@noy-db/hub 0.1.0-pre.9 → 0.2.0-pre.10

package/dist/{chunk-GOUT6DND.js → chunk-6YLPHBKR.js}

@@ -1,11 +1,13 @@
 import {
   evaluateClause,
   readPath
-} from "./chunk-M5INGEFC.js";
+} from "./chunk-TV3YZ35S.js";
 import {
+  CrossJoinSourceUnknownError,
+  CrossJoinTooLargeError,
   DanglingReferenceError,
   JoinTooLargeError
-} from "./chunk-ACLDOTNQ.js";
+} from "./chunk-O6EJ6WTI.js";
 
 // src/query/join.ts
 var DEFAULT_JOIN_MAX_ROWS = 5e4;
@@ -244,6 +246,9 @@ var NO_AGGREGATE = {
   groupBy() {
     throw NOT_ENABLED;
   },
+  groupByN() {
+    throw NOT_ENABLED;
+  },
   scanAggregate() {
     throw NOT_ENABLED;
   }
@@ -257,16 +262,89 @@ var EMPTY_PLAN = {
   offset: 0,
   joins: []
 };
+var DEFAULT_CROSS_JOIN_MAX_ROWS = 5e4;
 var Query = class _Query {
   source;
   plan;
   joinContext;
   aggregateStrategy;
-  constructor(source, plan = EMPTY_PLAN, joinContext, aggregateStrategy = NO_AGGREGATE) {
+  predicates;
+  constructor(source, plan = EMPTY_PLAN, joinContext, aggregateStrategy = NO_AGGREGATE, predicates) {
     this.source = source;
     this.plan = plan;
     this.joinContext = joinContext;
     this.aggregateStrategy = aggregateStrategy;
+    this.predicates = predicates;
+  }
+  /**
+   * @internal — accessor for the materialized-view dependency
+   * analyzer. Not part of the public API; consumers should use the
+   * builder methods, not inspect the plan directly.
+   */
+  _plan() {
+    return this.plan;
+  }
+  /**
+   * @internal — accessor for the materialized-view dependency
+   * analyzer. Returns the join resolution context (or `undefined` for
+   * queries constructed without a Collection backing).
+   */
+  _joinContext() {
+    return this.joinContext;
+  }
+  /**
+   * @internal — clone this Query with a declared-predicate map
+   * attached. Used by the materialized-view registry to enable
+   * `.wherePredicate(name, ctx?)` for the MV's query callback.
+   * Consumers don't call this directly.
+   */
+  _withPredicates(predicates) {
+    return new _Query(
+      this.source,
+      this.plan,
+      this.joinContext,
+      this.aggregateStrategy,
+      predicates
+    );
+  }
+  /**
+   * Filter by a registered deterministic predicate. Requires
+   * the Query to have been augmented with a predicates map (typically
+   * via the materialized-view registry — bare Queries constructed
+   * outside an MV throw on `.wherePredicate()`).
+   *
+   * `ctx` is an optional opaque value passed verbatim to the predicate
+   * function. Both `predicateHash` (from the registration) and a
+   * canonical-JSON hash of `ctx` fold into the MV's `queryHash`, so
+   * either changing forces refresh on next visit.
