| 1 | /* |
| 2 | This file is part of the KDE Baloo Project |
| 3 | SPDX-FileCopyrightText: 2013-2015 Vishesh Handa <vhanda@kde.org> |
| 4 | |
| 5 | SPDX-License-Identifier: LGPL-2.1-only OR LGPL-3.0-only OR LicenseRef-KDE-Accepted-LGPL |
| 6 | */ |
| 7 | |
| 8 | #include "baloodebug.h" |
| 9 | #include "searchstore.h" |
| 10 | #include "global.h" |
| 11 | |
| 12 | #include "database.h" |
| 13 | #include "term.h" |
| 14 | #include "transaction.h" |
| 15 | #include "enginequery.h" |
| 16 | #include "termgenerator.h" |
| 17 | #include "andpostingiterator.h" |
| 18 | #include "orpostingiterator.h" |
| 19 | |
| 20 | #include <QDateTime> |
| 21 | |
| 22 | #include <KFileMetaData/PropertyInfo> |
| 23 | #include <KFileMetaData/TypeInfo> |
| 24 | #include <KFileMetaData/Types> |
| 25 | |
| 26 | #include <algorithm> |
| 27 | #include <array> |
| 28 | #include <tuple> |
| 29 | |
| 30 | namespace Baloo { |
| 31 | |
| 32 | namespace { |
| 33 | QPair<quint32, quint32> calculateTimeRange(const QDateTime& dt, Term::Comparator com) |
| 34 | { |
| 35 | Q_ASSERT(dt.isValid()); |
| 36 | |
| 37 | if (com == Term::Equal) { |
| 38 | // Timestamps in DB are quint32 relative to Epoch (1970...2106) |
| 39 | auto start = static_cast<quint32>(dt.date().startOfDay().toSecsSinceEpoch()); |
| 40 | auto end = static_cast<quint32>(dt.date().endOfDay().toSecsSinceEpoch()); |
| 41 | return {start, end}; |
| 42 | } |
| 43 | |
| 44 | quint32 timet = dt.toSecsSinceEpoch(); |
| 45 | if (com == Term::LessEqual) { |
| 46 | return {0, timet}; |
| 47 | } |
| 48 | if (com == Term::Less) { |
| 49 | return {0, timet - 1}; |
| 50 | } |
| 51 | if (com == Term::GreaterEqual) { |
| 52 | return {timet, std::numeric_limits<quint32>::max()}; |
| 53 | } |
| 54 | if (com == Term::Greater) { |
| 55 | return {timet + 1, std::numeric_limits<quint32>::max()}; |
| 56 | } |
| 57 | |
| 58 | Q_ASSERT_X(0, __func__, "mtime query must contain a valid comparator" ); |
| 59 | return {0, 0}; |
| 60 | } |
| 61 | |
| 62 | struct InternalProperty { |
| 63 | const char* propertyName; |
| 64 | const char* prefix; |
| 65 | QMetaType::Type valueType; |
| 66 | }; |
| 67 | constexpr std::array<InternalProperty, 7> internalProperties{._M_elems: {{.propertyName: "content" , .prefix: "" , .valueType: QMetaType::QString}, |
| 68 | {.propertyName: "filename" , .prefix: "F" , .valueType: QMetaType::QString}, |
| 69 | {.propertyName: "mimetype" , .prefix: "M" , .valueType: QMetaType::QString}, |
| 70 | {.propertyName: "rating" , .prefix: "R" , .valueType: QMetaType::Int}, |
| 71 | {.propertyName: "tag" , .prefix: "TAG-" , .valueType: QMetaType::QString}, |
| 72 | {.propertyName: "tags" , .prefix: "TA" , .valueType: QMetaType::QString}, |
| 73 | {.propertyName: "usercomment" , .prefix: "C" , .valueType: QMetaType::QString}}}; |
| 74 | |
| 75 | std::pair<QByteArray, QMetaType::Type> propertyInfo(const QByteArray &property) |
| 76 | { |
| 77 | auto it = std::find_if(first: std::begin(cont: internalProperties), last: std::end(cont: internalProperties), |
| 78 | pred: [&property] (const InternalProperty& entry) { return property == entry.propertyName; }); |
