I am using TR1 Regular Expressions (for VS2010) and what I'm trying to do is search for specific pattern for a group called "name", and another pattern for a group called "value". I think what I want is called a capture group, but I'm not sure if that's the right terminology. I want to assign matches to the pattern "[^:\r\n]+):\s" to a list of matches called "name", and matches of the pattern "[^\r\n]+)\r\n)+" to a list of matches called "value".
The regex pattern I have so far is
string pattern = "((?<name>[^:\r\n]+):\s(?<value>[^\r\n]+)\r\n)+";
But the regex T4R1 header keeps throwing an exception when the program runs. What's wrong with the syntax of the pattern I have? Can someone show an example pattern that would do what I'm trying to accomplish?
Also, how would it be possible to include a substring within the pattern to match, but not actually include that substring in the results? For example, I want to match all strings of the pattern
"http://[[:alpha:]]\r\n"
, but I don't want to include the substring "http://" in the returned results of matches.
The C++ TR1 and C++11 regular expression grammars don't support named capture groups. You'll have to do unnamed capture groups.
Also, make sure you don't run into escaping issues. You'll have to escape some characters twice: one for being in a C++ string, and another for being in a regex. The pattern (([^:\r\n]+):\s\s([^\r\n]+)\r\n)+ can be written as a C++ string literal like this:
"([^:\\r\\n]+:\\s\\s([^\\r\\n]+)\\r\\n)+"
// or in C++11
R"xxx(([^:\r\n]+:\s\s([^\r\n]+)\r\n)+)xxx"
Lookbehinds are not supported either. You'll have to work around this limitation by using capture groups: use the pattern (http://)([[:alpha:]]\r\n) and grab only the second capture group.
来源：https://stackoverflow.com/questions/6398278/tr1-regex-capture-groups