Пример #1
0
__C_LINK int __DPSCALL DpsBaseClose(DPS_BASE_PARAM *P) {
  TRACE_IN(P->A, "DpsBaseClose");
/*  if (P->opened && (P->mode == DPS_WRITE_LOCK) ) {
      fsync(P->Sfd);
      fsync(P->Ifd);
  }*/
  if (!P->A->Flags.cold_var && P->locked) {
      DpsUnLock(P->Sfd);
      DpsUnLock(P->Ifd); 
#if 1
      DPS_RELEASELOCK(P->A, DPS_LOCK_BASE_N(P->FileNo));
#endif
      P->locked = 0;
  }
  if (P->opened){
    DpsClose(P->Sfd); 
    DpsClose(P->Ifd); 
    P->opened = 0;
  }
  DPS_FREE(P->Ifilename);
  DPS_FREE(P->Sfilename);
  TRACE_OUT(P->A);
  return DPS_OK;
}
Пример #2
0
static int MakeNestedIndex(DPS_AGENT *Indexer, DPS_UINT8URLIDLIST *L, const char *lim_name, DPS_DB *db) {
     DPS_ENV   *Conf = Indexer->Conf;
     size_t    k, prev;
     urlid_t   *data=NULL;
     DPS_UINT8_POS_LEN *ind=NULL;
     size_t    mind=1000, nind=0, ndata;
     char fname[PATH_MAX];
     int  dat_fd=0, ind_fd=0;
     int  rc=DPS_OK;
     const char	*vardir = (db->vardir) ? db->vardir : DpsVarListFindStr(&Conf->Vars, "VarDir", DPS_VAR_DIR);
     
     if(!L->Item)return(1);
     
     if (L->nitems > 1) DpsSort(L->Item, L->nitems, sizeof(DPS_UINT8URLID), (qsort_cmp)cmp_ind8);
     
     data = (urlid_t*)DpsMalloc((L->nitems + 1) * sizeof(urlid_t));
     if(!data){
       DpsLog(Indexer, DPS_LOG_ERROR, "Can't alloc %d bytes [%s:%d]", (L->nitems + 1) * sizeof(urlid_t), __FILE__, __LINE__);
       goto err1;
     }
     ind=(DPS_UINT8_POS_LEN*)DpsMalloc(mind*sizeof(DPS_UINT8_POS_LEN));
     if(!ind){
       DpsLog(Indexer, DPS_LOG_ERROR, "Can't alloc %d bytes [%s:%d]", mind * sizeof(DPS_UINT8_POS_LEN), __FILE__, __LINE__);
       goto err1;
     }
     prev=0;
     for(k=0; k < L->nitems; k++) {
          data[k] = L->Item[k].url_id;
          if((k == L->nitems-1) || (L->Item[k].hi != L->Item[prev].hi) || (L->Item[k].lo != L->Item[prev].lo)) {
               if(nind==mind){
                    mind+=1000;
                    ind=(DPS_UINT8_POS_LEN*)DpsRealloc(ind,mind*sizeof(DPS_UINT8_POS_LEN));
                    if(!ind) {
		      DpsLog(Indexer, DPS_LOG_ERROR, "Can't alloc %d bytes [%s:%d]", mind * sizeof(DPS_UINT8_POS_LEN), __FILE__, __LINE__);
		      goto err1;
                    }
               }
               /* Fill index */
               ind[nind].hi = L->Item[prev].hi;
               ind[nind].lo = L->Item[prev].lo;
               ind[nind].pos = prev * sizeof(*data);
               if (k == L->nitems - 1) ind[nind].len = (k - prev + 1) * sizeof(*data);
               else ind[nind].len = (k - prev) * sizeof(*data);
               DpsLog(Indexer, DPS_LOG_DEBUG, "%08X%08X - %d %d\n", ind[nind].hi, ind[nind].lo, (int)ind[nind].pos, ind[nind].len);
               nind++;
               
               prev=k;
          }
     }
     ndata = L->nitems;
     ClearIndex8(L);
     
