Merge pull request #8 from sboh1214/feature/section

Parse SectionArray

.swiftlint.yml

@@ -1,6 +1,7 @@
 excluded:
   - Carthage
   - Pods
+  - .build
 
 line_length: 100
 

README.md

@@ -8,6 +8,9 @@
 ![Lint](https://github.com/sboh1214/HwpKit/workflows/Lint/badge.svg)
 [![codecov](https://codecov.io/gh/sboh1214/HwpKit/branch/master/graph/badge.svg)](https://codecov.io/gh/sboh1214/HwpKit)
 
+[![Quality Gate Status](https://sonarcloud.io/api/project_badges/measure?project=sboh1214_HwpKit&metric=alert_status)](https://sonarcloud.io/dashboard?id=sboh1214_HwpKit)
+[![CodeFactor](https://www.codefactor.io/repository/github/sboh1214/hwpkit/badge)](https://www.codefactor.io/repository/github/sboh1214/hwpkit)
+
 Swift Package for Reading & Writing HWP File
 
 ## Install

Sources/HwpKit/Enums/HwpCharType.swift

@@ -0,0 +1,7 @@
+import Foundation
+
+public enum HwpCharType: String, Codable {
+    case char
+    case inline
+    case extended
+}

Sources/HwpKit/Enums/HwpStreamName.swift

@@ -1,6 +1,7 @@
 public enum HwpStreamName: String {
     case fileHeader = "FileHeader"
     case docInfo = "DocInfo"
+    case bodyText = "BodyText"
     case summary = "\005HwpSummaryInformation"
     case previewText = "PrvText"
     case previewImage = "PrvImage"

Sources/HwpKit/Enums/HwpTag.swift

@@ -1,67 +1,65 @@
-// swiftlint:disable identifier_name
-
 let BEGIN: UInt32 = 0x10
 
 /**
  ’문서 정보’의 데이터 레코드
  */
 final class HwpDocInfoTag {
-    static let DOCUMENT_PROPERTIES: UInt32 = BEGIN
-    static let ID_MAPPINGS: UInt32 = BEGIN + 1
-    static let BIN_DATA: UInt32 = BEGIN + 2
-    static let FACE_NAME: UInt32 = BEGIN + 3
-    static let BORDER_FILL: UInt32 = BEGIN + 4
-    static let CHAR_SHAPE: UInt32 = BEGIN + 5
-    static let TAB_DEF: UInt32 = BEGIN + 6
-    static let NUMBERING: UInt32 = BEGIN + 7
-    static let BULLET: UInt32 = BEGIN + 8
-    static let PARA_SHAPE: UInt32 = BEGIN + 9
-    static let STYLE: UInt32 = BEGIN + 10
-    static let DOC_DATA: UInt32 = BEGIN + 11
-    static let DISTRIBUTE_DOC_DATA: UInt32 = BEGIN + 12
-    static let RESERVED: UInt32 = BEGIN + 13
-    static let COMPATIBLE_DOCUMENT: UInt32 = BEGIN + 14
-    static let LAYOUT_COMPATIBILITY: UInt32 = BEGIN + 15
-    static let TRACKCHANGE: UInt32 = BEGIN + 16
-    static let MEMO_SHAPE: UInt32 = BEGIN + 76
-    static let FORBIDDEN_CHAR: UInt32 = BEGIN + 78
-    static let TRACK_CHANGE: UInt32 = BEGIN + 80
-    static let TRACK_CHANGE_AUTHOR: UInt32 = BEGIN + 81
+    static let documentProperties: UInt32 = BEGIN
+    static let idMappings: UInt32 = BEGIN + 1
+    static let binData: UInt32 = BEGIN + 2
+    static let faceName: UInt32 = BEGIN + 3
+    static let borderFill: UInt32 = BEGIN + 4
+    static let charShape: UInt32 = BEGIN + 5
+    static let tabDef: UInt32 = BEGIN + 6
+    static let numbering: UInt32 = BEGIN + 7
+    static let bullet: UInt32 = BEGIN + 8
+    static let paraShape: UInt32 = BEGIN + 9
+    static let style: UInt32 = BEGIN + 10
+    static let docData: UInt32 = BEGIN + 11
+    static let distributeDocData: UInt32 = BEGIN + 12
+    static let reserved: UInt32 = BEGIN + 13
+    static let compatibleDocument: UInt32 = BEGIN + 14
+    static let layoutCompatibility: UInt32 = BEGIN + 15
+    static let trackChange: UInt32 = BEGIN + 16
+    static let memoShape: UInt32 = BEGIN + 76
+    static let forbiddenChar: UInt32 = BEGIN + 78
+    static let trackChangeContent: UInt32 = BEGIN + 80
+    static let trackChangeAuthor: UInt32 = BEGIN + 81
 }
 
