contrib/mdocml/cgi.c

   1 /*      $Id: cgi.c,v 1.45 2013/06/05 02:00:26 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 #include "apropos_db.h"
  38 #include "mandoc.h"
  39 #include "mdoc.h"
  40 #include "man.h"
  41 #include "main.h"
  42 #include "manpath.h"
  43 #include "mandocdb.h"
  44
  45 #ifdef __linux__
  46 # include <db_185.h>
  47 #else
  48 # include <db.h>
  49 #endif
  50
  51 enum    page {
  52         PAGE_INDEX,
  53         PAGE_SEARCH,
  54         PAGE_SHOW,
  55         PAGE__MAX
  56 };
  57
  58 struct  paths {
  59         char            *name;
  60         char            *path;
  61 };
  62
  63 /*
  64  * A query as passed to the search function.
  65  */
  66 struct  query {
  67         const char      *arch; /* architecture */
  68         const char      *sec; /* manual section */
  69         const char      *expr; /* unparsed expression string */
  70         int              manroot; /* manroot index (or -1)*/
  71         int              legacy; /* whether legacy mode */
  72 };
  73
  74 struct  req {
  75         struct query     q;
  76         struct paths    *p;
  77         size_t           psz;
  78         enum page        page;
  79 };
  80
  81 static  int              atou(const char *, unsigned *);
  82 static  void             catman(const struct req *, const char *);
  83 static  int              cmp(const void *, const void *);
  84 static  void             format(const struct req *, const char *);
  85 static  void             html_print(const char *);
  86 static  void             html_printquery(const struct req *);
  87 static  void             html_putchar(char);
  88 static  int              http_decode(char *);
  89 static  void             http_parse(struct req *, char *);
  90 static  void             http_print(const char *);
  91 static  void             http_putchar(char);
  92 static  void             http_printquery(const struct req *);
  93 static  int              pathstop(DIR *);
  94 static  void             pathgen(DIR *, char *, struct req *);
  95 static  void             pg_index(const struct req *, char *);
  96 static  void             pg_search(const struct req *, char *);
  97 static  void             pg_show(const struct req *, char *);
  98 static  void             resp_bad(void);
  99 static  void             resp_baddb(void);
 100 static  void             resp_error400(void);
 101 static  void             resp_error404(const char *);
 102 static  void             resp_begin_html(int, const char *);
 103 static  void             resp_begin_http(int, const char *);
 104 static  void             resp_end_html(void);
 105 static  void             resp_index(const struct req *);
 106 static  void             resp_search(struct res *, size_t, void *);
 107 static  void             resp_searchform(const struct req *);
 108
 109 static  const char       *progname; /* cgi script name */
 110 static  const char       *cache; /* cache directory */
 111 static  const char       *css; /* css directory */
 112 static  const char       *host; /* hostname */
 113
 114 static  const char * const pages[PAGE__MAX] = {
 115         "index", /* PAGE_INDEX */
 116         "search", /* PAGE_SEARCH */
 117         "show", /* PAGE_SHOW */
 118 };
 119
 120 /*
 121  * This is just OpenBSD's strtol(3) suggestion.
 122  * I use it instead of strtonum(3) for portability's sake.
 123  */
 124 static int
 125 atou(const char *buf, unsigned *v)
 126 {
 127         char            *ep;
 128         long             lval;
 129
 130         errno = 0;
 131         lval = strtol(buf, &ep, 10);
 132         if (buf[0] == '\0' || *ep != '\0')
 133                 return(0);
 134         if ((errno == ERANGE && (lval == LONG_MAX ||
 135                                         lval == LONG_MIN)) ||
 136                         (lval > INT_MAX || lval < 0))
 137                 return(0);
 138
 139         *v = (unsigned int)lval;
 140         return(1);
 141 }
 142
 143 /*
 144  * Print a character, escaping HTML along the way.
 145  * This will pass non-ASCII straight to output: be warned!
 146  */
 147 static void
 148 html_putchar(char c)
 149 {
 150
 151         switch (c) {
 152         case ('"'):
 153                 printf("&quote;");
 154                 break;
 155         case ('&'):
 156                 printf("&amp;");
 157                 break;
 158         case ('>'):
 159                 printf("&gt;");
 160                 break;
 161         case ('<'):
 162                 printf("&lt;");
 163                 break;
 164         default:
 165                 putchar((unsigned char)c);
 166                 break;
 167         }
 168 }
 169 static void
 170 http_printquery(const struct req *req)
 171 {
 172
 173         printf("&expr=");
 174         http_print(req->q.expr ? req->q.expr : "");
 175         printf("&sec=");
 176         http_print(req->q.sec ? req->q.sec : "");
 177         printf("&arch=");
 178         http_print(req->q.arch ? req->q.arch : "");
 179 }
 180
 181
 182 static void
 183 html_printquery(const struct req *req)
 184 {
 185
 186         printf("&amp;expr=");
 187         html_print(req->q.expr ? req->q.expr : "");
 188         printf("&amp;sec=");
 189         html_print(req->q.sec ? req->q.sec : "");
 190         printf("&amp;arch=");
 191         html_print(req->q.arch ? req->q.arch : "");
 192 }
 193
 194 static void
 195 http_print(const char *p)
 196 {
 197
 198         if (NULL == p)
 199                 return;
 200         while ('\0' != *p)
 201                 http_putchar(*p++);
 202 }
 203
 204 /*
 205  * Call through to html_putchar().
