src/modules/m_nationalchars.cpp

   1 /*       +------------------------------------+
   2  *       | Inspire Internet Relay Chat Daemon |
   3  *       +------------------------------------+
   4  *
   5  *  InspIRCd: (C) 2002-2009 InspIRCd Development Team
   6  * See: http://www.inspircd.org/wiki/index.php/Credits
   7  *
   8  * This program is free but copyrighted software; see
   9  *            the file COPYING for details.
  10  *
  11  * ---------------------------------------------------
  12  */
  13
  14 /* Contains a code of Unreal IRCd + Bynets patch ( http://www.unrealircd.com/ and http://www.bynets.org/ )
  15    Changed at 2008-06-15 - 2008-12-15
  16    by Chernov-Phoenix Alexey (Phoenix@RusNet) mailto:phoenix /email address separator/ pravmail.ru */
  17
  18 #include "inspircd.h"
  19 #include "caller.h"
  20 #include <fstream>
  21
  22 /* $ModDesc: Provides an ability to have non-RFC1459 nicks & support for national CASEMAPPING */
  23
  24 DEFINE_HANDLER2(lwbNickHandler, bool, const char*, size_t);
  25
  26                                                                  /*,m_reverse_additionalUp[256];*/
  27 static unsigned char m_reverse_additional[256],m_additionalMB[256],m_additionalUtf8[256],m_additionalUtf8range[256];
  28
  29 void SearchAndReplace(std::string& newline, const std::string &find, const std::string &replace)
  30 {
  31         std::string::size_type x = newline.find(find);
  32         while (x != std::string::npos)
  33         {
  34                 newline.erase(x, find.length());
  35                 newline.insert(x, replace);
  36                 x = newline.find(find);
  37         }
  38 }
  39
  40
  41 char utf8checkrest(unsigned char * mb, unsigned char cnt)
  42 {
  43         for (unsigned char * tmp=mb; tmp<mb+cnt; tmp++)
  44         {
  45                 if ((*tmp<128)||(*tmp>191))
  46                         return -1;
  47         }
  48         return cnt+1;
  49 }
  50
  51
  52 char utf8size(unsigned char * mb)
  53 {
  54         if (!*mb)
  55                 return -1;
  56         if (!(*mb & 128))
  57                 return 1;
  58         if ((*mb & 224)==192)
  59                 return utf8checkrest(mb+1,1);
  60         if ((*mb & 240)==224)
  61                 return utf8checkrest(mb+1,2);
  62         if ((*mb & 248)==240)
  63                 return utf8checkrest(mb+1,3);
  64         return -1;
  65 }
  66
  67
  68 /* Conditions added */
  69 bool lwbNickHandler::Call(const char* n, size_t max)
  70 {
  71         if (!n || !*n)
  72                 return false;
  73
  74         unsigned int p = 0;
  75         for (const char* i = n; *i; i++, p++)
  76         {
  77                 /* 1. Multibyte encodings support:  */
  78                 /* 1.1. 16bit char. areas, e.g. chinese:*/
  79
  80                 /* if current character is the last, we DO NOT check it against multibyte table */
  81                                                                  /* if there are mbtable ranges, use ONLY them. No 8bit at all */
  82                 if (i[1] && m_additionalMB[0])
  83                 {
  84                         /* otherwise let's take a look at the current character and the following one */
  85                         bool found=false;
  86                         for(unsigned char * mb=m_additionalMB; (*mb) && (mb<m_additionalMB+sizeof(m_additionalMB)); mb+=4)
  87                         {
  88                                 if ( (i[0]>=mb[0]) && (i[0]<=mb[1]) && (i[1]>=mb[2]) && (i[1]<=mb[3]) )
  89                                 {
  90                                         /* multibyte range character found */
  91                                         i++;p++;
  92                                         found=true;
  93                                         break;
  94                                 }
  95                         }
  96                         if (found)
  97                                 /* next char! */
  98                                 continue;
  99                         else
 100                                 /* there are ranges, but incorrect char (8bit?) given, sorry */
 101                                 return false;
 102                 }
 103
 104                 /* 2. 8bit character support */
 105
 106                 if ( ((*i >= 'A') && (*i <= '}'))
 107                         || m_reverse_additional[(unsigned char)*i])
 108                 {
 109                         /* "A"-"}" can occur anywhere in a nickname */
 110                         continue;
 111                 }
 112
 113                 if ((((*i >= '0') && (*i <= '9')) || (*i == '-')) && (i > n))
 114                 {
 115                         /* "0"-"9", "-" can occur anywhere BUT the first char of a nickname */
 116                         continue;
 117                 }
 118
 119                 /* 3.1. Check against a simple UTF-8 characters enumeration */
 120                 int cursize,ncursize;    /*size of a current character*/
