contrib/mdocml/cgi.c

   1 /*      $Id: cgi.c,v 1.46 2013/10/11 00:06:48 schwarze Exp $ */
   2 /*
   3  * Copyright (c) 2011, 2012 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/wait.h>
  22
  23 #include <assert.h>
  24 #include <ctype.h>
  25 #include <errno.h>
  26 #include <dirent.h>
  27 #include <fcntl.h>
  28 #include <limits.h>
  29 #include <regex.h>
  30 #include <stdio.h>
  31 #include <stdarg.h>
  32 #include <stdint.h>
  33 #include <stdlib.h>
  34 #include <string.h>
  35 #include <unistd.h>
  36
  37 #if defined(__sun)
  38 /* for stat() */
  39 #include <fcntl.h>
  40 #include <sys/types.h>
  41 #include <sys/stat.h>
  42 #endif
  43
  44 #include "apropos_db.h"
  45 #include "mandoc.h"
  46 #include "mdoc.h"
  47 #include "man.h"
  48 #include "main.h"
  49 #include "manpath.h"
  50 #include "mandocdb.h"
  51
  52 #if defined(__linux__) || defined(__sun)
  53 # include <db_185.h>
  54 #else
  55 # include <db.h>
  56 #endif
  57
  58 enum    page {
  59         PAGE_INDEX,
  60         PAGE_SEARCH,
  61         PAGE_SHOW,
  62         PAGE__MAX
  63 };
  64
  65 struct  paths {
  66         char            *name;
  67         char            *path;
  68 };
  69
  70 /*
  71  * A query as passed to the search function.
  72  */
  73 struct  query {
  74         const char      *arch; /* architecture */
  75         const char      *sec; /* manual section */
  76         const char      *expr; /* unparsed expression string */
  77         int              manroot; /* manroot index (or -1)*/
  78         int              legacy; /* whether legacy mode */
  79 };
  80
  81 struct  req {
  82         struct query     q;
  83         struct paths    *p;
  84         size_t           psz;
  85         enum page        page;
  86 };
  87
  88 static  int              atou(const char *, unsigned *);
  89 static  void             catman(const struct req *, const char *);
  90 static  int              cmp(const void *, const void *);
  91 static  void             format(const struct req *, const char *);
  92 static  void             html_print(const char *);
  93 static  void             html_printquery(const struct req *);
  94 static  void             html_putchar(char);
  95 static  int              http_decode(char *);
  96 static  void             http_parse(struct req *, char *);
  97 static  void             http_print(const char *);
  98 static  void             http_putchar(char);
  99 static  void             http_printquery(const struct req *);
 100 static  int              pathstop(DIR *);
 101 static  void             pathgen(DIR *, char *, struct req *);
 102 static  void             pg_index(const struct req *, char *);
 103 static  void             pg_search(const struct req *, char *);
 104 static  void             pg_show(const struct req *, char *);
 105 static  void             resp_bad(void);
 106 static  void             resp_baddb(void);
 107 static  void             resp_error400(void);
 108 static  void             resp_error404(const char *);
 109 static  void             resp_begin_html(int, const char *);
 110 static  void             resp_begin_http(int, const char *);
 111 static  void             resp_end_html(void);
 112 static  void             resp_index(const struct req *);
 113 static  void             resp_search(struct res *, size_t, void *);
 114 static  void             resp_searchform(const struct req *);
 115
 116 static  const char       *progname; /* cgi script name */
 117 static  const char       *cache; /* cache directory */
 118 static  const char       *css; /* css directory */
 119 static  const char       *host; /* hostname */
 120
 121 static  const char * const pages[PAGE__MAX] = {
 122         "index", /* PAGE_INDEX */
 123         "search", /* PAGE_SEARCH */
 124         "show", /* PAGE_SHOW */
 125 };
 126
 127 /*
 128  * This is just OpenBSD's strtol(3) suggestion.
 129  * I use it instead of strtonum(3) for portability's sake.
 130  */
 131 static int
 132 atou(const char *buf, unsigned *v)
 133 {
 134         char            *ep;
 135         long             lval;
 136
 137         errno = 0;
 138         lval = strtol(buf, &ep, 10);
 139         if (buf[0] == '\0' || *ep != '\0')
 140                 return(0);
 141         if ((errno == ERANGE && (lval == LONG_MAX ||
 142                                         lval == LONG_MIN)) ||
 143                         (lval > INT_MAX || lval < 0))
 144                 return(0);
 145
 146         *v = (unsigned int)lval;
 147         return(1);
 148 }
 149
 150 /*
 151  * Print a character, escaping HTML along the way.
 152  * This will pass non-ASCII straight to output: be warned!
