fs_layoutrecognition.h

浏览该文件的文档.

#ifndef FS_LAYOUTRECOGNITION_H_
#define FS_LAYOUTRECOGNITION_H_

#include "common/fs_common.h"
#include "pdf/fs_pdfdoc.h"
#include "pdf/fs_pdfpage.h"

namespace foxit {
namespace addon {
namespace layoutrecognition {
class LRElement : public Base {
 public:
  typedef enum _ElementType {
    e_ElementTypeDocument = 0x0000,
    e_ElementTypeDocumentFragment = 0x0001,
    e_ElementTypePart = 0x0002,
    e_ElementTypeArt = 0x0101,
    e_ElementTypeSect = 0x0102,
    e_ElementTypeDiv = 0x0100,
    e_ElementTypeBlockQuote = 0x0103,
    e_ElementTypeCaption = 0x0104,
    e_ElementTypeTOC = 0x0105,
    e_ElementTypeTOCI = 0x0106,
    e_ElementTypeIndex = 0x0107,
    e_ElementTypeAside = 0x0108,
    e_ElementTypeTitle = 0x0109,
    e_ElementTypeFENote = 0x010A,
    e_ElementTypeNonStruct = 0x010F,
    e_ElementTypePrivate = 0x0110,
    e_ElementTypeParagraph = 0x0200,
    e_ElementTypeHeading = 0x0201,
    e_ElementTypeHeading1 = 0x0202,
    e_ElementTypeHeading2 = 0x0203,
    e_ElementTypeHeading3 = 0x0204,
    e_ElementTypeHeading4 = 0x0205,
    e_ElementTypeHeading5 = 0x0206,
    e_ElementTypeHeading6 = 0x0207,
    e_ElementTypeHeadingN = 0x0208,
    e_ElementTypeList = 0x0209,
    e_ElementTypeListItem = 0x020A,
    e_ElementTypeListLabel = 0x020B,
    e_ElementTypeListBody = 0x020C,
    e_ElementTypeTable = 0x020D,
    e_ElementTypeTableRow = 0x020E,
    e_ElementTypeTableHeaderCell = 0x020F,
    e_ElementTypeTableDataCell = 0x0210,
    e_ElementTypeTableHeaderGroup = 0x0211,
    e_ElementTypeTableBodyGroup = 0x0212,
    e_ElementTypeTableFootGroup = 0x0213,
    e_ElementTypeSpan = 0x0300,
    e_ElementTypeQuote = 0x0301,
    e_ElementTypeNote = 0x0302,
    e_ElementTypeReference = 0x0303,
    e_ElementTypeBibEntry = 0x0304,
    e_ElementTypeCode = 0x0305,
    e_ElementTypeLink = 0x0306,
    e_ElementTypeAnnot = 0x0307,
    e_ElementTypeRuby = 0x0308,
    e_ElementTypeRubyBase = 0x0309,
    e_ElementTypeRubyAnnot = 0x030A,
    e_ElementTypeRubyPunc = 0x030B,
    e_ElementTypeWarichu = 0x030C,
    e_ElementTypeWarichuText = 0x030D,
    e_ElementTypeWarichuPunc = 0x030E,
    e_ElementTypeEm = 0x030F,
    e_ElementTypeStrong = 0x0310,
    e_ElementTypeSub = 0x0311,
    e_ElementTypeFigure = 0x0400,
    e_ElementTypeFormula = 0x0401,
    e_ElementTypeForm = 0x0402,
    e_ElementTypeArtifact = 0x1000,
    e_ElementTypeTextObject = 0x80000001,
    e_ElementTypePathObject = 0x80000002,
    e_ElementTypeImageObject = 0x80000003,
    e_ElementTypeShadingObject = 0x80000004,
    e_ElementTypeFormObject = 0x80000005,
    e_ElementTypeContainerObject = 0x8000000C,
    e_ElementTypeAnnotObject = 0x8000000E,
    e_ElementTypePageObject = 0x8000000F,
    e_ElementTypeTextContent = 0xC0000001,
    e_ElementTypePathContent = 0xC0000002,
    e_ElementTypeImageContent = 0xC0000003,
    e_ElementTypeShadingContent = 0xC0000004,
    e_ElementTypeComplexContent = 0xC000000D,
    e_ElementTypeAnnotContent = 0xC000000E
  } ElementType;
  

