annotate import/Import.cpp @ 10:d35e5d769c87 classical-rdf

* some experiments with composer name matching
author Chris Cannam
date Wed, 17 Feb 2010 19:26:48 +0000
parents df999875c53b
children c8ef23d3888c
rev   line source
Chris@0 1 /* -*- c-basic-offset: 4 indent-tabs-mode: nil -*- vi:set ts=8 sts=4 sw=4: */
Chris@0 2
Chris@0 3 #include "Objects.h"
Chris@0 4
Chris@0 5 #include <dataquay/BasicStore.h>
Chris@0 6 #include <dataquay/RDFException.h>
Chris@0 7 #include <dataquay/objectmapper/ObjectMapper.h>
Chris@0 8 #include <dataquay/objectmapper/ObjectBuilder.h>
Chris@0 9 #include <dataquay/objectmapper/ContainerBuilder.h>
Chris@0 10
Chris@0 11 #include "ImportClassicalComposersOrg.h"
Chris@0 12 #include "ImportClassicalDotNet.h"
Chris@4 13 #include "ImportClassicalArchives.h"
Chris@0 14 #include "ImportWikipediaComposers.h"
Chris@0 15 #include "ImportWikipediaWorks.h"
Chris@0 16 #include "ImportWikipediaWorksK.h"
Chris@0 17 #include "ImportWikipediaWorksList.h"
Chris@0 18 #include "ImportHoboken.h"
Chris@0 19
Chris@7 20 #include "TypeRegistrar.h"
Chris@7 21
Chris@0 22 #include <dataquay/Debug.h>
Chris@0 23
Chris@0 24 using namespace ClassicalData;
Chris@0 25 using namespace Dataquay;
Chris@0 26
Chris@0 27 #include <iostream>
Chris@0 28 #include <set>
Chris@0 29
Chris@0 30 typedef QMap<QString, QSet<Composer *> > ComposerMap; // name -> composers
Chris@0 31
Chris@0 32 void
Chris@0 33 addMiscExpansions(Composer *c)
Chris@0 34 {
Chris@0 35 QString n = c->name();
Chris@0 36
Chris@0 37 DEBUG << "addMiscExpansions: n = " << n << endl;
Chris@0 38
Chris@0 39 // lovely hard-coded special cases go here! some of these are
Chris@0 40 // needed for works->composer assignments
Chris@0 41 if (n == "Balakirev, Milii") {
Chris@0 42 c->addAlias("Mily Balakirev");
Chris@0 43 }
Chris@0 44 if (n.startsWith("Cui, C")) {
Chris@0 45 c->addAlias(QString::fromUtf8("C\303\251sar Cui"));
Chris@0 46 }
Chris@0 47 if (n == "Handel, George Frideric") {
Chris@0 48 c->addAlias("Handel, Georg Friedrich");
Chris@0 49 c->addAlias("Handel");
Chris@0 50 }
Chris@1 51 if (n == "Prokofiev, Sergey") {
Chris@1 52 c->addAlias("Prokofieff, Sergei");
Chris@1 53 c->addAlias("Sergei Prokofieff");
Chris@1 54 }
Chris@1 55 if (n == "Rossini, Gioacchino") {
Chris@1 56 c->addAlias("Rossini, Gioachino");
Chris@1 57 c->addAlias("Gioachino Rossini");
Chris@1 58 }
Chris@1 59 if (n == "Edwards, Richard") {
Chris@1 60 c->addAlias("Edwardes, Richard");
Chris@1 61 c->addAlias("Richard Edwardes");
Chris@1 62 c->addAlias("Richard Edwards");
Chris@1 63 }
Chris@1 64 if (n == "Rimsky-Korsakov, Nikolay Andreyevich") {
Chris@1 65 c->addAlias("Nikolai Rimsky-Korsakov");
Chris@1 66 }
Chris@1 67 if (n.startsWith("Piccinni, Nico")) {
Chris@1 68 c->addAlias(n);
Chris@1 69 c->setName(QString::fromUtf8("Piccinni, Niccol\303\262"));
Chris@1 70 }
Chris@1 71 if (n == "Tchaikovsky, Pyotr Ilyich") {
Chris@1 72 c->addAlias("Tchaikovsky, Piotr Ilyitch");
Chris@1 73 }
Chris@1 74 if (n == "Wilhelm Stenhammar") {
Chris@1 75 c->addAlias("Stenhammar, Vilhelm Eugene");
Chris@1 76 c->setName("Stenhammar, Wilhelm");
Chris@1 77 c->addAlias(n);
Chris@1 78 }
Chris@1 79 if (n == "Mercadante, Saverio Rafaele") {
Chris@1 80 c->addAlias("Mercadante, Giuseppe");
Chris@1 81 }
Chris@1 82 if (n == "Johann Wenzel Anton Stamitz") {
Chris@1 83 c->addAlias(n);
Chris@1 84 c->setName("Stamitz, Johann Wenzel Anton");
