mandoc-db.c

   1 /*      $Id: mandoc-db.c,v 1.3 2011/04/03 14:18:29 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 };
  53
  54 #define MDOC_ARGS         DB *db, \
  55                           const char *dbn, \
  56                           DBT *key, size_t *ksz, \
  57                           DBT *val, \
  58                           const struct mdoc_node *n
  59
  60 static  void              dbt_append(DBT *, size_t *, const char *);
  61 static  void              dbt_appendb(DBT *, size_t *,
  62                                 const void *, size_t);
  63 static  void              dbt_init(DBT *, size_t *);
  64 static  void              version(void);
  65 static  void              usage(void);
  66 static  void              pmdoc(DB *, const char *,
  67                                 DBT *, size_t *, DBT *,
  68                                 const char *, struct mdoc *);
  69 static  void              pmdoc_node(MDOC_ARGS);
  70 static  void              pmdoc_Fd(MDOC_ARGS);
  71 static  void              pmdoc_In(MDOC_ARGS);
  72 static  void              pmdoc_Fn(MDOC_ARGS);
  73 static  void              pmdoc_Fo(MDOC_ARGS);
  74 static  void              pmdoc_Nm(MDOC_ARGS);
  75 static  void              pmdoc_Vt(MDOC_ARGS);
  76
  77 typedef void            (*pmdoc_nf)(MDOC_ARGS);
  78
  79 static  const char       *progname;
  80
  81 static  const pmdoc_nf    mdocs[MDOC_MAX] = {
  82         NULL, /* Ap */
  83         NULL, /* Dd */
  84         NULL, /* Dt */
  85         NULL, /* Os */
  86         NULL, /* Sh */
  87         NULL, /* Ss */
  88         NULL, /* Pp */
  89         NULL, /* D1 */
  90         NULL, /* Dl */
  91         NULL, /* Bd */
  92         NULL, /* Ed */
  93         NULL, /* Bl */
  94         NULL, /* El */
  95         NULL, /* It */
  96         NULL, /* Ad */
  97         NULL, /* An */
  98         NULL, /* Ar */
  99         NULL, /* Cd */
 100         NULL, /* Cm */
 101         NULL, /* Dv */
 102         NULL, /* Er */
 103         NULL, /* Ev */
 104         NULL, /* Ex */
 105         NULL, /* Fa */
 106         pmdoc_Fd, /* Fd */
 107         NULL, /* Fl */
 108         pmdoc_Fn, /* Fn */
 109         NULL, /* Ft */
 110         NULL, /* Ic */
 111         pmdoc_In, /* In */
 112         NULL, /* Li */
 113         NULL, /* Nd */
 114         pmdoc_Nm, /* Nm */
 115         NULL, /* Op */
 116         NULL, /* Ot */
 117         NULL, /* Pa */
 118         NULL, /* Rv */
 119         NULL, /* St */
 120         pmdoc_Vt, /* Va */
 121         pmdoc_Vt, /* Vt */
 122         NULL, /* Xr */
 123         NULL, /* %A */
 124         NULL, /* %B */
 125         NULL, /* %D */
 126         NULL, /* %I */
 127         NULL, /* %J */
 128         NULL, /* %N */
 129         NULL, /* %O */
 130         NULL, /* %P */
 131         NULL, /* %R */
 132         NULL, /* %T */
 133         NULL, /* %V */
 134         NULL, /* Ac */
 135         NULL, /* Ao */
 136         NULL, /* Aq */
 137         NULL, /* At */
 138         NULL, /* Bc */
 139         NULL, /* Bf */
 140         NULL, /* Bo */
 141         NULL, /* Bq */
 142         NULL, /* Bsx */
 143         NULL, /* Bx */
 144         NULL, /* Db */
 145         NULL, /* Dc */
 146         NULL, /* Do */
 147         NULL, /* Dq */
 148         NULL, /* Ec */
 149         NULL, /* Ef */
 150         NULL, /* Em */
 151         NULL, /* Eo */
 152         NULL, /* Fx */
 153         NULL, /* Ms */
 154         NULL, /* No */
 155         NULL, /* Ns */
 156         NULL, /* Nx */
 157         NULL, /* Ox */
 158         NULL, /* Pc */
 159         NULL, /* Pf */
 160         NULL, /* Po */
 161         NULL, /* Pq */
 162         NULL, /* Qc */
 163         NULL, /* Ql */
 164         NULL, /* Qo */
 165         NULL, /* Qq */
 166         NULL, /* Re */
 167         NULL, /* Rs */
 168         NULL, /* Sc */
 169         NULL, /* So */
 170         NULL, /* Sq */
 171         NULL, /* Sm */
 172         NULL, /* Sx */
 173         NULL, /* Sy */
 174         NULL, /* Tn */
 175         NULL, /* Ux */
 176         NULL, /* Xc */
