Annotation of mandoc/cgi.c, Revision 1.131
1.131 ! schwarze 1: /* $Id: cgi.c,v 1.130 2016/04/28 17:59:14 schwarze Exp $ */
1.6 kristaps 2: /*
1.42 kristaps 3: * Copyright (c) 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
1.118 schwarze 4: * Copyright (c) 2014, 2015, 2016 Ingo Schwarze <schwarze@usta.de>
1.6 kristaps 5: *
6: * Permission to use, copy, modify, and distribute this software for any
7: * purpose with or without fee is hereby granted, provided that the above
8: * copyright notice and this permission notice appear in all copies.
9: *
1.105 schwarze 10: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1.6 kristaps 11: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1.105 schwarze 12: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1.6 kristaps 13: * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
14: * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
15: * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
16: * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
17: */
18: #include "config.h"
1.93 schwarze 19:
20: #include <sys/types.h>
1.95 schwarze 21: #include <sys/time.h>
1.6 kristaps 22:
23: #include <ctype.h>
1.128 schwarze 24: #include <err.h>
1.58 schwarze 25: #include <errno.h>
1.1 kristaps 26: #include <fcntl.h>
1.6 kristaps 27: #include <limits.h>
1.92 schwarze 28: #include <stdint.h>
1.1 kristaps 29: #include <stdio.h>
30: #include <stdlib.h>
31: #include <string.h>
1.6 kristaps 32: #include <unistd.h>
1.1 kristaps 33:
1.108 schwarze 34: #include "mandoc_aux.h"
1.4 schwarze 35: #include "mandoc.h"
1.108 schwarze 36: #include "roff.h"
1.111 schwarze 37: #include "mdoc.h"
1.112 schwarze 38: #include "man.h"
1.8 kristaps 39: #include "main.h"
1.105 schwarze 40: #include "manconf.h"
1.52 schwarze 41: #include "mansearch.h"
1.67 schwarze 42: #include "cgi.h"
1.1 kristaps 43:
1.20 kristaps 44: /*
45: * A query as passed to the search function.
46: */
47: struct query {
1.83 schwarze 48: char *manpath; /* desired manual directory */
49: char *arch; /* architecture */
50: char *sec; /* manual section */
1.85 schwarze 51: char *query; /* unparsed query expression */
1.65 schwarze 52: int equal; /* match whole names, not substrings */
1.20 kristaps 53: };
54:
1.1 kristaps 55: struct req {
1.58 schwarze 56: struct query q;
57: char **p; /* array of available manpaths */
58: size_t psz; /* number of available manpaths */
1.121 schwarze 59: int isquery; /* QUERY_STRING used, not PATH_INFO */
1.1 kristaps 60: };
61:
1.131 ! schwarze 62: enum focus {
! 63: FOCUS_NONE = 0,
! 64: FOCUS_QUERY
! 65: };
! 66:
1.6 kristaps 67: static void html_print(const char *);
1.10 kristaps 68: static void html_putchar(char);
1.104 schwarze 69: static int http_decode(char *);
1.129 schwarze 70: static void parse_manpath_conf(struct req *);
71: static void parse_path_info(struct req *req, const char *path);
72: static void parse_query_string(struct req *, const char *);
1.72 schwarze 73: static void pg_error_badrequest(const char *);
74: static void pg_error_internal(void);
75: static void pg_index(const struct req *);
76: static void pg_noresult(const struct req *, const char *);
1.66 schwarze 77: static void pg_search(const struct req *);
1.72 schwarze 78: static void pg_searchres(const struct req *,
79: struct manpage *, size_t);
1.79 schwarze 80: static void pg_show(struct req *, const char *);
1.6 kristaps 81: static void resp_begin_html(int, const char *);
82: static void resp_begin_http(int, const char *);
1.129 schwarze 83: static void resp_catman(const struct req *, const char *);
1.114 schwarze 84: static void resp_copy(const char *);
1.6 kristaps 85: static void resp_end_html(void);
1.129 schwarze 86: static void resp_format(const struct req *, const char *);
1.131 ! schwarze 87: static void resp_searchform(const struct req *, enum focus);
1.70 schwarze 88: static void resp_show(const struct req *, const char *);
1.85 schwarze 89: static void set_query_attr(char **, char **);
90: static int validate_filename(const char *);
91: static int validate_manpath(const struct req *, const char *);
92: static int validate_urifrag(const char *);
1.6 kristaps 93:
1.119 schwarze 94: static const char *scriptname = SCRIPT_NAME;
1.1 kristaps 95:
1.70 schwarze 96: static const int sec_prios[] = {1, 4, 5, 8, 6, 3, 7, 2, 9};
1.68 schwarze 97: static const char *const sec_numbers[] = {
98: "0", "1", "2", "3", "3p", "4", "5", "6", "7", "8", "9"
99: };
100: static const char *const sec_names[] = {
101: "All Sections",
102: "1 - General Commands",
103: "2 - System Calls",
1.96 schwarze 104: "3 - Library Functions",
105: "3p - Perl Library",
106: "4 - Device Drivers",
1.68 schwarze 107: "5 - File Formats",
108: "6 - Games",
1.96 schwarze 109: "7 - Miscellaneous Information",
110: "8 - System Manager\'s Manual",
111: "9 - Kernel Developer\'s Manual"
1.68 schwarze 112: };
113: static const int sec_MAX = sizeof(sec_names) / sizeof(char *);
114:
115: static const char *const arch_names[] = {
116: "amd64", "alpha", "armish", "armv7",
1.125 schwarze 117: "hppa", "hppa64", "i386", "landisk",
118: "loongson", "luna88k", "macppc", "mips64",
119: "octeon", "sgi", "socppc", "sparc",
120: "sparc64", "zaurus",
1.68 schwarze 121: "amiga", "arc", "arm32", "atari",
1.125 schwarze 122: "aviion", "beagle", "cats", "hp300",
123: "ia64", "mac68k", "mvme68k", "mvme88k",
124: "mvmeppc", "palm", "pc532", "pegasos",
125: "pmax", "powerpc", "solbourne", "sun3",
126: "vax", "wgrisc", "x68k"
1.68 schwarze 127: };
128: static const int arch_MAX = sizeof(arch_names) / sizeof(char *);
129:
1.6 kristaps 130: /*
1.20 kristaps 131: * Print a character, escaping HTML along the way.
