V3 backport[#9048]: Validate input file for Vectorize inserts (#9093)

* fix: Validate input file for Vectorize inserts

* VS-398: Throw user errors instead of logging error and returning early for Vectorize commands

---------

Co-authored-by: Garvit Gupta <garvit@cloudflare.com>

Author: CarmenPopoviciu

.changeset/stupid-paws-slide.md

@@ -0,0 +1,5 @@
+---
+"wrangler": patch
+---
+
+fix: Validate input file for Vectorize inserts

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

@@ -339,13 +339,9 @@ describe("vectorize commands", () => {
 
 		await expect(
 			runWrangler("vectorize create test-index --dimensions=1536")
-		).resolves.toBeUndefined();
-
-		expect(std.err).toMatchInlineSnapshot(`
-				"X [ERROR] You must provide both dimensions and a metric, or a known model preset when creating an index.
-
-"
-			`);
+		).rejects.toThrowErrorMatchingInlineSnapshot(
+			`[Error: 🚨 You must provide both dimensions and a metric, or a known model preset when creating an index.]`
+		);
 	});
 
 	it("should handle listing vectorize V1 indexes", async () => {
@@ -501,13 +497,12 @@ describe("vectorize commands", () => {
 
 	it("should log error when getByIds does not receive ids", async () => {
 		mockVectorizeV2Request();
-		await runWrangler("vectorize get-vectors test-index --ids");
-
-		expect(std.err).toMatchInlineSnapshot(`
-			"X [ERROR] 🚨 Please provide valid vector identifiers.
 
-"
-		`);
+		await expect(
+			runWrangler("vectorize get-vectors test-index --ids")
+		).rejects.toThrowErrorMatchingInlineSnapshot(
+			`[Error: 🚨 Please provide valid vector identifiers.]`
+		);
 	});
 
 	it("should handle a deleteByIds on a vectorize index", async () => {
@@ -521,13 +516,12 @@ describe("vectorize commands", () => {
 
 	it("should log error when deleteByIds does not receive ids", async () => {
 		mockVectorizeV2Request();
-		await runWrangler("vectorize delete-vectors test-index --ids");
-
-		expect(std.err).toMatchInlineSnapshot(`
-			"X [ERROR] 🚨 Please provide valid vector identifiers for deletion.
 
-"
-		`);
+		await expect(
+			runWrangler("vectorize delete-vectors test-index --ids")
+		).rejects.toThrowErrorMatchingInlineSnapshot(
+			`[Error: 🚨 Please provide valid vector identifiers for deletion.]`
+		);
 	});
 
 	it("should handle a query on a vectorize index", async () => {
@@ -590,30 +584,22 @@ describe("vectorize commands", () => {
 
 	it("should fail query when neither vector nor vector-id is provided", async () => {
 		mockVectorizeV2RequestError();
-		await runWrangler(
-			"vectorize query test-index --top-k=2 --return-values=true"
+		await expect(
+			runWrangler("vectorize query test-index --top-k=2 --return-values=true")
+		).rejects.toThrowErrorMatchingInlineSnapshot(
+			`[Error: 🚨 Either vector or vector-id parameter must be provided, but not both.]`
 		);
-		expect(std.out).toMatchInlineSnapshot(`""`);
-
-		expect(std.err).toMatchInlineSnapshot(`
-			"X [ERROR] 🚨 Either vector or vector-id parameter must be provided, but not both.
-
-"
-		`);
 	});
 
 	it("should fail query when both vector and vector-id are provided", async () => {
 		mockVectorizeV2RequestError();
-		await runWrangler(
-			"vectorize query test-index --vector 1 2 3 '4' --vector-id some-vector-id"
+		await expect(
+			runWrangler(
+				"vectorize query test-index --vector 1 2 3 '4' --vector-id some-vector-id"
+			)
+		).rejects.toThrowErrorMatchingInlineSnapshot(
+			`[Error: 🚨 Either vector or vector-id parameter must be provided, but not both.]`
 		);
-		expect(std.out).toMatchInlineSnapshot(`""`);
-
-		expect(std.err).toMatchInlineSnapshot(`
-			"X [ERROR] 🚨 Either vector or vector-id parameter must be provided, but not both.
-
-"
-		`);
 	});
 