| 79 | if (it != std::end(cont: internalProperties)) { |
| 80 | return { (*it).prefix, (*it).valueType }; |
| 81 | } else { |
| 82 | KFileMetaData::PropertyInfo pi = KFileMetaData::PropertyInfo::fromName(name: QString::fromUtf8(ba: property)); |
| 83 | if (pi.property() == KFileMetaData::Property::Empty) { |
| 84 | return {QByteArray(), QMetaType::UnknownType}; |
| 85 | } |
| 86 | int propPrefix = static_cast<int>(pi.property()); |
| 87 | return {QByteArray('X' + QByteArray::number(propPrefix) + '-'), pi.valueType()}; |
| 88 | } |
| 89 | } |
| 90 | |
| 91 | EngineQuery constructEqualsQuery(const QByteArray& prefix, const QString& value) |
| 92 | { |
| 93 | // We use the TermGenerator to normalize the words in the value and to |
| 94 | // split it into other words. If we split the words, we then add them as a |
| 95 | // phrase query. |
| 96 | const QByteArrayList terms = TermGenerator::termList(text: value); |
| 97 | |
| 98 | QVector<EngineQuery> queries; |
| 99 | queries.reserve(asize: terms.size()); |
| 100 | for (const QByteArray& term : terms) { |
| 101 | QByteArray arr = prefix + term; |
| 102 | // FIXME - compatibility hack, to find truncated terms with old |
| 103 | // DBs, remove on next DB bump |
| 104 | if (arr.size() > 25) { |
| 105 | queries << EngineQuery(arr.left(n: 25), EngineQuery::StartsWith); |
| 106 | } else { |
| 107 | queries << EngineQuery(arr); |
| 108 | } |
| 109 | } |
| 110 | |
| 111 | if (queries.isEmpty()) { |
| 112 | return EngineQuery(); |
| 113 | } else if (queries.size() == 1) { |
| 114 | return queries.first(); |
| 115 | } else { |
| 116 | return EngineQuery(queries); |
| 117 | } |
| 118 | } |
| 119 | |
| 120 | EngineQuery constructContainsQuery(const QByteArray& prefix, const QString& value) |
| 121 | { |
| 122 | auto query = constructEqualsQuery(prefix, value); |
| 123 | if (query.op() == EngineQuery::Equal) { |
| 124 | if (query.term().size() >= 3) { |
| 125 | query.setOp(EngineQuery::StartsWith); |
| 126 | } |
| 127 | } |
| 128 | return query; |
| 129 | } |
| 130 | |
| 131 | EngineQuery constructTypeQuery(const QString& value) |
| 132 | { |
| 133 | Q_ASSERT(!value.isEmpty()); |
| 134 | |
| 135 | KFileMetaData::TypeInfo ti = KFileMetaData::TypeInfo::fromName(name: value); |
| 136 | if (ti == KFileMetaData::Type::Empty) { |
| 137 | qCDebug(BALOO) << "Type" << value << "does not exist" ; |
| 138 | return EngineQuery(); |
| 139 | } |
| 140 | int num = static_cast<int>(ti.type()); |
| 141 | |
| 142 | return EngineQuery('T' + QByteArray::number(num)); |
| 143 | } |
| 144 | } // namespace |
| 145 | |
| 146 | SearchStore::SearchStore() |
| 147 | : m_db(nullptr) |
| 148 | { |
| 149 | m_db = globalDatabaseInstance(); |
| 150 | if (m_db->open(mode: Database::ReadOnlyDatabase) != Database::OpenResult::Success) { |
| 151 | m_db = nullptr; |
| 152 | } |
| 153 | } |
| 154 | |
| 155 | SearchStore::~SearchStore() |
| 156 | { |
| 157 | } |
| 158 | |
| 159 | // Return the result with-in [offset, offset + limit) |