132: * This will pass non-ASCII straight to output: be warned!
133: */
1.10 kristaps 134: static void
135: html_putchar(char c)
136: {
137:
138: switch (c) {
139: case ('"'):
140: printf(""e;");
141: break;
142: case ('&'):
143: printf("&");
144: break;
145: case ('>'):
146: printf(">");
147: break;
148: case ('<'):
149: printf("<");
150: break;
151: default:
152: putchar((unsigned char)c);
153: break;
154: }
155: }
1.57 schwarze 156:
1.6 kristaps 157: /*
1.20 kristaps 158: * Call through to html_putchar().
159: * Accepts NULL strings.
1.6 kristaps 160: */
1.1 kristaps 161: static void
1.6 kristaps 162: html_print(const char *p)
1.1 kristaps 163: {
1.104 schwarze 164:
1.6 kristaps 165: if (NULL == p)
166: return;
1.1 kristaps 167: while ('\0' != *p)
1.10 kristaps 168: html_putchar(*p++);
1.1 kristaps 169: }
170:
171: /*
1.83 schwarze 172: * Transfer the responsibility for the allocated string *val
173: * to the query structure.
1.1 kristaps 174: */
175: static void
1.83 schwarze 176: set_query_attr(char **attr, char **val)
1.1 kristaps 177: {
178:
1.83 schwarze 179: free(*attr);
180: if (**val == '\0') {
181: *attr = NULL;
182: free(*val);
183: } else
184: *attr = *val;
185: *val = NULL;
186: }
187:
188: /*
189: * Parse the QUERY_STRING for key-value pairs
190: * and store the values into the query structure.
191: */
192: static void
1.129 schwarze 193: parse_query_string(struct req *req, const char *qs)
1.83 schwarze 194: {
195: char *key, *val;
196: size_t keysz, valsz;
197:
1.121 schwarze 198: req->isquery = 1;
1.83 schwarze 199: req->q.manpath = NULL;
200: req->q.arch = NULL;
201: req->q.sec = NULL;
1.85 schwarze 202: req->q.query = NULL;
1.83 schwarze 203: req->q.equal = 1;
204:
205: key = val = NULL;
206: while (*qs != '\0') {
1.24 kristaps 207:
1.83 schwarze 208: /* Parse one key. */
209:
210: keysz = strcspn(qs, "=;&");
211: key = mandoc_strndup(qs, keysz);
212: qs += keysz;
213: if (*qs != '=')
214: goto next;
215:
216: /* Parse one value. */
217:
218: valsz = strcspn(++qs, ";&");
219: val = mandoc_strndup(qs, valsz);
220: qs += valsz;
221:
222: /* Decode and catch encoding errors. */
1.1 kristaps 223:
1.83 schwarze 224: if ( ! (http_decode(key) && http_decode(val)))
225: goto next;
1.1 kristaps 226:
1.83 schwarze 227: /* Handle key-value pairs. */
1.1 kristaps 228:
1.83 schwarze 229: if ( ! strcmp(key, "query"))
1.85 schwarze 230: set_query_attr(&req->q.query, &val);
1.1 kristaps 231:
1.83 schwarze 232: else if ( ! strcmp(key, "apropos"))
233: req->q.equal = !strcmp(val, "0");
234:
235: else if ( ! strcmp(key, "manpath")) {
1.73 schwarze 236: #ifdef COMPAT_OLDURI
1.83 schwarze 237: if ( ! strncmp(val, "OpenBSD ", 8)) {
1.73 schwarze 238: val[7] = '-';
239: if ('C' == val[8])
240: val[8] = 'c';
241: }
242: #endif
1.83 schwarze 243: set_query_attr(&req->q.manpath, &val);
244: }
245:
246: else if ( ! (strcmp(key, "sec")
1.73 schwarze 247: #ifdef COMPAT_OLDURI
1.83 schwarze 248: && strcmp(key, "sektion")
1.73 schwarze 249: #endif
1.83 schwarze 250: )) {
251: if ( ! strcmp(val, "0"))
252: *val = '\0';
253: set_query_attr(&req->q.sec, &val);
1.65 schwarze 254: }
1.83 schwarze 255:
256: else if ( ! strcmp(key, "arch")) {
257: if ( ! strcmp(val, "default"))
258: *val = '\0';
259: set_query_attr(&req->q.arch, &val);
260: }
261:
262: /*
263: * The key must be freed in any case.
264: * The val may have been handed over to the query
265: * structure, in which case it is now NULL.
266: */
267: next:
268: free(key);
269: key = NULL;
270: free(val);
271: val = NULL;
272:
273: if (*qs != '\0')
274: qs++;
1.24 kristaps 275: }
1.1 kristaps 276: }
277:
278: /*
1.6 kristaps 279: * HTTP-decode a string. The standard explanation is that this turns
280: * "%4e+foo" into "n foo" in the regular way. This is done in-place
281: * over the allocated string.
