MarkdownBlockParser.hpp

浏览该文件的文档.

 
 
#pragma once
 
#include "AstGlobal.h"
#include "MarkdownInlineParser.hpp"
#include "MarkdownTableParser.hpp"
#include <vector>
#include <memory>
 
AST_NAMESPACE_BEGIN
 
class MarkdownSax;
 
class AST_UTIL_API MarkdownBlockParser
{
public:
    MarkdownBlockParser(MarkdownSax& sax);
    ~MarkdownBlockParser();
 
    void feed(StringView data);
 
    void finish();
 
    void reset();
 
private:
    // ---- 块级状态 ----
    enum class EState
    {
        eLineStart,          
        eHeadingHashes,      
        eHeadingContent,     
        eCodeFenceOpen,      
        eCodeBlockContent,   
        eBlockquoteContent,  
        eListMarker,         
        eListItemContent,    
        eParagraph,          
        eTable,              
    };
 
    // ---- 块栈帧 ----
    enum class EBlockType
    {
        Paragraph,
        Heading,
        CodeBlock,
        List,
        ListItem,
        Blockquote,
    };
    struct BlockFrame
    {
        EBlockType type;
        int  level;    
        bool ordered;  
    };
 
    // ---- 核心方法 ----
    void feedChar(char c);
 
    void flushInlineContent();
    void flushInlineContent(bool force);
 
    // ---- 行首分类 ----
    void classifyFirstChar(char c);
 
    // ---- 各状态字符处理 ----
    void handleHeadingHashes(char c);
    void handleHeadingChar(char c);
    void handleCodeFenceOpen(char c);
    void handleCodeBlockChar(char c);
    void handleBlockquoteChar(char c);
    void handleListMarker(char c);
    void handleParagraphChar(char c);
    void handleTableChar(char c);
 
    // ---- 块栈管理 ----
    void openBlock(EBlockType type, int level = 0, bool ordered = false);
    void closeTop();
    void closeToType(EBlockType type);
    void closeAllBlocks();
    bool hasBlock(EBlockType type) const;
 
    // ---- 段落辅助 ----
    void ensureParagraph();
    void endParagraph();
    void closeParagraph();       
    void ensureListBlock(bool ordered, int indent = 0); 
 
    // ---- 文档辅助 ----
    void ensureDocStarted();
 
    // ============================================================
    // 状态成员
    // ============================================================
    MarkdownSax& sax_;
    MarkdownInlineParser inlineSM_;
    std::unique_ptr<MarkdownTableParser> tableSM_;
 
    EState state_ = EState::eLineStart;
    std::vector<BlockFrame> blockStack_;
 
    // 分类缓冲（行首判定用，最多缓冲几个字符）
    std::string classBuf_;
    // 行内内容缓冲（分类确认后，累积待喂入 inlineSM 的内容）
    std::string contentBuf_;
    // 代码块行缓冲（仅代码块内逐行检测关界围栏）
    std::string codeLineBuf_;
 
    // ---- 标题状态 ----
    int headingLevel_ = 0;
 
    // ---- 代码围栏状态 ----
    char codeFenceChar_ = 0;
    int  codeFenceCount_ = 0;
    std::string codeFenceLang_;
 
    // ---- 列表状态 ----
    bool inList_           = false;
    bool listOrdered_      = false;
    int  currentLineIndent_ = 0;      
    bool listItemAfterNL_  = false;   
    bool lineFresh_         = true;   
 
    int currentListIndent() const;
    void closePendingListItem();
 
    // ---- 段落跨行追踪 ----
    bool paraAfterNL_ = false;   
    bool paraHadContent_ = false;
 
    // ---- 文档状态 ----
    bool docStarted_ = false;
};
 
 
 
AST_NAMESPACE_END