source: cpp/frams/genetics/geno.cpp @ 972

// This file is a part of Framsticks SDK.  http://www.framsticks.com/
// Copyright (C) 1999-2020  Maciej Komosinski and Szymon Ulatowski.
// See LICENSE.txt for details.

#include "geno.h"
#include "genoconv.h"
#include <common/loggers/loggers.h>
#include <common/util-string.h>
#include <frams/model/model.h>

THREAD_LOCAL_DEF_PTR(Geno::Validators, geno_validators);
THREAD_LOCAL_DEF_PTR(GenoConvManager, geno_converters);

Geno::Validators* Geno::getValidators() { return tlsGetPtr(geno_validators); }
GenoConvManager* Geno::getConverters() { return tlsGetPtr(geno_converters); }

Geno::Validators* Geno::useValidators(Validators* val)
{
        return tlsSetPtr(geno_validators, val);
}
GenoConvManager* Geno::useConverters(GenoConvManager* gcm)
{
        return tlsSetPtr(geno_converters, gcm);
}

bool Geno::formatIsOneOf(const SString& format, const SString& format_list)
{
        if (strchr(format_list.c_str(), ',') == NULL)
                return format == format_list;
        else
        {
                SString item; int pos = 0;
                while (format_list.getNextToken(pos, item, ','))
                        if (item == format)
                                return true;
        }
        return false;
}

const SString Geno::INVALID_FORMAT = "invalid";
const SString Geno::UNKNOWN_FORMAT = "";
const SString Geno::F0_FORMAT_LIST = "0,0s";

bool Geno::isF0Format(const SString& format_list)
{
        if (strchr(format_list.c_str(), ',') == NULL)
                return formatIsOneOf(format_list, F0_FORMAT_LIST);
        SString item; int pos = 0;
        while (format_list.getNextToken(pos, item, ','))
                if (!formatIsOneOf(item, F0_FORMAT_LIST))
                        return false;
        return true;
}

void Geno::init(const SString& genstring, const SString& genformat, const SString& genname, const SString& comment)
{
        refcount = 1;
        owner = 0;
        f0gen = 0;
        isvalid = -1;
        name = genname;
        txt = comment;
        setGenesAndFormat(genstring, genformat);
}

static SString trimAndValidateFormat(const SString& input) //the new requirement for genotype format name: no whitespace
{
        SString format = trim(input);
        if (format.length() == 0 || strContainsOneOf(format.c_str(), " \r\n\t"))
                return Geno::INVALID_FORMAT;
        return format;
}

void Geno::setGenesAndFormat(const SString& genstring, const SString& in_genformat)
{
        mapinshift = 0;
        mapoutshift = 0;
        SString gencopy(genstring);
        SString genformat = in_genformat;
        if (genformat == UNKNOWN_FORMAT)
        { // unknown format
                genformat = "1";
                if (genstring.charAt(0) == '/')
                {
                        int end, error_end = -1;
                        switch (genstring.charAt(1))
                        {
                        case '/':
                                if ((end = genstring.indexOf('\n')) >= 0)
                                {
                                        genformat = trimAndValidateFormat(genstring.substr(2, end - 2));
                                        mapinshift = end + 1;
                                        gencopy = genstring.substr(end + 1);
                                        if ((end > 0) && (genstring[end - 1] == '\r')) end--;
                                        error_end = end;
                                }
                                else
                                {
                                        genformat = trimAndValidateFormat(genstring.substr(2));
                                        gencopy = "";
                                        mapinshift = genstring.length();
                                }
                                break;
                        case '*':
                                if ((end = genstring.indexOf("*/")) >= 0)
                                {
                                        genformat = trimAndValidateFormat(genstring.substr(2, end - 2));
                                        error_end = end + 2;
                                        gencopy = genstring.substr(end + 2);
                                        mapinshift = end + 2;
                                }
                                else
                                {
                                        genformat = trimAndValidateFormat(genstring.substr(2));
                                        gencopy = "";
                                        mapinshift = genstring.length();
                                }
                                break;
                        }
                        if (genformat == INVALID_FORMAT)
                        {
                                SString cut;
                                if (error_end < 0) error_end = genstring.length();
                                static const int MAX_ERROR = 20;
                                if (error_end > MAX_ERROR)
                                        cut = genstring.substr(0, MAX_ERROR) + "...";
                                else
                                        cut = genstring.substr(0, error_end);
                                int lf = cut.indexOf('\n');
                                if (lf >= 0) { if ((lf > 0) && (cut[lf - 1] == '\r')) lf--; cut = cut.substr(0, lf); }
                                sstringQuote(cut);
                                logPrintf("Geno", "init", LOG_ERROR, "Invalid genotype format declaration: '%s'%s", cut.c_str(), name.length() ? SString::sprintf(" in '%s'", name.c_str()).c_str() : "");
                        }