+   */
+  wherePredicate(name, ctx) {
+    if (!this.predicates) {
+      throw new Error(
+        `.wherePredicate("${name}"): no predicates registered on this Query. Function-based predicates require the Query to be obtained from inside a materialized-view query() callback whose strategy declares \`predicates: { ${name}: { hash, fn } }\`.`
+      );
+    }
+    const decl = this.predicates.get(name);
+    if (!decl) {
+      throw new Error(
+        `.wherePredicate("${name}"): predicate not registered. Available: ${[...this.predicates.keys()].join(", ") || "(none)"}.`
+      );
+    }
+    const clause = {
+      type: "wherePredicate",
+      name,
+      ctx,
+      predicateHash: decl.hash,
+      ctxHash: canonicalCtxHash(ctx),
+      fn: decl.fn
+    };
+    return new _Query(
+      this.source,
+      { ...this.plan, clauses: [...this.plan.clauses, clause] },
+      this.joinContext,
+      this.aggregateStrategy,
+      this.predicates
+    );
   }
   /** Add a field comparison. Multiple where() calls are AND-combined. */
   where(field, op, value) {
@@ -275,7 +353,8 @@ var Query = class _Query {
       this.source,
       { ...this.plan, clauses: [...this.plan.clauses, clause] },
       this.joinContext,
-      this.aggregateStrategy
+      this.aggregateStrategy,
+      this.predicates
     );
   }
   /**
@@ -285,7 +364,7 @@ var Query = class _Query {
    */
   or(builder) {
     const sub = builder(
-      new _Query(this.source, EMPTY_PLAN, this.joinContext, this.aggregateStrategy)
+      new _Query(this.source, EMPTY_PLAN, this.joinContext, this.aggregateStrategy, this.predicates)
     );
     const group = {
       type: "group",
@@ -296,7 +375,8 @@ var Query = class _Query {
       this.source,
       { ...this.plan, clauses: [...this.plan.clauses, group] },
       this.joinContext,
-      this.aggregateStrategy
+      this.aggregateStrategy,
+      this.predicates
     );
   }
   /**
@@ -305,7 +385,7 @@ var Query = class _Query {
    */
   and(builder) {
     const sub = builder(
-      new _Query(this.source, EMPTY_PLAN, this.joinContext, this.aggregateStrategy)
+      new _Query(this.source, EMPTY_PLAN, this.joinContext, this.aggregateStrategy, this.predicates)
     );
     const group = {
       type: "group",
@@ -316,7 +396,8 @@ var Query = class _Query {
       this.source,
       { ...this.plan, clauses: [...this.plan.clauses, group] },
       this.joinContext,
-      this.aggregateStrategy
+      this.aggregateStrategy,
+      this.predicates
     );
   }
   /** Escape hatch: add an arbitrary predicate function. Not serializable. */
@@ -329,7 +410,8 @@ var Query = class _Query {
       this.source,
       { ...this.plan, clauses: [...this.plan.clauses, clause] },
       this.joinContext,
-      this.aggregateStrategy
+      this.aggregateStrategy,
+      this.predicates
     );
   }
   /** Sort by a field. Subsequent calls are tie-breakers. */
@@ -338,7 +420,8 @@ var Query = class _Query {
       this.source,
       { ...this.plan, orderBy: [...this.plan.orderBy, { field, direction }] },
       this.joinContext,
-      this.aggregateStrategy
+      this.aggregateStrategy,
+      this.predicates
     );
   }
   /** Cap the result size. */
@@ -347,7 +430,8 @@ var Query = class _Query {
       this.source,
       { ...this.plan, limit: n },
       this.joinContext,
-      this.aggregateStrategy
+      this.aggregateStrategy,
+      this.predicates
     );
   }
   /** Skip the first N matching records (after ordering). */
@@ -356,7 +440,8 @@ var Query = class _Query {
       this.source,
       { ...this.plan, offset: n },
       this.joinContext,
-      this.aggregateStrategy
+      this.aggregateStrategy,
+      this.predicates
     );
   }
   /**
@@ -455,7 +540,80 @@ var Query = class _Query {
       this.source,
       { ...this.plan, joins: [...this.plan.joins, leg] },
       this.joinContext,
-      this.aggregateStrategy
+      this.aggregateStrategy,
+      this.predicates
+    );
+  }
+  /**
+   * Cartesian-product cross-join against `target` collection. Each result row
+   * carries the original `T` fields plus `result[as]` populated from every
+   * right-side row (or the filtered subset when `on:` is supplied).
+   *
+   * **Order matters:** `.where().crossJoin()` filters BEFORE expanding (cheaper);
+   * `.crossJoin().where('alias.field', ...)` filters AFTER (required when the
+   * where clause references the aliased fields).
+   *
+   * **Cost ceiling:** `CrossJoinTooLargeError` fires before allocation when
+   * `leftRows × rightRows` (or the cumulative lateral count) exceeds the limit.
+   * Default: 50,000 rows. Override per-clause with `{ maxRows: N }`.
+   *
+   * **`on:` shapes:**
+   *   - `on: (left) => TTarget[]`              — subset form (most efficient)
+   *   - `on: (left) => (right) => boolean`     — predicate form
+   *   - `on: { predicate: 'name' }`            — MV-safe, hash-tracked form
+   *     (requires the Query to have been augmented via `_withPredicates`)
+   *
+   * Requires a JoinContext (constructed via `collection.query()`).