     dps_snprintf(fname,sizeof(fname)-1,"%s%c%s%c%s.dat", vardir,DPSSLASH, DPS_TREEDIR,DPSSLASH, lim_name);
     if((dat_fd = DpsOpen3(fname, O_CREAT | O_WRONLY | O_TRUNC | DPS_BINARY, DPS_IWRITE)) < 0) {
       DpsLog(Indexer, DPS_LOG_ERROR, "Can't open '%s': %s [%s:%d]", fname, strerror(errno), __FILE__, __LINE__);
       goto err1;
     }
     DpsWriteLock(dat_fd);
     if((ndata * sizeof(*data)) != (size_t)write(dat_fd, data, ndata * sizeof(*data))) {
       DpsLog(Indexer, DPS_LOG_ERROR, "Can't write '%s': %s [%s:%d]", fname, strerror(errno), __FILE__, __LINE__);
       goto err1;
     }
     DpsUnLock(dat_fd);
     DpsClose(dat_fd);
     DPS_FREE(data);

     dps_snprintf(fname,sizeof(fname)-1,"%s%c%s%c%s.ind", vardir, DPSSLASH,DPS_TREEDIR, DPSSLASH, lim_name);
     if((ind_fd = DpsOpen3(fname, O_CREAT | O_WRONLY | O_TRUNC | DPS_BINARY, DPS_IWRITE)) < 0) {
       DpsLog(Indexer, DPS_LOG_ERROR, "Can't open '%s': %s [%s:%d]", fname, strerror(errno), __FILE__, __LINE__);
       goto err1;
     }
     DpsWriteLock(ind_fd);
     if((nind*sizeof(DPS_UINT8_POS_LEN)) != (size_t)write(ind_fd,ind,nind*sizeof(DPS_UINT8_POS_LEN))){
       DpsLog(Indexer, DPS_LOG_ERROR, "Can't write '%s': %s [%s:%d]", fname, strerror(errno), __FILE__, __LINE__);
          goto err1;
     }
     DpsUnLock(ind_fd);
     DpsClose(ind_fd);
     DPS_FREE(ind);
     
     return(0);
     
err1:
     ClearIndex8(L);
     DPS_FREE(data);
     DPS_FREE(ind);
     if(dat_fd) DpsClose(dat_fd);
     if(ind_fd) DpsClose(ind_fd);
     return(1);
}
Пример #3
0
static int MakeLinearIndex(DPS_AGENT *Indexer, const char *field, const char *lim_name, int type, DPS_DB *db) {
    DPS_ENV *Conf = Indexer->Conf;
    DPS_UINT4URLIDLIST  L;
    size_t    k,prev;
    urlid_t   *data = NULL;
    DPS_UINT4_POS_LEN *ind=NULL;
    size_t    mind=1000,nind=0;
    char fname[PATH_MAX];
    int  dat_fd=0, ind_fd=0, rc;
    const char	*vardir = (db->vardir) ? db->vardir : DpsVarListFindStr(&Conf->Vars, "VarDir", DPS_VAR_DIR);

    bzero(&L, sizeof(DPS_UINT4URLIDLIST));

    rc = DpsLimit4(Indexer, &L, field, type, db);

    if(rc != DPS_OK) {
        DpsLog(Indexer, DPS_LOG_ERROR, "Error: %s [%s:%d]", DpsEnvErrMsg(Conf), __FILE__, __LINE__);
        goto err1;
    }

    if(!L.Item)return(1);

    if (L.nitems > 1) DpsSort(L.Item, L.nitems, sizeof(DPS_UINT4URLID), (qsort_cmp)cmp_ind4);

    data = (urlid_t*)DpsMalloc((L.nitems + 1) * sizeof(*data));
    if(!data) {
        fprintf(stderr,"Error1: %s\n",strerror(errno));
        goto err1;
    }
    ind=(DPS_UINT4_POS_LEN*)DpsMalloc(mind*sizeof(DPS_UINT4_POS_LEN));
    if(!ind) {
        fprintf(stderr,"Error2: %s\n",strerror(errno));
        goto err1;
    }
    prev=0;
    for(k=0; k<L.nitems; k++) {
        data[k]=L.Item[k].url_id;
        if((k==L.nitems-1) || (L.Item[k].val!=L.Item[prev].val)) {
            if(nind==mind) {
                mind+=1000;
                ind=(DPS_UINT4_POS_LEN*)DpsRealloc(ind,mind*sizeof(DPS_UINT4_POS_LEN));
                if(!ind) {
                    fprintf(stderr,"Error3: %s\n",strerror(errno));
                    goto err1;
                }
            }
            /* Fill index */
            ind[nind].val=L.Item[prev].val;
            ind[nind].pos = prev * sizeof(*data);
            if (k == L.nitems - 1) ind[nind].len = (k - prev + 1) * sizeof(*data);
            else ind[nind].len = (k - prev) * sizeof(*data);
            DpsLog(Indexer, DPS_LOG_DEBUG, "%d - pos:%x len:%d\n", ind[nind].val, (int)ind[nind].pos, ind[nind].len);
            nind++;