1.1 kristaps 282: */
283: static int
1.24 kristaps 284: http_decode(char *p)
1.1 kristaps 285: {
286: char hex[3];
1.63 schwarze 287: char *q;
1.1 kristaps 288: int c;
289:
290: hex[2] = '\0';
291:
1.63 schwarze 292: q = p;
293: for ( ; '\0' != *p; p++, q++) {
1.1 kristaps 294: if ('%' == *p) {
295: if ('\0' == (hex[0] = *(p + 1)))
1.109 schwarze 296: return 0;
1.1 kristaps 297: if ('\0' == (hex[1] = *(p + 2)))
1.109 schwarze 298: return 0;
1.1 kristaps 299: if (1 != sscanf(hex, "%x", &c))
1.109 schwarze 300: return 0;
1.1 kristaps 301: if ('\0' == c)
1.109 schwarze 302: return 0;
1.1 kristaps 303:
1.63 schwarze 304: *q = (char)c;
305: p += 2;
1.1 kristaps 306: } else
1.63 schwarze 307: *q = '+' == *p ? ' ' : *p;
1.1 kristaps 308: }
309:
1.63 schwarze 310: *q = '\0';
1.109 schwarze 311: return 1;
1.1 kristaps 312: }
313:
1.6 kristaps 314: static void
315: resp_begin_http(int code, const char *msg)
316: {
317:
318: if (200 != code)
1.62 schwarze 319: printf("Status: %d %s\r\n", code, msg);
1.6 kristaps 320:
1.62 schwarze 321: printf("Content-Type: text/html; charset=utf-8\r\n"
322: "Cache-Control: no-cache\r\n"
323: "Pragma: no-cache\r\n"
324: "\r\n");
1.6 kristaps 325:
326: fflush(stdout);
327: }
328:
329: static void
1.114 schwarze 330: resp_copy(const char *filename)
331: {
332: char buf[4096];
333: ssize_t sz;
334: int fd;
335:
336: if ((fd = open(filename, O_RDONLY)) != -1) {
337: fflush(stdout);
338: while ((sz = read(fd, buf, sizeof(buf))) > 0)
339: write(STDOUT_FILENO, buf, sz);
340: }
341: }
342:
343: static void
1.6 kristaps 344: resp_begin_html(int code, const char *msg)
345: {
346:
347: resp_begin_http(code, msg);
348:
1.98 kristaps 349: printf("<!DOCTYPE html>\n"
1.126 schwarze 350: "<html>\n"
351: "<head>\n"
352: "<meta charset=\"UTF-8\"/>\n"
353: "<link rel=\"stylesheet\" href=\"%s/mandoc.css\""
354: " type=\"text/css\" media=\"all\">\n"
355: "<title>%s</title>\n"
356: "</head>\n"
357: "<body>\n"
1.58 schwarze 358: "<!-- Begin page content. //-->\n",
1.113 schwarze 359: CSS_DIR, CUSTOMIZE_TITLE);
1.114 schwarze 360:
361: resp_copy(MAN_DIR "/header.html");
1.6 kristaps 362: }
363:
364: static void
365: resp_end_html(void)
366: {
367:
1.114 schwarze 368: resp_copy(MAN_DIR "/footer.html");
369:
1.126 schwarze 370: puts("</body>\n"
371: "</html>");
1.6 kristaps 372: }
373:
374: static void
1.131 ! schwarze 375: resp_searchform(const struct req *req, enum focus focus)
1.6 kristaps 376: {
1.27 kristaps 377: int i;
1.13 kristaps 378:
1.6 kristaps 379: puts("<!-- Begin search form. //-->");
1.126 schwarze 380: printf("<div id=\"mancgi\">\n"
381: "<form action=\"/%s\" method=\"get\">\n"
382: "<fieldset>\n"
383: "<legend>Manual Page Search Parameters</legend>\n",
1.58 schwarze 384: scriptname);
1.68 schwarze 385:
386: /* Write query input box. */
387:
1.126 schwarze 388: printf( "<table><tr><td>\n"
389: "<input type=\"text\" name=\"query\" value=\"");
1.131 ! schwarze 390: if (req->q.query != NULL)
1.85 schwarze 391: html_print(req->q.query);
1.131 ! schwarze 392: printf( "\" size=\"40\"");
! 393: if (focus == FOCUS_QUERY)
! 394: printf(" autofocus");
! 395: puts(">");
1.68 schwarze 396:
397: /* Write submission and reset buttons. */
398:
1.126 schwarze 399: printf( "<input type=\"submit\" value=\"Submit\">\n"
400: "<input type=\"reset\" value=\"Reset\">\n");
1.68 schwarze 401:
402: /* Write show radio button */
403:
1.126 schwarze 404: printf( "</td><td>\n"
405: "<input type=\"radio\" ");
1.65 schwarze 406: if (req->q.equal)
1.126 schwarze 407: printf("checked=\"checked\" ");
408: printf( "name=\"apropos\" id=\"show\" value=\"0\">\n"
409: "<label for=\"show\">Show named manual page</label>\n");
1.68 schwarze 410:
411: /* Write section selector. */
412:
1.126 schwarze 413: puts( "</td></tr><tr><td>\n"
414: "<select name=\"sec\">");
1.68 schwarze 415: for (i = 0; i < sec_MAX; i++) {
1.126 schwarze 416: printf("<option value=\"%s\"", sec_numbers[i]);
1.68 schwarze 417: if (NULL != req->q.sec &&
418: 0 == strcmp(sec_numbers[i], req->q.sec))
1.126 schwarze 419: printf(" selected=\"selected\"");
420: printf(">%s</option>\n", sec_names[i]);
1.68 schwarze 421: }
1.126 schwarze 422: puts("</select>");
1.68 schwarze 423:
424: /* Write architecture selector. */
425:
1.126 schwarze 426: printf( "<select name=\"arch\">\n"
427: "<option value=\"default\"");
1.81 schwarze 428: if (NULL == req->q.arch)
1.126 schwarze 429: printf(" selected=\"selected\"");
430: puts(">All Architectures</option>");
1.68 schwarze 431: for (i = 0; i < arch_MAX; i++) {
1.126 schwarze 432: printf("<option value=\"%s\"", arch_names[i]);
1.68 schwarze 433: if (NULL != req->q.arch &&
434: 0 == strcmp(arch_names[i], req->q.arch))
1.126 schwarze 435: printf(" selected=\"selected\"");
436: printf(">%s</option>\n", arch_names[i]);