Chris@1 85 c->addAlias("Stamitz, Jan Vaclav");
Chris@1 86 }
Chris@1 87 if (n == "Mario Castelnuovo-Tedesco") {
Chris@1 88 c->addAlias("Castelnuovo Tedesco, Mario");
Chris@1 89 }
Chris@0 90 if (n == "Mayr, Simon") {
Chris@0 91 c->addAlias("Mayr");
Chris@0 92 }
Chris@0 93
Chris@0 94 n.replace(", Sr.", " Sr.");
Chris@0 95 n.replace(", Jr.", " Jr.");
Chris@0 96
Chris@0 97 int comma = n.indexOf(", ");
Chris@0 98 if (comma > 0 && comma + 2 < n.length()) {
Chris@0 99
Chris@0 100 QString left = n.left(comma);
Chris@0 101 QString right = n.right(n.length() - comma - 2);
Chris@0 102
Chris@0 103 QRegExp jrsr("( (Sr\\.|Jr\\.|I|II))$");
Chris@0 104 if (jrsr.indexIn(right) >= 0) {
Chris@0 105 left = left + jrsr.cap(1);
Chris@0 106 right = right.left(right.length()-jrsr.matchedLength());
Chris@0 107 }
Chris@0 108 n = right + " " + left;
Chris@0 109 }
Chris@0 110
Chris@0 111 if (n != c->name()) c->addAlias(n);
Chris@0 112
Chris@0 113 if (n.contains("Sergey")) {
Chris@0 114 QString nn(n);
Chris@0 115 nn.replace("Sergey", "Sergei");
Chris@0 116 c->addAlias(nn);
Chris@1 117 } else if (n.contains("Sergei")) {
Chris@1 118 QString nn(n);
Chris@1 119 nn.replace("Sergei", "Sergey");
Chris@1 120 c->addAlias(nn);
Chris@0 121 }
Chris@0 122
Chris@0 123 QRegExp sr("((, )?Sr\\.|Senior|\\(?the elder\\)?)", Qt::CaseInsensitive);
Chris@0 124 if (sr.indexIn(n) >= 0) {
Chris@0 125 QString nr = n;
Chris@0 126 nr.replace(sr.pos(0), sr.matchedLength(), " I");
Chris@0 127 nr.replace(" ", " ");
Chris@0 128 DEBUG << "addMiscExpansions: trying " << nr << " for " << n << endl;
Chris@0 129 c->addAlias(nr);
Chris@0 130 }
Chris@0 131 QRegExp jr("((, )?Jr\\.|Junior|\\(?the younger\\)?)", Qt::CaseInsensitive);
Chris@0 132 if (jr.indexIn(n) >= 0) {
Chris@0 133 QString nr = n;
Chris@0 134 nr.replace(jr.pos(0), jr.matchedLength(), " II");
Chris@0 135 nr.replace(" ", " ");
Chris@0 136 DEBUG << "addMiscExpansions: trying " << nr << " for " << n << endl;
Chris@0 137 c->addAlias(nr);
Chris@0 138 }
Chris@0 139 QString nr = n;
Chris@0 140 nr.replace("(I)", "I");
Chris@0 141 nr.replace("(II)", "II");
Chris@0 142 nr.replace("(III)", "III");
Chris@0 143 c->addAlias(nr);
Chris@0 144 }
Chris@0 145
Chris@0 146 bool
Chris@0 147 hasBetterName(Composer *c, Composer *other)
Chris@0 148 {
Chris@0 149 if (c->name() == other->name()) return false;
Chris@0 150
Chris@0 151 // Try to guess which of c and other is more likely to have a good
Chris@0 152 // "canonical form" of the composer's name
Chris@0 153
Chris@0 154 if (c->name().startsWith("van ")) {
Chris@0 155 return false; // wrong choice of sort for e.g. LvB; should be
Chris@0 156 // Beethoven, Ludwig van, not van Beethoven, Ludwig
Chris@0 157 }
Chris@0 158 if (other->name().startsWith("van ")) {
Chris@0 159 return true;
Chris@0 160 }
Chris@0 161
Chris@0 162 if (c->aliases().size() != other->aliases().size()) {
Chris@0 163 // a rather weak heuristic
Chris@0 164 return c->aliases().size() > other->aliases().size();
Chris@0 165 }
Chris@0 166
Chris@0 167 if (c->name().contains(',') && !other->name().contains(',')) {
Chris@0 168 // another rather weak heuristic
Chris@0 169 return true;
Chris@0 170 }
Chris@0 171
Chris@0 172 return false;
Chris@0 173 }
Chris@0 174
Chris@0 175 void mergeComposer(Composer *c, ComposerMap &composers)
Chris@0 176 {
Chris@0 177 QString name = c->name();
Chris@0 178
Chris@0 179 QSet<QString> allNames = c->aliases();
