Skip to content
Tutorials Center
My learning

Topics

C++ Reference ยท Strings and I/O

C++ std::regex class

Reference for std::regex and <regex>: building patterns, ECMAScript syntax and flags, error handling with regex_error, raw strings and performance costs.

std::regex (header <regex>, C++11) is the standard library's regular expression engine. A std::regex object holds a compiled pattern. You pass it to the algorithms std::regex_match, std::regex_search and std::regex_replace, or to the iterators std::sregex_iterator and std::sregex_token_iterator, which do the actual matching on a string.

Compiling the pattern is the expensive part, so construct a regex once and reuse it instead of creating it inside a loop. The default grammar is ECMAScript, which is the same family of syntax JavaScript uses: character classes, groups, alternation, quantifiers, anchors, lazy quantifiers and look-ahead. Look-behind and named groups are not supported.

C++
#include <regex>

std::regex re(R"((\d+)-(\d+))");   // raw string: no doubled backslashes

Try it

C++ Compiling, matching and a bad pattern

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with std::cin or std::getline), it gets the next line from this box.

C++ Grammars and flags

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with std::cin or std::getline), it gets the next line from this box.

Syntax

C++
std::basic_regex<char> regex;     // typedef std::regex  (std::wregex for wchar_t)

explicit regex(const std::string& pattern,
               std::regex_constants::syntax_option_type flags = std::regex_constants::ECMAScript);
Flag (std::regex::...) Description
ECMAScript The default grammar.
basic, extended, awk, grep, egrep POSIX-style grammars. Choose one grammar only.
icase Case-insensitive matching.
nosubs Do not store sub-matches: only the whole match is reported.
optimize Hint to favor matching speed over compile speed (often no effect).
collate Make [a-z] ranges locale-sensitive.
multiline ^ and $ match at line boundaries (C++17).

Errors: an invalid pattern throws std::regex_error, whose code() is one of the constants in std::regex_constants (error_paren, error_brack, error_badrepeat, error_escape, ...).

Usage notes

  • Use raw string literals. R"(\d+)" is far easier to read than "\\d+". If the pattern contains )", add a delimiter: R"re(...)re".
  • Compile once. Make the regex const, static or a class member; never build it per iteration of a loop or per call of a hot function.
  • Never pass a temporary string to regex_search with a match_results. The overload taking const std::string&& is deleted, because the result would point into a destroyed string. Store the string in a variable first.
  • Known limits. Large inputs and certain patterns can recurse deeply in libstdc++ and overflow the stack; avoid regexes on megabytes of data. Compilation and matching are slower than in dedicated libraries such as RE2 or PCRE.
  • Not a parser. HTML, JSON, nested parentheses and similar recursive grammars cannot be handled reliably with regular expressions. Use a real parser.
  • Unicode. std::regex works on char units, so a multi-byte UTF-8 character counts as several characters. . may match half of a character.
  • Compile time. <regex> is a heavy header. In our editor, programs that use it take noticeably longer to compile than most examples (several seconds).
  • C++17 adds the multiline flag. The library has no named capture groups or look-behind in any standard.