+   */
+  crossJoin(target, opts) {
+    if (!this.joinContext) {
+      throw new Error(
+        `Query.crossJoin("${target}"): requires a join context. Use collection.query() to construct a cross-join-capable Query instead of the Query constructor directly.`
+      );
+    }
+    let onFn;
+    let onPredicateName;
+    if (opts.on !== void 0) {
+      if (typeof opts.on === "function") {
+        onFn = opts.on;
+        if (this.predicates) {
+          console.warn(
+            `Query.crossJoin("${target}", { on: callback }): inline on: callback inside a withMaterializedView query() disables queryHash drift detection for this cross-join. Use on: { predicate: '<name>' } to enable it.`
+          );
+        }
+      } else {
+        const predName = opts.on.predicate;
+        if (!this.predicates) {
+          throw new Error(
+            `Query.crossJoin("${target}", { on: { predicate: "${predName}" } }): the { predicate } form requires a predicates map. Use this form inside a withMaterializedView query() callback that declares predicates: { ${predName}: { hash, fn } }.`
+          );
+        }
+        const decl = this.predicates.get(predName);
+        if (!decl) {
+          throw new Error(
+            `Query.crossJoin("${target}"): predicate "${predName}" not registered. Available: ${[...this.predicates.keys()].join(", ") || "(none)"}.`
+          );
+        }
+        const as = opts.as;
+        const predicateFn = decl.fn;
+        onFn = (_left) => (right) => predicateFn({ ..._left, [as]: right });
+        onPredicateName = predName;
+      }
+    }
+    const clause = {
+      type: "crossJoin",
+      target,
+      as: opts.as,
+      ...onFn !== void 0 && { on: onFn },
+      ...onPredicateName !== void 0 && { onPredicateName },
+      ...opts.maxRows !== void 0 && { maxRows: opts.maxRows }
+    };
+    return new _Query(
+      this.source,
+      { ...this.plan, clauses: [...this.plan.clauses, clause] },
+      this.joinContext,
+      this.aggregateStrategy,
+      this.predicates
     );
   }
   /**
@@ -465,7 +623,7 @@ var Query = class _Query {
    * for the ordering rationale.
    */
   toArray() {
-    const base = executePlanWithSource(this.source, this.plan);
+    const base = executePlanWithSource(this.source, this.plan, this.joinContext);
     if (this.plan.joins.length === 0) return base;
     if (!this.joinContext) {
       throw new Error(
@@ -489,6 +647,14 @@ var Query = class _Query {
    * intent is purely to count.
    */
   count() {
+    if (this.plan.clauses.some((c) => c.type === "crossJoin")) {
+      if (!this.joinContext) {
+        throw new Error(
+          `Query.count(): plan contains crossJoin clauses but no JoinContext is attached.`
+        );
+      }
+      return executeClausePipeline(this.source, this.plan.clauses, this.joinContext).length;
+    }
     const { candidates, remainingClauses } = candidateRecords(this.source, this.plan.clauses);
     if (remainingClauses.length === 0) return candidates.length;
     return filterRecords(candidates, remainingClauses).length;
@@ -536,7 +702,13 @@ var Query = class _Query {
   aggregate(spec) {
     const source = this.source;
     const clauses = this.plan.clauses;
+    const joinCtx = this.joinContext;
+    const hasCrossJoins = clauses.some((c) => c.type === "crossJoin");
     const executeRecords = () => {
+      if (hasCrossJoins) {
+        if (!joinCtx) throw new Error("Query.aggregate(): crossJoin requires a join context");
+        return executeClausePipeline(source, clauses, joinCtx);
+      }
       const { candidates, remainingClauses } = candidateRecords(source, clauses);
       return remainingClauses.length === 0 ? candidates : filterRecords(candidates, remainingClauses);
     };
@@ -547,56 +719,19 @@ var Query = class _Query {
     }
     return this.aggregateStrategy.aggregate(executeRecords, spec, upstreams);
   }
-  /**
-   * Partition matching records into buckets keyed by a field, then
-   * terminate with `.aggregate(spec)` to compute per-bucket
-   * reducers..
