@berthojoris/mcp-mysql-server 1.15.0 → 1.16.0

package/dist/tools/smartDiscoveryTools.js

@@ -0,0 +1,750 @@
+"use strict";
+var __importDefault = (this && this.__importDefault) || function (mod) {
+    return (mod && mod.__esModule) ? mod : { "default": mod };
+};
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.SmartDiscoveryTools = void 0;
+const connection_1 = __importDefault(require("../db/connection"));
+const config_1 = require("../config/config");
+/**
+ * Smart Data Discovery Agent
+ * Finds relevant tables/columns using semantic search and pattern matching
+ * Discovers hidden relationships automatically
+ */
+class SmartDiscoveryTools {
+    constructor(security) {
+        this.db = connection_1.default.getInstance();
+        this.security = security;
+    }
+    /**
+     * Validate database access - ensures only the connected database can be accessed
+     */
+    validateDatabaseAccess(requestedDatabase) {
+        const connectedDatabase = config_1.dbConfig.database;
+        if (!connectedDatabase) {
+            return {
+                valid: false,
+                database: "",
+                error: "No database specified in connection string. Cannot access any database.",
+            };
+        }
+        if (!requestedDatabase) {
+            return {
+                valid: true,
+                database: connectedDatabase,
+            };
+        }
+        if (requestedDatabase !== connectedDatabase) {
+            return {
+                valid: false,
+                database: "",
+                error: `Access denied. You can only access the connected database '${connectedDatabase}'. Requested database '${requestedDatabase}' is not allowed.`,
+            };
+        }
+        return {
+            valid: true,
+            database: connectedDatabase,
+        };
+    }
+    /**
+     * Smart search across database objects (tables, columns, data patterns)
+     */
+    async smartSearch(params) {
+        const startTime = Date.now();
+        try {
+            const dbValidation = this.validateDatabaseAccess(params?.database);
+            if (!dbValidation.valid) {
+                return { status: "error", error: dbValidation.error };
+            }
+            const { search_term, search_type = "all", similarity_threshold = 0.3, include_sample_data = false, max_results = 20, } = params;
+            const database = dbValidation.database;
+            if (!search_term?.trim()) {
+                return { status: "error", error: "search_term is required" };
+            }
+            const searchTermLower = search_term.toLowerCase().trim();
+            const searchTokens = this.tokenize(searchTermLower);
+            // Get all tables and columns
+            const tablesResult = await this.db.query(`SELECT TABLE_NAME, TABLE_ROWS
+         FROM INFORMATION_SCHEMA.TABLES
+         WHERE TABLE_SCHEMA = ? AND TABLE_TYPE = 'BASE TABLE'`, [database]);
+            const columnsResult = await this.db.query(`SELECT TABLE_NAME, COLUMN_NAME, DATA_TYPE, COLUMN_KEY
+         FROM INFORMATION_SCHEMA.COLUMNS
+         WHERE TABLE_SCHEMA = ?
