@tanstack/db 0.1.5 → 0.1.7

package/dist/esm/query/optimizer.js.map

@@ -1 +1 @@
-{"version":3,"file":"optimizer.js","sources":["../../../src/query/optimizer.ts"],"sourcesContent":["/**\n * # Query Optimizer\n *\n * The query optimizer improves query performance by implementing predicate pushdown optimization.\n * It rewrites the intermediate representation (IR) to push WHERE clauses as close to the data\n * source as possible, reducing the amount of data processed during joins.\n *\n * ## How It Works\n *\n * The optimizer follows a 4-step process:\n *\n * ### 1. AND Clause Splitting\n * Splits AND clauses at the root level into separate WHERE clauses for granular optimization.\n * ```javascript\n * // Before: WHERE and(eq(users.department_id, 1), gt(users.age, 25))\n * // After:  WHERE eq(users.department_id, 1) + WHERE gt(users.age, 25)\n * ```\n *\n * ### 2. Source Analysis\n * Analyzes each WHERE clause to determine which table sources it references:\n * - Single-source clauses: Touch only one table (e.g., `users.department_id = 1`)\n * - Multi-source clauses: Touch multiple tables (e.g., `users.id = posts.user_id`)\n *\n * ### 3. Clause Grouping\n * Groups WHERE clauses by the sources they touch:\n * - Single-source clauses are grouped by their respective table\n * - Multi-source clauses are combined for the main query\n *\n * ### 4. Subquery Creation\n * Lifts single-source WHERE clauses into subqueries that wrap the original table references.\n *\n * ## Safety & Edge Cases\n *\n * The optimizer includes targeted safety checks to prevent predicate pushdown when it could\n * break query semantics:\n *\n * ### Always Safe Operations\n * - **Creating new subqueries**: Wrapping collection references in subqueries with WHERE clauses\n * - **Main query optimizations**: Moving single-source WHERE clauses from main query to subqueries\n * - **Queries with aggregates/ORDER BY/HAVING**: Can still create new filtered subqueries\n *\n * ### Unsafe Operations (blocked by safety checks)\n * Pushing WHERE clauses **into existing subqueries** that have:\n * - **Aggregates**: GROUP BY, HAVING, or aggregate functions in SELECT (would change aggregation)\n * - **Ordering + Limits**: ORDER BY combined with LIMIT/OFFSET (would change result set)\n * - **Functional Operations**: fnSelect, fnWhere, fnHaving (potential side effects)\n *\n * The optimizer tracks which clauses were actually optimized and only removes those from the\n * main query. Subquery reuse is handled safely through immutable query copies.\n *\n * ## Example Optimizations\n *\n * ### Basic Query with Joins\n * **Original Query:**\n * ```javascript\n * query\n *   .from({ users: usersCollection })\n *   .join({ posts: postsCollection }, ({users, posts}) => eq(users.id, posts.user_id))\n *   .where(({users}) => eq(users.department_id, 1))\n *   .where(({posts}) => gt(posts.views, 100))\n *   .where(({users, posts}) => eq(users.id, posts.author_id))\n * ```\n *\n * **Optimized Query:**\n * ```javascript\n * query\n *   .from({\n *     users: subquery\n *       .from({ users: usersCollection })\n *       .where(({users}) => eq(users.department_id, 1))\n *   })\n *   .join({\n *     posts: subquery\n *       .from({ posts: postsCollection })\n *       .where(({posts}) => gt(posts.views, 100))\n *   }, ({users, posts}) => eq(users.id, posts.user_id))\n *   .where(({users, posts}) => eq(users.id, posts.author_id))\n * ```\n *\n * ### Query with Aggregates (Now Optimizable!)\n * **Original Query:**\n * ```javascript\n * query\n *   .from({ users: usersCollection })\n *   .join({ posts: postsCollection }, ({users, posts}) => eq(users.id, posts.user_id))\n *   .where(({users}) => eq(users.department_id, 1))\n *   .groupBy(['users.department_id'])\n *   .select({ count: agg('count', '*') })\n * ```\n *\n * **Optimized Query:**\n * ```javascript\n * query\n *   .from({\n *     users: subquery\n *       .from({ users: usersCollection })\n *       .where(({users}) => eq(users.department_id, 1))\n *   })\n *   .join({ posts: postsCollection }, ({users, posts}) => eq(users.id, posts.user_id))\n *   .groupBy(['users.department_id'])\n *   .select({ count: agg('count', '*') })\n * ```\n *\n * ## Benefits\n *\n * - **Reduced Data Processing**: Filters applied before joins reduce intermediate result size\n * - **Better Performance**: Smaller datasets lead to faster query execution\n * - **Automatic Optimization**: No manual query rewriting required\n * - **Preserves Semantics**: Optimized queries return identical results\n * - **Safe by Design**: Comprehensive checks prevent semantic-breaking optimizations\n *\n * ## Integration\n *\n * The optimizer is automatically called during query compilation before the IR is\n * transformed into a D2Mini pipeline.\n */\n\nimport { deepEquals } from \"../utils.js\"\nimport { CannotCombineEmptyExpressionListError } from \"../errors.js\"\nimport {\n  CollectionRef as CollectionRefClass,\n  Func,\n  QueryRef as QueryRefClass,\n} from \"./ir.js\"\nimport { isConvertibleToCollectionFilter } from \"./compiler/expressions.js\"\nimport type { BasicExpression, From, QueryIR } from \"./ir.js\"\n\n/**\n * Represents a WHERE clause after source analysis\n */\nexport interface AnalyzedWhereClause {\n  /** The WHERE expression */\n  expression: BasicExpression<boolean>\n  /** Set of table/source aliases that this WHERE clause touches */\n  touchedSources: Set<string>\n}\n\n/**\n * Represents WHERE clauses grouped by the sources they touch\n */\nexport interface GroupedWhereClauses {\n  /** WHERE clauses that touch only a single source, grouped by source alias */\n  singleSource: Map<string, BasicExpression<boolean>>\n  /** WHERE clauses that touch multiple sources, combined into one expression */\n  multiSource?: BasicExpression<boolean>\n}\n\n/**\n * Result of query optimization including both the optimized query and collection-specific WHERE clauses\n */\nexport interface OptimizationResult {\n  /** The optimized query with WHERE clauses potentially moved to subqueries */\n  optimizedQuery: QueryIR\n  /** Map of collection aliases to their extracted WHERE clauses for index optimization */\n  collectionWhereClauses: Map<string, BasicExpression<boolean>>\n}\n\n/**\n * Main query optimizer entry point that lifts WHERE clauses into subqueries.\n *\n * This function implements multi-level predicate pushdown optimization by recursively\n * moving WHERE clauses through nested subqueries to get them as close to the data\n * sources as possible, then removing redundant subqueries.\n *\n * @param query - The QueryIR to optimize\n * @returns An OptimizationResult with the optimized query and collection WHERE clause mapping\n *\n * @example\n * ```typescript\n * const originalQuery = {\n *   from: new CollectionRef(users, 'u'),\n *   join: [{ from: new CollectionRef(posts, 'p'), ... }],\n *   where: [eq(u.dept_id, 1), gt(p.views, 100)]\n * }\n *\n * const { optimizedQuery, collectionWhereClauses } = optimizeQuery(originalQuery)\n * // Result: Single-source clauses moved to deepest possible subqueries\n * // collectionWhereClauses: Map { 'u' => eq(u.dept_id, 1), 'p' => gt(p.views, 100) }\n * ```\n */\nexport function optimizeQuery(query: QueryIR): OptimizationResult {\n  // First, extract collection WHERE clauses before optimization\n  const collectionWhereClauses = extractCollectionWhereClauses(query)\n\n  // Apply multi-level predicate pushdown with iterative convergence\n  let optimized = query\n  let previousOptimized: QueryIR | undefined\n  let iterations = 0\n  const maxIterations = 10 // Prevent infinite loops\n\n  // Keep optimizing until no more changes occur or max iterations reached\n  while (\n    iterations < maxIterations &&\n    !deepEquals(optimized, previousOptimized)\n  ) {\n    previousOptimized = optimized\n    optimized = applyRecursiveOptimization(optimized)\n    iterations++\n  }\n\n  // Remove redundant subqueries\n  const cleaned = removeRedundantSubqueries(optimized)\n\n  return {\n    optimizedQuery: cleaned,\n    collectionWhereClauses,\n  }\n}\n\n/**\n * Extracts collection-specific WHERE clauses from a query for index optimization.\n * This analyzes the original query to identify WHERE clauses that can be pushed down\n * to specific collections, but only for simple queries without joins.\n *\n * @param query - The original QueryIR to analyze\n * @returns Map of collection aliases to their WHERE clauses\n */\nfunction extractCollectionWhereClauses(\n  query: QueryIR\n): Map<string, BasicExpression<boolean>> {\n  const collectionWhereClauses = new Map<string, BasicExpression<boolean>>()\n\n  // Only analyze queries that have WHERE clauses\n  if (!query.where || query.where.length === 0) {\n    return collectionWhereClauses\n  }\n\n  // Split all AND clauses at the root level for granular analysis\n  const splitWhereClauses = splitAndClauses(query.where)\n\n  // Analyze each WHERE clause to determine which sources it touches\n  const analyzedClauses = splitWhereClauses.map((clause) =>\n    analyzeWhereClause(clause)\n  )\n\n  // Group clauses by single-source vs multi-source\n  const groupedClauses = groupWhereClauses(analyzedClauses)\n\n  // Only include single-source clauses that reference collections directly\n  // and can be converted to BasicExpression format for collection indexes\n  for (const [sourceAlias, whereClause] of groupedClauses.singleSource) {\n    // Check if this source alias corresponds to a collection reference\n    if (isCollectionReference(query, sourceAlias)) {\n      // Check if the WHERE clause can be converted to collection-compatible format\n      if (isConvertibleToCollectionFilter(whereClause)) {\n        collectionWhereClauses.set(sourceAlias, whereClause)\n      }\n    }\n  }\n\n  return collectionWhereClauses\n}\n\n/**\n * Determines if a source alias refers to a collection reference (not a subquery).\n * This is used to identify WHERE clauses that can be pushed down to collection subscriptions.\n *\n * @param query - The query to analyze\n * @param sourceAlias - The source alias to check\n * @returns True if the alias refers to a collection reference\n */\nfunction isCollectionReference(query: QueryIR, sourceAlias: string): boolean {\n  // Check the FROM clause\n  if (query.from.alias === sourceAlias) {\n    return query.from.type === `collectionRef`\n  }\n\n  // Check JOIN clauses\n  if (query.join) {\n    for (const joinClause of query.join) {\n      if (joinClause.from.alias === sourceAlias) {\n        return joinClause.from.type === `collectionRef`\n      }\n    }\n  }\n\n  return false\n}\n\n/**\n * Applies recursive predicate pushdown optimization.