Annotation of mandoc/mandocdb.c, Revision 1.1
1.1 ! kristaps 1: /* $Id: makewhatis.c,v 1.21 2011/07/12 15:26:35 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: #include <fcntl.h>
! 25: #include <getopt.h>
! 26: #include <stdio.h>
! 27: #include <stdint.h>
! 28: #include <stdlib.h>
! 29: #include <string.h>
! 30:
! 31: #ifdef __linux__
! 32: # include <db_185.h>
! 33: #else
! 34: # include <db.h>
! 35: #endif
! 36:
! 37: #include "man.h"
! 38: #include "mdoc.h"
! 39: #include "mandoc.h"
! 40:
! 41: #define MANDOC_DB "mandoc.db"
! 42: #define MANDOC_IDX "mandoc.index"
! 43: #define MANDOC_BUFSZ BUFSIZ
! 44: #define MANDOC_FLAGS O_CREAT|O_TRUNC|O_RDWR
! 45: #define MANDOC_SLOP 1024
! 46:
! 47: /* Bit-fields. See makewhatis.1. */
! 48:
! 49: #define TYPE_NAME 0x01
! 50: #define TYPE_FUNCTION 0x02
! 51: #define TYPE_UTILITY 0x04
! 52: #define TYPE_INCLUDES 0x08
! 53: #define TYPE_VARIABLE 0x10
! 54: #define TYPE_STANDARD 0x20
! 55: #define TYPE_AUTHOR 0x40
! 56: #define TYPE_CONFIG 0x80
! 57: #define TYPE_DESC 0x100
! 58: #define TYPE_XREF 0x200
! 59: #define TYPE_PATH 0x400
! 60: #define TYPE_ENV 0x800
! 61: #define TYPE_ERR 0x1000
! 62:
! 63: /* Buffer for storing growable data. */
! 64:
! 65: struct buf {
! 66: char *cp;
! 67: size_t len;
! 68: size_t size;
! 69: };
! 70:
! 71: /* Operation we're going to perform. */
! 72:
! 73: enum op {
! 74: OP_NEW = 0, /* new database */
! 75: OP_UPDATE, /* update entries in existing database */
! 76: OP_DELETE /* delete entries from existing database */
! 77: };
! 78:
! 79: #define MAN_ARGS DB *hash, \
! 80: struct buf *buf, \
! 81: struct buf *dbuf, \
! 82: const struct man_node *n
! 83: #define MDOC_ARGS DB *hash, \
! 84: struct buf *buf, \
! 85: struct buf *dbuf, \
! 86: const struct mdoc_node *n, \
! 87: const struct mdoc_meta *m
! 88:
! 89: static void buf_appendmdoc(struct buf *,
! 90: const struct mdoc_node *, int);
! 91: static void buf_append(struct buf *, const char *);
! 92: static void buf_appendb(struct buf *,
! 93: const void *, size_t);
! 94: static void dbt_put(DB *, const char *, DBT *, DBT *);
! 95: static void hash_put(DB *, const struct buf *, int);
! 96: static void hash_reset(DB **);
! 97: static int pman_node(MAN_ARGS);
! 98: static void pmdoc_node(MDOC_ARGS);
! 99: static void pmdoc_An(MDOC_ARGS);
! 100: static void pmdoc_Cd(MDOC_ARGS);
! 101: static void pmdoc_Er(MDOC_ARGS);
! 102: static void pmdoc_Ev(MDOC_ARGS);
! 103: static void pmdoc_Fd(MDOC_ARGS);
! 104: static void pmdoc_In(MDOC_ARGS);
! 105: static void pmdoc_Fn(MDOC_ARGS);
! 106: static void pmdoc_Fo(MDOC_ARGS);
! 107: static void pmdoc_Nd(MDOC_ARGS);
! 108: static void pmdoc_Nm(MDOC_ARGS);
! 109: static void pmdoc_Pa(MDOC_ARGS);
! 110: static void pmdoc_St(MDOC_ARGS);
! 111: static void pmdoc_Vt(MDOC_ARGS);
! 112: static void pmdoc_Xr(MDOC_ARGS);
! 113: static void usage(void);
! 114:
! 115: typedef void (*pmdoc_nf)(MDOC_ARGS);
! 116:
! 117: static const pmdoc_nf mdocs[MDOC_MAX] = {
! 118: NULL, /* Ap */
! 119: NULL, /* Dd */
! 120: NULL, /* Dt */
! 121: NULL, /* Os */
! 122: NULL, /* Sh */
! 123: NULL, /* Ss */
! 124: NULL, /* Pp */
! 125: NULL, /* D1 */
! 126: NULL, /* Dl */
! 127: NULL, /* Bd */
! 128: NULL, /* Ed */
! 129: NULL, /* Bl */
! 130: NULL, /* El */
! 131: NULL, /* It */
! 132: NULL, /* Ad */
! 133: pmdoc_An, /* An */
! 134: NULL, /* Ar */
! 135: pmdoc_Cd, /* Cd */
! 136: NULL, /* Cm */
! 137: NULL, /* Dv */
! 138: pmdoc_Er, /* Er */
! 139: pmdoc_Ev, /* Ev */
! 140: NULL, /* Ex */
! 141: NULL, /* Fa */
! 142: pmdoc_Fd, /* Fd */
! 143: NULL, /* Fl */
! 144: pmdoc_Fn, /* Fn */
! 145: NULL, /* Ft */
! 146: NULL, /* Ic */
! 147: pmdoc_In, /* In */
! 148: NULL, /* Li */
! 149: pmdoc_Nd, /* Nd */
! 150: pmdoc_Nm, /* Nm */
! 151: NULL, /* Op */
! 152: NULL, /* Ot */
! 153: pmdoc_Pa, /* Pa */
! 154: NULL, /* Rv */
! 155: pmdoc_St, /* St */
! 156: pmdoc_Vt, /* Va */
! 157: pmdoc_Vt, /* Vt */
! 158: pmdoc_Xr, /* Xr */
! 159: NULL, /* %A */
! 160: NULL, /* %B */
! 161: NULL, /* %D */
! 162: NULL, /* %I */
! 163: NULL, /* %J */
! 164: NULL, /* %N */
! 165: NULL, /* %O */
! 166: NULL, /* %P */
! 167: NULL, /* %R */
! 168: NULL, /* %T */
! 169: NULL, /* %V */
! 170: NULL, /* Ac */
! 171: NULL, /* Ao */
! 172: NULL, /* Aq */
! 173: NULL, /* At */
! 174: NULL, /* Bc */
! 175: NULL, /* Bf */
! 176: NULL, /* Bo */
! 177: NULL, /* Bq */
! 178: NULL, /* Bsx */
! 179: NULL, /* Bx */
! 180: NULL, /* Db */
! 181: NULL, /* Dc */
! 182: NULL, /* Do */
! 183: NULL, /* Dq */
! 184: NULL, /* Ec */
! 185: NULL, /* Ef */
! 186: NULL, /* Em */
! 187: NULL, /* Eo */
! 188: NULL, /* Fx */
! 189: NULL, /* Ms */
! 190: NULL, /* No */
! 191: NULL, /* Ns */
! 192: NULL, /* Nx */
! 193: NULL, /* Ox */
! 194: NULL, /* Pc */
! 195: NULL, /* Pf */
! 196: NULL, /* Po */
! 197: NULL, /* Pq */
! 198: NULL, /* Qc */
! 199: NULL, /* Ql */
! 200: NULL, /* Qo */
! 201: NULL, /* Qq */
! 202: NULL, /* Re */
! 203: NULL, /* Rs */
! 204: NULL, /* Sc */
! 205: NULL, /* So */
! 206: NULL, /* Sq */
! 207: NULL, /* Sm */
! 208: NULL, /* Sx */
! 209: NULL, /* Sy */
! 210: NULL, /* Tn */
! 211: NULL, /* Ux */
! 212: NULL, /* Xc */
! 213: NULL, /* Xo */
! 214: pmdoc_Fo, /* Fo */
! 215: NULL, /* Fc */
! 216: NULL, /* Oo */
! 217: NULL, /* Oc */
! 218: NULL, /* Bk */
! 219: NULL, /* Ek */
! 220: NULL, /* Bt */
! 221: NULL, /* Hf */
! 222: NULL, /* Fr */
! 223: NULL, /* Ud */
! 224: NULL, /* Lb */
! 225: NULL, /* Lp */
! 226: NULL, /* Lk */
! 227: NULL, /* Mt */
! 228: NULL, /* Brq */
! 229: NULL, /* Bro */
! 230: NULL, /* Brc */
! 231: NULL, /* %C */
! 232: NULL, /* Es */
! 233: NULL, /* En */
! 234: NULL, /* Dx */
! 235: NULL, /* %Q */
! 236: NULL, /* br */
! 237: NULL, /* sp */
! 238: NULL, /* %U */
! 239: NULL, /* Ta */
! 240: };
! 241:
! 242: static const char *progname;
! 243:
! 244: int
! 245: main(int argc, char *argv[])
! 246: {
! 247: struct mparse *mp; /* parse sequence */
! 248: struct mdoc *mdoc; /* resulting mdoc */
! 249: struct man *man; /* resulting man */
! 250: enum op op; /* current operation */
! 251: char *fn; /* current file being parsed */
! 252: const char *msec, /* manual section */
! 253: *mtitle, /* manual title */
! 254: *arch, /* manual architecture */
! 255: *dir; /* result dir (default: cwd) */
! 256: char ibuf[MAXPATHLEN], /* index fname */
! 257: fbuf[MAXPATHLEN], /* btree fname */