1.68 schwarze 437: }
1.126 schwarze 438: puts("</select>");
1.68 schwarze 439:
440: /* Write manpath selector. */
441:
1.27 kristaps 442: if (req->psz > 1) {
1.126 schwarze 443: puts("<select name=\"manpath\">");
1.27 kristaps 444: for (i = 0; i < (int)req->psz; i++) {
1.126 schwarze 445: printf("<option ");
1.102 schwarze 446: if (strcmp(req->q.manpath, req->p[i]) == 0)
1.126 schwarze 447: printf("selected=\"selected\" ");
448: printf("value=\"");
1.52 schwarze 449: html_print(req->p[i]);
1.27 kristaps 450: printf("\">");
1.52 schwarze 451: html_print(req->p[i]);
1.126 schwarze 452: puts("</option>");
1.27 kristaps 453: }
1.126 schwarze 454: puts("</select>");
1.27 kristaps 455: }
1.68 schwarze 456:
457: /* Write search radio button */
458:
1.126 schwarze 459: printf( "</td><td>\n"
460: "<input type=\"radio\" ");
1.68 schwarze 461: if (0 == req->q.equal)
1.126 schwarze 462: printf("checked=\"checked\" ");
463: printf( "name=\"apropos\" id=\"search\" value=\"1\">\n"
464: "<label for=\"search\">Search with apropos query</label>\n");
465:
466: puts("</td></tr></table>\n"
467: "</fieldset>\n"
468: "</form>\n"
469: "</div>");
1.20 kristaps 470: puts("<!-- End search form. //-->");
1.6 kristaps 471: }
472:
1.76 schwarze 473: static int
1.80 schwarze 474: validate_urifrag(const char *frag)
475: {
476:
477: while ('\0' != *frag) {
478: if ( ! (isalnum((unsigned char)*frag) ||
479: '-' == *frag || '.' == *frag ||
480: '/' == *frag || '_' == *frag))
1.109 schwarze 481: return 0;
1.80 schwarze 482: frag++;
483: }
1.109 schwarze 484: return 1;
1.80 schwarze 485: }
486:
487: static int
1.77 schwarze 488: validate_manpath(const struct req *req, const char* manpath)
489: {
490: size_t i;
491:
492: if ( ! strcmp(manpath, "mandoc"))
1.109 schwarze 493: return 1;
1.77 schwarze 494:
495: for (i = 0; i < req->psz; i++)
496: if ( ! strcmp(manpath, req->p[i]))
1.109 schwarze 497: return 1;
1.77 schwarze 498:
1.109 schwarze 499: return 0;
1.77 schwarze 500: }
501:
502: static int
1.76 schwarze 503: validate_filename(const char *file)
504: {
505:
506: if ('.' == file[0] && '/' == file[1])
507: file += 2;
508:
1.109 schwarze 509: return ! (strstr(file, "../") || strstr(file, "/..") ||
510: (strncmp(file, "man", 3) && strncmp(file, "cat", 3)));
1.76 schwarze 511: }
512:
1.6 kristaps 513: static void
1.72 schwarze 514: pg_index(const struct req *req)
1.6 kristaps 515: {
516:
517: resp_begin_html(200, NULL);
1.131 ! schwarze 518: resp_searchform(req, FOCUS_QUERY);
1.126 schwarze 519: printf("<p>\n"
1.86 schwarze 520: "This web interface is documented in the\n"
1.126 schwarze 521: "<a href=\"/%s%smandoc/man8/man.cgi.8\">man.cgi</a>\n"
1.86 schwarze 522: "manual, and the\n"
1.126 schwarze 523: "<a href=\"/%s%smandoc/man1/apropos.1\">apropos</a>\n"
1.69 schwarze 524: "manual explains the query syntax.\n"
1.126 schwarze 525: "</p>\n",
1.119 schwarze 526: scriptname, *scriptname == '\0' ? "" : "/",
527: scriptname, *scriptname == '\0' ? "" : "/");
1.6 kristaps 528: resp_end_html();
529: }
530:
531: static void
1.72 schwarze 532: pg_noresult(const struct req *req, const char *msg)
1.59 schwarze 533: {
534: resp_begin_html(200, NULL);
1.131 ! schwarze 535: resp_searchform(req, FOCUS_QUERY);
1.126 schwarze 536: puts("<p>");
1.59 schwarze 537: puts(msg);
1.126 schwarze 538: puts("</p>");
1.59 schwarze 539: resp_end_html();
540: }
541:
542: static void
1.72 schwarze 543: pg_error_badrequest(const char *msg)
1.9 kristaps 544: {
545:
1.59 schwarze 546: resp_begin_html(400, "Bad Request");
1.126 schwarze 547: puts("<h1>Bad Request</h1>\n"
548: "<p>\n");
1.59 schwarze 549: puts(msg);
550: printf("Try again from the\n"
1.126 schwarze 551: "<a href=\"/%s\">main page</a>.\n"
552: "</p>", scriptname);
1.9 kristaps 553: resp_end_html();
554: }
555:
556: static void
1.72 schwarze 557: pg_error_internal(void)
1.7 kristaps 558: {
559: resp_begin_html(500, "Internal Server Error");
1.126 schwarze 560: puts("<p>Internal Server Error</p>");
1.7 kristaps 561: resp_end_html();
562: }
563:
564: static void
1.72 schwarze 565: pg_searchres(const struct req *req, struct manpage *r, size_t sz)
1.1 kristaps 566: {
1.81 schwarze 567: char *arch, *archend;
1.120 schwarze 568: const char *sec;
569: size_t i, iuse;
1.81 schwarze 570: int archprio, archpriouse;
1.70 schwarze 571: int prio, priouse;
1.19 kristaps 572:
1.76 schwarze 573: for (i = 0; i < sz; i++) {
574: if (validate_filename(r[i].file))
575: continue;
1.128 schwarze 576: warnx("invalid filename %s in %s database",
1.76 schwarze 577: r[i].file, req->q.manpath);
578: pg_error_internal();
579: return;
580: }
581:
1.121 schwarze 582: if (req->isquery && sz == 1) {
1.6 kristaps 583: /*
584: * If we have just one result, then jump there now
585: * without any delay.
