From 96dd4de8768b2b64085b848cbe21a77794f9c20f Mon Sep 17 00:00:00 2001
From: Gianni Carafa <gianni.carafa@srf.ch>
Date: Fri, 3 Nov 2023 21:46:57 +0100
Subject: [PATCH] improve Rules file

---
 README.md         |  1 +
 handlers/proxy.go | 20 +++++++++++---------
 ruleset.yaml      | 20 +++++++++++++-------
 3 files changed, 25 insertions(+), 16 deletions(-)

diff --git a/README.md b/README.md
index fca6e45..72c019e 100644
--- a/README.md
+++ b/README.md
@@ -82,3 +82,4 @@ http://localhost:8080/raw/https://www.google.com
 | `USERPASS` | Enables Basic Auth, format `admin:123456` | `` |
 | `LOG_URLS` | Log fetched URL's | `true` |
 | `DISABLE_FORM` | Disables URL Form Frontpage | `false` |
+| `RULES_URL` | URL to a ruleset file | `https://raw.githubusercontent.com/kubero-dev/ladder/main/ruleset.yaml` |
diff --git a/handlers/proxy.go b/handlers/proxy.go
index 6fe4462..1e0d4b9 100644
--- a/handlers/proxy.go
+++ b/handlers/proxy.go
@@ -97,7 +97,10 @@ func rewriteHtml(bodyB []byte, u *url.URL) string {
 	body = strings.ReplaceAll(body, "url(/", "url(/https://"+u.Host+"/")
 	body = strings.ReplaceAll(body, "href=\"https://"+u.Host, "href=\"/https://"+u.Host+"/")
 
-	body = applyRules(u.Host, u.Path, body)
+	if os.Getenv("RULES_URL") != "" {
+		log.Println("Applying rules")
+		body = applyRules(u.Host, u.Path, body)
+	}
 	return body
 }
 
@@ -123,17 +126,21 @@ func loadRules() RuleSet {
 	}
 	defer resp.Body.Close()
 
-	bodyB, err := io.ReadAll(resp.Body)
+	if resp.StatusCode >= 400 {
+		log.Println("ERROR:", resp.StatusCode, rulesUrl)
+	}
+
+	body, err := io.ReadAll(resp.Body)
 	if err != nil {
 		log.Println("ERROR:", err)
 	}
 
 	var ruleSet RuleSet
-	yaml.Unmarshal(bodyB, &ruleSet)
+	yaml.Unmarshal(body, &ruleSet)
 	if err != nil {
 		log.Println("ERROR:", err)
 	}
-
+	log.Println(ruleSet)
 	return ruleSet
 }
 
@@ -149,11 +156,6 @@ func applyRules(domain string, path string, body string) string {
 		if rule.Path != "" && rule.Path != path {
 			continue
 		}
-		/*
-			for _, domRule := range rule.DomRules {
-				// run the dom rules
-			}
-		*/
 		for _, regexRule := range rule.RegexRules {
 			re := regexp.MustCompile(regexRule.Match)
 			body = re.ReplaceAllString(body, regexRule.Replace)
diff --git a/ruleset.yaml b/ruleset.yaml
index fe1eeed..cc03f01 100644
--- a/ruleset.yaml
+++ b/ruleset.yaml
@@ -1,15 +1,21 @@
 - domain: www.example.com
-  path: /article
   regexRules:
     - match: <script\s+([^>]*\s+)?src="(/)([^"]*)"
       replace: <script $1 script="/https://www.example.com/$3"
-  domRules:
-    - match: "tobe.defined"
-      replace: "test"
+  injectCode: | 
+    <script>
+      window.localStorage.clear();
+      console.log("test");
+    </script>
 - domain: www.anotherdomain.com
+  path: /article
+  googleCache: false
   regexRules:
     - match: <script\s+([^>]*\s+)?src="(/)([^"]*)"
       replace: <script $1 script="/https://www.example.com/$3"
-  domRules:
-    - match: "tobe.defined"
-      replace: "test"
\ No newline at end of file
+  injectCode: | 
+    <script>
+      window.localStorage.clear();
+      console.log("test");
+    </script>
+  
\ No newline at end of file