diff --git a/servatrice/src/servatrice_database_interface.cpp b/servatrice/src/servatrice_database_interface.cpp index a4f184e38..e25ae5e6c 100644 --- a/servatrice/src/servatrice_database_interface.cpp +++ b/servatrice/src/servatrice_database_interface.cpp @@ -34,6 +34,7 @@ #include #include #include +#include #include #include #include @@ -1977,7 +1978,66 @@ QList Servatrice_DatabaseInterface::getUserSessions(cons return results; } -QList Servatrice_DatabaseInterface::getUserAlts(const QString &userName) +namespace +{ +constexpr int MAX_USER_ALT_DEPTH = 3; +constexpr int MAX_USER_ALTS_PER_LEVEL = 200; +// Bounds per-level work and the prepared-statement cache (keyed by query text, +// so every distinct IN-list size is a new entry on the long-lived server). +// Identifier-heavy frontiers, e.g. VPN users with hundreds of recent IPs, are +// truncated: this tool is deliberately a first-pass sweep, not an audit. +constexpr int MAX_USER_ALT_IDENTIFIERS = 200; + +QString altInClause(const QString &prefix, const QList &values) +{ + QStringList placeholders; + placeholders.reserve(values.size()); + for (int i = 0; i < values.size(); ++i) { + placeholders << prefix + QString::number(i); + } + return QStringLiteral("(") + placeholders.join(QStringLiteral(",")) + QStringLiteral(")"); +} + +void altBindValues(QSqlQuery *query, const QString &prefix, const QList &values) +{ + for (int i = 0; i < values.size(); ++i) { + query->bindValue(prefix + QString::number(i), values.at(i)); + } +} + +struct AltCandidate +{ + QString name; + QString email; + QString clientId; + qint64 registrationTime = 0; + qint64 lastLogin = 0; + bool hasLastLogin = false; + int warnCount = 0; + int banCount = 0; + bool isActive = false; +}; + +// Keeps the first-seen order: callers add the strongest evidence first so the +// cap below drops the least relevant identifiers rather than an arbitrary slice. +void appendIdentifier(QList &list, QSet &seen, const QString &value) +{ + if (!value.isEmpty() && !seen.contains(value)) { + seen.insert(value); + list << value; + } +} + +QList cappedIdentifiers(const QList &values) +{ + if (values.size() <= MAX_USER_ALT_IDENTIFIERS) { + return values; + } + return values.mid(0, MAX_USER_ALT_IDENTIFIERS); +} +} // namespace + +QList Servatrice_DatabaseInterface::getUserAlts(const QString &userName, int depth) { QList results; @@ -1985,63 +2045,344 @@ QList Servatrice_DatabaseInterface::getUserAlts(const QStrin return results; } - // Seed account identifiers used to find related accounts - QSqlQuery *seedQuery = prepareQuery("SELECT email, clientid FROM {prefix}_users WHERE name = :user_name"); + if (depth < 1) { + depth = 1; + } else if (depth > MAX_USER_ALT_DEPTH) { + depth = MAX_USER_ALT_DEPTH; + } + + // Seed must exist; its row anchors the result at depth 0. + QSqlQuery *seedQuery = prepareQuery("SELECT u.name, u.email, u.clientid, UNIX_TIMESTAMP(u.registrationDate), " + "UNIX_TIMESTAMP(a.last_login), " + "(SELECT COUNT(*) FROM {prefix}_warnings w WHERE w.user_id = u.id), " + "(SELECT COUNT(*) FROM {prefix}_bans b WHERE b.user_name = u.name), " + "u.active " + "FROM {prefix}_users u " + "LEFT JOIN {prefix}_user_analytics a ON a.id = u.id " + "WHERE u.name = :user_name"); seedQuery->bindValue(":user_name", userName); if (!execSqlQuery(seedQuery) || !seedQuery->next()) { return results; } - const QString seedEmail = seedQuery->value(0).toString(); - const QString seedClientId = seedQuery->value(1).toString(); - QString