586: */
1.62 schwarze 587: printf("Status: 303 See Other\r\n");
1.119 schwarze 588: printf("Location: http://%s/%s%s%s/%s",
589: HTTP_HOST, scriptname,
590: *scriptname == '\0' ? "" : "/",
591: req->q.manpath, r[0].file);
1.62 schwarze 592: printf("\r\n"
593: "Content-Type: text/html; charset=utf-8\r\n"
594: "\r\n");
1.6 kristaps 595: return;
596: }
597:
1.12 kristaps 598: resp_begin_html(200, NULL);
1.131 ! schwarze 599: resp_searchform(req,
! 600: req->q.equal || sz == 1 ? FOCUS_NONE : FOCUS_QUERY);
1.1 kristaps 601:
1.123 schwarze 602: if (sz > 1) {
1.126 schwarze 603: puts("<div class=\"results\">");
604: puts("<table>");
1.123 schwarze 605:
606: for (i = 0; i < sz; i++) {
1.126 schwarze 607: printf("<tr>\n"
608: "<td class=\"title\">\n"
609: "<a href=\"/%s%s%s/%s",
1.123 schwarze 610: scriptname, *scriptname == '\0' ? "" : "/",
611: req->q.manpath, r[i].file);
612: printf("\">");
613: html_print(r[i].names);
1.126 schwarze 614: printf("</a>\n"
615: "</td>\n"
616: "<td class=\"desc\">");
1.123 schwarze 617: html_print(r[i].output);
1.126 schwarze 618: puts("</td>\n"
619: "</tr>");
1.123 schwarze 620: }
621:
1.126 schwarze 622: puts("</table>\n"
623: "</div>");
1.1 kristaps 624: }
1.16 kristaps 625:
1.70 schwarze 626: /*
627: * In man(1) mode, show one of the pages
628: * even if more than one is found.
629: */
630:
1.123 schwarze 631: if (req->q.equal || sz == 1) {
1.126 schwarze 632: puts("<hr>");
1.70 schwarze 633: iuse = 0;
1.120 schwarze 634: priouse = 20;
1.81 schwarze 635: archpriouse = 3;
1.70 schwarze 636: for (i = 0; i < sz; i++) {
1.120 schwarze 637: sec = r[i].file;
638: sec += strcspn(sec, "123456789");
639: if (sec[0] == '\0')
1.70 schwarze 640: continue;
1.120 schwarze 641: prio = sec_prios[sec[0] - '1'];
642: if (sec[1] != '/')
643: prio += 10;
644: if (req->q.arch == NULL) {
1.81 schwarze 645: archprio =
1.120 schwarze 646: ((arch = strchr(sec + 1, '/'))
647: == NULL) ? 3 :
648: ((archend = strchr(arch + 1, '/'))
649: == NULL) ? 0 :
1.81 schwarze 650: strncmp(arch, "amd64/",
651: archend - arch) ? 2 : 1;
652: if (archprio < archpriouse) {
653: archpriouse = archprio;
654: priouse = prio;
655: iuse = i;
656: continue;
657: }
658: if (archprio > archpriouse)
659: continue;
660: }
1.70 schwarze 661: if (prio >= priouse)
662: continue;
663: priouse = prio;
664: iuse = i;
665: }
666: resp_show(req, r[iuse].file);
667: }
668:
1.6 kristaps 669: resp_end_html();
670: }
671:
1.1 kristaps 672: static void
1.129 schwarze 673: resp_catman(const struct req *req, const char *file)
1.9 kristaps 674: {
1.10 kristaps 675: FILE *f;
1.115 schwarze 676: char *p;
677: size_t sz;
678: ssize_t len;
1.10 kristaps 679: int i;
680: int italic, bold;
1.9 kristaps 681:
1.115 schwarze 682: if ((f = fopen(file, "r")) == NULL) {
1.126 schwarze 683: puts("<p>You specified an invalid manual file.</p>");
1.9 kristaps 684: return;
685: }
686:
1.126 schwarze 687: puts("<div class=\"catman\">\n"
688: "<pre>");
1.10 kristaps 689:
1.115 schwarze 690: p = NULL;
691: sz = 0;
692:
693: while ((len = getline(&p, &sz, f)) != -1) {
1.10 kristaps 694: bold = italic = 0;
1.115 schwarze 695: for (i = 0; i < len - 1; i++) {
1.104 schwarze 696: /*
1.10 kristaps 697: * This means that the catpage is out of state.
698: * Ignore it and keep going (although the
699: * catpage is bogus).
700: */
701:
702: if ('\b' == p[i] || '\n' == p[i])
703: continue;
704:
705: /*
706: * Print a regular character.
707: * Close out any bold/italic scopes.
708: * If we're in back-space mode, make sure we'll
709: * have something to enter when we backspace.
710: */
711:
712: if ('\b' != p[i + 1]) {
713: if (italic)
1.126 schwarze 714: printf("</i>");
1.10 kristaps 715: if (bold)
1.126 schwarze 716: printf("</b>");
1.10 kristaps 717: italic = bold = 0;
718: html_putchar(p[i]);
719: continue;
1.115 schwarze 720: } else if (i + 2 >= len)
1.10 kristaps 721: continue;
722:
723: /* Italic mode. */
724:
725: if ('_' == p[i]) {
726: if (bold)
1.126 schwarze 727: printf("</b>");
1.10 kristaps 728: if ( ! italic)
1.126 schwarze 729: printf("<i>");
1.10 kristaps 730: bold = 0;
731: italic = 1;
732: i += 2;
733: html_putchar(p[i]);
734: continue;
735: }
736:
1.104 schwarze 737: /*
1.10 kristaps 738: * Handle funny behaviour troff-isms.
739: * These grok'd from the original man2html.c.