+         ORDER BY TABLE_NAME, ORDINAL_POSITION`, [database]);
+            // Get foreign key relationships
+            const fkResult = await this.db.query(`SELECT TABLE_NAME, COLUMN_NAME, REFERENCED_TABLE_NAME, REFERENCED_COLUMN_NAME
+         FROM INFORMATION_SCHEMA.KEY_COLUMN_USAGE
+         WHERE TABLE_SCHEMA = ? AND REFERENCED_TABLE_NAME IS NOT NULL`, [database]);
+            const results = {
+                tables: [],
+                columns: [],
+                data_patterns: [],
+                relationships: [],
+            };
+            // Search tables
+            if (search_type === "all" || search_type === "table") {
+                for (const table of tablesResult) {
+                    const tableName = table.TABLE_NAME.toLowerCase();
+                    const score = this.calculateRelevanceScore(searchTokens, tableName, searchTermLower);
+                    if (score >= similarity_threshold) {
+                        const tableColumns = columnsResult.filter((c) => c.TABLE_NAME === table.TABLE_NAME);
+                        const matchingCols = tableColumns
+                            .filter((c) => this.calculateRelevanceScore(searchTokens, c.COLUMN_NAME.toLowerCase(), searchTermLower) >= similarity_threshold)
+                            .map((c) => c.COLUMN_NAME);
+                        results.tables.push({
+                            name: table.TABLE_NAME,
+                            relevance_score: Math.round(score * 100) / 100,
+                            match_reason: this.getMatchReason(searchTokens, tableName),
+                            column_count: tableColumns.length,
+                            row_estimate: parseInt(table.TABLE_ROWS || "0", 10) || 0,
+                            matching_columns: matchingCols.length > 0 ? matchingCols : undefined,
+                        });
+                    }
+                }
+            }
+            // Search columns
+            if (search_type === "all" || search_type === "column") {
+                for (const column of columnsResult) {
+                    const colName = column.COLUMN_NAME.toLowerCase();
+                    const score = this.calculateRelevanceScore(searchTokens, colName, searchTermLower);
+                    if (score >= similarity_threshold) {
+                        const colResult = {
+                            table_name: column.TABLE_NAME,
+                            column_name: column.COLUMN_NAME,
+                            data_type: column.DATA_TYPE,
+                            relevance_score: Math.round(score * 100) / 100,
+                            match_reason: this.getMatchReason(searchTokens, colName),
+                        };
+                        // Include sample data if requested
+                        if (include_sample_data) {
+                            try {
+                                const samples = await this.db.query(`SELECT DISTINCT \`${column.COLUMN_NAME}\` 
+                   FROM \`${database}\`.\`${column.TABLE_NAME}\`
+                   WHERE \`${column.COLUMN_NAME}\` IS NOT NULL
+                   LIMIT 5`);
+                                colResult.sample_values = samples.map((s) => s[column.COLUMN_NAME]);
+                            }
+                            catch {
+                                // Ignore errors when fetching samples
+                            }
+                        }
+                        results.columns.push(colResult);
+                    }
+                }
+            }
+            // Search for data patterns (look for the search term in actual data)
+            if (search_type === "all" || search_type === "data_pattern") {
+                // Only search text and varchar columns for data patterns
+                const textColumns = columnsResult.filter((c) => ["varchar", "text", "char", "longtext", "mediumtext", "tinytext", "enum", "set"].includes(c.DATA_TYPE.toLowerCase()));
+                // Limit to prevent too many queries
+                const columnsToSearch = textColumns.slice(0, 20);
+                for (const column of columnsToSearch) {
+                    try {
+                        const patternQuery = `
+              SELECT DISTINCT \`${column.COLUMN_NAME}\`
+              FROM \`${database}\`.\`${column.TABLE_NAME}\`
+              WHERE LOWER(\`${column.COLUMN_NAME}\`) LIKE ?