 153  */
 154 static void
 155 html_putchar(char c)
 156 {
 157
 158         switch (c) {
 159         case ('"'):
 160                 printf("&quote;");
 161                 break;
 162         case ('&'):
 163                 printf("&amp;");
 164                 break;
 165         case ('>'):
 166                 printf("&gt;");
 167                 break;
 168         case ('<'):
 169                 printf("&lt;");
 170                 break;
 171         default:
 172                 putchar((unsigned char)c);
 173                 break;
 174         }
 175 }
 176 static void
 177 http_printquery(const struct req *req)
 178 {
 179
 180         printf("&expr=");
 181         http_print(req->q.expr ? req->q.expr : "");
 182         printf("&sec=");
 183         http_print(req->q.sec ? req->q.sec : "");
 184         printf("&arch=");
 185         http_print(req->q.arch ? req->q.arch : "");
 186 }
 187
 188
 189 static void
 190 html_printquery(const struct req *req)
 191 {
 192
 193         printf("&amp;expr=");
 194         html_print(req->q.expr ? req->q.expr : "");
 195         printf("&amp;sec=");
 196         html_print(req->q.sec ? req->q.sec : "");
 197         printf("&amp;arch=");
 198         html_print(req->q.arch ? req->q.arch : "");
 199 }
 200
 201 static void
 202 http_print(const char *p)
 203 {
 204
 205         if (NULL == p)
 206                 return;
 207         while ('\0' != *p)
 208                 http_putchar(*p++);
 209 }
 210
 211 /*
 212  * Call through to html_putchar().
 213  * Accepts NULL strings.
 214  */
 215 static void
 216 html_print(const char *p)
 217 {
 218
 219         if (NULL == p)
 220                 return;
 221         while ('\0' != *p)
 222                 html_putchar(*p++);
 223 }
 224
 225 /*
 226  * Parse out key-value pairs from an HTTP request variable.
 227  * This can be either a cookie or a POST/GET string, although man.cgi
 228  * uses only GET for simplicity.
 229  */
 230 static void
 231 http_parse(struct req *req, char *p)
 232 {
 233         char            *key, *val, *manroot;
 234         int              i, legacy;
 235
 236         memset(&req->q, 0, sizeof(struct query));
 237
 238         legacy = -1;
 239         manroot = NULL;
 240
 241         while ('\0' != *p) {
 242                 key = p;
 243                 val = NULL;
 244
 245                 p += (int)strcspn(p, ";&");
 246                 if ('\0' != *p)
 247                         *p++ = '\0';
 248                 if (NULL != (val = strchr(key, '=')))
 249                         *val++ = '\0';
 250
 251                 if ('\0' == *key || NULL == val || '\0' == *val)
 252                         continue;
 253
 254                 /* Just abort handling. */
 255
 256                 if ( ! http_decode(key))
 257                         break;
 258                 if (NULL != val && ! http_decode(val))
 259                         break;
 260
 261                 if (0 == strcmp(key, "expr"))
 262                         req->q.expr = val;
 263                 else if (0 == strcmp(key, "query"))
 264                         req->q.expr = val;
 265                 else if (0 == strcmp(key, "sec"))
 266                         req->q.sec = val;
 267                 else if (0 == strcmp(key, "sektion"))
 268                         req->q.sec = val;
 269                 else if (0 == strcmp(key, "arch"))
 270                         req->q.arch = val;
 271                 else if (0 == strcmp(key, "manpath"))
 272                         manroot = val;
 273                 else if (0 == strcmp(key, "apropos"))
 274                         legacy = 0 == strcmp(val, "0");
 275         }
 276
 277         /* Test for old man.cgi compatibility mode. */
 278
 279         req->q.legacy = legacy > 0;
 280
 281         /*
 282          * Section "0" means no section when in legacy mode.
 283          * For some man.cgi scripts, "default" arch is none.
 284          */
 285
 286         if (req->q.legacy && NULL != req->q.sec)
 287                 if (0 == strcmp(req->q.sec, "0"))
 288                         req->q.sec = NULL;
 289         if (req->q.legacy && NULL != req->q.arch)
 290                 if (0 == strcmp(req->q.arch, "default"))
 291                         req->q.arch = NULL;
 292
 293         /* Default to first manroot. */
 294
 295         if (NULL != manroot) {
 296                 for (i = 0; i < (int)req->psz; i++)
 297                         if (0 == strcmp(req->p[i].name, manroot))
 298                                 break;
 299                 req->q.manroot = i < (int)req->psz ? i : -1;
 300         }
 301 }
 302
 303 static void
 304 http_putchar(char c)
 305 {
 306
 307         if (isalnum((unsigned char)c)) {
 308                 putchar((unsigned char)c);
 309                 return;
 310         } else if (' ' == c) {
 311                 putchar('+');
 312                 return;
 313         }
 314         printf("%%%.2x", c);
 315 }
 316
 317 /*
 318  * HTTP-decode a string.  The standard explanation is that this turns
 319  * "%4e+foo" into "n foo" in the regular way.  This is done in-place
 320  * over the allocated string.
