parser_unicoil_json.h

Go to the documentation of this file.

/*
    PARSER_UNICOIL_JSON.H
    ---------------------
    Copyright (c) 2021 Andrew Trotman
    Released under the 2-clause BSD license (See:https://en.wikipedia.org/wiki/BSD_licenses)
*/
#pragma once

#include <charconv>

#include "ascii.h"
#include "parser.h"
#include "unicode.h"

namespace JASS
    {
    /*
        CLASS PARSER_UNICOIL_JSON
        -------------------------
    */
    class parser_unicoil_json : public parser
        {
        public:
            /*
                PARSER_UNICOIL_JSON::PARSER_UNICOIL_JSON()
                ------------------------------------------
            */
            parser_unicoil_json() :
                parser()
                {
                /* Nothing */
                }

            /*
                PARSER_UNICOIL_JSON::~PARSER_UNICOIL_JSON()
                -------------------------------------------
            */
            virtual ~parser_unicoil_json()
                {
                /*
                    Nothing
                */
                }

            /*
                PARSER_UNICOIL_JSON::GET_NEXT_TOKEN()
                -------------------------------------
            */
            virtual const class parser::token &get_next_token(void)
                {
                /*
                    Check for EOF
                */
                if (current >= end_of_document)
                    return eof_token;

                /*
                    Set up the parser
                */
                uint8_t *buffer_pos = current_token.buffer;
                uint8_t *buffer_end = current_token.buffer + sizeof(current_token.buffer);

                /*
                    Find the term
                */
                while (current < end_of_document)
                    {
                    if (*current == '"')
                        break;
                    current++;
                    }

                current++;

                /*
                    Copy the term to the token buffer
                */
                while (current < end_of_document)
                    {
                    if (*current == '\\')
                        {
                        if (buffer_pos < buffer_end)
                            *buffer_pos++ = *current;
                        current++;              // escaped characters in JSON
                        }
                    else if (*current == '"')
                        break;
                    if (buffer_pos < buffer_end)
                        *buffer_pos++ = *current;
                    current++;
                    }

                /*
                    Find the term count
                */
                while (current < end_of_document)
                    {
                    if (ascii::isdigit(*current))
                        break;
                    current++;
                    }

                const uint8_t *count_start = current;

                while (current < end_of_document)
                    {
                    if (!ascii::isdigit(*current))
                        break;
                    current++;
                    }

                const uint8_t *count_finish = current;

                /*
                    Construct a token object
                */
                current_token.lexeme = slice((void *)current_token.buffer, (void *)buffer_pos);
                current_token.type = token::alpha;
                std::from_chars((char *)count_start, (char *)count_finish, current_token.count);

                if (current_token.count > index_postings_impact::largest_impact)
                    {
                    std::cout << current_token.lexeme;
                    std::cout << " " << current_token.count << "\n";
                    current_token.count = index_postings_impact::largest_impact;
                    }

                return current_token.count > 0 ? current_token : get_next_token();
                }

            /*
                PARSER_UNICOIL_JSON::UNITTEST()
                -------------------------------
            */
            static void unittest(void);
        };
    }