parser_fasta.h

Go to the documentation of this file.

/*
    PARSER_FASTA.H
    --------------
    Copyright (c) 2019 Andrew Trotman
    Released under the 2-clause BSD license (See:https://en.wikipedia.org/wiki/BSD_licenses)
*/
#pragma once

#include <stdint.h>

#include "slice.h"
#include "parser.h"
#include "document.h"

namespace JASS
    {
    /*
        CLASS PARSER_FASTA
        ------------------
    */
    class parser_fasta : public parser
        {
        private:
            enum parser_mode
                {
                TEXT,
                DNA
                };

        private:
            size_t kmer_length;                 
            parser_mode mode;                       
            uint8_t *end_of_fasta_document;     

        protected:
            /*
                PARSER_FASTA::GET_NEXT_TOKEN_DNA()
                ----------------------------------
            */
            const class parser::token &get_next_token_dna(void);

        public:
            /*
                PARSER_FASTA::PARSER_FASTA()
                ----------------------------
            */
            parser_fasta(size_t kmer_length) :
                kmer_length(kmer_length),
                mode(TEXT),
                end_of_fasta_document(nullptr)
                {
                /*
                    We can only return alpha tokens or eof tokens so set the token type here to alpha
                */
                current_token.type = current_token.alpha;
                current_token.count = 1;
                }
            
            /*
                PARSER_FASTA::~PARSER_FASTA()
                -----------------------------
            */
            virtual ~parser_fasta()
                {
                /*
                    Nothing
                */
                }

            /*
                PARSER_FASTA::SET_DOCUMENT()
                ----------------------------
            */
            virtual void set_document(const class document &document)
                {
                the_document = &document;
                current = (uint8_t *)document.contents.address();
                end_of_fasta_document = (uint8_t *)document.contents.address() + document.contents.size();

                /*
                    Start in TEXT mode and "pretend" that the document stops at the start of the DNA (the end of the first line)
                */
                mode = TEXT;
                end_of_document = static_cast<uint8_t *>(std::find(static_cast<uint8_t *>(document.contents.address()), end_of_fasta_document, '\n'));
                }

            /*
                PARSER_FASTA::GET_NEXT_TOKEN()
                ------------------------------
            */
            virtual const class parser::token &get_next_token(void)
                {
                if (mode == DNA)
                    return get_next_token_dna();
                else
                    {
                    const token &got = parser::get_next_token();
                    if (got.type == token::token_type::eof)
                        {
                        mode = DNA;
                        end_of_document = end_of_fasta_document;        // shift to the end of document being the end of the FASTA document not just the end of the text.
                        return get_next_token_dna();
                        }
                    return got;
                    }
                }

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