 121                 ncursize=utf8size((unsigned char *)i);
 122                 /* do check only if current multibyte character is valid UTF-8 only */
 123                 if (ncursize!=-1)
 124                 {
 125                         bool found=false;
 126                         for(unsigned char * mb=m_additionalUtf8;
 127                                 (utf8size(mb)!=-1) && (mb<m_additionalUtf8+sizeof(m_additionalUtf8));
 128                                 mb+=cursize)
 129                         {
 130                                 cursize=utf8size(mb);
 131                                 /* Size differs? Pick the next! */
 132                                 if (cursize!=ncursize)
 133                                         continue;
 134                                 if (!strncmp(i,(char *)mb,cursize))
 135                                 {
 136                                         i+=cursize-1;
 137                                         p+=cursize-1;
 138                                         found=true;
 139                                         break;
 140                                 }
 141                         }
 142                         if (found)
 143                                 continue;
 144                         /* 3.2. Check against an UTF-8 ranges: <start character> and <lenght of the range>.
 145                         Also char. is to be checked if it is a valid UTF-8 one */
 146
 147                         found=false;
 148                         for(unsigned char * mb=m_additionalUtf8range;
 149                                 (utf8size(mb)!=-1) && (mb<m_additionalUtf8range+sizeof(m_additionalUtf8range));
 150                                 mb+=cursize+1)
 151                         {
 152                                 cursize=utf8size(mb);
 153                                 /* Size differs? Pick the next! */
 154                                 if ((cursize!=ncursize)||(!mb[cursize]))
 155                                         continue;
 156
 157                                 unsigned char uright[5]={0,0,0,0,0};
 158                                 strncpy((char *)uright,(char *)mb, cursize);
 159
 160                                 if((uright[cursize-1]+mb[cursize]-1>0xff) && (cursize!=1))
 161                                 {
 162                                         uright[cursize-2]+=1;
 163                                 }
 164                                 uright[cursize-1]=(uright[cursize-1]+mb[cursize]-1) % 0x100;
 165
 166                                 if ((strncmp(i,(char *)mb,cursize)>=0) && (strncmp(i,(char *)uright,cursize)<=0))
 167                                 {
 168                                         i+=cursize-1;
 169                                         p+=cursize-1;
 170                                         found=true;
 171                                         break;
 172                                 }
 173                         }
 174                         if (found)
 175                                 continue;
 176                 }
 177
 178                 /* invalid character! abort */
 179                 return false;
 180         }
 181
 182         /* too long? or not -- pointer arithmetic rocks */
 183         return (p < max);
 184 }
 185
 186
 187 class ModuleNationalChars : public Module
 188 {
 189         private:
 190                 InspIRCd* ServerInstance;
 191                 lwbNickHandler * myhandler;
 192                 std::string charset,casemapping;
 193                 unsigned char  m_additional[256],m_additionalUp[256],m_lower[256], m_upper[256];
 194                 caller2<bool, const char*, size_t> * rememberer;
 195                 bool forcequit;
 196                 const unsigned char * lowermap_rememberer;
 197         public:
 198                 ModuleNationalChars(InspIRCd* Me)
 199                         : Module(Me)
 200                 {
 201                         rememberer=(caller2<bool, const char*, size_t> *) malloc(sizeof(rememberer));
 202                         lowermap_rememberer=national_case_insensitive_map;
 203                         memcpy(m_lower,rfc_case_insensitive_map,256);
 204                         national_case_insensitive_map=m_lower;
 205
 206                         ServerInstance=Me;
 207                         *rememberer=ServerInstance->IsNick;
 208                         myhandler=new lwbNickHandler(ServerInstance);
 209                         ServerInstance->IsNick=myhandler;
 210                         Implementation eventlist[] = { I_OnRehash, I_On005Numeric };
 211                         ServerInstance->Modules->Attach(eventlist, this, 2);
 212                         OnRehash(NULL, "");
 213                 }
 214
 215                 virtual void On005Numeric(std::string &output)
 216                 {
 217                         std::string tmp(casemapping);
 218                         tmp.insert(0,"CASEMAPPING=");