queryString = "SELECT u.name, u.email, u.clientid, UNIX_TIMESTAMP(u.registrationDate), " - "UNIX_TIMESTAMP(a.last_login), " - "(SELECT COUNT(*) FROM {prefix}_warnings w WHERE w.user_id = u.id), " - "(SELECT COUNT(*) FROM {prefix}_bans b WHERE b.user_name = u.name), " - "u.active " - "FROM {prefix}_users u " - "LEFT JOIN {prefix}_user_analytics a ON a.id = u.id " - "WHERE u.name = :user_name"; - if (!seedEmail.isEmpty()) { - queryString.append(" OR u.email = :seed_email"); - } - if (!seedClientId.isEmpty()) { - queryString.append(" OR u.clientid = :seed_clientid"); - } - queryString.append(" OR u.name IN (SELECT DISTINCT s.user_name FROM {prefix}_sessions s " - "WHERE s.ip_address IN (SELECT DISTINCT s2.ip_address FROM {prefix}_sessions s2 " - "WHERE s2.user_name = :user_name" - " AND s2.start_time >= DATE_SUB(NOW(), INTERVAL 6 MONTH))" - " AND s.start_time >= DATE_SUB(NOW(), INTERVAL 6 MONTH)) " - "ORDER BY u.name LIMIT 200"); - - QSqlQuery *query = prepareQuery(queryString); - query->bindValue(":user_name", userName); - if (!seedEmail.isEmpty()) { - query->bindValue(":seed_email", seedEmail); - } - if (!seedClientId.isEmpty()) { - query->bindValue(":seed_clientid", seedClientId); - } - - if (!execSqlQuery(query)) { - qCWarning(DatabaseInterfaceLog) << "Failed to collect user alt information: SQL Error"; - return results; - } - - while (query->next()) { + auto appendAlt = [&results](const AltCandidate &candidate, const QString &matchReason, int matchDepth, + int sharedCount) { ServerInfo_UserAlt altDetails; - altDetails.set_user_name(query->value(0).toString().toStdString()); - altDetails.set_email(query->value(1).toString().toStdString()); - altDetails.set_clientid(query->value(2).toString().toStdString()); - altDetails.set_registration_time(query->value(3).toLongLong()); - if (!query->value(4).isNull()) { - altDetails.set_last_login(query->value(4).toLongLong()); + altDetails.set_user_name(candidate.name.toStdString()); + altDetails.set_email(candidate.email.toStdString()); + altDetails.set_clientid(candidate.clientId.toStdString()); + altDetails.set_registration_time(candidate.registrationTime); + if (candidate.hasLastLogin) { + altDetails.set_last_login(candidate.lastLogin); } - altDetails.set_warn_count(query->value(5).toInt()); - altDetails.set_ban_count(query->value(6).toInt()); - altDetails.set_is_active(query->value(7).toBool()); + altDetails.set_warn_count(candidate.warnCount); + altDetails.set_ban_count(candidate.banCount); + altDetails.set_is_active(candidate.isActive); + altDetails.set_match_reason(matchReason.toStdString()); + altDetails.set_depth(static_cast(matchDepth)); + altDetails.set_shared_count(static_cast(sharedCount)); results << altDetails; + }; + + AltCandidate seed{.name = seedQuery->value(0).toString(), + .email = seedQuery->value(1).toString(), + .clientId = seedQuery->value(2).toString(), + .registrationTime = seedQuery->value(3).toLongLong(), + .lastLogin = seedQuery->value(4).toLongLong(), + .hasLastLogin = !seedQuery->value(4).isNull(), + .warnCount = seedQuery->value(5).toInt(), + .banCount = seedQuery->value(6).toInt(), + .isActive = seedQuery->value(7).toBool()}; + appendAlt(seed, QStringLiteral("self"), 0, 1); + + // Identifier lookups are case-insensitive in SQL, so the seed must be keyed by + // its stored casing: otherwise the level-1 query re-lists it as an alt. + QSet visited{seed.name}; + QList frontier{seed.name}; + + // Dynamic IN-lists bypass