data/FolderPredictor.cpp
| Line | Branch | Exec | Source |
|---|---|---|---|
| 1 | #include "data/FolderPredictor.h" | ||
| 2 | #include "data/DatabaseSecurity.h" | ||
| 3 | #include <QMutexLocker> | ||
| 4 | #include "data/MailCache.h" | ||
| 5 | |||
| 6 | #include <QLoggingCategory> | ||
| 7 | #include <QRegularExpression> | ||
| 8 | #include <QSqlError> | ||
| 9 | #include <QSqlQuery> | ||
| 10 | #include <QScopeGuard> | ||
| 11 | #include <QUuid> | ||
| 12 | #include <QtMath> | ||
| 13 | |||
| 14 | #include <algorithm> | ||
| 15 | #include <optional> | ||
| 16 | |||
| 17 |
4/7✓ Branch 2 → 3 taken 4 times.
✓ Branch 2 → 8 taken 31 times.
✓ Branch 4 → 5 taken 4 times.
✗ Branch 4 → 8 not taken.
✓ Branch 5 → 6 taken 4 times.
✗ Branch 10 → 11 not taken.
✗ Branch 10 → 12 not taken.
|
35 | Q_LOGGING_CATEGORY(lcFolderPredictor, "mailjd.folderpredictor") |
| 18 | |||
| 19 | // German + English stop words (common words that don't help classification) | ||
| 20 | const QSet<QString> FolderPredictor::s_stopWords = { | ||
| 21 | // German | ||
| 22 | "der", "die", "das", "und", "oder", "ein", "eine", "ist", "sind", "war", | ||
| 23 | "hat", "haben", "wird", "werden", "mit", "von", "für", "auf", "aus", "bei", | ||
| 24 | "nach", "über", "unter", "vor", "zum", "zur", "dem", "den", "des", "sich", | ||
| 25 | "nicht", "auch", "noch", "nur", "wie", "aber", "als", "wenn", "ich", "wir", | ||
| 26 | "sie", "ihr", "uns", "mir", "dir", "mein", "dein", "sein", "kann", "sehr", | ||
| 27 | // English | ||
| 28 | "the", "a", "an", "and", "or", "is", "are", "was", "has", "have", "will", | ||
| 29 | "with", "from", "for", "on", "at", "by", "to", "in", "of", "not", "but", | ||
| 30 | "this", "that", "it", "be", "do", "if", "so", "no", "up", "out", "we", | ||
| 31 | "you", "he", "she", "my", "your", "his", "her", "our", "can", | ||
| 32 | // Common mail prefixes (already stripped, but just in case) | ||
| 33 | "re", "fwd", "aw", "wg", | ||
| 34 | }; | ||
| 35 | |||
| 36 |
1/1✓ Branch 4 → 5 taken 1175 times.
|
1175 | FolderPredictor::FolderPredictor(QObject *parent) : QObject(parent) {} |
| 37 | |||
| 38 | 1184 | FolderPredictor::~FolderPredictor() { | |
| 39 | 1175 | close(); | |
| 40 | 1184 | } | |
| 41 | |||
| 42 | 861 | bool FolderPredictor::open(const QString &dbPath) { | |
| 43 |
1/1✓ Branch 2 → 3 taken 861 times.
|
861 | close(); |
| 44 |
3/3✓ Branch 3 → 4 taken 861 times.
✓ Branch 4 → 5 taken 5 times.
✓ Branch 4 → 17 taken 856 times.
|
861 | if (!DatabaseSecurity::preparePath(dbPath)) { |
| 45 |
4/4✓ Branch 5 → 6 taken 5 times.
✓ Branch 9 → 10 taken 5 times.
✓ Branch 15 → 7 taken 5 times.
✓ Branch 15 → 16 taken 5 times.
|
10 | qCWarning(lcFolderPredictor) |
| 46 |
2/2✓ Branch 10 → 11 taken 5 times.
✓ Branch 11 → 12 taken 5 times.
|
5 | << "failed to create private database file" << dbPath; |
| 47 | 5 | return false; | |
| 48 | } | ||
| 49 | m_connectionName = | ||
| 50 |
3/3✓ Branch 17 → 18 taken 856 times.
✓ Branch 18 → 19 taken 856 times.
✓ Branch 22 → 23 taken 856 times.
|
1712 | QStringLiteral("FolderPredictor_") + QUuid::createUuid().toString(); |
| 51 |
2/2✓ Branch 31 → 32 taken 856 times.
✓ Branch 32 → 33 taken 856 times.
|
1712 | m_db = QSqlDatabase::addDatabase(QStringLiteral("QSQLITE"), m_connectionName); |
| 52 |
1/1✓ Branch 36 → 37 taken 856 times.
|
856 | m_db.setDatabaseName(dbPath); |
| 53 | |||
| 54 |
2/3✓ Branch 37 → 38 taken 856 times.
✗ Branch 38 → 39 not taken.
✓ Branch 38 → 57 taken 856 times.
|
856 | if (!m_db.open()) { |
| 55 | ✗ | qCWarning(lcFolderPredictor) << "failed to open" << dbPath << ':' // GCOVR_EXCL_BR_LINE — Qt logging-macro cleanup only; database result remains measured. | |
| 56 | ✗ | << m_db.lastError().text(); | |
| 57 | ✗ | return false; | |
| 58 | } | ||
| 59 |
2/3✓ Branch 57 → 58 taken 856 times.
✗ Branch 58 → 59 not taken.
✓ Branch 58 → 72 taken 856 times.
|
856 | if (!DatabaseSecurity::restrictExistingFile(dbPath)) { |
| 60 | ✗ | qCWarning(lcFolderPredictor) // GCOVR_EXCL_BR_LINE — Qt logging-macro cleanup only; security result remains measured. | |
| 61 | ✗ | << "failed to restrict database permissions for" << dbPath; | |
| 62 | ✗ | m_db.close(); | |
| 63 | ✗ | return false; | |
| 64 | } | ||
| 65 | |||
| 66 | // WAL for fast writes | ||
| 67 |
1/1✓ Branch 72 → 73 taken 856 times.
|
856 | QSqlQuery q(m_db); |
| 68 |
1/1✓ Branch 76 → 77 taken 856 times.
|
856 | q.exec(QStringLiteral("PRAGMA journal_mode=WAL")); |
| 69 |
1/1✓ Branch 82 → 83 taken 856 times.
|
856 | q.exec(QStringLiteral("PRAGMA synchronous=NORMAL")); |
| 70 | // T-619/SEC-19: Prevent SQLITE_BUSY errors during concurrent access | ||
| 71 |
1/1✓ Branch 88 → 89 taken 856 times.
|
856 | q.exec(QStringLiteral("PRAGMA busy_timeout=5000")); |
| 72 | |||
| 73 |
1/1✓ Branch 91 → 92 taken 856 times.
|
856 | return createSchema(); |
| 74 | 856 | } | |
| 75 | |||
| 76 | 317 | bool FolderPredictor::openReadOnly(const QString &dbPath, std::shared_ptr<StatisticsCache> cache) { | |
| 77 | 317 | close(); | |
| 78 |
1/2✗ Branch 4 → 5 not taken.
✓ Branch 4 → 6 taken 317 times.
|
317 | if (dbPath.isEmpty()) return false; |
| 79 | 317 | m_path = dbPath; | |
| 80 | 317 | m_statisticsCache = std::move(cache); | |
| 81 |
3/3✓ Branch 9 → 10 taken 317 times.
✓ Branch 10 → 11 taken 317 times.
✓ Branch 14 → 15 taken 317 times.
|
634 | m_connectionName = QStringLiteral("FolderPredictionRead_") + QUuid::createUuid().toString(); |
| 82 |
2/2✓ Branch 23 → 24 taken 317 times.
✓ Branch 24 → 25 taken 317 times.
|
634 | m_db = QSqlDatabase::addDatabase(QStringLiteral("QSQLITE"), m_connectionName); |
| 83 | 317 | m_db.setDatabaseName(dbPath); | |
| 84 |
1/1✓ Branch 32 → 33 taken 317 times.
|
634 | m_db.setConnectOptions(QStringLiteral("QSQLITE_OPEN_READONLY;QSQLITE_BUSY_TIMEOUT=250")); |
| 85 |
2/2✓ Branch 36 → 37 taken 1 time.
✓ Branch 36 → 39 taken 316 times.
|
317 | if (!m_db.open()) { close(); return false; } |
| 86 | 316 | bool valid = false; | |
| 87 | { | ||
| 88 |
1/1✓ Branch 39 → 40 taken 316 times.
|
316 | QSqlQuery query(m_db); |
| 89 |
1/6✓ Branch 76 → 77 taken 316 times.
✗ Branch 76 → 78 not taken.
✗ Branch 133 → 134 not taken.
✗ Branch 133 → 135 not taken.
✗ Branch 137 → 138 not taken.
✗ Branch 137 → 139 not taken.
|
632 | valid = query.exec(QStringLiteral("SELECT token,folder,count FROM token_counts LIMIT 0")) && |
| 90 |
5/11✓ Branch 48 → 49 taken 315 times.
✓ Branch 49 → 50 taken 315 times.
✗ Branch 49 → 61 not taken.
✓ Branch 72 → 73 taken 315 times.
✓ Branch 72 → 74 taken 1 time.
✓ Branch 74 → 75 taken 316 times.
✗ Branch 74 → 76 not taken.
✗ Branch 125 → 126 not taken.
✗ Branch 125 → 127 not taken.
✗ Branch 129 → 130 not taken.
✗ Branch 129 → 131 not taken.
|
631 | query.exec(QStringLiteral("SELECT folder,doc_count FROM folder_doc_counts LIMIT 0")) && |
| 91 |
9/14✓ Branch 43 → 44 taken 316 times.
✓ Branch 44 → 45 taken 315 times.
✓ Branch 44 → 61 taken 1 time.
✓ Branch 53 → 54 taken 315 times.
✓ Branch 54 → 55 taken 315 times.
✗ Branch 54 → 61 not taken.
✓ Branch 68 → 69 taken 315 times.
✓ Branch 68 → 70 taken 1 time.
✓ Branch 70 → 71 taken 315 times.
✓ Branch 70 → 72 taken 1 time.
✗ Branch 117 → 118 not taken.
✗ Branch 117 → 119 not taken.
✗ Branch 121 → 122 not taken.
✗ Branch 121 → 123 not taken.
|
1578 | query.exec(QStringLiteral("SELECT key,value FROM meta LIMIT 0")) && |
| 92 |
8/13✓ Branch 58 → 59 taken 315 times.
✓ Branch 59 → 60 taken 315 times.
✗ Branch 59 → 61 not taken.
✓ Branch 62 → 63 taken 315 times.
✓ Branch 62 → 64 taken 1 time.
✓ Branch 64 → 65 taken 315 times.
✓ Branch 64 → 66 taken 1 time.
✓ Branch 66 → 67 taken 315 times.
✓ Branch 66 → 68 taken 1 time.
✗ Branch 109 → 110 not taken.
✗ Branch 109 → 111 not taken.
✗ Branch 113 → 114 not taken.
✗ Branch 113 → 115 not taken.
|
631 | query.exec(QStringLiteral("BEGIN")); |
| 93 | 316 | } | |
| 94 |
2/2✓ Branch 79 → 80 taken 1 time.
✓ Branch 79 → 81 taken 315 times.
|
316 | if (!valid) close(); |
| 95 | 316 | m_readOnly = valid; | |
| 96 | 316 | return valid; | |
| 97 | } | ||
| 98 | |||
| 99 | 4318 | FolderSuggestion FolderPredictor::suggestions(const MailHeader &header, | |
| 100 | const QString &junkFolder, QString *error) const { | ||
| 101 | // F-75 (SPRINT-104): the same direct-evidence guard predict() applies — | ||
| 102 | // a sender no folder has a from-token for gets NO suggestion, instead of | ||
| 103 | // the least-trained folder with near-certain softmax confidence (the | ||
| 104 | // smoothing favours the folder with the fewest total tokens). Spam | ||
| 105 | // classification is server metadata, independent of token evidence: an | ||
| 106 | // unknown spam-flagged sender still suggests the Junk folder. | ||
| 107 | // S106-199: a mail without a readable sender address has no evidence at | ||
| 108 | // all — it used to skip the guard and get a suggestion from subject/to | ||
| 109 | // tokens alone (the exact F-75 mechanism, reachable with an empty From). | ||
| 110 |
1/1✓ Branch 2 → 3 taken 4318 times.
|
4318 | const QString senderEmail = extractEmail(header.from); |
| 111 | 4318 | bool known = false; | |
| 112 |
2/2✓ Branch 4 → 5 taken 4312 times.
✓ Branch 4 → 31 taken 6 times.
|
4318 | if (!senderEmail.isEmpty()) { |
| 113 |
1/1✓ Branch 5 → 6 taken 4312 times.
|
4312 | QSqlQuery evidence(m_db); |
| 114 |
1/1✓ Branch 9 → 10 taken 4312 times.
|
4312 | evidence.prepare(QStringLiteral( |
| 115 | "SELECT 1 FROM token_counts WHERE token = ? LIMIT 1")); | ||
| 116 |
2/2✓ Branch 16 → 17 taken 4312 times.
✓ Branch 18 → 19 taken 4312 times.
|
4312 | evidence.addBindValue(QStringLiteral("from:") + senderEmail); |
| 117 |
5/6✓ Branch 23 → 24 taken 4312 times.
✓ Branch 24 → 25 taken 4312 times.
✗ Branch 24 → 28 not taken.
✓ Branch 25 → 26 taken 4312 times.
✓ Branch 26 → 27 taken 2724 times.
✓ Branch 26 → 28 taken 1588 times.
|
4312 | known = evidence.exec() && evidence.next(); |
| 118 | 4312 | } | |
| 119 |
2/2✓ Branch 31 → 32 taken 1594 times.
✓ Branch 31 → 44 taken 2724 times.
|
4318 | if (!known) { |
| 120 |
6/8✓ Branch 32 → 33 taken 3 times.
✓ Branch 32 → 38 taken 1591 times.
✓ Branch 34 → 35 taken 3 times.
✗ Branch 34 → 38 not taken.
✓ Branch 36 → 37 taken 3 times.
✗ Branch 36 → 38 not taken.
✓ Branch 39 → 40 taken 3 times.
✓ Branch 39 → 43 taken 1591 times.
|
1594 | if (header.isSpam && !junkFolder.isEmpty() && junkFolder.size() <= 4096) |
| 121 | 3 | return FolderSuggestion{{junkFolder, 0.95}, {}}; | |
| 122 | 1591 | return FolderSuggestion{}; | |
| 123 | } | ||
| 124 | 2724 | return FolderSuggestion::fromScores(predictTop(header.from, header.subject, header.to, 2, error), | |
| 125 |
2/2✓ Branch 44 → 45 taken 2724 times.
✓ Branch 45 → 46 taken 2724 times.
|
2724 | header.isSpam, junkFolder); |
| 126 | 4318 | } | |
| 127 | |||
| 128 | 2390 | void FolderPredictor::close() { | |
| 129 | 2390 | m_statistics = {}; | |
| 130 | 2390 | m_readOnly = false; | |
| 131 | 2390 | m_path.clear(); | |
| 132 | 2390 | m_statisticsCache.reset(); | |
| 133 |
2/2✓ Branch 7 → 8 taken 1172 times.
✓ Branch 7 → 9 taken 1218 times.
|
2390 | if (m_db.isOpen()) { |
| 134 | 1172 | m_db.close(); | |
| 135 | } | ||
| 136 |
2/2✓ Branch 10 → 11 taken 1173 times.
✓ Branch 10 → 19 taken 1217 times.
|
2390 | if (!m_connectionName.isEmpty()) { |
| 137 | // Must drop the QSqlDatabase handle before removeDatabase — | ||
| 138 | // otherwise the member still holds a reference and Qt warns | ||
| 139 | // "connection still in use" (classic Qt gotcha). | ||
| 140 | 1173 | QString connName = m_connectionName; | |
| 141 | 1173 | m_connectionName.clear(); | |
| 142 |
2/2✓ Branch 13 → 14 taken 1173 times.
✓ Branch 14 → 15 taken 1173 times.
|
1173 | m_db = QSqlDatabase(); // Release reference |
| 143 |
1/1✓ Branch 16 → 17 taken 1173 times.
|
1173 | QSqlDatabase::removeDatabase(connName); |
| 144 | 1173 | } | |
| 145 | 2390 | } | |
| 146 | |||
| 147 | 6 | bool FolderPredictor::isOpen() const { | |
| 148 | 6 | return m_db.isOpen(); | |
| 149 | } | ||
| 150 | |||
| 151 | 856 | bool FolderPredictor::createSchema() { | |
| 152 |
1/1✓ Branch 2 → 3 taken 856 times.
|
856 | QSqlQuery q(m_db); |
| 153 | |||
| 154 | // Token counts: how often a token was seen in a specific folder | ||
| 155 |
1/1✓ Branch 6 → 7 taken 856 times.
|
856 | bool ok = q.exec(QStringLiteral( |
| 156 | "CREATE TABLE IF NOT EXISTS token_counts (" | ||
| 157 | " id INTEGER PRIMARY KEY AUTOINCREMENT," | ||
| 158 | " token TEXT NOT NULL," | ||
| 159 | " folder TEXT NOT NULL," | ||
| 160 | " count INTEGER NOT NULL DEFAULT 1," | ||
| 161 | " UNIQUE(token, folder)" | ||
| 162 | ")")); | ||
| 163 |
2/2✓ Branch 9 → 10 taken 1 time.
✓ Branch 9 → 26 taken 855 times.
|
856 | if (!ok) { |
| 164 | 2 | qCWarning(lcFolderPredictor) // GCOVR_EXCL_BR_LINE — Qt logging-macro cleanup only; schema result remains measured. | |
| 165 |
4/4✓ Branch 15 → 16 taken 1 time.
✓ Branch 16 → 17 taken 1 time.
✓ Branch 17 → 18 taken 1 time.
