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 class lwbNickHandler : public HandlerBase2<bool, const char*, size_t>
  25 {
  26         InspIRCd* Server;
  27  public:
  28         lwbNickHandler(InspIRCd* Srv) : Server(Srv) { }
  29         virtual ~lwbNickHandler() { }
  30         virtual bool Call(const char*, size_t);
  31 };
  32
  33                                                                  /*,m_reverse_additionalUp[256];*/
  34 static unsigned char m_reverse_additional[256],m_additionalMB[256],m_additionalUtf8[256],m_additionalUtf8range[256];
  35
  36 char utf8checkrest(unsigned char * mb, unsigned char cnt)
  37 {
  38         for (unsigned char * tmp=mb; tmp<mb+cnt; tmp++)
  39         {
  40                 if ((*tmp < 128) || (*tmp > 191))
  41                         return -1;
  42         }
  43         return cnt + 1;
  44 }
  45
  46
  47 char utf8size(unsigned char * mb)
  48 {
  49         if (!*mb)
  50                 return -1;
  51         if (!(*mb & 128))
  52                 return 1;
  53         if ((*mb & 224) == 192)
  54                 return utf8checkrest(mb + 1,1);
  55         if ((*mb & 240) == 224)
  56                 return utf8checkrest(mb + 1,2);
  57         if ((*mb & 248) == 240)
  58                 return utf8checkrest(mb + 1,3);
  59         return -1;
  60 }
  61
  62
  63 /* Conditions added */
  64 bool lwbNickHandler::Call(const char* n, size_t max)
  65 {
  66         if (!n || !*n)
  67                 return false;
  68
  69         unsigned int p = 0;
  70         for (const char* i = n; *i; i++, p++)
  71         {
  72                 /* 1. Multibyte encodings support:  */
  73                 /* 1.1. 16bit char. areas, e.g. chinese:*/
  74
  75                 /* if current character is the last, we DO NOT check it against multibyte table */
  76                 /* if there are mbtable ranges, use ONLY them. No 8bit at all */
  77                 if (i[1] && m_additionalMB[0])
  78                 {
  79                         /* otherwise let's take a look at the current character and the following one */
  80                         bool found = false;
  81                         for(unsigned char * mb = m_additionalMB; (*mb) && (mb < m_additionalMB + sizeof(m_additionalMB)); mb += 4)
  82                         {
  83                                 if ( (i[0] >= mb[0]) && (i[0] <= mb[1]) && (i[1] >= mb[2]) && (i[1] <= mb[3]) )
  84                                 {
  85                                         /* multibyte range character found */
  86                                         i++;
  87                                         p++;
  88                                         found = true;
  89                                         break;
  90                                 }
  91                         }
  92                         if (found)
  93                                 /* next char! */
  94                                 continue;
  95                         else
  96                                 /* there are ranges, but incorrect char (8bit?) given, sorry */
  97                                 return false;
  98                 }
  99
 100                 /* 2. 8bit character support */
 101                 if (((*i >= 'A') && (*i <= '}')) || m_reverse_additional[(unsigned char)*i])
 102                         /* "A"-"}" can occur anywhere in a nickname */
 103                         continue;
 104
 105                 if ((((*i >= '0') && (*i <= '9')) || (*i == '-')) && (i > n))
 106                         /* "0"-"9", "-" can occur anywhere BUT the first char of a nickname */
 107                         continue;
 108
 109                 /* 3.1. Check against a simple UTF-8 characters enumeration */
 110                 int cursize, ncursize = utf8size((unsigned char *)i);
 111                 /* do check only if current multibyte character is valid UTF-8 only */
 112                 if (ncursize != -1)
 113                 {
 114                         bool found = false;
 115                         for (unsigned char * mb = m_additionalUtf8; (utf8size(mb) != -1) && (mb < m_additionalUtf8 + sizeof(m_additionalUtf8)); mb += cursize)
 116                         {
 117                                 cursize = utf8size(mb);
 118                                 /* Size differs? Pick the next! */
 119                                 if (cursize != ncursize)
 120                                         continue;
 121
 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
 133                         /* 3.2. Check against an UTF-8 ranges: <start character> and <lenght of the range>.