  LRElement() {}

  LRElement(const LRElement& other);

  // User is strongly recommended NOT to use this method; otherwise unknown situation may occur.
  explicit LRElement(FS_HANDLE handle);

  ~LRElement();

  LRElement& operator = (const LRElement& other);

  bool operator == (const LRElement& other) const;

  bool operator != (const LRElement& other) const;

  bool IsEmpty() const;

  ElementType GetElementType();

  bool IsContentElement() const;

  bool IsGraphicsObjectElement() const;

  bool IsStructureElement() const;

  String StringifyType() const;

  static String StringifyElementType(ElementType type);
};

class LRStructureElement FS_FINAL : public LRElement {
 public:
  typedef enum _AttributeType {
    e_AttributeTypeInvalid = 0x0,
    e_AttributeTypeRole = 0x524f4c45,
    e_AttributeTypePlacement = 0x504c4143,
    e_AttributeTypeRotation = 0x524f5441,
    e_AttributeTypeWritingMode = 0x574d4f44,
    e_AttributeTypeDestination = 0x44455354,
    e_AttributeTypeBBox = 0x42424f58,
    e_AttributeTypeWidth = 0x57445448,
    e_AttributeTypeHeight = 0x48474854,
    e_AttributeTypeBackgroundColor = 0x424b4743,
    e_AttributeTypeBorderColor = 0x42445243,
    e_AttributeTypeBorderStyle = 0x42445253,
    e_AttributeTypeBorderThickness = 0x42445254,
    e_AttributeTypeAttached = 0x41545448,
    e_AttributeTypeSubType = 0x53554254,
    e_AttributeTypePadding = 0x50414447,
    e_AttributeTypeColor = 0x434f4c52,
    e_AttributeTypeSpaceBefore = 0x53504246,
    e_AttributeTypeSpaceAfter = 0x53504146,
    e_AttributeTypeStartIndent = 0x53494e44,
    e_AttributeTypeEndIndent = 0x45494e44,
    e_AttributeTypeTextIndent = 0x54494e44,
    e_AttributeTypeTextAlign = 0x54414c4e,
    e_AttributeTypeBlockAlign = 0x42414c4e,
    e_AttributeTypeInlineAlign = 0x49414c4e,
    e_AttributeTypeTBorderStyle = 0x54424453,
    e_AttributeTypeTPadding = 0x54504144,
    e_AttributeTypeBaselineShift = 0x42534846,
    e_AttributeTypeLineHeight = 0x4c484754,
    e_AttributeTypeTextDecorationColor = 0x5444434c,
    e_AttributeTypeTextDecorationThickness = 0x54445448,
    e_AttributeTypeTextDecorationType = 0x54445459,
    e_AttributeTypeRubyAlign = 0x5242414c,
    e_AttributeTypeRubyPosition = 0x52425053,
    e_AttributeTypeGlyphOrientationVertical = 0x474f5654,
    e_AttributeTypeColumnCount = 0x434c434e,
    e_AttributeTypeColumnGap = 0x434c4750,
    e_AttributeTypeColumnWidths = 0x434c5744,
    e_AttributeTypeListNumbering = 0x4c4e554d,
    e_AttributeTypeFieldRole = 0x46524f4c,
    e_AttributeTypeFieldChecked = 0x43484b44,
    e_AttributeTypeFieldDesc = 0x46445343,
    e_AttributeTypeRowSpan = 0x5253504e,
    e_AttributeTypeColSpan = 0x4353504e,
    e_AttributeTypeTableHeaders = 0x54484453,
    e_AttributeTypeTableHeaderScope = 0x54485343,
    e_AttributeTypeTableSummary = 0x5442534d,
    e_AttributeTypeShort        = 0x53484f52,
    e_AttributeTypeDirection = 0x44495245,
    e_AttributeTypeABox = 0x414c4258,
    e_AttributeTypeVersion = 0x5645534e
  } AttributeType;
  