✓ Branch 18 → 19 taken 1 time.
|
1 | << "create token_counts:" << q.lastError().text(); |
| 166 | 1 | return false; | |
| 167 | } | ||
| 168 | |||
| 169 |
1/1✓ Branch 29 → 30 taken 855 times.
|
855 | q.exec(QStringLiteral( |
| 170 | "CREATE INDEX IF NOT EXISTS idx_token ON token_counts(token)")); | ||
| 171 |
1/1✓ Branch 35 → 36 taken 855 times.
|
855 | q.exec(QStringLiteral( |
| 172 | "CREATE INDEX IF NOT EXISTS idx_folder ON token_counts(folder)")); | ||
| 173 | |||
| 174 |
1/1✓ Branch 41 → 42 taken 855 times.
|
855 | ok = q.exec(QStringLiteral( |
| 175 | "CREATE TABLE IF NOT EXISTS trained_messages (" | ||
| 176 | " account TEXT NOT NULL," | ||
| 177 | " folder TEXT NOT NULL," | ||
| 178 | " uid INTEGER NOT NULL," | ||
| 179 | " uid_validity INTEGER NOT NULL DEFAULT 0," | ||
| 180 | " PRIMARY KEY(account, folder, uid)" | ||
| 181 | ")")); | ||
| 182 |
1/2✗ Branch 44 → 45 not taken.
✓ Branch 44 → 61 taken 855 times.
|
855 | if (!ok) { |
| 183 | ✗ | qCWarning(lcFolderPredictor) // GCOVR_EXCL_BR_LINE — Qt logging-macro cleanup only; schema result remains measured. | |
| 184 | ✗ | << "create trained_messages:" << q.lastError().text(); | |
| 185 | ✗ | return false; | |
| 186 | } | ||
| 187 | |||
| 188 | // A defaulted column preserves the old ledger and learned counts without | ||
| 189 | // copying the table. Legacy identities adopt their first known epoch below. | ||
| 190 |
2/3✓ Branch 64 → 65 taken 855 times.
✗ Branch 67 → 68 not taken.
✓ Branch 67 → 69 taken 855 times.
|
855 | if (!q.exec(QStringLiteral("PRAGMA table_info(trained_messages)"))) return false; |
| 191 | 855 | bool hasEpoch = false; | |
| 192 |
5/5✓ Branch 73 → 74 taken 3419 times.
✓ Branch 74 → 75 taken 3419 times.
✓ Branch 81 → 82 taken 4274 times.
✓ Branch 82 → 70 taken 3419 times.
✓ Branch 82 → 83 taken 855 times.
|
4274 | while (q.next()) hasEpoch |= q.value(1).toString() == QStringLiteral("uid_validity"); |
| 193 |
1/1✓ Branch 83 → 84 taken 855 times.
|
855 | q.finish(); |
| 194 |
9/15✓ Branch 84 → 85 taken 1 time.
✓ Branch 84 → 91 taken 854 times.
✓ Branch 88 → 89 taken 1 time.
✗ Branch 89 → 90 not taken.
✓ Branch 89 → 91 taken 1 time.
✓ Branch 92 → 93 taken 1 time.
✓ Branch 92 → 94 taken 854 times.
✓ Branch 94 → 95 taken 1 time.
✓ Branch 94 → 96 taken 854 times.
✗ Branch 96 → 97 not taken.
✓ Branch 96 → 113 taken 855 times.
✗ Branch 316 → 317 not taken.
✗ Branch 316 → 318 not taken.
✗ Branch 320 → 321 not taken.
✗ Branch 320 → 322 not taken.
|
856 | if (!hasEpoch && !q.exec(QStringLiteral( |
| 195 | "ALTER TABLE trained_messages ADD COLUMN uid_validity INTEGER NOT NULL DEFAULT 0"))) { | ||
| 196 | ✗ | qCWarning(lcFolderPredictor) << "Cannot add training mailbox epoch:" << q.lastError().text(); | |
| 197 | ✗ | return false; | |
| 198 | } | ||
| 199 | |||
| 200 |
2/3✓ Branch 116 → 117 taken 855 times.
✗ Branch 119 → 120 not taken.
✓ Branch 119 → 121 taken 855 times.
|
855 | if (!q.exec(QStringLiteral( |
| 201 | "CREATE TABLE IF NOT EXISTS completed_training_folders (" | ||
| 202 | "account TEXT NOT NULL,folder TEXT NOT NULL,folder_id INTEGER NOT NULL," | ||
| 203 | "uid_validity INTEGER NOT NULL,revision TEXT NOT NULL," | ||
| 204 | ✗ | "PRIMARY KEY(account,folder))"))) return false; | |
| 205 | |||
| 206 | // Folder document counts: how many mails were moved to each folder | ||
| 207 |
1/1✓ Branch 124 → 125 taken 855 times.
|
855 | ok = q.exec(QStringLiteral( |
| 208 | "CREATE TABLE IF NOT EXISTS folder_doc_counts (" | ||
| 209 | " folder TEXT PRIMARY KEY," | ||
| 210 | " doc_count INTEGER NOT NULL DEFAULT 0" | ||
| 211 | ")")); | ||
| 212 |
1/2✗ Branch 127 → 128 not taken.
✓ Branch 127 → 144 taken 855 times.
|
855 | if (!ok) { |
| 213 | ✗ | qCWarning(lcFolderPredictor) | |
| 214 | ✗ | << "create folder_doc_counts:" << q.lastError().text(); | |
| 215 | ✗ | return false; | |
| 216 | } | ||
| 217 | |||
| 218 | // Metadata (total_docs, pretrained flag) | ||
| 219 |
1/1✓ Branch 147 → 148 taken 855 times.
|
855 | ok = q.exec(QStringLiteral( |
| 220 | "CREATE TABLE IF NOT EXISTS meta (" | ||
| 221 | " key TEXT PRIMARY KEY," | ||
| 222 | " value TEXT" | ||
| 223 | ")")); | ||
| 224 |
1/2✗ Branch 150 → 151 not taken.
✓ Branch 150 → 167 taken 855 times.
|
855 | if (!ok) { |
| 225 | ✗ | qCWarning(lcFolderPredictor) << "create meta:" << q.lastError().text(); // GCOVR_EXCL_BR_LINE — Qt logging-macro cleanup only; schema result remains measured. | |
| 226 | ✗ | return false; | |
| 227 | } | ||
| 228 | |||
| 229 | // Initialize total_docs if not present | ||
| 230 |
1/1✓ Branch 170 → 171 taken 855 times.
|
855 | q.exec(QStringLiteral( |
| 231 | "INSERT OR IGNORE INTO meta (key, value) VALUES ('total_docs', '0')")); | ||
| 232 | // F100-50: databases from before the statistics revision get one here; | ||
| 233 | // until then readers simply do not cache their statistics. | ||
| 234 |
1/1✓ Branch 176 → 177 taken 855 times.
|
855 | q.prepare(QStringLiteral("INSERT OR IGNORE INTO meta (key, value) VALUES ('statistics_revision', ?)")); |
| 235 |
3/3✓ Branch 180 → 181 taken 855 times.
✓ Branch 181 → 182 taken 855 times.
✓ Branch 183 → 184 taken 855 times.
|
855 | q.addBindValue(QUuid::createUuid().toString(QUuid::WithoutBraces)); |
| 236 |
1/1✓ Branch 186 → 187 taken 855 times.
|
855 | q.exec(); |
| 237 | |||
| 238 | // Schema version check — if algorithm changed, force re-training | ||
| 239 | static constexpr int kSchemaVersion = 4; | ||
| 240 |
1/1✓ Branch 190 → 191 taken 855 times.
|
855 | q.exec(QStringLiteral("SELECT value FROM meta WHERE key = 'schema_version'")); |
| 241 | 855 | int storedVersion = 0; | |
| 242 |
3/3✓ Branch 193 → 194 taken 855 times.
✓ Branch 194 → 195 taken 52 times.
✓ Branch 194 → 199 taken 803 times.
|
855 | if (q.next()) |
| 243 |
2/2✓ Branch 195 → 196 taken 52 times.
✓ Branch 196 → 197 taken 52 times.
|
52 | storedVersion = q.value(0).toInt(); |
| 244 | |||
| 245 |
2/2✓ Branch 199 → 200 taken 804 times.
✓ Branch 199 → 250 taken 51 times.
|
855 | if (storedVersion != kSchemaVersion) { |
| 246 | // Algorithm changed — reset all training data | ||
| 247 |
1/1✓ Branch 203 → 204 taken 804 times.
|
804 | q.exec(QStringLiteral("DELETE FROM token_counts")); |
| 248 |
1/1✓ Branch 209 → 210 taken 804 times.
|
804 | q.exec(QStringLiteral("DELETE FROM trained_messages")); |
| 249 |
1/1✓ Branch 215 → 216 taken 804 times.
|
804 | q.exec(QStringLiteral("DELETE FROM completed_training_folders")); |
| 250 |
1/1✓ Branch 221 → 222 taken 804 times.
|
804 | q.exec(QStringLiteral("DELETE FROM folder_doc_counts")); |
| 251 |
1/1✓ Branch 227 → 228 taken 804 times.
|
804 | q.exec(QStringLiteral( |
| 252 | "UPDATE meta SET value = '0' WHERE key = 'total_docs'")); | ||
| 253 |
1/1✓ Branch 233 → 234 taken 804 times.
|
804 | q.exec(QStringLiteral("DELETE FROM meta WHERE key = 'pretrained_account'")); |
| 254 |
1/1✓ Branch 236 → 237 taken 804 times.
|
804 | touchStatisticsRevision(); |
| 255 |
1/1✓ Branch 240 → 241 taken 804 times.
|
804 | q.prepare(QStringLiteral( |
| 256 | "INSERT OR REPLACE INTO meta (key, value) VALUES ('schema_version', ?)")); | ||
| 257 |
2/2✓ Branch 244 → 245 taken 804 times.
✓ Branch 246 → 247 taken 804 times.
|
804 | q.bindValue(0, QString::number(kSchemaVersion)); |
| 258 |
1/1✓ Branch 249 → 250 taken 804 times.
|
804 | q.exec(); |
| 259 | } | ||
| 260 | |||
| 261 | 855 | return true; | |
| 262 | 856 | } | |
| 263 | |||
| 264 | 6 | void FolderPredictor::resetDatabase() { | |
| 265 |
3/3✓ Branch 2 → 3 taken 6 times.
✓ Branch 3 → 4 taken 1 time.
✓ Branch 3 → 5 taken 5 times.
|
6 | if (!m_db.isOpen()) |
| 266 | 1 | return; | |
| 267 | |||
| 268 |
1/1✓ Branch 5 → 6 taken 5 times.
|
5 | m_db.transaction(); |
| 269 |
1/1✓ Branch 6 → 7 taken 5 times.
|
5 | QSqlQuery q(m_db); |
| 270 |
1/1✓ Branch 10 → 11 taken 5 times.
|
5 | q.exec(QStringLiteral("DELETE FROM token_counts")); |
| 271 |
1/1✓ Branch 16 → 17 taken 5 times.
|
5 | q.exec(QStringLiteral("DELETE FROM trained_messages")); |
| 272 |
1/1✓ Branch 22 → 23 taken 5 times.
|
5 | q.exec(QStringLiteral("DELETE FROM completed_training_folders")); |
| 273 |
1/1✓ Branch 28 → 29 taken 5 times.
|
5 | q.exec(QStringLiteral("DELETE FROM folder_doc_counts")); |
| 274 |
1/1✓ Branch 34 → 35 taken 5 times.
|
5 | q.exec(QStringLiteral( |
| 275 | "UPDATE meta SET value = '0' WHERE key = 'total_docs'")); | ||
| 276 |
1/1✓ Branch 40 → 41 taken 5 times.
|
5 | q.exec(QStringLiteral("DELETE FROM meta WHERE key = 'pretrained_account'")); |
| 277 |
1/1✓ Branch 43 → 44 taken 5 times.
|
5 | touchStatisticsRevision(); |
| 278 |
1/1✓ Branch 44 → 45 taken 5 times.
|
5 | m_db.commit(); |
| 279 | 5 | } | |
| 280 | |||
| 281 | 5840 | bool FolderPredictor::touchStatisticsRevision() { | |
| 282 |
1/1✓ Branch 2 → 3 taken 5840 times.
|
5840 | QSqlQuery q(m_db); |
| 283 |
1/1✓ Branch 6 → 7 taken 5840 times.
|
5840 | q.prepare(QStringLiteral("INSERT OR REPLACE INTO meta (key, value) VALUES ('statistics_revision', ?)")); |
| 284 |
3/3✓ Branch 10 → 11 taken 5840 times.
✓ Branch 11 → 12 taken 5840 times.
✓ Branch 13 → 14 taken 5840 times.
|
5840 | q.addBindValue(QUuid::createUuid().toString(QUuid::WithoutBraces)); |
| 285 |
3/3✓ Branch 16 → 17 taken 5840 times.
✓ Branch 17 → 18 taken 5835 times.
✓ Branch 17 → 19 taken 5 times.
|
5840 | if (q.exec()) return true; |
| 286 |
8/8✓ Branch 19 → 20 taken 5 times.
✓ Branch 23 → 24 taken 5 times.
✓ Branch 24 → 25 taken 5 times.
✓ Branch 25 → 26 taken 5 times.
✓ Branch 26 → 27 taken 5 times.
✓ Branch 27 → 28 taken 5 times.
✓ Branch 33 → 21 taken 5 times.
✓ Branch 33 → 34 taken 5 times.
|
10 | qCWarning(lcFolderPredictor) << "statistics revision not stored:" << q.lastError().text(); |
| 287 | 5 | return false; | |
| 288 | 5840 | } | |
| 289 | |||
| 290 | // T-287: Number of training documents for a specific folder | ||
| 291 | 41 | int FolderPredictor::documentsForFolder(const QString &folderPath) const { | |
| 292 |
3/3✓ Branch 2 → 3 taken 41 times.
✓ Branch 3 → 4 taken 2 times.
✓ Branch 3 → 5 taken 39 times.
|
41 | if (!m_db.isOpen()) |
| 293 | 2 | return 0; | |
| 294 | |||
| 295 |
1/1✓ Branch 5 → 6 taken 39 times.
|
39 | QSqlQuery q(m_db); |
| 296 |
1/1✓ Branch 9 → 10 taken 39 times.
|
39 | q.prepare(QStringLiteral( |
| 297 | "SELECT doc_count FROM folder_doc_counts WHERE folder = :folder")); | ||
| 298 |
1/1✓ Branch 17 → 18 taken 39 times.
|
78 | q.bindValue(QStringLiteral(":folder"), folderPath); |
| 299 |
8/8✓ Branch 21 → 22 taken 39 times.
✓ Branch 22 → 23 taken 36 times.
✓ Branch 22 → 26 taken 3 times.
✓ Branch 23 → 24 taken 36 times.
✓ Branch 24 → 25 taken 23 times.
✓ Branch 24 → 26 taken 13 times.
✓ Branch 27 → 28 taken 23 times.
✓ Branch 27 → 33 taken 16 times.
|
39 | if (q.exec() && q.next()) |
| 300 |
2/2✓ Branch 28 → 29 taken 23 times.
✓ Branch 29 → 30 taken 23 times.
|
23 | return q.value(0).toInt(); |
| 301 | 16 | return 0; | |
| 302 | 39 | } | |
| 303 | |||
| 304 | // T-287: Whether a folder has enough training data for reliable predictions | ||
| 305 | 2 | bool FolderPredictor::hasSufficientData(const QString &folderPath) const { | |
| 306 | // Need at least 5 training documents for basic predictions | ||
| 307 | 2 | return documentsForFolder(folderPath) >= 5; | |
| 308 | } | ||
| 309 | |||
| 310 | // T-287: Total unique tokens trained for a specific folder | ||
| 311 | 15 | int FolderPredictor::tokenCountForFolder(const QString &folderPath) const { | |
| 312 |
3/3✓ Branch 2 → 3 taken 15 times.
✓ Branch 3 → 4 taken 1 time.
✓ Branch 3 → 5 taken 14 times.
|
15 | if (!m_db.isOpen()) |
| 313 | 1 | return 0; | |
| 314 | |||
| 315 |
1/1✓ Branch 5 → 6 taken 14 times.
|
14 | QSqlQuery q(m_db); |
| 316 |
1/1✓ Branch 9 → 10 taken 14 times.
|
14 | q.prepare(QStringLiteral( |
| 317 | "SELECT COUNT(*) FROM token_counts WHERE folder = :folder")); | ||
| 318 |
1/1✓ Branch 17 → 18 taken 14 times.
|
28 | q.bindValue(QStringLiteral(":folder"), folderPath); |
| 319 |
7/8✓ Branch 21 → 22 taken 14 times.
✓ Branch 22 → 23 taken 13 times.
✓ Branch 22 → 26 taken 1 time.
✓ Branch 23 → 24 taken 13 times.
✓ Branch 24 → 25 taken 13 times.
✗ Branch 24 → 26 not taken.
✓ Branch 27 → 28 taken 13 times.
✓ Branch 27 → 33 taken 1 time.
|
14 | if (q.exec() && q.next()) |
| 320 |
2/2✓ Branch 28 → 29 taken 13 times.
✓ Branch 29 → 30 taken 13 times.
|
13 | return q.value(0).toInt(); |
| 321 | 1 | return 0; | |
| 322 | 14 | } | |
| 323 | |||
| 324 | // T-287: Reset training data for a single folder (not the entire database) | ||
| 325 | 7 | void FolderPredictor::resetFolder(const QString &folderPath) { | |
| 326 |
3/3✓ Branch 2 → 3 taken 7 times.
✓ Branch 3 → 4 taken 1 time.
