Home | History | Annotate | Download | only in fpdftext
      1 // Copyright 2014 PDFium Authors. All rights reserved.
      2 // Use of this source code is governed by a BSD-style license that can be
      3 // found in the LICENSE file.
      4 
      5 // Original code copyright 2014 Foxit Software Inc. http://www.foxitsoftware.com
      6 
      7 #ifndef CORE_SRC_FPDFTEXT_TEXT_INT_H_
      8 #define CORE_SRC_FPDFTEXT_TEXT_INT_H_
      9 
     10 class CPDF_TextParseOptions
     11 {
     12 public:
     13     CPDF_TextParseOptions();
     14     FX_BOOL			m_bCheckObjectOrder;
     15     FX_BOOL			m_bCheckDirection;
     16     int				m_nCheckSameObject;
     17 };
     18 class CPDF_TextPage;
     19 class CPDF_LinkExtract;
     20 class CPDF_TextPageFind;
     21 class CPDF_DocProgressiveSearch;
     22 #define FPDFTEXT_CHAR_ERROR			-1
     23 #define FPDFTEXT_CHAR_NORMAL		0
     24 #define FPDFTEXT_CHAR_GENERATED		1
     25 #define FPDFTEXT_CHAR_UNUNICODE		2
     26 #define FPDFTEXT_CHAR_HYPHEN		3
     27 #define FPDFTEXT_CHAR_PIECE			4
     28 #define FPDFTEXT_MC_PASS			0
     29 #define FPDFTEXT_MC_DONE			1
     30 #define FPDFTEXT_MC_DELAY			2
     31 typedef struct _PAGECHAR_INFO {
     32     int					m_CharCode;
     33     FX_WCHAR			m_Unicode;
     34     FX_FLOAT			m_OriginX;
     35     FX_FLOAT			m_OriginY;
     36     FX_INT32			m_Flag;
     37     CFX_FloatRect		m_CharBox;
     38     CPDF_TextObject*	m_pTextObj;
     39     CFX_AffineMatrix	m_Matrix;
     40     int					m_Index;
     41 } PAGECHAR_INFO;
     42 typedef	CFX_SegmentedArray<PAGECHAR_INFO> PAGECHAR_InfoArray;
     43 typedef struct {
     44     int	m_Start;
     45     int m_nCount;
     46 } FPDF_SEGMENT;
     47 typedef CFX_ArrayTemplate<FPDF_SEGMENT> SEGMENT_Array;
     48 typedef struct {
     49     CPDF_TextObject*	m_pTextObj;
     50     CFX_AffineMatrix	m_formMatrix;
     51 } PDFTEXT_Obj;
     52 typedef CFX_ArrayTemplate<PDFTEXT_Obj> LINEOBJ;
     53 class CPDF_TextPage: public IPDF_TextPage
     54 {
     55 public:
     56     CPDF_TextPage(const CPDF_Page* pPage, int flags = 0);
     57     CPDF_TextPage(const CPDF_PageObjects* pPage, int flags = 0);
     58     CPDF_TextPage(const CPDF_Page* pPage, CPDFText_ParseOptions ParserOptions);
     59     virtual FX_BOOL					ParseTextPage();
     60     virtual void					NormalizeObjects(FX_BOOL bNormalize);
     61     virtual	FX_BOOL					IsParsered() const
     62     {
     63         return m_IsParsered;
     64     }
     65     virtual ~CPDF_TextPage() {};
     66 public:
     67     virtual int CharIndexFromTextIndex(int TextIndex)const ;
     68     virtual int TextIndexFromCharIndex(int CharIndex)const;
     69     virtual int						CountChars() const;
     70     virtual	void					GetCharInfo(int index, FPDF_CHAR_INFO & info) const;
     71     virtual void					GetRectArray(int start, int nCount, CFX_RectArray& rectArray) const;
     72     virtual int						GetIndexAtPos(CPDF_Point point, FX_FLOAT xTorelance, FX_FLOAT yTorelance) const;
     73     virtual int						GetIndexAtPos(FX_FLOAT x, FX_FLOAT y, FX_FLOAT xTorelance,
     74             FX_FLOAT yTorelance) const;
     75     virtual CFX_WideString			GetTextByRect(const CFX_FloatRect& rect) const;
     76     virtual void					GetRectsArrayByRect(const CFX_FloatRect& rect, CFX_RectArray& resRectArray) const;
     77     virtual	int						GetOrderByDirection(int order, int direction) const;
     78     virtual	CFX_WideString			GetPageText(int start = 0, int nCount = -1) const;
     79 
     80     virtual int						CountRects(int start, int nCount);
