2007-09-03 19:52:46 +00:00
|
|
|
/***************************************************************************
|
|
|
|
* Copyright (C) 2006 by Tobias Koenig <tokoe@kde.org> *
|
|
|
|
* Copyright (C) 2007 by Pino Toscano <pino@kde.org> *
|
|
|
|
* *
|
|
|
|
* This program is free software; you can redistribute it and/or modify *
|
|
|
|
* it under the terms of the GNU General Public License as published by *
|
|
|
|
* the Free Software Foundation; either version 2 of the License, or *
|
|
|
|
* (at your option) any later version. *
|
|
|
|
***************************************************************************/
|
|
|
|
|
|
|
|
#ifndef _OKULAR_TEXTPAGE_P_H_
|
|
|
|
#define _OKULAR_TEXTPAGE_P_H_
|
|
|
|
|
2008-05-01 19:03:02 +00:00
|
|
|
#include <QtCore/QList>
|
2007-09-03 19:52:46 +00:00
|
|
|
#include <QtCore/QMap>
|
|
|
|
#include <QtGui/QMatrix>
|
|
|
|
|
|
|
|
class SearchPoint;
|
2008-05-01 19:03:02 +00:00
|
|
|
class TinyTextEntity;
|
2011-07-08 18:12:39 +00:00
|
|
|
class RegionText;
|
2007-09-03 19:52:46 +00:00
|
|
|
|
|
|
|
namespace Okular
|
|
|
|
{
|
|
|
|
|
2007-09-04 22:37:39 +00:00
|
|
|
class PagePrivate;
|
2008-05-01 19:03:02 +00:00
|
|
|
typedef QList< TinyTextEntity* > TextList;
|
2007-09-04 22:37:39 +00:00
|
|
|
|
2009-12-27 14:22:13 +00:00
|
|
|
typedef bool ( *TextComparisonFunction )( const QStringRef & from, const QStringRef & to,
|
|
|
|
int *fromLength, int *toLength );
|
|
|
|
|
2011-06-29 12:17:15 +00:00
|
|
|
/**
|
2011-08-21 04:44:42 +00:00
|
|
|
* Make a line of TextList and store the bounding rectangle of line
|
|
|
|
*/
|
2011-06-29 12:17:15 +00:00
|
|
|
typedef QList<TextList> SortedTextList;
|
|
|
|
typedef QList<QRect> LineRect;
|
|
|
|
|
2011-08-21 04:44:42 +00:00
|
|
|
/**
|
|
|
|
* A list of RegionText. It keeps a bunch of TextList with their bounding rectangles
|
|
|
|
*/
|
2011-07-13 11:16:03 +00:00
|
|
|
typedef QList<RegionText> RegionTextList;
|
2011-06-29 12:17:15 +00:00
|
|
|
|
2007-09-03 19:52:46 +00:00
|
|
|
class TextPagePrivate
|
|
|
|
{
|
|
|
|
public:
|
2008-04-18 09:57:55 +00:00
|
|
|
TextPagePrivate();
|
2007-09-03 19:52:46 +00:00
|
|
|
~TextPagePrivate();
|
|
|
|
|
|
|
|
RegularAreaRect * findTextInternalForward( int searchID, const QString &query,
|
|
|
|
Qt::CaseSensitivity caseSensitivity,
|
2009-12-27 14:22:13 +00:00
|
|
|
TextComparisonFunction comparer,
|
2008-05-01 19:03:02 +00:00
|
|
|
const TextList::ConstIterator &start,
|
|
|
|
const TextList::ConstIterator &end );
|
2008-03-14 14:14:06 +00:00
|
|
|
RegularAreaRect * findTextInternalBackward( int searchID, const QString &query,
|
|
|
|
Qt::CaseSensitivity caseSensitivity,
|
2009-12-27 14:22:13 +00:00
|
|
|
TextComparisonFunction comparer,
|
2008-05-01 19:03:02 +00:00
|
|
|
const TextList::ConstIterator &start,
|
|
|
|
const TextList::ConstIterator &end );
|
2007-09-03 19:52:46 +00:00
|
|
|
|
2011-08-17 04:45:46 +00:00
|
|
|
/**
|
2011-08-23 14:10:43 +00:00
|
|
|
* Copy a TextList to m_words, the pointers of list are adopted
|
2011-08-17 04:45:46 +00:00