 /**
  ‘본문’의 데이터 레코드
  */
 final class HwpSectionTag {
-    static let PARA_HEADER: UInt32 = BEGIN + 50
-    static let PARA_TEXT: UInt32 = BEGIN + 51
-    static let PARA_CHAR_SHAPE: UInt32 = BEGIN + 52
-    static let PARA_LINE_SEG: UInt32 = BEGIN + 53
-    static let PARA_RANGE_TAG: UInt32 = BEGIN + 54
-    static let CTRL_HEADER: UInt32 = BEGIN + 55
-    static let LIST_HEADER: UInt32 = BEGIN + 56
-    static let PAGE_DEF: UInt32 = BEGIN + 57
-    static let FOOTNOTE_SHAPE: UInt32 = BEGIN + 58
-    static let PAGE_BORDER_FILL: UInt32 = BEGIN + 59
-    static let SHAPE_COMPONENT: UInt32 = BEGIN + 60
-    static let TABLE: UInt32 = BEGIN + 61
-    static let SHAPE_COMPONENT_LINE: UInt32 = BEGIN + 62
-    static let SHAPE_COMPONENT_RECTANGLE: UInt32 = BEGIN + 63
-    static let SHAPE_COMPONENT_ELLIPSE: UInt32 = BEGIN + 64
-    static let SHAPE_COMPONENT_ARC: UInt32 = BEGIN + 65
-    static let SHAPE_COMPONENT_POLYGON: UInt32 = BEGIN + 66
-    static let SHAPE_COMPONENT_CURVE: UInt32 = BEGIN + 67
-    static let SHAPE_COMPONENT_OLE: UInt32 = BEGIN + 68
-    static let SHAPE_COMPONENT_PICTURE: UInt32 = BEGIN + 69
-    static let SHAPE_COMPONENT_CONTAINER: UInt32 = BEGIN + 70
-    static let CTRL_DATA: UInt32 = BEGIN + 71
-    static let EQEDIT: UInt32 = BEGIN + 72
-    static let RESERVED: UInt32 = BEGIN + 73
-    static let SHAPE_COMPONENT_TEXTART: UInt32 = BEGIN + 74
-    static let FORM_OBJECT: UInt32 = BEGIN + 75
-    static let MEMO_SHAPE: UInt32 = BEGIN + 76
-    static let MEMO_LIST: UInt32 = BEGIN + 77
-    static let CHART_DATA: UInt32 = BEGIN + 79
-    static let VIDEO_DATA: UInt32 = BEGIN + 82
-    static let SHAPE_COMPONENT_UNKNOWN: UInt32 = BEGIN + 99
+    static let paraHeader: UInt32 = BEGIN + 50
+    static let paraText: UInt32 = BEGIN + 51
+    static let paraCharShape: UInt32 = BEGIN + 52
+    static let paraLineSeg: UInt32 = BEGIN + 53
+    static let paraRangeTag: UInt32 = BEGIN + 54
+    static let ctrlHeader: UInt32 = BEGIN + 55
+    static let listHeader: UInt32 = BEGIN + 56
+    static let pageDef: UInt32 = BEGIN + 57
+    static let footnoteShape: UInt32 = BEGIN + 58
+    static let pageBorderFill: UInt32 = BEGIN + 59
+    static let shapeComponent: UInt32 = BEGIN + 60
+    static let table: UInt32 = BEGIN + 61
+    static let shapeComponentLine: UInt32 = BEGIN + 62
+    static let shapeComponentRectangle: UInt32 = BEGIN + 63
+    static let shapeComponentEllipse: UInt32 = BEGIN + 64
+    static let shapeComponentArc: UInt32 = BEGIN + 65
+    static let shapeComponentPolygon: UInt32 = BEGIN + 66
+    static let shapeComponentCurve: UInt32 = BEGIN + 67
+    static let shapeComponentOle: UInt32 = BEGIN + 68
+    static let shapeComponentPicture: UInt32 = BEGIN + 69
+    static let shapeComponentContainer: UInt32 = BEGIN + 70
+    static let ctrlData: UInt32 = BEGIN + 71
+    static let eqEdit: UInt32 = BEGIN + 72
+    static let reserved: UInt32 = BEGIN + 73
+    static let shapeComponentTextart: UInt32 = BEGIN + 74
+    static let formObject: UInt32 = BEGIN + 75
+    static let memoShape: UInt32 = BEGIN + 76
+    static let memoList: UInt32 = BEGIN + 77
+    static let chartData: UInt32 = BEGIN + 79
+    static let videoData: UInt32 = BEGIN + 82
+    static let shapeComponentUnknown: UInt32 = BEGIN + 99
 }