 321  */
 322 static int
 323 http_decode(char *p)
 324 {
 325         char             hex[3];
 326         int              c;
 327
 328         hex[2] = '\0';
 329
 330         for ( ; '\0' != *p; p++) {
 331                 if ('%' == *p) {
 332                         if ('\0' == (hex[0] = *(p + 1)))
 333                                 return(0);
 334                         if ('\0' == (hex[1] = *(p + 2)))
 335                                 return(0);
 336                         if (1 != sscanf(hex, "%x", &c))
 337                                 return(0);
 338                         if ('\0' == c)
 339                                 return(0);
 340
 341                         *p = (char)c;
 342                         memmove(p + 1, p + 3, strlen(p + 3) + 1);
 343                 } else
 344                         *p = '+' == *p ? ' ' : *p;
 345         }
 346
 347         *p = '\0';
 348         return(1);
 349 }
 350
 351 static void
 352 resp_begin_http(int code, const char *msg)
 353 {
 354
 355         if (200 != code)
 356                 printf("Status: %d %s\n", code, msg);
 357
 358         puts("Content-Type: text/html; charset=utf-8\n"
 359              "Cache-Control: no-cache\n"
 360              "Pragma: no-cache\n"
 361              "");
 362
 363         fflush(stdout);
 364 }
 365
 366 static void
 367 resp_begin_html(int code, const char *msg)
 368 {
 369
 370         resp_begin_http(code, msg);
 371
 372         printf("<!DOCTYPE HTML PUBLIC "
 373                " \"-//W3C//DTD HTML 4.01//EN\""
 374                " \"http://www.w3.org/TR/html4/strict.dtd\">\n"
 375                "<HTML>\n"
 376                "<HEAD>\n"
 377                "<META HTTP-EQUIV=\"Content-Type\""
 378                " CONTENT=\"text/html; charset=utf-8\">\n"
 379                "<LINK REL=\"stylesheet\" HREF=\"%s/man-cgi.css\""
 380                " TYPE=\"text/css\" media=\"all\">\n"
 381                "<LINK REL=\"stylesheet\" HREF=\"%s/man.css\""
 382                " TYPE=\"text/css\" media=\"all\">\n"
 383                "<TITLE>System Manpage Reference</TITLE>\n"
 384                "</HEAD>\n"
 385                "<BODY>\n"
 386                "<!-- Begin page content. //-->\n", css, css);
 387 }
 388
 389 static void
 390 resp_end_html(void)
 391 {
 392
 393         puts("</BODY>\n"
 394              "</HTML>");
 395 }
 396
 397 static void
 398 resp_searchform(const struct req *req)
 399 {
 400         int              i;
 401
 402         puts("<!-- Begin search form. //-->");
 403         printf("<DIV ID=\"mancgi\">\n"
 404                "<FORM ACTION=\"%s/search.html\" METHOD=\"get\">\n"
 405                "<FIELDSET>\n"
 406                "<LEGEND>Search Parameters</LEGEND>\n"
 407                "<INPUT TYPE=\"submit\" "
 408                " VALUE=\"Search\"> for manuals satisfying \n"
 409                "<INPUT TYPE=\"text\" NAME=\"expr\" VALUE=\"",
 410                progname);
 411         html_print(req->q.expr ? req->q.expr : "");
 412         printf("\">, section "
 413                "<INPUT TYPE=\"text\""
 414                " SIZE=\"4\" NAME=\"sec\" VALUE=\"");
 415         html_print(req->q.sec ? req->q.sec : "");
 416         printf("\">, arch "
 417                "<INPUT TYPE=\"text\""
 418                " SIZE=\"8\" NAME=\"arch\" VALUE=\"");
 419         html_print(req->q.arch ? req->q.arch : "");
 420         printf("\">");
 421         if (req->psz > 1) {
 422                 puts(", <SELECT NAME=\"manpath\">");
 423                 for (i = 0; i < (int)req->psz; i++) {
 424                         printf("<OPTION %s VALUE=\"",
 425                                 (i == req->q.manroot) ||
 426                                 (0 == i && -1 == req->q.manroot) ?
 427                                 "SELECTED=\"selected\"" : "");
 428                         html_print(req->p[i].name);
 429                         printf("\">");
 430                         html_print(req->p[i].name);
 431                         puts("</OPTION>");
 432                 }
 433                 puts("</SELECT>");
 434         }
 435         puts(".\n"
 436              "<INPUT TYPE=\"reset\" VALUE=\"Reset\">\n"
 437              "</FIELDSET>\n"
 438              "</FORM>\n"
 439              "</DIV>");
 440         puts("<!-- End search form. //-->");
 441 }
 442
 443 static void
 444 resp_index(const struct req *req)
 445 {
 446
 447         resp_begin_html(200, NULL);
 448         resp_searchform(req);
 449         resp_end_html();
 450 }
 451
 452 static void
 453 resp_error400(void)
 454 {
 455
 456         resp_begin_html(400, "Query Malformed");
 457         printf("<H1>Malformed Query</H1>\n"
 458                "<P>\n"
 459                "The query your entered was malformed.\n"
 460                "Try again from the\n"
 461                "<A HREF=\"%s/index.html\">main page</A>.\n"
 462                "</P>", progname);
 463         resp_end_html();
 464 }
 465
 466 static void
 467 resp_error404(const char *page)
 468 {
 469
 470         resp_begin_html(404, "Not Found");
 471         puts("<H1>Page Not Found</H1>\n"
 472              "<P>\n"
 473              "The page you're looking for, ");
 474         printf("<B>");
 475         html_print(page);
 476         printf("</B>,\n"
 477                "could not be found.\n"
 478                "Try searching from the\n"
 479                "<A HREF=\"%s/index.html\">main page</A>.\n"
 480                "</P>", progname);
 481         resp_end_html();
 482 }
 483
 484 static void
 485 resp_bad(void)
 486 {
 487         resp_begin_html(500, "Internal Server Error");
 488         puts("<P>Generic badness happened.</P>");
 489         resp_end_html();
 490 }
 491
 492 static void
 493 resp_baddb(void)
 494 {
 495
 496         resp_begin_html(500, "Internal Server Error");
 497         puts("<P>Your database is broken.</P>");
 498         resp_end_html();
 499 }
 500
 501 static void
 502 resp_search(struct res *r, size_t sz, void *arg)
 503 {
 504         size_t           i, matched;
 505         const struct req *req;
 506
 507         req = (const struct req *)arg;
 508
 509         if (sz > 0)
 510                 assert(req->q.manroot >= 0);
 511
 512         for (matched = i = 0; i < sz; i++)
 513                 if (r[i].matched)
 514                         matched++;
 515
 516         if (1 == matched) {
 517                 for (i = 0; i < sz; i++)
 518                         if (r[i].matched)
 519                                 break;
 520                 /*
 521                  * If we have just one result, then jump there now
 522                  * without any delay.
 523                  */
 524                 puts("Status: 303 See Other");
 525                 printf("Location: http://%s%s/show/%d/%u/%u.html?",
 526                                 host, progname, req->q.manroot,
 527                                 r[i].volume, r[i].rec);
 528                 http_printquery(req);
 529                 puts("\n"
 530                      "Content-Type: text/html; charset=utf-8\n");
 531                 return;
 532         }
 533
 534         resp_begin_html(200, NULL);
 535         resp_searchform(req);
 536
 537         puts("<DIV CLASS=\"results\">");
 538
 539         if (0 == matched) {
 540                 puts("<P>\n"
 541                      "No results found.\n"
 542                      "</P>\n"
 543                      "</DIV>");
 544                 resp_end_html();
 545                 return;
 546         }
 547
 548         qsort(r, sz, sizeof(struct res), cmp);
 549
 550         puts("<TABLE>");
 551
 552         for (i = 0; i < sz; i++) {
 553                 if ( ! r[i].matched)
 554                         continue;
 555                 printf("<TR>\n"
 556                        "<TD CLASS=\"title\">\n"
 557                        "<A HREF=\"%s/show/%d/%u/%u.html?",
 558                                 progname, req->q.manroot,
 559                                 r[i].volume, r[i].rec);
 560                 html_printquery(req);
 561                 printf("\">");
 562                 html_print(r[i].title);
 563                 putchar('(');
 564                 html_print(r[i].cat);
 565                 if (r[i].arch && '\0' != *r[i].arch) {
 566                         putchar('/');
 567                         html_print(r[i].arch);
 568                 }
 569                 printf(")</A>\n"
 570                        "</TD>\n"
 571                        "<TD CLASS=\"desc\">");
 572                 html_print(r[i].desc);
 573                 puts("</TD>\n"
 574                      "</TR>");
 575         }
 576
 577         puts("</TABLE>\n"
 578              "</DIV>");
 579         resp_end_html();
 580 }
 581
 582 /* ARGSUSED */
 583 static void
 584 pg_index(const struct req *req, char *path)
 585 {
 586
 587         resp_index(req);
 588 }
 589
 590 static void
 591 catman(const struct req *req, const char *file)
 592 {
 593         FILE            *f;
 594         size_t           len;
 595         int              i;
 596         char            *p;
 597         int              italic, bold;
 598
 599         if (NULL == (f = fopen(file, "r"))) {
 600                 resp_baddb();
 601                 return;
 602         }
 603
 604         resp_begin_html(200, NULL);
 605         resp_searchform(req);
 606         puts("<DIV CLASS=\"catman\">\n"
 607              "<PRE>");
 608
 609         while (NULL != (p = fgetln(f, &len))) {
 610                 bold = italic = 0;
 611                 for (i = 0; i < (int)len - 1; i++) {
 612                         /*
 613                          * This means that the catpage is out of state.
 614                          * Ignore it and keep going (although the
 615                          * catpage is bogus).
 616                          */
 617
 618                         if ('\b' == p[i] || '\n' == p[i])
 619                                 continue;
 620
 621                         /*
 622                          * Print a regular character.
 623                          * Close out any bold/italic scopes.
 624                          * If we're in back-space mode, make sure we'll
 625                          * have something to enter when we backspace.
 626                          */
 627
 628                         if ('\b' != p[i + 1]) {
 629                                 if (italic)
 630                                         printf("</I>");
 631                                 if (bold)
 632                                         printf("</B>");
 633                                 italic = bold = 0;
 634                                 html_putchar(p[i]);
 635                                 continue;
 636                         } else if (i + 2 >= (int)len)
 637                                 continue;
 638
 639                         /* Italic mode. */
 640
 641                         if ('_' == p[i]) {
 642                                 if (bold)
 643                                         printf("</B>");
 644                                 if ( ! italic)
 645                                         printf("<I>");
 646                                 bold = 0;
 647                                 italic = 1;
 648                                 i += 2;
 649                                 html_putchar(p[i]);
 650                                 continue;
 651                         }
 652
 653                         /*
 654                          * Handle funny behaviour troff-isms.
 655                          * These grok'd from the original man2html.c.
 656                          */
 657
 658                         if (('+' == p[i] && 'o' == p[i + 2]) ||
 659                                         ('o' == p[i] && '+' == p[i + 2]) ||
 660                                         ('|' == p[i] && '=' == p[i + 2]) ||
 661                                         ('=' == p[i] && '|' == p[i + 2]) ||
 662                                         ('*' == p[i] && '=' == p[i + 2]) ||
 663                                         ('=' == p[i] && '*' == p[i + 2]) ||
 664                                         ('*' == p[i] && '|' == p[i + 2]) ||
 665                                         ('|' == p[i] && '*' == p[i + 2]))  {
 666                                 if (italic)
 667                                         printf("</I>");
 668                                 if (bold)
 669                                         printf("</B>");
 670                                 italic = bold = 0;
 671                                 putchar('*');
 672                                 i += 2;
 673                                 continue;
 674                         } else if (('|' == p[i] && '-' == p[i + 2]) ||
 675                                         ('-' == p[i] && '|' == p[i + 1]) ||
 676                                         ('+' == p[i] && '-' == p[i + 1]) ||
 677                                         ('-' == p[i] && '+' == p[i + 1]) ||
 678                                         ('+' == p[i] && '|' == p[i + 1]) ||
 679                                         ('|' == p[i] && '+' == p[i + 1]))  {
 680                                 if (italic)
 681                                         printf("</I>");
 682                                 if (bold)
 683                                         printf("</B>");
 684                                 italic = bold = 0;
 685                                 putchar('+');
 686                                 i += 2;
 687                                 continue;
 688                         }
 689
 690                         /* Bold mode. */
 691
 692                         if (italic)
 693                                 printf("</I>");
 694                         if ( ! bold)
 695                                 printf("<B>");
 696                         bold = 1;
 697                         italic = 0;
 698                         i += 2;
 699                         html_putchar(p[i]);
 700                 }
 701
 702                 /*
 703                  * Clean up the last character.
 704                  * We can get to a newline; don't print that.
 705                  */
 706
 707                 if (italic)
 708                         printf("</I>");
 709                 if (bold)
 710                         printf("</B>");
 711
 712                 if (i == (int)len - 1 && '\n' != p[i])
 713                         html_putchar(p[i]);
 714
 715                 putchar('\n');
 716         }
 717
 718         puts("</PRE>\n"
 719              "</DIV>\n"
 720              "</BODY>\n"
 721              "</HTML>");
 722
 723         fclose(f);
 724 }
 725
 726 static void
 727 format(const struct req *req, const char *file)
 728 {
 729         struct mparse   *mp;
 730         int              fd;
 731         struct mdoc     *mdoc;
 732         struct man      *man;
 733         void            *vp;
 734         enum mandoclevel rc;
 735         char             opts[PATH_MAX + 128];
 736
 737         if (-1 == (fd = open(file, O_RDONLY, 0))) {
 738                 resp_baddb();
 739                 return;
 740         }
 741
 742         mp = mparse_alloc(MPARSE_AUTO, MANDOCLEVEL_FATAL, NULL, NULL, NULL);
 743         rc = mparse_readfd(mp, fd, file);
 744         close(fd);
 745
 746         if (rc >= MANDOCLEVEL_FATAL) {
 747                 resp_baddb();
 748                 return;
 749         }
 750
 751         snprintf(opts, sizeof(opts), "fragment,"
 752                         "man=%s/search.html?sec=%%S&expr=Nm~^%%N$,"
 753                         /*"includes=/cgi-bin/man.cgi/usr/include/%%I"*/,
 754                         progname);
 755
 756         mparse_result(mp, &mdoc, &man);
 757         if (NULL == man && NULL == mdoc) {
 758                 resp_baddb();
 759                 mparse_free(mp);
 760                 return;
 761         }
 762
 763         resp_begin_html(200, NULL);
 764         resp_searchform(req);
 765
 766         vp = html_alloc(opts);
 767
 768         if (NULL != mdoc)
 769                 html_mdoc(vp, mdoc);
 770         else
 771                 html_man(vp, man);
 772
 773         puts("</BODY>\n"
 774              "</HTML>");
 775
 776         html_free(vp);
 777         mparse_free(mp);
 778 }
 779
 780 static void
 781 pg_show(const struct req *req, char *path)
 782 {
 783         struct manpaths  ps;
 784         size_t           sz;
 785         char            *sub;
 786         char             file[PATH_MAX];
 787         const char      *cp;
 788         int              rc, catm;
 789         unsigned int     vol, rec, mr;
 790         DB              *idx;
 791         DBT              key, val;
 792
 793         idx = NULL;
 794
 795         /* Parse out mroot, volume, and record from the path. */
 796
 797         if (NULL == path || NULL == (sub = strchr(path, '/'))) {
 798                 resp_error400();
 799                 return;
 800         }
 801         *sub++ = '\0';
 802         if ( ! atou(path, &mr)) {
 803                 resp_error400();
 804                 return;
 805         }
 806         path = sub;
 807         if (NULL == (sub = strchr(path, '/'))) {
 808                 resp_error400();
 809                 return;
 810         }
 811         *sub++ = '\0';
 812         if ( ! atou(path, &vol) || ! atou(sub, &rec)) {
 813                 resp_error400();
 814                 return;
 815         } else if (mr >= (unsigned int)req->psz) {
 816                 resp_error400();
 817                 return;
 818         }
 819
 820         /*
 821          * Begin by chdir()ing into the manroot.
