| 1 | // This file is a part of Framsticks SDK. http://www.framsticks.com/ |
|---|
| 2 | // Copyright (C) 1999-2020 Maciej Komosinski and Szymon Ulatowski. |
|---|
| 3 | // See LICENSE.txt for details. |
|---|
| 4 | |
|---|
| 5 | #include "geno.h" |
|---|
| 6 | #include "genoconv.h" |
|---|
| 7 | #include <common/loggers/loggers.h> |
|---|
| 8 | #include <common/util-string.h> |
|---|
| 9 | #include <frams/model/model.h> |
|---|
| 10 | |
|---|
| 11 | THREAD_LOCAL_DEF_PTR(Geno::Validators, geno_validators); |
|---|
| 12 | THREAD_LOCAL_DEF_PTR(GenoConvManager, geno_converters); |
|---|
| 13 | |
|---|
| 14 | Geno::Validators* Geno::getValidators() { return tlsGetPtr(geno_validators); } |
|---|
| 15 | GenoConvManager* Geno::getConverters() { return tlsGetPtr(geno_converters); } |
|---|
| 16 | |
|---|
| 17 | Geno::Validators* Geno::useValidators(Validators* val) |
|---|
| 18 | { |
|---|
| 19 | return tlsSetPtr(geno_validators, val); |
|---|
| 20 | } |
|---|
| 21 | GenoConvManager* Geno::useConverters(GenoConvManager* gcm) |
|---|
| 22 | { |
|---|
| 23 | return tlsSetPtr(geno_converters, gcm); |
|---|
| 24 | } |
|---|
| 25 | |
|---|
| 26 | bool Geno::formatIsOneOf(const SString& format, const SString& format_list) |
|---|
| 27 | { |
|---|
| 28 | if (strchr(format_list.c_str(), ',') == NULL) |
|---|
| 29 | return format == format_list; |
|---|
| 30 | else |
|---|
| 31 | { |
|---|
| 32 | SString item; int pos = 0; |
|---|
| 33 | while (format_list.getNextToken(pos, item, ',')) |
|---|
| 34 | if (item == format) |
|---|
| 35 | return true; |
|---|
| 36 | } |
|---|
| 37 | return false; |
|---|
| 38 | } |
|---|
| 39 | |
|---|
| 40 | const SString Geno::FORMAT_INVALID = "invalid"; |
|---|
| 41 | const SString Geno::FORMAT_UNKNOWN = ""; |
|---|
| 42 | const SString Geno::F0_FORMAT_LIST = "0,0s"; |
|---|
| 43 | |
|---|
| 44 | bool Geno::isF0Format(const SString& format_list) |
|---|
| 45 | { |
|---|
| 46 | if (strchr(format_list.c_str(), ',') == NULL) |
|---|
| 47 | return formatIsOneOf(format_list, F0_FORMAT_LIST); |
|---|
| 48 | SString item; int pos = 0; |
|---|
| 49 | while (format_list.getNextToken(pos, item, ',')) |
|---|
| 50 | if (!formatIsOneOf(item, F0_FORMAT_LIST)) |
|---|
| 51 | return false; |
|---|
| 52 | return true; |
|---|
| 53 | } |
|---|
| 54 | |
|---|
| 55 | void Geno::init(const SString& genstring, const SString& genformat, const SString& genname, const SString& comment) |
|---|
| 56 | { |
|---|
| 57 | refcount = 1; |
|---|
| 58 | owner = 0; |
|---|
| 59 | f0gen = 0; |
|---|
| 60 | isvalid = -1; |
|---|
| 61 | name = genname; |
|---|
| 62 | txt = comment; |
|---|
| 63 | setGenesAndFormat(genstring, genformat); |
|---|
| 64 | } |
|---|
| 65 | |
|---|
| 66 | static SString trimAndValidateFormat(const SString& input) //the new requirement for genotype format name: no whitespace |
|---|
| 67 | { |
|---|
| 68 | SString format = trim(input); |
