mandoc-db.c

   1 /*      $Id: mandoc-db.c,v 1.18 2011/05/04 08:21:17 kristaps Exp $ */
   2 /*
   3  * Copyright (c) 2011 Kristaps Dzonsons <kristaps@bsd.lv>
   4  *
   5  * Permission to use, copy, modify, and distribute this software for any
   6  * purpose with or without fee is hereby granted, provided that the above
   7  * copyright notice and this permission notice appear in all copies.
   8  *
   9  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
  10  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
  11  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
  12  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
  13  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
  14  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
  15  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
  16  */
  17 #ifdef HAVE_CONFIG_H
  18 #include "config.h"
  19 #endif
  20
  21 #include <sys/param.h>
  22
  23 #include <assert.h>
  24 #ifdef __linux__
  25 # include <db_185.h>
  26 #else
  27 # include <db.h>
  28 #endif
  29 #include <fcntl.h>
  30 #include <getopt.h>
  31 #include <stdio.h>
  32 #include <stdint.h>
  33 #include <stdlib.h>
  34 #include <string.h>
  35
  36 #include "man.h"
  37 #include "mdoc.h"
  38 #include "mandoc.h"
  39
  40 #define MANDOC_DB        "mandoc.db"
  41 #define MANDOC_IDX       "mandoc.index"
  42 #define MANDOC_BUFSZ      BUFSIZ
  43 #define MANDOC_FLAGS      O_CREAT|O_TRUNC|O_RDWR
  44
  45 enum    type {
  46         MANDOC_NONE = 0,
  47         MANDOC_NAME,
  48         MANDOC_FUNCTION,
  49         MANDOC_UTILITY,
  50         MANDOC_INCLUDES,
  51         MANDOC_VARIABLE,
  52         MANDOC_STANDARD,
  53         MANDOC_AUTHOR,
  54         MANDOC_CONFIG
  55 };
  56
  57 #define MAN_ARGS          DB *db, \
  58                           const char *dbn, \
  59                           DBT *key, size_t *ksz, \
  60                           DBT *val, \
  61                           DBT *rval, size_t *rsz, \
  62                           const struct man_node *n
  63 #define MDOC_ARGS         DB *db, \
  64                           const char *dbn, \
  65                           DBT *key, size_t *ksz, \
  66                           DBT *val, \
  67                           DBT *rval, size_t *rsz, \
  68                           const struct mdoc_node *n
  69
  70 static  void              dbt_append(DBT *, size_t *, const char *);
  71 static  void              dbt_appendb(DBT *, size_t *,
  72                                 const void *, size_t);
  73 static  void              dbt_init(DBT *, size_t *);
  74 static  void              dbt_put(DB *, const char *, DBT *, DBT *);
  75 static  void              usage(void);
  76 static  void              pman(DB *, const char *, DBT *, size_t *,
  77                                 DBT *, DBT *, size_t *, struct man *);
  78 static  int               pman_node(MAN_ARGS);
  79 static  void              pmdoc(DB *, const char *, DBT *, size_t *,
  80                                 DBT *, DBT *, size_t *, struct mdoc *);
  81 static  void              pmdoc_node(MDOC_ARGS);
  82 static  void              pmdoc_An(MDOC_ARGS);
  83 static  void              pmdoc_Cd(MDOC_ARGS);
  84 static  void              pmdoc_Fd(MDOC_ARGS);
  85 static  void              pmdoc_In(MDOC_ARGS);
  86 static  void              pmdoc_Fn(MDOC_ARGS);
  87 static  void              pmdoc_Fo(MDOC_ARGS);
  88 static  void              pmdoc_Nd(MDOC_ARGS);
  89 static  void              pmdoc_Nm(MDOC_ARGS);
  90 static  void              pmdoc_St(MDOC_ARGS);
