mirror of
https://github.com/JakubMelka/PDF4QT.git
synced 2025-01-30 17:14:50 +01:00
612 lines
20 KiB
C++
612 lines
20 KiB
C++
// Copyright (C) 2021 Jakub Melka
|
|
//
|
|
// This file is part of PDF4QT.
|
|
//
|
|
// PDF4QT is free software: you can redistribute it and/or modify
|
|
// it under the terms of the GNU Lesser General Public License as published by
|
|
// the Free Software Foundation, either version 3 of the License, or
|
|
// with the written consent of the copyright owner, any later version.
|
|
//
|
|
// PDF4QT is distributed in the hope that it will be useful,
|
|
// but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
// GNU Lesser General Public License for more details.
|
|
//
|
|
// You should have received a copy of the GNU Lesser General Public License
|
|
// along with PDF4QT. If not, see <https://www.gnu.org/licenses/>.
|
|
|
|
#include "pdfdiff.h"
|
|
#include "pdfrenderer.h"
|
|
#include "pdfdocumenttextflow.h"
|
|
#include "pdfexecutionpolicy.h"
|
|
#include "pdffont.h"
|
|
#include "pdfcms.h"
|
|
#include "pdfcompiler.h"
|
|
#include "pdfconstants.h"
|
|
#include "pdfalgorithmlcs.h"
|
|
|
|
#include <QtConcurrent/QtConcurrent>
|
|
|
|
namespace pdf
|
|
{
|
|
|
|
class PDFDiffHelper
|
|
{
|
|
public:
|
|
using GraphicPieceInfo = PDFPrecompiledPage::GraphicPieceInfo;
|
|
using GraphicPieceInfos = PDFPrecompiledPage::GraphicPieceInfos;
|
|
using PageSequence = PDFAlgorithmLongestCommonSubsequenceBase::Sequence;
|
|
|
|
|
|
struct Differences
|
|
{
|
|
GraphicPieceInfos left;
|
|
GraphicPieceInfos right;
|
|
|
|
bool isEmpty() const { return left.empty() && right.empty(); }
|
|
};
|
|
|
|
static Differences calculateDifferences(const GraphicPieceInfos& left, const GraphicPieceInfos& right, PDFReal epsilon);
|
|
static std::vector<size_t> getLeftUnmatched(const PageSequence& sequence);
|
|
static std::vector<size_t> getRightUnmatched(const PageSequence& sequence);
|
|
static void matchPage(PageSequence& sequence, size_t leftPage, size_t rightPage);
|
|
};
|
|
|
|
PDFDiff::PDFDiff(QObject* parent) :
|
|
BaseClass(parent),
|
|
m_progress(nullptr),
|
|
m_leftDocument(nullptr),
|
|
m_rightDocument(nullptr),
|
|
m_options(Asynchronous | PC_Text | PC_VectorGraphics | PC_Images),
|
|
m_epsilon(0.001),
|
|
m_cancelled(false)
|
|
{
|
|
|
|
}
|
|
|
|
PDFDiff::~PDFDiff()
|
|
{
|
|
stop();
|
|
}
|
|
|
|
void PDFDiff::setLeftDocument(const PDFDocument* leftDocument)
|
|
{
|
|
if (m_leftDocument != leftDocument)
|
|
{
|
|
stop();
|
|
m_leftDocument = leftDocument;
|
|
}
|
|
}
|
|
|
|
void PDFDiff::setRightDocument(const PDFDocument* rightDocument)
|
|
{
|
|
if (m_rightDocument != rightDocument)
|
|
{
|
|
stop();
|
|
m_rightDocument = rightDocument;
|
|
}
|
|
}
|
|
|
|
void PDFDiff::setPagesForLeftDocument(PDFClosedIntervalSet pagesForLeftDocument)
|
|
{
|
|
stop();
|
|
m_pagesForLeftDocument = std::move(pagesForLeftDocument);
|
|
}
|
|
|
|
void PDFDiff::setPagesForRightDocument(PDFClosedIntervalSet pagesForRightDocument)
|
|
{
|
|
stop();
|
|
m_pagesForRightDocument = std::move(pagesForRightDocument);
|
|
}
|
|
|
|
void PDFDiff::start()
|
|
{
|
|
// Jakub Melka: First, we must ensure, that comparation
|
|
// process is finished, otherwise we must wait for end.