 177         NULL, /* Xo */
 178         pmdoc_Fo, /* Fo */
 179         NULL, /* Fc */
 180         NULL, /* Oo */
 181         NULL, /* Oc */
 182         NULL, /* Bk */
 183         NULL, /* Ek */
 184         NULL, /* Bt */
 185         NULL, /* Hf */
 186         NULL, /* Fr */
 187         NULL, /* Ud */
 188         NULL, /* Lb */
 189         NULL, /* Lp */
 190         NULL, /* Lk */
 191         NULL, /* Mt */
 192         NULL, /* Brq */
 193         NULL, /* Bro */
 194         NULL, /* Brc */
 195         NULL, /* %C */
 196         NULL, /* Es */
 197         NULL, /* En */
 198         NULL, /* Dx */
 199         NULL, /* %Q */
 200         NULL, /* br */
 201         NULL, /* sp */
 202         NULL, /* %U */
 203         NULL, /* Ta */
 204 };
 205
 206 int
 207 main(int argc, char *argv[])
 208 {
 209         struct mparse   *mp; /* parse sequence */
 210         struct mdoc     *mdoc; /* resulting mdoc */
 211         char            *fn;
 212         const char      *dir; /* result dir (default: cwd) */
 213         char             ibuf[MAXPATHLEN], /* index fname */
 214                          ibbuf[MAXPATHLEN], /* index backup fname */
 215                          fbuf[MAXPATHLEN],  /* btree fname */
 216                          fbbuf[MAXPATHLEN]; /* btree backup fname */
 217         int              c;
 218         DB              *index, /* index database */
 219                         *db; /* keyword database */
 220         DBT              rkey, rval, /* recno entries */
 221                          key, val; /* persistent keyword entries */
 222         size_t           ksz; /* entry buffer size */
 223         char             vbuf[8];
 224         BTREEINFO        info; /* btree configuration */
 225         recno_t          rec;
 226         extern int       optind;
 227         extern char     *optarg;
 228
 229         progname = strrchr(argv[0], '/');
 230         if (progname == NULL)
 231                 progname = argv[0];
 232         else
 233                 ++progname;
 234
 235         dir = "";
 236
 237         while (-1 != (c = getopt(argc, argv, "d:V")))
 238                 switch (c) {
 239                 case ('d'):
 240                         dir = optarg;
 241                         break;
 242                 case ('V'):
 243                         version();
 244                         return((int)MANDOCLEVEL_OK);
 245                 default:
 246                         usage();
 247                         return((int)MANDOCLEVEL_BADARG);
 248                 }
 249
 250         argc -= optind;
 251         argv += optind;
 252
 253         /*
 254          * Set up temporary file-names into which we're going to write
 255          * all of our data (both for the index and database).  These
 256          * will be securely renamed to the real file-names after we've
 257          * written all of our data.
 258          */
 259
 260         ibuf[0] = ibuf[MAXPATHLEN - 2] =
 261                 ibbuf[0] = ibbuf[MAXPATHLEN - 2] =
 262                 fbuf[0] = fbuf[MAXPATHLEN - 2] =
 263                 fbbuf[0] = fbbuf[MAXPATHLEN - 2] = '\0';
 264
 265         strlcat(fbuf, dir, MAXPATHLEN);
 266         strlcat(fbuf, MANDOC_DB, MAXPATHLEN);
 267
 268         strlcat(fbbuf, fbuf, MAXPATHLEN);
 269         strlcat(fbbuf, "~", MAXPATHLEN);
 270
 271         strlcat(ibuf, dir, MAXPATHLEN);
 272         strlcat(ibuf, MANDOC_IDX, MAXPATHLEN);
 273
 274         strlcat(ibbuf, ibuf, MAXPATHLEN);
 275         strlcat(ibbuf, "~", MAXPATHLEN);
 276
 277         if ('\0' != fbuf[MAXPATHLEN - 2] ||
 278                         '\0' != fbbuf[MAXPATHLEN - 2] ||
 279                         '\0' != ibuf[MAXPATHLEN - 2] ||
 280                         '\0' != ibbuf[MAXPATHLEN - 2]) {
 281                 fprintf(stderr, "%s: Path too long\n", progname);
 282                 exit((int)MANDOCLEVEL_SYSERR);
 283         }
 284
 285         /*
 286          * For the keyword database, open a BTREE database that allows
 287          * duplicates.  For the index database, use a standard RECNO
 288          * database type.
