Changeset 797 for cpp/frams/genetics/fB

Timestamp:

06/06/18 01:45:18 (7 years ago)

Author:

Maciej Komosinski

Message:

A more complete implementation of fB, fH, fL

Location:

cpp/frams/genetics/fB

Files:

• fB_conv.cpp (modified) (6 diffs)
• fB_conv.h (modified) (2 diffs)
• fB_general.h (modified) (4 diffs)
• fB_oper.cpp (modified) (8 diffs)
• fB_oper.h (modified) (3 diffs)

cpp/frams/genetics/fB/fB_conv.cpp

@@ +1 @@
+// This file is a part of Framsticks SDK.  http://www.framsticks.com/
+// Copyright (C) 1999-2018  Maciej Komosinski and Szymon Ulatowski.
+// See LICENSE.txt for details.
+
 #include "fB_conv.h"
 
@@ -89 +93 @@
                         continue;
                 }
-                fH_Handle *handle = convertCharacterToHandle(gene[2], dims, start, end, ranges);
+                int hclasspos = 2;
+                if (gene[2] == '"')
+                {
+                        hclasspos--;
+                        if (!getNextCharId(gene, hclasspos))
+                        {
+                                return "";
+                        }
+                }
+                fH_Handle *handle = convertCharacterToHandle(gene[hclasspos], dims, start, end, ranges);
                 ParamEntry *tab = creature.getParamTab(handle->type);
                 void *obj = ParamObject::makeObject(tab);
@@ -97 +110 @@
 
                 int propindex = 0;
-                int z = 3;
+                int z = hclasspos;
+                if (gene[z] == '"')
+                {
+                        z--;
+                        if (!getNextCharId(gene, z))
+                        {
+                                delete handle;
+                                ParamObject::freeObject(obj);
+                                return "";
+                        }
+                }
                 endoffset = 0;
                 if (gene.indexOf("zz", 0) != -1) endoffset = 2;
+                int nclassdefcount = 1;
                 while (z < gene.len() - endoffset)
                 {
-                        if (processNextLetter(creature, handle, par, gene, propindex, z, ranges) == -1)
-                        {
-                                logMessage("GenoConv_fBH", "convert", LOG_WARN, "Property of fH could not be parsed");
-                        }
+                        if (processNextLetter(creature, handle, par, gene, propindex, z, ranges, nclassdefcount) == -1)
+                        {
+                                logMessage("GenoConv_fBH", "convert", LOG_ERROR, "Property of fH could not be parsed");
+                                delete handle;
+                                ParamObject::freeObject(obj);
+                                return "";
+                        }
+                }
+                if (handle->type == fHBodyType::NEURON && propindex < par.getPropCount())
+                {
+                        SString nclass;
+                        if (!getNeuroClass(gene, nclass, nclassdefcount))
+                        {
+                                delete handle;
+                                ParamObject::freeObject(obj);
+                                return "";
+                        }
+                        par.setStringById(FH_PE_NEURO_DET, nclass);
                 }
                 handle->loadProperties(par);
@@ -132 +170 @@
 }
 
