xref: /aosp_15_r20/external/pdfium/xfa/fxfa/cxfa_textparser.h (revision 3ac0a46f773bac49fa9476ec2b1cf3f8da5ec3a4)
1*3ac0a46fSAndroid Build Coastguard Worker // Copyright 2017 The PDFium Authors
2*3ac0a46fSAndroid Build Coastguard Worker // Use of this source code is governed by a BSD-style license that can be
3*3ac0a46fSAndroid Build Coastguard Worker // found in the LICENSE file.
4*3ac0a46fSAndroid Build Coastguard Worker 
5*3ac0a46fSAndroid Build Coastguard Worker // Original code copyright 2014 Foxit Software Inc. http://www.foxitsoftware.com
6*3ac0a46fSAndroid Build Coastguard Worker 
7*3ac0a46fSAndroid Build Coastguard Worker #ifndef XFA_FXFA_CXFA_TEXTPARSER_H_
8*3ac0a46fSAndroid Build Coastguard Worker #define XFA_FXFA_CXFA_TEXTPARSER_H_
9*3ac0a46fSAndroid Build Coastguard Worker 
10*3ac0a46fSAndroid Build Coastguard Worker #include <stdint.h>
11*3ac0a46fSAndroid Build Coastguard Worker 
12*3ac0a46fSAndroid Build Coastguard Worker #include <map>
13*3ac0a46fSAndroid Build Coastguard Worker #include <memory>
14*3ac0a46fSAndroid Build Coastguard Worker #include <vector>
15*3ac0a46fSAndroid Build Coastguard Worker 
16*3ac0a46fSAndroid Build Coastguard Worker #include "core/fxcrt/css/cfx_css.h"
17*3ac0a46fSAndroid Build Coastguard Worker #include "core/fxcrt/css/cfx_csscomputedstyle.h"
18*3ac0a46fSAndroid Build Coastguard Worker #include "core/fxcrt/css/cfx_cssdeclaration.h"
19*3ac0a46fSAndroid Build Coastguard Worker #include "core/fxcrt/retain_ptr.h"
20*3ac0a46fSAndroid Build Coastguard Worker #include "core/fxcrt/widestring.h"
21*3ac0a46fSAndroid Build Coastguard Worker #include "core/fxge/dib/fx_dib.h"
22*3ac0a46fSAndroid Build Coastguard Worker #include "fxjs/gc/heap.h"
23*3ac0a46fSAndroid Build Coastguard Worker #include "third_party/abseil-cpp/absl/types/optional.h"
24*3ac0a46fSAndroid Build Coastguard Worker #include "v8/include/cppgc/garbage-collected.h"
25*3ac0a46fSAndroid Build Coastguard Worker #include "xfa/fxfa/fxfa_basic.h"
26*3ac0a46fSAndroid Build Coastguard Worker 
27*3ac0a46fSAndroid Build Coastguard Worker class CFGAS_GEFont;
28*3ac0a46fSAndroid Build Coastguard Worker class CFX_CSSStyleSelector;
29*3ac0a46fSAndroid Build Coastguard Worker class CFX_CSSStyleSheet;
30*3ac0a46fSAndroid Build Coastguard Worker class CFX_XMLNode;
31*3ac0a46fSAndroid Build Coastguard Worker class CXFA_FFDoc;
32*3ac0a46fSAndroid Build Coastguard Worker class CXFA_TextProvider;
33*3ac0a46fSAndroid Build Coastguard Worker class CXFA_TextTabstopsContext;
34*3ac0a46fSAndroid Build Coastguard Worker 
35*3ac0a46fSAndroid Build Coastguard Worker class CXFA_TextParser : public cppgc::GarbageCollected<CXFA_TextParser> {
36*3ac0a46fSAndroid Build Coastguard Worker  public:
37*3ac0a46fSAndroid Build Coastguard Worker   class Context {
38*3ac0a46fSAndroid Build Coastguard Worker    public:
39*3ac0a46fSAndroid Build Coastguard Worker     Context();
40*3ac0a46fSAndroid Build Coastguard Worker     ~Context();
41*3ac0a46fSAndroid Build Coastguard Worker 
42*3ac0a46fSAndroid Build Coastguard Worker     void SetParentStyle(RetainPtr<const CFX_CSSComputedStyle> style);
GetParentStyle()43*3ac0a46fSAndroid Build Coastguard Worker     RetainPtr<const CFX_CSSComputedStyle> GetParentStyle() const {
44*3ac0a46fSAndroid Build Coastguard Worker       return m_pParentStyle;
