From 501dfb106ad5e025f19b9cc2c88ebea52b3b9480 Mon Sep 17 00:00:00 2001
From: joncrangle <94425204+joncrangle@users.noreply.github.com>
Date: Tue, 7 Nov 2023 21:15:24 -0500
Subject: [PATCH 1/4] feat: enable multiple domains to apply same rule

---
 handlers/proxy.go | 52 +++++++++++++++++++++++++----------------------
 1 file changed, 28 insertions(+), 24 deletions(-)

diff --git a/handlers/proxy.go b/handlers/proxy.go
index 89d8fa8..664e9c8 100644
--- a/handlers/proxy.go
+++ b/handlers/proxy.go
@@ -172,33 +172,37 @@ func applyRules(domain string, path string, body string) string {
 	}
 
 	for _, rule := range rulesSet {
-		if rule.Domain != domain {
-			continue
-		}
-		if len(rule.Paths) > 0 && !StringInSlice(path, rule.Paths) {
-			continue
-		}
-		for _, regexRule := range rule.RegexRules {
-			re := regexp.MustCompile(regexRule.Match)
-			body = re.ReplaceAllString(body, regexRule.Replace)
-		}
-		for _, injection := range rule.Injections {
-			doc, err := goquery.NewDocumentFromReader(strings.NewReader(body))
-			if err != nil {
-				log.Fatal(err)
-			}
-			if injection.Replace != "" {
-				doc.Find(injection.Position).ReplaceWithHtml(injection.Replace)
+		// rule.Domain can be multiple domains delimited by "|"
+		domains := strings.Split(rule.Domain, "|")
+		for _, ruleDomain := range domains {
+			if ruleDomain != domain {
+				continue
 			}
-			if injection.Append != "" {
-				doc.Find(injection.Position).AppendHtml(injection.Append)
+			if len(rule.Paths) > 0 && !StringInSlice(path, rule.Paths) {
+				continue
 			}
-			if injection.Prepend != "" {
-				doc.Find(injection.Position).PrependHtml(injection.Prepend)
+			for _, regexRule := range rule.RegexRules {
+				re := regexp.MustCompile(regexRule.Match)
+				body = re.ReplaceAllString(body, regexRule.Replace)
 			}
-			body, err = doc.Html()
-			if err != nil {
-				log.Fatal(err)
+			for _, injection := range rule.Injections {
+				doc, err := goquery.NewDocumentFromReader(strings.NewReader(body))
+				if err != nil {
+					log.Fatal(err)
+				}
+				if injection.Replace != "" {
+					doc.Find(injection.Position).ReplaceWithHtml(injection.Replace)
+				}
+				if injection.Append != "" {
+					doc.Find(injection.Position).AppendHtml(injection.Append)
+				}
+				if injection.Prepend != "" {
+					doc.Find(injection.Position).PrependHtml(injection.Prepend)
+				}
+				body, err = doc.Html()
+				if err != nil {
+					log.Fatal(err)
+				}
 			}
 		}
 	}

From 84617b32e326d8714d8a49997481e1a8e99e960d Mon Sep 17 00:00:00 2001
From: joncrangle <94425204+joncrangle@users.noreply.github.com>
Date: Tue, 7 Nov 2023 22:44:28 -0500
Subject: [PATCH 2/4] add conde nytimes torstar usatoday washingtonpost

---
 ruleset.yaml | 83 +++++++++++++++++++++++++++++++++++++++++++++++-----
 1 file changed, 76 insertions(+), 7 deletions(-)