-   *
-   * ```ts
-   * const byClient = invoices.query()
-   *   .where('status', '==', 'open')
-   *   .groupBy('clientId')
-   *   .aggregate({ total: sum('amount'), n: count() })
-   *   .run()
-   * // → [ { clientId: 'c1', total: 5250, n: 3 }, … ]
-   * ```
-   *
-   * Result rows carry the group key value under the grouping field
-   * name plus every reducer output from the spec. Buckets are
-   * emitted in first-seen order — consumers who want a specific
-   * ordering should `.sort()` downstream.
-   *
-   * **Cardinality caps:** a one-shot warning fires at 10_000
-   * distinct groups; `GroupCardinalityError` throws at 100_000.
-   * Grouping on a high-uniqueness field like `id` or `createdAt` is
-   * almost always a query mistake — the error message names the
-   * field and observed cardinality and suggests narrowing with
-   * `.where()` first.
-   *
-   * **Null / undefined keys:** records with a missing or explicitly
-   * `null` group field get their own buckets. `Map`-based
-   * partitioning distinguishes `undefined` from `null`, so the two
-   * cases do NOT merge. Consumers who want them merged should
-   * coalesce upstream with `.filter()`.
-   *
-   * **Joins are not applied** — same rationale as `.count()` and
-   * `.aggregate()`. Joined fields in are projection-only, so
-   * running a join inside a grouping pipeline would be wasteful and
-   * could trigger `DanglingReferenceError` in strict mode for a
-   * call whose intent is purely to bucket-and-reduce. Grouping by
-   * a joined field is explicitly out of scope for — file an
-   * issue if a real consumer needs it.
-   *
-   * **Filter clauses (`.filter(fn)`):** grouped queries still
-   * support filter clauses in the underlying plan — they run in
-   * the same candidate/filter pipeline that `.aggregate()` uses.
-   * The performance caveat is the same: filter clauses cost O(N)
-   * per record and can't be index-accelerated.
-   */
-  groupBy(field) {
+  groupBy(...fields) {
+    if (fields.length === 0) {
+      throw new Error(".groupBy() requires at least one field");
+    }
     const source = this.source;
     const clauses = this.plan.clauses;
+    const joinCtx = this.joinContext;
+    const hasCrossJoins = clauses.some((c) => c.type === "crossJoin");
     const executeRecords = () => {
+      if (hasCrossJoins) {
+        if (!joinCtx) throw new Error("Query.groupBy(): crossJoin requires a join context");
+        return executeClausePipeline(source, clauses, joinCtx);
+      }
       const { candidates, remainingClauses } = candidateRecords(source, clauses);
       return remainingClauses.length === 0 ? candidates : filterRecords(candidates, remainingClauses);
     };
@@ -605,36 +740,21 @@ var Query = class _Query {
       const subscribe = source.subscribe.bind(source);
       upstreams.push({ subscribe: (cb) => subscribe(cb) });
     }
-    const joinCtx = this.joinContext;
-    const dictLabelResolver = joinCtx?.resolveDictSource ? (() => {
-      const dictSource = joinCtx.resolveDictSource(field);
-      if (!dictSource) return void 0;
-      const snapshot = dictSource.snapshot();
-      const dictMap = /* @__PURE__ */ new Map();
-      for (const entry of snapshot) {
-        const k = entry["key"];
-        const labels = entry["labels"];
-        if (typeof k === "string" && labels && typeof labels === "object") {
-          dictMap.set(k, labels);
-        }
-      }
-      return async (key, locale, fallback) => {
-        const labels = dictMap.get(key);
-        if (!labels) return void 0;
-        if (labels[locale] !== void 0) return labels[locale];
-        const chain = Array.isArray(fallback) ? fallback : fallback ? [fallback] : [];
-        for (const fb of chain) {
-          if (fb === "any") {
-            const any = Object.values(labels)[0];
-            if (any !== void 0) return any;
-          } else if (labels[fb] !== void 0) {
-            return labels[fb];
-          }
-        }
-        return void 0;
-      };
-    })() : void 0;
-    return this.aggregateStrategy.groupBy(executeRecords, field, upstreams, dictLabelResolver);
+    if (fields.length === 1) {
+      const field = fields[0];
+      const dictLabelResolver = buildDictLabelResolver(this.joinContext, field);
+      return this.aggregateStrategy.groupBy(
+        executeRecords,
+        field,
+        upstreams,
+        dictLabelResolver
+      );
+    }
+    return this.aggregateStrategy.groupByN(
+      executeRecords,
+      fields,
+      upstreams
+    );
   }
   /**
    * Re-run the query whenever the source notifies of changes.