 206  * Accepts NULL strings.
 207  */
 208 static void
 209 html_print(const char *p)
 210 {
 211
 212         if (NULL == p)
 213                 return;
 214         while ('\0' != *p)
 215                 html_putchar(*p++);
 216 }
 217
 218 /*
 219  * Parse out key-value pairs from an HTTP request variable.
 220  * This can be either a cookie or a POST/GET string, although man.cgi
 221  * uses only GET for simplicity.
 222  */
 223 static void
 224 http_parse(struct req *req, char *p)
 225 {
 226         char            *key, *val, *manroot;
 227         int              i, legacy;
 228
 229         memset(&req->q, 0, sizeof(struct query));
 230
 231         legacy = -1;
 232         manroot = NULL;
 233
 234         while ('\0' != *p) {
 235                 key = p;
 236                 val = NULL;
 237
 238                 p += (int)strcspn(p, ";&");
 239                 if ('\0' != *p)
 240                         *p++ = '\0';
 241                 if (NULL != (val = strchr(key, '=')))
 242                         *val++ = '\0';
 243
 244                 if ('\0' == *key || NULL == val || '\0' == *val)
 245                         continue;
 246
 247                 /* Just abort handling. */
 248
 249                 if ( ! http_decode(key))
 250                         break;
 251                 if (NULL != val && ! http_decode(val))
 252                         break;
 253
 254                 if (0 == strcmp(key, "expr"))
 255                         req->q.expr = val;
 256                 else if (0 == strcmp(key, "query"))
 257                         req->q.expr = val;
 258                 else if (0 == strcmp(key, "sec"))
 259                         req->q.sec = val;
 260                 else if (0 == strcmp(key, "sektion"))
 261                         req->q.sec = val;
 262                 else if (0 == strcmp(key, "arch"))
 263                         req->q.arch = val;
 264                 else if (0 == strcmp(key, "manpath"))
 265                         manroot = val;
 266                 else if (0 == strcmp(key, "apropos"))
 267                         legacy = 0 == strcmp(val, "0");
 268         }
 269
 270         /* Test for old man.cgi compatibility mode. */
 271
 272         req->q.legacy = legacy > 0;
 273
 274         /*
 275          * Section "0" means no section when in legacy mode.
 276          * For some man.cgi scripts, "default" arch is none.
 277          */
 278
 279         if (req->q.legacy && NULL != req->q.sec)
 280                 if (0 == strcmp(req->q.sec, "0"))
 281                         req->q.sec = NULL;
 282         if (req->q.legacy && NULL != req->q.arch)
 283                 if (0 == strcmp(req->q.arch, "default"))
 284                         req->q.arch = NULL;
 285
 286         /* Default to first manroot. */
 287
 288         if (NULL != manroot) {
 289                 for (i = 0; i < (int)req->psz; i++)
 290                         if (0 == strcmp(req->p[i].name, manroot))
 291                                 break;
 292                 req->q.manroot = i < (int)req->psz ? i : -1;
 293         }
 294 }
 295
 296 static void
 297 http_putchar(char c)
 298 {
 299
 300         if (isalnum((unsigned char)c)) {
 301                 putchar((unsigned char)c);
 302                 return;
 303         } else if (' ' == c) {
 304                 putchar('+');
 305                 return;
 306         }
 307         printf("%%%.2x", c);
 308 }
 309
 310 /*
 311  * HTTP-decode a string.  The standard explanation is that this turns
 312  * "%4e+foo" into "n foo" in the regular way.  This is done in-place
 313  * over the allocated string.