|---|
| 69 | if (format.length() == 0 || strContainsOneOf(format.c_str(), " \r\n\t")) |
|---|
| 70 | return Geno::FORMAT_INVALID; |
|---|
| 71 | return format; |
|---|
| 72 | } |
|---|
| 73 | |
|---|
| 74 | void Geno::setGenesAndFormat(const SString& genstring, const SString& in_genformat) |
|---|
| 75 | { |
|---|
| 76 | mapinshift = 0; |
|---|
| 77 | mapoutshift = 0; |
|---|
| 78 | SString gencopy(genstring); |
|---|
| 79 | SString genformat = in_genformat; |
|---|
| 80 | if (genformat == FORMAT_UNKNOWN) |
|---|
| 81 | { // unknown format |
|---|
| 82 | genformat = "1"; |
|---|
| 83 | if (genstring.charAt(0) == '/') |
|---|
| 84 | { |
|---|
| 85 | int end, error_end = -1; |
|---|
| 86 | switch (genstring.charAt(1)) |
|---|
| 87 | { |
|---|
| 88 | case '/': |
|---|
| 89 | if ((end = genstring.indexOf('\n')) >= 0) |
|---|
| 90 | { |
|---|
| 91 | genformat = trimAndValidateFormat(genstring.substr(2, end - 2)); |
|---|
| 92 | mapinshift = end + 1; |
|---|
| 93 | gencopy = genstring.substr(end + 1); |
|---|
| 94 | if ((end > 0) && (genstring[end - 1] == '\r')) end--; |
|---|
| 95 | error_end = end; |
|---|
| 96 | } |
|---|
| 97 | else |
|---|
| 98 | { |
|---|
| 99 | genformat = trimAndValidateFormat(genstring.substr(2)); |
|---|
| 100 | gencopy = ""; |
|---|
| 101 | mapinshift = genstring.length(); |
|---|
| 102 | } |
|---|
| 103 | break; |
|---|
| 104 | case '*': |
|---|
| 105 | if ((end = genstring.indexOf("*/")) >= 0) |
|---|
| 106 | { |
|---|
| 107 | genformat = trimAndValidateFormat(genstring.substr(2, end - 2)); |
|---|
| 108 | error_end = end + 2; |
|---|
| 109 | gencopy = genstring.substr(end + 2); |
|---|
| 110 | mapinshift = end + 2; |
|---|
| 111 | } |
|---|
| 112 | else |
|---|
| 113 | { |
|---|
| 114 | genformat = trimAndValidateFormat(genstring.substr(2)); |
|---|
| 115 | gencopy = ""; |
|---|
| 116 | mapinshift = genstring.length(); |
|---|
| 117 | } |
|---|
| 118 | break; |
|---|
| 119 | } |
|---|
| 120 | if (genformat == FORMAT_INVALID) |
|---|
| 121 | { |
|---|
| 122 | SString cut; |
|---|
| 123 | if (error_end < 0) error_end = genstring.length(); |
|---|
| 124 | static const int MAX_ERROR = 20; |
|---|
| 125 | if (error_end > MAX_ERROR) |
|---|
| 126 | cut = genstring.substr(0, MAX_ERROR) + "..."; |
|---|
| 127 | else |
|---|
| 128 | cut = genstring.substr(0, error_end); |
|---|
| 129 | int lf = cut.indexOf('\n'); |
|---|
| 130 | if (lf >= 0) { if ((lf > 0) && (cut[lf - 1] == '\r')) lf--; cut = cut.substr(0, lf); } |
|---|
| 131 | sstringQuote(cut); |
|---|
| 132 | 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() : ""); |
|---|
| 133 | } |
|---|
| 134 | |
|---|
| 135 | } |
|---|
| 136 | } |
|---|
| 137 | gen = gencopy; |
|---|
| 138 | multiline = (strchr(gen.c_str(), '\n') != 0); |
|---|
| 139 | format = genformat; |
|---|
| 140 | freeF0(); |
|---|
| 141 | isvalid = -1; |
|---|
| 142 | // mapoutshift...? |
|---|