@@ -725,6 +845,21 @@ var Query = class _Query {
         }
       }
     }
+    if (this.joinContext) {
+      const subscribedCross = /* @__PURE__ */ new Set();
+      for (const clause of this.plan.clauses) {
+        if (clause.type !== "crossJoin") continue;
+        if (subscribedCross.has(clause.target)) continue;
+        subscribedCross.add(clause.target);
+        const rightSource = this.joinContext.resolveSource(clause.target);
+        if (rightSource?.subscribe) {
+          const rightSubscribe = rightSource.subscribe.bind(rightSource);
+          upstreams.push({
+            subscribe: (cb) => rightSubscribe(cb)
+          });
+        }
+      }
+    }
     return buildLiveQuery(() => this.toArray(), upstreams);
   }
   /**
@@ -736,9 +871,20 @@ var Query = class _Query {
     return serializePlan(this.plan);
   }
 };
-function executePlanWithSource(source, plan) {
-  const { candidates, remainingClauses } = candidateRecords(source, plan.clauses);
-  let result = remainingClauses.length === 0 ? [...candidates] : filterRecords(candidates, remainingClauses);
+function executePlanWithSource(source, plan, joinContext) {
+  const hasCrossJoins = plan.clauses.some((c) => c.type === "crossJoin");
+  let result;
+  if (hasCrossJoins) {
+    if (!joinContext) {
+      throw new Error(
+        `Query.toArray(): plan contains crossJoin clauses but no JoinContext is attached. Use collection.query() instead of new Query() for cross-join support.`
+      );
+    }
+    result = executeClausePipeline(source, plan.clauses, joinContext);
+  } else {
+    const { candidates, remainingClauses } = candidateRecords(source, plan.clauses);
+    result = remainingClauses.length === 0 ? [...candidates] : filterRecords(candidates, remainingClauses);
+  }
   if (plan.orderBy.length > 0) {
     result = sortRecords(result, plan.orderBy);
   }
@@ -788,6 +934,11 @@ function materializeIds(ids, lookupById) {
   return out;
 }
 function executePlan(records, plan) {
+  if (plan.clauses.some((c) => c.type === "crossJoin")) {
+    throw new Error(
+      `executePlan(): does not support crossJoin clauses. executePlan is a stateless pure function \u2014 it cannot resolve cross-join right-side collections. Use Query.toArray() (via collection.query()) instead.`
+    );
+  }
   let result = filterRecords(records, plan.clauses);
   if (plan.orderBy.length > 0) {
     result = sortRecords(result, plan.orderBy);
@@ -815,6 +966,74 @@ function filterRecords(records, clauses) {
   }
   return out;
 }
+function executeClausePipeline(source, clauses, joinContext) {
+  let rel = [...source.snapshot()];
+  let filterBatch = [];
+  for (const clause of clauses) {
+    if (clause.type === "crossJoin") {
+      if (filterBatch.length > 0) {
+        rel = filterRecords(rel, filterBatch);
+        filterBatch = [];
+      }
+      const rightSource = joinContext.resolveSource(clause.target);
+      if (!rightSource) {
+        throw new CrossJoinSourceUnknownError(clause.target, joinContext.leftCollection);
+      }
+      rel = applyCrossJoin(rel, clause, rightSource);
+    } else {
+      filterBatch.push(clause);
+    }
+  }
+  if (filterBatch.length > 0) {
+    rel = filterRecords(rel, filterBatch);
+  }
+  return rel;
+}
+function applyCrossJoin(leftRel, clause, rightSource) {
+  const rightRows = rightSource.snapshot();
+  const maxRows = clause.maxRows ?? DEFAULT_CROSS_JOIN_MAX_ROWS;
+  const { as } = clause;
+  if (!clause.on) {
+    const product = leftRel.length * rightRows.length;
+    if (product > maxRows) {
+      throw new CrossJoinTooLargeError({ target: clause.target, expected: product, limit: maxRows });
+    }
+    const expanded2 = [];
+    for (const left of leftRel) {
+      const leftObj = left;