 289          */
 290
 291         memset(&info, 0, sizeof(BTREEINFO));
 292         info.flags = R_DUP;
 293         db = dbopen(fbbuf, MANDOC_FLAGS, 0644, DB_BTREE, &info);
 294
 295         if (NULL == db) {
 296                 perror(fbbuf);
 297                 exit((int)MANDOCLEVEL_SYSERR);
 298         }
 299
 300         index = dbopen(ibbuf, MANDOC_FLAGS, 0644, DB_RECNO, NULL);
 301
 302         if (NULL == db) {
 303                 perror(ibbuf);
 304                 (*db->close)(db);
 305                 exit((int)MANDOCLEVEL_SYSERR);
 306         }
 307
 308         /*
 309          * Try parsing the manuals given on the command line.  If we
 310          * totally fail, then just keep on going.  Take resulting trees
 311          * and push them down into the database code.
 312          * Use the auto-parser and don't report any errors.
 313          */
 314
 315         mp = mparse_alloc(MPARSE_AUTO, MANDOCLEVEL_FATAL, NULL, NULL);
 316
 317         memset(&key, 0, sizeof(DBT));
 318         memset(&val, 0, sizeof(DBT));
 319         memset(&rkey, 0, sizeof(DBT));
 320         memset(&rval, 0, sizeof(DBT));
 321
 322         val.size = sizeof(vbuf);
 323         val.data = vbuf;
 324         rkey.size = sizeof(recno_t);
 325
 326         rec = 1;
 327         ksz = 0;
 328
 329         while (NULL != (fn = *argv++)) {
 330                 mparse_reset(mp);
 331
 332                 if (mparse_readfd(mp, -1, fn) >= MANDOCLEVEL_FATAL)
 333                         continue;
 334
 335                 mparse_result(mp, &mdoc, NULL);
 336                 if (NULL == mdoc)
 337                         continue;
 338
 339                 rkey.data = &rec;
 340                 rval.data = fn;
 341                 rval.size = strlen(fn) + 1;
 342
 343                 if (-1 == (*index->put)(index, &rkey, &rval, 0)) {
 344                         perror(ibbuf);
 345                         break;
 346                 }
 347
 348                 memset(val.data, 0, sizeof(uint32_t));
 349                 memcpy(val.data + 4, &rec, sizeof(uint32_t));
 350
 351                 pmdoc(db, fbbuf, &key, &ksz, &val, fn, mdoc);
 352                 rec++;
 353         }
 354
 355         (*db->close)(db);
 356         (*index->close)(index);
 357
 358         mparse_free(mp);
 359
 360         free(key.data);
 361
 362         /* Atomically replace the file with our temporary one. */
 363
 364         if (-1 == rename(fbbuf, fbuf))
 365                 perror(fbuf);
 366         if (-1 == rename(ibbuf, ibuf))
 367                 perror(fbuf);
 368
 369         return((int)MANDOCLEVEL_OK);
 370 }
 371
 372 /*
 373  * Initialise the stored database key whose data buffer is shared
 374  * between uses (as the key must sometimes be constructed from an array
 375  * of
 376  */
 377 static void
 378 dbt_init(DBT *key, size_t *ksz)
 379 {
 380
 381         if (0 == *ksz) {
 382                 assert(0 == key->size);
 383                 assert(NULL == key->data);
 384                 key->data = mandoc_malloc(MANDOC_BUFSZ);
 385                 *ksz = MANDOC_BUFSZ;
 386         }
 387
 388         key->size = 0;
 389 }
 390
 391 /*
 392  * Append a binary value to a database entry.  This can be invoked
 393  * multiple times; the buffer is automatically resized.
