Fix extraction completion logic failing due to undefined job

client/src/components/app/extractions/detail.tsx

@@ -35,6 +35,7 @@ import {
   RecipeDetectionStatus,
   concisePrintDate,
   prettyPrintDate,
+  resolveCrawlPageUrl,
   trpc,
 } from "@/utils";
 import { CookingPot, LibraryBig, List } from "lucide-react";
@@ -711,7 +712,10 @@ export default function ExtractionDetail() {
                                       </TableCell>
                                       <TableCell className="break-all align-top">
                                         <a
-                                          href={p.url}
+                                          href={resolveCrawlPageUrl(
+                                            p.url,
+                                            extraction.recipe.url
+                                          )}
                                           target="_blank"
                                           rel="noreferrer"
                                           className="underline"
@@ -784,7 +788,10 @@ export default function ExtractionDetail() {
                                     </TableCell>
                                     <TableCell className="break-all align-top">
                                       <a
-                                        href={p.url}
+                                        href={resolveCrawlPageUrl(
+                                          p.url,
+                                          extraction.recipe.url
+                                        )}
                                         target="_blank"
                                         rel="noreferrer"
                                         className="underline"

client/src/components/app/extractions/page.tsx

@@ -16,7 +16,7 @@ import {
   TableRow,
 } from "@/components/ui/table";
 import { Tabs, TabsContent, TabsList, TabsTrigger } from "@/components/ui/tabs";
-import { concisePrintDate, prettyPrintDate, trpc } from "@/utils";
+import { concisePrintDate, prettyPrintDate, resolveCrawlPageUrl, trpc } from "@/utils";
 import { ExternalLink } from "lucide-react";
 import { useState } from "react";
 import { useParams } from "wouter";
@@ -244,7 +244,14 @@ export default function CrawlPageDetail() {
           <Button
             variant="outline"
             size="sm"
-            onClick={() => window.open(item.crawlPage.url, "_blank", "noopener,noreferrer")}
+            onClick={() => {
+              const url = item.crawlPage.url;
+              const baseUrl = item.crawlPage.extraction?.recipe?.url;
+              const resolved = baseUrl
+                ? resolveCrawlPageUrl(url, baseUrl)
+                : url;
+              window.open(resolved, "_blank", "noopener,noreferrer");
+            }}
           >
             <ExternalLink className="w-4 h-4 mr-2" />
             Open Page URL

client/src/components/app/extractions/step.tsx

@@ -10,7 +10,7 @@ import {
   TableHeader,
   TableRow,
 } from "@/components/ui/table";
-import { concisePrintDate, trpc } from "@/utils";
+import { concisePrintDate, resolveCrawlPageUrl, trpc } from "@/utils";
 import { Link, useParams } from "wouter";
 import { PageType } from "../../../../../common/types";
 import usePagination from "../usePagination";
@@ -85,7 +85,14 @@ export default function CrawlStepDetail() {
                   </TableCell>
                   <TableCell>{concisePrintDate(item.createdAt)}</TableCell>
                   <TableCell className="max-w-40 overflow-hidden whitespace-nowrap text-ellipsis text-blue-800 underline">
-                    <a href={item.url} target="_blank">
+                    <a
+                      href={resolveCrawlPageUrl(
+                        item.url,
+                        extractionQuery.data.recipe.url
+                      )}
+                      target="_blank"
+                      rel="noreferrer"
+                    >
                       {item.url}
                     </a>
                   </TableCell>

client/src/utils.ts

@@ -108,6 +108,19 @@ export function formatCatalogueType(catalogueType: string): string {
   return typeMap[catalogueType] || catalogueType;
 }
 
+/**
+ * Resolves a crawl page URL to an absolute URL. Relative URLs (e.g. /courses/math)
+ * are resolved against the catalogue base URL so they open on the extracted
+ * website rather than the app origin.
+ */
+export function resolveCrawlPageUrl(url: string, baseUrl: string): string {
+  try {
+    return new URL(url, baseUrl).href;
+  } catch {
+    return url;
+  }
+}
+
 export type IterableElement<TargetIterable> =
   TargetIterable extends Iterable<infer ElementType>
   ? ElementType

server/src/data/extractions.ts

@@ -608,6 +608,11 @@ export async function findPage(crawlPageId: number) {
     where: (crawlPages, { eq }) => eq(crawlPages.id, crawlPageId),
     with: {
       crawlStep: true,
+      extraction: {
+        with: {
+          recipe: true,
+        },
+      },
     },
   });
   return result;

server/src/utils.ts

@@ -1,4 +1,4 @@
-import { BrowserFetchError, BrowserTaskResult } from "./extraction/browser";
+import { BrowserFetchError } from "./extraction/browser";
 import getLogger from "./logging";
 
 const logger = getLogger("utils");

server/src/workers/index.ts

@@ -204,6 +204,9 @@ export async function detectExtractionJobs(extractionId: number) {
         break;
       }
       for (const job of jobs) {
+        if (!job || !job.data)
+          continue;
+
         if (job.data.extractionId === extractionId) {
           return true;
         }