prepareQuery(): it caches by query text, so every + // distinct list size would leave a prepared statement in the long-lived cache + // and can exhaust MySQL's max_prepared_stmt_count. + const QString dbPrefix = server->getDbPrefix(); + auto prepareDynamicQuery = [&dbPrefix](QSqlQuery &query, const QString &queryText) { + QString text = queryText; + text.replace(QStringLiteral("{prefix}"), dbPrefix); + query.prepare(text); + }; + + for (int level = 1; level <= depth && !frontier.isEmpty(); ++level) { + const QString frontierClause = altInClause(QStringLiteral(":alt_name"), frontier); + + // Current identifiers of the frontier: registration email + client id. + // Kept strongest-first so cappedIdentifiers() drops the least useful ones. + QList emails; + QList clientIds; + QList ips; + QSet emailKeys; + QSet clientIdKeys; + QSet ipKeys; + + QSqlQuery frontierQuery(sqlDatabase); + prepareDynamicQuery(frontierQuery, QStringLiteral("SELECT email, clientid FROM {prefix}_users WHERE name IN ") + + frontierClause); + altBindValues(&frontierQuery, QStringLiteral(":alt_name"), frontier); + if (execSqlQuery(&frontierQuery)) { + while (frontierQuery.next()) { + // Email comparison is case-insensitive in SQL, so normalise the + // keys too or the membership tests below miss real matches. + appendIdentifier(emails, emailKeys, frontierQuery.value(0).toString().toLower()); + appendIdentifier(clientIds, clientIdKeys, frontierQuery.value(1).toString()); + } + } + + // Recent session identifiers: IPs plus historic client ids, most recent + // first so a capped list keeps what the account actually used lately. + QSqlQuery sessionQuery(sqlDatabase); + prepareDynamicQuery(sessionQuery, QStringLiteral("SELECT ip_address, clientid, MAX(start_time) AS last_seen " + "FROM {prefix}_sessions WHERE user_name IN ") + + frontierClause + + QStringLiteral(" AND start_time >= DATE_SUB(NOW(), INTERVAL 6 MONTH) " + "GROUP BY ip_address, clientid ORDER BY last_seen DESC")); + altBindValues(&sessionQuery, QStringLiteral(":alt_name"), frontier); + if (execSqlQuery(&sessionQuery)) { + while (sessionQuery.next()) { + appendIdentifier(ips, ipKeys, sessionQuery.value(0).toString()); + appendIdentifier(clientIds, clientIdKeys, sessionQuery.value(1).toString()); + } + } + + if (emails.isEmpty() && clientIds.isEmpty() && ips.isEmpty()) { + break; + } + + // Accounts sharing any frontier identifier. Historic client ids are matched + // through past sessions so identifier changes do not hide older alts. + QString candidateString = + QStringLiteral("SELECT u.name, u.email, u.clientid, UNIX_TIMESTAMP(u.registrationDate), " + "UNIX_TIMESTAMP(a.last_login), " + "(SELECT COUNT(*) FROM {prefix}_warnings w WHERE w.user_id = u.id), " + "(SELECT COUNT(*) FROM {prefix}_bans b WHERE b.user_name = u.name), " + "u.active " + "FROM {prefix}_users u " + "LEFT JOIN {prefix}_user_analytics a ON a.id = u.id " + "WHERE "); + QStringList conditions; + const QList emailList = cappedIdentifiers(emails); + const QList clientIdList = cappedIdentifiers(clientIds); + const QList ipList = cappedIdentifiers(ips); + if (!emailList.isEmpty()) { + conditions << QStringLiteral("u.email IN ") + altInClause(QStringLiteral(":alt_email"), emailList); + } + if (!clientIdList.isEmpty()) { + conditions << QStringLiteral("u.clientid IN ") + altInClause(QStringLiteral(":alt_cid"), clientIdList); + } + if (!ipList.isEmpty()) { + conditions << QStringLiteral("u.name IN (SELECT DISTINCT s.user_name