740: */
741:
742: if (('+' == p[i] && 'o' == p[i + 2]) ||
743: ('o' == p[i] && '+' == p[i + 2]) ||
744: ('|' == p[i] && '=' == p[i + 2]) ||
745: ('=' == p[i] && '|' == p[i + 2]) ||
746: ('*' == p[i] && '=' == p[i + 2]) ||
747: ('=' == p[i] && '*' == p[i + 2]) ||
748: ('*' == p[i] && '|' == p[i + 2]) ||
749: ('|' == p[i] && '*' == p[i + 2])) {
750: if (italic)
1.126 schwarze 751: printf("</i>");
1.10 kristaps 752: if (bold)
1.126 schwarze 753: printf("</b>");
1.10 kristaps 754: italic = bold = 0;
755: putchar('*');
756: i += 2;
757: continue;
758: } else if (('|' == p[i] && '-' == p[i + 2]) ||
759: ('-' == p[i] && '|' == p[i + 1]) ||
760: ('+' == p[i] && '-' == p[i + 1]) ||
761: ('-' == p[i] && '+' == p[i + 1]) ||
762: ('+' == p[i] && '|' == p[i + 1]) ||
763: ('|' == p[i] && '+' == p[i + 1])) {
764: if (italic)
1.126 schwarze 765: printf("</i>");
1.10 kristaps 766: if (bold)
1.126 schwarze 767: printf("</b>");
1.10 kristaps 768: italic = bold = 0;
769: putchar('+');
770: i += 2;
771: continue;
772: }
773:
774: /* Bold mode. */
1.104 schwarze 775:
1.10 kristaps 776: if (italic)
1.126 schwarze 777: printf("</i>");
1.10 kristaps 778: if ( ! bold)
1.126 schwarze 779: printf("<b>");
1.10 kristaps 780: bold = 1;
781: italic = 0;
782: i += 2;
783: html_putchar(p[i]);
784: }
785:
1.104 schwarze 786: /*
1.10 kristaps 787: * Clean up the last character.
1.104 schwarze 788: * We can get to a newline; don't print that.
1.10 kristaps 789: */
1.9 kristaps 790:
1.10 kristaps 791: if (italic)
1.126 schwarze 792: printf("</i>");
1.10 kristaps 793: if (bold)
1.126 schwarze 794: printf("</b>");
1.9 kristaps 795:
1.115 schwarze 796: if (i == len - 1 && p[i] != '\n')
1.10 kristaps 797: html_putchar(p[i]);
1.9 kristaps 798:
1.10 kristaps 799: putchar('\n');
800: }
1.115 schwarze 801: free(p);
1.10 kristaps 802:
1.126 schwarze 803: puts("</pre>\n"
804: "</div>");
1.10 kristaps 805:
806: fclose(f);
1.9 kristaps 807: }
808:
809: static void
1.129 schwarze 810: resp_format(const struct req *req, const char *file)
1.7 kristaps 811: {
1.106 schwarze 812: struct manoutput conf;
1.8 kristaps 813: struct mparse *mp;
1.107 schwarze 814: struct roff_man *man;
1.8 kristaps 815: void *vp;
1.90 schwarze 816: int fd;
817: int usepath;
1.7 kristaps 818:
1.8 kristaps 819: if (-1 == (fd = open(file, O_RDONLY, 0))) {
1.126 schwarze 820: puts("<p>You specified an invalid manual file.</p>");
1.7 kristaps 821: return;
822: }
823:
1.110 schwarze 824: mchars_alloc();
825: mp = mparse_alloc(MPARSE_SO, MANDOCLEVEL_BADARG, NULL, req->q.manpath);
1.103 schwarze 826: mparse_readfd(mp, fd, file);
1.8 kristaps 827: close(fd);
1.7 kristaps 828:
1.106 schwarze 829: memset(&conf, 0, sizeof(conf));
830: conf.fragment = 1;
1.90 schwarze 831: usepath = strcmp(req->q.manpath, req->p[0]);
1.122 schwarze 832: mandoc_asprintf(&conf.man, "/%s%s%%N.%%S",
833: usepath ? req->q.manpath : "", usepath ? "/" : "");
1.10 kristaps 834:
1.108 schwarze 835: mparse_result(mp, &man, NULL);
836: if (man == NULL) {
1.128 schwarze 837: warnx("fatal mandoc error: %s/%s", req->q.manpath, file);
1.72 schwarze 838: pg_error_internal();
1.32 kristaps 839: mparse_free(mp);
1.110 schwarze 840: mchars_free();
1.32 kristaps 841: return;
842: }
843:
1.110 schwarze 844: vp = html_alloc(&conf);
1.7 kristaps 845:
1.111 schwarze 846: if (man->macroset == MACROSET_MDOC) {
847: mdoc_validate(man);
1.108 schwarze 848: html_mdoc(vp, man);
1.112 schwarze 849: } else {
850: man_validate(man);
1.8 kristaps 851: html_man(vp, man);
1.112 schwarze 852: }
1.32 kristaps 853:
1.8 kristaps 854: html_free(vp);
855: mparse_free(mp);
1.110 schwarze 856: mchars_free();
1.106 schwarze 857: free(conf.man);
1.7 kristaps 858: }
859:
860: static void
1.70 schwarze 861: resp_show(const struct req *req, const char *file)
862: {
1.76 schwarze 863:
864: if ('.' == file[0] && '/' == file[1])
1.71 schwarze 865: file += 2;
1.70 schwarze 866:
867: if ('c' == *file)
1.129 schwarze 868: resp_catman(req, file);
1.70 schwarze 869: else
1.129 schwarze 870: resp_format(req, file);
1.70 schwarze 871: }
872:
873: static void
1.84 schwarze 874: pg_show(struct req *req, const char *fullpath)
1.1 kristaps 875: {
1.84 schwarze 876: char *manpath;
877: const char *file;
1.25 kristaps 878:
1.84 schwarze 879: if ((file = strchr(fullpath, '/')) == NULL) {
1.72 schwarze 880: pg_error_badrequest(
1.59 schwarze 881: "You did not specify a page to show.");
1.25 kristaps 882: return;
1.104 schwarze 883: }
1.84 schwarze 884: manpath = mandoc_strndup(fullpath, file - fullpath);
885: file++;
1.6 kristaps 886:
1.84 schwarze 887: if ( ! validate_manpath(req, manpath)) {
1.77 schwarze 888: pg_error_badrequest(
889: "You specified an invalid manpath.");
1.84 schwarze 890: free(manpath);
1.77 schwarze 891: return;
892: }
893:
1.24 kristaps 894: /*
1.58 schwarze 895: * Begin by chdir()ing into the manpath.
1.24 kristaps 896: * This way we can pick up the database files, which are
897: * relative to the manpath root.