Chris@0 180 allNames.insert(name);
Chris@0 181
Chris@0 182 QString dates;
Chris@0 183 if (c->birth()) {
Chris@0 184 if (c->death()) {
Chris@0 185 dates = QString("%1-%2").arg(c->birth()->year()).arg(c->death()->year());
Chris@0 186 } else {
Chris@0 187 dates = QString("%1-").arg(c->birth()->year());
Chris@0 188 }
Chris@0 189 }
Chris@0 190 if (dates != "") {
Chris@0 191 allNames.insert(dates);
Chris@0 192 }
Chris@0 193
Chris@0 194 QSet<Composer *> matches;
Chris@0 195
Chris@0 196 foreach (QString candidateName, allNames) {
Chris@10 197 QString key = Composer::reduceName(candidateName);
Chris@0 198 if (composers.contains(key)) {
Chris@0 199 foreach (Composer *candidate, composers[key]) {
Chris@0 200 if (candidateName == dates) {
Chris@5 201 if (c->name() == candidate->name()) {
Chris@5 202 DEBUG << "mergeComposer: Exact name match for " << c->name() << " with date(s) " << dates << endl;
Chris@10 203 } else if (!candidate->matchCatalogueName(c->name()) &&
Chris@10 204 !c->matchCatalogueName(candidate->name())) {
Chris@0 205 DEBUG << "mergeComposer: Names differ for " << c->name() << " and " << candidate->name() << " (having matched date(s) " << dates << ")" << endl;
Chris@0 206 continue;
Chris@0 207 } else {
Chris@0 208 DEBUG << "mergeComposer: Note: Fuzzy name match for " << c->name() << " and " << candidate->name() << " with date(s) " << dates << endl;
Chris@0 209 }
Chris@0 210 } else {
Chris@10 211 if (!c->matchDates(candidate)) {
Chris@0 212 DEBUG << "mergeComposer: Dates differ for " << c->name() << " and " << candidate->name() << endl;
Chris@0 213 continue;
Chris@0 214 }
Chris@0 215 }
Chris@0 216 matches.insert(candidate);
Chris@0 217 }
Chris@0 218 }
Chris@0 219 }
Chris@0 220
Chris@0 221 if (matches.empty()) {
Chris@0 222 DEBUG << "mergeComposer: No existing composer with alias matching any alias of " << c->name() << ", adding" << endl;
Chris@0 223
Chris@0 224 if (!c->birth() && !c->death()) {
Chris@5 225 DEBUG << "Composer has no dates, laboriously searching for all names" << endl;
Chris@0 226 // laboriously look for fuzzy match across _all_ composers
Chris@0 227 for (ComposerMap::iterator i = composers.begin();
Chris@0 228 i != composers.end(); ++i) {
Chris@0 229 foreach (Composer *candidate, *i) {
Chris@10 230 if (candidate->matchCatalogueName(c->name())) {
Chris@0 231 DEBUG << "mergeComposer: Found fuzzy match for undated composer " << c->name() << " as " << candidate->name() << ", daringly merging" << endl;
Chris@0 232 matches.insert(candidate);
Chris@0 233 break;
Chris@0 234 }
Chris@0 235 }
Chris@0 236 if (!matches.empty()) break;
Chris@0 237 }
Chris@0 238 }
Chris@0 239
Chris@0 240 if (matches.empty()) {
Chris@0 241 foreach (QString candidateName, allNames) {
Chris@10 242 QString key = Composer::reduceName(candidateName);
Chris@5 243 composers[key].insert(c);
Chris@0 244 DEBUG << "added for alias or date " << candidateName << endl;
Chris@0 245 }
Chris@0 246 return;
Chris@0 247 }
Chris@0 248 }
Chris@0 249
Chris@0 250 if (matches.size() > 1) {
Chris@0 251 DEBUG << "mergeComposer: More than one composer matches name and date(s) for " << c->name() << " -- something fishy here" << endl;
Chris@0 252 }
Chris@0 253
Chris@0 254 Composer *other = *matches.begin();
Chris@0 255
Chris@0 256 DEBUG << "mergeComposer: Merging " << c->name() << " with " << other->name() << endl;
Chris@0 257