  91 static  void              pmdoc_Vt(MDOC_ARGS);
  92
  93 typedef void            (*pmdoc_nf)(MDOC_ARGS);
  94
  95 static  const char       *progname;
  96
  97 static  const pmdoc_nf    mdocs[MDOC_MAX] = {
  98         NULL, /* Ap */
  99         NULL, /* Dd */
 100         NULL, /* Dt */
 101         NULL, /* Os */
 102         NULL, /* Sh */
 103         NULL, /* Ss */
 104         NULL, /* Pp */
 105         NULL, /* D1 */
 106         NULL, /* Dl */
 107         NULL, /* Bd */
 108         NULL, /* Ed */
 109         NULL, /* Bl */
 110         NULL, /* El */
 111         NULL, /* It */
 112         NULL, /* Ad */
 113         pmdoc_An, /* An */
 114         NULL, /* Ar */
 115         pmdoc_Cd, /* Cd */
 116         NULL, /* Cm */
 117         NULL, /* Dv */
 118         NULL, /* Er */
 119         NULL, /* Ev */
 120         NULL, /* Ex */
 121         NULL, /* Fa */
 122         pmdoc_Fd, /* Fd */
 123         NULL, /* Fl */
 124         pmdoc_Fn, /* Fn */
 125         NULL, /* Ft */
 126         NULL, /* Ic */
 127         pmdoc_In, /* In */
 128         NULL, /* Li */
 129         pmdoc_Nd, /* Nd */
 130         pmdoc_Nm, /* Nm */
 131         NULL, /* Op */
 132         NULL, /* Ot */
 133         NULL, /* Pa */
 134         NULL, /* Rv */
 135         pmdoc_St, /* St */
 136         pmdoc_Vt, /* Va */
 137         pmdoc_Vt, /* Vt */
 138         NULL, /* Xr */
 139         NULL, /* %A */
 140         NULL, /* %B */
 141         NULL, /* %D */
 142         NULL, /* %I */
 143         NULL, /* %J */
 144         NULL, /* %N */
 145         NULL, /* %O */
 146         NULL, /* %P */
 147         NULL, /* %R */
 148         NULL, /* %T */
 149         NULL, /* %V */
 150         NULL, /* Ac */
 151         NULL, /* Ao */
 152         NULL, /* Aq */
 153         NULL, /* At */
 154         NULL, /* Bc */
 155         NULL, /* Bf */
 156         NULL, /* Bo */
 157         NULL, /* Bq */
 158         NULL, /* Bsx */
 159         NULL, /* Bx */
 160         NULL, /* Db */
 161         NULL, /* Dc */
 162         NULL, /* Do */
 163         NULL, /* Dq */
 164         NULL, /* Ec */
 165         NULL, /* Ef */
 166         NULL, /* Em */
 167         NULL, /* Eo */
 168         NULL, /* Fx */
 169         NULL, /* Ms */
 170         NULL, /* No */
 171         NULL, /* Ns */
 172         NULL, /* Nx */
 173         NULL, /* Ox */
 174         NULL, /* Pc */
 175         NULL, /* Pf */
 176         NULL, /* Po */
 177         NULL, /* Pq */
 178         NULL, /* Qc */
 179         NULL, /* Ql */
 180         NULL, /* Qo */
 181         NULL, /* Qq */
 182         NULL, /* Re */
 183         NULL, /* Rs */
 184         NULL, /* Sc */
 185         NULL, /* So */
 186         NULL, /* Sq */
 187         NULL, /* Sm */
 188         NULL, /* Sx */
 189         NULL, /* Sy */
 190         NULL, /* Tn */
 191         NULL, /* Ux */
 192         NULL, /* Xc */
 193         NULL, /* Xo */
 194         pmdoc_Fo, /* Fo */
 195         NULL, /* Fc */
 196         NULL, /* Oo */
 197         NULL, /* Oc */
 198         NULL, /* Bk */
 199         NULL, /* Ek */
 200         NULL, /* Bt */
 201         NULL, /* Hf */
 202         NULL, /* Fr */
 203         NULL, /* Ud */
 204         NULL, /* Lb */
 205         NULL, /* Lp */
 206         NULL, /* Lk */
 207         NULL, /* Mt */
 208         NULL, /* Brq */
 209         NULL, /* Bro */
 210         NULL, /* Brc */
 211         NULL, /* %C */
 212         NULL, /* Es */
 213         NULL, /* En */
 214         NULL, /* Dx */
 215         NULL, /* %Q */
 216         NULL, /* br */
 217         NULL, /* sp */