 822          * This way we can pick up the database files, which are
 823          * relative to the manpath root.
 824          */
 825
 826         if (-1 == chdir(req->p[(int)mr].path)) {
 827                 perror(req->p[(int)mr].path);
 828                 resp_baddb();
 829                 return;
 830         }
 831
 832         memset(&ps, 0, sizeof(struct manpaths));
 833         manpath_manconf(&ps, "etc/catman.conf");
 834
 835         if (vol >= (unsigned int)ps.sz) {
 836                 resp_error400();
 837                 goto out;
 838         }
 839
 840         sz = strlcpy(file, ps.paths[vol], PATH_MAX);
 841         assert(sz < PATH_MAX);
 842         strlcat(file, "/", PATH_MAX);
 843         strlcat(file, MANDOC_IDX, PATH_MAX);
 844
 845         /* Open the index recno(3) database. */
 846
 847         idx = dbopen(file, O_RDONLY, 0, DB_RECNO, NULL);
 848         if (NULL == idx) {
 849                 perror(file);
 850                 resp_baddb();
 851                 goto out;
 852         }
 853
 854         key.data = &rec;
 855         key.size = 4;
 856
 857         if (0 != (rc = (*idx->get)(idx, &key, &val, 0))) {
 858                 rc < 0 ? resp_baddb() : resp_error400();
 859                 goto out;
 860         } else if (0 == val.size) {
 861                 resp_baddb();
 862                 goto out;
 863         }
 864
 865         cp = (char *)val.data;
 866         catm = 'c' == *cp++;
 867
 868         if (NULL == memchr(cp, '\0', val.size - 1))
 869                 resp_baddb();
 870         else {
 871                 file[(int)sz] = '\0';
 872                 strlcat(file, "/", PATH_MAX);
 873                 strlcat(file, cp, PATH_MAX);
 874                 if (catm)
 875                         catman(req, file);
 876                 else
 877                         format(req, file);
 878         }
 879 out:
 880         if (idx)
 881                 (*idx->close)(idx);
 882         manpath_free(&ps);
 883 }
 884
 885 static void
 886 pg_search(const struct req *req, char *path)
 887 {
 888         size_t            tt, ressz;
 889         struct manpaths   ps;
 890         int               i, sz, rc;
 891         const char       *ep, *start;
 892         struct res      *res;
 893         char            **cp;
 894         struct opts       opt;
 895         struct expr      *expr;
 896
 897         if (req->q.manroot < 0 || 0 == req->psz) {
 898                 resp_search(NULL, 0, (void *)req);
 899                 return;
 900         }
 901
 902         memset(&opt, 0, sizeof(struct opts));
 903
 904         ep       = req->q.expr;
 905         opt.arch = req->q.arch;
 906         opt.cat  = req->q.sec;
 907         rc       = -1;
 908         sz       = 0;
 909         cp       = NULL;
 910         ressz    = 0;
 911         res      = NULL;
 912
 913         /*
 914          * Begin by chdir()ing into the root of the manpath.
 915          * This way we can pick up the database files, which are
 916          * relative to the manpath root.
 917          */
 918
 919         assert(req->q.manroot < (int)req->psz);
 920         if (-1 == (chdir(req->p[req->q.manroot].path))) {
 921                 perror(req->p[req->q.manroot].path);
 922                 resp_search(NULL, 0, (void *)req);
 923                 return;
 924         }
 925
 926         memset(&ps, 0, sizeof(struct manpaths));
 927         manpath_manconf(&ps, "etc/catman.conf");
 928
 929         /*
 930          * Poor man's tokenisation: just break apart by spaces.
