| 1 | /* ifupd.c - Update an inverted file |
|---|
| 2 | */ |
|---|
| 3 | |
|---|
| 4 | #include <stdio.h> |
|---|
| 5 | #include <string.h> |
|---|
| 6 | |
|---|
| 7 | #include "cisis.h" |
|---|
| 8 | #include "ciupi.h" |
|---|
| 9 | #include "cirun.h" |
|---|
| 10 | |
|---|
| 11 | char *fstarea=NULL; |
|---|
| 12 | char *stwp=NULL; |
|---|
| 13 | int nstws; |
|---|
| 14 | |
|---|
| 15 | FST_CODE *fstpgmp; |
|---|
| 16 | LONGX maxlk1=1000; |
|---|
| 17 | LONGX maxlk2=500; |
|---|
| 18 | |
|---|
| 19 | |
|---|
| 20 | void main(argc,argv) |
|---|
| 21 | int argc; |
|---|
| 22 | char *argv[]; |
|---|
| 23 | |
|---|
| 24 | { |
|---|
| 25 | char *dbnp,*ifnp; |
|---|
| 26 | DBXSTRU *dbxp; |
|---|
| 27 | char *p,*q,qline[CIMPL+1]; |
|---|
| 28 | int argnext=1; |
|---|
| 29 | int parmtrace=0; |
|---|
| 30 | LONGX parmfrom,parmto,parmcount,tell,rangtell,mfn; |
|---|
| 31 | int pstflag=IFUPISIS; |
|---|
| 32 | int rc=0; |
|---|
| 33 | |
|---|
| 34 | #if LIND |
|---|
| 35 | LONGX parmxmfn=0L; |
|---|
| 36 | char *parmlogp=NULL; |
|---|
| 37 | #endif |
|---|
| 38 | |
|---|
| 39 | if (argc < 2) { |
|---|
| 40 | #if LINDLUX |
|---|
| 41 | printf("%s",cicopyr("Utility IFUPD/LINDLUX")); |
|---|
| 42 | #else |
|---|
| 43 | printf("%s",cicopyr("Utility IFUPD")); |
|---|
| 44 | #endif |
|---|
| 45 | printf("\n"); |
|---|
| 46 | printf("ifupd "); |
|---|
| 47 | #if MULTI |
|---|
| 48 | printf("[mono|full] "); |
|---|
| 49 | #endif |
|---|
| 50 | printf("[create=]<ifname> [<option> [...]]\n"); |
|---|
| 51 | printf(" \n"); |
|---|
| 52 | #if MULTI |
|---|
| 53 | printf(" [mono|full] -> single/multi user operation \n"); |
|---|
| 54 | #endif |
|---|
| 55 | printf(" [create=] -> to delete+create <ifname> \n"); |
|---|
| 56 | printf(" <ifname> -> output inverted file \n"); |
|---|
| 57 | printf(" \n"); |
|---|
| 58 | printf("options: \n"); |
|---|
| 59 | printf("\n"); |
|---|
| 60 | printf(" fst=<fstspec>|@[fstfile] -> field select table \n"); |
|---|
| 61 | printf(" stw=<stwspec>|@[stwfile] -> stop words \n"); |
|---|
| 62 | printf(" -posts -> (init and) do not load .ifp \n"); |
|---|
| 63 | #if !LIND |
|---|
| 64 | printf(" master=<name> -> alternate master file \n"); |
|---|
| 65 | #else |
|---|
| 66 | printf(" maxmfn=<n> -> max master file number \n"); |
|---|
| 67 | printf(" master=<dbn> -> master file to get maxmfn \n"); |
|---|
| 68 | printf(" log=<file> -> log file name \n"); |
|---|
| 69 | #endif |
|---|
| 70 | printf(" actab=<file> -> alphabetic chars table \n"); |
|---|
| 71 | printf(" uctab=<file> -> upper case chars table \n"); |
