mdw@git.distorted.org.uk Git - ezmlm/blob - ezmlm-archive.c

   1 /*$Id: ezmlm-archive.c,v 1.13 1999/11/28 20:13:32 lindberg Exp $*/
   2 /*$Name: ezmlm-idx-040 $*/
   3
   4 #include "alloc.h"
   5 #include "error.h"
   6 #include "stralloc.h"
   7 #include "str.h"
   8 #include "sig.h"
   9 #include "getconf.h"
  10 #include "strerr.h"
  11 #include "getln.h"
  12 #include "substdio.h"
  13 #include "readwrite.h"
  14 #include "fmt.h"
  15 #include "sgetopt.h"
  16 #include "idxthread.h"
  17 #include "makehash.h"
  18 #include "idx.h"
  19 #include "errtxt.h"
  20
  21 #define FATAL "ezmlm-archive: fatal: "
  22 #define WARNING "ezmlm-archive: warning: inconsistent index: "
  23
  24 substdio ssin;
  25 char inbuf[1024];
  26 substdio ssout;
  27 char outbuf[1024];
  28 substdio ssnum;
  29 char numbuf[16];
  30
  31 stralloc line = {0};
  32 stralloc num = {0};
  33 stralloc fn = {0};
  34 stralloc fnn = {0};
  35
  36 char strnum[FMT_ULONG];
  37 int flagerror = 0;
  38 int flagsync = 1;       /* sync() by default, not for -c or -f or -t */
  39 char *dir;
  40
  41 struct ca {
  42   char *s;              /* start */
  43   unsigned int l;       /* length */
  44 } ca;
  45
  46 void die_usage() {
  47   strerr_die1x(100,
  48     "ezmlm-archive: usage: "
  49         "ezmlm-archive [-cCFsSTvV] [-f min_msg] [-t max_msg] dir");
  50 }
  51
  52 void die_nomem() { strerr_die2x(111,FATAL,ERR_NOMEM); }
  53
  54 void close_proper(ss,s,sn)
  55 /* flush,sync,close,move sn->s) */
  56 substdio *ss;
  57 char *s, *sn;
  58 {
  59    if (substdio_flush(ss) == -1)
  60      strerr_die6sys(111,FATAL,ERR_FLUSH,dir,"/",s,": ");
  61   if (flagsync)
  62     if (fsync(ss->fd) == -1)
  63        strerr_die6sys(111,FATAL,ERR_SYNC,dir,"/",s,": ");
  64   if (close(ss->fd) == -1)
  65      strerr_die6sys(111,FATAL,ERR_CLOSE,dir,"/",s,": ");
  66   if (rename(sn,s) == -1)
  67      strerr_die6sys(111,FATAL,ERR_MOVE,dir,"/",sn,": ");
  68 }
  69
  70 void write_threads(msgtable,subtable,authtable,datetable,from,to)
  71 /* Add the current threading data to the thread database without dups */
  72 /* Writes the subject index first, then processes the individual files */
  73 msgentry *msgtable; subentry *subtable; authentry *authtable;
  74 dateentry *datetable;
  75 unsigned long from,to;
  76 {
  77   msgentry *pmsgt;
  78   subentry *psubt,*psubtm, *psubtlast;
  79   subentry *presubt = (subentry *)0;
  80   authentry *pautht;
  81   dateentry *pdatet;
  82   char *cp,*cp1;
  83   unsigned long msg;
  84   unsigned long ulmsginthread;
  85   unsigned long subnum;
  86   unsigned long authnum;
  87   unsigned long msgnum;
  88   unsigned int pos,l;
  89   unsigned int startdate,nextdate;
  90   unsigned int startmsg,nextmsg;
  91   int fd = -1;
  92   int fdn = -1;
  93   int match;
  94   int ffound;
  95   int lineno;
  96   int res;
  97
  98   psubtm = subtable;            /* now for new threads */
  99   pdatet = datetable;
 100   nextmsg = 0L;
 101   nextdate = pdatet->date;
 102   while (psubtm->sub) {         /* these are in msgnum order */
 103     if (!presubt)               /* for rewind */
 104       if (psubtm->lastmsg >= nextmsg)
 105         presubt = psubtm;       /* this thread extends beyond current month */
