fenix/PDF4QT
1
0
Fork 0
mirror of https://github.com/JakubMelka/PDF4QT.git synced 2025-06-05 21:59:17 +02:00
Code Issues Packages Projects Releases Wiki Activity

Issue #80: Optimizer is slow on large PDF files

Browse Source
This commit is contained in:
Jakub Melka
2023-08-21 20:53:03 +02:00
parent 16c35d5f01
commit 7a77bd796a
1 changed files with 43 additions and 33 deletions
Download Patch File Download Diff File Expand all files Collapse all files

76
Pdf4QtLib/sources/pdfoptimizer.cpp
View File

@ -24,6 +24,7 @@
#include "pdfdocumentbuilder.h" #include "pdfdocumentbuilder.h"
#include "pdfstreamfilters.h" #include "pdfstreamfilters.h"
#include "pdfdbgheap.h" #include "pdfdbgheap.h"
#include "pdfdocumentwriter.h"
namespace pdf namespace pdf
{ {
@ -246,48 +247,57 @@ bool PDFOptimizer::performMergeIdenticalObjects()
{ {
std::atomic<PDFInteger> counter = 0; std::atomic<PDFInteger> counter = 0;
std::map<PDFObjectReference, PDFObjectReference> replacementMap; std::map<PDFObjectReference, PDFObjectReference> replacementMap;
std::map<QByteArray, PDFObjectReference> serializedObjectToReference;
PDFObjectStorage::PDFObjects objects = m_storage.getObjects(); PDFObjectStorage::PDFObjects objects = m_storage.getObjects();
std::vector<QByteArray> serializedObjects(objects.size());
// Find same objects
QMutex mutex;
PDFIntegerRange<size_t> range(0, objects.size()); PDFIntegerRange<size_t> range(0, objects.size());
auto processEntry = [this, &counter, &objects, &mutex, &replacementMap](size_t index) auto serializeEntry = [&, this](size_t index)
{ {
const PDFObjectStorage::Entry& entry = objects[index]; const PDFObjectStorage::Entry& entry = objects[index];
// We do not merge special objects, such as pages if (!entry.object.isNull())
if (const PDFDictionary* dictionary = m_storage.getDictionaryFromObject(entry.object))
{ {
PDFObject nameObject = m_storage.getObject(dictionary->get("Type")); serializedObjects[index] = PDFDocumentWriter::getSerializedObject(entry.object);
if (nameObject.isName())
{
if (nameObject.getString() == "Page")
{
return;
}
}
}
for (size_t i = 0; i < index; ++i)
{
if (objects[i].object.isNull())
{
// Jakub Melka: we do not merge null objects, they are just removed
continue;
}
if (objects[i].object == entry.object)
{
QMutexLocker lock(&mutex);
PDFObjectReference oldReference = PDFObjectReference(PDFInteger(index), objects[index].generation);
PDFObjectReference newReference = PDFObjectReference(PDFInteger(i), objects[i].generation);
replacementMap[oldReference] = newReference;
++counter;
break;
}
} }
}; };
PDFExecutionPolicy::execute(PDFExecutionPolicy::Scope::Unknown, range.begin(), range.end(), processEntry); PDFExecutionPolicy::execute(PDFExecutionPolicy::Scope::Unknown, range.begin(), range.end(), serializeEntry);
// Find same object
for (PDFInteger index : range)
{
const PDFObjectStorage::Entry& entry = objects[index];
if (!entry.object.isNull())
{
// We do not merge special objects, such as pages
if (const PDFDictionary* dictionary = m_storage.getDictionaryFromObject(entry.object))
{
PDFObject nameObject = m_storage.getObject(dictionary->get("Type"));
if (nameObject.isName())
{
if (nameObject.getString() == "Page")
{
continue;
}
}
}
PDFObjectReference currentReference(PDFInteger(index), objects[index].generation);
const QByteArray& serializedObject = serializedObjects[index];
if (!serializedObjectToReference.count(serializedObject))
{
serializedObjectToReference[serializedObject] = currentReference;
}
else
{
PDFObjectReference oldReference = currentReference;
PDFObjectReference newReference = serializedObjectToReference.at(serializedObject);
replacementMap[oldReference] = newReference;
++counter;
}
}
}
// Replace objects // Replace objects
if (!replacementMap.empty()) if (!replacementMap.empty())