contrib/mdocml/cgi.c

   1 /*      $Id: cgi.c,v 1.42 2012/03/24 01:46:25 kristaps 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/param.h>
  22 #include <sys/wait.h>
  23
  24 #include <assert.h>
  25 #include <ctype.h>
  26 #include <errno.h>
  27 #include <dirent.h>
  28 #include <fcntl.h>
  29 #include <limits.h>
  30 #include <regex.h>
  31 #include <stdio.h>
  32 #include <stdarg.h>
  33 #include <stdint.h>
  34 #include <stdlib.h>
  35 #include <string.h>
  36 #include <unistd.h>
  37
  38 #include "apropos_db.h"
  39 #include "mandoc.h"
  40 #include "mdoc.h"
  41 #include "man.h"
  42 #include "main.h"
  43 #include "manpath.h"
  44 #include "mandocdb.h"
  45
  46 #ifdef __linux__
  47 # include <db_185.h>
  48 #else
  49 # include <db.h>
  50 #endif
  51
  52 enum    page {
  53         PAGE_INDEX,
  54         PAGE_SEARCH,
  55         PAGE_SHOW,
  56         PAGE__MAX
  57 };
  58
  59 struct  paths {
  60         char            *name;
  61         char            *path;
  62 };
  63
  64 /*
  65  * A query as passed to the search function.
  66  */
  67 struct  query {
  68         const char      *arch; /* architecture */
  69         const char      *sec; /* manual section */
  70         const char      *expr; /* unparsed expression string */
  71         int              manroot; /* manroot index (or -1)*/
  72         int              legacy; /* whether legacy mode */
  73 };
  74
  75 struct  req {
  76         struct query     q;
  77         struct paths    *p;
  78         size_t           psz;
  79         enum page        page;
  80 };
  81
  82 static  int              atou(const char *, unsigned *);
  83 static  void             catman(const struct req *, const char *);
  84 static  int              cmp(const void *, const void *);
  85 static  void             format(const struct req *, const char *);
  86 static  void             html_print(const char *);
  87 static  void             html_printquery(const struct req *);
  88 static  void             html_putchar(char);
  89 static  int              http_decode(char *);
  90 static  void             http_parse(struct req *, char *);
  91 static  void             http_print(const char *);
  92 static  void             http_putchar(char);
  93 static  void             http_printquery(const struct req *);
  94 static  int              pathstop(DIR *);
  95 static  void             pathgen(DIR *, char *, struct req *);
  96 static  void             pg_index(const struct req *, char *);
  97 static  void             pg_search(const struct req *, char *);
  98 static  void             pg_show(const struct req *, char *);
  99 static  void             resp_bad(void);
 100 static  void             resp_baddb(void);
 101 static  void             resp_error400(void);
 102 static  void             resp_error404(const char *);
 103 static  void             resp_begin_html(int, const char *);
 104 static  void             resp_begin_http(int, const char *);
 105 static  void             resp_end_html(void);
 106 static  void             resp_index(const struct req *);
 107 static  void             resp_search(struct res *, size_t, void *);
 108 static  void             resp_searchform(const struct req *);
 109
 110 static  const char       *progname; /* cgi script name */
 111 static  const char       *cache; /* cache directory */
 112 static  const char       *css; /* css directory */
 113 static  const char       *host; /* hostname */
 114
 115 static  const char * const pages[PAGE__MAX] = {
 116         "index", /* PAGE_INDEX */
 117         "search", /* PAGE_SEARCH */
 118         "show", /* PAGE_SHOW */
 119 };
 120
 121 /*
 122  * This is just OpenBSD's strtol(3) suggestion.
 123  * I use it instead of strtonum(3) for portability's sake.
 124  */
 125 static int
 126 atou(const char *buf, unsigned *v)
 127 {
 128         char            *ep;
 129         long             lval;
 130
 131         errno = 0;
 132         lval = strtol(buf, &ep, 10);
 133         if (buf[0] == '\0' || *ep != '\0')
 134                 return(0);
 135         if ((errno == ERANGE && (lval == LONG_MAX ||
 136                                         lval == LONG_MIN)) ||
 137                         (lval > INT_MAX || lval < 0))
 138                 return(0);
 139
 140         *v = (unsigned int)lval;
 141         return(1);
 142 }
 143
 144 /*
 145  * Print a character, escaping HTML along the way.
 146  * This will pass non-ASCII straight to output: be warned!