| 143 | } |
|---|
| 144 | |
|---|
| 145 | void Geno::freeF0() |
|---|
| 146 | { |
|---|
| 147 | if (f0gen) { delete f0gen; f0gen = 0; } |
|---|
| 148 | } |
|---|
| 149 | |
|---|
| 150 | Geno::Geno(const char *genstring, const char* genformat, const char *genname, const char *comment) |
|---|
| 151 | { |
|---|
| 152 | init(SString(genstring), SString(genformat), SString(genname), SString(comment)); |
|---|
| 153 | } |
|---|
| 154 | |
|---|
| 155 | Geno::Geno(const char *genstring, char genformat, const char *genname, const char *comment) |
|---|
| 156 | { |
|---|
| 157 | SString genformat_string; |
|---|
| 158 | if (genformat > 0) |
|---|
| 159 | genformat_string = SString(&genformat, 1); |
|---|
| 160 | init(genstring, genformat_string, genname, comment); |
|---|
| 161 | } |
|---|
| 162 | |
|---|
| 163 | Geno::Geno(const SString& genstring, const SString& genformat, const SString& genname, const SString& comment) |
|---|
| 164 | { |
|---|
| 165 | init(genstring, genformat, genname, comment); |
|---|
| 166 | } |
|---|
| 167 | |
|---|
| 168 | Geno::Geno(const Geno& src) |
|---|
| 169 | :gen(src.gen), name(src.name), format(src.format), txt(src.txt), isvalid(src.isvalid), |
|---|
| 170 | f0gen(0), mapinshift(src.mapinshift), mapoutshift(src.mapinshift), |
|---|
| 171 | multiline(src.multiline), owner(0) |
|---|
| 172 | { |
|---|
| 173 | f0gen = src.f0gen ? new Geno(*src.f0gen) : 0; refcount = 1; |
|---|
| 174 | } |
|---|
| 175 | |
|---|
| 176 | void Geno::operator=(const Geno& src) |
|---|
| 177 | { |
|---|
| 178 | freeF0(); |
|---|
| 179 | gen = src.gen; |
|---|
| 180 | name = src.name; |
|---|
| 181 | format = src.format; |
|---|
| 182 | txt = src.txt; |
|---|
| 183 | isvalid = src.isvalid; |
|---|
| 184 | mapinshift = src.mapinshift; |
|---|
| 185 | mapoutshift = src.mapinshift; |
|---|
| 186 | multiline = src.multiline; |
|---|
| 187 | f0gen = src.f0gen ? new Geno(*src.f0gen) : 0; |
|---|
| 188 | owner = 0; |
|---|
| 189 | } |
|---|
| 190 | |
|---|
| 191 | Geno::Geno(const SString& src) |
|---|
| 192 | { |
|---|
| 193 | init(src, FORMAT_UNKNOWN, SString::empty(), SString::empty()); |
|---|
| 194 | } |
|---|
| 195 | |
|---|
| 196 | void Geno::setGenesAssumingSameFormat(const SString& g) |
|---|
| 197 | { |
|---|
| 198 | gen = g; |
|---|
| 199 | isvalid = -1; |
|---|
| 200 | freeF0(); |
|---|
| 201 | } |
|---|
| 202 | |
|---|
| 203 | void Geno::setString(const SString& g) |
|---|
| 204 | { |
|---|
| 205 | freeF0(); |
|---|
| 206 | init(g, FORMAT_UNKNOWN, SString::empty(), SString::empty()); |
|---|
| 207 | } |
|---|
| 208 | |
|---|
| 209 | void Geno::setName(const SString& n) |
|---|
| 210 | { |
|---|
| 211 | name = n; |
|---|
| 212 | } |
|---|
| 213 | |
|---|
| 214 | void Geno::setComment(const SString& c) |
|---|
| 215 | { |
|---|
| 216 | txt = c; |
|---|
| 217 | } |
|---|
| 218 | |
|---|
| 219 | SString Geno::getGenesAndFormat(void) const |
