kirograph 0.16.0 → 0.17.0

package/dist/data/filters.js

@@ -0,0 +1,105 @@
+"use strict";
+var __defProp = Object.defineProperty;
+var __getOwnPropDesc = Object.getOwnPropertyDescriptor;
+var __getOwnPropNames = Object.getOwnPropertyNames;
+var __hasOwnProp = Object.prototype.hasOwnProperty;
+var __export = (target, all) => {
+  for (var name in all)
+    __defProp(target, name, { get: all[name], enumerable: true });
+};
+var __copyProps = (to, from, except, desc) => {
+  if (from && typeof from === "object" || typeof from === "function") {
+    for (let key of __getOwnPropNames(from))
+      if (!__hasOwnProp.call(to, key) && key !== except)
+        __defProp(to, key, { get: () => from[key], enumerable: !(desc = __getOwnPropDesc(from, key)) || desc.enumerable });
+  }
+  return to;
+};
+var __toCommonJS = (mod) => __copyProps(__defProp({}, "__esModule", { value: true }), mod);
+var filters_exports = {};
+__export(filters_exports, {
+  buildWhereClause: () => buildWhereClause
+});
+module.exports = __toCommonJS(filters_exports);
+function validateColumn(column, validColumns) {
+  if (!validColumns.has(column)) {
+    throw new Error(`Invalid column: "${column}". Available columns: ${[...validColumns].join(", ")}`);
+  }
+  return `"${column.replace(/"/g, '""')}"`;
+}
+function buildWhereClause(filters, validColumns) {
+  if (filters.length === 0) {
+    return { where: "", params: [] };
+  }
+  const conditions = [];
+  const params = [];
+  for (const filter of filters) {
+    const col = validateColumn(filter.column, validColumns);
+    switch (filter.op) {
+      case "eq":
+        conditions.push(`${col} = ?`);
+        params.push(filter.value);
+        break;
+      case "neq":
+        conditions.push(`${col} != ?`);
+        params.push(filter.value);
+        break;
+      case "gt":
+        conditions.push(`${col} > ?`);
+        params.push(filter.value);
+        break;
+      case "gte":
+        conditions.push(`${col} >= ?`);
+        params.push(filter.value);
+        break;
+      case "lt":
+        conditions.push(`${col} < ?`);
+        params.push(filter.value);
+        break;
+      case "lte":
+        conditions.push(`${col} <= ?`);
+        params.push(filter.value);
+        break;
+      case "contains":
+        conditions.push(`${col} LIKE ?`);
+        params.push(`%${filter.value}%`);
+        break;
+      case "in": {
+        if (!Array.isArray(filter.value) || filter.value.length === 0) {
+          conditions.push("0");
+        } else {
+          const placeholders = filter.value.map(() => "?").join(",");
+          conditions.push(`${col} IN (${placeholders})`);
+          params.push(...filter.value);
+        }
+        break;
+      }
+      case "is_null":
+        if (filter.value === true || filter.value === "true") {
+          conditions.push(`${col} IS NULL`);
+        } else {
+          conditions.push(`${col} IS NOT NULL`);
+        }
+        break;
+      case "between": {
+        if (!Array.isArray(filter.value) || filter.value.length !== 2) {
+          throw new Error(`"between" filter requires a [min, max] array`);
+        }
+        conditions.push(`${col} BETWEEN ? AND ?`);
+        params.push(filter.value[0], filter.value[1]);
+        break;
+      }
+      default:
+        throw new Error(`Unknown filter operator: "${filter.op}"`);
+    }
+  }
+  return {
+    where: `WHERE ${conditions.join(" AND ")}`,
+    params
+  };
+}
+// Annotate the CommonJS export names for ESM import in node:
+0 && (module.exports = {
+  buildWhereClause
+});
+//# sourceMappingURL=filters.js.map

package/dist/data/filters.js.map

@@ -0,0 +1,7 @@
+{
+  "version": 3,
+  "sources": ["../../src/data/filters.ts"],
+  "sourcesContent": ["/**\n * Structured Filter \u2192 SQL WHERE Clause Builder\n *\n * Converts QueryFilter objects into parameterized SQL.\n * Zero injection surface \u2014 column names are validated against schema,\n * values are always parameterized.\n */\n\nimport type { QueryFilter } from './types';\n\ninterface SQLClause {\n  where: string;\n  params: any[];\n}\n\n/**\n * Validate that a column name exists in the schema.\n * Prevents SQL injection via column names.\n */\nfunction validateColumn(column: string, validColumns: Set<string>): string {\n  if (!validColumns.has(column)) {\n    throw new Error(`Invalid column: \"${column}\". Available columns: ${[...validColumns].join(', ')}`);\n  }\n  // Quote column name to handle spaces and reserved words\n  return `\"${column.replace(/\"/g, '\"\"')}\"`;\n}\n\n/**\n * Build a parameterized WHERE clause from structured filters.\n * All filters are ANDed together.\n *\n * @param filters - Array of QueryFilter objects\n * @param validColumns - Set of valid column names (from schema)\n * @returns SQL WHERE clause and parameter array\n */\nexport function buildWhereClause(filters: QueryFilter[], validColumns: Set<string>): SQLClause {\n  if (filters.length === 0) {\n    return { where: '', params: [] };\n  }\n\n  const conditions: string[] = [];\n  const params: any[] = [];\n\n  for (const filter of filters) {\n    const col = validateColumn(filter.column, validColumns);\n\n    switch (filter.op) {\n      case 'eq':\n        conditions.push(`${col} = ?`);\n        params.push(filter.value);\n        break;\n\n      case 'neq':\n        conditions.push(`${col} != ?`);\n        params.push(filter.value);\n        break;\n\n      case 'gt':\n        conditions.push(`${col} > ?`);\n        params.push(filter.value);\n        break;\n\n      case 'gte':\n        conditions.push(`${col} >= ?`);\n        params.push(filter.value);\n        break;\n\n      case 'lt':\n        conditions.push(`${col} < ?`);\n        params.push(filter.value);\n        break;\n\n      case 'lte':\n        conditions.push(`${col} <= ?`);\n        params.push(filter.value);\n        break;\n\n      case 'contains':\n        conditions.push(`${col} LIKE ?`);\n        params.push(`%${filter.value}%`);\n        break;\n\n      case 'in': {\n        if (!Array.isArray(filter.value) || filter.value.length === 0) {\n          conditions.push('0'); // Always false\n        } else {\n          const placeholders = filter.value.map(() => '?').join(',');\n          conditions.push(`${col} IN (${placeholders})`);\n          params.push(...filter.value);\n        }\n        break;\n      }\n\n      case 'is_null':\n        if (filter.value === true || filter.value === 'true') {\n          conditions.push(`${col} IS NULL`);\n        } else {\n          conditions.push(`${col} IS NOT NULL`);\n        }\n        break;\n\n      case 'between': {\n        if (!Array.isArray(filter.value) || filter.value.length !== 2) {\n          throw new Error(`\"between\" filter requires a [min, max] array`);\n        }\n        conditions.push(`${col} BETWEEN ? AND ?`);\n        params.push(filter.value[0], filter.value[1]);\n        break;\n      }\n\n      default:\n        throw new Error(`Unknown filter operator: \"${(filter as any).op}\"`);\n    }\n  }\n\n  return {\n    where: `WHERE ${conditions.join(' AND ')}`,\n    params,\n  };\n}\n"],