\n *\n * @param query - The QueryIR to optimize\n * @returns A new QueryIR with optimizations applied\n */\nfunction applyRecursiveOptimization(query: QueryIR): QueryIR {\n  // First, recursively optimize any existing subqueries\n  const subqueriesOptimized = {\n    ...query,\n    from:\n      query.from.type === `queryRef`\n        ? new QueryRefClass(\n            applyRecursiveOptimization(query.from.query),\n            query.from.alias\n          )\n        : query.from,\n    join: query.join?.map((joinClause) => ({\n      ...joinClause,\n      from:\n        joinClause.from.type === `queryRef`\n          ? new QueryRefClass(\n              applyRecursiveOptimization(joinClause.from.query),\n              joinClause.from.alias\n            )\n          : joinClause.from,\n    })),\n  }\n\n  // Then apply single-level optimization to this query\n  return applySingleLevelOptimization(subqueriesOptimized)\n}\n\n/**\n * Applies single-level predicate pushdown optimization (existing logic)\n */\nfunction applySingleLevelOptimization(query: QueryIR): QueryIR {\n  // Skip optimization if no WHERE clauses exist\n  if (!query.where || query.where.length === 0) {\n    return query\n  }\n\n  // Skip optimization if there are no joins - predicate pushdown only benefits joins\n  // Single-table queries don't benefit from this optimization\n  if (!query.join || query.join.length === 0) {\n    return query\n  }\n\n  // Step 1: Split all AND clauses at the root level for granular optimization\n  const splitWhereClauses = splitAndClauses(query.where)\n\n  // Step 2: Analyze each WHERE clause to determine which sources it touches\n  const analyzedClauses = splitWhereClauses.map((clause) =>\n    analyzeWhereClause(clause)\n  )\n\n  // Step 3: Group clauses by single-source vs multi-source\n  const groupedClauses = groupWhereClauses(analyzedClauses)\n\n  // Step 4: Apply optimizations by lifting single-source clauses into subqueries\n  return applyOptimizations(query, groupedClauses)\n}\n\n/**\n * Removes redundant subqueries that don't add value.\n * A subquery is redundant if it only wraps another query without adding\n * WHERE, SELECT, GROUP BY, HAVING, ORDER BY, or LIMIT/OFFSET clauses.\n *\n * @param query - The QueryIR to process\n * @returns A new QueryIR with redundant subqueries removed\n */\nfunction removeRedundantSubqueries(query: QueryIR): QueryIR {\n  return {\n    ...query,\n    from: removeRedundantFromClause(query.from),\n    join: query.join?.map((joinClause) => ({\n      ...joinClause,\n      from: removeRedundantFromClause(joinClause.from),\n    })),\n  }\n}\n\n/**\n * Removes redundant subqueries from a FROM clause.\n *\n * @param from - The FROM clause to process\n * @returns A FROM clause with redundant subqueries removed\n */\nfunction removeRedundantFromClause(from: From): From {\n  if (from.type === `collectionRef`) {\n    return from\n  }\n\n  const processedQuery = removeRedundantSubqueries(from.query)\n\n  // Check if this subquery is redundant\n  if (isRedundantSubquery(processedQuery)) {\n    // Return the inner query's FROM clause with this alias\n    const innerFrom = removeRedundantFromClause(processedQuery.from)\n    if (innerFrom.type === `collectionRef`) {\n      return new CollectionRefClass(innerFrom.collection, from.alias)\n    } else {\n      return new QueryRefClass(innerFrom.query, from.alias)\n    }\n  }\n\n  return new QueryRefClass(processedQuery, from.alias)\n}\n\n/**\n * Determines if a subquery is redundant (adds no value).\n *\n * @param query - The query to check\n * @returns True if the query is redundant and can be removed\n */\nfunction isRedundantSubquery(query: QueryIR): boolean {\n  return (\n    (!query.where || query.where.length === 0) &&\n    !query.select &&\n    (!query.groupBy || query.groupBy.length === 0) &&\n    (!query.having || query.having.length === 0) &&\n    (!query.orderBy || query.orderBy.length === 0) &&\n    (!query.join || query.join.length === 0) &&\n    query.limit === undefined &&\n    query.offset === undefined &&\n    !query.fnSelect &&\n    (!query.fnWhere || query.fnWhere.length === 0) &&\n    (!query.fnHaving || query.fnHaving.length === 0)\n  )\n}\n\n/**\n * Step 1: Split all AND clauses recursively into separate WHERE clauses.\n *\n * This enables more granular optimization by treating each condition independently.\n * OR clauses are preserved as they cannot be split without changing query semantics.\n *\n * @param whereClauses - Array of WHERE expressions to split\n * @returns Flattened array with AND clauses split into separate expressions\n *\n * @example\n * ```typescript\n * // Input: [and(eq(a, 1), gt(b, 2)), eq(c, 3)]\n * // Output: [eq(a, 1), gt(b, 2), eq(c, 3)]\n * ```\n */\nfunction splitAndClauses(\n  whereClauses: Array<BasicExpression<boolean>>\n): Array<BasicExpression<boolean>> {\n  const result: Array<BasicExpression<boolean>> = []\n\n  for (const clause of whereClauses) {\n    if (clause.type === `func` && clause.name === `and`) {\n      // Recursively split nested AND clauses to handle complex expressions\n      const splitArgs = splitAndClauses(\n        clause.args as Array<BasicExpression<boolean>>\n      )\n      result.push(...splitArgs)\n    } else {\n      // Preserve non-AND clauses as-is (including OR clauses)\n      result.push(clause)\n    }\n  }\n\n  return result\n}\n\n/**\n * Step 2: Analyze which table sources a WHERE clause touches.\n *\n * This determines whether a clause can be pushed down to a specific table\n * or must remain in the main query (for multi-source clauses like join conditions).\n *\n * @param clause - The WHERE expression to analyze\n * @returns Analysis result with the expression and touched source aliases\n *\n * @example\n * ```typescript\n * // eq(users.department_id, 1) -> touches ['users']\n * // eq(users.id, posts.user_id) -> touches ['users', 'posts']\n * ```\n */\nfunction analyzeWhereClause(\n  clause: BasicExpression<boolean>\n): AnalyzedWhereClause {\n  const touchedSources = new Set<string>()\n\n  /**\n   * Recursively collect all table aliases referenced in an expression\n   */\n  function collectSources(expr: BasicExpression | any): void {\n    switch (expr.type) {\n      case `ref`:\n        // PropRef path has the table alias as the first element\n        if (expr.path && expr.path.length > 0) {\n          const firstElement = expr.path[0]\n          if (firstElement) {\n            touchedSources.add(firstElement)\n          }\n        }\n        break\n      case `func`:\n        // Recursively analyze function arguments (e.g., eq, gt, and, or)\n        if (expr.args) {\n          expr.args.forEach(collectSources)\n        }\n        break\n      case `val`:\n        // Values don't reference any sources\n        break\n      case `agg`:\n        // Aggregates can reference sources in their arguments\n        if (expr.args) {\n          expr.args.forEach(collectSources)\n        }\n        break\n    }\n  }\n\n  collectSources(clause)\n\n  return {\n    expression: clause,\n    touchedSources,\n  }\n}\n\n/**\n * Step 3: Group WHERE clauses by the sources they touch.\n *\n * Single-source clauses can be pushed down to subqueries for optimization.\n * Multi-source clauses must remain in the main query to preserve join semantics.\n *\n * @param analyzedClauses - Array of analyzed WHERE clauses\n * @returns Grouped clauses ready for optimization\n */\nfunction groupWhereClauses(\n  analyzedClauses: Array<AnalyzedWhereClause>\n): GroupedWhereClauses {\n  const singleSource = new Map<string, Array<BasicExpression<boolean>>>()\n  const multiSource: Array<BasicExpression<boolean>> = []\n\n  // Categorize each clause based on how many sources it touches\n  for (const clause of analyzedClauses) {\n    if (clause.touchedSources.size === 1) {\n      // Single source clause - can be optimized\n      const source = Array.from(clause.touchedSources)[0]!\n      if (!singleSource.has(source)) {\n        singleSource.set(source, [])\n      }\n      singleSource.get(source)!.push(clause.expression)\n    } else if (clause.touchedSources.size > 1) {\n      // Multi-source clause - must stay in main query\n      multiSource.push(clause.expression)\n    }\n    // Skip clauses that touch no sources (constants) - they don't need optimization\n  }\n\n  // Combine multiple clauses for each source with AND\n  const combinedSingleSource = new Map<string, BasicExpression<boolean>>()\n  for (const [source, clauses] of singleSource) {\n    combinedSingleSource.set(source, combineWithAnd(clauses))\n  }\n\n  // Combine multi-source clauses with AND\n  const combinedMultiSource =\n    multiSource.length > 0 ? combineWithAnd(multiSource) : undefined\n\n  return {\n    singleSource: combinedSingleSource,\n    multiSource: combinedMultiSource,\n  }\n}\n\n/**\n * Step 4: Apply optimizations by lifting single-source clauses into subqueries.\n *\n * Creates a new QueryIR with single-source WHERE clauses moved to subqueries\n * that wrap the original table references. This ensures immutability and prevents\n * infinite recursion issues.\n *\n * @param query - Original QueryIR to optimize\n * @param groupedClauses - WHERE clauses grouped by optimization strategy\n * @returns New QueryIR with optimizations applied\n */\nfunction applyOptimizations(\n  query: QueryIR,\n  groupedClauses: GroupedWhereClauses\n): QueryIR {\n  // Track which single-source clauses were actually optimized\n  const actuallyOptimized = new Set<string>()\n\n  // Optimize the main FROM clause and track what was optimized\n  const optimizedFrom = optimizeFromWithTracking(\n    query.from,\n    groupedClauses.singleSource,\n    actuallyOptimized\n  )\n\n  // Optimize JOIN clauses and track what was optimized\n  const optimizedJoins = query.join\n    ? query.join.map((joinClause) => ({\n        ...joinClause,\n        from: optimizeFromWithTracking(\n          joinClause.from,\n          groupedClauses.singleSource,\n          actuallyOptimized\n        ),\n      }))\n    : undefined\n\n  // Build the remaining WHERE clauses: multi-source + any single-source that weren't optimized\n  const remainingWhereClauses: Array<BasicExpression<boolean>> = []\n\n  // Add multi-source clauses\n  if (groupedClauses.multiSource) {\n    remainingWhereClauses.push(groupedClauses.multiSource)\n  }\n\n  // Add single-source clauses that weren't actually optimized\n  for (const [source, clause] of groupedClauses.singleSource) {\n    if (!actuallyOptimized.has(source)) {\n      remainingWhereClauses.push(clause)\n    }\n  }\n\n  // Create a completely new query object to ensure immutability\n  const optimizedQuery: QueryIR = {\n    // Copy all non-optimized fields as-is\n    select: query.select,\n    groupBy: query.groupBy ? [...query.groupBy] : undefined,\n    having: query.having ? [...query.having] : undefined,\n    orderBy: query.orderBy ? [...query.orderBy] : undefined,\n    limit: query.limit,\n    offset: query.offset,\n    fnSelect: query.fnSelect,\n    fnWhere: query.fnWhere ? [...query.fnWhere] : undefined,\n    fnHaving: query.fnHaving ? [...query.fnHaving] : undefined,\n\n    // Use the optimized FROM and JOIN clauses\n    from: optimizedFrom,\n    join: optimizedJoins,\n\n    // Only include WHERE clauses that weren't successfully optimized\n    where: remainingWhereClauses.length > 0 ? remainingWhereClauses : [],\n  }\n\n  return optimizedQuery\n}\n\n/**\n * Helper function to create a deep copy of a QueryIR object for immutability.