| 160 | ResultList SearchStore::exec(const Term& term, uint offset, int limit, bool sortResults) |
| 161 | { |
| 162 | if (!m_db) { |
| 163 | return ResultList(); |
| 164 | } |
| 165 | |
| 166 | Transaction tr(m_db, Transaction::ReadOnly); |
| 167 | std::unique_ptr<PostingIterator> it(constructQuery(tr: &tr, term)); |
| 168 | if (!it) { |
| 169 | return ResultList(); |
| 170 | } |
| 171 | |
| 172 | if (sortResults) { |
| 173 | QVector<std::pair<quint64, quint32>> resultIds; |
| 174 | while (it->next()) { |
| 175 | quint64 id = it->docId(); |
| 176 | quint32 mtime = tr.documentTimeInfo(id).mTime; |
| 177 | resultIds << std::pair<quint64, quint32>{id, mtime}; |
| 178 | |
| 179 | Q_ASSERT(id > 0); |
| 180 | } |
| 181 | |
| 182 | // Not enough results within range, no need to sort. |
| 183 | if (offset >= static_cast<uint>(resultIds.size())) { |
| 184 | return ResultList(); |
| 185 | } |
| 186 | |
| 187 | auto compFunc = [](const std::pair<quint64, quint32>& lhs, |
| 188 | const std::pair<quint64, quint32>& rhs) { |
| 189 | return lhs.second > rhs.second; |
| 190 | }; |
| 191 | |
| 192 | std::sort(first: resultIds.begin(), last: resultIds.end(), comp: compFunc); |
| 193 | if (limit < 0) { |
| 194 | limit = resultIds.size(); |
| 195 | } |
| 196 | |
| 197 | ResultList results; |
| 198 | const uint end = qMin(a: static_cast<uint>(resultIds.size()), b: offset + static_cast<uint>(limit)); |
| 199 | results.reserve(n: end - offset); |
| 200 | for (uint i = offset; i < end; i++) { |
| 201 | const quint64 id = resultIds[i].first; |
| 202 | Result res{tr.documentUrl(id), id}; |
| 203 | |
| 204 | results.emplace_back(args&: res); |
| 205 | } |
| 206 | |
| 207 | return results; |
| 208 | } |
| 209 | else { |
| 210 | ResultList results; |
| 211 | uint ulimit = limit < 0 ? UINT_MAX : limit; |
| 212 | |
| 213 | while (offset && it->next()) { |
| 214 | offset--; |
| 215 | } |
| 216 | |
| 217 | while (ulimit && it->next()) { |
| 218 | const quint64 id = it->docId(); |
| 219 | Q_ASSERT(id > 0); |
| 220 | Result res{tr.documentUrl(id), id}; |
| 221 | Q_ASSERT(!res.filePath.isEmpty()); |
| 222 | |
| 223 | results.emplace_back(args&: res); |
| 224 | |
| 225 | ulimit--; |
| 226 | } |
| 227 | |
| 228 | return results; |
| 229 | } |
| 230 | } |
| 231 | |
| 232 | PostingIterator* SearchStore::constructQuery(Transaction* tr, const Term& term) |
| 233 | { |
| 234 | Q_ASSERT(tr); |
| 235 | |
| 236 | if (term.operation() == Term::And || term.operation() == Term::Or) { |
| 237 | const QList<Term> subTerms = term.subTerms(); |
| 238 | QVector<PostingIterator*> vec; |
| 239 | vec.reserve(asize: subTerms.size()); |
| 240 | |
| 241 | for (const Term& t : subTerms) { |
| 242 | auto iterator = constructQuery(tr, term: t); |
| 243 | // constructQuery returns a nullptr to signal an empty list |
| 244 | if (iterator) { |
| 245 | vec << iterator; |
| 246 | } else if (term.operation() == Term::And) { |
| 247 | return nullptr; |
| 248 | } |
| 249 | } |
| 250 | |
| 251 | if (vec.isEmpty()) { |
| 252 | return nullptr; |
| 253 | } else if (vec.size() == 1) { |
| 254 | return vec.takeFirst(); |
| 255 | } |
| 256 | |