✓ Branch 3 → 5 taken 6 times.
|
7 | if (!m_db.isOpen()) |
| 327 | 1 | return; | |
| 328 | |||
| 329 |
1/1✓ Branch 5 → 6 taken 6 times.
|
6 | m_db.transaction(); |
| 330 | |||
| 331 |
1/1✓ Branch 6 → 7 taken 6 times.
|
6 | QSqlQuery q(m_db); |
| 332 | |||
| 333 | // Get the document count for this folder to subtract from total | ||
| 334 |
1/1✓ Branch 7 → 8 taken 6 times.
|
6 | int docCount = documentsForFolder(folderPath); |
| 335 | |||
| 336 | // Delete token counts for this folder | ||
| 337 |
1/1✓ Branch 11 → 12 taken 6 times.
|
6 | q.prepare(QStringLiteral( |
| 338 | "DELETE FROM token_counts WHERE folder = :folder")); | ||
| 339 |
1/1✓ Branch 19 → 20 taken 6 times.
|
12 | q.bindValue(QStringLiteral(":folder"), folderPath); |
| 340 |
1/1✓ Branch 23 → 24 taken 6 times.
|
6 | q.exec(); |
| 341 | |||
| 342 |
1/1✓ Branch 27 → 28 taken 6 times.
|
6 | q.prepare(QStringLiteral( |
| 343 | "DELETE FROM trained_messages WHERE folder = :folder")); | ||
| 344 |
1/1✓ Branch 35 → 36 taken 6 times.
|
12 | q.bindValue(QStringLiteral(":folder"), folderPath); |
| 345 |
1/1✓ Branch 39 → 40 taken 6 times.
|
6 | q.exec(); |
| 346 | |||
| 347 |
1/1✓ Branch 43 → 44 taken 6 times.
|
6 | q.prepare(QStringLiteral("DELETE FROM completed_training_folders WHERE folder = :folder")); |
| 348 |
1/1✓ Branch 51 → 52 taken 6 times.
|
12 | q.bindValue(QStringLiteral(":folder"), folderPath); |
| 349 |
1/1✓ Branch 55 → 56 taken 6 times.
|
6 | q.exec(); |
| 350 | |||
| 351 | // Delete folder document count | ||
| 352 |
1/1✓ Branch 59 → 60 taken 6 times.
|
6 | q.prepare(QStringLiteral( |
| 353 | "DELETE FROM folder_doc_counts WHERE folder = :folder")); | ||
| 354 |
1/1✓ Branch 67 → 68 taken 6 times.
|
12 | q.bindValue(QStringLiteral(":folder"), folderPath); |
| 355 |
1/1✓ Branch 71 → 72 taken 6 times.
|
6 | q.exec(); |
| 356 | |||
| 357 | // Subtract from total_docs | ||
| 358 |
2/2✓ Branch 72 → 73 taken 3 times.
✓ Branch 72 → 89 taken 3 times.
|
6 | if (docCount > 0) { |
| 359 |
1/1✓ Branch 76 → 77 taken 3 times.
|
3 | q.prepare(QStringLiteral( |
| 360 | "UPDATE meta SET value = CAST(MAX(0, CAST(value AS INTEGER) - :count) " | ||
| 361 | "AS TEXT) WHERE key = 'total_docs'")); | ||
| 362 |
1/1✓ Branch 84 → 85 taken 3 times.
|
6 | q.bindValue(QStringLiteral(":count"), docCount); |
| 363 |
1/1✓ Branch 88 → 89 taken 3 times.
|
3 | q.exec(); |
| 364 | } | ||
| 365 |
1/1✓ Branch 89 → 90 taken 6 times.
|
6 | touchStatisticsRevision(); |
| 366 | |||
| 367 |
1/1✓ Branch 90 → 91 taken 6 times.
|
6 | m_db.commit(); |
| 368 | 6 | } | |
| 369 | |||
| 370 | // T-287: Rename folder in training data (for folder rename/move operations). | ||
| 371 | // F-51 (SPRINT-104): the rename carries the folder AND its children, and | ||
| 372 | // the completed-training checkpoint table moves with it (a stale checkpoint | ||
| 373 | // under the old path kept the next background training re-scanning the | ||
| 374 | // renamed folder from scratch — noted by the audit's rules/search pass). | ||
| 375 | // S106-211: one checked transaction that MERGES into rows the new path may | ||
| 376 | // already have (a training job or an earlier rename can have written them): | ||
| 377 | // the former UPDATE failed on UNIQUE(token, folder) or on a busy writer | ||
| 378 | // lock without a word, and the rows stayed under the old path, where the | ||
| 379 | // next bootstrap's ledger no longer found them and counted the mail again. | ||
| 380 | // Lengths are code points like SQLite's substr() (S106-112). | ||
| 381 | 7 | bool FolderPredictor::renameFolderData(const QString &oldPath, | |
| 382 | const QString &newPath, | ||
| 383 | const QString &delimiter) { | ||
| 384 | 7 | QMutexLocker locker(&m_mutex); | |
| 385 |
8/11✓ Branch 3 → 4 taken 7 times.
✓ Branch 4 → 5 taken 6 times.
✓ Branch 4 → 11 taken 1 time.
✓ Branch 6 → 7 taken 6 times.
✗ Branch 6 → 11 not taken.
✓ Branch 8 → 9 taken 6 times.
✗ Branch 8 → 11 not taken.
✗ Branch 10 → 11 not taken.
✓ Branch 10 → 12 taken 6 times.
✓ Branch 13 → 14 taken 1 time.
✓ Branch 13 → 15 taken 6 times.
|
7 | if (!m_db.isOpen() || oldPath.isEmpty() || newPath.isEmpty() || oldPath == newPath) |
| 386 | 1 | return false; | |
| 387 |
1/1✓ Branch 15 → 16 taken 6 times.
|
6 | QSqlQuery q(m_db); |
| 388 |
2/3✓ Branch 19 → 20 taken 6 times.
✗ Branch 22 → 23 not taken.
✓ Branch 22 → 39 taken 6 times.
|
6 | if (!q.exec(QStringLiteral("BEGIN IMMEDIATE"))) { |
| 389 | ✗ | qCWarning(lcFolderPredictor) << "Folder rename not applied:" << q.lastError().text(); | |
| 390 | ✗ | return false; | |
| 391 | } | ||
| 392 | 7 | auto rollback = qScopeGuard([this] { m_db.rollback(); }); | |
| 393 |
1/1✓ Branch 40 → 41 taken 6 times.
|
6 | const QString prefix = oldPath + delimiter; |
| 394 |
1/1✓ Branch 41 → 42 taken 6 times.
|
6 | const qsizetype oldLength = oldPath.toUcs4().size(); |
| 395 | 6 | const QString matches = QStringLiteral("(folder = :oldPath OR substr(folder, 1, :oldLen) = :oldPrefix)"); | |
| 396 | 6 | const QString renamed = QStringLiteral(":newPath || substr(folder, :oldLen)"); | |
| 397 | 40 | const auto run = [&](const QString &sql) { | |
| 398 |
1/2✗ Branch 3 → 4 not taken.
✓ Branch 3 → 5 taken 40 times.
|
40 | if (!q.prepare(sql)) return false; |
| 399 | // Only the placeholders this statement has. | ||
| 400 |
4/4✓ Branch 8 → 9 taken 40 times.
✓ Branch 11 → 12 taken 20 times.
✓ Branch 11 → 22 taken 20 times.
✓ Branch 17 → 18 taken 20 times.
|
60 | if (sql.contains(QStringLiteral(":newPath"))) q.bindValue(QStringLiteral(":newPath"), newPath); |
| 401 |
1/1✓ Branch 27 → 28 taken 40 times.
|
80 | q.bindValue(QStringLiteral(":oldPath"), oldPath); |
| 402 |
1/1✓ Branch 36 → 37 taken 40 times.
|
80 | q.bindValue(QStringLiteral(":oldLen"), oldLength + 1); |
| 403 |
1/1✓ Branch 45 → 46 taken 40 times.
|
80 | q.bindValue(QStringLiteral(":oldPrefix"), prefix); |
| 404 | 40 | return q.exec(); | |
| 405 | 6 | }; | |
| 406 | // Each table: copy the matching rows with their new path aside, delete | ||
| 407 | // them, then merge the copies back. The new path may lie below the old | ||
| 408 | // one ("Projects" → "Projects/2026"), so a plain INSERT-then-DELETE would | ||
| 409 | // delete the rows it had just written. | ||
| 410 | 21 | const auto move = [&](const QString &table, const QString &columns, const QString &renamedColumns, | |
| 411 | const QString &onConflict) { | ||
| 412 |
1/6✓ Branch 59 → 60 taken 21 times.
✗ Branch 59 → 61 not taken.
✗ Branch 111 → 112 not taken.
✗ Branch 111 → 113 not taken.
✗ Branch 115 → 116 not taken.
✗ Branch 115 → 117 not taken.
|
42 | return q.exec(QStringLiteral("CREATE TEMP TABLE predictor_rename AS SELECT %1 FROM %2 WHERE 0") |
| 413 |
2/5✓ Branch 5 → 6 taken 21 times.
✓ Branch 57 → 58 taken 21 times.
✗ Branch 57 → 59 not taken.
✗ Branch 107 → 108 not taken.
✗ Branch 107 → 109 not taken.
|
42 | .arg(columns, table)) && |
| 414 |
5/11✓ Branch 12 → 13 taken 20 times.
✓ Branch 13 → 14 taken 20 times.
✗ Branch 13 → 32 not taken.
✓ Branch 53 → 54 taken 20 times.
✓ Branch 53 → 55 taken 1 time.
✓ Branch 55 → 56 taken 21 times.
✗ Branch 55 → 57 not taken.
✗ Branch 99 → 100 not taken.
✗ Branch 99 → 101 not taken.
✗ Branch 103 → 104 not taken.
✗ Branch 103 → 105 not taken.
|
81 | run(QStringLiteral("INSERT INTO predictor_rename(%1) SELECT %2 FROM %3 WHERE %4") |
| 415 |
3/5✓ Branch 11 → 12 taken 20 times.
✓ Branch 51 → 52 taken 20 times.
✓ Branch 51 → 53 taken 1 time.
✗ Branch 95 → 96 not taken.
✗ Branch 95 → 97 not taken.
|
41 | .arg(columns, renamedColumns, table, matches)) && |
| 416 |
9/16✓ Branch 17 → 18 taken 20 times.
✓ Branch 18 → 19 taken 20 times.
✓ Branch 19 → 20 taken 20 times.
✗ Branch 19 → 32 not taken.
✓ Branch 45 → 46 taken 20 times.
✓ Branch 45 → 47 taken 1 time.
✓ Branch 47 → 48 taken 20 times.
✓ Branch 47 → 49 taken 1 time.
✓ Branch 49 → 50 taken 20 times.
✓ Branch 49 → 51 taken 1 time.
✗ Branch 83 → 84 not taken.
✗ Branch 83 → 85 not taken.
✗ Branch 87 → 88 not taken.
✗ Branch 87 → 89 not taken.
✗ Branch 91 → 92 not taken.
✗ Branch 91 → 93 not taken.
|
61 | run(QStringLiteral("DELETE FROM %1 WHERE %2").arg(table, matches)) && |
| 417 |
6/11✓ Branch 24 → 25 taken 20 times.
✓ Branch 25 → 26 taken 20 times.
✗ Branch 25 → 32 not taken.
✓ Branch 41 → 42 taken 20 times.
✓ Branch 41 → 43 taken 1 time.
✓ Branch 43 → 44 taken 20 times.
✓ Branch 43 → 45 taken 1 time.
✗ Branch 75 → 76 not taken.
✗ Branch 75 → 77 not taken.
✗ Branch 79 → 80 not taken.
✗ Branch 79 → 81 not taken.
|
81 | q.exec(QStringLiteral("INSERT INTO %1(%2) SELECT %2 FROM predictor_rename WHERE true %3") |
| 418 |
6/8✓ Branch 6 → 7 taken 21 times.
✓ Branch 7 → 8 taken 20 times.
✓ Branch 7 → 32 taken 1 time.
✓ Branch 23 → 24 taken 20 times.
✓ Branch 39 → 40 taken 20 times.
✓ Branch 39 → 41 taken 1 time.
✗ Branch 71 → 72 not taken.
✗ Branch 71 → 73 not taken.
|
103 | .arg(table, columns, onConflict)) && |
| 419 |
8/13✓ Branch 29 → 30 taken 20 times.
✓ Branch 30 → 31 taken 20 times.
✗ Branch 30 → 32 not taken.
✓ Branch 33 → 34 taken 20 times.
✓ Branch 33 → 35 taken 1 time.
✓ Branch 35 → 36 taken 20 times.
✓ Branch 35 → 37 taken 1 time.
✓ Branch 37 → 38 taken 20 times.
✓ Branch 37 → 39 taken 1 time.
✗ Branch 63 → 64 not taken.
✗ Branch 63 → 65 not taken.
✗ Branch 67 → 68 not taken.
✗ Branch 67 → 69 not taken.
|
82 | q.exec(QStringLiteral("DROP TABLE predictor_rename")); |
| 420 | 6 | }; | |
| 421 | const bool ok = | ||
| 422 |
3/14✓ Branch 174 → 175 taken 6 times.
✗ Branch 174 → 176 not taken.
✓ Branch 176 → 177 taken 6 times.
✗ Branch 176 → 178 not taken.
✓ Branch 178 → 179 taken 6 times.
✗ Branch 178 → 180 not taken.
✗ Branch 355 → 356 not taken.
✗ Branch 355 → 357 not taken.
✗ Branch 359 → 360 not taken.
✗ Branch 359 → 361 not taken.
✗ Branch 363 → 364 not taken.
✗ Branch 363 → 365 not taken.
✗ Branch 367 → 368 not taken.
✗ Branch 367 → 369 not taken.
|
18 | move(QStringLiteral("token_counts"), QStringLiteral("token, folder, count"), |
| 423 |
4/13✓ Branch 59 → 60 taken 6 times.
✓ Branch 180 → 181 taken 6 times.
✗ Branch 180 → 182 not taken.
✓ Branch 182 → 183 taken 6 times.
✗ Branch 182 → 184 not taken.
✓ Branch 184 → 185 taken 6 times.
✗ Branch 184 → 186 not taken.
✗ Branch 371 → 372 not taken.
✗ Branch 371 → 373 not taken.
✗ Branch 376 → 377 not taken.
✗ Branch 376 → 378 not taken.
✗ Branch 380 → 381 not taken.
✗ Branch 380 → 382 not taken.
|
12 | QStringLiteral("token, %1 AS folder, count").arg(renamed), |
| 424 |
2/8✓ Branch 186 → 187 taken 6 times.
✗ Branch 186 → 188 not taken.
✓ Branch 188 → 189 taken 6 times.
✗ Branch 188 → 190 not taken.
✗ Branch 384 → 385 not taken.
✗ Branch 384 → 386 not taken.
✗ Branch 388 → 389 not taken.
✗ Branch 388 → 390 not taken.
|
27 | QStringLiteral("ON CONFLICT(token, folder) DO UPDATE SET count = count + excluded.count")) && |
| 425 |
8/17✓ Branch 82 → 83 taken 5 times.
✓ Branch 83 → 84 taken 5 times.
✗ Branch 83 → 117 not taken.
✓ Branch 156 → 157 taken 5 times.
✓ Branch 156 → 158 taken 1 time.
✓ Branch 158 → 159 taken 5 times.
✓ Branch 158 → 160 taken 1 time.
✓ Branch 160 → 161 taken 5 times.
✓ Branch 160 → 162 taken 1 time.
✗ Branch 318 → 319 not taken.
✗ Branch 318 → 320 not taken.
✗ Branch 322 → 323 not taken.
✗ Branch 322 → 324 not taken.
✗ Branch 326 → 327 not taken.
✗ Branch 326 → 328 not taken.
✗ Branch 330 → 331 not taken.
✗ Branch 330 → 332 not taken.
|
16 | move(QStringLiteral("folder_doc_counts"), QStringLiteral("folder, doc_count"), |
| 426 |
7/13✓ Branch 75 → 76 taken 5 times.
✓ Branch 162 → 163 taken 5 times.
✓ Branch 162 → 164 taken 1 time.
✓ Branch 164 → 165 taken 5 times.
✓ Branch 164 → 166 taken 1 time.
✓ Branch 166 → 167 taken 5 times.
✓ Branch 166 → 168 taken 1 time.
✗ Branch 334 → 335 not taken.
✗ Branch 334 → 336 not taken.
✗ Branch 339 → 340 not taken.
✗ Branch 339 → 341 not taken.
✗ Branch 343 → 344 not taken.
✗ Branch 343 → 345 not taken.
|
11 | QStringLiteral("%1 AS folder, doc_count").arg(renamed), |
| 427 |
5/10✓ Branch 168 → 169 taken 5 times.
✓ Branch 168 → 170 taken 1 time.
✓ Branch 170 → 171 taken 5 times.
✓ Branch 170 → 172 taken 1 time.
✓ Branch 172 → 173 taken 6 times.
✗ Branch 172 → 174 not taken.
✗ Branch 347 → 348 not taken.
✗ Branch 347 → 349 not taken.
✗ Branch 351 → 352 not taken.
✗ Branch 351 → 353 not taken.
|
26 | QStringLiteral("ON CONFLICT(folder) DO UPDATE SET doc_count = doc_count + excluded.doc_count")) && |
| 428 |
8/17✓ Branch 98 → 99 taken 5 times.
✓ Branch 99 → 100 taken 5 times.
✗ Branch 99 → 117 not taken.
✓ Branch 138 → 139 taken 5 times.
✓ Branch 138 → 140 taken 1 time.
✓ Branch 140 → 141 taken 5 times.
✓ Branch 140 → 142 taken 1 time.
✓ Branch 142 → 143 taken 5 times.
✓ Branch 142 → 144 taken 1 time.
✗ Branch 281 → 282 not taken.
✗ Branch 281 → 283 not taken.
✗ Branch 285 → 286 not taken.
✗ Branch 285 → 287 not taken.
✗ Branch 289 → 290 not taken.
✗ Branch 289 → 291 not taken.
✗ Branch 293 → 294 not taken.
✗ Branch 293 → 295 not taken.
|
16 | move(QStringLiteral("trained_messages"), QStringLiteral("account, folder, uid, uid_validity"), |
| 429 |
7/13✓ Branch 91 → 92 taken 5 times.
✓ Branch 144 → 145 taken 5 times.
✓ Branch 144 → 146 taken 1 time.
✓ Branch 146 → 147 taken 5 times.
✓ Branch 146 → 148 taken 1 time.
✓ Branch 148 → 149 taken 5 times.
✓ Branch 148 → 150 taken 1 time.
✗ Branch 297 → 298 not taken.
✗ Branch 297 → 299 not taken.
✗ Branch 302 → 303 not taken.
✗ Branch 302 → 304 not taken.