|---|
| 72 | printf(" from=<n> -> initial mfn \n"); |
|---|
| 73 | printf(" to=<n> -> final mfn \n"); |
|---|
| 74 | printf(" count=<n> -> max mfns \n"); |
|---|
| 75 | printf(" tell=<n> -> tell <n>%% loaded \n"); |
|---|
| 76 | printf(" \n"); |
|---|
| 77 | /* |
|---|
| 78 | printf("FMT=%ld \n",fst_fmtby); |
|---|
| 79 | printf(" \n"); |
|---|
| 80 | */ |
|---|
| 81 | exit(1); |
|---|
| 82 | } |
|---|
| 83 | |
|---|
| 84 | /* get dbn */ |
|---|
| 85 | dbnp=argv[argnext++]; |
|---|
| 86 | #if MULTI |
|---|
| 87 | if (strcmp(dbnp,"mono") == 0) { |
|---|
| 88 | cipnetws=MONONETS; |
|---|
| 89 | dbnp=argv[argnext++]; |
|---|
| 90 | } else |
|---|
| 91 | if (strcmp(dbnp,"mast") == 0) { |
|---|
| 92 | cipnetws=MASTNETS; |
|---|
| 93 | dbnp=argv[argnext++]; |
|---|
| 94 | } |
|---|
| 95 | if (strcmp(dbnp,"full") == 0) { |
|---|
| 96 | cipnetws=FULLNETS; |
|---|
| 97 | dbnp=argv[argnext++]; |
|---|
| 98 | } |
|---|
| 99 | #endif |
|---|
| 100 | if (strncmp(dbnp,"create?",7) == 0 || strncmp(dbnp,"create=",7) == 0) { |
|---|
| 101 | dbnp+=7; |
|---|
| 102 | trmisis0(dbnp); |
|---|
| 103 | } |
|---|
| 104 | ifnp=dbnp; |
|---|
| 105 | |
|---|
| 106 | parmfrom=1L; parmto=parmcount=9999999L; tell=0L; |
|---|
| 107 | for (; argnext < argc; ) { |
|---|
| 108 | |
|---|
| 109 | p=argv[argnext++]; |
|---|
| 110 | |
|---|
| 111 | if (strncmp(p,"reset?",6) == 0 || strncmp(p,"reset=",6) == 0) { |
|---|
| 112 | if (sscanf(p+6,"%d",&ifupd_reset) != 1) fatal(p); |
|---|
| 113 | continue; |
|---|
| 114 | } |
|---|
| 115 | if (strncmp(p,"master?",7) == 0 || strncmp(p,"master=",7) == 0) { |
|---|
| 116 | dbnp=p+7; |
|---|
| 117 | if (!*dbnp) fatal(p); |
|---|
| 118 | continue; |
|---|
| 119 | } |
|---|
| 120 | |
|---|
| 121 | #if LIND |
|---|
| 122 | if (strncmp(p,"maxmfn?",7) == 0 || strncmp(p,"maxmfn=",7) == 0) { |
|---|
| 123 | if (sscanf(p+7,"%ld",&parmxmfn) != 1) fatal(p); |
|---|
| 124 | if (parmxmfn < 1) fatal(p); |
|---|
| 125 | continue; |
|---|
| 126 | } |
|---|
| 127 | if (strncmp(p,"log?",4) == 0 || strncmp(p,"log=",4) == 0) { |
|---|
| 128 | parmlogp=p+4; |
|---|
| 129 | if (!*parmlogp) fatal(p); |
|---|
| 130 | continue; |
|---|
| 131 | } |
|---|
| 132 | #endif |
|---|
| 133 | if (strncmp(p,"from?",5) == 0 || strncmp(p,"from=",5) == 0) { |
|---|
| 134 | if (sscanf(p+5,"%ld",&parmfrom) != 1) fatal(p); |
|---|
| 135 | continue; |
|---|
| 136 | } |
|---|
| 137 | if (strncmp(p,"to?",3) == 0 || strncmp(p,"to=",3) == 0) { |
|---|
| 138 | if (sscanf(p+3,"%ld",&parmto) != 1) fatal(p); |
|---|