+  "mappings": ";;;;;;;;;;;;;;;;;;AAAA;AAAA;AAAA;AAAA;AAAA;AAmBA,SAAS,eAAe,QAAgB,cAAmC;AACzE,MAAI,CAAC,aAAa,IAAI,MAAM,GAAG;AAC7B,UAAM,IAAI,MAAM,oBAAoB,MAAM,yBAAyB,CAAC,GAAG,YAAY,EAAE,KAAK,IAAI,CAAC,EAAE;AAAA,EACnG;AAEA,SAAO,IAAI,OAAO,QAAQ,MAAM,IAAI,CAAC;AACvC;AAUO,SAAS,iBAAiB,SAAwB,cAAsC;AAC7F,MAAI,QAAQ,WAAW,GAAG;AACxB,WAAO,EAAE,OAAO,IAAI,QAAQ,CAAC,EAAE;AAAA,EACjC;AAEA,QAAM,aAAuB,CAAC;AAC9B,QAAM,SAAgB,CAAC;AAEvB,aAAW,UAAU,SAAS;AAC5B,UAAM,MAAM,eAAe,OAAO,QAAQ,YAAY;AAEtD,YAAQ,OAAO,IAAI;AAAA,MACjB,KAAK;AACH,mBAAW,KAAK,GAAG,GAAG,MAAM;AAC5B,eAAO,KAAK,OAAO,KAAK;AACxB;AAAA,MAEF,KAAK;AACH,mBAAW,KAAK,GAAG,GAAG,OAAO;AAC7B,eAAO,KAAK,OAAO,KAAK;AACxB;AAAA,MAEF,KAAK;AACH,mBAAW,KAAK,GAAG,GAAG,MAAM;AAC5B,eAAO,KAAK,OAAO,KAAK;AACxB;AAAA,MAEF,KAAK;AACH,mBAAW,KAAK,GAAG,GAAG,OAAO;AAC7B,eAAO,KAAK,OAAO,KAAK;AACxB;AAAA,MAEF,KAAK;AACH,mBAAW,KAAK,GAAG,GAAG,MAAM;AAC5B,eAAO,KAAK,OAAO,KAAK;AACxB;AAAA,MAEF,KAAK;AACH,mBAAW,KAAK,GAAG,GAAG,OAAO;AAC7B,eAAO,KAAK,OAAO,KAAK;AACxB;AAAA,MAEF,KAAK;AACH,mBAAW,KAAK,GAAG,GAAG,SAAS;AAC/B,eAAO,KAAK,IAAI,OAAO,KAAK,GAAG;AAC/B;AAAA,MAEF,KAAK,MAAM;AACT,YAAI,CAAC,MAAM,QAAQ,OAAO,KAAK,KAAK,OAAO,MAAM,WAAW,GAAG;AAC7D,qBAAW,KAAK,GAAG;AAAA,QACrB,OAAO;AACL,gBAAM,eAAe,OAAO,MAAM,IAAI,MAAM,GAAG,EAAE,KAAK,GAAG;AACzD,qBAAW,KAAK,GAAG,GAAG,QAAQ,YAAY,GAAG;AAC7C,iBAAO,KAAK,GAAG,OAAO,KAAK;AAAA,QAC7B;AACA;AAAA,MACF;AAAA,MAEA,KAAK;AACH,YAAI,OAAO,UAAU,QAAQ,OAAO,UAAU,QAAQ;AACpD,qBAAW,KAAK,GAAG,GAAG,UAAU;AAAA,QAClC,OAAO;AACL,qBAAW,KAAK,GAAG,GAAG,cAAc;AAAA,QACtC;AACA;AAAA,MAEF,KAAK,WAAW;AACd,YAAI,CAAC,MAAM,QAAQ,OAAO,KAAK,KAAK,OAAO,MAAM,WAAW,GAAG;AAC7D,gBAAM,IAAI,MAAM,8CAA8C;AAAA,QAChE;AACA,mBAAW,KAAK,GAAG,GAAG,kBAAkB;AACxC,eAAO,KAAK,OAAO,MAAM,CAAC,GAAG,OAAO,MAAM,CAAC,CAAC;AAC5C;AAAA,MACF;AAAA,MAEA;AACE,cAAM,IAAI,MAAM,6BAA8B,OAAe,EAAE,GAAG;AAAA,IACtE;AAAA,EACF;AAEA,SAAO;AAAA,IACL,OAAO,SAAS,WAAW,KAAK,OAAO,CAAC;AAAA,IACxC;AAAA,EACF;AACF;",
+  "names": []
+}

package/dist/data/indexer.js

@@ -0,0 +1,225 @@
+"use strict";
+var __create = Object.create;
+var __defProp = Object.defineProperty;
+var __getOwnPropDesc = Object.getOwnPropertyDescriptor;
+var __getOwnPropNames = Object.getOwnPropertyNames;
+var __getProtoOf = Object.getPrototypeOf;
+var __hasOwnProp = Object.prototype.hasOwnProperty;
+var __export = (target, all) => {
+  for (var name in all)
+    __defProp(target, name, { get: all[name], enumerable: true });
+};
+var __copyProps = (to, from, except, desc) => {
+  if (from && typeof from === "object" || typeof from === "function") {
+    for (let key of __getOwnPropNames(from))
+      if (!__hasOwnProp.call(to, key) && key !== except)
+        __defProp(to, key, { get: () => from[key], enumerable: !(desc = __getOwnPropDesc(from, key)) || desc.enumerable });
+  }
+  return to;
+};
+var __toESM = (mod, isNodeMode, target) => (target = mod != null ? __create(__getProtoOf(mod)) : {}, __copyProps(
+  // If the importer is in node compatibility mode or this is not an ESM
+  // file that has been converted to a CommonJS file using a Babel-
+  // compatible transform (i.e. "__esModule" has not been set), then set
+  // "default" to the CommonJS "module.exports" for node compatibility.
+  isNodeMode || !mod || !mod.__esModule ? __defProp(target, "default", { value: mod, enumerable: true }) : target,
+  mod
+));
+var __toCommonJS = (mod) => __copyProps(__defProp({}, "__esModule", { value: true }), mod);
+var indexer_exports = {};
+__export(indexer_exports, {
+  DataIndexer: () => DataIndexer
+});
+module.exports = __toCommonJS(indexer_exports);
+var fs = __toESM(require("fs"));
+var path = __toESM(require("path"));
+var crypto = __toESM(require("crypto"));
+var import_picomatch = __toESM(require("picomatch"));
+var import_parsers = require("./parsers/index");
+var import_profiler = require("./profiler");
+function generateDatasetId(filePath) {
+  return filePath.replace(/\.[^.]+$/, "").replace(/[/\\]/g, "-").replace(/[^a-zA-Z0-9-_]/g, "-").replace(/-+/g, "-").replace(/^-|-$/g, "").toLowerCase();
+}
+class DataIndexer {
+  constructor(db, config, projectRoot) {
+    this.db = db;
+    this.config = config;
+    this.projectRoot = projectRoot;
+  }
+  async indexAll(opts) {
+    const start = Date.now();
+    const result = {
+      datasetsIndexed: 0,
+      rowsIndexed: 0,
+      columnsProfiled: 0,
+      errors: [],
+      duration: 0
+    };
+    const files = this.scanDataFiles();
+    opts?.onProgress?.(`data: found ${files.length} data files`);
+    for (const relPath of files) {
+      try {
+        const indexed = await this.indexFile(relPath, opts?.force ?? false);
+        if (indexed) {
+          result.datasetsIndexed++;
+          result.rowsIndexed += indexed.rows;
+          result.columnsProfiled += indexed.columns;
+        }
+      } catch (err) {
+        result.errors.push(`${relPath}: ${err instanceof Error ? err.message : String(err)}`);
+      }
+    }
+    this.removeDeletedDatasets(files);
+    if (this.config.dataLinkCode !== false && result.datasetsIndexed > 0) {
+      try {
+        const { DataCodeLinker } = await Promise.resolve().then(() => require("./linker.js"));
+        const linker = new DataCodeLinker(this.db, this.projectRoot);
+        const refsCreated = linker.linkAll();
+        opts?.onProgress?.(`data: linked ${refsCreated} code references`);
+      } catch {
+      }
+    }
+    result.duration = Date.now() - start;
+    return result;
+  }
+  async indexFile(relPath, force = false) {
+    const absPath = path.join(this.projectRoot, relPath);
+    if (!fs.existsSync(absPath)) return null;
+    const stat = fs.statSync(absPath);
+    if (stat.size > this.config.dataMaxFileSize) return null;
+    const fileBuffer = fs.readFileSync(absPath);
+    const contentHash = crypto.createHash("sha256").update(fileBuffer).digest("hex");
+    const datasetId = generateDatasetId(relPath);
+    if (!force) {
+      const existing = this.db.get("SELECT content_hash FROM data_datasets WHERE id = ?", [datasetId]);
+      if (existing?.content_hash === contentHash) return null;
+    }
+    const parser = (0, import_parsers.getDataParser)(relPath);
+    if (!parser) return null;
+    const profiler = new import_profiler.ColumnProfiler();
+    const allRows = [];
+    const maxRows = this.config.dataMaxRows ?? 1e6;
+    const { columns, totalRows } = await parser.parse(absPath, {
+      maxRows,
+      onBatch: (rows, cols) => {
+        profiler.addBatch(rows, cols);
+        allRows.push(...rows);
+      }
+    });
+    if (columns.length === 0) return null;
+    const columnProfiles = profiler.finalize(datasetId);
+    this.db.run("BEGIN");
+    try {
+      const rowTableName = `data_rows_${datasetId.replace(/[^a-zA-Z0-9_]/g, "_")}`;
+      this.db.run(`DROP TABLE IF EXISTS "${rowTableName}"`);
+      this.db.run("DELETE FROM data_columns WHERE dataset_id = ?", [datasetId]);
+      this.db.run("DELETE FROM data_datasets WHERE id = ?", [datasetId]);
+      this.db.run(`
+        INSERT INTO data_datasets (id, file_path, format, row_count, column_count, file_size, content_hash, summary, indexed_at)
+        VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?)