-int GenoConv_fBH::processNextLetter(fH_Builder &creature, fH_Handle *&currhandle, Param &par, SString gene, int &propindex, int &i, std::vector<IRange> ranges[3])
+bool GenoConv_fBH::getNextCharId(SString genotype, int &i)
+{
+        i++;
+        if (genotype[i] == '"')
+        {
+                int nextid = i + 1;
+                do
+                {
+                        nextid = genotype.indexOf('"', nextid);
+                        if (nextid == -1)
+                        {
+                                return false;
+                        }
+                        nextid++;
+                }
+                while (genotype[nextid] == '"');
+                i = nextid;
+        }
+        return true;
+}
+
+bool GenoConv_fBH::getNeuroClass(SString gene, SString &def, int nclassdefcount)
+{
+        SString lastdef = "N";
+        int nclass = 0;
+        int pos = 0;
+        while (nclass < nclassdefcount)
+        {
+                pos = gene.indexOf('\"', pos);
+                if (pos == -1)
+                {
+                        def = lastdef;
+                        return true;
+                }
+                pos++;
+                SString currdef;
+                if (gene.indexOf('\"', pos) == -1 || !gene.getNextToken(pos, currdef, '\"'))
+                {
+                        def = lastdef;
+                        return false;
+                }
+                lastdef = currdef;
+                nclass++;
+        }
+        def = lastdef;
+        return true;
+}
+
+int GenoConv_fBH::processNextLetter(fH_Builder &creature, fH_Handle *&currhandle, Param &par, SString gene, int &propindex, int &i, std::vector<IRange> ranges[3], int &nclassdefcount)
 {
         if (propindex >= par.getPropCount())
@@ -147 +233 @@
                 par.setDefault();
                 propindex = 0;
-                i++;
+                if (!getNextCharId(gene, i))
+                        return -1;
                 return 0;
         }
@@ -166 +253 @@
                         }
                         propindex++;
-                        i++;
+                        if (!getNextCharId(gene, i))
+                                return -1;
                         return 0;
                 }
                 else if (currhandle->type == fHBodyType::NEURON && *par.id(propindex) == 'd')
                 {
-                        //TODO: handle neuron classes and properties
+                        //When 'd' property appears for i-th element in gene, the method
+                        //looks for i-th neuron definition
+                        SString nclass;
+                        if (!getNeuroClass(gene, nclass, nclassdefcount)) return -1;
+                        par.setString(propindex, nclass);
                         propindex++;
-                        i++;
+                        nclassdefcount++;
+                        if (!getNextCharId(gene, i))
+                                return -1;
                         return 0;
                 }

cpp/frams/genetics/fB/fB_conv.h

@@ +1 @@
+// This file is a part of Framsticks SDK.  http://www.framsticks.com/
+// Copyright (C) 1999-2018  Maciej Komosinski and Szymon Ulatowski.
+// See LICENSE.txt for details.
+
 #ifndef _FB_CONV_H_
 #define _FB_CONV_H_
@@ -8 +12 @@
 {
 private:
+        bool getNextCharId(SString genotype, int &i);
         double convertCharacterTo01(char c);
         double convertCharacterToWeight(char c);
         static fH_Handle* convertCharacterToHandle(char c, int dims, int start, int end, std::vector<IRange> ranges[3]);
-        int processNextLetter(fH_Builder &creature, fH_Handle *&currhandle, Param &par, SString gene, int &propindex, int &i, std::vector<IRange> ranges[3]);
+        int processNextLetter(fH_Builder &creature, fH_Handle *&currhandle, Param &par, SString gene, int &propindex, int &i, std::vector<IRange> ranges[3], int &nclassdefcount);
+        bool getNeuroClass(SString gene, SString &def, int nclassdefcount);
 
 public:

cpp/frams/genetics/fB/fB_general.h

@@ +1 @@
+// This file is a part of Framsticks SDK.  http://www.framsticks.com/
+// Copyright (C) 1999-2018  Maciej Komosinski and Szymon Ulatowski.
+// See LICENSE.txt for details.
+
 #ifndef _FB_GENERAL_H_
 #define _FB_GENERAL_H_
@@ -10 +14 @@
         {
                 int start = 0;
+                int prev = 0;
                 int count = -1;
                 do {
                         count++;
                         start = geno.indexOf("aa", start) + 1; // +1 is for progress, starting codons can overlap
+                        int quotecount = 0;
+                        for (int q = prev; q < start; q++) if (geno[q] == '\"') quotecount++;
+                        prev = start;
+                        if (quotecount % 2 != 0) count--; // 'aa' sequence is within quotes
                 } while (start - 1 != -1);
                 return count;
@@ -25 +34 @@
                         count++;
                         start = genotype.indexOf("aa", start) + 1;
+                        int quotecount = 0;
+                        for (int q = 0; q < start; q++) if (genotype[q] == '\"') quotecount++;
+                        if (quotecount % 2 != 0) count--; // 'aa' sequence is within quotes
                 } while (start - 1 != -1 && count != i);
                 if (start - 1 == -1) // there is no gene with a given "i"
@@ -33 +45 @@
                 }
                 end = start;
-                end = genotype.indexOf("zz", end);
+                int quotecount = 0;
+                do {
+                        quotecount = 0;
+                        end = genotype.indexOf("zz", end);
+                        if (end != -1)
+                        {
+                                for (int q = start; q < end; q++) if (genotype[q] == '\"') quotecount++;
+                                if (quotecount % 2 != 0) end++;
+                        }
+                } while (quotecount % 2 != 0 && end != -1);
+
                 if (end == -1) end = genotype.len();
                 else end += 2;

cpp/frams/genetics/fB/fB_oper.cpp

@@ +1 @@
+// This file is a part of Framsticks SDK.  http://www.framsticks.com/
+// Copyright (C) 1999-2018  Maciej Komosinski and Szymon Ulatowski.
+// See LICENSE.txt for details.
+
 #include <frams/util/sstring.h>
 #include <vector>
@@ -5 +9 @@
 #include "fB_general.h"
 #include "fB_oper.h"
+#include "../fH/fH_oper.h"
 
