Pull request #3509: JS_Norm: distinct arrow functions handling

Merge in SNORT/snort3 from ~ASERBENI/snort3:arrow_scope to master

Squashed commit of the following:

commit fa93f3dd0ff971447de8b2d85876b514a33dee85
Author: Andrii Serbeniuk <aserbeni@cisco.com>
Date:   Mon Jul 11 15:31:19 2022 +0300

    utils: fix arrow functions parsing

diff --git a/src/utils/js_tokenizer.h b/src/utils/js_tokenizer.h
index f0a16027ad528b56892c1039394f52366fc62cee..fd985c6c8d9b35be6a93089a13fcb0710ac40ccc 100644 (file)
--- a/src/utils/js_tokenizer.h
+++ b/src/utils/js_tokenizer.h
@@ -83,9 +83,10 @@ private:
     enum ScopeMetaType
     {
         NOT_SET = 0,
-        FUNCTION,   // function, arrow function
-        BLOCK,      // if, else, for, while, do, with, switch, try, catch, finally, block of code
-        OBJECT,     // object definition, class definition
+        ARROW_FUNCTION, // arrow function
+        FUNCTION,       // function
+        BLOCK,          // if, else, for, while, do, with, switch, try, catch, finally, block of code
+        OBJECT,         // object definition, class definition
         SCOPE_META_TYPE_MAX
     };
     enum FuncType

diff --git a/src/utils/js_tokenizer.l b/src/utils/js_tokenizer.l
index 26b9df90bdf1aa25d62f813d25077e66d62ea432..45a4bcb873c50d63f6210f629cda79d99d224768 100644 (file)
--- a/src/utils/js_tokenizer.l
+++ b/src/utils/js_tokenizer.l
@@ -1352,6 +1352,7 @@ static std::string unescape_unicode(const char* lexeme)
 const char* JSTokenizer::p_scope_codes[] =
 {
     "invalid",
+    "arrow function",
     "function",
     "block",
     "object",
@@ -1863,6 +1864,7 @@ JSTokenizer::JSRet JSTokenizer::scope_push(ScopeType t)
 
         break;
     }
+    case ScopeMetaType::ARROW_FUNCTION: break;
     case ScopeMetaType::BLOCK: break;
     case ScopeMetaType::NOT_SET: break;
     default: assert(false); return BAD_TOKEN;
@@ -2045,6 +2047,7 @@ JSProgramScopeType JSTokenizer::m2p(ScopeMetaType mt)
 {
     switch (mt)
     {
+    case ScopeMetaType::ARROW_FUNCTION:
     case ScopeMetaType::FUNCTION:
         return JSProgramScopeType::FUNCTION;
     case ScopeMetaType::BLOCK:
@@ -2549,7 +2552,7 @@ JSTokenizer::JSRet JSTokenizer::punctuator_arrow()
     set_ident_norm(true);
     if (meta_type() == ScopeMetaType::NOT_SET)
     {
-        set_meta_type(ScopeMetaType::FUNCTION);
+        set_meta_type(ScopeMetaType::ARROW_FUNCTION);
         EXEC(p_scope_push(meta_type()))
     }
     return EOS;

diff --git a/src/utils/test/js_normalizer_test.cc b/src/utils/test/js_normalizer_test.cc
index 5b44df3e0365541a066533c4f87ac15685a55e0f..d6709f6bc0add29bb34fe5d71d873c3ccd49b0a8 100644 (file)
--- a/src/utils/test/js_normalizer_test.cc
+++ b/src/utils/test/js_normalizer_test.cc
@@ -4769,6 +4769,9 @@ TEST_CASE("Scope tracking - basic","[JSNormalizer]")
     SECTION("Function scope - arrow function without scope")
         test_scope("var f = (a,b)=> a",{GLOBAL,FUNCTION});
 
+    SECTION("Function scope - function call in an arrow function without scope")
+        test_scope("var f = (a,b)=> call(",{GLOBAL,FUNCTION});
+
     SECTION("Function scope - method in object initialization")
         test_scope("var o = { f(){",{GLOBAL,BLOCK,BLOCK});
 
@@ -4917,6 +4920,9 @@ TEST_CASE("Scope tracking - closing","[JSNormalizer]")
     SECTION("Function scope - arrow function without scope")
         test_scope("var f = (a,b)=>a;",{GLOBAL});
 
+    SECTION("Function scope - function call in an arrow function without scope")
+        test_scope("var f = a=>call();",{GLOBAL});
+
     SECTION("Function scope - arrow function as a function parameter")
         test_scope("console.log(a=>c)",{GLOBAL});
 

diff --git a/src/utils/test/js_unescape_test.cc b/src/utils/test/js_unescape_test.cc
index 0df3e2c34262a472b0217b8386e5b7b4af2baf52..3b8cdb58a336a984c2387127ce755529b55cf26e 100644 (file)
--- a/src/utils/test/js_unescape_test.cc
+++ b/src/utils/test/js_unescape_test.cc
@@ -1217,5 +1217,24 @@ TEST_CASE("Internal limits", "[JSNormalizer]")
     }
 }
 
+TEST_CASE("Function type detection", "[JSNormalizer]")
+{
+    SECTION("in arrow function")
+    {
+        test_normalization(
+            "var func = () => unescape('%62%61%72');"
+            "func = () => String.fromCodePoint(0x0062, 0x0061, 0x0072);"
+            "func = () => String.fromCharCode(0x0062, 0x0061, 0x0072);"
+            "func = () => decodeURIComponent('%62%61%72');"
+            "func = () => decodeURI('%62%61%72');",
+            "var var_0000=()=>'bar';"
+            "var_0000=()=>'bar';"
+            "var_0000=()=>'bar';"
+            "var_0000=()=>'bar';"
+            "var_0000=()=>'bar';"
+        );
+    }
+}
+
 #endif // CATCH_TEST_BUILD