|---|
| 220 | { |
|---|
| 221 | SString out; |
|---|
| 222 | if (format != "1") |
|---|
| 223 | { |
|---|
| 224 | if (multiline) |
|---|
| 225 | out += "//"; |
|---|
| 226 | else |
|---|
| 227 | out += "/*"; |
|---|
| 228 | out += format; |
|---|
| 229 | if (multiline) |
|---|
| 230 | out += "\n"; |
|---|
| 231 | else |
|---|
| 232 | out += "*/"; |
|---|
| 233 | } |
|---|
| 234 | out += gen; |
|---|
| 235 | return out; |
|---|
| 236 | } |
|---|
| 237 | |
|---|
| 238 | int Geno::mapGenToString(int genpos) const |
|---|
| 239 | { |
|---|
| 240 | if (genpos > gen.length()) return -2; |
|---|
| 241 | if (genpos < 0) return -1; |
|---|
| 242 | return mapinshift + genpos; |
|---|
| 243 | } |
|---|
| 244 | |
|---|
| 245 | int Geno::mapStringToGen(int stringpos) const |
|---|
| 246 | { |
|---|
| 247 | stringpos -= mapinshift; |
|---|
| 248 | if (stringpos > gen.length()) return -2; |
|---|
| 249 | if (stringpos < 0) return -1; |
|---|
| 250 | return stringpos; |
|---|
| 251 | } |
|---|
| 252 | |
|---|
| 253 | SString Geno::getGenes(void) const { return gen; } |
|---|
| 254 | SString Geno::getName(void) const { return name; } |
|---|
| 255 | SString Geno::getFormat(void) const { return format; } |
|---|
| 256 | SString Geno::getComment(void) const { return txt; } |
|---|
| 257 | |
|---|
| 258 | int ModelGenoValidator::testGenoValidity(Geno& g) |
|---|
| 259 | { |
|---|
| 260 | if (Geno::formatIsOneOf(g.getFormat(), Geno::F0_FORMAT_LIST)) |
|---|
| 261 | { |
|---|
| 262 | Model mod(g, Model::SHAPETYPE_UNKNOWN); |
|---|
| 263 | return mod.isValid(); |
|---|
| 264 | } |
|---|
| 265 | else |
|---|
| 266 | { |
|---|
| 267 | bool converter_missing; |
|---|
| 268 | Geno f0geno = g.getConverted(Geno::F0_FORMAT_LIST, NULL, false, &converter_missing); |
|---|
| 269 | if (converter_missing) |
|---|
| 270 | return -1;//no result |
|---|
| 271 | return f0geno.isValid(); |
|---|
| 272 | } |
|---|
| 273 | } |
|---|
| 274 | |
|---|
| 275 | void Geno::validate() |
|---|
| 276 | { |
|---|
| 277 | if (isvalid >= 0) return; |
|---|
| 278 | if (gen.length() == 0) { isvalid = 0; return; } |
|---|
| 279 | if (format == FORMAT_INVALID) { isvalid = 0; return; } |
|---|
| 280 | Validators* vals = getValidators(); |
|---|
| 281 | if (vals != NULL) |
|---|
| 282 | { |
|---|
| 283 | #ifdef WARN_VALIDATION_INCONSISTENCY |
|---|
| 284 | vector<int> results; |
|---|
| 285 | int first_result = -1; |
|---|
| 286 | FOREACH(GenoValidator*, v, (*vals)) |
|---|
| 287 | { |
|---|
| 288 | int r = v->testGenoValidity(*this); |
|---|
| 289 | if (first_result < 0) first_result = r; |
|---|
| 290 | results.push_back(r); |
|---|
| 291 | } |
|---|
| 292 | int N = vals->size(); |
|---|
| 293 | for (int i = 1; i < N; i++) |
|---|
| 294 | if (results[i] != results[0]) |
|---|
| 295 | { |
|---|
| 296 | SString txt = "Inconsistent validation results"; |
|---|
| 297 | for (int i = 0; i < N; i++) |