 #define FIELDSTRUCT Geno_fB
@@ -10 +15 @@
 static ParamEntry GENOfBparam_tab[] =
 {
-        { "Genetics: fB", 3, FB_MUT_COUNT + FB_XOVER_COUNT, }, // ask about it
+        { "Genetics: fB", 3, FB_MUT_COUNT + FB_XOVER_COUNT, },
         { "Genetics: fB: Mutation", },
         { "Genetics: fB: Crossover", },
         { "fB_mut_substitution", 1, 0, "Substitution", "f 0 1 0.6", FIELD(mutationprobs[FB_SUBSTITUTION]), "Probability of mutation by changing single random letter in genotype", },
-        { "fB_mut_insertion", 1, 0, "Insertion", "f 0 1 0.1", FIELD(mutationprobs[FB_INSERTION]), "Probability of mutation by inserting characters in random place of genotype", },
+        { "fB_mut_insertion", 1, 0, "Insertion", "f 0 1 0.95", FIELD(mutationprobs[FB_INSERTION]), "Probability of mutation by inserting characters in random place of genotype", },
+        { "fB_mut_nclassins", 1, 0, "Insertion of neuron class definition", "f 0 1 0.05", FIELD(mutationprobs[FB_NCLASSINS]), "Probability of mutation by inserting neuron class definition in random place of genotype", },
         { "fB_mut_deletion", 1, 0, "Deletion", "f 0 1 0.1", FIELD(mutationprobs[FB_DELETION]), "Probability of mutation by deleting random characters in genotype", },
         { "fB_mut_duplication", 1, 0, "Duplication", "f 0 1 0.05", FIELD(mutationprobs[FB_DUPLICATION]), "Probability of mutation by copying single *gene* of genotype and appending it to the beginning of this genotype", },
@@ -81 +87 @@
                 if (!islower(genotype[i]))
                 {
-                        return i + 1;
+                        if (genotype[i] == '"')
+                        {
+                                SString neuclassdef;
+                                int nextid = i + 1;
+                                if (!genotype.getNextToken(nextid, neuclassdef, '"'))
+                                {
+                                        return i + 1;
+                                }
+                                Neuro *neu = new Neuro();
+                                neu->setDetails(neuclassdef);
+
+                                bool isclass = neu->getClass() ? true : false;
+                                delete neu;
+                                if (!isclass)
+                                {
+                                        return i + 1;
+                                }
+                                i = nextid;
+                        }
+                        else
+                        {
+                                return i + 1;
+                        }
                 }
         }
@@ -121 +149 @@
                 if (!isalpha(genotype[i]))
                 {
-                        return GENOPER_OPFAIL;
+                        if (genotype[i] == '"')
+                        {
+                                SString neuclassdef;
+                                int nextid = i + 1;
+                                if (!genotype.getNextToken(nextid, neuclassdef, '"'))
+                                {
+                                        return i + 1;
+                                }
+                                Neuro *neu = new Neuro();
+                                neu->setDetails(neuclassdef);
+
+                                bool isclass = neu->getClass() ? true : false;
+                                delete neu;
+                                if (!isclass)
+                                {
+                                        return i + 1;
+                                }
+                                i = nextid;
+                        }
+                        else
+                        {
+                                return GENOPER_OPFAIL;
+                        }
                 }
                 // if character is uppercase, then convert it to lowercase
-                if (isupper(genotype[i]))
+                else if (isupper(genotype[i]))
                 {
                         genotype.directWrite()[i] = tolower(genotype[i]);
@@ -142 +192 @@
         }
         return GENOPER_OK;
+}
+
+SString Geno_fB::detokenizeSequence(std::list<SString> tokenlist)
+{
+        SString res = "";
+        for (std::list<SString>::iterator it = tokenlist.begin(); it != tokenlist.end(); it++)
+        {
+                res += (*it);
+        }
+        return res;
+}
+
+std::list<SString> Geno_fB::tokenizeSequence(SString genotype)
+{
+        std::list<SString> res;
+        int i = 0;
+        while (i < genotype.len())
+        {
+                // if character is not alphabetic - error
+                if (isalpha(genotype[i]))
+                {
+                        SString el = "";
+                        el += genotype[i];
+                        res.push_back(el);
+                        i++;
+                }
+                else
+                {
+                        SString neuclassdef;
+                        i++;
+                        genotype.getNextToken(i, neuclassdef, '"');
+                        SString ndef = "\"";
+                        ndef += neuclassdef;
+                        ndef += "\"";
+                        res.push_back(ndef);
+                }
+        }
+        return res;
 }
 
