util-src/encodings.c
author Kim Alvefur <zash@zash.se>
Mon, 23 Mar 2015 14:26:34 +0100
changeset 6594 fe3018a2f187
parent 4302 bbb0bf0a09f5
child 6595 141afe8a167b
permissions -rw-r--r--
util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
2923
b7049746bd29 Update copyright headers for 2010
Matthew Wild <mwild1@gmail.com>
parents: 2572
diff changeset
     1
/* Prosody IM
b7049746bd29 Update copyright headers for 2010
Matthew Wild <mwild1@gmail.com>
parents: 2572
diff changeset
     2
-- Copyright (C) 2008-2010 Matthew Wild
b7049746bd29 Update copyright headers for 2010
Matthew Wild <mwild1@gmail.com>
parents: 2572
diff changeset
     3
-- Copyright (C) 2008-2010 Waqas Hussain
6594
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
     4
-- Copyright (C) 1994-2015 Lua.org, PUC-Rio.
520
e96ac4bb6dd8 and the C files too
Matthew Wild <mwild1@gmail.com>
parents: 474
diff changeset
     5
-- 
766
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
     6
-- This project is MIT/X11 licensed. Please see the
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
     7
-- COPYING file in the source package for more information.
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
     8
--
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
     9
*/
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    10
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    11
/*
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    12
* encodings.c
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    13
* Lua library for base64, stringprep and idna encodings
520
e96ac4bb6dd8 and the C files too
Matthew Wild <mwild1@gmail.com>
parents: 474
diff changeset
    14
*/
e96ac4bb6dd8 and the C files too
Matthew Wild <mwild1@gmail.com>
parents: 474
diff changeset
    15
3965
4ae4b2c0e99d util.encodings: Switch comment styles to build ok as ANSI C
Matthew Wild <mwild1@gmail.com>
parents: 3769
diff changeset
    16
/* Newer MSVC compilers deprecate strcpy as unsafe, but we use it in a safe way */
766
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    17
#define _CRT_SECURE_NO_DEPRECATE
520
e96ac4bb6dd8 and the C files too
Matthew Wild <mwild1@gmail.com>
parents: 474
diff changeset
    18
766
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    19
#include <string.h>
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    20
#include <stdlib.h>
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    21
#include "lua.h"
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    22
#include "lauxlib.h"
3769
9338d0785277 util-src/Makefile, util-src/encodings.c{,pp}: Port ICU code to C, rename encodings.cpp back to .c and amend the Makefile accordingly
Matthew Wild <mwild1@gmail.com>
parents: 3764
diff changeset
    23
3762
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
    24
/***************** BASE64 *****************/
766
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    25
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    26
static const char code[]=
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    27
"ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/";
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    28
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    29
static void base64_encode(luaL_Buffer *b, unsigned int c1, unsigned int c2, unsigned int c3, int n)
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    30
{
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    31
	unsigned long tuple=c3+256UL*(c2+256UL*c1);
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    32
	int i;
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    33
	char s[4];
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    34
	for (i=0; i<4; i++) {
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    35
		s[3-i] = code[tuple % 64];
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    36
		tuple /= 64;
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    37
	}
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    38
	for (i=n+1; i<4; i++) s[i]='=';
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    39
	luaL_addlstring(b,s,4);
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    40
}
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    41
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    42
static int Lbase64_encode(lua_State *L)		/** encode(s) */
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    43
{
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    44
	size_t l;
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    45
	const unsigned char *s=(const unsigned char*)luaL_checklstring(L,1,&l);
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    46
	luaL_Buffer b;
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    47
	int n;
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    48
	luaL_buffinit(L,&b);
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    49
	for (n=l/3; n--; s+=3) base64_encode(&b,s[0],s[1],s[2],3);
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    50
	switch (l%3)
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    51
	{
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    52
		case 1: base64_encode(&b,s[0],0,0,1);		break;
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    53
		case 2: base64_encode(&b,s[0],s[1],0,2);		break;
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    54
	}
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    55
	luaL_pushresult(&b);
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    56
	return 1;
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    57
}
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    58
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    59
static void base64_decode(luaL_Buffer *b, int c1, int c2, int c3, int c4, int n)
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    60
{
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    61
	unsigned long tuple=c4+64L*(c3+64L*(c2+64L*c1));
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    62
	char s[3];
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    63
	switch (--n)
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    64
	{
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    65
		case 3: s[2]=(char) tuple;
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    66
		case 2: s[1]=(char) (tuple >> 8);
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    67
		case 1: s[0]=(char) (tuple >> 16);
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    68
	}
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    69
	luaL_addlstring(b,s,n);
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    70
}
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    71
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    72
static int Lbase64_decode(lua_State *L)		/** decode(s) */
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    73
{
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    74
	size_t l;
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    75
	const char *s=luaL_checklstring(L,1,&l);
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    76
	luaL_Buffer b;
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    77
	int n=0;
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    78
	char t[4];
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    79
	luaL_buffinit(L,&b);
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    80
	for (;;)
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    81
	{
601
6cb908ef01c8 Fixed util.encodings.base64.decode to not truncate results when encountering an '=' before the end of the given input.