diff --git a/ruleset.yaml b/ruleset.yaml
index 8ca8ce7..28af5b6 100644
--- a/ruleset.yaml
+++ b/ruleset.yaml
@@ -4,21 +4,21 @@
       replace: <script $1 script="/https://www.example.com/$3"
   injections:
     - position: head # Position where to inject the code
-      append: | 
+      append: |
         <script>
           window.localStorage.clear();
           console.log("test");
           alert("Hello!");
         </script>
     - position: h1
-      replace: | 
+      replace: |
         <h1>An example with a ladder ;-)</h1>
 - domain: www.americanbanker.com
-  paths: 
+  paths:
     - /news
   injections:
     - position: head
-      append: | 
+      append: |
         <script>
           document.addEventListener("DOMContentLoaded", () => {
             const inlineGate = document.querySelector('.inline-gate');
@@ -30,7 +30,7 @@
           });
         </script>
 - domain: www.nzz.ch
-  paths: 
+  paths:
     - /international
     - /sport
     - /wirtschaft
@@ -46,10 +46,79 @@
     - /finanze
   injections:
     - position: head
-      append: | 
+      append: |
         <script>
           document.addEventListener("DOMContentLoaded", () => {
             const paywall = document.querySelector('.dynamic-regwall');
             removeDOMElement(paywall)
           });
-        </script>
\ No newline at end of file
+        </script>
+- domain: www.architecturaldigest.com|www.bonappetit.com|www.cntraveler.com|www.epicurious.com|www.gq.com|www.newyorker.com|www.vanityfair.com|www.vogue.com|www.wired.com
+  injections:
+    - position: head
+      append: |
+        <script>
+          document.addEventListener("DOMContentLoaded", () => {
+            const banners = document.querySelectorAll('.paywall-bar, div[class^="MessageBannerWrapper-"');
+            banners.forEach(el => { el.remove(); });
+          });
+        </script>
+- domain: www.nytimes.com|www.time.com
+  injections:
+    - position: head
+      append: |
+        <script>
+          window.localStorage.clear();
+          document.addEventListener("DOMContentLoaded", () => {
+            const banners = document.querySelectorAll('div[data-testid="inline-message"], div[id^="ad-"], div[id^="leaderboard-"], div.expanded-dock, div.pz-ad-box, div[id="top-wrapper"], div[id="bottom-wrapper"]');
+            banners.forEach(el => { el.remove(); });
+          });
+        </script>
+- domain: www.thestar.com|www.niagarafallsreview.ca|www.stcatharinesstandard.ca|www.thepeterboroughexaminer.com|www.therecord.com|www.thespec.com|www.wellandtribune.ca
+  injections:
+    - position: head
+      append: |
+        <script>
+          window.localStorage.clear();
+          document.addEventListener("DOMContentLoaded", () => {
+            const paywall = document.querySelectorAll('div.subscriber-offers');
+            paywall.forEach(el => { el.remove(); });
+            const subscriber_only = document.querySelectorAll('div.subscriber-only');
+            for (const elem of subscriber_only) {
+              if (elem.classList.contains('encrypted-content') && dompurify_loaded) {
+                const parser = new DOMParser();
+                const doc = parser.parseFromString('<div>' + DOMPurify.sanitize(unscramble(elem.innerText)) + '</div>', 'text/html');
+                const content_new = doc.querySelector('div');
+                elem.parentNode.replaceChild(content_new, elem);
+              }
+              elem.removeAttribute('style');
+              elem.removeAttribute('class');
+            }
+            const banners = document.querySelectorAll('div.subscription-required, div.redacted-overlay, div.subscriber-hide, div.tnt-ads-container');
+            banners.forEach(el => { el.remove(); });
+            const ads = document.querySelectorAll('div.tnt-ads-container, div[class*="adLabelWrapper"]');
+            ads.forEach(el => { el.remove(); });
+            const recommendations = document.querySelectorAll('div[id^="tncms-region-article"]');
+            recommendations.forEach(el => { el.remove(); });
+          });
+        </script>
+- domain: www.usatoday.com
+  injections:
+    - position: head
+      append: |
+        <script>
+          document.addEventListener("DOMContentLoaded", () => {
+            const banners = document.querySelectorAll('div.roadblock-container, .gnt_nb, [aria-label="advertisement"], div[id="main-frame-error"]');
+            banners.forEach(el => { el.remove(); });
+          });
+        </script>
+- domain: www.washingtonpost.com
+  injections:
+    - position: head
+      append: |
+        <script>
+          document.addEventListener("DOMContentLoaded", () => {
+            let paywall = document.querySelectorAll('div[data-qa$="-ad"], div[id="leaderboard-wrapper"], div[data-qa="subscribe-promo"]');
+            paywall.forEach(el => { el.remove(); });
+          });
+        </script>

From 1aa917e0c1549cf5798146a0954e8f686ecc1db3 Mon Sep 17 00:00:00 2001
From: joncrangle <94425204+joncrangle@users.noreply.github.com>
Date: Wed, 8 Nov 2023 00:14:10 -0500
Subject: [PATCH 3/4] unblur images on washingtonpost

---
 ruleset.yaml | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/ruleset.yaml b/ruleset.yaml
index 28af5b6..8f4796c 100644
--- a/ruleset.yaml
+++ b/ruleset.yaml
@@ -120,5 +120,7 @@
           document.addEventListener("DOMContentLoaded", () => {
             let paywall = document.querySelectorAll('div[data-qa$="-ad"], div[id="leaderboard-wrapper"], div[data-qa="subscribe-promo"]');
             paywall.forEach(el => { el.remove(); });
+            const images = document.querySelectorAll('img');
+            images.forEach(image => { image.parentElement.style.filter = ''; });
           });
         </script>

From 5442da81b9c85497350aaefbf80ece25bc227011 Mon Sep 17 00:00:00 2001
From: Gianni Carafa <gianni.carafa@srf.ch>
Date: Wed, 8 Nov 2023 23:36:06 +0100
Subject: [PATCH 4/4] allow domain list in rules

