Annotation of src/usr.bin/mandoc/cgi.c, Revision 1.82
1.82 ! schwarze 1: /* $OpenBSD: cgi.c,v 1.81 2017/01/19 13:34:59 schwarze Exp $ */
1.1 schwarze 2: /*
3: * Copyright (c) 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
1.81 schwarze 4: * Copyright (c) 2014, 2015, 2016, 2017 Ingo Schwarze <schwarze@usta.de>
1.1 schwarze 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: *
1.44 schwarze 10: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1.1 schwarze 11: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1.44 schwarze 12: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1.1 schwarze 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: */
1.33 schwarze 18: #include <sys/types.h>
19: #include <sys/time.h>
20:
1.1 schwarze 21: #include <ctype.h>
1.67 schwarze 22: #include <err.h>
1.1 schwarze 23: #include <errno.h>
24: #include <fcntl.h>
25: #include <limits.h>
1.32 schwarze 26: #include <stdint.h>
1.1 schwarze 27: #include <stdio.h>
28: #include <stdlib.h>
29: #include <string.h>
30: #include <unistd.h>
31:
1.47 schwarze 32: #include "mandoc_aux.h"
1.1 schwarze 33: #include "mandoc.h"
1.47 schwarze 34: #include "roff.h"
1.50 schwarze 35: #include "mdoc.h"
1.51 schwarze 36: #include "man.h"
1.1 schwarze 37: #include "main.h"
1.44 schwarze 38: #include "manconf.h"
1.1 schwarze 39: #include "mansearch.h"
1.7 schwarze 40: #include "cgi.h"
1.1 schwarze 41:
42: /*
43: * A query as passed to the search function.
44: */
45: struct query {
1.23 schwarze 46: char *manpath; /* desired manual directory */
47: char *arch; /* architecture */
48: char *sec; /* manual section */
1.25 schwarze 49: char *query; /* unparsed query expression */
1.5 schwarze 50: int equal; /* match whole names, not substrings */
1.1 schwarze 51: };
52:
53: struct req {
54: struct query q;
55: char **p; /* array of available manpaths */
56: size_t psz; /* number of available manpaths */
1.60 schwarze 57: int isquery; /* QUERY_STRING used, not PATH_INFO */
1.1 schwarze 58: };
59:
1.70 schwarze 60: enum focus {
61: FOCUS_NONE = 0,
62: FOCUS_QUERY
63: };
64:
1.1 schwarze 65: static void html_print(const char *);
66: static void html_putchar(char);
1.43 schwarze 67: static int http_decode(char *);
1.68 schwarze 68: static void parse_manpath_conf(struct req *);
69: static void parse_path_info(struct req *req, const char *path);
70: static void parse_query_string(struct req *, const char *);
1.12 schwarze 71: static void pg_error_badrequest(const char *);
72: static void pg_error_internal(void);
73: static void pg_index(const struct req *);
74: static void pg_noresult(const struct req *, const char *);
1.6 schwarze 75: static void pg_search(const struct req *);
1.12 schwarze 76: static void pg_searchres(const struct req *,
77: struct manpage *, size_t);
1.19 schwarze 78: static void pg_show(struct req *, const char *);
1.1 schwarze 79: static void resp_begin_html(int, const char *);
80: static void resp_begin_http(int, const char *);
1.68 schwarze 81: static void resp_catman(const struct req *, const char *);
1.53 schwarze 82: static void resp_copy(const char *);
1.1 schwarze 83: static void resp_end_html(void);
1.68 schwarze 84: static void resp_format(const struct req *, const char *);
1.70 schwarze 85: static void resp_searchform(const struct req *, enum focus);
1.10 schwarze 86: static void resp_show(const struct req *, const char *);
1.25 schwarze 87: static void set_query_attr(char **, char **);
88: static int validate_filename(const char *);
89: static int validate_manpath(const struct req *, const char *);
90: static int validate_urifrag(const char *);
1.1 schwarze 91:
1.58 schwarze 92: static const char *scriptname = SCRIPT_NAME;
1.1 schwarze 93:
1.10 schwarze 94: static const int sec_prios[] = {1, 4, 5, 8, 6, 3, 7, 2, 9};
1.8 schwarze 95: static const char *const sec_numbers[] = {
96: "0", "1", "2", "3", "3p", "4", "5", "6", "7", "8", "9"
97: };
98: static const char *const sec_names[] = {
99: "All Sections",
100: "1 - General Commands",
101: "2 - System Calls",
1.34 schwarze 102: "3 - Library Functions",
103: "3p - Perl Library",
104: "4 - Device Drivers",
1.8 schwarze 105: "5 - File Formats",
106: "6 - Games",
1.34 schwarze 107: "7 - Miscellaneous Information",
108: "8 - System Manager\'s Manual",
109: "9 - Kernel Developer\'s Manual"
1.8 schwarze 110: };
111: static const int sec_MAX = sizeof(sec_names) / sizeof(char *);
112:
113: static const char *const arch_names[] = {
1.76 schwarze 114: "amd64", "alpha", "armv7",
115: "hppa", "i386", "landisk",
1.64 schwarze 116: "loongson", "luna88k", "macppc", "mips64",
1.79 schwarze 117: "octeon", "sgi", "socppc", "sparc64",
1.76 schwarze 118: "amiga", "arc", "armish", "arm32",
119: "atari", "aviion", "beagle", "cats",
120: "hppa64", "hp300",
1.64 schwarze 121: "ia64", "mac68k", "mvme68k", "mvme88k",
122: "mvmeppc", "palm", "pc532", "pegasos",
1.78 schwarze 123: "pmax", "powerpc", "solbourne", "sparc",
1.79 schwarze 124: "sun3", "vax", "wgrisc", "x68k",
125: "zaurus"
1.8 schwarze 126: };
127: static const int arch_MAX = sizeof(arch_names) / sizeof(char *);
128:
1.1 schwarze 129: /*
130: * Print a character, escaping HTML along the way.
