Add set and range filters to filter validation when using Wrangler CLI (#7592)

Author: garrettgu10

.changeset/lemon-years-sniff.md

@@ -0,0 +1,5 @@
+---
+"wrangler": minor
+---
+
+New filter validation logic supporting set and range queries in Vectorize CLI

packages/wrangler/src/__tests__/vectorize/vectorize.test.ts

@@ -727,11 +727,11 @@ describe("vectorize commands", () => {
 describe("vectorize query filter", () => {
 	it("should parse correctly", async () => {
 		let jsonString =
-			'{ "p1": "abc", "p2": { "$ne": true }, "p3": 10, "p4": false, "nested.p5": "abcd" }'; // Successful parse
+			'{ "p1": "abc", "p2": { "$ne": true }, "p3": 10, "p4": false, "nested.p5": "abcd", "p6": { "$in": ["a", 3, 4] }, "p7": {"$gt": 4, "$lte": "aaa"} }'; // Successful parse
 		expect(
 			JSON.stringify(validateQueryFilter(JSON.parse(jsonString)))
 		).toMatchInlineSnapshot(
-			`"{\\"p1\\":\\"abc\\",\\"p2\\":{\\"$ne\\":true},\\"p3\\":10,\\"p4\\":false,\\"nested.p5\\":\\"abcd\\"}"`
+			`"{\\"p1\\":\\"abc\\",\\"p2\\":{\\"$ne\\":true},\\"p3\\":10,\\"p4\\":false,\\"nested.p5\\":\\"abcd\\",\\"p6\\":{\\"$in\\":[\\"a\\",3,4]},\\"p7\\":{\\"$gt\\":4,\\"$lte\\":\\"aaa\\"}}"`
 		);
 
 		jsonString =

packages/wrangler/src/vectorize/query.ts

@@ -8,13 +8,11 @@ import type {
 } from "../yargs-types";
 import type {
 	VectorizeMatches,
-	VectorizeMetadataFilterInnerValue,
 	VectorizeMetadataFilterValue,
 	VectorizeMetadataRetrievalLevel,
 	VectorizeQueryOptions,
 	VectorizeVectorMetadataFilter,
 	VectorizeVectorMetadataFilterOp,
-	VectorizeVectorMetadataValue,
 } from "./types";
 
 export function options(yargs: CommonYargsArgv) {
@@ -155,6 +153,12 @@ export async function handler(
 	logger.log(JSON.stringify(res, null, 2));
 }
 
+function validateQueryFilterInnerValue(
+	innerValue: VectorizeMetadataFilterValue
+) {
+	return ["string", "number", "boolean"].includes(typeof innerValue);
+}
+
 export function validateQueryFilter(
 	input: object
 ): VectorizeVectorMetadataFilter | null {
@@ -176,7 +180,7 @@ export function validateQueryFilter(
 		for (const field in parsedObj) {
 			if (Object.prototype.hasOwnProperty.call(parsedObj, field)) {
 				const value = (
-					parsedObj as Record<string, VectorizeMetadataFilterValue>
+					parsedObj as Record<string, VectorizeVectorMetadataFilter>
 				)[field];
 
 				if (Array.isArray(value)) {
@@ -186,38 +190,28 @@ export function validateQueryFilter(
 
 				if (typeof value === "object" && value !== null) {
 					// Handle nested objects
-					const innerObj: Partial<{
-						[Op in VectorizeVectorMetadataFilterOp]?: Exclude<
-							VectorizeVectorMetadataValue,
-							string[]
-						> | null;
-					}> = {};
+					const innerObj: VectorizeVectorMetadataFilter = {};
 					let validInnerObj = true;
 
 					for (const op in value) {
 						if (Object.prototype.hasOwnProperty.call(value, op)) {
-							if (!["$eq", "$ne"].includes(op)) {
-								// Skip objects with invalid operators
-								validInnerObj = false;
-								break;
-							}
-							const innerValue = (value as VectorizeMetadataFilterInnerValue)[
-								op as VectorizeVectorMetadataFilterOp
-							];
-							if (Array.isArray(innerValue)) {
-								// Skip arrays in nested objects
-								validInnerObj = false;
-								break;
-							}
-							if (
-								typeof innerValue === "object" &&
-								innerValue !== null &&
-								Object.keys(innerValue).length === 0
-							) {
-								// Skip empty objects in nested objects
+							const innerValue = value[op];
+							if (["$eq", "$ne", "$lt", "$lte", "$gt", "gte"].includes(op)) {
+								if (!validateQueryFilterInnerValue(innerValue)) {
+									validInnerObj = false;
+								}
+							} else if (["$in", "$nin"].includes(op)) {
+								if (!Array.isArray(innerValue)) {
+									validInnerObj = false;
+								} else {
+									if (!innerValue.every(validateQueryFilterInnerValue)) {
+										validInnerObj = false;
+									}
+								}
+							} else {
 								validInnerObj = false;
-								break;
 							}
+
 							innerObj[op as VectorizeVectorMetadataFilterOp] = innerValue;
 						}
 					}

packages/wrangler/src/vectorize/types.ts

@@ -25,11 +25,19 @@ export type VectorFloatArray = Float32Array | Float64Array;
  *
  * This list is expected to grow as support for more operations are released.
  */
-export type VectorizeVectorMetadataFilterOp = "$eq" | "$ne";
+export type VectorizeMetadataFilterEqualityOp = "$eq" | "$ne";
+export type VectorizeMetadataFilterSetOp = "$in" | "$nin";
+export type VectorizeMetadataFilterRangeOp = "$lt" | "$lte" | "$gt" | "$gte";
+
+export type VectorizeVectorMetadataFilterOp =
+	| VectorizeMetadataFilterEqualityOp
+	| VectorizeMetadataFilterSetOp
+	| VectorizeMetadataFilterRangeOp;
 
 export type VectorizeMetadataFilterInnerValue = Record<
 	VectorizeVectorMetadataFilterOp,
-	Exclude<VectorizeVectorMetadataValue, string[]>
+	| Exclude<VectorizeVectorMetadataValue, string[]>
+	| Exclude<VectorizeVectorMetadataValue, string[]>[]
 >;
 
 export type VectorizeMetadataFilterValue =