|
|
// Then, create a new future watcher.
|
|
stop();
|
|
|
|
m_cancelled = false;
|
|
|
|
if (m_options.testFlag(Asynchronous))
|
|
{
|
|
m_futureWatcher = std::nullopt;
|
|
m_futureWatcher.emplace();
|
|
|
|
m_future = QtConcurrent::run(std::bind(&PDFDiff::perform, this));
|
|
connect(&*m_futureWatcher, &QFutureWatcher<PDFDiffResult>::finished, this, &PDFDiff::onComparationPerformed);
|
|
m_futureWatcher->setFuture(m_future);
|
|
}
|
|
else
|
|
{
|
|
// Just do comparation immediately
|
|
m_result = perform();
|
|
emit comparationFinished();
|
|
}
|
|
}
|
|
|
|
void PDFDiff::stop()
|
|
{
|
|
if (m_futureWatcher && !m_futureWatcher->isFinished())
|
|
{
|
|
// Do stop only if process doesn't finished already.
|
|
// If we are finished, we do not want to set cancelled state.
|
|
m_cancelled = true;
|
|
m_futureWatcher->waitForFinished();
|
|
}
|
|
}
|
|
|
|
PDFDiffResult PDFDiff::perform()
|
|
{
|
|
PDFDiffResult result;
|
|
|
|
if (!m_leftDocument || !m_rightDocument)
|
|
{
|
|
result.setResult(tr("No document to be compared."));
|
|
return result;
|
|
}
|
|
|
|
if (m_pagesForLeftDocument.isEmpty() || m_pagesForRightDocument.isEmpty())
|
|
{
|
|
result.setResult(tr("No page to be compared."));
|
|
return result;
|
|
}
|
|
|
|
auto leftPages = m_pagesForLeftDocument.unfold();
|
|
auto rightPages = m_pagesForRightDocument.unfold();
|
|
|
|
const size_t leftDocumentPageCount = m_leftDocument->getCatalog()->getPageCount();
|
|
const size_t rightDocumentPageCount = m_rightDocument->getCatalog()->getPageCount();
|
|
|
|
if (leftPages.front() < 0 ||
|
|
leftPages.back() >= PDFInteger(leftDocumentPageCount) ||
|
|
rightPages.front() < 0 ||
|
|
rightPages.back() >= PDFInteger(rightDocumentPageCount))
|
|
{
|
|
result.setResult(tr("Invalid page range."));
|
|
return result;
|
|
}
|
|
|
|
if (m_progress)
|
|
{
|
|
ProgressStartupInfo info;
|
|
info.showDialog = false;
|
|
info.text = tr("Comparing documents.");
|
|
m_progress->start(StepLast, std::move(info));
|
|
}
|
|
|
|
performSteps(leftPages, rightPages);
|
|
|
|
if (m_progress)
|
|
{
|
|
m_progress->finish();
|
|
}
|
|
|
|
return result;
|
|
}
|
|
|
|
void PDFDiff::stepProgress()
|
|
{
|
|
if (m_progress)
|
|
{
|
|
m_progress->step();
|
|
}
|
|
}
|
|
|
|
struct PDFDiffPageContext
|
|
{
|
|
PDFInteger pageIndex = 0;
|
|
std::array<uint8_t, 64> pageHash = { };
|
|
PDFPrecompiledPage::GraphicPieceInfos graphicPieces;
|
|
};
|
|
|
|
void PDFDiff::performPageMatching(const std::vector<PDFDiffPageContext>& leftPreparedPages,
|
|
const std::vector<PDFDiffPageContext>& rightPreparedPages,
|
|
PDFAlgorithmLongestCommonSubsequenceBase::Sequence& pageSequence)
|
|
{
|
|
// Match pages. We will use following algorithm: exact solution can fail, because
|
|
// we are using hashes and due to numerical instability, hashes can be different
|
|
// even for exactly the same page. But if hashes are the same, the page must be the same.
|
|
// So, we use longest common subsequence algorithm to detect same page ranges,
|
|
// and then we match the rest. We assume the number of failing pages is relatively small.