Waqas Hussain <waqas20@gmail.com>
parents: 520
diff changeset
    82
		int c=*s++;
766
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    83
		switch (c)
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    84
		{
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    85
			const char *p;
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    86
			default:
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    87
				p=strchr(code,c); if (p==NULL) return 0;
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    88
				t[n++]= (char) (p-code);
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    89
				if (n==4)
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    90
				{
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    91
					base64_decode(&b,t[0],t[1],t[2],t[3],4);
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    92
					n=0;
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    93
				}
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    94
				break;
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    95
			case '=':
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    96
				switch (n)
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    97
				{
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    98
					case 1: base64_decode(&b,t[0],0,0,0,1);		break;
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
    99
					case 2: base64_decode(&b,t[0],t[1],0,0,2);	break;
601
6cb908ef01c8 Fixed util.encodings.base64.decode to not truncate results when encountering an '=' before the end of the given input.
Waqas Hussain <waqas20@gmail.com>
parents: 520
diff changeset
   100
					case 3: base64_decode(&b,t[0],t[1],t[2],0,3);	break;
6cb908ef01c8 Fixed util.encodings.base64.decode to not truncate results when encountering an '=' before the end of the given input.
Waqas Hussain <waqas20@gmail.com>
parents: 520
diff changeset
   101
				}
6cb908ef01c8 Fixed util.encodings.base64.decode to not truncate results when encountering an '=' before the end of the given input.
Waqas Hussain <waqas20@gmail.com>
parents: 520
diff changeset
   102
				n=0;
766
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   103
				break;
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   104
			case 0:
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   105
				luaL_pushresult(&b);
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   106
				return 1;
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   107
			case '\n': case '\r': case '\t': case ' ': case '\f': case '\b':
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   108
				break;
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   109
		}
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   110
	}
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   111
}
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   112
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   113
static const luaL_Reg Reg_base64[] =
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   114
{
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   115
	{ "encode",	Lbase64_encode	},
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   116
	{ "decode",	Lbase64_decode	},
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   117
	{ NULL,		NULL	}
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   118
};
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   119
6594
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   120
/******************* UTF-8 ********************/
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   121
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   122
/*
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   123
 * Adapted from Lua 5.3
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   124
 * Needed because libidn does not validate that input is valid UTF-8
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   125
 */
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   126
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   127
#define MAXUNICODE	0x10FFFF
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   128
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   129
/*
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   130
 * Decode one UTF-8 sequence, returning NULL if byte sequence is invalid.
