Allow configuring casemaps

11 files changed, 272 insertions, 115 deletions

diff --git a/data/CMakeLists.txt b/data/CMakeLists.txt
index 8bb635dcf..2515bc082 100644
--- a/data/CMakeLists.txt
+++ b/data/CMakeLists.txt
@@ -1,6 +1,4 @@
-# Only install example.chk and example.conf from this directory
-# NOTE: I would've had this just find all files in the directory, but that would include files not needed (like this file)
-set(DATA example.chk botserv.example.conf example.conf hostserv.example.conf modules.example.conf operserv.example.conf chanserv.example.conf global.example.conf memoserv.example.conf nickserv.example.conf chanstats.example.conf irc2sql.example.conf stats.standalone.example.conf)
-install(FILES ${DATA}
+file(GLOB DATA_EXAMPLECONFS RELATIVE ${CMAKE_CURRENT_SOURCE_DIR} "*.conf")
+install(FILES example.chk ${DATA_EXAMPLECONFS}
   DESTINATION ${CONF_DIR}
 )
diff --git a/data/ascii.conf b/data/ascii.conf
new file mode 100644
index 000000000..1bfcb33f0
--- /dev/null
+++ b/data/ascii.conf
@@ -0,0 +1,157 @@
+
+casemap
+{
+        lower = "a"
+        upper = "A"
+}
+
+casemap
+{
+        lower = "b"
+        upper = "B"
+}
+
+casemap
+{
+        lower = "c"
+        upper = "C"
+}
+
+casemap
+{
+        lower = "d"
+        upper = "D"
+}
+
+casemap
+{
+        lower = "e"
+        upper = "E"
+}
+
+casemap
+{
+        lower = "f"
+        upper = "F"
+}
+
+casemap
+{
+        lower = "g"
+        upper = "G"
+}
+
+casemap
+{
+        lower = "h"
+        upper = "H"
+}
+
+casemap
+{
+        lower = "i"
+        upper = "I"
+}
+
+casemap
+{
+        lower = "j"
+        upper = "J"
+}
+
+casemap
+{
+        lower = "k"
+        upper = "K"
+}
+
+casemap
+{
+        lower = "l"
+        upper = "L"
+}
+
+casemap
+{
+        lower = "m"
+        upper = "M"
+}
+
+casemap
+{
+        lower = "n"
+        upper = "N"
+}
+
+casemap
+{
+        lower = "o"
+        upper = "O"
+}
+
+casemap
+{
+        lower = "p"
+        upper = "P"
+}
+
+casemap
+{
+        lower = "q"
+        upper = "Q"
+}
+
+casemap
+{
+        lower = "r"
+        upper = "R"
+}
+
+casemap
+{
+        lower = "s"
+        upper = "S"
+}
+
+casemap
+{
+        lower = "t"
+        upper = "T"
+}
+
+casemap
+{
+        lower = "u"
+        upper = "U"
+}
+
+casemap
+{
+        lower = "v"
+        upper = "V"
+}
+
+casemap
+{
+        lower = "w"
+        upper = "W"
+}
+
+casemap
+{
+        lower = "x"
+        upper = "X"
+}
+
+casemap
+{
+        lower = "y"
+        upper = "Y"
+}
+
+casemap
+{
+        lower = "z"
+        upper = "Z"
+}
+
diff --git a/data/example.conf b/data/example.conf
index f10c62fb2..f746a2595 100644
--- a/data/example.conf
+++ b/data/example.conf
@@ -362,20 +362,6 @@ options
 	#group = "anope"
 
 	/*
-	 * The case mapping used by services. This must be set to a valid locale name
-	 * installed on your machine. Services use this case map to compare, with
-	 * case insensitivity, things such as nick names, channel names, etc.
-	 *
-	 * We provide two special casemaps shipped with Anope, ascii and rfc1459.
-	 *
-	 * This value should be set to what your IRCd uses, which is probably rfc1459,
-	 * however Anope has always used ascii for comparison, so the default is ascii.
-	 *
-	 * Changing this value once set is not recommended.
-	 */
-	casemap = "ascii"
-
-	/*
 	 * This key is used to initiate the random number generator. This number
 	 * MUST be random as you want your passcodes to be random. Don't give this
 	 * key to anyone! Keep it private!
@@ -539,6 +525,26 @@ options
 }
 
