introduces support for custom page load wait via 'awaitPageLoad(ctx)'…

… function.

- if implemented, this function, allows behavior to add custom wait to the page load
- implement awaitPageLoad for instagram behavior, ensure 'main' element is loaded on the page

package.json

@@ -1,6 +1,6 @@
 {
   "name": "browsertrix-behaviors",
-  "version": "0.5.3",
+  "version": "0.5.4",
   "main": "index.js",
   "author": "Webrecorder Software",
   "license": "AGPL-3.0-or-later",

src/index.ts

@@ -89,7 +89,7 @@ export class BehaviorManager {
       this.behaviors.push(new Autoplay(this.autofetch, opts.startEarly));
     }
 
-    if (self.window.top !== self.window && window["__WB_replay_top"] !== self.window) {
+    if (!this.isInTopFrame()) {
       return;
     }
 
@@ -105,6 +105,9 @@ export class BehaviorManager {
   }
 
   selectMainBehavior() {
+    if (this.mainBehavior) {
+      return;
+    }
     const opts = this.opts;
     let siteMatch = false;
 
@@ -179,6 +182,13 @@ export class BehaviorManager {
     }
   }
 
+  async awaitPageLoad() {
+    this.selectMainBehavior();
+    if (this.mainBehavior?.awaitPageLoad) {
+      await this.mainBehavior.awaitPageLoad();
+    }
+  }
+
   async run(opts: BehaviorManagerOpts = DEFAULT_OPTS, restart = false) {
     if (restart) {
       this.started = false;
@@ -190,9 +200,7 @@ export class BehaviorManager {
     }
 
     this.init(opts, restart);
-    if (!this.mainBehavior) {
-      this.selectMainBehavior();
-    }
+    this.selectMainBehavior();
 
     await awaitLoad();
 
@@ -250,6 +258,10 @@ export class BehaviorManager {
     behaviorLog("Queueing Async Fetch Url: " + url);
     return this.autofetch.queueUrl(url);
   }
+
+  isInTopFrame() {
+    return self.window.top === self.window || window["__WB_replay_top"] === self.window;
+  }
 }
 
 

src/lib/behavior.ts

@@ -87,6 +87,10 @@ export class Behavior extends BackgroundBehavior {
 
   }
 
+  async awaitPageLoad() {
+    // wait for initial page load here
+  }
+
   static load() {
     if (self["__bx_behaviors"]) {
       self["__bx_behaviors"].load(this);
@@ -102,11 +106,13 @@ export class Behavior extends BackgroundBehavior {
   }
 }
 
-// WIP: BehaviorRunner class allows for arbitrary behavirs outside of the
+// WIP: BehaviorRunner class allows for arbitrary behaviors outside of the
 // library to be run through the BehaviorManager
 
 abstract class AbstractBehaviorInst {
   abstract run: (ctx: any) => AsyncIterable<any>;
+
+  abstract awaitPageLoad?: (ctx: any) => Promise<void>;
 }
 
 interface StaticAbstractBehavior {
@@ -193,6 +199,12 @@ export class BehaviorRunner extends BackgroundBehavior {
 
   }
 
+  async awaitPageLoad() {
+    if (this.inst.awaitPageLoad) {
+      await this.inst.awaitPageLoad(this.ctx);
+    }
+  }
+
   static load() {
     if (self["__bx_behaviors"]) {
       self["__bx_behaviors"].load(this);

src/site/instagram.ts

@@ -13,6 +13,7 @@ const Q = {
   commentRoot: "//article[@role='presentation']/div[1]/div[2]//ul",
   viewReplies: "//li//button[span[not(count(*)) and text()!='$1']]",
   loadMore: "//button[span[@aria-label]]",
+  pageLoadWaitUntil: "//main"
 };
 
 export class InstagramPostsBehavior {
@@ -246,4 +247,13 @@ export class InstagramPostsBehavior {
       await sleep(waitUnit * 5);
     }
   }
+
+  async awaitPageLoad(ctx: any) {
+    const { Lib, log } = ctx;
+    const { waitUntilNode } = Lib;
+
+    log("Waiting for Instagram to fully load", "info");
+
+    await waitUntilNode(Q.pageLoadWaitUntil, document, null, 30000);
+  }
 }