 219                         SearchAndReplace(output,"CASEMAPPING=rfc1459",tmp);
 220                 }
 221
 222                 virtual void OnRehash(User* user, const std::string &parameter)
 223                 {
 224                         ConfigReader* conf = new ConfigReader(ServerInstance);
 225                         charset = conf->ReadValue("nationalchars", "file", 0);
 226                         casemapping = conf->ReadValue("nationalchars", "casemapping", charset, 0, false);
 227                         charset.insert(0,"../locales/");
 228                         unsigned char * tables[7]=
 229                         {
 230                                 m_additional,m_additionalMB,m_additionalUp,m_lower,m_upper,
 231                                 m_additionalUtf8,m_additionalUtf8range
 232                         };
 233                         loadtables(charset,tables,7,5);
 234                         forcequit = conf->ReadFlag("nationalchars", "forcequit", 0);
 235                         CheckForceQuit("National character set changed");
 236                         delete conf;
 237                 }
 238
 239                 void CheckForceQuit(const char * message)
 240                 {
 241                         if (!forcequit)
 242                                 return;
 243
 244                         std::vector<User*> purge;
 245                         std::vector<User*>::iterator iter;
 246                         purge.clear();
 247                         for (iter=ServerInstance->Users->local_users.begin();iter!=ServerInstance->Users->local_users.end();++iter)
 248                         {
 249                                 if (!ServerInstance->IsNick((*iter)->nick.c_str(), ServerInstance->Config->Limits.NickMax))
 250                                         purge.push_back(*iter);
 251                         }
 252                         for (iter=purge.begin();iter!=purge.end();++iter)
 253                         {
 254                                 ServerInstance->Users->QuitUser((*iter), message);
 255                         }
 256                 }
 257                 virtual ~ModuleNationalChars()
 258                 {
 259                         delete myhandler;
 260                         ServerInstance->IsNick= *rememberer;
 261                         free(rememberer);
 262                         national_case_insensitive_map=lowermap_rememberer;
 263                         CheckForceQuit("National characters module unloaded");
 264                 }
 265
 266                 virtual Version GetVersion()
 267                 {
 268                         return Version("$Id: m_nationalchars.cpp 0 2008-12-15 14:24:12SAMT phoenix $",VF_COMMON,API_VERSION);
 269                 }
 270
 271                 /*make an array to check against it 8bit characters a bit faster. Whether allowed or uppercase (for your needs).*/
 272
 273                 void makereverse(unsigned char * from, unsigned  char * to, unsigned int cnt)
 274                 {
 275                         memset(to, 0, cnt);
 276                         for(unsigned char * n=from; (*n) && ((*n)<cnt) && (n<from+cnt); n++)
 277                         {
 278                                 to[*n]=1;
 279                         }
 280                 }
 281
 282                 /*so Bynets Unreal distribution stuff*/
 283                 void loadtables(std::string filename, unsigned char ** tables, unsigned char cnt, char faillimit)
 284                 {
 285                         std::ifstream ifs(filename.c_str());
 286                         if (ifs.fail())
 287                         {
 288                                 ServerInstance->Logs->Log("m_nationalchars",DEFAULT,"loadtables() called for missing file: %s", filename.c_str());
 289                                 return;
 290                         }
 291
 292                         unsigned char n;
 293                         for (n=0;n<cnt;n++)
 294                         {
 295                                 memset(tables[n], 0, 256);
 296                         }
 297
 298                         memcpy(m_lower,rfc_case_insensitive_map,256);
 299
 300                         for (n=0;n<cnt;n++)
 301                         {
 302                                 if (loadtable(ifs, tables[n], 255) && (n<faillimit))
 303                                 {
 304                                         ServerInstance->Logs->Log("m_nationalchars",DEFAULT,"loadtables() called for illegal file: %s (line %d)", filename.c_str(), n+1);
 305                                         return;
 306                                 }
 307                         }
 308                         /*                      ServerInstance->Logs->Log("m_nationalchars",DEFAULT,"loadtables() : %s", ((char *)national_case_insensitive_map)+1);*/
 309
 310                         makereverse(m_additional, m_reverse_additional, sizeof(m_additional));
