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