\n *\n * This ensures that all optimizations create new objects rather than modifying\n * existing ones, preventing infinite recursion and shared reference issues.\n *\n * @param query - QueryIR to deep copy\n * @returns New QueryIR object with all nested objects copied\n */\nfunction deepCopyQuery(query: QueryIR): QueryIR {\n  return {\n    // Recursively copy the FROM clause\n    from:\n      query.from.type === `collectionRef`\n        ? new CollectionRefClass(query.from.collection, query.from.alias)\n        : new QueryRefClass(deepCopyQuery(query.from.query), query.from.alias),\n\n    // Copy all other fields, creating new arrays where necessary\n    select: query.select,\n    join: query.join\n      ? query.join.map((joinClause) => ({\n          type: joinClause.type,\n          left: joinClause.left,\n          right: joinClause.right,\n          from:\n            joinClause.from.type === `collectionRef`\n              ? new CollectionRefClass(\n                  joinClause.from.collection,\n                  joinClause.from.alias\n                )\n              : new QueryRefClass(\n                  deepCopyQuery(joinClause.from.query),\n                  joinClause.from.alias\n                ),\n        }))\n      : undefined,\n    where: query.where ? [...query.where] : undefined,\n    groupBy: query.groupBy ? [...query.groupBy] : undefined,\n    having: query.having ? [...query.having] : undefined,\n    orderBy: query.orderBy ? [...query.orderBy] : undefined,\n    limit: query.limit,\n    offset: query.offset,\n    fnSelect: query.fnSelect,\n    fnWhere: query.fnWhere ? [...query.fnWhere] : undefined,\n    fnHaving: query.fnHaving ? [...query.fnHaving] : undefined,\n  }\n}\n\n/**\n * Helper function to optimize a FROM clause while tracking what was actually optimized.\n *\n * @param from - FROM clause to optimize\n * @param singleSourceClauses - Map of source aliases to their WHERE clauses\n * @param actuallyOptimized - Set to track which sources were actually optimized\n * @returns New FROM clause, potentially wrapped in a subquery\n */\nfunction optimizeFromWithTracking(\n  from: From,\n  singleSourceClauses: Map<string, BasicExpression<boolean>>,\n  actuallyOptimized: Set<string>\n): From {\n  const whereClause = singleSourceClauses.get(from.alias)\n\n  if (!whereClause) {\n    // No optimization needed, but return a copy to maintain immutability\n    if (from.type === `collectionRef`) {\n      return new CollectionRefClass(from.collection, from.alias)\n    }\n    // Must be queryRef due to type system\n    return new QueryRefClass(deepCopyQuery(from.query), from.alias)\n  }\n\n  if (from.type === `collectionRef`) {\n    // Create a new subquery with the WHERE clause for the collection\n    // This is always safe since we're creating a new subquery\n    const subQuery: QueryIR = {\n      from: new CollectionRefClass(from.collection, from.alias),\n      where: [whereClause],\n    }\n    actuallyOptimized.add(from.alias) // Mark as successfully optimized\n    return new QueryRefClass(subQuery, from.alias)\n  }\n\n  // Must be queryRef due to type system\n\n  // SAFETY CHECK: Only check safety when pushing WHERE clauses into existing subqueries\n  // We need to be careful about pushing WHERE clauses into subqueries that already have\n  // aggregates, HAVING, or ORDER BY + LIMIT since that could change their semantics\n  if (!isSafeToPushIntoExistingSubquery(from.query)) {\n    // Return a copy without optimization to maintain immutability\n    // Do NOT mark as optimized since we didn't actually optimize it\n    return new QueryRefClass(deepCopyQuery(from.query), from.alias)\n  }\n\n  // Add the WHERE clause to the existing subquery\n  // Create a deep copy to ensure immutability\n  const existingWhere = from.query.where || []\n  const optimizedSubQuery: QueryIR = {\n    ...deepCopyQuery(from.query),\n    where: [...existingWhere, whereClause],\n  }\n  actuallyOptimized.add(from.alias) // Mark as successfully optimized\n  return new QueryRefClass(optimizedSubQuery, from.alias)\n}\n\n/**\n * Determines if it's safe to push WHERE clauses into an existing subquery.\n *\n * Pushing WHERE clauses into existing subqueries can break semantics in several cases:\n *\n * 1. **Aggregates**: Pushing predicates before GROUP BY changes what gets aggregated\n * 2. **ORDER BY + LIMIT/OFFSET**: Pushing predicates before sorting+limiting changes the result set\n * 3. **HAVING clauses**: These operate on aggregated data, predicates should not be pushed past them\n * 4. **Functional operations**: fnSelect, fnWhere, fnHaving could have side effects\n *\n * Note: This safety check only applies when pushing WHERE clauses into existing subqueries.\n * Creating new subqueries from collection references is always safe.\n *\n * @param query - The existing subquery to check for safety\n * @returns True if it's safe to push WHERE clauses into this subquery, false otherwise\n *\n * @example\n * ```typescript\n * // UNSAFE: has GROUP BY - pushing WHERE could change aggregation\n * { from: users, groupBy: [dept], select: { count: agg('count', '*') } }\n *\n * // UNSAFE: has ORDER BY + LIMIT - pushing WHERE could change \"top 10\"\n * { from: users, orderBy: [salary desc], limit: 10 }\n *\n * // SAFE: plain SELECT without aggregates/limits\n * { from: users, select: { id, name } }\n * ```\n */\nfunction isSafeToPushIntoExistingSubquery(query: QueryIR): boolean {\n  // Check for aggregates in SELECT clause\n  if (query.select) {\n    const hasAggregates = Object.values(query.select).some(\n      (expr) => expr.type === `agg`\n    )\n    if (hasAggregates) {\n      return false\n    }\n  }\n\n  // Check for GROUP BY clause\n  if (query.groupBy && query.groupBy.length > 0) {\n    return false\n  }\n\n  // Check for HAVING clause\n  if (query.having && query.having.length > 0) {\n    return false\n  }\n\n  // Check for ORDER BY with LIMIT or OFFSET (dangerous combination)\n  if (query.orderBy && query.orderBy.length > 0) {\n    if (query.limit !== undefined || query.offset !== undefined) {\n      return false\n    }\n  }\n\n  // Check for functional variants that might have side effects\n  if (\n    query.fnSelect ||\n    (query.fnWhere && query.fnWhere.length > 0) ||\n    (query.fnHaving && query.fnHaving.length > 0)\n  ) {\n    return false\n  }\n\n  // If none of the unsafe conditions are present, it's safe to optimize\n  return true\n}\n\n/**\n * Helper function to combine multiple expressions with AND.\n *\n * If there's only one expression, it's returned as-is.\n * If there are multiple expressions, they're combined with an AND function.\n *\n * @param expressions - Array of expressions to combine\n * @returns Single expression representing the AND combination\n * @throws Error if the expressions array is empty\n */\nfunction combineWithAnd(\n  expressions: Array<BasicExpression<boolean>>\n): BasicExpression<boolean> {\n  if (expressions.length === 0) {\n    throw new CannotCombineEmptyExpressionListError()\n  }\n\n  if (expressions.length === 1) {\n    return expressions[0]!\n  }\n\n  // Create an AND function with all expressions as arguments\n  return new Func(`and`, expressions)\n}\n"],"names":["QueryRefClass","CollectionRefClass"],"mappings":";;;;AAoLO,SAAS,cAAc,OAAoC;AAEhE,QAAM,yBAAyB,8BAA8B,KAAK;AAGlE,MAAI,YAAY;AAChB,MAAI;AACJ,MAAI,aAAa;AACjB,QAAM,gBAAgB;AAGtB,SACE,aAAa,iBACb,CAAC,WAAW,WAAW,iBAAiB,GACxC;AACA,wBAAoB;AACpB,gBAAY,2BAA2B,SAAS;AAChD;AAAA,EACF;AAGA,QAAM,UAAU,0BAA0B,SAAS;AAEnD,SAAO;AAAA,IACL,gBAAgB;AAAA,IAChB;AAAA,EAAA;AAEJ;AAUA,SAAS,8BACP,OACuC;AACvC,QAAM,6CAA6B,IAAA;AAGnC,MAAI,CAAC,MAAM,SAAS,MAAM,MAAM,WAAW,GAAG;AAC5C,WAAO;AAAA,EACT;AAGA,QAAM,oBAAoB,gBAAgB,MAAM,KAAK;AAGrD,QAAM,kBAAkB,kBAAkB;AAAA,IAAI,CAAC,WAC7C,mBAAmB,MAAM;AAAA,EAAA;AAI3B,QAAM,iBAAiB,kBAAkB,eAAe;AAIxD,aAAW,CAAC,aAAa,WAAW,KAAK,eAAe,cAAc;AAEpE,QAAI,sBAAsB,OAAO,WAAW,GAAG;AAE7C,UAAI,gCAAgC,WAAW,GAAG;AAChD,+BAAuB,IAAI,aAAa,WAAW;AAAA,MACrD;AAAA,IACF;AAAA,EACF;AAEA,SAAO;AACT;AAUA,SAAS,sBAAsB,OAAgB,aAA8B;AAE3E,MAAI,MAAM,KAAK,UAAU,aAAa;AACpC,WAAO,MAAM,KAAK,SAAS;AAAA,EAC7B;AAGA,MAAI,MAAM,MAAM;AACd,eAAW,cAAc,MAAM,MAAM;AACnC,UAAI,WAAW,KAAK,UAAU,aAAa;AACzC,eAAO,WAAW,KAAK,SAAS;AAAA,MAClC;AAAA,IACF;AAAA,EACF;AAEA,SAAO;AACT;AAQA,SAAS,2BAA2B,OAAyB;;AAE3D,QAAM,sBAAsB;AAAA,IAC1B,GAAG;AAAA,IACH,MACE,MAAM,KAAK,SAAS,aAChB,IAAIA;AAAAA,MACF,2BAA2B,MAAM,KAAK,KAAK;AAAA,MAC3C,MAAM,KAAK;AAAA,IAAA,IAEb,MAAM;AAAA,IACZ,OAAM,WAAM,SAAN,mBAAY,IAAI,CAAC,gBAAgB;AAAA,MACrC,GAAG;AAAA,MACH,MACE,WAAW,KAAK,SAAS,aACrB,IAAIA;AAAAA,QACF,2BAA2B,WAAW,KAAK,KAAK;AAAA,QAChD,WAAW,KAAK;AAAA,MAAA,IAElB,WAAW;AAAA,IAAA;AAAA,EACjB;AAIJ,SAAO,6BAA6B,mBAAmB;AACzD;AAKA,SAAS,6BAA6B,OAAyB;AAE7D,MAAI,CAAC,MAAM,SAAS,MAAM,MAAM,WAAW,GAAG;AAC5C,WAAO;AAAA,EACT;AAIA,MAAI,CAAC,MAAM,QAAQ,MAAM,KAAK,WAAW,GAAG;AAC1C,WAAO;AAAA,EACT;AAGA,QAAM,oBAAoB,gBAAgB,MAAM,KAAK;AAGrD,QAAM,kBAAkB,kBAAkB;AAAA,IAAI,CAAC,WAC7C,mBAAmB,MAAM;AAAA,EAAA;AAI3B,QAAM,iBAAiB,kBAAkB,eAAe;AAGxD,SAAO,mBAAmB,OAAO,cAAc;AACjD;AAUA,SAAS,0BAA0B,OAAyB;;AAC1D,SAAO;AAAA,IACL,GAAG;AAAA,IACH,MAAM,0BAA0B,MAAM,IAAI;AAAA,IAC1C,OAAM,WAAM,SAAN,mBAAY,IAAI,CAAC,gBAAgB;AAAA,MACrC,GAAG;AAAA,MACH,MAAM,0BAA0B,WAAW,IAAI;AAAA,IAAA;AAAA,EAC/C;AAEN;AAQA,SAAS,0BAA0B,MAAkB;AACnD,MAAI,KAAK,SAAS,iBAAiB;AACjC,WAAO;AAAA,EACT;AAEA,QAAM,iBAAiB,0BAA0B,KAAK,KAAK;AAG3D,MAAI,oBAAoB,cAAc,GAAG;AAEvC,UAAM,YAAY,0BAA0B,eAAe,IAAI;AAC/D,QAAI,UAAU,SAAS,iBAAiB;AACtC,aAAO,IAAIC,cAAmB,UAAU,YAAY,KAAK,KAAK;AAAA,IAChE,OAAO;AACL,aAAO,IAAID,SAAc,UAAU,OAAO,KAAK,KAAK;AAAA,IACtD;AAAA,EACF;AAEA,SAAO,IAAIA,SAAc,gBAAgB,KAAK,KAAK;AACrD;AAQA,SAAS,oBAAoB,OAAyB;AACpD,UACG,CAAC,MAAM,SAAS,MAAM,MAAM,WAAW,MACxC,CAAC,MAAM,WACN,CAAC,MAAM,WAAW,MAAM,QAAQ,WAAW,OAC3C,CAAC,MAAM,UAAU,MAAM,OAAO,WAAW,OACzC,CAAC,MAAM,WAAW,MAAM,QAAQ,WAAW,OAC3C,CAAC,MAAM,QAAQ,MAAM,KAAK,WAAW,MACtC,MAAM,UAAU,UAChB,MAAM,WAAW,UACjB,CAAC,MAAM,aACN,CAAC,MAAM,WAAW,MAAM,QAAQ,WAAW,OAC3C,CAAC,MAAM,YAAY,MAAM,SAAS,WAAW;AAElD;AAiBA,SAAS,gBACP,cACiC;AACjC,QAAM,SAA0C,CAAA;AAEhD,aAAW,UAAU,cAAc;AACjC,QAAI,OAAO,SAAS,UAAU,OAAO,SAAS,OAAO;AAEnD,YAAM,YAAY;AAAA,QAChB,OAAO;AAAA,MAAA;AAET,aAAO,KAAK,GAAG,SAAS;AAAA,IAC1B,OAAO;AAEL,aAAO,KAAK,MAAM;AAAA,IACpB;AAAA,EACF;AAEA,SAAO;AACT;AAiBA,SAAS,mBACP,QACqB;AACrB,QAAM,qCAAqB,IAAA;AAK3B,WAAS,eAAe,MAAmC;AACzD,YAAQ,KAAK,MAAA;AAAA,MACX,KAAK;AAEH,YAAI,KAAK,QAAQ,KAAK,KAAK,SAAS,GAAG;AACrC,gBAAM,eAAe,KAAK,KAAK,CAAC;AAChC,cAAI,cAAc;AAChB,2BAAe,IAAI,YAAY;AAAA,UACjC;AAAA,QACF;AACA;AAAA,MACF,KAAK;AAEH,YAAI,KAAK,MAAM;AACb,eAAK,KAAK,QAAQ,cAAc;AAAA,QAClC;AACA;AAAA,MACF,KAAK;AAEH;AAAA,MACF,KAAK;AAEH,YAAI,KAAK,MAAM;AACb,eAAK,KAAK,QAAQ,cAAc;AAAA,QAClC;AACA;AAAA,IAAA;AAAA,EAEN;AAEA,iBAAe,MAAM;AAErB,SAAO;AAAA,IACL,YAAY;AAAA,IACZ;AAAA,EAAA;AAEJ;AAWA,SAAS,kBACP,iBACqB;AACrB,QAAM,mCAAmB,IAAA;AACzB,QAAM,cAA+C,CAAA;AAGrD,aAAW,UAAU,iBAAiB;AACpC,QAAI,OAAO,eAAe,SAAS,GAAG;AAEpC,YAAM,SAAS,MAAM,KAAK,OAAO,cAAc,EAAE,CAAC;AAClD,UAAI,CAAC,aAAa,IAAI,MAAM,GAAG;AAC7B,qBAAa,IAAI,QAAQ,EAAE;AAAA,MAC7B;AACA,mBAAa,IAAI,MAAM,EAAG,KAAK,OAAO,UAAU;AAAA,IAClD,WAAW,OAAO,eAAe,OAAO,GAAG;AAEzC,kBAAY,KAAK,OAAO,UAAU;AAAA,IACpC;AAAA,EAEF;AAGA,QAAM,2CAA2B,IAAA;AACjC,aAAW,CAAC,QAAQ,OAAO,KAAK,cAAc;AAC5C,yBAAqB,IAAI,QAAQ,eAAe,OAAO,CAAC;AAAA,EAC1D;AAGA,QAAM,sBACJ,YAAY,SAAS,IAAI,eAAe,WAAW,IAAI;AAEzD,SAAO;AAAA,IACL,cAAc;AAAA,IACd,aAAa;AAAA,EAAA;AAEjB;AAaA,SAAS,mBACP,OACA,gBACS;AAET,QAAM,wCAAwB,IAAA;AAG9B,QAAM,gBAAgB;AAAA,IACpB,MAAM;AAAA,IACN,eAAe;AAAA,IACf;AAAA,EAAA;AAIF,QAAM,iBAAiB,MAAM,OACzB,MAAM,KAAK,IAAI,CAAC,gBAAgB;AAAA,IAC9B,GAAG;AAAA,IACH,MAAM;AAAA,MACJ,WAAW;AAAA,MACX,eAAe;AAAA,MACf;AAAA,IAAA;AAAA,EACF,EACA,IACF;AAGJ,QAAM,wBAAyD,CAAA;AAG/D,MAAI,eAAe,aAAa;AAC9B,0BAAsB,KAAK,eAAe,WAAW;AAAA,EACvD;AAGA,aAAW,CAAC,QAAQ,MAAM,KAAK,eAAe,cAAc;AAC1D,QAAI,CAAC,kBAAkB,IAAI,MAAM,GAAG;AAClC,4BAAsB,KAAK,MAAM;AAAA,IACnC;AAAA,EACF;AAGA,QAAM,iBAA0B;AAAA;AAAA,IAE9B,QAAQ,MAAM;AAAA,IACd,SAAS,MAAM,UAAU,CAAC,GAAG,MAAM,OAAO,IAAI;AAAA,IAC9C,QAAQ,MAAM,SAAS,CAAC,GAAG,MAAM,MAAM,IAAI;AAAA,IAC3C,SAAS,MAAM,UAAU,CAAC,GAAG,MAAM,OAAO,IAAI;AAAA,IAC9C,OAAO,MAAM;AAAA,IACb,QAAQ,MAAM;AAAA,IACd,UAAU,MAAM;AAAA,IAChB,SAAS,MAAM,UAAU,CAAC,GAAG,MAAM,OAAO,IAAI;AAAA,IAC9C,UAAU,MAAM,WAAW,CAAC,GAAG,MAAM,QAAQ,IAAI;AAAA;AAAA,IAGjD,MAAM;AAAA,IACN,MAAM;AAAA;AAAA,IAGN,OAAO,sBAAsB,SAAS,IAAI,wBAAwB,CAAA;AAAA,EAAC;AAGrE,SAAO;AACT;AAWA,SAAS,cAAc,OAAyB;AAC9C,SAAO;AAAA;AAAA,IAEL,MACE,MAAM,KAAK,SAAS,kBAChB,IAAIC,cAAmB,MAAM,KAAK,YAAY,MAAM,KAAK,KAAK,IAC9D,IAAID,SAAc,cAAc,MAAM,KAAK,KAAK,GAAG,MAAM,KAAK,KAAK;AAAA;AAAA,IAGzE,QAAQ,MAAM;AAAA,IACd,MAAM,MAAM,OACR,MAAM,KAAK,IAAI,CAAC,gBAAgB;AAAA,MAC9B,MAAM,WAAW;AAAA,MACjB,MAAM,WAAW;AAAA,MACjB,OAAO,WAAW;AAAA,MAClB,MACE,WAAW,KAAK,SAAS,kBACrB,IAAIC;AAAAA,QACF,WAAW,KAAK;AAAA,QAChB,WAAW,KAAK;AAAA,MAAA,IAElB,IAAID;AAAAA,QACF,cAAc,WAAW,KAAK,KAAK;AAAA,QACnC,WAAW,KAAK;AAAA,MAAA;AAAA,IAClB,EACN,IACF;AAAA,IACJ,OAAO,MAAM,QAAQ,CAAC,GAAG,MAAM,KAAK,IAAI;AAAA,IACxC,SAAS,MAAM,UAAU,CAAC,GAAG,MAAM,OAAO,IAAI;AAAA,IAC9C,QAAQ,MAAM,SAAS,CAAC,GAAG,MAAM,MAAM,IAAI;AAAA,IAC3C,SAAS,MAAM,UAAU,CAAC,GAAG,MAAM,OAAO,IAAI;AAAA,IAC9C,OAAO,MAAM;AAAA,IACb,QAAQ,MAAM;AAAA,IACd,UAAU,MAAM;AAAA,IAChB,SAAS,MAAM,UAAU,CAAC,GAAG,MAAM,OAAO,IAAI;AAAA,IAC9C,UAAU,MAAM,WAAW,CAAC,GAAG,MAAM,QAAQ,IAAI;AAAA,EAAA;AAErD;AAUA,SAAS,yBACP,MACA,qBACA,mBACM;AACN,QAAM,cAAc,oBAAoB,IAAI,KAAK,KAAK;AAEtD,MAAI,CAAC,aAAa;AAEhB,QAAI,KAAK,SAAS,iBAAiB;AACjC,aAAO,IAAIC,cAAmB,KAAK,YAAY,KAAK,KAAK;AAAA,IAC3D;AAEA,WAAO,IAAID,SAAc,cAAc,KAAK,KAAK,GAAG,KAAK,KAAK;AAAA,EAChE;AAEA,MAAI,KAAK,SAAS,iBAAiB;AAGjC,UAAM,WAAoB;AAAA,MACxB,MAAM,IAAIC,cAAmB,KAAK,YAAY,KAAK,KAAK;AAAA,MACxD,OAAO,CAAC,WAAW;AAAA,IAAA;AAErB,sBAAkB,IAAI,KAAK,KAAK;AAChC,WAAO,IAAID,SAAc,UAAU,KAAK,KAAK;AAAA,EAC/C;AAOA,MAAI,CAAC,iCAAiC,KAAK,KAAK,GAAG;AAGjD,WAAO,IAAIA,SAAc,cAAc,KAAK,KAAK,GAAG,KAAK,KAAK;AAAA,EAChE;AAIA,QAAM,gBAAgB,KAAK,MAAM,SAAS,CAAA;AAC1C,QAAM,oBAA6B;AAAA,IACjC,GAAG,cAAc,KAAK,KAAK;AAAA,IAC3B,OAAO,CAAC,GAAG,eAAe,WAAW;AAAA,EAAA;AAEvC,oBAAkB,IAAI,KAAK,KAAK;AAChC,SAAO,IAAIA,SAAc,mBAAmB,KAAK,KAAK;AACxD;AA8BA,SAAS,iCAAiC,OAAyB;AAEjE,MAAI,MAAM,QAAQ;AAChB,UAAM,gBAAgB,OAAO,OAAO,MAAM,MAAM,EAAE;AAAA,MAChD,CAAC,SAAS,KAAK,SAAS;AAAA,IAAA;AAE1B,QAAI,eAAe;AACjB,aAAO;AAAA,IACT;AAAA,EACF;AAGA,MAAI,MAAM,WAAW,MAAM,QAAQ,SAAS,GAAG;AAC7C,WAAO;AAAA,EACT;AAGA,MAAI,MAAM,UAAU,MAAM,OAAO,SAAS,GAAG;AAC3C,WAAO;AAAA,EACT;AAGA,MAAI,MAAM,WAAW,MAAM,QAAQ,SAAS,GAAG;AAC7C,QAAI,MAAM,UAAU,UAAa,MAAM,WAAW,QAAW;AAC3D,aAAO;AAAA,IACT;AAAA,EACF;AAGA,MACE,MAAM,YACL,MAAM,WAAW,MAAM,QAAQ,SAAS,KACxC,MAAM,YAAY,MAAM,SAAS,SAAS,GAC3C;AACA,WAAO;AAAA,EACT;AAGA,SAAO;AACT;AAYA,SAAS,eACP,aAC0B;AAC1B,MAAI,YAAY,WAAW,GAAG;AAC5B,UAAM,IAAI,sCAAA;AAAA,EACZ;AAEA,MAAI,YAAY,WAAW,GAAG;AAC5B,WAAO,YAAY,CAAC;AAAA,EACtB;AAGA,SAAO,IAAI,KAAK,OAAO,WAAW;AACpC;"}
+{"version":3,"file":"optimizer.js","sources":["../../../src/query/optimizer.ts"],"sourcesContent":["/**\n * # Query Optimizer\n *\n * The query optimizer improves query performance by implementing predicate pushdown optimization.\n * It rewrites the intermediate representation (IR) to push WHERE clauses as close to the data\n * source as possible, reducing the amount of data processed during joins.\n *\n * ## How It Works\n *\n * The optimizer follows a 4-step process:\n *\n * ### 1. AND Clause Splitting\n * Splits AND clauses at the root level into separate WHERE clauses for granular optimization.\n * ```javascript\n * // Before: WHERE and(eq(users.department_id, 1), gt(users.age, 25))\n * // After:  WHERE eq(users.department_id, 1) + WHERE gt(users.age, 25)\n * ```\n *\n * ### 2. Source Analysis\n * Analyzes each WHERE clause to determine which table sources it references:\n * - Single-source clauses: Touch only one table (e.g., `users.department_id = 1`)\n * - Multi-source clauses: Touch multiple tables (e.g., `users.id = posts.user_id`)\n *\n * ### 3. Clause Grouping\n * Groups WHERE clauses by the sources they touch:\n * - Single-source clauses are grouped by their respective table\n * - Multi-source clauses are combined for the main query\n *\n * ### 4. Subquery Creation\n * Lifts single-source WHERE clauses into subqueries that wrap the original table references.\n *\n * ## Safety & Edge Cases\n *\n * The optimizer includes targeted safety checks to prevent predicate pushdown when it could\n * break query semantics:\n *\n * ### Always Safe Operations\n * - **Creating new subqueries**: Wrapping collection references in subqueries with WHERE clauses\n * - **Main query optimizations**: Moving single-source WHERE clauses from main query to subqueries\n * - **Queries with aggregates/ORDER BY/HAVING**: Can still create new filtered subqueries\n *\n * ### Unsafe Operations (blocked by safety checks)\n * Pushing WHERE clauses **into existing subqueries** that have:\n * - **Aggregates**: GROUP BY, HAVING, or aggregate functions in SELECT (would change aggregation)\n * - **Ordering + Limits**: ORDER BY combined with LIMIT/OFFSET (would change result set)\n * - **Functional Operations**: fnSelect, fnWhere, fnHaving (potential side effects)\n *\n * ### Residual WHERE Clauses\n * For outer joins (LEFT, RIGHT, FULL), WHERE clauses are copied to subqueries for optimization\n * but also kept as \"residual\" clauses in the main query to preserve semantics. This ensures\n * that NULL values from outer joins are properly filtered according to SQL standards.\n *\n * The optimizer tracks which clauses were actually optimized and only removes those from the\n * main query. Subquery reuse is handled safely through immutable query copies.\n *\n * ## Example Optimizations\n *\n * ### Basic Query with Joins\n * **Original Query:**\n * ```javascript\n * query\n *   .from({ users: usersCollection })\n *   .join({ posts: postsCollection }, ({users, posts}) => eq(users.id, posts.user_id))\n *   .where(({users}) => eq(users.department_id, 1))\n *   .where(({posts}) => gt(posts.views, 100))\n *   .where(({users, posts}) => eq(users.id, posts.author_id))\n * ```\n *\n * **Optimized Query:**\n * ```javascript\n * query\n *   .from({\n *     users: subquery\n *       .from({ users: usersCollection })\n *       .where(({users}) => eq(users.department_id, 1))\n *   })\n *   .join({\n *     posts: subquery\n *       .from({ posts: postsCollection })\n *       .where(({posts}) => gt(posts.views, 100))\n *   }, ({users, posts}) => eq(users.id, posts.user_id))\n *   .where(({users, posts}) => eq(users.id, posts.author_id))\n * ```\n *\n * ### Query with Aggregates (Now Optimizable!)\n * **Original Query:**\n * ```javascript\n * query\n *   .from({ users: usersCollection })\n *   .join({ posts: postsCollection }, ({users, posts}) => eq(users.id, posts.user_id))\n *   .where(({users}) => eq(users.department_id, 1))\n *   .groupBy(['users.department_id'])\n *   .select({ count: agg('count', '*') })\n * ```\n *\n * **Optimized Query:**\n * ```javascript\n * query\n *   .from({\n *     users: subquery\n *       .from({ users: usersCollection })\n *       .where(({users}) => eq(users.department_id, 1))\n *   })\n *   .join({ posts: postsCollection }, ({users, posts}) => eq(users.id, posts.user_id))\n *   .groupBy(['users.department_id'])\n *   .select({ count: agg('count', '*') })\n * ```\n *\n * ## Benefits\n *\n * - **Reduced Data Processing**: Filters applied before joins reduce intermediate result size\n * - **Better Performance**: Smaller datasets lead to faster query execution\n * - **Automatic Optimization**: No manual query rewriting required\n * - **Preserves Semantics**: Optimized queries return identical results\n * - **Safe by Design**: Comprehensive checks prevent semantic-breaking optimizations\n *\n * ## Integration\n *\n * The optimizer is automatically called during query compilation before the IR is\n * transformed into a D2Mini pipeline.\n */\n\nimport { deepEquals } from \"../utils.js\"\nimport { CannotCombineEmptyExpressionListError } from \"../errors.js\"\nimport {\n  CollectionRef as CollectionRefClass,\n  Func,\n  QueryRef as QueryRefClass,\n  createResidualWhere,\n  getWhereExpression,\n  isResidualWhere,\n} from \"./ir.js\"\nimport { isConvertibleToCollectionFilter } from \"./compiler/expressions.js\"\nimport type { BasicExpression, From, QueryIR, Where } from \"./ir.js\"\n\n/**\n * Represents a WHERE clause after source analysis\n */\nexport interface AnalyzedWhereClause {\n  /** The WHERE expression */\n  expression: BasicExpression<boolean>\n  /** Set of table/source aliases that this WHERE clause touches */\n  touchedSources: Set<string>\n}\n\n/**\n * Represents WHERE clauses grouped by the sources they touch\n */\nexport interface GroupedWhereClauses {\n  /** WHERE clauses that touch only a single source, grouped by source alias */\n  singleSource: Map<string, BasicExpression<boolean>>\n  /** WHERE clauses that touch multiple sources, combined into one expression */\n  multiSource?: BasicExpression<boolean>\n}\n\n/**\n * Result of query optimization including both the optimized query and collection-specific WHERE clauses\n */\nexport interface OptimizationResult {\n  /** The optimized query with WHERE clauses potentially moved to subqueries */\n  optimizedQuery: QueryIR\n  /** Map of collection aliases to their extracted WHERE clauses for index optimization */\n  collectionWhereClauses: Map<string, BasicExpression<boolean>>\n}\n\n/**\n * Main query optimizer entry point that lifts WHERE clauses into subqueries.