 218         NULL, /* %U */
 219         NULL, /* Ta */
 220 };
 221
 222 int
 223 main(int argc, char *argv[])
 224 {
 225         struct mparse   *mp; /* parse sequence */
 226         struct mdoc     *mdoc; /* resulting mdoc */
 227         struct man      *man; /* resulting man */
 228         char            *fn; /* current file being parsed */
 229         const char      *msec, /* manual section */
 230                         *mtitle, /* manual title */
 231                         *dir; /* result dir (default: cwd) */
 232         char             ibuf[MAXPATHLEN], /* index fname */
 233                          ibbuf[MAXPATHLEN], /* index backup fname */
 234                          fbuf[MAXPATHLEN],  /* btree fname */
 235                          fbbuf[MAXPATHLEN]; /* btree backup fname */
 236         int              ch;
 237         DB              *idx, /* index database */
 238                         *db; /* keyword database */
 239         DBT              rkey, rval, /* recno entries */
 240                          key, val; /* persistent keyword entries */
 241         size_t           sv,
 242                          ksz, rsz; /* entry buffer size */
 243         char             vbuf[8]; /* stringified record number */
 244         BTREEINFO        info; /* btree configuration */
 245         recno_t          rec; /* current record number */
 246         extern int       optind;
 247         extern char     *optarg;
 248
 249         progname = strrchr(argv[0], '/');
 250         if (progname == NULL)
 251                 progname = argv[0];
 252         else
 253                 ++progname;
 254
 255         dir = "";
 256
 257         while (-1 != (ch = getopt(argc, argv, "d:")))
 258                 switch (ch) {
 259                 case ('d'):
 260                         dir = optarg;
 261                         break;
 262                 default:
 263                         usage();
 264                         return((int)MANDOCLEVEL_BADARG);
 265                 }
 266
 267         argc -= optind;
 268         argv += optind;
 269
 270         /*
 271          * Set up temporary file-names into which we're going to write
 272          * all of our data (both for the index and database).  These
 273          * will be securely renamed to the real file-names after we've
 274          * written all of our data.
 275          */
 276
 277         ibuf[0] = ibuf[MAXPATHLEN - 2] =
 278                 ibbuf[0] = ibbuf[MAXPATHLEN - 2] =
 279                 fbuf[0] = fbuf[MAXPATHLEN - 2] =
 280                 fbbuf[0] = fbbuf[MAXPATHLEN - 2] = '\0';
 281
 282         strlcat(fbuf, dir, MAXPATHLEN);
 283         strlcat(fbuf, MANDOC_DB, MAXPATHLEN);
 284
 285         strlcat(fbbuf, fbuf, MAXPATHLEN);
 286         strlcat(fbbuf, "~", MAXPATHLEN);
 287
 288         strlcat(ibuf, dir, MAXPATHLEN);
 289         strlcat(ibuf, MANDOC_IDX, MAXPATHLEN);
 290
 291         strlcat(ibbuf, ibuf, MAXPATHLEN);
 292         strlcat(ibbuf, "~", MAXPATHLEN);
 293
 294         if ('\0' != fbuf[MAXPATHLEN - 2] ||
 295                         '\0' != fbbuf[MAXPATHLEN - 2] ||
 296                         '\0' != ibuf[MAXPATHLEN - 2] ||
 297                         '\0' != ibbuf[MAXPATHLEN - 2]) {
 298                 fprintf(stderr, "%s: Path too long\n", progname);
 299                 exit((int)MANDOCLEVEL_SYSERR);
 300         }
 301
 302         /*
 303          * For the keyword database, open a BTREE database that allows
 304          * duplicates.  For the index database, use a standard RECNO
 305          * database type.