Chris@0 258 if (hasBetterName(c, other)) {
Chris@0 259 other->addAlias(other->name());
Chris@0 260 other->setName(c->name());
Chris@0 261 } else {
Chris@0 262 other->addAlias(c->name());
Chris@0 263 }
Chris@10 264 composers[Composer::reduceName(c->name())].insert(other);
Chris@0 265 DEBUG << "linking from alias " << c->name() << endl;
Chris@0 266
Chris@0 267 foreach (QString alias, c->aliases()) {
Chris@0 268 if (alias != other->name() &&
Chris@0 269 !other->aliases().contains(alias)) {
Chris@0 270 other->addAlias(alias);
Chris@10 271 composers[Composer::reduceName(alias)].insert(other);
Chris@0 272 DEBUG << "linking from alias " << alias << endl;
Chris@0 273 }
Chris@0 274 }
Chris@0 275
Chris@0 276 foreach (Document *d, c->pages()) {
Chris@0 277 bool found = false;
Chris@0 278 foreach (Document *dd, other->pages()) {
Chris@0 279 if (d->uri() == dd->uri()) {
Chris@0 280 found = true;
Chris@0 281 break;
Chris@0 282 }
Chris@0 283 }
Chris@0 284 if (!found) {
Chris@0 285 d->setTopic(other);
Chris@0 286 other->addPage(d);
Chris@0 287 }
Chris@0 288 }
Chris@0 289
Chris@0 290 //!!! actually the "approximate" bits of the following are bogus;
Chris@0 291 // a source reporting birth or death date as approx is probably
Chris@0 292 // more accurate than one reporting an exact date
Chris@0 293
Chris@0 294 if (c->birth()) {
Chris@0 295 if (!other->birth() || other->birth()->approximate()) {
Chris@0 296 other->setBirth(c->birth());
Chris@0 297 }
Chris@0 298 }
Chris@0 299
Chris@0 300 if (c->death()) {
Chris@0 301 if (!other->death() || other->death()->approximate()) {
Chris@0 302 other->setDeath(c->death());
Chris@0 303 }
Chris@0 304 }
Chris@0 305
Chris@0 306 if (c->gender() != "") other->setGender(c->gender());
Chris@4 307
Chris@4 308 foreach (QString s, c->nationality()) {
Chris@4 309 other->addNationality(s);
Chris@4 310 }
Chris@4 311
Chris@4 312 foreach (QUrl s, c->geonameURIs()) {
Chris@4 313 other->addGeonameURI(s);
Chris@4 314 }
Chris@4 315
Chris@0 316 if (c->remarks() != "") other->setRemarks(c->remarks());
Chris@0 317 if (c->period() != "") other->setPeriod(c->period());
Chris@0 318
Chris@0 319 }
Chris@0 320
Chris@0 321 QString
Chris@0 322 asciify(QString field)
Chris@0 323 {
Chris@0 324 // accented characters etc -- add "ascii version" for dumb search purposes
Chris@0 325 QString ascii;
Chris@0 326 for (int i = 0; i < field.length(); ++i) {
Chris@0 327 QString dc = field[i].decomposition();
Chris@0 328 if (dc != "") ascii += dc[0];
Chris@0 329 else if (field[i] == QChar(0x00DF)) {
Chris@0 330 ascii += "ss";
Chris@0 331 } else {
Chris@0 332 ascii += field[i];
Chris@0 333 }
Chris@0 334 }
Chris@0 335 ascii.replace(QString::fromUtf8("\342\200\231"), "'"); // apostrophe
Chris@0 336 ascii.replace(QString::fromUtf8("\342\200\222"), "-");
Chris@0 337 ascii.replace(QString::fromUtf8("\342\200\223"), "-");
Chris@0 338 ascii.replace(QString::fromUtf8("\342\200\224"), "-");
Chris@0 339 ascii.replace(QString::fromUtf8("\342\200\225"), "-");
Chris@0 340 return ascii;
Chris@0 341 }
Chris@0 342
Chris@0 343 void
Chris@0 344 asciify(Composer *c)
Chris@0 345 {
Chris@0 346 QString n = c->name();
Chris@0 347 QString asc = asciify(n);
Chris@0 348 if (asc != n && !c->aliases().contains(asc)) c->addAlias(asc);
Chris@0 349 foreach (QString alias, c->aliases()) {
Chris@0 350 asc = asciify(alias);
Chris@0 351 if (asc != alias && !c->aliases().contains(asc)) c->addAlias(asc);
