separate fetch api for autofetch bbehavior + additional improvements on partial responses: (#736)

Chromium now interrupts fetch() if abort() is called or page is
navigated, so autofetch behavior using native fetch() is less than
ideal. This PR adds support for __bx_fetch() command for autofetch
behavior (supported in browsertrix-behaviors 0.6.6) to fetch separately
from browser's reguar fetch()
- __bx_fetch() starts a fetch, but does not return content to browser,
doesn't need abort(), unaffected by page navigation, but will still try
to use browser network stack when possible, making it more efficient for
background fetching.
- if network stack fetch fails, fallback to regular node fetch() in the
crawler.
Additional improvements for interrupted fetch:
- don't store truncated media responses, even for 200
- avoid doing duplicate async fetching if response already handled (eg.
fetch handled in multiple contexts)
- fixes #735, where fetch was interrupted, resulted in an empty response

.github/workflows/ci.yaml

@@ -61,7 +61,7 @@ jobs:
         run: yarn add -D http-server
 
       - name: install py-wacz as root for tests
-        run: sudo pip install wacz
+        run: sudo pip install wacz --ignore-installed
 
       - name: run all tests as root
         run: sudo DOCKER_HOST_NAME=172.17.0.1 CI=true yarn test -validate

package.json

@@ -18,7 +18,7 @@
   "dependencies": {
     "@novnc/novnc": "1.4.0",
     "@webrecorder/wabac": "^2.20.7",
-    "browsertrix-behaviors": "^0.6.5",
+    "browsertrix-behaviors": "^0.6.6",
     "client-zip": "^2.4.5",
     "css-selector-parser": "^3.0.5",
     "fetch-socks": "^1.3.0",
@@ -37,7 +37,7 @@
     "tsc": "^2.0.4",
     "undici": "^6.18.2",
     "uuid": "8.3.2",
-    "warcio": "^2.4.2",
+    "warcio": "^2.4.3",
     "ws": "^7.4.4",
     "yargs": "^17.7.2"
   },
@@ -67,7 +67,7 @@
   },
   "resolutions": {
     "wrap-ansi": "7.0.0",
-    "warcio": "^2.4.2",
+    "warcio": "^2.4.3",
     "@novnc/novnc": "1.4.0"
   }
 }

src/crawler.ts

@@ -46,6 +46,7 @@ import { Browser } from "./util/browser.js";
 import {
   ADD_LINK_FUNC,
   BEHAVIOR_LOG_FUNC,
+  FETCH_FUNC,
   DISPLAY,
   ExtractSelector,
   PAGE_OP_TIMEOUT_SECS,
@@ -693,6 +694,7 @@ export class Crawler {
     cdp,
     workerid,
     callbacks,
+    recorder,
     frameIdToExecId,
   }: WorkerOpts) {
     await this.browser.setupPage({ page, cdp });
@@ -767,6 +769,10 @@ self.__bx_behaviors.selectMainBehavior();
         this.behaviorsChecked = true;
       }
 
+      await page.exposeFunction(FETCH_FUNC, (url: string) => {
+        return recorder ? recorder.addExternalFetch(url, cdp) : true;
+      });
+
       await this.browser.addInitScript(page, initScript);
     }
   }

src/util/constants.ts

@@ -24,6 +24,8 @@ export const EXTRACT_TEXT_TYPES = ["to-pages", "to-warc", "final-to-warc"];
 
 export const BEHAVIOR_LOG_FUNC = "__bx_log";
 export const ADD_LINK_FUNC = "__bx_addLink";
+export const FETCH_FUNC = "__bx_fetch";
+
 export const MAX_DEPTH = 1000000;
 
 export const FETCH_HEADERS_TIMEOUT_SECS = 30;

src/util/recorder.ts

@@ -127,6 +127,7 @@ export class Recorder {
   pendingRequests!: Map<string, RequestResponseInfo>;
   skipIds!: Set<string>;
   pageInfo!: PageInfoRecord;
+  mainFrameId: string | null = null;
   skipRangeUrls!: Map<string, number>;
 
   swTargetId?: string | null;
@@ -449,10 +450,14 @@ export class Recorder {
             { url, ...this.logDetails },
             "recorder",
           );
+          reqresp.deleteRange();
+          reqresp.requestId = "0";
+
           const fetcher = new AsyncFetcher({
             reqresp,
+            expectedSize: reqresp.expectedSize ? reqresp.expectedSize : -1,
             recorder: this,
-            networkId: requestId,
+            networkId: "0",
           });
           void this.fetcherQ.add(() => fetcher.load());
           return;
@@ -574,6 +579,12 @@ export class Recorder {
 
     const networkId = params.networkId || requestId;
 