 147  */
 148 static void
 149 html_putchar(char c)
 150 {
 151
 152         switch (c) {
 153         case ('"'):
 154                 printf("&quote;");
 155                 break;
 156         case ('&'):
 157                 printf("&amp;");
 158                 break;
 159         case ('>'):
 160                 printf("&gt;");
 161                 break;
 162         case ('<'):
 163                 printf("&lt;");
 164                 break;
 165         default:
 166                 putchar((unsigned char)c);
 167                 break;
 168         }
 169 }
 170 static void
 171 http_printquery(const struct req *req)
 172 {
 173
 174         printf("&expr=");
 175         http_print(req->q.expr ? req->q.expr : "");
 176         printf("&sec=");
 177         http_print(req->q.sec ? req->q.sec : "");
 178         printf("&arch=");
 179         http_print(req->q.arch ? req->q.arch : "");
 180 }
 181
 182
 183 static void
 184 html_printquery(const struct req *req)
 185 {
 186
 187         printf("&amp;expr=");
 188         html_print(req->q.expr ? req->q.expr : "");
 189         printf("&amp;sec=");
 190         html_print(req->q.sec ? req->q.sec : "");
 191         printf("&amp;arch=");
 192         html_print(req->q.arch ? req->q.arch : "");
 193 }
 194
 195 static void
 196 http_print(const char *p)
 197 {
 198
 199         if (NULL == p)
 200                 return;
 201         while ('\0' != *p)
 202                 http_putchar(*p++);
 203 }
 204
 205 /*
 206  * Call through to html_putchar().
 207  * Accepts NULL strings.
 208  */
 209 static void
 210 html_print(const char *p)
 211 {
 212
 213         if (NULL == p)
 214                 return;
 215         while ('\0' != *p)
 216                 html_putchar(*p++);
 217 }
 218
 219 /*
 220  * Parse out key-value pairs from an HTTP request variable.
 221  * This can be either a cookie or a POST/GET string, although man.cgi
 222  * uses only GET for simplicity.
 223  */
 224 static void
 225 http_parse(struct req *req, char *p)
 226 {
 227         char            *key, *val, *manroot;
 228         int              i, legacy;
 229
 230         memset(&req->q, 0, sizeof(struct query));
 231
 232         legacy = -1;
 233         manroot = NULL;
 234
 235         while ('\0' != *p) {
 236                 key = p;
 237                 val = NULL;
 238
 239                 p += (int)strcspn(p, ";&");
 240                 if ('\0' != *p)
 241                         *p++ = '\0';
 242                 if (NULL != (val = strchr(key, '=')))
 243                         *val++ = '\0';
 244
 245                 if ('\0' == *key || NULL == val || '\0' == *val)
 246                         continue;
 247
 248                 /* Just abort handling. */
 249
 250                 if ( ! http_decode(key))
 251                         break;
 252                 if (NULL != val && ! http_decode(val))
 253                         break;
 254
 255                 if (0 == strcmp(key, "expr"))
 256                         req->q.expr = val;
 257                 else if (0 == strcmp(key, "query"))
 258                         req->q.expr = val;
 259                 else if (0 == strcmp(key, "sec"))
 260                         req->q.sec = val;
 261                 else if (0 == strcmp(key, "sektion"))
 262                         req->q.sec = val;
 263                 else if (0 == strcmp(key, "arch"))
 264                         req->q.arch = val;
 265                 else if (0 == strcmp(key, "manpath"))
 266                         manroot = val;
 267                 else if (0 == strcmp(key, "apropos"))
 268                         legacy = 0 == strcmp(val, "0");
 269         }
 270
 271         /* Test for old man.cgi compatibility mode. */
 272
 273         req->q.legacy = legacy > 0;
 274
 275         /*
 276          * Section "0" means no section when in legacy mode.
 277          * For some man.cgi scripts, "default" arch is none.
 278          */
 279
 280         if (req->q.legacy && NULL != req->q.sec)
 281                 if (0 == strcmp(req->q.sec, "0"))
 282                         req->q.sec = NULL;
 283         if (req->q.legacy && NULL != req->q.arch)
 284                 if (0 == strcmp(req->q.arch, "default"))
 285                         req->q.arch = NULL;
 286
 287         /* Default to first manroot. */
 288
 289         if (NULL != manroot) {
 290                 for (i = 0; i < (int)req->psz; i++)
 291                         if (0 == strcmp(req->p[i].name, manroot))
 292                                 break;
 293                 req->q.manroot = i < (int)req->psz ? i : -1;
 294         }
 295 }
 296
 297 static void
 298 http_putchar(char c)
 299 {
 300
 301         if (isalnum((unsigned char)c)) {
 302                 putchar((unsigned char)c);
 303                 return;
 304         } else if (' ' == c) {
 305                 putchar('+');
 306                 return;
 307         }
 308         printf("%%%.2x", c);
 309 }
 310
 311 /*
 312  * HTTP-decode a string.  The standard explanation is that this turns
 313  * "%4e+foo" into "n foo" in the regular way.  This is done in-place
 314  * over the allocated string.