! 258: vbuf[8]; /* stringified record number */
! 259: int ch, seq, sseq, verb, i;
! 260: DB *idx, /* index database */
! 261: *db, /* keyword database */
! 262: *hash; /* temporary keyword hashtable */
! 263: DBT key, val;
! 264: enum mandoclevel ec; /* exit status */
! 265: size_t sv;
! 266: BTREEINFO info; /* btree configuration */
! 267: recno_t rec,
! 268: maxrec; /* supremum of all records */
! 269: recno_t *recs; /* buffer of empty records */
! 270: size_t recsz, /* buffer size of recs */
! 271: reccur; /* valid number of recs */
! 272: struct buf buf, /* keyword buffer */
! 273: dbuf; /* description buffer */
! 274: extern int optind;
! 275: extern char *optarg;
! 276:
! 277: progname = strrchr(argv[0], '/');
! 278: if (progname == NULL)
! 279: progname = argv[0];
! 280: else
! 281: ++progname;
! 282:
! 283: dir = "";
! 284: verb = 0;
! 285: db = idx = NULL;
! 286: mp = NULL;
! 287: hash = NULL;
! 288: recs = NULL;
! 289: recsz = reccur = 0;
! 290: maxrec = 0;
! 291: op = OP_NEW;
! 292: ec = MANDOCLEVEL_SYSERR;
! 293:
! 294: memset(&buf, 0, sizeof(struct buf));
! 295: memset(&dbuf, 0, sizeof(struct buf));
! 296:
! 297: while (-1 != (ch = getopt(argc, argv, "d:ruv")))
! 298: switch (ch) {
! 299: case ('d'):
! 300: dir = optarg;
! 301: break;
! 302: case ('r'):
! 303: op = OP_DELETE;
! 304: break;
! 305: case ('u'):
! 306: op = OP_UPDATE;
! 307: break;
! 308: case ('v'):
! 309: verb++;
! 310: break;
! 311: default:
! 312: usage();
! 313: return((int)MANDOCLEVEL_BADARG);
! 314: }
! 315:
! 316: argc -= optind;
! 317: argv += optind;
! 318:
! 319: ibuf[0] = ibuf[MAXPATHLEN - 2] =
! 320: fbuf[0] = fbuf[MAXPATHLEN - 2] = '\0';
! 321:
! 322: strlcat(fbuf, dir, MAXPATHLEN);
! 323: strlcat(fbuf, MANDOC_DB, MAXPATHLEN);
! 324:
! 325: strlcat(ibuf, dir, MAXPATHLEN);
! 326: strlcat(ibuf, MANDOC_IDX, MAXPATHLEN);
! 327:
! 328: if ('\0' != fbuf[MAXPATHLEN - 2] ||
! 329: '\0' != ibuf[MAXPATHLEN - 2]) {
! 330: fprintf(stderr, "%s: Path too long\n", dir);
! 331: goto out;
! 332: }
! 333:
! 334: /*
! 335: * For the keyword database, open a BTREE database that allows
! 336: * duplicates.
! 337: * For the index database, use a standard RECNO database type.
! 338: * Truncate the database if we're creating a new one.
! 339: */
! 340:
! 341: memset(&info, 0, sizeof(BTREEINFO));
! 342: info.flags = R_DUP;
! 343:
! 344: if (OP_NEW == op) {
! 345: db = dbopen(fbuf, MANDOC_FLAGS, 0644, DB_BTREE, &info);
! 346: idx = dbopen(ibuf, MANDOC_FLAGS, 0644, DB_RECNO, NULL);
! 347: } else {
! 348: db = dbopen(fbuf, O_CREAT|O_RDWR, 0644, DB_BTREE, &info);
! 349: idx = dbopen(ibuf, O_CREAT|O_RDWR, 0644, DB_RECNO, NULL);
! 350: }
! 351:
! 352: if (NULL == db) {
! 353: perror(fbuf);
! 354: goto out;
! 355: } else if (NULL == db) {
! 356: perror(ibuf);
! 357: goto out;
! 358: }
! 359:
! 360: /*
! 361: * If we're going to delete or update a database, remove the
! 362: * entries now (both the index and all keywords pointing to it).
! 363: * This doesn't actually remove them: it only sets their record
! 364: * value lengths to zero.
! 365: * While doing so, add the empty records to a list we'll access
! 366: * later in re-adding entries to the database.
! 367: */
! 368:
! 369: if (OP_DELETE == op || OP_UPDATE == op) {
! 370: seq = R_FIRST;
! 371: while (0 == (ch = (*idx->seq)(idx, &key, &val, seq))) {
! 372: seq = R_NEXT;
! 373: maxrec = *(recno_t *)key.data;
! 374: if (0 == val.size && OP_UPDATE == op) {
! 375: if (reccur >= recsz) {
! 376: recsz += MANDOC_SLOP;
! 377: recs = mandoc_realloc
! 378: (recs, recsz * sizeof(recno_t));
! 379: }
! 380: recs[(int)reccur] = maxrec;
! 381: reccur++;
! 382: continue;
! 383: }
! 384:
! 385: fn = (char *)val.data;
! 386: for (i = 0; i < argc; i++)
! 387: if (0 == strcmp(fn, argv[i]))
! 388: break;
! 389:
! 390: if (i == argc)
! 391: continue;
! 392:
! 393: sseq = R_FIRST;
! 394: while (0 == (ch = (*db->seq)(db, &key, &val, sseq))) {
! 395: sseq = R_NEXT;
! 396: assert(8 == val.size);
! 397: if (maxrec != *(recno_t *)(val.data + 4))
! 398: continue;
! 399: if (verb > 1)
! 400: printf("%s: Deleted keyword: %s\n",
! 401: fn, (char *)key.data);
! 402: ch = (*db->del)(db, &key, R_CURSOR);
! 403: if (ch < 0)
! 404: break;
! 405: }
! 406: if (ch < 0) {
! 407: perror(fbuf);
! 408: exit((int)MANDOCLEVEL_SYSERR);
! 409: }
! 410:
! 411: if (verb)
! 412: printf("%s: Deleted index\n", fn);
! 413:
! 414: val.size = 0;
! 415: ch = (*idx->put)(idx, &key, &val, R_CURSOR);
! 416: if (ch < 0) {
! 417: perror(ibuf);
! 418: exit((int)MANDOCLEVEL_SYSERR);
! 419: }
! 420:
! 421: if (OP_UPDATE == op) {
! 422: if (reccur >= recsz) {
! 423: recsz += MANDOC_SLOP;
! 424: recs = mandoc_realloc
! 425: (recs, recsz * sizeof(recno_t));
! 426: }
! 427: recs[(int)reccur] = maxrec;
! 428: reccur++;
! 429: }
! 430: }
! 431: maxrec++;
! 432: }
! 433:
! 434: if (OP_DELETE == op) {
! 435: ec = MANDOCLEVEL_OK;
! 436: goto out;
! 437: }
! 438:
! 439: /*
! 440: * Add records to the database.
! 441: * Try parsing each manual given on the command line.
! 442: * If we fail, then emit an error and keep on going.
! 443: * Take resulting trees and push them down into the database code.
! 444: * Use the auto-parser and don't report any errors.
! 445: */
! 446:
! 447: mp = mparse_alloc(MPARSE_AUTO, MANDOCLEVEL_FATAL, NULL, NULL);
! 448:
! 449: buf.size = dbuf.size = MANDOC_BUFSZ;
! 450: buf.cp = mandoc_malloc(buf.size);
! 451: dbuf.cp = mandoc_malloc(dbuf.size);
! 452:
! 453: for (rec = 0, i = 0; i < argc; i++) {
! 454: fn = argv[i];
! 455: if (OP_UPDATE == op) {
! 456: if (reccur > 0) {
! 457: --reccur;
! 458: rec = recs[(int)reccur];
! 459: } else if (maxrec > 0) {
! 460: rec = maxrec;
! 461: maxrec = 0;
! 462: } else
! 463: rec++;
! 464: } else
! 465: rec++;
! 466:
! 467: mparse_reset(mp);
! 468: hash_reset(&hash);
! 469:
! 470: if (mparse_readfd(mp, -1, fn) >= MANDOCLEVEL_FATAL) {
! 471: fprintf(stderr, "%s: Parse failure\n", fn);
! 472: continue;
! 473: }
! 474:
! 475: mparse_result(mp, &mdoc, &man);
! 476: if (NULL == mdoc && NULL == man)
! 477: continue;
! 478:
! 479: msec = NULL != mdoc ?
! 480: mdoc_meta(mdoc)->msec : man_meta(man)->msec;
! 481: mtitle = NULL != mdoc ?
! 482: mdoc_meta(mdoc)->title : man_meta(man)->title;
! 483: arch = NULL != mdoc ? mdoc_meta(mdoc)->arch : NULL;
! 484:
! 485: if (NULL == arch)
! 486: arch = "";
! 487:
! 488: /*
! 489: * The index record value consists of a nil-terminated
! 490: * filename, a nil-terminated manual section, and a
! 491: * nil-terminated description. Since the description
! 492: * may not be set, we set a sentinel to see if we're
! 493: * going to write a nil byte in its place.
! 494: */
! 495:
! 496: dbuf.len = 0;
! 497: buf_appendb(&dbuf, fn, strlen(fn) + 1);
! 498: buf_appendb(&dbuf, msec, strlen(msec) + 1);
! 499: buf_appendb(&dbuf, mtitle, strlen(mtitle) + 1);
! 500: buf_appendb(&dbuf, arch, strlen(arch) + 1);
! 501:
! 502: sv = dbuf.len;
! 503:
! 504: /* Fix the record number in the btree value. */
! 505:
! 506: if (mdoc)
! 507: pmdoc_node(hash, &buf, &dbuf,
! 508: mdoc_node(mdoc), mdoc_meta(mdoc));
! 509: else
! 510: pman_node(hash, &buf, &dbuf, man_node(man));
! 511:
! 512: /*
! 513: * Copy from the in-memory hashtable of pending keywords
! 514: * into the database.