+      `, [datasetId, relPath, parser.name, totalRows, columns.length, stat.size, contentHash, null, Date.now()]);
+      for (const col of columnProfiles) {
+        this.db.run(`
+          INSERT INTO data_columns (id, dataset_id, name, position, inferred_type, nullable, null_count, null_pct, cardinality, min_value, max_value, mean_value, sample_values, summary, updated_at)
+          VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
+        `, [
+          col.id,
+          col.datasetId,
+          col.name,
+          col.position,
+          col.inferredType,
+          col.nullable ? 1 : 0,
+          col.nullCount,
+          col.nullPct,
+          col.cardinality,
+          col.minValue,
+          col.maxValue,
+          col.meanValue,
+          JSON.stringify(col.sampleValues),
+          col.summary,
+          col.updatedAt
+        ]);
+      }
+      const colDefs = columns.map((c) => `"${c.replace(/"/g, '""')}" TEXT`).join(", ");
+      this.db.run(`CREATE TABLE "${rowTableName}" (rowid INTEGER PRIMARY KEY AUTOINCREMENT, ${colDefs})`);
+      const colNames = columns.map((c) => `"${c.replace(/"/g, '""')}"`).join(", ");
+      const placeholders = columns.map(() => "?").join(", ");
+      const insertStmt = `INSERT INTO "${rowTableName}" (${colNames}) VALUES (${placeholders})`;
+      for (const row of allRows) {
+        const values = columns.map((c) => {
+          const v = row[c];
+          if (v === null || v === void 0) return null;
+          if (typeof v === "object") return JSON.stringify(v);
+          return v;
+        });
+        this.db.run(insertStmt, values);
+      }
+      const historyColumns = columnProfiles.map((col) => ({
+        name: col.name,
+        type: col.inferredType,
+        cardinality: col.cardinality,
+        nullPct: col.nullPct
+      }));
+      this.db.run(`
+        INSERT INTO data_dataset_history (dataset_id, snapshot_at, row_count, column_count, columns_json, content_hash)
+        VALUES (?, ?, ?, ?, ?, ?)
+      `, [datasetId, Date.now(), totalRows, columns.length, JSON.stringify(historyColumns), contentHash]);
+      this.db.run("COMMIT");
+    } catch (err) {
+      this.db.run("ROLLBACK");
+      throw err;
+    }
+    return { rows: totalRows, columns: columns.length };
+  }
+  // ── Private ─────────────────────────────────────────────────────────────────
+  scanDataFiles() {
+    const includePatterns = this.config.dataInclude ?? [];
+    const excludePatterns = this.config.dataExclude ?? [];
+    const includeMatchers = includePatterns.map((p) => (0, import_picomatch.default)(p));
+    const excludeMatchers = excludePatterns.map((p) => (0, import_picomatch.default)(p));
+    const results = [];
+    const walk = (dir) => {
+      let entries;
+      try {
+        entries = fs.readdirSync(dir, { withFileTypes: true });
+      } catch {
+        return;
+      }
+      for (const entry of entries) {
+        const full = path.join(dir, entry.name);
+        const rel = path.relative(this.projectRoot, full).replace(/\\/g, "/");
+        if (entry.isDirectory()) {
+          if (excludeMatchers.some((m) => m(rel + "/"))) continue;
+          walk(full);
+        } else if (entry.isFile()) {
+          if (excludeMatchers.some((m) => m(rel))) continue;
+          if (includeMatchers.some((m) => m(rel))) {
+            if ((0, import_parsers.getDataParser)(rel)) {
+              results.push(rel);
+            }
+          }
+        }
+      }
+    };
+    walk(this.projectRoot);
+    return results;
+  }
+  removeDeletedDatasets(currentFiles) {
+    const currentSet = new Set(currentFiles.map((f) => generateDatasetId(f)));
+    const indexed = this.db.all("SELECT id FROM data_datasets").map((r) => r.id);
+    for (const id of indexed) {
+      if (!currentSet.has(id)) {
+        const rowTableName = `data_rows_${id.replace(/[^a-zA-Z0-9_]/g, "_")}`;
+        this.db.run(`DROP TABLE IF EXISTS "${rowTableName}"`);
+        this.db.run("DELETE FROM data_columns WHERE dataset_id = ?", [id]);
+        this.db.run("DELETE FROM data_code_refs WHERE dataset_id = ?", [id]);
+        this.db.run("DELETE FROM data_datasets WHERE id = ?", [id]);
+      }
+    }
+  }
+}
+// Annotate the CommonJS export names for ESM import in node:
+0 && (module.exports = {
+  DataIndexer
+});
+//# sourceMappingURL=indexer.js.map

package/dist/data/indexer.js.map

@@ -0,0 +1,7 @@
+{
+  "version": 3,
+  "sources": ["../../src/data/indexer.ts"],
+  "sourcesContent": ["/**\n * Data Indexer\n *\n * Orchestrates: scan data files \u2192 parse \u2192 profile columns \u2192 persist rows to SQLite.\n * Supports incremental re-indexing via content hashes.\n */\n\nimport * as fs from 'fs';\nimport * as path from 'path';\nimport * as crypto from 'crypto';\nimport picomatch from 'picomatch';\nimport type { KiroGraphConfig } from '../config';\nimport type { DataSet, DataColumn, DataIndexResult, ParsedRow } from './types';\nimport { getDataParser } from './parsers/index';\nimport { ColumnProfiler } from './profiler';\n\n/**\n * Generate a dataset ID from a file path.\n * Slugifies the relative path: tests/fixtures/users.csv \u2192 tests-fixtures-users\n */\nfunction generateDatasetId(filePath: string): string {\n  return filePath\n    .replace(/\\.[^.]+$/, '')  // remove extension\n    .replace(/[/\\\\]/g, '-')   // replace path separators\n    .replace(/[^a-zA-Z0-9-_]/g, '-') // replace special chars\n    .replace(/-+/g, '-')      // collapse multiple dashes\n    .replace(/^-|-$/g, '')    // trim leading/trailing dashes\n    .toLowerCase();\n}\n\nexport class DataIndexer {\n  private readonly db: any;\n  private readonly config: KiroGraphConfig;\n  private readonly projectRoot: string;\n\n  constructor(db: any, config: KiroGraphConfig, projectRoot: string) {\n    this.db = db;\n    this.config = config;\n    this.projectRoot = projectRoot;\n  }\n\n  async indexAll(opts?: { force?: boolean; onProgress?: (msg: string) => void }): Promise<DataIndexResult> {\n    const start = Date.now();\n    const result: DataIndexResult = {\n      datasetsIndexed: 0, rowsIndexed: 0, columnsProfiled: 0, errors: [], duration: 0,\n    };\n\n    const files = this.scanDataFiles();\n    opts?.onProgress?.