+              LIMIT 5
+            `;
+                        const matches = await this.db.query(patternQuery, [
+                            `%${searchTermLower}%`,
+                        ]);
+                        if (matches.length > 0) {
+                            results.data_patterns.push({
+                                table_name: column.TABLE_NAME,
+                                column_name: column.COLUMN_NAME,
+                                pattern_type: "CONTAINS",
+                                description: `Found ${matches.length}+ values containing "${search_term}"`,
+                                sample_matches: matches.map((m) => m[column.COLUMN_NAME]),
+                            });
+                        }
+                    }
+                    catch {
+                        // Ignore errors when searching patterns
+                    }
+                }
+            }
+            // Search relationships
+            if (search_type === "all" || search_type === "relationship") {
+                // Explicit foreign keys
+                for (const fk of fkResult) {
+                    const score = Math.max(this.calculateRelevanceScore(searchTokens, fk.TABLE_NAME.toLowerCase(), searchTermLower), this.calculateRelevanceScore(searchTokens, fk.REFERENCED_TABLE_NAME.toLowerCase(), searchTermLower), this.calculateRelevanceScore(searchTokens, fk.COLUMN_NAME.toLowerCase(), searchTermLower));
+                    if (score >= similarity_threshold) {
+                        results.relationships.push({
+                            from_table: fk.TABLE_NAME,
+                            from_column: fk.COLUMN_NAME,
+                            to_table: fk.REFERENCED_TABLE_NAME,
+                            to_column: fk.REFERENCED_COLUMN_NAME,
+                            relationship_type: "FOREIGN_KEY",
+                            confidence: 1.0,
+                        });
+                    }
+                }
+                // Discover implicit relationships (naming conventions)
+                const implicitRels = this.discoverImplicitRelationships(tablesResult, columnsResult, searchTokens, similarity_threshold);
+                results.relationships.push(...implicitRels);
+            }
+            // Sort and limit results
+            results.tables.sort((a, b) => b.relevance_score - a.relevance_score);
+            results.columns.sort((a, b) => b.relevance_score - a.relevance_score);
+            // Apply max_results limit
+            results.tables = results.tables.slice(0, max_results);
+            results.columns = results.columns.slice(0, max_results);
+            results.data_patterns = results.data_patterns.slice(0, max_results);
+            results.relationships = results.relationships.slice(0, max_results);
+            const totalMatches = results.tables.length +
+                results.columns.length +
+                results.data_patterns.length +
+                results.relationships.length;
+            return {
+                status: "success",
+                data: {
+                    search_term,
+                    search_type,
+                    results,
+                    total_matches: totalMatches,
+                    search_time_ms: Date.now() - startTime,
+                },
+            };
+        }
+        catch (error) {
+            return {
+                status: "error",
+                error: error.message,
+            };
+        }
+    }
+    /**
+     * Find similar columns across tables (potential join candidates)
+     */
+    async findSimilarColumns(params) {
+        try {
+            const dbValidation = this.validateDatabaseAccess(params?.database);
+            if (!dbValidation.valid) {
+                return { status: "error", error: dbValidation.error };
+            }
+            const { column_name, table_name, include_data_comparison = false, max_results = 20, } = params;
+            const database = dbValidation.database;
+            // Get all columns
+            const allColumns = await this.db.query(`SELECT TABLE_NAME, COLUMN_NAME, DATA_TYPE, COLUMN_KEY
+         FROM INFORMATION_SCHEMA.COLUMNS
+         WHERE TABLE_SCHEMA = ?
+         ORDER BY TABLE_NAME, ORDINAL_POSITION`, [database]);
+            const similarColumns = [];
+            const potentialJoins = [];
+            let referenceColumn;
+            if (column_name && table_name) {
+                // Find reference column
+                const refCol = allColumns.find((c) => c.TABLE_NAME.toLowerCase() === table_name.toLowerCase() &&
+                    c.COLUMN_NAME.toLowerCase() === column_name.toLowerCase());
+                if (!refCol) {
+                    return {
+                        status: "error",
+                        error: `Column '${column_name}' not found in table '${table_name}'`,