! 515: */
! 516:
! 517: memset(vbuf, 0, sizeof(uint32_t));
! 518: memcpy(vbuf + 4, &rec, sizeof(uint32_t));
! 519:
! 520: seq = R_FIRST;
! 521: while (0 == (ch = (*hash->seq)(hash, &key, &val, seq))) {
! 522: seq = R_NEXT;
! 523:
! 524: memcpy(vbuf, val.data, sizeof(uint32_t));
! 525: val.size = sizeof(vbuf);
! 526: val.data = vbuf;
! 527:
! 528: if (verb > 1)
! 529: printf("%s: Added keyword: %s, 0x%x\n",
! 530: fn, (char *)key.data,
! 531: *(int *)val.data);
! 532: dbt_put(db, fbuf, &key, &val);
! 533: }
! 534: if (ch < 0) {
! 535: perror("hash");
! 536: exit((int)MANDOCLEVEL_SYSERR);
! 537: }
! 538:
! 539: /*
! 540: * Apply to the index. If we haven't had a description
! 541: * set, put an empty one in now.
! 542: */
! 543:
! 544: if (dbuf.len == sv)
! 545: buf_appendb(&dbuf, "", 1);
! 546:
! 547: key.data = &rec;
! 548: key.size = sizeof(recno_t);
! 549:
! 550: val.data = dbuf.cp;
! 551: val.size = dbuf.len;
! 552:
! 553: if (verb > 0)
! 554: printf("%s: Added index\n", fn);
! 555:
! 556: dbt_put(idx, ibuf, &key, &val);
! 557: }
! 558:
! 559: ec = MANDOCLEVEL_OK;
! 560: out:
! 561: if (db)
! 562: (*db->close)(db);
! 563: if (idx)
! 564: (*idx->close)(idx);
! 565: if (hash)
! 566: (*hash->close)(hash);
! 567: if (mp)
! 568: mparse_free(mp);
! 569:
! 570: free(buf.cp);
! 571: free(dbuf.cp);
! 572: free(recs);
! 573:
! 574: return((int)ec);
! 575: }
! 576:
! 577: /*
! 578: * Grow the buffer (if necessary) and copy in a binary string.
! 579: */
! 580: static void
! 581: buf_appendb(struct buf *buf, const void *cp, size_t sz)
! 582: {
! 583:
! 584: /* Overshoot by MANDOC_BUFSZ. */
! 585:
! 586: while (buf->len + sz >= buf->size) {
! 587: buf->size = buf->len + sz + MANDOC_BUFSZ;
! 588: buf->cp = mandoc_realloc(buf->cp, buf->size);
! 589: }
! 590:
! 591: memcpy(buf->cp + (int)buf->len, cp, sz);
! 592: buf->len += sz;
! 593: }
! 594:
! 595: /*
! 596: * Append a nil-terminated string to the buffer.
! 597: * This can be invoked multiple times.
! 598: * The buffer string will be nil-terminated.
! 599: * If invoked multiple times, a space is put between strings.
! 600: */
! 601: static void
! 602: buf_append(struct buf *buf, const char *cp)
! 603: {
! 604: size_t sz;
! 605:
! 606: if (0 == (sz = strlen(cp)))
! 607: return;
! 608:
! 609: if (buf->len)
! 610: buf->cp[(int)buf->len - 1] = ' ';
! 611:
! 612: buf_appendb(buf, cp, sz + 1);
! 613: }
! 614:
! 615: /*
! 616: * Recursively add all text from a given node.
! 617: * This is optimised for general mdoc nodes in this context, which do
! 618: * not consist of subexpressions and having a recursive call for n->next
! 619: * would be wasteful.
! 620: * The "f" variable should be 0 unless called from pmdoc_Nd for the
! 621: * description buffer, which does not start at the beginning of the
! 622: * buffer.
! 623: */
! 624: static void
! 625: buf_appendmdoc(struct buf *buf, const struct mdoc_node *n, int f)
! 626: {
! 627:
! 628: for ( ; n; n = n->next) {
! 629: if (n->child)
! 630: buf_appendmdoc(buf, n->child, f);
! 631:
! 632: if (MDOC_TEXT == n->type && f) {
! 633: f = 0;
! 634: buf_appendb(buf, n->string,
! 635: strlen(n->string) + 1);
! 636: } else if (MDOC_TEXT == n->type)
! 637: buf_append(buf, n->string);
! 638:
! 639: }
! 640: }
! 641:
! 642: /* ARGSUSED */
! 643: static void
! 644: pmdoc_An(MDOC_ARGS)
! 645: {
! 646:
! 647: if (SEC_AUTHORS != n->sec)
! 648: return;
! 649:
! 650: buf_appendmdoc(buf, n->child, 0);
! 651: hash_put(hash, buf, TYPE_AUTHOR);
! 652: }
! 653:
! 654: static void
! 655: hash_reset(DB **db)
! 656: {
! 657: DB *hash;
! 658:
! 659: if (NULL != (hash = *db))
! 660: (*hash->close)(hash);
! 661:
! 662: *db = dbopen(NULL, MANDOC_FLAGS, 0644, DB_HASH, NULL);
! 663: if (NULL == *db) {
! 664: perror("hash");
! 665: exit((int)MANDOCLEVEL_SYSERR);
! 666: }
! 667: }
! 668:
! 669: /* ARGSUSED */
! 670: static void
! 671: pmdoc_Fd(MDOC_ARGS)
! 672: {
! 673: const char *start, *end;
! 674: size_t sz;
! 675:
! 676: if (SEC_SYNOPSIS != n->sec)
! 677: return;
! 678: if (NULL == (n = n->child) || MDOC_TEXT != n->type)
! 679: return;
! 680:
! 681: /*
! 682: * Only consider those `Fd' macro fields that begin with an
! 683: * "inclusion" token (versus, e.g., #define).