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   131
 */
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   132
static const char *utf8_decode (const char *o, int *val) {
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   133
	static unsigned int limits[] = {0xFF, 0x7F, 0x7FF, 0xFFFF};
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   134
	const unsigned char *s = (const unsigned char *)o;
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   135
	unsigned int c = s[0];
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   136
	unsigned int res = 0;  /* final result */
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   137
	if (c < 0x80)  /* ascii? */
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   138
		res = c;
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   139
	else {
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   140
		int count = 0;  /* to count number of continuation bytes */
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   141
		while (c & 0x40) {  /* still have continuation bytes? */
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   142
			int cc = s[++count];  /* read next byte */
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   143
			if ((cc & 0xC0) != 0x80)  /* not a continuation byte? */
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   144
				return NULL;  /* invalid byte sequence */
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   145
			res = (res << 6) | (cc & 0x3F);  /* add lower 6 bits from cont. byte */
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   146
			c <<= 1;  /* to test next bit */
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   147
		}
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   148
		res |= ((c & 0x7F) << (count * 5));  /* add first byte */
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   149
		if (count > 3 || res > MAXUNICODE || res <= limits[count] || (0xd800 <= res && res <= 0xdfff) )
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   150
			return NULL;  /* invalid byte sequence */
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   151
		s += count;  /* skip continuation bytes read */
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   152
	}
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   153
	if (val) *val = res;
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   154
	return (const char *)s + 1;  /* +1 to include first byte */
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   155
}
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   156
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   157
/*
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   158
 * Check that a string is valid UTF-8
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   159
 * Returns NULL if not
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   160
 */
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   161
const char* check_utf8 (lua_State *L, int idx, size_t *l) {
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   162
	size_t pos, len;
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   163
	const char *s = luaL_checklstring(L, 1, &len);
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   164
	pos = 0;
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   165
	while (pos <= len) {
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   166
		const char *s1 = utf8_decode(s + pos, NULL);
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   167
		if (s1 == NULL) {  /* conversion error? */
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   168
			return NULL;
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   169
		}
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   170
		pos = s1 - s;
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   171
	}
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   172
	if(l != NULL) {
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   173
		*l = len;
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   174
	}
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   175
	return s;
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   176
}
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   177
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   178
766
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   179
/***************** STRINGPREP *****************/
4272
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   180
#ifdef USE_STRINGPREP_ICU
3769
9338d0785277 util-src/Makefile, util-src/encodings.c{,pp}: Port ICU code to C, rename encodings.cpp back to .c and amend the Makefile accordingly
Matthew Wild <mwild1@gmail.com>
parents: 3764
diff changeset
   181
3762
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   182
#include <unicode/usprep.h>
3769
9338d0785277 util-src/Makefile, util-src/encodings.c{,pp}: Port ICU code to C, rename encodings.cpp back to .c and amend the Makefile accordingly
Matthew Wild <mwild1@gmail.com>
parents: 3764
diff changeset
   183
#include <unicode/ustring.h>
3762
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   184
#include <unicode/utrace.h>
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   185
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   186
static int icu_stringprep_prep(lua_State *L, const UStringPrepProfile *profile)
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   187
{
3769
9338d0785277 util-src/Makefile, util-src/encodings.c{,pp}: Port ICU code to C, rename encodings.cpp back to .c and amend the Makefile accordingly
Matthew Wild <mwild1@gmail.com>
parents: 3764
diff changeset
   188
	size_t input_len;
9338d0785277 util-src/Makefile, util-src/encodings.c{,pp}: Port ICU code to C, rename encodings.cpp back to .c and amend the Makefile accordingly
Matthew Wild <mwild1@gmail.com>
parents: 3764
diff changeset
   189
	int32_t unprepped_len, prepped_len, output_len;
9338d0785277 util-src/Makefile, util-src/encodings.c{,pp}: Port ICU code to C, rename encodings.cpp back to .c and amend the Makefile accordingly