131: * This will pass non-ASCII straight to output: be warned!
132: */
133: static void
134: html_putchar(char c)
135: {
136:
137: switch (c) {
138: case ('"'):
1.80 bentley 139: printf(""");
1.1 schwarze 140: break;
141: case ('&'):
142: printf("&");
143: break;
144: case ('>'):
145: printf(">");
146: break;
147: case ('<'):
148: printf("<");
149: break;
150: default:
151: putchar((unsigned char)c);
152: break;
153: }
154: }
155:
156: /*
157: * Call through to html_putchar().
158: * Accepts NULL strings.
159: */
160: static void
161: html_print(const char *p)
162: {
1.43 schwarze 163:
1.1 schwarze 164: if (NULL == p)
165: return;
166: while ('\0' != *p)
167: html_putchar(*p++);
168: }
169:
170: /*
1.23 schwarze 171: * Transfer the responsibility for the allocated string *val
172: * to the query structure.
1.1 schwarze 173: */
174: static void
1.23 schwarze 175: set_query_attr(char **attr, char **val)
1.1 schwarze 176: {
177:
1.23 schwarze 178: free(*attr);
179: if (**val == '\0') {
180: *attr = NULL;
181: free(*val);
182: } else
183: *attr = *val;
184: *val = NULL;
185: }
186:
187: /*
188: * Parse the QUERY_STRING for key-value pairs
189: * and store the values into the query structure.
190: */
191: static void
1.68 schwarze 192: parse_query_string(struct req *req, const char *qs)
1.23 schwarze 193: {
194: char *key, *val;
195: size_t keysz, valsz;
196:
1.60 schwarze 197: req->isquery = 1;
1.23 schwarze 198: req->q.manpath = NULL;
199: req->q.arch = NULL;
200: req->q.sec = NULL;
1.25 schwarze 201: req->q.query = NULL;
1.23 schwarze 202: req->q.equal = 1;
203:
204: key = val = NULL;
205: while (*qs != '\0') {
1.1 schwarze 206:
1.23 schwarze 207: /* Parse one key. */
208:
209: keysz = strcspn(qs, "=;&");
210: key = mandoc_strndup(qs, keysz);
211: qs += keysz;
212: if (*qs != '=')
213: goto next;
214:
215: /* Parse one value. */
216:
217: valsz = strcspn(++qs, ";&");
218: val = mandoc_strndup(qs, valsz);
219: qs += valsz;
220:
221: /* Decode and catch encoding errors. */
1.1 schwarze 222:
1.23 schwarze 223: if ( ! (http_decode(key) && http_decode(val)))
224: goto next;
1.1 schwarze 225:
1.23 schwarze 226: /* Handle key-value pairs. */
1.1 schwarze 227:
1.23 schwarze 228: if ( ! strcmp(key, "query"))
1.25 schwarze 229: set_query_attr(&req->q.query, &val);
1.1 schwarze 230:
1.23 schwarze 231: else if ( ! strcmp(key, "apropos"))
232: req->q.equal = !strcmp(val, "0");
233:
234: else if ( ! strcmp(key, "manpath")) {
1.13 schwarze 235: #ifdef COMPAT_OLDURI
1.23 schwarze 236: if ( ! strncmp(val, "OpenBSD ", 8)) {
1.13 schwarze 237: val[7] = '-';
238: if ('C' == val[8])
239: val[8] = 'c';
240: }
241: #endif
1.23 schwarze 242: set_query_attr(&req->q.manpath, &val);
243: }
244:
245: else if ( ! (strcmp(key, "sec")
1.13 schwarze 246: #ifdef COMPAT_OLDURI
1.23 schwarze 247: && strcmp(key, "sektion")
1.13 schwarze 248: #endif
1.23 schwarze 249: )) {
250: if ( ! strcmp(val, "0"))
251: *val = '\0';
252: set_query_attr(&req->q.sec, &val);
1.5 schwarze 253: }
1.23 schwarze 254:
255: else if ( ! strcmp(key, "arch")) {
256: if ( ! strcmp(val, "default"))
257: *val = '\0';
258: set_query_attr(&req->q.arch, &val);
259: }
260:
261: /*
262: * The key must be freed in any case.
263: * The val may have been handed over to the query
264: * structure, in which case it is now NULL.
265: */
266: next:
267: free(key);
268: key = NULL;
269: free(val);
270: val = NULL;
271:
272: if (*qs != '\0')
273: qs++;
1.1 schwarze 274: }
275: }
276:
277: /*
278: * HTTP-decode a string. The standard explanation is that this turns
279: * "%4e+foo" into "n foo" in the regular way. This is done in-place
280: * over the allocated string.
