Add the regex_stdlib regular expression engine.

Unlike other regular expression engines this is available on all
systems.

4 files changed, 132 insertions, 2 deletions

diff --git a/data/anope.example.conf b/data/anope.example.conf
index 118f2f2f5..62eaa8154 100644
--- a/data/anope.example.conf
+++ b/data/anope.example.conf
@@ -527,7 +527,7 @@ options
 	 *
 	 * Note for this to work the regex module providing the regex engine must be loaded.
 	 */
-	#regexengine = "regex/pcre"
+	#regexengine = "regex/stdlib"
 
 	/*
 	 * A list of languages to load on startup that will be available in /NICKSERV SET LANGUAGE.
diff --git a/data/modules.example.conf b/data/modules.example.conf
index 6630bc750..fdf5b57be 100644
--- a/data/modules.example.conf
+++ b/data/modules.example.conf
@@ -412,11 +412,33 @@ module { name = "help" }
  * m_regex_posix [EXTRA]
  *
  * Provides the regex engine regex/posix, which uses the POSIX compliant regular expressions.
- * This is likely the only regex module you will not need extra libraries for.
  */
 #module { name = "m_regex_posix" }
 
 /*
+ * m_regex_stdlib
+ *
+ * Provides the regex engine regex/stdlib, which uses the regular expression library that is part of
+ * the C++ standard library.
+*/
+module
+{
+	name = "m_regex_stdlib"
+
+	/*
+	 * The syntax scheme to use. Can be set to awk to use the regular expression grammar used by the
+	 * awk utility in POSIX, basic to use the basic POSIX regular expression grammar, ecmascript to
+	 * use the modified ECMAScript regular expression grammar, egrep to use the regular expression
+	 * grammar used by the grep utility with the -E option in POSIX, extended to use the extended
+	 * POSIX regular expression grammar, or grep to use the regular expression grammar used by the
+	 * grep utility in POSIX.
+	 *
+	 * See https://en.cppreference.com/w/cpp/regex/syntax_option_type for more information.
+	 */
+	syntax = "ecmascript"
+}
+
+/*
  * m_regex_tre [EXTRA]
  *
  * Provides the regex engine regex/tre, which uses the TRE regex library.
diff --git a/docs/Changes.conf b/docs/Changes.conf
index fdd729b04..7397a8588 100644
--- a/docs/Changes.conf
+++ b/docs/Changes.conf
@@ -1,5 +1,6 @@
 Anope Version 2.1.1-git
 -----------------------
+Added the m_regex_stdlib module.
 Removed the m_regex_pcre module (use m_regex_pcre2 instead).
 
 Anope Version 2.1.0
diff --git a/modules/m_regex_stdlib.cpp b/modules/m_regex_stdlib.cpp
new file mode 100644
index 000000000..67fceedae
--- /dev/null
+++ b/modules/m_regex_stdlib.cpp
@@ -0,0 +1,107 @@
+/*
+ *
+ * (C) 2012-2023 Anope Team
+ * Contact us at team@anope.org
+ *
+ * Please read COPYING and README for further details.
+ */
+
+#include "module.h"
+
+#include <regex>
+
+class StdLibRegex final
+	: public Regex
+{
+private:
+	std::regex regex;
+
+public:
+	StdLibRegex(const Anope::string &expr, std::regex::flag_type type)
+		: Regex(expr)
+	{
+		try
+		{
+			regex.assign(expr.str(), type | std::regex::optimize);
+		}
+		catch (const std::regex_error& error)
+		{
+			throw RegexException("Error in regex " + expr + ": " + error.what());
+		}
+	}
+
+	bool Matches(const Anope::string &str)
+	{
+		return std::regex_search(str.str(), regex);
+	}
+};
+
+class StdLibRegexProvider final
+	: public RegexProvider
+{
+public:
+	std::regex_constants::syntax_option_type type;
+
+	StdLibRegexProvider(Module *creator)
+		: RegexProvider(creator, "regex/stdlib")
+	{
+	}
+
+	Regex *Compile(const Anope::string &expression) override
+	{
+		return new StdLibRegex(expression, type);
+	}
+};
+
+class ModuleRegexStdLib final
+	: public Module
+{
+private:
+	StdLibRegexProvider stdlib_regex_provider;
+
+public:
+	ModuleRegexStdLib(const Anope::string &modname, const Anope::string &creator)
+		: Module(modname, creator, EXTRA | VENDOR)
+		, stdlib_regex_provider(this)
+	{
+		this->SetPermanent(true);
+	}
+
+	void OnReload(Configuration::Conf *conf) override
+	{
+		Configuration::Block *block = conf->GetModule(this);
+
+		const Anope::string syntax = block->Get<const Anope::string>("syntax", "ecmascript");
+		if (syntax == "awk")
+			stdlib_regex_provider.type = std::regex::awk;
+		else if (syntax == "basic")
+			stdlib_regex_provider.type = std::regex::basic;
+		else if (syntax == "ecmascript")
+			stdlib_regex_provider.type = std::regex::ECMAScript;
+		else if (syntax == "egrep")
+			stdlib_regex_provider.type = std::regex::egrep;
+		else if (syntax == "extended")
+			stdlib_regex_provider.type = std::regex::extended;
+		else if (syntax == "grep")
+			stdlib_regex_provider.type = std::regex::grep;
+		else
+			throw ConfigException(this->name + ": syntax must be set to awk, basic, ecmascript, egrep, extended, or grep.");
+	}
+
+	~ModuleRegexStdLib()
+	{
+		for (auto *xlm : XLineManager::XLineManagers)
+		{
+			for (auto *x : xlm->GetList())
+			{
+				if (x->regex && dynamic_cast<StdLibRegex *>(x->regex))
+				{
+					delete x->regex;
+					x->regex = NULL;
+				}
+			}
+		}
+	}
+};
+
+MODULE_INIT(ModuleRegexStdLib)