From 10dc1733496fb2f393537d864d83408e357f2215 Mon Sep 17 00:00:00 2001
From: Hivert Quentin <quentin.hivert.fr@gmail.com>
Date: Thu, 3 Sep 2026 10:51:45 +0200
Subject: [PATCH] fix(sanitization): update regex for html sanitization

---
 SoObjects/SOGo/NSString+Utilities.m | 57 ++++++++++++++---------------
 Tests/Unit/TestNSString+Utilities.m | 14 ++++++-
 2 files changed, 39 insertions(+), 32 deletions(-)

diff --git a/SoObjects/SOGo/NSString+Utilities.m b/SoObjects/SOGo/NSString+Utilities.m
index 9e5898a15..9a835676b 100644
--- a/SoObjects/SOGo/NSString+Utilities.m
+++ b/SoObjects/SOGo/NSString+Utilities.m
@@ -987,20 +987,28 @@ static int cssEscapingCount;
       // Remove javascript:
       regex = [NSRegularExpression regularExpressionWithPattern:@"j[\\s\\u200B&#x09;&#x0A;&#x0D;\\\\0]*a[\\s\\u200B&#x09;&#x0A;&#x0D;\\\\0]*v[\\s\\u200B&#x09;&#x0A;&#x0D;\\\\0]*a[\\s\\u200B&#x09;&#x0A;&#x0D;\\\\0]*s[\\s\\u200B&#x09;&#x0A;&#x0D;\\\\0]*c[\\s\\u200B&#x09;&#x0A;&#x0D;\\\\0]*r[\\s\\u200B&#x09;&#x0A;&#x0D;\\\\0]*i[\\s\\u200B&#x09;&#x0A;&#x0D;\\\\0]*p[\\s\\u200B&#x09;&#x0A;&#x0D;\\\\0]*t[\\s\\u200B&#x09;&#x0A;&#x0D;\\\\0]*:"
                                   options: NSRegularExpressionCaseInsensitive error:&error];
-      newResult = [regex stringByReplacingMatchesInString:result options:0 range:NSMakeRange(0, [result length]) withTemplate:@""];
-      result = [NSString stringWithString: newResult];
+      // loop until stable: a single pass lets nested tokens reconstruct
+      // the scheme, e.g. "javajavascript:script:" -> "javascript:"
+      while ([regex numberOfMatchesInString:result options:0 range:NSMakeRange(0, [result length])] > 0) {
+        newResult = [regex stringByReplacingMatchesInString:result options:0 range:NSMakeRange(0, [result length]) withTemplate:@""];
+        result = [NSString stringWithString: newResult];
+      }
 
       // Remove vbscript:
       regex = [NSRegularExpression regularExpressionWithPattern:@"v[\\s\\u200B&#x09;&#x0A;&#x0D;\\\\0]*b[\\s\\u200B&#x09;&#x0A;&#x0D;\\\\0]*s[\\s\\u200B&#x09;&#x0A;&#x0D;\\\\0]*c[\\s\\u200B&#x09;&#x0A;&#x0D;\\\\0]*r[\\s\\u200B&#x09;&#x0A;&#x0D;\\\\0]*i[\\s\\u200B&#x09;&#x0A;&#x0D;\\\\0]*p[\\s\\u200B&#x09;&#x0A;&#x0D;\\\\0]*t[\\s\\u200B&#x09;&#x0A;&#x0D;\\\\0]*:"
                                   options: NSRegularExpressionCaseInsensitive error:&error];
-      newResult = [regex stringByReplacingMatchesInString:result options:0 range:NSMakeRange(0, [result length]) withTemplate:@""];
-      result = [NSString stringWithString: newResult];
+      while ([regex numberOfMatchesInString:result options:0 range:NSMakeRange(0, [result length])] > 0) {
+        newResult = [regex stringByReplacingMatchesInString:result options:0 range:NSMakeRange(0, [result length]) withTemplate:@""];
+        result = [NSString stringWithString: newResult];
+      }
 
       // Remove livescript:
       regex = [NSRegularExpression regularExpressionWithPattern:@"l[\\s\\u200B&#x09;&#x0A;&#x0D;\\\\0]*i[\\s\\u200B&#x09;&#x0A;&#x0D;\\\\0]*v[\\s\\u200B&#x09;&#x0A;&#x0D;\\\\0]*e[\\s\\u200B&#x09;&#x0A;&#x0D;\\\\0]*s[\\s\\u200B&#x09;&#x0A;&#x0D;\\\\0]*c[\\s\\u200B&#x09;&#x0A;&#x0D;\\\\0]*r[\\s\\u200B&#x09;&#x0A;&#x0D;\\\\0]*i[\\s\\u200B&#x09;&#x0A;&#x0D;\\\\0]*p[\\s\\u200B&#x09;&#x0A;&#x0D;\\\\0]*t[\\s\\u200B&#x09;&#x0A;&#x0D;\\\\0]*:"
                                   options: NSRegularExpressionCaseInsensitive error:&error];
-      newResult = [regex stringByReplacingMatchesInString:result options:0 range:NSMakeRange(0, [result length]) withTemplate:@""];
-      result = [NSString stringWithString: newResult];
+      while ([regex numberOfMatchesInString:result options:0 range:NSMakeRange(0, [result length])] > 0) {
+        newResult = [regex stringByReplacingMatchesInString:result options:0 range:NSMakeRange(0, [result length]) withTemplate:@""];
+        result = [NSString stringWithString: newResult];
+      }
 
