prt.h

Go to the documentation of this file.

#ifndef PRT_H
#define PRT_H

#include <iostream>
#include <fstream>
#include <string>
#include <map>
#include <vector>
#include "dbtime.h"
#include <set>
#include "util.h"
#include "refloc.h"
#include "book.h"
using namespace std;

// simple helper, break up into LINE CODE (2 letters) first
// Content into second
void separate(const string &ln, string &first, string &second);

#define KEY_LIMIT 25
//#define DEBUG

class ref
{
        public:
                ref() { if (field.empty()) init_field(); };
                ref(const ref &r);

                ref& operator=(const ref &r);

                friend ostream& operator<<(ostream &os, const ref &r);
                bool next(string &ln, istream &in);

                //string makeID() const; // create a unique ID for this object

                void writeRaw(ostream &os) const;

                const string* get(const string &code) const;

                string getKey() const { return key; }
                bool bad() { return key.empty(); }

                vector<string> process_volpp(string &s);

                string process_year(const string &s);
                string firstAuthor() const;

                static void init_field();
                //static int getPaperKey();
                static void dumpkeymap(const string &file);
                static void readkeymap(const string &file);
        private:
                void process();  

                void constructKey();

                map<string, string> record;
                
                /* Derived information 
                 * digest includes:
                        medline, pubmed, journal type,
                        Digested information
                */
                map<string,string> digest; // stores processed information

                vector<string> editor;  // all editors
                string key;

                static map<string,string> field;

                static string field_str[14];  // string version used to compose
                // the map version at start up.  

                static map<string, int> paperkey;
};

class prt
{
        public:
                prt(const string &infile);

                bool next();  // read the next object
                void clear(); // start a new object
                void processID(); // helper function to process the first line
                vector<string> getAC();
                string getSWID() { return swpid; }
                void writeAce(ostream &os1, ostream &os2);
                ostream& writeSeq(ostream &os, const int len=75);
                void writeMappedTag(ostream &os);

                static const int content_idx = 5;     // 5
                static const int ftdescript_idx = 34;  // 34

                static const int cc_dash;     // 74 dashes start end of trash

                // in pairs index {at,1_afterend}
                // feature_key, from, to, description (optional note)
                //static const int ftidx[8];
                static map<string, string> field;
                static string record_separator;

                static map<string, string> swtag2ace;
                static string swtag2ace_str[];  // variable sized

                static void init();  // initialize class-wise parameters

        private:
                void read_upto(const string &code);  // helper function to read into the record map
                void tossccJunk();
                //vector<string> splitFT(); // work on next_line

                ifstream input;           // stay the same

                map<string,string> record;
                string next_line;  // internal buffer
                string dates[3];  // overwrite

                string seq;
                vector<ref> papers;  // all the papers stored under this field
                vector< vector<string> > dbxref;  // database cross reference
                vector< vector<string> > feature;  // database cross reference

                // processed infomation, all initialized to default
                string swpid;  // should use this one as the id in ACE
                char type;   // p for protein, or d for DNA
                char swprt_status; // s for standard, p for un_curated
                int length;  // length in aa

                static set<string> orgset;  // should it be
                static set<string> papset;  // all papers processed
};

#endif