 134                         Also char. is to be checked if it is a valid UTF-8 one */
 135                         found = false;
 136                         for (unsigned char * mb = m_additionalUtf8range; (utf8size(mb) != -1) && (mb < m_additionalUtf8range + sizeof(m_additionalUtf8range)); mb += cursize + 1)
 137                         {
 138                                 cursize = utf8size(mb);
 139                                 /* Size differs? Pick the next! */
 140                                 if ((cursize != ncursize) || (!mb[cursize]))
 141                                         continue;
 142
 143                                 unsigned char uright[5] = {0,0,0,0,0};
 144
 145                                 strncpy((char* ) uright, (char *) mb, cursize);
 146
 147                                 if ((uright[cursize-1] + mb[cursize]-1>0xff) && (cursize != 1))
 148                                 {
 149                                         uright[cursize - 2]+=1;
 150                                 }
 151                                 uright[cursize - 1] = (uright[cursize - 1]+mb[cursize] - 1) % 0x100;
 152
 153                                 if ((strncmp(i, (char *) mb, cursize) >= 0) && (strncmp(i, (char *) uright, cursize) <= 0))
 154                                 {
 155                                         i += cursize - 1;
 156                                         p += cursize - 1;
 157                                         found = true;
 158                                         break;
 159                                 }
 160                         }
 161                         if (found)
 162                                 continue;
 163                 }
 164
 165                 /* invalid character! abort */
 166                 return false;
 167         }
 168
 169         /* too long? or not -- pointer arithmetic rocks */
 170         return (p < max);
 171 }
 172
 173
 174 class ModuleNationalChars : public Module
 175 {
 176  private:
 177
 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
 186  public:
 187         ModuleNationalChars(InspIRCd* Me) : Module(Me)
 188         {
 189                 rememberer = (caller2<bool, const char*, size_t> *) malloc(sizeof(rememberer));
 190
 191                 lowermap_rememberer = national_case_insensitive_map;
 192                 memcpy(m_lower, rfc_case_insensitive_map, 256);
 193                 national_case_insensitive_map = m_lower;
 194
 195                 ServerInstance = Me;
 196
 197                 *rememberer = ServerInstance->IsNick;
 198                 myhandler = new lwbNickHandler(ServerInstance);
 199                 ServerInstance->IsNick = myhandler;
 200
 201                 Implementation eventlist[] = { I_OnRehash, I_On005Numeric };
 202                 ServerInstance->Modules->Attach(eventlist, this, 2);
 203                 OnRehash(NULL, "");
 204         }
 205
 206         virtual void On005Numeric(std::string &output)
 207         {
 208                 std::string tmp(casemapping);
 209                 tmp.insert(0, "CASEMAPPING=");
 210                 SearchAndReplace(output, std::string("CASEMAPPING=rfc1459"), tmp);
 211         }
 212
 213         virtual void OnRehash(User* user, const std::string &parameter)
 214         {
 215                 ConfigReader* conf = new ConfigReader(ServerInstance);
 216                 charset = conf->ReadValue("nationalchars", "file", 0);
 217                 casemapping = conf->ReadValue("nationalchars", "casemapping", charset, 0, false);
 218                 charset.insert(0, "../locales/");
 219                 unsigned char * tables[7] = { m_additional, m_additionalMB, m_additionalUp, m_lower, m_upper, m_additionalUtf8, m_additionalUtf8range };
 220                 loadtables(charset, tables, 7, 5);
 221                 forcequit = conf->ReadFlag("nationalchars", "forcequit", 0);
 222                 CheckForceQuit("National character set changed");
 223                 delete conf;
 224         }
 225
 226         void CheckForceQuit(const char * message)
 227         {
 228                 if (!forcequit)
 229                         return;
 230
 231                 for (std::vector<User*>::iterator iter = ServerInstance->Users->local_users.begin(); iter != ServerInstance->Users->local_users.end(); ++iter)
 232                 {
 233                         /* Fix by Brain: Dont quit UID users */
 234                         User* n = *iter;
 235                         if (!isdigit(n->nick[0]) && !ServerInstance->IsNick(n->nick.c_str(), ServerInstance->Config->Limits.NickMax))
 236                                 ServerInstance->Users->QuitUser(n, message);
 237                 }
 238         }
 239
 240         virtual ~ModuleNationalChars()
 241         {
 242                 delete myhandler;
 243                 ServerInstance->IsNick = *rememberer;
 244                 free(rememberer);
 245                 national_case_insensitive_map = lowermap_rememberer;
 246                 CheckForceQuit("National characters module unloaded");
 247         }
 248
 249         virtual Version GetVersion()
 250         {
 251                 return Version("$Id: m_nationalchars.cpp 0 2008-12-15 14:24:12SAMT phoenix $",VF_COMMON,API_VERSION);
 252         }
 253
 254         /*make an array to check against it 8bit characters a bit faster. Whether allowed or uppercase (for your needs).*/
 255         void makereverse(unsigned char * from, unsigned  char * to, unsigned int cnt)
 256         {
 257                 memset(to, 0, cnt);
 258                 for(unsigned char * n=from; (*n) && ((*n)<cnt) && (n<from+cnt); n++)
 259                         to[*n] = 1;
 260         }