| 139 | continue; |
|---|
| 140 | } |
|---|
| 141 | if (strncmp(p,"count?",6) == 0 || strncmp(p,"count=",6) == 0) { |
|---|
| 142 | if (sscanf(p+6,"%ld",&parmcount) != 1) fatal(p); |
|---|
| 143 | continue; |
|---|
| 144 | } |
|---|
| 145 | if (strncmp(p,"tell=",5) == 0) { |
|---|
| 146 | if (sscanf(p+5,"%ld",&tell) != 1) fatal(p); |
|---|
| 147 | if (tell < 0) fatal(p); |
|---|
| 148 | continue; |
|---|
| 149 | } |
|---|
| 150 | if (strcmp(p,"trace?all") == 0|| strcmp(p,"trace=all") == 0) { |
|---|
| 151 | dbxtrace=rectrace=parmtrace=fsttrace=multrace=trmtrace=1; |
|---|
| 152 | continue; |
|---|
| 153 | } |
|---|
| 154 | if (strcmp(p,"trace") == 0) { |
|---|
| 155 | dbxtrace=rectrace=multrace=parmtrace=1; |
|---|
| 156 | continue; |
|---|
| 157 | } |
|---|
| 158 | |
|---|
| 159 | if (strncmp(p,"maxlk1?",7) == 0 || strncmp(p,"maxlk1=",7) == 0) { |
|---|
| 160 | if (sscanf(p+7,"%ld",&maxlk1) != 1) fatal(p); |
|---|
| 161 | continue; |
|---|
| 162 | } |
|---|
| 163 | if (strncmp(p,"maxlk2?",7) == 0 || strncmp(p,"maxlk2=",7) == 0) { |
|---|
| 164 | if (sscanf(p+7,"%ld",&maxlk2) != 1) fatal(p); |
|---|
| 165 | continue; |
|---|
| 166 | } |
|---|
| 167 | |
|---|
| 168 | if (strncmp(p,"fst?",4) == 0 || strncmp(p,"fst=",4) == 0) { |
|---|
| 169 | q=p+4; |
|---|
| 170 | if (strcmp(q,"@") == 0) sprintf(q=qline,"@%s.fst",ifnp); |
|---|
| 171 | fstarea=loadfile(ifnp,'@',q,fstarea,0L,'\n'); |
|---|
| 172 | if (!fstarea) fatal(q); |
|---|
| 173 | if (parmtrace) { showcore("+++ fst"); printf("%s.\n",fstarea); } |
|---|
| 174 | fst_gener(&fstpgmp,fstarea); |
|---|
| 175 | if (parmtrace) showcore("+++ fst"); |
|---|
| 176 | if (fst_error) { |
|---|
| 177 | printf("\n*** FST error#%ld at line %ld\n",fst_error,fst_errl); |
|---|
| 178 | if (fst_error == FSTERRF) |
|---|
| 179 | printf("\n*** FMT error#%ld at offset %ld\n", |
|---|
| 180 | fmt_error,fmt_errof); |
|---|
| 181 | fatal(fst_errp); |
|---|
| 182 | } |
|---|
| 183 | continue; |
|---|
| 184 | } |
|---|
| 185 | |
|---|
| 186 | if (strncmp(p,"stw?",4) == 0 || strncmp(p,"stw=",4) == 0) { |
|---|
| 187 | q=p+4; |
|---|
| 188 | if (*q++ != '@') fatal(p); |
|---|
| 189 | if (!*q) sprintf(q=qline,"%s.stw",ifnp); |
|---|
| 190 | if (parmtrace) showcore("+++ stw"); |
|---|
| 191 | stwp=loadstw(ifnp,q,stwp,0L,&nstws); |
|---|
| 192 | if (parmtrace) { |
|---|
| 193 | for (q=stwp; *q; q+=LE1+1) printf("%s/",q); |
|---|
| 194 | printf("=%d\n",nstws); |
|---|
| 195 | showcore("+++ stw"); |
|---|
| 196 | } |
|---|
| 197 | continue; |
|---|
| 198 | } |
|---|
| 199 | |
|---|
