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

DocDiff application: filters

Browse Source
This commit is contained in:
Jakub Melka
2021-09-29 16:59:13 +02:00
parent 8508fe9ef2
commit dee2535983
5 changed files with 569 additions and 49 deletions
Download Patch File Download Diff File Expand all files Collapse all files

191
Pdf4QtLib/sources/pdfdiff.cpp
View File

@@ -196,7 +196,7 @@ PDFDiffResult PDFDiff::perform()
m_progress->start(StepLast, std::move(info));
}
performSteps(leftPages, rightPages);
performSteps(leftPages, rightPages, result);
if (m_progress)
{
@@ -327,7 +327,9 @@ void PDFDiff::performPageMatching(const std::vector<PDFDiffPageContext>& leftPre
PDFAlgorithmLongestCommonSubsequenceBase::markSequence(pageSequence, leftPagesMoved, rightPagesMoved);
}
void PDFDiff::performSteps(const std::vector<PDFInteger>& leftPages, const std::vector<PDFInteger>& rightPages)
void PDFDiff::performSteps(const std::vector<PDFInteger>& leftPages,
const std::vector<PDFInteger>& rightPages,
PDFDiffResult& result)
{
std::vector<PDFDiffPageContext> leftPreparedPages;
std::vector<PDFDiffPageContext> rightPreparedPages;
@@ -449,7 +451,7 @@ void PDFDiff::performSteps(const std::vector<PDFInteger>& leftPages, const std::
// StepCompare
if (!m_cancelled)
{
performCompare(leftPreparedPages, rightPreparedPages, pageSequence, pageMatches);
performCompare(leftPreparedPages, rightPreparedPages, pageSequence, pageMatches, result);
stepProgress();
}
}
@@ -457,7 +459,8 @@ void PDFDiff::performSteps(const std::vector<PDFInteger>& leftPages, const std::
void PDFDiff::performCompare(const std::vector<PDFDiffPageContext>& leftPreparedPages,
const std::vector<PDFDiffPageContext>& rightPreparedPages,
PDFAlgorithmLongestCommonSubsequenceBase::Sequence& pageSequence,
const std::map<size_t, size_t>& pageMatches)
const std::map<size_t, size_t>& pageMatches,
PDFDiffResult& result)
{
using AlgorithmLCS = PDFAlgorithmLongestCommonSubsequenceBase;
@@ -471,11 +474,11 @@ void PDFDiff::performCompare(const std::vector<PDFDiffPageContext>& leftPrepared
Q_ASSERT(pageMatches.contains(leftPreparedPages.at(item.index1).pageIndex));
const PDFInteger leftIndex = leftPreparedPages[item.index1].pageIndex;
const PDFInteger rightIndex = pageMatches.at(leftIndex);
m_result.addPageMoved(leftIndex, rightIndex);
result.addPageMoved(leftIndex, rightIndex);
}
if (item.isMoved())
{
m_result.addPageMoved(leftPreparedPages[item.index1].pageIndex, rightPreparedPages[item.index2].pageIndex);
result.addPageMoved(leftPreparedPages[item.index1].pageIndex, rightPreparedPages[item.index2].pageIndex);
}
}
@@ -527,20 +530,20 @@ void PDFDiff::performCompare(const std::vector<PDFDiffPageContext>& leftPrepared
case PDFDiffHelper::GraphicPieceInfo::Type::Text:
if (isTextComparedAsVectorGraphics)
{
m_result.addRemovedTextCharContent(leftPageContext.pageIndex, info.boundingRect);
result.addRemovedTextCharContent(leftPageContext.pageIndex, info.boundingRect);
}
break;
case PDFDiffHelper::GraphicPieceInfo::Type::VectorGraphics:
m_result.addRemovedVectorGraphicContent(leftPageContext.pageIndex, info.boundingRect);
result.addRemovedVectorGraphicContent(leftPageContext.pageIndex, info.boundingRect);
break;
case PDFDiffHelper::GraphicPieceInfo::Type::Image:
m_result.addRemovedImageContent(leftPageContext.pageIndex, info.boundingRect);
result.addRemovedImageContent(leftPageContext.pageIndex, info.boundingRect);
break;
case PDFDiffHelper::GraphicPieceInfo::Type::Shading:
m_result.addRemovedShadingContent(leftPageContext.pageIndex, info.boundingRect);
result.addRemovedShadingContent(leftPageContext.pageIndex, info.boundingRect);
break;
default:
@@ -556,20 +559,20 @@ void PDFDiff::performCompare(const std::vector<PDFDiffPageContext>& leftPrepared
case PDFDiffHelper::GraphicPieceInfo::Type::Text:
if (isTextComparedAsVectorGraphics)
{
m_result.addAddedTextCharContent(rightPageContext.pageIndex, info.boundingRect);
result.addAddedTextCharContent(rightPageContext.pageIndex, info.boundingRect);
