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://wiki.inspircd.org/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    Original patch is made by Dmitry "Killer{R}" Kononko. ( http://killprog.com/ )
  16    Changed at 2008-06-15 - 2009-02-11
  17    by Chernov-Phoenix Alexey (Phoenix@RusNet) mailto:phoenix /email address separator/ pravmail.ru */
  18
  19 #include "inspircd.h"
  20 #include "caller.h"
  21 #include <fstream>
  22
  23 /* $ModDesc: Provides an ability to have non-RFC1459 nicks & support for national CASEMAPPING */
  24
  25 class lwbNickHandler : public HandlerBase2<bool, const char*, size_t>
  26 {
  27         InspIRCd* Server;
  28  public:
  29         lwbNickHandler(InspIRCd* Srv) : Server(Srv) { }
  30         virtual ~lwbNickHandler() { }
  31         virtual bool Call(const char*, size_t);
  32 };
  33
  34                                                                  /*,m_reverse_additionalUp[256];*/
  35 static unsigned char m_reverse_additional[256],m_additionalMB[256],m_additionalUtf8[256],m_additionalUtf8range[256],m_additionalUtf8interval[256];
  36
  37 char utf8checkrest(unsigned char * mb, unsigned char cnt)
  38 {
  39         for (unsigned char * tmp=mb; tmp<mb+cnt; tmp++)
  40         {
  41                 /* & is faster! -- Phoenix (char & b11000000 == b10000000) */
  42                 if ((*tmp & 192) != 128)
  43                         return -1;
  44         }
  45         return cnt + 1;
  46 }
  47
  48
  49 char utf8size(unsigned char * mb)
  50 {
  51         if (!*mb)
  52                 return -1;
  53         if (!(*mb & 128))
  54                 return 1;
  55         if ((*mb & 224) == 192)
  56                 return utf8checkrest(mb + 1,1);
  57         if ((*mb & 240) == 224)
  58                 return utf8checkrest(mb + 1,2);
  59         if ((*mb & 248) == 240)
  60                 return utf8checkrest(mb + 1,3);
  61         return -1;
  62 }
  63
  64
  65 /* Conditions added */
  66 bool lwbNickHandler::Call(const char* n, size_t max)
  67 {
  68         if (!n || !*n)
  69                 return false;
  70
  71         unsigned int p = 0;
  72         for (const char* i = n; *i; i++, p++)
  73         {
  74                 /* 1. Multibyte encodings support:  */
  75                 /* 1.1. 16bit char. areas, e.g. chinese:*/
  76
  77                 /* if current character is the last, we DO NOT check it against multibyte table */
  78                 /* if there are mbtable ranges, use ONLY them. No 8bit at all */
  79                 if (i[1] && m_additionalMB[0])
  80                 {
  81                         /* otherwise let's take a look at the current character and the following one */
  82                         bool found = false;
  83                         for(unsigned char * mb = m_additionalMB; (*mb) && (mb < m_additionalMB + sizeof(m_additionalMB)); mb += 4)
  84                         {
  85                                 if ( (i[0] >= mb[0]) && (i[0] <= mb[1]) && (i[1] >= mb[2]) && (i[1] <= mb[3]) )
  86                                 {
  87                                         /* multibyte range character found */
  88                                         i++;
  89                                         p++;
  90                                         found = true;
  91                                         break;
  92                                 }
  93                         }
  94                         if (found)
  95                                 /* next char! */
  96                                 continue;
  97                         else
  98                                 /* there are ranges, but incorrect char (8bit?) given, sorry */
  99                                 return false;
 100                 }
 101
 102                 /* 2. 8bit character support */
 103                 if (((*i >= 'A') && (*i <= '}')) || m_reverse_additional[(unsigned char)*i])
 104                         /* "A"-"}" can occur anywhere in a nickname */
 105                         continue;
 106
 107                 if ((((*i >= '0') && (*i <= '9')) || (*i == '-')) && (i > n))
 108                         /* "0"-"9", "-" can occur anywhere BUT the first char of a nickname */
 109                         continue;
 110
 111                 /* 3.1. Check against a simple UTF-8 characters enumeration */
 112                 int cursize, cursize2, ncursize = utf8size((unsigned char *)i);
 113                 /* do check only if current multibyte character is valid UTF-8 only */
 114                 if (ncursize != -1)
 115                 {
 116                         bool found = false;
 117                         for (unsigned char * mb = m_additionalUtf8; (utf8size(mb) != -1) && (mb < m_additionalUtf8 + sizeof(m_additionalUtf8)); mb += cursize)
 118                         {
 119                                 cursize = utf8size(mb);
 120                                 /* Size differs? Pick the next! */