+      for (const right of rightRows) {
+        expanded2.push({ ...leftObj, [as]: right });
+      }
+    }
+    return expanded2;
+  }
+  const expanded = [];
+  let cumulative = 0;
+  for (const left of leftRel) {
+    const callbackResult = clause.on(left);
+    let filteredRight;
+    if (Array.isArray(callbackResult)) {
+      filteredRight = callbackResult;
+    } else {
+      filteredRight = rightRows.filter(
+        callbackResult
+      );
+    }
+    cumulative += filteredRight.length;
+    if (cumulative > maxRows) {
+      throw new CrossJoinTooLargeError({
+        target: clause.target,
+        expected: cumulative,
+        limit: maxRows
+      });
+    }
+    const leftObj = left;
+    for (const right of filteredRight) {
+      expanded.push({ ...leftObj, [as]: right });
+    }
+  }
+  return expanded;
+}
 function sortRecords(records, orderBy) {
   return [...records].sort((a, b) => {
     for (const { field, direction } of orderBy) {
@@ -860,6 +1079,16 @@ function serializeClause(clause) {
   if (clause.type === "filter") {
     return { type: "filter", fn: "[function]" };
   }
+  if (clause.type === "wherePredicate") {
+    return {
+      type: "wherePredicate",
+      name: clause.name,
+      ctx: clause.ctx,
+      predicateHash: clause.predicateHash,
+      ctxHash: clause.ctxHash,
+      fn: "[function]"
+    };
+  }
   if (clause.type === "group") {
     return {
       type: "group",
@@ -867,8 +1096,65 @@ function serializeClause(clause) {
       clauses: clause.clauses.map(serializeClause)
     };
   }
+  if (clause.type === "crossJoin") {
+    return {
+      type: "crossJoin",
+      target: clause.target,
+      as: clause.as,
+      on: clause.on ? "[function]" : void 0,
+      onPredicateName: clause.onPredicateName,
+      maxRows: clause.maxRows
+    };
+  }
   return clause;
 }
+function canonicalCtxHash(ctx) {
+  if (ctx === void 0) return "";
+  const canonical = JSON.stringify(ctx, (_key, value) => {
+    if (value && typeof value === "object" && !Array.isArray(value)) {
+      const sorted = {};
+      for (const k of Object.keys(value).sort()) {
+        sorted[k] = value[k];
+      }
+      return sorted;
+    }
+    return value;
+  });
+  let h = 5381;
+  for (let i = 0; i < canonical.length; i++) {
+    h = (h << 5) + h ^ canonical.charCodeAt(i);
+  }
+  return (h >>> 0).toString(16).padStart(8, "0");
+}
+function buildDictLabelResolver(joinCtx, field) {
+  if (!joinCtx?.resolveDictSource) return void 0;
+  const dictSource = joinCtx.resolveDictSource(field);
+  if (!dictSource) return void 0;
+  const snapshot = dictSource.snapshot();
+  const dictMap = /* @__PURE__ */ new Map();
+  for (const entry of snapshot) {
+    const k = entry["key"];
+    const labels = entry["labels"];
+    if (typeof k === "string" && labels && typeof labels === "object") {
+      dictMap.set(k, labels);
+    }
+  }
+  return async (key, locale, fallback) => {
+    const labels = dictMap.get(key);
+    if (!labels) return void 0;
+    if (labels[locale] !== void 0) return labels[locale];
+    const chain = Array.isArray(fallback) ? fallback : fallback ? [fallback] : [];
+    for (const fb of chain) {
+      if (fb === "any") {
+        const any = Object.values(labels)[0];
+        if (any !== void 0) return any;
+      } else if (labels[fb] !== void 0) {
+        return labels[fb];
+      }
+    }
+    return void 0;
+  };
+}
 
 // src/query/scan-builder.ts
 var DEFAULT_SCAN_PAGE_SIZE = 100;
@@ -1278,8 +1564,9 @@ export {
   resetJoinWarnings,
   buildLiveQuery,
   NO_AGGREGATE,
+  DEFAULT_CROSS_JOIN_MAX_ROWS,
   Query,
   executePlan,
   ScanBuilder
 };
-//# sourceMappingURL=chunk-GOUT6DND.js.map
+//# sourceMappingURL=chunk-6YLPHBKR.js.map