 306          */
 307
 308         memset(&info, 0, sizeof(BTREEINFO));
 309         info.flags = R_DUP;
 310         db = dbopen(fbbuf, MANDOC_FLAGS, 0644, DB_BTREE, &info);
 311
 312         if (NULL == db) {
 313                 perror(fbbuf);
 314                 exit((int)MANDOCLEVEL_SYSERR);
 315         }
 316
 317         idx = dbopen(ibbuf, MANDOC_FLAGS, 0644, DB_RECNO, NULL);
 318
 319         if (NULL == db) {
 320                 perror(ibbuf);
 321                 (*db->close)(db);
 322                 exit((int)MANDOCLEVEL_SYSERR);
 323         }
 324
 325         /*
 326          * Try parsing the manuals given on the command line.  If we
 327          * totally fail, then just keep on going.  Take resulting trees
 328          * and push them down into the database code.
 329          * Use the auto-parser and don't report any errors.
 330          */
 331
 332         mp = mparse_alloc(MPARSE_AUTO, MANDOCLEVEL_FATAL, NULL, NULL);
 333
 334         memset(&key, 0, sizeof(DBT));
 335         memset(&val, 0, sizeof(DBT));
 336         memset(&rkey, 0, sizeof(DBT));
 337         memset(&rval, 0, sizeof(DBT));
 338
 339         val.size = sizeof(vbuf);
 340         val.data = vbuf;
 341         rkey.size = sizeof(recno_t);
 342
 343         rec = 1;
 344         ksz = rsz = 0;
 345
 346         while (NULL != (fn = *argv++)) {
 347                 mparse_reset(mp);
 348
 349                 /* Parse and get (non-empty) AST. */
 350
 351                 if (mparse_readfd(mp, -1, fn) >= MANDOCLEVEL_FATAL) {
 352                         fprintf(stderr, "%s: Parse failure\n", fn);
 353                         continue;
 354                 }
 355                 mparse_result(mp, &mdoc, &man);
 356                 if (NULL == mdoc && NULL == man)
 357                         continue;
 358
 359                 /* Manual section: can be empty string. */
 360
 361                 msec = NULL != mdoc ?
 362                         mdoc_meta(mdoc)->msec :
 363                         man_meta(man)->msec;
 364                 mtitle = NULL != mdoc ?
 365                         mdoc_meta(mdoc)->title :
 366                         man_meta(man)->title;
 367
 368                 assert(msec);
 369                 assert(mtitle);
 370
 371                 /*
 372                  * The index record value consists of a nil-terminated
 373                  * filename, a nil-terminated manual section, and a
 374                  * nil-terminated description.  Since the description
 375                  * may not be set, we set a sentinel to see if we're
 376                  * going to write a nil byte in its place.
 377                  */
 378
 379                 dbt_init(&rval, &rsz);
 380                 dbt_appendb(&rval, &rsz, fn, strlen(fn) + 1);
 381                 dbt_appendb(&rval, &rsz, msec, strlen(msec) + 1);
 382                 dbt_appendb(&rval, &rsz, mtitle, strlen(mtitle) + 1);
 383                 sv = rval.size;
 384
 385                 /* Fix the record number in the btree value. */
 386
 387                 memset(val.data, 0, sizeof(uint32_t));
 388                 memcpy(val.data + 4, &rec, sizeof(uint32_t));
 389
 390                 if (mdoc)
 391                         pmdoc(db, fbbuf, &key, &ksz,
 392                                 &val, &rval, &rsz, mdoc);
 393                 else
 394                         pman(db, fbbuf, &key, &ksz,
 395                                 &val, &rval, &rsz, man);
 396
 397                 /*
 398                  * Apply this to the index.  If we haven't had a
 399                  * description set, put an empty one in now.
 400                  */
 401
 402                 if (rval.size == sv)
 403                         dbt_appendb(&rval, &rsz, "", 1);
 404
 405                 rkey.data = &rec;
 406                 dbt_put(idx, ibbuf, &rkey, &rval);
 407
 408                 printf("Indexed: %s\n", fn);
 409                 rec++;
 410         }
 411
 412         (*db->close)(db);
 413         (*idx->close)(idx);
 414
 415         mparse_free(mp);
 416
 417         free(key.data);
 418         free(rval.data);
 419
 420         /* Atomically replace the file with our temporary one. */
 421
 422         if (-1 == rename(fbbuf, fbuf))
 423                 perror(fbuf);
 424         if (-1 == rename(ibbuf, ibuf))
 425                 perror(fbuf);
 426
 427         return((int)MANDOCLEVEL_OK);
 428 }
 429
 430 /*
 431  * Initialise the stored database key whose data buffer is shared
 432  * between uses (as the key must sometimes be constructed from an array
 433  * of
 434  */
 435 static void
 436 dbt_init(DBT *key, size_t *ksz)
 437 {
 438
 439         if (0 == *ksz) {
 440                 assert(0 == key->size);
 441                 assert(NULL == key->data);
 442                 key->data = mandoc_malloc(MANDOC_BUFSZ);
 443                 *ksz = MANDOC_BUFSZ;
 444         }
 445
 446         key->size = 0;
 447 }
 448
 449 /*
 450  * Append a binary value to a database entry.  This can be invoked
 451  * multiple times; the buffer is automatically resized.