Chris@0 352 }
Chris@0 353 }
Chris@0 354
Chris@0 355 void
Chris@0 356 asciify(Work *w)
Chris@0 357 {
Chris@0 358 QString n = w->name();
Chris@0 359 QString asc = asciify(n);
Chris@0 360 if (asc != n && !w->aliases().contains(asc)) w->addAlias(asc);
Chris@0 361 foreach (QString alias, w->aliases()) {
Chris@0 362 asc = asciify(alias);
Chris@0 363 if (asc != alias && !w->aliases().contains(asc)) w->addAlias(asc);
Chris@0 364 }
Chris@0 365 }
Chris@0 366
Chris@0 367 void
Chris@0 368 assignUri(Store *s, Composer *c)
Chris@0 369 {
Chris@0 370 static QSet<QString> convSet;
Chris@0 371 QString conv = c->name();
Chris@0 372 if (!conv.contains(",")) {
Chris@0 373 QStringList sl = conv.split(" ");
Chris@0 374 if (!sl.empty()) {
Chris@0 375 sl.push_front(sl[sl.size()-1]);
Chris@0 376 sl.removeLast();
Chris@0 377 conv = sl.join(" ");
Chris@0 378 DEBUG << "assignUri: " << c->name() << " -> " << conv << endl;
Chris@0 379 }
Chris@0 380 }
Chris@0 381 conv = asciify(conv);
Chris@0 382 conv.replace(" ", "_");
Chris@0 383 conv.replace("-", "_");
Chris@0 384 conv.replace(QRegExp("[^a-zA-Z0-9_-]"), "");
Chris@0 385 conv = conv.toLower();
Chris@0 386 QString initial = conv;
Chris@1 387 int i = 2;
Chris@0 388 while (convSet.contains(conv)) {
Chris@0 389 conv = QString("%1__%2").arg(initial).arg(i);
Chris@0 390 i++;
Chris@0 391 }
Chris@0 392 convSet.insert(conv);
Chris@1 393 c->setProperty("uri", s->expand(":composer/" + conv));
Chris@0 394 }
Chris@0 395
Chris@0 396 void
Chris@0 397 assignUri(Store *s, Work *w, Composer *c)
Chris@0 398 {
Chris@0 399 QString pfx = c->property("uri").toUrl().toString();
Chris@0 400 DEBUG << "pfx = " << pfx << endl;
Chris@2 401 if (!pfx.contains("composer/")) pfx = ":work/";
Chris@2 402 else {
Chris@2 403 pfx.replace("composer/", "work/");
Chris@2 404 pfx += "/";
Chris@2 405 }
Chris@0 406
Chris@0 407 static QSet<QString> convSet;
Chris@1 408
Chris@0 409 QString conv = w->catalogue();
Chris@0 410 if (conv == "") conv = w->opus();
Chris@0 411 conv = conv.replace(".", "");
Chris@0 412 bool hasOpus = (conv != "");
Chris@1 413 if (conv == "") conv = w->name().toLower();
Chris@0 414 if (w->number() != "") conv = conv + "_no" + w->number();
Chris@0 415 conv = asciify(conv);
Chris@0 416 conv.replace(" ", "_");
Chris@0 417 conv.replace("-", "_");
Chris@0 418 conv.replace(":", "_");
Chris@0 419 conv.replace(QRegExp("[^a-zA-Z0-9_-]"), "");
Chris@1 420
Chris@2 421 if (pfx != "") conv = pfx + conv;
Chris@1 422
Chris@0 423 // I think actually for works we want to merge duplicates rather than
Chris@0 424 // assign them separate URIs, _unless_ they lack a viable opus number
Chris@0 425 if (!hasOpus) {
Chris@0 426 QString initial = conv;
Chris@1 427 int i = 2;
Chris@0 428 while (convSet.contains(conv)) {
Chris@0 429 conv = QString("%1__%2").arg(initial).arg(i);
Chris@0 430 i++;
Chris@0 431 }
Chris@0 432 }
Chris@0 433 convSet.insert(conv);
Chris@1 434
Chris@1 435 w->setProperty("uri", conv);
Chris@0 436 }
Chris@0 437
Chris@0 438 void
Chris@0 439 addDbpediaResource(Store *store, QObject *o, QString s)
Chris@0 440 {
Chris@0 441 QUrl u = o->property("uri").toUrl();
Chris@0 442 if (u == QUrl()) return;
Chris@0 443 if (s.startsWith("http://en.wikipedia.org/wiki/")) {
Chris@0 444 store->add(Triple(u,
Chris@0 445 "mo:wikipedia",
Chris@0 446 QUrl(s)));
Chris@0 447 s.replace("http://en.wikipedia.org/wiki/",
