From 103f321dd4dad494a994a3b5eb7fb5bd5d036b10 Mon Sep 17 00:00:00 2001
From: Ilya Kreymer <ikreymer@gmail.com>
Date: Wed, 3 Apr 2024 18:48:56 -0700
Subject: [PATCH] add pixelRatio storing to extraOpts

---
 src/util/recorder.ts | 25 ++++++++++++++++++++-----
 src/util/reqresp.ts  | 17 ++++++++++++++---
 2 files changed, 34 insertions(+), 8 deletions(-)

diff --git a/src/util/recorder.ts b/src/util/recorder.ts
index ec2a24c1..15596749 100644
--- a/src/util/recorder.ts
+++ b/src/util/recorder.ts
@@ -20,7 +20,7 @@ import { WARCRecord } from "warcio";
 import { TempFileBuffer, WARCSerializer } from "warcio/node";
 import { WARCWriter } from "./warcwriter.js";
 import { RedisCrawlState, WorkerId } from "./state.js";
-import { CDPSession, Protocol } from "puppeteer-core";
+import { CDPSession, Page, Protocol } from "puppeteer-core";
 import { Crawler } from "../crawler.js";
 
 const MAX_BROWSER_DEFAULT_FETCH_SIZE = 5_000_000;
@@ -118,6 +118,8 @@ export class Recorder {
   pageUrl!: string;
   pageid!: string;
 
+  pixelRatio: number = 0;
+
   constructor({
     workerid,
     writer,
@@ -142,7 +144,7 @@ export class Recorder {
     this.fetcherQ = new PQueue({ concurrency: 1 });
   }
 
-  async onCreatePage({ cdp }: { cdp: CDPSession }) {
+  async onCreatePage({ cdp, page }: { cdp: CDPSession; page: Page }) {
     // Fetch
     cdp.on("Fetch.requestPaused", async (params) => {
       this.handleRequestPaused(params, cdp);
@@ -204,6 +206,16 @@ export class Recorder {
       }
     });
 
+    // just set the first time
+    if (!this.pixelRatio) {
+      try {
+        const res = await page.evaluate("window.devicePixelRatio");
+        this.pixelRatio = typeof res === "number" ? res : 1;
+      } catch (e) {
+        this.pixelRatio = 1;
+      }
+    }
+
     await cdp.send("Target.setAutoAttach", {
       autoAttach: true,
       waitForDebuggerOnStart: false,
@@ -550,9 +562,12 @@ export class Recorder {
       return false;
     }
 
-    if (url === this.pageUrl && !this.pageInfo.ts) {
-      logger.debug("Setting page timestamp", { ts: reqresp.ts, url });
-      this.pageInfo.ts = reqresp.ts;
+    if (url === this.pageUrl) {
+      if (!this.pageInfo.ts) {
+        logger.debug("Setting page timestamp", { ts: reqresp.ts, url });
+        this.pageInfo.ts = reqresp.ts;
+      }
+      reqresp.extraOpts.pixelRatio = this.pixelRatio;
     }
 
     reqresp.fillFetchRequestPaused(params);
diff --git a/src/util/reqresp.ts b/src/util/reqresp.ts
index d8c36ee1..145e55d9 100644
--- a/src/util/reqresp.ts
+++ b/src/util/reqresp.ts
@@ -14,6 +14,19 @@ export const MAX_URL_LENGTH = 4096;
 // max length for single query arg for post/put converted URLs
 const MAX_ARG_LEN = 512;
 
+// ===========================================================================
+export type ExtraOptsType = {
+  cert?: { issuer: string; ctc: string };
+  ipType?: string;
+  rewritten?: number;
+  pixelRatio?: number;
+
+  // for wabac.js video rewriting
+  maxRes?: number;
+  maxBand?: number;
+  maxBitrate?: number;
+};
+
 // ===========================================================================
 export class RequestResponseInfo {
   ts: Date = new Date();
@@ -55,9 +68,7 @@ export class RequestResponseInfo {
 
   resourceType?: string;
 
-  // TODO: Fix this the next time the file is edited.
-  // eslint-disable-next-line @typescript-eslint/no-explicit-any
-  extraOpts: Record<string, any> = {};
+  extraOpts: ExtraOptsType = {};
 
   // stats
   readSize: number = 0;