            prev=k;
        }
    }
    if (L.mapped) {
#ifdef HAVE_SYS_MMAN_H
        if (munmap(L.Item, (L.nitems + 1) * sizeof(DPS_UINT4URLID))) {
            fprintf(stderr, "Can't shmdt '%s': %s\n", L.shm_name, strerror(errno));
        }
#elif defined(HAVE_SYS_SHM_H)
        if (shmdt(L.Item)) {
            fprintf(stderr, "Can't shmdt '%s': %s\n", L.shm_name, strerror(errno));
        }
#endif
        unlink(L.shm_name);
    } else {
        DPS_FREE(L.Item);
    }

    dps_snprintf(fname,sizeof(fname),"%s%c%s%c%s.dat", vardir,DPSSLASH, DPS_TREEDIR, DPSSLASH, lim_name);
    if((dat_fd = DpsOpen3(fname, O_CREAT | O_WRONLY | O_TRUNC | DPS_BINARY, DPS_IWRITE)) < 0) {
        fprintf(stderr,"Can't open '%s': %s\n",fname,strerror(errno));
        goto err1;
    }
    DpsWriteLock(dat_fd);
    if((L.nitems * sizeof(*data)) != (size_t)write(dat_fd, data, L.nitems * sizeof(*data))) {
        fprintf(stderr,"Can't write '%s': %s\n",fname,strerror(errno));
        goto err1;
    }
    DpsUnLock(dat_fd);
    DpsClose(dat_fd);
    DPS_FREE(data);

    dps_snprintf(fname,sizeof(fname),"%s%c%s%c%s.ind", vardir,DPSSLASH, DPS_TREEDIR, DPSSLASH, lim_name);
    if((ind_fd = DpsOpen3(fname, O_CREAT | O_WRONLY | O_TRUNC | DPS_BINARY, DPS_IWRITE)) < 0) {
        fprintf(stderr,"Can't open '%s': %s\n",fname,strerror(errno));
        goto err1;
    }
    DpsWriteLock(ind_fd);
    if((nind*sizeof(DPS_UINT4_POS_LEN)) != (size_t)write(ind_fd,ind,nind*sizeof(DPS_UINT4_POS_LEN))) {
        fprintf(stderr,"Can't write '%s': %s\n",fname,strerror(errno));
        goto err1;
    }
    DpsUnLock(ind_fd);
    DpsClose(ind_fd);
    DPS_FREE(ind);

    return(0);

err1:
    if (L.mapped) {
#ifdef HAVE_SYS_MMAN_H
        if (munmap(L.Item, (L.nitems + 1) * sizeof(DPS_UINT4URLID))) {
            fprintf(stderr, "Can't shmdt '%s': %s\n", L.shm_name, strerror(errno));
        }
#elif defined(HAVE_SYS_SHM_H)
        if (shmdt(L.Item)) {
            fprintf(stderr, "Can't shmdt '%s': %s\n", L.shm_name, strerror(errno));
        }
#endif
        unlink(L.shm_name);
    } else {
        DPS_FREE(L.Item);
    }
    DPS_FREE(data);
    DPS_FREE(ind);
    if(dat_fd) DpsClose(dat_fd);
    if(ind_fd) DpsClose(ind_fd);
    return(1);
}
Пример #4
0
int main(int argc,char **argv, char **envp) {
  int ch, sleeps = 1, optimize = 0, obi = 0;
  unsigned int from = 0, to = 0xFFF, p_to = 0;
	DPS_ENV * Env;
	const char * config_name = DPS_CONF_DIR "/cached.conf";

	DpsInit(argc, argv, envp); /* Initialize library */
	
	DpsInitMutexes();
	Env=DpsEnvInit(NULL);
	if (Env == NULL) exit(1);
	DpsSetLockProc(Env, DpsLockProc);