 /*
+ * [REQUIRED] Casemapping configuration
+ *
+ * Includes the configuration for the casemap services should use.
+ * Services use this case map to compare, with case insensitivity,
+ * things such as nick names, channel names, etc.
+ *
+ * Two casemaps shipped with Anope are ascii and rfc1459.
+ *
+ * This value should be set to what your IRCd uses, which is probably rfc1459,
+ * however Anope has always used ascii for comparison, so the default is ascii.
+ *
+ * Changing this value once set is not recommended.
+ */
+include
+{
+	name = "ascii.conf"
+	#name = "rfc1459.conf"
+}
+
+/*
  * [OPTIONAL] BotServ
  *
  * Includes botserv.example.conf, which is necessary for BotServ functionality.
diff --git a/data/rfc1459.conf b/data/rfc1459.conf
new file mode 100644
index 000000000..81ced1ac0
--- /dev/null
+++ b/data/rfc1459.conf
@@ -0,0 +1,24 @@
+
+include
+{
+	name = "ascii.conf"
+}
+
+casemap
+{
+	lower = "{"
+	upper = "["
+}
+
+casemap
+{
+	lower = "}"
+	upper = "]"
+}
+
+casemap
+{
+	lower = "|"
+	upper = "\""
+}
+
diff --git a/include/config.h b/include/config.h
index fe4a6fded..fdeddf6dc 100644
--- a/include/config.h
+++ b/include/config.h
@@ -60,9 +60,10 @@ namespace Configuration
 		const item_map* GetItems() const;
 	};
 
-	template<> CoreExport const Anope::string Block::Get(const Anope::string &tag, const Anope::string& def) const;
+	template<> CoreExport Anope::string Block::Get(const Anope::string &tag, const Anope::string& def) const;
 	template<> CoreExport time_t Block::Get(const Anope::string &tag, const Anope::string &def) const;
 	template<> CoreExport bool Block::Get(const Anope::string &tag, const Anope::string &def) const;
+	template<> CoreExport unsigned int Block::Get(const Anope::string &tag, const Anope::string &def) const;
 
 	/** Represents a configuration file
 	 */
@@ -122,6 +123,7 @@ namespace Configuration
 		std::vector<Anope::string> ModulesAutoLoad;
 		/* After how many characters do we wrap lines? */
 		unsigned int LineWrap;
+		unsigned char CaseMapUpper[256], CaseMapLower[256];
 
 		/* module configuration blocks */
 		std::map<Anope::string, Block *> modules;
diff --git a/include/hashcomp.h b/include/hashcomp.h
index 2364975b9..a9a1105ee 100644
--- a/include/hashcomp.h
+++ b/include/hashcomp.h
@@ -19,56 +19,8 @@ namespace Anope
 {
 	class string;
 
-	/* Casemap in use by Anope. ci::string's comparation functions use this (and thus Anope::string) */
-	extern std::locale casemap;
-
-	extern void CaseMapRebuild();
 	extern unsigned char tolower(unsigned char);
 	extern unsigned char toupper(unsigned char);
-
-	/* ASCII case insensitive ctype. */
-	template<typename char_type>
-	class ascii_ctype : public std::ctype<char_type>
-	{
-	 public:
-		char_type do_toupper(char_type c) const override
-		{
-			if (c >= 'a' && c <= 'z')
-				return c - 32;
-			else
-				return c;
-		}
-
-		char_type do_tolower(char_type c) const override
-		{
-			if (c >= 'A' && c <= 'Z')
-				return c + 32;
-			else
-				return c;
-		}
-	};
-
-	/* rfc1459 case insensitive ctype, { = [, } = ], and | = \ */
-	template<typename char_type>
-	class rfc1459_ctype : public ascii_ctype<char_type>
-	{
-	 public:
-		char_type do_toupper(char_type c) const override
-		{
-			if (c == '{' || c == '}' || c == '|')
-				return c - 32;
-			else
-				return ascii_ctype<char_type>::do_toupper(c);
-		}
-
-		char_type do_tolower(char_type c) const override
-		{
-			if (c == '[' || c == ']' || c == '\\')
-				return c + 32;
-			else
-				return ascii_ctype<char_type>::do_tolower(c);
-		}
-	};
 }
 