(`data: found ${files.length} data files`);\n\n    for (const relPath of files) {\n      try {\n        const indexed = await this.indexFile(relPath, opts?.force ?? false);\n        if (indexed) {\n          result.datasetsIndexed++;\n          result.rowsIndexed += indexed.rows;\n          result.columnsProfiled += indexed.columns;\n        }\n      } catch (err) {\n        result.errors.push(`${relPath}: ${err instanceof Error ? err.message : String(err)}`);\n      }\n    }\n\n    // Remove datasets for files that no longer exist\n    this.removeDeletedDatasets(files);\n\n    // Link data files to code (if enabled)\n    if ((this.config as any).dataLinkCode !== false && result.datasetsIndexed > 0) {\n      try {\n        const { DataCodeLinker } = await Promise.resolve().then(() => require('./linker.js'));\n        const linker = new DataCodeLinker(this.db, this.projectRoot);\n        const refsCreated = linker.linkAll();\n        opts?.onProgress?.(`data: linked ${refsCreated} code references`);\n      } catch {\n        // Non-critical \u2014 linking failure shouldn't block indexing\n      }\n    }\n\n    result.duration = Date.now() - start;\n    return result;\n  }\n\n  async indexFile(relPath: string, force = false): Promise<{ rows: number; columns: number } | null> {\n    const absPath = path.join(this.projectRoot, relPath);\n    if (!fs.existsSync(absPath)) return null;\n\n    const stat = fs.statSync(absPath);\n    if (stat.size > (this.config as any).dataMaxFileSize) return null;\n\n    // Check content hash for incremental\n    const fileBuffer = fs.readFileSync(absPath);\n    const contentHash = crypto.createHash('sha256').update(fileBuffer).digest('hex');\n\n    const datasetId = generateDatasetId(relPath);\n\n    if (!force) {\n      const existing = this.db.get('SELECT content_hash FROM data_datasets WHERE id = ?', [datasetId]);\n      if (existing?.content_hash === contentHash) return null; // unchanged\n    }\n\n    // Get parser\n    const parser = getDataParser(relPath);\n    if (!parser) return null;\n\n    // Parse and profile\n    const profiler = new ColumnProfiler();\n    const allRows: ParsedRow[] = [];\n    const maxRows = (this.config as any).dataMaxRows ?? 1_000_000;\n\n    const { columns, totalRows } = await parser.parse(absPath, {\n      maxRows,\n      onBatch: (rows, cols) => {\n        profiler.addBatch(rows, cols);\n        allRows.push(...rows);\n      },\n    });\n\n    if (columns.length === 0) return null;\n\n    // Finalize profiles\n    const columnProfiles = profiler.finalize(datasetId);\n\n    // Persist in transaction\n    this.db.run('BEGIN');\n    try {\n      // Delete old dataset data\n      const rowTableName = `data_rows_${datasetId.replace(/[^a-zA-Z0-9_]/g, '_')}`;\n      this.db.run(`DROP TABLE IF EXISTS \"${rowTableName}\"`);\n      this.db.run('DELETE FROM data_columns WHERE dataset_id = ?', [datasetId]);\n      this.db.run('DELETE FROM data_datasets WHERE id = ?', [datasetId]);\n\n      // Insert dataset\n      this.db.run(`\n        INSERT INTO data_datasets (id, file_path, format, row_count, column_count, file_size, content_hash, summary, indexed_at)\n        VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?)\n      `, [datasetId, relPath, parser.name, totalRows, columns.length, stat.size, contentHash, null, Date.now()]);\n\n      // Insert column profiles\n      for (const col of columnProfiles) {\n        this.db.run(`\n          INSERT INTO data_columns (id, dataset_id, name, position, inferred_type, nullable, null_count, null_pct, cardinality, min_value, max_value, mean_value, sample_values, summary, updated_at)\n          VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)\n        `, [\n          col.id, col.datasetId, col.name, col.position, col.inferredType,\n          col.nullable ? 1 : 0, col.nullCount, col.nullPct, col.cardinality,\n          col.minValue, col.maxValue, col.meanValue,\n          JSON.stringify(col.sampleValues), col.summary, col.updatedAt,\n        ]);\n      }\n\n      // Create row table with dynamic schema\n      const colDefs = columns.map(c => `\"${c.replace(/\"/g, '\"\"')}\" TEXT`).join(', ');\n      this.db.run(`CREATE TABLE \"${rowTableName}\" (rowid INTEGER PRIMARY KEY AUTOINCREMENT, ${colDefs})`);\n\n      // Insert rows in batches\n      const colNames = columns.map(c => `\"${c.replace(/\"/g, '\"\"')}\"`).join(', ');\n      const placeholders = columns.map(() => '?').join(', ');\n      const insertStmt = `INSERT INTO \"${rowTableName}\" (${colNames}) VALUES (${placeholders})`;\n\n      for (const row of allRows) {\n        const values = columns.map(c => {\n          const v = row[c];\n          if (v === null || v === undefined) return null;\n          if (typeof v === 'object') return JSON.stringify(v);\n          return v;\n        });\n        this.db.run(insertStmt, values);\n      }\n\n      // Save history snapshot for drift detection\n      const historyColumns = columnProfiles.map(col => ({\n        name: col.name,\n        type: col.inferredType,\n        cardinality: col.cardinality,\n        nullPct: col.nullPct,\n      }));\n      this.db.run(`\n        INSERT INTO data_dataset_history (dataset_id, snapshot_at, row_count, column_count, columns_json, content_hash)\n        VALUES (?, ?, ?, ?, ?, ?)