 	it("should fail query with invalid return-metadata flag", async () => {

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

@@ -222,4 +222,22 @@ describe("dataset upsert", () => {
 		✅ Successfully enqueued 5 vectors into index 'my-index' for upsertion."
 	`);
 	});
+
+	it("should reject an invalid file param", async () => {
+		await expect(
+			runWrangler("vectorize upsert my-index --file invalid_vectors.ndjson")
+		).rejects.toThrowErrorMatchingInlineSnapshot(
+			`[Error: 🚨 Cannot read invalid or empty file: invalid_vectors.ndjson.]`
+		);
+	});
+
+	it("should reject an empty file param", async () => {
+		writeFileSync("empty_vectors.ndjson", "");
+
+		await expect(
+			runWrangler("vectorize upsert my-index --file empty_vectors.ndjson")
+		).rejects.toThrowErrorMatchingInlineSnapshot(
+			`[Error: 🚨 Cannot read invalid or empty file: empty_vectors.ndjson.]`
+		);
+	});
 });

packages/wrangler/src/vectorize/common.ts

@@ -1,3 +1,4 @@
+import { access, constants, stat } from "node:fs/promises";
 import { logger } from "../logger";
 import type { Interface as RLInterface } from "node:readline";
 
@@ -18,6 +19,20 @@ export const VECTORIZE_MAX_BATCH_SIZE = 5_000;
 export const VECTORIZE_UPSERT_BATCH_SIZE = VECTORIZE_V1_MAX_BATCH_SIZE;
 export const VECTORIZE_MAX_UPSERT_VECTOR_RECORDS = 100_000;
 
+// Helper function to carry out initial validations to a file supplied for
+// Vectorize upserts/inserts. This function returns true only if the file exists,
+// can be read, and is non-empty.
+export async function isValidFile(path: string): Promise<boolean> {
+	try {
+		await access(path, constants.R_OK);
+
+		const fileStat = await stat(path);
+		return fileStat.isFile() && fileStat.size > 0;
+	} catch (err) {
+		return false;
+	}
+}
+
 // helper method that reads an ndjson file line by line in batches. not this doesn't
 // actually do any parsing - that will be handled on the backend
 // https://nodejs.org/docs/latest-v16.x/api/readline.html#rlsymbolasynciterator

packages/wrangler/src/vectorize/create.ts

@@ -1,4 +1,5 @@
 import { configFileName, formatConfigSnippet, readConfig } from "../config";
+import { UserError } from "../errors";
 import { logger } from "../logger";
 import { createIndex } from "./client";
 import { deprecatedV1DefaultFlag, vectorizeGABanner } from "./common";
@@ -83,10 +84,9 @@ export async function handler(
 			dimensions: args.dimensions,
 		};
 	} else {
-		logger.error(
-			"You must provide both dimensions and a metric, or a known model preset when creating an index."
+		throw new UserError(
+			"🚨 You must provide both dimensions and a metric, or a known model preset when creating an index."
 		);
-		return;
 	}
 
 	if (args.deprecatedV1) {

packages/wrangler/src/vectorize/deleteByIds.ts

@@ -1,4 +1,5 @@
 import { readConfig } from "../config";
+import { UserError } from "../errors";
 import { logger } from "../logger";
 import { deleteByIds } from "./client";
 import { vectorizeGABanner } from "./common";
@@ -33,8 +34,9 @@ export async function handler(
 	const config = readConfig(args);
 
 	if (args.ids.length === 0) {
-		logger.error("🚨 Please provide valid vector identifiers for deletion.");
-		return;
+		throw new UserError(
+			"🚨 Please provide valid vector identifiers for deletion."
+		);
 	}
 
 	logger.log(`📋 Deleting vectors...`);

packages/wrangler/src/vectorize/getByIds.ts

@@ -1,4 +1,5 @@
 import { readConfig } from "../config";
+import { UserError } from "../errors";
 import { logger } from "../logger";
 import { getByIds } from "./client";
 import { vectorizeGABanner } from "./common";
@@ -33,8 +34,7 @@ export async function handler(
 	const config = readConfig(args);
 
 	if (args.ids.length === 0) {
-		logger.error("🚨 Please provide valid vector identifiers.");
-		return;
+		throw new UserError("🚨 Please provide valid vector identifiers.");
 	}
 
 	logger.log(`📋 Fetching vectors...`);

