v0:feat: add request-scoped services context enhance logging across processors

lumpinif · lumpinif · commit 68762f197e22 · 2025-07-26T15:29:23.000+08:00
diff --git a/apps/workers/v0/src/app.ts b/apps/workers/v0/src/app.ts
@@ -25,6 +25,11 @@ app.get('/', (c) => {
     runtime: getRuntimeKey(),
     nodeEnv: c.env.WORKER_NODE_ENV,
     connInfo: info,
+    services: {
+      scrapeService: !!c.var.scrapeService,
+      linkService: !!c.var.linkService,
+      markdownConverter: !!c.var.markdownConverter,
+    },
   });
 });
 
diff --git a/apps/workers/v0/src/lib/context.ts b/apps/workers/v0/src/lib/context.ts
@@ -3,7 +3,10 @@ import type { ResponseHeadersPluginContext } from '@orpc/server/plugins';
 import type { Ratelimit } from '@upstash/ratelimit';
 import type { Context as HonoContext } from 'hono';
 import type { RequestIdVariables } from 'hono/request-id';
+import type { NodeHtmlMarkdown } from 'node-html-markdown';
 import type { getAuthClient } from '@/middlewares/client.auth';
+import type { LinkService } from '@/services/link/link.service';
+import type { ScrapeService } from '@/services/scrape/scrape.service';
 
 export interface AppVariables extends RequestIdVariables {
   /** Auth Instance */
@@ -20,6 +23,10 @@ export interface AppVariables extends RequestIdVariables {
   userIP: string | null;
   /** Rate Limiter */
   rateLimiter: Ratelimit;
+  /** Shared Services - Created at app level for performance */
+  scrapeService: ScrapeService;
+  linkService: LinkService;
+  markdownConverter: NodeHtmlMarkdown;
 }
 
 export interface AppBindings {
diff --git a/apps/workers/v0/src/lib/hono/create-hono-app.tsx b/apps/workers/v0/src/lib/hono/create-hono-app.tsx
@@ -10,6 +10,7 @@ import { apiKeyAuthMiddleware } from '@/middlewares/api-key-auth.hono';
 import { connInfoMiddleware } from '@/middlewares/connInfo.hono';
 import { cookieAuthMiddleware } from '@/middlewares/cookie-auth.hono';
 import deepCrawlCors from '@/middlewares/cors.hono';
+import { servicesAppMiddleware } from '@/middlewares/serivces.app';
 import { serviceFetcherMiddleware } from '@/middlewares/service-fetchers.hono';
 
 export default function createHonoApp() {
@@ -31,6 +32,7 @@ export default function createHonoApp() {
     .use('*', cookieAuthMiddleware)
     .use('*', connInfoMiddleware)
 
+    .use('*', servicesAppMiddleware)
     .use('*', prettyJSON());
 
   // Register check-auth route
diff --git a/apps/workers/v0/src/middlewares/serivces.app.ts b/apps/workers/v0/src/middlewares/serivces.app.ts
@@ -0,0 +1,47 @@
+import { createMiddleware } from 'hono/factory';
+import { NodeHtmlMarkdown } from 'node-html-markdown';
+import type { AppBindings } from '@/lib/context';
+import { LinkService } from '@/services/link/link.service';
+import { ScrapeService } from '@/services/scrape/scrape.service';
+import { logDebug } from '@/utils/loggers';
+import { nhmCustomTranslators, nhmTranslators } from '@/utils/markdown';
+
+// Request-scoped service factory to prevent resource leaks in workerd
+// This ensures proper IoContext lifecycle management while maintaining performance
+function createRequestScopedServices() {
+  const serviceCreationStart = Date.now();
+
+  const services = {
+    scrapeService: new ScrapeService(),
+    linkService: new LinkService(),
+    markdownConverter: new NodeHtmlMarkdown(
+      {}, // options
+      nhmCustomTranslators, // customTransformers
+      nhmTranslators, // customCodeBlockTranslators
+    ),
+  };
+
+  const serviceCreationTime = Date.now() - serviceCreationStart;
+
+  logDebug('[PERF] Request-scoped services created:', {
+    timestamp: new Date().toISOString(),
+    services: Object.keys(services),
+    serviceCreationTime,
+  });
+
+  return services;
+}
+
+export const servicesAppMiddleware = createMiddleware<AppBindings>(
+  async (c, next) => {
+    // Create services per request to respect IoContext boundaries
+    // This prevents resource leaks in production workerd
+    const services = createRequestScopedServices();
+
+    c.set('scrapeService', services.scrapeService);
+    c.set('linkService', services.linkService);
+    c.set('markdownConverter', services.markdownConverter);
+
+    return next();
+  },
+);
diff --git a/apps/workers/v0/src/routers/links/links.processor.ts b/apps/workers/v0/src/routers/links/links.processor.ts
@@ -22,18 +22,14 @@ import {
 } from '@/config/constants';
 import { DEFAULT_LINK_OPTIONS } from '@/config/default-options';
 import type { ORPCContext } from '@/lib/context';
-import { type _linksSets, LinkService } from '@/services/link/link.service';
-import { ScrapeService } from '@/services/scrape/scrape.service';
+import type { _linksSets } from '@/services/link/link.service';
 import { formatDuration } from '@/utils/formater';
 import { kvPutWithRetry } from '@/utils/kv/retry';
 import * as helpers from '@/utils/links/helpers';
+import { logDebug, logError, logWarn } from '@/utils/loggers';
 import { cleanEmptyValues } from '@/utils/response/clean-empty-values';
 import { targetUrlHelper } from '@/utils/url/target-url-helper';
 
-// Create service instances at module level for reuse across requests
-const scrapeService = new ScrapeService();
-const linkService = new LinkService();
-
 // Helper function to check if a URL exists in the Set of visited URLs
 function isUrlInVisitedSet(
   urlSet: Set<VisitedUrl>,
@@ -90,6 +86,21 @@ export async function processLinksRequest(
   const timestamp = new Date().toISOString();
   const startRequestTime = performance.now();
 
+  // Use app-level services from context for optimal performance
+  const serviceAccessStart = Date.now();
+  const scrapeService = c.var.scrapeService;
+  const linkService = c.var.linkService;
+  const serviceAccessTime = Date.now() - serviceAccessStart;
+
+  logDebug('[PERF] Links processor using request-scoped services:', {
+    url,
+    serviceAccessTime,
+    hasScrapeService: !!scrapeService,
+    hasLinkService: !!linkService,
+    timestamp: new Date().toISOString(),
+    requestId: c.var.requestId,
+  });
+
   // config
   const withTree = isTree !== false; // True by default, false only if explicitly set to false
   const linkExtractionOptions: LinkExtractionOptions = {
@@ -311,7 +322,7 @@ export async function processLinksRequest(
             linkService.mergeLinks(extractedKinLinks, _linksSets);
           } catch (error) {
             // Log the error and add to skipped URLs
-            console.error(`Error processing path ${kin}:`, error);
+            logError(`Error processing path ${kin}:`, error);
             const errorMessage =
               error instanceof Error ? error.message : String(error);
             skippedUrls.set(kin, `Failed to process: ${errorMessage}`);
@@ -367,7 +378,7 @@ export async function processLinksRequest(
         }
       } catch (error) {
         // Log the error and add to skipped URLs
-        console.error(`Error processing root URL ${rootUrl}:`, error);
+        logError(`Error processing root URL ${rootUrl}:`, error);
         const errorMessage =
           error instanceof Error ? error.message : String(error);
         skippedUrls.set(rootUrl, `Failed to process: ${errorMessage}`);
@@ -407,7 +418,7 @@ export async function processLinksRequest(
         }
       }
     } catch (error) {
-      console.error(
+      logError(
         `Error reading from DEEPCRAWL_V0_LINKS_STORE for ${rootUrl}:`,
         error,
       );
@@ -572,7 +583,7 @@ export async function processLinksRequest(
           },
         );
       } catch (error) {
-        console.warn(
+        logWarn(
           `[LINKS Endpoint] Failed to store sitemap in KV for ${rootUrl}. Error: ${error instanceof Error ? error.message : String(error)}`,
         );
         // Skip KV put on error, allowing the function to continue
@@ -667,7 +678,7 @@ export async function processLinksRequest(
 
     return linksPostResponse as LinksSuccessResponse;
   } catch (error) {
-    console.error('❌ [LINKS PROCESSOR] error:', error);
+    logError('❌ [LINKS PROCESSOR] error:', error);
 
     const err =
       error instanceof Error
diff --git a/apps/workers/v0/src/routers/read/read.processor.ts b/apps/workers/v0/src/routers/read/read.processor.ts
@@ -6,36 +6,22 @@ import type {
   ReadSuccessResponse,
   ScrapedData,
 } from '@deepcrawl/types/index';
-
-import { NodeHtmlMarkdown } from 'node-html-markdown';
-
+import type { NodeHtmlMarkdown } from 'node-html-markdown';
 import { KV_CACHE_EXPIRATION_TTL } from '@/config/constants';
 import { ENABLE_READ_CACHE } from '@/config/default-options';
 import type { ORPCContext } from '@/lib/context';
-import { ScrapeService } from '@/services/scrape/scrape.service';
 import { formatDuration } from '@/utils/formater';
 import { getReadCacheKey } from '@/utils/kv/read-kv-key';
 import { kvPutWithRetry } from '@/utils/kv/retry';
+import { logDebug, logError } from '@/utils/loggers';
 import {
   fixCodeBlockFormatting,
-  nhmCustomTranslators,
-  nhmTranslators,
   processMultiLineLinks,
   removeNavigationAidLinks,
 } from '@/utils/markdown';
 import { cleanEmptyValues } from '@/utils/response/clean-empty-values';
 import { targetUrlHelper } from '@/utils/url/target-url-helper';
 
-// Create service instance at module level for reuse across requests
-const scrapeService = new ScrapeService();
-
-// Pre-configure NodeHtmlMarkdown at module level to avoid per-request setup
-const markdownConverter = new NodeHtmlMarkdown(
-  /* options */ {},
-  /* customTransformers */ nhmCustomTranslators,
-  /* customCodeBlockTranslators */ nhmTranslators,
-);
-
 /**
  * Calculates performance metrics for the read operation.
  * @param startTime - Timestamp in milliseconds when the operation started.
@@ -109,15 +95,19 @@ function hasMeaningfulMarkdown(markdown: string): boolean {
 }
 
 /**
- * Convert HTML to Markdown
- * @param param0 - HTML content to convert
+ * Convert HTML to Markdown using app-level converter
+ * @param param0 - HTML content and markdown converter
  * @returns Markdown content
  */
-function getMarkdown({ html }: { html: string }): string {
+function getMarkdown({
+  html,
+  markdownConverter,
+}: {
+  html: string;
+  markdownConverter: NodeHtmlMarkdown;
+}): string {
   try {
-    const nhm = markdownConverter;
-
-    let nhmMarkdown = nhm.translate(html);
+    let nhmMarkdown = markdownConverter.translate(html);
     nhmMarkdown = processMultiLineLinks(nhmMarkdown);
     nhmMarkdown = removeNavigationAidLinks(nhmMarkdown);
     nhmMarkdown = fixCodeBlockFormatting(nhmMarkdown);
@@ -303,6 +293,19 @@ export async function processReadRequest(
       requestId: c.var.requestId,
     });
 