 /** The ci namespace contains a number of helper classes relevant to case insensitive strings.
diff --git a/modules/database/db_flatfile.cpp b/modules/database/db_flatfile.cpp
index 0793269ea..2486a2307 100644
--- a/modules/database/db_flatfile.cpp
+++ b/modules/database/db_flatfile.cpp
@@ -101,7 +101,6 @@ class DBFlatFile : public Module
 		Serialize::Object *obj = nullptr;
 		for (Anope::string buf; std::getline(fd, buf.str());)
 		{
-			Log() << buf;
 			if (buf.find("OBJECT ") == 0)
 			{
 				Anope::string t = buf.substr(7);
diff --git a/src/config.cpp b/src/config.cpp
index 6ac777a98..323a48519 100644
--- a/src/config.cpp
+++ b/src/config.cpp
@@ -72,7 +72,7 @@ const Block::item_map* Block::GetItems() const
 		return NULL;
 }
 
-template<> const Anope::string Block::Get(const Anope::string &tag, const Anope::string& def) const
+template<> Anope::string Block::Get(const Anope::string &tag, const Anope::string& def) const
 {
 	if (!this)
 		return def;
@@ -95,6 +95,27 @@ template<> bool Block::Get(const Anope::string &tag, const Anope::string &def) c
 	return !str.empty() && !str.equals_ci("no") && !str.equals_ci("off") && !str.equals_ci("false") && !str.equals_ci("0");
 }
 
+template<> unsigned int Block::Get(const Anope::string &tag, const Anope::string &def) const
+{
+	const Anope::string &str = Get<Anope::string>(tag, def);
+	std::size_t pos = str.length();
+	unsigned long l;
+
+	try
+	{
+		l = std::stoul(str.str(), &pos, 0);
+	}
+	catch (...)
+	{
+		return 0;
+	}
+
+	if (pos != str.length())
+		return 0;
+
+	return l;
+}
+
 static void ValidateNotEmpty(const Anope::string &block, const Anope::string &name, const Anope::string &value)
 {
 	if (value.empty())
@@ -127,6 +148,8 @@ Conf::Conf() : Block("")
 		const Anope::string &type = include->Get<Anope::string>("type"),
 					&file = include->Get<Anope::string>("name");
 
+		ValidateNotEmpty("include", "name", file);
+
 		File f(file, type == "executable");
 		this->LoadConf(f);
 	}
@@ -489,6 +512,34 @@ Conf::Conf() : Block("")
 		this->CommandGroups.push_back(gr);
 	}
 
+	for (int i = 0; i < this->CountBlock("casemap"); ++i)
+	{
+		Block *casemap = this->GetBlock("casemap", i);
+
+		unsigned char upper = casemap->Get<unsigned int>("upper"),
+				lower = casemap->Get<unsigned int>("lower");
+
+		if (!upper)
+		{
+			Anope::string s = casemap->Get<Anope::string>("upper");
+			if (s.length() == 1)
+				upper = s[0];
+		}
+
+		if (!lower)
+		{
+			Anope::string s = casemap->Get<Anope::string>("lower");
+			if (s.length() == 1)
+				lower = s[0];
+		}
+
+		if (upper && lower)
+		{
+			CaseMapUpper[lower] = CaseMapUpper[upper] = upper;
+			CaseMapLower[lower] = CaseMapLower[upper] = lower;
+		}
+	}
+
 	/* Below here can't throw */
 
 	/* Clear existing conf opers */
@@ -508,23 +559,6 @@ Conf::Conf() : Block("")
 			Log() << "Tied oper " << na->GetAccount()->GetDisplay() << " to type " << o->GetType()->GetName();
 		}
 
-	if (options->Get<Anope::string>("casemap", "ascii") == "ascii")
-		Anope::casemap = std::locale(std::locale(), new Anope::ascii_ctype<char>());
-	else if (options->Get<Anope::string>("casemap") == "rfc1459")
-		Anope::casemap = std::locale(std::locale(), new Anope::rfc1459_ctype<char>());
-	else
-	{
-		try
-		{
-			Anope::casemap = std::locale(options->Get<Anope::string>("casemap").c_str());
-		}
-		catch (const std::runtime_error &)
-		{
-			Log() << "Unknown casemap " << options->Get<Anope::string>("casemap") << " - casemap not changed";
-		}
-	}
-	Anope::CaseMapRebuild();
-
 	/* Check the user keys */
 	if (!options->Get<unsigned>("seed"))
 		Log() << "Configuration option options:seed should be set. It's for YOUR safety! Remember that!";
diff --git a/src/hashcomp.cpp b/src/hashcomp.cpp
index 62bcc404f..e17ae6176 100644
--- a/src/hashcomp.cpp
+++ b/src/hashcomp.cpp
@@ -10,32 +10,22 @@
 #include "services.h"
 #include "hashcomp.h"
 #include "anope.h"
-
-/* Case map in use by Anope */
-std::locale Anope::casemap = std::locale(std::locale(), new Anope::ascii_ctype<char>());
-/* Cache of the above case map, forced upper */
-static unsigned char case_map_upper[256], case_map_lower[256];
-
-/* called whenever Anope::casemap is modified to rebuild the casemap cache */
-void Anope::CaseMapRebuild()
-{
-	const std::ctype<char> &ct = std::use_facet<std::ctype<char> >(Anope::casemap);
-
-	for (unsigned i = 0; i < sizeof(case_map_upper); ++i)
-	{
-		case_map_upper[i] = ct.toupper(i);
-		case_map_lower[i] = ct.tolower(i);
-	}
-}
+#include "config.h"
 
