使用MSVC 2013正则表达式错误 [英] Regex error using MSVC 2013

问题描述

我有一块xml像代码来解析，在MSVC 2013中使用std :: regex

 < GLVertex& 
 #version 450 core 
 layout（location = 0）in vec3 pos; 
 in VertexInfo {
 vec2 uv; 
} vertexInfo; 
 void main（）{
 gl_Position = vec4（pos，1.0）; 
 vertexInfo.uv = pos.xy; 
} 
< GLVertex /> 
< GLFragment> 
 #version 450 core 
 layout（location = 0）uniform sampler2D g_map; 
 uniform uniform {
 vec4 color; 
}; 
 layout（location = 0）out vec4 fragColor; 
 void main（）{
 fragColor = texture（g_map，vertexInfo.uv）; 
} 
< GLFragment /> 
  

以下是模式：

 < GLVertex>（（。| \\\
）+）< GLVertex\ /> 
  

但程序总是崩溃！在我的正则表达式有任何错误？我在regex101测试。

PS。当我删除第五行：

  vec2 uv; 
  

它工作正常！

方案

由于模式不高效，你会得到堆栈溢出（参数：0x00000001，0x00312FFC）异常。我认为它是如何 std :: regex 处理重复的组（你已经定义一个 + - 量化组（。| \\\
）+ ）。此模式匹配不是换行符（。）或换行符（ \\\
）的每个字符，然后将该匹配存储在缓冲区中。然后，只有在调试模式中，才会发生迭代器调试问题。 std :: _ Orphan_Me 是发生断点的地方，在匹配字符串时，它被认为是最昂贵的方法。请参见

I have a piece of xml like code to parse, using std::regex in MSVC 2013

<GLVertex>
#version 450 core
layout(location = 0) in vec3 pos;
in VertexInfo{
    vec2 uv;
}vertexInfo;
void main(){
    gl_Position = vec4(pos, 1.0);
    vertexInfo.uv = pos.xy;
}
<GLVertex/>
<GLFragment>
#version 450 core
layout(location = 0) uniform sampler2D g_map;
uniform Color {
    vec4 color;
};
layout(location = 0) out vec4 fragColor;
void main(){
    fragColor = texture(g_map, vertexInfo.uv);
}
<GLFragment/>

Here is the pattern:

<GLVertex>((.|\n)+)<GLVertex\/>

But the program always crash! Is there any bug in my regex? I've tested on regex101.

PS. when i delete the 5th line:

vec2 uv;

it works OK!

解决方案

You get a Stack overflow (parameters: 0x00000001, 0x00312FFC) exception as the pattern is not efficient. I think it is related to how std::regex processes repeated groups (you have define one with a +-quantifier group (.|\n)+). This pattern matches each and every character that is not a newline (.), or a newline (\n), and then stores the match inside a buffer. Then, an issue with the iterator debugging happens only in Debug mode. The std::_Orphan_Me is the place where the break occurs, and it is considered the most "expensive" method when matching strings. See performance killer -Debug Iterator Support in Visual studio

You should either switch to the Release mode, or test with a regex that does not require the use of the repeated groups, like any non-null character [^\x00] with lazy quantifier *?:

std::string str1 = "<GLVertex>\n#version 450 core\nlayout(location = 0) in vec3 pos;\nin VertexInfo{\n    vec2 uv;\n}vertexInfo;\nvoid main(){\n    gl_Position = vec4(pos, 1.0);\n    vertexInfo.uv = pos.xy;\n}\n<GLVertex/>\n<GLFragment>\n#version 450 core\nlayout(location = 0) uniform sampler2D g_map;\nuniform Color {\n    vec4 color;\n};\nlayout(location = 0) out vec4 fragColor;\nvoid main(){\n    fragColor = texture(g_map, vertexInfo.uv);\n}\n<GLFragment/>"; 
std::regex reg1("<GLVertex>([^\\x00]*?)<GLVertex/>");
std::smatch find1;
if (std::regex_search(str1, find1, reg1)){
    std::cout << find1[1].str();
}

这篇关于使用MSVC 2013正则表达式错误的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！