+    // Use app-level service instance for optimal performance
+    const serviceAccessStart = Date.now();
+    const scrapeService = c.var.scrapeService;
+    const serviceAccessTime = Date.now() - serviceAccessStart;
+
+    console.log('[PERF] Read processor using request-scoped service:', {
+      url: targetUrl,
+      serviceAccessTime,
+      hasService: !!scrapeService,
+      timestamp: new Date().toISOString(),
+      requestId: c.var.requestId,
+    });
+
     const {
       title,
       rawHtml,
@@ -342,17 +345,20 @@ export async function processReadRequest(
       const markdownStart = Date.now();
 
       if (cleanedHtml) {
-        console.log('[PERF] Read processor markdown conversion started:', {
+        logDebug('[PERF] Read processor markdown conversion started:', {
           url: targetUrl,
           htmlSize: cleanedHtml.length,
           timestamp: new Date().toISOString(),
           requestId: c.var.requestId,
         });
 
-        const convertedMarkdown = getMarkdown({ html: cleanedHtml });
+        const convertedMarkdown = getMarkdown({
+          html: cleanedHtml,
+          markdownConverter: c.var.markdownConverter,
+        });
         const markdownTime = Date.now() - markdownStart;
 
-        console.log('[PERF] Read processor markdown conversion completed:', {
+        logDebug('[PERF] Read processor markdown conversion completed:', {
           url: targetUrl,
           markdownTime,
           markdownSize: convertedMarkdown.length,
@@ -368,7 +374,7 @@ export async function processReadRequest(
           // For POST requests with markdown=true but no meaningful content,
           // provide informative default markdown instead of undefined
           markdown = getDefaultMarkdown(title, targetUrl, description);
-          console.log(
+          logDebug(
             '[PERF] Read processor using default markdown (no meaningful content):',
             {
               url: targetUrl,
@@ -381,7 +387,7 @@ export async function processReadRequest(
         // For POST requests with markdown=true but no extractable content,
         // provide informative default markdown instead of undefined
         markdown = getDefaultMarkdown(title, targetUrl, description);
-        console.log(
+        logDebug(
           '[PERF] Read processor using default markdown (no cleaned HTML):',
           {
             url: targetUrl,
@@ -418,7 +424,7 @@ export async function processReadRequest(
     });
     const responseTime = Date.now() - responseStart;
 
-    console.log('[PERF] Read processor response prepared:', {
+    logDebug('[PERF] Read processor response prepared:', {
       url: targetUrl,
       responseTime,
       hasResponse: !!readResponse,
@@ -443,7 +449,7 @@ export async function processReadRequest(
             )
           : JSON.stringify(readResponse);
 
-        console.log('[PERF] Read processor cache write started:', {
+        logDebug('[PERF] Read processor cache write started:', {
           url: targetUrl,
           cacheValueSize: valueToCache.length,
           isGETRequest,
@@ -466,15 +472,15 @@ export async function processReadRequest(
         );
 
         const cacheWriteTime = Date.now() - cacheWriteStart;
-        console.log('[PERF] Read processor cache write completed:', {
+        logDebug('[PERF] Read processor cache write completed:', {
           url: targetUrl,
           cacheWriteTime,
           timestamp: new Date().toISOString(),
           requestId: c.var.requestId,
         });
       } catch (error) {
         const cacheWriteErrorTime = Date.now() - processorStart;
-        console.error('[PERF] Read processor cache write error:', {
+        logError('[PERF] Read processor cache write error:', {
           url: targetUrl,
           cacheWriteErrorTime,
           error: error instanceof Error ? error.message : String(error),
@@ -487,7 +493,7 @@ export async function processReadRequest(
 
     const totalProcessorTime = Date.now() - processorStart;
 
-    console.log('[PERF] Read processor completed:', {
+    logDebug('[PERF] Read processor completed:', {
       url: targetUrl,
       totalProcessorTime,
       scrapeTime,
diff --git a/apps/workers/v0/src/services/link/link.service.ts b/apps/workers/v0/src/services/link/link.service.ts
diff --git a/apps/workers/v0/src/services/scrape/scrape.service.ts b/apps/workers/v0/src/services/scrape/scrape.service.ts
diff --git a/apps/workers/v0/src/utils/loggers.ts b/apps/workers/v0/src/utils/loggers.ts