 314  */
 315 static int
 316 http_decode(char *p)
 317 {
 318         char             hex[3];
 319         int              c;
 320
 321         hex[2] = '\0';
 322
 323         for ( ; '\0' != *p; p++) {
 324                 if ('%' == *p) {
 325                         if ('\0' == (hex[0] = *(p + 1)))
 326                                 return(0);
 327                         if ('\0' == (hex[1] = *(p + 2)))
 328                                 return(0);
 329                         if (1 != sscanf(hex, "%x", &c))
 330                                 return(0);
 331                         if ('\0' == c)
 332                                 return(0);
 333
 334                         *p = (char)c;
 335                         memmove(p + 1, p + 3, strlen(p + 3) + 1);
 336                 } else
 337                         *p = '+' == *p ? ' ' : *p;
 338         }
 339
 340         *p = '\0';
 341         return(1);
 342 }
 343
 344 static void
 345 resp_begin_http(int code, const char *msg)
 346 {
 347
 348         if (200 != code)
 349                 printf("Status: %d %s\n", code, msg);
 350
 351         puts("Content-Type: text/html; charset=utf-8\n"
 352              "Cache-Control: no-cache\n"
 353              "Pragma: no-cache\n"
 354              "");
 355
 356         fflush(stdout);
 357 }
 358
 359 static void
 360 resp_begin_html(int code, const char *msg)
 361 {
 362
 363         resp_begin_http(code, msg);
 364
 365         printf("<!DOCTYPE HTML PUBLIC "
 366                " \"-//W3C//DTD HTML 4.01//EN\""
 367                " \"http://www.w3.org/TR/html4/strict.dtd\">\n"
 368                "<HTML>\n"
 369                "<HEAD>\n"
 370                "<META HTTP-EQUIV=\"Content-Type\""
 371                " CONTENT=\"text/html; charset=utf-8\">\n"
 372                "<LINK REL=\"stylesheet\" HREF=\"%s/man-cgi.css\""
 373                " TYPE=\"text/css\" media=\"all\">\n"
 374                "<LINK REL=\"stylesheet\" HREF=\"%s/man.css\""
 375                " TYPE=\"text/css\" media=\"all\">\n"
 376                "<TITLE>System Manpage Reference</TITLE>\n"
 377                "</HEAD>\n"
 378                "<BODY>\n"
 379                "<!-- Begin page content. //-->\n", css, css);
 380 }
 381
 382 static void
 383 resp_end_html(void)
 384 {
 385
 386         puts("</BODY>\n"
 387              "</HTML>");
 388 }
 389
 390 static void
 391 resp_searchform(const struct req *req)
 392 {
 393         int              i;
 394
 395         puts("<!-- Begin search form. //-->");
 396         printf("<DIV ID=\"mancgi\">\n"
 397                "<FORM ACTION=\"%s/search.html\" METHOD=\"get\">\n"
 398                "<FIELDSET>\n"
 399                "<LEGEND>Search Parameters</LEGEND>\n"
 400                "<INPUT TYPE=\"submit\" "
 401                " VALUE=\"Search\"> for manuals satisfying \n"
 402                "<INPUT TYPE=\"text\" NAME=\"expr\" VALUE=\"",
 403                progname);
 404         html_print(req->q.expr ? req->q.expr : "");
 405         printf("\">, section "
 406                "<INPUT TYPE=\"text\""
 407                " SIZE=\"4\" NAME=\"sec\" VALUE=\"");
 408         html_print(req->q.sec ? req->q.sec : "");
 409         printf("\">, arch "
 410                "<INPUT TYPE=\"text\""
 411                " SIZE=\"8\" NAME=\"arch\" VALUE=\"");
 412         html_print(req->q.arch ? req->q.arch : "");
 413         printf("\">");
 414         if (req->psz > 1) {
 415                 puts(", <SELECT NAME=\"manpath\">");
 416                 for (i = 0; i < (int)req->psz; i++) {
 417                         printf("<OPTION %s VALUE=\"",
 418                                 (i == req->q.manroot) ||
 419                                 (0 == i && -1 == req->q.manroot) ?