|
|
|
|
std::map<size_t, size_t> pageMatches;
|
|
auto comparePages = [&](const PDFDiffPageContext& left, const PDFDiffPageContext& right)
|
|
{
|
|
if (left.pageHash == right.pageHash)
|
|
{
|
|
return true;
|
|
}
|
|
|
|
auto it = pageMatches.find(left.pageIndex);
|
|
if (it != pageMatches.cend())
|
|
{
|
|
return it->second == right.pageIndex;
|
|
}
|
|
|
|
return false;
|
|
};
|
|
PDFAlgorithmLongestCommonSubsequence algorithm(leftPreparedPages.cbegin(), leftPreparedPages.cend(),
|
|
rightPreparedPages.cbegin(), rightPreparedPages.cend(),
|
|
comparePages);
|
|
algorithm.perform();
|
|
pageSequence = algorithm.getSequence();
|
|
|
|
std::vector<size_t> leftUnmatched = PDFDiffHelper::getLeftUnmatched(pageSequence);
|
|
std::vector<size_t> rightUnmatched = PDFDiffHelper::getRightUnmatched(pageSequence);
|
|
|
|
// We are matching left pages to the right ones
|
|
std::map<size_t, std::vector<size_t>> matchedPages;
|
|
|
|
for (const size_t index : leftUnmatched)
|
|
{
|
|
matchedPages[index] = std::vector<size_t>();
|
|
}
|
|
|
|
auto matchLeftPage = [&, this](size_t leftIndex)
|
|
{
|
|
const PDFDiffPageContext& leftPageContext = leftPreparedPages[leftIndex];
|
|
|
|
auto page = m_leftDocument->getCatalog()->getPage(leftPageContext.pageIndex);
|
|
PDFReal epsilon = calculateEpsilonForPage(page);
|
|
|
|
for (const size_t rightIndex : rightUnmatched)
|
|
{
|
|
const PDFDiffPageContext& rightPageContext = rightPreparedPages[rightIndex];
|
|
if (leftPageContext.graphicPieces.size() != rightPageContext.graphicPieces.size())
|
|
{
|
|
// Match cannot exist, graphic pieces have different size
|
|
continue;
|
|
}
|
|
|
|
PDFDiffHelper::Differences differences = PDFDiffHelper::calculateDifferences(leftPageContext.graphicPieces, rightPageContext.graphicPieces, epsilon);
|
|
|
|
if (differences.isEmpty())
|
|
{
|
|
// Jakub Melka: we have a match
|
|
matchedPages[leftIndex].push_back(rightIndex);
|
|
}
|
|
}
|
|
};
|
|
|
|
PDFExecutionPolicy::execute(PDFExecutionPolicy::Scope::Page, leftUnmatched.begin(), leftUnmatched.end(), matchLeftPage);
|
|
|
|
std::set<size_t> matchedRightPages;
|
|
for (const auto& matchedPage : matchedPages)
|
|
{
|
|
for (size_t rightContextIndex : matchedPage.second)
|
|
{
|
|
if (!matchedRightPages.count(rightContextIndex))
|
|
{
|
|
matchedRightPages.insert(rightContextIndex);
|
|
const PDFDiffPageContext& leftPageContext = leftPreparedPages[matchedPage.first];
|
|
const PDFDiffPageContext& rightPageContext = rightPreparedPages[rightContextIndex];
|
|
|
|
pageMatches[leftPageContext.pageIndex] = rightPageContext.pageIndex;
|
|
}
|
|
}
|
|
}
|
|
|
|
if (!pageMatches.empty())
|
|
{
|
|
algorithm.perform();
|
|
pageSequence = algorithm.getSequence();
|
|
}
|
|
}
|
|
|
|
void PDFDiff::performSteps(const std::vector<PDFInteger>& leftPages, const std::vector<PDFInteger>& rightPages)
|
|
{
|
|
std::vector<PDFDiffPageContext> leftPreparedPages;
|
|
std::vector<PDFDiffPageContext> rightPreparedPages;
|
|
|
|
PDFDiffHelper::PageSequence pageSequence;
|
|
|
|
auto createDiffPageContext = [](auto pageIndex)
|
|
{
|
|
PDFDiffPageContext context;
|
|
context.pageIndex = pageIndex;
|
|
return context;
|
|
};
|
|