45*3ac0a46fSAndroid Build Coastguard Worker     }
46*3ac0a46fSAndroid Build Coastguard Worker 
SetDisplay(CFX_CSSDisplay eDisplay)47*3ac0a46fSAndroid Build Coastguard Worker     void SetDisplay(CFX_CSSDisplay eDisplay) { m_eDisplay = eDisplay; }
GetDisplay()48*3ac0a46fSAndroid Build Coastguard Worker     CFX_CSSDisplay GetDisplay() const { return m_eDisplay; }
49*3ac0a46fSAndroid Build Coastguard Worker 
50*3ac0a46fSAndroid Build Coastguard Worker     void SetDecls(std::vector<const CFX_CSSDeclaration*>&& decl);
GetDecls()51*3ac0a46fSAndroid Build Coastguard Worker     const std::vector<const CFX_CSSDeclaration*>& GetDecls() const {
52*3ac0a46fSAndroid Build Coastguard Worker       return decls_;
53*3ac0a46fSAndroid Build Coastguard Worker     }
54*3ac0a46fSAndroid Build Coastguard Worker 
55*3ac0a46fSAndroid Build Coastguard Worker    private:
56*3ac0a46fSAndroid Build Coastguard Worker     RetainPtr<const CFX_CSSComputedStyle> m_pParentStyle;
57*3ac0a46fSAndroid Build Coastguard Worker     CFX_CSSDisplay m_eDisplay = CFX_CSSDisplay::None;
58*3ac0a46fSAndroid Build Coastguard Worker     std::vector<const CFX_CSSDeclaration*> decls_;
59*3ac0a46fSAndroid Build Coastguard Worker   };
60*3ac0a46fSAndroid Build Coastguard Worker 
61*3ac0a46fSAndroid Build Coastguard Worker   CONSTRUCT_VIA_MAKE_GARBAGE_COLLECTED;
62*3ac0a46fSAndroid Build Coastguard Worker   virtual ~CXFA_TextParser();
63*3ac0a46fSAndroid Build Coastguard Worker 
Trace(cppgc::Visitor * visitor)64*3ac0a46fSAndroid Build Coastguard Worker   void Trace(cppgc::Visitor* visitor) const {}
65*3ac0a46fSAndroid Build Coastguard Worker 
66*3ac0a46fSAndroid Build Coastguard Worker   void Reset();
67*3ac0a46fSAndroid Build Coastguard Worker   void DoParse(const CFX_XMLNode* pXMLContainer,
68*3ac0a46fSAndroid Build Coastguard Worker                CXFA_TextProvider* pTextProvider);
69*3ac0a46fSAndroid Build Coastguard Worker 
70*3ac0a46fSAndroid Build Coastguard Worker   RetainPtr<CFX_CSSComputedStyle> CreateRootStyle(
71*3ac0a46fSAndroid Build Coastguard Worker       CXFA_TextProvider* pTextProvider);
72*3ac0a46fSAndroid Build Coastguard Worker   RetainPtr<CFX_CSSComputedStyle> ComputeStyle(
73*3ac0a46fSAndroid Build Coastguard Worker       const CFX_XMLNode* pXMLNode,
74*3ac0a46fSAndroid Build Coastguard Worker       RetainPtr<const CFX_CSSComputedStyle> pParentStyle);
75*3ac0a46fSAndroid Build Coastguard Worker 
IsParsed()76*3ac0a46fSAndroid Build Coastguard Worker   bool IsParsed() const { return m_bParsed; }
77*3ac0a46fSAndroid Build Coastguard Worker 
78*3ac0a46fSAndroid Build Coastguard Worker   XFA_AttributeValue GetVAlign(CXFA_TextProvider* pTextProvider) const;
79*3ac0a46fSAndroid Build Coastguard Worker 
80*3ac0a46fSAndroid Build Coastguard Worker   float GetTabInterval(const CFX_CSSComputedStyle* pStyle) const;
81*3ac0a46fSAndroid Build Coastguard Worker   int32_t CountTabs(const CFX_CSSComputedStyle* pStyle) const;
82*3ac0a46fSAndroid Build Coastguard Worker 
83*3ac0a46fSAndroid Build Coastguard Worker   bool IsSpaceRun(const CFX_CSSComputedStyle* pStyle) const;
84*3ac0a46fSAndroid Build Coastguard Worker   bool GetTabstops(const CFX_CSSComputedStyle* pStyle,
85*3ac0a46fSAndroid Build Coastguard Worker                    CXFA_TextTabstopsContext* pTabstopContext);
86*3ac0a46fSAndroid Build Coastguard Worker 