}
break;
case PDFDiffHelper::GraphicPieceInfo::Type::VectorGraphics:
m_result.addAddedVectorGraphicContent(rightPageContext.pageIndex, info.boundingRect);
result.addAddedVectorGraphicContent(rightPageContext.pageIndex, info.boundingRect);
break;
case PDFDiffHelper::GraphicPieceInfo::Type::Image:
m_result.addAddedImageContent(rightPageContext.pageIndex, info.boundingRect);
result.addAddedImageContent(rightPageContext.pageIndex, info.boundingRect);
break;
case PDFDiffHelper::GraphicPieceInfo::Type::Shading:
m_result.addAddedShadingContent(rightPageContext.pageIndex, info.boundingRect);
result.addAddedShadingContent(rightPageContext.pageIndex, info.boundingRect);
break;
default:
@@ -588,7 +591,7 @@ void PDFDiff::performCompare(const std::vector<PDFDiffPageContext>& leftPrepared
rightTextFlow.append(rightPageContext.text);
}
m_result.addPageAdded(rightPageContext.pageIndex);
result.addPageAdded(rightPageContext.pageIndex);
}
if (item.isRemoved())
{
@@ -599,7 +602,7 @@ void PDFDiff::performCompare(const std::vector<PDFDiffPageContext>& leftPrepared
leftTextFlow.append(leftPageContext.text);
}
m_result.addPageRemoved(leftPageContext.pageIndex);
result.addPageRemoved(leftPageContext.pageIndex);
}
}
@@ -628,11 +631,11 @@ void PDFDiff::performCompare(const std::vector<PDFDiffPageContext>& leftPrepared
if (item.isAdded())
{
m_result.addPageAdded(rightPreparedPages[item.index2].pageIndex);
result.addPageAdded(rightPreparedPages[item.index2].pageIndex);
}
if (item.isRemoved())
{
m_result.addPageRemoved(leftPreparedPages[item.index1].pageIndex);
result.addPageRemoved(leftPreparedPages[item.index1].pageIndex);
}
}
}
@@ -641,7 +644,7 @@ void PDFDiff::performCompare(const std::vector<PDFDiffPageContext>& leftPrepared
QMutex mutex;
// Jakub Melka: try to compare text differences
auto compareTexts = [this, &mutex](PDFDiffHelper::TextFlowDifferences& context)
auto compareTexts = [this, &mutex, &result](PDFDiffHelper::TextFlowDifferences& context)
{
using TextCompareItem = PDFDiffHelper::TextCompareItem;
const bool isWordsComparingMode = m_options.testFlag(CompareWords);
@@ -804,25 +807,27 @@ void PDFDiff::performCompare(const std::vector<PDFDiffPageContext>& leftPrepared
QMutexLocker locker(&mutex);
if (!leftString.isEmpty() && !rightString.isEmpty())
{
m_result.addTextReplaced(pageIndex1, pageIndex2, leftString, rightString, leftRectInfos, rightRectInfos);
result.addTextReplaced(pageIndex1, pageIndex2, leftString, rightString, leftRectInfos, rightRectInfos);
}
else
{
if (!leftString.isEmpty())
{
m_result.addTextRemoved(pageIndex1, leftString, leftRectInfos);
result.addTextRemoved(pageIndex1, leftString, leftRectInfos);
}
if (!rightString.isEmpty())
{
m_result.addTextAdded(pageIndex2, rightString, rightRectInfos);
result.addTextAdded(pageIndex2, rightString, rightRectInfos);
}
}
}
};
PDFExecutionPolicy::execute(PDFExecutionPolicy::Scope::Page, textFlowDifferences.begin(), textFlowDifferences.end(), compareTexts);
//std::for_each(textFlowDifferences.begin(), textFlowDifferences.end(), compareTexts);
// Jakub Melka: sort results
result.finalize();
}
void PDFDiff::finalizeGraphicsPieces(PDFDiffPageContext& context)
@@ -1053,6 +1058,22 @@ void PDFDiffResult::addTextReplaced(PDFInteger pageIndex1,
m_differences.emplace_back(std::move(difference));
}
void PDFDiffResult::finalize()
{
auto predicate = [](const Difference& l, const Difference& r)
{
return qMax(l.pageIndex1, l.pageIndex2) < qMax(r.pageIndex1, r.pageIndex2);
};
std::stable_sort(m_differences.begin(), m_differences.end(), predicate);
m_typeFlags = 0;
for (const Difference& difference : m_differences)
{
m_typeFlags |= static_cast<uint32_t>(difference.type);
}
}
QString PDFDiffResult::getMessage(size_t index) const
{
if (index >= m_differences.size())
@@ -1113,6 +1134,50 @@ QString PDFDiffResult::getMessage(size_t index) const
return QString();
}
PDFDiffResult PDFDiffResult::filter(bool filterPageMoveDifferences,