✗ Branch 306 → 307 not taken.
✗ Branch 306 → 308 not taken.
|
11 | QStringLiteral("account, %1 AS folder, uid, uid_validity").arg(renamed), |
| 430 |
9/13✓ Branch 66 → 67 taken 6 times.
✓ Branch 67 → 68 taken 5 times.
✓ Branch 67 → 117 taken 1 time.
✓ Branch 150 → 151 taken 5 times.
✓ Branch 150 → 152 taken 1 time.
✓ Branch 152 → 153 taken 5 times.
✓ Branch 152 → 154 taken 1 time.
✓ Branch 154 → 155 taken 5 times.
✓ Branch 154 → 156 taken 1 time.
✗ Branch 310 → 311 not taken.
✗ Branch 310 → 312 not taken.
✗ Branch 314 → 315 not taken.
✗ Branch 314 → 316 not taken.
|
61 | QStringLiteral("ON CONFLICT DO NOTHING")) && |
| 431 |
4/9✓ Branch 114 → 115 taken 5 times.
✓ Branch 115 → 116 taken 5 times.
✗ Branch 115 → 117 not taken.
✓ Branch 120 → 121 taken 5 times.
✓ Branch 120 → 122 taken 1 time.
✗ Branch 244 → 245 not taken.
✗ Branch 244 → 246 not taken.
✗ Branch 248 → 249 not taken.
✗ Branch 248 → 250 not taken.
|
11 | move(QStringLiteral("completed_training_folders"), |
| 432 |
4/8✓ Branch 122 → 123 taken 5 times.
✓ Branch 122 → 124 taken 1 time.
✓ Branch 124 → 125 taken 5 times.
✓ Branch 124 → 126 taken 1 time.
✗ Branch 252 → 253 not taken.
✗ Branch 252 → 254 not taken.
✗ Branch 256 → 257 not taken.
✗ Branch 256 → 258 not taken.
|
11 | QStringLiteral("account, folder, folder_id, uid_validity, revision"), |
| 433 |
7/13✓ Branch 107 → 108 taken 5 times.
✓ Branch 126 → 127 taken 5 times.
✓ Branch 126 → 128 taken 1 time.
✓ Branch 128 → 129 taken 5 times.
✓ Branch 128 → 130 taken 1 time.
✓ Branch 130 → 131 taken 5 times.
✓ Branch 130 → 132 taken 1 time.
✗ Branch 260 → 261 not taken.
✗ Branch 260 → 262 not taken.
✗ Branch 265 → 266 not taken.
✗ Branch 265 → 267 not taken.
✗ Branch 269 → 270 not taken.
✗ Branch 269 → 271 not taken.
|
11 | QStringLiteral("account, %1 AS folder, folder_id, uid_validity, revision").arg(renamed), |
| 434 |
8/12✓ Branch 118 → 119 taken 5 times.
✓ Branch 118 → 120 taken 1 time.
✓ Branch 132 → 133 taken 5 times.
✓ Branch 132 → 134 taken 1 time.
✓ Branch 134 → 135 taken 5 times.
✓ Branch 134 → 136 taken 1 time.
✓ Branch 136 → 137 taken 5 times.
✓ Branch 136 → 138 taken 1 time.
✗ Branch 273 → 274 not taken.
✗ Branch 273 → 275 not taken.
✗ Branch 277 → 278 not taken.
✗ Branch 277 → 279 not taken.
|
17 | QStringLiteral("ON CONFLICT DO NOTHING")); |
| 435 |
2/2✓ Branch 190 → 191 taken 1 time.
✓ Branch 190 → 207 taken 5 times.
|
6 | if (!ok) |
| 436 |
8/8✓ Branch 191 → 192 taken 1 time.
✓ Branch 195 → 196 taken 1 time.
✓ Branch 196 → 197 taken 1 time.
✓ Branch 197 → 198 taken 1 time.
✓ Branch 198 → 199 taken 1 time.
✓ Branch 199 → 200 taken 1 time.
✓ Branch 205 → 193 taken 1 time.
✓ Branch 205 → 206 taken 1 time.
|
2 | qCWarning(lcFolderPredictor) << "Folder rename not applied:" << q.lastError().text(); |
| 437 |
8/10✓ Branch 207 → 208 taken 5 times.
✓ Branch 207 → 212 taken 1 time.
✓ Branch 208 → 209 taken 5 times.
✓ Branch 209 → 210 taken 5 times.
✗ Branch 209 → 212 not taken.
✓ Branch 210 → 211 taken 5 times.
✗ Branch 211 → 212 not taken.
✓ Branch 211 → 213 taken 5 times.
✓ Branch 214 → 215 taken 1 time.
✓ Branch 214 → 216 taken 5 times.
|
6 | if (!ok || !touchStatisticsRevision() || !m_db.commit()) |
| 438 | 1 | return false; | |
| 439 | 5 | rollback.dismiss(); | |
| 440 | 5 | return true; | |
| 441 | 7 | } | |
| 442 | |||
| 443 | // --- Tokenizer --- | ||
| 444 | |||
| 445 | 29696 | QString FolderPredictor::extractEmail(const QString &fromField) const { | |
| 446 | // "Max Mustermann <max@example.com>" → "max@example.com" | ||
| 447 |
3/3✓ Branch 2 → 3 taken 354 times.
✓ Branch 2 → 12 taken 29342 times.
✓ Branch 7 → 8 taken 354 times.
|
30050 | thread_local static QRegularExpression angleRe(QStringLiteral("<([^>]+)>")); |
| 448 |
1/1✓ Branch 13 → 14 taken 29696 times.
|
29696 | auto match = angleRe.match(fromField); |
| 449 |
3/3✓ Branch 14 → 15 taken 29696 times.
✓ Branch 15 → 16 taken 6913 times.
✓ Branch 15 → 23 taken 22783 times.
|
29696 | if (match.hasMatch()) { |
| 450 |
3/3✓ Branch 16 → 17 taken 6913 times.
✓ Branch 17 → 18 taken 6913 times.
✓ Branch 18 → 19 taken 6913 times.
|
6913 | return match.captured(1).trimmed().toLower(); |
| 451 | } | ||
| 452 |
2/2✓ Branch 23 → 24 taken 22783 times.
✓ Branch 24 → 25 taken 22783 times.
|
22783 | return fromField.trimmed().toLower(); |
| 453 | 29696 | } | |
| 454 | |||
| 455 | 13341 | QStringList FolderPredictor::tokenize(const QString &from, | |
| 456 | const QString &subject, | ||
| 457 | const QString &to) const { | ||
| 458 | 13341 | QStringList tokens; | |
| 459 | |||
| 460 | // --- From tokens (boosted 3× — sender is strongest signal) --- | ||
| 461 |
1/1✓ Branch 2 → 3 taken 13341 times.
|
13341 | QString email = extractEmail(from); |
| 462 |
2/2✓ Branch 4 → 5 taken 13332 times.
✓ Branch 4 → 52 taken 9 times.
|
13341 | if (!email.isEmpty()) { |
| 463 | // Full email address (3× boost) | ||
| 464 |
2/2✓ Branch 15 → 6 taken 39996 times.
✓ Branch 15 → 16 taken 13332 times.
|
53328 | for (int b = 0; b < 3; ++b) |
| 465 |
2/2✓ Branch 9 → 10 taken 39996 times.
✓ Branch 10 → 11 taken 39996 times.
|
39996 | tokens.append(QStringLiteral("from:") + email); |
| 466 | |||
| 467 | 13332 | int atIdx = email.indexOf(QLatin1Char('@')); | |
| 468 |
2/2✓ Branch 19 → 20 taken 13328 times.
✓ Branch 19 → 52 taken 4 times.
|
13332 | if (atIdx > 0) { |
| 469 |
1/1✓ Branch 20 → 21 taken 13328 times.
|
13328 | QString user = email.left(atIdx); |
| 470 |
1/1✓ Branch 21 → 22 taken 13328 times.
|
13328 | QString domain = email.mid(atIdx + 1); |
| 471 |
2/2✓ Branch 25 → 26 taken 13328 times.
✓ Branch 26 → 27 taken 13328 times.
|
13328 | tokens.append(QStringLiteral("from:") + user); |
| 472 | // Domain (2× boost — most mails from same domain go to same folder) | ||
| 473 |
2/2✓ Branch 48 → 31 taken 26656 times.
✓ Branch 48 → 49 taken 13328 times.
|
39984 | for (int b = 0; b < 2; ++b) { |
| 474 |
2/2✓ Branch 34 → 35 taken 26656 times.
✓ Branch 35 → 36 taken 26656 times.
|
26656 | tokens.append(QStringLiteral("from:") + domain); |
| 475 |
2/2✓ Branch 42 → 43 taken 26656 times.
✓ Branch 43 → 44 taken 26656 times.
|
26656 | tokens.append(QStringLiteral("from_domain:") + domain); |
| 476 | } | ||
| 477 | 13328 | } | |
| 478 | } | ||
| 479 | |||
| 480 | // --- Subject tokens --- | ||
| 481 | 13341 | QString subj = subject; | |
| 482 | // Strip Re:/Fwd:/AW:/WG: prefixes | ||
| 483 | thread_local static QRegularExpression prefixRe( | ||
| 484 | 264 | QStringLiteral("^\\s*(Re|Fwd|AW|WG)\\s*:\\s*"), | |
| 485 |
3/3✓ Branch 53 → 54 taken 132 times.
✓ Branch 53 → 63 taken 13209 times.
✓ Branch 58 → 59 taken 132 times.
|
13605 | QRegularExpression::CaseInsensitiveOption); |
| 486 |
3/3✓ Branch 68 → 69 taken 14227 times.
✓ Branch 69 → 64 taken 886 times.
✓ Branch 69 → 70 taken 13341 times.
|
14227 | while (subj.contains(prefixRe)) { |
| 487 |
1/1✓ Branch 65 → 66 taken 886 times.
|
886 | subj.replace(prefixRe, QString()); |
| 488 | } | ||
| 489 | |||
| 490 | // Split on whitespace + punctuation | ||
| 491 |
3/3✓ Branch 70 → 71 taken 132 times.
✓ Branch 70 → 80 taken 13209 times.
✓ Branch 75 → 76 taken 132 times.
|
13473 | thread_local static QRegularExpression splitRe(QStringLiteral("[\\s,.;:!?()\\[\\]{}<>\"'/\\\\]+")); |
| 492 |
1/1✓ Branch 81 → 82 taken 13341 times.
|
13341 | const QStringList words = subj.split(splitRe, Qt::SkipEmptyParts); |
| 493 |
2/2✓ Branch 105 → 84 taken 32691 times.
✓ Branch 105 → 106 taken 13341 times.
|
46032 | for (const QString &word : words) { |
| 494 |
1/1✓ Branch 85 → 86 taken 32691 times.
|
32691 | QString lower = word.toLower(); |
| 495 |
6/6✓ Branch 87 → 88 taken 31065 times.
✓ Branch 87 → 91 taken 1626 times.
✓ Branch 89 → 90 taken 30865 times.
✓ Branch 89 → 91 taken 200 times.
✓ Branch 92 → 93 taken 30865 times.
✓ Branch 92 → 102 taken 1826 times.
|
32691 | if (lower.size() >= 2 && !s_stopWords.contains(lower)) { |
| 496 |
2/2✓ Branch 96 → 97 taken 30865 times.
✓ Branch 97 → 98 taken 30865 times.
|
30865 | tokens.append(QStringLiteral("subj:") + lower); |
| 497 | } | ||
| 498 | 32691 | } | |
| 499 | |||
| 500 | // --- To tokens --- | ||
| 501 | // May contain multiple addresses separated by comma/semicolon | ||
| 502 |
3/3✓ Branch 106 → 107 taken 132 times.
✓ Branch 106 → 116 taken 13209 times.
✓ Branch 111 → 112 taken 132 times.
|
13473 | thread_local static QRegularExpression addrSplitRe(QStringLiteral("[,;]+")); |
| 503 |
1/1✓ Branch 117 → 118 taken 13341 times.
|
13341 | const QStringList toAddrs = to.split(addrSplitRe, Qt::SkipEmptyParts); |
| 504 |
2/2✓ Branch 143 → 120 taken 12019 times.
✓ Branch 143 → 144 taken 13341 times.
|
25360 | for (const QString &addr : toAddrs) { |
| 505 |
1/1✓ Branch 121 → 122 taken 12019 times.
|
12019 | QString toEmail = extractEmail(addr); |
| 506 |
6/7✓ Branch 123 → 124 taken 12019 times.
✗ Branch 123 → 129 not taken.
✓ Branch 126 → 127 taken 12019 times.
✓ Branch 127 → 128 taken 12016 times.
✓ Branch 127 → 129 taken 3 times.
✓ Branch 130 → 131 taken 12016 times.
✓ Branch 130 → 140 taken 3 times.
|
12019 | if (!toEmail.isEmpty() && toEmail.contains(QLatin1Char('@'))) { |
| 507 |
2/2✓ Branch 134 → 135 taken 12016 times.
✓ Branch 135 → 136 taken 12016 times.
|
12016 | tokens.append(QStringLiteral("to:") + toEmail); |
| 508 | } | ||
| 509 | 12019 | } | |
| 510 | |||
| 511 | 13341 | return tokens; | |
| 512 | 13341 | } | |
| 513 | |||
| 514 | // --- Training --- | ||
| 515 | |||
| 516 | 4765 | void FolderPredictor::train(const QString &from, const QString &subject, | |
| 517 | const QString &to, const QString &targetFolder) { | ||
| 518 | 4765 | QMutexLocker locker(&m_mutex); | |
| 519 |
7/7✓ Branch 3 → 4 taken 4765 times.
✓ Branch 4 → 5 taken 4764 times.
✓ Branch 4 → 7 taken 1 time.
✓ Branch 6 → 7 taken 1 time.
✓ Branch 6 → 8 taken 4763 times.
✓ Branch 9 → 10 taken 2 times.
✓ Branch 9 → 11 taken 4763 times.
|
4765 | if (!m_db.isOpen() || targetFolder.isEmpty()) |
| 520 | 2 | return; | |
| 521 | |||
| 522 |
1/1✓ Branch 11 → 12 taken 4763 times.
|
4763 | const QStringList tokens = tokenize(from, subject, to); |
| 523 |
2/2✓ Branch 13 → 14 taken 1 time.
✓ Branch 13 → 15 taken 4762 times.
|
4763 | if (tokens.isEmpty()) |
| 524 | 1 | return; | |
| 525 | |||
| 526 |
1/1✓ Branch 15 → 16 taken 4762 times.
|
4762 | QSqlQuery q(m_db); |
| 527 |
1/1✓ Branch 16 → 17 taken 4762 times.
|
4762 | m_db.transaction(); |
| 528 | |||
| 529 | // Update token counts | ||
| 530 |
1/1✓ Branch 20 → 21 taken 4762 times.
|
4762 | q.prepare(QStringLiteral( |
| 531 | "INSERT INTO token_counts (token, folder, count) VALUES (?, ?, 1) " | ||
| 532 | "ON CONFLICT(token, folder) DO UPDATE SET count = count + 1")); | ||
| 533 |
2/2✓ Branch 37 → 25 taken 56313 times.
✓ Branch 37 → 38 taken 4762 times.
|
61075 | for (const QString &token : tokens) { |
| 534 |
1/1✓ Branch 28 → 29 taken 56313 times.
|
56313 | q.bindValue(0, token); |
| 535 |
1/1✓ Branch 32 → 33 taken 56313 times.
|
56313 | q.bindValue(1, targetFolder); |
| 536 |
1/1✓ Branch 34 → 35 taken 56313 times.
|
56313 | q.exec(); |
| 537 | } | ||
| 538 | |||
| 539 | // Update folder document count | ||
| 540 |
1/1✓ Branch 41 → 42 taken 4762 times.
|
4762 | q.prepare(QStringLiteral( |
| 541 | "INSERT INTO folder_doc_counts (folder, doc_count) VALUES (?, 1) " | ||
| 542 | "ON CONFLICT(folder) DO UPDATE SET doc_count = doc_count + 1")); | ||
| 543 |
1/1✓ Branch 46 → 47 taken 4762 times.
|
4762 | q.bindValue(0, targetFolder); |
| 544 |
1/1✓ Branch 48 → 49 taken 4762 times.
|
4762 | q.exec(); |
| 545 | |||
| 546 | // Increment total docs | ||
| 547 |
1/1✓ Branch 52 → 53 taken 4762 times.
|
4762 | q.exec(QStringLiteral( |
| 548 | "UPDATE meta SET value = CAST(CAST(value AS INTEGER) + 1 AS TEXT) " | ||
| 549 | "WHERE key = 'total_docs'")); | ||
| 550 |
1/1✓ Branch 55 → 56 taken 4762 times.
|
4762 | touchStatisticsRevision(); |
| 551 | |||
| 552 |
1/1✓ Branch 56 → 57 taken 4762 times.
|
4762 | m_db.commit(); |
| 553 | 4766 | } | |
| 554 | |||
| 555 | 25 | void FolderPredictor::untrain(const QString &from, const QString &subject, | |
| 556 | const QString &to, const QString &wrongFolder) { | ||
| 557 | 25 | QMutexLocker locker(&m_mutex); | |
| 558 |
7/7✓ Branch 3 → 4 taken 25 times.
✓ Branch 4 → 5 taken 24 times.
✓ Branch 4 → 7 taken 1 time.
✓ Branch 6 → 7 taken 1 time.
✓ Branch 6 → 8 taken 23 times.
✓ Branch 9 → 10 taken 2 times.
✓ Branch 9 → 11 taken 23 times.
|
25 | if (!m_db.isOpen() || wrongFolder.isEmpty()) |
| 559 | 2 | return; | |
| 560 | |||
| 561 |
1/1✓ Branch 11 → 12 taken 23 times.
|
23 | const QStringList tokens = tokenize(from, subject, to); |
| 562 |
2/2✓ Branch 13 → 14 taken 1 time.