\n *\n * This function implements multi-level predicate pushdown optimization by recursively\n * moving WHERE clauses through nested subqueries to get them as close to the data\n * sources as possible, then removing redundant subqueries.\n *\n * @param query - The QueryIR to optimize\n * @returns An OptimizationResult with the optimized query and collection WHERE clause mapping\n *\n * @example\n * ```typescript\n * const originalQuery = {\n *   from: new CollectionRef(users, 'u'),\n *   join: [{ from: new CollectionRef(posts, 'p'), ... }],\n *   where: [eq(u.dept_id, 1), gt(p.views, 100)]\n * }\n *\n * const { optimizedQuery, collectionWhereClauses } = optimizeQuery(originalQuery)\n * // Result: Single-source clauses moved to deepest possible subqueries\n * // collectionWhereClauses: Map { 'u' => eq(u.dept_id, 1), 'p' => gt(p.views, 100) }\n * ```\n */\nexport function optimizeQuery(query: QueryIR): OptimizationResult {\n  // First, extract collection WHERE clauses before optimization\n  const collectionWhereClauses = extractCollectionWhereClauses(query)\n\n  // Apply multi-level predicate pushdown with iterative convergence\n  let optimized = query\n  let previousOptimized: QueryIR | undefined\n  let iterations = 0\n  const maxIterations = 10 // Prevent infinite loops\n\n  // Keep optimizing until no more changes occur or max iterations reached\n  while (\n    iterations < maxIterations &&\n    !deepEquals(optimized, previousOptimized)\n  ) {\n    previousOptimized = optimized\n    optimized = applyRecursiveOptimization(optimized)\n    iterations++\n  }\n\n  // Remove redundant subqueries\n  const cleaned = removeRedundantSubqueries(optimized)\n\n  return {\n    optimizedQuery: cleaned,\n    collectionWhereClauses,\n  }\n}\n\n/**\n * Extracts collection-specific WHERE clauses from a query for index optimization.\n * This analyzes the original query to identify WHERE clauses that can be pushed down\n * to specific collections, but only for simple queries without joins.\n *\n * @param query - The original QueryIR to analyze\n * @returns Map of collection aliases to their WHERE clauses\n */\nfunction extractCollectionWhereClauses(\n  query: QueryIR\n): Map<string, BasicExpression<boolean>> {\n  const collectionWhereClauses = new Map<string, BasicExpression<boolean>>()\n\n  // Only analyze queries that have WHERE clauses\n  if (!query.where || query.where.length === 0) {\n    return collectionWhereClauses\n  }\n\n  // Split all AND clauses at the root level for granular analysis\n  const splitWhereClauses = splitAndClauses(query.where)\n\n  // Analyze each WHERE clause to determine which sources it touches\n  const analyzedClauses = splitWhereClauses.map((clause) =>\n    analyzeWhereClause(clause)\n  )\n\n  // Group clauses by single-source vs multi-source\n  const groupedClauses = groupWhereClauses(analyzedClauses)\n\n  // Only include single-source clauses that reference collections directly\n  // and can be converted to BasicExpression format for collection indexes\n  for (const [sourceAlias, whereClause] of groupedClauses.singleSource) {\n    // Check if this source alias corresponds to a collection reference\n    if (isCollectionReference(query, sourceAlias)) {\n      // Check if the WHERE clause can be converted to collection-compatible format\n      if (isConvertibleToCollectionFilter(whereClause)) {\n        collectionWhereClauses.set(sourceAlias, whereClause)\n      }\n    }\n  }\n\n  return collectionWhereClauses\n}\n\n/**\n * Determines if a source alias refers to a collection reference (not a subquery).\n * This is used to identify WHERE clauses that can be pushed down to collection subscriptions.\n *\n * @param query - The query to analyze\n * @param sourceAlias - The source alias to check\n * @returns True if the alias refers to a collection reference\n */\nfunction isCollectionReference(query: QueryIR, sourceAlias: string): boolean {\n  // Check the FROM clause\n  if (query.from.alias === sourceAlias) {\n    return query.from.type === `collectionRef`\n  }\n\n  // Check JOIN clauses\n  if (query.join) {\n    for (const joinClause of query.join) {\n      if (joinClause.from.alias === sourceAlias) {\n        return joinClause.from.type === `collectionRef`\n      }\n    }\n  }\n\n  return false\n}\n\n/**\n * Applies recursive predicate pushdown optimization.\n *\n * @param query - The QueryIR to optimize\n * @returns A new QueryIR with optimizations applied\n */\nfunction applyRecursiveOptimization(query: QueryIR): QueryIR {\n  // First, recursively optimize any existing subqueries\n  const subqueriesOptimized = {\n    ...query,\n    from:\n      query.from.type === `queryRef`\n        ? new QueryRefClass(\n            applyRecursiveOptimization(query.from.query),\n            query.from.alias\n          )\n        : query.from,\n    join: query.join?.map((joinClause) => ({\n      ...joinClause,\n      from:\n        joinClause.from.type === `queryRef`\n          ? new QueryRefClass(\n              applyRecursiveOptimization(joinClause.from.query),\n              joinClause.from.alias\n            )\n          : joinClause.from,\n    })),\n  }\n\n  // Then apply single-level optimization to this query\n  return applySingleLevelOptimization(subqueriesOptimized)\n}\n\n/**\n * Applies single-level predicate pushdown optimization (existing logic)\n */\nfunction applySingleLevelOptimization(query: QueryIR): QueryIR {\n  // Skip optimization if no WHERE clauses exist\n  if (!query.where || query.where.length === 0) {\n    return query\n  }\n\n  // Skip optimization if there are no joins - predicate pushdown only benefits joins\n  // Single-table queries don't benefit from this optimization\n  if (!query.join || query.join.length === 0) {\n    return query\n  }\n\n  // Filter out residual WHERE clauses to prevent them from being optimized again\n  const nonResidualWhereClauses = query.where.filter(\n    (where) => !isResidualWhere(where)\n  )\n\n  // Step 1: Split all AND clauses at the root level for granular optimization\n  const splitWhereClauses = splitAndClauses(nonResidualWhereClauses)\n\n  // Step 2: Analyze each WHERE clause to determine which sources it touches\n  const analyzedClauses = splitWhereClauses.map((clause) =>\n    analyzeWhereClause(clause)\n  )\n\n  // Step 3: Group clauses by single-source vs multi-source\n  const groupedClauses = groupWhereClauses(analyzedClauses)\n\n  // Step 4: Apply optimizations by lifting single-source clauses into subqueries\n  const optimizedQuery = applyOptimizations(query, groupedClauses)\n\n  // Add back any residual WHERE clauses that were filtered out\n  const residualWhereClauses = query.where.filter((where) =>\n    isResidualWhere(where)\n  )\n  if (residualWhereClauses.length > 0) {\n    optimizedQuery.where = [\n      ...(optimizedQuery.where || []),\n      ...residualWhereClauses,\n    ]\n  }\n\n  return optimizedQuery\n}\n\n/**\n * Removes redundant subqueries that don't add value.\n * A subquery is redundant if it only wraps another query without adding\n * WHERE, SELECT, GROUP BY, HAVING, ORDER BY, or LIMIT/OFFSET clauses.\n *\n * @param query - The QueryIR to process\n * @returns A new QueryIR with redundant subqueries removed\n */\nfunction removeRedundantSubqueries(query: QueryIR): QueryIR {\n  return {\n    ...query,\n    from: removeRedundantFromClause(query.from),\n    join: query.join?.map((joinClause) => ({\n      ...joinClause,\n      from: removeRedundantFromClause(joinClause.from),\n    })),\n  }\n}\n\n/**\n * Removes redundant subqueries from a FROM clause.\n *\n * @param from - The FROM clause to process\n * @returns A FROM clause with redundant subqueries removed\n */\nfunction removeRedundantFromClause(from: From): From {\n  if (from.type === `collectionRef`) {\n    return from\n  }\n\n  const processedQuery = removeRedundantSubqueries(from.query)\n\n  // Check if this subquery is redundant\n  if (isRedundantSubquery(processedQuery)) {\n    // Return the inner query's FROM clause with this alias\n    const innerFrom = removeRedundantFromClause(processedQuery.from)\n    if (innerFrom.type === `collectionRef`) {\n      return new CollectionRefClass(innerFrom.collection, from.alias)\n    } else {\n      return new QueryRefClass(innerFrom.query, from.alias)\n    }\n  }\n\n  return new QueryRefClass(processedQuery, from.alias)\n}\n\n/**\n * Determines if a subquery is redundant (adds no value).\n *\n * @param query - The query to check\n * @returns True if the query is redundant and can be removed\n */\nfunction isRedundantSubquery(query: QueryIR): boolean {\n  return (\n    (!query.where || query.where.length === 0) &&\n    !query.select &&\n    (!query.groupBy || query.groupBy.length === 0) &&\n    (!query.having || query.having.length === 0) &&\n    (!query.orderBy || query.orderBy.length === 0) &&\n    (!query.join || query.join.length === 0) &&\n    query.limit === undefined &&\n    query.offset === undefined &&\n    !query.fnSelect &&\n    (!query.fnWhere || query.fnWhere.length === 0) &&\n    (!query.fnHaving || query.fnHaving.length === 0)\n  )\n}\n\n/**\n * Step 1: Split all AND clauses recursively into separate WHERE clauses.\n *\n * This enables more granular optimization by treating each condition independently.\n * OR clauses are preserved as they cannot be split without changing query semantics.\n *\n * @param whereClauses - Array of WHERE expressions to split\n * @returns Flattened array with AND clauses split into separate expressions\n *\n * @example\n * ```typescript\n * // Input: [and(eq(a, 1), gt(b, 2)), eq(c, 3)]\n * // Output: [eq(a, 1), gt(b, 2), eq(c, 3)]\n * ```\n */\nfunction splitAndClauses(\n  whereClauses: Array<Where>\n): Array<BasicExpression<boolean>> {\n  const result: Array<BasicExpression<boolean>> = []\n\n  for (const whereClause of whereClauses) {\n    const clause = getWhereExpression(whereClause)\n    result.push(...splitAndClausesRecursive(clause))\n  }\n\n  return result\n}\n\n// Helper function for recursive splitting of BasicExpression arrays\nfunction splitAndClausesRecursive(\n  clause: BasicExpression<boolean>\n): Array<BasicExpression<boolean>> {\n  if (clause.type === `func` && clause.name === `and`) {\n    // Recursively split nested AND clauses to handle complex expressions\n    const result: Array<BasicExpression<boolean>> = []\n    for (const arg of clause.args as Array<BasicExpression<boolean>>) {\n      result.push(...splitAndClausesRecursive(arg))\n    }\n    return result\n  } else {\n    // Preserve non-AND clauses as-is (including OR clauses)\n    return [clause]\n  }\n}\n\n/**\n * Step 2: Analyze which table sources a WHERE clause touches.