 106     if (psubtm->firstmsg >= nextmsg) {  /* done with this month */
 107       if (fdn != -1) close_proper(&ssout,fn.s,fnn.s);
 108       psubtlast = psubtm;               /* last thread done */
 109       if (presubt)                      /* need to rewind? */
 110         psubtm = presubt;               /* do it */
 111       psubt = psubtm;                   /* tmp pointer to reset done flag */
 112       presubt = (subentry *)0;          /* reset rewind pointer */
 113       pdatet++;                         /* next month */
 114       startdate = nextdate;             /* startdate */
 115       nextdate = pdatet->date;          /* end date */
 116       startmsg = nextmsg;               /* first message in month */
 117       nextmsg = pdatet->msg;            /* first message in next month */
 118       if (!stralloc_copys(&fn,"archive/threads/")) die_nomem();
 119       if (!stralloc_catb(&fn,strnum,fmt_uint(strnum,startdate))) die_nomem();
 120       if (!stralloc_copy(&fnn,&fn)) die_nomem();
 121       if (!stralloc_0(&fn)) die_nomem();
 122       if (!stralloc_cats(&fnn,"n")) die_nomem();
 123       if (!stralloc_0(&fnn)) die_nomem();
 124       if ((fdn = open_trunc(fnn.s)) == -1)
 125         strerr_die6sys(111,FATAL,ERR_CREATE,dir,"/",fnn.s,": ");
 126       substdio_fdbuf(&ssout,write,fdn,outbuf,sizeof(outbuf));
 127       if ((fd = open_read(fn.s)) == -1) {
 128       if (errno != error_noent)
 129              strerr_die6sys(111,FATAL,ERR_OPEN,dir,"/",fn.s,": ");
 130       } else {
 131         substdio_fdbuf(&ssin,read,fd,inbuf,sizeof(inbuf));
 132       for (;;) {
 133       if (getln(&ssin,&line,&match,'\n') == -1)
 134              strerr_die6sys(111,FATAL,ERR_READ,dir,"/",fn.s,": ");
 135       if (!match) break;
 136       pos = scan_ulong(line.s,&msgnum);
 137       pos++;                    /* skip ':' */
 138       if (msgnum >= from)
 139         continue;               /* ignore entries from threading range */
 140       if (line.len < pos + HASHLEN) {
 141         flagerror = -1;         /* and bad ones */
 142         continue;
 143       }
 144       psubt = subtable;
 145       cp = line.s + pos;
 146       ffound = 0;               /* search among already known subjects */
 147       for (;;) {
 148         res = str_diffn(psubt->sub,cp,HASHLEN);
 149         if (res < 0) {
 150           if (psubt->higher)
 151             psubt = psubt->higher;
 152          else
 153            break;
 154         } else if (res > 0) {
 155           if (psubt->lower)
 156             psubt = psubt->lower;
 157           else
 158             break;
 159         } else {
 160           ffound = 1;
 161           break;
 162         }
 163       }
 164       if (!ffound) {
 165         if (substdio_put(&ssout,line.s,line.len) == -1)
 166           strerr_die6sys(111,FATAL,ERR_WRITE,dir,"/",fnn.s,": ");
 167       } else {                  /* new # of msg in thread */
 168         cp += HASHLEN;          /* HASHLEN [#] Subject always \n at end */
 169         if (*(cp++) == ' ' && *(cp++) == '[') {
 170           cp += scan_ulong(cp,&ulmsginthread);
 171           if (*cp == ']') {
 172             psubt->msginthread += (unsigned char) (ulmsginthread & 0xff);
 173           }
 174         } else