+                    };
+                }
+                referenceColumn = {
+                    table: refCol.TABLE_NAME,
+                    column: refCol.COLUMN_NAME,
+                    data_type: refCol.DATA_TYPE,
+                };
+                // Find similar columns
+                const refNameTokens = this.tokenize(refCol.COLUMN_NAME.toLowerCase());
+                for (const col of allColumns) {
+                    if (col.TABLE_NAME === refCol.TABLE_NAME &&
+                        col.COLUMN_NAME === refCol.COLUMN_NAME) {
+                        continue;
+                    }
+                    const colNameLower = col.COLUMN_NAME.toLowerCase();
+                    const nameSimilarity = this.calculateNameSimilarity(refCol.COLUMN_NAME.toLowerCase(), colNameLower);
+                    const typeSimilarity = col.DATA_TYPE === refCol.DATA_TYPE ? 0.3 : 0;
+                    const totalScore = nameSimilarity * 0.7 + typeSimilarity;
+                    if (totalScore >= 0.3) {
+                        const simCol = {
+                            table_name: col.TABLE_NAME,
+                            column_name: col.COLUMN_NAME,
+                            data_type: col.DATA_TYPE,
+                            similarity_score: Math.round(totalScore * 100) / 100,
+                            similarity_type: this.getSimilarityType(refCol.COLUMN_NAME, col.COLUMN_NAME),
+                        };
+                        // Compare data if requested
+                        if (include_data_comparison &&
+                            col.DATA_TYPE === refCol.DATA_TYPE) {
+                            try {
+                                const overlapResult = await this.calculateDataOverlap(database, refCol.TABLE_NAME, refCol.COLUMN_NAME, col.TABLE_NAME, col.COLUMN_NAME);
+                                simCol.data_overlap_percentage = overlapResult;
+                            }
+                            catch {
+                                // Ignore data comparison errors
+                            }
+                        }
+                        similarColumns.push(simCol);
+                        // Add as potential join if high confidence
+                        if (totalScore >= 0.6 || (simCol.data_overlap_percentage || 0) > 50) {
+                            potentialJoins.push({
+                                table1: refCol.TABLE_NAME,
+                                column1: refCol.COLUMN_NAME,
+                                table2: col.TABLE_NAME,
+                                column2: col.COLUMN_NAME,
+                                confidence: Math.round(Math.max(totalScore, (simCol.data_overlap_percentage || 0) / 100) * 100) / 100,
+                                reason: (simCol.data_overlap_percentage || 0) > 50
+                                    ? "High data overlap"
+                                    : "Similar column names",
+                            });
+                        }
+                    }
+                }
+            }
+            else {
+                // Find all potential join candidates based on naming patterns
+                const columnGroups = new Map();
+                for (const col of allColumns) {
+                    // Group by normalized column name
+                    const normalized = this.normalizeColumnName(col.COLUMN_NAME);
+                    if (!columnGroups.has(normalized)) {
+                        columnGroups.set(normalized, []);
+                    }
+                    columnGroups.get(normalized).push(col);
+                }
+                // Find groups with multiple columns (potential joins)
+                for (const [normalizedName, columns] of columnGroups) {
+                    if (columns.length > 1) {
+                        // Generate pairs
+                        for (let i = 0; i < columns.length; i++) {
+                            for (let j = i + 1; j < columns.length; j++) {
+                                const col1 = columns[i];
+                                const col2 = columns[j];
+                                if (col1.TABLE_NAME !== col2.TABLE_NAME &&
+                                    col1.DATA_TYPE === col2.DATA_TYPE) {
+                                    const confidence = col1.COLUMN_NAME === col2.COLUMN_NAME ? 0.9 : 0.7;
+                                    potentialJoins.push({
+                                        table1: col1.TABLE_NAME,
+                                        column1: col1.COLUMN_NAME,
+                                        table2: col2.TABLE_NAME,
+                                        column2: col2.COLUMN_NAME,
+                                        confidence,
+                                        reason: col1.COLUMN_NAME === col2.COLUMN_NAME
+                                            ? "Identical column names"