@@ -157 +245 @@
         case FB_SUBSTITUTION:
         {
-                int rndid = randomN(line.len()); // select random letter from genotype
+                std::list<SString> tokenized = tokenizeSequence(line);
+                int rndid = randomN(tokenized.size()); // select random letter from genotype
                 // increment/decrement character - when overflow happens, this method
                 // uses reflect method
-                if (randomN(2) == 0)
-                {
-                        if (line[rndid] == 'a') line.directWrite()[rndid] = 'b';
-                        else line.directWrite()[rndid] = line[rndid] - 1;
+                std::list<SString>::iterator it = tokenized.begin();
+                std::advance(it, rndid);
+                SString t = (*it);
+                if ((*it).len() == 1)
+                {
+                        if (randomN(2) == 0)
+                        {
+                                if ((*it)[0] == 'a') (*it).directWrite()[0] = 'b';
+                                else (*it).directWrite()[0] = (*it)[0] - 1;
+                        }
+                        else
+                        {
+                                if ((*it)[0] == 'z') (*it).directWrite()[0] = 'y';
+                                else (*it).directWrite()[0] = (*it)[0] + 1;
+                        }
+                        chg = 1.0 / line.len();
                 }
                 else
                 {
-                        if (line[rndid] == 'z') line.directWrite()[rndid] = 'y';
-                        else line.directWrite()[rndid] = line[rndid] + 1;
-                }
+                        // first method needs to extract quotes
+                        SString def = (*it);
+                        def = def.substr(1, def.len() - 2);
+                        Geno_fH::mutateNeuronProperties(def);
+                        SString res = "\"";
+                        res += def;
+                        res += "\"";
+                        (*it) = res;
+                        chg = (double)def.len() / line.len();
+                }
+                line = detokenizeSequence(tokenized);
+                break;
+        }
+        case FB_INSERTION:
+        {
                 chg = 1.0 / line.len();
-                break;
-        }
-        case FB_INSERTION:
+                std::list<SString> tokenized = tokenizeSequence(line);
+                int rndid = randomN(tokenized.size()); // select random insertion point
+                std::list<SString>::iterator it = tokenized.begin();
+                std::advance(it, rndid);
+                SString letter = "a";
+                letter.directWrite()[0] = 'a' + randomN(26);
+                tokenized.insert(it, letter);
+                line = detokenizeSequence(tokenized);
+                break;
+        }
+        case FB_NCLASSINS:
+        {
+                std::list<SString> tokenized = tokenizeSequence(line);
+                std::list<SString>::iterator it = tokenized.begin();
+                int rndid = randomN(tokenized.size()); // select random insertion point
+                std::advance(it, rndid);
+                NeuroClass *cls = getRandomNeuroClass();
+                SString classdef = cls->getName();
+                Geno_fH::mutateNeuronProperties(classdef);
+                SString res = "\"";
+                res += classdef;
+                res += "\"";
+                tokenized.insert(it, res);
+                chg = (double)classdef.len() / line.len();
+                line = detokenizeSequence(tokenized);
+                break;
+        }
+        case FB_DELETION:
         {
                 chg = 1.0 / line.len();
-                int rndid = randomN(genotype.len()); // select random insertion point
-                char letter = 'a' + randomN(26);
-                SString result = line.substr(0, rndid);
-                result += letter;
-                result += line.substr(rndid);
-                line = result;
-                break;
-        }
-        case FB_DELETION:
-        {
-                chg = 1.0 / line.len();
-                int rndid = randomN(line.len()); // select random insertion point
-                if (rndid == line.len() - 1)
-                {
-                        line = line.substr(0, line.len() - 1);
-                }
-                else
-                {
-                        line = line.substr(0, rndid) + line.substr(rndid + 1);
-                }
+                std::list<SString> tokenized = tokenizeSequence(line);
+                std::list<SString>::iterator it = tokenized.begin();
+                int rndid = randomN(tokenized.size()); // select random deletion point
+                std::advance(it, rndid);
+                tokenized.erase(it);
+                line = detokenizeSequence(tokenized);
                 break;
         }
@@ -210 +334 @@
         case FB_TRANSLOCATION:
         {
-                std::vector<int> cuts(4);
+                std::list<SString> tokenized = tokenizeSequence(line);
+                std::vector<unsigned int> cuts(4);
                 for (int i = 0; i < 4; i++)
                 {
-                        cuts[i] = randomN(line.len());
+                        cuts[i] = randomN(tokenized.size());
                 }
                 std::sort(cuts.begin(), cuts.end());
-                SString first = line.substr(cuts[0], cuts[1] - cuts[0]);
-                SString second = line.substr(cuts[2], cuts[3] - cuts[2]);
-                SString result = line.substr(0, cuts[0]) + second +
-                        line.substr(cuts[1], cuts[2] - cuts[1]) + first + line.substr(cuts[3]);
-                line = result;
+                std::vector<std::list<SString>::iterator> iters(4);
+                for (int i = 0; i < 4; i++)
+                {
+                        iters[i] = tokenized.begin();
+                        std::advance(iters[i], cuts[i]);
+                }
+
+                std::list<SString> res;
+                res.insert(res.end(), tokenized.begin(), iters[0]);
+                res.insert(res.end(), iters[2], iters[3]);
+                res.insert(res.end(), iters[1], iters[2]);
+                res.insert(res.end(), iters[0], iters[1]);
+                res.insert(res.end(), iters[3], tokenized.end());
+
+//              SString first = line.substr(cuts[0], cuts[1] - cuts[0]);
+//              SString second = line.substr(cuts[2], cuts[3] - cuts[2]);
+//              SString result = line.substr(0, cuts[0]) + second +
+//                      line.substr(cuts[1], cuts[2] - cuts[1]) + first + line.substr(cuts[3]);
+                line = detokenizeSequence(res);
                 chg = (float)(cuts[3] - cuts[2] + cuts[1] - cuts[0]) / line.len();
                 break;