87*3ac0a46fSAndroid Build Coastguard Worker   RetainPtr<CFGAS_GEFont> GetFont(CXFA_FFDoc* doc,
88*3ac0a46fSAndroid Build Coastguard Worker                                   CXFA_TextProvider* pTextProvider,
89*3ac0a46fSAndroid Build Coastguard Worker                                   const CFX_CSSComputedStyle* pStyle) const;
90*3ac0a46fSAndroid Build Coastguard Worker   float GetFontSize(CXFA_TextProvider* pTextProvider,
91*3ac0a46fSAndroid Build Coastguard Worker                     const CFX_CSSComputedStyle* pStyle) const;
92*3ac0a46fSAndroid Build Coastguard Worker   int32_t GetHorScale(CXFA_TextProvider* pTextProvider,
93*3ac0a46fSAndroid Build Coastguard Worker                       const CFX_CSSComputedStyle* pStyle,
94*3ac0a46fSAndroid Build Coastguard Worker                       const CFX_XMLNode* pXMLNode) const;
95*3ac0a46fSAndroid Build Coastguard Worker   int32_t GetVerScale(CXFA_TextProvider* pTextProvider,
96*3ac0a46fSAndroid Build Coastguard Worker                       const CFX_CSSComputedStyle* pStyle) const;
97*3ac0a46fSAndroid Build Coastguard Worker   int32_t GetUnderline(CXFA_TextProvider* pTextProvider,
98*3ac0a46fSAndroid Build Coastguard Worker                        const CFX_CSSComputedStyle* pStyle) const;
99*3ac0a46fSAndroid Build Coastguard Worker   XFA_AttributeValue GetUnderlinePeriod(
100*3ac0a46fSAndroid Build Coastguard Worker       CXFA_TextProvider* pTextProvider,
101*3ac0a46fSAndroid Build Coastguard Worker       const CFX_CSSComputedStyle* pStyle) const;
102*3ac0a46fSAndroid Build Coastguard Worker   int32_t GetLinethrough(CXFA_TextProvider* pTextProvider,
103*3ac0a46fSAndroid Build Coastguard Worker                          const CFX_CSSComputedStyle* pStyle) const;
104*3ac0a46fSAndroid Build Coastguard Worker   FX_ARGB GetColor(CXFA_TextProvider* pTextProvider,
105*3ac0a46fSAndroid Build Coastguard Worker                    const CFX_CSSComputedStyle* pStyle) const;
106*3ac0a46fSAndroid Build Coastguard Worker   float GetBaseline(CXFA_TextProvider* pTextProvider,
107*3ac0a46fSAndroid Build Coastguard Worker                     const CFX_CSSComputedStyle* pStyle) const;
108*3ac0a46fSAndroid Build Coastguard Worker   float GetLineHeight(CXFA_TextProvider* pTextProvider,
109*3ac0a46fSAndroid Build Coastguard Worker                       const CFX_CSSComputedStyle* pStyle,
110*3ac0a46fSAndroid Build Coastguard Worker                       bool bFirst,
111*3ac0a46fSAndroid Build Coastguard Worker                       float fVerScale) const;
112*3ac0a46fSAndroid Build Coastguard Worker 
113*3ac0a46fSAndroid Build Coastguard Worker   absl::optional<WideString> GetEmbeddedObj(
114*3ac0a46fSAndroid Build Coastguard Worker       const CXFA_TextProvider* pTextProvider,
115*3ac0a46fSAndroid Build Coastguard Worker       const CFX_XMLNode* pXMLNode);
116*3ac0a46fSAndroid Build Coastguard Worker   Context* GetParseContextFromMap(const CFX_XMLNode* pXMLNode);
117*3ac0a46fSAndroid Build Coastguard Worker 
118*3ac0a46fSAndroid Build Coastguard Worker  protected:
119*3ac0a46fSAndroid Build Coastguard Worker   CXFA_TextParser();
120*3ac0a46fSAndroid Build Coastguard Worker 
121*3ac0a46fSAndroid Build Coastguard Worker   bool TagValidate(const WideString& str) const;
122*3ac0a46fSAndroid Build Coastguard Worker 
123*3ac0a46fSAndroid Build Coastguard Worker  private:
124*3ac0a46fSAndroid Build Coastguard Worker   class TagProvider {
125*3ac0a46fSAndroid Build Coastguard Worker    public:
126*3ac0a46fSAndroid Build Coastguard Worker     TagProvider();
127*3ac0a46fSAndroid Build Coastguard Worker     ~TagProvider();
128*3ac0a46fSAndroid Build Coastguard Worker 
GetTagName()129*3ac0a46fSAndroid Build Coastguard Worker     WideString GetTagName() { return m_wsTagName; }
130*3ac0a46fSAndroid Build Coastguard Worker 
SetTagName(const WideString & wsName)131*3ac0a46fSAndroid Build Coastguard Worker     void SetTagName(const WideString& wsName) { m_wsTagName = wsName; }
SetAttribute(const WideString & wsAttr,const WideString & wsValue)132*3ac0a46fSAndroid Build Coastguard Worker     void SetAttribute(const WideString& wsAttr, const WideString& wsValue) {
133*3ac0a46fSAndroid Build Coastguard Worker       m_Attributes.insert({wsAttr, wsValue});
134*3ac0a46fSAndroid Build Coastguard Worker     }
135*3ac0a46fSAndroid Build Coastguard Worker 
GetAttribute(const WideString & wsAttr)136*3ac0a46fSAndroid Build Coastguard Worker     WideString GetAttribute(const WideString& wsAttr) {
137*3ac0a46fSAndroid Build Coastguard Worker       return m_Attributes[wsAttr];
138*3ac0a46fSAndroid Build Coastguard Worker     }
139*3ac0a46fSAndroid Build Coastguard Worker 
140*3ac0a46fSAndroid Build Coastguard Worker     bool m_bTagAvailable = false;
141*3ac0a46fSAndroid Build Coastguard Worker     bool m_bContent = false;
142*3ac0a46fSAndroid Build Coastguard Worker 
143*3ac0a46fSAndroid Build Coastguard Worker    private:
144*3ac0a46fSAndroid Build Coastguard Worker     WideString m_wsTagName;
145*3ac0a46fSAndroid Build Coastguard Worker     std::map<WideString, WideString> m_Attributes;
146*3ac0a46fSAndroid Build Coastguard Worker   };
147*3ac0a46fSAndroid Build Coastguard Worker 
148*3ac0a46fSAndroid Build Coastguard Worker   // static
149*3ac0a46fSAndroid Build Coastguard Worker   std::unique_ptr<TagProvider> ParseTagInfo(const CFX_XMLNode* pXMLNode);
150*3ac0a46fSAndroid Build Coastguard Worker 
151*3ac0a46fSAndroid Build Coastguard Worker   void InitCSSData(CXFA_TextProvider* pTextProvider);
152*3ac0a46fSAndroid Build Coastguard Worker   void ParseRichText(const CFX_XMLNode* pXMLNode,
153*3ac0a46fSAndroid Build Coastguard Worker                      const CFX_CSSComputedStyle* pParentStyle);
154*3ac0a46fSAndroid Build Coastguard Worker   std::unique_ptr<CFX_CSSStyleSheet> LoadDefaultSheetStyle();
155*3ac0a46fSAndroid Build Coastguard Worker   RetainPtr<CFX_CSSComputedStyle> CreateStyle(
156*3ac0a46fSAndroid Build Coastguard Worker       const CFX_CSSComputedStyle* pParentStyle);
157*3ac0a46fSAndroid Build Coastguard Worker 
158*3ac0a46fSAndroid Build Coastguard Worker   bool m_bParsed = false;
159*3ac0a46fSAndroid Build Coastguard Worker   bool m_cssInitialized = false;
160*3ac0a46fSAndroid Build Coastguard Worker   std::unique_ptr<CFX_CSSStyleSelector> m_pSelector;
161*3ac0a46fSAndroid Build Coastguard Worker   std::map<const CFX_XMLNode*, std::unique_ptr<Context>>
162*3ac0a46fSAndroid Build Coastguard Worker       m_mapXMLNodeToParseContext;
163*3ac0a46fSAndroid Build Coastguard Worker };
164*3ac0a46fSAndroid Build Coastguard Worker 
165*3ac0a46fSAndroid Build Coastguard Worker #endif  // XFA_FXFA_CXFA_TEXTPARSER_H_
166