✓ Branch 13 → 15 taken 22 times.
|
23 | if (tokens.isEmpty()) |
| 563 | 1 | return; | |
| 564 | |||
| 565 |
1/1✓ Branch 15 → 16 taken 22 times.
|
22 | QSqlQuery q(m_db); |
| 566 |
1/1✓ Branch 16 → 17 taken 22 times.
|
22 | m_db.transaction(); |
| 567 | |||
| 568 | // Decrement token counts (don't go below 0) | ||
| 569 |
1/1✓ Branch 20 → 21 taken 22 times.
|
22 | q.prepare(QStringLiteral( |
| 570 | "UPDATE token_counts SET count = MAX(count - 1, 0) " | ||
| 571 | "WHERE token = ? AND folder = ?")); | ||
| 572 |
2/2✓ Branch 37 → 25 taken 226 times.
✓ Branch 37 → 38 taken 22 times.
|
248 | for (const QString &token : tokens) { |
| 573 |
1/1✓ Branch 28 → 29 taken 226 times.
|
226 | q.bindValue(0, token); |
| 574 |
1/1✓ Branch 32 → 33 taken 226 times.
|
226 | q.bindValue(1, wrongFolder); |
| 575 |
1/1✓ Branch 34 → 35 taken 226 times.
|
226 | q.exec(); |
| 576 | } | ||
| 577 | |||
| 578 | // Clean up zero-count entries | ||
| 579 |
1/1✓ Branch 41 → 42 taken 22 times.
|
22 | q.exec(QStringLiteral("DELETE FROM token_counts WHERE count = 0")); |
| 580 | |||
| 581 | // Decrement folder doc count | ||
| 582 |
1/1✓ Branch 47 → 48 taken 22 times.
|
22 | q.prepare(QStringLiteral( |
| 583 | "UPDATE folder_doc_counts SET doc_count = MAX(doc_count - 1, 0) " | ||
| 584 | "WHERE folder = ?")); | ||
| 585 |
1/1✓ Branch 52 → 53 taken 22 times.
|
22 | q.bindValue(0, wrongFolder); |
| 586 |
1/1✓ Branch 54 → 55 taken 22 times.
|
22 | q.exec(); |
| 587 | |||
| 588 | // Decrement total docs | ||
| 589 |
1/1✓ Branch 58 → 59 taken 22 times.
|
22 | q.exec(QStringLiteral( |
| 590 | "UPDATE meta SET value = CAST(MAX(CAST(value AS INTEGER) - 1, 0) AS TEXT) " | ||
| 591 | "WHERE key = 'total_docs'")); | ||
| 592 |
1/1✓ Branch 61 → 62 taken 22 times.
|
22 | touchStatisticsRevision(); |
| 593 | |||
| 594 |
1/1✓ Branch 62 → 63 taken 22 times.
|
22 | m_db.commit(); |
| 595 | 26 | } | |
| 596 | |||
| 597 | // --- Pre-training from cache (T-167) --- | ||
| 598 | |||
| 599 | 776 | bool FolderPredictor::trainFromCache(MailCache *cache, const QString &account, | |
| 600 | const QStringList &excludeFolders, QString *error, | ||
| 601 | const std::atomic_bool *cancelled, | ||
| 602 | int *trainedDocuments) { | ||
| 603 | 776 | QMutexLocker locker(&m_mutex); | |
| 604 |
2/2✓ Branch 3 → 4 taken 767 times.
✓ Branch 3 → 5 taken 9 times.
|
776 | if (error) error->clear(); |
| 605 |
2/2✓ Branch 5 → 6 taken 763 times.
✓ Branch 5 → 7 taken 13 times.
|
776 | if (trainedDocuments) *trainedDocuments = 0; |
| 606 | 11 | const auto fail = [error](const QString &message) { | |
| 607 |
2/2✓ Branch 2 → 3 taken 8 times.
✓ Branch 2 → 4 taken 3 times.
|
11 | if (error) *error = message; |
| 608 | 11 | return false; | |
| 609 | 776 | }; | |
| 610 | 2178 | const auto interrupted = [cancelled] { | |
| 611 |
4/4✓ Branch 2 → 3 taken 2006 times.
✓ Branch 2 → 6 taken 172 times.
✓ Branch 4 → 5 taken 3 times.
✓ Branch 4 → 6 taken 2003 times.
|
2178 | return cancelled && cancelled->load(std::memory_order_relaxed); |
| 612 | 776 | }; | |
| 613 |
9/10✓ Branch 7 → 8 taken 776 times.
✓ Branch 8 → 9 taken 774 times.
✓ Branch 8 → 12 taken 2 times.
✓ Branch 9 → 10 taken 773 times.
✓ Branch 9 → 12 taken 1 time.
✓ Branch 10 → 11 taken 773 times.
✗ Branch 11 → 12 not taken.
✓ Branch 11 → 13 taken 773 times.
✓ Branch 14 → 15 taken 3 times.
✓ Branch 14 → 22 taken 773 times.
|
776 | if (!m_db.isOpen() || !cache || !cache->isOpen()) |
| 614 | 3 | return fail(QStringLiteral("Predictor or training cache is not open")); | |
| 615 |
2/2✓ Branch 23 → 24 taken 2 times.
✓ Branch 23 → 31 taken 771 times.
|
775 | if (interrupted()) return fail(QStringLiteral("Predictor training cancelled")); |
| 616 | |||
| 617 |
1/1✓ Branch 31 → 32 taken 771 times.
|
771 | QSqlQuery q(m_db); |
| 618 | |||
| 619 | // Get all folder paths | ||
| 620 | 771 | QString cacheError; | |
| 621 |
1/1✓ Branch 32 → 33 taken 771 times.
|
771 | QStringList folderPaths = cache->allFolderPaths(account, &cacheError); |
| 622 |
2/2✓ Branch 34 → 35 taken 1 time.
✓ Branch 34 → 36 taken 770 times.
|
771 | if (!cacheError.isEmpty()) return fail(cacheError); |
| 623 |
2/2✓ Branch 37 → 38 taken 140 times.
✓ Branch 37 → 39 taken 630 times.
|
770 | if (folderPaths.isEmpty()) |
| 624 | 140 | return true; | |
| 625 | |||
| 626 | // Filter excluded folders (case-insensitive match) | ||
| 627 | 630 | QStringList trainFolders; | |
| 628 |
4/4✓ Branch 39 → 40 taken 630 times.
✓ Branch 40 → 41 taken 630 times.
✓ Branch 87 → 42 taken 1882 times.
✓ Branch 87 → 88 taken 630 times.
|
2512 | for (const QString &path : folderPaths) { |
| 629 | 1882 | bool excluded = false; | |
| 630 |
2/2✓ Branch 82 → 45 taken 8861 times.
✓ Branch 82 → 83 taken 1047 times.
|
9908 | for (const QString &ex : excludeFolders) { |
| 631 | 33790 | if (path.compare(ex, Qt::CaseInsensitive) == 0 || | |
| 632 |
6/8✓ Branch 50 → 51 taken 8034 times.
✓ Branch 51 → 52 taken 8034 times.
✓ Branch 52 → 53 taken 8032 times.
✓ Branch 52 → 68 taken 2 times.
✓ Branch 78 → 79 taken 835 times.
✓ Branch 78 → 80 taken 8026 times.
✗ Branch 338 → 339 not taken.
✗ Branch 338 → 340 not taken.
|
16895 | path.endsWith(QLatin1Char('.') + ex, Qt::CaseInsensitive) || |
| 633 |
6/8✓ Branch 55 → 56 taken 8032 times.
✓ Branch 56 → 57 taken 8032 times.
✓ Branch 57 → 58 taken 8030 times.
✓ Branch 57 → 68 taken 2 times.
✓ Branch 76 → 77 taken 8034 times.
✓ Branch 76 → 78 taken 827 times.
✗ Branch 332 → 333 not taken.
✗ Branch 332 → 334 not taken.
|
16893 | path.endsWith(QLatin1Char('/') + ex, Qt::CaseInsensitive) || |
| 634 |
8/10✓ Branch 47 → 48 taken 8034 times.
✓ Branch 47 → 68 taken 827 times.
✓ Branch 60 → 61 taken 8030 times.
✓ Branch 61 → 62 taken 8030 times.
✓ Branch 62 → 63 taken 8028 times.
✓ Branch 62 → 68 taken 2 times.
✓ Branch 74 → 75 taken 8032 times.
✓ Branch 74 → 76 taken 829 times.
✗ Branch 326 → 327 not taken.
✗ Branch 326 → 328 not taken.
|
33784 | path.startsWith(ex + QLatin1Char('.'), Qt::CaseInsensitive) || |
| 635 |
8/10✓ Branch 65 → 66 taken 8028 times.
✓ Branch 66 → 67 taken 8028 times.
✓ Branch 67 → 68 taken 2 times.
✓ Branch 67 → 69 taken 8026 times.
✓ Branch 70 → 71 taken 8028 times.
✓ Branch 70 → 72 taken 833 times.
✓ Branch 72 → 73 taken 8030 times.
✓ Branch 72 → 74 taken 831 times.
✗ Branch 320 → 321 not taken.
✗ Branch 320 → 322 not taken.
|
16889 | path.startsWith(ex + QLatin1Char('/'), Qt::CaseInsensitive)) { |
| 636 | 835 | excluded = true; | |
| 637 | 835 | break; | |
| 638 | } | ||
| 639 | } | ||
| 640 |
2/2✓ Branch 83 → 84 taken 1047 times.
✓ Branch 83 → 85 taken 835 times.
|
1882 | if (!excluded) |
| 641 |
1/1✓ Branch 84 → 85 taken 1047 times.
|
1047 | trainFolders.append(path); |
| 642 | } | ||
| 643 | |||
| 644 | // Count total headers for progress | ||
| 645 | 630 | int totalHeaders = 0; | |
| 646 |
4/4✓ Branch 88 → 89 taken 630 times.
✓ Branch 89 → 90 taken 630 times.
✓ Branch 99 → 91 taken 1047 times.
✓ Branch 99 → 100 taken 630 times.
|
1677 | for (const QString &folder : trainFolders) { |
| 647 |
1/1✓ Branch 93 → 94 taken 1047 times.
|
1047 | totalHeaders += cache->payloadLimits().cachedHeaderCount(account, folder, &cacheError); |
| 648 |
1/2✗ Branch 95 → 96 not taken.
✓ Branch 95 → 97 taken 1047 times.
|
1047 | if (!cacheError.isEmpty()) return fail(cacheError); |
| 649 | } | ||
| 650 | |||
| 651 |
2/2✓ Branch 100 → 102 taken 476 times.
✓ Branch 100 → 103 taken 154 times.
|
630 | if (totalHeaders == 0) |
| 652 | 476 | return true; | |
| 653 | |||
| 654 |
1/1✓ Branch 103 → 104 taken 154 times.
|
154 | q.finish(); |
| 655 | 154 | int processed = 0; | |
| 656 |
4/4✓ Branch 104 → 105 taken 154 times.
✓ Branch 105 → 106 taken 154 times.
✓ Branch 284 → 107 taken 359 times.
✓ Branch 284 → 285 taken 144 times.
|
503 | for (const QString &folder : trainFolders) { |
| 657 |
1/2✗ Branch 109 → 110 not taken.
✓ Branch 109 → 117 taken 359 times.
|
359 | if (interrupted()) return fail(QStringLiteral("Predictor training cancelled")); |
| 658 |
1/1✓ Branch 118 → 119 taken 359 times.
|
359 | const auto snapshot = cache->training().trainingSnapshot(account, folder); |
| 659 |
1/2✗ Branch 120 → 121 not taken.
✓ Branch 120 → 123 taken 359 times.
|
359 | if (!snapshot.error.isEmpty()) return fail(snapshot.error); |
| 660 |
1/1✓ Branch 126 → 127 taken 359 times.
|
359 | q.prepare(QStringLiteral("SELECT folder_id,uid_validity,revision FROM completed_training_folders " |
| 661 | "WHERE account=? AND folder=?")); | ||
| 662 |
2/2✓ Branch 131 → 132 taken 359 times.
✓ Branch 135 → 136 taken 359 times.
|
359 | q.addBindValue(account); q.addBindValue(folder); |
| 663 |
2/5✓ Branch 137 → 138 taken 359 times.
✗ Branch 138 → 139 not taken.
✓ Branch 138 → 145 taken 359 times.
✗ Branch 139 → 140 not taken.
✗ Branch 140 → 141 not taken.
|
359 | if (!q.exec()) return fail(q.lastError().text()); |
| 664 |
3/6✓ Branch 147 → 148 taken 15 times.
✓ Branch 148 → 149 taken 15 times.
✓ Branch 149 → 150 taken 15 times.
✗ Branch 149 → 158 not taken.
✗ Branch 377 → 378 not taken.
✗ Branch 377 → 379 not taken.
|
30 | const bool unchanged = q.next() && q.value(0).toLongLong() == snapshot.folderId && |
| 665 |
18/25✓ Branch 145 → 146 taken 359 times.
✓ Branch 146 → 147 taken 15 times.
✓ Branch 146 → 158 taken 344 times.
✓ Branch 150 → 151 taken 15 times.
✓ Branch 151 → 152 taken 15 times.
✓ Branch 152 → 153 taken 15 times.
✗ Branch 152 → 158 not taken.
✓ Branch 153 → 154 taken 15 times.
✓ Branch 154 → 155 taken 15 times.
✓ Branch 156 → 157 taken 13 times.
✓ Branch 156 → 158 taken 2 times.
✓ Branch 159 → 160 taken 15 times.
✓ Branch 159 → 161 taken 344 times.
✓ Branch 161 → 162 taken 15 times.
✓ Branch 161 → 163 taken 344 times.
✓ Branch 163 → 164 taken 15 times.
✓ Branch 163 → 165 taken 344 times.
✓ Branch 165 → 166 taken 15 times.
✓ Branch 165 → 167 taken 344 times.
✗ Branch 365 → 366 not taken.
✗ Branch 365 → 367 not taken.
✗ Branch 369 → 370 not taken.
✗ Branch 369 → 371 not taken.
✗ Branch 373 → 374 not taken.
✗ Branch 373 → 375 not taken.
|
374 | q.value(1).toUInt() == snapshot.uidValidity && q.value(2).toString() == snapshot.revision; |
| 666 |
1/1✓ Branch 167 → 168 taken 359 times.
|
359 | q.finish(); |
| 667 |
2/2✓ Branch 168 → 169 taken 13 times.
✓ Branch 168 → 170 taken 346 times.
|
359 | if (unchanged) continue; |
| 668 | 346 | qint64 afterUid = 0; | |
| 669 | 346 | std::optional<quint32> epoch; | |
| 670 | while (true) { | ||
| 671 |
2/2✓ Branch 172 → 173 taken 1 time.
✓ Branch 172 → 180 taken 564 times.
|
566 | if (interrupted()) return fail(QStringLiteral("Predictor training cancelled")); |
| 672 |
1/1✓ Branch 181 → 182 taken 564 times.
|
564 | const auto page = cache->training().trainingPage(account, folder, afterUid); |
| 673 |
2/2✓ Branch 183 → 184 taken 2 times.
✓ Branch 183 → 186 taken 562 times.
|
564 | if (!page.error.isEmpty()) return fail(page.error); |
| 674 |
2/2✓ Branch 187 → 188 taken 337 times.
✓ Branch 187 → 189 taken 225 times.
|
562 | if (page.headers.isEmpty()) break; |
| 675 |
5/8✓ Branch 189 → 190 taken 225 times.
✗ Branch 189 → 195 not taken.
✓ Branch 190 → 191 taken 225 times.
✗ Branch 190 → 195 not taken.
✓ Branch 192 → 193 taken 25 times.
✓ Branch 192 → 196 taken 200 times.
✗ Branch 197 → 198 not taken.
✓ Branch 197 → 205 taken 225 times.
|
475 | if (page.folderId != snapshot.folderId || page.uidValidity != snapshot.uidValidity || |
| 676 |
1/2✗ Branch 194 → 195 not taken.
✓ Branch 194 → 196 taken 25 times.
|
250 | (epoch && *epoch != page.uidValidity)) |
| 677 | ✗ | return fail(QStringLiteral("Predictor training mailbox identity changed")); | |
| 678 | 225 | epoch = page.uidValidity; | |
| 679 | 225 | int batchTrained = 0; | |
| 680 |
1/1✓ Branch 207 → 208 taken 225 times.
|
225 | if (!trainCachedBatch(account, folder, page.uidValidity, page.headers, |
| 681 |
3/3✓ Branch 206 → 207 taken 225 times.
✓ Branch 209 → 210 taken 6 times.
✓ Branch 209 → 211 taken 219 times.
|
675 | [&] { return cache->training().trainingIdentityCurrent(account, page.folderId, |
| 682 | 225 | folder, page.uidValidity); }, | |
| 683 | 6 | error, cancelled, &batchTrained)) return false; | |
| 684 |
2/2✓ Branch 211 → 212 taken 211 times.
✓ Branch 211 → 213 taken 8 times.
|
219 | if (trainedDocuments) *trainedDocuments += batchTrained; |
| 685 | 219 | afterUid = page.headers.last().uid; | |
| 686 | 219 | processed += page.headers.size(); | |
| 687 |
1/1✓ Branch 215 → 216 taken 219 times.
|
219 | emit trainingProgress(processed, totalHeaders); |
| 688 | 783 | } | |
| 689 |
1/2✗ Branch 223 → 224 not taken.
✓ Branch 223 → 231 taken 337 times.
|
337 | if (interrupted()) return fail(QStringLiteral("Predictor training cancelled")); |
| 690 |
2/3✓ Branch 232 → 233 taken 337 times.
✗ Branch 233 → 234 not taken.
✓ Branch 233 → 241 taken 337 times.
|
337 | if (!cache->training().trainingIdentityCurrent(account, snapshot.folderId, folder, snapshot.uidValidity)) |
| 691 | ✗ | return fail(QStringLiteral("Predictor training mailbox identity changed")); | |
| 692 | // Save the starting revision, never a later one: arrivals below the paging | ||
| 693 | // cursor during this scan must remain discoverable on the next recovery. | ||
| 694 |
1/1✓ Branch 244 → 245 taken 337 times.
|
337 | q.prepare(QStringLiteral("INSERT INTO completed_training_folders VALUES(?,?,?,?,?) " |
| 695 | "ON CONFLICT(account,folder) DO UPDATE SET folder_id=excluded.folder_id," | ||
| 696 | "uid_validity=excluded.uid_validity,revision=excluded.revision")); | ||
| 697 |
3/3✓ Branch 249 → 250 taken 337 times.