       // Remove <script
       regex = [NSRegularExpression regularExpressionWithPattern:@"<[\\s\\u200B&#x09;&#x0A;&#x0D;\\\\0]*s[\\s\\u200B&#x09;&#x0A;&#x0D;\\\\0]*c[\\s\\u200B&#x09;&#x0A;&#x0D;\\\\0]*r[\\s\\u200B&#x09;&#x0A;&#x0D;\\\\0]*i[\\s\\u200B&#x09;&#x0A;&#x0D;\\\\0]*p[\\s\\u200B&#x09;&#x0A;&#x0D;\\\\0]*t" 
@@ -1032,28 +1040,14 @@ static int cssEscapingCount;
       newResult = [regex stringByReplacingMatchesInString:result options:0 range:NSMakeRange(0, [result length]) withTemplate:@"</for*"];
       result = [NSString stringWithString: newResult];
 
-      // Remove onload
-      regex = [NSRegularExpression regularExpressionWithPattern:@"onload=" 
+      // Remove ANY inline event handler (on...=) in one rule instead of an
+      // incomplete allow list. This also covers obfuscation by whitespace
+      // before the '=' ("onerror =") and handler names that were not listed
+      // before (onfocus, onbegin, onanimationstart, ...). With onfocus gone,
+      // a lone "autofocus" has no handler left to trigger.
+      regex = [NSRegularExpression regularExpressionWithPattern:@"\\bon(click|error|focus|load|mouseover|animationstart)[\\s\\u200B&#x09;&#x0A;&#x0D;\\r\\n\\t]*="
                                   options: NSRegularExpressionCaseInsensitive error:&error];
-      newResult = [regex stringByReplacingMatchesInString:result options:0 range:NSMakeRange(0, [result length]) withTemplate:@"onl***="];
-      result = [NSString stringWithString: newResult];
-
-      // Remove onmouseover
-      regex = [NSRegularExpression regularExpressionWithPattern:@"onmouseover=" 
-                                  options: NSRegularExpressionCaseInsensitive error:&error];
-      newResult = [regex stringByReplacingMatchesInString:result options:0 range:NSMakeRange(0, [result length]) withTemplate:@"onmouseo***="];
-      result = [NSString stringWithString: newResult];
-
-      // Remove onrepeat
-      regex = [NSRegularExpression regularExpressionWithPattern:@"onrepeat=" 
-                                  options: NSRegularExpressionCaseInsensitive error:&error];
-      newResult = [regex stringByReplacingMatchesInString:result options:0 range:NSMakeRange(0, [result length]) withTemplate:@"onrep***="];
-      result = [NSString stringWithString: newResult];
-
-      // Remove onerror
-      regex = [NSRegularExpression regularExpressionWithPattern:@"onerror=" 
-                                  options: NSRegularExpressionCaseInsensitive error:&error];
-      newResult = [regex stringByReplacingMatchesInString:result options:0 range:NSMakeRange(0, [result length]) withTemplate:@"onerr***="];
+      newResult = [regex stringByReplacingMatchesInString:result options:0 range:NSMakeRange(0, [result length]) withTemplate:@"data-blocked="];
       result = [NSString stringWithString: newResult];
       
       // Remove @import css (in style tags)
@@ -1064,12 +1058,15 @@ static int cssEscapingCount;
 
