fix: return suggested results and add resultType to dataset items (#73)

* return also suggested results, add resultType to output dataset items

* improve readability

Author: MQ37

.actor/actor.json

@@ -16,10 +16,11 @@
                     "title": "Overview",
                     "description": "An view showing just basic properties for simplicity.",
                     "transformation": {
-                        "flatten": ["metadata"],
+                        "flatten": ["metadata", "searchResult"],
                         "fields": [
                             "metadata.url",
                             "metadata.title",
+                            "searchResult.resultType",
                             "markdown"
                         ]
                     },
@@ -34,6 +35,10 @@
                                 "label": "Page title",
                                 "format": "text"
                             },
+                            "searchResult.resultType": {
+                                "label": "Result type",
+                                "format": "text"
+                            },
                             "text": {
                                 "label": "Extracted Markdown",
                                 "format": "text"
@@ -49,6 +54,7 @@
                         "fields": [
                             "searchResult.title",
                             "searchResult.description",
+                            "searchResult.resultType",
                             "searchResult.url"
                         ]
                     },
@@ -63,6 +69,10 @@
                                 "label": "Title",
                                 "format": "text"
                             },
+                            "searchResult.resultType": {
+                                "label": "Result type",
+                                "format": "text"
+                            },
                             "searchResult.url": {
                                 "label": "URL",
                                 "format": "text"

src/google-search/google-extractors-urls.ts

@@ -1,7 +1,7 @@
 import type { CheerioAPI } from 'cheerio';
 import type { Element } from 'domhandler';
 
-import type { OrganicResult } from '../types.js';
+import type { OrganicResult, SearchResultType } from '../types.js';
 
 /**
  * Deduplicates search results based on their title and URL (source @apify/google-search).
@@ -64,7 +64,7 @@ const areTheResultsSuggestions = ($: CheerioAPI) => {
 /**
  * Extracts organic search results from the given Cheerio instance (source: @apify/google-search).
  */
-export const scrapeOrganicResults = ($: CheerioAPI) => {
+export const scrapeOrganicResults = ($: CheerioAPI): OrganicResult[] => {
     const resultSelectors2023January = [
         '.hlcw0c', // Top result with site links
         '.g.Ww4FFb', // General search results
@@ -75,10 +75,14 @@ export const scrapeOrganicResults = ($: CheerioAPI) => {
         '.sATSHe', // another new selector in March 2025
     ];
 
+    const searchResults = extractResultsFromSelectors($, resultSelectors2023January);
+    const deduplicatedResults = deduplicateResults(searchResults);
+    let resultType: SearchResultType = 'ORGANIC';
     if (areTheResultsSuggestions($)) {
-        return [];
+        resultType = 'SUGGESTED';
     }
-
-    const searchResults = extractResultsFromSelectors($, resultSelectors2023January);
-    return deduplicateResults(searchResults);
+    return deduplicatedResults.map((result) => ({
+        ...result,
+        resultType,
+    }));
 };

src/types.ts

@@ -32,11 +32,14 @@ export type Input = {
     scrapingTool: ScrapingTool;
 };
 
+export type SearchResultType = 'ORGANIC' | 'SUGGESTED';
+
 export type OrganicResult = {
     description?: string;
     title?: string;
     rank?: number;
     url?: string;
+    resultType?: SearchResultType;
 };
 
 export interface TimeMeasure {