\n *\n * This determines whether a clause can be pushed down to a specific table\n * or must remain in the main query (for multi-source clauses like join conditions).\n *\n * @param clause - The WHERE expression to analyze\n * @returns Analysis result with the expression and touched source aliases\n *\n * @example\n * ```typescript\n * // eq(users.department_id, 1) -> touches ['users']\n * // eq(users.id, posts.user_id) -> touches ['users', 'posts']\n * ```\n */\nfunction analyzeWhereClause(\n  clause: BasicExpression<boolean>\n): AnalyzedWhereClause {\n  const touchedSources = new Set<string>()\n\n  /**\n   * Recursively collect all table aliases referenced in an expression\n   */\n  function collectSources(expr: BasicExpression | any): void {\n    switch (expr.type) {\n      case `ref`:\n        // PropRef path has the table alias as the first element\n        if (expr.path && expr.path.length > 0) {\n          const firstElement = expr.path[0]\n          if (firstElement) {\n            touchedSources.add(firstElement)\n          }\n        }\n        break\n      case `func`:\n        // Recursively analyze function arguments (e.g., eq, gt, and, or)\n        if (expr.args) {\n          expr.args.forEach(collectSources)\n        }\n        break\n      case `val`:\n        // Values don't reference any sources\n        break\n      case `agg`:\n        // Aggregates can reference sources in their arguments\n        if (expr.args) {\n          expr.args.forEach(collectSources)\n        }\n        break\n    }\n  }\n\n  collectSources(clause)\n\n  return {\n    expression: clause,\n    touchedSources,\n  }\n}\n\n/**\n * Step 3: Group WHERE clauses by the sources they touch.\n *\n * Single-source clauses can be pushed down to subqueries for optimization.\n * Multi-source clauses must remain in the main query to preserve join semantics.\n *\n * @param analyzedClauses - Array of analyzed WHERE clauses\n * @returns Grouped clauses ready for optimization\n */\nfunction groupWhereClauses(\n  analyzedClauses: Array<AnalyzedWhereClause>\n): GroupedWhereClauses {\n  const singleSource = new Map<string, Array<BasicExpression<boolean>>>()\n  const multiSource: Array<BasicExpression<boolean>> = []\n\n  // Categorize each clause based on how many sources it touches\n  for (const clause of analyzedClauses) {\n    if (clause.touchedSources.size === 1) {\n      // Single source clause - can be optimized\n      const source = Array.from(clause.touchedSources)[0]!\n      if (!singleSource.has(source)) {\n        singleSource.set(source, [])\n      }\n      singleSource.get(source)!.push(clause.expression)\n    } else if (clause.touchedSources.size > 1) {\n      // Multi-source clause - must stay in main query\n      multiSource.push(clause.expression)\n    }\n    // Skip clauses that touch no sources (constants) - they don't need optimization\n  }\n\n  // Combine multiple clauses for each source with AND\n  const combinedSingleSource = new Map<string, BasicExpression<boolean>>()\n  for (const [source, clauses] of singleSource) {\n    combinedSingleSource.set(source, combineWithAnd(clauses))\n  }\n\n  // Combine multi-source clauses with AND\n  const combinedMultiSource =\n    multiSource.length > 0 ? combineWithAnd(multiSource) : undefined\n\n  return {\n    singleSource: combinedSingleSource,\n    multiSource: combinedMultiSource,\n  }\n}\n\n/**\n * Step 4: Apply optimizations by lifting single-source clauses into subqueries.\n *\n * Creates a new QueryIR with single-source WHERE clauses moved to subqueries\n * that wrap the original table references. This ensures immutability and prevents\n * infinite recursion issues.\n *\n * @param query - Original QueryIR to optimize\n * @param groupedClauses - WHERE clauses grouped by optimization strategy\n * @returns New QueryIR with optimizations applied\n */\nfunction applyOptimizations(\n  query: QueryIR,\n  groupedClauses: GroupedWhereClauses\n): QueryIR {\n  // Track which single-source clauses were actually optimized\n  const actuallyOptimized = new Set<string>()\n\n  // Optimize the main FROM clause and track what was optimized\n  const optimizedFrom = optimizeFromWithTracking(\n    query.from,\n    groupedClauses.singleSource,\n    actuallyOptimized\n  )\n\n  // Optimize JOIN clauses and track what was optimized\n  const optimizedJoins = query.join\n    ? query.join.map((joinClause) => ({\n        ...joinClause,\n        from: optimizeFromWithTracking(\n          joinClause.from,\n          groupedClauses.singleSource,\n          actuallyOptimized\n        ),\n      }))\n    : undefined\n\n  // Build the remaining WHERE clauses: multi-source + residual single-source clauses\n  const remainingWhereClauses: Array<Where> = []\n\n  // Add multi-source clauses\n  if (groupedClauses.multiSource) {\n    remainingWhereClauses.push(groupedClauses.multiSource)\n  }\n\n  // Determine if we need residual clauses (when query has outer JOINs)\n  const hasOuterJoins =\n    query.join &&\n    query.join.some(\n      (join) =>\n        join.type === `left` || join.type === `right` || join.type === `full`\n    )\n\n  // Add single-source clauses\n  for (const [source, clause] of groupedClauses.singleSource) {\n    if (!actuallyOptimized.has(source)) {\n      // Wasn't optimized at all - keep as regular WHERE clause\n      remainingWhereClauses.push(clause)\n    } else if (hasOuterJoins) {\n      // Was optimized AND query has outer JOINs - keep as residual WHERE clause\n      remainingWhereClauses.push(createResidualWhere(clause))\n    }\n    // If optimized and no outer JOINs - don't keep (original behavior)\n  }\n\n  // Create a completely new query object to ensure immutability\n  const optimizedQuery: QueryIR = {\n    // Copy all non-optimized fields as-is\n    select: query.select,\n    groupBy: query.groupBy ? [...query.groupBy] : undefined,\n    having: query.having ? [...query.having] : undefined,\n    orderBy: query.orderBy ? [...query.orderBy] : undefined,\n    limit: query.limit,\n    offset: query.offset,\n    fnSelect: query.fnSelect,\n    fnWhere: query.fnWhere ? [...query.fnWhere] : undefined,\n    fnHaving: query.fnHaving ? [...query.fnHaving] : undefined,\n\n    // Use the optimized FROM and JOIN clauses\n    from: optimizedFrom,\n    join: optimizedJoins,\n\n    // Only include WHERE clauses that weren't successfully optimized\n    where: remainingWhereClauses.length > 0 ? remainingWhereClauses : [],\n  }\n\n  return optimizedQuery\n}\n\n/**\n * Helper function to create a deep copy of a QueryIR object for immutability.\n *\n * This ensures that all optimizations create new objects rather than modifying\n * existing ones, preventing infinite recursion and shared reference issues.\n *\n * @param query - QueryIR to deep copy\n * @returns New QueryIR object with all nested objects copied\n */\nfunction deepCopyQuery(query: QueryIR): QueryIR {\n  return {\n    // Recursively copy the FROM clause\n    from:\n      query.from.type === `collectionRef`\n        ? new CollectionRefClass(query.from.collection, query.from.alias)\n        : new QueryRefClass(deepCopyQuery(query.from.query), query.from.alias),\n\n    // Copy all other fields, creating new arrays where necessary\n    select: query.select,\n    join: query.join\n      ? query.join.map((joinClause) => ({\n          type: joinClause.type,\n          left: joinClause.left,\n          right: joinClause.right,\n          from:\n            joinClause.from.type === `collectionRef`\n              ? new CollectionRefClass(\n                  joinClause.from.collection,\n                  joinClause.from.alias\n                )\n              : new QueryRefClass(\n                  deepCopyQuery(joinClause.from.query),\n                  joinClause.from.alias\n                ),\n        }))\n      : undefined,\n    where: query.where ? [...query.where] : undefined,\n    groupBy: query.groupBy ? [...query.groupBy] : undefined,\n    having: query.having ? [...query.having] : undefined,\n    orderBy: query.orderBy ? [...query.orderBy] : undefined,\n    limit: query.limit,\n    offset: query.offset,\n    fnSelect: query.fnSelect,\n    fnWhere: query.fnWhere ? [...query.fnWhere] : undefined,\n    fnHaving: query.fnHaving ? [...query.fnHaving] : undefined,\n  }\n}\n\n/**\n * Helper function to optimize a FROM clause while tracking what was actually optimized.\n *\n * @param from - FROM clause to optimize\n * @param singleSourceClauses - Map of source aliases to their WHERE clauses\n * @param actuallyOptimized - Set to track which sources were actually optimized\n * @returns New FROM clause, potentially wrapped in a subquery\n */\nfunction optimizeFromWithTracking(\n  from: From,\n  singleSourceClauses: Map<string, BasicExpression<boolean>>,\n  actuallyOptimized: Set<string>\n): From {\n  const whereClause = singleSourceClauses.get(from.alias)\n\n  if (!whereClause) {\n    // No optimization needed, but return a copy to maintain immutability\n    if (from.type === `collectionRef`) {\n      return new CollectionRefClass(from.collection, from.alias)\n    }\n    // Must be queryRef due to type system\n    return new QueryRefClass(deepCopyQuery(from.query), from.alias)\n  }\n\n  if (from.type === `collectionRef`) {\n    // Create a new subquery with the WHERE clause for the collection\n    // This is always safe since we're creating a new subquery\n    const subQuery: QueryIR = {\n      from: new CollectionRefClass(from.collection, from.alias),\n      where: [whereClause],\n    }\n    actuallyOptimized.add(from.alias) // Mark as successfully optimized\n    return new QueryRefClass(subQuery, from.alias)\n  }\n\n  // Must be queryRef due to type system\n\n  // SAFETY CHECK: Only check safety when pushing WHERE clauses into existing subqueries\n  // We need to be careful about pushing WHERE clauses into subqueries that already have\n  // aggregates, HAVING, or ORDER BY + LIMIT since that could change their semantics\n  if (!isSafeToPushIntoExistingSubquery(from.query)) {\n    // Return a copy without optimization to maintain immutability\n    // Do NOT mark as optimized since we didn't actually optimize it\n    return new QueryRefClass(deepCopyQuery(from.query), from.alias)\n  }\n\n  // Add the WHERE clause to the existing subquery\n  // Create a deep copy to ensure immutability\n  const existingWhere = from.query.where || []\n  const optimizedSubQuery: QueryIR = {\n    ...deepCopyQuery(from.query),\n    where: [...existingWhere, whereClause],\n  }\n  actuallyOptimized.add(from.alias) // Mark as successfully optimized\n  return new QueryRefClass(optimizedSubQuery, from.alias)\n}\n\n/**\n * Determines if it's safe to push WHERE clauses into an existing subquery.