Merge pull request #5506 from rspamd/vstakhov-idna-tests

[Test] Detect libicu idna behaviour and select the appropriate tests

1 files changed, 48 insertions, 14 deletions

diff --git a/test/lua/unit/rspamd_resolver.lua b/test/lua/unit/rspamd_resolver.lua
index e987ff00b..2fdec2c4b 100644
--- a/test/lua/unit/rspamd_resolver.lua
+++ b/test/lua/unit/rspamd_resolver.lua
@@ -6,24 +6,58 @@ context("Check punycoding UTF-8 URL", function()
 
   local resolver = rspamd_resolver.init(rspamd_util.create_event_base(), rspamd_config)
 
-  local cases = {
-    -- https://unicode.org/reports/tr46/#Deviations
-    ['faß.de'] = 'fass.de', -- IDNA2008 result: xn--fa-hia.de
-    ['βόλος.com'] = 'xn--nxasmq6b.com', -- IDNA2008 result: xn--nxasmm1c.com
-    ['نامه‌ای.com'] = 'xn--mgba3gch31f.com', -- IDNA2008 result: xn--mgba3gch31f060k.com
-    ['ශ්‍රී.com'] = 'xn--10cl1a0b.com', -- IDNA2008 result: xn--10cl1a0b660p.com 
-
-    -- https://unicode.org/reports/tr46/#Table_Example_Processing
-    ['日本語。ＪＰ'] = 'xn--wgv71a119e.jp', -- Fullwidth characters are remapped, including 。
-    --['u¨.com'] = 'xn--tda.com', -- Normalize changes u + umlaut to ü
-    ['☕.us'] = 'xn--53h.us', -- Post-Unicode 3.2 characters are allowed
-
-    -- Other
+  -- Helper function to detect IDNA behavior by testing a known conversion
+  local function detect_idna_behavior()
+    -- Use faß.de as a test case - different results in IDNA2003 vs IDNA2008
+    local test_result = resolver:idna_convert_utf8('faß.de')
+    if test_result == 'fass.de' then
+      return 'transitional' -- IDNA2003/transitional behavior
+    elseif test_result == 'xn--fa-hia.de' then
+      return 'nontransitional' -- IDNA2008/nontransitional behavior
+    else
+      return 'unknown'
+    end
+  end
+
+  local idna_behavior = detect_idna_behavior()
+
+  -- Define test cases with both expected results
+  local cases_transitional = {
+    -- IDNA2003/transitional results (ICU < 76 default)
+    ['faß.de'] = 'fass.de',
+    ['βόλος.com'] = 'xn--nxasmq6b.com',
+    ['نامه‌ای.com'] = 'xn--mgba3gch31f.com',
+    ['ශ්‍රී.com'] = 'xn--10cl1a0b.com',
+    ['日本語。ＪＰ'] = 'xn--wgv71a119e.jp',
+    ['☕.us'] = 'xn--53h.us',
+    ['example.рф'] = 'example.xn--p1ai',
+  }
+
+  local cases_nontransitional = {
+    -- IDNA2008/nontransitional results (ICU >= 76 default)
+    ['faß.de'] = 'xn--fa-hia.de',
+    ['βόλος.com'] = 'xn--nxasmm1c.com',
+    ['نامه‌ای.com'] = 'xn--mgba3gch31f060k.com',
+    ['ශ්‍රී.com'] = 'xn--10cl1a0b660p.com',
+    ['日本語。ＪＰ'] = 'xn--wgv71a119e.jp',
+    ['☕.us'] = 'xn--53h.us',
     ['example.рф'] = 'example.xn--p1ai',
   }
 
+  -- Choose appropriate test cases based on detected behavior
+  local cases
+  if idna_behavior == 'transitional' then
+    cases = cases_transitional
+    print("Detected IDNA transitional behavior (ICU < 76 or configured for IDNA2003)")
+  elseif idna_behavior == 'nontransitional' then
+    cases = cases_nontransitional
+    print("Detected IDNA nontransitional behavior (ICU >= 76 default)")
+  else
+    error("Could not detect IDNA behavior - unexpected result for test case")
+  end
+
   for k, v in pairs(cases) do
-    test(string.format("punycode %s -> %s", k, v), function()
+    test(string.format("punycode %s -> %s (%s)", k, v, idna_behavior), function()
       local res = resolver:idna_convert_utf8(k)
       assert_equal(res, v)
     end)