|---|
| 298 | txt += SString::sprintf(" %d", results[i]); |
|---|
| 299 | txt += " for genotype '"; |
|---|
| 300 | txt += getGene(); |
|---|
| 301 | txt += "'"; |
|---|
| 302 | logPrintf("Geno", "validate", LOG_WARN, txt.c_str()); |
|---|
| 303 | break; |
|---|
| 304 | } |
|---|
| 305 | isvalid = first_result; |
|---|
| 306 | if (isvalid >= 0) |
|---|
| 307 | return; |
|---|
| 308 | #else |
|---|
| 309 | FOREACH(GenoValidator*, v, (*vals)) |
|---|
| 310 | if ((isvalid = v->testGenoValidity(*this)) >= 0) |
|---|
| 311 | return; |
|---|
| 312 | #endif |
|---|
| 313 | } |
|---|
| 314 | isvalid = 0; |
|---|
| 315 | logPrintf("Geno", "validate", LOG_WARN, "Wrong configuration? No genotype validators defined for genetic format 'f%s'.", format.c_str()); |
|---|
| 316 | } |
|---|
| 317 | |
|---|
| 318 | bool Geno::isValid(void) |
|---|
| 319 | { |
|---|
| 320 | if (isvalid < 0) |
|---|
| 321 | { |
|---|
| 322 | LoggerToMemory err(LoggerBase::Enable | LoggerToMemory::StoreAllMessages, LOG_INFO); |
|---|
| 323 | validate(); |
|---|
| 324 | err.disable(); |
|---|
| 325 | string msg = err.getCountSummary(); |
|---|
| 326 | if (msg.size() > 0) |
|---|
| 327 | { |
|---|
| 328 | msg += ssprintf(" while checking validity of '%s'", getName().c_str()); |
|---|
| 329 | msg += "\n"; |
|---|
| 330 | msg += err.getMessages(); |
|---|
| 331 | logMessage("Geno", "isValid", err.getErrorLevel(), msg.c_str()); |
|---|
| 332 | } |
|---|
| 333 | } |
|---|
| 334 | return isvalid > 0; |
|---|
| 335 | } |
|---|
| 336 | |
|---|
| 337 | Geno Geno::getConverted(SString otherformat_list, MultiMap *m, bool using_checkpoints, bool *converter_missing) |
|---|
| 338 | { |
|---|
| 339 | if (formatIsOneOf(getFormat(), otherformat_list)) { if (converter_missing) *converter_missing = false; return *this; } |
|---|
| 340 | #ifndef NO_GENOCONVMANAGER |
|---|
| 341 | GenoConvManager *converters = getConverters(); |
|---|
| 342 | if (converters) |
|---|
| 343 | { |
|---|
| 344 | if ((isF0Format(otherformat_list)) && (!m) && (!using_checkpoints)) |
|---|
| 345 | { |
|---|
| 346 | if (!f0gen) |
|---|
| 347 | f0gen = new Geno(converters->convert(*this, otherformat_list, NULL, using_checkpoints, converter_missing)); |
|---|
| 348 | else |
|---|
| 349 | { |
|---|
| 350 | if (converter_missing) *converter_missing = false; |
|---|
| 351 | } |
|---|
| 352 | return *f0gen; |
|---|
| 353 | } |
|---|
| 354 | else |
|---|
| 355 | return converters->convert(*this, otherformat_list, m, using_checkpoints, converter_missing); |
|---|
| 356 | } |
|---|
| 357 | #endif |
|---|
| 358 | if (converter_missing) *converter_missing = true; |
|---|
| 359 | return (formatIsOneOf(getFormat(), otherformat_list)) ? *this : Geno("", "", "", "GenConvManager not available"); |
|---|
| 360 | } |
|---|
| 361 | |
|---|
| 362 | Geno::~Geno() |
|---|
| 363 | { |
|---|
| 364 | if (f0gen) delete f0gen; |
|---|
| 365 | } |
|---|