\n *\n * Pushing WHERE clauses into existing subqueries can break semantics in several cases:\n *\n * 1. **Aggregates**: Pushing predicates before GROUP BY changes what gets aggregated\n * 2. **ORDER BY + LIMIT/OFFSET**: Pushing predicates before sorting+limiting changes the result set\n * 3. **HAVING clauses**: These operate on aggregated data, predicates should not be pushed past them\n * 4. **Functional operations**: fnSelect, fnWhere, fnHaving could have side effects\n *\n * Note: This safety check only applies when pushing WHERE clauses into existing subqueries.\n * Creating new subqueries from collection references is always safe.\n *\n * @param query - The existing subquery to check for safety\n * @returns True if it's safe to push WHERE clauses into this subquery, false otherwise\n *\n * @example\n * ```typescript\n * // UNSAFE: has GROUP BY - pushing WHERE could change aggregation\n * { from: users, groupBy: [dept], select: { count: agg('count', '*') } }\n *\n * // UNSAFE: has ORDER BY + LIMIT - pushing WHERE could change \"top 10\"\n * { from: users, orderBy: [salary desc], limit: 10 }\n *\n * // SAFE: plain SELECT without aggregates/limits\n * { from: users, select: { id, name } }\n * ```\n */\nfunction isSafeToPushIntoExistingSubquery(query: QueryIR): boolean {\n  // Check for aggregates in SELECT clause\n  if (query.select) {\n    const hasAggregates = Object.values(query.select).some(\n      (expr) => expr.type === `agg`\n    )\n    if (hasAggregates) {\n      return false\n    }\n  }\n\n  // Check for GROUP BY clause\n  if (query.groupBy && query.groupBy.length > 0) {\n    return false\n  }\n\n  // Check for HAVING clause\n  if (query.having && query.having.length > 0) {\n    return false\n  }\n\n  // Check for ORDER BY with LIMIT or OFFSET (dangerous combination)\n  if (query.orderBy && query.orderBy.length > 0) {\n    if (query.limit !== undefined || query.offset !== undefined) {\n      return false\n    }\n  }\n\n  // Check for functional variants that might have side effects\n  if (\n    query.fnSelect ||\n    (query.fnWhere && query.fnWhere.length > 0) ||\n    (query.fnHaving && query.fnHaving.length > 0)\n  ) {\n    return false\n  }\n\n  // If none of the unsafe conditions are present, it's safe to optimize\n  return true\n}\n\n/**\n * Helper function to combine multiple expressions with AND.\n *\n * If there's only one expression, it's returned as-is.\n * If there are multiple expressions, they're combined with an AND function.\n *\n * @param expressions - Array of expressions to combine\n * @returns Single expression representing the AND combination\n * @throws Error if the expressions array is empty\n */\nfunction combineWithAnd(\n  expressions: Array<BasicExpression<boolean>>\n): BasicExpression<boolean> {\n  if (expressions.length === 0) {\n    throw new CannotCombineEmptyExpressionListError()\n  }\n\n  if (expressions.length === 1) {\n    return expressions[0]!\n  }\n\n  // Create an AND function with all expressions as arguments\n  return new Func(`and`, expressions)\n}\n"],"names":["QueryRefClass","CollectionRefClass"],"mappings":";;;;AA4LO,SAAS,cAAc,OAAoC;AAEhE,QAAM,yBAAyB,8BAA8B,KAAK;AAGlE,MAAI,YAAY;AAChB,MAAI;AACJ,MAAI,aAAa;AACjB,QAAM,gBAAgB;AAGtB,SACE,aAAa,iBACb,CAAC,WAAW,WAAW,iBAAiB,GACxC;AACA,wBAAoB;AACpB,gBAAY,2BAA2B,SAAS;AAChD;AAAA,EACF;AAGA,QAAM,UAAU,0BAA0B,SAAS;AAEnD,SAAO;AAAA,IACL,gBAAgB;AAAA,IAChB;AAAA,EAAA;AAEJ;AAUA,SAAS,8BACP,OACuC;AACvC,QAAM,6CAA6B,IAAA;AAGnC,MAAI,CAAC,MAAM,SAAS,MAAM,MAAM,WAAW,GAAG;AAC5C,WAAO;AAAA,EACT;AAGA,QAAM,oBAAoB,gBAAgB,MAAM,KAAK;AAGrD,QAAM,kBAAkB,kBAAkB;AAAA,IAAI,CAAC,WAC7C,mBAAmB,MAAM;AAAA,EAAA;AAI3B,QAAM,iBAAiB,kBAAkB,eAAe;AAIxD,aAAW,CAAC,aAAa,WAAW,KAAK,eAAe,cAAc;AAEpE,QAAI,sBAAsB,OAAO,WAAW,GAAG;AAE7C,UAAI,gCAAgC,WAAW,GAAG;AAChD,+BAAuB,IAAI,aAAa,WAAW;AAAA,MACrD;AAAA,IACF;AAAA,EACF;AAEA,SAAO;AACT;AAUA,SAAS,sBAAsB,OAAgB,aAA8B;AAE3E,MAAI,MAAM,KAAK,UAAU,aAAa;AACpC,WAAO,MAAM,KAAK,SAAS;AAAA,EAC7B;AAGA,MAAI,MAAM,MAAM;AACd,eAAW,cAAc,MAAM,MAAM;AACnC,UAAI,WAAW,KAAK,UAAU,aAAa;AACzC,eAAO,WAAW,KAAK,SAAS;AAAA,MAClC;AAAA,IACF;AAAA,EACF;AAEA,SAAO;AACT;AAQA,SAAS,2BAA2B,OAAyB;;AAE3D,QAAM,sBAAsB;AAAA,IAC1B,GAAG;AAAA,IACH,MACE,MAAM,KAAK,SAAS,aAChB,IAAIA;AAAAA,MACF,2BAA2B,MAAM,KAAK,KAAK;AAAA,MAC3C,MAAM,KAAK;AAAA,IAAA,IAEb,MAAM;AAAA,IACZ,OAAM,WAAM,SAAN,mBAAY,IAAI,CAAC,gBAAgB;AAAA,MACrC,GAAG;AAAA,MACH,MACE,WAAW,KAAK,SAAS,aACrB,IAAIA;AAAAA,QACF,2BAA2B,WAAW,KAAK,KAAK;AAAA,QAChD,WAAW,KAAK;AAAA,MAAA,IAElB,WAAW;AAAA,IAAA;AAAA,EACjB;AAIJ,SAAO,6BAA6B,mBAAmB;AACzD;AAKA,SAAS,6BAA6B,OAAyB;AAE7D,MAAI,CAAC,MAAM,SAAS,MAAM,MAAM,WAAW,GAAG;AAC5C,WAAO;AAAA,EACT;AAIA,MAAI,CAAC,MAAM,QAAQ,MAAM,KAAK,WAAW,GAAG;AAC1C,WAAO;AAAA,EACT;AAGA,QAAM,0BAA0B,MAAM,MAAM;AAAA,IAC1C,CAAC,UAAU,CAAC,gBAAgB,KAAK;AAAA,EAAA;AAInC,QAAM,oBAAoB,gBAAgB,uBAAuB;AAGjE,QAAM,kBAAkB,kBAAkB;AAAA,IAAI,CAAC,WAC7C,mBAAmB,MAAM;AAAA,EAAA;AAI3B,QAAM,iBAAiB,kBAAkB,eAAe;AAGxD,QAAM,iBAAiB,mBAAmB,OAAO,cAAc;AAG/D,QAAM,uBAAuB,MAAM,MAAM;AAAA,IAAO,CAAC,UAC/C,gBAAgB,KAAK;AAAA,EAAA;AAEvB,MAAI,qBAAqB,SAAS,GAAG;AACnC,mBAAe,QAAQ;AAAA,MACrB,GAAI,eAAe,SAAS,CAAA;AAAA,MAC5B,GAAG;AAAA,IAAA;AAAA,EAEP;AAEA,SAAO;AACT;AAUA,SAAS,0BAA0B,OAAyB;;AAC1D,SAAO;AAAA,IACL,GAAG;AAAA,IACH,MAAM,0BAA0B,MAAM,IAAI;AAAA,IAC1C,OAAM,WAAM,SAAN,mBAAY,IAAI,CAAC,gBAAgB;AAAA,MACrC,GAAG;AAAA,MACH,MAAM,0BAA0B,WAAW,IAAI;AAAA,IAAA;AAAA,EAC/C;AAEN;AAQA,SAAS,0BAA0B,MAAkB;AACnD,MAAI,KAAK,SAAS,iBAAiB;AACjC,WAAO;AAAA,EACT;AAEA,QAAM,iBAAiB,0BAA0B,KAAK,KAAK;AAG3D,MAAI,oBAAoB,cAAc,GAAG;AAEvC,UAAM,YAAY,0BAA0B,eAAe,IAAI;AAC/D,QAAI,UAAU,SAAS,iBAAiB;AACtC,aAAO,IAAIC,cAAmB,UAAU,YAAY,KAAK,KAAK;AAAA,IAChE,OAAO;AACL,aAAO,IAAID,SAAc,UAAU,OAAO,KAAK,KAAK;AAAA,IACtD;AAAA,EACF;AAEA,SAAO,IAAIA,SAAc,gBAAgB,KAAK,KAAK;AACrD;AAQA,SAAS,oBAAoB,OAAyB;AACpD,UACG,CAAC,MAAM,SAAS,MAAM,MAAM,WAAW,MACxC,CAAC,MAAM,WACN,CAAC,MAAM,WAAW,MAAM,QAAQ,WAAW,OAC3C,CAAC,MAAM,UAAU,MAAM,OAAO,WAAW,OACzC,CAAC,MAAM,WAAW,MAAM,QAAQ,WAAW,OAC3C,CAAC,MAAM,QAAQ,MAAM,KAAK,WAAW,MACtC,MAAM,UAAU,UAChB,MAAM,WAAW,UACjB,CAAC,MAAM,aACN,CAAC,MAAM,WAAW,MAAM,QAAQ,WAAW,OAC3C,CAAC,MAAM,YAAY,MAAM,SAAS,WAAW;AAElD;AAiBA,SAAS,gBACP,cACiC;AACjC,QAAM,SAA0C,CAAA;AAEhD,aAAW,eAAe,cAAc;AACtC,UAAM,SAAS,mBAAmB,WAAW;AAC7C,WAAO,KAAK,GAAG,yBAAyB,MAAM,CAAC;AAAA,EACjD;AAEA,SAAO;AACT;AAGA,SAAS,yBACP,QACiC;AACjC,MAAI,OAAO,SAAS,UAAU,OAAO,SAAS,OAAO;AAEnD,UAAM,SAA0C,CAAA;AAChD,eAAW,OAAO,OAAO,MAAyC;AAChE,aAAO,KAAK,GAAG,yBAAyB,GAAG,CAAC;AAAA,IAC9C;AACA,WAAO;AAAA,EACT,OAAO;AAEL,WAAO,CAAC,MAAM;AAAA,EAChB;AACF;AAiBA,SAAS,mBACP,QACqB;AACrB,QAAM,qCAAqB,IAAA;AAK3B,WAAS,eAAe,MAAmC;AACzD,YAAQ,KAAK,MAAA;AAAA,MACX,KAAK;AAEH,YAAI,KAAK,QAAQ,KAAK,KAAK,SAAS,GAAG;AACrC,gBAAM,eAAe,KAAK,KAAK,CAAC;AAChC,cAAI,cAAc;AAChB,2BAAe,IAAI,YAAY;AAAA,UACjC;AAAA,QACF;AACA;AAAA,MACF,KAAK;AAEH,YAAI,KAAK,MAAM;AACb,eAAK,KAAK,QAAQ,cAAc;AAAA,QAClC;AACA;AAAA,MACF,KAAK;AAEH;AAAA,MACF,KAAK;AAEH,YAAI,KAAK,MAAM;AACb,eAAK,KAAK,QAAQ,cAAc;AAAA,QAClC;AACA;AAAA,IAAA;AAAA,EAEN;AAEA,iBAAe,MAAM;AAErB,SAAO;AAAA,IACL,YAAY;AAAA,IACZ;AAAA,EAAA;AAEJ;AAWA,SAAS,kBACP,iBACqB;AACrB,QAAM,mCAAmB,IAAA;AACzB,QAAM,cAA+C,CAAA;AAGrD,aAAW,UAAU,iBAAiB;AACpC,QAAI,OAAO,eAAe,SAAS,GAAG;AAEpC,YAAM,SAAS,MAAM,KAAK,OAAO,cAAc,EAAE,CAAC;AAClD,UAAI,CAAC,aAAa,IAAI,MAAM,GAAG;AAC7B,qBAAa,IAAI,QAAQ,EAAE;AAAA,MAC7B;AACA,mBAAa,IAAI,MAAM,EAAG,KAAK,OAAO,UAAU;AAAA,IAClD,WAAW,OAAO,eAAe,OAAO,GAAG;AAEzC,kBAAY,KAAK,OAAO,UAAU;AAAA,IACpC;AAAA,EAEF;AAGA,QAAM,2CAA2B,IAAA;AACjC,aAAW,CAAC,QAAQ,OAAO,KAAK,cAAc;AAC5C,yBAAqB,IAAI,QAAQ,eAAe,OAAO,CAAC;AAAA,EAC1D;AAGA,QAAM,sBACJ,YAAY,SAAS,IAAI,eAAe,WAAW,IAAI;AAEzD,SAAO;AAAA,IACL,cAAc;AAAA,IACd,aAAa;AAAA,EAAA;AAEjB;AAaA,SAAS,mBACP,OACA,gBACS;AAET,QAAM,wCAAwB,IAAA;AAG9B,QAAM,gBAAgB;AAAA,IACpB,MAAM;AAAA,IACN,eAAe;AAAA,IACf;AAAA,EAAA;AAIF,QAAM,iBAAiB,MAAM,OACzB,MAAM,KAAK,IAAI,CAAC,gBAAgB;AAAA,IAC9B,GAAG;AAAA,IACH,MAAM;AAAA,MACJ,WAAW;AAAA,MACX,eAAe;AAAA,MACf;AAAA,IAAA;AAAA,EACF,EACA,IACF;AAGJ,QAAM,wBAAsC,CAAA;AAG5C,MAAI,eAAe,aAAa;AAC9B,0BAAsB,KAAK,eAAe,WAAW;AAAA,EACvD;AAGA,QAAM,gBACJ,MAAM,QACN,MAAM,KAAK;AAAA,IACT,CAAC,SACC,KAAK,SAAS,UAAU,KAAK,SAAS,WAAW,KAAK,SAAS;AAAA,EAAA;AAIrE,aAAW,CAAC,QAAQ,MAAM,KAAK,eAAe,cAAc;AAC1D,QAAI,CAAC,kBAAkB,IAAI,MAAM,GAAG;AAElC,4BAAsB,KAAK,MAAM;AAAA,IACnC,WAAW,eAAe;AAExB,4BAAsB,KAAK,oBAAoB,MAAM,CAAC;AAAA,IACxD;AAAA,EAEF;AAGA,QAAM,iBAA0B;AAAA;AAAA,IAE9B,QAAQ,MAAM;AAAA,IACd,SAAS,MAAM,UAAU,CAAC,GAAG,MAAM,OAAO,IAAI;AAAA,IAC9C,QAAQ,MAAM,SAAS,CAAC,GAAG,MAAM,MAAM,IAAI;AAAA,IAC3C,SAAS,MAAM,UAAU,CAAC,GAAG,MAAM,OAAO,IAAI;AAAA,IAC9C,OAAO,MAAM;AAAA,IACb,QAAQ,MAAM;AAAA,IACd,UAAU,MAAM;AAAA,IAChB,SAAS,MAAM,UAAU,CAAC,GAAG,MAAM,OAAO,IAAI;AAAA,IAC9C,UAAU,MAAM,WAAW,CAAC,GAAG,MAAM,QAAQ,IAAI;AAAA;AAAA,IAGjD,MAAM;AAAA,IACN,MAAM;AAAA;AAAA,IAGN,OAAO,sBAAsB,SAAS,IAAI,wBAAwB,CAAA;AAAA,EAAC;AAGrE,SAAO;AACT;AAWA,SAAS,cAAc,OAAyB;AAC9C,SAAO;AAAA;AAAA,IAEL,MACE,MAAM,KAAK,SAAS,kBAChB,IAAIC,cAAmB,MAAM,KAAK,YAAY,MAAM,KAAK,KAAK,IAC9D,IAAID,SAAc,cAAc,MAAM,KAAK,KAAK,GAAG,MAAM,KAAK,KAAK;AAAA;AAAA,IAGzE,QAAQ,MAAM;AAAA,IACd,MAAM,MAAM,OACR,MAAM,KAAK,IAAI,CAAC,gBAAgB;AAAA,MAC9B,MAAM,WAAW;AAAA,MACjB,MAAM,WAAW;AAAA,MACjB,OAAO,WAAW;AAAA,MAClB,MACE,WAAW,KAAK,SAAS,kBACrB,IAAIC;AAAAA,QACF,WAAW,KAAK;AAAA,QAChB,WAAW,KAAK;AAAA,MAAA,IAElB,IAAID;AAAAA,QACF,cAAc,WAAW,KAAK,KAAK;AAAA,QACnC,WAAW,KAAK;AAAA,MAAA;AAAA,IAClB,EACN,IACF;AAAA,IACJ,OAAO,MAAM,QAAQ,CAAC,GAAG,MAAM,KAAK,IAAI;AAAA,IACxC,SAAS,MAAM,UAAU,CAAC,GAAG,MAAM,OAAO,IAAI;AAAA,IAC9C,QAAQ,MAAM,SAAS,CAAC,GAAG,MAAM,MAAM,IAAI;AAAA,IAC3C,SAAS,MAAM,UAAU,CAAC,GAAG,MAAM,OAAO,IAAI;AAAA,IAC9C,OAAO,MAAM;AAAA,IACb,QAAQ,MAAM;AAAA,IACd,UAAU,MAAM;AAAA,IAChB,SAAS,MAAM,UAAU,CAAC,GAAG,MAAM,OAAO,IAAI;AAAA,IAC9C,UAAU,MAAM,WAAW,CAAC,GAAG,MAAM,QAAQ,IAAI;AAAA,EAAA;AAErD;AAUA,SAAS,yBACP,MACA,qBACA,mBACM;AACN,QAAM,cAAc,oBAAoB,IAAI,KAAK,KAAK;AAEtD,MAAI,CAAC,aAAa;AAEhB,QAAI,KAAK,SAAS,iBAAiB;AACjC,aAAO,IAAIC,cAAmB,KAAK,YAAY,KAAK,KAAK;AAAA,IAC3D;AAEA,WAAO,IAAID,SAAc,cAAc,KAAK,KAAK,GAAG,KAAK,KAAK;AAAA,EAChE;AAEA,MAAI,KAAK,SAAS,iBAAiB;AAGjC,UAAM,WAAoB;AAAA,MACxB,MAAM,IAAIC,cAAmB,KAAK,YAAY,KAAK,KAAK;AAAA,MACxD,OAAO,CAAC,WAAW;AAAA,IAAA;AAErB,sBAAkB,IAAI,KAAK,KAAK;AAChC,WAAO,IAAID,SAAc,UAAU,KAAK,KAAK;AAAA,EAC/C;AAOA,MAAI,CAAC,iCAAiC,KAAK,KAAK,GAAG;AAGjD,WAAO,IAAIA,SAAc,cAAc,KAAK,KAAK,GAAG,KAAK,KAAK;AAAA,EAChE;AAIA,QAAM,gBAAgB,KAAK,MAAM,SAAS,CAAA;AAC1C,QAAM,oBAA6B;AAAA,IACjC,GAAG,cAAc,KAAK,KAAK;AAAA,IAC3B,OAAO,CAAC,GAAG,eAAe,WAAW;AAAA,EAAA;AAEvC,oBAAkB,IAAI,KAAK,KAAK;AAChC,SAAO,IAAIA,SAAc,mBAAmB,KAAK,KAAK;AACxD;AA8BA,SAAS,iCAAiC,OAAyB;AAEjE,MAAI,MAAM,QAAQ;AAChB,UAAM,gBAAgB,OAAO,OAAO,MAAM,MAAM,EAAE;AAAA,MAChD,CAAC,SAAS,KAAK,SAAS;AAAA,IAAA;AAE1B,QAAI,eAAe;AACjB,aAAO;AAAA,IACT;AAAA,EACF;AAGA,MAAI,MAAM,WAAW,MAAM,QAAQ,SAAS,GAAG;AAC7C,WAAO;AAAA,EACT;AAGA,MAAI,MAAM,UAAU,MAAM,OAAO,SAAS,GAAG;AAC3C,WAAO;AAAA,EACT;AAGA,MAAI,MAAM,WAAW,MAAM,QAAQ,SAAS,GAAG;AAC7C,QAAI,MAAM,UAAU,UAAa,MAAM,WAAW,QAAW;AAC3D,aAAO;AAAA,IACT;AAAA,EACF;AAGA,MACE,MAAM,YACL,MAAM,WAAW,MAAM,QAAQ,SAAS,KACxC,MAAM,YAAY,MAAM,SAAS,SAAS,GAC3C;AACA,WAAO;AAAA,EACT;AAGA,SAAO;AACT;AAYA,SAAS,eACP,aAC0B;AAC1B,MAAI,YAAY,WAAW,GAAG;AAC5B,UAAM,IAAI,sCAAA;AAAA,EACZ;AAEA,MAAI,YAAY,WAAW,GAAG;AAC5B,WAAO,YAAY,CAAC;AAAA,EACtB;AAGA,SAAO,IAAI,KAAK,OAAO,WAAW;AACpC;"}