 175           flagerror = -5;
 176       }
 177     }
 178     close(fd);
 179   }
 180   continue;
 181   }
 182
 183     if (psubtm->firstmsg < nextmsg && psubtm->lastmsg >= startmsg) {
 184     if (!stralloc_copyb(&line,strnum,fmt_ulong(strnum,psubtm->lastmsg)))
 185                 die_nomem();
 186     if (!stralloc_cats(&line,":")) die_nomem();
 187     if (!stralloc_catb(&line,psubtm->sub,HASHLEN)) die_nomem();
 188     if (!stralloc_cats(&line," [")) die_nomem();
 189     if (!stralloc_catb(&line,strnum,
 190         fmt_ulong(strnum,(unsigned long) psubtm->msginthread)))
 191                 die_nomem();
 192     if (!stralloc_cats(&line,"]")) die_nomem();
 193     if (!stralloc_catb(&line,psubtm->sub + HASHLEN,psubtm->sublen - HASHLEN))
 194                          die_nomem();   /* has \n */
 195     if (substdio_put(&ssout,line.s,line.len) == -1)
 196         strerr_die6sys(111,FATAL,ERR_WRITE,dir,"/",fnn.s,": ");
 197     }
 198   psubtm++;
 199   }
 200   if (fdn != -1)
 201     close_proper(&ssout,fn.s,fnn.s);
 202
 203   psubt = subtable;
 204   while (psubt->sub) {          /* now the threads */
 205     if (!stralloc_copys(&fn,"archive/subjects/")) die_nomem();
 206     if (!stralloc_catb(&fn,psubt->sub,2)) die_nomem();
 207     if (!stralloc_0(&fn)) die_nomem();
 208     if (mkdir(fn.s,0755) == -1)
 209     if (errno != error_exist)
 210       strerr_die6sys(111,FATAL,ERR_CREATE,dir,"/",fn.s,": ");
 211     fn.s[fn.len - 1] = '/';
 212     if (!stralloc_catb(&fn,psubt->sub+2,HASHLEN-2)) die_nomem();
 213     if (!stralloc_copy(&fnn,&fn)) die_nomem();
 214     if (!stralloc_cats(&fnn,"n")) die_nomem();
 215     if (!stralloc_0(&fn)) die_nomem();
 216     if (!stralloc_0(&fnn)) die_nomem();
 217     if ((fdn = open_trunc(fnn.s)) == -1)
 218       strerr_die4sys(111,FATAL,ERR_CREATE,fnn.s,": ");
 219     substdio_fdbuf(&ssout,write,fdn,outbuf,sizeof(outbuf));
 220     if ((fd = open_read(fn.s)) == -1) {
 221       if (errno != error_noent)
 222           strerr_die4sys(111,FATAL,ERR_OPEN,fn.s,": ");
 223       if (substdio_puts(&ssout,psubt->sub) == -1)       /* write subject */
 224              strerr_die6sys(111,FATAL,ERR_WRITE,dir,"/",fnn.s,": ");
 225     } else {                                    /* copy data */
 226         substdio_fdbuf(&ssin,read,fd,inbuf,sizeof(inbuf));
 227         lineno = 0;
 228         for (;;) {
 229           if (getln(&ssin,&line,&match,'\n') == -1)
 230              strerr_die6sys(111,FATAL,ERR_READ,dir,"/",fn.s,": ");
 231           if (!match) break;
 232           if (!lineno) {                        /* write subject */
 233             if (line.len < HASHLEN + 1 || line.s[HASHLEN] != ' ')
 234                 flagerror = -3;
 235             if (substdio_put(&ssout,line.s,line.len) == -1)
 236                strerr_die6sys(111,FATAL,ERR_WRITE,dir,"/",fnn.s,": ");
 237             lineno = 1;
 238             continue;
 239           }
 240           (void) scan_ulong(line.s,&msgnum);
 241           if (msgnum >= from) break;
 242           if (substdio_put(&ssout,line.s,line.len) == -1)
 243              strerr_die6sys(111,FATAL,ERR_WRITE,dir,"/",fnn.s,": ");
 244         }
 245         (void) close(fd);       /* close old index */
 246       }
 247