Chris@0 448 "http://dbpedia.org/resource/");
Chris@0 449 store->add(Triple(u,
Chris@0 450 "owl:sameAs",
Chris@0 451 QUrl(s)));
Chris@0 452 }
Chris@0 453 }
Chris@0 454
Chris@0 455 int main(int argc, char **argv)
Chris@0 456 {
Chris@0 457 qRegisterMetaType<ClassicalComposersOrgImporter *>
Chris@0 458 ("ClassicalData::ClassicalComposersOrgImporter*");
Chris@0 459 qRegisterMetaType<ClassicalDotNetImporter *>
Chris@0 460 ("ClassicalData::ClassicalDotNetImporter*");
Chris@4 461 qRegisterMetaType<ClassicalArchivesImporter *>
Chris@4 462 ("ClassicalData::ClassicalArchivesImporter*");
Chris@0 463 qRegisterMetaType<WikipediaComposersImporter *>
Chris@0 464 ("ClassicalData::WikipediaComposersImporter*");
Chris@0 465 qRegisterMetaType<WikipediaWorksImporter *>
Chris@0 466 ("ClassicalData::WikipediaWorksImporter*");
Chris@0 467 qRegisterMetaType<WikipediaWorksKImporter *>
Chris@0 468 ("ClassicalData::WikipediaWorksKImporter*");
Chris@0 469 qRegisterMetaType<WikipediaWorksListImporter *>
Chris@0 470 ("ClassicalData::WikipediaWorksListImporter*");
Chris@0 471 qRegisterMetaType<HobokenImporter *>
Chris@0 472 ("ClassicalData::HobokenImporter*");
Chris@0 473
Chris@0 474 ObjectBuilder::getInstance()->registerClass
Chris@0 475 <ClassicalComposersOrgImporter>("ClassicalData::ClassicalComposersOrgImporter*");
Chris@0 476 ObjectBuilder::getInstance()->registerClass
Chris@0 477 <ClassicalDotNetImporter>("ClassicalData::ClassicalDotNetImporter*");
Chris@0 478 ObjectBuilder::getInstance()->registerClass
Chris@4 479 <ClassicalArchivesImporter>("ClassicalData::ClassicalArchivesImporter*");
Chris@4 480 ObjectBuilder::getInstance()->registerClass
Chris@0 481 <WikipediaComposersImporter>("ClassicalData::WikipediaComposersImporter*");
Chris@0 482 ObjectBuilder::getInstance()->registerClass
Chris@0 483 <WikipediaWorksImporter>("ClassicalData::WikipediaWorksImporter*");
Chris@0 484 ObjectBuilder::getInstance()->registerClass
Chris@0 485 <WikipediaWorksKImporter>("ClassicalData::WikipediaWorksKImporter*");
Chris@0 486 ObjectBuilder::getInstance()->registerClass
Chris@0 487 <WikipediaWorksListImporter>("ClassicalData::WikipediaWorksListImporter*");
Chris@0 488 ObjectBuilder::getInstance()->registerClass
Chris@0 489 <HobokenImporter>("ClassicalData::HobokenImporter*");
Chris@0 490
Chris@0 491 BasicStore *store = BasicStore::load("file:importers.ttl");
Chris@0 492 ObjectMapper mapper(store);
Chris@0 493 QObject *parentObject = mapper.loadAllObjects(new QObject());
Chris@0 494
Chris@0 495 BasicStore *outstore = new BasicStore();
Chris@1 496 outstore->setBaseUri("http://dbtune.org/classical/resource/");
Chris@0 497 ObjectMapper outmapper(outstore);
Chris@0 498
Chris@7 499 TypeRegistrar::addMappings(outstore, &outmapper);
Chris@7 500
Chris@0 501 outmapper.setPropertyStorePolicy(ObjectMapper::StoreIfChanged);
Chris@3 502 outmapper.setObjectStorePolicy(ObjectMapper::StoreAllObjects);
Chris@3 503 outmapper.setBlankNodePolicy(ObjectMapper::NoBlankNodes);
Chris@1 504
Chris@0 505 QList<Importer *> importers = parentObject->findChildren<Importer *>();
Chris@0 506 std::cerr << "have " << importers.size() << " importers" << std::endl;
Chris@0 507
Chris@0 508 ComposerMap composers;
Chris@0 509
Chris@0 510 QList<Composer *> dated;
Chris@0 511 QList<Composer *> undated;
Chris@0 512
Chris@0 513 QList<Work *> works;
Chris@0 514 QList<Composition *> compositions;