281: */
282: static int
283: http_decode(char *p)
284: {
285: char hex[3];
1.3 tedu 286: char *q;
1.1 schwarze 287: int c;
288:
289: hex[2] = '\0';
290:
1.3 tedu 291: q = p;
292: for ( ; '\0' != *p; p++, q++) {
1.1 schwarze 293: if ('%' == *p) {
294: if ('\0' == (hex[0] = *(p + 1)))
1.48 schwarze 295: return 0;
1.1 schwarze 296: if ('\0' == (hex[1] = *(p + 2)))
1.48 schwarze 297: return 0;
1.1 schwarze 298: if (1 != sscanf(hex, "%x", &c))
1.48 schwarze 299: return 0;
1.1 schwarze 300: if ('\0' == c)
1.48 schwarze 301: return 0;
1.1 schwarze 302:
1.3 tedu 303: *q = (char)c;
304: p += 2;
1.1 schwarze 305: } else
1.3 tedu 306: *q = '+' == *p ? ' ' : *p;
1.1 schwarze 307: }
308:
1.3 tedu 309: *q = '\0';
1.48 schwarze 310: return 1;
1.1 schwarze 311: }
312:
313: static void
314: resp_begin_http(int code, const char *msg)
315: {
316:
317: if (200 != code)
1.2 tedu 318: printf("Status: %d %s\r\n", code, msg);
1.1 schwarze 319:
1.2 tedu 320: printf("Content-Type: text/html; charset=utf-8\r\n"
321: "Cache-Control: no-cache\r\n"
322: "Pragma: no-cache\r\n"
323: "\r\n");
1.1 schwarze 324:
325: fflush(stdout);
326: }
327:
328: static void
1.53 schwarze 329: resp_copy(const char *filename)
330: {
331: char buf[4096];
332: ssize_t sz;
333: int fd;
334:
335: if ((fd = open(filename, O_RDONLY)) != -1) {
336: fflush(stdout);
337: while ((sz = read(fd, buf, sizeof(buf))) > 0)
338: write(STDOUT_FILENO, buf, sz);
1.77 jsg 339: close(fd);
1.53 schwarze 340: }
341: }
342:
343: static void
1.1 schwarze 344: resp_begin_html(int code, const char *msg)
345: {
346:
347: resp_begin_http(code, msg);
348:
1.37 schwarze 349: printf("<!DOCTYPE html>\n"
1.65 schwarze 350: "<html>\n"
351: "<head>\n"
1.82 ! schwarze 352: " <meta charset=\"UTF-8\"/>\n"
! 353: " <link rel=\"stylesheet\" href=\"%s/mandoc.css\""
1.65 schwarze 354: " type=\"text/css\" media=\"all\">\n"
1.82 ! schwarze 355: " <title>%s</title>\n"
1.65 schwarze 356: "</head>\n"
1.81 schwarze 357: "<body>\n",
1.52 schwarze 358: CSS_DIR, CUSTOMIZE_TITLE);
1.53 schwarze 359:
360: resp_copy(MAN_DIR "/header.html");
1.1 schwarze 361: }
362:
363: static void
364: resp_end_html(void)
365: {
366:
1.53 schwarze 367: resp_copy(MAN_DIR "/footer.html");
368:
1.65 schwarze 369: puts("</body>\n"
370: "</html>");
1.1 schwarze 371: }
372:
373: static void
1.70 schwarze 374: resp_searchform(const struct req *req, enum focus focus)
1.1 schwarze 375: {
376: int i;
377:
1.81 schwarze 378: printf("<form action=\"/%s\" method=\"get\">\n"
1.82 ! schwarze 379: " <fieldset>\n"
! 380: " <legend>Manual Page Search Parameters</legend>\n",
1.1 schwarze 381: scriptname);
1.8 schwarze 382:
383: /* Write query input box. */
384:
1.82 ! schwarze 385: printf(" <input type=\"text\" name=\"query\" value=\"");
1.70 schwarze 386: if (req->q.query != NULL)
1.25 schwarze 387: html_print(req->q.query);
1.70 schwarze 388: printf( "\" size=\"40\"");
389: if (focus == FOCUS_QUERY)
390: printf(" autofocus");
391: puts(">");
1.8 schwarze 392:
1.71 schwarze 393: /* Write submission buttons. */
1.8 schwarze 394:
1.82 ! schwarze 395: printf( " <button type=\"submit\" name=\"apropos\" value=\"0\">"
1.71 schwarze 396: "man</button>\n"
1.82 ! schwarze 397: " <button type=\"submit\" name=\"apropos\" value=\"1\">"
! 398: "apropos</button>\n"
! 399: " <br/>\n");
1.8 schwarze 400:
401: /* Write section selector. */
402:
1.82 ! schwarze 403: puts(" <select name=\"sec\">");
1.8 schwarze 404: for (i = 0; i < sec_MAX; i++) {
1.82 ! schwarze 405: printf(" <option value=\"%s\"", sec_numbers[i]);
1.8 schwarze 406: if (NULL != req->q.sec &&
407: 0 == strcmp(sec_numbers[i], req->q.sec))
1.65 schwarze 408: printf(" selected=\"selected\"");
409: printf(">%s</option>\n", sec_names[i]);
1.8 schwarze 410: }
1.82 ! schwarze 411: puts(" </select>");
1.8 schwarze 412:
413: /* Write architecture selector. */
414:
1.82 ! schwarze 415: printf( " <select name=\"arch\">\n"
! 416: " <option value=\"default\"");
1.21 schwarze 417: if (NULL == req->q.arch)
1.65 schwarze 418: printf(" selected=\"selected\"");
419: puts(">All Architectures</option>");
1.8 schwarze 420: for (i = 0; i < arch_MAX; i++) {
1.82 ! schwarze 421: printf(" <option value=\"%s\"", arch_names[i]);
1.8 schwarze 422: if (NULL != req->q.arch &&
423: 0 == strcmp(arch_names[i], req->q.arch))
1.65 schwarze 424: printf(" selected=\"selected\"");
425: printf(">%s</option>\n", arch_names[i]);
1.8 schwarze 426: }
1.82 ! schwarze 427: puts(" </select>");
1.8 schwarze 428:
429: /* Write manpath selector. */
430:
1.1 schwarze 431: if (req->psz > 1) {
1.82 ! schwarze 432: puts(" <select name=\"manpath\">");
1.1 schwarze 433: for (i = 0; i < (int)req->psz; i++) {
1.82 ! schwarze 434: printf(" <option ");
1.41 schwarze 435: if (strcmp(req->q.manpath, req->p[i]) == 0)
1.65 schwarze 436: printf("selected=\"selected\" ");
437: printf("value=\"");
1.1 schwarze 438: html_print(req->p[i]);
439: printf("\">");
440: html_print(req->p[i]);