/*#ifndef HAVE_SETPROCTITLE*/
	ARGV = argv;
	ARGC = argc;
/*#endif*/
	while ((ch = getopt(argc, argv, "blt:f:op:w:v:h?")) != -1){
		switch (ch) {
			case 'f':
				sscanf(optarg, "%x", &from);
				break;	
			case 't': 
				sscanf(optarg, "%x", &p_to);
				break;
			case 'w':
			        DpsVarListReplaceStr(&Env->Vars, "VarDir", optarg);
				break;
                        case 'v': DpsSetLogLevel(NULL, atoi(optarg)); break;
                        case 'b': obi++; break;
                        case 'o': optimize++; break;
                        case 'p': sleeps = atoi(optarg); break;
			case 'h':
			case '?':
			default:
			  usage();
			  DpsEnvFree(Env);
			  DpsDeInit();
			  DpsDestroyMutexes();
				return 1;
				break;
		}
	}
	argc -= optind;
	argv += optind;

	if(argc > 1) {
		usage();
		DpsEnvFree(Env);
		DpsDeInit();
		DpsDestroyMutexes();
		return 1;
	} else if (argc == 1) {
	        config_name = argv[0];
	}
	{
		DPS_LOGDEL *del_buf=NULL;
		size_t del_count = 0, log, bytes, n = 0;
		int dd, log_fd;
		struct stat sb;
		char dname[PATH_MAX] = "";
		DPS_BASE_PARAM P;
		DPS_LOGWORD *log_buf = NULL;
		DPS_AGENT *Indexer = DpsAgentInit(NULL, Env, 0);

		log2stderr = 1;
		if (Indexer == NULL) {
		  fprintf(stderr, "Can't alloc Agent at %s:%d\n", __FILE__, __LINE__);
		  exit(DPS_ERROR);
		}
		
		if(DPS_OK != DpsEnvLoad(Indexer, config_name, (dps_uint8)0)){
		  fprintf(stderr, "%s\n", DpsEnvErrMsg(Env));
		  DpsEnvFree(Env);
		  DpsDeInit();
		  DpsDestroyMutexes();
		  return DPS_ERROR;
		}
		DpsOpenLog("splitter", Env, log2stderr);
		Indexer->flags = Env->flags = DPS_FLAG_UNOCON;
		DpsVarListAddLst(&Indexer->Vars, &Env->Vars, NULL, "*");

		bzero(&P, sizeof(P));
		P.subdir = DPS_TREEDIR;
		P.basename = "wrd";
		P.indname = "wrd";
		P.mode = DPS_WRITE_LOCK;
		P.NFiles = DpsVarListFindInt(&Indexer->Conf->Vars, "WrdFiles", 0x300);
		P.vardir = DpsStrdup(DpsVarListFindStr(&Indexer->Conf->Vars, "VarDir", DPS_VAR_DIR));
		P.A = Indexer;
		if (p_to != 0) to = p_to;
		else to = P.NFiles - 1;
#ifdef HAVE_ZLIB
		P.zlib_method = Z_DEFLATED;
		P.zlib_level = 9;
		P.zlib_windowBits = DPS_BASE_WRD_WINDOWBITS;
		P.zlib_memLevel = 9;
		P.zlib_strategy = DPS_BASE_WRD_STRATEGY;
#endif

		/* Open del log file */
		dps_snprintf(dname,sizeof(dname),"%s%c%s%cdel-split.log", P.vardir, DPSSLASH, DPS_SPLDIR, DPSSLASH);
		if((dd = DpsOpen2(dname, O_RDONLY | DPS_BINARY)) < 0) {
		  dps_strerror(NULL, 0, "Can't open del log '%s'", dname);
		  exit(DPS_ERROR);
		}

		DpsLog(Indexer, DPS_LOG_DEBUG, "VarDir: %s, WrdFiles: %d [%x]", P.vardir, P.NFiles, P.NFiles);

		/* Allocate del buffer */
		fstat(dd, &sb);
		if (sb.st_size != 0) {
		  del_buf=(DPS_LOGDEL*)DpsMalloc((size_t)sb.st_size + 1);
		  if (del_buf == NULL) {
		    fprintf(stderr, "Can't alloc %d bytes at %s:%d\n", (int)sb.st_size, __FILE__, __LINE__);
		    exit(0);
		  }
		  del_count=read(dd,del_buf,(size_t)sb.st_size)/sizeof(DPS_LOGDEL);
		}
		DpsClose(dd);