packages/wrangler/src/vectorize/insert.ts

@@ -2,11 +2,13 @@ import { createReadStream } from "node:fs";
 import { createInterface } from "node:readline";
 import { File, FormData } from "undici";
 import { readConfig } from "../config";
+import { UserError } from "../errors";
 import { logger } from "../logger";
 import { insertIntoIndex, insertIntoIndexV1 } from "./client";
 import {
 	deprecatedV1DefaultFlag,
 	getBatchFromFile,
+	isValidFile,
 	VECTORIZE_MAX_BATCH_SIZE,
 	VECTORIZE_MAX_UPSERT_VECTOR_RECORDS,
 	VECTORIZE_UPSERT_BATCH_SIZE,
@@ -56,25 +58,27 @@ export function options(yargs: CommonYargsArgv) {
 export async function handler(
 	args: StrictYargsOptionsToInterface<typeof options>
 ) {
+	if (!(await isValidFile(args.file))) {
+		throw new UserError(`🚨 Cannot read invalid or empty file: ${args.file}.`);
+	}
+
 	const config = readConfig(args);
 	const rl = createInterface({ input: createReadStream(args.file) });
 
 	if (
 		args.deprecatedV1 &&
 		Number(args.batchSize) > VECTORIZE_V1_MAX_BATCH_SIZE
 	) {
-		logger.error(
+		throw new UserError(
 			`🚨 Vectorize currently limits upload batches to ${VECTORIZE_V1_MAX_BATCH_SIZE} records at a time.`
 		);
-		return;
 	} else if (
 		!args.deprecatedV1 &&
 		Number(args.batchSize) > VECTORIZE_MAX_BATCH_SIZE
 	) {
-		logger.error(
-			`🚨 The global rate limit for the Cloudflare API is 1200 requests per five minutes. Vectorize V2 indexes currently limit upload batches to ${VECTORIZE_MAX_BATCH_SIZE} records at a time to stay within the service limits`
+		throw new UserError(
+			`🚨 The global rate limit for the Cloudflare API is 1200 requests per five minutes. Vectorize V2 indexes currently limit upload batches to ${VECTORIZE_MAX_BATCH_SIZE} records at a time to stay within the service limits.`
 		);
-		return;
 	}
 
 	let vectorInsertCount = 0;

packages/wrangler/src/vectorize/query.ts

@@ -1,4 +1,5 @@
 import { readConfig } from "../config";
+import { UserError } from "../errors";
 import { logger } from "../logger";
 import { queryIndexByVector, queryIndexByVectorId } from "./client";
 import { vectorizeGABanner } from "./common";
@@ -121,10 +122,9 @@ export async function handler(
 		(args.vector === undefined && args.vectorId === undefined) ||
 		(args.vector !== undefined && args.vectorId !== undefined)
 	) {
-		logger.error(
+		throw new UserError(
 			"🚨 Either vector or vector-id parameter must be provided, but not both."
 		);
-		return;
 	}
 
 	logger.log(`📋 Searching for relevant vectors...`);

packages/wrangler/src/vectorize/upsert.ts

@@ -2,10 +2,12 @@ import { createReadStream } from "node:fs";
 import { createInterface } from "node:readline";
 import { File, FormData } from "undici";
 import { readConfig } from "../config";
+import { UserError } from "../errors";
 import { logger } from "../logger";
 import { upsertIntoIndex } from "./client";
 import {
 	getBatchFromFile,
+	isValidFile,
 	VECTORIZE_MAX_BATCH_SIZE,
 	VECTORIZE_MAX_UPSERT_VECTOR_RECORDS,
 	vectorizeGABanner,
@@ -47,14 +49,17 @@ export function options(yargs: CommonYargsArgv) {
 export async function handler(
 	args: StrictYargsOptionsToInterface<typeof options>
 ) {
+	if (!(await isValidFile(args.file))) {
+		throw new UserError(`🚨 Cannot read invalid or empty file: ${args.file}.`);
+	}
+
 	const config = readConfig(args);
 	const rl = createInterface({ input: createReadStream(args.file) });
 
 	if (Number(args.batchSize) > VECTORIZE_MAX_BATCH_SIZE) {
-		logger.error(
-			`🚨 The global rate limit for the Cloudflare API is 1200 requests per five minutes. Vectorize indexes currently limit upload batches to ${VECTORIZE_MAX_BATCH_SIZE} records at a time to stay within the service limits`
+		throw new UserError(
+			`🚨 The global rate limit for the Cloudflare API is 1200 requests per five minutes. Vectorize indexes currently limit upload batches to ${VECTORIZE_MAX_BATCH_SIZE} records at a time to stay within the service limits.`
 		);
-		return;
 	}
 
 	let vectorUpsertCount = 0;