| 1 | /* |
| 2 | This file is part of the KDE Baloo project. |
| 3 | SPDX-FileCopyrightText: 2015 Vishesh Handa <me@vhanda.in> |
| 4 | |
| 5 | SPDX-License-Identifier: LGPL-2.1-or-later |
| 6 | */ |
| 7 | |
| 8 | #include "postingdb.h" |
| 9 | #include "enginedebug.h" |
| 10 | #include "orpostingiterator.h" |
| 11 | #include "postingcodec.h" |
| 12 | #include "vectorpostingiterator.h" |
| 13 | |
| 14 | using namespace Baloo; |
| 15 | |
| 16 | PostingDB::PostingDB(MDB_dbi dbi, MDB_txn* txn) |
| 17 | : m_txn(txn) |
| 18 | , m_dbi(dbi) |
| 19 | { |
| 20 | Q_ASSERT(txn != nullptr); |
| 21 | Q_ASSERT(dbi != 0); |
| 22 | } |
| 23 | |
| 24 | PostingDB::~PostingDB() |
| 25 | { |
| 26 | } |
| 27 | |
| 28 | MDB_dbi PostingDB::create(MDB_txn* txn) |
| 29 | { |
| 30 | MDB_dbi dbi = 0; |
| 31 | int rc = mdb_dbi_open(txn, name: "postingdb" , MDB_CREATE, dbi: &dbi); |
| 32 | if (rc) { |
| 33 | qCWarning(ENGINE) << "PostingDB::create" << mdb_strerror(err: rc); |
| 34 | return 0; |
| 35 | } |
| 36 | |
| 37 | return dbi; |
| 38 | } |
| 39 | |
| 40 | MDB_dbi PostingDB::open(MDB_txn* txn) |
| 41 | { |
| 42 | MDB_dbi dbi = 0; |
| 43 | int rc = mdb_dbi_open(txn, name: "postingdb" , flags: 0, dbi: &dbi); |
| 44 | if (rc) { |
| 45 | qCWarning(ENGINE) << "PostingDB::open" << mdb_strerror(err: rc); |
| 46 | return 0; |
| 47 | } |
| 48 | |
| 49 | return dbi; |
| 50 | } |
| 51 | |
| 52 | void PostingDB::put(const QByteArray& term, const PostingList& list) |
| 53 | { |
| 54 | Q_ASSERT(!term.isEmpty()); |
| 55 | Q_ASSERT(!list.isEmpty()); |
| 56 | |
| 57 | MDB_val key; |
| 58 | key.mv_size = term.size(); |
| 59 | key.mv_data = static_cast<void*>(const_cast<char*>(term.constData())); |
| 60 | |
| 61 | QByteArray arr = PostingCodec::encode(list); |
| 62 | |
| 63 | MDB_val val; |
| 64 | val.mv_size = arr.size(); |
| 65 | val.mv_data = static_cast<void*>(arr.data()); |
| 66 | |
| 67 | int rc = mdb_put(txn: m_txn, dbi: m_dbi, key: &key, data: &val, flags: 0); |
| 68 | if (rc) { |
| 69 | qCWarning(ENGINE) << "PostingDB::put" << mdb_strerror(err: rc); |
| 70 | } |
| 71 | } |
| 72 | |
| 73 | PostingList PostingDB::get(const QByteArray& term) |
| 74 | { |
| 75 | Q_ASSERT(!term.isEmpty()); |
| 76 | |
| 77 | MDB_val key; |
| 78 | key.mv_size = term.size(); |
| 79 | key.mv_data = static_cast<void*>(const_cast<char*>(term.constData())); |
| 80 | |
| 81 | MDB_val val{.mv_size: 0, .mv_data: nullptr}; |
| 82 | int rc = mdb_get(txn: m_txn, dbi: m_dbi, key: &key, data: &val); |
| 83 | if (rc) { |
| 84 | if (rc != MDB_NOTFOUND) { |
| 85 | qCDebug(ENGINE) << "PostingDB::get" << term << mdb_strerror(err: rc); |
| 86 | } |
| 87 | return PostingList(); |
| 88 | } |
| 89 | |
| 90 | QByteArray arr = QByteArray::fromRawData(data: static_cast<char*>(val.mv_data), size: val.mv_size); |
| 91 | |
| 92 | return PostingCodec::decode(arr); |
| 93 | } |
| 94 | |
| 95 | void PostingDB::del(const QByteArray& term) |
| 96 | { |
| 97 | Q_ASSERT(!term.isEmpty()); |
| 98 | |
| 99 | MDB_val key; |
| 100 | key.mv_size = term.size(); |
| 101 | key.mv_data = static_cast<void*>(const_cast<char*>(term.constData())); |
| 102 | |
| 103 | int rc = mdb_del(txn: m_txn, dbi: m_dbi, key: &key, data: nullptr); |
| 104 | if (rc != 0 && rc != MDB_NOTFOUND) { |
| 105 | qCDebug(ENGINE) << "PostingDB::del" << term << mdb_strerror(err: rc); |
| 106 | } |
| 107 | } |
| 108 | |
| 109 | QVector< QByteArray > PostingDB::fetchTermsStartingWith(const QByteArray& term) |