Matthew Wild <mwild1@gmail.com>
parents: 3764
diff changeset
   190
	const char *input;
9338d0785277 util-src/Makefile, util-src/encodings.c{,pp}: Port ICU code to C, rename encodings.cpp back to .c and amend the Makefile accordingly
Matthew Wild <mwild1@gmail.com>
parents: 3764
diff changeset
   191
	char output[1024];
9338d0785277 util-src/Makefile, util-src/encodings.c{,pp}: Port ICU code to C, rename encodings.cpp back to .c and amend the Makefile accordingly
Matthew Wild <mwild1@gmail.com>
parents: 3764
diff changeset
   192
9338d0785277 util-src/Makefile, util-src/encodings.c{,pp}: Port ICU code to C, rename encodings.cpp back to .c and amend the Makefile accordingly
Matthew Wild <mwild1@gmail.com>
parents: 3764
diff changeset
   193
	UChar unprepped[1024]; /* Temporary unicode buffer (1024 characters) */
9338d0785277 util-src/Makefile, util-src/encodings.c{,pp}: Port ICU code to C, rename encodings.cpp back to .c and amend the Makefile accordingly
Matthew Wild <mwild1@gmail.com>
parents: 3764
diff changeset
   194
	UChar prepped[1024];
9338d0785277 util-src/Makefile, util-src/encodings.c{,pp}: Port ICU code to C, rename encodings.cpp back to .c and amend the Makefile accordingly
Matthew Wild <mwild1@gmail.com>
parents: 3764
diff changeset
   195
	
3762
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   196
	UErrorCode err = U_ZERO_ERROR;
3769
9338d0785277 util-src/Makefile, util-src/encodings.c{,pp}: Port ICU code to C, rename encodings.cpp back to .c and amend the Makefile accordingly
Matthew Wild <mwild1@gmail.com>
parents: 3764
diff changeset
   197
3762
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   198
	if(!lua_isstring(L, 1)) {
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   199
		lua_pushnil(L);
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   200
		return 1;
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   201
	}
3769
9338d0785277 util-src/Makefile, util-src/encodings.c{,pp}: Port ICU code to C, rename encodings.cpp back to .c and amend the Makefile accordingly
Matthew Wild <mwild1@gmail.com>
parents: 3764
diff changeset
   202
	input = lua_tolstring(L, 1, &input_len);
9338d0785277 util-src/Makefile, util-src/encodings.c{,pp}: Port ICU code to C, rename encodings.cpp back to .c and amend the Makefile accordingly
Matthew Wild <mwild1@gmail.com>
parents: 3764
diff changeset
   203
	if (input_len >= 1024) {
3762
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   204
		lua_pushnil(L);
3769
9338d0785277 util-src/Makefile, util-src/encodings.c{,pp}: Port ICU code to C, rename encodings.cpp back to .c and amend the Makefile accordingly
Matthew Wild <mwild1@gmail.com>
parents: 3764
diff changeset
   205
		return 1;
3762
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   206
	}
3769
9338d0785277 util-src/Makefile, util-src/encodings.c{,pp}: Port ICU code to C, rename encodings.cpp back to .c and amend the Makefile accordingly
Matthew Wild <mwild1@gmail.com>
parents: 3764
diff changeset
   207
	u_strFromUTF8(unprepped, 1024, &unprepped_len, input, input_len, &err);
4273
7f789266b741 util.encodings: Check return values before proceeding
Paul Aurich <paul@darkrain42.org>
parents: 4272
diff changeset
   208
	if (U_FAILURE(err)) {
4302
bbb0bf0a09f5 util.encodings: Fix small typo introduced in 7f789266b741
Matthew Wild <mwild1@gmail.com>
parents: 4273
diff changeset
   209
		lua_pushnil(L);
4273
7f789266b741 util.encodings: Check return values before proceeding
Paul Aurich <paul@darkrain42.org>
parents: 4272
diff changeset
   210
		return 1;
7f789266b741 util.encodings: Check return values before proceeding
Paul Aurich <paul@darkrain42.org>
parents: 4272
diff changeset
   211
	}
3769
9338d0785277 util-src/Makefile, util-src/encodings.c{,pp}: Port ICU code to C, rename encodings.cpp back to .c and amend the Makefile accordingly
Matthew Wild <mwild1@gmail.com>
parents: 3764
diff changeset
   212
	prepped_len = usprep_prepare(profile, unprepped, unprepped_len, prepped, 1024, 0, NULL, &err);
3762
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   213
	if (U_FAILURE(err)) {
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   214
		lua_pushnil(L);
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   215
		return 1;
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   216
	} else {
3769
9338d0785277 util-src/Makefile, util-src/encodings.c{,pp}: Port ICU code to C, rename encodings.cpp back to .c and amend the Makefile accordingly
Matthew Wild <mwild1@gmail.com>
parents: 3764
diff changeset
   217
		u_strToUTF8(output, 1024, &output_len, prepped, prepped_len, &err);
4273
7f789266b741 util.encodings: Check return values before proceeding
Paul Aurich <paul@darkrain42.org>
parents: 4272
diff changeset
   218
		if (U_SUCCESS(err) && output_len < 1024)
3769
9338d0785277 util-src/Makefile, util-src/encodings.c{,pp}: Port ICU code to C, rename encodings.cpp back to .c and amend the Makefile accordingly
Matthew Wild <mwild1@gmail.com>
parents: 3764
diff changeset
   219
			lua_pushlstring(L, output, output_len);
9338d0785277 util-src/Makefile, util-src/encodings.c{,pp}: Port ICU code to C, rename encodings.cpp back to .c and amend the Makefile accordingly
Matthew Wild <mwild1@gmail.com>
parents: 3764
diff changeset
   220
		else
9338d0785277 util-src/Makefile, util-src/encodings.c{,pp}: Port ICU code to C, rename encodings.cpp back to .c and amend the Makefile accordingly
Matthew Wild <mwild1@gmail.com>
parents: 3764
diff changeset
   221
			lua_pushnil(L);
3762
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   222
		return 1;
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   223
	}
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   224
}
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   225
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   226
UStringPrepProfile *icu_nameprep;
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   227
UStringPrepProfile *icu_nodeprep;
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   228