 452  */
 453 static void
 454 dbt_appendb(DBT *key, size_t *ksz, const void *cp, size_t sz)
 455 {
 456
 457         assert(key->data);
 458
 459         /* Overshoot by MANDOC_BUFSZ. */
 460
 461         while (key->size + sz >= *ksz) {
 462                 *ksz = key->size + sz + MANDOC_BUFSZ;
 463                 key->data = mandoc_realloc(key->data, *ksz);
 464         }
 465
 466 #if 0
 467         dstp = key->data + (int)key->size;
 468
 469         while (NULL != (endp = memchr(cp, '\\', sz))) {
 470                 ssz = endp - cp;
 471                 memcpy(dstp, cp, ssz);
 472
 473                 dstp += ssz;
 474                 key->size += ssz;
 475                 sz -= ssz;
 476
 477                 cp = endp++;
 478                 /* FIXME: expects nil-terminated string! */
 479                 esc = mandoc_escape((const char **)&endp, NULL, NULL);
 480
 481                 switch (esc) {
 482                 case (ESCAPE_ERROR):
 483                         /* Nil-terminate this point. */
 484                         memcpy(dstp, "", 1);
 485                         key->size++;
 486                         return;
 487                 case (ESCAPE_PREDEF):
 488                         /* FALLTHROUGH */
 489                 case (ESCAPE_SPECIAL):
 490                         break;
 491                 default:
 492                         sz -= endp - cp;
 493                         cp = endp;
 494                         continue;
 495                 }
 496
 497                 ssz = endp - cp;
 498                 memcpy(dstp, cp, ssz);
 499
 500                 dstp += ssz;
 501                 key->size += ssz;
 502                 sz -= ssz;
 503
 504                 cp = endp;
 505         }
 506 #endif
 507
 508         memcpy(key->data + (int)key->size, cp, sz);
 509         key->size += sz;
 510 }
 511
 512 /*
 513  * Append a nil-terminated string to the database entry.  This can be
 514  * invoked multiple times.  The database entry will be nil-terminated as
 515  * well; if invoked multiple times, a space is put between strings.
 516  */
 517 static void
 518 dbt_append(DBT *key, size_t *ksz, const char *cp)
 519 {
 520         size_t           sz;
 521
 522         if (0 == (sz = strlen(cp)))
 523                 return;
 524
 525         assert(key->data);
 526
 527         if (key->size)
 528                 ((char *)key->data)[(int)key->size - 1] = ' ';
 529
 530         dbt_appendb(key, ksz, cp, sz + 1);
 531 }
 532
 533 /* ARGSUSED */
 534 static void
 535 pmdoc_An(MDOC_ARGS)
 536 {
 537         uint32_t         fl;
 538
 539         if (SEC_AUTHORS != n->sec)
 540                 return;
 541
 542         for (n = n->child; n; n = n->next)
 543                 if (MDOC_TEXT == n->type)
 544                         dbt_append(key, ksz, n->string);
 545
 546         fl = MANDOC_AUTHOR;
 547         memcpy(val->data, &fl, 4);
 548 }
 549
 550 /* ARGSUSED */
 551 static void
 552 pmdoc_Fd(MDOC_ARGS)
 553 {
 554         uint32_t         fl;
 555         const char      *start, *end;
 556         size_t           sz;
 557
 558         if (SEC_SYNOPSIS != n->sec)
 559                 return;
 560         if (NULL == (n = n->child) || MDOC_TEXT != n->type)
 561                 return;
 562
 563         /*
 564          * Only consider those `Fd' macro fields that begin with an
 565          * "inclusion" token (versus, e.g., #define).
