使用boost::smatch的valgrind中的读取大小1无效

Question

我正在使用boost:：smatch对象从boost：：regex_match中获取匹配模式，并使用该匹配模式字符串进行进一步操作。它在正常运行中运行良好，但当我使用Valgrind进行内存检查时，它在我读取匹配字符串的行中显示错误，即match.str(1(

我试图通过添加对smatch对象大小的检查来消除这个错误，但错误仍然发生。

//[Modified function name because of policy]
if(boost::regex_match(arg_str,match,dcv::func_name_regex)
&& match.size() > 1) {
std::string func_name = match.str(1);
std::string modified_arg_str = replaceText(arg_str,func_name);
lines.replace(
func_lines.find(arg_str),
arg_str.length(),
modified_arg_str
);
continue;
}

Valgrind输出为：

[Modified some filename and path because of policy]
3201  ==7723== Invalid read of size 1
3202  ==7723==    at 0x517259: snps_boost_1_72_0::sub_match<__gnu_cxx::__normal_iterator<char const*, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > > >::str() const (../../include/boost/regex/v4/sub_match.hpp:84)
3203  ==7723==    by 0x516205: snps_boost_1_72_0::match_results<__gnu_cxx::__normal_iterator<char const*, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > >, std::allocator<snps_boost_1_72_0::sub_match<__gnu_cxx::__normal_iterator<char const*, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > > > > >::str(int) const (../../include/boost/regex/v4/match_results.hpp:207)
3204  ==7723==    by 0x50DDF7: Modifier::replaceText(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >) (/path/to/file.cpp:634)
3205  ==7723==    by 0x50D227: Modifier::functionArgument(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >, unsigned int, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >&) (/path/to/file.cpp:577)

无效读取位于读取匹配字符串的同一行。

std::string func_name = match.str(1);

我使用的是gcc-7.3.0和boost_1_72_0。

Answer 1

使用ASAN/UBSAN/Valgrind跑步值得称赞。你知道为什么它很重要：(

您正在读取未成功匹配的子组。你可能想检查一下它是否做到了，检查size()>1是不够的：

在Coliru上直播

#include <boost/regex.hpp>
#include <iostream>
#include <iomanip>
namespace dcv {
boost::regex func_name_regex("z|(.)");
}
int main() {
boost::smatch match;

for (std::string arg_str: {"a","z"}) {
if(boost::regex_match(arg_str, match, dcv::func_name_regex)) {
std::cout << "Matched: " << std::quoted(arg_str) << " (" << match.size() << ")n";
if (match[1].matched) {
std::string func_name = match.str(1);
std::cout << "Subgroup matched: " << std::quoted(func_name) << "n";
} else {
std::cout << "Subgroup NOT matched.n";
}
}
}
}

打印

Matched: "a" (2)
Subgroup matched: "a"
Matched: "z" (2)
Subgroup NOT matched.