UStringPrepProfile *icu_resourceprep; 
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   229
UStringPrepProfile *icu_saslprep;
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   230
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   231
/* initialize global ICU stringprep profiles */
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   232
void init_icu()
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   233
{
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   234
	UErrorCode err = U_ZERO_ERROR;
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   235
	utrace_setLevel(UTRACE_VERBOSE);
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   236
	icu_nameprep = usprep_openByType(USPREP_RFC3491_NAMEPREP, &err);
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   237
	icu_nodeprep = usprep_openByType(USPREP_RFC3920_NODEPREP, &err);
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   238
	icu_resourceprep = usprep_openByType(USPREP_RFC3920_RESOURCEPREP, &err);
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   239
	icu_saslprep = usprep_openByType(USPREP_RFC4013_SASLPREP, &err);
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   240
	if (U_FAILURE(err)) fprintf(stderr, "[c] util.encodings: error: %s\n", u_errorName((UErrorCode)err));
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   241
}
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   242
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   243
#define MAKE_PREP_FUNC(myFunc, prep) \
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   244
static int myFunc(lua_State *L) { return icu_stringprep_prep(L, prep); }
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   245
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   246
MAKE_PREP_FUNC(Lstringprep_nameprep, icu_nameprep)		/** stringprep.nameprep(s) */
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   247
MAKE_PREP_FUNC(Lstringprep_nodeprep, icu_nodeprep)		/** stringprep.nodeprep(s) */
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   248
MAKE_PREP_FUNC(Lstringprep_resourceprep, icu_resourceprep)		/** stringprep.resourceprep(s) */
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   249
MAKE_PREP_FUNC(Lstringprep_saslprep, icu_saslprep)		/** stringprep.saslprep(s) */
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   250
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   251
static const luaL_Reg Reg_stringprep[] =
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   252
{
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   253
	{ "nameprep",	Lstringprep_nameprep	},
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   254
	{ "nodeprep",	Lstringprep_nodeprep	},
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   255
	{ "resourceprep",	Lstringprep_resourceprep	},
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   256
	{ "saslprep",	Lstringprep_saslprep	},
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   257
	{ NULL,		NULL	}
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   258
};
4272
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   259
#else /* USE_STRINGPREP_ICU */
3762
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   260
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   261
/****************** libidn ********************/
3769
9338d0785277 util-src/Makefile, util-src/encodings.c{,pp}: Port ICU code to C, rename encodings.cpp back to .c and amend the Makefile accordingly
Matthew Wild <mwild1@gmail.com>
parents: 3764
diff changeset
   262
4272
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   263
#include <stringprep.h>
3769
9338d0785277 util-src/Makefile, util-src/encodings.c{,pp}: Port ICU code to C, rename encodings.cpp back to .c and amend the Makefile accordingly
Matthew Wild <mwild1@gmail.com>
parents: 3764
diff changeset
   264
4272
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   265
static int stringprep_prep(lua_State *L, const Stringprep_profile *profile)
766
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   266
{
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   267
	size_t len;
4272
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   268
	const char *s;
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   269
	char string[1024];
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   270
	int ret;
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   271
	if(!lua_isstring(L, 1)) {
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   272
		lua_pushnil(L);
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   273
		return 1;
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   274
	}
6594
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   275
	s = check_utf8(L, 1, &len);
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   276
	if (s == NULL || len >= 1024 || len != strlen(s)) {
4272
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   277
		lua_pushnil(L);
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   278
		return 1; /* TODO return error message */
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   279
	}
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   280
	strcpy(string, s);
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   281
	ret = stringprep(string, 1024, (Stringprep_profile_flags)0, profile);
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   282
	if (ret == STRINGPREP_OK) {
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   283
		lua_pushstring(L, string);
766
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   284
		return 1;
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   285
	} else {
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   286
		lua_pushnil(L);
3965
4ae4b2c0e99d util.encodings: Switch comment styles to build ok as ANSI C
Matthew Wild <mwild1@gmail.com>
parents: 3769
diff changeset
   287
		return 1; /* TODO return error message */
766
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   288
	}
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   289
}