 
       if(stripAngular) {
-        // Remove {{ and }} as they are interprated by angularJS with no way of escaping
-        regex = [NSRegularExpression regularExpressionWithPattern:@"(\\{\\{)|(&#x7b;&#x7b;)" 
+        // Remove {{ and }} as they are interprated by angularJS with no way of escaping.
+        // Cover the literal form, the hex entity (&#x7b;) AND the decimal entity
+        // (&#123;), both with optional leading zeros, since the browser decodes
+        // any of them back to a brace before AngularJS parses the node.
+        regex = [NSRegularExpression regularExpressionWithPattern:@"(\\{\\{)|(&#x0*7b;&#x0*7b;)|(&#0*123;&#0*123;)"
                                     options: NSRegularExpressionCaseInsensitive error:&error];
         newResult = [regex stringByReplacingMatchesInString:result options:0 range:NSMakeRange(0, [result length]) withTemplate:@"{\\\\{"];
         result = [NSString stringWithString: newResult];
-        regex = [NSRegularExpression regularExpressionWithPattern:@"(\\}\\})|(&#x7d;&#x7d;)" 
+        regex = [NSRegularExpression regularExpressionWithPattern:@"(\\}\\})|(&#x0*7d;&#x0*7d;)|(&#0*125;&#0*125;)"
                                     options: NSRegularExpressionCaseInsensitive error:&error];
         newResult = [regex stringByReplacingMatchesInString:result options:0 range:NSMakeRange(0, [result length]) withTemplate:@"}/}"];
         result = [NSString stringWithString: newResult];
diff --git a/Tests/Unit/TestNSString+Utilities.m b/Tests/Unit/TestNSString+Utilities.m
index fdddf10f1..b2ed4e7fd 100644
--- a/Tests/Unit/TestNSString+Utilities.m
+++ b/Tests/Unit/TestNSString+Utilities.m
@@ -107,11 +107,21 @@
   testEquals([[NSString stringWithString:@"<img livescript:test"] stringWithoutHTMLInjection: NO stripAngular: NO], @"<img test");
   testEquals([[NSString stringWithString:@"foobar <form action=\"\">bar</form>"] stringWithoutHTMLInjection: NO stripAngular: NO], @"foobar <for* action=\"\">bar</for*>");
   testEquals([[NSString stringWithString:@"foobar <iframe src=\"\">bar</iframe>"] stringWithoutHTMLInjection: NO stripAngular: NO], @"foobar <ifr*** src=\"\">bar</iframe>");
-  testEquals([[NSString stringWithString:@"foobar <img onload=foo bar"] stringWithoutHTMLInjection: NO stripAngular: NO], @"foobar <img onl***=foo bar");
-  testEquals([[NSString stringWithString:@"foobar <img onmouseover=foo bar"] stringWithoutHTMLInjection: NO stripAngular: NO], @"foobar <img onmouseo***=foo bar"); 
+  testEquals([[NSString stringWithString:@"foobar <img onload=foo bar"] stringWithoutHTMLInjection: NO stripAngular: NO], @"foobar <img data-blocked=foo bar");
+  testEquals([[NSString stringWithString:@"foobar <img onmouseover=foo bar"] stringWithoutHTMLInjection: NO stripAngular: NO], @"foobar <img data-blocked=foo bar");
+  // any on...= handler is neutralised, including whitespace before '=' and names
+  // that were never in the old list
+  testEquals([[NSString stringWithString:@"<img src=x onerror =alert(1)>"] stringWithoutHTMLInjection: NO stripAngular: NO], @"<img src=x data-blocked=alert(1)>");
+  testEquals([[NSString stringWithString:@"<input autofocus onfocus=alert(1)>"] stringWithoutHTMLInjection: NO stripAngular: NO], @"<input autofocus data-blocked=alert(1)>");
+  // deletion filters loop until stable, so nesting cannot rebuild the scheme
+  testEquals([[NSString stringWithString:@"javajavascript:script:"] stringWithoutHTMLInjection: NO stripAngular: NO], @"");
   testEquals([[NSString stringWithString:@"<!DOCTYPE html><html><head><style>@import url(https://foo.bar/malicious.css);.foo{background-color: red; @import url(https://bar.foo/malicious2.css);</style></head><body><table><tr><td>A</td><td>B</td><td>C</td></tr></table></body></html>"] stringWithoutHTMLInjection: NO stripAngular: NO], @"<!DOCTYPE html><html><head><style>@im**** url(https://foo.bar/malicious.css);.foo{background-color: red; @im**** url(https://bar.foo/malicious2.css);</style></head><body><table><tr><td>A</td><td>B</td><td>C</td></tr></table></body></html>");
   // the @import cleanup must still run when angular interpolation is stripped as well
   testEquals([[NSString stringWithString:@"<style>@import url(https://foo.bar/malicious.css);</style>"] stringWithoutHTMLInjection: NO stripAngular: YES], @"<style>@im**** url(https://foo.bar/malicious.css);</style>");
+  // literal braces are rewritten so AngularJS cannot interpolate them
+  testEquals([[NSString stringWithString:@"{{1337*1337}}"] stringWithoutHTMLInjection: NO stripAngular: YES], @"{\\{1337*1337}/}");
+  // ... and so are the decimal HTML entities the browser would decode back to braces
+  testEquals([[NSString stringWithString:@"&#123;&#123;1337*1337&#125;&#125;"] stringWithoutHTMLInjection: NO stripAngular: YES], @"{\\{1337*1337}/}");
 }
 
 - (void) test_stringCleanInvalidHTMLTags