 121                                 if (cursize != ncursize)
 122                                         continue;
 123
 124                                 if (!strncmp(i, (char *)mb, cursize))
 125                                 {
 126                                         i += cursize - 1;
 127                                         p += cursize - 1;
 128                                         found = true;
 129                                         break;
 130                                 }
 131                         }
 132                         if (found)
 133                                 continue;
 134
 135                         /* 3.2. Check against an UTF-8 ranges: <start character> and <length of the range>. */
 136                         found = false;
 137                         for (unsigned char * mb = m_additionalUtf8range; (utf8size(mb) != -1) && (mb < m_additionalUtf8range + sizeof(m_additionalUtf8range)); mb += cursize + 1)
 138                         {
 139                                 cursize = utf8size(mb);
 140                                 /* Size differs (or lengthbyte is zero)? Pick the next! */
 141                                 if ((cursize != ncursize) || (!mb[cursize]))
 142                                         continue;
 143
 144                                 unsigned char uright[5] = {0,0,0,0,0}, range = mb[cursize] - 1;
 145                                 strncpy((char* ) uright, (char *) mb, cursize);
 146
 147                                 for (int temp = cursize - 1; (temp >= 0) && range; --temp)
 148                                 {
 149                                         /* all but the first char are 64-based */
 150                                         if (temp)
 151                                         {
 152                                                 char part64 = range & 63; /* i.e. % 64 */
 153                                                 /* handle carrying over */
 154                                                 if (uright[temp] + part64 - 1 > 191)
 155                                                 {
 156                                                         uright[temp] -= 64;
 157                                                         range += 64;
 158                                                 }
 159                                                 uright[temp] += part64;
 160                                                 range >>= 6; /* divide it on a 64 */
 161                                         }
 162                                         /* the first char of UTF-8 doesn't follow the rule */
 163                                         else
 164                                         {
 165                                                 uright[temp] += range;
 166                                         }
 167                                 }
 168
 169                                 if ((strncmp(i, (char *) mb, cursize) >= 0) && (strncmp(i, (char *) uright, cursize) <= 0))
 170                                 {
 171                                         i += cursize - 1;
 172                                         p += cursize - 1;
 173                                         found = true;
 174                                         break;
 175                                 }
 176                         }
 177                         if (found)
 178                                 continue;
 179
 180                         /* 3.3. Check against an UTF-8 intervals: <start character> and <end character>. */
 181                         found = false;
 182                         for (unsigned char * mb = m_additionalUtf8interval; (utf8size(mb) != -1) && (utf8size(mb+utf8size(mb)) != -1)
 183                                 && (mb < m_additionalUtf8interval + sizeof(m_additionalUtf8interval)); mb += (cursize+cursize2) )
 184                         {
 185                                 cursize = utf8size(mb);
 186                                 cursize2= utf8size(mb+cursize);
 187
 188                                 int minlen  = cursize  > ncursize ? ncursize : cursize;
 189                                 int minlen2 = cursize2 > ncursize ? ncursize : cursize2;
 190
 191                                 unsigned char* uright = mb + cursize;
 192
 193                                 if ((strncmp(i, (char *) mb, minlen) >= 0) && (strncmp(i, (char *) uright, minlen2) <= 0))
 194                                 {
 195                                         i += cursize - 1;
 196                                         p += cursize - 1;
 197                                         found = true;
 198                                         break;
 199                                 }
 200                         }
 201                         if (found)
 202                                 continue;
 203                 }
 204
 205                 /* invalid character! abort */
 206                 return false;