 248     subnum = (unsigned long) (psubt - subtable + 1);    /* idx of this subj */
 249     pmsgt = msgtable + psubt->firstmsg - from;  /* first message entry */
 250     for (msg = psubt->firstmsg; msg <= psubt->lastmsg; msg++) {
 251       if (pmsgt->subnum == subnum) {
 252         if (!stralloc_copyb(&line,strnum,fmt_ulong(strnum,msg))) die_nomem();
 253         if (!stralloc_cats(&line,":")) die_nomem();
 254         if (!stralloc_catb(&line,strnum,fmt_uint(strnum,pmsgt->date)))
 255                 die_nomem();
 256         if (!stralloc_cats(&line,":")) die_nomem();
 257         if (pmsgt->authnum) {
 258           pautht = authtable + pmsgt->authnum - 1;
 259           cp = pautht->auth;
 260           cp1 = cp + str_chr(cp,' ');
 261           if (cp + HASHLEN != cp1)
 262             strerr_die1x(100,ERR_BAD_INDEX);
 263           if (!stralloc_cats(&line,cp))
 264                 die_nomem();                            /* hash */
 265         } else
 266           if (!stralloc_cats(&line,"\n")) die_nomem();
 267         if (substdio_put(&ssout,line.s,line.len) == -1)
 268           strerr_die6sys(111,FATAL,ERR_WRITE,dir,"/",fnn.s,": ");
 269       }
 270       pmsgt++;
 271     }
 272     close_proper(&ssout,fn.s,fnn.s);
 273     psubt++;
 274   }
 275
 276                                         /* (no master author index) */
 277   pautht = authtable;
 278   while (pautht->auth) {                /* now the authors */
 279     if (!stralloc_copys(&fn,"archive/authors/")) die_nomem();
 280     if (!stralloc_catb(&fn,pautht->auth,2)) die_nomem();
 281     if (!stralloc_0(&fn)) die_nomem();
 282     if (mkdir(fn.s,0755) == -1)
 283     if (errno != error_exist)
 284       strerr_die6sys(111,FATAL,ERR_CREATE,dir,"/",fn.s,": ");
 285     fn.s[fn.len - 1] = '/';
 286     if (!stralloc_catb(&fn,pautht->auth+2,HASHLEN-2)) die_nomem();
 287     if (!stralloc_copy(&fnn,&fn)) die_nomem();
 288     if (!stralloc_cats(&fnn,"n")) die_nomem();
 289     if (!stralloc_0(&fn)) die_nomem();
 290     if (!stralloc_0(&fnn)) die_nomem();
 291     if ((fdn = open_trunc(fnn.s)) == -1)
 292       strerr_die4sys(111,FATAL,ERR_CREATE,fnn.s,": ");
 293     substdio_fdbuf(&ssout,write,fdn,outbuf,sizeof(outbuf));
 294       if ((fd = open_read(fn.s)) == -1) {
 295         if (errno != error_noent)
 296           strerr_die4sys(111,FATAL,ERR_OPEN,fn.s,": ");
 297         else {                  /* didn't exist before: write author */
 298           if (substdio_put(&ssout,pautht->auth,pautht->authlen) == -1)
 299              strerr_die6sys(111,FATAL,ERR_WRITE,dir,"/",fnn.s,": ");
 300         }
 301       } else {                                  /* copy data */
 302         substdio_fdbuf(&ssin,read,fd,inbuf,sizeof(inbuf));
 303         lineno = 0;
 304         for (;;) {
 305           if (getln(&ssin,&line,&match,'\n') == -1)
 306              strerr_die6sys(111,FATAL,ERR_READ,dir,"/",fn.s,": ");
 307           if (!match) break;
 308           if (!lineno) {                        /* write author */
 309             if (line.len < HASHLEN + 1 || line.s[HASHLEN] != ' ')
 310                 flagerror = - 4;
 311             if (substdio_put(&ssout,line.s,line.len) == -1)
 312                strerr_die6sys(111,FATAL,ERR_WRITE,dir,"/",fnn.s,": ");