 420                                 "SELECTED=\"selected\"" : "");
 421                         html_print(req->p[i].name);
 422                         printf("\">");
 423                         html_print(req->p[i].name);
 424                         puts("</OPTION>");
 425                 }
 426                 puts("</SELECT>");
 427         }
 428         puts(".\n"
 429              "<INPUT TYPE=\"reset\" VALUE=\"Reset\">\n"
 430              "</FIELDSET>\n"
 431              "</FORM>\n"
 432              "</DIV>");
 433         puts("<!-- End search form. //-->");
 434 }
 435
 436 static void
 437 resp_index(const struct req *req)
 438 {
 439
 440         resp_begin_html(200, NULL);
 441         resp_searchform(req);
 442         resp_end_html();
 443 }
 444
 445 static void
 446 resp_error400(void)
 447 {
 448
 449         resp_begin_html(400, "Query Malformed");
 450         printf("<H1>Malformed Query</H1>\n"
 451                "<P>\n"
 452                "The query your entered was malformed.\n"
 453                "Try again from the\n"
 454                "<A HREF=\"%s/index.html\">main page</A>.\n"
 455                "</P>", progname);
 456         resp_end_html();
 457 }
 458
 459 static void
 460 resp_error404(const char *page)
 461 {
 462
 463         resp_begin_html(404, "Not Found");
 464         puts("<H1>Page Not Found</H1>\n"
 465              "<P>\n"
 466              "The page you're looking for, ");
 467         printf("<B>");
 468         html_print(page);
 469         printf("</B>,\n"
 470                "could not be found.\n"
 471                "Try searching from the\n"
 472                "<A HREF=\"%s/index.html\">main page</A>.\n"
 473                "</P>", progname);
 474         resp_end_html();
 475 }
 476
 477 static void
 478 resp_bad(void)
 479 {
 480         resp_begin_html(500, "Internal Server Error");
 481         puts("<P>Generic badness happened.</P>");
 482         resp_end_html();
 483 }
 484
 485 static void
 486 resp_baddb(void)
 487 {
 488
 489         resp_begin_html(500, "Internal Server Error");
 490         puts("<P>Your database is broken.</P>");
 491         resp_end_html();
 492 }
 493
 494 static void
 495 resp_search(struct res *r, size_t sz, void *arg)
 496 {
 497         size_t           i, matched;
 498         const struct req *req;
 499
 500         req = (const struct req *)arg;
 501
 502         if (sz > 0)
 503                 assert(req->q.manroot >= 0);
 504
 505         for (matched = i = 0; i < sz; i++)
 506                 if (r[i].matched)
 507                         matched++;
 508
 509         if (1 == matched) {
 510                 for (i = 0; i < sz; i++)
 511                         if (r[i].matched)
 512                                 break;
 513                 /*
 514                  * If we have just one result, then jump there now
 515                  * without any delay.
 516                  */
 517                 puts("Status: 303 See Other");
 518                 printf("Location: http://%s%s/show/%d/%u/%u.html?",
 519                                 host, progname, req->q.manroot,
 520                                 r[i].volume, r[i].rec);
 521                 http_printquery(req);
 522                 puts("\n"
 523                      "Content-Type: text/html; charset=utf-8\n");
 524                 return;
 525         }
 526
 527         resp_begin_html(200, NULL);
 528         resp_searchform(req);
 529
 530         puts("<DIV CLASS=\"results\">");
 531
 532         if (0 == matched) {
 533                 puts("<P>\n"
 534                      "No results found.\n"
 535                      "</P>\n"
 536                      "</DIV>");
 537                 resp_end_html();
 538                 return;
 539         }
 540
 541         qsort(r, sz, sizeof(struct res), cmp);
 542
 543         puts("<TABLE>");
 544
 545         for (i = 0; i < sz; i++) {
 546                 if ( ! r[i].matched)
 547                         continue;
 548                 printf("<TR>\n"
 549                        "<TD CLASS=\"title\">\n"
 550                        "<A HREF=\"%s/show/%d/%u/%u.html?",
 551                                 progname, req->q.manroot,
 552                                 r[i].volume, r[i].rec);
 553                 html_printquery(req);
 554                 printf("\">");
 555                 html_print(r[i].title);
 556                 putchar('(');
 557                 html_print(r[i].cat);
 558                 if (r[i].arch && '\0' != *r[i].arch) {
 559                         putchar('/');
 560                         html_print(r[i].arch);
 561                 }
 562                 printf(")</A>\n"
 563                        "</TD>\n"
 564                        "<TD CLASS=\"desc\">");
 565                 html_print(r[i].desc);
 566                 puts("</TD>\n"
 567                      "</TR>");
 568         }
 569
 570         puts("</TABLE>\n"
 571              "</DIV>");
 572         resp_end_html();
 573 }
 574
 575 /* ARGSUSED */
 576 static void
 577 pg_index(const struct req *req, char *path)
 578 {
 579
 580         resp_index(req);
 581 }
 582
 583 static void
 584 catman(const struct req *req, const char *file)
 585 {
 586         FILE            *f;
 587         size_t           len;
 588         int              i;
 589         char            *p;
 590         int              italic, bold;
 591
 592         if (NULL == (f = fopen(file, "r"))) {
 593                 resp_baddb();
 594                 return;
 595         }
 596
 597         resp_begin_html(200, NULL);
 598         resp_searchform(req);
 599         puts("<DIV CLASS=\"catman\">\n"
 600              "<PRE>");
 601
 602         while (NULL != (p = fgetln(f, &len))) {
 603                 bold = italic = 0;
 604                 for (i = 0; i < (int)len - 1; i++) {
 605                         /*
 606                          * This means that the catpage is out of state.