 566          */
 567         if (strcmp("#include", n->string))
 568                 return;
 569
 570         if (NULL == (n = n->next) || MDOC_TEXT != n->type)
 571                 return;
 572
 573         /*
 574          * Strip away the enclosing angle brackets and make sure we're
 575          * not zero-length.
 576          */
 577
 578         start = n->string;
 579         if ('<' == *start || '"' == *start)
 580                 start++;
 581
 582         if (0 == (sz = strlen(start)))
 583                 return;
 584
 585         end = &start[(int)sz - 1];
 586         if ('>' == *end || '"' == *end)
 587                 end--;
 588
 589         dbt_appendb(key, ksz, start, end - start + 1);
 590         dbt_appendb(key, ksz, "", 1);
 591
 592         fl = MANDOC_INCLUDES;
 593         memcpy(val->data, &fl, 4);
 594 }
 595
 596 /* ARGSUSED */
 597 static void
 598 pmdoc_Cd(MDOC_ARGS)
 599 {
 600         uint32_t         fl;
 601
 602         if (SEC_SYNOPSIS != n->sec)
 603                 return;
 604
 605         for (n = n->child; n; n = n->next)
 606                 if (MDOC_TEXT == n->type)
 607                         dbt_append(key, ksz, n->string);
 608
 609         fl = MANDOC_CONFIG;
 610         memcpy(val->data, &fl, 4);
 611 }
 612
 613 /* ARGSUSED */
 614 static void
 615 pmdoc_In(MDOC_ARGS)
 616 {
 617         uint32_t         fl;
 618
 619         if (SEC_SYNOPSIS != n->sec)
 620                 return;
 621         if (NULL == n->child || MDOC_TEXT != n->child->type)
 622                 return;
 623
 624         dbt_append(key, ksz, n->child->string);
 625         fl = MANDOC_INCLUDES;
 626         memcpy(val->data, &fl, 4);
 627 }
 628
 629 /* ARGSUSED */
 630 static void
 631 pmdoc_Fn(MDOC_ARGS)
 632 {
 633         uint32_t         fl;
 634         const char      *cp;
 635
 636         if (SEC_SYNOPSIS != n->sec)
 637                 return;
 638         if (NULL == n->child || MDOC_TEXT != n->child->type)
 639                 return;
 640
 641         /* .Fn "struct type *arg" "foo" */
 642
 643         cp = strrchr(n->child->string, ' ');
 644         if (NULL == cp)
 645                 cp = n->child->string;
 646
 647         /* Strip away pointer symbol. */
 648
 649         while ('*' == *cp)
 650                 cp++;
 651
 652         dbt_append(key, ksz, cp);
 653         fl = MANDOC_FUNCTION;
 654         memcpy(val->data, &fl, 4);
 655 }
 656
 657 /* ARGSUSED */
 658 static void
 659 pmdoc_St(MDOC_ARGS)
 660 {
 661         uint32_t         fl;
 662
 663         if (SEC_STANDARDS != n->sec)
 664                 return;
 665         if (NULL == n->child || MDOC_TEXT != n->child->type)
 666                 return;
 667
 668         dbt_append(key, ksz, n->child->string);
 669         fl = MANDOC_STANDARD;
 670         memcpy(val->data, &fl, 4);
 671 }
 672
 673 /* ARGSUSED */
 674 static void
 675 pmdoc_Vt(MDOC_ARGS)
 676 {
 677         uint32_t         fl;
 678         const char      *start;
 679         size_t           sz;
 680
 681         if (SEC_SYNOPSIS != n->sec)
 682                 return;
 683         if (MDOC_Vt == n->tok && MDOC_BODY != n->type)
 684                 return;
 685         if (NULL == n->last || MDOC_TEXT != n->last->type)
 686                 return;
 687
 688         /*
 689          * Strip away leading pointer symbol '*' and trailing ';'.