 394  */
 395 static void
 396 dbt_appendb(DBT *key, size_t *ksz, const void *cp, size_t sz)
 397 {
 398
 399         assert(key->data);
 400
 401         /* Overshoot by MANDOC_BUFSZ. */
 402
 403         while (key->size + sz >= *ksz) {
 404                 *ksz = key->size + sz + MANDOC_BUFSZ;
 405                 key->data = mandoc_realloc(key->data, *ksz);
 406         }
 407
 408         memcpy(key->data + (int)key->size, cp, sz);
 409         key->size += sz;
 410 }
 411
 412 /*
 413  * Append a nil-terminated string to the database entry.  This can be
 414  * invoked multiple times.  The database entry will be nil-terminated as
 415  * well; if invoked multiple times, a space is put between strings.
 416  */
 417 static void
 418 dbt_append(DBT *key, size_t *ksz, const char *cp)
 419 {
 420         size_t           sz;
 421
 422         if (0 == (sz = strlen(cp)))
 423                 return;
 424
 425         assert(key->data);
 426
 427         if (key->size)
 428                 ((char *)key->data)[(int)key->size - 1] = ' ';
 429
 430         dbt_appendb(key, ksz, cp, sz + 1);
 431 }
 432
 433 /* ARGSUSED */
 434 static void
 435 pmdoc_Fd(MDOC_ARGS)
 436 {
 437         uint32_t         fl;
 438         const char      *start, *end;
 439         size_t           sz;
 440         char             nil;
 441
 442         if (SEC_SYNOPSIS != n->sec)
 443                 return;
 444         if (NULL == (n = n->child) || MDOC_TEXT != n->type)
 445                 return;
 446         if (strcmp("#include", n->string))
 447                 return;
 448         if (NULL == (n = n->next) || MDOC_TEXT != n->type)
 449                 return;
 450
 451         start = n->string;
 452         if ('<' == *start)
 453                 start++;
 454
 455         if (0 == (sz = strlen(start)))
 456                 return;
 457
 458         end = &start[(int)sz - 1];
 459         if ('>' == *end)
 460                 end--;
 461
 462         nil = '\0';
 463         dbt_appendb(key, ksz, start, end - start + 1);
 464         dbt_appendb(key, ksz, &nil, 1);
 465
 466         fl = MANDOC_INCLUDES;
 467         memcpy(val->data, &fl, 4);
 468 }
 469
 470 /* ARGSUSED */
 471 static void
 472 pmdoc_In(MDOC_ARGS)
 473 {
 474         uint32_t         fl;
 475
 476         if (SEC_SYNOPSIS != n->sec)
 477                 return;
 478         if (NULL == n->child || MDOC_TEXT != n->child->type)
 479                 return;
 480
 481         dbt_append(key, ksz, n->child->string);
 482         fl = MANDOC_INCLUDES;
 483         memcpy(val->data, &fl, 4);
 484 }
 485
 486 /* ARGSUSED */
 487 static void
 488 pmdoc_Fn(MDOC_ARGS)
 489 {
 490         uint32_t         fl;
 491         const char      *cp;
 492
 493         if (SEC_SYNOPSIS != n->sec)
 494                 return;
 495         if (NULL == n->child || MDOC_TEXT != n->child->type)
 496                 return;
 497
 498         /* .Fn "struct type *arg" "foo" */
 499
 500         cp = strrchr(n->child->string, ' ');
 501         if (NULL == cp)
 502                 cp = n->child->string;
 503
 504         /* Ignore pointers. */
 505
 506         while ('*' == *cp)
 507                 cp++;
 508
 509         dbt_append(key, ksz, cp);
 510         fl = MANDOC_FUNCTION;
 511         memcpy(val->data, &fl, 4);
 512 }
 513
 514 /* ARGSUSED */
 515 static void
 516 pmdoc_Vt(MDOC_ARGS)
 517 {
 518         uint32_t         fl;
 519         const char      *start, *end;
 520         size_t           sz;
 521         char             nil;
 522
 523         if (SEC_SYNOPSIS != n->sec)
 524                 return;
 525         if (MDOC_Vt == n->tok && MDOC_BODY != n->type)
 526                 return;
 527         if (NULL == n->child || MDOC_TEXT != n->child->type)
 528                 return;
 529
 530         /*
 531          * Strip away leading '*' and trailing ';'.