| 257 | if (term.operation() == Term::And) { |
| 258 | return new AndPostingIterator(vec); |
| 259 | } else { |
| 260 | return new OrPostingIterator(vec); |
| 261 | } |
| 262 | } |
| 263 | |
| 264 | if (term.value().isNull()) { |
| 265 | return nullptr; |
| 266 | } |
| 267 | Q_ASSERT(term.value().isValid()); |
| 268 | Q_ASSERT(term.comparator() != Term::Auto); |
| 269 | Q_ASSERT(term.comparator() == Term::Contains ? term.value().typeId() == QMetaType::QString : true); |
| 270 | |
| 271 | const QVariant value = term.value(); |
| 272 | const QByteArray property = term.property().toLower().toUtf8(); |
| 273 | |
| 274 | if (property == "type" || property == "kind" ) { |
| 275 | EngineQuery q = constructTypeQuery(value: value.toString()); |
| 276 | return tr->postingIterator(query: q); |
| 277 | } |
| 278 | else if (property == "includefolder" ) { |
| 279 | const QByteArray folder = value.toString().toUtf8(); |
| 280 | |
| 281 | if (folder.isEmpty()) { |
| 282 | return nullptr; |
| 283 | } |
| 284 | if (!folder.startsWith(c: '/')) { |
| 285 | return nullptr; |
| 286 | } |
| 287 | |
| 288 | quint64 id = tr->documentId(path: folder); |
| 289 | if (!id) { |
| 290 | qCDebug(BALOO) << "Folder" << value.toString() << "not indexed" ; |
| 291 | return nullptr; |
| 292 | } |
| 293 | |
| 294 | return tr->docUrlIter(id); |
| 295 | } |
| 296 | else if (property == "modified" || property == "mtime" ) { |
| 297 | if (value.typeId() == QMetaType::QByteArray) { |
| 298 | // Used by Baloo::Query |
| 299 | QByteArray ba = value.toByteArray(); |
| 300 | Q_ASSERT(ba.size() >= 4); |
| 301 | |
| 302 | int year = ba.mid(index: 0, len: 4).toInt(); |
| 303 | int month = ba.mid(index: 4, len: 2).toInt(); |
| 304 | int day = ba.mid(index: 6, len: 2).toInt(); |
| 305 | |
| 306 | Q_ASSERT(year); |
| 307 | |
| 308 | // uses 0 to represent whole month or whole year |
| 309 | month = month >= 0 && month <= 12 ? month : 0; |
| 310 | day = day >= 0 && day <= 31 ? day : 0; |
| 311 | |
| 312 | QDate startDate(year, month ? month : 1, day ? day : 1); |
| 313 | QDate endDate(startDate); |
| 314 | |
| 315 | if (month == 0) { |
| 316 | endDate.setDate(year: endDate.year(), month: 12, day: 31); |
| 317 | } else if (day == 0) { |
| 318 | endDate.setDate(year: endDate.year(), month: endDate.month(), day: endDate.daysInMonth()); |
| 319 | } |
| 320 | |
| 321 | return tr->mTimeRangeIter(beginTime: startDate.startOfDay().toSecsSinceEpoch(), endTime: endDate.endOfDay().toSecsSinceEpoch()); |
| 322 | } else if (value.typeId() == QMetaType::QString) { |
| 323 | const QDateTime dt = value.toDateTime(); |
| 324 | QPair<quint32, quint32> timerange = calculateTimeRange(dt, com: term.comparator()); |
| 325 | if ((timerange.first == 0) && (timerange.second == 0)) { |
| 326 | return nullptr; |
| 327 | } |
| 328 | return tr->mTimeRangeIter(beginTime: timerange.first, endTime: timerange.second); |
| 329 | } else { |
| 330 | Q_ASSERT_X(0, "SearchStore::constructQuery" , "modified property must contain date/datetime values" ); |
| 331 | return nullptr; |
| 332 | } |