! 684: */
! 685: if (strcmp("#include", n->string))
! 686: return;
! 687:
! 688: if (NULL == (n = n->next) || MDOC_TEXT != n->type)
! 689: return;
! 690:
! 691: /*
! 692: * Strip away the enclosing angle brackets and make sure we're
! 693: * not zero-length.
! 694: */
! 695:
! 696: start = n->string;
! 697: if ('<' == *start || '"' == *start)
! 698: start++;
! 699:
! 700: if (0 == (sz = strlen(start)))
! 701: return;
! 702:
! 703: end = &start[(int)sz - 1];
! 704: if ('>' == *end || '"' == *end)
! 705: end--;
! 706:
! 707: assert(end >= start);
! 708:
! 709: buf_appendb(buf, start, (size_t)(end - start + 1));
! 710: buf_appendb(buf, "", 1);
! 711:
! 712: hash_put(hash, buf, TYPE_INCLUDES);
! 713: }
! 714:
! 715: /* ARGSUSED */
! 716: static void
! 717: pmdoc_Cd(MDOC_ARGS)
! 718: {
! 719:
! 720: if (SEC_SYNOPSIS != n->sec)
! 721: return;
! 722:
! 723: buf_appendmdoc(buf, n->child, 0);
! 724: hash_put(hash, buf, TYPE_CONFIG);
! 725: }
! 726:
! 727: /* ARGSUSED */
! 728: static void
! 729: pmdoc_In(MDOC_ARGS)
! 730: {
! 731:
! 732: if (SEC_SYNOPSIS != n->sec)
! 733: return;
! 734: if (NULL == n->child || MDOC_TEXT != n->child->type)
! 735: return;
! 736:
! 737: buf_append(buf, n->child->string);
! 738: hash_put(hash, buf, TYPE_INCLUDES);
! 739: }
! 740:
! 741: /* ARGSUSED */
! 742: static void
! 743: pmdoc_Fn(MDOC_ARGS)
! 744: {
! 745: const char *cp;
! 746:
! 747: if (SEC_SYNOPSIS != n->sec)
! 748: return;
! 749: if (NULL == n->child || MDOC_TEXT != n->child->type)
! 750: return;
! 751:
! 752: /* .Fn "struct type *arg" "foo" */
! 753:
! 754: cp = strrchr(n->child->string, ' ');
! 755: if (NULL == cp)
! 756: cp = n->child->string;
! 757:
! 758: /* Strip away pointer symbol. */
! 759:
! 760: while ('*' == *cp)
! 761: cp++;
! 762:
! 763: buf_append(buf, cp);
! 764: hash_put(hash, buf, TYPE_FUNCTION);
! 765: }
! 766:
! 767: /* ARGSUSED */
! 768: static void
! 769: pmdoc_St(MDOC_ARGS)
! 770: {
! 771:
! 772: if (SEC_STANDARDS != n->sec)
! 773: return;
! 774: if (NULL == n->child || MDOC_TEXT != n->child->type)
! 775: return;
! 776:
! 777: buf_append(buf, n->child->string);
! 778: hash_put(hash, buf, TYPE_STANDARD);
! 779: }
! 780:
! 781: /* ARGSUSED */
! 782: static void
! 783: pmdoc_Xr(MDOC_ARGS)
! 784: {
! 785:
! 786: if (NULL == (n = n->child))
! 787: return;
! 788:
! 789: buf_appendb(buf, n->string, strlen(n->string));
! 790:
! 791: if (NULL != (n = n->next)) {
! 792: buf_appendb(buf, ".", 1);
! 793: buf_appendb(buf, n->string, strlen(n->string) + 1);
! 794: } else
! 795: buf_appendb(buf, ".", 2);
! 796:
! 797: hash_put(hash, buf, TYPE_XREF);
! 798: }
! 799:
! 800: /* ARGSUSED */
! 801: static void
! 802: pmdoc_Vt(MDOC_ARGS)
! 803: {
! 804: const char *start;
! 805: size_t sz;
! 806:
! 807: if (SEC_SYNOPSIS != n->sec)
! 808: return;
! 809: if (MDOC_Vt == n->tok && MDOC_BODY != n->type)
! 810: return;
! 811: if (NULL == n->last || MDOC_TEXT != n->last->type)
! 812: return;
! 813:
! 814: /*
! 815: * Strip away leading pointer symbol '*' and trailing ';'.