		/* Remove duplicates URLs in DEL log     */
		/* Keep only oldest records for each URL */
		if (del_count > 0) {
		  DpsLog(Indexer, DPS_LOG_DEBUG, "Sorting del_buf: %d items", del_count);
		  if (del_count > 1) DpsSort(del_buf, (size_t)del_count, sizeof(DPS_LOGDEL), DpsCmpurldellog);
		    DpsLog(Indexer, DPS_LOG_DEBUG, "Removing DelLogDups");
		  del_count = DpsRemoveDelLogDups(del_buf, del_count);
		}

		DpsLog(Indexer, DPS_LOG_DEBUG, "Processing Bufs from %d [%x] to %d [%x]", from, from, to, to);

		for(log = from; log <= to; log++) {

		  /* Open log file */
		  dps_snprintf(dname, sizeof(dname), "%s%c%s%c%03X-split.log", P.vardir, DPSSLASH, DPS_SPLDIR, DPSSLASH, log);
		  if((log_fd = DpsOpen2(dname, O_RDWR|DPS_BINARY)) < 0){
		    if (errno == ENOENT) {
		      dps_strerror(Indexer, DPS_LOG_DEBUG, "Can't open '%s'", dname);
		      n = 0;
/*		      continue;*/
		    } else {
		      dps_strerror(Indexer, DPS_LOG_ERROR, "Can't open '%s'", dname);
		      continue;
		    }
		  } else {
		    DpsWriteLock(log_fd); 
		    DpsLog(Indexer, DPS_LOG_DEBUG, "Processing Log: %x", log);
		    fstat(log_fd, &sb);
		    log_buf = (sb.st_size > 0) ? (DPS_LOGWORD*)DpsMalloc((size_t)sb.st_size + 1) : NULL;
		    if (log_buf != NULL) {
		      unlink(dname);
		      bytes = read(log_fd,log_buf,(size_t)sb.st_size);
		      (void)ftruncate(log_fd, (off_t)0);
		      DpsUnLock(log_fd);
		      DpsClose(log_fd);
		      
		      n = bytes / sizeof(DPS_LOGWORD);
		      DpsLog(Indexer, DPS_LOG_DEBUG, "Sorting log_buf: %d items", n);
		      if (n > 1) DpsSort(log_buf, n, sizeof(DPS_LOGWORD), (qsort_cmp)DpsCmplog);
		      DpsLog(Indexer, DPS_LOG_DEBUG, "Removing OldWords");
		      n = DpsRemoveOldWords(log_buf, n, del_buf, del_count);
		      if (n > 1) DpsSort(log_buf, n, sizeof(DPS_LOGWORD), (qsort_cmp)DpsCmplog_wrd);
		      
		    } else {
		      n = 0;
		      DpsUnLock(log_fd);
		      DpsClose(log_fd);
		    }
		  }

		  DpsLog(Indexer, DPS_LOG_DEBUG, "Processing Buf, optimize: %d", optimize);
		  if (obi) DpsBaseOptimize(&P, log);
		  DpsProcessBuf(Indexer, &P, log, log_buf, n, del_buf, del_count);
		  if (optimize) DpsBaseOptimize(&P, log);
		  DpsBaseClose(&P);
		  DPS_FREE(log_buf);

		  DpsLog(Indexer, DPS_LOG_DEBUG, "pas done: %d from %d to %d", log, from, to);
		  DPSSLEEP(sleeps);
		}
		DPS_FREE(del_buf);
		DpsAgentFree(Indexer);
		DPS_FREE(P.vardir);
	}

	fprintf(stderr, "Splitting done.\n");
	
	DpsEnvFree(Env);
	DpsDeInit();
	DpsDestroyMutexes();

