Dev environment runs on M1 mac

Pupeteer does not play well with M1 and ARM in general. While they [just released](puppeteer/puppeteer#7546) experimental support for ARM macs, they still don't support [ARM machines for Debian/Docker](puppeteer/puppeteer#7546 (comment)). This PR switches from puppeteer to [zombie](https://github.com/assaf/zombie). Ideally I would like to move back, as Zombie does not appear to be actively maintained.
olaven · May 12, 2022 · ea01032 · ea01032
1 parent 7a5824b
commit ea01032
Show file tree

Hide file tree

Showing 7 changed files with 299 additions and 101 deletions.
diff --git a/docker-compose.yml b/docker-compose.yml
@@ -19,9 +19,7 @@ services:
       - LOG_LEVEL=$LOG_LEVEL
     working_dir: /paperpod
   api:
-    build:
-      context: .
-      dockerfile: ./packages/api/Dockerfile
+    build: .
     env_file:
       - .env
     ports:
@@ -78,7 +76,7 @@ services:
       - .env
     ports:
       - $WEB_PORT:$WEB_PORT
-    command: yarn web dev
+    command: yarn web dev --port $WEB_PORT
     volumes:
       - .:/paperpod
     environment:

diff --git a/packages/api/Dockerfile b/packages/api/Dockerfile
diff --git a/packages/authentication/src/app.ts b/packages/authentication/src/app.ts
@@ -5,7 +5,6 @@ import { subscriptionManagementRoutes } from "./routes/public/subscription-manag
 import { userRoutes } from "./routes/public/user-routes";
 
 export const publicAuthenticationApp = server.app
-
   .appWithEnvironment(
     server.app.appWithBodyParser(server.app.appWithCookieParser())
   )

diff --git a/packages/converter/package.json b/packages/converter/package.json
@@ -14,9 +14,9 @@
     "@paperpod/common": "0.0.1",
     "node-kall": "^1.0.83",
     "pdf-parse": "^1.1.1",
-    "puppeteer": "^5.4.1",
     "serialize-xml": "^0.4.0",
-    "unfluff": "^3.2.0"
+    "unfluff": "^3.2.0",
+    "zombie": "^6.1.4"
   },
   "devDependencies": {
     "@types/faker": "^5.5.3",

diff --git a/packages/converter/src/text/web.ts b/packages/converter/src/text/web.ts
@@ -1,6 +1,7 @@
 import unfluff from "unfluff";
-import puppeteer from "puppeteer";
 import { models } from "@paperpod/common";
+import { logger } from "@paperpod/common";
+import zombie from "zombie";
 
 /**
  * returns the publication timestamp, if any.
@@ -22,30 +23,47 @@ export const extractTextFromWeb = async (
     author: extracted.author().join(", "),
     description: extracted.description(),
     publication_time: date(extracted),
-    added_time: new Date()
+    added_time: new Date(),
   };
 };
 
 /**
- * Using Puppeteer (or a browser-emulator in general) makes it possible for
- * me to access text that is not directly provided, but client side rendered.
+ * Using Zombie (or a browser-emulator in general) makes it possible
+ * to access text that is not directly provided, but client side rendered.
  */
-const getHtml = async (url: string) => {
-  const browser = await puppeteer.launch({
-    executablePath: process.env.PUPPETEER_EXEC_PATH,
-    headless: true,
-    //FIXME: security considerations without sandbox? Read up on this.
-    args: ["--disable-setuid-sandbox", "--no-sandbox"],
-    ignoreHTTPSErrors: true,
-  });
+const getHtml = (url: string): Promise<string> =>
+  new Promise((resolve, _reject) => {
+    const browser = new zombie({
+      debug: true,
+      waitFor: 15000,
+    });
 
-  const page = await browser.newPage();
-  await page.goto(url);
+    browser.visit(url, function () {
+      const html = browser.html();
+      logger.debug({ message: `Got HTML from zombie`, html, url });
+      resolve(html);
+    });
 
-  await page.waitFor("*");
+    /*
+      NOTE: replaced puppeteer in favour of zombie because puppeteer 
+      does not run on arm systems in Docker..
+      [They're working on it though](https://github.com/garris/BackstopJS/issues/1300#issuecomment-1096969710)
 
-  const html = await page.content();
-  await browser.close();
+      TODO: Consider switching back to puppeteer; Zombie is not actively maintained.
 
-  return html;
-};
+      const browser = await puppeteer.launch({
+        executablePath: process.env.PUPPETEER_EXEC_PATH,
+        headless: true,
+        //FIXME: security considerations without sandbox? Read up on this.
+        args: ["--disable-setuid-sandbox", "--no-sandbox"],
+        ignoreHTTPSErrors: true,
+      });
+      const page = await browser.newPage();
+      await page.goto(url);
+
+      await page.waitFor("*");
+
+      const html = await page.content();
+      await browser.close();
+    */
+  });
diff --git a/packages/extension/src/background/fetch_token.ts b/packages/extension/src/background/fetch_token.ts
@@ -25,7 +25,6 @@ const run = async () => {
   }
 };
 
-logger.debug("AM her at all");
 run().then(() => {
   logger.debug(`Running token background script`);
 });