| 110 | { |
| 111 | MDB_val key; |
| 112 | key.mv_size = term.size(); |
| 113 | key.mv_data = static_cast<void*>(const_cast<char*>(term.constData())); |
| 114 | |
| 115 | MDB_cursor* cursor; |
| 116 | int rc = mdb_cursor_open(txn: m_txn, dbi: m_dbi, cursor: &cursor); |
| 117 | if (rc) { |
| 118 | qCWarning(ENGINE) << "PostingDB::fetchTermsStartingWith" << mdb_strerror(err: rc); |
| 119 | return {}; |
| 120 | } |
| 121 | |
| 122 | QVector<QByteArray> terms; |
| 123 | rc = mdb_cursor_get(cursor, key: &key, data: nullptr, op: MDB_SET_RANGE); |
| 124 | while (rc == 0) { |
| 125 | const QByteArray arr(static_cast<char*>(key.mv_data), key.mv_size); |
| 126 | if (!arr.startsWith(bv: term)) { |
| 127 | break; |
| 128 | } |
| 129 | terms << arr; |
| 130 | rc = mdb_cursor_get(cursor, key: &key, data: nullptr, op: MDB_NEXT); |
| 131 | } |
| 132 | if (rc != MDB_NOTFOUND) { |
| 133 | qCDebug(ENGINE) << "PostingDB::fetchTermsStartingWith" << mdb_strerror(err: rc); |
| 134 | } |
| 135 | |
| 136 | mdb_cursor_close(cursor); |
| 137 | return terms; |
| 138 | } |
| 139 | |
| 140 | PostingIterator* PostingDB::iter(const QByteArray& term) |
| 141 | { |
| 142 | MDB_val key; |
| 143 | key.mv_size = term.size(); |
| 144 | key.mv_data = static_cast<void*>(const_cast<char*>(term.constData())); |
| 145 | |
| 146 | MDB_val val; |
| 147 | int rc = mdb_get(txn: m_txn, dbi: m_dbi, key: &key, data: &val); |
| 148 | if (rc) { |
| 149 | qCDebug(ENGINE) << "PostingDB::iter" << term << mdb_strerror(err: rc); |
| 150 | return nullptr; |
| 151 | } |
| 152 | |
| 153 | auto values = QByteArray::fromRawData(data: static_cast<char *>(val.mv_data), size: val.mv_size); |
| 154 | return new VectorPostingIterator(PostingCodec().decode(arr: values)); |
| 155 | } |
| 156 | |
| 157 | template <typename Validator> |
| 158 | PostingIterator* PostingDB::iter(const QByteArray& prefix, Validator validate) |
| 159 | { |
| 160 | Q_ASSERT(!prefix.isEmpty()); |
| 161 | |
| 162 | MDB_val key; |
| 163 | key.mv_size = prefix.size(); |
| 164 | key.mv_data = static_cast<void*>(const_cast<char*>(prefix.constData())); |
| 165 | |
| 166 | MDB_cursor* cursor; |
| 167 | int rc = mdb_cursor_open(txn: m_txn, dbi: m_dbi, cursor: &cursor); |
| 168 | |
| 169 | if (rc) { |
| 170 | qCWarning(ENGINE) << "PostingDB::regexpIter" << mdb_strerror(err: rc); |
| 171 | return nullptr; |
| 172 | } |
| 173 | |
| 174 | QVector<PostingIterator*> termIterators; |
| 175 | |
| 176 | MDB_val val; |
| 177 | rc = mdb_cursor_get(cursor, key: &key, data: &val, op: MDB_SET_RANGE); |
| 178 | while (rc == 0) { |
| 179 | const QByteArray arr(static_cast<char*>(key.mv_data), key.mv_size); |
| 180 | if (!arr.startsWith(bv: prefix)) { |
| 181 | break; |
| 182 | } |
| 183 | if (validate(arr)) { |
| 184 | auto values = QByteArray::fromRawData(data: static_cast<char *>(val.mv_data), size: val.mv_size); |
| 185 | termIterators << new VectorPostingIterator(PostingCodec().decode(arr: values)); |
| 186 | } |
| 187 | rc = mdb_cursor_get(cursor, key: &key, data: &val, op: MDB_NEXT); |
| 188 | } |
| 189 | |
| 190 | if (rc != 0 && rc != MDB_NOTFOUND) { |
| 191 | qCWarning(ENGINE) << "PostingDB::regexpIter" << mdb_strerror(err: rc); |
| 192 | } |
| 193 | |
| 194 | mdb_cursor_close(cursor); |
| 195 | if (termIterators.isEmpty()) { |
| 196 | return nullptr; |
| 197 | } |
| 198 | return new OrPostingIterator(termIterators); |
| 199 | } |
| 200 | |
