Nick: partial docs working, cleaner

apps/api/src/controllers/crawl-status.ts

@@ -19,7 +19,7 @@ export async function crawlStatusController(req: Request, res: Response) {
       return res.status(404).json({ error: "Job not found" });
     }
 
-    const { current, current_url, total, current_step } = await job.progress();
+    const { current, current_url, total, current_step, partialDocs } = await job.progress();
     res.json({
       status: await job.getState(),
       // progress: job.progress(),
@@ -28,6 +28,7 @@ export async function crawlStatusController(req: Request, res: Response) {
       current_step: current_step,
       total: total,
       data: job.returnvalue,
+      partial_docs: partialDocs ?? [],
     });
   } catch (error) {
     console.error(error);

apps/api/src/controllers/search.ts

@@ -147,7 +147,7 @@ export async function searchController(req: Request, res: Response) {
     logJob({
       success: result.success,
       message: result.error,
-      num_docs: result.data.length,
+      num_docs: result.data ? result.data.length : 0,
       docs: result.data,
       time_taken: timeTakenInSeconds,
       team_id: team_id,

apps/api/src/lib/entities.ts

@@ -7,6 +7,7 @@ export interface Progress {
     [key: string]: any;
   };
   currentDocumentUrl?: string;
+  currentDocument?: Document;
 }
 
 export type PageOptions = {

apps/api/src/main/runWebScraper.ts

@@ -10,13 +10,15 @@ export async function startWebScraperPipeline({
 }: {
   job: Job<WebScraperOptions>;
 }) {
+  let partialDocs: Document[] = [];
   return (await runWebScraper({
     url: job.data.url,
     mode: job.data.mode,
     crawlerOptions: job.data.crawlerOptions,
     pageOptions: job.data.pageOptions,
     inProgress: (progress) => {
-      job.progress(progress);
+      partialDocs.push(progress.currentDocument);
+      job.progress({...progress, partialDocs: partialDocs});
     },
     onSuccess: (result) => {
       job.moveToCompleted(result);
@@ -69,6 +71,7 @@ export async function runWebScraper({
     }
     const docs = (await provider.getDocuments(false, (progress: Progress) => {
       inProgress(progress);
+      
     })) as Document[];
 
     if (docs.length === 0) {

apps/api/src/scraper/WebScraper/index.ts

@@ -54,6 +54,7 @@ export class WebScraperDataProvider {
               total: totalUrls,
               status: "SCRAPING",
               currentDocumentUrl: url,
+              currentDocument: result
             });
           }
           results[i + index] = result;
@@ -114,9 +115,7 @@ export class WebScraperDataProvider {
   }
 
   private async handleSingleUrlsMode(inProgress?: (progress: Progress) => void): Promise<Document[]> {
-    let documents = await this.convertUrlsToDocuments(this.urls, inProgress);
-    documents = await this.applyPathReplacements(documents);
-    documents = await this.applyImgAltText(documents);
+    let documents = await this.processLinks(this.urls, inProgress);
     return documents;
   }
 
@@ -153,6 +152,13 @@ export class WebScraperDataProvider {
     documents = await this.getSitemapData(this.urls[0], documents);
     documents = this.applyPathReplacements(documents);
     documents = await this.applyImgAltText(documents);
+    
+    if(this.extractorOptions.mode === "llm-extraction" && this.mode === "single_urls") {
+      documents = await generateCompletions(
+        documents,
+        this.extractorOptions
+      )
+    }
     return documents.concat(pdfDocuments);
   }
 
@@ -275,7 +281,7 @@ export class WebScraperDataProvider {
         documents.push(cachedDocument);
 
         // get children documents
-        for (const childUrl of cachedDocument.childrenLinks) {
+        for (const childUrl of (cachedDocument.childrenLinks || [])) {
           const normalizedChildUrl = this.normalizeUrl(childUrl);
           const childCachedDocumentString = await getValue(
             "web-scraper-cache:" + normalizedChildUrl