 207         }
 208
 209         /* too long? or not -- pointer arithmetic rocks */
 210         return (p < max);
 211 }
 212
 213
 214 class ModuleNationalChars : public Module
 215 {
 216  private:
 217
 218         InspIRCd* ServerInstance;
 219         lwbNickHandler* myhandler;
 220         std::string charset, casemapping;
 221         unsigned char m_additional[256], m_additionalUp[256], m_lower[256], m_upper[256];
 222         caller2<bool, const char*, size_t> * rememberer;
 223         bool forcequit;
 224         const unsigned char * lowermap_rememberer;
 225
 226  public:
 227         ModuleNationalChars(InspIRCd* Me) : Module(Me)
 228         {
 229                 rememberer = (caller2<bool, const char*, size_t> *) malloc(sizeof(caller2<bool, const char*, size_t>));
 230
 231                 lowermap_rememberer = national_case_insensitive_map;
 232                 memcpy(m_lower, rfc_case_insensitive_map, 256);
 233                 national_case_insensitive_map = m_lower;
 234
 235                 ServerInstance = Me;
 236
 237                 *rememberer = ServerInstance->IsNick;
 238                 myhandler = new lwbNickHandler(ServerInstance);
 239                 ServerInstance->IsNick = myhandler;
 240
 241                 Implementation eventlist[] = { I_OnRehash, I_On005Numeric };
 242                 ServerInstance->Modules->Attach(eventlist, this, 2);
 243                 OnRehash(NULL);
 244         }
 245
 246         virtual void On005Numeric(std::string &output)
 247         {
 248                 std::string tmp(casemapping);
 249                 tmp.insert(0, "CASEMAPPING=");
 250                 SearchAndReplace(output, std::string("CASEMAPPING=rfc1459"), tmp);
 251         }
 252
 253         virtual void OnRehash(User* user)
 254         {
 255                 ConfigReader* conf = new ConfigReader(ServerInstance);
 256                 charset = conf->ReadValue("nationalchars", "file", 0);
 257                 casemapping = conf->ReadValue("nationalchars", "casemapping", charset, 0, false);
 258                 charset.insert(0, "../locales/");
 259                 unsigned char * tables[8] = { m_additional, m_additionalMB, m_additionalUp, m_lower, m_upper, m_additionalUtf8, m_additionalUtf8range, m_additionalUtf8interval };
 260                 loadtables(charset, tables, 8, 5);
 261                 forcequit = conf->ReadFlag("nationalchars", "forcequit", 0);
 262                 CheckForceQuit("National character set changed");
 263                 delete conf;
 264         }
 265
 266         void CheckForceQuit(const char * message)
 267         {
 268                 if (!forcequit)
 269                         return;
 270
 271                 for (std::vector<User*>::iterator iter = ServerInstance->Users->local_users.begin(); iter != ServerInstance->Users->local_users.end(); ++iter)
 272                 {
 273                         /* Fix by Brain: Dont quit UID users */
 274                         User* n = *iter;
 275                         if (!isdigit(n->nick[0]) && !ServerInstance->IsNick(n->nick.c_str(), ServerInstance->Config->Limits.NickMax))
 276                                 ServerInstance->Users->QuitUser(n, message);
 277                 }
 278         }
 279
 280         virtual ~ModuleNationalChars()
 281         {
 282                 delete myhandler;
 283                 ServerInstance->IsNick = *rememberer;
 284                 free(rememberer);
 285                 national_case_insensitive_map = lowermap_rememberer;
 286                 CheckForceQuit("National characters module unloaded");
 287         }
 288
 289         virtual Version GetVersion()
 290         {
 291                 return Version("Provides an ability to have non-RFC1459 nicks & support for national CASEMAPPING",VF_COMMON,API_VERSION);
 292         }
 293
 294         /*make an array to check against it 8bit characters a bit faster. Whether allowed or uppercase (for your needs).*/
 295         void makereverse(unsigned char * from, unsigned  char * to, unsigned int cnt)
 296         {
 297                 memset(to, 0, cnt);
 298                 for(unsigned char * n=from; (*n) && ((*n)<cnt) && (n<from+cnt); n++)
 299                         to[*n] = 1;
 300         }
 301
 302         /*so Bynets Unreal distribution stuff*/
 303         void loadtables(std::string filename, unsigned char ** tables, unsigned char cnt, char faillimit)
 304         {
 305                 std::ifstream ifs(filename.c_str());
 306                 if (ifs.fail())
 307                 {
 308                         ServerInstance->Logs->Log("m_nationalchars",DEFAULT,"loadtables() called for missing file: %s", filename.c_str());
 309                         return;
 310                 }
 311
 312                 for (unsigned char n=0; n< cnt; n++)
 313                 {
 314                         memset(tables[n], 0, 256);
 315                 }
 316
 317                 memcpy(m_lower, rfc_case_insensitive_map, 256);