Chris@0 515 QList<QObject *> other;
Chris@0 516
Chris@0 517 foreach (Importer *importer, importers) {
Chris@0 518 QObjectList objects = importer->getImportedObjects();
Chris@0 519 foreach (QObject *o, objects) {
Chris@0 520 Composer *c;
Chris@0 521 if ((c = qobject_cast<Composer *>(o))) {
Chris@0 522 addMiscExpansions(c);
Chris@0 523 asciify(c);
Chris@0 524 if (c->birth() || c->death()) dated.push_back(c);
Chris@0 525 else undated.push_back(c);
Chris@0 526 continue;
Chris@0 527 }
Chris@0 528 Work *w;
Chris@0 529 if ((w = qobject_cast<Work *>(o))) {
Chris@0 530 asciify(w);
Chris@0 531 works.push_back(w);
Chris@0 532 continue;
Chris@0 533 }
Chris@0 534 Composition *cn;
Chris@0 535 if ((cn = qobject_cast<Composition *>(o))) {
Chris@0 536 compositions.push_back(cn);
Chris@0 537 continue;
Chris@0 538 }
Chris@0 539 }
Chris@0 540 }
Chris@0 541
Chris@0 542 // get all the dated composers merged before attempting to match
Chris@0 543 // the undated ones
Chris@0 544 foreach (Composer *c, dated) {
Chris@0 545 mergeComposer(c, composers);
Chris@0 546 }
Chris@0 547 foreach (Composer *c, undated) {
Chris@0 548 mergeComposer(c, composers);
Chris@0 549 }
Chris@0 550
Chris@0 551 QObjectList toStore;
Chris@0 552
Chris@0 553 QSet<Composer *> cset;
Chris@0 554 for (ComposerMap::iterator i = composers.begin(); i != composers.end(); ++i) {
Chris@0 555 foreach (Composer *c, i.value()) {
Chris@0 556 if (!cset.contains(c)) {
Chris@0 557 assignUri(outstore, c);
Chris@0 558 toStore.push_back(c);
Chris@0 559 cset.insert(c);
Chris@0 560 }
Chris@0 561 foreach (Document *d, c->pages()) {
Chris@0 562 QString s = d->uri().toString();
Chris@0 563 addDbpediaResource(outstore, c, s);
Chris@0 564 }
Chris@0 565 }
Chris@0 566 }
Chris@0 567
Chris@0 568 QSet<QString> storedUris;
Chris@0 569
Chris@0 570 foreach (Work *w, works) {
Chris@0 571 Composition *cn = w->composition();
Chris@0 572 if (!cn) continue;
Chris@0 573 if (!cn->composer()) {
Chris@0 574 QString cname = cn->composerName();
Chris@10 575 QString key = Composer::reduceName(cname);
Chris@0 576 if (cname != "") {
Chris@5 577 if (!composers.contains(key)) {
Chris@0 578 DEBUG << "Failed to assign Composition to composer: no composer matches name " << cname << endl;
Chris@0 579 } else {
Chris@5 580 QSet<Composer *> cs = composers[key];
Chris@0 581 if (cs.empty()) {
Chris@0 582 DEBUG << "Failed to assign Composition to composer: no composer matches name " << cname << endl;
Chris@0 583 } else if (cs.size() > 1) {
Chris@0 584 DEBUG << "Failed to assign Composition to composer: "
Chris@0 585 << cs.size() << " composers match name " << cname << endl;
Chris@0 586 } else {
Chris@0 587 cn->setComposer(*cs.begin());
Chris@0 588 }
Chris@0 589 }
Chris@0 590 } else {
Chris@0 591 DEBUG << "Failed to assign Composition to composer: composer name is empty" << endl;
Chris@0 592 }
Chris@0 593 }
Chris@0 594
Chris@0 595 if (cn->composer()) {
Chris@0 596 assignUri(outstore, w, cn->composer());
Chris@0 597 }
Chris@0 598
Chris@0 599 foreach (Document *d, w->pages()) {
Chris@0 600 QString s = d->uri().toString();
Chris@0 601 addDbpediaResource(outstore, w, s);
Chris@1 602 if (!storedUris.contains(s)) {
Chris@1 603 toStore.push_back(d);
Chris@1 604 storedUris.insert(s);
Chris@1 605 }
Chris@0 606 }
Chris@0 607
Chris@0 608 QString u = w->property("uri").toUrl().toString();
Chris@0 609 if (u == "" || !storedUris.contains(u)) {