898: */
899:
1.84 schwarze 900: if (chdir(manpath) == -1) {
1.128 schwarze 901: warn("chdir %s", manpath);
1.77 schwarze 902: pg_error_internal();
1.84 schwarze 903: free(manpath);
1.76 schwarze 904: return;
905: }
906:
1.84 schwarze 907: if (strcmp(manpath, "mandoc")) {
908: free(req->q.manpath);
909: req->q.manpath = manpath;
910: } else
911: free(manpath);
912:
913: if ( ! validate_filename(file)) {
1.76 schwarze 914: pg_error_badrequest(
915: "You specified an invalid manual file.");
1.24 kristaps 916: return;
917: }
1.79 schwarze 918:
1.70 schwarze 919: resp_begin_html(200, NULL);
1.131 ! schwarze 920: resp_searchform(req, FOCUS_NONE);
1.84 schwarze 921: resp_show(req, file);
1.70 schwarze 922: resp_end_html();
1.6 kristaps 923: }
924:
925: static void
1.66 schwarze 926: pg_search(const struct req *req)
1.6 kristaps 927: {
1.52 schwarze 928: struct mansearch search;
929: struct manpaths paths;
930: struct manpage *res;
1.97 schwarze 931: char **argv;
932: char *query, *rp, *wp;
1.52 schwarze 933: size_t ressz;
1.97 schwarze 934: int argc;
1.6 kristaps 935:
936: /*
1.24 kristaps 937: * Begin by chdir()ing into the root of the manpath.
938: * This way we can pick up the database files, which are
939: * relative to the manpath root.
940: */
941:
1.128 schwarze 942: if (chdir(req->q.manpath) == -1) {
943: warn("chdir %s", req->q.manpath);
1.77 schwarze 944: pg_error_internal();
1.24 kristaps 945: return;
946: }
947:
1.52 schwarze 948: search.arch = req->q.arch;
949: search.sec = req->q.sec;
1.94 schwarze 950: search.outkey = "Nd";
951: search.argmode = req->q.equal ? ARG_NAME : ARG_EXPR;
1.101 schwarze 952: search.firstmatch = 1;
1.52 schwarze 953:
954: paths.sz = 1;
955: paths.paths = mandoc_malloc(sizeof(char *));
956: paths.paths[0] = mandoc_strdup(".");
1.24 kristaps 957:
958: /*
1.97 schwarze 959: * Break apart at spaces with backslash-escaping.
1.6 kristaps 960: */
961:
1.97 schwarze 962: argc = 0;
963: argv = NULL;
964: rp = query = mandoc_strdup(req->q.query);
965: for (;;) {
966: while (isspace((unsigned char)*rp))
967: rp++;
968: if (*rp == '\0')
969: break;
970: argv = mandoc_reallocarray(argv, argc + 1, sizeof(char *));
971: argv[argc++] = wp = rp;
972: for (;;) {
973: if (isspace((unsigned char)*rp)) {
974: *wp = '\0';
975: rp++;
976: break;
977: }
978: if (rp[0] == '\\' && rp[1] != '\0')
979: rp++;
980: if (wp != rp)
981: *wp = *rp;
982: if (*rp == '\0')
983: break;
984: wp++;
985: rp++;
986: }
1.6 kristaps 987: }
988:
1.97 schwarze 989: if (0 == mansearch(&search, &paths, argc, argv, &res, &ressz))
1.72 schwarze 990: pg_noresult(req, "You entered an invalid query.");
1.59 schwarze 991: else if (0 == ressz)
1.72 schwarze 992: pg_noresult(req, "No results found.");
1.59 schwarze 993: else
1.72 schwarze 994: pg_searchres(req, res, ressz);
1.6 kristaps 995:
1.97 schwarze 996: free(query);
997: mansearch_free(res, ressz);
1.52 schwarze 998: free(paths.paths[0]);
999: free(paths.paths);
1.1 kristaps 1000: }
1001:
1002: int
1003: main(void)
1004: {
1.66 schwarze 1005: struct req req;
1.95 schwarze 1006: struct itimerval itimer;
1.66 schwarze 1007: const char *path;
1.83 schwarze 1008: const char *querystring;
1.1 kristaps 1009: int i;
1.95 schwarze 1010:
1011: /* Poor man's ReDoS mitigation. */
1012:
1.99 schwarze 1013: itimer.it_value.tv_sec = 2;
1.95 schwarze 1014: itimer.it_value.tv_usec = 0;
1.99 schwarze 1015: itimer.it_interval.tv_sec = 2;
1.95 schwarze 1016: itimer.it_interval.tv_usec = 0;
1017: if (setitimer(ITIMER_VIRTUAL, &itimer, NULL) == -1) {
1.128 schwarze 1018: warn("setitimer");
1.80 schwarze 1019: pg_error_internal();
1.109 schwarze 1020: return EXIT_FAILURE;
1.80 schwarze 1021: }
1022:
1.24 kristaps 1023: /*
1.67 schwarze 1024: * First we change directory into the MAN_DIR so that
1.24 kristaps 1025: * subsequent scanning for manpath directories is rooted
1026: * relative to the same position.