  typedef enum _AttributeValueEnum {
    e_AttributeValueEnumInvalid = 0x0,
    e_AttributeValueEnumBlock = 0x424c434b,
    e_AttributeValueEnumInline = 0x494e4c4e,
    e_AttributeValueEnumWarichu = 0x57415249,
    e_AttributeValueEnumBefore = 0x42454652,
    e_AttributeValueEnumAfter = 0x41465445,
    e_AttributeValueEnumStart = 0x53545254,
    e_AttributeValueEnumEnd = 0x454e4400,
    e_AttributeValueEnumLrTb = 0x4c525442,
    e_AttributeValueEnumRlTb = 0x524c5442,
    e_AttributeValueEnumTbRl = 0x5442524c,
    e_AttributeValueEnumTbLr = 0x54424c52,
    e_AttributeValueEnumNone = 0x4e4f4e45,
    e_AttributeValueEnumHidden = 0x4849444e,
    e_AttributeValueEnumDotted = 0x444f5444,
    e_AttributeValueEnumDashed = 0x44534844,
    e_AttributeValueEnumSolid = 0x534f4c44,
    e_AttributeValueEnumDouble = 0x44424c45,
    e_AttributeValueEnumGroove = 0x41525459,
    e_AttributeValueEnumRidge = 0x52444745,
    e_AttributeValueEnumInset = 0x494e5354,
    e_AttributeValueEnumOutset = 0x4f545354,
    e_AttributeValueEnumNormal = 0x4e524d4c,
    e_AttributeValueEnumAuto = 0x4155544f,
    e_AttributeValueEnumCenter = 0x434e5452,
    e_AttributeValueEnumJustify = 0x4a535459,
    e_AttributeValueEnumMiddle = 0x4d49444c,
    e_AttributeValueEnumUnderline = 0x55444c4e,
    e_AttributeValueEnumOverline = 0x4f564c4e,
    e_AttributeValueEnumLineThrough = 0x4c4e5448,
    e_AttributeValueEnumDistribute = 0x44535442,
    e_AttributeValueEnumDisc = 0x44495343,
    e_AttributeValueEnumCircle = 0x4352434c,
    e_AttributeValueEnumSquare = 0x53515552,
    e_AttributeValueEnumDecimal = 0x44434d4c,
    e_AttributeValueEnumUpperRoman = 0x5550524d,
    e_AttributeValueEnumLowerRoman = 0x4c57524d,
    e_AttributeValueEnumUpperAlpha = 0x5550414c,
    e_AttributeValueEnumLowerAlpha = 0x4c57414c,
    e_AttributeValueEnumUnordered = 0x554e4f52,
    e_AttributeValueEnumOrdered = 0x4f524445,
    e_AttributeValueEnumFloat = 0x464c4f54,
    e_AttributeValueEnumRB = 0x52420000,
    e_AttributeValueEnumCB = 0x43420000,
    e_AttributeValueEnumPB = 0x50440000,
    e_AttributeValueEnumTV = 0x54560000,
    e_AttributeValueEnumOn = 0x4f4e0000,
    e_AttributeValueEnumOff = 0x4f464600,
    e_AttributeValueEnumNeutral = 0x4e555452,
    e_AttributeValueEnumRow = 0x524f5700,
    e_AttributeValueEnumColumn = 0x434c4d4e,
    e_AttributeValueEnumBoth = 0x424f5448,
    e_AttributeValueEnumLeft = 0x4c454654,
    e_AttributeValueEnumTop = 0x544f5000,
    e_AttributeValueEnumBottom = 0x42544f4d,
    e_AttributeValueEnumRight = 0x52474854,
    e_AttributeValueEnumPagination = 0x5041474e,
    e_AttributeValueEnumLayout = 0x4c41594f,
    e_AttributeValueEnumPage = 0x50414745,
    e_AttributeValueEnumBackground = 0x424b474e,
    e_AttributeValueEnumHeader = 0x48454144,
    e_AttributeValueEnumFooter = 0x464f4f54,
    e_AttributeValueEnumWatermark = 0x57544d4b,
    e_AttributeValueEnumPageNum = 0x504e554d,
    e_AttributeValueEnumBates = 0x42415445,
    e_AttributeValueEnumOther = 0x4f544845
  } AttributeValueEnum;
  
