From 8a6b35827fb57b5ed72a695de713c668a268060d Mon Sep 17 00:00:00 2001
From: m-holger <m.holger@qpdf.org>
Date: Sun, 20 Jul 2025 15:27:55 +0100
Subject: [PATCH] Refactor `QPDF_Stream` and related classes to replace `Buffer` usage with `std::string`, improving performance, simplifying stream data handling, and reducing dependency on `BufferInputSource`.

---
 libqpdf/QPDF_Stream.cc                   | 19 +++++++++++--------
 libqpdf/QPDF_objects.cc                  | 29 ++++++++++++-----------------
 libqpdf/qpdf/QPDFObjectHandle_private.hh |  4 ++--
 3 files changed, 25 insertions(+), 27 deletions(-)

diff --git a/libqpdf/QPDF_Stream.cc b/libqpdf/QPDF_Stream.cc
index b71a439..e6dd873 100644
--- a/libqpdf/QPDF_Stream.cc
+++ b/libqpdf/QPDF_Stream.cc
@@ -3,6 +3,7 @@
 #include <qpdf/ContentNormalizer.hh>
 #include <qpdf/JSON_writer.hh>
 #include <qpdf/Pipeline.hh>
+#include <qpdf/Pipeline_private.hh>
 #include <qpdf/Pl_Base64.hh>
 #include <qpdf/Pl_Buffer.hh>
 #include <qpdf/Pl_Count.hh>
@@ -319,10 +320,11 @@ qpdf::Stream::setDictDescription()
     }
 }
 
-std::shared_ptr<Buffer>
+std::string
 Stream::getStreamData(qpdf_stream_decode_level_e decode_level)
 {
-    Pl_Buffer buf("stream data buffer");
+    std::string result;
+    pl::String buf(result);
     bool filtered;
     pipeStreamData(&buf, &filtered, 0, decode_level, false, false);
     if (!filtered) {
@@ -334,13 +336,14 @@ Stream::getStreamData(qpdf_stream_decode_level_e decode_level)
             "getStreamData called on unfilterable stream");
     }
     QTC::TC("qpdf", "QPDF_Stream getStreamData");
-    return buf.getBufferSharedPointer();
+    return result;
 }
 
-std::shared_ptr<Buffer>
+std::string
 Stream::getRawStreamData()
 {
-    Pl_Buffer buf("stream data buffer");
+    std::string result;
+    pl::String buf(result);
     if (!pipeStreamData(&buf, nullptr, 0, qpdf_dl_none, false, false)) {
         throw QPDFExc(
             qpdf_e_unsupported,
@@ -350,7 +353,7 @@ Stream::getRawStreamData()
             "error getting raw stream data");
     }
     QTC::TC("qpdf", "QPDF_Stream getRawStreamData");
-    return buf.getBufferSharedPointer();
+    return result;
 }
 
 bool
@@ -683,13 +686,13 @@ QPDFObjectHandle::isRootMetadata() const
 std::shared_ptr<Buffer>
 QPDFObjectHandle::getStreamData(qpdf_stream_decode_level_e level)
 {
-    return as_stream(error).getStreamData(level);
+    return std::make_shared<Buffer>(as_stream(error).getStreamData(level));
 }
 
 std::shared_ptr<Buffer>
 QPDFObjectHandle::getRawStreamData()
 {
-    return as_stream(error).getRawStreamData();
+    return std::make_shared<Buffer>(as_stream(error).getRawStreamData());
 }
 
 bool
diff --git a/libqpdf/QPDF_objects.cc b/libqpdf/QPDF_objects.cc
index 70eea69..b2a352d 100644
--- a/libqpdf/QPDF_objects.cc
+++ b/libqpdf/QPDF_objects.cc
@@ -2,18 +2,7 @@
 
 #include <qpdf/QPDF_private.hh>
 
-#include <array>
-#include <atomic>
-#include <cstring>
-#include <limits>
-#include <map>
-#include <regex>
-#include <sstream>
-#include <vector>
-
-#include <qpdf/BufferInputSource.hh>
 #include <qpdf/InputSource_private.hh>
-#include <qpdf/OffsetInputSource.hh>
 #include <qpdf/Pipeline.hh>
 #include <qpdf/QPDFExc.hh>
 #include <qpdf/QPDFLogger.hh>
@@ -24,6 +13,13 @@
 #include <qpdf/QUtil.hh>
 #include <qpdf/Util.hh>
 
+#include <array>
+#include <atomic>
+#include <cstring>
+#include <limits>
+#include <map>
+#include <vector>
+
 using namespace qpdf;
 using namespace std::literals;
 
@@ -1675,13 +1671,13 @@ QPDF::resolveObjectsInStream(int obj_stream_number)
     // id, offset, size
     std::vector<std::tuple<int, qpdf_offset_t, size_t>> offsets;
 
-    auto bp = obj_stream.getStreamData(qpdf_dl_specialized);
+    auto stream_data = obj_stream.getStreamData(qpdf_dl_specialized);
 
-    BufferInputSource input("", bp.get());
+    is::OffsetBuffer input("", stream_data);
 
-    const auto b_size = bp->getSize();
+    const auto b_size = stream_data.size();
     const auto end_offset = static_cast<qpdf_offset_t>(b_size);
-    auto b_start = bp->getBuffer();
+    auto b_start = stream_data.data();
 
     if (first >= end_offset) {
         throw damagedPDF(
@@ -1762,8 +1758,7 @@ QPDF::resolveObjectsInStream(int obj_stream_number)
         auto entry = m->xref_table.find(og);
         if (entry != m->xref_table.end() && entry->second.getType() == 2 &&
             entry->second.getObjStreamNumber() == obj_stream_number) {
-            Buffer obj_buffer{b_start + obj_offset, obj_size};
-            is::OffsetBuffer in("", &obj_buffer, obj_offset);
+            is::OffsetBuffer in("", {b_start + obj_offset, obj_size}, obj_offset);
             auto oh = readObjectInStream(in, obj_stream_number, obj_id);
             updateCache(og, oh.getObj(), end_before_space, end_after_space);
         } else {
diff --git a/libqpdf/qpdf/QPDFObjectHandle_private.hh b/libqpdf/qpdf/QPDFObjectHandle_private.hh
index ddba62f..c73ea1b 100644
--- a/libqpdf/qpdf/QPDFObjectHandle_private.hh
+++ b/libqpdf/qpdf/QPDFObjectHandle_private.hh
@@ -254,8 +254,8 @@ namespace qpdf
             qpdf_stream_decode_level_e decode_level,
             bool suppress_warnings,
             bool will_retry);
-        std::shared_ptr<Buffer> getStreamData(qpdf_stream_decode_level_e level);
-        std::shared_ptr<Buffer> getRawStreamData();
+        std::string getStreamData(qpdf_stream_decode_level_e level);
+        std::string getRawStreamData();
         void replaceStreamData(
             std::shared_ptr<Buffer> data,
             QPDFObjectHandle const& filter,
--
libgit2 0.21.4