 607                          * Ignore it and keep going (although the
 608                          * catpage is bogus).
 609                          */
 610
 611                         if ('\b' == p[i] || '\n' == p[i])
 612                                 continue;
 613
 614                         /*
 615                          * Print a regular character.
 616                          * Close out any bold/italic scopes.
 617                          * If we're in back-space mode, make sure we'll
 618                          * have something to enter when we backspace.
 619                          */
 620
 621                         if ('\b' != p[i + 1]) {
 622                                 if (italic)
 623                                         printf("</I>");
 624                                 if (bold)
 625                                         printf("</B>");
 626                                 italic = bold = 0;
 627                                 html_putchar(p[i]);
 628                                 continue;
 629                         } else if (i + 2 >= (int)len)
 630                                 continue;
 631
 632                         /* Italic mode. */
 633
 634                         if ('_' == p[i]) {
 635                                 if (bold)
 636                                         printf("</B>");
 637                                 if ( ! italic)
 638                                         printf("<I>");
 639                                 bold = 0;
 640                                 italic = 1;
 641                                 i += 2;
 642                                 html_putchar(p[i]);
 643                                 continue;
 644                         }
 645
 646                         /*
 647                          * Handle funny behaviour troff-isms.
 648                          * These grok'd from the original man2html.c.
 649                          */
 650
 651                         if (('+' == p[i] && 'o' == p[i + 2]) ||
 652                                         ('o' == p[i] && '+' == p[i + 2]) ||
 653                                         ('|' == 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                                 if (italic)
 660                                         printf("</I>");
 661                                 if (bold)
 662                                         printf("</B>");
 663                                 italic = bold = 0;
 664                                 putchar('*');
 665                                 i += 2;
 666                                 continue;
 667                         } else if (('|' == p[i] && '-' == p[i + 2]) ||
 668                                         ('-' == p[i] && '|' == p[i + 1]) ||
 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                                 if (italic)
 674                                         printf("</I>");
 675                                 if (bold)
 676                                         printf("</B>");
 677                                 italic = bold = 0;
 678                                 putchar('+');
 679                                 i += 2;
 680                                 continue;
 681                         }
 682
 683                         /* Bold mode. */
 684
 685                         if (italic)
 686                                 printf("</I>");
 687                         if ( ! bold)
 688                                 printf("<B>");
 689                         bold = 1;
 690                         italic = 0;
 691                         i += 2;
 692                         html_putchar(p[i]);
 693                 }
 694
 695                 /*
 696                  * Clean up the last character.
 697                  * We can get to a newline; don't print that.
 698                  */
 699
 700                 if (italic)
 701                         printf("</I>");
 702                 if (bold)
 703                         printf("</B>");
 704
 705                 if (i == (int)len - 1 && '\n' != p[i])
 706                         html_putchar(p[i]);
 707
 708                 putchar('\n');
 709         }
 710
 711         puts("</PRE>\n"
 712              "</DIV>\n"
 713              "</BODY>\n"
 714              "</HTML>");
 715
 716         fclose(f);
 717 }
 718
 719 static void
 720 format(const struct req *req, const char *file)
 721 {
 722         struct mparse   *mp;
 723         int              fd;
 724         struct mdoc     *mdoc;
 725         struct man      *man;
 726         void            *vp;
 727         enum mandoclevel rc;
 728         char             opts[PATH_MAX + 128];
 729
 730         if (-1 == (fd = open(file, O_RDONLY, 0))) {
 731                 resp_baddb();
 732                 return;
 733         }
 734
 735         mp = mparse_alloc(MPARSE_AUTO, MANDOCLEVEL_FATAL, NULL, NULL, NULL);
 736         rc = mparse_readfd(mp, fd, file);
 737         close(fd);
 738
 739         if (rc >= MANDOCLEVEL_FATAL) {
 740                 resp_baddb();
 741                 return;
 742         }
 743
 744         snprintf(opts, sizeof(opts), "fragment,"
 745                         "man=%s/search.html?sec=%%S&expr=Nm~^%%N$,"
 746                         /*"includes=/cgi-bin/man.cgi/usr/include/%%I"*/,
 747                         progname);
 748
 749         mparse_result(mp, &mdoc, &man);
 750         if (NULL == man && NULL == mdoc) {
 751                 resp_baddb();
 752                 mparse_free(mp);
 753                 return;
 754         }
 755
 756         resp_begin_html(200, NULL);
 757         resp_searchform(req);
 758
 759         vp = html_alloc(opts);
 760
 761         if (NULL != mdoc)
 762                 html_mdoc(vp, mdoc);
 763         else
 764                 html_man(vp, man);
 765
 766         puts("</BODY>\n"
 767              "</HTML>");
 768
 769         html_free(vp);
 770         mparse_free(mp);
 771 }
 772
 773 static void
 774 pg_show(const struct req *req, char *path)
 775 {
 776         struct manpaths  ps;
 777         size_t           sz;
 778         char            *sub;
 779         char             file[PATH_MAX];
 780         const char      *cp;
 781         int              rc, catm;
 782         unsigned int     vol, rec, mr;
 783         DB              *idx;
 784         DBT              key, val;
 785
 786         idx = NULL;
 787
 788         /* Parse out mroot, volume, and record from the path. */
 789
 790         if (NULL == path || NULL == (sub = strchr(path, '/'))) {
 791                 resp_error400();
 792                 return;
 793         }
 794         *sub++ = '\0';
 795         if ( ! atou(path, &mr)) {
 796                 resp_error400();
 797                 return;
 798         }
 799         path = sub;
 800         if (NULL == (sub = strchr(path, '/'))) {
 801                 resp_error400();
 802                 return;
 803         }
 804         *sub++ = '\0';
 805         if ( ! atou(path, &vol) || ! atou(sub, &rec)) {
 806                 resp_error400();
 807                 return;
 808         } else if (mr >= (unsigned int)req->psz) {
 809                 resp_error400();
 810                 return;
 811         }
 812
 813         /*
 814          * Begin by chdir()ing into the manroot.
 815          * This way we can pick up the database files, which are
 816          * relative to the manpath root.
 817          */
 818
 819         if (-1 == chdir(req->p[(int)mr].path)) {
 820                 perror(req->p[(int)mr].path);
 821                 resp_baddb();
 822                 return;
 823         }
 824
 825         memset(&ps, 0, sizeof(struct manpaths));
 826         manpath_manconf(&ps, "etc/catman.conf");
 827
 828         if (vol >= (unsigned int)ps.sz) {
 829                 resp_error400();
 830                 goto out;
 831         }
 832
 833         sz = strlcpy(file, ps.paths[vol], PATH_MAX);
 834         assert(sz < PATH_MAX);
 835         strlcat(file, "/", PATH_MAX);
 836         strlcat(file, MANDOC_IDX, PATH_MAX);
 837
 838         /* Open the index recno(3) database. */
 839
 840         idx = dbopen(file, O_RDONLY, 0, DB_RECNO, NULL);
 841         if (NULL == idx) {
 842                 perror(file);
 843                 resp_baddb();
 844                 goto out;
 845         }
 846
 847         key.data = &rec;
 848         key.size = 4;
 849
 850         if (0 != (rc = (*idx->get)(idx, &key, &val, 0))) {
 851                 rc < 0 ? resp_baddb() : resp_error400();
 852                 goto out;
 853         } else if (0 == val.size) {
 854                 resp_baddb();
 855                 goto out;
 856         }
 857
 858         cp = (char *)val.data;
 859         catm = 'c' == *cp++;
 860
 861         if (NULL == memchr(cp, '\0', val.size - 1))
 862                 resp_baddb();
 863         else {
 864                 file[(int)sz] = '\0';
 865                 strlcat(file, "/", PATH_MAX);
 866                 strlcat(file, cp, PATH_MAX);
 867                 if (catm)
 868                         catman(req, file);
 869                 else
 870                         format(req, file);
 871         }
 872 out:
 873         if (idx)
 874                 (*idx->close)(idx);
 875         manpath_free(&ps);
 876 }
 877
 878 static void
 879 pg_search(const struct req *req, char *path)
 880 {
 881         size_t            tt, ressz;
 882         struct manpaths   ps;
 883         int               i, sz, rc;
 884         const char       *ep, *start;
 885         struct res      *res;
 886         char            **cp;
 887         struct opts       opt;
 888         struct expr      *expr;
 889
 890         if (req->q.manroot < 0 || 0 == req->psz) {
 891                 resp_search(NULL, 0, (void *)req);
 892                 return;
 893         }
 894
 895         memset(&opt, 0, sizeof(struct opts));
 896
 897         ep       = req->q.expr;
 898         opt.arch = req->q.arch;
 899         opt.cat  = req->q.sec;
 900         rc       = -1;
 901         sz       = 0;
 902         cp       = NULL;
 903         ressz    = 0;
 904         res      = NULL;
 905
 906         /*
 907          * Begin by chdir()ing into the root of the manpath.
 908          * This way we can pick up the database files, which are
 909          * relative to the manpath root.
 910          */
 911
 912         assert(req->q.manroot < (int)req->psz);
 913         if (-1 == (chdir(req->p[req->q.manroot].path))) {
 914                 perror(req->p[req->q.manroot].path);
 915                 resp_search(NULL, 0, (void *)req);
 916                 return;
 917         }
 918
 919         memset(&ps, 0, sizeof(struct manpaths));
 920         manpath_manconf(&ps, "etc/catman.conf");
 921
 922         /*
 923          * Poor man's tokenisation: just break apart by spaces.
