Bug 1886953 - Introduce deCOMtaminated IDNA functions and benchmark them r=necko-reviewers,valentin

The plan is to replace the implementations of these functions with an
ICU4X/Rust back end and then make the XPCOM versions call into these.

Migrating existing C++ callers is out of scope for this changeset.

Differential Revision: https://phabricator.services.mozilla.com/D205420

netwerk/base/nsNetUtil.cpp

@@ -693,17 +693,6 @@ int32_t NS_GetDefaultPort(const char* scheme,
   return NS_SUCCEEDED(rv) ? port : -1;
 }
 
-/**
- * This function is a helper function to apply the ToAscii conversion
- * to a string
- */
-bool NS_StringToACE(const nsACString& idn, nsACString& result) {
-  nsCOMPtr<nsIIDNService> idnSrv = do_GetService(NS_IDNSERVICE_CONTRACTID);
-  if (!idnSrv) return false;
-  nsresult rv = idnSrv->ConvertUTF8toACE(idn, result);
-  return NS_SUCCEEDED(rv);
-}
-
 int32_t NS_GetRealPort(nsIURI* aURI) {
   int32_t port;
   nsresult rv = aURI->GetPort(&port);
@@ -721,6 +710,20 @@ int32_t NS_GetRealPort(nsIURI* aURI) {
   return NS_GetDefaultPort(scheme.get());
 }
 
+nsresult NS_DomainToASCII(const nsACString& aHost, nsACString& aASCII) {
+  return nsStandardURL::GetIDNService()->ConvertUTF8toACE(aHost, aASCII);
+}
+
+nsresult NS_DomainToDisplay(const nsACString& aHost, nsACString& aDisplay) {
+  bool ignored;
+  return nsStandardURL::GetIDNService()->ConvertToDisplayIDN(aHost, &ignored,
+                                                             aDisplay);
+}
+
+nsresult NS_DomainToUnicode(const nsACString& aHost, nsACString& aUnicode) {
+  return nsStandardURL::GetIDNService()->ConvertACEtoUTF8(aHost, aUnicode);
+}
+
 nsresult NS_NewInputStreamChannelInternal(
     nsIChannel** outChannel, nsIURI* aUri,
     already_AddRefed<nsIInputStream> aStream, const nsACString& aContentType,

netwerk/base/nsNetUtil.h

@@ -294,10 +294,28 @@ int32_t NS_GetDefaultPort(const char* scheme,
                           nsIIOService* ioService = nullptr);
 
 /**
- * This function is a helper function to apply the ToAscii conversion
- * to a string
+ * The UTS #46 ToASCII operation as parametrized by the WHATWG URL Standard.
+ *
+ * Use this function to prepare a host name for network protocols.
  */
-bool NS_StringToACE(const nsACString& idn, nsACString& result);
+nsresult NS_DomainToASCII(const nsACString& aHost, nsACString& aASCII);
+
+/**
+ * The UTS #46 ToUnicode operation as parametrized by the WHATWG URL Standard,
+ * except potentially misleading labels are treated according to ToASCII
+ * instead.
+ *
+ * Use this function to prepare a host name for display to the user.
+ */
+nsresult NS_DomainToDisplay(const nsACString& aHost, nsACString& aDisplay);
+
+/**
+ * The UTS #46 ToUnicode operation as parametrized by the WHATWG URL Standard.
+ *
+ * It's most likely incorrect to call this function, and `NS_DomainToDisplay`
+ * should typically be called instead.
+ */
+nsresult NS_DomainToUnicode(const nsACString& aHost, nsACString& aUnicode);
 
 /**
  * This function is a helper function to get a protocol's default port if the

netwerk/base/nsStandardURL.cpp

@@ -622,6 +622,8 @@ nsresult nsStandardURL::NormalizeIPv4(const nsACString& host,
   return NS_OK;
 }
 
+nsIIDNService* nsStandardURL::GetIDNService() { return gIDN.get(); }
+
 nsresult nsStandardURL::NormalizeIDN(const nsCString& host, nsCString& result) {
   result.Truncate();
   mDisplayHost.Truncate();

netwerk/base/nsStandardURL.h

@@ -198,6 +198,8 @@ class nsStandardURL : public nsIFileURL,
 
   static nsresult NormalizeIPv4(const nsACString& host, nsCString& result);
 
+  static nsIIDNService* GetIDNService();
+
  protected:
   // enum used in a few places to specify how .ref attribute should be handled
   enum RefHandlingEnum { eIgnoreRef, eHonorRef, eReplaceRef };

