From 78f58a92a1d84d2036502d03578ae595fe662a88 Mon Sep 17 00:00:00 2001
From: Pavel Pivovarov <pavel.pivovarov@cba.com.au>
Date: Fri, 9 Jan 2026 11:39:27 +1100
Subject: [PATCH] Switched to startpage

---
 tools/search.go | 103 +++++++++++++++++++++++++++++++++++-------------
 1 file changed, 75 insertions(+), 28 deletions(-)

diff --git a/tools/search.go b/tools/search.go
index e3a37ec..d40b460 100644
--- a/tools/search.go
+++ b/tools/search.go
@@ -24,7 +24,7 @@ type SearchResult struct {
 	Content string `json:"content"`
 }
 
-// WebSearch performs a web search using DuckDuckGo
+// WebSearch performs a web search using Startpage
 func WebSearch(query string) (string, error) {
 	ctx, cancel := context.WithTimeout(context.Background(), timeout)
 	defer cancel()
@@ -40,9 +40,9 @@ func WebSearch(query string) (string, error) {
 	}
 
 	// Perform search with retry logic
-	results, err := searchDuckDuckGo(ctx, client, query)
+	results, err := searchStartpage(ctx, client, query)
 	if err != nil {
-		return "", fmt.Errorf("DuckDuckGo search failed: %w", err)
+		return "", fmt.Errorf("Startpage search failed: %w", err)
 	}
 
 	// Format results as text
@@ -68,10 +68,10 @@ func WebSearch(query string) (string, error) {
 	return output.String(), nil
 }
 
-// searchDuckDuckGo performs the actual DuckDuckGo search
-func searchDuckDuckGo(ctx context.Context, client *http.Client, query string) ([]SearchResult, error) {
-	// Build DuckDuckGo search URL
-	searchURL := fmt.Sprintf("https://html.duckduckgo.com/html/?q=%s", url.QueryEscape(query))
+// searchStartpage performs the actual Startpage search
+func searchStartpage(ctx context.Context, client *http.Client, query string) ([]SearchResult, error) {
+	// Build Startpage search URL
+	searchURL := fmt.Sprintf("https://www.startpage.com/sp/search?query=%s", url.QueryEscape(query))
 
 	// Create request with proper headers
 	req, err := http.NewRequestWithContext(ctx, "GET", searchURL, nil)
@@ -85,7 +85,7 @@ func searchDuckDuckGo(ctx context.Context, client *http.Client, query string) ([
 	req.Header.Set("Accept-Language", "en-US,en;q=0.5")
 
 	// Execute request with retry logic
-	resp, err := executeWithRetry(ctx, client, req, "search DuckDuckGo")
+	resp, err := executeWithRetry(ctx, client, req, "search Startpage")
 	if err != nil {
 		return nil, err
 	}
@@ -102,31 +102,78 @@ func searchDuckDuckGo(ctx context.Context, client *http.Client, query string) ([
 		return nil, fmt.Errorf("failed to parse search results: %w", err)
 	}
 
-	// Extract search results
+	// Extract search results from Startpage
 	var results []SearchResult
-	doc.Find(".web-result").Each(func(i int, s *goquery.Selection) {
-		titleNode := s.Find(".result__a")
-		title := strings.TrimSpace(titleNode.Text())
-		info := strings.TrimSpace(s.Find(".result__snippet").Text())
-
-		// Extract URL
-		var resultURL string
-		if titleNode.Length() > 0 {
-			if href, exists := titleNode.Attr("href"); exists {
-				// Validate URL format
-				if strings.HasPrefix(href, "http://") || strings.HasPrefix(href, "https://") {
-					resultURL = href
+	
+	// Find all search result links in the "Web results" section
+	doc.Find("a[href]").Each(func(i int, s *goquery.Selection) {
+		href, exists := s.Attr("href")
+		if !exists {
+			return
+		}
+		
+		// Skip internal Startpage links and Anonymous View links
+		if strings.Contains(href, "startpage.com") || 
+		   strings.Contains(href, "/av/proxy") ||
+		   strings.HasPrefix(href, "#") ||
+		   strings.HasPrefix(href, "/") {
+			return
+		}
+		
+		// Only process HTTP/HTTPS URLs
+		if !strings.HasPrefix(href, "http://") && !strings.HasPrefix(href, "https://") {
+			return
+		}
+		
+		title := strings.TrimSpace(s.Text())
+		if title == "" {
+			return
+		}
+		
+		// Skip very short titles (likely navigation or other non-content links)
+		if len(title) < 10 {
+			return
+		}
+		
+		// Try to find description text near the link
+		var description string
+		parent := s.Parent()
+		for parent.Length() > 0 {
+			// Look for text content in siblings or parent elements
+			text := strings.TrimSpace(parent.Text())
+			if len(text) > len(title)+20 { // Found longer text that includes description
+				// Extract the part that's not the title
+				if idx := strings.Index(text, title); idx >= 0 {
+					remainder := strings.TrimSpace(text[idx+len(title):])
+					if len(remainder) > 20 { // Good description length
+						description = remainder
+						break
+					}
 				}
 			}
+			parent = parent.Parent()
+			if parent.Length() == 0 {
+				break
+			}
 		}
-
-		if title != "" && resultURL != "" {
-			results = append(results, SearchResult{
-				Title:   title,
-				URL:     resultURL,
-				Content: info,
-			})
+		
+		// Limit description length
+		if len(description) > 200 {
+			description = description[:200] + "..."
 		}
+		
+		// Check if we already have this URL (avoid duplicates)
+		for _, existing := range results {
+			if existing.URL == href {
+				return
+			}
+		}
+		
+		results = append(results, SearchResult{
+			Title:   title,
+			URL:     href,
+			Content: description,
+		})
 	})
 
 	return results, nil