+                                            : `Similar names: ${col1.COLUMN_NAME} ~ ${col2.COLUMN_NAME}`,
+                                    });
+                                }
+                            }
+                        }
+                    }
+                }
+                // Also look for id/foreign key patterns
+                for (const col of allColumns) {
+                    if (col.COLUMN_NAME.toLowerCase().endsWith("_id")) {
+                        const potentialTable = col.COLUMN_NAME.slice(0, -3);
+                        const matchingTable = allColumns.find((c) => c.TABLE_NAME.toLowerCase() === potentialTable.toLowerCase() &&
+                            c.COLUMN_KEY === "PRI");
+                        if (matchingTable && matchingTable.TABLE_NAME !== col.TABLE_NAME) {
+                            potentialJoins.push({
+                                table1: col.TABLE_NAME,
+                                column1: col.COLUMN_NAME,
+                                table2: matchingTable.TABLE_NAME,
+                                column2: matchingTable.COLUMN_NAME,
+                                confidence: 0.85,
+                                reason: "Foreign key naming convention (_id suffix)",
+                            });
+                        }
+                    }
+                }
+            }
+            // Sort by similarity/confidence
+            similarColumns.sort((a, b) => b.similarity_score - a.similarity_score);
+            potentialJoins.sort((a, b) => b.confidence - a.confidence);
+            return {
+                status: "success",
+                data: {
+                    reference_column: referenceColumn,
+                    similar_columns: similarColumns.slice(0, max_results),
+                    potential_joins: potentialJoins.slice(0, max_results),
+                },
+            };
+        }
+        catch (error) {
+            return {
+                status: "error",
+                error: error.message,
+            };
+        }
+    }
+    /**
+     * Discover data relationships and patterns
+     */
+    async discoverDataPatterns(params) {
+        try {
+            const dbValidation = this.validateDatabaseAccess(params?.database);
+            if (!dbValidation.valid) {
+                return { status: "error", error: dbValidation.error };
+            }
+            const { table_name, pattern_types = ["unique", "null", "duplicate", "format", "range"], max_columns = 20, } = params;
+            const database = dbValidation.database;
+            // Validate table name
+            if (!this.security.validateIdentifier(table_name).valid) {
+                return { status: "error", error: "Invalid table name" };
+            }
+            // Get columns
+            const columns = await this.db.query(`SELECT COLUMN_NAME, DATA_TYPE, IS_NULLABLE, COLUMN_KEY
+         FROM INFORMATION_SCHEMA.COLUMNS
+         WHERE TABLE_SCHEMA = ? AND TABLE_NAME = ?
+         ORDER BY ORDINAL_POSITION
+         LIMIT ?`, [database, table_name, max_columns]);
+            if (columns.length === 0) {
+                return {
+                    status: "error",
+                    error: `Table '${table_name}' not found or has no columns`,
+                };
+            }
+            // Get row count
+            const countResult = await this.db.query(`SELECT COUNT(*) as cnt FROM \`${database}\`.\`${table_name}\``);
+            const totalRows = countResult[0]?.cnt || 0;
+            const patterns = [];
+            let qualityScore = 100;
+            for (const col of columns) {
+                const colName = col.COLUMN_NAME;
+                // Check for NULL patterns
+                if (pattern_types.includes("null")) {
+                    const nullResult = await this.db.query(`SELECT COUNT(*) as null_count
+             FROM \`${database}\`.\`${table_name}\`
+             WHERE \`${colName}\` IS NULL`);
+                    const nullCount = nullResult[0]?.null_count || 0;
+                    const nullPercentage = totalRows > 0 ? (nullCount / totalRows) * 100 : 0;
+                    if (nullPercentage > 50) {
+                        patterns.push({
+                            column_name: colName,
+                            pattern_type: "HIGH_NULL_RATE",
+                            description: `${nullPercentage.toFixed(1)}% of values are NULL`,
+                            metrics: { null_count: nullCount, null_percentage: nullPercentage },
+                            recommendations: [
+                                "Review if this column is necessary",
+                                "Consider setting a default value",
+                            ],
+                        });
+                        qualityScore -= 5;
+                    }