! 816: */
! 817:
! 818: start = n->last->string;
! 819:
! 820: while ('*' == *start)
! 821: start++;
! 822:
! 823: if (0 == (sz = strlen(start)))
! 824: return;
! 825:
! 826: if (';' == start[(int)sz - 1])
! 827: sz--;
! 828:
! 829: if (0 == sz)
! 830: return;
! 831:
! 832: buf_appendb(buf, start, sz);
! 833: buf_appendb(buf, "", 1);
! 834: hash_put(hash, buf, TYPE_VARIABLE);
! 835: }
! 836:
! 837: /* ARGSUSED */
! 838: static void
! 839: pmdoc_Fo(MDOC_ARGS)
! 840: {
! 841:
! 842: if (SEC_SYNOPSIS != n->sec || MDOC_HEAD != n->type)
! 843: return;
! 844: if (NULL == n->child || MDOC_TEXT != n->child->type)
! 845: return;
! 846:
! 847: buf_append(buf, n->child->string);
! 848: hash_put(hash, buf, TYPE_FUNCTION);
! 849: }
! 850:
! 851:
! 852: /* ARGSUSED */
! 853: static void
! 854: pmdoc_Nd(MDOC_ARGS)
! 855: {
! 856:
! 857: if (MDOC_BODY != n->type)
! 858: return;
! 859:
! 860: buf_appendmdoc(dbuf, n->child, 1);
! 861: buf_appendmdoc(buf, n->child, 0);
! 862:
! 863: hash_put(hash, buf, TYPE_DESC);
! 864: }
! 865:
! 866: /* ARGSUSED */
! 867: static void
! 868: pmdoc_Er(MDOC_ARGS)
! 869: {
! 870:
! 871: if (SEC_ERRORS != n->sec)
! 872: return;
! 873:
! 874: buf_appendmdoc(buf, n->child, 0);
! 875: hash_put(hash, buf, TYPE_ERR);
! 876: }
! 877:
! 878: /* ARGSUSED */
! 879: static void
! 880: pmdoc_Ev(MDOC_ARGS)
! 881: {
! 882:
! 883: if (SEC_ENVIRONMENT != n->sec)
! 884: return;
! 885:
! 886: buf_appendmdoc(buf, n->child, 0);
! 887: hash_put(hash, buf, TYPE_ENV);
! 888: }
! 889:
! 890: /* ARGSUSED */
! 891: static void
! 892: pmdoc_Pa(MDOC_ARGS)
! 893: {
! 894:
! 895: if (SEC_FILES != n->sec)
! 896: return;
! 897:
! 898: buf_appendmdoc(buf, n->child, 0);
! 899: hash_put(hash, buf, TYPE_PATH);
! 900: }
! 901:
! 902: /* ARGSUSED */
! 903: static void
! 904: pmdoc_Nm(MDOC_ARGS)
! 905: {
! 906:
! 907: if (SEC_NAME == n->sec) {
! 908: buf_appendmdoc(buf, n->child, 0);
! 909: hash_put(hash, buf, TYPE_NAME);
! 910: return;
! 911: } else if (SEC_SYNOPSIS != n->sec || MDOC_HEAD != n->type)
! 912: return;
! 913:
! 914: if (NULL == n->child)
! 915: buf_append(buf, m->name);
! 916:
! 917: buf_appendmdoc(buf, n->child, 0);
! 918: hash_put(hash, buf, TYPE_UTILITY);
! 919: }
! 920:
! 921: static void
! 922: hash_put(DB *db, const struct buf *buf, int mask)
! 923: {
! 924: DBT key, val;
! 925: int rc;
! 926:
! 927: if (buf->len < 2)
! 928: return;
! 929:
! 930: key.data = buf->cp;
! 931: key.size = buf->len;
! 932:
! 933: if ((rc = (*db->get)(db, &key, &val, 0)) < 0) {
! 934: perror("hash");
! 935: exit((int)MANDOCLEVEL_SYSERR);
! 936: } else if (0 == rc)
! 937: mask |= *(int *)val.data;
! 938:
! 939: val.data = &mask;
! 940: val.size = sizeof(int);
! 941:
! 942: if ((rc = (*db->put)(db, &key, &val, 0)) < 0) {
! 943: perror("hash");
! 944: exit((int)MANDOCLEVEL_SYSERR);
! 945: }
! 946: }
! 947:
! 948: static void
! 949: dbt_put(DB *db, const char *dbn, DBT *key, DBT *val)
! 950: {
! 951:
! 952: assert(key->size);
! 953: assert(val->size);
! 954:
! 955: if (0 == (*db->put)(db, key, val, 0))
! 956: return;
! 957:
! 958: perror(dbn);
! 959: exit((int)MANDOCLEVEL_SYSERR);
! 960: /* NOTREACHED */
! 961: }
! 962:
! 963: /*
! 964: * Call out to per-macro handlers after clearing the persistent database
! 965: * key. If the macro sets the database key, flush it to the database.