 318
 319                 for (unsigned char n = 0; n < cnt; n++)
 320                 {
 321                         if (loadtable(ifs, tables[n], 255) && (n < faillimit))
 322                         {
 323                                 ServerInstance->Logs->Log("m_nationalchars",DEFAULT,"loadtables() called for illegal file: %s (line %d)", filename.c_str(), n+1);
 324                                 return;
 325                         }
 326                 }
 327
 328                 makereverse(m_additional, m_reverse_additional, sizeof(m_additional));
 329         }
 330
 331         unsigned char symtoi(const char *t,unsigned char base)
 332         /* base = 16 for hexadecimal, 10 for decimal, 8 for octal ;) */
 333         {
 334                 unsigned char tmp = 0, current;
 335                 while ((*t) && (*t !=' ') && (*t != 13) && (*t != 10) && (*t != ','))
 336                 {
 337                         tmp *= base;
 338                         current = ascii_case_insensitive_map[(unsigned char)*t];
 339                         if (current >= 'a')
 340                                 current = current - 'a' + 10;
 341                         else
 342                                 current = current - '0';
 343                         tmp+=current;
 344                         t++;
 345                 }
 346                 return tmp;
 347         }
 348
 349         int loadtable(std::ifstream &ifs , unsigned char *chartable, unsigned int maxindex)
 350         {
 351                 std::string buf;
 352                 getline(ifs, buf);
 353
 354                 unsigned int i = 0;
 355                 int fail = 0;
 356
 357                 buf.erase(buf.find_last_not_of("\n") + 1);
 358
 359                 if (buf[0] == '.')      /* simple plain-text string after dot */
 360                 {
 361                         i = buf.size() - 1;
 362
 363                         if (i > (maxindex + 1))
 364                                 i = maxindex + 1;
 365
 366                         memcpy(chartable, buf.c_str() + 1, i);
 367                 }
 368                 else
 369                 {
 370                         const char * p = buf.c_str();
 371                         while (*p)
 372                         {
 373                                 if (i > maxindex)
 374                                 {
 375                                         fail = 1;
 376                                         break;
 377                                 }
 378
 379                                 if (*p != '\'')         /* decimal or hexadecimal char code */
 380                                 {
 381                                         if (*p == '0')
 382                                         {
 383                                                 if (p[1] == 'x')
 384                                                          /* hex with the leading "0x" */
 385                                                         chartable[i] = symtoi(p + 2, 16);
 386                                                 else
 387                                                         chartable[i] = symtoi(p + 1, 8);
 388                                         }
 389                                         /* hex form */
 390                                         else if (*p == 'x')
 391                                         {
 392                                                 chartable[i] = symtoi(p + 1, 16);
 393                                         }else    /* decimal form */
 394                                         {
 395                                                 chartable[i] = symtoi(p, 10);
 396                                         }
 397                                 }
 398                                 else             /* plain-text char between '' */
 399                                 {
 400                                         if (*(p + 1) == '\\')
 401                                         {
 402                                                 chartable[i] = *(p + 2);
 403                                                 p += 3;
 404                                         }else
 405                                         {
 406                                                 chartable[i] = *(p + 1);
 407                                                 p += 2;
 408                                         }
 409                                 }
 410                                 while (*p && (*p != ',') && (*p != ' ') && (*p != 13) && (*p != 10))
 411                                         p++;
 412                                 while (*p && ((*p == ',') || (*p == ' ') || (*p == 13) || (*p == 10)))
 413                                         p++;
 414                                 i++;
 415                         }
 416                 }
 417                 return fail;
 418         }
 419 };
 420
 421 MODULE_INIT(ModuleNationalChars)