✓ Branch 253 → 254 taken 337 times.
✓ Branch 257 → 258 taken 337 times.
|
337 | q.addBindValue(account); q.addBindValue(folder); q.addBindValue(snapshot.folderId); |
| 698 |
2/2✓ Branch 261 → 262 taken 337 times.
✓ Branch 265 → 266 taken 337 times.
|
337 | q.addBindValue(snapshot.uidValidity); q.addBindValue(snapshot.revision); |
| 699 |
5/5✓ Branch 267 → 268 taken 337 times.
✓ Branch 268 → 269 taken 1 time.
✓ Branch 268 → 275 taken 336 times.
✓ Branch 269 → 270 taken 1 time.
✓ Branch 270 → 271 taken 1 time.
|
337 | if (!q.exec()) return fail(q.lastError().text()); |
| 700 | 359 | } | |
| 701 |
1/2✗ Branch 286 → 287 not taken.
✓ Branch 286 → 294 taken 144 times.
|
144 | if (interrupted()) return fail(QStringLiteral("Predictor training cancelled")); |
| 702 |
1/1✓ Branch 297 → 298 taken 144 times.
|
144 | q.prepare(QStringLiteral( |
| 703 | "INSERT OR REPLACE INTO meta (key, value) VALUES ('pretrained_account', ?)")); | ||
| 704 |
1/1✓ Branch 302 → 303 taken 144 times.
|
144 | q.bindValue(0, account); |
| 705 |
5/5✓ Branch 304 → 305 taken 144 times.
✓ Branch 305 → 306 taken 1 time.
✓ Branch 305 → 312 taken 143 times.
✓ Branch 306 → 307 taken 1 time.
✓ Branch 307 → 308 taken 1 time.
|
144 | if (!q.exec()) return fail(q.lastError().text()); |
| 706 | 143 | return true; | |
| 707 | 776 | } | |
| 708 | |||
| 709 | 44 | bool FolderPredictor::trainCachedHeaders( | |
| 710 | const QString &account, const QString &folder, quint32 uidValidity, | ||
| 711 | const QList<MailHeader> &headers, const std::function<bool()> &identityCurrent, | ||
| 712 | QString *error, const std::atomic_bool *cancelled, int *trainedDocuments) { | ||
| 713 | 44 | QMutexLocker locker(&m_mutex); | |
| 714 |
1/1✓ Branch 3 → 4 taken 44 times.
|
44 | return trainCachedBatch(account, folder, uidValidity, headers, identityCurrent, |
| 715 | 88 | error, cancelled, trainedDocuments); | |
| 716 | 44 | } | |
| 717 | |||
| 718 | 269 | bool FolderPredictor::trainCachedBatch( | |
| 719 | const QString &account, const QString &folder, quint32 uidValidity, | ||
| 720 | const QList<MailHeader> &headers, const std::function<bool()> &identityCurrent, | ||
| 721 | QString *error, const std::atomic_bool *cancelled, int *trainedDocuments) { | ||
| 722 |
2/2✓ Branch 2 → 3 taken 264 times.
✓ Branch 2 → 4 taken 5 times.
|
269 | if (error) error->clear(); |
| 723 |
2/2✓ Branch 4 → 5 taken 267 times.
✓ Branch 4 → 6 taken 2 times.
|
269 | if (trainedDocuments) *trainedDocuments = 0; |
| 724 | 16 | const auto fail = [error](const QString &message) { | |
| 725 |
1/2✓ Branch 2 → 3 taken 16 times.
✗ Branch 2 → 4 not taken.
|
16 | if (error) *error = message; |
| 726 | 16 | return false; | |
| 727 | 269 | }; | |
| 728 |
4/4✓ Branch 2 → 3 taken 5389 times.
✓ Branch 2 → 6 taken 2224 times.
✓ Branch 4 → 5 taken 1 time.
✓ Branch 4 → 6 taken 5388 times.
|
7613 | const auto interrupted = [cancelled] { return cancelled && cancelled->load(); }; |
| 729 |
9/9✓ Branch 6 → 7 taken 269 times.
✓ Branch 7 → 8 taken 268 times.
✓ Branch 7 → 12 taken 1 time.
✓ Branch 9 → 10 taken 267 times.
✓ Branch 9 → 12 taken 1 time.
✓ Branch 11 → 12 taken 1 time.
✓ Branch 11 → 13 taken 266 times.
✓ Branch 14 → 15 taken 3 times.
✓ Branch 14 → 22 taken 266 times.
|
269 | if (!m_db.isOpen() || folder.isEmpty() || headers.size() > 128) |
| 730 | 3 | return fail(QStringLiteral("Predictor is closed or the training batch is invalid")); | |
| 731 |
2/2✓ Branch 23 → 24 taken 1 time.
✓ Branch 23 → 31 taken 265 times.
|
267 | if (interrupted()) return fail(QStringLiteral("Predictor training cancelled")); |
| 732 |
2/2✓ Branch 32 → 33 taken 1 time.
✓ Branch 32 → 34 taken 264 times.
|
265 | if (headers.isEmpty()) return true; |
| 733 |
1/1✓ Branch 34 → 35 taken 264 times.
|
264 | QSqlQuery begin(m_db); |
| 734 |
2/5✓ Branch 38 → 39 taken 264 times.
✗ Branch 41 → 42 not taken.
✓ Branch 41 → 48 taken 264 times.
✗ Branch 42 → 43 not taken.
✗ Branch 43 → 44 not taken.
|
264 | if (!begin.exec(QStringLiteral("BEGIN IMMEDIATE"))) return fail(begin.lastError().text()); |
| 735 | 276 | auto rollback = qScopeGuard([this] { m_db.rollback(); }); | |
| 736 | // Recheck the cache epoch after acquiring the predictor writer lock. A job | ||
| 737 | // delayed by another writer must not restore an obsolete ledger epoch. | ||
| 738 |
6/7✓ Branch 50 → 51 taken 264 times.
✗ Branch 50 → 53 not taken.
✓ Branch 51 → 52 taken 264 times.
✓ Branch 52 → 53 taken 1 time.
✓ Branch 52 → 54 taken 263 times.
✓ Branch 55 → 56 taken 1 time.
✓ Branch 55 → 63 taken 263 times.
|
264 | if (!identityCurrent || !identityCurrent()) |
| 739 | 1 | return fail(QStringLiteral("Predictor training mailbox identity changed")); | |
| 740 | 263 | int trained = 0; | |
| 741 |
2/2✓ Branch 263 → 65 taken 7095 times.
✓ Branch 263 → 264 taken 252 times.
|
7347 | for (const MailHeader &header : headers) { |
| 742 |
1/2✗ Branch 67 → 68 not taken.
✓ Branch 67 → 75 taken 7095 times.
|
7095 | if (interrupted()) return fail(QStringLiteral("Predictor training cancelled")); |
| 743 |
7/9✓ Branch 76 → 77 taken 7094 times.
✓ Branch 78 → 79 taken 7093 times.
✓ Branch 78 → 85 taken 1 time.
✓ Branch 81 → 82 taken 7092 times.
✓ Branch 81 → 85 taken 1 time.
✓ Branch 93 → 94 taken 4 times.
✓ Branch 93 → 101 taken 7091 times.
✗ Branch 320 → 321 not taken.
✗ Branch 320 → 322 not taken.
|
21282 | if (header.uid <= 0 || header.from.toUtf8().size() > 8192 || |
| 744 |
12/16✓ Branch 75 → 76 taken 7094 times.
✓ Branch 75 → 85 taken 1 time.
✓ Branch 79 → 80 taken 7093 times.
✓ Branch 82 → 83 taken 7092 times.
✓ Branch 84 → 85 taken 1 time.
✓ Branch 84 → 86 taken 7091 times.
✓ Branch 87 → 88 taken 7092 times.
✓ Branch 87 → 89 taken 3 times.
✓ Branch 89 → 90 taken 7093 times.
✓ Branch 89 → 91 taken 2 times.
✓ Branch 91 → 92 taken 7094 times.
✓ Branch 91 → 93 taken 1 time.
✗ Branch 312 → 313 not taken.
✗ Branch 312 → 314 not taken.
✗ Branch 316 → 317 not taken.
✗ Branch 316 → 318 not taken.
|
21282 | header.subject.toUtf8().size() > 8192 || header.to.toUtf8().size() > 8192) |
| 745 | 4 | return fail(QStringLiteral("Predictor training header is invalid or oversized")); | |
| 746 |
1/1✓ Branch 101 → 102 taken 7091 times.
|
7091 | QSqlQuery seen(m_db); |
| 747 |
1/1✓ Branch 105 → 106 taken 7091 times.
|
7091 | seen.prepare(QStringLiteral( |
| 748 | "SELECT uid_validity FROM trained_messages WHERE account=? AND folder=? AND uid=?")); | ||
| 749 |
3/3✓ Branch 110 → 111 taken 7091 times.
✓ Branch 114 → 115 taken 7091 times.
✓ Branch 118 → 119 taken 7091 times.
|
7091 | seen.addBindValue(account); seen.addBindValue(folder); seen.addBindValue(header.uid); |
| 750 |
2/5✓ Branch 120 → 121 taken 7091 times.
✗ Branch 121 → 122 not taken.
✓ Branch 121 → 128 taken 7091 times.
✗ Branch 122 → 123 not taken.
✗ Branch 123 → 124 not taken.
|
7091 | if (!seen.exec()) return fail(seen.lastError().text()); |
| 751 |
1/1✓ Branch 128 → 129 taken 7091 times.
|
7091 | const bool exists = seen.next(); |
| 752 |
6/8✓ Branch 129 → 130 taken 1302 times.
✓ Branch 129 → 133 taken 5789 times.
✓ Branch 130 → 131 taken 1302 times.
✓ Branch 131 → 132 taken 1302 times.
✓ Branch 134 → 135 taken 1302 times.
✓ Branch 134 → 136 taken 5789 times.
✗ Branch 346 → 347 not taken.
✗ Branch 346 → 348 not taken.
|
7091 | const quint32 previousEpoch = exists ? seen.value(0).toUInt() : 0; |
| 753 |
1/1✓ Branch 136 → 137 taken 7091 times.
|
7091 | seen.finish(); |
| 754 |
4/4✓ Branch 137 → 138 taken 1302 times.
✓ Branch 137 → 140 taken 5789 times.
✓ Branch 138 → 139 taken 1298 times.
✓ Branch 138 → 140 taken 4 times.
|
7091 | if (exists && previousEpoch == uidValidity) continue; |
| 755 |
1/1✓ Branch 140 → 141 taken 5793 times.
|
5793 | const auto tokens = tokenize(header.from, header.subject, header.to); |
| 756 |
2/2✓ Branch 142 → 143 taken 1 time.
✓ Branch 142 → 144 taken 5792 times.
|
5793 | if (tokens.isEmpty()) continue; |
| 757 |
1/1✓ Branch 147 → 148 taken 5792 times.
|
5792 | seen.prepare(QStringLiteral( |
| 758 | "INSERT INTO trained_messages(account,folder,uid,uid_validity) VALUES(?,?,?,?) " | ||
| 759 | "ON CONFLICT(account,folder,uid) DO UPDATE SET uid_validity=excluded.uid_validity")); | ||
| 760 |
2/2✓ Branch 152 → 153 taken 5792 times.
✓ Branch 156 → 157 taken 5792 times.
|
5792 | seen.addBindValue(account); seen.addBindValue(folder); |
| 761 |
2/2✓ Branch 160 → 161 taken 5792 times.
✓ Branch 164 → 165 taken 5792 times.
|
5792 | seen.addBindValue(header.uid); seen.addBindValue(uidValidity); |
| 762 |
5/5✓ Branch 166 → 167 taken 5792 times.
✓ Branch 167 → 168 taken 2 times.
✓ Branch 167 → 174 taken 5790 times.
✓ Branch 168 → 169 taken 2 times.
✓ Branch 169 → 170 taken 2 times.
|
5792 | if (!seen.exec()) return fail(seen.lastError().text()); |
| 763 | // Old ledgers have no reconstructible UIDVALIDITY. Adopt that identity | ||
| 764 | // once rather than double-counting all existing learned documents. | ||
| 765 |
4/4✓ Branch 174 → 175 taken 4 times.
✓ Branch 174 → 177 taken 5786 times.
✓ Branch 175 → 176 taken 2 times.
✓ Branch 175 → 177 taken 2 times.
|
5790 | if (exists && previousEpoch == 0) continue; |
| 766 |
1/1✓ Branch 177 → 178 taken 5788 times.
|
5788 | QSqlQuery update(m_db); |
| 767 |
1/1✓ Branch 181 → 182 taken 5788 times.
|
5788 | update.prepare(QStringLiteral( |
| 768 | "INSERT INTO token_counts(token,folder,count) VALUES(?,?,1) " | ||
| 769 | "ON CONFLICT(token,folder) DO UPDATE SET count=count+1")); | ||
| 770 |
2/2✓ Branch 205 → 186 taken 59931 times.
✓ Branch 205 → 206 taken 5787 times.
|
65718 | for (const auto &token : tokens) { |
| 771 |
2/2✓ Branch 189 → 190 taken 59931 times.
✓ Branch 193 → 194 taken 59931 times.
|
59931 | update.bindValue(0, token); update.bindValue(1, folder); |
| 772 |
5/5✓ Branch 195 → 196 taken 59931 times.
✓ Branch 196 → 197 taken 1 time.
✓ Branch 196 → 203 taken 59930 times.
✓ Branch 197 → 198 taken 1 time.
✓ Branch 198 → 199 taken 1 time.
|
59931 | if (!update.exec()) return fail(update.lastError().text()); |
| 773 | } | ||
| 774 |
1/1✓ Branch 209 → 210 taken 5787 times.
|
5787 | update.prepare(QStringLiteral( |
| 775 | "INSERT INTO folder_doc_counts(folder,doc_count) VALUES(?,1) " | ||
| 776 | "ON CONFLICT(folder) DO UPDATE SET doc_count=doc_count+1")); | ||
| 777 |
1/1✓ Branch 214 → 215 taken 5787 times.
|
5787 | update.addBindValue(folder); |
| 778 |
5/5✓ Branch 216 → 217 taken 5787 times.
✓ Branch 217 → 218 taken 1 time.
✓ Branch 217 → 224 taken 5786 times.
✓ Branch 218 → 219 taken 1 time.
✓ Branch 219 → 220 taken 1 time.
|
5787 | if (!update.exec()) return fail(update.lastError().text()); |
| 779 |
3/8✓ Branch 235 → 236 taken 5786 times.
✗ Branch 235 → 237 not taken.
✓ Branch 237 → 238 taken 3 times.
✓ Branch 237 → 245 taken 5783 times.
✗ Branch 409 → 410 not taken.
✗ Branch 409 → 411 not taken.
✗ Branch 413 → 414 not taken.
✗ Branch 413 → 415 not taken.
|
11572 | if (!update.exec(QStringLiteral( |
| 780 |
4/5✓ Branch 227 → 228 taken 5786 times.
✓ Branch 228 → 229 taken 5784 times.
✓ Branch 228 → 231 taken 2 times.
✓ Branch 233 → 234 taken 5786 times.
✗ Branch 233 → 235 not taken.
|
17356 | "UPDATE meta SET value=CAST(CAST(value AS INTEGER)+1 AS TEXT) WHERE key='total_docs'")) || |
| 781 |
3/3✓ Branch 229 → 230 taken 5784 times.
✓ Branch 230 → 231 taken 1 time.
✓ Branch 230 → 232 taken 5783 times.
|
5784 | update.numRowsAffected() != 1) |
| 782 | 3 | return fail(QStringLiteral("Failed to update predictor document count")); | |
| 783 | 5783 | ++trained; | |
| 784 | 7106 | } | |
| 785 |
1/2✗ Branch 265 → 266 not taken.
✓ Branch 265 → 273 taken 252 times.
|
252 | if (interrupted()) return fail(QStringLiteral("Predictor training cancelled")); |
| 786 |
5/7✓ Branch 273 → 274 taken 236 times.
✓ Branch 273 → 277 taken 16 times.
✓ Branch 274 → 275 taken 236 times.
✗ Branch 275 → 276 not taken.
✓ Branch 275 → 277 taken 236 times.
✗ Branch 278 → 279 not taken.
✓ Branch 278 → 286 taken 252 times.
|
252 | if (trained > 0 && !touchStatisticsRevision()) |
| 787 | ✗ | return fail(QStringLiteral("Failed to update the predictor statistics revision")); | |
| 788 |
2/5✓ Branch 286 → 287 taken 252 times.
✗ Branch 287 → 288 not taken.
✓ Branch 287 → 294 taken 252 times.
✗ Branch 288 → 289 not taken.
✗ Branch 289 → 290 not taken.
|
252 | if (!m_db.commit()) return fail(m_db.lastError().text()); |
| 789 | 252 | rollback.dismiss(); | |
| 790 |
2/2✓ Branch 295 → 296 taken 251 times.
✓ Branch 295 → 297 taken 1 time.
|
252 | if (trainedDocuments) *trainedDocuments = trained; |
| 791 | 252 | return true; | |
| 792 | 264 | } | |
| 793 | |||
| 794 | // --- Prediction --- | ||
| 795 | |||
| 796 | 2759 | QMap<QString, double> FolderPredictor::computeScores( | |
| 797 | const QStringList &tokens, QString *error) const { | ||
| 798 | 2759 | QMap<QString, double> scores; | |
| 799 |
2/2✓ Branch 2 → 3 taken 31 times.
✓ Branch 2 → 4 taken 2728 times.
|
2759 | if (error) error->clear(); |
| 800 | ✗ | const auto fail = [error](const QSqlQuery &query) -> QMap<QString, double> { | |
| 801 | ✗ | if (error) *error = query.lastError().isValid() ? query.lastError().text() | |
| 802 | ✗ | : QStringLiteral("Folder predictor metadata is missing"); | |
| 803 | ✗ | return {}; | |
| 804 | 2759 | }; | |
| 805 | 2759 | Statistics loaded; | |
| 806 | 2759 | const Statistics *statistics = &m_statistics; | |
| 807 |
2/2✓ Branch 4 → 5 taken 31 times.