1.65 schwarze 441: puts("</option>");
1.1 schwarze 442: }
1.82 ! schwarze 443: puts(" </select>");
1.1 schwarze 444: }
1.8 schwarze 445:
1.82 ! schwarze 446: puts(" </fieldset>\n"
1.81 schwarze 447: "</form>");
1.1 schwarze 448: }
449:
1.16 schwarze 450: static int
1.20 schwarze 451: validate_urifrag(const char *frag)
452: {
453:
454: while ('\0' != *frag) {
455: if ( ! (isalnum((unsigned char)*frag) ||
456: '-' == *frag || '.' == *frag ||
457: '/' == *frag || '_' == *frag))
1.48 schwarze 458: return 0;
1.20 schwarze 459: frag++;
460: }
1.48 schwarze 461: return 1;
1.20 schwarze 462: }
463:
464: static int
1.17 schwarze 465: validate_manpath(const struct req *req, const char* manpath)
466: {
467: size_t i;
468:
469: for (i = 0; i < req->psz; i++)
470: if ( ! strcmp(manpath, req->p[i]))
1.48 schwarze 471: return 1;
1.17 schwarze 472:
1.48 schwarze 473: return 0;
1.17 schwarze 474: }
475:
476: static int
1.16 schwarze 477: validate_filename(const char *file)
478: {
479:
480: if ('.' == file[0] && '/' == file[1])
481: file += 2;
482:
1.48 schwarze 483: return ! (strstr(file, "../") || strstr(file, "/..") ||
484: (strncmp(file, "man", 3) && strncmp(file, "cat", 3)));
1.16 schwarze 485: }
486:
1.1 schwarze 487: static void
1.12 schwarze 488: pg_index(const struct req *req)
1.1 schwarze 489: {
490:
491: resp_begin_html(200, NULL);
1.70 schwarze 492: resp_searchform(req, FOCUS_QUERY);
1.65 schwarze 493: printf("<p>\n"
1.26 schwarze 494: "This web interface is documented in the\n"
1.73 schwarze 495: "<a href=\"/%s%sman.cgi.8\">man.cgi(8)</a>\n"
1.26 schwarze 496: "manual, and the\n"
1.73 schwarze 497: "<a href=\"/%s%sapropos.1\">apropos(1)</a>\n"
1.9 schwarze 498: "manual explains the query syntax.\n"
1.65 schwarze 499: "</p>\n",
1.58 schwarze 500: scriptname, *scriptname == '\0' ? "" : "/",
501: scriptname, *scriptname == '\0' ? "" : "/");
1.1 schwarze 502: resp_end_html();
503: }
504:
505: static void
1.12 schwarze 506: pg_noresult(const struct req *req, const char *msg)
1.1 schwarze 507: {
508: resp_begin_html(200, NULL);
1.70 schwarze 509: resp_searchform(req, FOCUS_QUERY);
1.65 schwarze 510: puts("<p>");
1.1 schwarze 511: puts(msg);
1.65 schwarze 512: puts("</p>");
1.1 schwarze 513: resp_end_html();
514: }
515:
516: static void
1.12 schwarze 517: pg_error_badrequest(const char *msg)
1.1 schwarze 518: {
519:
520: resp_begin_html(400, "Bad Request");
1.65 schwarze 521: puts("<h1>Bad Request</h1>\n"
522: "<p>\n");
1.1 schwarze 523: puts(msg);
524: printf("Try again from the\n"
1.65 schwarze 525: "<a href=\"/%s\">main page</a>.\n"
526: "</p>", scriptname);
1.1 schwarze 527: resp_end_html();
528: }
529:
530: static void
1.12 schwarze 531: pg_error_internal(void)
1.1 schwarze 532: {
533: resp_begin_html(500, "Internal Server Error");
1.65 schwarze 534: puts("<p>Internal Server Error</p>");
1.1 schwarze 535: resp_end_html();
536: }
537:
538: static void
1.12 schwarze 539: pg_searchres(const struct req *req, struct manpage *r, size_t sz)
1.1 schwarze 540: {
1.21 schwarze 541: char *arch, *archend;
1.59 schwarze 542: const char *sec;
543: size_t i, iuse;
1.21 schwarze 544: int archprio, archpriouse;
1.10 schwarze 545: int prio, priouse;
1.1 schwarze 546:
1.16 schwarze 547: for (i = 0; i < sz; i++) {
548: if (validate_filename(r[i].file))
549: continue;
1.67 schwarze 550: warnx("invalid filename %s in %s database",
1.16 schwarze 551: r[i].file, req->q.manpath);
552: pg_error_internal();
553: return;
554: }
555:
1.60 schwarze 556: if (req->isquery && sz == 1) {
1.1 schwarze 557: /*
558: * If we have just one result, then jump there now
559: * without any delay.
560: */
1.2 tedu 561: printf("Status: 303 See Other\r\n");
1.58 schwarze 562: printf("Location: http://%s/%s%s%s/%s",
563: HTTP_HOST, scriptname,
564: *scriptname == '\0' ? "" : "/",
565: req->q.manpath, r[0].file);
1.2 tedu 566: printf("\r\n"
567: "Content-Type: text/html; charset=utf-8\r\n"
568: "\r\n");
1.1 schwarze 569: return;
570: }
571:
572: resp_begin_html(200, NULL);
1.70 schwarze 573: resp_searchform(req,
574: req->q.equal || sz == 1 ? FOCUS_NONE : FOCUS_QUERY);
1.1 schwarze 575:
1.62 schwarze 576: if (sz > 1) {
1.65 schwarze 577: puts("<div class=\"results\">");
578: puts("<table>");
1.62 schwarze 579:
580: for (i = 0; i < sz; i++) {
1.82 ! schwarze 581: printf(" <tr>\n"
! 582: " <td class=\"title\">"
1.65 schwarze 583: "<a href=\"/%s%s%s/%s",
1.62 schwarze 584: scriptname, *scriptname == '\0' ? "" : "/",
585: req->q.manpath, r[i].file);
586: printf("\">");
587: html_print(r[i].names);
1.82 ! schwarze 588: printf("</a></td>\n"
! 589: " <td class=\"desc\">");
1.62 schwarze 590: html_print(r[i].output);
1.65 schwarze 591: puts("</td>\n"
1.82 ! schwarze 592: " </tr>");
1.62 schwarze 593: }
594:
1.65 schwarze 595: puts("</table>\n"
596: "</div>");
1.1 schwarze 597: }
598:
1.10 schwarze 599: /*
600: * In man(1) mode, show one of the pages
601: * even if more than one is found.
