Add method for finding the matching ; corresponding to a reference L

Author: thk123

src/java_bytecode/java_types.cpp

@@ -320,6 +320,39 @@ build_class_name(const std::string src, const std::string &class_name_prefix)
   return java_reference_type(symbol_type);
 }
 
+/// Finds the closing semi-colon ending a ClassTypeSignature that starts at
+/// \p starting_point.
+/// \param src: The input string to work on.
+/// \param starting_point: The string position where the opening 'L' we want to
+///   find the closing ';' for.
+/// \return The string position corresponding to the matching ';'. For example:
+/// LA;, we would return 2. For LA<TT;>; we would return 7.
+/// See unit/java_bytecode/java_util_tests.cpp for more examples.
+size_t find_closing_semi_colon_for_reference_type(
+  const std::string src,
+  size_t starting_point = 0)
+{
+  PRECONDITION(src[starting_point] == 'L');
+  size_t next_semi_colon = src.find(';', starting_point);
+  INVARIANT(
+    next_semi_colon != std::string::npos,
+    "There must be a semi-colon somewhere in the input");
+  size_t next_angle_bracket = src.find('<', starting_point);
+
+  while(next_angle_bracket < next_semi_colon)
+  {
+    size_t end_bracket =
+      find_closing_delimiter(src, next_angle_bracket, '<', '>');
+    INVARIANT(
+      end_bracket != std::string::npos, "Must find matching angle bracket");
+
+    next_semi_colon = src.find(';', end_bracket + 1);
+    next_angle_bracket = src.find('<', end_bracket + 1);
+  }
+
+  return next_semi_colon;
+}
+
 /// Transforms a string representation of a Java type into an internal type
 /// representation thereof.
 ///

unit/java_bytecode/java_utils_test.cpp

@@ -221,3 +221,88 @@ SCENARIO("gather_full_class_name")
     }
   }
 }
+
+SCENARIO("find_closing_semi_colon_for_reference_type", "[core][java_util_test]")
+{
+  GIVEN("A simple reference type")
+  {
+    std::string descriptor = "LA;";
+    //                        |
+    //                      012
+    REQUIRE(find_closing_semi_colon_for_reference_type(descriptor, 0) == 2);
+  }
+  GIVEN("A generic reference type")
+  {
+    std::string descriptor = "LA<TT;>;";
+    //                             |
+    //                      01234567
+    REQUIRE(find_closing_semi_colon_for_reference_type(descriptor, 0) == 7);
+  }
+  GIVEN("A generic reference type with multiple generic params")
+  {
+    std::string descriptor = "LA<TT;TU;>;";
+    //                                |
+    //                      01234567890
+    REQUIRE(find_closing_semi_colon_for_reference_type(descriptor, 0) == 10);
+  }
+  GIVEN("A descriptor with multiple reference type")
+  {
+    std::string descriptor = "LA;LB;";
+    //                        |
+    //                      012
+    REQUIRE(find_closing_semi_colon_for_reference_type(descriptor, 0) == 2);
+  }
+  GIVEN("A descriptor with multiple reference types parsing the second")
+  {
+    std::string descriptor = "LA;LB;";
+    //                         | |
+    //                      012345
+    REQUIRE(find_closing_semi_colon_for_reference_type(descriptor, 3) == 5);
+  }
+  GIVEN("A descriptor inner class")
+  {
+    std::string descriptor = "LA$B;";
+    //                          |
+    //                      01234
+    REQUIRE(find_closing_semi_colon_for_reference_type(descriptor, 0) == 4);
+  }
+  GIVEN("A signature inner class")
+  {
+    std::string descriptor = "LA.B;";
+    //                          |
+    //                      01234
+    REQUIRE(find_closing_semi_colon_for_reference_type(descriptor, 0) == 4);
+  }
+  GIVEN("A inner class of a generic class")
+  {
+    std::string descriptor = "LA<TT;>.B;";
+    //                               |
+    //                      0123456789
+    REQUIRE(find_closing_semi_colon_for_reference_type(descriptor, 0) == 9);
+  }
+  GIVEN("A inner class of a instantiated generic class")
+  {
+    std::string descriptor = "LA<LFoo;>.B;";
+    //                                 |
+    //                      012345678901
+    REQUIRE(find_closing_semi_colon_for_reference_type(descriptor, 0) == 11);
+  }
+  GIVEN(
+    "A signature with multiple references and an inner class of a generic "
+    "class")
+  {
+    std::string descriptor = "LA<TT;>.B;LA<TT;>.B;";
+    //                               |
+    //                      0123456789
+    REQUIRE(find_closing_semi_colon_for_reference_type(descriptor, 0) == 9);
+  }
+  GIVEN(
+    "A signature with multiple references and an inner class of a generic "
+    "class")
+  {
+    std::string descriptor = "LA<TT;>.B;LA<TT;>.B;";
+    //                                |        |
+    //                      01234567890123456789
+    REQUIRE(find_closing_semi_colon_for_reference_type(descriptor, 10) == 19);
+  }
+}