 532          */
 533
 534         start = n->last->string;
 535
 536         while ('*' == *start)
 537                 start++;
 538
 539         if (0 == (sz = strlen(start)))
 540                 return;
 541
 542         end = &start[sz - 1];
 543         while (end > start && ';' == *end)
 544                 end--;
 545
 546         if (end == start)
 547                 return;
 548
 549         nil = '\0';
 550         dbt_appendb(key, ksz, start, end - start + 1);
 551         dbt_appendb(key, ksz, &nil, 1);
 552         fl = MANDOC_VARIABLE;
 553         memcpy(val->data, &fl, 4);
 554 }
 555
 556 /* ARGSUSED */
 557 static void
 558 pmdoc_Fo(MDOC_ARGS)
 559 {
 560         uint32_t         fl;
 561
 562         if (SEC_SYNOPSIS != n->sec || MDOC_HEAD != n->type)
 563                 return;
 564         if (NULL == n->child || MDOC_TEXT != n->child->type)
 565                 return;
 566
 567         dbt_append(key, ksz, n->child->string);
 568         fl = MANDOC_FUNCTION;
 569         memcpy(val->data, &fl, 4);
 570 }
 571
 572 /* ARGSUSED */
 573 static void
 574 pmdoc_Nm(MDOC_ARGS)
 575 {
 576         uint32_t         fl;
 577
 578         if (SEC_NAME == n->sec) {
 579                 for (n = n->child; n; n = n->next) {
 580                         if (MDOC_TEXT != n->type)
 581                                 continue;
 582                         dbt_append(key, ksz, n->string);
 583                 }
 584                 fl = MANDOC_NAME;
 585                 memcpy(val->data, &fl, 4);
 586                 return;
 587         } else if (SEC_SYNOPSIS != n->sec || MDOC_HEAD != n->type)
 588                 return;
 589
 590         for (n = n->child; n; n = n->next) {
 591                 if (MDOC_TEXT != n->type)
 592                         continue;
 593                 dbt_append(key, ksz, n->string);
 594         }
 595
 596         fl = MANDOC_UTILITY;
 597         memcpy(val->data, &fl, 4);
 598 }
 599
 600 /*
 601  * Call out to per-macro handlers after clearing the persistent database
 602  * key.  If the macro sets the database key, flush it to the database.
 603  */
 604 static void
 605 pmdoc_node(MDOC_ARGS)
 606 {
 607
 608         if (NULL == n)
 609                 return;
 610
 611         switch (n->type) {
 612         case (MDOC_HEAD):
 613                 /* FALLTHROUGH */
 614         case (MDOC_BODY):
 615                 /* FALLTHROUGH */
 616         case (MDOC_TAIL):
 617                 /* FALLTHROUGH */
 618         case (MDOC_BLOCK):
 619                 /* FALLTHROUGH */
 620         case (MDOC_ELEM):
 621                 if (NULL == mdocs[n->tok])
 622                         break;
 623
 624                 dbt_init(key, ksz);
 625                 (*mdocs[n->tok])(db, dbn, key, ksz, val, n);
 626
 627                 if (0 == key->size)
 628                         break;
 629                 if (0 == (*db->put)(db, key, val, 0))
 630                         break;
 631
 632                 perror(dbn);
 633                 exit((int)MANDOCLEVEL_SYSERR);
 634                 /* NOTREACHED */
 635         default:
 636                 break;
 637         }
 638
 639         pmdoc_node(db, dbn, key, ksz, val, n->child);
 640         pmdoc_node(db, dbn, key, ksz, val, n->next);
 641 }
 642
 643 static void
 644 pmdoc(DB *db, const char *dbn,
 645                 DBT *key, size_t *ksz, DBT *val,
 646                 const char *path, struct mdoc *m)
 647 {
 648
 649         pmdoc_node(db, dbn, key, ksz, val, mdoc_node(m));
 650 }
 651
 652 static void
 653 version(void)
 654 {
 655
 656         printf("%s %s\n", progname, VERSION);
 657 }
 658
 659 static void
 660 usage(void)
 661 {
 662
 663         fprintf(stderr, "usage: %s "
 664                         "[-V] "
 665                         "[-d path] "
 666                         "[file...]\n",
 667                         progname);
 668 }