FROM {prefix}_sessions s WHERE " + "s.ip_address IN ") + + altInClause(QStringLiteral(":alt_ip"), ipList) + + QStringLiteral(" AND s.start_time >= DATE_SUB(NOW(), INTERVAL 6 MONTH))"); + } + if (!clientIdList.isEmpty()) { + conditions << QStringLiteral("u.name IN (SELECT DISTINCT s.user_name FROM {prefix}_sessions s WHERE " + "s.clientid IN ") + + altInClause(QStringLiteral(":alt_hcid"), clientIdList) + + QStringLiteral(" AND s.start_time >= DATE_SUB(NOW(), INTERVAL 6 MONTH))"); + } + candidateString += conditions.join(QStringLiteral(" OR ")); + // Already-visited accounts match their own identifiers at every level and + // are dropped below, so widen the row limit by that count: otherwise a + // second level can return only visited rows and no new candidates. + candidateString += + QStringLiteral(" ORDER BY u.name LIMIT ") + QString::number(MAX_USER_ALTS_PER_LEVEL + visited.size()); + + QSqlQuery candidateQuery(sqlDatabase); + prepareDynamicQuery(candidateQuery, candidateString); + altBindValues(&candidateQuery, QStringLiteral(":alt_email"), emailList); + altBindValues(&candidateQuery, QStringLiteral(":alt_cid"), clientIdList); + altBindValues(&candidateQuery, QStringLiteral(":alt_ip"), ipList); + altBindValues(&candidateQuery, QStringLiteral(":alt_hcid"), clientIdList); + if (!execSqlQuery(&candidateQuery)) { + qCWarning(DatabaseInterfaceLog) << "Failed to collect user alt information: SQL Error"; + return results; + } + + QList candidates; + while (candidateQuery.next()) { + AltCandidate candidate; + candidate.name = candidateQuery.value(0).toString(); + if (visited.contains(candidate.name)) { + continue; + } + candidate.email = candidateQuery.value(1).toString(); + candidate.clientId = candidateQuery.value(2).toString(); + candidate.registrationTime = candidateQuery.value(3).toLongLong(); + candidate.hasLastLogin = !candidateQuery.value(4).isNull(); + candidate.lastLogin = candidateQuery.value(4).toLongLong(); + candidate.warnCount = candidateQuery.value(5).toInt(); + candidate.banCount = candidateQuery.value(6).toInt(); + candidate.isActive = candidateQuery.value(7).toBool(); + candidates << candidate; + } + + // Resolve session-level evidence (shared IP or historic client id) for + // candidates that do not match on current email or client id. + QMap candidateIp; + QMap candidateHistoricClientId; + QList unexplained; + for (const AltCandidate &candidate : candidates) { + const bool emailMatch = !candidate.email.isEmpty() && emailKeys.contains(candidate.email.toLower()); + const bool clientIdMatch = !candidate.clientId.isEmpty() && clientIdKeys.contains(candidate.clientId); + if (!emailMatch && !clientIdMatch) { + unexplained << candidate.name; + } + } + if (!unexplained.isEmpty() && (!ipKeys.isEmpty() || !clientIdKeys.isEmpty())) { + QString evidenceString = + QStringLiteral("SELECT DISTINCT user_name, ip_address, clientid FROM {prefix}_sessions WHERE " + "user_name IN ") + + altInClause(QStringLiteral(":alt_ev"), unexplained) + QStringLiteral(" AND ("); + QStringList evidenceConditions; + if (!ipList.isEmpty()) { + evidenceConditions << QStringLiteral("ip_address IN ") + + altInClause(QStringLiteral(":alt_eip"), ipList); + } + if (!clientIdList.isEmpty()) { + evidenceConditions << QStringLiteral("clientid IN ") + + altInClause(QStringLiteral(":alt_ecid"), clientIdList); + } + evidenceString += evidenceConditions.join(QStringLiteral(" OR ")); + evidenceString += QStringLiteral(") AND start_time >= DATE_SUB(NOW(), INTERVAL 6 MONTH)"); + + QSqlQuery evidenceQuery(sqlDatabase); + prepareDynamicQuery(evidenceQuery, evidenceString); + altBindValues(&evidenceQuery, QStringLiteral(":alt_ev"), unexplained); + altBindValues(&evidenceQuery, QStringLiteral(":alt_eip"), ipList); + altBindValues(&evidenceQuery, QStringLiteral(":alt_ecid"), clientIdList); + if (execSqlQuery(&evidenceQuery)) { + while (evidenceQuery.next()) { + const QString name = evidenceQuery.value(0).toString(); + const QString ip = evidenceQuery.value(1).toString(); + const QString clientId = evidenceQuery.value(2).toString(); + if (!ip.isEmpty() && ipKeys.contains(ip) && !candidateIp.contains(name)) { + candidateIp.insert(name, ip); + } + if (!clientId.isEmpty() && clientIdKeys.contains(clientId) && + !candidateHistoricClientId.contains(name)) { + candidateHistoricClientId.insert(name, clientId); + } + } + } + } + + // Fan-out counts so moderators can spot VPN/shared-network false positives. + QMap emailCounts; + QMap clientIdCounts; + QMap historicClientIdCounts; + QMap ipCounts; + QSet neededEmails; + QSet neededClientIds; + QSet neededHistoricClientIds; + QSet neededIps; + for (const AltCandidate &candidate : candidates) { + if (!candidate.email.isEmpty() && emailKeys.contains(candidate.email.toLower())) { + neededEmails << candidate.email.toLower(); + } else if (!candidate.clientId.isEmpty() && clientIdKeys.contains(candidate.clientId)) { + neededClientIds << candidate.clientId; + } else if (candidateHistoricClientId.contains(candidate.name)) { + neededHistoricClientIds << candidateHistoricClientId.value(candidate.name); + } else if (candidateIp.contains(candidate.name)) { + neededIps << candidateIp.value(candidate.name); + } + } + if (!neededEmails.isEmpty()) { + const QList neededEmailList = neededEmails.values(); + QSqlQuery countQuery(sqlDatabase); + prepareDynamicQuery(countQuery, QStringLiteral("SELECT email, COUNT(*) FROM {prefix}_users WHERE " + "email IN ") + + altInClause(QStringLiteral(":alt_ce"), neededEmailList) + + QStringLiteral(" GROUP BY email")); + altBindValues(&countQuery, QStringLiteral(":alt_ce"), neededEmailList); + if (execSqlQuery(&countQuery)) { + while (countQuery.next()) { + emailCounts.insert(countQuery.value(0).toString().toLower(), countQuery.value(1).toInt()); + } + } + } + if (!neededClientIds.isEmpty()) { + const QList neededClientIdList = neededClientIds.values(); + QSqlQuery countQuery(sqlDatabase); + prepareDynamicQuery(countQuery, QStringLiteral("SELECT clientid, COUNT(*) FROM {prefix}_users WHERE " + "clientid IN ") + + altInClause(QStringLiteral(":alt_cc"), neededClientIdList) + + QStringLiteral(" GROUP BY clientid")); + altBindValues(&countQuery, QStringLiteral(":alt_cc"), neededClientIdList); + if (execSqlQuery(&countQuery)) { + while (countQuery.next()) { + clientIdCounts.insert(countQuery.value(0).toString(), countQuery.value(1).toInt()); + } + } + } + // Historic client ids are matched through sessions, so their fan-out must + // come from _sessions too: _users only knows each account's current id. + if (!neededHistoricClientIds.isEmpty()) { + const QList neededHistoricClientIdList = neededHistoricClientIds.values(); + QSqlQuery countQuery(sqlDatabase); + prepareDynamicQuery(countQuery, + QStringLiteral("SELECT clientid, COUNT(DISTINCT user_name) FROM {prefix}_sessions " + "WHERE clientid IN ") + + altInClause(QStringLiteral(":alt_hcc"), neededHistoricClientIdList) + + QStringLiteral(" AND