 924          * Yes, this is half-ass.  But it works for now.
 925          */
 926
 927         while (ep && isspace((unsigned char)*ep))
 928                 ep++;
 929
 930         while (ep && '\0' != *ep) {
 931                 cp = mandoc_realloc(cp, (sz + 1) * sizeof(char *));
 932                 start = ep;
 933                 while ('\0' != *ep && ! isspace((unsigned char)*ep))
 934                         ep++;
 935                 cp[sz] = mandoc_malloc((ep - start) + 1);
 936                 memcpy(cp[sz], start, ep - start);
 937                 cp[sz++][ep - start] = '\0';
 938                 while (isspace((unsigned char)*ep))
 939                         ep++;
 940         }
 941
 942         /*
 943          * Pump down into apropos backend.
 944          * The resp_search() function is called with the results.
 945          */
 946
 947         expr = req->q.legacy ?
 948                 termcomp(sz, cp, &tt) : exprcomp(sz, cp, &tt);
 949
 950         if (NULL != expr)
 951                 rc = apropos_search
 952                         (ps.sz, ps.paths, &opt, expr, tt,
 953                          (void *)req, &ressz, &res, resp_search);
 954
 955         /* ...unless errors occured. */
 956
 957         if (0 == rc)
 958                 resp_baddb();
 959         else if (-1 == rc)
 960                 resp_search(NULL, 0, NULL);
 961
 962         for (i = 0; i < sz; i++)
 963                 free(cp[i]);
 964
 965         free(cp);
 966         resfree(res, ressz);
 967         exprfree(expr);
 968         manpath_free(&ps);
 969 }
 970
 971 int
 972 main(void)
 973 {
 974         int              i;
 975         char             buf[PATH_MAX];
 976         DIR             *cwd;
 977         struct req       req;
 978         char            *p, *path, *subpath;
 979
 980         /* Scan our run-time environment. */
 981
 982         if (NULL == (cache = getenv("CACHE_DIR")))
 983                 cache = "/cache/man.cgi";
 984
 985         if (NULL == (progname = getenv("SCRIPT_NAME")))
 986                 progname = "";
 987
 988         if (NULL == (css = getenv("CSS_DIR")))
 989                 css = "";
 990
 991         if (NULL == (host = getenv("HTTP_HOST")))
 992                 host = "localhost";
 993
 994         /*
 995          * First we change directory into the cache directory so that
 996          * subsequent scanning for manpath directories is rooted
 997          * relative to the same position.
 998          */
 999
1000         if (-1 == chdir(cache)) {
1001                 perror(cache);
1002                 resp_bad();
1003                 return(EXIT_FAILURE);
1004         } else if (NULL == (cwd = opendir(cache))) {
1005                 perror(cache);
1006                 resp_bad();
1007                 return(EXIT_FAILURE);
1008         }
1009
1010         memset(&req, 0, sizeof(struct req));
1011
1012         strlcpy(buf, ".", PATH_MAX);
1013         pathgen(cwd, buf, &req);
1014         closedir(cwd);
1015
1016         /* Next parse out the query string. */
1017
1018         if (NULL != (p = getenv("QUERY_STRING")))
1019                 http_parse(&req, p);
1020
1021         /*
1022          * Now juggle paths to extract information.
1023          * We want to extract our filetype (the file suffix), the
1024          * initial path component, then the trailing component(s).
1025          * Start with leading subpath component.
1026          */
1027
1028         subpath = path = NULL;
1029         req.page = PAGE__MAX;
1030
1031         if (NULL == (path = getenv("PATH_INFO")) || '\0' == *path)
1032                 req.page = PAGE_INDEX;
1033
1034         if (NULL != path && '/' == *path && '\0' == *++path)
1035                 req.page = PAGE_INDEX;
1036
1037         /* Strip file suffix. */
1038
1039         if (NULL != path && NULL != (p = strrchr(path, '.')))