 313             lineno = 1;
 314             continue;
 315           }
 316           (void) scan_ulong(line.s,&msgnum);
 317           if (msgnum >= from) break;
 318           if (substdio_put(&ssout,line.s,line.len) == -1)
 319              strerr_die6sys(111,FATAL,ERR_WRITE,dir,"/",fnn.s,": ");
 320         }
 321         (void) close(fd);                       /* close old index */
 322       }
 323
 324     authnum = (unsigned long) (pautht - authtable + 1); /* idx of this auth */
 325     pmsgt = msgtable + pautht->firstmsg - from; /* first message entry */
 326     for (msg = pautht->firstmsg; msg <= to; msg++) {
 327       if (pmsgt->authnum == authnum) {
 328         if (!stralloc_copyb(&line,strnum,fmt_ulong(strnum,msg))) die_nomem();
 329         if (!stralloc_cats(&line,":")) die_nomem();
 330         if (!stralloc_catb(&line,strnum,fmt_uint(strnum,pmsgt->date)))
 331                 die_nomem();
 332         if (!stralloc_cats(&line,":")) die_nomem();
 333         if (pmsgt->subnum) {
 334           psubt = subtable + pmsgt->subnum - 1;
 335           if (!stralloc_catb(&line,psubt->sub,psubt->sublen))
 336                 die_nomem();
 337         }
 338         if (substdio_put(&ssout,line.s,line.len) == -1)
 339           strerr_die6sys(111,FATAL,ERR_WRITE,dir,"/",fnn.s,": ");
 340       }
 341       pmsgt++;
 342     }
 343     close_proper(&ssout,fn.s,fnn.s);
 344     pautht++;
 345   }
 346 }
 347
 348 int main(argc,argv)
 349 int argc;
 350 char **argv;
 351 {
 352   unsigned long archnum = 0L;
 353   unsigned long to = 0L;
 354   unsigned long max;
 355   int fd;
 356   int fdlock;
 357   int flagcreate = 0;
 358   int flagsyncall = 0;
 359   int opt;
 360   msgentry *msgtable;
 361   subentry *subtable;
 362   authentry *authtable;
 363   dateentry *datetable;
 364
 365   (void) umask(022);
 366   sig_pipeignore();
 367
 368   while ((opt = getopt(argc,argv,"cCf:FsSt:TvV")) != opteof)
 369     switch (opt) {
 370       case 'c': flagcreate = 1;
 371                 flagsync = 0;
 372                 break;                  /* start at beginning of archive */
 373       case 'C': flagcreate = 0;
 374                 break;  /* Do only archnum+1 => num */
 375       case 'f': if (optarg) {
 376                   (void) scan_ulong(optarg,&archnum);
 377                   archnum = (archnum / 100) * 100;
 378                 }
 379                 flagsync = 0;
 380                 break;
 381       case 'F': archnum = 0; break;
 382       case 's': flagsyncall = 1; break;
 383       case 'S': flagsyncall = 0; break;
 384       case 't': if (optarg) {
 385                   (void) scan_ulong(optarg,&to);
 386                 }
 387                 flagsync = 0;
 388                 break;
 389       case 'T': to = 0; break;
 390       case 'v':
 391       case 'V': strerr_die2x(0,"ezmlm-archive version: ",EZIDX_VERSION);
 392       default:
 393         die_usage();
 394     }
 395
 396   if (flagsyncall) flagsync = 1;        /* overrides */
 397   dir = argv[optind++];
 398   if (!dir) die_usage();
 399   if (chdir(dir) == -1)
 400     strerr_die4sys(111,FATAL,ERR_SWITCH,dir,": ");
 401
 402   if (mkdir("archive/threads",0755) == -1)
 403     if (errno != error_exist)