bool filterTextDifferences,
bool filterVectorGraphicsDifferences,
bool filterImageDifferences,
bool filterShadingDifferences)
{
PDFDiffResult filteredResult = *this;
uint32_t typeFlags = 0;
if (filterPageMoveDifferences)
{
typeFlags |= FLAGS_PAGE_MOVE;
}
if (filterTextDifferences)
{
typeFlags |= FLAGS_TEXT;
}
if (filterVectorGraphicsDifferences)
{
typeFlags |= FLAGS_VECTOR_GRAPHICS;
}
if (filterImageDifferences)
{
typeFlags |= FLAGS_IMAGE;
}
if (filterShadingDifferences)
{
typeFlags |= FLAGS_SHADING;
}
auto remove = [typeFlags](const Difference& difference)
{
return (uint32_t(difference.type) & typeFlags) == 0;
};
filteredResult.m_differences.erase(std::remove_if(filteredResult.m_differences.begin(), filteredResult.m_differences.end(), remove), filteredResult.m_differences.end());
return filteredResult;
}
void PDFDiffResult::addRectLeft(Difference& difference, QRectF rect)
{
difference.leftRectIndex = m_rects.size();
@@ -1376,4 +1441,84 @@ void PDFDiffHelper::refineTextRectangles(PDFDiffResult::RectInfos& items)
items = std::move(refinedItems);
}
PDFDiffResultNavigator::PDFDiffResultNavigator(QObject* parent) :
QObject(parent),
m_diffResult(nullptr),
m_currentIndex(0)
{
}
PDFDiffResultNavigator::~PDFDiffResultNavigator()
{
}
void PDFDiffResultNavigator::setResult(const PDFDiffResult* diffResult)
{
if (m_diffResult != diffResult)
{
m_diffResult = diffResult;
emit selectionChanged(m_currentIndex);
}
}
bool PDFDiffResultNavigator::isSelected() const
{
const size_t limit = getLimit();
return m_currentIndex >= 0 && m_currentIndex < limit;
}
bool PDFDiffResultNavigator::canGoNext() const
{
const size_t limit = getLimit();
return limit > 0 && m_currentIndex + 1 < limit;
}
bool PDFDiffResultNavigator::canGoPrevious() const
{
const size_t limit = getLimit();
return limit > 0 && m_currentIndex > 0;
}
void PDFDiffResultNavigator::goNext()
{
if (!canGoNext())
{
return;
}
++m_currentIndex;
emit selectionChanged(m_currentIndex);
}
void PDFDiffResultNavigator::goPrevious()
{
if (!canGoPrevious())
{
return;
}
const size_t limit = getLimit();
if (m_currentIndex >= limit)
{
m_currentIndex = limit - 1;
}
else
{
--m_currentIndex;
}
emit selectionChanged(m_currentIndex);
}
void PDFDiffResultNavigator::update()
{
const size_t limit = getLimit();
if (limit > 0 && m_currentIndex >= limit)
{
m_currentIndex = limit - 1;
emit selectionChanged(m_currentIndex);
}
}
} // namespace pdf

110
Pdf4QtLib/sources/pdfdiff.h
View File

@@ -35,28 +35,28 @@ namespace pdf
struct PDFDiffPageContext;
class PDFDiffResult
class PDF4QTLIBSHARED_EXPORT PDFDiffResult
{
public:
explicit PDFDiffResult();
enum class Type
enum class Type : uint32_t
{
Invalid,
PageMoved,
PageAdded,
PageRemoved,
RemovedTextCharContent,
RemovedVectorGraphicContent,
RemovedImageContent,
RemovedShadingContent,
AddedTextCharContent,
AddedVectorGraphicContent,
AddedImageContent,
AddedShadingContent,
TextReplaced,
TextAdded,
TextRemoved,
Invalid = 0x0000,
PageMoved = 0x0001,
PageAdded = 0x0002,
PageRemoved = 0x0004,
RemovedTextCharContent = 0x0008,
RemovedVectorGraphicContent = 0x0010,
RemovedImageContent = 0x0020,
RemovedShadingContent = 0x0040,
AddedTextCharContent = 0x0080,
AddedVectorGraphicContent = 0x0100,
AddedImageContent = 0x0200,
AddedShadingContent = 0x0400,
TextReplaced = 0x0800,
TextAdded = 0x1000,
TextRemoved = 0x2000,
};
using RectInfos = std::vector<std::pair<PDFInteger, QRectF>>;
@@ -77,9 +77,33 @@ public:
/// \param index Index
QString getMessage(size_t index) const;
bool hasPageMoveDifferences() const { return m_typeFlags & FLAGS_PAGE_MOVE; }
bool hasTextDifferences() const { return m_typeFlags & FLAGS_TEXT; }
bool hasVectorGraphicsDifferences() const { return m_typeFlags & FLAGS_VECTOR_GRAPHICS; }
bool hasImageDifferences() const { return m_typeFlags & FLAGS_IMAGE; }
bool hasShadingDifferences() const { return m_typeFlags & FLAGS_SHADING; }
/// Filters results using given critera
/// \param filterPageMoveDifferences Filter page move differences?