602: */
603:
1.62 schwarze 604: if (req->q.equal || sz == 1) {
1.65 schwarze 605: puts("<hr>");
1.10 schwarze 606: iuse = 0;
1.59 schwarze 607: priouse = 20;
1.21 schwarze 608: archpriouse = 3;
1.10 schwarze 609: for (i = 0; i < sz; i++) {
1.59 schwarze 610: sec = r[i].file;
611: sec += strcspn(sec, "123456789");
612: if (sec[0] == '\0')
1.10 schwarze 613: continue;
1.59 schwarze 614: prio = sec_prios[sec[0] - '1'];
615: if (sec[1] != '/')
616: prio += 10;
617: if (req->q.arch == NULL) {
1.21 schwarze 618: archprio =
1.59 schwarze 619: ((arch = strchr(sec + 1, '/'))
620: == NULL) ? 3 :
621: ((archend = strchr(arch + 1, '/'))
622: == NULL) ? 0 :
1.21 schwarze 623: strncmp(arch, "amd64/",
624: archend - arch) ? 2 : 1;
625: if (archprio < archpriouse) {
626: archpriouse = archprio;
627: priouse = prio;
628: iuse = i;
629: continue;
630: }
631: if (archprio > archpriouse)
632: continue;
633: }
1.10 schwarze 634: if (prio >= priouse)
635: continue;
636: priouse = prio;
637: iuse = i;
638: }
639: resp_show(req, r[iuse].file);
640: }
641:
1.1 schwarze 642: resp_end_html();
643: }
644:
645: static void
1.68 schwarze 646: resp_catman(const struct req *req, const char *file)
1.1 schwarze 647: {
648: FILE *f;
1.54 schwarze 649: char *p;
650: size_t sz;
651: ssize_t len;
1.1 schwarze 652: int i;
653: int italic, bold;
654:
1.54 schwarze 655: if ((f = fopen(file, "r")) == NULL) {
1.65 schwarze 656: puts("<p>You specified an invalid manual file.</p>");
1.1 schwarze 657: return;
658: }
659:
1.65 schwarze 660: puts("<div class=\"catman\">\n"
661: "<pre>");
1.1 schwarze 662:
1.54 schwarze 663: p = NULL;
664: sz = 0;
665:
666: while ((len = getline(&p, &sz, f)) != -1) {
1.1 schwarze 667: bold = italic = 0;
1.54 schwarze 668: for (i = 0; i < len - 1; i++) {
1.43 schwarze 669: /*
1.1 schwarze 670: * This means that the catpage is out of state.
671: * Ignore it and keep going (although the
672: * catpage is bogus).
673: */
674:
675: if ('\b' == p[i] || '\n' == p[i])
676: continue;
677:
678: /*
679: * Print a regular character.
680: * Close out any bold/italic scopes.
681: * If we're in back-space mode, make sure we'll
682: * have something to enter when we backspace.
683: */
684:
685: if ('\b' != p[i + 1]) {
686: if (italic)
1.65 schwarze 687: printf("</i>");
1.1 schwarze 688: if (bold)
1.65 schwarze 689: printf("</b>");
1.1 schwarze 690: italic = bold = 0;
691: html_putchar(p[i]);
692: continue;
1.54 schwarze 693: } else if (i + 2 >= len)
1.1 schwarze 694: continue;
695:
696: /* Italic mode. */
697:
698: if ('_' == p[i]) {
699: if (bold)
1.65 schwarze 700: printf("</b>");
1.1 schwarze 701: if ( ! italic)
1.65 schwarze 702: printf("<i>");
1.1 schwarze 703: bold = 0;
704: italic = 1;
705: i += 2;
706: html_putchar(p[i]);
707: continue;
708: }
709:
1.43 schwarze 710: /*
1.1 schwarze 711: * Handle funny behaviour troff-isms.
712: * These grok'd from the original man2html.c.
713: */
714:
715: if (('+' == p[i] && 'o' == p[i + 2]) ||
716: ('o' == p[i] && '+' == p[i + 2]) ||
717: ('|' == p[i] && '=' == p[i + 2]) ||
718: ('=' == p[i] && '|' == p[i + 2]) ||
719: ('*' == p[i] && '=' == p[i + 2]) ||
720: ('=' == p[i] && '*' == p[i + 2]) ||
721: ('*' == p[i] && '|' == p[i + 2]) ||
722: ('|' == p[i] && '*' == p[i + 2])) {
723: if (italic)
1.65 schwarze 724: printf("</i>");
1.1 schwarze 725: if (bold)
1.65 schwarze 726: printf("</b>");
1.1 schwarze 727: italic = bold = 0;
728: putchar('*');
729: i += 2;
730: continue;
731: } else if (('|' == p[i] && '-' == p[i + 2]) ||
732: ('-' == p[i] && '|' == p[i + 1]) ||
733: ('+' == p[i] && '-' == p[i + 1]) ||
734: ('-' == p[i] && '+' == p[i + 1]) ||
735: ('+' == p[i] && '|' == p[i + 1]) ||
736: ('|' == p[i] && '+' == p[i + 1])) {
737: if (italic)
1.65 schwarze 738: printf("</i>");
1.1 schwarze 739: if (bold)
1.65 schwarze 740: printf("</b>");
1.1 schwarze 741: italic = bold = 0;
742: putchar('+');
743: i += 2;
744: continue;
745: }
746:
747: /* Bold mode. */
1.43 schwarze 748:
1.1 schwarze 749: if (italic)
1.65 schwarze 750: printf("</i>");
1.1 schwarze 751: if ( ! bold)
1.65 schwarze 752: printf("<b>");
1.1 schwarze 753: bold = 1;
754: italic = 0;
755: i += 2;
756: html_putchar(p[i]);
757: }
758:
1.43 schwarze 759: /*
1.1 schwarze 760: * Clean up the last character.