 315  */
 316 static int
 317 http_decode(char *p)
 318 {
 319         char             hex[3];
 320         int              c;
 321
 322         hex[2] = '\0';
 323
 324         for ( ; '\0' != *p; p++) {
 325                 if ('%' == *p) {
 326                         if ('\0' == (hex[0] = *(p + 1)))
 327                                 return(0);
 328                         if ('\0' == (hex[1] = *(p + 2)))
 329                                 return(0);
 330                         if (1 != sscanf(hex, "%x", &c))
 331                                 return(0);
 332                         if ('\0' == c)
 333                                 return(0);
 334
 335                         *p = (char)c;
 336                         memmove(p + 1, p + 3, strlen(p + 3) + 1);
 337                 } else
 338                         *p = '+' == *p ? ' ' : *p;
 339         }
 340
 341         *p = '\0';
 342         return(1);
 343 }
 344
 345 static void
 346 resp_begin_http(int code, const char *msg)
 347 {
 348
 349         if (200 != code)
 350                 printf("Status: %d %s\n", code, msg);
 351
 352         puts("Content-Type: text/html; charset=utf-8\n"
 353              "Cache-Control: no-cache\n"
 354              "Pragma: no-cache\n"
 355              "");
 356
 357         fflush(stdout);
 358 }
 359
 360 static void
 361 resp_begin_html(int code, const char *msg)
 362 {
 363
 364         resp_begin_http(code, msg);
 365
 366         printf("<!DOCTYPE HTML PUBLIC "
 367                " \"-//W3C//DTD HTML 4.01//EN\""
 368                " \"http://www.w3.org/TR/html4/strict.dtd\">\n"
 369                "<HTML>\n"
 370                "<HEAD>\n"
 371                "<META HTTP-EQUIV=\"Content-Type\""
 372                " CONTENT=\"text/html; charset=utf-8\">\n"
 373                "<LINK REL=\"stylesheet\" HREF=\"%s/man-cgi.css\""
 374                " TYPE=\"text/css\" media=\"all\">\n"
 375                "<LINK REL=\"stylesheet\" HREF=\"%s/man.css\""
 376                " TYPE=\"text/css\" media=\"all\">\n"
 377                "<TITLE>System Manpage Reference</TITLE>\n"
 378                "</HEAD>\n"
 379                "<BODY>\n"
 380                "<!-- Begin page content. //-->\n", css, css);
 381 }
 382
 383 static void
 384 resp_end_html(void)
 385 {
 386
 387         puts("</BODY>\n"
 388              "</HTML>");
 389 }
 390
 391 static void
 392 resp_searchform(const struct req *req)
 393 {
 394         int              i;
 395
 396         puts("<!-- Begin search form. //-->");
 397         printf("<DIV ID=\"mancgi\">\n"
 398                "<FORM ACTION=\"%s/search.html\" METHOD=\"get\">\n"
 399                "<FIELDSET>\n"
 400                "<LEGEND>Search Parameters</LEGEND>\n"
 401                "<INPUT TYPE=\"submit\" "
 402                " VALUE=\"Search\"> for manuals satisfying \n"
 403                "<INPUT TYPE=\"text\" NAME=\"expr\" VALUE=\"",
 404                progname);
 405         html_print(req->q.expr ? req->q.expr : "");
 406         printf("\">, section "
 407                "<INPUT TYPE=\"text\""
 408                " SIZE=\"4\" NAME=\"sec\" VALUE=\"");
 409         html_print(req->q.sec ? req->q.sec : "");
 410         printf("\">, arch "
 411                "<INPUT TYPE=\"text\""
 412                " SIZE=\"8\" NAME=\"arch\" VALUE=\"");
 413         html_print(req->q.arch ? req->q.arch : "");
 414         printf("\">");
 415         if (req->psz > 1) {
 416                 puts(", <SELECT NAME=\"manpath\">");
 417                 for (i = 0; i < (int)req->psz; i++) {
 418                         printf("<OPTION %s VALUE=\"",
 419                                 (i == req->q.manroot) ||
 420                                 (0 == i && -1 == req->q.manroot) ?
 421                                 "SELECTED=\"selected\"" : "");
 422                         html_print(req->p[i].name);
 423                         printf("\">");
 424                         html_print(req->p[i].name);
 425                         puts("</OPTION>");
 426                 }
 427                 puts("</SELECT>");
 428         }
 429         puts(".\n"
 430              "<INPUT TYPE=\"reset\" VALUE=\"Reset\">\n"
 431              "</FIELDSET>\n"
 432              "</FORM>\n"
 433              "</DIV>");
 434         puts("<!-- End search form. //-->");
 435 }
 436
 437 static void
 438 resp_index(const struct req *req)
 439 {
 440
 441         resp_begin_html(200, NULL);
 442         resp_searchform(req);
 443         resp_end_html();
 444 }
 445
 446 static void
 447 resp_error400(void)
 448 {
 449
 450         resp_begin_html(400, "Query Malformed");
 451         printf("<H1>Malformed Query</H1>\n"
 452                "<P>\n"
 453                "The query your entered was malformed.\n"
 454                "Try again from the\n"
 455                "<A HREF=\"%s/index.html\">main page</A>.\n"
 456                "</P>", progname);
 457         resp_end_html();
 458 }
 459
 460 static void
 461 resp_error404(const char *page)
 462 {
 463
 464         resp_begin_html(404, "Not Found");
 465         puts("<H1>Page Not Found</H1>\n"
 466              "<P>\n"
 467              "The page you're looking for, ");
 468         printf("<B>");
 469         html_print(page);
 470         printf("</B>,\n"
 471                "could not be found.\n"
 472                "Try searching from the\n"
 473                "<A HREF=\"%s/index.html\">main page</A>.\n"
 474                "</P>", progname);
 475         resp_end_html();
 476 }
 477
 478 static void
 479 resp_bad(void)
 480 {
 481         resp_begin_html(500, "Internal Server Error");
 482         puts("<P>Generic badness happened.</P>");
 483         resp_end_html();
 484 }
 485
 486 static void
 487 resp_baddb(void)
 488 {
 489
 490         resp_begin_html(500, "Internal Server Error");
 491         puts("<P>Your database is broken.</P>");
 492         resp_end_html();
 493 }
 494
 495 static void
 496 resp_search(struct res *r, size_t sz, void *arg)
 497 {
 498         size_t           i, matched;
 499         const struct req *req;
 500
 501         req = (const struct req *)arg;
 502
 503         if (sz > 0)
 504                 assert(req->q.manroot >= 0);
 505
 506         for (matched = i = 0; i < sz; i++)
 507                 if (r[i].matched)
 508                         matched++;
 509
 510         if (1 == matched) {
 511                 for (i = 0; i < sz; i++)
 512                         if (r[i].matched)
 513                                 break;
 514                 /*
 515                  * If we have just one result, then jump there now
 516                  * without any delay.
 517                  */
 518                 puts("Status: 303 See Other");
 519                 printf("Location: http://%s%s/show/%d/%u/%u.html?",
 520                                 host, progname, req->q.manroot,
 521                                 r[i].volume, r[i].rec);
 522                 http_printquery(req);
 523                 puts("\n"
 524                      "Content-Type: text/html; charset=utf-8\n");
 525                 return;
 526         }
 527
 528         resp_begin_html(200, NULL);
 529         resp_searchform(req);
 530
 531         puts("<DIV CLASS=\"results\">");
 532
 533         if (0 == matched) {
 534                 puts("<P>\n"
 535                      "No results found.\n"
 536                      "</P>\n"
 537                      "</DIV>");
 538                 resp_end_html();
 539                 return;
 540         }
 541
 542         qsort(r, sz, sizeof(struct res), cmp);
 543
 544         puts("<TABLE>");
 545
 546         for (i = 0; i < sz; i++) {
 547                 if ( ! r[i].matched)
 548                         continue;
 549                 printf("<TR>\n"
 550                        "<TD CLASS=\"title\">\n"
 551                        "<A HREF=\"%s/show/%d/%u/%u.html?",
 552                                 progname, req->q.manroot,
 553                                 r[i].volume, r[i].rec);
 554                 html_printquery(req);
 555                 printf("\">");
 556                 html_print(r[i].title);
 557                 putchar('(');
 558                 html_print(r[i].cat);
 559                 if (r[i].arch && '\0' != *r[i].arch) {
 560                         putchar('/');
 561                         html_print(r[i].arch);
 562                 }
 563                 printf(")</A>\n"
 564                        "</TD>\n"
 565                        "<TD CLASS=\"desc\">");
 566                 html_print(r[i].desc);
 567                 puts("</TD>\n"
 568                      "</TR>");
 569         }
 570
 571         puts("</TABLE>\n"
 572              "</DIV>");
 573         resp_end_html();
 574 }
 575
 576 /* ARGSUSED */
 577 static void
 578 pg_index(const struct req *req, char *path)
 579 {
 580
 581         resp_index(req);
 582 }
 583
 584 static void
 585 catman(const struct req *req, const char *file)
 586 {
 587         FILE            *f;
 588         size_t           len;
 589         int              i;
 590         char            *p;
 591         int              italic, bold;
 592
 593         if (NULL == (f = fopen(file, "r"))) {
 594                 resp_baddb();
 595                 return;
 596         }
 597
 598         resp_begin_html(200, NULL);
 599         resp_searchform(req);
 600         puts("<DIV CLASS=\"catman\">\n"
 601              "<PRE>");
 602
 603         while (NULL != (p = fgetln(f, &len))) {
 604                 bold = italic = 0;
 605                 for (i = 0; i < (int)len - 1; i++) {
 606                         /*
 607                          * This means that the catpage is out of state.
 608                          * Ignore it and keep going (although the
 609                          * catpage is bogus).