\n      `, [datasetId, Date.now(), totalRows, columns.length, JSON.stringify(historyColumns), contentHash]);\n\n      this.db.run('COMMIT');\n    } catch (err) {\n      this.db.run('ROLLBACK');\n      throw err;\n    }\n\n    return { rows: totalRows, columns: columns.length };\n  }\n\n  // \u2500\u2500 Private \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n\n  private scanDataFiles(): string[] {\n    const includePatterns = (this.config as any).dataInclude ?? [];\n    const excludePatterns = (this.config as any).dataExclude ?? [];\n    const includeMatchers = includePatterns.map((p: string) => picomatch(p));\n    const excludeMatchers = excludePatterns.map((p: string) => picomatch(p));\n    const results: string[] = [];\n\n    const walk = (dir: string) => {\n      let entries: fs.Dirent[];\n      try { entries = fs.readdirSync(dir, { withFileTypes: true }); } catch { return; }\n\n      for (const entry of entries) {\n        const full = path.join(dir, entry.name);\n        const rel = path.relative(this.projectRoot, full).replace(/\\\\/g, '/');\n\n        if (entry.isDirectory()) {\n          if (excludeMatchers.some((m: any) => m(rel + '/'))) continue;\n          walk(full);\n        } else if (entry.isFile()) {\n          if (excludeMatchers.some((m: any) => m(rel))) continue;\n          if (includeMatchers.some((m: any) => m(rel))) {\n            // Verify a parser exists for this format\n            if (getDataParser(rel)) {\n              results.push(rel);\n            }\n          }\n        }\n      }\n    };\n\n    walk(this.projectRoot);\n    return results;\n  }\n\n  private removeDeletedDatasets(currentFiles: string[]): void {\n    const currentSet = new Set(currentFiles.map(f => generateDatasetId(f)));\n    const indexed = (this.db.all('SELECT id FROM data_datasets') as Array<{ id: string }>).map(r => r.id);\n\n    for (const id of indexed) {\n      if (!currentSet.has(id)) {\n        const rowTableName = `data_rows_${id.replace(/[^a-zA-Z0-9_]/g, '_')}`;\n        this.db.run(`DROP TABLE IF EXISTS \"${rowTableName}\"`);\n        this.db.run('DELETE FROM data_columns WHERE dataset_id = ?', [id]);\n        this.db.run('DELETE FROM data_code_refs WHERE dataset_id = ?', [id]);\n        this.db.run('DELETE FROM data_datasets WHERE id = ?', [id]);\n      }\n    }\n  }\n}\n"],
+  "mappings": ";;;;;;;;;;;;;;;;;;;;;;;;;;;;AAAA;AAAA;AAAA;AAAA;AAAA;AAOA,SAAoB;AACpB,WAAsB;AACtB,aAAwB;AACxB,uBAAsB;AAGtB,qBAA8B;AAC9B,sBAA+B;AAM/B,SAAS,kBAAkB,UAA0B;AACnD,SAAO,SACJ,QAAQ,YAAY,EAAE,EACtB,QAAQ,UAAU,GAAG,EACrB,QAAQ,mBAAmB,GAAG,EAC9B,QAAQ,OAAO,GAAG,EAClB,QAAQ,UAAU,EAAE,EACpB,YAAY;AACjB;AAEO,MAAM,YAAY;AAAA,EAKvB,YAAY,IAAS,QAAyB,aAAqB;AACjE,SAAK,KAAK;AACV,SAAK,SAAS;AACd,SAAK,cAAc;AAAA,EACrB;AAAA,EAEA,MAAM,SAAS,MAA0F;AACvG,UAAM,QAAQ,KAAK,IAAI;AACvB,UAAM,SAA0B;AAAA,MAC9B,iBAAiB;AAAA,MAAG,aAAa;AAAA,MAAG,iBAAiB;AAAA,MAAG,QAAQ,CAAC;AAAA,MAAG,UAAU;AAAA,IAChF;AAEA,UAAM,QAAQ,KAAK,cAAc;AACjC,UAAM,aAAa,eAAe,MAAM,MAAM,aAAa;AAE3D,eAAW,WAAW,OAAO;AAC3B,UAAI;AACF,cAAM,UAAU,MAAM,KAAK,UAAU,SAAS,MAAM,SAAS,KAAK;AAClE,YAAI,SAAS;AACX,iBAAO;AACP,iBAAO,eAAe,QAAQ;AAC9B,iBAAO,mBAAmB,QAAQ;AAAA,QACpC;AAAA,MACF,SAAS,KAAK;AACZ,eAAO,OAAO,KAAK,GAAG,OAAO,KAAK,eAAe,QAAQ,IAAI,UAAU,OAAO,GAAG,CAAC,EAAE;AAAA,MACtF;AAAA,IACF;AAGA,SAAK,sBAAsB,KAAK;AAGhC,QAAK,KAAK,OAAe,iBAAiB,SAAS,OAAO,kBAAkB,GAAG;AAC7E,UAAI;AACF,cAAM,EAAE,eAAe,IAAI,MAAM,QAAQ,QAAQ,EAAE,KAAK,MAAM,QAAQ,aAAa,CAAC;AACpF,cAAM,SAAS,IAAI,eAAe,KAAK,IAAI,KAAK,WAAW;AAC3D,cAAM,cAAc,OAAO,QAAQ;AACnC,cAAM,aAAa,gBAAgB,WAAW,kBAAkB;AAAA,MAClE,QAAQ;AAAA,MAER;AAAA,IACF;AAEA,WAAO,WAAW,KAAK,IAAI,IAAI;AAC/B,WAAO;AAAA,EACT;AAAA,EAEA,MAAM,UAAU,SAAiB,QAAQ,OAA0D;AACjG,UAAM,UAAU,KAAK,KAAK,KAAK,aAAa,OAAO;AACnD,QAAI,CAAC,GAAG,WAAW,OAAO,EAAG,QAAO;AAEpC,UAAM,OAAO,GAAG,SAAS,OAAO;AAChC,QAAI,KAAK,OAAQ,KAAK,OAAe,gBAAiB,QAAO;AAG7D,UAAM,aAAa,GAAG,aAAa,OAAO;AAC1C,UAAM,cAAc,OAAO,WAAW,QAAQ,EAAE,OAAO,UAAU,EAAE,OAAO,KAAK;AAE/E,UAAM,YAAY,kBAAkB,OAAO;AAE3C,QAAI,CAAC,OAAO;AACV,YAAM,WAAW,KAAK,GAAG,IAAI,uDAAuD,CAAC,SAAS,CAAC;AAC/F,UAAI,UAAU,iBAAiB,YAAa,QAAO;AAAA,IACrD;AAGA,UAAM,aAAS,8BAAc,OAAO;AACpC,QAAI,CAAC,OAAQ,QAAO;AAGpB,UAAM,WAAW,IAAI,+BAAe;AACpC,UAAM,UAAuB,CAAC;AAC9B,UAAM,UAAW,KAAK,OAAe,eAAe;AAEpD,UAAM,EAAE,SAAS,UAAU,IAAI,MAAM,OAAO,MAAM,SAAS;AAAA,MACzD;AAAA,MACA,SAAS,CAAC,MAAM,SAAS;AACvB,iBAAS,SAAS,MAAM,IAAI;AAC5B,gBAAQ,KAAK,GAAG,IAAI;AAAA,MACtB;AAAA,IACF,CAAC;AAED,QAAI,QAAQ,WAAW,EAAG,QAAO;AAGjC,UAAM,iBAAiB,SAAS,SAAS,SAAS;AAGlD,SAAK,GAAG,IAAI,OAAO;AACnB,QAAI;AAEF,YAAM,eAAe,aAAa,UAAU,QAAQ,kBAAkB,GAAG,CAAC;AAC1E,WAAK,GAAG,IAAI,yBAAyB,YAAY,GAAG;AACpD,WAAK,GAAG,IAAI,iDAAiD,CAAC,SAAS,CAAC;AACxE,WAAK,GAAG,IAAI,0CAA0C,CAAC,SAAS,CAAC;AAGjE,WAAK,GAAG,IAAI;AAAA;AAAA;AAAA,SAGT,CAAC,WAAW,SAAS,OAAO,MAAM,WAAW,QAAQ,QAAQ,KAAK,MAAM,aAAa,MAAM,KAAK,IAAI,CAAC,CAAC;AAGzG,iBAAW,OAAO,gBAAgB;AAChC,aAAK,GAAG,IAAI;AAAA;AAAA;AAAA,WAGT;AAAA,UACD,IAAI;AAAA,UAAI,IAAI;AAAA,UAAW,IAAI;AAAA,UAAM,IAAI;AAAA,UAAU,IAAI;AAAA,UACnD,IAAI,WAAW,IAAI;AAAA,UAAG,IAAI;AAAA,UAAW,IAAI;AAAA,UAAS,IAAI;AAAA,UACtD,IAAI;AAAA,UAAU,IAAI;AAAA,UAAU,IAAI;AAAA,UAChC,KAAK,UAAU,IAAI,YAAY;AAAA,UAAG,IAAI;AAAA,UAAS,IAAI;AAAA,QACrD,CAAC;AAAA,MACH;AAGA,YAAM,UAAU,QAAQ,IAAI,OAAK,IAAI,EAAE,QAAQ,MAAM,IAAI,CAAC,QAAQ,EAAE,KAAK,IAAI;AAC7E,WAAK,GAAG,IAAI,iBAAiB,YAAY,+CAA+C,OAAO,GAAG;AAGlG,YAAM,WAAW,QAAQ,IAAI,OAAK,IAAI,EAAE,QAAQ,MAAM,IAAI,CAAC,GAAG,EAAE,KAAK,IAAI;AACzE,YAAM,eAAe,QAAQ,IAAI,MAAM,GAAG,EAAE,KAAK,IAAI;AACrD,YAAM,aAAa,gBAAgB,YAAY,MAAM,QAAQ,aAAa,YAAY;AAEtF,iBAAW,OAAO,SAAS;AACzB,cAAM,SAAS,QAAQ,IAAI,OAAK;AAC9B,gBAAM,IAAI,IAAI,CAAC;AACf,cAAI,MAAM,QAAQ,MAAM,OAAW,QAAO;AAC1C,cAAI,OAAO,MAAM,SAAU,QAAO,KAAK,UAAU,CAAC;AAClD,iBAAO;AAAA,QACT,CAAC;AACD,aAAK,GAAG,IAAI,YAAY,MAAM;AAAA,MAChC;AAGA,YAAM,iBAAiB,eAAe,IAAI,UAAQ;AAAA,QAChD,MAAM,IAAI;AAAA,QACV,MAAM,IAAI;AAAA,QACV,aAAa,IAAI;AAAA,QACjB,SAAS,IAAI;AAAA,MACf,EAAE;AACF,WAAK,GAAG,IAAI;AAAA;AAAA;AAAA,SAGT,CAAC,WAAW,KAAK,IAAI,GAAG,WAAW,QAAQ,QAAQ,KAAK,UAAU,cAAc,GAAG,WAAW,CAAC;AAElG,WAAK,GAAG,IAAI,QAAQ;AAAA,IACtB,SAAS,KAAK;AACZ,WAAK,GAAG,IAAI,UAAU;AACtB,YAAM;AAAA,IACR;AAEA,WAAO,EAAE,MAAM,WAAW,SAAS,QAAQ,OAAO;AAAA,EACpD;AAAA;AAAA,EAIQ,gBAA0B;AAChC,UAAM,kBAAmB,KAAK,OAAe,eAAe,CAAC;AAC7D,UAAM,kBAAmB,KAAK,OAAe,eAAe,CAAC;AAC7D,UAAM,kBAAkB,gBAAgB,IAAI,CAAC,UAAc,iBAAAA,SAAU,CAAC,CAAC;AACvE,UAAM,kBAAkB,gBAAgB,IAAI,CAAC,UAAc,iBAAAA,SAAU,CAAC,CAAC;AACvE,UAAM,UAAoB,CAAC;AAE3B,UAAM,OAAO,CAAC,QAAgB;AAC5B,UAAI;AACJ,UAAI;AAAE,kBAAU,GAAG,YAAY,KAAK,EAAE,eAAe,KAAK,CAAC;AAAA,MAAG,QAAQ;AAAE;AAAA,MAAQ;AAEhF,iBAAW,SAAS,SAAS;AAC3B,cAAM,OAAO,KAAK,KAAK,KAAK,MAAM,IAAI;AACtC,cAAM,MAAM,KAAK,SAAS,KAAK,aAAa,IAAI,EAAE,QAAQ,OAAO,GAAG;AAEpE,YAAI,MAAM,YAAY,GAAG;AACvB,cAAI,gBAAgB,KAAK,CAAC,MAAW,EAAE,MAAM,GAAG,CAAC,EAAG;AACpD,eAAK,IAAI;AAAA,QACX,WAAW,MAAM,OAAO,GAAG;AACzB,cAAI,gBAAgB,KAAK,CAAC,MAAW,EAAE,GAAG,CAAC,EAAG;AAC9C,cAAI,gBAAgB,KAAK,CAAC,MAAW,EAAE,GAAG,CAAC,GAAG;AAE5C,oBAAI,8BAAc,GAAG,GAAG;AACtB,sBAAQ,KAAK,GAAG;AAAA,YAClB;AAAA,UACF;AAAA,QACF;AAAA,MACF;AAAA,IACF;AAEA,SAAK,KAAK,WAAW;AACrB,WAAO;AAAA,EACT;AAAA,EAEQ,sBAAsB,cAA8B;AAC1D,UAAM,aAAa,IAAI,IAAI,aAAa,IAAI,OAAK,kBAAkB,CAAC,CAAC,CAAC;AACtE,UAAM,UAAW,KAAK,GAAG,IAAI,8BAA8B,EAA4B,IAAI,OAAK,EAAE,EAAE;AAEpG,eAAW,MAAM,SAAS;AACxB,UAAI,CAAC,WAAW,IAAI,EAAE,GAAG;AACvB,cAAM,eAAe,aAAa,GAAG,QAAQ,kBAAkB,GAAG,CAAC;AACnE,aAAK,GAAG,IAAI,yBAAyB,YAAY,GAAG;AACpD,aAAK,GAAG,IAAI,iDAAiD,CAAC,EAAE,CAAC;AACjE,aAAK,GAAG,IAAI,mDAAmD,CAAC,EAAE,CAAC;AACnE,aAAK,GAAG,IAAI,0CAA0C,CAAC,EAAE,CAAC;AAAA,MAC5D;AAAA,IACF;AAAA,EACF;AACF;",
+  "names": ["picomatch"]
+}

package/dist/data/linker.js

@@ -0,0 +1,150 @@
+"use strict";
+var __defProp = Object.defineProperty;
+var __getOwnPropDesc = Object.getOwnPropertyDescriptor;
+var __getOwnPropNames = Object.getOwnPropertyNames;
+var __hasOwnProp = Object.prototype.hasOwnProperty;
+var __export = (target, all) => {
+  for (var name in all)
+    __defProp(target, name, { get: all[name], enumerable: true });
+};
+var __copyProps = (to, from, except, desc) => {
+  if (from && typeof from === "object" || typeof from === "function") {
+    for (let key of __getOwnPropNames(from))
+      if (!__hasOwnProp.call(to, key) && key !== except)
+        __defProp(to, key, { get: () => from[key], enumerable: !(desc = __getOwnPropDesc(from, key)) || desc.enumerable });
+  }
+  return to;
+};
+var __toCommonJS = (mod) => __copyProps(__defProp({}, "__esModule", { value: true }), mod);
+var linker_exports = {};
+__export(linker_exports, {
+  DataCodeLinker: () => DataCodeLinker,
+  detectDataReferences: () => detectDataReferences
+});
+module.exports = __toCommonJS(linker_exports);
+const READ_PATTERNS = [
+  // Node.js / TypeScript
+  /readFileSync\s*\(\s*['"`]([^'"`]+)['"`]/g,
+  /readFile\s*\(\s*['"`]([^'"`]+)['"`]/g,
+  /createReadStream\s*\(\s*['"`]([^'"`]+)['"`]/g,
+  // Python pandas
+  /read_csv\s*\(\s*['"`]([^'"`]+)['"`]/g,
+  /read_excel\s*\(\s*['"`]([^'"`]+)['"`]/g,
+  /read_json\s*\(\s*['"`]([^'"`]+)['"`]/g,
+  /read_parquet\s*\(\s*['"`]([^'"`]+)['"`]/g,
+  /read_table\s*\(\s*['"`]([^'"`]+)['"`]/g,
+  // Python open
+  /open\s*\(\s*['"`]([^'"`]+\.(?:csv|tsv|jsonl|ndjson|json|xlsx|xls|parquet))['"`]/g,
+  // SQL COPY
+  /COPY\s+\w+\s+FROM\s+['"`]([^'"`]+)['"`]/gi,
+  /\\copy\s+\w+\s+from\s+['"`]([^'"`]+)['"`]/gi,
+  // Generic path references in strings (data file extensions)
+  /['"`]((?:\.\.?\/)?(?:[\w\-./]+\/)?[\w\-]+\.(?:csv|tsv|jsonl|ndjson|xlsx|xls|parquet))['"`]/g
+];
+const WRITE_PATTERNS = [
+  /writeFileSync\s*\(\s*['"`]([^'"`]+\.(?:csv|tsv|jsonl|json|xlsx|parquet))['"`]/g,
+  /writeFile\s*\(\s*['"`]([^'"`]+\.(?:csv|tsv|jsonl|json|xlsx|parquet))['"`]/g,
+  /to_csv\s*\(\s*['"`]([^'"`]+)['"`]/g,
+  /to_excel\s*\(\s*['"`]([^'"`]+)['"`]/g,
+  /to_json\s*\(\s*['"`]([^'"`]+)['"`]/g,
+  /to_parquet\s*\(\s*['"`]([^'"`]+)['"`]/g,
+  /createWriteStream\s*\(\s*['"`]([^'"`]+\.(?:csv|tsv|jsonl|json|xlsx|parquet))['"`]/g,
+  /COPY\s+\w+\s+TO\s+['"`]([^'"`]+)['"`]/gi
+];
+function detectDataReferences(sourceContent, knownDataPaths) {
+  const refs = /* @__PURE__ */ new Map();
+  for (const pattern of READ_PATTERNS) {
+    pattern.lastIndex = 0;
+    let match;
+    while ((match = pattern.exec(sourceContent)) !== null) {
+      const filePath = normalizePath(match[1]);
+      if (isDataFilePath(filePath, knownDataPaths)) {
+        const existing = refs.get(filePath);
+        if (!existing || existing.confidence < 0.9) {
+          const confidence = knownDataPaths.has(filePath) ? 1 : 0.7;
+          refs.set(filePath, { refType: "reads", confidence });
+        }
+      }
+    }
+  }
+  for (const pattern of WRITE_PATTERNS) {
+    pattern.lastIndex = 0;
+    let match;
+    while ((match = pattern.exec(sourceContent)) !== null) {
+      const filePath = normalizePath(match[1]);
+      if (isDataFilePath(filePath, knownDataPaths)) {
+        const confidence = knownDataPaths.has(filePath) ? 1 : 0.7;
+        refs.set(filePath, { refType: "writes", confidence });
+      }
+    }
+  }
+  return Array.from(refs.entries()).map(([filePath, info]) => ({
+    filePath,
+    ...info
+  }));
+}
+function generateDatasetId(filePath) {
+  return filePath.replace(/\.[^.]+$/, "").replace(/[/\\]/g, "-").replace(/[^a-zA-Z0-9-_]/g, "-").replace(/-+/g, "-").replace(/^-|-$/g, "").toLowerCase();
+}
+function normalizePath(p) {
+  return p.replace(/^\.\//, "").replace(/\\/g, "/");
+}
+function isDataFilePath(filePath, knownPaths) {
+  if (knownPaths.has(filePath)) return true;
+  const ext = filePath.split(".").pop()?.toLowerCase();
+  return ["csv", "tsv", "jsonl", "ndjson", "json", "xlsx", "xls", "parquet"].includes(ext ?? "");
+}
+class DataCodeLinker {
+  constructor(db, projectRoot) {
+    this.db = db;
+    this.projectRoot = projectRoot;
+  }
+  /**
+   * Populate data_code_refs by scanning source files for data file references.
+   * Returns the number of refs created.