 690          */
 691
 692         start = n->last->string;
 693
 694         while ('*' == *start)
 695                 start++;
 696
 697         if (0 == (sz = strlen(start)))
 698                 return;
 699
 700         if (';' == start[sz - 1])
 701                 sz--;
 702
 703         if (0 == sz)
 704                 return;
 705
 706         dbt_appendb(key, ksz, start, sz);
 707         dbt_appendb(key, ksz, "", 1);
 708
 709         fl = MANDOC_VARIABLE;
 710         memcpy(val->data, &fl, 4);
 711 }
 712
 713 /* ARGSUSED */
 714 static void
 715 pmdoc_Fo(MDOC_ARGS)
 716 {
 717         uint32_t         fl;
 718
 719         if (SEC_SYNOPSIS != n->sec || MDOC_HEAD != n->type)
 720                 return;
 721         if (NULL == n->child || MDOC_TEXT != n->child->type)
 722                 return;
 723
 724         dbt_append(key, ksz, n->child->string);
 725         fl = MANDOC_FUNCTION;
 726         memcpy(val->data, &fl, 4);
 727 }
 728
 729
 730 /* ARGSUSED */
 731 static void
 732 pmdoc_Nd(MDOC_ARGS)
 733 {
 734         int              first;
 735
 736         for (first = 1, n = n->child; n; n = n->next) {
 737                 if (MDOC_TEXT != n->type)
 738                         continue;
 739                 if (first)
 740                         dbt_appendb(rval, rsz, n->string, strlen(n->string) + 1);
 741                 else
 742                         dbt_append(rval, rsz, n->string);
 743                 first = 0;
 744         }
 745 }
 746
 747 /* ARGSUSED */
 748 static void
 749 pmdoc_Nm(MDOC_ARGS)
 750 {
 751         uint32_t         fl;
 752
 753         if (SEC_NAME == n->sec) {
 754                 for (n = n->child; n; n = n->next) {
 755                         if (MDOC_TEXT != n->type)
 756                                 continue;
 757                         dbt_append(key, ksz, n->string);
 758                 }
 759                 fl = MANDOC_NAME;
 760                 memcpy(val->data, &fl, 4);
 761                 return;
 762         } else if (SEC_SYNOPSIS != n->sec || MDOC_HEAD != n->type)
 763                 return;
 764
 765         for (n = n->child; n; n = n->next) {
 766                 if (MDOC_TEXT != n->type)
 767                         continue;
 768                 dbt_append(key, ksz, n->string);
 769         }
 770
 771         fl = MANDOC_UTILITY;
 772         memcpy(val->data, &fl, 4);
 773 }
 774
 775 static void
 776 dbt_put(DB *db, const char *dbn, DBT *key, DBT *val)
 777 {
 778
 779         if (0 == key->size)
 780                 return;
 781
 782         assert(key->data);
 783         assert(val->size);
 784         assert(val->data);
 785
 786         if (0 == (*db->put)(db, key, val, 0))
 787                 return;
 788
 789         perror(dbn);
 790         exit((int)MANDOCLEVEL_SYSERR);
 791         /* NOTREACHED */
 792 }
 793
 794 /*
 795  * Call out to per-macro handlers after clearing the persistent database
 796  * key.  If the macro sets the database key, flush it to the database.
 797  */
 798 static void
 799 pmdoc_node(MDOC_ARGS)
 800 {
 801
 802         if (NULL == n)
 803                 return;
 804
 805         switch (n->type) {
 806         case (MDOC_HEAD):
 807                 /* FALLTHROUGH */
 808         case (MDOC_BODY):
 809                 /* FALLTHROUGH */
 810         case (MDOC_TAIL):
 811                 /* FALLTHROUGH */
 812         case (MDOC_BLOCK):
 813                 /* FALLTHROUGH */
 814         case (MDOC_ELEM):
 815                 if (NULL == mdocs[n->tok])
 816                         break;
 817
 818                 dbt_init(key, ksz);
 819
 820                 (*mdocs[n->tok])(db, dbn, key, ksz, val, rval, rsz, n);
 821                 dbt_put(db, dbn, key, val);
 822                 break;
 823         default:
 824                 break;
 825         }
 826
 827         pmdoc_node(db, dbn, key, ksz, val, rval, rsz, n->child);
 828         pmdoc_node(db, dbn, key, ksz, val, rval, rsz, n->next);
 829 }
 830
 831 static int
 832 pman_node(MAN_ARGS)
 833 {
 834         const struct man_node *head, *body;
 835         const char      *start, *sv;
 836         size_t           sz;
 837         uint32_t         fl;
 838
 839         if (NULL == n)
 840                 return(0);
 841
 842         /*
 843          * We're only searching for one thing: the first text child in
 844          * the BODY of a NAME section.  Since we don't keep track of
 845          * sections in -man, run some hoops to find out whether we're in
 846          * the correct section or not.
