undo again

Author: jordan-homan

src/hooks/custom/SplitPdfHook.ts

@@ -1,5 +1,6 @@
 import async from "async";
 
+import { HTTPClient } from "../../lib/http.js";
 import {
   AfterErrorContext,
   AfterErrorHook,
@@ -24,12 +25,10 @@ import {
   stringToBoolean,
 } from "./utils/index.js";
 import {
-  HTTPClientExtension,
   MIN_PAGES_PER_THREAD,
   PARTITION_FORM_FILES_KEY,
   PARTITION_FORM_SPLIT_PDF_PAGE_KEY,
 } from "./common.js";
-import {retry, RetryConfig} from "../../lib/retries";
 
 /**
  * Represents a hook for splitting and sending PDF files as per page requests.
@@ -40,7 +39,8 @@ export class SplitPdfHook
   /**
    * The HTTP client used for making requests.
    */
-  client: HTTPClientExtension | undefined;
+  client: HTTPClient | undefined;
+
 
   /**
    * Keeps the strict-mode setting for splitPdfPage feature.
@@ -68,16 +68,9 @@ export class SplitPdfHook
    * @returns The initialized SDK options.
    */
   sdkInit(opts: SDKInitOptions): SDKInitOptions {
-    const { baseURL } = opts;
-    this.client = new HTTPClientExtension();
-
-    this.client.addHook("response", (res) => {
-        if (res.status != 200) {
-            console.error("Request failed with status code", `${res.status}`);
-        }
-    });
-
-    return { baseURL: baseURL, client: this.client };
+    const { baseURL, client } = opts;
+    this.client = client;
+    return { baseURL: baseURL, client: client };
   }
 
   /**
@@ -185,48 +178,23 @@ export class SplitPdfHook
         file.name,
         firstPageNumber
       );
-      const timeoutInMs = 60 * 10 * 1000;
       const req = new Request(requestClone, {
         headers,
         body,
-        signal: AbortSignal.timeout(timeoutInMs)
       });
       requests.push(req);
       setIndex+=1;
     }
 
     this.partitionSuccessfulResponses[operationID] = new Array(requests.length);
-    this.partitionFailedResponses[operationID] = new Array(requests.length);
 
     const allowFailed = this.allowFailed;
 
-    // These are the retry values from our api spec
-    // We need to hardcode them here until we're able to reuse the SDK
-    // from within this hook
-    const oneSecond = 1000;
-    const oneMinute = 1000 * 60;
-    const retryConfig = {
-        strategy: "backoff",
-        backoff: {
-            initialInterval: oneSecond * 3,
-            maxInterval: oneMinute * 12,
-            exponent: 1.88,
-            maxElapsedTime: oneMinute * 30,
-        },
-    } as RetryConfig;
-
-    const retryCodes = ["502", "503", "504"];
-
     this.partitionRequests[operationID] = async.parallelLimit(
-      requests.map((req, pageIndex) => async () => {
+      requests.slice(0, -1).map((req, pageIndex) => async () => {
         const pageNumber = pageIndex + startingPageNumber;
         try {
-         const response = await retry(
-              async () => {
-                return await this.client!.request(req.clone());
-              },
-              { config: retryConfig, statusCodes: retryCodes }
-          );
+          const response = await this.client!.request(req);
           if (response.status === 200) {
             (this.partitionSuccessfulResponses[operationID] as Response[])[pageIndex] =
               response.clone();
@@ -238,7 +206,7 @@ export class SplitPdfHook
               }
           }
         } catch (e) {
-          console.error(`Failed to send request for page ${pageNumber}.`, e);
+          console.error(`Failed to send request for page ${pageNumber}.`);
           if (!allowFailed) {
             throw e;
           }
@@ -247,8 +215,7 @@ export class SplitPdfHook
       concurrencyLevel
     );
 
-    const dummyRequest = new Request("https://no-op/");
-    return dummyRequest;
+    return requests.at(-1) as Request;
   }
 
     /**
@@ -263,39 +230,28 @@ export class SplitPdfHook
                     successfulResponses: Response[],
                     failedResponses: Response[]
   ): Promise<Response>  {
-    let realResponse = response.clone();
-    const firstSuccessfulResponse = successfulResponses.at(0);
-    const isFakeResponse = response.headers.has("fake-response");
-    if (firstSuccessfulResponse !== undefined && isFakeResponse) {
-      realResponse = firstSuccessfulResponse.clone();
-    }
-
     let responseBody, responseStatus, responseStatusText;
     const numFailedResponses = failedResponses?.length ?? 0;
-    const headers = prepareResponseHeaders(realResponse);
+    const headers = prepareResponseHeaders(response);
 
     if (!this.allowFailed && failedResponses && failedResponses.length > 0) {
        const failedResponse = failedResponses[0]?.clone();
        if (failedResponse) {
             responseBody = await failedResponse.text();
+            responseStatus = failedResponse.status;
             responseStatusText = failedResponse.statusText;
         } else {
             responseBody = JSON.stringify({"details:": "Unknown error"});
+            responseStatus = 503
             responseStatusText = "Unknown error"
         }
-        // if the response status is unknown or was 502, 503, 504, set back to 500 to ensure we don't cause more retries
-        responseStatus = 500;
         console.warn(
             `${numFailedResponses} requests failed. The partition operation is cancelled.`
           );
       } else {
-        if (isFakeResponse) {
-          responseBody = await prepareResponseBody([...successfulResponses]);
-        } else {
-          responseBody = await prepareResponseBody([...successfulResponses, response]);
-        }
-        responseStatus = realResponse.status
-        responseStatusText = realResponse.statusText
+        responseBody = await prepareResponseBody([...successfulResponses, response]);
+        responseStatus = response.status
+        responseStatusText = response.statusText
         if (numFailedResponses > 0) {
           console.warn(
             `${numFailedResponses} requests failed. The results might miss some pages.`

src/hooks/custom/common.ts

@@ -1,5 +1,3 @@
-import {HTTPClient} from "../../lib/http";
-
 /**
  * Regular expression pattern for matching base hostnames in the form of "*.unstructuredapp.io".
  */
@@ -25,22 +23,3 @@ export const MAX_NUMBER_OF_PARALLEL_REQUESTS = 15;
 
 export const MIN_PAGES_PER_THREAD = 2;
 export const MAX_PAGES_PER_THREAD = 20;
-
-export class HTTPClientExtension extends HTTPClient {
-  constructor() {
-    super();
-  }
-
-  override async request(request: Request): Promise<Response> {
-    if (request.url === "https://no-op/") {
-      return new Response('{}', {
-        headers: [
-            ["fake-response", "fake-response"]
-        ],
-        status: 200,
-        statusText: 'OK_NO_OP'
-      });
-    }
-    return super.request(request);
-   }
-}