 610                          */
 611
 612                         if ('\b' == p[i] || '\n' == p[i])
 613                                 continue;
 614
 615                         /*
 616                          * Print a regular character.
 617                          * Close out any bold/italic scopes.
 618                          * If we're in back-space mode, make sure we'll
 619                          * have something to enter when we backspace.
 620                          */
 621
 622                         if ('\b' != p[i + 1]) {
 623                                 if (italic)
 624                                         printf("</I>");
 625                                 if (bold)
 626                                         printf("</B>");
 627                                 italic = bold = 0;
 628                                 html_putchar(p[i]);
 629                                 continue;
 630                         } else if (i + 2 >= (int)len)
 631                                 continue;
 632
 633                         /* Italic mode. */
 634
 635                         if ('_' == p[i]) {
 636                                 if (bold)
 637                                         printf("</B>");
 638                                 if ( ! italic)
 639                                         printf("<I>");
 640                                 bold = 0;
 641                                 italic = 1;
 642                                 i += 2;
 643                                 html_putchar(p[i]);
 644                                 continue;
 645                         }
 646
 647                         /*
 648                          * Handle funny behaviour troff-isms.
 649                          * These grok'd from the original man2html.c.
 650                          */
 651
 652                         if (('+' == p[i] && 'o' == p[i + 2]) ||
 653                                         ('o' == p[i] && '+' == p[i + 2]) ||
 654                                         ('|' == p[i] && '=' == p[i + 2]) ||
 655                                         ('=' == p[i] && '|' == p[i + 2]) ||
 656                                         ('*' == p[i] && '=' == p[i + 2]) ||
 657                                         ('=' == p[i] && '*' == p[i + 2]) ||
 658                                         ('*' == p[i] && '|' == p[i + 2]) ||
 659                                         ('|' == p[i] && '*' == p[i + 2]))  {
 660                                 if (italic)
 661                                         printf("</I>");
 662                                 if (bold)
 663                                         printf("</B>");
 664                                 italic = bold = 0;
 665                                 putchar('*');
 666                                 i += 2;
 667                                 continue;
 668                         } else if (('|' == p[i] && '-' == p[i + 2]) ||
 669                                         ('-' == p[i] && '|' == p[i + 1]) ||
 670                                         ('+' == p[i] && '-' == p[i + 1]) ||
 671                                         ('-' == p[i] && '+' == p[i + 1]) ||
 672                                         ('+' == p[i] && '|' == p[i + 1]) ||
 673                                         ('|' == p[i] && '+' == p[i + 1]))  {
 674                                 if (italic)
 675                                         printf("</I>");
 676                                 if (bold)
 677                                         printf("</B>");
 678                                 italic = bold = 0;
 679                                 putchar('+');
 680                                 i += 2;
 681                                 continue;
 682                         }
 683
 684                         /* Bold mode. */
 685
 686                         if (italic)
 687                                 printf("</I>");
 688                         if ( ! bold)
 689                                 printf("<B>");
 690                         bold = 1;
 691                         italic = 0;
 692                         i += 2;
 693                         html_putchar(p[i]);
 694                 }
 695
 696                 /*
 697                  * Clean up the last character.
 698                  * We can get to a newline; don't print that.
 699                  */
 700
 701                 if (italic)
 702                         printf("</I>");
 703                 if (bold)
 704                         printf("</B>");
 705
 706                 if (i == (int)len - 1 && '\n' != p[i])
 707                         html_putchar(p[i]);
 708
 709                 putchar('\n');
 710         }
 711
 712         puts("</PRE>\n"
 713              "</DIV>\n"
 714              "</BODY>\n"
 715              "</HTML>");
 716
 717         fclose(f);
 718 }
 719
 720 static void
 721 format(const struct req *req, const char *file)
 722 {
 723         struct mparse   *mp;
 724         int              fd;
 725         struct mdoc     *mdoc;
 726         struct man      *man;
 727         void            *vp;
 728         enum mandoclevel rc;
 729         char             opts[MAXPATHLEN + 128];
 730
 731         if (-1 == (fd = open(file, O_RDONLY, 0))) {
 732                 resp_baddb();
 733                 return;
 734         }
 735
 736         mp = mparse_alloc(MPARSE_AUTO, MANDOCLEVEL_FATAL, NULL, NULL);