     81     virtual	void					GetRect(int rectIndex, FX_FLOAT& left, FX_FLOAT& top
     82                                             , FX_FLOAT& right, FX_FLOAT &bottom) const;
     83     virtual FX_BOOL					GetBaselineRotate(int rectIndex, int& Rotate);
     84     virtual FX_BOOL					GetBaselineRotate(const CFX_FloatRect& rect, int& Rotate);
     85     virtual	int						CountBoundedSegments(FX_FLOAT left, FX_FLOAT top,
     86             FX_FLOAT right, FX_FLOAT bottom, FX_BOOL bContains = FALSE);
     87     virtual	void					GetBoundedSegment(int index, int& start, int& count) const;
     88     virtual int						GetWordBreak(int index, int direction) const;
     89 public:
     90     const	PAGECHAR_InfoArray*		GetCharList() const
     91     {
     92         return &m_charList;
     93     }
     94     static	FX_BOOL					IsRectIntersect(const CFX_FloatRect& rect1, const CFX_FloatRect& rect2);
     95     static	FX_BOOL					IsLetter(FX_WCHAR unicode);
     96 private:
     97     FX_BOOL							IsHyphen(FX_WCHAR curChar);
     98     FX_BOOL							IsControlChar(PAGECHAR_INFO* pCharInfo);
     99     FX_BOOL							GetBaselineRotate(int start, int end, int& Rotate);
    100     void							ProcessObject();
    101     void							ProcessFormObject(CPDF_FormObject*	pFormObj, const CFX_AffineMatrix& formMatrix);
    102     void							ProcessTextObject(PDFTEXT_Obj pObj);
    103     void							ProcessTextObject(CPDF_TextObject*	pTextObj, const CFX_AffineMatrix& formMatrix, FX_POSITION ObjPos);
    104     int								ProcessInsertObject(const CPDF_TextObject* pObj, const CFX_AffineMatrix& formMatrix);
    105     FX_BOOL							GenerateCharInfo(FX_WCHAR unicode, PAGECHAR_INFO& info);
    106     FX_BOOL							IsSameAsPreTextObject(CPDF_TextObject* pTextObj, FX_POSITION ObjPos);
    107     FX_BOOL							IsSameTextObject(CPDF_TextObject* pTextObj1, CPDF_TextObject* pTextObj2);
    108     int								GetCharWidth(FX_DWORD charCode, CPDF_Font* pFont) const;
    109     void							CloseTempLine();
    110     void							OnPiece(IFX_BidiChar* pBidi, CFX_WideString& str);
    111     FX_INT32	PreMarkedContent(PDFTEXT_Obj pObj);
    112     void		ProcessMarkedContent(PDFTEXT_Obj pObj);
    113     void		CheckMarkedContentObject(FX_INT32& start, FX_INT32& nCount) const;
    114     void		FindPreviousTextObject(void);
    115     void		AddCharInfoByLRDirection(CFX_WideString& str, int i);
    116     void		AddCharInfoByRLDirection(CFX_WideString& str, int i);
    117     FX_INT32	GetTextObjectWritingMode(const CPDF_TextObject* pTextObj);
    118     FX_INT32	FindTextlineFlowDirection();
    119     void SwapTempTextBuf(FX_INT32 iCharListStartAppend,
    120                          FX_INT32 iBufStartAppend);
    121     FX_BOOL IsRightToLeft(const CPDF_TextObject* pTextObj,
    122                           const CPDF_Font* pFont,
    123                           int nItems) const;
    124 protected:
    125     CPDFText_ParseOptions			m_ParseOptions;
    126     CFX_WordArray					m_CharIndex;
    127     const CPDF_PageObjects*			m_pPage;
    128     PAGECHAR_InfoArray				m_charList;
    129     CFX_WideTextBuf					m_TextBuf;
    130     PAGECHAR_InfoArray				m_TempCharList;
    131     CFX_WideTextBuf					m_TempTextBuf;
    132     int								m_parserflag;
    133     CPDF_TextObject*				m_pPreTextObj;
    134     CFX_AffineMatrix				m_perMatrix;
    135     FX_BOOL							m_IsParsered;
    136     CFX_AffineMatrix				m_DisplayMatrix;
    137 
    138     SEGMENT_Array					m_Segment;
    139     CFX_RectArray					m_SelRects;
    140     LINEOBJ							m_LineObj;