#ifdef EFENCE
	fprintf(stderr, "Memory leaks checking\n");
	DpsEfenceCheckLeaks();
#endif
#ifdef FILENCE
	fprintf(stderr, "FD leaks checking\n");
	DpsFilenceCheckLeaks(NULL);
#endif
	return 0;
}
Пример #5
0
__C_LINK int __DPSCALL DpsImportDictionary(DPS_ENV * Conf, const char *lang, const char *charset,
				   const char *filename, int skip_noflag, const char *first_letters){
        struct stat     sb;
	char *str, *data = NULL, *cur_n = NULL;	
	char *lstr;	
	dpsunicode_t *ustr;	
	DPS_CHARSET *sys_int;
	DPS_CHARSET *dict_charset;
	DPS_CONV touni;
	DPS_CONV fromuni;
	int             fd;
	char            savebyte;

	if ((lstr = (char*) DpsMalloc(2048)) == NULL) {
	  DPS_FREE(str);
	  return DPS_ERROR; 
	}
	if ((ustr = (dpsunicode_t*) DpsMalloc(8192)) == NULL) {
	  DPS_FREE(lstr);
	  return DPS_ERROR; 
	}

	dict_charset = DpsGetCharSet(charset);
	sys_int = DpsGetCharSet("sys-int");
	if ((dict_charset == NULL) || (sys_int == NULL)) {
	  DPS_FREE(lstr);
	  DPS_FREE(ustr);
	  return DPS_ERROR;
	}
	
	DpsConvInit(&touni, dict_charset, sys_int, Conf->CharsToEscape, 0);
	DpsConvInit(&fromuni, sys_int, dict_charset, Conf->CharsToEscape, 0);
	
	if (stat(filename, &sb)) {
	  fprintf(stderr, "Unable to stat synonyms file '%s': %s", filename, strerror(errno));
	  DPS_FREE(lstr);
	  DPS_FREE(ustr);
	  return DPS_ERROR;
	}
	if ((fd = DpsOpen2(filename, O_RDONLY)) <= 0) {
	  fprintf(stderr, "Unable to open synonyms file '%s': %s", filename, strerror(errno));
	  return DPS_ERROR;
	}
	if ((data = (char*)DpsMalloc(sb.st_size + 1)) == NULL) {
	  fprintf(stderr, "Unable to alloc %ld bytes", (long)sb.st_size);
	  DpsClose(fd);
	  DPS_FREE(lstr);
	  DPS_FREE(ustr);
	  return DPS_ERROR;
	}
	if (read(fd, data, sb.st_size) != (ssize_t)sb.st_size) {
	  fprintf(stderr, "Unable to read synonym file '%s': %s", filename, strerror(errno));
	  DPS_FREE(data);
	  DpsClose(fd);
	  DPS_FREE(lstr);
	  DPS_FREE(ustr);
	  return DPS_ERROR;
	}
	data[sb.st_size] = '\0';
	str = data;
	cur_n = strchr(str, '\n');
	if (cur_n != NULL) {
	  cur_n++;
	  savebyte = *cur_n;
	  *cur_n = '\0';
	}

	DpsClose(fd);
	while(str != NULL) {
		char *s;
		const char *flag;
		int res;
		
	        flag = NULL;
		s = str;
		while(*s){
			if(*s == '\r') *s = '\0';
			if(*s == '\n') *s = '\0';
			s++;
		}
		if((s=strchr(str,'/'))){
			*s=0;
			s++;flag=s;
			while(*s){
				if(((*s>='A')&&(*s<='Z'))||((*s>='a')&&(*s<='z')))s++;
				else{
					*s=0;
					break;
				}
			}
		}else{
			if(skip_noflag)	goto loop_continue;
			flag="";
		}

		res = DpsConv(&touni, (char*)ustr, 8192, str, 1024);
		DpsUniStrToLower(ustr);