 737         rc = mparse_readfd(mp, fd, file);
 738         close(fd);
 739
 740         if (rc >= MANDOCLEVEL_FATAL) {
 741                 resp_baddb();
 742                 return;
 743         }
 744
 745         snprintf(opts, sizeof(opts), "fragment,"
 746                         "man=%s/search.html?sec=%%S&expr=%%N,"
 747                         /*"includes=/cgi-bin/man.cgi/usr/include/%%I"*/,
 748                         progname);
 749
 750         mparse_result(mp, &mdoc, &man);
 751         if (NULL == man && NULL == mdoc) {
 752                 resp_baddb();
 753                 mparse_free(mp);
 754                 return;
 755         }
 756
 757         resp_begin_html(200, NULL);
 758         resp_searchform(req);
 759
 760         vp = html_alloc(opts);
 761
 762         if (NULL != mdoc)
 763                 html_mdoc(vp, mdoc);
 764         else
 765                 html_man(vp, man);
 766
 767         puts("</BODY>\n"
 768              "</HTML>");
 769
 770         html_free(vp);
 771         mparse_free(mp);
 772 }
 773
 774 static void
 775 pg_show(const struct req *req, char *path)
 776 {
 777         struct manpaths  ps;
 778         size_t           sz;
 779         char            *sub;
 780         char             file[MAXPATHLEN];
 781         const char      *cp;
 782         int              rc, catm;
 783         unsigned int     vol, rec, mr;
 784         DB              *idx;
 785         DBT              key, val;
 786
 787         idx = NULL;
 788
 789         /* Parse out mroot, volume, and record from the path. */
 790
 791         if (NULL == path || NULL == (sub = strchr(path, '/'))) {
 792                 resp_error400();
 793                 return;
 794         }
 795         *sub++ = '\0';
 796         if ( ! atou(path, &mr)) {
 797                 resp_error400();
 798                 return;
 799         }
 800         path = sub;
 801         if (NULL == (sub = strchr(path, '/'))) {
 802                 resp_error400();
 803                 return;
 804         }
 805         *sub++ = '\0';
 806         if ( ! atou(path, &vol) || ! atou(sub, &rec)) {
 807                 resp_error400();
 808                 return;
 809         } else if (mr >= (unsigned int)req->psz) {
 810                 resp_error400();
 811                 return;
 812         }
 813
 814         /*
 815          * Begin by chdir()ing into the manroot.
 816          * This way we can pick up the database files, which are
 817          * relative to the manpath root.
 818          */
 819
 820         if (-1 == chdir(req->p[(int)mr].path)) {
 821                 perror(req->p[(int)mr].path);
 822                 resp_baddb();
 823                 return;
 824         }
 825
 826         memset(&ps, 0, sizeof(struct manpaths));
 827         manpath_manconf(&ps, "etc/catman.conf");
 828
 829         if (vol >= (unsigned int)ps.sz) {
 830                 resp_error400();
 831                 goto out;
 832         }
 833
 834         sz = strlcpy(file, ps.paths[vol], MAXPATHLEN);
 835         assert(sz < MAXPATHLEN);
 836         strlcat(file, "/", MAXPATHLEN);
 837         strlcat(file, MANDOC_IDX, MAXPATHLEN);
 838
 839         /* Open the index recno(3) database. */
 840
 841         idx = dbopen(file, O_RDONLY, 0, DB_RECNO, NULL);
 842         if (NULL == idx) {
 843                 perror(file);
 844                 resp_baddb();
 845                 goto out;
 846         }
 847
 848         key.data = &rec;
 849         key.size = 4;
 850
 851         if (0 != (rc = (*idx->get)(idx, &key, &val, 0))) {
 852                 rc < 0 ? resp_baddb() : resp_error400();
 853                 goto out;
 854         } else if (0 == val.size) {
 855                 resp_baddb();
 856                 goto out;
 857         }
 858
 859         cp = (char *)val.data;
 860         catm = 'c' == *cp++;
 861
 862         if (NULL == memchr(cp, '\0', val.size - 1))
 863                 resp_baddb();
 864         else {
 865                 file[(int)sz] = '\0';
 866                 strlcat(file, "/", MAXPATHLEN);
 867                 strlcat(file, cp, MAXPATHLEN);
 868                 if (catm)
 869                         catman(req, file);
 870                 else
 871                         format(req, file);
 872         }
 873 out:
 874         if (idx)
 875                 (*idx->close)(idx);
 876         manpath_free(&ps);
 877 }
 878
 879 static void
 880 pg_search(const struct req *req, char *path)
 881 {
 882         size_t            tt, ressz;
 883         struct manpaths   ps;
 884         int               i, sz, rc;
 885         const char       *ep, *start;
 886         struct res      *res;
 887         char            **cp;
 888         struct opts       opt;
 889         struct expr      *expr;
 890
 891         if (req->q.manroot < 0 || 0 == req->psz) {
 892                 resp_search(NULL, 0, (void *)req);
 893                 return;
 894         }
 895
 896         memset(&opt, 0, sizeof(struct opts));
 897
 898         ep       = req->q.expr;
 899         opt.arch = req->q.arch;
 900         opt.cat  = req->q.sec;
 901         rc       = -1;
 902         sz       = 0;
 903         cp       = NULL;
 904         ressz    = 0;
 905         res      = NULL;
 906
 907         /*
 908          * Begin by chdir()ing into the root of the manpath.
 909          * This way we can pick up the database files, which are
 910          * relative to the manpath root.
