Bug 1358225: Validate characters with a cached lookup array. About 10-15% improvement on TestStandardURL.Perf gtest. r=valentin.gosu

MozReview-Commit-ID: KMABJ3X6IZ1

--HG--
extra : rebase_source : c5bde6642a193e47628226a38d6ec9d8366aaa90

netwerk/base/nsStandardURL.cpp

@@ -6,6 +6,7 @@
 
 #include "IPCMessageUtils.h"
 
+#include "nsASCIIMask.h"
 #include "nsStandardURL.h"
 #include "nsCRT.h"
 #include "nsEscape.h"
@@ -142,6 +143,25 @@ nsIIDNService *nsStandardURL::gIDN = nullptr;
 bool nsStandardURL::gInitialized = false;
 char nsStandardURL::gHostLimitDigits[] = { '/', '\\', '?', '#', 0 };
 
+// Invalid host characters
+// We still allow % because it is in the ID of addons.
+// Any percent encoded ASCII characters that are not allowed in the
+// hostname are not percent decoded, and will be parsed just fine.
+//
+// Note that the array below will be initialized at compile time,
+// so we do not need to "optimize" TestForInvalidHostCharacters.
+//
+constexpr bool TestForInvalidHostCharacters(char c)
+{
+    // Testing for these:
+    // CONTROL_CHARACTERS " #/:?@[\\]*<>|\"";
+    return (c > 0 && c < 32) || // The control characters are [1, 31]
+           c == ' ' || c == '#' || c == '/' || c == ':' || c == '?' ||
+           c == '@' || c == '[' || c == '\\' || c == ']' || c == '*' ||
+           c == '<' || c == '>' || c == '|' || c == '"';
+}
+constexpr ASCIIMaskArray sInvalidHostChars = CreateASCIIMask(TestForInvalidHostCharacters);
+
 //----------------------------------------------------------------------------
 
 #define ENSURE_MUTABLE() \
@@ -690,14 +710,13 @@ nsStandardURL::ValidIPv6orHostname(const char *host, uint32_t length)
         return false;
     }
 
-    const char *end = host + length;
-    if (end != net_FindCharInSet(host, end, CONTROL_CHARACTERS " #/:?@[\\]*<>|\"")) {
-        // We still allow % because it is in the ID of addons.
-        // Any percent encoded ASCII characters that are not allowed in the
-        // hostname are not percent decoded, and will be parsed just fine.
+    const char* end = host + length;
+    const char* iter = host;
+    for (; iter != end && *iter; ++iter) {
+        if (ASCIIMask::IsMasked(sInvalidHostChars, *iter)) {
             return false;
         }
-
+    }
     return true;
 }