 931          * Yes, this is half-ass.  But it works for now.
 932          */
 933
 934         while (ep && isspace((unsigned char)*ep))
 935                 ep++;
 936
 937         while (ep && '\0' != *ep) {
 938                 cp = mandoc_realloc(cp, (sz + 1) * sizeof(char *));
 939                 start = ep;
 940                 while ('\0' != *ep && ! isspace((unsigned char)*ep))
 941                         ep++;
 942                 cp[sz] = mandoc_malloc((ep - start) + 1);
 943                 memcpy(cp[sz], start, ep - start);
 944                 cp[sz++][ep - start] = '\0';
 945                 while (isspace((unsigned char)*ep))
 946                         ep++;
 947         }
 948
 949         /*
 950          * Pump down into apropos backend.
 951          * The resp_search() function is called with the results.
 952          */
 953
 954         expr = req->q.legacy ?
 955                 termcomp(sz, cp, &tt) : exprcomp(sz, cp, &tt);
 956
 957         if (NULL != expr)
 958                 rc = apropos_search
 959                         (ps.sz, ps.paths, &opt, expr, tt,
 960                          (void *)req, &ressz, &res, resp_search);
 961
 962         /* ...unless errors occured. */
 963
 964         if (0 == rc)
 965                 resp_baddb();
 966         else if (-1 == rc)
 967                 resp_search(NULL, 0, NULL);
 968
 969         for (i = 0; i < sz; i++)
 970                 free(cp[i]);
 971
 972         free(cp);
 973         resfree(res, ressz);
 974         exprfree(expr);
 975         manpath_free(&ps);
 976 }
 977
 978 int
 979 main(void)
 980 {
 981         int              i;
 982         char             buf[PATH_MAX];
 983         DIR             *cwd;
 984         struct req       req;
 985         char            *p, *path, *subpath;
 986
 987         /* Scan our run-time environment. */
 988
 989         if (NULL == (cache = getenv("CACHE_DIR")))
 990                 cache = "/cache/man.cgi";
 991
 992         if (NULL == (progname = getenv("SCRIPT_NAME")))
 993                 progname = "";
 994
 995         if (NULL == (css = getenv("CSS_DIR")))
 996                 css = "";
 997
 998         if (NULL == (host = getenv("HTTP_HOST")))
 999                 host = "localhost";
1000
1001         /*
1002          * First we change directory into the cache directory so that
1003          * subsequent scanning for manpath directories is rooted
1004          * relative to the same position.
1005          */
1006
1007         if (-1 == chdir(cache)) {
1008                 perror(cache);
1009                 resp_bad();
1010                 return(EXIT_FAILURE);
1011         } else if (NULL == (cwd = opendir(cache))) {
1012                 perror(cache);
1013                 resp_bad();
1014                 return(EXIT_FAILURE);
1015         }
1016
1017         memset(&req, 0, sizeof(struct req));
1018
1019         strlcpy(buf, ".", PATH_MAX);
1020         pathgen(cwd, buf, &req);
1021         closedir(cwd);
1022
1023         /* Next parse out the query string. */
1024
1025         if (NULL != (p = getenv("QUERY_STRING")))
1026                 http_parse(&req, p);
1027
1028         /*
1029          * Now juggle paths to extract information.
1030          * We want to extract our filetype (the file suffix), the
1031          * initial path component, then the trailing component(s).
1032          * Start with leading subpath component.
1033          */
1034
1035         subpath = path = NULL;
1036         req.page = PAGE__MAX;
1037
1038         if (NULL == (path = getenv("PATH_INFO")) || '\0' == *path)
1039                 req.page = PAGE_INDEX;
1040
1041         if (NULL != path && '/' == *path && '\0' == *++path)
1042                 req.page = PAGE_INDEX;
1043
1044         /* Strip file suffix. */
1045
1046         if (NULL != path && NULL != (p = strrchr(path, '.')))
1047                 if (NULL != p && NULL == strchr(p, '/'))
1048                         *p++ = '\0';
1049
1050         /* Resolve subpath component. */
1051
1052         if (NULL != path && NULL != (subpath = strchr(path, '/')))
1053                 *subpath++ = '\0';
1054
1055         /* Map path into one we recognise. */
1056
1057         if (NULL != path && '\0' != *path)
1058                 for (i = 0; i < (int)PAGE__MAX; i++)
1059                         if (0 == strcmp(pages[i], path)) {
1060                                 req.page = (enum page)i;
1061                                 break;
1062                         }
1063
1064         /* Route pages. */
1065
1066         switch (req.page) {
1067         case (PAGE_INDEX):
1068                 pg_index(&req, subpath);
1069                 break;
1070         case (PAGE_SEARCH):
1071                 pg_search(&req, subpath);
1072                 break;
1073         case (PAGE_SHOW):
1074                 pg_show(&req, subpath);
1075                 break;
1076         default:
1077                 resp_error404(path);
1078                 break;
1079         }
1080
1081         for (i = 0; i < (int)req.psz; i++) {
1082                 free(req.p[i].path);
1083                 free(req.p[i].name);
1084         }
1085
1086         free(req.p);
1087         return(EXIT_SUCCESS);
1088 }
1089
1090 static int
1091 cmp(const void *p1, const void *p2)
1092 {
1093
1094         return(strcasecmp(((const struct res *)p1)->title,
1095                                 ((const struct res *)p2)->title));
1096 }
1097
1098 /*
1099  * Check to see if an "etc" path consists of a catman.conf file.  If it
1100  * does, that means that the path contains a tree created by catman(8)
1101  * and should be used for indexing.