netwerk/test/gtest/TestIDNA.cpp

@@ -0,0 +1,77 @@
+#include "gtest/gtest.h"
+#include "gtest/MozGTestBench.h"  // For MOZ_GTEST_BENCH
+#include "gtest/BlackBox.h"
+
+#include "nsNetUtil.h"
+
+#define TEST_COUNT 10000
+
+class TestIDNA : public ::testing::Test {
+ protected:
+  void SetUp() override {
+    // Intentionally Assign and not AssignLiteral
+    // to simulate the usual heap case.
+    mPlainASCII.Assign("example.com");
+    mLeadingDigitASCII.Assign("1test.example");
+    mUnicodeMixed.Assign("مثال.example");
+    mPunycodeMixed.Assign("xn--mgbh0fb.example");
+    mUnicodeLTR.Assign("නම.උදාහරණ");
+    mPunycodeLTR.Assign("xn--r0co.xn--ozc8dl2c3bxd");
+    mUnicodeRTL.Assign("الاسم.مثال");
+    mPunycodeRTL.Assign("xn--mgba0b1dh.xn--mgbh0fb");
+    // Intentionally not assigning to mEmpty
+  }
+
+ public:
+  nsCString mPlainASCII;
+  nsCString mLeadingDigitASCII;
+  nsCString mUnicodeMixed;
+  nsCString mPunycodeMixed;
+  nsCString mUnicodeLTR;
+  nsCString mPunycodeLTR;
+  nsCString mUnicodeRTL;
+  nsCString mPunycodeRTL;
+  nsCString mEmpty;  // Extremely suspicious measurement!
+};
+
+#define IDNA_ITERATIONS 50000
+
+#define IDNA_BENCH(name, func, src)                             \
+  MOZ_GTEST_BENCH_F(TestIDNA, name, [this] {                    \
+    for (int i = 0; i < IDNA_ITERATIONS; i++) {                 \
+      nsCString dst;                                            \
+      func(*mozilla::BlackBox(&src), *mozilla::BlackBox(&dst)); \
+    }                                                           \
+  });
+
+IDNA_BENCH(BenchToASCIIPlainASCII, NS_DomainToASCII, mPlainASCII);
+IDNA_BENCH(BenchToASCIILeadingDigitASCII, NS_DomainToASCII, mLeadingDigitASCII);
+IDNA_BENCH(BenchToASCIIUnicodeMixed, NS_DomainToASCII, mUnicodeMixed);
+IDNA_BENCH(BenchToASCIIPunycodeMixed, NS_DomainToASCII, mPunycodeMixed);
+IDNA_BENCH(BenchToASCIIUnicodeLTR, NS_DomainToASCII, mUnicodeLTR);
+IDNA_BENCH(BenchToASCIIPunycodeLTR, NS_DomainToASCII, mPunycodeLTR);
+IDNA_BENCH(BenchToASCIIUnicodeRTL, NS_DomainToASCII, mUnicodeRTL);
+IDNA_BENCH(BenchToASCIIPunycodeRTL, NS_DomainToASCII, mPunycodeRTL);
+IDNA_BENCH(BenchToASCIIEmpty, NS_DomainToASCII, mEmpty);
+
+IDNA_BENCH(BenchToDisplayPlainASCII, NS_DomainToDisplay, mPlainASCII);
+IDNA_BENCH(BenchToDisplayLeadingDigitASCII, NS_DomainToDisplay,
+           mLeadingDigitASCII);
+IDNA_BENCH(BenchToDisplayUnicodeMixed, NS_DomainToDisplay, mUnicodeMixed);
+IDNA_BENCH(BenchToDisplayPunycodeMixed, NS_DomainToDisplay, mPunycodeMixed);
+IDNA_BENCH(BenchToDisplayUnicodeLTR, NS_DomainToDisplay, mUnicodeLTR);
+IDNA_BENCH(BenchToDisplayPunycodeLTR, NS_DomainToDisplay, mPunycodeLTR);
+IDNA_BENCH(BenchToDisplayUnicodeRTL, NS_DomainToDisplay, mUnicodeRTL);
+IDNA_BENCH(BenchToDisplayPunycodeRTL, NS_DomainToDisplay, mPunycodeRTL);
+IDNA_BENCH(BenchToDisplayEmpty, NS_DomainToDisplay, mEmpty);
+
+IDNA_BENCH(BenchToUnicodePlainASCII, NS_DomainToUnicode, mPlainASCII);
+IDNA_BENCH(BenchToUnicodeLeadingDigitASCII, NS_DomainToUnicode,
+           mLeadingDigitASCII);
+IDNA_BENCH(BenchToUnicodeUnicodeMixed, NS_DomainToUnicode, mUnicodeMixed);
+IDNA_BENCH(BenchToUnicodePunycodeMixed, NS_DomainToUnicode, mPunycodeMixed);
+IDNA_BENCH(BenchToUnicodeUnicodeLTR, NS_DomainToUnicode, mUnicodeLTR);
+IDNA_BENCH(BenchToUnicodePunycodeLTR, NS_DomainToUnicode, mPunycodeLTR);
+IDNA_BENCH(BenchToUnicodeUnicodeRTL, NS_DomainToUnicode, mUnicodeRTL);
+IDNA_BENCH(BenchToUnicodePunycodeRTL, NS_DomainToUnicode, mPunycodeRTL);
+IDNA_BENCH(BenchToUnicodeEmpty, NS_DomainToUnicode, mEmpty);

netwerk/test/gtest/moz.build

@@ -17,6 +17,7 @@ UNIFIED_SOURCES += [
     "TestHttpAuthUtils.cpp",
     "TestHttpChannel.cpp",
     "TestHttpResponseHead.cpp",
+    "TestIDNA.cpp",
     "TestInputStreamTransport.cpp",
     "TestIsValidIp.cpp",
     "TestLinkHeader.cpp",