+    const reqresp = this.pendingReqResp(networkId);
+
+    if (!reqresp) {
+      return false;
+    }
+
     if (responseErrorReason) {
       logger.warn(
         "Skipping failed response",
@@ -601,24 +612,23 @@ export class Recorder {
         );
         this.removeReqResp(networkId);
 
-        const reqresp = new RequestResponseInfo("0");
-        reqresp.fillRequest(params.request, params.resourceType);
-        if (reqresp.requestHeaders) {
-          delete reqresp.requestHeaders["range"];
-          delete reqresp.requestHeaders["Range"];
-        }
-        reqresp.frameId = params.frameId;
+        if (!reqresp.fetchContinued) {
+          const reqrespNew = new RequestResponseInfo("0");
+          reqrespNew.fillRequest(params.request, params.resourceType);
+          reqrespNew.deleteRange();
+          reqrespNew.frameId = params.frameId;
 
-        this.addAsyncFetch(
-          {
-            reqresp,
-            expectedSize: parseInt(range.split("/")[1]),
-            recorder: this,
-            networkId: "0",
-            cdp,
-          },
-          contentLen,
-        );
+          this.addAsyncFetch(
+            {
+              reqresp: reqrespNew,
+              expectedSize: parseInt(range.split("/")[1]),
+              recorder: this,
+              networkId: "0",
+              cdp,
+            },
+            contentLen,
+          );
+        }
 
         return false;
       } else {
@@ -651,27 +661,23 @@ export class Recorder {
           "recorder",
         );
 
-        const reqresp = new RequestResponseInfo("0");
-        reqresp.fillRequest(params.request, params.resourceType);
-        reqresp.url = filteredUrl;
-        reqresp.frameId = params.frameId;
+        if (!reqresp.fetchContinued) {
+          const reqrespNew = new RequestResponseInfo("0");
+          reqrespNew.fillRequest(params.request, params.resourceType);
+          reqrespNew.url = filteredUrl;
+          reqrespNew.frameId = params.frameId;
 
-        this.addAsyncFetch({
-          reqresp,
-          recorder: this,
-          networkId: "0",
-          cdp,
-        });
+          this.addAsyncFetch({
+            reqresp: reqrespNew,
+            recorder: this,
+            networkId: "0",
+            cdp,
+          });
+        }
         return false;
       }
     }
 
-    const reqresp = this.pendingReqResp(networkId);
-
-    if (!reqresp) {
-      return false;
-    }
-
     // indicate that this is intercepted in the page context
     if (!isBrowserContext) {
       reqresp.inPageContext = true;
@@ -696,6 +702,7 @@ export class Recorder {
       });
       this.pageInfo.ts = reqresp.ts;
       this.pageInfo.tsStatus = responseStatusCode!;
+      this.mainFrameId = params.frameId;
     }
 
     reqresp.fillFetchRequestPaused(params);
@@ -842,6 +849,22 @@ export class Recorder {
     void this.fetcherQ.add(() => fetcher.load());
   }
 
+  addExternalFetch(url: string, cdp: CDPSession) {
+    const reqresp = new RequestResponseInfo("0");
+    reqresp.url = url;
+    reqresp.method = "GET";
+    reqresp.frameId = this.mainFrameId || undefined;
+    const fetcher = new NetworkLoadStreamAsyncFetcher({
+      reqresp,
+      recorder: this,
+      cdp,
+      networkId: "0",
+    });
+    void this.fetcherQ.add(() => fetcher.load());
+    // return true if successful
+    return true;
+  }
+
   startPage({ pageid, url }: { pageid: string; url: string }) {
     this.pageid = pageid;
     this.pageUrl = url;
@@ -864,6 +887,7 @@ export class Recorder {
       counts: { jsErrors: 0 },
       tsStatus: 999,
     };
+    this.mainFrameId = null;
   }
 
