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