Sources/HwpKit/HwpFile.swift

@@ -5,6 +5,7 @@ public class HwpFile {
     public let fileHeader: HwpFileHeader
     public let docInfo: HwpDocInfo
     public let previewText: HwpPreviewText
+    public let sectionArray: [HwpSection]
 
     public init(filePath: String) throws {
         let ole: OLEFile
@@ -16,11 +17,15 @@ public class HwpFile {
         let streams = Dictionary(uniqueKeysWithValues: ole.root.children.map { ($0.name, $0 ) })
         let reader = StreamReader(ole, streams)
 
-        fileHeader = try HwpFileHeader(reader.getDataFromStream(.fileHeader, false))
+        let fileHeader = try HwpFileHeader(reader.getDataFromStream(.fileHeader, false))
+        self.fileHeader = fileHeader
 
         let docInfoData = try reader.getDataFromStream(.docInfo, fileHeader.isCompressed)
         docInfo = try HwpDocInfo(docInfoData, fileHeader.version)
 
+        sectionArray = try reader.getDataFromStorage(.bodyText, fileHeader.isCompressed)
+            .map {try HwpSection($0, fileHeader.version)}
+
         guard let previewTextStream = streams[HwpStreamName.previewText.rawValue] else {
             throw HwpError.streamDoesNotExist(name: HwpStreamName.previewText)
         }

Sources/HwpKit/Models/HwpParaShape.swift → Sources/HwpKit/Model/DocInfo/HwpParaShape.swift

@@ -7,7 +7,7 @@ import Foundation
  */
 public struct HwpParaShape: HwpDataWithVersion {
     /**속성 1*/
-    public let property1:UInt32
+    public let property1: UInt32
     /**왼쪽 여백*/
     public let marginLeft: Int32
     /**오른쪽 여백*/
@@ -19,7 +19,7 @@ public struct HwpParaShape: HwpDataWithVersion {
     /**문단 간격 아래*/
     public let paragraphSpacingBottom: Int32
     /**줄 간격. 한글 2007 이하 버전(5.0.2.5 버전 미만)에서 사용.*/
-    public let lineSpacing:Int32
+    public let lineSpacing: Int32
     /**탭 정의 아이디(TabDef ID) 참조 값*/
     public let tabDefId: UInt16
     /**번호 문단 ID(Numbering ID) 또는 글머리표 문단 모양 ID(Bullet ID) 참조 값*/
@@ -35,12 +35,12 @@ public struct HwpParaShape: HwpDataWithVersion {
     /**문단 테두리 아래쪽 간격*/
     public let borderSpacingBottom: Int16
     /**속성 2(표 40 참조) (5.0.1.7 버전 이상)*/
-    public var property2: UInt32? = nil
+    public var property2: UInt32?
     /**속성 3(표 41 참조) (5.0.2.5 버전 이상)*/
-    public var property3: UInt32? = nil
+    public var property3: UInt32?
     /**줄 간격(5.0.2.5 버전 이상)*/
-    public var lineSpacing2: UInt32? = nil
-    
+    public var lineSpacing2: UInt32?
+
     init(_ data: Data, _ version: HwpVersion) throws {
         var reader = DataReader(data)
         defer {
@@ -60,10 +60,10 @@ public struct HwpParaShape: HwpDataWithVersion {
         borderSpacingRight = reader.read(Int16.self)
         borderSpacingTop = reader.read(Int16.self)
         borderSpacingBottom = reader.read(Int16.self)
-        if version >= HwpVersion(5,0,1,7) {
+        if version >= HwpVersion(5, 0, 1, 7) {
             property2 = reader.read(UInt32.self)
         }
-        if version >= HwpVersion(5,0,2,5) {
+        if version >= HwpVersion(5, 0, 2, 5) {
             property3 = reader.read(UInt32.self)
             lineSpacing2 = reader.read(UInt32.self)
         }

Sources/HwpKit/Model/Section/HwpChar.swift

@@ -0,0 +1,6 @@
+import Foundation
+
+public struct HwpChar: Codable {
+    public let type: HwpCharType
+    public let value: WCHAR
+}

Sources/HwpKit/Model/Section/HwpCtrlHeader.swift

@@ -0,0 +1,16 @@
+import Foundation
+
+/**
+ 컨트롤 헤더
+
+ 컨트롤 문자가 존재하면 컨트롤 문자로부터 존재하는 컨트롤 정보를 생성한다.
+ Tag ID : HWPTAG_CTRL_HEADER
+ */
+public struct HwpCtrlHeader: HwpData {
+    public let ctrlId: UInt32
+
+    init(_ data: Data) throws {
+        var reader = DataReader(data)
+        ctrlId = reader.read(UInt32.self)
+    }
+}

