compile-time-regular-expressions

Compile Time Regular Expressions v3

Fast compile-time regular expressions with support for matching/searching/capturing during compile-time or runtime.

You can use the single header version from directory single-header. This header can be regenerated with make single-header. If you are using cmake, you can add this directory as subdirectory and link to target ctre.

More info at compile-time.re

What this library can do

ctre::match<"REGEX">(subject); // C++20
"REGEX"_ctre.match(subject); // C++17 + N3599 extension

• Matching
• Searching (search or starts_with)
• Capturing content (named captures are supported too, but only with syntax (?<name>...))
• Back-Reference (\g{N} syntax, and \1...\9 syntax too)
• Multiline support (with multiline_) functions
• Unicode properties and UTF-8 support

The library is implementing most of the PCRE syntax with a few exceptions:

• callouts
• comments
• conditional patterns
• control characters (\cX)
• match point reset (\K)
• named characters
• octal numbers
• options / modes
• subroutines
• unicode grapheme cluster (\X)

More documentation on pcre.org.

Unknown character escape behaviour

Not all escaped characters are automatically inserted as self, behaviour of the library is escaped characters are with special meaning, unknown escaped character is a syntax error.

Explicitly allowed character escapes which insert only the character are:

\-\"\<\>

Basic API

This is approximated API specification from a user perspective (omitting constexpr and noexcept which are everywhere, and using C++20 syntax even the API is C++17 compatible):

…

Range outputting API

…

Functors

All the functions (ctre::match, ctre::search, ctre::starts_with, ctre::range, ctre::tokenize, ctre::split) are functors and can be used without parenthesis:

auto matcher = ctre::match<"regex">;
if (matcher(input)) ...

Possible subjects (inputs)

• std::string-like objects (std::string_view or your own string if it's providing begin/end functions with forward iterators)
• pairs of forward iterators

Unicode support

To enable you need to include:

• <ctre-unicode.hpp>
• or <ctre.hpp> and <unicode-db.hpp>

Otherwise you will get missing symbols if you try to use the unicode support without enabling it.

Supported compilers

• clang 14.0+ (template UDL, C++17 syntax, C++20 cNTTP syntax)
• xcode clang 15.0+ (template UDL, C++17 syntax, C++20 cNTTP syntax)
• gcc 9.0+ (C++17 & C++20 cNTTP syntax)
• MSVC 14.29+ (Visual Studio 16.11+) (C++20 cNTTP syntax)

Template UDL syntax

The compiler must support extension N3599, for example as GNU extension in gcc (not in GCC 9.1+) and clang.

constexpr auto match(std::string_view sv) noexcept {
    using namespace ctre::literals;
    return "h.*"_ctre.match(sv);
}

If you need extension N3599 in GCC 9.1+, you can't use -pedantic. Also, you need to define macro CTRE_ENABLE_LITERALS.

C++17 syntax

You can provide a pattern as a constexpr ctll::fixed_string variable.

static constexpr auto pattern = ctll::fixed_string{ "h.*" };

constexpr auto match(std::string_view sv) noexcept {
    return ctre::match<pattern>(sv);
}

(this is tested in MSVC 15.8.8)

C++20 syntax

Currently, the only compiler which supports cNTTP syntax ctre::match<PATTERN>(subject) is GCC 9+.

constexpr auto match(std::string_view sv) noexcept {
    return ctre::match<"h.*">(sv);
}

Examples

Extracting number from input

std::optional<std::string_view> extract_number(std::string_view s) noexcept {
    if (auto m = ctre::match<"[a-z]+([0-9]+)">(s)) {
        return m.get<1>().to_view();
    } else {
        return std::nullopt;
    }
}

link to compiler explorer

Extracting values from date

…

link to compiler explorer

Using captures

auto result = ctre::match<"(?<year>\\d{4})/(?<month>\\d{1,2})/(?<day>\\d{1,2})">(s);
return date{result.get<"year">(), result.get<"month">, result.get<"day">};

// or in C++ emulation, but the object must have a linkage
static constexpr ctll::fixed_string year = "year";
static constexpr ctll::fixed_string month = "month";
static constexpr ctll::fixed_string day = "day";
return date{result.get<year>(), result.get<month>(), result.get<day>()};

// or use numbered access
// capture 0 is the whole match
return date{result.get<1>(), result.get<2>(), result.get<3>()};

Lexer

enum class type {
    unknown, identifier, number
};

struct lex_item {
    type t;
    std::string_view c;
};

std::optional<lex_item> lexer(std::string_view v) noexcept {
    if (auto [m,id,num] = ctre::match<"([a-z]+)|([0-9]+)">(v); m) {
        if (id) {
            return lex_item{type::identifier, id};
        } else if (num) {
            return lex_item{type::number, num};
        }
    }
    return std::nullopt;
}

link to compiler explorer

Range over input

This support is preliminary, probably the API will be changed.

auto input = "123,456,768"sv;

for (auto match: ctre::search_all<"([0-9]+),?">(input))
    std::cout << std::string_view{match.get<0>()} << "\n";

Unicode

#include <ctre-unicode.hpp>
#include <iostream>

// needed if you want to output to the terminal
std::string_view cast_from_unicode(std::u8string_view input) noexcept {
    return std::string_view(reinterpret_cast<const char *>(input.data()), input.size());
}

int main() {
    using namespace std::literals;
    std::u8string_view original = u8"Tu es un génie"sv;

    for (auto match: ctre::search_all<"\\p{Letter}+">(original))
        std::cout << cast_from_unicode(match) << std::endl;
    return 0;
}

link to compiler explorer

Installing ctre using vcpkg

You can download and install ctre using the vcpkg dependency manager:

git clone https://github.com/Microsoft/vcpkg.git
cd vcpkg
./bootstrap-vcpkg.sh
./vcpkg integrate install
./vcpkg install ctre

The ctre port in vcpkg is kept up to date by Microsoft team members and community contributors. If the version is out of date, please create an issue or pull request on the vcpkg repository.

Running tests (for developers)

Just run make in root of this project.