code filters out objects that are not allowed to be in object streams. In addition to
+    // removing objects that were erroneously included in object streams in the source PDF, it also
+    // prevents unreferenced objects from being included.
+    if (m->preserve_unreferenced_objects) {
+        QTC::TC("qpdf", "QPDFWriter preserve object streams preserve unreferenced");
+        for (auto [id, stream]: xref.compressed_objects()) {
+            m->obj[id].object_stream = stream;
         }
     } else {
-        // Start by scanning for first compressed object in case we don't have any object streams to
-        // process.
-        for (auto iter = xref.cbegin(); iter != end; ++iter) {
-            if (iter->second.getType() == 2) {
-                // Pdf contains object streams.
         QTC::TC("qpdf", "QPDFWriter preserve object streams");
-                m->obj.streams_empty = false;
         auto eligible = QPDF::Writer::getCompressibleObjSet(m->pdf);
-                // The object pointed to by iter may be a previous generation, in which case it is
+        for (auto [id, stream]: xref.compressed_objects()) {
-                // removed by getCompressibleObjSet. We need to restart the loop (while the object
+            if (eligible[id]) {
-                // table may contain multiple generations of an object).
+                m->obj[id].object_stream = stream;
-                for (iter = xref.cbegin(); iter != end; ++iter) {
-                    if (iter->second.getType() == 2) {
-                        auto id = static_cast<size_t>(iter->first.getObj());
-                        if (id < eligible.size() && eligible[id]) {
-                            m->obj[iter->first].object_stream = iter->second.getObjStreamNumber();
             } else {
                 QTC::TC("qpdf", "QPDFWriter exclude from object stream");
             }
         }
     }
-                return;
-            }
-        }
-    }
 }
 
 void

libqpdf/qpdf/ObjTable.hh

@@ -45,6 +45,12 @@ class ObjTable: public std::vector<T>
         return element(static_cast<size_t>(idx));
     }
 
+    inline T const&
+    operator[](unsigned int idx) const
+    {
+        return element(idx);
+    }
+
     inline T const&
     operator[](QPDFObjGen og) const
     {

libqpdf/qpdf/QPDF_private.hh

@@ -112,6 +112,33 @@ class QPDF::Xref_table
         return result;
     }
 
+    bool
+    object_streams() const noexcept
+    {
+        return object_streams_;
+    }
+
+    // Return a vector of object id and stream number for each compressed object.
+    std::vector<std::pair<unsigned int, int>>
+    compressed_objects() const
+    {
+        if (!initialized()) {
+            throw std::logic_error("Xref_table::compressed_objects called before parsing.");
+        }
+
+        std::vector<std::pair<unsigned int, int>> result;
+        result.reserve(table.size());
+
+        unsigned int i{0};
+        for (auto const& item: table) {
+            if (item.type() == 2) {
+                result.emplace_back(i, item.stream_number());
+            }
+            ++i;
+        }
+        return result;
+    }
+
     // Temporary access to underlying table size
     size_t
     size() const noexcept
@@ -282,6 +309,7 @@ class QPDF::Xref_table
     bool initialized_{false};
     bool ignore_streams_{false};
     bool reconstructed_{false};
+    bool object_streams_{false};
     // Before the xref table is initialized, max_id_ is an upper bound on the possible object ids
     // that could be present in the PDF file. Once the trailer has been read, max_id_ is set to the
     // value of /Size. If the file is damaged, max_id_ becomes the maximum object id in the xref
@@ -293,72 +321,6 @@ class QPDF::Xref_table
     qpdf_offset_t first_item_offset_{0}; // actual value from file
 };
 
-// Writer class is restricted to QPDFWriter so that only it can call certain methods.
-class QPDF::Writer
-{
-    friend class QPDFWriter;
-
-  private:
-    static void
-    optimize(
-        QPDF& qpdf,
-        QPDFWriter::ObjTable const& obj,
-        std::function<int(QPDFObjectHandle&)> skip_stream_parameters)
-    {
-        return qpdf.optimize(obj, skip_stream_parameters);
-    }
-
-    static void
-    getLinearizedParts(
-        QPDF& qpdf,
-        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(obj, part4, part6, part7, part8, part9);
-    }
-
-    static void
-    generateHintStream(
-        QPDF& qpdf,
-        QPDFWriter::NewObjTable const& new_obj,
-        QPDFWriter::ObjTable const& obj,
-        std::shared_ptr<Buffer>& hint_stream,
-        int& S,
-        int& O,
-        bool compressed)
-    {
-        return qpdf.generateHintStream(new_obj, obj, hint_stream, S, O, compressed);
-    }
-
-    static std::vector<QPDFObjGen>
-    getCompressibleObjGens(QPDF& qpdf)
-    {
-        return qpdf.getCompressibleObjVector();
-    }
-
-    static std::vector<bool>
-    getCompressibleObjSet(QPDF& qpdf)
-    {
-        return qpdf.getCompressibleObjSet();
-    }
-
-    static std::map<QPDFObjGen, QPDFXRefEntry>
-    getXRefTable(QPDF& qpdf)
-    {
-        return qpdf.getXRefTableInternal();
-    }
-
-    static size_t
-    tableSize(QPDF& qpdf)
-    {
-        return qpdf.tableSize();
-    }
-};
-
 // The Resolver class is restricted to QPDFObject so that only it can resolve indirect
 // references.
 class QPDF::Resolver
@@ -841,4 +803,70 @@ class QPDF::ResolveRecorder
     std::set<QPDFObjGen>::const_iterator iter;
 };
 
+// Writer class is restricted to QPDFWriter so that only it can call certain methods.
+class QPDF::Writer
+{
+    friend class QPDFWriter;
+
+  private:
+    static void
+    optimize(
+        QPDF& qpdf,
+        QPDFWriter::ObjTable const& obj,
+        std::function<int(QPDFObjectHandle&)> skip_stream_parameters)
+    {
+        return qpdf.optimize(obj, skip_stream_parameters);
+    }
+
+    static void
+    getLinearizedParts(
+        QPDF& qpdf,
+        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(obj, part4, part6, part7, part8, part9);
+    }
+
+    static void
+    generateHintStream(
+        QPDF& qpdf,
+        QPDFWriter::NewObjTable const& new_obj,
+        QPDFWriter::ObjTable const& obj,
+        std::shared_ptr<Buffer>& hint_stream,
+        int& S,
+        int& O,
+        bool compressed)
+    {
+        return qpdf.generateHintStream(new_obj, obj, hint_stream, S, O, compressed);
+    }
+
+    static std::vector<QPDFObjGen>
+    getCompressibleObjGens(QPDF& qpdf)
+    {
+        return qpdf.getCompressibleObjVector();
+    }
+
+    static std::vector<bool>
+    getCompressibleObjSet(QPDF& qpdf)
+    {
+        return qpdf.getCompressibleObjSet();
+    }
+
+    static Xref_table const&
+    getXRefTable(QPDF& qpdf)
+    {
+        return qpdf.m->xref_table;
+    }
+
+    static size_t
+    tableSize(QPDF& qpdf)
+    {
+        return qpdf.tableSize();
+    }
+};
+
 #endif // QPDF_PRIVATE_HH