! 966: */
! 967: static void
! 968: pmdoc_node(MDOC_ARGS)
! 969: {
! 970:
! 971: if (NULL == n)
! 972: return;
! 973:
! 974: switch (n->type) {
! 975: case (MDOC_HEAD):
! 976: /* FALLTHROUGH */
! 977: case (MDOC_BODY):
! 978: /* FALLTHROUGH */
! 979: case (MDOC_TAIL):
! 980: /* FALLTHROUGH */
! 981: case (MDOC_BLOCK):
! 982: /* FALLTHROUGH */
! 983: case (MDOC_ELEM):
! 984: if (NULL == mdocs[n->tok])
! 985: break;
! 986:
! 987: buf->len = 0;
! 988: (*mdocs[n->tok])(hash, buf, dbuf, n, m);
! 989: break;
! 990: default:
! 991: break;
! 992: }
! 993:
! 994: pmdoc_node(hash, buf, dbuf, n->child, m);
! 995: pmdoc_node(hash, buf, dbuf, n->next, m);
! 996: }
! 997:
! 998: static int
! 999: pman_node(MAN_ARGS)
! 1000: {
! 1001: const struct man_node *head, *body;
! 1002: const char *start, *sv;
! 1003: size_t sz;
! 1004:
! 1005: if (NULL == n)
! 1006: return(0);
! 1007:
! 1008: /*
! 1009: * We're only searching for one thing: the first text child in
! 1010: * the BODY of a NAME section. Since we don't keep track of
! 1011: * sections in -man, run some hoops to find out whether we're in
! 1012: * the correct section or not.
! 1013: */
! 1014:
! 1015: if (MAN_BODY == n->type && MAN_SH == n->tok) {
! 1016: body = n;
! 1017: assert(body->parent);
! 1018: if (NULL != (head = body->parent->head) &&
! 1019: 1 == head->nchild &&
! 1020: NULL != (head = (head->child)) &&
! 1021: MAN_TEXT == head->type &&
! 1022: 0 == strcmp(head->string, "NAME") &&
! 1023: NULL != (body = body->child) &&
! 1024: MAN_TEXT == body->type) {
! 1025:
! 1026: assert(body->string);
! 1027: start = sv = body->string;
! 1028:
! 1029: /*
! 1030: * Go through a special heuristic dance here.
! 1031: * This is why -man manuals are great!
! 1032: * (I'm being sarcastic: my eyes are bleeding.)
! 1033: * Conventionally, one or more manual names are
! 1034: * comma-specified prior to a whitespace, then a
! 1035: * dash, then a description. Try to puzzle out
! 1036: * the name parts here.
! 1037: */
! 1038:
! 1039: for ( ;; ) {
! 1040: sz = strcspn(start, " ,");
! 1041: if ('\0' == start[(int)sz])
! 1042: break;
! 1043:
! 1044: buf->len = 0;
! 1045: buf_appendb(buf, start, sz);
! 1046: buf_appendb(buf, "", 1);
! 1047:
! 1048: hash_put(hash, buf, TYPE_NAME);
! 1049:
! 1050: if (' ' == start[(int)sz]) {
! 1051: start += (int)sz + 1;
! 1052: break;
! 1053: }
! 1054:
! 1055: assert(',' == start[(int)sz]);
! 1056: start += (int)sz + 1;
! 1057: while (' ' == *start)
! 1058: start++;
! 1059: }
! 1060:
! 1061: buf->len = 0;
! 1062:
! 1063: if (sv == start) {
! 1064: buf_append(buf, start);
! 1065: return(1);
! 1066: }
! 1067:
! 1068: while (' ' == *start)
! 1069: start++;
! 1070:
! 1071: if (0 == strncmp(start, "-", 1))
! 1072: start += 1;
! 1073: else if (0 == strncmp(start, "\\-", 2))
! 1074: start += 2;
! 1075: else if (0 == strncmp(start, "\\(en", 4))
! 1076: start += 4;
! 1077: else if (0 == strncmp(start, "\\(em", 4))
! 1078: start += 4;
! 1079:
! 1080: while (' ' == *start)
! 1081: start++;
! 1082:
! 1083: sz = strlen(start) + 1;
! 1084: buf_appendb(dbuf, start, sz);
! 1085: buf_appendb(buf, start, sz);
! 1086:
! 1087: hash_put(hash, buf, TYPE_DESC);
! 1088: }
! 1089: }
! 1090:
! 1091: if (pman_node(hash, buf, dbuf, n->child))
! 1092: return(1);
! 1093: if (pman_node(hash, buf, dbuf, n->next))
! 1094: return(1);
! 1095:
! 1096: return(0);
! 1097: }
! 1098:
! 1099: static void
! 1100: usage(void)
! 1101: {
! 1102:
! 1103: fprintf(stderr, "usage: %s [-ruv] [-d path] [file...]\n",
! 1104: progname);
! 1105: }
CVSweb