Sources/HwpKit/Model/Section/HwpListHeader.swift

@@ -0,0 +1,26 @@
+import Foundation
+
+/**
+ 문단 리스트 헤더
+
+ Tag ID : HWPTAG_LIST_HEADER
+ */
+public struct HwpListHeader: HwpData {
+    /**
+     문단 수
+
+     한글문서에선 Int16으로 되어있으나 대부분의 경우 Int32 으로 읽어야 문제가 없다
+     */
+    public let paragraphCount: Int32
+    public let property: UInt32
+
+    init(_ data: Data) throws {
+        var reader = DataReader(data)
+        defer {
+            precondition(reader.isEOF())
+        }
+
+        paragraphCount = reader.read(Int32.self)
+        property = reader.read(UInt32.self)
+    }
+}

Sources/HwpKit/Model/Section/HwpParaCharShape.swift

@@ -0,0 +1,29 @@
+import Foundation
+
+/**
+ 문단의 글자 모양
+
+ Tag ID : HWPTAG_PARA_CHAR_SHAPE
+ */
+public struct HwpParaCharShape: HwpData {
+    /**글자 모양이 바뀌는 시작 위치*/
+    public var startingIndex: [UInt32]
+    /**글자 모양 ID*/
+    public var shapeId: [UInt32]
+
+    init(_ data: Data) throws {
+        var reader = DataReader(data)
+        defer {
+            precondition(reader.isEOF())
+        }
+
+        var startingIndex = [UInt32]()
+        var shapeId = [UInt32]()
+        while !reader.isEOF() {
+            startingIndex.append(reader.read(UInt32.self))
+            shapeId.append(reader.read(UInt32.self))
+        }
+        self.startingIndex = startingIndex
+        self.shapeId = shapeId
+    }
+}

Sources/HwpKit/Model/Section/HwpParaHeader.swift

@@ -0,0 +1,60 @@
+import Foundation
+
+/**
+ 문단 헤더
+
+ Tag ID : HWPTAG_PARA_HEADER
+ 텍스트의 수가 1 이상이면 문자 수만큼 텍스트를 로드하고 그렇지 않을 경우 PARA_BREAK로 문단을 생성한다.
+ */
+public struct HwpParaHeader: HwpDataWithVersion {
+    /**if (nchars & 0x80000000) { nchars &= 0x7fffffff;}*/
+    public let isLastInList: Bool
+    /**text(=chars)*/
+    public let charCount: UInt32
+    /**
+     control mask
+
+     (UINT32)(1<<ctrlch) 조합
+     ctrlch는 HwpCtrlAPI.Hwp 2.1 CtrlCh 참고
+     */
+    public let controlMask: UInt32
+    /**문단 모양 아이디 참조값*/
+    public let paraShapeId: UInt16
+    /**문단 스타일 아이디 참조값*/
+    public let paraStyleId: UInt8
+    /**단 나누기 종류*/
+    public let columnType: UInt8
+    /**글자 모양 정보 수*/
+    public let charShapeInfoCount: UInt16
+    /**range tag 정보 수*/
+    public let rangeTagInfoCount: UInt16
+    /**각 줄에 대한 align에 대한 정보 수*/
+    public let alignInfoCount: UInt16
+    /**문단 Instance ID (unique ID)*/
+    public let paraId: UInt32
+    /**변경추적 병합 문단여부. (5.0.3.2 버전 이상)*/
+    public var isTraceChange: UInt16?
+
+    init(_ data: Data, _ version: HwpVersion) throws {
+        var reader = DataReader(data)
+        let value = reader.read(UInt32.self)
+        if (value & 0x80000000) == 0x80000000 {
+            isLastInList = true
+            charCount = (value & 0x7fffffff)
+        } else {
+            isLastInList = false
+            charCount = value
+        }
+        controlMask = reader.read(UInt32.self)
+        paraShapeId = reader.read(UInt16.self)
+        paraStyleId = reader.read(UInt8.self)
+        columnType = reader.read(UInt8.self)
+        charShapeInfoCount = reader.read(UInt16.self)
+        rangeTagInfoCount = reader.read(UInt16.self)
+        alignInfoCount = reader.read(UInt16.self)
+        paraId = reader.read(UInt32.self)
+        if version >= HwpVersion(5, 0, 3, 2) {
+            isTraceChange = reader.read(UInt16.self)
+        }
+    }
+}