 404       strerr_die4sys(111,FATAL,ERR_CREATE,dir,"/archive/threads: ");
 405   if (mkdir("archive/subjects",0755) == -1)
 406     if (errno != error_exist)
 407       strerr_die4sys(111,FATAL,ERR_CREATE,dir,"/archive/subjects: ");
 408   if (mkdir("archive/authors",0755) == -1)
 409     if (errno != error_exist)
 410       strerr_die4sys(111,FATAL,ERR_CREATE,dir,"/archive/authors: ");
 411
 412         /* Lock list to assure that no ezmlm-send is working on it */
 413         /* and that the "num" message is final */
 414   fdlock = open_append("lock");
 415   if (fdlock == -1)
 416     strerr_die2sys(111,FATAL,ERR_OPEN_LOCK);
 417   if (lock_ex(fdlock) == -1) {
 418     (void) close(fdlock);
 419     strerr_die2sys(111,FATAL,ERR_OBTAIN_LOCK);
 420   }
 421                                         /* get num */
 422   if (!getconf_line(&num,"num",0,FATAL,dir))
 423     strerr_die1x(100,ERR_EMPTY_LIST);
 424   (void) close(fdlock);
 425
 426   if (!stralloc_0(&num)) die_nomem();   /* parse num */
 427   (void) scan_ulong(num.s,&max);
 428   if (!to || to > max) to = max;
 429
 430   fdlock = open_append("archive/lock"); /* lock index */
 431   if (fdlock == -1)
 432     strerr_die4sys(111,FATAL,ERR_OPEN,dir,"/archive/lock: ");
 433   if (lock_ex(fdlock) == -1) {
 434     (void) close(fdlock);
 435     strerr_die4sys(111,FATAL,ERR_OBTAIN,dir,"/archive/lock: ");
 436   }
 437   if (!flagcreate && !archnum) {        /* adjust archnum (from) / to */
 438     if (getconf_line(&num,"archnum",0,FATAL,dir)) {
 439       if (!stralloc_0(&num)) die_nomem();
 440       (void) scan_ulong(num.s,&archnum);
 441       archnum++;
 442     }
 443   }
 444
 445   if (archnum > to)
 446     _exit(0);                           /* nothing to do */
 447
 448                                         /* do the subject threading */
 449   idx_mkthreads(&msgtable,&subtable,&authtable,&datetable,
 450         archnum,to,max,0,FATAL);
 451                                         /* update the index */
 452   write_threads(msgtable,subtable,authtable,datetable,archnum,to);
 453                                         /* update archnum */
 454   if ((fd = open_trunc("archnumn")) == -1)
 455     strerr_die4sys(111,FATAL,ERR_CREATE,dir,"/archnumn: ");
 456   substdio_fdbuf(&ssnum,write,fd,numbuf,sizeof(numbuf));
 457   if (substdio_put(&ssnum,strnum,fmt_ulong(strnum,to)) == -1)
 458      strerr_die6sys(111,FATAL,ERR_WRITE,dir,"/",fnn.s,": ");
 459   if (substdio_puts(&ssnum,"\n") == -1)
 460      strerr_die6sys(111,FATAL,ERR_WRITE,dir,"/",fnn.s,": ");
 461   close_proper(&ssnum,"archnum","archnumn");
 462   switch (flagerror) {
 463     case 0:
 464        _exit(0);                                /* go bye-bye */
 465     case -1:
 466        strerr_die2x(99,WARNING,"threads entry with illegal format");
 467     case -2:
 468        strerr_die2x(99,WARNING,"thread in index, but threadfile missing");
 469     case -3:
 470        strerr_die2x(99,WARNING,"a subject file lacks subject");
 471     case -4:
 472        strerr_die2x(99,WARNING,"an author file lacks author/hash");
 473     case -5:
 474        strerr_die2x(99,WARNING,"threads entry lacks message count");
 475     default:
 476        strerr_die2x(99,WARNING,"something happened that isn't quite right");
 477   }
 478 }
 479