|
|
|
*/
|
2011-08-23 14:10:43 +00:00
|
|
|
void setWordList(const TextList &list);
|
2011-07-20 07:04:59 +00:00
|
|
|
|
2011-08-17 04:45:46 +00:00
|
|
|
/**
|
2011-08-23 14:10:43 +00:00
|
|
|
* Copy m_words to a new TextList, it is the caller responsability to free the pointers
|
2011-08-17 04:45:46 +00:00
|
|
|
*/
|
2011-08-23 14:10:43 +00:00
|
|
|
TextList duplicateWordList() const;
|
2011-07-10 14:54:38 +00:00
|
|
|
|
2011-10-17 17:09:26 +00:00
|
|
|
/**
|
|
|
|
* Make necessary modifications in the TextList to make the text order correct, so
|
|
|
|
* that textselection works fine
|
|
|
|
*/
|
|
|
|
void correctTextOrder();
|
|
|
|
|
2011-07-13 13:27:04 +00:00
|
|
|
/**
|
2011-08-17 04:45:46 +00:00
|
|
|
* Remove odd spaces which are much bigger than normal spaces from m_words
|
|
|
|
*/
|
2011-07-13 13:27:04 +00:00
|
|
|
void removeSpace();
|
|
|
|
|
|
|
|
/**
|
2011-08-17 04:45:46 +00:00
|
|
|
* Create words from characters
|
|
|
|
*/
|
2011-11-03 18:11:23 +00:00
|
|
|
QHash<QRect, RegionText> makeWordFromCharacters();
|
2011-07-13 13:27:04 +00:00
|
|
|
|
|
|
|
/**
|
2011-08-17 04:45:46 +00:00
|
|
|
* Create lines from TextList and sort them according to their position
|
|
|
|
*/
|
2011-11-30 22:13:34 +00:00
|
|
|
void makeAndSortLines(const TextList &words, SortedTextList *lines, LineRect *line_rects) const;
|
2011-07-13 13:27:04 +00:00
|
|
|
|
|
|
|
/**
|
2011-08-17 04:45:46 +00:00
|
|
|
* Caluclate statistical info like, word spacing, column spacing, line spacing from the Lines
|
|
|
|
* we made
|
|
|
|
*/
|
2011-11-30 22:50:01 +00:00
|
|
|
void calculateStatisticalInformation(const TextList &words, int *word_spacing,
|
2011-11-30 22:21:51 +00:00
|
|
|
int *line_spacing, int *column_spacing) const;
|
2011-07-13 13:27:04 +00:00
|
|
|
|
|
|
|
/**
|
2011-08-17 04:45:46 +00:00
|
|
|
* Functions necessary for document file segmentation into text regions for document layout
|
|
|
|
* analysis.
|
|
|
|
*/
|
2011-11-30 22:36:10 +00:00
|
|
|
RegionTextList XYCutForBoundingBoxes();
|
2011-07-13 13:27:04 +00:00
|
|
|
|
2011-07-12 14:24:17 +00:00
|
|
|
/**
|
2011-08-17 04:45:46 +00:00
|
|
|
* Add additional spaces between words, if necessary, which can make the words valuable
|
|
|
|
* while copying after selection
|
|
|
|
*/
|
2011-08-23 15:15:12 +00:00
|
|
|
void addNecessarySpace(RegionTextList tree);
|
2011-07-13 13:27:04 +00:00
|
|
|
|
|
|
|
/**
|
2011-08-17 04:45:46 +00:00
|
|
|
* Break the words into characters, so the text selection wors fine
|
|
|
|
*/
|
2011-11-03 18:11:23 +00:00
|
|
|
void breakWordIntoCharacters(const QHash<QRect, RegionText> &words_char_map);
|
2011-07-13 13:27:04 +00:00
|
|
|
|
|
|
|
// variables those can be accessed directly from TextPage
|
2008-05-01 19:03:02 +00:00
|
|
|
TextList m_words;
|
2007-09-03 19:52:46 +00:00
|
|
|
QMap< int, SearchPoint* > m_searchPoints;
|
2007-09-04 22:37:39 +00:00
|
|
|
PagePrivate *m_page;
|
2007-09-03 19:52:46 +00:00
|
|
|
};
|
|
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
#endif
|