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   290
4272
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   291
#define MAKE_PREP_FUNC(myFunc, prep) \
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   292
static int myFunc(lua_State *L) { return stringprep_prep(L, prep); }
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   293
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   294
MAKE_PREP_FUNC(Lstringprep_nameprep, stringprep_nameprep)		/** stringprep.nameprep(s) */
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   295
MAKE_PREP_FUNC(Lstringprep_nodeprep, stringprep_xmpp_nodeprep)		/** stringprep.nodeprep(s) */
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   296
MAKE_PREP_FUNC(Lstringprep_resourceprep, stringprep_xmpp_resourceprep)		/** stringprep.resourceprep(s) */
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   297
MAKE_PREP_FUNC(Lstringprep_saslprep, stringprep_saslprep)		/** stringprep.saslprep(s) */
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   298
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   299
static const luaL_Reg Reg_stringprep[] =
766
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   300
{
4272
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   301
	{ "nameprep",	Lstringprep_nameprep	},
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   302
	{ "nodeprep",	Lstringprep_nodeprep	},
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   303
	{ "resourceprep",	Lstringprep_resourceprep	},
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   304
	{ "saslprep",	Lstringprep_saslprep	},
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   305
	{ NULL,		NULL	}
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   306
};
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   307
#endif
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   308
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   309
/***************** IDNA *****************/
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   310
#ifdef USE_STRINGPREP_ICU
3762
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   311
#include <unicode/ustdio.h>
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   312
#include <unicode/uidna.h>
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   313
/* IDNA2003 or IDNA2008 ? ? ? */
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   314
static int Lidna_to_ascii(lua_State *L)		/** idna.to_ascii(s) */
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   315
{
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   316
	size_t len;
3769
9338d0785277 util-src/Makefile, util-src/encodings.c{,pp}: Port ICU code to C, rename encodings.cpp back to .c and amend the Makefile accordingly
Matthew Wild <mwild1@gmail.com>
parents: 3764
diff changeset
   317
	int32_t ulen, dest_len, output_len;
3762
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   318
	const char *s = luaL_checklstring(L, 1, &len);
3769
9338d0785277 util-src/Makefile, util-src/encodings.c{,pp}: Port ICU code to C, rename encodings.cpp back to .c and amend the Makefile accordingly
Matthew Wild <mwild1@gmail.com>
parents: 3764
diff changeset
   319
	UChar ustr[1024];
3762
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   320
	UErrorCode err = U_ZERO_ERROR;
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   321
	UChar dest[1024];
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   322
	char output[1024];
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   323
3769
9338d0785277 util-src/Makefile, util-src/encodings.c{,pp}: Port ICU code to C, rename encodings.cpp back to .c and amend the Makefile accordingly
Matthew Wild <mwild1@gmail.com>
parents: 3764
diff changeset
   324
	u_strFromUTF8(ustr, 1024, &ulen, s, len, &err);
4273
7f789266b741 util.encodings: Check return values before proceeding
Paul Aurich <paul@darkrain42.org>
parents: 4272
diff changeset
   325
	if (U_FAILURE(err)) {
7f789266b741 util.encodings: Check return values before proceeding
Paul Aurich <paul@darkrain42.org>
parents: 4272
diff changeset
   326
		lua_pushnil(L);
7f789266b741 util.encodings: Check return values before proceeding
Paul Aurich <paul@darkrain42.org>
parents: 4272
diff changeset
   327
		return 1;
7f789266b741 util.encodings: Check return values before proceeding
Paul Aurich <paul@darkrain42.org>
parents: 4272
diff changeset
   328
	}
7f789266b741 util.encodings: Check return values before proceeding
Paul Aurich <paul@darkrain42.org>
parents: 4272
diff changeset
   329
3769
9338d0785277 util-src/Makefile, util-src/encodings.c{,pp}: Port ICU code to C, rename encodings.cpp back to .c and amend the Makefile accordingly
Matthew Wild <mwild1@gmail.com>
parents: 3764
diff changeset
   330
	dest_len = uidna_IDNToASCII(ustr, ulen, dest, 1024, UIDNA_USE_STD3_RULES, NULL, &err);
3762
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   331
	if (U_FAILURE(err)) {
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   332
		lua_pushnil(L);
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   333
		return 1;
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   334
	} else {
3769
9338d0785277 util-src/Makefile, util-src/encodings.c{,pp}: Port ICU code to C, rename encodings.cpp back to .c and amend the Makefile accordingly
Matthew Wild <mwild1@gmail.com>
parents: 3764
diff changeset
   335
		u_strToUTF8(output, 1024, &output_len, dest, dest_len, &err);
4273
7f789266b741 util.encodings: Check return values before proceeding
Paul Aurich <paul@darkrain42.org>
parents: 4272
diff changeset
   336
		if (U_SUCCESS(err) && output_len < 1024)
3769
9338d0785277 util-src/Makefile, util-src/encodings.c{,pp}: Port ICU code to C, rename encodings.cpp back to .c and amend the Makefile accordingly
Matthew Wild <mwild1@gmail.com>
parents: 3764
diff changeset
   337
			lua_pushlstring(L, output, output_len);
9338d0785277 util-src/Makefile, util-src/encodings.c{,pp}: Port ICU code to C, rename encodings.cpp back to .c and amend the Makefile accordingly
Matthew Wild <mwild1@gmail.com>
