refactor: Improve robots.txt parsing and caching

gemini/robots.go

@@ -5,27 +5,49 @@ import (
 	"strings"
 )
 
-// Takes robots.txt content and a host, and
+func ParseRobotsTxt(content string) *RobotsData {
-// returns a list of full URLs that shouldn't
+    data := &RobotsData{}
-// be visited.
+    var currentUserAgent string
-// TODO Also take into account the user agent?
+
-// Check gemini://geminiprotocol.net/docs/companion/robots.gmi
-func ParseRobotsTxt(content string, host string) []string {
-	var disallowedPaths []string
     for _, line := range strings.Split(content, "\n") {
         line = strings.TrimSpace(line)
-		line = strings.ToLower(line)
+        if line == "" || strings.HasPrefix(line, "#") {
-		if strings.HasPrefix(line, "disallow:") {
+            continue
+        }
+
         parts := strings.SplitN(line, ":", 2)
-			if len(parts) == 2 {
+        if len(parts) != 2 {
-				path := strings.TrimSpace(parts[1])
+            continue
-				if path != "" {
+        }
-					// Construct full Gemini URL
+
-					disallowedPaths = append(disallowedPaths,
+        directive := strings.TrimSpace(strings.ToLower(parts[0]))
-						fmt.Sprintf("gemini://%s%s", host, path))
+        value := strings.TrimSpace(parts[1])
+
+        switch directive {
+        case "user-agent":
+            currentUserAgent = value
+        case "allow":
+            if value != "" {
+                data.Rules = append(data.Rules, RobotRule{
+                    UserAgent: currentUserAgent,
+                    Allow:     true,
+                    Path:      value,
+                })
+            }
+        case "disallow":
+            if value != "" {
+                data.Rules = append(data.Rules, RobotRule{
+                    UserAgent: currentUserAgent,
+                    Allow:     false,
+                    Path:      value,
+                })
+            }
+        case "crawl-delay":
+            if delay, err := strconv.Atoi(value); err == nil {
+                data.CrawlDelay = delay
             }
         }
     }
-	}
+
-	return disallowedPaths
+    return data
 }