1102  */
1103 static int
1104 pathstop(DIR *dir)
1105 {
1106         struct dirent   *d;
1107 #if defined(__sun)
1108         struct stat      sb;
1109 #endif
1110
1111         while (NULL != (d = readdir(dir))) {
1112 #if defined(__sun)
1113                 stat(d->d_name, &sb);
1114                 if (S_IFREG & sb.st_mode)
1115 #else
1116                 if (DT_REG == d->d_type)
1117 #endif
1118                         if (0 == strcmp(d->d_name, "catman.conf"))
1119                                 return(1);
1120   }
1121
1122         return(0);
1123 }
1124
1125 /*
1126  * Scan for indexable paths.
1127  * This adds all paths with "etc/catman.conf" to the buffer.
1128  */
1129 static void
1130 pathgen(DIR *dir, char *path, struct req *req)
1131 {
1132         struct dirent   *d;
1133         char            *cp;
1134         DIR             *cd;
1135         int              rc;
1136         size_t           sz, ssz;
1137 #if defined(__sun)
1138         struct stat      sb;
1139 #endif
1140
1141         sz = strlcat(path, "/", PATH_MAX);
1142         if (sz >= PATH_MAX) {
1143                 fprintf(stderr, "%s: Path too long", path);
1144                 return;
1145         }
1146
1147         /*
1148          * First, scan for the "etc" directory.
1149          * If it's found, then see if it should cause us to stop.  This
1150          * happens when a catman.conf is found in the directory.
1151          */
1152
1153         rc = 0;
1154         while (0 == rc && NULL != (d = readdir(dir))) {
1155 #if defined(__sun)
1156                 stat(d->d_name, &sb);
1157                 if (!(S_IFDIR & sb.st_mode)
1158 #else
1159                 if (DT_DIR != d->d_type
1160 #endif
1161         || strcmp(d->d_name, "etc"))
1162                         continue;
1163
1164                 path[(int)sz] = '\0';
1165                 ssz = strlcat(path, d->d_name, PATH_MAX);
1166
1167                 if (ssz >= PATH_MAX) {
1168                         fprintf(stderr, "%s: Path too long", path);
1169                         return;
1170                 } else if (NULL == (cd = opendir(path))) {
1171                         perror(path);
1172                         return;
1173                 }
1174
1175                 rc = pathstop(cd);
1176                 closedir(cd);
1177         }
1178
1179         if (rc > 0) {
1180                 /* This also strips the trailing slash. */
1181                 path[(int)--sz] = '\0';
1182                 req->p = mandoc_realloc
1183                         (req->p,
1184                          (req->psz + 1) * sizeof(struct paths));
1185                 /*
1186                  * Strip out the leading "./" unless we're just a ".",
1187                  * in which case use an empty string as our name.
1188                  */
1189                 req->p[(int)req->psz].path = mandoc_strdup(path);
1190                 req->p[(int)req->psz].name =
1191                         cp = mandoc_strdup(path + (1 == sz ? 1 : 2));
1192                 req->psz++;
1193                 /*
1194                  * The name is just the path with all the slashes taken
1195                  * out of it.  Simple but effective.
1196                  */
1197                 for ( ; '\0' != *cp; cp++)
1198                         if ('/' == *cp)
1199                                 *cp = ' ';
1200                 return;
1201         }
1202
1203         /*
1204          * If no etc/catman.conf was found, recursively enter child
1205          * directory and continue scanning.
1206          */
1207
1208         rewinddir(dir);
1209         while (NULL != (d = readdir(dir))) {
1210 #if defined(__sun)
1211                 stat(d->d_name, &sb);
1212                 if (!(S_IFDIR & sb.st_mode)
1213 #else
1214                 if (DT_DIR != d->d_type
1215 #endif
1216         || '.' == d->d_name[0])
1217                         continue;
1218
1219                 path[(int)sz] = '\0';
1220                 ssz = strlcat(path, d->d_name, PATH_MAX);
1221
1222                 if (ssz >= PATH_MAX) {
1223                         fprintf(stderr, "%s: Path too long", path);
1224                         return;
1225                 } else if (NULL == (cd = opendir(path))) {
1226                         perror(path);
1227                         return;
1228                 }
1229
1230                 pathgen(cd, path, req);
1231                 closedir(cd);
1232         }
1233 }