From 664d4b1c9f65f1712a7545a484b35d6fc795dda5 Mon Sep 17 00:00:00 2001
From: Yanlong Wang <yanlong.wang@naiver.org>
Date: Fri, 12 Apr 2024 09:25:19 +0800
Subject: [PATCH] fix

---
 .../functions/src/cloud-functions/crawler.ts  | 39 +++++++++----------
 backend/functions/src/services/puppeteer.ts   |  7 ++--
 2 files changed, 23 insertions(+), 23 deletions(-)

diff --git a/backend/functions/src/cloud-functions/crawler.ts b/backend/functions/src/cloud-functions/crawler.ts
index f58f8ee..d8e754e 100644
--- a/backend/functions/src/cloud-functions/crawler.ts
+++ b/backend/functions/src/cloud-functions/crawler.ts
@@ -32,13 +32,21 @@ export class CrawlerHost extends RPCHost {
         const toBeTurnedToMd = snapshot.parsed?.content;
         const contentText = toBeTurnedToMd ? this.turnDownService.turndown(toBeTurnedToMd) : snapshot.text;
 
-        const formatted = `Title: ${(snapshot.parsed?.title || snapshot.title || '').trim()}
+        const formatted = {
+            title: (snapshot.parsed?.title || snapshot.title || '').trim(),
+            urlSource: snapshot.href.trim(),
+            markdownContent: contentText.trim(),
 
-URL Source: ${snapshot.href.trim()}
+            toString() {
+                return `Title: ${this.title}
+
+URL Source: ${this.urlSource}
 
 Markdown Content:
-${contentText.trim()}
+${contentText}
 `;
+            }
+        };
 
         return formatted;
     }
@@ -47,6 +55,7 @@ ${contentText.trim()}
         runtime: {
             memory: '4GiB',
             timeoutSeconds: 540,
+            concurrency: 4,
         },
         httpMethod: ['get', 'post'],
         returnType: [String, OutputServerEventStream],
@@ -60,20 +69,22 @@ ${contentText.trim()}
     ) {
         const noSlashURL = ctx.req.url.slice(1);
         const urlToCrawl = new URL(normalizeUrl(noSlashURL));
+        const screenshotEnabled = Boolean(ctx.req.headers['x-screenshot']);
+        const noCache = Boolean(ctx.req.headers['x-no-cache']);
 
         if (!ctx.req.accepts('text/plain') && ctx.req.accepts('text/event-stream')) {
             const sseStream = new OutputServerEventStream();
             rpcReflect.return(sseStream);
 
             try {
-                for await (const scrapped of this.puppeteerControl.scrap(urlToCrawl.toString())) {
+                for await (const scrapped of this.puppeteerControl.scrap(urlToCrawl.toString(), noCache)) {
                     if (!scrapped) {
                         continue;
                     }
 
                     const formatted = this.formatSnapshot(scrapped);
 
-                    if (scrapped.screenshot) {
+                    if (scrapped.screenshot && screenshotEnabled) {
                         sseStream.write({
                             event: 'screenshot',
                             data: scrapped.screenshot.toString('base64'),
@@ -99,37 +110,25 @@ ${contentText.trim()}
         }
 
         if (!ctx.req.accepts('text/plain') && (ctx.req.accepts('text/json') || ctx.req.accepts('application/json'))) {
-            for await (const scrapped of this.puppeteerControl.scrap(urlToCrawl.toString())) {
+            for await (const scrapped of this.puppeteerControl.scrap(urlToCrawl.toString(), noCache)) {
                 if (!scrapped?.parsed?.content) {
                     continue;
                 }
 
                 const formatted = this.formatSnapshot(scrapped);
 
-                if (scrapped.screenshot) {
-
-                    return [
-                        {
-                            type: 'image_url', image_url: {
-                                url: `data:image/jpeg;base64,${scrapped.screenshot.toString('base64')}`,
-                            }
-                        },
-                        { type: 'text', content: formatted },
-                    ];
-                }
-
                 return formatted;
             }
         }
 
-        for await (const scrapped of this.puppeteerControl.scrap(urlToCrawl.toString())) {
+        for await (const scrapped of this.puppeteerControl.scrap(urlToCrawl.toString(), noCache)) {
             if (!scrapped?.parsed?.content) {
                 continue;
             }
 
             const formatted = this.formatSnapshot(scrapped);
 
-            return assignTransferProtocolMeta(formatted, { contentType: 'text/plain', envelope: null });
+            return assignTransferProtocolMeta(`${formatted}`, { contentType: 'text/plain', envelope: null });
         }
 
         throw new Error('Unreachable');
diff --git a/backend/functions/src/services/puppeteer.ts b/backend/functions/src/services/puppeteer.ts
index 6fd24fd..8545f25 100644
--- a/backend/functions/src/services/puppeteer.ts
+++ b/backend/functions/src/services/puppeteer.ts
@@ -49,9 +49,9 @@ export class PuppeteerControl extends AsyncService {
             return page.browser().connected && !page.isClosed();
         }
     }, {
-        max: 1 + Math.floor(os.freemem() / 1024 * 1024 * 1024),
+        max: Math.max(1 + Math.floor(os.freemem() / 1024 * 1024 * 1024), 4),
         min: 1,
-        acquireTimeoutMillis: 15_000,
+        acquireTimeoutMillis: 60_000,
         testOnBorrow: true,
         testOnReturn: true,
     });
@@ -72,7 +72,7 @@ export class PuppeteerControl extends AsyncService {
         }
         this.browser = await puppeteer.launch({
             headless: true,
-            timeout: 300_000
+            timeout: 60_000
         });
         this.browser.once('disconnected', () => {
             this.logger.warn(`Browser disconnected`);
@@ -91,6 +91,7 @@ export class PuppeteerControl extends AsyncService {
         const preparations = [];
 
         preparations.push(page.setUserAgent(`Slackbot-LinkExpanding 1.0 (+https://api.slack.com/robots)`));
+        preparations.push(page.setBypassCSP(true));
         preparations.push(page.setViewport({ width: 1920, height: 1080 }));
         preparations.push(page.exposeFunction('reportSnapshot', (snapshot: any) => {
             page.emit('snapshot', snapshot);