From 0df0d00c588f3ca5f71524f96817a84c4b98bc2c Mon Sep 17 00:00:00 2001
From: m-holger <m-holger@kubitscheck.org>
Date: Sun, 25 Feb 2024 22:40:09 +0000
Subject: [PATCH] Add method QPDF::Writer::getCompressibleObjSet

Create set without creation of an intermediate vector.
---
 include/qpdf/QPDF.hh  | 13 +++++++++++--
 libqpdf/QPDF.cc       | 28 ++++++++++++++++++++++++++--
 libqpdf/QPDFWriter.cc | 26 ++++++++++----------------
 3 files changed, 47 insertions(+), 20 deletions(-)

diff --git a/include/qpdf/QPDF.hh b/include/qpdf/QPDF.hh
index 10fa404a..b366b17b 100644
--- a/include/qpdf/QPDF.hh
+++ b/include/qpdf/QPDF.hh
@@ -756,7 +756,13 @@ class QPDF
         static std::vector<QPDFObjGen>
         getCompressibleObjGens(QPDF& qpdf)
         {
-            return qpdf.getCompressibleObjGens();
+            return qpdf.getCompressibleObjVector();
+        }
+
+        static std::vector<bool>
+        getCompressibleObjSet(QPDF& qpdf)
+        {
+            return qpdf.getCompressibleObjSet();
         }
 
         static std::map<QPDFObjGen, QPDFXRefEntry> const&
@@ -1110,7 +1116,10 @@ class QPDF
         bool compressed);
 
     // Get a list of objects that would be permitted in an object stream.
-    std::vector<QPDFObjGen> getCompressibleObjGens();
+    template <typename T>
+    std::vector<T> getCompressibleObjGens();
+    std::vector<QPDFObjGen> getCompressibleObjVector();
+    std::vector<bool> getCompressibleObjSet();
 
     // methods to support page handling
 
diff --git a/libqpdf/QPDF.cc b/libqpdf/QPDF.cc
index 5417c9ba..3f776852 100644
--- a/libqpdf/QPDF.cc
+++ b/libqpdf/QPDF.cc
@@ -2397,6 +2397,19 @@ QPDF::tableSize()
 }
 
 std::vector<QPDFObjGen>
+QPDF::getCompressibleObjVector()
+{
+    return getCompressibleObjGens<QPDFObjGen>();
+}
+
+std::vector<bool>
+QPDF::getCompressibleObjSet()
+{
+    return getCompressibleObjGens<bool>();
+}
+
+template <typename T>
+std::vector<T>
 QPDF::getCompressibleObjGens()
 {
     // Return a list of objects that are allowed to be in object streams.  Walk through the objects
@@ -2414,7 +2427,14 @@ QPDF::getCompressibleObjGens()
     std::vector<QPDFObjectHandle> queue;
     queue.reserve(512);
     queue.push_back(m->trailer);
-    std::vector<QPDFObjGen> result;
+    std::vector<T> result;
+    if constexpr (std::is_same_v<T, QPDFObjGen>) {
+        result.reserve(m->obj_cache.size());
+    } else if constexpr (std::is_same_v<T, bool>) {
+        result.resize(max_obj + 1U, false);
+    } else {
+        throw std::logic_error("Unsupported type in QPDF::getCompressibleObjGens");
+    }
     while (!queue.empty()) {
         auto obj = queue.back();
         queue.pop_back();
@@ -2446,7 +2466,11 @@ QPDF::getCompressibleObjGens()
             } else if (!(obj.isStream() ||
                          (obj.isDictionaryOfType("/Sig") && obj.hasKey("/ByteRange") &&
                           obj.hasKey("/Contents")))) {
-                result.push_back(og);
+                if constexpr (std::is_same_v<T, QPDFObjGen>) {
+                    result.push_back(og);
+                } else if constexpr (std::is_same_v<T, bool>) {
+                    result[id + 1U] = true;
+                }
             }
         }
         if (obj.isStream()) {
diff --git a/libqpdf/QPDFWriter.cc b/libqpdf/QPDFWriter.cc
index aaf954e2..2ad4d786 100644
--- a/libqpdf/QPDFWriter.cc
+++ b/libqpdf/QPDFWriter.cc
@@ -1964,13 +1964,11 @@ QPDFWriter::preserveObjectStreams()
                     }
                 }
             } else {
-                std::set<QPDFObjGen> eligible;
-                std::vector<QPDFObjGen> eligible_v = QPDF::Writer::getCompressibleObjGens(m->pdf);
-                eligible = std::set<QPDFObjGen>(eligible_v.begin(), eligible_v.end());
+                auto eligible = QPDF::Writer::getCompressibleObjSet(m->pdf);
                 for (; iter != end; ++iter) {
                     if (iter->second.getType() == 2) {
-                        QPDFObjGen og(iter->first.getObj(), 0);
-                        if (eligible.count(og)) {
+                        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");
@@ -2009,22 +2007,18 @@ QPDFWriter::generateObjectStreams()
         ++n_per;
     }
     unsigned int n = 0;
-    int cur_ostream = 0;
-    for (auto const& iter: eligible) {
-        if ((n % n_per) == 0) {
-            if (n > 0) {
-                QTC::TC("qpdf", "QPDFWriter generate >1 ostream");
-            }
+    int cur_ostream = m->pdf.newIndirectNull().getObjectID();
+    for (auto const& item: eligible) {
+        if (n == n_per) {
+            QTC::TC("qpdf", "QPDFWriter generate >1 ostream");
             n = 0;
-        }
-        if (n == 0) {
             // Construct a new null object as the "original" object stream.  The rest of the code
             // knows that this means we're creating the object stream from scratch.
-            cur_ostream = m->pdf.makeIndirectObject(QPDFObjectHandle::newNull()).getObjectID();
+            cur_ostream = m->pdf.newIndirectNull().getObjectID();
         }
-        auto& obj = m->obj[iter];
+        auto& obj = m->obj[item];
         obj.object_stream = cur_ostream;
-        obj.gen = iter.getGen();
+        obj.gen = item.getGen();
         ++n;
     }
 }