1.43 schwarze 761: * We can get to a newline; don't print that.
1.1 schwarze 762: */
763:
764: if (italic)
1.65 schwarze 765: printf("</i>");
1.1 schwarze 766: if (bold)
1.65 schwarze 767: printf("</b>");
1.1 schwarze 768:
1.54 schwarze 769: if (i == len - 1 && p[i] != '\n')
1.1 schwarze 770: html_putchar(p[i]);
771:
772: putchar('\n');
773: }
1.54 schwarze 774: free(p);
1.1 schwarze 775:
1.65 schwarze 776: puts("</pre>\n"
777: "</div>");
1.1 schwarze 778:
779: fclose(f);
780: }
781:
782: static void
1.68 schwarze 783: resp_format(const struct req *req, const char *file)
1.1 schwarze 784: {
1.45 schwarze 785: struct manoutput conf;
1.1 schwarze 786: struct mparse *mp;
1.46 schwarze 787: struct roff_man *man;
1.1 schwarze 788: void *vp;
1.30 schwarze 789: int fd;
790: int usepath;
1.1 schwarze 791:
792: if (-1 == (fd = open(file, O_RDONLY, 0))) {
1.65 schwarze 793: puts("<p>You specified an invalid manual file.</p>");
1.1 schwarze 794: return;
795: }
796:
1.49 schwarze 797: mchars_alloc();
1.75 schwarze 798: mp = mparse_alloc(MPARSE_SO | MPARSE_UTF8 | MPARSE_LATIN1,
799: MANDOCLEVEL_BADARG, NULL, req->q.manpath);
1.42 schwarze 800: mparse_readfd(mp, fd, file);
1.1 schwarze 801: close(fd);
802:
1.45 schwarze 803: memset(&conf, 0, sizeof(conf));
804: conf.fragment = 1;
1.30 schwarze 805: usepath = strcmp(req->q.manpath, req->p[0]);
1.61 schwarze 806: mandoc_asprintf(&conf.man, "/%s%s%%N.%%S",
807: usepath ? req->q.manpath : "", usepath ? "/" : "");
1.1 schwarze 808:
1.47 schwarze 809: mparse_result(mp, &man, NULL);
810: if (man == NULL) {
1.67 schwarze 811: warnx("fatal mandoc error: %s/%s", req->q.manpath, file);
1.12 schwarze 812: pg_error_internal();
1.1 schwarze 813: mparse_free(mp);
1.49 schwarze 814: mchars_free();
1.1 schwarze 815: return;
816: }
817:
1.49 schwarze 818: vp = html_alloc(&conf);
1.1 schwarze 819:
1.50 schwarze 820: if (man->macroset == MACROSET_MDOC) {
821: mdoc_validate(man);
1.47 schwarze 822: html_mdoc(vp, man);
1.51 schwarze 823: } else {
824: man_validate(man);
1.1 schwarze 825: html_man(vp, man);
1.51 schwarze 826: }
1.1 schwarze 827:
828: html_free(vp);
829: mparse_free(mp);
1.49 schwarze 830: mchars_free();
1.45 schwarze 831: free(conf.man);
1.1 schwarze 832: }
833:
834: static void
1.10 schwarze 835: resp_show(const struct req *req, const char *file)
836: {
1.16 schwarze 837:
838: if ('.' == file[0] && '/' == file[1])
1.11 schwarze 839: file += 2;
1.10 schwarze 840:
841: if ('c' == *file)
1.68 schwarze 842: resp_catman(req, file);
1.10 schwarze 843: else
1.68 schwarze 844: resp_format(req, file);
1.10 schwarze 845: }
846:
847: static void
1.24 schwarze 848: pg_show(struct req *req, const char *fullpath)
1.1 schwarze 849: {
1.24 schwarze 850: char *manpath;
851: const char *file;
1.1 schwarze 852:
1.24 schwarze 853: if ((file = strchr(fullpath, '/')) == NULL) {
1.12 schwarze 854: pg_error_badrequest(
1.1 schwarze 855: "You did not specify a page to show.");
856: return;
1.43 schwarze 857: }
1.24 schwarze 858: manpath = mandoc_strndup(fullpath, file - fullpath);
859: file++;
1.1 schwarze 860:
1.24 schwarze 861: if ( ! validate_manpath(req, manpath)) {
1.17 schwarze 862: pg_error_badrequest(
863: "You specified an invalid manpath.");
1.24 schwarze 864: free(manpath);
1.17 schwarze 865: return;
866: }
867:
1.1 schwarze 868: /*
869: * Begin by chdir()ing into the manpath.
870: * This way we can pick up the database files, which are
871: * relative to the manpath root.
872: */
873:
1.24 schwarze 874: if (chdir(manpath) == -1) {
1.67 schwarze 875: warn("chdir %s", manpath);
1.17 schwarze 876: pg_error_internal();
1.24 schwarze 877: free(manpath);
1.16 schwarze 878: return;
879: }
1.73 schwarze 880: free(manpath);
1.24 schwarze 881:
882: if ( ! validate_filename(file)) {
1.16 schwarze 883: pg_error_badrequest(
884: "You specified an invalid manual file.");
1.1 schwarze 885: return;
886: }
1.19 schwarze 887:
1.10 schwarze 888: resp_begin_html(200, NULL);
1.70 schwarze 889: resp_searchform(req, FOCUS_NONE);
1.24 schwarze 890: resp_show(req, file);
1.10 schwarze 891: resp_end_html();
1.1 schwarze 892: }
893:
894: static void
1.6 schwarze 895: pg_search(const struct req *req)
1.1 schwarze 896: {
897: struct mansearch search;
898: struct manpaths paths;
899: struct manpage *res;
1.36 schwarze 900: char **argv;
901: char *query, *rp, *wp;
1.1 schwarze 902: size_t ressz;
1.36 schwarze 903: int argc;
1.1 schwarze 904:
905: /*
906: * Begin by chdir()ing into the root of the manpath.