  typedef enum _AttributeValueType {
    e_AttributeValueTypeEmpty = 0x0,
    e_AttributeValueTypeEnum = 0x1,
    e_AttributeValueTypeInt32 = 0x2,
    e_AttributeValueTypeFloat = 0x3,
    e_AttributeValueTypeARGB = 0x4,
    e_AttributeValueTypeWStr = 0x5,
    e_AttributeValueTypeElement = 0x6,
    e_AttributeValueTypeEnumArray = 0x101,
    e_AttributeValueTypeInt32Array = 0x102,
    e_AttributeValueTypeFloatArray = 0x103,
    e_AttributeValueTypeARGBArray = 0x104,
    e_AttributeValueTypeWStrArray = 0x105,
    e_AttributeValueTypeElementArray = 0x106
  } AttributeValueType;
  

  LRStructureElement() {}

  LRStructureElement(const LRStructureElement& other);

  LRStructureElement(const LRElement& element);

  // User is strongly recommended NOT to use this method; otherwise unknown situation may occur.
  explicit LRStructureElement(FS_HANDLE handle);

  ~LRStructureElement() {}

  static String StringifyAttributeType(AttributeType type);

  static AttributeType ParseAttributeType(String type_string);

  static String StringifyEnumVariant(AttributeValueEnum type);

  static bool IsArrayAttributeValueType(AttributeValueType type);

  int GetSupportedAttributeCount();

  AttributeType GetSupportedAttribute(int index);

  int GetAttributeValueCount(AttributeType type);

  AttributeValueType GetAttributeValueType(AttributeType type);

  AttributeValueEnum GetAttributeValueEnum(AttributeType type, int index);

  int32 GetAttributeValueInt32(AttributeType type, int index);

  float GetAttributeValueFloat(AttributeType type, int index);

  ARGB GetAttributeValueARGB(AttributeType type, int index);

  WString GetAttributeValueString(AttributeType type, int index);

  LRStructureElement GetAttributeValueElement(AttributeType type, int index);

  LRStructureElement GetParentElement() const;

  int GetChildCount();

  LRElement GetChild(int index);

  RectF GetBBox() const;
};

class LRGraphicsObjectElement FS_FINAL : public LRElement {
 public:
  LRGraphicsObjectElement() {}

  LRGraphicsObjectElement(const LRGraphicsObjectElement& other);

  LRGraphicsObjectElement(const LRElement& element);

  // User is strongly recommended NOT to use this method; otherwise unknown situation may occur.
  explicit LRGraphicsObjectElement(FS_HANDLE handle);

  ~LRGraphicsObjectElement() {}

  LRGraphicsObjectElement GetParentGraphicsObjectElement() const;

  pdf::graphics::GraphicsObject* GetGraphicsObject() const;

  pdf::objects::PDFDictionary* GetDict() const;
  
  int GetGraphicsObjectIndex() const;
  
  RectF GetBBox() const;

  Matrix GetMatrix() const;
};

class LRContentElement FS_FINAL : public LRElement {
 public:
  LRContentElement() {}

  LRContentElement(const LRContentElement& other);

  LRContentElement(const LRElement& element);

  // User is strongly recommended NOT to use this method; otherwise unknown situation may occur.
  explicit LRContentElement(FS_HANDLE handle);

  ~LRContentElement() {}

  LRStructureElement GetParentElement() const;

  LRGraphicsObjectElement GetGraphicsObjectElement() const;

  common::Range GetGraphicsObjectRange();

  RectF GetBBox() const;

  Matrix GetMatrix() const;
};

class LRContext FS_FINAL : public Base {
 public:
  LRContext(const pdf::PDFPage& page);

  LRContext(const LRContext& other);

  // User is strongly recommended NOT to use this method; otherwise unknown situation may occur.
  explicit LRContext(FS_HANDLE handle);

  ~LRContext();

  LRContext& operator=(const LRContext& other);

  bool operator==(const LRContext& other) const;

  bool operator!=(const LRContext& other) const;

  bool IsEmpty() const;
  
  common::Progressive StartParse(common::PauseCallback* pause = NULL);

  LRStructureElement GetRootElement();
};
}  // namespace layoutrecognition
}  // namespace addon
}  // namespace foxit
#endif  // FS_LAYOUTRECOGNITION_H_