    141     FX_BOOL							m_TextlineDir;
    142     CFX_FloatRect					m_CurlineRect;
    143 };
    144 class CPDF_TextPageFind: public IPDF_TextPageFind
    145 {
    146 public:
    147     CPDF_TextPageFind(const IPDF_TextPage* pTextPage);
    148     virtual							~CPDF_TextPageFind() {};
    149 public:
    150     virtual	FX_BOOL					FindFirst(const CFX_WideString& findwhat, int flags, int startPos = 0);
    151     virtual	FX_BOOL					FindNext();
    152     virtual	FX_BOOL					FindPrev();
    153 
    154     virtual void					GetRectArray(CFX_RectArray& rects) const;
    155     virtual int						GetCurOrder() const;
    156     virtual int						GetMatchedCount()const;
    157 protected:
    158     void							ExtractFindWhat(const CFX_WideString& findwhat);
    159     FX_BOOL							IsMatchWholeWord(const CFX_WideString& csPageText, int startPos, int endPos);
    160     FX_BOOL							ExtractSubString(CFX_WideString& rString, FX_LPCWSTR lpszFullString,
    161             int iSubString, FX_WCHAR chSep);
    162     CFX_WideString					MakeReverse(const CFX_WideString& str);
    163     int								ReverseFind(const CFX_WideString& csPageText, const CFX_WideString& csWord, int nStartPos, int& WordLength);
    164     int								GetCharIndex(int index) const;
    165 private:
    166     CFX_WordArray					m_CharIndex;
    167     const IPDF_TextPage*			m_pTextPage;
    168     CFX_WideString					m_strText;
    169     CFX_WideString					m_findWhat;
    170     int								m_flags;
    171     CFX_WideStringArray				m_csFindWhatArray;
    172     int								m_findNextStart;
    173     int								m_findPreStart;
    174     FX_BOOL							m_bMatchCase;
    175     FX_BOOL							m_bMatchWholeWord;
    176     int								m_resStart;
    177     int								m_resEnd;
    178     CFX_RectArray					m_resArray;
    179     FX_BOOL							m_IsFind;
    180 };
    181 class CPDF_LinkExt
    182 {
    183 public:
    184     CPDF_LinkExt() {};
    185     int								m_Start;
    186     int								m_Count;
    187     CFX_WideString					m_strUrl;
    188     virtual							~CPDF_LinkExt() {};
    189 };
    190 typedef CFX_ArrayTemplate<CPDF_LinkExt*> LINK_InfoArray;
    191 class CPDF_LinkExtract: public IPDF_LinkExtract
    192 {
    193 public:
    194     CPDF_LinkExtract();
    195     virtual							~CPDF_LinkExtract();
    196     virtual FX_BOOL					ExtractLinks(const IPDF_TextPage* pTextPage);
    197     virtual	FX_BOOL					IsExtract() const
    198     {
    199         return m_IsParserd;
    200     }
    201 public:
    202     virtual int						CountLinks() const;
    203     virtual	CFX_WideString			GetURL(int index) const;
    204     virtual	void					GetBoundedSegment(int index, int& start, int& count) const;
    205     virtual	void					GetRects(int index, CFX_RectArray& rects)const;
    206 protected:
    207     void							parserLink();
    208     void							DeleteLinkList();
    209     FX_BOOL							CheckWebLink(CFX_WideString& strBeCheck);
    210     FX_BOOL							CheckMailLink(CFX_WideString& str);
    211     FX_BOOL							AppendToLinkList(int start, int count, const CFX_WideString& strUrl);
    212 private:
    213     LINK_InfoArray					m_LinkList;
    214     const CPDF_TextPage*			m_pTextPage;
    215     CFX_WideString					m_strPageText;
    216     FX_BOOL							m_IsParserd;
    217 };
    218 FX_STRSIZE FX_Unicode_GetNormalization(FX_WCHAR wch, FX_LPWSTR pDst);
    219 void NormalizeString(CFX_WideString& str);
    220 void NormalizeCompositeChar(FX_WCHAR wChar, CFX_WideString& sDest);
    221 
    222 #endif  // CORE_SRC_FPDFTEXT_TEXT_INT_H_
    223