907: * This way we can pick up the database files, which are
908: * relative to the manpath root.
909: */
910:
1.67 schwarze 911: if (chdir(req->q.manpath) == -1) {
912: warn("chdir %s", req->q.manpath);
1.17 schwarze 913: pg_error_internal();
1.1 schwarze 914: return;
915: }
916:
917: search.arch = req->q.arch;
918: search.sec = req->q.sec;
1.35 schwarze 919: search.outkey = "Nd";
920: search.argmode = req->q.equal ? ARG_NAME : ARG_EXPR;
1.40 schwarze 921: search.firstmatch = 1;
1.1 schwarze 922:
923: paths.sz = 1;
924: paths.paths = mandoc_malloc(sizeof(char *));
925: paths.paths[0] = mandoc_strdup(".");
926:
927: /*
1.36 schwarze 928: * Break apart at spaces with backslash-escaping.
1.1 schwarze 929: */
930:
1.36 schwarze 931: argc = 0;
932: argv = NULL;
933: rp = query = mandoc_strdup(req->q.query);
934: for (;;) {
935: while (isspace((unsigned char)*rp))
936: rp++;
937: if (*rp == '\0')
938: break;
939: argv = mandoc_reallocarray(argv, argc + 1, sizeof(char *));
940: argv[argc++] = wp = rp;
941: for (;;) {
942: if (isspace((unsigned char)*rp)) {
943: *wp = '\0';
944: rp++;
945: break;
946: }
947: if (rp[0] == '\\' && rp[1] != '\0')
948: rp++;
949: if (wp != rp)
950: *wp = *rp;
951: if (*rp == '\0')
952: break;
953: wp++;
954: rp++;
955: }
1.1 schwarze 956: }
957:
1.36 schwarze 958: if (0 == mansearch(&search, &paths, argc, argv, &res, &ressz))
1.12 schwarze 959: pg_noresult(req, "You entered an invalid query.");
1.1 schwarze 960: else if (0 == ressz)
1.12 schwarze 961: pg_noresult(req, "No results found.");
1.1 schwarze 962: else
1.12 schwarze 963: pg_searchres(req, res, ressz);
1.1 schwarze 964:
1.36 schwarze 965: free(query);
966: mansearch_free(res, ressz);
1.1 schwarze 967: free(paths.paths[0]);
968: free(paths.paths);
969: }
970:
971: int
972: main(void)
973: {
1.6 schwarze 974: struct req req;
1.33 schwarze 975: struct itimerval itimer;
1.6 schwarze 976: const char *path;
1.23 schwarze 977: const char *querystring;
1.1 schwarze 978: int i;
1.33 schwarze 979:
980: /* Poor man's ReDoS mitigation. */
981:
1.38 schwarze 982: itimer.it_value.tv_sec = 2;
1.33 schwarze 983: itimer.it_value.tv_usec = 0;
1.38 schwarze 984: itimer.it_interval.tv_sec = 2;
1.33 schwarze 985: itimer.it_interval.tv_usec = 0;
986: if (setitimer(ITIMER_VIRTUAL, &itimer, NULL) == -1) {
1.67 schwarze 987: warn("setitimer");
1.20 schwarze 988: pg_error_internal();
1.48 schwarze 989: return EXIT_FAILURE;
1.20 schwarze 990: }
991:
1.1 schwarze 992: /*
1.7 schwarze 993: * First we change directory into the MAN_DIR so that
1.1 schwarze 994: * subsequent scanning for manpath directories is rooted
995: * relative to the same position.
996: */
997:
1.67 schwarze 998: if (chdir(MAN_DIR) == -1) {
999: warn("MAN_DIR: %s", MAN_DIR);
1.12 schwarze 1000: pg_error_internal();
1.48 schwarze 1001: return EXIT_FAILURE;
1.43 schwarze 1002: }
1.1 schwarze 1003:
1004: memset(&req, 0, sizeof(struct req));
1.57 schwarze 1005: req.q.equal = 1;
1.68 schwarze 1006: parse_manpath_conf(&req);
1.1 schwarze 1007:
1.56 schwarze 1008: /* Parse the path info and the query string. */
1.1 schwarze 1009:
1.56 schwarze 1010: if ((path = getenv("PATH_INFO")) == NULL)
1011: path = "";
1012: else if (*path == '/')
1013: path++;
1014:
1.63 schwarze 1015: if (*path != '\0') {
1.68 schwarze 1016: parse_path_info(&req, path);
1.72 schwarze 1017: if (req.q.manpath == NULL || access(path, F_OK) == -1)
1.63 schwarze 1018: path = "";
1.56 schwarze 1019: } else if ((querystring = getenv("QUERY_STRING")) != NULL)
1.68 schwarze 1020: parse_query_string(&req, querystring);
1.17 schwarze 1021:
1.56 schwarze 1022: /* Validate parsed data and add defaults. */
1023:
1.41 schwarze 1024: if (req.q.manpath == NULL)
1025: req.q.manpath = mandoc_strdup(req.p[0]);
1026: else if ( ! validate_manpath(&req, req.q.manpath)) {
1.17 schwarze 1027: pg_error_badrequest(
1028: "You specified an invalid manpath.");
1.48 schwarze 1029: return EXIT_FAILURE;
1.17 schwarze 1030: }
1.1 schwarze 1031:
1.20 schwarze 1032: if ( ! (NULL == req.q.arch || validate_urifrag(req.q.arch))) {
1033: pg_error_badrequest(
1034: "You specified an invalid architecture.");
1.48 schwarze 1035: return EXIT_FAILURE;
1.20 schwarze 1036: }
1037:
1.6 schwarze 1038: /* Dispatch to the three different pages. */
1.1 schwarze 1039:
1.6 schwarze 1040: if ('\0' != *path)
1041: pg_show(&req, path);
1.25 schwarze 1042: else if (NULL != req.q.query)
1.6 schwarze 1043: pg_search(&req);
1044: else
1.12 schwarze 1045: pg_index(&req);
1.1 schwarze 1046:
1.23 schwarze 1047: free(req.q.manpath);
1048: free(req.q.arch);
1049: free(req.q.sec);
1.25 schwarze 1050: free(req.q.query);
1.1 schwarze 1051: for (i = 0; i < (int)req.psz; i++)
1052: free(req.p[i]);
1053: free(req.p);
1.48 schwarze 1054: return EXIT_SUCCESS;
1.56 schwarze 1055: }
1056:
1057: /*
1058: * If PATH_INFO is not a file name, translate it to a query.