+                    else if (nullPercentage > 0 && col.IS_NULLABLE === "NO") {
+                        patterns.push({
+                            column_name: colName,
+                            pattern_type: "NULLABLE_MISMATCH",
+                            description: "Column is marked NOT NULL but has NULL values",
+                            metrics: { null_count: nullCount },
+                            recommendations: ["Check data integrity constraints"],
+                        });
+                        qualityScore -= 10;
+                    }
+                }
+                // Check for uniqueness patterns
+                if (pattern_types.includes("unique")) {
+                    const uniqueResult = await this.db.query(`SELECT COUNT(DISTINCT \`${colName}\`) as distinct_count
+             FROM \`${database}\`.\`${table_name}\``);
+                    const distinctCount = uniqueResult[0]?.distinct_count || 0;
+                    const uniqueRatio = totalRows > 0 ? distinctCount / totalRows : 0;
+                    if (uniqueRatio === 1 && col.COLUMN_KEY !== "PRI" && col.COLUMN_KEY !== "UNI") {
+                        patterns.push({
+                            column_name: colName,
+                            pattern_type: "POTENTIALLY_UNIQUE",
+                            description: "All values are unique but column is not marked as UNIQUE",
+                            metrics: { distinct_count: distinctCount, total_rows: totalRows },
+                            recommendations: [
+                                "Consider adding a UNIQUE constraint",
+                                "Could be a natural key candidate",
+                            ],
+                        });
+                    }
+                    else if (uniqueRatio < 0.1 && totalRows > 100) {
+                        patterns.push({
+                            column_name: colName,
+                            pattern_type: "LOW_CARDINALITY",
+                            description: `Only ${distinctCount} distinct values across ${totalRows} rows`,
+                            metrics: { distinct_count: distinctCount, cardinality_ratio: uniqueRatio },
+                            recommendations: [
+                                "Consider using ENUM if values are fixed",
+                                "May be a good candidate for indexing",
+                            ],
+                        });
+                    }
+                }
+                // Check for duplicate patterns
+                if (pattern_types.includes("duplicate") && totalRows > 10) {
+                    const duplicateResult = await this.db.query(`SELECT \`${colName}\`, COUNT(*) as cnt
+             FROM \`${database}\`.\`${table_name}\`
+             WHERE \`${colName}\` IS NOT NULL
+             GROUP BY \`${colName}\`
+             HAVING cnt > 1
+             ORDER BY cnt DESC
+             LIMIT 5`);
+                    if (duplicateResult.length > 0) {
+                        const topDuplicates = duplicateResult.map((r) => ({
+                            value: r[colName],
+                            count: r.cnt,
+                        }));
+                        if (col.COLUMN_KEY === "UNI" || col.COLUMN_KEY === "PRI") {
+                            patterns.push({
+                                column_name: colName,
+                                pattern_type: "DUPLICATE_IN_UNIQUE",
+                                description: "Duplicates found in supposedly unique column",
+                                metrics: { top_duplicates: topDuplicates },
+                                recommendations: ["Critical: Fix data integrity issue"],
+                            });
+                            qualityScore -= 20;
+                        }
+                        else {
+                            patterns.push({
+                                column_name: colName,
+                                pattern_type: "DUPLICATES_FOUND",
+                                description: `Found ${duplicateResult.length} values with duplicates`,
+                                metrics: { duplicate_groups: duplicateResult.length, top_duplicates: topDuplicates },
+                            });
+                        }
+                    }
+                }
+                // Check for format patterns (for string columns)
+                if (pattern_types.includes("format") &&
+                    ["varchar", "char", "text"].includes(col.DATA_TYPE.toLowerCase())) {
+                    // Check for email pattern
+                    const emailResult = await this.db.query(`SELECT COUNT(*) as cnt
+             FROM \`${database}\`.\`${table_name}\`
+             WHERE \`${colName}\` REGEXP '^[A-Za-z0-9._%+-]+@[A-Za-z0-9.-]+\\.[A-Za-z]{2,}$'`);
+                    if (emailResult[0]?.cnt > totalRows * 0.5) {