                }
        }
        gen = gencopy;
        multiline = (strchr(gen.c_str(), '\n') != 0);
        format = genformat;
        freeF0();
        isvalid = -1;
        // mapoutshift...?
}

void Geno::freeF0()
{
        if (f0gen) { delete f0gen; f0gen = 0; }
}

Geno::Geno(const char *genstring, const char* genformat, const char *genname, const char *comment)
{
        init(SString(genstring), SString(genformat), SString(genname), SString(comment));
}

Geno::Geno(const char *genstring, char genformat, const char *genname, const char *comment)
{
        SString genformat_string;
        if (genformat > 0)
                genformat_string = SString(&genformat, 1);
        init(genstring, genformat_string, genname, comment);
}

Geno::Geno(const SString& genstring, const SString& genformat, const SString& genname, const SString& comment)
{
        init(genstring, genformat, genname, comment);
}

Geno::Geno(const Geno& src)
        :gen(src.gen), name(src.name), format(src.format), txt(src.txt), isvalid(src.isvalid),
        f0gen(0), mapinshift(src.mapinshift), mapoutshift(src.mapinshift),
        multiline(src.multiline), owner(0)
{
        f0gen = src.f0gen ? new Geno(*src.f0gen) : 0; refcount = 1;
}

void Geno::operator=(const Geno& src)
{
        freeF0();
        gen = src.gen;
        name = src.name;
        format = src.format;
        txt = src.txt;
        isvalid = src.isvalid;
        mapinshift = src.mapinshift;
        mapoutshift = src.mapinshift;
        multiline = src.multiline;
        f0gen = src.f0gen ? new Geno(*src.f0gen) : 0;
        owner = 0;
}

Geno::Geno(const SString& src)
{
        init(src, UNKNOWN_FORMAT, SString::empty(), SString::empty());
}

void Geno::setGenesAssumingSameFormat(const SString& g)
{
        gen = g;
        isvalid = -1;
        freeF0();
}

void Geno::setString(const SString& g)
{
        freeF0();
        init(g, UNKNOWN_FORMAT, SString::empty(), SString::empty());
}

void Geno::setName(const SString& n)
{
        name = n;
}

void Geno::setComment(const SString& c)
{
        txt = c;
}

SString Geno::getGenesAndFormat(void) const
{
        SString out;
        if (format != "1")
        {
                if (multiline)
                        out += "//";
                else
                        out += "/*";
                out += format;
                if (multiline)
                        out += "\n";
                else
                        out += "*/";
        }
        out += gen;
        return out;
}

int Geno::mapGenToString(int genpos) const
{
        if (genpos > gen.length()) return -2;
        if (genpos < 0) return -1;
        return mapinshift + genpos;
}

int Geno::mapStringToGen(int stringpos) const
{
        stringpos -= mapinshift;
        if (stringpos > gen.length()) return -2;
        if (stringpos < 0) return -1;
        return stringpos;
}