✓ Branch 4 → 12 taken 2728 times.
|
2759 | if (m_readOnly) { |
| 808 |
6/7✓ Branch 5 → 6 taken 31 times.
✗ Branch 5 → 9 not taken.
✓ Branch 6 → 7 taken 31 times.
✓ Branch 7 → 8 taken 1 time.
✓ Branch 7 → 9 taken 30 times.
✓ Branch 10 → 11 taken 1 time.
✓ Branch 10 → 16 taken 30 times.
|
31 | if (!m_statistics.valid && !loadSnapshotStatistics(error)) return {}; |
| 809 | } else { | ||
| 810 |
3/3✓ Branch 12 → 13 taken 2728 times.
✓ Branch 13 → 14 taken 2 times.
✓ Branch 13 → 15 taken 2726 times.
|
2728 | if (!readStatistics(&loaded, error)) return {}; |
| 811 | 2726 | statistics = &loaded; | |
| 812 | } | ||
| 813 | 2756 | const int totalDocs = statistics->totalDocs; | |
| 814 |
2/2✓ Branch 16 → 17 taken 7 times.
✓ Branch 16 → 18 taken 2749 times.
|
2756 | if (totalDocs < kMinDocsForPrediction) |
| 815 | 7 | return scores; | |
| 816 | 2749 | const QMap<QString, int> &folderDocs = statistics->folderDocs; | |
| 817 |
3/3✓ Branch 18 → 19 taken 2749 times.
✓ Branch 19 → 20 taken 2 times.
✓ Branch 19 → 21 taken 2747 times.
|
2749 | if (folderDocs.isEmpty()) |
| 818 | 2 | return scores; | |
| 819 | 2747 | const int vocabSize = statistics->vocabSize; | |
| 820 | 2747 | const QMap<QString, int> &folderTotalTokens = statistics->folderTotalTokens; | |
| 821 | |||
| 822 | // --- Batch query: load all matching token counts in one SQL call --- | ||
| 823 | // Build IN-clause placeholders | ||
| 824 | 2747 | QStringList uniqueTokens = tokens; | |
| 825 |
1/1✓ Branch 22 → 23 taken 2747 times.
|
2747 | uniqueTokens.removeDuplicates(); |
| 826 | 2747 | QString placeholders; | |
| 827 |
2/2✓ Branch 34 → 24 taken 21816 times.
✓ Branch 34 → 35 taken 2747 times.
|
24563 | for (int i = 0; i < uniqueTokens.size(); ++i) { |
| 828 |
2/2✓ Branch 24 → 25 taken 19069 times.
✓ Branch 24 → 29 taken 2747 times.
|
21816 | if (i > 0) |
| 829 |
1/1✓ Branch 27 → 28 taken 19069 times.
|
19069 | placeholders += QLatin1Char(','); |
| 830 |
1/1✓ Branch 31 → 32 taken 21816 times.
|
21816 | placeholders += QLatin1Char('?'); |
| 831 | } | ||
| 832 | |||
| 833 |
1/1✓ Branch 35 → 36 taken 2747 times.
|
2747 | QSqlQuery batchQ(m_db); |
| 834 |
1/1✓ Branch 41 → 42 taken 2747 times.
|
2747 | batchQ.prepare( |
| 835 | 5494 | QStringLiteral("SELECT token, folder, count FROM token_counts " | |
| 836 | "WHERE token IN (%1)") | ||
| 837 |
1/1✓ Branch 40 → 41 taken 2747 times.
|
5494 | .arg(placeholders)); |
| 838 |
2/2✓ Branch 53 → 46 taken 21816 times.
✓ Branch 53 → 54 taken 2747 times.
|
24563 | for (int i = 0; i < uniqueTokens.size(); ++i) { |
| 839 |
2/2✓ Branch 47 → 48 taken 21816 times.
✓ Branch 49 → 50 taken 21816 times.
|
21816 | batchQ.bindValue(i, uniqueTokens[i]); |
| 840 | } | ||
| 841 |
2/4✓ Branch 54 → 55 taken 2747 times.
✗ Branch 55 → 56 not taken.
✓ Branch 55 → 58 taken 2747 times.
✗ Branch 56 → 57 not taken.
|
2747 | if (!batchQ.exec()) return fail(batchQ); |
| 842 | |||
| 843 | // Build lookup map: (token, folder) → count | ||
| 844 | 2747 | QMap<QPair<QString, QString>, int> tokenCounts; | |
| 845 |
3/3✓ Branch 74 → 75 taken 126980 times.
✓ Branch 75 → 59 taken 124233 times.
✓ Branch 75 → 76 taken 2747 times.
|
126980 | while (batchQ.next()) { |
| 846 |
4/4✓ Branch 62 → 63 taken 124233 times.
✓ Branch 63 → 64 taken 124233 times.
✓ Branch 64 → 65 taken 124233 times.
✓ Branch 66 → 67 taken 124233 times.
|
124233 | tokenCounts[{batchQ.value(0).toString(), batchQ.value(1).toString()}] = |
| 847 |
3/3✓ Branch 59 → 60 taken 124233 times.
✓ Branch 60 → 61 taken 124233 times.
✓ Branch 61 → 62 taken 124233 times.
|
248466 | batchQ.value(2).toInt(); |
| 848 | } | ||
| 849 |
3/5✓ Branch 76 → 77 taken 2747 times.
✓ Branch 77 → 78 taken 2747 times.
✗ Branch 79 → 80 not taken.
✓ Branch 79 → 82 taken 2747 times.
✗ Branch 80 → 81 not taken.
|
2747 | if (batchQ.lastError().isValid()) return fail(batchQ); |
| 850 | |||
| 851 | // Compute log P(folder|tokens) for each folder | ||
| 852 | 2747 | constexpr double kAlpha = 0.1; | |
| 853 |
4/4✓ Branch 82 → 83 taken 2747 times.
✓ Branch 100 → 101 taken 29755 times.
✓ Branch 102 → 84 taken 27008 times.
✓ Branch 102 → 103 taken 2747 times.
|
29755 | for (auto it = folderDocs.constBegin(); it != folderDocs.constEnd(); ++it) { |
| 854 | 27008 | const QString &folder = it.key(); | |
| 855 | 27008 | int docCount = it.value(); | |
| 856 | |||
| 857 | // Prior: P(folder) = doc_count / total_docs | ||
| 858 | 27008 | double logPrior = qLn(static_cast<double>(docCount) / totalDocs); | |
| 859 | |||
| 860 | // Likelihood: P(token|folder) for each token with Laplace smoothing | ||
| 861 | 27008 | double logLikelihood = 0.0; | |
| 862 |
1/1✓ Branch 87 → 88 taken 27008 times.
|
27008 | int totalTokensInFolder = folderTotalTokens.value(folder, 0); |
| 863 | |||
| 864 |
2/2✓ Branch 97 → 90 taken 323949 times.
✓ Branch 97 → 98 taken 27008 times.
|
350957 | for (const QString &token : tokens) { |
| 865 |
1/1✓ Branch 92 → 93 taken 323949 times.
|
323949 | int tokenCount = tokenCounts.value({token, folder}, 0); |
| 866 | |||
| 867 | 323949 | double prob = static_cast<double>(tokenCount + kAlpha) / | |
| 868 | 323949 | (totalTokensInFolder + kAlpha * vocabSize); | |
| 869 | 323949 | logLikelihood += qLn(prob); | |
| 870 | } | ||
| 871 | |||
| 872 |
1/1✓ Branch 98 → 99 taken 27008 times.
|
27008 | scores[folder] = logPrior + logLikelihood; |
| 873 | } | ||
| 874 | |||
| 875 | 2747 | return scores; | |
| 876 | 2759 | } | |
| 877 | |||
| 878 | 31 | bool FolderPredictor::loadSnapshotStatistics(QString *error) const { | |
| 879 | 31 | QString revision; | |
| 880 | { | ||
| 881 |
1/1✓ Branch 2 → 3 taken 31 times.
|
31 | QSqlQuery q(m_db); |
| 882 |
9/16✓ Branch 6 → 7 taken 31 times.
✓ Branch 7 → 8 taken 31 times.
✗ Branch 7 → 11 not taken.
✓ Branch 8 → 9 taken 31 times.
✓ Branch 9 → 10 taken 29 times.
✓ Branch 9 → 11 taken 2 times.
✓ Branch 12 → 13 taken 31 times.
✗ Branch 12 → 14 not taken.
✓ Branch 14 → 15 taken 31 times.
✗ Branch 14 → 16 not taken.
✓ Branch 16 → 17 taken 29 times.
✓ Branch 16 → 23 taken 2 times.
✗ Branch 100 → 101 not taken.
✗ Branch 100 → 102 not taken.
✗ Branch 104 → 105 not taken.
✗ Branch 104 → 106 not taken.
|
62 | if (q.exec(QStringLiteral("SELECT value FROM meta WHERE key='statistics_revision'")) && q.next()) |
| 883 |
2/2✓ Branch 17 → 18 taken 29 times.
✓ Branch 18 → 19 taken 29 times.
|
29 | revision = q.value(0).toString(); |
| 884 | 31 | } | |
| 885 |
6/6✓ Branch 25 → 26 taken 26 times.
✓ Branch 25 → 29 taken 5 times.
✓ Branch 27 → 28 taken 24 times.
✓ Branch 27 → 29 taken 2 times.
✓ Branch 30 → 31 taken 24 times.
✓ Branch 30 → 53 taken 7 times.
|
31 | if (m_statisticsCache && !revision.isEmpty()) { |
| 886 | 24 | QMutexLocker locker(&m_statisticsCache->mutex); | |
| 887 |
6/6✓ Branch 35 → 36 taken 11 times.
✓ Branch 35 → 42 taken 13 times.
✓ Branch 38 → 39 taken 7 times.
✓ Branch 38 → 42 taken 4 times.
✓ Branch 43 → 44 taken 7 times.
✓ Branch 43 → 47 taken 17 times.
|
31 | if (m_statisticsCache->path == m_path && m_statisticsCache->revision == revision && |
| 888 |
1/2✓ Branch 40 → 41 taken 7 times.
✗ Branch 40 → 42 not taken.
|
7 | m_statisticsCache->statistics.valid) { |
| 889 | 7 | m_statistics = m_statisticsCache->statistics; | |
| 890 | 7 | return true; | |
| 891 | } | ||
| 892 | 24 | } | |
| 893 | 24 | Statistics loaded; | |
| 894 |
3/3✓ Branch 53 → 54 taken 24 times.
✓ Branch 54 → 55 taken 1 time.
✓ Branch 54 → 56 taken 23 times.
|
24 | if (!readStatistics(&loaded, error)) return false; |
| 895 | 23 | m_statistics = loaded; | |
| 896 |
6/6✓ Branch 58 → 59 taken 18 times.
✓ Branch 58 → 62 taken 5 times.
✓ Branch 60 → 61 taken 16 times.
✓ Branch 60 → 62 taken 2 times.
✓ Branch 63 → 64 taken 16 times.
✓ Branch 63 → 74 taken 7 times.
|
23 | if (m_statisticsCache && !revision.isEmpty()) { |
| 897 | 16 | QMutexLocker locker(&m_statisticsCache->mutex); | |
| 898 | 16 | m_statisticsCache->path = m_path; | |
| 899 | 16 | m_statisticsCache->revision = revision; | |
| 900 | 16 | m_statisticsCache->statistics = loaded; | |
| 901 | 16 | } | |
| 902 |
8/8✓ Branch 74 → 75 taken 23 times.
✓ Branch 78 → 79 taken 23 times.
✓ Branch 79 → 80 taken 23 times.
✓ Branch 80 → 81 taken 23 times.
✓ Branch 81 → 82 taken 23 times.
✓ Branch 82 → 83 taken 23 times.
✓ Branch 94 → 76 taken 23 times.
✓ Branch 94 → 95 taken 23 times.
|
46 | qCDebug(lcFolderPredictor) << "statistics read docs" << loaded.totalDocs << "vocabulary" << loaded.vocabSize |
| 903 |
6/6✓ Branch 83 → 84 taken 23 times.
✓ Branch 85 → 86 taken 18 times.
✓ Branch 85 → 89 taken 5 times.
✓ Branch 87 → 88 taken 16 times.
✓ Branch 87 → 89 taken 2 times.
✓ Branch 90 → 91 taken 23 times.
|
23 | << "cached" << (m_statisticsCache && !revision.isEmpty()); |
| 904 | 23 | return true; | |
| 905 | 31 | } | |
| 906 | |||
| 907 | 2752 | bool FolderPredictor::readStatistics(Statistics *statistics, QString *error) const { | |
| 908 | 3 | const auto fail = [error](const QSqlQuery &query) { | |
| 909 |
5/10✓ Branch 2 → 3 taken 1 time.
✓ Branch 2 → 19 taken 2 times.
✓ Branch 3 → 4 taken 1 time.
✓ Branch 4 → 5 taken 1 time.
✗ Branch 5 → 6 not taken.
✓ Branch 5 → 8 taken 1 time.
✗ Branch 6 → 7 not taken.
✗ Branch 7 → 11 not taken.
✗ Branch 25 → 26 not taken.
✗ Branch 25 → 27 not taken.
|
6 | if (error) *error = query.lastError().isValid() ? query.lastError().text() |
| 910 |
2/6✓ Branch 13 → 14 taken 1 time.
✗ Branch 13 → 15 not taken.
✗ Branch 15 → 16 not taken.
✓ Branch 15 → 17 taken 1 time.
✗ Branch 21 → 22 not taken.
✗ Branch 21 → 23 not taken.
|
2 | : QStringLiteral("Folder predictor metadata is missing"); |
| 911 | 3 | return false; | |
| 912 | 2752 | }; | |
| 913 |
1/1✓ Branch 2 → 3 taken 2752 times.
|
2752 | QSqlQuery q(m_db); |
| 914 |
11/17✓ Branch 6 → 7 taken 2752 times.
✓ Branch 7 → 8 taken 2750 times.
✓ Branch 7 → 10 taken 2 times.
✓ Branch 8 → 9 taken 2750 times.
✓ Branch 9 → 10 taken 1 time.
✓ Branch 9 → 11 taken 2749 times.
✓ Branch 12 → 13 taken 2752 times.
✗ Branch 12 → 14 not taken.
✓ Branch 14 → 15 taken 2752 times.
✗ Branch 14 → 16 not taken.
✓ Branch 16 → 17 taken 3 times.
✓ Branch 16 → 19 taken 2749 times.
✓ Branch 17 → 18 taken 3 times.
✗ Branch 100 → 101 not taken.
✗ Branch 100 → 102 not taken.
✗ Branch 104 → 105 not taken.
✗ Branch 104 → 106 not taken.
|
5504 | if (!q.exec(QStringLiteral("SELECT value FROM meta WHERE key='total_docs'")) || !q.next()) return fail(q); |
| 915 |
2/2✓ Branch 19 → 20 taken 2749 times.
✓ Branch 20 → 21 taken 2749 times.
|
2749 | statistics->totalDocs = q.value(0).toInt(); |
| 916 |
2/4✓ Branch 25 → 26 taken 2749 times.
✗ Branch 28 → 29 not taken.
✓ Branch 28 → 31 taken 2749 times.
✗ Branch 29 → 30 not taken.
|
2749 | if (!q.exec(QStringLiteral("SELECT folder, doc_count FROM folder_doc_counts WHERE doc_count > 0"))) return fail(q); |
| 917 |
8/8✓ Branch 32 → 33 taken 27001 times.
✓ Branch 33 → 34 taken 27001 times.
✓ Branch 34 → 35 taken 27001 times.
✓ Branch 35 → 36 taken 27001 times.
✓ Branch 36 → 37 taken 27001 times.
✓ Branch 41 → 42 taken 29750 times.
✓ Branch 42 → 32 taken 27001 times.
✓ Branch 42 → 43 taken 2749 times.
|
29750 | while (q.next()) statistics->folderDocs[q.value(0).toString()] = q.value(1).toInt(); |
| 918 |
3/5✓ Branch 43 → 44 taken 2749 times.
✓ Branch 44 → 45 taken 2749 times.
✗ Branch 46 → 47 not taken.
✓ Branch 46 → 49 taken 2749 times.
✗ Branch 47 → 48 not taken.
|
2749 | if (q.lastError().isValid()) return fail(q); |
| 919 | // Vocabulary size (distinct tokens), at least 1 against division by zero. | ||
| 920 |
7/17✓ Branch 52 → 53 taken 2749 times.
✓ Branch 53 → 54 taken 2749 times.
✗ Branch 53 → 56 not taken.
✓ Branch 54 → 55 taken 2749 times.
✗ Branch 55 → 56 not taken.
✓ Branch 55 → 57 taken 2749 times.
✓ Branch 58 → 59 taken 2749 times.
✗ Branch 58 → 60 not taken.
✓ Branch 60 → 61 taken 2749 times.
✗ Branch 60 → 62 not taken.
✗ Branch 62 → 63 not taken.
✓ Branch 62 → 65 taken 2749 times.
✗ Branch 63 → 64 not taken.
✗ Branch 129 → 130 not taken.
✗ Branch 129 → 131 not taken.
✗ Branch 133 → 134 not taken.
✗ Branch 133 → 135 not taken.
|
5498 | if (!q.exec(QStringLiteral("SELECT COUNT(DISTINCT token) FROM token_counts")) || !q.next()) return fail(q); |
| 921 |
2/2✓ Branch 65 → 66 taken 2749 times.
✓ Branch 66 → 67 taken 2749 times.
|
2749 | statistics->vocabSize = qMax(1, q.value(0).toInt()); |
| 922 |
2/4✓ Branch 72 → 73 taken 2749 times.
✗ Branch 75 → 76 not taken.
✓ Branch 75 → 78 taken 2749 times.
✗ Branch 76 → 77 not taken.
|
2749 | if (!q.exec(QStringLiteral("SELECT folder, SUM(count) FROM token_counts GROUP BY folder"))) return fail(q); |
| 923 |
8/8✓ Branch 79 → 80 taken 27002 times.
✓ Branch 80 → 81 taken 27002 times.