+                        patterns.push({
+                            column_name: colName,
+                            pattern_type: "EMAIL_FORMAT",
+                            description: "Contains email-formatted data",
+                            metrics: { email_count: emailResult[0].cnt },
+                        });
+                    }
+                    // Check for phone pattern
+                    const phoneResult = await this.db.query(`SELECT COUNT(*) as cnt
+             FROM \`${database}\`.\`${table_name}\`
+             WHERE \`${colName}\` REGEXP '^[+]?[0-9]{10,15}$'`);
+                    if (phoneResult[0]?.cnt > totalRows * 0.3) {
+                        patterns.push({
+                            column_name: colName,
+                            pattern_type: "PHONE_FORMAT",
+                            description: "Contains phone number-formatted data",
+                            metrics: { phone_count: phoneResult[0].cnt },
+                        });
+                    }
+                }
+                // Check for range patterns (for numeric columns)
+                if (pattern_types.includes("range") &&
+                    ["int", "bigint", "decimal", "float", "double"].includes(col.DATA_TYPE.toLowerCase())) {
+                    const rangeResult = await this.db.query(`SELECT MIN(\`${colName}\`) as min_val, MAX(\`${colName}\`) as max_val, AVG(\`${colName}\`) as avg_val
+             FROM \`${database}\`.\`${table_name}\``);
+                    if (rangeResult[0]) {
+                        const { min_val, max_val, avg_val } = rangeResult[0];
+                        const range = max_val - min_val;
+                        patterns.push({
+                            column_name: colName,
+                            pattern_type: "NUMERIC_RANGE",
+                            description: `Values range from ${min_val} to ${max_val}`,
+                            metrics: {
+                                min: min_val,
+                                max: max_val,
+                                avg: avg_val,
+                                range: range,
+                            },
+                        });
+                        // Check for potential outliers
+                        if (range > 0 && (max_val > avg_val * 10 || min_val < avg_val / 10)) {
+                            patterns.push({
+                                column_name: colName,
+                                pattern_type: "POTENTIAL_OUTLIERS",
+                                description: "Large variance detected, may contain outliers",
+                                recommendations: [
+                                    "Review extreme values for data quality",
+                                    "Consider outlier detection",
+                                ],
+                            });
+                        }
+                    }
+                }
+            }
+            return {
+                status: "success",
+                data: {
+                    table_name,
+                    patterns,
+                    summary: {
+                        columns_analyzed: columns.length,
+                        patterns_found: patterns.length,
+                        data_quality_score: Math.max(0, qualityScore),
+                    },
+                },
+            };
+        }
+        catch (error) {
+            return {
+                status: "error",
+                error: error.message,
+            };
+        }
+    }
+    // ==================== Helper Methods ====================
+    /**
+     * Tokenize a string into searchable tokens
+     */
+    tokenize(text) {
+        return text
+            .toLowerCase()
+            .replace(/[-_]/g, " ")
+            .split(/\s+/)
+            .filter((t) => t.length > 1);
+    }
+    /**
+     * Calculate relevance score between search tokens and target
+     */
+    calculateRelevanceScore(tokens, target, originalTerm) {
+        const targetLower = target.toLowerCase();
+        // Exact match
+        if (targetLower === originalTerm)
+            return 1.0;
+        // Contains full term
+        if (targetLower.includes(originalTerm))
+            return 0.9;
+        if (originalTerm.includes(targetLower))
+            return 0.85;
+        // Token-based scoring
+        let matchedTokens = 0;
+        for (const token of tokens) {
+            if (targetLower.includes(token))
+                matchedTokens++;
+        }
+        if (tokens.length > 0) {
+            const tokenScore = matchedTokens / tokens.length;
+            if (tokenScore > 0)
+                return 0.3 + tokenScore * 0.5;
+        }
+        // Similarity-based scoring
+        const similarity = this.calculateNameSimilarity(originalTerm, targetLower);
+        if (similarity > 0.5)