parents: 3764
diff changeset
   338
		else
9338d0785277 util-src/Makefile, util-src/encodings.c{,pp}: Port ICU code to C, rename encodings.cpp back to .c and amend the Makefile accordingly
Matthew Wild <mwild1@gmail.com>
parents: 3764
diff changeset
   339
			lua_pushnil(L);
3762
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   340
		return 1;
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   341
	}
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   342
}
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   343
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   344
static int Lidna_to_unicode(lua_State *L)		/** idna.to_unicode(s) */
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   345
{
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   346
	size_t len;
3769
9338d0785277 util-src/Makefile, util-src/encodings.c{,pp}: Port ICU code to C, rename encodings.cpp back to .c and amend the Makefile accordingly
Matthew Wild <mwild1@gmail.com>
parents: 3764
diff changeset
   347
	int32_t ulen, dest_len, output_len;
3762
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   348
	const char *s = luaL_checklstring(L, 1, &len);
4271
18d888c8c12d util.encodings: Fix idna.to_unicode
Paul Aurich <paul@darkrain42.org>
parents: 3965
diff changeset
   349
	UChar ustr[1024];
3762
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   350
	UErrorCode err = U_ZERO_ERROR;
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   351
	UChar dest[1024];
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   352
	char output[1024];
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   353
3769
9338d0785277 util-src/Makefile, util-src/encodings.c{,pp}: Port ICU code to C, rename encodings.cpp back to .c and amend the Makefile accordingly
Matthew Wild <mwild1@gmail.com>
parents: 3764
diff changeset
   354
	u_strFromUTF8(ustr, 1024, &ulen, s, len, &err);
4273
7f789266b741 util.encodings: Check return values before proceeding
Paul Aurich <paul@darkrain42.org>
parents: 4272
diff changeset
   355
	if (U_FAILURE(err)) {
7f789266b741 util.encodings: Check return values before proceeding
Paul Aurich <paul@darkrain42.org>
parents: 4272
diff changeset
   356
		lua_pushnil(L);
7f789266b741 util.encodings: Check return values before proceeding
Paul Aurich <paul@darkrain42.org>
parents: 4272
diff changeset
   357
		return 1;
7f789266b741 util.encodings: Check return values before proceeding
Paul Aurich <paul@darkrain42.org>
parents: 4272
diff changeset
   358
	}
7f789266b741 util.encodings: Check return values before proceeding
Paul Aurich <paul@darkrain42.org>
parents: 4272
diff changeset
   359
3769
9338d0785277 util-src/Makefile, util-src/encodings.c{,pp}: Port ICU code to C, rename encodings.cpp back to .c and amend the Makefile accordingly
Matthew Wild <mwild1@gmail.com>
parents: 3764
diff changeset
   360
	dest_len = uidna_IDNToUnicode(ustr, ulen, dest, 1024, UIDNA_USE_STD3_RULES, NULL, &err);
3762
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   361
	if (U_FAILURE(err)) {
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   362
		lua_pushnil(L);
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   363
		return 1;
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   364
	} else {
3769
9338d0785277 util-src/Makefile, util-src/encodings.c{,pp}: Port ICU code to C, rename encodings.cpp back to .c and amend the Makefile accordingly
Matthew Wild <mwild1@gmail.com>
parents: 3764
diff changeset
   365
		u_strToUTF8(output, 1024, &output_len, dest, dest_len, &err);
4273
7f789266b741 util.encodings: Check return values before proceeding
Paul Aurich <paul@darkrain42.org>
parents: 4272
diff changeset
   366
		if (U_SUCCESS(err) && output_len < 1024)
3769
9338d0785277 util-src/Makefile, util-src/encodings.c{,pp}: Port ICU code to C, rename encodings.cpp back to .c and amend the Makefile accordingly
Matthew Wild <mwild1@gmail.com>
parents: 3764
diff changeset
   367
			lua_pushlstring(L, output, output_len);
9338d0785277 util-src/Makefile, util-src/encodings.c{,pp}: Port ICU code to C, rename encodings.cpp back to .c and amend the Makefile accordingly
Matthew Wild <mwild1@gmail.com>
parents: 3764
diff changeset
   368
		else
9338d0785277 util-src/Makefile, util-src/encodings.c{,pp}: Port ICU code to C, rename encodings.cpp back to .c and amend the Makefile accordingly
Matthew Wild <mwild1@gmail.com>
parents: 3764
diff changeset
   369
			lua_pushnil(L);
3762
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   370
		return 1;
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   371
	}
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   372
}
4272
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   373
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   374
#else /* USE_STRINGPREP_ICU */
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   375
/****************** libidn ********************/
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   376
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   377
#include <idna.h>
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   378
#include <idn-free.h>
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   379
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   380
static int Lidna_to_ascii(lua_State *L)		/** idna.to_ascii(s) */
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   381
{
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   382
	size_t len;
6594
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   383
	const char *s = check_utf8(L, 1, &len);
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   384
	if (s == NULL || len != strlen(s)) {
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   385
		lua_pushnil(L);
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   386
		return 1; /* TODO return error message */
fe3018a2f187 util.encodings: Perform validation of UTF-8 strings before passing to libidn (Based on code from the utf8 library in Lua 5.3)
Kim Alvefur <zash@zash.se>
parents: 4302
diff changeset
   387