start_time >= DATE_SUB(NOW(), INTERVAL 6 MONTH) " + "GROUP BY clientid")); + altBindValues(&countQuery, QStringLiteral(":alt_hcc"), neededHistoricClientIdList); + if (execSqlQuery(&countQuery)) { + while (countQuery.next()) { + historicClientIdCounts.insert(countQuery.value(0).toString(), countQuery.value(1).toInt()); + } + } + } + if (!neededIps.isEmpty()) { + const QList neededIpList = neededIps.values(); + QSqlQuery countQuery(sqlDatabase); + prepareDynamicQuery(countQuery, + QStringLiteral("SELECT ip_address, COUNT(DISTINCT user_name) FROM {prefix}_sessions " + "WHERE ip_address IN ") + + altInClause(QStringLiteral(":alt_ci"), neededIpList) + + QStringLiteral(" AND start_time >= DATE_SUB(NOW(), INTERVAL 6 MONTH) " + "GROUP BY ip_address")); + altBindValues(&countQuery, QStringLiteral(":alt_ci"), neededIpList); + if (execSqlQuery(&countQuery)) { + while (countQuery.next()) { + ipCounts.insert(countQuery.value(0).toString(), countQuery.value(1).toInt()); + } + } + } + + QList nextFrontier; + for (const AltCandidate &candidate : candidates) { + QString matchReason; + int sharedCount = 1; + if (!candidate.email.isEmpty() && emailKeys.contains(candidate.email.toLower())) { + matchReason = QStringLiteral("email"); + sharedCount = qMax(1, emailCounts.value(candidate.email.toLower(), 1)); + } else if (!candidate.clientId.isEmpty() && clientIdKeys.contains(candidate.clientId)) { + matchReason = QStringLiteral("clientid:") + candidate.clientId; + sharedCount = qMax(1, clientIdCounts.value(candidate.clientId, 1)); + } else if (candidateHistoricClientId.contains(candidate.name)) { + const QString historicId = candidateHistoricClientId.value(candidate.name); + matchReason = QStringLiteral("clientid:") + historicId; + sharedCount = qMax(1, historicClientIdCounts.value(historicId, 1)); + } else if (candidateIp.contains(candidate.name)) { + const QString ip = candidateIp.value(candidate.name); + matchReason = QStringLiteral("ip:") + ip; + sharedCount = qMax(1, ipCounts.value(ip, 1)); + } else { + continue; + } + appendAlt(candidate, matchReason, level, sharedCount); + visited.insert(candidate.name); + nextFrontier << candidate.name; + } + frontier = nextFrontier; } return results; diff --git a/servatrice/src/servatrice_database_interface.h b/servatrice/src/servatrice_database_interface.h index eb74063ac..327906b3c 100644 --- a/servatrice/src/servatrice_database_interface.h +++ b/servatrice/src/servatrice_database_interface.h @@ -224,7 +224,7 @@ public: const QString &beforeTime = QString(), bool *hasMore = nullptr); QList getUserSessions(const QString &userName, int limit); - QList getUserAlts(const QString &userName); + QList getUserAlts(const QString &userName, int depth = 1); QList getModeratorLastLogins(); // Uptime snapshot as recorded by Servatrice::statusUpdate() into the diff --git a/servatrice/src/serversocketinterface.cpp b/servatrice/src/serversocketinterface.cpp index c38d790b1..b49b94ec0 100644 --- a/servatrice/src/serversocketinterface.cpp +++ b/servatrice/src/serversocketinterface.cpp @@ -2595,7 +2595,8 @@ Response::ResponseCode AbstractServerSocketInterface::cmdGetUserAlts(const Comma } Response_UserAlts *re = new Response_UserAlts; - const QList alts = sqlInterface->getUserAlts(userName); + const int depth = cmd.has_depth() ? static_cast(cmd.depth()) : 1; + const QList alts = sqlInterface->getUserAlts(userName, depth); for (const ServerInfo_UserAlt &alt : alts) { re->add_alts()->CopyFrom(alt); }