+            return similarity * 0.6;
+        return 0;
+    }
+    /**
+     * Get a human-readable match reason
+     */
+    getMatchReason(tokens, target) {
+        const targetLower = target.toLowerCase();
+        for (const token of tokens) {
+            if (targetLower === token)
+                return `Exact match: "${token}"`;
+            if (targetLower.includes(token))
+                return `Contains: "${token}"`;
+        }
+        return "Similar name pattern";
+    }
+    /**
+     * Calculate name similarity using Levenshtein-like approach
+     */
+    calculateNameSimilarity(a, b) {
+        if (a === b)
+            return 1;
+        if (a.length === 0 || b.length === 0)
+            return 0;
+        // Simple character overlap ratio
+        const setA = new Set(a.split(""));
+        const setB = new Set(b.split(""));
+        const intersection = [...setA].filter((x) => setB.has(x)).length;
+        const union = new Set([...setA, ...setB]).size;
+        const charSimilarity = union > 0 ? intersection / union : 0;
+        // Prefix matching bonus
+        let prefixLength = 0;
+        const minLen = Math.min(a.length, b.length);
+        for (let i = 0; i < minLen; i++) {
+            if (a[i] === b[i])
+                prefixLength++;
+            else
+                break;
+        }
+        const prefixBonus = prefixLength / Math.max(a.length, b.length);
+        return charSimilarity * 0.6 + prefixBonus * 0.4;
+    }
+    /**
+     * Normalize column name for comparison
+     */
+    normalizeColumnName(name) {
+        return name
+            .toLowerCase()
+            .replace(/^(fk_|pk_|idx_)/, "")
+            .replace(/_id$/, "")
+            .replace(/[_-]/g, "");
+    }
+    /**
+     * Get similarity type description
+     */
+    getSimilarityType(name1, name2) {
+        if (name1.toLowerCase() === name2.toLowerCase())
+            return "EXACT_MATCH";
+        const n1 = this.normalizeColumnName(name1);
+        const n2 = this.normalizeColumnName(name2);
+        if (n1 === n2)
+            return "NORMALIZED_MATCH";
+        if (n1.includes(n2) || n2.includes(n1))
+            return "SUBSTRING_MATCH";
+        return "SIMILAR_PATTERN";
+    }
+    /**
+     * Calculate data overlap between two columns
+     */
+    async calculateDataOverlap(database, table1, col1, table2, col2) {
+        const overlapQuery = `
+      SELECT COUNT(DISTINCT t1.\`${col1}\`) as overlap_count
+      FROM \`${database}\`.\`${table1}\` t1
+      INNER JOIN \`${database}\`.\`${table2}\` t2 ON t1.\`${col1}\` = t2.\`${col2}\`
+      WHERE t1.\`${col1}\` IS NOT NULL
+    `;
+        const totalQuery = `
+      SELECT COUNT(DISTINCT \`${col1}\`) as total
+      FROM \`${database}\`.\`${table1}\`
+      WHERE \`${col1}\` IS NOT NULL
+    `;
+        const [overlapResult, totalResult] = await Promise.all([
+            this.db.query(overlapQuery),
+            this.db.query(totalQuery),
+        ]);
+        const overlap = overlapResult[0]?.overlap_count || 0;
+        const total = totalResult[0]?.total || 1;
+        return Math.round((overlap / total) * 100);
+    }
+    /**
+     * Discover implicit relationships based on naming conventions
+     */
+    discoverImplicitRelationships(tables, columns, searchTokens, threshold) {
+        const relationships = [];
+        const tableNames = new Set(tables.map((t) => t.TABLE_NAME.toLowerCase()));
+        const primaryKeys = new Map();
+        // Find primary keys
+        for (const col of columns) {
+            if (col.COLUMN_KEY === "PRI") {
+                primaryKeys.set(col.TABLE_NAME.toLowerCase(), col.COLUMN_NAME);
+            }
+        }
+        // Look for _id columns that match table names
+        for (const col of columns) {
+            const colLower = col.COLUMN_NAME.toLowerCase();
+            if (colLower.endsWith("_id")) {
+                const potentialTable = colLower.slice(0, -3);
+                // Check if the tokens match
+                const score = this.calculateRelevanceScore(searchTokens, potentialTable, searchTokens.join(" "));
+                if (score >= threshold || searchTokens.length === 0) {
+                    if (tableNames.has(potentialTable)) {
+                        const pk = primaryKeys.get(potentialTable) || "id";
+                        relationships.push({
+                            from_table: col.TABLE_NAME,
+                            from_column: col.COLUMN_NAME,
+                            to_table: potentialTable,
+                            to_column: pk,
+                            relationship_type: "IMPLICIT_FK",
+                            confidence: 0.8,
+                        });
+                    }
+                }
+            }
+        }
+        return relationships;
+    }
+}
+exports.SmartDiscoveryTools = SmartDiscoveryTools;