# HG changeset patch # User Tobias Markmann # Date 1292335417 -3600 # Node ID f02bac902a1e62ba8f9538e54ceff97601979a7b # Parent 72a917d910a48923c692422840b76b4991efd05e util.encodings: Support for ICU for IDNA operations. diff -r 72a917d910a4 -r f02bac902a1e configure --- a/configure Fri Dec 10 00:23:47 2010 +0500 +++ b/configure Tue Dec 14 15:03:37 2010 +0100 @@ -11,8 +11,10 @@ LUA_INCDIR="/usr/include" LUA_LIBDIR="/usr/lib" IDN_LIB=idn +ICU_FLAGS="-licui18n -licudata -licuuc" OPENSSL_LIB=crypto CC=gcc +CXX=g++ LD=gcc CFLAGS="-fPIC -Wall" @@ -43,6 +45,9 @@ Default is \$LUA_DIR/lib --with-idn=LIB The name of the IDN library to link with. Default is $IDN_LIB +--idn-library=(idn|icu) Select library to use for IDNA functionality. + idn: use GNU libidn (default) + icu: use ICU from IBM --with-ssl=LIB The name of the SSL to link with. Default is $OPENSSL_LIB --cflags=FLAGS Flags to pass to the compiler @@ -96,7 +101,6 @@ LUA_INCDIR_SET=yes LUA_LIBDIR=/usr/local/lib LUA_LIBDIR_SET=yes - CFLAGS="-Wall" LDFLAGS="-bundle -undefined dynamic_lookup" fi if [ "$OSTYPE" = "linux" ] @@ -143,6 +147,9 @@ --with-idn=*) IDN_LIB="$value" ;; + --idn-library=*) + IDN_LIBRARY="$value" + ;; --with-ssl=*) OPENSSL_LIB="$value" ;; @@ -261,6 +268,21 @@ LUA_BINDIR="$LUA_DIR/bin" fi +if [ "$IDN_LIBRARY" = "icu" ] +then + IDNA_LIBS="$ICU_FLAGS" + CFLAGS="$CFLAGS -DUSE_STRINGPREP_ICU" +else + if [ "$IDN_LIBRARY" = "idn" ] + then + IDNA_LIBS="-l$IDN_LIB" + else + echo "Must use idn or icu as argument for --idn-library!" + exit 1 + fi +fi +LDFLAGS="$LDFLAGS -llua" + echo -n "Checking Lua includes... " lua_h="$LUA_INCDIR/lua.h" if [ -e "$lua_h" ] @@ -311,10 +333,12 @@ LUA_BINDIR=$LUA_BINDIR REQUIRE_CONFIG=$REQUIRE_CONFIG IDN_LIB=$IDN_LIB +IDNA_LIBS=$IDNA_LIBS OPENSSL_LIB=$OPENSSL_LIB CFLAGS=$CFLAGS LDFLAGS=$LDFLAGS CC=$CC +CXX=$CXX LD=$LD EOF diff -r 72a917d910a4 -r f02bac902a1e util-src/Makefile --- a/util-src/Makefile Fri Dec 10 00:23:47 2010 +0500 +++ b/util-src/Makefile Tue Dec 14 15:03:37 2010 +0100 @@ -7,10 +7,18 @@ IDN_LIB?=idn OPENSSL_LIB?=crypto CC?=gcc +CXX?=g++ LD?=gcc .SUFFIXES: .c .o .so +encodings.o: + $(CXX) $(CFLAGS) -I$(LUA_INCDIR) -c -o encodings.o encodings.cpp + +encodings.so: encodings.o + MACOSX_DEPLOYMENT_TARGET="10.3"; export MACOSX_DEPLOYMENT_TARGET; + $(CXX) $(LDFLAGS) $(IDNA_LIBS) -o encodings.so encodings.o -lcrypto + .c.o: $(CC) $(CFLAGS) -I$(LUA_INCDIR) -c -o $@ $< diff -r 72a917d910a4 -r f02bac902a1e util-src/encodings.c --- a/util-src/encodings.c Fri Dec 10 00:23:47 2010 +0500 +++ /dev/null Thu Jan 01 00:00:00 1970 +0000 @@ -1,243 +0,0 @@ -/* Prosody IM --- Copyright (C) 2008-2010 Matthew Wild --- Copyright (C) 2008-2010 Waqas Hussain --- --- This project is MIT/X11 licensed. Please see the --- COPYING file in the source package for more information. --- -*/ - -/* -* encodings.c -* Lua library for base64, stringprep and idna encodings -*/ - -// Newer MSVC compilers deprecate strcpy