 261
 262         /*so Bynets Unreal distribution stuff*/
 263         void loadtables(std::string filename, unsigned char ** tables, unsigned char cnt, char faillimit)
 264         {
 265                 std::ifstream ifs(filename.c_str());
 266                 if (ifs.fail())
 267                 {
 268                         ServerInstance->Logs->Log("m_nationalchars",DEFAULT,"loadtables() called for missing file: %s", filename.c_str());
 269                         return;
 270                 }
 271
 272                 for (unsigned char n=0; n< cnt; n++)
 273                 {
 274                         memset(tables[n], 0, 256);
 275                 }
 276
 277                 memcpy(m_lower, rfc_case_insensitive_map, 256);
 278
 279                 for (unsigned char n = 0; n < cnt; n++)
 280                 {
 281                         if (loadtable(ifs, tables[n], 255) && (n < faillimit))
 282                         {
 283                                 ServerInstance->Logs->Log("m_nationalchars",DEFAULT,"loadtables() called for illegal file: %s (line %d)", filename.c_str(), n+1);
 284                                 return;
 285                         }
 286                 }
 287
 288                 makereverse(m_additional, m_reverse_additional, sizeof(m_additional));
 289         }
 290
 291         unsigned char symtoi(const char *t,unsigned char base)
 292         /* base = 16 for hexadecimal, 10 for decimal, 8 for octal ;) */
 293         {
 294                 unsigned char tmp = 0, current;
 295                 while ((*t) && (*t !=' ') && (*t != 13) && (*t != 10) && (*t != ','))
 296                 {
 297                         tmp *= base;
 298                         current = ascii_case_insensitive_map[(unsigned char)*t];
 299                         if (current >= 'a')
 300                                 current = current - 'a' + 10;
 301                         else
 302                                 current = current - '0';
 303                         tmp+=current;
 304                         t++;
 305                 }
 306                 return tmp;
 307         }
 308
 309         int loadtable(std::ifstream &ifs , unsigned char *chartable, unsigned int maxindex)
 310         {
 311                 std::string buf;
 312                 getline(ifs, buf);
 313
 314                 unsigned int i = 0;
 315                 int fail = 0;
 316
 317                 buf.erase(buf.find_last_not_of("\n") + 1);
 318
 319                 if (buf[0] == '.')      /* simple plain-text string after dot */
 320                 {
 321                         i = buf.size() - 1;
 322
 323                         if (i > (maxindex + 1))
 324                                 i = maxindex + 1;
 325
 326                         memcpy(chartable, buf.c_str() + 1, i);
 327                 }
 328                 else
 329                 {
 330                         const char * p = buf.c_str();
 331                         while (*p)
 332                         {
 333                                 if (i > maxindex)
 334                                 {
 335                                         fail = 1;
 336                                         break;
 337                                 }
 338
 339                                 if (*p != '\'')         /* decimal or hexadecimal char code */
 340                                 {
 341                                         if (*p == '0')
 342                                         {
 343                                                 if (p[1] == 'x')
 344                                                          /* hex with the leading "0x" */
 345                                                         chartable[i] = symtoi(p + 2, 16);
 346                                                 else
 347                                                         chartable[i] = symtoi(p + 1, 8);
 348                                         }
 349                                         /* hex form */
 350                                         else if (*p == 'x')
 351                                         {
 352                                                 chartable[i] = symtoi(p + 1, 16);
 353                                         }else    /* decimal form */
 354                                         {
 355                                                 chartable[i] = symtoi(p, 10);
 356                                         }
 357                                 }
 358                                 else             /* plain-text char between '' */
 359                                 {
 360                                         if (*(p + 1) == '\\')
 361                                         {
 362                                                 chartable[i] = *(p + 2);
 363                                                 p += 3;
 364                                         }else
 365                                         {
 366                                                 chartable[i] = *(p + 1);
 367                                                 p += 2;
 368                                         }
 369                                 }
 370                                 while (*p && (*p != ',') && (*p != ' ') && (*p != 13) && (*p != 10))
 371                                         p++;
 372                                 while (*p && ((*p == ',') || (*p == ' ') || (*p == 13) || (*p == 10)))
 373                                         p++;
 374                                 i++;
 375                         }
 376                 }
 377                 return fail;
 378         }
 379 };
 380
 381 MODULE_INIT(ModuleNationalChars)