   addPageRecord(reqresp: RequestResponseInfo) {
@@ -1485,6 +1509,7 @@ class AsyncFetcher {
         logger.warn(
           "Async fetch: possible response size mismatch",
           {
+            type: this.constructor.name,
             size: reqresp.readSize,
             expected: reqresp.expectedSize,
             url,
@@ -1492,7 +1517,8 @@ class AsyncFetcher {
           },
           "recorder",
         );
-        if (status === 206) {
+        if (status === 206 || status === 200) {
+          void serializer.externalBuffer?.purge();
           await crawlState.removeDupe(ASYNC_FETCH_DUPE_KEY, url, status);
           return "notfetched";
         }

src/util/reqresp.ts

@@ -335,6 +335,13 @@ export class RequestResponseInfo {
     return this.fromCache && !this.payload;
   }
 
+  deleteRange() {
+    if (this.requestHeaders) {
+      delete this.requestHeaders["range"];
+      delete this.requestHeaders["Range"];
+    }
+  }
+
   shouldSkipSave() {
     // skip cached, OPTIONS/HEAD responses, and 304 responses
     if (

src/util/worker.ts

@@ -27,6 +27,7 @@ export type WorkerOpts = {
   directFetchCapture:
     | ((request: DirectFetchRequest) => Promise<DirectFetchResponse>)
     | null;
+  recorder: Recorder | null;
   markPageUsed: () => void;
   frameIdToExecId: Map<string, number>;
   isAuthSet?: boolean;
@@ -183,6 +184,7 @@ export class PageWorker {
           cdp,
           workerid,
           callbacks: this.callbacks,
+          recorder: this.recorder,
           directFetchCapture,
           frameIdToExecId: new Map<string, number>(),
           markPageUsed: () => {

yarn.lock

@@ -1435,10 +1435,10 @@ browserslist@^4.24.0:
     node-releases "^2.0.18"
     update-browserslist-db "^1.1.1"
 
-browsertrix-behaviors@^0.6.5:
-  version "0.6.5"
-  resolved "https://registry.yarnpkg.com/browsertrix-behaviors/-/browsertrix-behaviors-0.6.5.tgz#a8e3da231caff8e54e34cac6ed3ff431c68b664d"
-  integrity sha512-URUMUPdU0O2J8rmgzrzY4BzT8vv/iYNQUf/B1Eif3ntMsMC51R4/MGgOC8d7pUDCfy5tnOkV1FGlhB9A5LLQrw==
+browsertrix-behaviors@^0.6.6:
+  version "0.6.6"
+  resolved "https://registry.yarnpkg.com/browsertrix-behaviors/-/browsertrix-behaviors-0.6.6.tgz#10bcccfb091c051f5c886d5f69487e6d184078de"
+  integrity sha512-UPNcU9dV0nAvUwJHKKYCkuqdYdlMjK7AWYDyr4xBpSq55xmEh2wQlwQyDyJuUUUrhJNII4NqXK24hVXPdvf5VA==
   dependencies:
     query-selector-shadow-dom "^1.0.1"
 
@@ -5006,10 +5006,10 @@ walker@^1.0.8:
   dependencies:
     makeerror "1.0.12"
 
-warcio@^2.4.0, warcio@^2.4.2:
-  version "2.4.2"
-  resolved "https://registry.yarnpkg.com/warcio/-/warcio-2.4.2.tgz#782d8dcb0769f271b0ae96521fb4969e2570e9b3"
-  integrity sha512-QYbZ3EGYtnAIrzL7Bajo7ak87pipilpkIfaFIzFQWUX4wuXNuKqnfQy/EAoi2tEIl3VJgsWcL+wjjk4+15MKbQ==
+warcio@^2.4.0, warcio@^2.4.2, warcio@^2.4.3:
+  version "2.4.3"
+  resolved "https://registry.yarnpkg.com/warcio/-/warcio-2.4.3.tgz#37ff95c2358d0d5ddb16e924fe200c4774b3903d"
+  integrity sha512-c397HNfLE7yJsyVF3XKXB+Yh3q3WKljhdYRPkKF9eyZMtB+HIxj1aBqgq0nTYz492KMKtzygBo0Gx+Gi0fJ9dg==
   dependencies:
     "@types/pako" "^1.0.7"
     "@types/stream-buffers" "^3.0.7"