| 200 | if (strcmp(p,"-posts") == 0) { |
|---|
| 201 | pstflag=IFUPDICT; |
|---|
| 202 | continue; |
|---|
| 203 | } |
|---|
| 204 | |
|---|
| 205 | if (strncmp(p,"uctab?",6) == 0 || strncmp(p,"uctab=",6) == 0) { |
|---|
| 206 | if (!loaductb(ifnp,isisuctab,p+6)) fatal(p); /* AOT, 02/04/2001 */ |
|---|
| 207 | continue; |
|---|
| 208 | } |
|---|
| 209 | if (strncmp(p,"actab?",6) == 0 || strncmp(p,"actab=",6) == 0) { |
|---|
| 210 | unsigned char ubuffup[256]; |
|---|
| 211 | memset(isiswctab,0x00,256); |
|---|
| 212 | if (!loadactb(ifnp,ubuffup,p+6)) fatal(p); /* AOT, 02/04/2001 */ |
|---|
| 213 | for (isiswctot=0; ubuffup[isiswctot]; isiswctot=isiswctot+1) |
|---|
| 214 | isiswctab[ubuffup[isiswctot]]=1; |
|---|
| 215 | continue; |
|---|
| 216 | } |
|---|
| 217 | |
|---|
| 218 | fatal(p); |
|---|
| 219 | } |
|---|
| 220 | |
|---|
| 221 | #if LIND |
|---|
| 222 | /* get maxmfn */ |
|---|
| 223 | if (!parmxmfn) { |
|---|
| 224 | parmxmfn=VMF0nxtmfn(record(nrecs,dbnp,0L))-1; |
|---|
| 225 | dbxflush(dbnp); |
|---|
| 226 | } |
|---|
| 227 | #endif |
|---|
| 228 | |
|---|
| 229 | dbxp=dbxstorp(ifnp); /* inverted file ! */ |
|---|
| 230 | |
|---|
| 231 | if (parmto == 9999999L) { |
|---|
| 232 | parmto=VMF0nxtmfn(record(nrecs,dbnp,0L))-1; |
|---|
| 233 | } |
|---|
| 234 | if (parmto > parmfrom+parmcount-1) parmto=parmfrom+parmcount-1; |
|---|
| 235 | |
|---|
| 236 | if (tell) { |
|---|
| 237 | rangtell=parmto-parmfrom+1; |
|---|
| 238 | if (rangtell <= 0) rangtell=1; |
|---|
| 239 | if (tell < 0 || tell > 100) tell=1; |
|---|
| 240 | tell=rangtell/(100/tell); |
|---|
| 241 | if (tell <= 0) tell=1; |
|---|
| 242 | } |
|---|
| 243 | |
|---|
| 244 | |
|---|
| 245 | if (tell) { DBXitell=tell; DBXirang=rangtell; } |
|---|
| 246 | if (parmtrace) DBXitrac=parmtrace; |
|---|
| 247 | |
|---|
| 248 | #if LIND |
|---|
| 249 | invsetup(ifnp,0L,0L,0L); |
|---|
| 250 | #if LINDLUX |
|---|
| 251 | DBXifmap->ltxthresh=1; /* noderead()/nodewrit() */ |
|---|
| 252 | DBXifmap->luxthresh=parmxmfn/8+1; /* leafread()/leafwrit()/threshold */ |
|---|
| 253 | DBXifmap->lvxiflind=0; /* ifupdat() */ |
|---|
| 254 | DBXifmap->lvxmaxmfn=parmxmfn; /* ifupdat() */ |
|---|
| 255 | #endif /* LINDLUX */ |
|---|
| 256 | #endif |
|---|
| 257 | |
|---|
| 258 | #if MULTI |
|---|
| 259 | if (DBXnetws != MONONETS) |
|---|
| 260 | for (mfn=parmfrom; mfn <= parmto; mfn++) { |
|---|
| 261 | rc=ifupdat(dbnp,mfn,mfn, |
|---|
| 262 | ifnp,fstpgmp,stwp,maxlk1,maxlk2,pstflag,IFUPWRIT); |
|---|
| 263 | if (rc) break; |
|---|
| 264 | } |
|---|
| 265 | else { |