+   */
+  linkAll() {
+    const fs = require("fs");
+    const path = require("path");
+    const datasets = this.db.all("SELECT id, file_path FROM data_datasets");
+    if (datasets.length === 0) return 0;
+    const knownPaths = new Set(datasets.map((d) => d.file_path));
+    const pathToDatasetId = new Map(datasets.map((d) => [d.file_path, d.id]));
+    const sourceFiles = this.db.all(
+      `SELECT file_path FROM files WHERE file_path NOT LIKE '%.csv' AND file_path NOT LIKE '%.tsv' AND file_path NOT LIKE '%.jsonl' AND file_path NOT LIKE '%.ndjson' AND file_path NOT LIKE '%.xlsx' AND file_path NOT LIKE '%.parquet'`
+    );
+    this.db.run("DELETE FROM data_code_refs");
+    let refsCreated = 0;
+    for (const file of sourceFiles) {
+      const absPath = path.join(this.projectRoot, file.file_path);
+      let content;
+      try {
+        content = fs.readFileSync(absPath, "utf-8");
+      } catch {
+        continue;
+      }
+      const refs = detectDataReferences(content, knownPaths);
+      if (refs.length === 0) continue;
+      const qualifiedName = file.file_path;
+      for (const ref of refs) {
+        const datasetId = pathToDatasetId.get(ref.filePath);
+        if (!datasetId) continue;
+        try {
+          this.db.run(
+            `INSERT OR REPLACE INTO data_code_refs (dataset_id, qualified_name, ref_type, confidence) VALUES (?, ?, ?, ?)`,
+            [datasetId, qualifiedName, ref.refType, ref.confidence]
+          );
+          refsCreated++;
+        } catch {
+        }
+      }
+    }
+    return refsCreated;
+  }
+}
+// Annotate the CommonJS export names for ESM import in node:
+0 && (module.exports = {
+  DataCodeLinker,
+  detectDataReferences
+});
+//# sourceMappingURL=linker.js.map

package/dist/data/linker.js.map

@@ -0,0 +1,7 @@
+{
+  "version": 3,
+  "sources": ["../../src/data/linker.ts"],
+  "sourcesContent": ["/**\n * Data \u2194 Code Linker\n *\n * Detects references to data files in source code and populates data_code_refs.\n * Patterns detected:\n *   - readFileSync('data/users.csv'), readFile('path/to/file.csv')\n *   - pd.read_csv('data/file.csv'), pd.read_excel(...)\n *   - open('data/file.csv'), csv.reader(open(...))\n *   - COPY FROM 'file.csv', \\copy ... from 'file.csv'\n *   - require('./data/file.json'), import ... from './data/file.json'\n *   - fs.createReadStream('data/file.csv')\n *   - String literals matching known data file paths\n */\n\nimport type { DataCodeRef } from './types';\n\n/** Patterns that indicate a code symbol reads a data file */\nconst READ_PATTERNS = [\n  // Node.js / TypeScript\n  /readFileSync\\s*\\(\\s*['\"`]([^'\"`]+)['\"`]/g,\n  /readFile\\s*\\(\\s*['\"`]([^'\"`]+)['\"`]/g,\n  /createReadStream\\s*\\(\\s*['\"`]([^'\"`]+)['\"`]/g,\n  // Python pandas\n  /read_csv\\s*\\(\\s*['\"`]([^'\"`]+)['\"`]/g,\n  /read_excel\\s*\\(\\s*['\"`]([^'\"`]+)['\"`]/g,\n  /read_json\\s*\\(\\s*['\"`]([^'\"`]+)['\"`]/g,\n  /read_parquet\\s*\\(\\s*['\"`]([^'\"`]+)['\"`]/g,\n  /read_table\\s*\\(\\s*['\"`]([^'\"`]+)['\"`]/g,\n  // Python open\n  /open\\s*\\(\\s*['\"`]([^'\"`]+\\.(?:csv|tsv|jsonl|ndjson|json|xlsx|xls|parquet))['\"`]/g,\n  // SQL COPY\n  /COPY\\s+\\w+\\s+FROM\\s+['\"`]([^'\"`]+)['\"`]/gi,\n  /\\\\copy\\s+\\w+\\s+from\\s+['\"`]([^'\"`]+)['\"`]/gi,\n  // Generic path references in strings (data file extensions)\n  /['\"`]((?:\\.\\.?\\/)?(?:[\\w\\-./]+\\/)?[\\w\\-]+\\.(?:csv|tsv|jsonl|ndjson|xlsx|xls|parquet))['\"`]/g,\n];\n\n/** Patterns that indicate a code symbol writes a data file */\nconst WRITE_PATTERNS = [\n  /writeFileSync\\s*\\(\\s*['\"`]([^'\"`]+\\.(?:csv|tsv|jsonl|json|xlsx|parquet))['\"`]/g,\n  /writeFile\\s*\\(\\s*['\"`]([^'\"`]+\\.(?:csv|tsv|jsonl|json|xlsx|parquet))['\"`]/g,\n  /to_csv\\s*\\(\\s*['\"`]([^'\"`]+)['\"`]/g,\n  /to_excel\\s*\\(\\s*['\"`]([^'\"`]+)['\"`]/g,\n  /to_json\\s*\\(\\s*['\"`]([^'\"`]+)['\"`]/g,\n  /to_parquet\\s*\\(\\s*['\"`]([^'\"`]+)['\"`]/g,\n  /createWriteStream\\s*\\(\\s*['\"`]([^'\"`]+\\.(?:csv|tsv|jsonl|json|xlsx|parquet))['\"`]/g,\n  /COPY\\s+\\w+\\s+TO\\s+['\"`]([^'\"`]+)['\"`]/gi,\n];\n\n/**\n * Scan source code content for references to data files.\n * Returns matched file paths with their reference type.\n */\nexport function detectDataReferences(\n  sourceContent: string,\n  knownDataPaths: Set<string>,\n): Array<{ filePath: string; refType: 'reads' | 'writes'; confidence: number }> {\n  const refs = new Map<string, { refType: 'reads' | 'writes'; confidence: number }>();\n\n  // Check read patterns\n  for (const pattern of READ_PATTERNS) {\n    pattern.lastIndex = 0;\n    let match: RegExpExecArray | null;\n    while ((match = pattern.exec(sourceContent)) !== null) {\n      const filePath = normalizePath(match[1]!);\n      if (isDataFilePath(filePath, knownDataPaths)) {\n        const existing = refs.get(filePath);\n        if (!existing || existing.confidence < 0.9) {\n          const confidence = knownDataPaths.has(filePath) ? 1.0 : 0.7;\n          refs.set(filePath, { refType: 'reads', confidence });\n        }\n      }\n    }\n  }\n\n  // Check write patterns\n  for (const pattern of WRITE_PATTERNS) {\n    pattern.lastIndex = 0;\n    let match: RegExpExecArray | null;\n    while ((match = pattern.exec(sourceContent)) !== null) {\n      const filePath = normalizePath(match[1]!);\n      if (isDataFilePath(filePath, knownDataPaths)) {\n        const confidence = knownDataPaths.has(filePath) ? 1.0 : 0.7;\n        refs.set(filePath, { refType: 'writes', confidence });\n      }\n    }\n  }\n\n  return Array.from(refs.entries()).map(([filePath, info]) => ({\n    filePath,\n    ...info,\n  }));\n}\n\n/**\n * Generate dataset ID from a file path (same logic as indexer).\n */\nfunction generateDatasetId(filePath: string): string {\n  return filePath\n    .replace(/\\.