SString Geno::getGenes(void) const { return gen; }
SString Geno::getName(void) const { return name; }
SString Geno::getFormat(void) const { return format; }
SString Geno::getComment(void) const { return txt; }

int ModelGenoValidator::testGenoValidity(Geno& g)
{
        if (Geno::formatIsOneOf(g.getFormat(), Geno::F0_FORMAT_LIST))
        {
                Model mod(g, Model::SHAPE_UNKNOWN);
                return mod.isValid();
        }
        else
        {
                bool converter_missing;
                Geno f0geno = g.getConverted(Geno::F0_FORMAT_LIST, NULL, false, &converter_missing);
                if (converter_missing)
                        return -1;//no result
                return f0geno.isValid();
        }
}

void Geno::validate()
{
        if (isvalid >= 0) return;
        if (gen.length() == 0) { isvalid = 0; return; }
        if (format == INVALID_FORMAT) { isvalid = 0; return; }
        Validators* vals = getValidators();
        if (vals != NULL)
        {
#ifdef WARN_VALIDATION_INCONSISTENCY
                vector<int> results;
                int first_result = -1;
                FOREACH(GenoValidator*, v, (*vals))
                {
                        int r = v->testGenoValidity(*this);
                        if (first_result < 0) first_result = r;
                        results.push_back(r);
                }
                int N = vals->size();
                for (int i = 1; i < N; i++)
                        if (results[i] != results[0])
                        {
                                SString txt = "Inconsistent validation results";
                                for (int i = 0; i < N; i++)
                                        txt += SString::sprintf(" %d", results[i]);
                                txt += " for genotype '";
                                txt += getGene();
                                txt += "'";
                                logPrintf("Geno", "validate", LOG_WARN, txt.c_str());
                                break;
                        }
                isvalid = first_result;
                if (isvalid >= 0)
                        return;
#else
                FOREACH(GenoValidator*, v, (*vals))
                        if ((isvalid = v->testGenoValidity(*this)) >= 0)
                                return;
#endif
        }
        isvalid = 0;
        logPrintf("Geno", "validate", LOG_WARN, "Wrong configuration? No genotype validators defined for genetic format 'f%s'.", format.c_str());
}

bool Geno::isValid(void)
{
        if (isvalid < 0)
        {
                LoggerToMemory err(LoggerBase::Enable | LoggerToMemory::StoreAllMessages, LOG_INFO);
                validate();
                err.disable();
                string msg = err.getCountSummary();
                if (msg.size() > 0)
                {
                        msg += ssprintf(" while checking validity of '%s'", getName().c_str());
                        msg += "\n";
                        msg += err.getMessages();
                        logMessage("Geno", "isValid", err.getErrorLevel(), msg.c_str());
                }
        }
        return isvalid > 0;
}

Geno Geno::getConverted(SString otherformat_list, MultiMap *m, bool using_checkpoints, bool *converter_missing)
{
        if (formatIsOneOf(getFormat(), otherformat_list)) { if (converter_missing) *converter_missing = false; return *this; }
#ifndef NO_GENOCONVMANAGER
        GenoConvManager *converters = getConverters();
        if (converters)
        {
                if ((isF0Format(otherformat_list)) && (!m) && (!using_checkpoints))
                {
                        if (!f0gen)
                                f0gen = new Geno(converters->convert(*this, otherformat_list, NULL, using_checkpoints, converter_missing));
                        else
                        {
                                if (converter_missing) *converter_missing = false;
                        }
                        return *f0gen;
                }
                else
                        return converters->convert(*this, otherformat_list, m, using_checkpoints, converter_missing);
        }
#endif
        if (converter_missing) *converter_missing = true;
        return (formatIsOneOf(getFormat(), otherformat_list)) ? *this : Geno("", "", "", "GenConvManager not available");
}

Geno::~Geno()
{
        if (f0gen) delete f0gen;
}