1059: */
1060: static void
1.68 schwarze 1061: parse_path_info(struct req *req, const char *path)
1.56 schwarze 1062: {
1.74 schwarze 1063: char *dir[4];
1064: int i;
1.56 schwarze 1065:
1.60 schwarze 1066: req->isquery = 0;
1.56 schwarze 1067: req->q.equal = 1;
1068: req->q.manpath = mandoc_strdup(path);
1.74 schwarze 1069: req->q.arch = NULL;
1.56 schwarze 1070:
1071: /* Mandatory manual page name. */
1072: if ((req->q.query = strrchr(req->q.manpath, '/')) == NULL) {
1073: req->q.query = req->q.manpath;
1074: req->q.manpath = NULL;
1075: } else
1076: *req->q.query++ = '\0';
1077:
1078: /* Optional trailing section. */
1079: if ((req->q.sec = strrchr(req->q.query, '.')) != NULL) {
1080: if(isdigit((unsigned char)req->q.sec[1])) {
1081: *req->q.sec++ = '\0';
1082: req->q.sec = mandoc_strdup(req->q.sec);
1083: } else
1084: req->q.sec = NULL;
1085: }
1086:
1087: /* Handle the case of name[.section] only. */
1.74 schwarze 1088: if (req->q.manpath == NULL)
1.56 schwarze 1089: return;
1.74 schwarze 1090: req->q.query = mandoc_strdup(req->q.query);
1091:
1092: /* Split directory components. */
1093: dir[i = 0] = req->q.manpath;
1094: while ((dir[i + 1] = strchr(dir[i], '/')) != NULL) {
1095: if (++i == 3) {
1096: pg_error_badrequest(
1097: "You specified too many directory components.");
1098: exit(EXIT_FAILURE);
1099: }
1100: *dir[i]++ = '\0';
1.56 schwarze 1101: }
1102:
1.74 schwarze 1103: /* Optional manpath. */
1104: if ((i = validate_manpath(req, req->q.manpath)) == 0)
1105: req->q.manpath = NULL;
1106: else if (dir[1] == NULL)
1107: return;
1.56 schwarze 1108:
1.74 schwarze 1109: /* Optional section. */
1110: if (strncmp(dir[i], "man", 3) == 0) {
1.66 schwarze 1111: free(req->q.sec);
1.74 schwarze 1112: req->q.sec = mandoc_strdup(dir[i++] + 3);
1.56 schwarze 1113: }
1.74 schwarze 1114: if (dir[i] == NULL) {
1115: if (req->q.manpath == NULL)
1116: free(dir[0]);
1117: return;
1118: }
1119: if (dir[i + 1] != NULL) {
1120: pg_error_badrequest(
1121: "You specified an invalid directory component.");
1122: exit(EXIT_FAILURE);
1123: }
1124:
1125: /* Optional architecture. */
1126: if (i) {
1127: req->q.arch = mandoc_strdup(dir[i]);
1128: if (req->q.manpath == NULL)
1129: free(dir[0]);
1130: } else
1131: req->q.arch = dir[0];
1.1 schwarze 1132: }
1133:
1134: /*
1135: * Scan for indexable paths.
1136: */
1137: static void
1.68 schwarze 1138: parse_manpath_conf(struct req *req)
1.1 schwarze 1139: {
1140: FILE *fp;
1141: char *dp;
1142: size_t dpsz;
1.54 schwarze 1143: ssize_t len;
1.1 schwarze 1144:
1.67 schwarze 1145: if ((fp = fopen("manpath.conf", "r")) == NULL) {
1146: warn("%s/manpath.conf", MAN_DIR);
1.14 schwarze 1147: pg_error_internal();
1148: exit(EXIT_FAILURE);
1149: }
1.1 schwarze 1150:
1.54 schwarze 1151: dp = NULL;
1152: dpsz = 0;
1153:
1154: while ((len = getline(&dp, &dpsz, fp)) != -1) {
1155: if (dp[len - 1] == '\n')
1156: dp[--len] = '\0';
1.1 schwarze 1157: req->p = mandoc_realloc(req->p,
1158: (req->psz + 1) * sizeof(char *));
1.20 schwarze 1159: if ( ! validate_urifrag(dp)) {
1.67 schwarze 1160: warnx("%s/manpath.conf contains "
1161: "unsafe path \"%s\"", MAN_DIR, dp);
1.20 schwarze 1162: pg_error_internal();
1163: exit(EXIT_FAILURE);
1164: }
1.67 schwarze 1165: if (strchr(dp, '/') != NULL) {
1166: warnx("%s/manpath.conf contains "
1167: "path with slash \"%s\"", MAN_DIR, dp);
1.20 schwarze 1168: pg_error_internal();
1169: exit(EXIT_FAILURE);
1170: }
1171: req->p[req->psz++] = dp;
1.54 schwarze 1172: dp = NULL;
1173: dpsz = 0;
1.14 schwarze 1174: }
1.54 schwarze 1175: free(dp);
1.14 schwarze 1176:
1.67 schwarze 1177: if (req->p == NULL) {
1178: warnx("%s/manpath.conf is empty", MAN_DIR);
1.14 schwarze 1179: pg_error_internal();
1180: exit(EXIT_FAILURE);
1.1 schwarze 1181: }
1182: }