[^.]+$/, '')\n    .replace(/[/\\\\]/g, '-')\n    .replace(/[^a-zA-Z0-9-_]/g, '-')\n    .replace(/-+/g, '-')\n    .replace(/^-|-$/g, '')\n    .toLowerCase();\n}\n\n/**\n * Normalize a path reference found in code.\n * Strips leading ./ and resolves simple relative paths.\n */\nfunction normalizePath(p: string): string {\n  return p.replace(/^\\.\\//, '').replace(/\\\\/g, '/');\n}\n\n/**\n * Check if a path looks like a data file (by extension or known path match).\n */\nfunction isDataFilePath(filePath: string, knownPaths: Set<string>): boolean {\n  if (knownPaths.has(filePath)) return true;\n  // Check extension\n  const ext = filePath.split('.').pop()?.toLowerCase();\n  return ['csv', 'tsv', 'jsonl', 'ndjson', 'json', 'xlsx', 'xls', 'parquet'].includes(ext ?? '');\n}\n\n/**\n * Link data files to code symbols.\n * Scans all indexed source files for references to known data file paths.\n */\nexport class DataCodeLinker {\n  private readonly db: any;\n  private readonly projectRoot: string;\n\n  constructor(db: any, projectRoot: string) {\n    this.db = db;\n    this.projectRoot = projectRoot;\n  }\n\n  /**\n   * Populate data_code_refs by scanning source files for data file references.\n   * Returns the number of refs created.\n   */\n  linkAll(): number {\n    const fs = require('fs');\n    const path = require('path');\n\n    // Get all known data file paths\n    const datasets = this.db.all('SELECT id, file_path FROM data_datasets') as Array<{ id: string; file_path: string }>;\n    if (datasets.length === 0) return 0;\n\n    const knownPaths = new Set(datasets.map(d => d.file_path));\n    const pathToDatasetId = new Map(datasets.map(d => [d.file_path, d.id]));\n\n    // Get all indexed source files (from the graph)\n    const sourceFiles = this.db.all(\n      `SELECT file_path FROM files WHERE file_path NOT LIKE '%.csv' AND file_path NOT LIKE '%.tsv' AND file_path NOT LIKE '%.jsonl' AND file_path NOT LIKE '%.ndjson' AND file_path NOT LIKE '%.xlsx' AND file_path NOT LIKE '%.parquet'`\n    ) as Array<{ file_path: string }>;\n\n    // Clear existing refs\n    this.db.run('DELETE FROM data_code_refs');\n\n    let refsCreated = 0;\n\n    for (const file of sourceFiles) {\n      const absPath = path.join(this.projectRoot, file.file_path);\n      let content: string;\n      try {\n        content = fs.readFileSync(absPath, 'utf-8');\n      } catch {\n        continue;\n      }\n\n      const refs = detectDataReferences(content, knownPaths);\n      if (refs.length === 0) continue;\n\n      // Find the qualified_name of the containing symbol (file-level)\n      // Use the file path as the qualified name for file-level refs\n      const qualifiedName = file.file_path;\n\n      for (const ref of refs) {\n        const datasetId = pathToDatasetId.get(ref.filePath);\n        if (!datasetId) continue;\n\n        try {\n          this.db.run(\n            `INSERT OR REPLACE INTO data_code_refs (dataset_id, qualified_name, ref_type, confidence) VALUES (?, ?, ?, ?)`,\n            [datasetId, qualifiedName, ref.refType, ref.confidence],\n          );\n          refsCreated++;\n        } catch {\n          // Ignore duplicates\n        }\n      }\n    }\n\n    return refsCreated;\n  }\n}\n"],
+  "mappings": ";;;;;;;;;;;;;;;;;;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAiBA,MAAM,gBAAgB;AAAA;AAAA,EAEpB;AAAA,EACA;AAAA,EACA;AAAA;AAAA,EAEA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA;AAAA,EAEA;AAAA;AAAA,EAEA;AAAA,EACA;AAAA;AAAA,EAEA;AACF;AAGA,MAAM,iBAAiB;AAAA,EACrB;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AACF;AAMO,SAAS,qBACd,eACA,gBAC8E;AAC9E,QAAM,OAAO,oBAAI,IAAiE;AAGlF,aAAW,WAAW,eAAe;AACnC,YAAQ,YAAY;AACpB,QAAI;AACJ,YAAQ,QAAQ,QAAQ,KAAK,aAAa,OAAO,MAAM;AACrD,YAAM,WAAW,cAAc,MAAM,CAAC,CAAE;AACxC,UAAI,eAAe,UAAU,cAAc,GAAG;AAC5C,cAAM,WAAW,KAAK,IAAI,QAAQ;AAClC,YAAI,CAAC,YAAY,SAAS,aAAa,KAAK;AAC1C,gBAAM,aAAa,eAAe,IAAI,QAAQ,IAAI,IAAM;AACxD,eAAK,IAAI,UAAU,EAAE,SAAS,SAAS,WAAW,CAAC;AAAA,QACrD;AAAA,MACF;AAAA,IACF;AAAA,EACF;AAGA,aAAW,WAAW,gBAAgB;AACpC,YAAQ,YAAY;AACpB,QAAI;AACJ,YAAQ,QAAQ,QAAQ,KAAK,aAAa,OAAO,MAAM;AACrD,YAAM,WAAW,cAAc,MAAM,CAAC,CAAE;AACxC,UAAI,eAAe,UAAU,cAAc,GAAG;AAC5C,cAAM,aAAa,eAAe,IAAI,QAAQ,IAAI,IAAM;AACxD,aAAK,IAAI,UAAU,EAAE,SAAS,UAAU,WAAW,CAAC;AAAA,MACtD;AAAA,IACF;AAAA,EACF;AAEA,SAAO,MAAM,KAAK,KAAK,QAAQ,CAAC,EAAE,IAAI,CAAC,CAAC,UAAU,IAAI,OAAO;AAAA,IAC3D;AAAA,IACA,GAAG;AAAA,EACL,EAAE;AACJ;AAKA,SAAS,kBAAkB,UAA0B;AACnD,SAAO,SACJ,QAAQ,YAAY,EAAE,EACtB,QAAQ,UAAU,GAAG,EACrB,QAAQ,mBAAmB,GAAG,EAC9B,QAAQ,OAAO,GAAG,EAClB,QAAQ,UAAU,EAAE,EACpB,YAAY;AACjB;AAMA,SAAS,cAAc,GAAmB;AACxC,SAAO,EAAE,QAAQ,SAAS,EAAE,EAAE,QAAQ,OAAO,GAAG;AAClD;AAKA,SAAS,eAAe,UAAkB,YAAkC;AAC1E,MAAI,WAAW,IAAI,QAAQ,EAAG,QAAO;AAErC,QAAM,MAAM,SAAS,MAAM,GAAG,EAAE,IAAI,GAAG,YAAY;AACnD,SAAO,CAAC,OAAO,OAAO,SAAS,UAAU,QAAQ,QAAQ,OAAO,SAAS,EAAE,SAAS,OAAO,EAAE;AAC/F;AAMO,MAAM,eAAe;AAAA,EAI1B,YAAY,IAAS,aAAqB;AACxC,SAAK,KAAK;AACV,SAAK,cAAc;AAAA,EACrB;AAAA;AAAA;AAAA;AAAA;AAAA,EAMA,UAAkB;AAChB,UAAM,KAAK,QAAQ,IAAI;AACvB,UAAM,OAAO,QAAQ,MAAM;AAG3B,UAAM,WAAW,KAAK,GAAG,IAAI,yCAAyC;AACtE,QAAI,SAAS,WAAW,EAAG,QAAO;AAElC,UAAM,aAAa,IAAI,IAAI,SAAS,IAAI,OAAK,EAAE,SAAS,CAAC;AACzD,UAAM,kBAAkB,IAAI,IAAI,SAAS,IAAI,OAAK,CAAC,EAAE,WAAW,EAAE,EAAE,CAAC,CAAC;AAGtE,UAAM,cAAc,KAAK,GAAG;AAAA,MAC1B;AAAA,IACF;AAGA,SAAK,GAAG,IAAI,4BAA4B;AAExC,QAAI,cAAc;AAElB,eAAW,QAAQ,aAAa;AAC9B,YAAM,UAAU,KAAK,KAAK,KAAK,aAAa,KAAK,SAAS;AAC1D,UAAI;AACJ,UAAI;AACF,kBAAU,GAAG,aAAa,SAAS,OAAO;AAAA,MAC5C,QAAQ;AACN;AAAA,MACF;AAEA,YAAM,OAAO,qBAAqB,SAAS,UAAU;AACrD,UAAI,KAAK,WAAW,EAAG;AAIvB,YAAM,gBAAgB,KAAK;AAE3B,iBAAW,OAAO,MAAM;AACtB,cAAM,YAAY,gBAAgB,IAAI,IAAI,QAAQ;AAClD,YAAI,CAAC,UAAW;AAEhB,YAAI;AACF,eAAK,GAAG;AAAA,YACN;AAAA,YACA,CAAC,WAAW,eAAe,IAAI,SAAS,IAAI,UAAU;AAAA,UACxD;AACA;AAAA,QACF,QAAQ;AAAA,QAER;AAAA,MACF;AAAA,IACF;AAEA,WAAO;AAAA,EACT;AACF;",
+  "names": []
+}