 847          */
 848
 849         if (MAN_BODY == n->type && MAN_SH == n->tok) {
 850                 body = n;
 851                 assert(body->parent);
 852                 if (NULL != (head = body->parent->head) &&
 853                                 1 == head->nchild &&
 854                                 NULL != (head = (head->child)) &&
 855                                 MAN_TEXT == head->type &&
 856                                 0 == strcmp(head->string, "NAME") &&
 857                                 NULL != (body = body->child) &&
 858                                 MAN_TEXT == body->type) {
 859
 860                         fl = MANDOC_NAME;
 861                         memcpy(val->data, &fl, 4);
 862
 863                         assert(body->string);
 864                         start = sv = body->string;
 865
 866                         /*
 867                          * Go through a special heuristic dance here.
 868                          * This is why -man manuals are great!
 869                          * Conventionally, one or more manual names are
 870                          * comma-specified prior to a whitespace, then a
 871                          * dash, then a description.  Try to puzzle out
 872                          * the name parts here.
 873                          */
 874
 875                         for ( ;; ) {
 876                                 sz = strcspn(start, " ,");
 877                                 if ('\0' == start[(int)sz])
 878                                         break;
 879
 880                                 dbt_init(key, ksz);
 881                                 dbt_appendb(key, ksz, start, sz);
 882                                 dbt_appendb(key, ksz, "", 1);
 883
 884                                 dbt_put(db, dbn, key, val);
 885
 886                                 if (' ' == start[(int)sz]) {
 887                                         start += (int)sz + 1;
 888                                         break;
 889                                 }
 890
 891                                 assert(',' == start[(int)sz]);
 892                                 start += (int)sz + 1;
 893                                 while (' ' == *start)
 894                                         start++;
 895                         }
 896
 897                         if (sv == start) {
 898                                 dbt_init(key, ksz);
 899                                 dbt_append(key, ksz, start);
 900                                 return(1);
 901                         }
 902
 903                         while (' ' == *start)
 904                                 start++;
 905
 906                         if ('\\' == *start && '-' == *(start + 1))
 907                                 start += 2;
 908                         else if ('-' == *start)
 909                                 start++;
 910
 911                         while (' ' == *start)
 912                                 start++;
 913
 914                         dbt_appendb(rval, rsz, start, strlen(start) + 1);
 915                 }
 916         }
 917
 918         if (pman_node(db, dbn, key, ksz, val, rval, rsz, n->child))
 919                 return(1);
 920         if (pman_node(db, dbn, key, ksz, val, rval, rsz, n->next))
 921                 return(1);
 922
 923         return(0);
 924 }
 925
 926 static void
 927 pman(DB *db, const char *dbn, DBT *key, size_t *ksz,
 928                 DBT *val, DBT *rval, size_t *rsz, struct man *m)
 929 {
 930
 931         pman_node(db, dbn, key, ksz, val, rval, rsz, man_node(m));
 932 }
 933
 934
 935 static void
 936 pmdoc(DB *db, const char *dbn, DBT *key, size_t *ksz,
 937                 DBT *val, DBT *rval, size_t *rsz, struct mdoc *m)
 938 {
 939
 940         pmdoc_node(db, dbn, key, ksz, val, rval, rsz, mdoc_node(m));
 941 }
 942
 943 static void
 944 usage(void)
 945 {
 946
 947         fprintf(stderr, "usage: %s "
 948                         "[-d path] "
 949                         "[file...]\n",
 950                         progname);
 951 }