10x performance improvement by optimizing zod parsing

Author: brendan-kellam

packages/web/package.json

@@ -51,6 +51,7 @@
     "@codemirror/search": "^6.5.6",
     "@codemirror/state": "^6.4.1",
     "@codemirror/view": "^6.33.0",
+    "@duplojs/zod-accelerator": "^2.6.2",
     "@floating-ui/react": "^0.27.2",
     "@hookform/resolvers": "^3.9.0",
     "@iconify/react": "^5.1.0",

packages/web/src/app/[domain]/search/components/searchResultsPage.tsx

@@ -35,7 +35,7 @@ import { FilterPanel } from "./filterPanel";
 import { useFilteredMatches } from "./filterPanel/useFilterMatches";
 import { SearchResultsPanel } from "./searchResultsPanel";
 
-const DEFAULT_MAX_MATCH_COUNT = 5000;
+const DEFAULT_MAX_MATCH_COUNT = 100_000;
 
 interface SearchResultsPageProps {
     searchQuery: string;

packages/web/src/features/search/schemas.ts

@@ -141,6 +141,7 @@ export const searchResponseSchema = z.object({
     repositoryInfo: z.array(repositoryInfoSchema),
     isBranchFilteringEnabled: z.boolean(),
     isSearchExhaustive: z.boolean(),
+    __debug_timings: z.record(z.string(), z.number()).optional(),
 });
 
 export const fileSourceRequestSchema = z.object({

packages/web/src/features/search/searchApi.ts

@@ -1,16 +1,21 @@
 'use server';
 
+import { sew } from "@/actions";
+import { withOptionalAuthV2 } from "@/withAuthV2";
+import { ZodAccelerator } from "@duplojs/zod-accelerator";
+import { PrismaClient, Repo } from "@sourcebot/db";
+import { base64Decode, createLogger } from "@sourcebot/shared";
+import { StatusCodes } from "http-status-codes";
+import z from "zod";
+import { ErrorCode } from "../../lib/errorCodes";
 import { invalidZoektResponse, ServiceError } from "../../lib/serviceError";
-import { isServiceError } from "../../lib/utils";
+import { isServiceError, measure } from "../../lib/utils";
+import { SearchRequest, SearchResponse, SourceRange } from "./types";
 import { zoektFetch } from "./zoektClient";
-import { ErrorCode } from "../../lib/errorCodes";
-import { StatusCodes } from "http-status-codes";
 import { zoektSearchResponseSchema } from "./zoektSchema";
-import { SearchRequest, SearchResponse, SourceRange } from "./types";
-import { PrismaClient, Repo } from "@sourcebot/db";
-import { sew } from "@/actions";
-import { base64Decode } from "@sourcebot/shared";
-import { withOptionalAuthV2 } from "@/withAuthV2";
+
+const acceleratedZoektSearchResponseSchema = ZodAccelerator.build(zoektSearchResponseSchema);
+const logger = createLogger("searchApi");
 
 // List of supported query prefixes in zoekt.
 // @see : https://github.com/sourcebot-dev/zoekt/blob/main/query/parse.go#L417
@@ -126,7 +131,7 @@ const getFileWebUrl = (template: string, branch: string, fileName: string): stri
     return encodeURI(url + optionalQueryParams);
 }
 
-export const search = async ({ query, matches, contextLines, whole }: SearchRequest) => sew(() =>
+export const search = async ({ query, matches, contextLines, whole }: SearchRequest): Promise<SearchResponse | ServiceError> => sew(() =>
     withOptionalAuthV2(async ({ org, prisma }) => {
         const transformedQuery = await transformZoektQuery(query, org.id, prisma);
         if (isServiceError(transformedQuery)) {
@@ -200,20 +205,22 @@ export const search = async ({ query, matches, contextLines, whole }: SearchRequ
             "X-Tenant-ID": org.id.toString()
         };
 
-        const searchResponse = await zoektFetch({
-            path: "/api/search",
-            body,
-            header,
-            method: "POST",
-        });
+        const { data: searchResponse, durationMs: fetchDurationMs } = await measure(
+            () => zoektFetch({
+                path: "/api/search",
+                body,
+                header,
+                method: "POST",
+            }),
+            "zoekt_fetch",
+            false
+        );
 
         if (!searchResponse.ok) {
             return invalidZoektResponse(searchResponse);
         }
 
-        const searchBody = await searchResponse.json();
-
-        const parser = zoektSearchResponseSchema.transform(async ({ Result }) => {
+        const transformZoektSearchResponse = async ({ Result }: z.infer<typeof zoektSearchResponseSchema>) => {
             // @note (2025-05-12): in zoekt, repositories are identified by the `RepositoryID` field
             // which corresponds to the `id` in the Repo table. In order to efficiently fetch repository
             // metadata when transforming (potentially thousands) of file matches, we aggregate a unique
@@ -379,7 +386,52 @@ export const search = async ({ query, matches, contextLines, whole }: SearchRequ
                     flushReason: Result.FlushReason,
                 }
             } satisfies SearchResponse;
-        });
+        }
+
+        const { data: rawZoektResponse, durationMs: parseJsonDurationMs } = await measure(
+            () => searchResponse.json(),
+            "parse_json",
+            false
+        );
 
-        return parser.parseAsync(searchBody);
+        const { data: zoektResponse, durationMs: parseZoektResponseDurationMs } = await measure(
+            () => acceleratedZoektSearchResponseSchema.parseAsync(rawZoektResponse),
+            "parse_zoekt_response",
+            false
+        );
+
+        const { data: response, durationMs: transformZoektResponseDurationMs } = await measure(
+            () => transformZoektSearchResponse(zoektResponse),
+            "transform_zoekt_response",
+            false
+        );
+
+        const totalDurationMs = fetchDurationMs + parseJsonDurationMs + parseZoektResponseDurationMs + transformZoektResponseDurationMs;
+
+        // Debug log: timing breakdown
+        const timings = [
+            { name: "zoekt_fetch", duration: fetchDurationMs },
+            { name: "parse_json", duration: parseJsonDurationMs },
+            { name: "parse_zoekt_response", duration: parseZoektResponseDurationMs },
+            { name: "transform_zoekt_response", duration: transformZoektResponseDurationMs },
+        ];
+
+        logger.debug(`Search timing breakdown (query: "${query}"):`);
+        timings.forEach(({ name, duration }) => {
+            const percentage = ((duration / totalDurationMs) * 100).toFixed(1);
+            const durationStr = duration.toFixed(2).padStart(8);
+            const percentageStr = percentage.padStart(5);
+            logger.debug(`  ${name.padEnd(25)} ${durationStr}ms (${percentageStr}%)`);
+        });
+        logger.debug(`  ${"TOTAL".padEnd(25)} ${totalDurationMs.toFixed(2).padStart(8)}ms (100.0%)`);
+
+        return {
+            ...response,
+            __debug_timings: {
+                zoekt_fetch: fetchDurationMs,
+                parse_json: parseJsonDurationMs,
+                parse_zoekt_response: parseZoektResponseDurationMs,
+                transform_zoekt_response: transformZoektResponseDurationMs,
+            }
+        } satisfies SearchResponse;
     }));

yarn.lock

@@ -1630,6 +1630,15 @@ __metadata:
   languageName: node
   linkType: hard
 
+"@duplojs/zod-accelerator@npm:^2.6.2":
+  version: 2.6.2
+  resolution: "@duplojs/zod-accelerator@npm:2.6.2"
+  peerDependencies:
+    zod: ">=3.0.0 <4.0.0"
+  checksum: 10c0/9b8a1dd6cc7c79df16d6e82b34a3f9f76242f513bb272265adbf159f565785f7c4be14d9cf492053fbacd806d0865b20ce1e9adff28f92f7ff9730121688e2b5
+  languageName: node
+  linkType: hard
+
 "@emnapi/core@npm:^1.3.1":
   version: 1.3.1
   resolution: "@emnapi/core@npm:1.3.1"
@@ -8058,6 +8067,7 @@ __metadata:
     "@codemirror/search": "npm:^6.5.6"
     "@codemirror/state": "npm:^6.4.1"
     "@codemirror/view": "npm:^6.33.0"
+    "@duplojs/zod-accelerator": "npm:^2.6.2"
     "@eslint/eslintrc": "npm:^3"
     "@floating-ui/react": "npm:^0.27.2"
     "@hookform/resolvers": "npm:^3.9.0"