✓ Branch 81 → 82 taken 27002 times.
✓ Branch 82 → 83 taken 27002 times.
✓ Branch 83 → 84 taken 27002 times.
✓ Branch 88 → 89 taken 29751 times.
✓ Branch 89 → 79 taken 27002 times.
✓ Branch 89 → 90 taken 2749 times.
|
29751 | while (q.next()) statistics->folderTotalTokens[q.value(0).toString()] = q.value(1).toInt(); |
| 924 |
3/5✓ Branch 90 → 91 taken 2749 times.
✓ Branch 91 → 92 taken 2749 times.
✗ Branch 93 → 94 not taken.
✓ Branch 93 → 96 taken 2749 times.
✗ Branch 94 → 95 not taken.
|
2749 | if (q.lastError().isValid()) return fail(q); |
| 925 | 2749 | statistics->valid = true; | |
| 926 | 2749 | return true; | |
| 927 | 2752 | } | |
| 928 | |||
| 929 | 28 | QString FolderPredictor::predict(const QString &from, const QString &subject, | |
| 930 | const QString &to, | ||
| 931 | double *confidence) const { | ||
| 932 | 28 | QMutexLocker locker(&m_mutex); | |
| 933 |
3/3✓ Branch 3 → 4 taken 28 times.
✓ Branch 4 → 5 taken 2 times.
✓ Branch 4 → 8 taken 26 times.
|
28 | if (!m_db.isOpen()) { |
| 934 |
2/2✓ Branch 5 → 6 taken 1 time.
✓ Branch 5 → 7 taken 1 time.
|
2 | if (confidence) |
| 935 | 1 | *confidence = 0.0; | |
| 936 | 2 | return {}; | |
| 937 | } | ||
| 938 | |||
| 939 |
1/1✓ Branch 8 → 9 taken 26 times.
|
26 | const QStringList tokens = tokenize(from, subject, to); |
| 940 |
2/2✓ Branch 10 → 11 taken 2 times.
✓ Branch 10 → 14 taken 24 times.
|
26 | if (tokens.isEmpty()) { |
| 941 |
2/2✓ Branch 11 → 12 taken 1 time.
✓ Branch 11 → 13 taken 1 time.
|
2 | if (confidence) |
| 942 | 1 | *confidence = 0.0; | |
| 943 | 2 | return {}; | |
| 944 | } | ||
| 945 | |||
| 946 |
1/1✓ Branch 14 → 15 taken 24 times.
|
24 | QMap<QString, double> scores = computeScores(tokens); |
| 947 |
3/3✓ Branch 15 → 16 taken 24 times.
✓ Branch 16 → 17 taken 6 times.
✓ Branch 16 → 20 taken 18 times.
|
24 | if (scores.isEmpty()) { |
| 948 |
2/2✓ Branch 17 → 18 taken 4 times.
✓ Branch 17 → 19 taken 2 times.
|
6 | if (confidence) |
| 949 | 4 | *confidence = 0.0; | |
| 950 | 6 | return {}; | |
| 951 | } | ||
| 952 | |||
| 953 | // Find best and second-best scores | ||
| 954 | 18 | QString bestFolder; | |
| 955 | 18 | double bestScore = -std::numeric_limits<double>::infinity(); | |
| 956 | 18 | double secondBestScore = -std::numeric_limits<double>::infinity(); | |
| 957 | |||
| 958 |
4/4✓ Branch 20 → 21 taken 18 times.
✓ Branch 32 → 33 taken 50 times.
✓ Branch 34 → 22 taken 32 times.
✓ Branch 34 → 35 taken 18 times.
|
50 | for (auto it = scores.constBegin(); it != scores.constEnd(); ++it) { |
| 959 |
2/2✓ Branch 23 → 24 taken 26 times.
✓ Branch 23 → 27 taken 6 times.
|
32 | if (it.value() > bestScore) { |
| 960 | 26 | secondBestScore = bestScore; | |
| 961 | 26 | bestScore = it.value(); | |
| 962 | 26 | bestFolder = it.key(); | |
| 963 |
1/2✓ Branch 28 → 29 taken 6 times.
✗ Branch 28 → 31 not taken.
|
6 | } else if (it.value() > secondBestScore) { |
| 964 | 6 | secondBestScore = it.value(); | |
| 965 | } | ||
| 966 | } | ||
| 967 | |||
| 968 | // Direct-evidence check: does the best folder have actual from-token data? | ||
| 969 | // Without this, senders unknown to the model get nonsensical predictions. | ||
| 970 |
1/1✓ Branch 35 → 36 taken 18 times.
|
18 | QString fullEmail = extractEmail(from); |
| 971 | // S106-199: no sender address, no evidence — no prediction. | ||
| 972 |
2/2✓ Branch 37 → 38 taken 2 times.
✓ Branch 37 → 41 taken 16 times.
|
18 | if (fullEmail.isEmpty()) { |
| 973 |
1/2✓ Branch 38 → 39 taken 2 times.
✗ Branch 38 → 40 not taken.
|
2 | if (confidence) |
| 974 | 2 | *confidence = 0.0; | |
| 975 | 2 | return {}; | |
| 976 | } | ||
| 977 | { | ||
| 978 |
1/1✓ Branch 44 → 45 taken 16 times.
|
16 | QString fullToken = QStringLiteral("from:") + fullEmail; |
| 979 |
1/1✓ Branch 47 → 48 taken 16 times.
|
16 | QSqlQuery evQ(m_db); |
| 980 |
1/1✓ Branch 51 → 52 taken 16 times.
|
16 | evQ.prepare(QStringLiteral( |
| 981 | "SELECT count FROM token_counts WHERE token = ? AND folder = ?")); | ||
| 982 |
1/1✓ Branch 56 → 57 taken 16 times.
|
16 | evQ.bindValue(0, fullToken); |
| 983 |
1/1✓ Branch 60 → 61 taken 16 times.
|
16 | evQ.bindValue(1, bestFolder); |
| 984 |
1/1✓ Branch 62 → 63 taken 16 times.
|
16 | evQ.exec(); |
| 985 |
8/11✓ Branch 63 → 64 taken 16 times.
✓ Branch 64 → 65 taken 11 times.
✓ Branch 64 → 69 taken 5 times.
✓ Branch 65 → 66 taken 11 times.
✓ Branch 66 → 67 taken 11 times.
✓ Branch 67 → 68 taken 11 times.
✗ Branch 67 → 69 not taken.
✓ Branch 70 → 71 taken 11 times.
✓ Branch 70 → 72 taken 5 times.
✗ Branch 147 → 148 not taken.
✗ Branch 147 → 149 not taken.
|
16 | bool hasEvidence = evQ.next() && evQ.value(0).toInt() > 0; |
| 986 | |||
| 987 |
2/2✓ Branch 72 → 73 taken 5 times.
✓ Branch 72 → 94 taken 11 times.
|
16 | if (!hasEvidence) { |
| 988 | // Check if ANY folder has this sender — if not, the model | ||
| 989 | // simply doesn't know this sender at all. | ||
| 990 |
1/1✓ Branch 75 → 76 taken 5 times.
|
5 | evQ.bindValue(0, fullToken); |
| 991 | // Re-prepare for single-column query | ||
| 992 |
1/1✓ Branch 80 → 81 taken 5 times.
|
5 | evQ.prepare(QStringLiteral( |
| 993 | "SELECT 1 FROM token_counts WHERE token = ? LIMIT 1")); | ||
| 994 |
1/1✓ Branch 85 → 86 taken 5 times.
|
5 | evQ.bindValue(0, fullToken); |
| 995 |
1/1✓ Branch 87 → 88 taken 5 times.
|
5 | evQ.exec(); |
| 996 |
3/3✓ Branch 88 → 89 taken 5 times.
✓ Branch 89 → 90 taken 4 times.
✓ Branch 89 → 94 taken 1 time.
|
5 | if (!evQ.next()) { |
| 997 | // Sender completely unknown — no prediction possible | ||
| 998 |
2/2✓ Branch 90 → 91 taken 3 times.
✓ Branch 90 → 92 taken 1 time.
|
4 | if (confidence) |
| 999 | 3 | *confidence = 0.0; | |
| 1000 | 4 | return {}; | |
| 1001 | } | ||
| 1002 | } | ||
| 1003 | 20 | } | |
| 1004 | |||
| 1005 | // Compute confidence using softmax probability | ||
| 1006 |
2/2✓ Branch 102 → 104 taken 10 times.
✓ Branch 102 → 117 taken 2 times.
|
12 | if (confidence) { |
| 1007 |
3/3✓ Branch 104 → 105 taken 10 times.
✓ Branch 105 → 106 taken 3 times.
✓ Branch 105 → 107 taken 7 times.
|
10 | if (scores.size() <= 1) { |
| 1008 | 3 | *confidence = 1.0; | |
| 1009 | } else { | ||
| 1010 | 7 | double sumExp = 0.0; | |
| 1011 |
4/4✓ Branch 107 → 108 taken 7 times.
✓ Branch 112 → 113 taken 21 times.
✓ Branch 114 → 109 taken 14 times.
✓ Branch 114 → 115 taken 7 times.
|
21 | for (auto it = scores.constBegin(); it != scores.constEnd(); ++it) { |
| 1012 | 14 | sumExp += qExp(it.value() - bestScore); | |
| 1013 | } | ||
| 1014 | 7 | *confidence = 1.0 / sumExp; | |
| 1015 |
1/1✓ Branch 115 → 116 taken 7 times.
|
7 | *confidence = qBound(0.0, *confidence, 1.0); |
| 1016 | } | ||
| 1017 | } | ||
| 1018 | |||
| 1019 | 12 | return bestFolder; | |
| 1020 | 28 | } | |
| 1021 | |||
| 1022 | 2738 | QList<QPair<QString, double>> FolderPredictor::predictTop( | |
| 1023 | const QString &from, const QString &subject, const QString &to, | ||
| 1024 | int n, QString *error) const { | ||
| 1025 | 2738 | QMutexLocker locker(&m_mutex); | |
| 1026 | 2738 | QList<QPair<QString, double>> result; | |
| 1027 | |||
| 1028 |
2/2✓ Branch 3 → 4 taken 32 times.
✓ Branch 3 → 5 taken 2706 times.
|
2738 | if (error) error->clear(); |
| 1029 |
3/3✓ Branch 5 → 6 taken 2738 times.
✓ Branch 6 → 7 taken 2 times.
✓ Branch 6 → 16 taken 2736 times.
|
2738 | if (!m_db.isOpen()) { |
| 1030 |
2/2✓ Branch 7 → 8 taken 1 time.
✓ Branch 7 → 15 taken 1 time.
|
3 | if (error) *error = QStringLiteral("Folder predictor is closed"); |
| 1031 | 2 | return result; | |
| 1032 | } | ||
| 1033 | |||
| 1034 |
1/1✓ Branch 16 → 17 taken 2736 times.
|
2736 | const QStringList tokens = tokenize(from, subject, to); |
| 1035 |
2/2✓ Branch 18 → 19 taken 1 time.
✓ Branch 18 → 20 taken 2735 times.
|
2736 | if (tokens.isEmpty()) |
| 1036 | 1 | return result; | |
| 1037 | |||
| 1038 |
1/1✓ Branch 20 → 21 taken 2735 times.
|
2735 | QMap<QString, double> scores = computeScores(tokens, error); |
| 1039 |
3/3✓ Branch 21 → 22 taken 2735 times.
✓ Branch 22 → 23 taken 6 times.
✓ Branch 22 → 24 taken 2729 times.
|
2735 | if (scores.isEmpty()) |
| 1040 | 6 | return result; | |
| 1041 | |||
| 1042 | // Convert to list and sort by score descending | ||
| 1043 |
4/4✓ Branch 24 → 25 taken 2729 times.
✓ Branch 32 → 33 taken 29705 times.
✓ Branch 34 → 26 taken 26976 times.
✓ Branch 34 → 35 taken 2729 times.
|
29705 | for (auto it = scores.constBegin(); it != scores.constEnd(); ++it) { |
| 1044 |
1/1✓ Branch 29 → 30 taken 26976 times.
|
26976 | result.append({it.key(), it.value()}); |
| 1045 | } | ||
| 1046 | |||
| 1047 |
3/3✓ Branch 35 → 36 taken 2729 times.
✓ Branch 36 → 37 taken 2729 times.
✓ Branch 37 → 38 taken 2729 times.
|
2729 | std::sort(result.begin(), result.end(), |
| 1048 | 83040 | [](const QPair<QString, double> &a, | |
| 1049 | const QPair<QString, double> &b) { | ||
| 1050 | 83040 | return a.second > b.second; | |
| 1051 | }); | ||
| 1052 | |||
| 1053 | // Normalize scores to [0, 1] using softmax | ||
| 1054 |
1/2✓ Branch 39 → 40 taken 2729 times.
✗ Branch 39 → 58 not taken.
|
2729 | if (!result.isEmpty()) { |
| 1055 |
1/1✓ Branch 40 → 41 taken 2729 times.
|
2729 | double maxScore = result.first().second; |
| 1056 | 2729 | double sumExp = 0.0; | |
| 1057 |
4/4✓ Branch 41 → 42 taken 2729 times.
✓ Branch 42 → 43 taken 2729 times.
✓ Branch 48 → 44 taken 26976 times.
✓ Branch 48 → 49 taken 2729 times.
|
29705 | for (const auto &pair : result) { |
| 1058 | 26976 | sumExp += qExp(pair.second - maxScore); | |
| 1059 | } | ||
| 1060 |
4/4✓ Branch 49 → 50 taken 2729 times.
✓ Branch 50 → 51 taken 2729 times.
✓ Branch 56 → 52 taken 26976 times.
✓ Branch 56 → 57 taken 2729 times.
|
29705 | for (auto &pair : result) { |
| 1061 | 26976 | pair.second = qExp(pair.second - maxScore) / sumExp; | |
| 1062 | } | ||
| 1063 | } | ||
| 1064 | |||
| 1065 | // Truncate to n entries | ||
| 1066 |
2/2✓ Branch 59 → 60 taken 2694 times.
✓ Branch 59 → 64 taken 35 times.
|
2729 | if (result.size() > n) { |
| 1067 |
1/1✓ Branch 60 → 61 taken 2694 times.
|
2694 | result = result.mid(0, n); |
| 1068 | } | ||
| 1069 | |||
| 1070 | 2729 | return result; | |
| 1071 | 2738 | } | |
| 1072 | |||
| 1073 | // --- Statistics --- | ||
| 1074 | |||
| 1075 | 1574 | int FolderPredictor::totalDocuments() const { | |
| 1076 |
3/3✓ Branch 2 → 3 taken 1574 times.
✓ Branch 3 → 4 taken 2 times.
✓ Branch 3 → 5 taken 1572 times.
|
1574 | if (!m_db.isOpen()) |
| 1077 | 2 | return 0; | |
| 1078 | |||
| 1079 |
1/1✓ Branch 5 → 6 taken 1572 times.
|
1572 | QSqlQuery q(m_db); |
| 1080 |
1/1✓ Branch 9 → 10 taken 1572 times.
|
1572 | q.exec(QStringLiteral( |
| 1081 | "SELECT value FROM meta WHERE key = 'total_docs'")); | ||
| 1082 |
3/3✓ Branch 12 → 13 taken 1572 times.
✓ Branch 13 → 14 taken 1571 times.
✓ Branch 13 → 19 taken 1 time.
|
1572 | if (q.next()) |
| 1083 |
2/2✓ Branch 14 → 15 taken 1571 times.
✓ Branch 15 → 16 taken 1571 times.
|
1571 | return q.value(0).toInt(); |
| 1084 | 1 | return 0; | |
| 1085 | 1572 | } | |
| 1086 | |||
| 1087 | 7 | int FolderPredictor::folderCount() const { | |
| 1088 |
3/3✓ Branch 2 → 3 taken 7 times.
✓ Branch 3 → 4 taken 1 time.
✓ Branch 3 → 5 taken 6 times.
|
7 | if (!m_db.isOpen()) |
| 1089 | 1 | return 0; | |
| 1090 | |||
| 1091 |
1/1✓ Branch 5 → 6 taken 6 times.
|
6 | QSqlQuery q(m_db); |
| 1092 |
1/1✓ Branch 9 → 10 taken 6 times.
|
6 | q.exec(QStringLiteral( |
| 1093 | "SELECT COUNT(*) FROM folder_doc_counts WHERE doc_count > 0")); | ||
| 1094 |
3/3✓ Branch 12 → 13 taken 6 times.
✓ Branch 13 → 14 taken 5 times.
✓ Branch 13 → 19 taken 1 time.
|
6 | if (q.next()) |
| 1095 |
2/2✓ Branch 14 → 15 taken 5 times.
✓ Branch 15 → 16 taken 5 times.
|
5 | return q.value(0).toInt(); |
| 1096 | 1 | return 0; | |
| 1097 | 6 | } | |
| 1098 | |||
| 1099 | 3 | QMap<QString, int> FolderPredictor::folderStats() const { | |
| 1100 | 3 | QMap<QString, int> stats; | |
| 1101 |
3/3✓ Branch 2 → 3 taken 3 times.
✓ Branch 3 → 4 taken 1 time.
✓ Branch 3 → 5 taken 2 times.
|
3 | if (!m_db.isOpen()) |
| 1102 | 1 | return stats; | |
| 1103 | |||
| 1104 |
1/1✓ Branch 5 → 6 taken 2 times.
|
2 | QSqlQuery q(m_db); |
| 1105 |
1/1✓ Branch 9 → 10 taken 2 times.
|
2 | q.exec(QStringLiteral( |
| 1106 | "SELECT folder, doc_count FROM folder_doc_counts " | ||
| 1107 | "WHERE doc_count > 0 ORDER BY doc_count DESC")); | ||
| 1108 |
2/3✓ Branch 22 → 23 taken 2 times.
✗ Branch 23 → 13 not taken.
✓ Branch 23 → 24 taken 2 times.
|
2 | while (q.next()) { |
| 1109 | ✗ | stats[q.value(0).toString()] = q.value(1).toInt(); | |
| 1110 | } | ||
| 1111 | 2 | return stats; | |
| 1112 | 2 | } | |
| 1113 |