std::transform(leftPages.cbegin(), leftPages.cend(), std::back_inserter(leftPreparedPages), createDiffPageContext);
|
|
std::transform(rightPages.cbegin(), rightPages.cend(), std::back_inserter(rightPreparedPages), createDiffPageContext);
|
|
|
|
// StepExtractContentLeftDocument
|
|
if (!m_cancelled)
|
|
{
|
|
PDFFontCache fontCache(DEFAULT_FONT_CACHE_LIMIT, DEFAULT_REALIZED_FONT_CACHE_LIMIT);
|
|
PDFOptionalContentActivity optionalContentActivity(m_leftDocument, pdf::OCUsage::View, nullptr);
|
|
fontCache.setDocument(pdf::PDFModifiedDocument(const_cast<pdf::PDFDocument*>(m_leftDocument), &optionalContentActivity));
|
|
|
|
PDFCMSManager cmsManager(nullptr);
|
|
cmsManager.setDocument(m_leftDocument);
|
|
PDFCMSPointer cms = cmsManager.getCurrentCMS();
|
|
|
|
auto fillPageContext = [&, this](PDFDiffPageContext& context)
|
|
{
|
|
PDFPrecompiledPage compiledPage;
|
|
constexpr PDFRenderer::Features features = PDFRenderer::IgnoreOptionalContent;
|
|
PDFRenderer renderer(m_leftDocument, &fontCache, cms.data(), &optionalContentActivity, features, pdf::PDFMeshQualitySettings());
|
|
renderer.compile(&compiledPage, context.pageIndex);
|
|
|
|
auto page = m_leftDocument->getCatalog()->getPage(context.pageIndex);
|
|
PDFReal epsilon = calculateEpsilonForPage(page);
|
|
context.graphicPieces = compiledPage.calculateGraphicPieceInfos(page->getMediaBox(), epsilon);
|
|
|
|
finalizeGraphicsPieces(context);
|
|
};
|
|
PDFExecutionPolicy::execute(PDFExecutionPolicy::Scope::Page, leftPreparedPages.begin(), leftPreparedPages.end(), fillPageContext);
|
|
stepProgress();
|
|
}
|
|
|
|
// StepExtractContentRightDocument
|
|
if (!m_cancelled)
|
|
{
|
|
PDFFontCache fontCache(DEFAULT_FONT_CACHE_LIMIT, DEFAULT_REALIZED_FONT_CACHE_LIMIT);
|
|
PDFOptionalContentActivity optionalContentActivity(m_rightDocument, pdf::OCUsage::View, nullptr);
|
|
fontCache.setDocument(pdf::PDFModifiedDocument(const_cast<pdf::PDFDocument*>(m_rightDocument), &optionalContentActivity));
|
|
|
|
PDFCMSManager cmsManager(nullptr);
|
|
cmsManager.setDocument(m_rightDocument);
|
|
PDFCMSPointer cms = cmsManager.getCurrentCMS();
|
|
|
|
auto fillPageContext = [&, this](PDFDiffPageContext& context)
|
|
{
|
|
PDFPrecompiledPage compiledPage;
|
|
constexpr PDFRenderer::Features features = PDFRenderer::IgnoreOptionalContent;
|
|
PDFRenderer renderer(m_rightDocument, &fontCache, cms.data(), &optionalContentActivity, features, pdf::PDFMeshQualitySettings());
|
|
renderer.compile(&compiledPage, context.pageIndex);
|
|
|
|
const PDFPage* page = m_leftDocument->getCatalog()->getPage(context.pageIndex);
|
|
PDFReal epsilon = calculateEpsilonForPage(page);
|
|
context.graphicPieces = compiledPage.calculateGraphicPieceInfos(page->getMediaBox(), epsilon);
|
|
|
|
finalizeGraphicsPieces(context);
|
|
};
|
|
|
|
PDFExecutionPolicy::execute(PDFExecutionPolicy::Scope::Page, rightPreparedPages.begin(), rightPreparedPages.end(), fillPageContext);
|
|
stepProgress();
|
|
}
|
|
|
|
// StepMatchPages
|
|
if (!m_cancelled)
|
|
{
|
|
performPageMatching(leftPreparedPages, rightPreparedPages, pageSequence);
|
|
stepProgress();
|
|
}
|
|
|
|
// StepExtractTextLeftDocument
|
|
if (!m_cancelled)
|
|
{
|
|
pdf::PDFDocumentTextFlowFactory factoryLeftDocumentTextFlow;