1040                 if (NULL != p && NULL == strchr(p, '/'))
1041                         *p++ = '\0';
1042
1043         /* Resolve subpath component. */
1044
1045         if (NULL != path && NULL != (subpath = strchr(path, '/')))
1046                 *subpath++ = '\0';
1047
1048         /* Map path into one we recognise. */
1049
1050         if (NULL != path && '\0' != *path)
1051                 for (i = 0; i < (int)PAGE__MAX; i++)
1052                         if (0 == strcmp(pages[i], path)) {
1053                                 req.page = (enum page)i;
1054                                 break;
1055                         }
1056
1057         /* Route pages. */
1058
1059         switch (req.page) {
1060         case (PAGE_INDEX):
1061                 pg_index(&req, subpath);
1062                 break;
1063         case (PAGE_SEARCH):
1064                 pg_search(&req, subpath);
1065                 break;
1066         case (PAGE_SHOW):
1067                 pg_show(&req, subpath);
1068                 break;
1069         default:
1070                 resp_error404(path);
1071                 break;
1072         }
1073
1074         for (i = 0; i < (int)req.psz; i++) {
1075                 free(req.p[i].path);
1076                 free(req.p[i].name);
1077         }
1078
1079         free(req.p);
1080         return(EXIT_SUCCESS);
1081 }
1082
1083 static int
1084 cmp(const void *p1, const void *p2)
1085 {
1086
1087         return(strcasecmp(((const struct res *)p1)->title,
1088                                 ((const struct res *)p2)->title));
1089 }
1090
1091 /*
1092  * Check to see if an "etc" path consists of a catman.conf file.  If it
1093  * does, that means that the path contains a tree created by catman(8)
1094  * and should be used for indexing.
1095  */
1096 static int
1097 pathstop(DIR *dir)
1098 {
1099         struct dirent   *d;
1100
1101         while (NULL != (d = readdir(dir)))
1102                 if (DT_REG == d->d_type)
1103                         if (0 == strcmp(d->d_name, "catman.conf"))
1104                                 return(1);
1105
1106         return(0);
1107 }
1108
1109 /*
1110  * Scan for indexable paths.
1111  * This adds all paths with "etc/catman.conf" to the buffer.
1112  */
1113 static void
1114 pathgen(DIR *dir, char *path, struct req *req)
1115 {
1116         struct dirent   *d;
1117         char            *cp;
1118         DIR             *cd;
1119         int              rc;
1120         size_t           sz, ssz;
1121
1122         sz = strlcat(path, "/", PATH_MAX);
1123         if (sz >= PATH_MAX) {
1124                 fprintf(stderr, "%s: Path too long", path);
1125                 return;
1126         }
1127
1128         /*
1129          * First, scan for the "etc" directory.
1130          * If it's found, then see if it should cause us to stop.  This
1131          * happens when a catman.conf is found in the directory.
1132          */
1133
1134         rc = 0;
1135         while (0 == rc && NULL != (d = readdir(dir))) {
1136                 if (DT_DIR != d->d_type || strcmp(d->d_name, "etc"))
1137                         continue;
1138
1139                 path[(int)sz] = '\0';
1140                 ssz = strlcat(path, d->d_name, PATH_MAX);
1141
1142                 if (ssz >= PATH_MAX) {
1143                         fprintf(stderr, "%s: Path too long", path);
1144                         return;
1145                 } else if (NULL == (cd = opendir(path))) {
1146                         perror(path);
1147                         return;
1148                 }
1149
1150                 rc = pathstop(cd);
1151                 closedir(cd);
1152         }
1153
1154         if (rc > 0) {
1155                 /* This also strips the trailing slash. */
1156                 path[(int)--sz] = '\0';
1157                 req->p = mandoc_realloc
1158                         (req->p,
1159                          (req->psz + 1) * sizeof(struct paths));
1160                 /*
1161                  * Strip out the leading "./" unless we're just a ".",
1162                  * in which case use an empty string as our name.
1163                  */
1164                 req->p[(int)req->psz].path = mandoc_strdup(path);
1165                 req->p[(int)req->psz].name =
1166                         cp = mandoc_strdup(path + (1 == sz ? 1 : 2));
1167                 req->psz++;
1168                 /*
1169                  * The name is just the path with all the slashes taken
1170                  * out of it.  Simple but effective.
1171                  */
1172                 for ( ; '\0' != *cp; cp++)
1173                         if ('/' == *cp)
1174                                 *cp = ' ';
1175                 return;
1176         }
1177
1178         /*
1179          * If no etc/catman.conf was found, recursively enter child
1180          * directory and continue scanning.
1181          */
1182
1183         rewinddir(dir);
1184         while (NULL != (d = readdir(dir))) {
1185                 if (DT_DIR != d->d_type || '.' == d->d_name[0])
1186                         continue;
1187
1188                 path[(int)sz] = '\0';
1189                 ssz = strlcat(path, d->d_name, PATH_MAX);
1190
1191                 if (ssz >= PATH_MAX) {
1192                         fprintf(stderr, "%s: Path too long", path);
1193                         return;
1194                 } else if (NULL == (cd = opendir(path))) {
1195                         perror(path);
1196                         return;
1197                 }
1198
1199                 pathgen(cd, path, req);
1200                 closedir(cd);
1201         }
1202 }