/// \param filterTextDifferences Filter text diffferences?
/// \param filterVectorGraphicsDifferences Filter vector graphics differences?
/// \param filterImageDifferences Filter image differences?
/// \param filterShadingDifferences Filter shading differences?
PDFDiffResult filter(bool filterPageMoveDifferences,
bool filterTextDifferences,
bool filterVectorGraphicsDifferences,
bool filterImageDifferences,
bool filterShadingDifferences);
private:
friend class PDFDiff;
static constexpr uint32_t FLAGS_PAGE_MOVE = uint32_t(Type::PageMoved) | uint32_t(Type::PageAdded) | uint32_t(Type::PageRemoved);
static constexpr uint32_t FLAGS_TEXT = uint32_t(Type::RemovedTextCharContent) | uint32_t(Type::AddedTextCharContent) | uint32_t(Type::TextReplaced) | uint32_t(Type::TextAdded) | uint32_t(Type::TextRemoved);
static constexpr uint32_t FLAGS_VECTOR_GRAPHICS = uint32_t(Type::RemovedVectorGraphicContent) | uint32_t(Type::AddedVectorGraphicContent);
static constexpr uint32_t FLAGS_IMAGE = uint32_t(Type::RemovedImageContent) | uint32_t(Type::AddedImageContent);
static constexpr uint32_t FLAGS_SHADING = uint32_t(Type::RemovedShadingContent) | uint32_t(Type::AddedShadingContent);
void addPageMoved(PDFInteger pageIndex1, PDFInteger pageIndex2);
void addPageAdded(PDFInteger pageIndex);
void addPageRemoved(PDFInteger pageIndex);
@@ -103,6 +127,8 @@ private:
const RectInfos& rectInfos1,
const RectInfos& rectInfos2);
void finalize();
/// Single content difference descriptor. It describes type
/// of difference (such as graphics, image, text change) on a page
/// or on a list of multiple pages.
@@ -131,6 +157,50 @@ private:
RectInfos m_rects; ///< Rectangles with page indices
PDFOperationResult m_result;
QStringList m_strings;
uint32_t m_typeFlags = 0;
};
/// Class for result navigation, can go to next, or previous result.
class PDF4QTLIBSHARED_EXPORT PDFDiffResultNavigator : public QObject
{
Q_OBJECT
public:
explicit PDFDiffResultNavigator(QObject* parent);
virtual ~PDFDiffResultNavigator() override;
void setResult(const PDFDiffResult* diffResult);
/// Returns true, if valid result is selected
bool isSelected() const;
/// Returns true if action go to next result can be performed,
/// otherwise false is returned.
bool canGoNext() const;
/// Returns true if action go to previous result can be performed,
/// otherwise false is returned.
bool canGoPrevious() const;
/// Goes to next result. If action cannot be performed,
/// nothing happens and signal is not emitted.
void goNext();
/// Goes to previous result. If action cannot be performed,
/// nothing happens and signal is not emitted.
void goPrevious();
/// Updates selection, if difference result was changed
void update();
signals:
void selectionChanged(size_t currentIndex);
private:
size_t getLimit() const { return m_diffResult ? m_diffResult->getDifferencesCount() : 0; }
const PDFDiffResult* m_diffResult;
size_t m_currentIndex;
};
/// Diff engine for comparing two pdf documents.
@@ -218,7 +288,8 @@ private:
PDFDiffResult perform();
void stepProgress();
void performSteps(const std::vector<PDFInteger>& leftPages,
const std::vector<PDFInteger>& rightPages);
const std::vector<PDFInteger>& rightPages,
PDFDiffResult& result);
void performPageMatching(const std::vector<PDFDiffPageContext>& leftPreparedPages,
const std::vector<PDFDiffPageContext>& rightPreparedPages,
PDFAlgorithmLongestCommonSubsequenceBase::Sequence& pageSequence,
@@ -226,7 +297,8 @@ private:
void performCompare(const std::vector<PDFDiffPageContext>& leftPreparedPages,
const std::vector<PDFDiffPageContext>& rightPreparedPages,
PDFAlgorithmLongestCommonSubsequenceBase::Sequence& pageSequence,
const std::map<size_t, size_t>& pageMatches);
const std::map<size_t, size_t>& pageMatches,
PDFDiffResult& result);
void finalizeGraphicsPieces(PDFDiffPageContext& context);
void onComparationPerformed();