HttpMultipartParser.h

// Copyright (c) Stefano Cristiano
// SPDX-License-Identifier: MIT
#pragma once
#include "../Foundation/Result.h"
#include "../Foundation/Span.h"
#include "../Foundation/StringSpan.h"
#include "HttpExport.h"
 
namespace SC
{
 
SC_HTTP_EXPORT bool HttpMultipartIsSafeFileName(StringSpan fileName);
 
struct SC_HTTP_EXPORT HttpMultipartContentDispositionView
{
    StringSpan disposition;
    StringSpan name;
    StringSpan fileName;
 
    bool hasName     = false;
    bool hasFileName = false;
 
    Result             parse(StringSpan headerValue);
    [[nodiscard]] bool isFormData() const;
};
 
struct SC_HTTP_EXPORT HttpMultipartPartHeadersView
{
    StringSpan contentDisposition;
    StringSpan contentType;
 
    HttpMultipartContentDispositionView disposition;
 
    void   reset();
    Result addHeader(StringSpan name, StringSpan value);
 
    [[nodiscard]] StringSpan fieldName() const { return disposition.name; }
    [[nodiscard]] StringSpan fileName() const { return disposition.fileName; }
    [[nodiscard]] bool       hasFileName() const { return disposition.hasFileName; }
    [[nodiscard]] bool       hasSafeFileName() const;
};
 
struct SC_HTTP_EXPORT HttpMultipartParser
{
    Result initWithBoundary(StringSpan boundary);
 
    void reset();
 
    enum class State
    {
        Parsing,  
        Result,   
        Finished, 
    };
 
    enum class Token
    {
        HeaderName,    
        HeaderValue,   
        PartBody,      
        Boundary,      
        PartHeaderEnd, 
        Finished,      
    };
 
    Token token = Token::Boundary; 
    State state = State::Parsing;  
 
    Result parse(Span<const char> data, size_t& readBytes, Span<const char>& parsedData);
 
  private:
    char boundaryStorage[71] = {0}; // MAX = 70 + NUL
    char boundaryBuffer[128] = {0};
 
    size_t tokenStart   = 0;
    size_t tokenLength  = 0;
    size_t globalStart  = 0;
    size_t globalLength = 0;
    size_t matchIndex   = 0; // For matching boundary and other fixed strings
 
    int topLevelCoroutine     = 0;
    int nestedParserCoroutine = 0;
 
    uint8_t boundaryMatchIndex      = 0;
    size_t  boundaryCandidateLength = 0;
    bool    emitBoundaryCandidate   = false;
 
    StringSpan boundary;
 
    // Helper functions for parsing
    [[nodiscard]] bool parseBoundary(char currentChar);
    [[nodiscard]] bool parseHeaders(char currentChar);
    [[nodiscard]] bool parseBody(char currentChar);
 
    // Specific parsers
    [[nodiscard]] bool parsePreamble(char currentChar);
    [[nodiscard]] bool parseBoundaryLine(char currentChar);
    [[nodiscard]] bool parseHeaderName(char currentChar);
    [[nodiscard]] bool parseHeaderValue(char currentChar);
    [[nodiscard]] bool parseBodyUntilBoundary(char currentChar);
 
    template <bool (HttpMultipartParser::*Func)(char), Token currentResult>
    Result process(Span<const char>& data, size_t& readBytes, Span<const char>& parsedData);
};
 
} // namespace SC