|
|
factoryLeftDocumentTextFlow.setCalculateBoundingBoxes(true);
|
|
PDFDocumentTextFlow leftTextFlow = factoryLeftDocumentTextFlow.create(m_leftDocument, leftPages, PDFDocumentTextFlowFactory::Algorithm::Auto);
|
|
stepProgress();
|
|
}
|
|
|
|
// StepExtractTextRightDocument
|
|
if (!m_cancelled)
|
|
{
|
|
pdf::PDFDocumentTextFlowFactory factoryRightDocumentTextFlow;
|
|
factoryRightDocumentTextFlow.setCalculateBoundingBoxes(true);
|
|
PDFDocumentTextFlow rightTextFlow = factoryRightDocumentTextFlow.create(m_rightDocument, rightPages, PDFDocumentTextFlowFactory::Algorithm::Auto);
|
|
stepProgress();
|
|
}
|
|
|
|
// StepCompare
|
|
if (!m_cancelled)
|
|
{
|
|
stepProgress();
|
|
}
|
|
}
|
|
|
|
void PDFDiff::finalizeGraphicsPieces(PDFDiffPageContext& context)
|
|
{
|
|
std::sort(context.graphicPieces.begin(), context.graphicPieces.end());
|
|
|
|
// Compute page hash using active settings
|
|
QCryptographicHash hasher(QCryptographicHash::Sha512);
|
|
hasher.reset();
|
|
|
|
for (const PDFPrecompiledPage::GraphicPieceInfo& info : context.graphicPieces)
|
|
{
|
|
if (info.isText() && !m_options.testFlag(PC_Text))
|
|
{
|
|
continue;
|
|
}
|
|
if (info.isVectorGraphics() && !m_options.testFlag(PC_VectorGraphics))
|
|
{
|
|
continue;
|
|
}
|
|
if (info.isImage() && !m_options.testFlag(PC_Images))
|
|
{
|
|
continue;
|
|
}
|
|
if (info.isShading() && !m_options.testFlag(PC_Mesh))
|
|
{
|
|
continue;
|
|
}
|
|
|
|
hasher.addData(reinterpret_cast<const char*>(info.hash.data()), int(info.hash.size()));
|
|
}
|
|
|
|
QByteArray hash = hasher.result();
|
|
Q_ASSERT(QCryptographicHash::hashLength(QCryptographicHash::Sha512) == 64);
|
|
|
|
size_t size = qMin<size_t>(hash.length(), context.pageHash.size());
|
|
std::copy(hash.data(), hash.data() + size, context.pageHash.data());
|
|
}
|
|
|
|
void PDFDiff::onComparationPerformed()
|
|
{
|
|
m_cancelled = false;
|
|
m_result = m_future.result();
|
|
emit comparationFinished();
|
|
}
|
|
|
|
PDFReal PDFDiff::calculateEpsilonForPage(const PDFPage* page) const
|
|
{
|
|
Q_ASSERT(page);
|
|
|
|
QRectF mediaBox = page->getMediaBox();
|
|
|
|
PDFReal width = mediaBox.width();
|
|
PDFReal height = mediaBox.height();
|
|
PDFReal factor = qMax(width, height);
|
|
|
|
return factor * m_epsilon;
|
|
}
|
|
|
|
PDFDiffResult::PDFDiffResult() :
|
|
m_result(true)
|
|
{
|
|
|
|
}
|
|
|
|
PDFDiffHelper::Differences PDFDiffHelper::calculateDifferences(const GraphicPieceInfos& left,
|
|
const GraphicPieceInfos& right,
|
|
PDFReal epsilon)
|
|
{
|
|
Differences differences;
|
|
|
|
Q_ASSERT(std::is_sorted(left.cbegin(), left.cend()));
|
|
Q_ASSERT(std::is_sorted(right.cbegin(), right.cend()));
|
|
|
|
for (const GraphicPieceInfo& info : left)
|
|
{
|
|
if (!std::binary_search(right.cbegin(), right.cend(), info))
|
|
{
|
|
differences.left.push_back(info);
|
|
}
|
|
}
|
|
|
|
for (const GraphicPieceInfo& info : right)
|
|
{
|
|
if (!std::binary_search(left.cbegin(), left.cend(), info))
|
|
{
|
|
differences.right.push_back(info);
|
|
}
|
|
}
|
|
|
|
const PDFReal epsilonSquared = epsilon * epsilon;
|
|
|
|
// If exact match fails, then try to use match with epsilon. For each
|
|
// item in left, we try to find matching item in right.