		/* Dont load words if first letter is not required */
		/* It allows to optimize loading at  search time   */
		if(*first_letters) {
			DpsConv(&fromuni, lstr, 2048, ((const char*)ustr),(size_t)res);
			if(!strchr(first_letters,lstr[0]))
				goto loop_continue;
		}
		res = DpsSpellAdd(&Conf->Spells,ustr,flag,lang);
		if (res != DPS_OK) {
		  DPS_FREE(lstr);
		  DPS_FREE(ustr); DPS_FREE(data);
		  return res;
		}
		if (Conf->Flags.use_accentext) {
		  dpsunicode_t *af_uwrd = DpsUniAccentStrip(ustr);
		  if (DpsUniStrCmp(af_uwrd, ustr) != 0) {
		    res = DpsSpellAdd(&Conf->Spells, af_uwrd, flag, lang);
		    if (res != DPS_OK) {
		      DPS_FREE(lstr);
		      DPS_FREE(ustr); DPS_FREE(data); DPS_FREE(af_uwrd);
		      return res;
		    }
		  }
		  DPS_FREE(af_uwrd);
		  if (strncasecmp(lang, "de", 2) == 0) {
		    dpsunicode_t *de_uwrd = DpsUniGermanReplace(ustr);
		    if (DpsUniStrCmp(de_uwrd, ustr) != 0) {
		      res = DpsSpellAdd(&Conf->Spells, de_uwrd, flag, lang);
		      if (res != DPS_OK) {
			DPS_FREE(lstr);
			DPS_FREE(ustr); DPS_FREE(data); DPS_FREE(de_uwrd);
			return res;
		      }
		    }
		    DPS_FREE(de_uwrd);
		  }
		}
	loop_continue:
		str = cur_n;
		if (str != NULL) {
		  *str = savebyte;
		  cur_n = strchr(str, '\n');
		  if (cur_n != NULL) {
		    cur_n++;
		    savebyte = *cur_n;
		    *cur_n = '\0';
		  }
		}
	}
	DPS_FREE(data);
	DPS_FREE(lstr);
	DPS_FREE(ustr);
	return DPS_OK;
}
Пример #6
0
__C_LINK int __DPSCALL DpsSynonymListLoad(DPS_ENV * Env,const char * filename){
     struct stat     sb;
     char      *str, *data = NULL, *cur_n = NULL;
     char      lang[64]="";
     DPS_CHARSET    *cs=NULL;
     DPS_CHARSET    *sys_int=DpsGetCharSet("sys-int");
     DPS_CONV  file_uni;
     DPS_WIDEWORD    *ww = NULL;
     size_t key = 1;
     int flag_th = 0;
     int             fd;
     char            savebyte;
     
     if (stat(filename, &sb)) {
       fprintf(stderr, "Unable to stat synonyms file '%s': %s", filename, strerror(errno));
       return DPS_ERROR;
     }
     if ((fd = DpsOpen2(filename, O_RDONLY)) <= 0) {
       dps_snprintf(Env->errstr,sizeof(Env->errstr)-1, "Unable to open synonyms file '%s': %s", filename, strerror(errno));
       return DPS_ERROR;
     }
     if ((data = (char*)DpsMalloc(sb.st_size + 1)) == NULL) {
       dps_snprintf(Env->errstr,sizeof(Env->errstr)-1, "Unable to alloc %d bytes", sb.st_size);
       DpsClose(fd);
       return DPS_ERROR;
     }
     if (read(fd, data, sb.st_size) != (ssize_t)sb.st_size) {
       dps_snprintf(Env->errstr,sizeof(Env->errstr)-1, "Unable to read synonym file '%s': %s", filename, strerror(errno));
       DPS_FREE(data);
       DpsClose(fd);
       return DPS_ERROR;
     }
     data[sb.st_size] = '\0';
     str = data;
     cur_n = strchr(str, '\n');
     if (cur_n != NULL) {
       cur_n++;
       savebyte = *cur_n;
       *cur_n = '\0';
     }

     while(str != NULL) {
          if(str[0]=='#'||str[0]==' '||str[0]=='\t'||str[0]=='\r'||str[0]=='\n') goto loop_continue;
          
          if(!strncasecmp(str,"Charset:",8)){
               char * lasttok;
               char * charset;
               if((charset = dps_strtok_r(str + 8, " \t\n\r", &lasttok))) {
                    cs=DpsGetCharSet(charset);
                    if(!cs){
                         dps_snprintf(Env->errstr, sizeof(Env->errstr), "Unknown charset '%s' in synonyms file '%s'",
                                   charset, filename);
                         DPS_FREE(data);
			 DpsClose(fd);
                         return DPS_ERROR;
                    }
                    DpsConvInit(&file_uni, cs, sys_int, Env->CharsToEscape, 0);
               }
          }else
          if(!strncasecmp(str,"Language:",9)){
               char * lasttok;
               char * l;
               if((l = dps_strtok_r(str + 9, " \t\n\r", &lasttok))) {
                    dps_strncpy(lang, l, sizeof(lang)-1);
               }
          }else
          if(!strncasecmp(str, "Thesaurus:", 10)) {
               char * lasttok;
	       char *tok = dps_strtok_r(str + 10, " \t\n\r", &lasttok);
	       flag_th = (strncasecmp(tok, "yes", 3) == 0) ? 1 : 0;
          }else{
               char      *av[255];
               size_t         ac, i, j;
	       dpsunicode_t *t;