1027: */
1028:
1.128 schwarze 1029: if (chdir(MAN_DIR) == -1) {
1030: warn("MAN_DIR: %s", MAN_DIR);
1.72 schwarze 1031: pg_error_internal();
1.109 schwarze 1032: return EXIT_FAILURE;
1.104 schwarze 1033: }
1.24 kristaps 1034:
1035: memset(&req, 0, sizeof(struct req));
1.118 schwarze 1036: req.q.equal = 1;
1.129 schwarze 1037: parse_manpath_conf(&req);
1.1 kristaps 1038:
1.117 schwarze 1039: /* Parse the path info and the query string. */
1.1 kristaps 1040:
1.117 schwarze 1041: if ((path = getenv("PATH_INFO")) == NULL)
1042: path = "";
1043: else if (*path == '/')
1044: path++;
1045:
1.124 schwarze 1046: if (*path != '\0') {
1.129 schwarze 1047: parse_path_info(&req, path);
1.124 schwarze 1048: if (access(path, F_OK) == -1)
1049: path = "";
1.117 schwarze 1050: } else if ((querystring = getenv("QUERY_STRING")) != NULL)
1.129 schwarze 1051: parse_query_string(&req, querystring);
1.77 schwarze 1052:
1.117 schwarze 1053: /* Validate parsed data and add defaults. */
1054:
1.102 schwarze 1055: if (req.q.manpath == NULL)
1056: req.q.manpath = mandoc_strdup(req.p[0]);
1057: else if ( ! validate_manpath(&req, req.q.manpath)) {
1.77 schwarze 1058: pg_error_badrequest(
1059: "You specified an invalid manpath.");
1.109 schwarze 1060: return EXIT_FAILURE;
1.77 schwarze 1061: }
1.1 kristaps 1062:
1.80 schwarze 1063: if ( ! (NULL == req.q.arch || validate_urifrag(req.q.arch))) {
1064: pg_error_badrequest(
1065: "You specified an invalid architecture.");
1.109 schwarze 1066: return EXIT_FAILURE;
1.80 schwarze 1067: }
1068:
1.66 schwarze 1069: /* Dispatch to the three different pages. */
1.1 kristaps 1070:
1.66 schwarze 1071: if ('\0' != *path)
1072: pg_show(&req, path);
1.85 schwarze 1073: else if (NULL != req.q.query)
1.66 schwarze 1074: pg_search(&req);
1075: else
1.72 schwarze 1076: pg_index(&req);
1.1 kristaps 1077:
1.83 schwarze 1078: free(req.q.manpath);
1079: free(req.q.arch);
1080: free(req.q.sec);
1.85 schwarze 1081: free(req.q.query);
1.52 schwarze 1082: for (i = 0; i < (int)req.psz; i++)
1083: free(req.p[i]);
1.24 kristaps 1084: free(req.p);
1.109 schwarze 1085: return EXIT_SUCCESS;
1.117 schwarze 1086: }
1087:
1088: /*
1089: * If PATH_INFO is not a file name, translate it to a query.
1090: */
1091: static void
1.129 schwarze 1092: parse_path_info(struct req *req, const char *path)
1.117 schwarze 1093: {
1.127 schwarze 1094: char *dir;
1.117 schwarze 1095:
1.121 schwarze 1096: req->isquery = 0;
1.117 schwarze 1097: req->q.equal = 1;
1098: req->q.manpath = mandoc_strdup(path);
1099:
1100: /* Mandatory manual page name. */
1101: if ((req->q.query = strrchr(req->q.manpath, '/')) == NULL) {
1102: req->q.query = req->q.manpath;
1103: req->q.manpath = NULL;
1104: } else
1105: *req->q.query++ = '\0';
1106:
1107: /* Optional trailing section. */
1108: if ((req->q.sec = strrchr(req->q.query, '.')) != NULL) {
1109: if(isdigit((unsigned char)req->q.sec[1])) {
1110: *req->q.sec++ = '\0';
1111: req->q.sec = mandoc_strdup(req->q.sec);
1112: } else
1113: req->q.sec = NULL;
1114: }
1115:
1116: /* Handle the case of name[.section] only. */
1117: if (req->q.manpath == NULL) {
1118: req->q.arch = NULL;
1119: return;
1120: }
1121: req->q.query = mandoc_strdup(req->q.query);
1122:
1123: /* Optional architecture. */
1.127 schwarze 1124: dir = strrchr(req->q.manpath, '/');
1125: if (dir != NULL && strncmp(dir + 1, "man", 3) != 0) {
1126: *dir++ = '\0';
1127: req->q.arch = mandoc_strdup(dir);
1128: dir = strrchr(req->q.manpath, '/');
1129: } else
1130: req->q.arch = NULL;
1.117 schwarze 1131:
1.127 schwarze 1132: /* Optional directory name. */
1133: if (dir != NULL && strncmp(dir + 1, "man", 3) == 0) {
1134: *dir++ = '\0';
1135: free(req->q.sec);
1136: req->q.sec = mandoc_strdup(dir + 3);
1.117 schwarze 1137: }
1.24 kristaps 1138: }
1139:
1140: /*
1141: * Scan for indexable paths.
1142: */
1143: static void
1.129 schwarze 1144: parse_manpath_conf(struct req *req)
1.24 kristaps 1145: {
1.54 schwarze 1146: FILE *fp;
1147: char *dp;
1148: size_t dpsz;
1.115 schwarze 1149: ssize_t len;
1.54 schwarze 1150:
1.128 schwarze 1151: if ((fp = fopen("manpath.conf", "r")) == NULL) {
1152: warn("%s/manpath.conf", MAN_DIR);
1.74 schwarze 1153: pg_error_internal();
1154: exit(EXIT_FAILURE);
1155: }
1.24 kristaps 1156:
1.115 schwarze 1157: dp = NULL;
1158: dpsz = 0;
1159:
1160: while ((len = getline(&dp, &dpsz, fp)) != -1) {
1161: if (dp[len - 1] == '\n')
1162: dp[--len] = '\0';
1.54 schwarze 1163: req->p = mandoc_realloc(req->p,
1164: (req->psz + 1) * sizeof(char *));
1.80 schwarze 1165: if ( ! validate_urifrag(dp)) {
1.128 schwarze 1166: warnx("%s/manpath.conf contains "
1167: "unsafe path \"%s\"", MAN_DIR, dp);
1.80 schwarze 1168: pg_error_internal();
1169: exit(EXIT_FAILURE);
1170: }
1.128 schwarze 1171: if (strchr(dp, '/') != NULL) {
1172: warnx("%s/manpath.conf contains "
1173: "path with slash \"%s\"", MAN_DIR, dp);
1.80 schwarze 1174: pg_error_internal();
1175: exit(EXIT_FAILURE);
1176: }
1177: req->p[req->psz++] = dp;
1.115 schwarze 1178: dp = NULL;
1179: dpsz = 0;
1.74 schwarze 1180: }
1.115 schwarze 1181: free(dp);
1.74 schwarze 1182:
1.128 schwarze 1183: if (req->p == NULL) {
1184: warnx("%s/manpath.conf is empty", MAN_DIR);
1.74 schwarze 1185: pg_error_internal();
1186: exit(EXIT_FAILURE);
1.24 kristaps 1187: }
1188: }
CVSweb