Chris@0 610 toStore.push_back(w);
Chris@0 611 if (u != "") storedUris.insert(u);
Chris@0 612 }
Chris@0 613 }
Chris@0 614
Chris@0 615 try {
Chris@0 616 outmapper.storeAllObjects(toStore);
Chris@0 617
Chris@0 618 } catch (RDFException e) {
Chris@0 619 std::cerr << "Caught RDF exception: " << e.what() << std::endl;
Chris@0 620 }
Chris@0 621
Chris@0 622 DEBUG << "Stored, now saving" << endl;
Chris@0 623
Chris@2 624 outstore->save("imported.ttl");
Chris@0 625
Chris@0 626 DEBUG << "Saved" << endl;
Chris@0 627
Chris@0 628
Chris@0 629 QMultiMap<QString, Composer *> cmap;
Chris@0 630 foreach (Composer *c, cset) {
Chris@0 631 QString n = c->getSortName(true);
Chris@0 632 cmap.insert(n, c);
Chris@0 633 }
Chris@0 634
Chris@0 635 std::cout << "Composers: " << cmap.size() << std::endl;
Chris@0 636
Chris@0 637 for (QMultiMap<QString, Composer *>::iterator i = cmap.begin();
Chris@0 638 i != cmap.end(); ++i) {
Chris@0 639
Chris@0 640 QString n = i.key();
Chris@0 641 Composer *c = i.value();
Chris@0 642
Chris@0 643 std::cout << n.toStdString();
Chris@0 644
Chris@0 645 QString d = c->getDisplayDates();
Chris@0 646 if (d != "") std::cout << " (" << d.toStdString() << ")";
Chris@0 647 std::cout << std::endl;
Chris@0 648 }
Chris@0 649
Chris@0 650 std::cout << std::endl;
Chris@0 651
Chris@0 652 std::cout << "Works by composer:" << std::endl;
Chris@0 653
Chris@0 654 for (QMultiMap<QString, Composer *>::iterator i = cmap.begin();
Chris@0 655 i != cmap.end(); ++i) {
Chris@0 656
Chris@0 657 QString n = i.key();
Chris@0 658 Composer *c = i.value();
Chris@0 659
Chris@0 660 std::set<Work *, Work::Ordering> wmap;
Chris@0 661 foreach (Work *w, works) {
Chris@0 662 Composition *cn = w->composition();
Chris@0 663 if (!cn) continue;
Chris@0 664 if (cn->composer() != c) continue;
Chris@0 665 if (w->partOf()) continue;
Chris@0 666 wmap.insert(w);
Chris@0 667 }
Chris@0 668
Chris@0 669 if (wmap.empty()) continue;
Chris@0 670
Chris@0 671 std::cout << n.toStdString() << std::endl;
Chris@0 672
Chris@0 673 foreach (Work *w, wmap) {
Chris@0 674 std::cout << " * ";
Chris@0 675 std::cout << w->name().toStdString();
Chris@0 676 if (w->catalogue() != "") {
Chris@0 677 std::cout << " [" << w->catalogue().toStdString() << "]";
Chris@0 678 }
Chris@0 679 if (w->opus() != "") {
Chris@0 680 std::cout << " [op. " << w->opus().toStdString() << "]";
Chris@0 681 }
Chris@0 682 std::cout << std::endl;
Chris@0 683 std::set<Work *, Work::Ordering> orderedParts;
Chris@0 684 foreach (Work *ww, w->parts()) {
Chris@0 685 orderedParts.insert(ww);
Chris@0 686 }
Chris@0 687 foreach (Work *ww, orderedParts) {
Chris@0 688 std::cout << " ";
Chris@0 689 if (ww->number() != "") {
Chris@0 690 std::cout << ww->number().toStdString() << ". ";
Chris@0 691 }
Chris@0 692 std::cout << ww->name().toStdString();
Chris@0 693 if (ww->catalogue() != "" && ww->catalogue() != w->catalogue()) {
Chris@0 694 std::cout << " [" << ww->catalogue().toStdString() << "]";
Chris@0 695 }
Chris@0 696 if (ww->opus() != "" && ww->opus() != w->opus()) {
Chris@0 697 std::cout << " [op. " << ww->opus().toStdString() << "]";
Chris@0 698 }
Chris@0 699 std::cout << std::endl;
Chris@0 700 }
Chris@0 701 }
Chris@0 702
Chris@0 703 std::cout << std::endl;
Chris@0 704 }
Chris@0 705
Chris@0 706 delete outstore;
Chris@0 707
Chris@0 708 DEBUG << "Done" << endl;
Chris@0 709
Chris@0 710
Chris@0 711 }
Chris@0 712
Chris@0 713