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