| 201 | PostingIterator* PostingDB::prefixIter(const QByteArray& prefix) |
| 202 | { |
| 203 | auto validate = [] (const QByteArray& arr) { |
| 204 | Q_UNUSED(arr); |
| 205 | return true; |
| 206 | }; |
| 207 | return iter(prefix, validate); |
| 208 | } |
| 209 | |
| 210 | PostingIterator* PostingDB::regexpIter(const QRegularExpression& regexp, const QByteArray& prefix) |
| 211 | { |
| 212 | int prefixLen = prefix.length(); |
| 213 | auto validate = [®exp, prefixLen] (const QByteArray& arr) { |
| 214 | QString term = QString::fromUtf8(ba: arr.mid(index: prefixLen)); |
| 215 | return regexp.match(subject: term).hasMatch(); |
| 216 | }; |
| 217 | |
| 218 | return iter(prefix, validate); |
| 219 | } |
| 220 | |
| 221 | PostingIterator* PostingDB::compIter(const QByteArray& prefix, qlonglong comVal, PostingDB::Comparator com) |
| 222 | { |
| 223 | int prefixLen = prefix.length(); |
| 224 | auto validate = [prefixLen, comVal, com] (const QByteArray& arr) { |
| 225 | bool ok = false; |
| 226 | auto val = QByteArray::fromRawData(data: arr.constData() + prefixLen, size: arr.length() - prefixLen).toLongLong(ok: &ok); |
| 227 | return ok && ((com == LessEqual && val <= comVal) || (com == GreaterEqual && val >= comVal)); |
| 228 | }; |
| 229 | return iter(prefix, validate); |
| 230 | } |
| 231 | |
| 232 | PostingIterator* PostingDB::compIter(const QByteArray& prefix, double comVal, PostingDB::Comparator com) |
| 233 | { |
| 234 | int prefixLen = prefix.length(); |
| 235 | auto validate = [prefixLen, comVal, com] (const QByteArray& arr) { |
| 236 | bool ok = false; |
| 237 | auto val = QByteArray::fromRawData(data: arr.constData() + prefixLen, size: arr.length() - prefixLen).toDouble(ok: &ok); |
| 238 | return ok && ((com == LessEqual && val <= comVal) || |
| 239 | (com == GreaterEqual && val >= comVal)); |
| 240 | }; |
| 241 | return iter(prefix, validate); |
| 242 | } |
| 243 | |
| 244 | PostingIterator* PostingDB::compIter(const QByteArray& prefix, const QByteArray& comVal, PostingDB::Comparator com) |
| 245 | { |
| 246 | int prefixLen = prefix.length(); |
| 247 | auto validate = [prefixLen, comVal, com] (const QByteArray& arr) { |
| 248 | auto val = QByteArray::fromRawData(data: arr.constData() + prefixLen, size: arr.length() - prefixLen); |
| 249 | return ((com == LessEqual && val <= comVal) || |
| 250 | (com == GreaterEqual && val >= comVal)); |
| 251 | }; |
| 252 | return iter(prefix, validate); |
| 253 | } |
| 254 | |
| 255 | QMap<QByteArray, PostingList> PostingDB::toTestMap() const |
| 256 | { |
| 257 | MDB_cursor* cursor; |
| 258 | mdb_cursor_open(txn: m_txn, dbi: m_dbi, cursor: &cursor); |
| 259 | |
| 260 | MDB_val key = {.mv_size: 0, .mv_data: nullptr}; |
| 261 | MDB_val val; |
| 262 | |
| 263 | QMap<QByteArray, PostingList> map; |
| 264 | while (1) { |
| 265 | int rc = mdb_cursor_get(cursor, key: &key, data: &val, op: MDB_NEXT); |
| 266 | if (rc == MDB_NOTFOUND) { |
| 267 | break; |
| 268 | } |
| 269 | if (rc) { |
| 270 | qCDebug(ENGINE) << "PostingDB::toTestMap" << mdb_strerror(err: rc); |
| 271 | break; |
| 272 | } |
| 273 | |
| 274 | const QByteArray ba(static_cast<char*>(key.mv_data), key.mv_size); |
| 275 | const PostingList plist = PostingCodec::decode(arr: QByteArray(static_cast<char*>(val.mv_data), val.mv_size)); |
| 276 | map.insert(key: ba, value: plist); |
| 277 | } |
| 278 | |
| 279 | mdb_cursor_close(cursor); |
| 280 | return map; |
| 281 | } |
| 282 | |