               if(!cs){
                    dps_snprintf(Env->errstr,sizeof(Env->errstr)-1,"No Charset command in synonyms file '%s'",filename);
                    DpsClose(fd); DPS_FREE(data);
                    return DPS_ERROR;
               }
               if(!lang[0]){
                    dps_snprintf(Env->errstr,sizeof(Env->errstr)-1,"No Language command in synonyms file '%s'",filename);
                    DpsClose(fd); DPS_FREE(data);
                    return DPS_ERROR;
               }

               ac = DpsGetArgs(str, av, 255);
               if (ac < 2) goto loop_continue;

               if ((ww = (DPS_WIDEWORD*)DpsRealloc(ww, ac * sizeof(DPS_WIDEWORD))) == NULL) return DPS_ERROR;

               for (i = 0; i < ac; i++) {
                 ww[i].word = av[i];
                 ww[i].len = dps_strlen(av[i]);
                 ww[i].uword = t = (dpsunicode_t*)DpsMalloc((3 * ww[i].len + 1) * sizeof(dpsunicode_t));
		 if (ww[i].uword == NULL) return DPS_ERROR;
                 DpsConv(&file_uni, (char*)ww[i].uword, sizeof(dpsunicode_t) * (3 * ww[i].len + 1), av[i], ww[i].len + 1);
                 DpsUniStrToLower(ww[i].uword);
		 ww[i].uword = DpsUniNormalizeNFC(NULL, ww[i].uword);
		 DPS_FREE(t);
               }

               for (i = 0; i < ac - 1; i++) {
                 for (j = i + 1; j < ac; j++) {

                   if((Env->Synonyms.nsynonyms + 1) >= Env->Synonyms.msynonyms){
                    Env->Synonyms.msynonyms += 64;
                    Env->Synonyms.Synonym = (DPS_SYNONYM*)DpsRealloc(Env->Synonyms.Synonym, 
                                                   sizeof(DPS_SYNONYM)*Env->Synonyms.msynonyms);
		    if (Env->Synonyms.Synonym == NULL) {
		      Env->Synonyms.msynonyms = Env->Synonyms.nsynonyms = 0;
		      return DPS_ERROR;
  		    }
                   }
               
                   bzero((void*)&Env->Synonyms.Synonym[Env->Synonyms.nsynonyms], sizeof(DPS_SYNONYM));
               
                   /* Add direct order */
                   Env->Synonyms.Synonym[Env->Synonyms.nsynonyms].p.uword = DpsUniDup(ww[i].uword);
                   Env->Synonyms.Synonym[Env->Synonyms.nsynonyms].s.uword = DpsUniDup(ww[j].uword);
		   Env->Synonyms.Synonym[Env->Synonyms.nsynonyms].p.count = 
		     Env->Synonyms.Synonym[Env->Synonyms.nsynonyms].s.count = (size_t)((flag_th) ? key : 0);
                   Env->Synonyms.nsynonyms++;
               
                   bzero((void*)&Env->Synonyms.Synonym[Env->Synonyms.nsynonyms], sizeof(DPS_SYNONYM));
               
                   /* Add reverse order */
                   Env->Synonyms.Synonym[Env->Synonyms.nsynonyms].p.uword = DpsUniDup(ww[j].uword);
                   Env->Synonyms.Synonym[Env->Synonyms.nsynonyms].s.uword = DpsUniDup(ww[i].uword);
		   Env->Synonyms.Synonym[Env->Synonyms.nsynonyms].p.count = 
		     Env->Synonyms.Synonym[Env->Synonyms.nsynonyms].s.count = (size_t)((flag_th) ? key : 0);
                   Env->Synonyms.nsynonyms++;
                 }
               }

               for (i = 0; i < ac; i++) {
                 DPS_FREE(ww[i].uword);
               }
               do { key++; } while (key == 0);
          }
     loop_continue:
	  str = cur_n;
	  if (str != NULL) {
	    *str = savebyte;
	    cur_n = strchr(str, '\n');
	    if (cur_n != NULL) {
	      cur_n++;
	      savebyte = *cur_n;
	      *cur_n = '\0';
	    }
	  }
     }
     DPS_FREE(data);
     DPS_FREE(ww);
     DpsClose(fd);
     return DPS_OK;
}