|---|
| 266 | rc=ifupdat(dbnp,parmfrom,parmto, |
|---|
| 267 | ifnp,fstpgmp,stwp,maxlk1,maxlk2,pstflag,IFUPCORE); |
|---|
| 268 | |
|---|
| 269 | rc+=ifupdat(dbnp,0L,0L, |
|---|
| 270 | ifnp,fstpgmp,stwp,maxlk1,maxlk2,pstflag,IFUPWRIT); |
|---|
| 271 | } |
|---|
| 272 | #else |
|---|
| 273 | rc=ifupdat(dbnp,parmfrom,parmto, |
|---|
| 274 | ifnp,fstpgmp,stwp,maxlk1,maxlk2,pstflag,IFUPCORE); |
|---|
| 275 | |
|---|
| 276 | rc+=ifupdat(dbnp,0L,0L, |
|---|
| 277 | ifnp,fstpgmp,stwp,maxlk1,maxlk2,pstflag,IFUPWRIT); |
|---|
| 278 | #endif |
|---|
| 279 | |
|---|
| 280 | #if !LIND /* v */ |
|---|
| 281 | if (tell) { /* 02/06/98 */ |
|---|
| 282 | printf("%ld records updated \n",DBXirecs); |
|---|
| 283 | printf("%ld+%ld=%ld postings added \n",DBXipadd[0],DBXipadd[1], |
|---|
| 284 | DBXipadd[0]+DBXipadd[1]); |
|---|
| 285 | printf("%ld+%ld=%ld postings deleted \n",DBXipdel[0],DBXipdel[1], |
|---|
| 286 | DBXipdel[0]+DBXipdel[1]); |
|---|
| 287 | } |
|---|
| 288 | #else /* !LIND v */ |
|---|
| 289 | printf("%ld records read \n",DBXirecs); |
|---|
| 290 | printf("%ld+%ld records on %s+%s \n",DBXifmap->cn[0].nmaxpos, |
|---|
| 291 | DBXifmap->cn[1].nmaxpos, |
|---|
| 292 | nx12extp[0],nx12extp[1]); |
|---|
| 293 | printf("%ld+%ld records on %s+%s \n",DBXifmap->cn[0].fmaxpos, |
|---|
| 294 | DBXifmap->cn[1].fmaxpos, |
|---|
| 295 | lx12extp[0],lx12extp[1]); |
|---|
| 296 | printf("%ld+%ld=%ld postings to add \n",DBXipadd[0],DBXipadd[1], |
|---|
| 297 | DBXipadd[0]+DBXipadd[1]); |
|---|
| 298 | #endif /* !LIND v */ |
|---|
| 299 | |
|---|
| 300 | #if LIND |
|---|
| 301 | if (pstflag != IFUPDICT) { |
|---|
| 302 | #if LINDLUX |
|---|
| 303 | DBXifmap->ltxthresh=1; /* noderead()/nodewrit() */ |
|---|
| 304 | DBXifmap->luxthresh=parmxmfn/8+1; /* leafread()/leafwrit()/threshold */ |
|---|
| 305 | DBXifmap->lvxiflind=1; /* svdifupd() */ |
|---|
| 306 | #endif /* LINDLUX */ |
|---|
| 307 | |
|---|
| 308 | ifupdat(dbnp,parmfrom,parmto, |
|---|
| 309 | ifnp,fstpgmp,stwp,maxlk1,maxlk2,pstflag,IFUPCORE); |
|---|
| 310 | |
|---|
| 311 | ifupdat(dbnp,0L,0L, |
|---|
| 312 | ifnp,fstpgmp,stwp,maxlk1,maxlk2,pstflag,IFUPWRIT); |
|---|
| 313 | |
|---|
| 314 | printf("%ld records read \n",DBXirecs); |
|---|
| 315 | printf("%ld+%ld=%ld postings added \n",DBXipadd[0],DBXipadd[1], |
|---|
| 316 | DBXipadd[0]+DBXipadd[1]); |
|---|
| 317 | } /* end if pstflag != IFUPDICT */ |
|---|
| 318 | #endif |
|---|
| 319 | |
|---|
| 320 | if (rc) fatal("error while updating the IF"); |
|---|
| 321 | |
|---|
| 322 | exit(0); |
|---|
| 323 | } |
|---|