| 333 | } else if (property == "tag" ) { |
| 334 | if (term.comparator() == Term::Equal) { |
| 335 | const QByteArray prefix = "TAG-" ; |
| 336 | EngineQuery q = EngineQuery(prefix + value.toByteArray()); |
| 337 | return tr->postingIterator(query: q); |
| 338 | } else if (term.comparator() == Term::Contains) { |
| 339 | const QByteArray prefix = "TA" ; |
| 340 | EngineQuery q = constructEqualsQuery(prefix, value: value.toString()); |
| 341 | return tr->postingIterator(query: q); |
| 342 | } else { |
| 343 | Q_ASSERT(0); |
| 344 | return nullptr; |
| 345 | } |
| 346 | } else if (property == "" ) { |
| 347 | Term cterm(QStringLiteral("content" ), term.value(), term.comparator()); |
| 348 | Term fterm(QStringLiteral("filename" ), term.value(), term.comparator()); |
| 349 | return constructQuery(tr, term: Term{cterm, Term::Operation::Or, fterm}); |
| 350 | } |
| 351 | |
| 352 | QByteArray prefix; |
| 353 | QMetaType::Type valueType = QMetaType::QString; |
| 354 | if (!property.isEmpty()) { |
| 355 | std::tie(args&: prefix, args&: valueType) = propertyInfo(property); |
| 356 | if (valueType == QMetaType::UnknownType) { |
| 357 | return nullptr; |
| 358 | } |
| 359 | } |
| 360 | |
| 361 | auto com = term.comparator(); |
| 362 | if (com == Term::Contains && valueType == QMetaType::Int) { |
| 363 | com = Term::Equal; |
| 364 | } |
| 365 | if (com == Term::Contains) { |
| 366 | EngineQuery q = constructContainsQuery(prefix, value: value.toString()); |
| 367 | return tr->postingIterator(query: q); |
| 368 | } |
| 369 | |
| 370 | if (com == Term::Equal) { |
| 371 | EngineQuery q = constructEqualsQuery(prefix, value: value.toString()); |
| 372 | return tr->postingIterator(query: q); |
| 373 | } |
| 374 | |
| 375 | PostingDB::Comparator pcom; |
| 376 | if (com == Term::Greater || com == Term::GreaterEqual) { |
| 377 | pcom = PostingDB::GreaterEqual; |
| 378 | } else if (com == Term::Less || com == Term::LessEqual) { |
| 379 | pcom = PostingDB::LessEqual; |
| 380 | } |
| 381 | |
| 382 | // FIXME -- has to be kept in sync with the code from |
| 383 | // Baloo::Result::add |
| 384 | if (valueType == QMetaType::Int) { |
| 385 | qlonglong intVal = value.toLongLong(); |
| 386 | |
| 387 | if (term.comparator() == Term::Greater) { |
| 388 | intVal++; |
| 389 | } else if (term.comparator() == Term::Less) { |
| 390 | intVal--; |
| 391 | } |
| 392 | |
| 393 | return tr->postingCompIterator(prefix, value: intVal, com: pcom); |
| 394 | |
| 395 | } else if (valueType == QMetaType::Double) { |
| 396 | double dVal = value.toDouble(); |
| 397 | return tr->postingCompIterator(prefix, value: dVal, com: pcom); |
| 398 | |
| 399 | } else if (valueType == QMetaType::QDateTime) { |
| 400 | QDateTime dt = value.toDateTime(); |
| 401 | const QByteArray ba = dt.toString(format: Qt::ISODate).toUtf8(); |
| 402 | return tr->postingCompIterator(prefix, value: ba, com: pcom); |
| 403 | |
| 404 | } else { |
| 405 | qCDebug(BALOO) << "Comparison must be with an integer" ; |
| 406 | } |
| 407 | |
| 408 | return nullptr; |
| 409 | } |
| 410 | |
| 411 | } // namespace Baloo |
| 412 | |