 311                         /*      Do you need faster access to additional 8bit uppercase table? No? Oh, sorry :( Let's comment this out */
 312                         /*      makereverse(m_additionalUp, m_reverse_additionalUp, sizeof(m_additional)); */
 313                 }
 314
 315                 unsigned char symtoi(const char *t,unsigned char base)
 316                 /* base = 16 for hexadecimal, 10 for decimal, 8 for octal ;) */
 317                 {
 318                         unsigned char tmp=0,current;
 319                         while ((*t)&&(*t!=' ')&&(*t!=13)&&(*t!=10)&&(*t!=','))
 320                         {
 321                                 tmp*=base;
 322                                 current=ascii_case_insensitive_map[(unsigned char)*t];
 323                                 if (current>='a')
 324                                         current=current-'a'+10;
 325                                 else
 326                                         current=current-'0';
 327                                 tmp+=current;
 328                                 t++;
 329                         }
 330                         return tmp;
 331                 }
 332
 333                 int loadtable(std::ifstream &ifs , unsigned char *chartable, unsigned int maxindex)
 334                 {
 335                         std::string buf;
 336                         getline(ifs, buf);
 337
 338                         unsigned int i=0;
 339                         int fail=0;
 340
 341                         buf.erase(buf.find_last_not_of("\n")+1);
 342
 343                         if (buf[0]=='.')         /* simple plain-text string after dot */
 344                         {
 345                                 i=buf.size()-1;
 346                                 if (i>(maxindex+1)) i=maxindex+1;
 347                                 memcpy(chartable,buf.c_str()+1,i);
 348                         }
 349                         else
 350                         {
 351                                 const char * p=buf.c_str();
 352                                 while (*p)
 353                                 {
 354                                         if (i>maxindex)
 355                                         {
 356                                                 fail=1;
 357                                                 break;
 358                                         }
 359
 360                                         if (*p!='\'')/* decimal or hexadecimal char code */
 361                                         {
 362                                                 if (*p=='0')
 363                                                 {
 364                                                         if (p[1]=='x')
 365                                                                  /* hex with the leading "0x" */
 366                                                                 chartable[i] = symtoi(p+2,16);
 367                                                         else
 368                                                                 chartable[i] = symtoi(p+1,8);
 369                                                 }
 370                                                                  /* hex form */
 371                                                 else if (*p=='x')
 372                                                 {
 373                                                         chartable[i] = symtoi(p+1,16);
 374                                                 }else    /* decimal form */
 375                                                 {
 376                                                         chartable[i] = symtoi(p,10);
 377                                                 }
 378                                         } else           /* plain-text char between '' */
 379                                         {
 380                                                 if (*(p+1)=='\\')
 381                                                 {
 382                                                         chartable[i] = *(p+2);
 383                                                         p+=3;
 384                                                 }else
 385                                                 {
 386                                                         chartable[i] = *(p+1);
 387                                                         p+=2;
 388                                                 }
 389                                         }
 390
 391                                         while (*p&& (*p!=',')&&(*p!=' ')&&(*p!=13)&&(*p!=10) ) p++;
 392                                         while (*p&&((*p==',')||(*p==' ')||(*p==13)||(*p==10))) p++;
 393
 394                                         i++;
 395
 396                                 }
 397                         }
 398
 399                         return fail;
 400                 }
 401
 402 };
 403
 404 MODULE_INIT(ModuleNationalChars)