as unsafe, but we use it in a safe way -#define _CRT_SECURE_NO_DEPRECATE - -#include -#include - -#include "lua.h" -#include "lauxlib.h" - -/***************** BASE64 *****************/ - -static const char code[]= -"ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/"; - -static void base64_encode(luaL_Buffer *b, unsigned int c1, unsigned int c2, unsigned int c3, int n) -{ - unsigned long tuple=c3+256UL*(c2+256UL*c1); - int i; - char s[4]; - for (i=0; i<4; i++) { - s[3-i] = code[tuple % 64]; - tuple /= 64; - } - for (i=n+1; i<4; i++) s[i]='='; - luaL_addlstring(b,s,4); -} - -static int Lbase64_encode(lua_State *L) /** encode(s) */ -{ - size_t l; - const unsigned char *s=(const unsigned char*)luaL_checklstring(L,1,&l); - luaL_Buffer b; - int n; - luaL_buffinit(L,&b); - for (n=l/3; n--; s+=3) base64_encode(&b,s[0],s[1],s[2],3); - switch (l%3) - { - case 1: base64_encode(&b,s[0],0,0,1); break; - case 2: base64_encode(&b,s[0],s[1],0,2); break; - } - luaL_pushresult(&b); - return 1; -} - -static void base64_decode(luaL_Buffer *b, int c1, int c2, int c3, int c4, int n) -{ - unsigned long tuple=c4+64L*(c3+64L*(c2+64L*c1)); - char s[3]; - switch (--n) - { - case 3: s[2]=(char) tuple; - case 2: s[1]=(char) (tuple >> 8); - case 1: s[0]=(char) (tuple >> 16); - } - luaL_addlstring(b,s,n); -} - -static int Lbase64_decode(lua_State *L) /** decode(s) */ -{ - size_t l; - const char *s=luaL_checklstring(L,1,&l); - luaL_Buffer b; - int n=0; - char t[4]; - luaL_buffinit(L,&b); - for (;;) - { - int c=*s++; - switch (c) - { - const char *p; - default: - p=strchr(code,c); if (p==NULL) return 0; - t[n++]= (char) (p-code); - if (n==4) - { - base64_decode(&b,t[0],t[1],t[2],t[3],4); - n=0; - } - break; - case '=': - switch (n) - { - case 1: base64_decode(&b,t[0],0,0,0,1); break; - case 2: base64_decode(&b,t[0],t[1],0,0,2); break; - case 3: base64_decode(&b,t[0],t[1],t[2],0,3); break; - } - n=0; - break; - case 0: - luaL_pushresult(&b); - return 1; - case '\n': case '\r': case '\t': case ' ': case '\f': case '\b': - break; - } - } -} - -static const luaL_Reg Reg_base64[] = -{ - { "encode", Lbase64_encode }, - { "decode", Lbase64_decode }, - { NULL, NULL } -}; - -/***************** STRINGPREP *****************/ - -#include - -static int stringprep_prep(lua_State *L, const Stringprep_profile *profile) -{ - size_t len; - const char *s; - char string[1024]; - int ret; - if(!lua_isstring(L, 1)) { - lua_pushnil(L); - return 1; - } - s = lua_tolstring(L, 1, &len); - if (len >= 1024) { - lua_pushnil(L); - return 1; // TODO return error message - } - strcpy(string, s); - ret = stringprep(string, 1024, 0, profile); - if (ret == STRINGPREP_OK) { - lua_pushstring(L, string); - return 1; - } else { - lua_pushnil(L); - return 1; // TODO return error message - } -} - -#define MAKE_PREP_FUNC(myFunc, prep) \ -static int myFunc(lua_State *L) { return stringprep_prep(L, prep); } - -MAKE_PREP_FUNC(Lstringprep_nameprep, stringprep_nameprep) /** stringprep.nameprep(s) */ -MAKE_PREP_FUNC(Lstringprep_nodeprep, stringprep_xmpp_nodeprep) /** stringprep.nodeprep(s) */ -MAKE_PREP_FUNC(Lstringprep_resourceprep, stringprep_xmpp_resourceprep) /** stringprep.resourceprep(s) */ -MAKE_PREP_FUNC(Lstringprep_saslprep, stringprep_saslprep) /** stringprep.saslprep(s) */ - -static const luaL_Reg Reg_stringprep[] = -{ - { "nameprep", Lstringprep_nameprep }, - { "nodeprep", Lstringprep_nodeprep }, - { "resourceprep", Lstringprep_resourceprep }, - { "saslprep", Lstringprep_saslprep }, - { NULL, NULL } -}; - -/***************** IDNA *****************/ - -#include -#include - -static int Lidna_to_ascii(lua_State *L) /** idna.to_ascii(s) */ -{ - size_t len; - const char *s = luaL_checklstring(L, 1, &len); - char* output = NULL; - int ret = idna_to_ascii_8z(s, &output, IDNA_USE_STD3_ASCII_RULES); - if (ret == IDNA_SUCCESS) { - lua_pushstring(L, output); - idn_free(output); - return 1; - } else { - lua_pushnil(L); - idn_free(output); - return 1; // TODO return error message - } -} - -static int Lidna_to_unicode(lua_State *L) /** idna.to_unicode(s) */ -{ - size_t len; - const char *s = luaL_checklstring(L, 1, &len); - char* output = NULL; - int ret = idna_to_unicode_8z8z(s, &output, 0); - if (ret == IDNA_SUCCESS) { - lua_pushstring(L, output); - idn_free(output); - return 1; - } else { - lua_pushnil(L); - idn_free(output); - return 1; // TODO return error message - } -} - -static const luaL_Reg Reg_idna[] = -{ - { "to_ascii", Lidna_to_ascii }, - { "to_unicode", Lidna_to_unicode }, - { NULL, NULL } -}; - -/***************** end *****************/ - -static const luaL_Reg Reg[] = -{ - { NULL, NULL } -}; - -LUALIB_API int luaopen_util_encodings(lua_State *L) -{ - luaL_register(L, "encodings", Reg); - - lua_pushliteral(L, "base64"); - lua_newtable(L); - luaL_register(L, NULL, Reg_base64); - lua_settable(L,-3); - - lua_pushliteral(L, "stringprep"); - lua_newtable(L); - luaL_register(L, NULL, Reg_stringprep); - lua_settable(L,-3); - - lua_pushliteral(L, "idna"); - lua_newtable(L); - luaL_register(L, NULL, Reg_idna); - lua_settable(L,-3); - - lua_pushliteral(L, "version"); /** version */ - lua_pushliteral(L, "-3.14"); - lua_settable(L,-3); - return 1; -} diff -r 72a917d910a4 -r f02bac902a1e util-src/encodings.cpp --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/util-src/encodings.cpp Tue Dec 14 15:03:37 2010 +0100 @@ -0,0 +1,373 @@ +/* Prosody IM +-- Copyright (C) 2008-2010 Matthew Wild +-- Copyright (C) 2008-2010 Waqas Hussain +-- +-- This project is MIT/X11 licensed. Please see the +-- COPYING file in the source package for more information. +-- +*/ + +/* +* encodings.c +* Lua library for base64, stringprep and idna encodings +*/ + +// Newer MSVC compilers deprecate strcpy as unsafe, but we use it in a safe way +#define _CRT_SECURE_NO_DEPRECATE + +extern "C" { +#include +#include +#include "lua.h" +#include "lauxlib.h" +} +/***************** BASE64 *****************/ + +#define LUALIB_API extern "C" + +static const char code[]= +"ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/"; + +static void base64_encode(luaL_Buffer *b, unsigned int c1, unsigned int c2, unsigned int c3, int n) +{ + unsigned long tuple=c3+256UL*(c2+256UL*c1); + int i; + char s[4]; + for (i=0; i<4; i++) { + s[3-i] = code[tuple % 64]; + tuple /= 64; + } + for (i=n+1; i<4; i++) s[i]='='; + luaL_addlstring(b,s,4); +} + +static int Lbase64_encode(lua_State *L) /** encode(s) */ +{ + size_t l; + const unsigned char *s=(const unsigned char*)luaL_checklstring(L,1,&l); + luaL_Buffer b; + int n; + luaL_buffinit(L,&b); + for (n=l/3; n--; s+=3) base64_encode(&b,s[0],s[1],s[2],3); + switch (l%3) + { + case 1: base64_encode(&b,s[0],0,0,1); break; + case 2: base64_encode(&b,s[0],s[1],0,2); break; + } + luaL_pushresult(&b); + return 1; +} + +static void base64_decode(luaL_Buffer *b, int c1, int c2, int c3, int c4, int n) +{ + unsigned long tuple=c4+64L*(c3+64L*(c2+64L*c1)); + char s[3]; + switch (--n) + { + case 3: s[2]=(char) tuple; + case 2: s[1]=(char) (tuple >> 8); + case 1: s[0]=(char) (tuple >> 16); + } + luaL_addlstring(b,s,n); +} + +static int Lbase64_decode(lua_State *L) /** decode(s) */ +{ + size_t l; + const char *s=luaL_checklstring(L,1,&l); + luaL_Buffer b; + int n=0; + char t[4]; + luaL_buffinit(L,&b); + for (;;) + { + int c=*s++; + switch (c) + { + const char *p; + default: + p=strchr(code,c); if (p==NULL) return 0; + t[n++]= (char) (p-code); + if (n==4) + { + base64_decode(&b,t[0],t[1],t[2],t[3],4); + n=0; + } + break; + case '=': + switch (n) + { + case 1: base64_decode(&b,t[0],0,0,0,1); break; + case 2: base64_decode(&b,t[0],t[1],0,0,2); break; + case 3: base64_decode(&b,t[0],t[1],t[2],0,3); break; + } + n=0; + break; + case 0: + luaL_pushresult(&b); + return 1; + case '\n': case '\r': case '\t': case ' ': case '\f': case '\b': + break; + } + } +} + +static const luaL_Reg Reg_base64[] = +{ + { "encode", Lbase64_encode }, + { "decode", Lbase64_decode }, + { NULL, NULL } +}; + +/***************** STRINGPREP *****************/ +#ifndef USE_STRINGPREP_ICU +/****************** libidn ********************/ +#include + +static int stringprep_prep(lua_State *L, const Stringprep_profile *profile) +{ + size_t len; + const char *s; + char string[1024]; + int ret; + if(!lua_isstring(L, 1)) { + lua_pushnil(L); + return 1; + } + s = lua_tolstring(L, 1, &len); + if (len >= 1024) { + lua_pushnil(L); + return 1; // TODO return error message + } + strcpy(string, s); + ret = stringprep(string, 1024, 0, profile); + if (ret == STRINGPREP_OK) { + lua_pushstring(L, string); + return 1; + } else { + lua_pushnil(L); + return 1; // TODO return error message + } +} + +#define MAKE_PREP_FUNC(myFunc, prep) \ +static int myFunc(lua_State *L) { return stringprep_prep(L, prep); } + +MAKE_PREP_FUNC(Lstringprep_nameprep, stringprep_nameprep) /** stringprep.nameprep(s) */ +MAKE_PREP_FUNC(Lstringprep_nodeprep, stringprep_xmpp_nodeprep) /** stringprep.nodeprep(s) */ +MAKE_PREP_FUNC(Lstringprep_resourceprep, stringprep_xmpp_resourceprep) /** stringprep.resourceprep(s) */ +MAKE_PREP_FUNC(Lstringprep_saslprep, stringprep_saslprep) /** stringprep.saslprep(s) */ + +static const luaL_Reg Reg_stringprep[] = +{ + { "nameprep", Lstringprep_nameprep }, + { "nodeprep", Lstringprep_nodeprep }, + { "resourceprep", Lstringprep_resourceprep }, + { "saslprep", Lstringprep_saslprep }, + { NULL, NULL } +}; + +#else +#include +//#include +#include +#include + +static int icu_stringprep_prep(lua_State *L, const UStringPrepProfile *profile) +{ + size_t len; + const char *s; + UnicodeString ustr; + UErrorCode err = U_ZERO_ERROR; + UChar dest[1024]; + char output[1024]; + if(!lua_isstring(L, 1)) { + lua_pushnil(L); + return 1; + } + s = lua_tolstring(L, 1, &len); + if (len >= 1024) { + lua_pushnil(L); + return 1; // TODO return error message + } + ustr = UnicodeString::fromUTF8(s); + len = usprep_prepare(profile, ustr.getBuffer(), ustr.length(), dest, 1024, 0, NULL, &err); + if (U_FAILURE(err)) { + lua_pushnil(L); + return 1; + } else { + CheckedArrayByteSink output_sink(output, 1024); + UnicodeString dest_str(TRUE, dest, len); + dest_str.toUTF8(output_sink); + lua_pushstring(L, output); + return 1; + } +} + +UStringPrepProfile *icu_nameprep; +UStringPrepProfile *icu_nodeprep; +UStringPrepProfile *icu_resourceprep; +UStringPrepProfile *icu_saslprep; + +/* initialize global ICU stringprep profiles */ +void init_icu() +{ + UErrorCode err = U_ZERO_ERROR; + utrace_setLevel(UTRACE_VERBOSE); + icu_nameprep = usprep_openByType(USPREP_RFC3491_NAMEPREP, &err); + icu_nodeprep = usprep_openByType(USPREP_RFC3920_NODEPREP, &err); + icu_resourceprep = usprep_openByType(USPREP_RFC3920_RESOURCEPREP, &err); + icu_saslprep = usprep_openByType(USPREP_RFC4013_SASLPREP, &err); + if (U_FAILURE(err)) fprintf(stderr, "[c] util.encodings: error: %s\n", u_errorName((UErrorCode)err)); +} + +#define MAKE_PREP_FUNC(myFunc, prep) \ +static int myFunc(lua_State *L) { return icu_stringprep_prep(L, prep); } + +MAKE_PREP_FUNC(Lstringprep_nameprep, icu_nameprep) /** stringprep.nameprep(s) */ +MAKE_PREP_FUNC(Lstringprep_nodeprep, icu_nodeprep) /** stringprep.nodeprep(s) */ +MAKE_PREP_FUNC(Lstringprep_resourceprep, icu_resourceprep) /** stringprep.resourceprep(s) */ +MAKE_PREP_FUNC(Lstringprep_saslprep, icu_saslprep) /** stringprep.saslprep(s) */ + +static const luaL_Reg Reg_stringprep[] = +{ + { "nameprep", Lstringprep_nameprep }, + { "nodeprep", Lstringprep_nodeprep }, + { "resourceprep", Lstringprep_resourceprep }, + { "saslprep", Lstringprep_saslprep }, + { NULL, NULL } +}; +#endif + +/***************** IDNA *****************/ +#ifndef USE_STRINGPREP_ICU +/****************** libidn ********************/ +#include +#include +static int Lidna_to_ascii(lua_State *L) /** idna.to_ascii(s) */ +{ + size_t len; + const char *s = luaL_checklstring(L, 1, &len); + char* output = NULL; + int ret = idna_to_ascii_8z(s, &output, IDNA_USE_STD3_ASCII_RULES); + if (ret == IDNA_SUCCESS) { + lua_pushstring(L, output); + idn_free(output); + return 1; + } else { + lua_pushnil(L); + idn_free(output); + return 1; // TODO return error message + } +} + +static int Lidna_to_unicode(lua_State *L) /** idna.to_unicode(s) */ +{ + size_t len; + const char *s = luaL_checklstring(L, 1, &len); + char* output = NULL; + int ret = idna_to_unicode_8z8z(s, &output, 0); + if (ret == IDNA_SUCCESS) { + lua_pushstring(L, output); + idn_free(output); + return 1; + } else { + lua_pushnil(L); + idn_free(output); + return 1; // TODO return error message + } +} +#else +#include +#include +/* IDNA2003 or IDNA2008 ? ? ? */ +static int Lidna_to_ascii(lua_State *L) /** idna.to_ascii(s) */ +{ + size_t len; + int32_t out_len; + const char *s = luaL_checklstring(L, 1, &len); + UnicodeString ustr; + UErrorCode err = U_ZERO_ERROR; + UChar dest[1024]; + char output[1024]; + + ustr = UnicodeString::fromUTF8(s); + out_len = uidna_IDNToASCII(ustr.getBuffer(), ustr.length(), dest, 1024, UIDNA_USE_STD3_RULES, NULL, &err); + if (U_FAILURE(err)) { + lua_pushnil(L); + return 1; + } else { + CheckedArrayByteSink output_sink(output, 1024); + UnicodeString dest_str(TRUE, dest, out_len); + dest_str.toUTF8(output_sink); + lua_pushstring(L, output); + return 1; + } +} + +static int Lidna_to_unicode(lua_State *L) /** idna.to_unicode(s) */ +{ + size_t len; + int32_t out_len; + const char *s = luaL_checklstring(L, 1, &len); + UnicodeString ustr; + UErrorCode err = U_ZERO_ERROR; + UChar dest[1024]; + char output[1024]; + + ustr = UnicodeString::fromUTF8(s); + out_len = uidna_IDNToUnicode(ustr.getBuffer(), ustr.length(), dest, 1024, UIDNA_USE_STD3_RULES, NULL, &err); + if (U_FAILURE(err)) { + lua_pushnil(L); + return 1; + } else { + CheckedArrayByteSink output_sink(output, 1024); + UnicodeString dest_str(TRUE, dest, out_len); + dest_str.toUTF8(output_sink); + lua_pushstring(L, output); + return 1; + } +} +#endif + +static const luaL_Reg Reg_idna[] = +{ + { "to_ascii", Lidna_to_ascii }, + { "to_unicode", Lidna_to_unicode }, + { NULL, NULL } +}; + +/***************** end *****************/ + +static const luaL_Reg Reg[] = +{ + { NULL, NULL } +}; + +LUALIB_API int luaopen_util_encodings(lua_State *L) +{ +#ifdef USE_STRINGPREP_ICU + init_icu(); +#endif + luaL_register(L, "encodings", Reg); + + lua_pushliteral(L, "base64"); + lua_newtable(L); + luaL_register(L, NULL, Reg_base64); + lua_settable(L,-3); + + lua_pushliteral(L, "stringprep"); + lua_newtable(L); + luaL_register(L, NULL, Reg_stringprep); + lua_settable(L,-3); + + lua_pushliteral(L, "idna"); + lua_newtable(L); + luaL_register(L, NULL, Reg_idna); + lua_settable(L,-3); + + lua_pushliteral(L, "version"); /** version */ + lua_pushliteral(L, "-3.14"); + lua_settable(L,-3); + return 1; +}