	}
4272
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   388
	char* output = NULL;
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   389
	int ret = idna_to_ascii_8z(s, &output, IDNA_USE_STD3_ASCII_RULES);
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   390
	if (ret == IDNA_SUCCESS) {
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   391
		lua_pushstring(L, output);
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   392
		idn_free(output);
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   393
		return 1;
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   394
	} else {
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   395
		lua_pushnil(L);
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   396
		idn_free(output);
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   397
		return 1; /* TODO return error message */
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   398
	}
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   399
}
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   400
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   401
static int Lidna_to_unicode(lua_State *L)		/** idna.to_unicode(s) */
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   402
{
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   403
	size_t len;
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   404
	const char *s = luaL_checklstring(L, 1, &len);
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   405
	char* output = NULL;
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   406
	int ret = idna_to_unicode_8z8z(s, &output, 0);
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   407
	if (ret == IDNA_SUCCESS) {
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   408
		lua_pushstring(L, output);
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   409
		idn_free(output);
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   410
		return 1;
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   411
	} else {
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   412
		lua_pushnil(L);
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   413
		idn_free(output);
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   414
		return 1; /* TODO return error message */
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   415
	}
0a4ce2086a88 util.encodings: Swap code order ("ifndef" bugs me)
Paul Aurich <paul@darkrain42.org>
parents: 4271
diff changeset
   416
}
3762
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   417
#endif
766
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   418
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   419
static const luaL_Reg Reg_idna[] =
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   420
{
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   421
	{ "to_ascii",	Lidna_to_ascii	},
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   422
	{ "to_unicode",	Lidna_to_unicode	},
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   423
	{ NULL,		NULL	}
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   424
};
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   425
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   426
/***************** end *****************/
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   427
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   428
static const luaL_Reg Reg[] =
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   429
{
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   430
	{ NULL,		NULL	}
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   431
};
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   432
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   433
LUALIB_API int luaopen_util_encodings(lua_State *L)
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   434
{
3762
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   435
#ifdef USE_STRINGPREP_ICU
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   436
	init_icu();
f02bac902a1e util.encodings: Support for ICU for IDNA operations.
Tobias Markmann <tm@ayena.de>
parents: 2923
diff changeset
   437
#endif
766
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   438
	luaL_register(L, "encodings", Reg);
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   439
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   440
	lua_pushliteral(L, "base64");
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   441
	lua_newtable(L);
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   442
	luaL_register(L, NULL, Reg_base64);
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   443
	lua_settable(L,-3);
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   444
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   445
	lua_pushliteral(L, "stringprep");
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   446
	lua_newtable(L);
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   447
	luaL_register(L, NULL, Reg_stringprep);
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   448
	lua_settable(L,-3);
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   449
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   450
	lua_pushliteral(L, "idna");
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   451
	lua_newtable(L);
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   452
	luaL_register(L, NULL, Reg_idna);
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   453
	lua_settable(L,-3);
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   454
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   455
	lua_pushliteral(L, "version");			/** version */
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   456
	lua_pushliteral(L, "-3.14");
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   457
	lua_settable(L,-3);
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   458
	return 1;
433a5226267f Licensing/version updates for some files (forgot to commit, doh...)
Matthew Wild <mwild1@gmail.com>
parents: 601
diff changeset
   459
}