|
|
for (auto it = differences.left.begin(); it != differences.left.end();)
|
|
{
|
|
bool hasMatch = false;
|
|
|
|
const GraphicPieceInfo& leftInfo = *it;
|
|
for (auto it2 = differences.right.begin(); it2 != differences.right.end();)
|
|
{
|
|
// Heuristically compare these items
|
|
|
|
const GraphicPieceInfo& rightInfo = *it2;
|
|
if (leftInfo.type != rightInfo.type || !leftInfo.boundingRect.intersects(rightInfo.boundingRect))
|
|
{
|
|
++it2;
|
|
continue;
|
|
}
|
|
|
|
const int elementCountPath1 = leftInfo.pagePath.elementCount();
|
|
const int elementCountPath2 = rightInfo.pagePath.elementCount();
|
|
|
|
if (elementCountPath1 != elementCountPath2)
|
|
{
|
|
++it2;
|
|
continue;
|
|
}
|
|
|
|
hasMatch = (leftInfo.type != GraphicPieceInfo::Type::Image) || (leftInfo.imageHash == rightInfo.imageHash);
|
|
const int elementCount = leftInfo.pagePath.elementCount();
|
|
for (int i = 0; i < elementCount && hasMatch; ++i)
|
|
{
|
|
QPainterPath::Element leftElement = leftInfo.pagePath.elementAt(i);
|
|
QPainterPath::Element rightElement = rightInfo.pagePath.elementAt(i);
|
|
|
|
PDFReal diffX = leftElement.x - rightElement.x;
|
|
PDFReal diffY = leftElement.y - rightElement.y;
|
|
PDFReal squaredDistance = diffX * diffX + diffY * diffY;
|
|
|
|
hasMatch = (leftElement.type == rightElement.type) &&
|
|
(squaredDistance < epsilonSquared);
|
|
}
|
|
|
|
if (hasMatch)
|
|
{
|
|
it2 = differences.right.erase(it2);
|
|
}
|
|
else
|
|
{
|
|
++it2;
|
|
}
|
|
}
|
|
|
|
if (hasMatch)
|
|
{
|
|
it = differences.left.erase(it);
|
|
}
|
|
else
|
|
{
|
|
++it;
|
|
}
|
|
}
|
|
|
|
return differences;
|
|
}
|
|
|
|
std::vector<size_t> PDFDiffHelper::getLeftUnmatched(const PageSequence& sequence)
|
|
{
|
|
std::vector<size_t> result;
|
|
|
|
for (const auto& item : sequence)
|
|
{
|
|
if (item.isLeft())
|
|
{
|
|
result.push_back(item.index1);
|
|
}
|
|
}
|
|
|
|
return result;
|
|
}
|
|
|
|
std::vector<size_t> PDFDiffHelper::getRightUnmatched(const PageSequence& sequence)
|
|
{
|
|
std::vector<size_t> result;
|
|
|
|
for (const auto& item : sequence)
|
|
{
|
|
if (item.isRight())
|
|
{
|
|
result.push_back(item.index2);
|
|
}
|
|
}
|
|
|
|
return result;
|
|
}
|
|
|
|
void PDFDiffHelper::matchPage(PageSequence& sequence,
|
|
size_t leftPage,
|
|
size_t rightPage)
|
|
{
|
|
for (auto it = sequence.begin(); it != sequence.end();)
|
|
{
|
|
auto& item = *it;
|
|
|
|
if (item.isLeft() && item.index1 == leftPage)
|
|
{
|
|
item.index2 = rightPage;
|
|
}
|
|
|
|
if (item.isRight() && item.index2 == rightPage)
|
|
{
|
|
it = sequence.erase(it);
|
|
}
|
|
else
|
|
{
|
|
++it;
|
|
}
|
|
}
|
|
}
|
|
|
|
} // namespace pdf
|