 911          */
 912
 913         assert(req->q.manroot < (int)req->psz);
 914         if (-1 == (chdir(req->p[req->q.manroot].path))) {
 915                 perror(req->p[req->q.manroot].path);
 916                 resp_search(NULL, 0, (void *)req);
 917                 return;
 918         }
 919
 920         memset(&ps, 0, sizeof(struct manpaths));
 921         manpath_manconf(&ps, "etc/catman.conf");
 922
 923         /*
 924          * Poor man's tokenisation: just break apart by spaces.
 925          * Yes, this is half-ass.  But it works for now.
 926          */
 927
 928         while (ep && isspace((unsigned char)*ep))
 929                 ep++;
 930
 931         while (ep && '\0' != *ep) {
 932                 cp = mandoc_realloc(cp, (sz + 1) * sizeof(char *));
 933                 start = ep;
 934                 while ('\0' != *ep && ! isspace((unsigned char)*ep))
 935                         ep++;
 936                 cp[sz] = mandoc_malloc((ep - start) + 1);
 937                 memcpy(cp[sz], start, ep - start);
 938                 cp[sz++][ep - start] = '\0';
 939                 while (isspace((unsigned char)*ep))
 940                         ep++;
 941         }
 942
 943         /*
 944          * Pump down into apropos backend.
 945          * The resp_search() function is called with the results.
 946          */
 947
 948         expr = req->q.legacy ?
 949                 termcomp(sz, cp, &tt) : exprcomp(sz, cp, &tt);
 950
 951         if (NULL != expr)
 952                 rc = apropos_search
 953                         (ps.sz, ps.paths, &opt, expr, tt,
 954                          (void *)req, &ressz, &res, resp_search);
 955
 956         /* ...unless errors occured. */
 957
 958         if (0 == rc)
 959                 resp_baddb();
 960         else if (-1 == rc)
 961                 resp_search(NULL, 0, NULL);
 962
 963         for (i = 0; i < sz; i++)
 964                 free(cp[i]);
 965
 966         free(cp);
 967         resfree(res, ressz);
 968         exprfree(expr);
 969         manpath_free(&ps);
 970 }
 971
 972 int
 973 main(void)
 974 {
 975         int              i;
 976         char             buf[MAXPATHLEN];
 977         DIR             *cwd;
 978         struct req       req;
 979         char            *p, *path, *subpath;
 980
 981         /* Scan our run-time environment. */
 982
 983         if (NULL == (cache = getenv("CACHE_DIR")))
 984                 cache = "/cache/man.cgi";
 985
 986         if (NULL == (progname = getenv("SCRIPT_NAME")))
 987                 progname = "";
 988
 989         if (NULL == (css = getenv("CSS_DIR")))
 990                 css = "";
 991
 992         if (NULL == (host = getenv("HTTP_HOST")))
 993                 host = "localhost";
 994
 995         /*
 996          * First we change directory into the cache directory so that
 997          * subsequent scanning for manpath directories is rooted
 998          * relative to the same position.
 999          */
1000
1001         if (-1 == chdir(cache)) {
1002                 perror(cache);
1003                 resp_bad();
1004                 return(EXIT_FAILURE);
1005         } else if (NULL == (cwd = opendir(cache))) {
1006                 perror(cache);
1007                 resp_bad();
1008                 return(EXIT_FAILURE);
1009         }
1010
1011         memset(&req, 0, sizeof(struct req));
1012
1013         strlcpy(buf, ".", MAXPATHLEN);
1014         pathgen(cwd, buf, &req);
1015         closedir(cwd);
1016
1017         /* Next parse out the query string. */
1018
1019         if (NULL != (p = getenv("QUERY_STRING")))
1020                 http_parse(&req, p);
1021
1022         /*
1023          * Now juggle paths to extract information.
1024          * We want to extract our filetype (the file suffix), the
1025          * initial path component, then the trailing component(s).
1026          * Start with leading subpath component.
1027          */
1028
1029         subpath = path = NULL;
1030         req.page = PAGE__MAX;
1031
1032         if (NULL == (path = getenv("PATH_INFO")) || '\0' == *path)
1033                 req.page = PAGE_INDEX;
1034
1035         if (NULL != path && '/' == *path && '\0' == *++path)
1036                 req.page = PAGE_INDEX;
1037
1038         /* Strip file suffix. */
1039
1040         if (NULL != path && NULL != (p = strrchr(path, '.')))