 unsigned char Anope::tolower(unsigned char c)
 {
-	return case_map_lower[c];
+	if (!Config || !Config->CaseMapLower[c])
+		return std::tolower(c);
+
+	return Config->CaseMapLower[c];
 }
 
 unsigned char Anope::toupper(unsigned char c)
 {
-	return case_map_upper[c];
+	if (!Config || !Config->CaseMapUpper[c])
+		return std::toupper(c);
+
+	return Config->CaseMapUpper[c];
 }
 
 /*
@@ -47,7 +37,7 @@ unsigned char Anope::toupper(unsigned char c)
 
 bool ci::ci_char_traits::eq(char c1st, char c2nd)
 {
-	return case_map_upper[static_cast<unsigned char>(c1st)] == case_map_upper[static_cast<unsigned char>(c2nd)];
+	return Anope::toupper(c1st) == Anope::toupper(c2nd);
 }
 
 bool ci::ci_char_traits::ne(char c1st, char c2nd)
@@ -57,15 +47,15 @@ bool ci::ci_char_traits::ne(char c1st, char c2nd)
 
 bool ci::ci_char_traits::lt(char c1st, char c2nd)
 {
-	return case_map_upper[static_cast<unsigned char>(c1st)] < case_map_upper[static_cast<unsigned char>(c2nd)];
+	return Anope::toupper(c1st) < Anope::toupper(c2nd);
 }
 
 int ci::ci_char_traits::compare(const char *str1, const char *str2, size_t n)
 {
 	for (unsigned i = 0; i < n; ++i)
 	{
-		unsigned char c1 = case_map_upper[static_cast<unsigned char>(*str1)],
-				c2 = case_map_upper[static_cast<unsigned char>(*str2)];
+		unsigned char c1 = Anope::toupper(*str1),
+				c2 = Anope::toupper(*str2);
 
 		if (c1 > c2)
 			return 1;
@@ -82,7 +72,7 @@ int ci::ci_char_traits::compare(const char *str1, const char *str2, size_t n)
 
 const char *ci::ci_char_traits::find(const char *s1, int n, char c)
 {
-	while (n-- > 0 && case_map_upper[static_cast<unsigned char>(*s1)] != case_map_upper[static_cast<unsigned char>(c)])
+	while (n-- > 0 && Anope::toupper(*s1) != Anope::toupper(c))
 		++s1;
 	return n >= 0 ? s1 : NULL;
 }
diff --git a/src/init.cpp b/src/init.cpp
index 853cf7209..0b381cbf6 100644
--- a/src/init.cpp
+++ b/src/init.cpp
@@ -290,8 +290,6 @@ void Anope::Init(int ac, char **av)
 	umask(DEFUMASK);
 #endif
 
-	//Serialize::RegisterTypes();
-
 	/* Parse command line arguments */
 	ParseCommandLineArguments(ac, av);
 
diff --git a/src/main.cpp b/src/main.cpp
index 1f4434f18..65fb8d3f6 100644
--- a/src/main.cpp
+++ b/src/main.cpp
@@ -116,9 +116,6 @@ static Anope::string GetFullProgDir(const Anope::string &argv0)
 
 int main(int ac, char **av, char **envp)
 {
-	/* String comparisons won't work until we build the case map cache, so do it first */
-	Anope::CaseMapRebuild();
-
 	BinaryDir = GetFullProgDir(av[0]);
 	if (BinaryDir[BinaryDir.length() - 1] == '.')
 		BinaryDir = BinaryDir.substr(0, BinaryDir.length() - 2);