fixed #12378 FN containerOutOfBounds with scope operator / Library: cleaned up usage of Container::startPattern2

[firewave]

No description provided.

[firewave]

This was wrong. The pattern is actually Vector < !!::.

The test now fails at the very first assert in this block so I think we might have a bug on our hands here as the pattern does not seem to make sense as code like Vector<:: should not appear.

Turns out this was a bug in introduced in an optimization in 16ebb90.

[firewave]

Actually - the code was wrong before. So it seems it wasn't my change after all.

The test also initially did not include a < - that was added by @danmar in f131a99.

The initial code to append this was also added by @danmar in 2a82968. The whole changes are unfortunately spread across several commits.

[chrchr-github]

Since this is a SDB test, why not replace &set with &settings1, Vector with std::vector, test::string with std::string?

            ASSERT_EQUALS("signed int", typeOf("std::vector<int> v; v[0]=3;", "[", "test.cpp", &settings1));
            ASSERT_EQUALS("container(std :: string|wstring|u16string|u32string)", typeOf("{return std::string();}", "(", "test.cpp", &settings1));
            ASSERT_EQUALS(
                "container(std :: string|wstring|u16string|u32string)",
                typeOf("void foo(std::vector<std::string> v) { for (auto s: v) { x=s+s; } }", "s", "test.cpp", &settings1));
            ASSERT_EQUALS(
                "container(std :: string|wstring|u16string|u32string)",
                typeOf("void foo(std::vector<std::string> v) { for (auto s: v) { x=s+s; } }", "+", "test.cpp", &settings1));
            ASSERT_EQUALS("container(std :: string|wstring|u16string|u32string) &",
                          typeOf("std::vector<std::string> v; x = v.front();", "(", "test.cpp", &settings1));
            ASSERT_EQUALS("container(std :: string|wstring|u16string|u32string) *",
                          typeOf("std::vector<std::string> v; x = v.data();", "(", "test.cpp", &settings1));
            ASSERT_EQUALS("signed int &", typeOf("std::vector<int> v; x = v.front();", "(", "test.cpp", &settings1));
            ASSERT_EQUALS("signed int *", typeOf("std::vector<int> v; x = v.data();", "(", "test.cpp", &settings1));
            ASSERT_EQUALS("signed int * *", typeOf("std::vector<int*> v; x = v.data();", "(", "test.cpp", &settings1));
            ASSERT_EQUALS("iterator(std :: vector <)", typeOf("std::vector<int> v; x = v.begin();", "(", "test.cpp", &settings1));
            ASSERT_EQUALS("signed int &", typeOf("std::vector<int> v; x = *v.begin();", "*", "test.cpp", &settings1));
            ASSERT_EQUALS("container(std :: string|wstring|u16string|u32string)",
                          typeOf("void foo(){std::string s; return \"x\"+s;}", "+", "test.cpp", &settings1));
            ASSERT_EQUALS("container(std :: string|wstring|u16string|u32string)",
                          typeOf("void foo(){std::string s; return s+\"x\";}", "+", "test.cpp", &settings1));
            ASSERT_EQUALS("container(std :: string|wstring|u16string|u32string)",
                          typeOf("void foo(){std::string s; return 'x'+s;}", "+", "test.cpp", &settings1));
            ASSERT_EQUALS("container(std :: string|wstring|u16string|u32string)",
                          typeOf("void foo(){std::string s; return s+'x';}", "+", "test.cpp", &settings1));

[firewave]

That might be a way. But I am not sure if that will fix the test (will try later).

It still seems there is something wrong here as the startPattern2 will never match in that case.

I assume if startPattern ends with < we should not generate startPattern2 at all. But the question is what we should do in that case. Just skip it or drop the trailing <, append !!:: and match it like the test previously did manually?

The original code before my optimization suggests that something different should have been matched since the Token::Match() was redundant which was most likely a copy&paste error.

[firewave]

The test strangely works when using std::vector which puzzles me. I want to know what the difference is.

Actually. The !!:: in startPattern2 would not make any sense since the last token would always match since it would never occur in real code. So in case it ends with a < we should not be appending ::!! at all.

[firewave]

I found the error in the changes. I forgot to terminate the encoded < character &lt - the ; is missing at the end.

[chrchr-github]

Actually. The !!:: in startPattern2 would not make any sense since the last token would always match since it would never occur in real code. So in case it ends with a < we should not be appending ::!! at all.

This will fix https://trac.cppcheck.net/ticket/12378
Feel free to take the test case from #5910

[firewave]

I have no idea how to fix this. So somebody else needs to take a look at this.

[firewave]

I already verified in the debugger that the values in the Library object after reading the XML are matching the ones we set explicitly beforehand.

If I disable the first assert it will succeed on the second one and fail on the third.