1041                 if (NULL != p && NULL == strchr(p, '/'))
1042                         *p++ = '\0';
1043
1044         /* Resolve subpath component. */
1045
1046         if (NULL != path && NULL != (subpath = strchr(path, '/')))
1047                 *subpath++ = '\0';
1048
1049         /* Map path into one we recognise. */
1050
1051         if (NULL != path && '\0' != *path)
1052                 for (i = 0; i < (int)PAGE__MAX; i++)
1053                         if (0 == strcmp(pages[i], path)) {
1054                                 req.page = (enum page)i;
1055                                 break;
1056                         }
1057
1058         /* Route pages. */
1059
1060         switch (req.page) {
1061         case (PAGE_INDEX):
1062                 pg_index(&req, subpath);
1063                 break;
1064         case (PAGE_SEARCH):
1065                 pg_search(&req, subpath);
1066                 break;
1067         case (PAGE_SHOW):
1068                 pg_show(&req, subpath);
1069                 break;
1070         default:
1071                 resp_error404(path);
1072                 break;
1073         }
1074
1075         for (i = 0; i < (int)req.psz; i++) {
1076                 free(req.p[i].path);
1077                 free(req.p[i].name);
1078         }
1079
1080         free(req.p);
1081         return(EXIT_SUCCESS);
1082 }
1083
1084 static int
1085 cmp(const void *p1, const void *p2)
1086 {
1087
1088         return(strcasecmp(((const struct res *)p1)->title,
1089                                 ((const struct res *)p2)->title));
1090 }
1091
1092 /*
1093  * Check to see if an "etc" path consists of a catman.conf file.  If it
1094  * does, that means that the path contains a tree created by catman(8)
1095  * and should be used for indexing.
1096  */
1097 static int
1098 pathstop(DIR *dir)
1099 {
1100         struct dirent   *d;
1101
1102         while (NULL != (d = readdir(dir)))
1103                 if (DT_REG == d->d_type)
1104                         if (0 == strcmp(d->d_name, "catman.conf"))
1105                                 return(1);
1106
1107         return(0);
1108 }
1109
1110 /*
1111  * Scan for indexable paths.
1112  * This adds all paths with "etc/catman.conf" to the buffer.
1113  */
1114 static void
1115 pathgen(DIR *dir, char *path, struct req *req)
1116 {
1117         struct dirent   *d;
1118         char            *cp;
1119         DIR             *cd;
1120         int              rc;
1121         size_t           sz, ssz;
1122
1123         sz = strlcat(path, "/", MAXPATHLEN);
1124         if (sz >= MAXPATHLEN) {
1125                 fprintf(stderr, "%s: Path too long", path);
1126                 return;
1127         }
1128
1129         /*
1130          * First, scan for the "etc" directory.
1131          * If it's found, then see if it should cause us to stop.  This
1132          * happens when a catman.conf is found in the directory.
1133          */
1134
1135         rc = 0;
1136         while (0 == rc && NULL != (d = readdir(dir))) {
1137                 if (DT_DIR != d->d_type || strcmp(d->d_name, "etc"))
1138                         continue;
1139
1140                 path[(int)sz] = '\0';
1141                 ssz = strlcat(path, d->d_name, MAXPATHLEN);
1142
1143                 if (ssz >= MAXPATHLEN) {
1144                         fprintf(stderr, "%s: Path too long", path);
1145                         return;
1146                 } else if (NULL == (cd = opendir(path))) {
1147                         perror(path);
1148                         return;
1149                 }
1150
1151                 rc = pathstop(cd);
1152                 closedir(cd);
1153         }
1154
1155         if (rc > 0) {
1156                 /* This also strips the trailing slash. */
1157                 path[(int)--sz] = '\0';
1158                 req->p = mandoc_realloc
1159                         (req->p,
1160                          (req->psz + 1) * sizeof(struct paths));
1161                 /*
1162                  * Strip out the leading "./" unless we're just a ".",
1163                  * in which case use an empty string as our name.
1164                  */
1165                 req->p[(int)req->psz].path = mandoc_strdup(path);
1166                 req->p[(int)req->psz].name =
1167                         cp = mandoc_strdup(path + (1 == sz ? 1 : 2));
1168                 req->psz++;
1169                 /*
1170                  * The name is just the path with all the slashes taken
1171                  * out of it.  Simple but effective.
1172                  */
1173                 for ( ; '\0' != *cp; cp++)
1174                         if ('/' == *cp)
1175                                 *cp = ' ';
1176                 return;
1177         }
1178
1179         /*
1180          * If no etc/catman.conf was found, recursively enter child
1181          * directory and continue scanning.
1182          */
1183
1184         rewinddir(dir);
1185         while (NULL != (d = readdir(dir))) {
1186                 if (DT_DIR != d->d_type || '.' == d->d_name[0])
1187                         continue;
1188
1189                 path[(int)sz] = '\0';
1190                 ssz = strlcat(path, d->d_name, MAXPATHLEN);
1191
1192                 if (ssz >= MAXPATHLEN) {
1193                         fprintf(stderr, "%s: Path too long", path);
1194                         return;
1195                 } else if (NULL == (cd = opendir(path))) {
1196                         perror(path);
1197                         return;
1198                 }
1199
1200                 pathgen(cd, path, req);
1201                 closedir(cd);
1202         }
1203 }