---
 README.md         |  5 ++++-
 handlers/proxy.go | 14 +++++++++-----
 ruleset.yaml      | 26 +++++++++++++++++++++++---
 3 files changed, 36 insertions(+), 9 deletions(-)

diff --git a/README.md b/README.md
index c2a63c7..25cea80 100644
--- a/README.md
+++ b/README.md
@@ -121,12 +121,15 @@ See in [ruleset.yaml](ruleset.yaml) for an example.
 
 ```yaml
 - domain: www.example.com
+  domains:                     # Additional domains to apply the rule
+    - www.example.com
+    - www.beispiel.de
   regexRules:
     - match: <script\s+([^>]*\s+)?src="(/)([^"]*)"
       replace: <script $1 script="/https://www.example.com/$3"
   injections:
     - position: head # Position where to inject the code
-      append: | 
+      append: |      # possible keys: append, prepend, replace
         <script>
           window.localStorage.clear();
           console.log("test");
diff --git a/handlers/proxy.go b/handlers/proxy.go
index 664e9c8..a39bdcc 100644
--- a/handlers/proxy.go
+++ b/handlers/proxy.go
@@ -155,14 +155,17 @@ func loadRules() RuleSet {
 		yaml.Unmarshal(yamlFile, &ruleSet)
 	}
 
+	domains := []string{}
 	for _, rule := range ruleSet {
-		//log.Println("Loaded rules for", rule.Domain)
+
+		domains = append(domains, rule.Domain)
+		domains = append(domains, rule.Domains...)
 		if os.Getenv("ALLOWED_DOMAINS_RULESET") == "true" {
-			allowedDomains = append(allowedDomains, rule.Domain)
+			allowedDomains = append(allowedDomains, domains...)
 		}
 	}
 
-	log.Println("Loaded rules for", len(ruleSet), "Domains")
+	log.Println("Loaded ", len(ruleSet), " rules for", len(domains), "Domains")
 	return ruleSet
 }
 
@@ -172,8 +175,8 @@ func applyRules(domain string, path string, body string) string {
 	}
 
 	for _, rule := range rulesSet {
-		// rule.Domain can be multiple domains delimited by "|"
-		domains := strings.Split(rule.Domain, "|")
+		domains := rule.Domains
+		domains = append(domains, rule.Domain)
 		for _, ruleDomain := range domains {
 			if ruleDomain != domain {
 				continue
@@ -217,6 +220,7 @@ type Rule struct {
 
 type RuleSet []struct {
 	Domain      string   `yaml:"domain"`
+	Domains     []string `yaml:"domains,omitempty"`
 	Paths       []string `yaml:"paths,omitempty"`
 	GoogleCache bool     `yaml:"googleCache,omitempty"`
 	RegexRules  []Rule   `yaml:"regexRules"`
diff --git a/ruleset.yaml b/ruleset.yaml
index 8f4796c..f7563ec 100644
--- a/ruleset.yaml
+++ b/ruleset.yaml
@@ -1,4 +1,6 @@
 - domain: www.example.com
+  domains: 
+  - www.beispiel.com
   regexRules:
     - match: <script\s+([^>]*\s+)?src="(/)([^"]*)"
       replace: <script $1 script="/https://www.example.com/$3"
@@ -53,7 +55,16 @@
             removeDOMElement(paywall)
           });
         </script>
-- domain: www.architecturaldigest.com|www.bonappetit.com|www.cntraveler.com|www.epicurious.com|www.gq.com|www.newyorker.com|www.vanityfair.com|www.vogue.com|www.wired.com
+- domains: 
+  - www.architecturaldigest.com
+  - www.bonappetit.com
+  - www.cntraveler.com
+  - www.epicurious.com
+  - www.gq.com
+  - www.newyorker.com
+  - www.vanityfair.com
+  - www.vogue.com
+  - www.wired.com
   injections:
     - position: head
       append: |
@@ -63,7 +74,9 @@
             banners.forEach(el => { el.remove(); });
           });
         </script>
-- domain: www.nytimes.com|www.time.com
+- domains: 
+  - www.nytimes.com
+  - www.time.com
   injections:
     - position: head
       append: |
@@ -74,7 +87,14 @@
             banners.forEach(el => { el.remove(); });
           });
         </script>
-- domain: www.thestar.com|www.niagarafallsreview.ca|www.stcatharinesstandard.ca|www.thepeterboroughexaminer.com|www.therecord.com|www.thespec.com|www.wellandtribune.ca
+- domains: 
+  - www.thestar.com
+  - www.niagarafallsreview.ca
+  - www.stcatharinesstandard.ca
+  - www.thepeterboroughexaminer.com
+  - www.therecord.com
+  - www.thespec.com
+  - www.wellandtribune.ca
   injections:
     - position: head
       append: |