package/dist/esm/types.d.ts

@@ -473,3 +473,15 @@ export interface CurrentStateAsChangesOptions<T extends object = Record<string,
  * }
  */
 export type ChangeListener<T extends object = Record<string, unknown>, TKey extends string | number = string | number> = (changes: Array<ChangeMessage<T, TKey>>) => void;
+type BuiltIns = null | undefined | string | number | boolean | symbol | bigint | void | Date | RegExp;
+type HasMultipleCallSignatures<T extends (...arguments_: Array<any>) => unknown> = T extends {
+    (...arguments_: infer A): unknown;
+    (...arguments_: infer B): unknown;
+} ? B extends A ? A extends B ? false : true : true : false;
+type WritableMapDeep<MapType extends ReadonlyMap<unknown, unknown>> = MapType extends ReadonlyMap<infer KeyType, infer ValueType> ? Map<WritableDeep<KeyType>, WritableDeep<ValueType>> : MapType;
+type WritableSetDeep<SetType extends ReadonlySet<unknown>> = SetType extends ReadonlySet<infer ItemType> ? Set<WritableDeep<ItemType>> : SetType;
+type WritableObjectDeep<ObjectType extends object> = {
+    -readonly [KeyType in keyof ObjectType]: WritableDeep<ObjectType[KeyType]>;
+};
+type WritableArrayDeep<ArrayType extends ReadonlyArray<unknown>> = ArrayType extends readonly [] ? [] : ArrayType extends readonly [...infer U, infer V] ? [...WritableArrayDeep<U>, WritableDeep<V>] : ArrayType extends readonly [infer U, ...infer V] ? [WritableDeep<U>, ...WritableArrayDeep<V>] : ArrayType extends ReadonlyArray<infer U> ? Array<WritableDeep<U>> : ArrayType extends Array<infer U> ? Array<WritableDeep<U>> : ArrayType;
+export type WritableDeep<T> = T extends BuiltIns ? T : T extends (...arguments_: Array<any>) => unknown ? {} extends WritableObjectDeep<T> ? T : HasMultipleCallSignatures<T> extends true ? T : ((...arguments_: Parameters<T>) => ReturnType<T>) & WritableObjectDeep<T> : T extends ReadonlyMap<unknown, unknown> ? WritableMapDeep<T> : T extends ReadonlySet<unknown> ? WritableSetDeep<T> : T extends ReadonlyArray<unknown> ? WritableArrayDeep<T> : T extends object ? WritableObjectDeep<T> : unknown;

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@tanstack/db",
   "description": "A reactive client store for building super fast apps on sync",
-  "version": "0.1.5",
+  "version": "0.1.7",
   "dependencies": {
     "@standard-schema/spec": "^1.0.0",
     "@tanstack/db-ivm": "0.1.2"

package/src/collection.ts

@@ -58,6 +58,7 @@ import type {
   Transaction as TransactionType,
   TransactionWithMutations,
   UtilsRecord,
+  WritableDeep,
 } from "./types"
 import type { IndexOptions } from "./indexes/index-options.js"
 import type { BaseIndex, IndexResolver } from "./indexes/base-index.js"
@@ -66,6 +67,7 @@ interface PendingSyncedTransaction<T extends object = Record<string, unknown>> {
   committed: boolean
   operations: Array<OptimisticChangeMessage<T>>
   truncate?: boolean
+  deletedKeys: Set<string | number>
 }
 
 /**
@@ -542,6 +544,7 @@ export class CollectionImpl<
           this.pendingSyncedTransactions.push({
             committed: false,
             operations: [],
+            deletedKeys: new Set(),
           })
         },
         write: (messageWithoutKey: Omit<ChangeMessage<T>, `key`>) => {
@@ -560,9 +563,8 @@ export class CollectionImpl<
           // Check if an item with this key already exists when inserting
           if (messageWithoutKey.type === `insert`) {
             const insertingIntoExistingSynced = this.syncedData.has(key)
-            const hasPendingDeleteForKey = pendingTransaction.operations.some(
-              (op) => op.key === key && op.type === `delete`
-            )
+            const hasPendingDeleteForKey =
+              pendingTransaction.deletedKeys.has(key)
             const isTruncateTransaction = pendingTransaction.truncate === true
             // Allow insert after truncate in the same transaction even if it existed in syncedData
             if (
@@ -579,6 +581,10 @@ export class CollectionImpl<
             key,
           }
           pendingTransaction.operations.push(message)
+
+          if (messageWithoutKey.type === `delete`) {
+            pendingTransaction.deletedKeys.add(key)
+          }
         },
         commit: () => {
           const pendingTransaction =
@@ -619,6 +625,7 @@ export class CollectionImpl<
 
           // Clear all operations from the current transaction
           pendingTransaction.operations = []
+          pendingTransaction.deletedKeys.clear()
 
           // Mark the transaction as a truncate operation. During commit, this triggers:
           // - Delete events for all previously synced keys (excluding optimistic-deleted keys)
@@ -1963,32 +1970,34 @@ export class CollectionImpl<
   // Overload 1: Update multiple items with a callback
   update<TItem extends object = T>(
     key: Array<TKey | unknown>,
-    callback: (drafts: Array<TItem>) => void
+    callback: (drafts: Array<WritableDeep<TItem>>) => void
   ): TransactionType
 
   // Overload 2: Update multiple items with config and a callback
   update<TItem extends object = T>(
     keys: Array<TKey | unknown>,
     config: OperationConfig,
-    callback: (drafts: Array<TItem>) => void
+    callback: (drafts: Array<WritableDeep<TItem>>) => void
   ): TransactionType
 
   // Overload 3: Update a single item with a callback
   update<TItem extends object = T>(
     id: TKey | unknown,
-    callback: (draft: TItem) => void
+    callback: (draft: WritableDeep<TItem>) => void
   ): TransactionType
 
   // Overload 4: Update a single item with config and a callback
   update<TItem extends object = T>(
     id: TKey | unknown,
     config: OperationConfig,
-    callback: (draft: TItem) => void
+    callback: (draft: WritableDeep<TItem>) => void
   ): TransactionType
 
   update<TItem extends object = T>(
     keys: (TKey | unknown) | Array<TKey | unknown>,
-    configOrCallback: ((draft: TItem | Array<TItem>) => void) | OperationConfig,
+    configOrCallback:
+      | ((draft: WritableDeep<TItem> | Array<WritableDeep<TItem>>) => void)
+      | OperationConfig,
     maybeCallback?: (draft: TItem | Array<TItem>) => void
   ) {
     if (typeof keys === `undefined`) {

package/src/query/compiler/group-by.ts

@@ -1,5 +1,5 @@
 import { filter, groupBy, groupByOperators, map } from "@tanstack/db-ivm"
-import { Func, PropRef } from "../ir.js"
+import { Func, PropRef, getHavingExpression } from "../ir.js"
 import {
   AggregateFunctionNotInSelectError,
   NonAggregateExpressionNotInGroupByError,
@@ -129,8 +129,9 @@ export function processGroupBy(
     // Apply HAVING clauses if present
     if (havingClauses && havingClauses.length > 0) {
       for (const havingClause of havingClauses) {
+        const havingExpression = getHavingExpression(havingClause)
         const transformedHavingClause = transformHavingClause(
-          havingClause,
+          havingExpression,
           selectClause || {}
         )
         const compiledHaving = compileExpression(transformedHavingClause)
@@ -263,8 +264,9 @@ export function processGroupBy(
   // Apply HAVING clauses if present
   if (havingClauses && havingClauses.length > 0) {
     for (const havingClause of havingClauses) {
+      const havingExpression = getHavingExpression(havingClause)
       const transformedHavingClause = transformHavingClause(
-        havingClause,
+        havingExpression,
         selectClause || {}
       )
       const compiledHaving = compileExpression(transformedHavingClause)

package/src/query/compiler/index.ts

@@ -7,7 +7,7 @@ import {
   LimitOffsetRequireOrderByError,
   UnsupportedFromTypeError,
 } from "../../errors.js"
-import { PropRef } from "../ir.js"
+import { PropRef, getWhereExpression } from "../ir.js"
 import { compileExpression } from "./evaluators.js"
 import { processJoins } from "./joins.js"
 import { processGroupBy } from "./group-by.js"
@@ -131,7 +131,8 @@ export function compileQuery(
   if (query.where && query.where.length > 0) {
     // Apply each WHERE condition as a filter (they are ANDed together)
     for (const where of query.where) {
-      const compiledWhere = compileExpression(where)
+      const whereExpression = getWhereExpression(where)
+      const compiledWhere = compileExpression(whereExpression)
       pipeline = pipeline.pipe(
         filter(([_key, namespacedRow]) => {
           return compiledWhere(namespacedRow)

package/src/query/index.ts

@@ -50,5 +50,6 @@ export { compileQuery } from "./compiler/index.js"
 export {
   createLiveQueryCollection,
   liveQueryCollectionOptions,
-  type LiveQueryCollectionConfig,
 } from "./live-query-collection.js"
+
+export { type LiveQueryCollectionConfig } from "./live/types.js"

package/src/query/ir.ts

@@ -39,7 +39,9 @@ export interface JoinClause {
   right: BasicExpression
 }
 
-export type Where = BasicExpression<boolean>
+export type Where =
+  | BasicExpression<boolean>
+  | { expression: BasicExpression<boolean>; residual?: boolean }
 
 export type GroupBy = Array<BasicExpression>
 
@@ -128,3 +130,48 @@ export class Aggregate<T = any> extends BaseExpression<T> {
     super()
   }
 }
+
+/**
+ * Helper functions for working with Where clauses
+ */
+
+/**
+ * Extract the expression from a Where clause
+ */
+export function getWhereExpression(where: Where): BasicExpression<boolean> {
+  return typeof where === `object` && `expression` in where
+    ? where.expression
+    : where
+}
+
+/**
+ * Extract the expression from a HAVING clause
+ * HAVING clauses can contain aggregates, unlike regular WHERE clauses
+ */
+export function getHavingExpression(
+  having: Having
+): BasicExpression | Aggregate {
+  return typeof having === `object` && `expression` in having
+    ? having.expression
+    : having
+}
+
+/**
+ * Check if a Where clause is marked as residual
+ */
+export function isResidualWhere(where: Where): boolean {
+  return (
+    typeof where === `object` &&
+    `expression` in where &&
+    where.residual === true
+  )
+}
+
+/**
+ * Create a residual Where clause from an expression
+ */
+export function createResidualWhere(
+  expression: BasicExpression<boolean>
+): Where {
+  return { expression, residual: true }
+}