cpp/frams/genetics/fB/fB_oper.h

@@ +1 @@
+// This file is a part of Framsticks SDK.  http://www.framsticks.com/
+// Copyright (C) 1999-2018  Maciej Komosinski and Szymon Ulatowski.
+// See LICENSE.txt for details.
+
 #ifndef _FB_OPER_H_
 #define _FB_OPER_H_
 
 #include "../genooperators.h"
+#include <list>
 
 /** @name Codes for general fB mutation types */
@@ -8 +13 @@
 #define FB_SUBSTITUTION  0 ///<Relative probability of mutation by changing single random letter in genotype (substitution)
 #define FB_INSERTION     1 ///<Relative probability of mutation by inserting characters in random place of genotype
-#define FB_DELETION      2 ///<Relative probability of mutation by deleting random characters in genotype
-#define FB_DUPLICATION   3 ///<Relative probability of mutation by copying single *gene* of genotype and appending it to the beginning of this genotype
-#define FB_TRANSLOCATION 4 ///<Relative probability of mutation by replacing two substrings in genotype
-#define FB_MUT_COUNT     5 ///<Count of mutation types
+#define FB_NCLASSINS     2 ///<Relative probability of mutation by inserting neuron class definition in random place of genotype
+#define FB_DELETION      3 ///<Relative probability of mutation by deleting random characters in genotype
+#define FB_DUPLICATION   4 ///<Relative probability of mutation by copying single *gene* of genotype and appending it to the beginning of this genotype
+#define FB_TRANSLOCATION 5 ///<Relative probability of mutation by replacing two substrings in genotype
+#define FB_MUT_COUNT     6 ///<Count of mutation types
 //@}
 
@@ -25 +31 @@
 private:
         bool hasStick(SString genotype);
+        SString detokenizeSequence(std::list<SString> tokenlist);
+        std::list<SString> tokenizeSequence(SString genotype);
 
 public: