Annotation of src/usr.bin/mandoc/cgi.c, Revision 1.79
1.79 ! schwarze 1: /* $OpenBSD: cgi.c,v 1.78 2016/09/01 15:07:19 schwarze Exp $ */
1.1 schwarze 2: /*
3: * Copyright (c) 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
1.57 schwarze 4: * Copyright (c) 2014, 2015, 2016 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 ('"'):
139: printf(""e;");
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"
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"
1.1 schwarze 358: "<!-- Begin page content. //-->\n",
1.52 schwarze 359: CSS_DIR, CUSTOMIZE_TITLE);
1.53 schwarze 360:
361: resp_copy(MAN_DIR "/header.html");
1.1 schwarze 362: }
363:
364: static void
365: resp_end_html(void)
366: {
367:
1.53 schwarze 368: resp_copy(MAN_DIR "/footer.html");
369:
1.65 schwarze 370: puts("</body>\n"
371: "</html>");
1.1 schwarze 372: }
373:
374: static void
1.70 schwarze 375: resp_searchform(const struct req *req, enum focus focus)
1.1 schwarze 376: {
377: int i;
378:
379: puts("<!-- Begin search form. //-->");
1.65 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",
1.1 schwarze 384: scriptname);
1.8 schwarze 385:
386: /* Write query input box. */
387:
1.71 schwarze 388: printf("<input type=\"text\" name=\"query\" value=\"");
1.70 schwarze 389: if (req->q.query != NULL)
1.25 schwarze 390: html_print(req->q.query);
1.70 schwarze 391: printf( "\" size=\"40\"");
392: if (focus == FOCUS_QUERY)
393: printf(" autofocus");
394: puts(">");
1.8 schwarze 395:
1.71 schwarze 396: /* Write submission buttons. */
1.8 schwarze 397:
1.71 schwarze 398: printf( "<button type=\"submit\" name=\"apropos\" value=\"0\">"
399: "man</button>\n"
400: "<button type=\"submit\" name=\"apropos\" value=\"1\">"
401: "apropos</button>\n<br/>\n");
1.8 schwarze 402:
403: /* Write section selector. */
404:
1.71 schwarze 405: puts("<select name=\"sec\">");
1.8 schwarze 406: for (i = 0; i < sec_MAX; i++) {
1.65 schwarze 407: printf("<option value=\"%s\"", sec_numbers[i]);
1.8 schwarze 408: if (NULL != req->q.sec &&
409: 0 == strcmp(sec_numbers[i], req->q.sec))
1.65 schwarze 410: printf(" selected=\"selected\"");
411: printf(">%s</option>\n", sec_names[i]);
1.8 schwarze 412: }
1.65 schwarze 413: puts("</select>");
1.8 schwarze 414:
415: /* Write architecture selector. */
416:
1.65 schwarze 417: printf( "<select name=\"arch\">\n"
418: "<option value=\"default\"");
1.21 schwarze 419: if (NULL == req->q.arch)
1.65 schwarze 420: printf(" selected=\"selected\"");
421: puts(">All Architectures</option>");
1.8 schwarze 422: for (i = 0; i < arch_MAX; i++) {
1.65 schwarze 423: printf("<option value=\"%s\"", arch_names[i]);
1.8 schwarze 424: if (NULL != req->q.arch &&
425: 0 == strcmp(arch_names[i], req->q.arch))
1.65 schwarze 426: printf(" selected=\"selected\"");
427: printf(">%s</option>\n", arch_names[i]);
1.8 schwarze 428: }
1.65 schwarze 429: puts("</select>");
1.8 schwarze 430:
431: /* Write manpath selector. */
432:
1.1 schwarze 433: if (req->psz > 1) {
1.65 schwarze 434: puts("<select name=\"manpath\">");
1.1 schwarze 435: for (i = 0; i < (int)req->psz; i++) {
1.65 schwarze 436: printf("<option ");
1.41 schwarze 437: if (strcmp(req->q.manpath, req->p[i]) == 0)
1.65 schwarze 438: printf("selected=\"selected\" ");
439: printf("value=\"");
1.1 schwarze 440: html_print(req->p[i]);
441: printf("\">");
442: html_print(req->p[i]);
1.65 schwarze 443: puts("</option>");
1.1 schwarze 444: }
1.65 schwarze 445: puts("</select>");
1.1 schwarze 446: }
1.8 schwarze 447:
1.71 schwarze 448: puts("</fieldset>\n"
1.65 schwarze 449: "</form>\n"
450: "</div>");
1.1 schwarze 451: puts("<!-- End search form. //-->");
452: }
453:
1.16 schwarze 454: static int
1.20 schwarze 455: validate_urifrag(const char *frag)
456: {
457:
458: while ('\0' != *frag) {
459: if ( ! (isalnum((unsigned char)*frag) ||
460: '-' == *frag || '.' == *frag ||
461: '/' == *frag || '_' == *frag))
1.48 schwarze 462: return 0;
1.20 schwarze 463: frag++;
464: }
1.48 schwarze 465: return 1;
1.20 schwarze 466: }
467:
468: static int
1.17 schwarze 469: validate_manpath(const struct req *req, const char* manpath)
470: {
471: size_t i;
472:
473: for (i = 0; i < req->psz; i++)
474: if ( ! strcmp(manpath, req->p[i]))
1.48 schwarze 475: return 1;
1.17 schwarze 476:
1.48 schwarze 477: return 0;
1.17 schwarze 478: }
479:
480: static int
1.16 schwarze 481: validate_filename(const char *file)
482: {
483:
484: if ('.' == file[0] && '/' == file[1])
485: file += 2;
486:
1.48 schwarze 487: return ! (strstr(file, "../") || strstr(file, "/..") ||
488: (strncmp(file, "man", 3) && strncmp(file, "cat", 3)));
1.16 schwarze 489: }
490:
1.1 schwarze 491: static void
1.12 schwarze 492: pg_index(const struct req *req)
1.1 schwarze 493: {
494:
495: resp_begin_html(200, NULL);
1.70 schwarze 496: resp_searchform(req, FOCUS_QUERY);
1.65 schwarze 497: printf("<p>\n"
1.26 schwarze 498: "This web interface is documented in the\n"
1.73 schwarze 499: "<a href=\"/%s%sman.cgi.8\">man.cgi(8)</a>\n"
1.26 schwarze 500: "manual, and the\n"
1.73 schwarze 501: "<a href=\"/%s%sapropos.1\">apropos(1)</a>\n"
1.9 schwarze 502: "manual explains the query syntax.\n"
1.65 schwarze 503: "</p>\n",
1.58 schwarze 504: scriptname, *scriptname == '\0' ? "" : "/",
505: scriptname, *scriptname == '\0' ? "" : "/");
1.1 schwarze 506: resp_end_html();
507: }
508:
509: static void
1.12 schwarze 510: pg_noresult(const struct req *req, const char *msg)
1.1 schwarze 511: {
512: resp_begin_html(200, NULL);
1.70 schwarze 513: resp_searchform(req, FOCUS_QUERY);
1.65 schwarze 514: puts("<p>");
1.1 schwarze 515: puts(msg);
1.65 schwarze 516: puts("</p>");
1.1 schwarze 517: resp_end_html();
518: }
519:
520: static void
1.12 schwarze 521: pg_error_badrequest(const char *msg)
1.1 schwarze 522: {
523:
524: resp_begin_html(400, "Bad Request");
1.65 schwarze 525: puts("<h1>Bad Request</h1>\n"
526: "<p>\n");
1.1 schwarze 527: puts(msg);
528: printf("Try again from the\n"
1.65 schwarze 529: "<a href=\"/%s\">main page</a>.\n"
530: "</p>", scriptname);
1.1 schwarze 531: resp_end_html();
532: }
533:
534: static void
1.12 schwarze 535: pg_error_internal(void)
1.1 schwarze 536: {
537: resp_begin_html(500, "Internal Server Error");
1.65 schwarze 538: puts("<p>Internal Server Error</p>");
1.1 schwarze 539: resp_end_html();
540: }
541:
542: static void
1.12 schwarze 543: pg_searchres(const struct req *req, struct manpage *r, size_t sz)
1.1 schwarze 544: {
1.21 schwarze 545: char *arch, *archend;
1.59 schwarze 546: const char *sec;
547: size_t i, iuse;
1.21 schwarze 548: int archprio, archpriouse;
1.10 schwarze 549: int prio, priouse;
1.1 schwarze 550:
1.16 schwarze 551: for (i = 0; i < sz; i++) {
552: if (validate_filename(r[i].file))
553: continue;
1.67 schwarze 554: warnx("invalid filename %s in %s database",
1.16 schwarze 555: r[i].file, req->q.manpath);
556: pg_error_internal();
557: return;
558: }
559:
1.60 schwarze 560: if (req->isquery && sz == 1) {
1.1 schwarze 561: /*
562: * If we have just one result, then jump there now
563: * without any delay.
564: */
1.2 tedu 565: printf("Status: 303 See Other\r\n");
1.58 schwarze 566: printf("Location: http://%s/%s%s%s/%s",
567: HTTP_HOST, scriptname,
568: *scriptname == '\0' ? "" : "/",
569: req->q.manpath, r[0].file);
1.2 tedu 570: printf("\r\n"
571: "Content-Type: text/html; charset=utf-8\r\n"
572: "\r\n");
1.1 schwarze 573: return;
574: }
575:
576: resp_begin_html(200, NULL);
1.70 schwarze 577: resp_searchform(req,
578: req->q.equal || sz == 1 ? FOCUS_NONE : FOCUS_QUERY);
1.1 schwarze 579:
1.62 schwarze 580: if (sz > 1) {
1.65 schwarze 581: puts("<div class=\"results\">");
582: puts("<table>");
1.62 schwarze 583:
584: for (i = 0; i < sz; i++) {
1.65 schwarze 585: printf("<tr>\n"
586: "<td class=\"title\">\n"
587: "<a href=\"/%s%s%s/%s",
1.62 schwarze 588: scriptname, *scriptname == '\0' ? "" : "/",
589: req->q.manpath, r[i].file);
590: printf("\">");
591: html_print(r[i].names);
1.65 schwarze 592: printf("</a>\n"
593: "</td>\n"
594: "<td class=\"desc\">");
1.62 schwarze 595: html_print(r[i].output);
1.65 schwarze 596: puts("</td>\n"
597: "</tr>");
1.62 schwarze 598: }
599:
1.65 schwarze 600: puts("</table>\n"
601: "</div>");
1.1 schwarze 602: }
603:
1.10 schwarze 604: /*
605: * In man(1) mode, show one of the pages
606: * even if more than one is found.
607: */
608:
1.62 schwarze 609: if (req->q.equal || sz == 1) {
1.65 schwarze 610: puts("<hr>");
1.10 schwarze 611: iuse = 0;
1.59 schwarze 612: priouse = 20;
1.21 schwarze 613: archpriouse = 3;
1.10 schwarze 614: for (i = 0; i < sz; i++) {
1.59 schwarze 615: sec = r[i].file;
616: sec += strcspn(sec, "123456789");
617: if (sec[0] == '\0')
1.10 schwarze 618: continue;
1.59 schwarze 619: prio = sec_prios[sec[0] - '1'];
620: if (sec[1] != '/')
621: prio += 10;
622: if (req->q.arch == NULL) {
1.21 schwarze 623: archprio =
1.59 schwarze 624: ((arch = strchr(sec + 1, '/'))
625: == NULL) ? 3 :
626: ((archend = strchr(arch + 1, '/'))
627: == NULL) ? 0 :
1.21 schwarze 628: strncmp(arch, "amd64/",
629: archend - arch) ? 2 : 1;
630: if (archprio < archpriouse) {
631: archpriouse = archprio;
632: priouse = prio;
633: iuse = i;
634: continue;
635: }
636: if (archprio > archpriouse)
637: continue;
638: }
1.10 schwarze 639: if (prio >= priouse)
640: continue;
641: priouse = prio;
642: iuse = i;
643: }
644: resp_show(req, r[iuse].file);
645: }
646:
1.1 schwarze 647: resp_end_html();
648: }
649:
650: static void
1.68 schwarze 651: resp_catman(const struct req *req, const char *file)
1.1 schwarze 652: {
653: FILE *f;
1.54 schwarze 654: char *p;
655: size_t sz;
656: ssize_t len;
1.1 schwarze 657: int i;
658: int italic, bold;
659:
1.54 schwarze 660: if ((f = fopen(file, "r")) == NULL) {
1.65 schwarze 661: puts("<p>You specified an invalid manual file.</p>");
1.1 schwarze 662: return;
663: }
664:
1.65 schwarze 665: puts("<div class=\"catman\">\n"
666: "<pre>");
1.1 schwarze 667:
1.54 schwarze 668: p = NULL;
669: sz = 0;
670:
671: while ((len = getline(&p, &sz, f)) != -1) {
1.1 schwarze 672: bold = italic = 0;
1.54 schwarze 673: for (i = 0; i < len - 1; i++) {
1.43 schwarze 674: /*
1.1 schwarze 675: * This means that the catpage is out of state.
676: * Ignore it and keep going (although the
677: * catpage is bogus).
678: */
679:
680: if ('\b' == p[i] || '\n' == p[i])
681: continue;
682:
683: /*
684: * Print a regular character.
685: * Close out any bold/italic scopes.
686: * If we're in back-space mode, make sure we'll
687: * have something to enter when we backspace.
688: */
689:
690: if ('\b' != p[i + 1]) {
691: if (italic)
1.65 schwarze 692: printf("</i>");
1.1 schwarze 693: if (bold)
1.65 schwarze 694: printf("</b>");
1.1 schwarze 695: italic = bold = 0;
696: html_putchar(p[i]);
697: continue;
1.54 schwarze 698: } else if (i + 2 >= len)
1.1 schwarze 699: continue;
700:
701: /* Italic mode. */
702:
703: if ('_' == p[i]) {
704: if (bold)
1.65 schwarze 705: printf("</b>");
1.1 schwarze 706: if ( ! italic)
1.65 schwarze 707: printf("<i>");
1.1 schwarze 708: bold = 0;
709: italic = 1;
710: i += 2;
711: html_putchar(p[i]);
712: continue;
713: }
714:
1.43 schwarze 715: /*
1.1 schwarze 716: * Handle funny behaviour troff-isms.
717: * These grok'd from the original man2html.c.
718: */
719:
720: if (('+' == p[i] && 'o' == p[i + 2]) ||
721: ('o' == p[i] && '+' == p[i + 2]) ||
722: ('|' == p[i] && '=' == p[i + 2]) ||
723: ('=' == p[i] && '|' == p[i + 2]) ||
724: ('*' == p[i] && '=' == p[i + 2]) ||
725: ('=' == p[i] && '*' == p[i + 2]) ||
726: ('*' == p[i] && '|' == p[i + 2]) ||
727: ('|' == p[i] && '*' == p[i + 2])) {
728: if (italic)
1.65 schwarze 729: printf("</i>");
1.1 schwarze 730: if (bold)
1.65 schwarze 731: printf("</b>");
1.1 schwarze 732: italic = bold = 0;
733: putchar('*');
734: i += 2;
735: continue;
736: } else if (('|' == p[i] && '-' == p[i + 2]) ||
737: ('-' == p[i] && '|' == p[i + 1]) ||
738: ('+' == p[i] && '-' == p[i + 1]) ||
739: ('-' == p[i] && '+' == p[i + 1]) ||
740: ('+' == p[i] && '|' == p[i + 1]) ||
741: ('|' == p[i] && '+' == p[i + 1])) {
742: if (italic)
1.65 schwarze 743: printf("</i>");
1.1 schwarze 744: if (bold)
1.65 schwarze 745: printf("</b>");
1.1 schwarze 746: italic = bold = 0;
747: putchar('+');
748: i += 2;
749: continue;
750: }
751:
752: /* Bold mode. */
1.43 schwarze 753:
1.1 schwarze 754: if (italic)
1.65 schwarze 755: printf("</i>");
1.1 schwarze 756: if ( ! bold)
1.65 schwarze 757: printf("<b>");
1.1 schwarze 758: bold = 1;
759: italic = 0;
760: i += 2;
761: html_putchar(p[i]);
762: }
763:
1.43 schwarze 764: /*
1.1 schwarze 765: * Clean up the last character.
1.43 schwarze 766: * We can get to a newline; don't print that.
1.1 schwarze 767: */
768:
769: if (italic)
1.65 schwarze 770: printf("</i>");
1.1 schwarze 771: if (bold)
1.65 schwarze 772: printf("</b>");
1.1 schwarze 773:
1.54 schwarze 774: if (i == len - 1 && p[i] != '\n')
1.1 schwarze 775: html_putchar(p[i]);
776:
777: putchar('\n');
778: }
1.54 schwarze 779: free(p);
1.1 schwarze 780:
1.65 schwarze 781: puts("</pre>\n"
782: "</div>");
1.1 schwarze 783:
784: fclose(f);
785: }
786:
787: static void
1.68 schwarze 788: resp_format(const struct req *req, const char *file)
1.1 schwarze 789: {
1.45 schwarze 790: struct manoutput conf;
1.1 schwarze 791: struct mparse *mp;
1.46 schwarze 792: struct roff_man *man;
1.1 schwarze 793: void *vp;
1.30 schwarze 794: int fd;
795: int usepath;
1.1 schwarze 796:
797: if (-1 == (fd = open(file, O_RDONLY, 0))) {
1.65 schwarze 798: puts("<p>You specified an invalid manual file.</p>");
1.1 schwarze 799: return;
800: }
801:
1.49 schwarze 802: mchars_alloc();
1.75 schwarze 803: mp = mparse_alloc(MPARSE_SO | MPARSE_UTF8 | MPARSE_LATIN1,
804: MANDOCLEVEL_BADARG, NULL, req->q.manpath);
1.42 schwarze 805: mparse_readfd(mp, fd, file);
1.1 schwarze 806: close(fd);
807:
1.45 schwarze 808: memset(&conf, 0, sizeof(conf));
809: conf.fragment = 1;
1.30 schwarze 810: usepath = strcmp(req->q.manpath, req->p[0]);
1.61 schwarze 811: mandoc_asprintf(&conf.man, "/%s%s%%N.%%S",
812: usepath ? req->q.manpath : "", usepath ? "/" : "");
1.1 schwarze 813:
1.47 schwarze 814: mparse_result(mp, &man, NULL);
815: if (man == NULL) {
1.67 schwarze 816: warnx("fatal mandoc error: %s/%s", req->q.manpath, file);
1.12 schwarze 817: pg_error_internal();
1.1 schwarze 818: mparse_free(mp);
1.49 schwarze 819: mchars_free();
1.1 schwarze 820: return;
821: }
822:
1.49 schwarze 823: vp = html_alloc(&conf);
1.1 schwarze 824:
1.50 schwarze 825: if (man->macroset == MACROSET_MDOC) {
826: mdoc_validate(man);
1.47 schwarze 827: html_mdoc(vp, man);
1.51 schwarze 828: } else {
829: man_validate(man);
1.1 schwarze 830: html_man(vp, man);
1.51 schwarze 831: }
1.1 schwarze 832:
833: html_free(vp);
834: mparse_free(mp);
1.49 schwarze 835: mchars_free();
1.45 schwarze 836: free(conf.man);
1.1 schwarze 837: }
838:
839: static void
1.10 schwarze 840: resp_show(const struct req *req, const char *file)
841: {
1.16 schwarze 842:
843: if ('.' == file[0] && '/' == file[1])
1.11 schwarze 844: file += 2;
1.10 schwarze 845:
846: if ('c' == *file)
1.68 schwarze 847: resp_catman(req, file);
1.10 schwarze 848: else
1.68 schwarze 849: resp_format(req, file);
1.10 schwarze 850: }
851:
852: static void
1.24 schwarze 853: pg_show(struct req *req, const char *fullpath)
1.1 schwarze 854: {
1.24 schwarze 855: char *manpath;
856: const char *file;
1.1 schwarze 857:
1.24 schwarze 858: if ((file = strchr(fullpath, '/')) == NULL) {
1.12 schwarze 859: pg_error_badrequest(
1.1 schwarze 860: "You did not specify a page to show.");
861: return;
1.43 schwarze 862: }
1.24 schwarze 863: manpath = mandoc_strndup(fullpath, file - fullpath);
864: file++;
1.1 schwarze 865:
1.24 schwarze 866: if ( ! validate_manpath(req, manpath)) {
1.17 schwarze 867: pg_error_badrequest(
868: "You specified an invalid manpath.");
1.24 schwarze 869: free(manpath);
1.17 schwarze 870: return;
871: }
872:
1.1 schwarze 873: /*
874: * Begin by chdir()ing into the manpath.
875: * This way we can pick up the database files, which are
876: * relative to the manpath root.
877: */
878:
1.24 schwarze 879: if (chdir(manpath) == -1) {
1.67 schwarze 880: warn("chdir %s", manpath);
1.17 schwarze 881: pg_error_internal();
1.24 schwarze 882: free(manpath);
1.16 schwarze 883: return;
884: }
1.73 schwarze 885: free(manpath);
1.24 schwarze 886:
887: if ( ! validate_filename(file)) {
1.16 schwarze 888: pg_error_badrequest(
889: "You specified an invalid manual file.");
1.1 schwarze 890: return;
891: }
1.19 schwarze 892:
1.10 schwarze 893: resp_begin_html(200, NULL);
1.70 schwarze 894: resp_searchform(req, FOCUS_NONE);
1.24 schwarze 895: resp_show(req, file);
1.10 schwarze 896: resp_end_html();
1.1 schwarze 897: }
898:
899: static void
1.6 schwarze 900: pg_search(const struct req *req)
1.1 schwarze 901: {
902: struct mansearch search;
903: struct manpaths paths;
904: struct manpage *res;
1.36 schwarze 905: char **argv;
906: char *query, *rp, *wp;
1.1 schwarze 907: size_t ressz;
1.36 schwarze 908: int argc;
1.1 schwarze 909:
910: /*
911: * Begin by chdir()ing into the root of the manpath.
912: * This way we can pick up the database files, which are
913: * relative to the manpath root.
914: */
915:
1.67 schwarze 916: if (chdir(req->q.manpath) == -1) {
917: warn("chdir %s", req->q.manpath);
1.17 schwarze 918: pg_error_internal();
1.1 schwarze 919: return;
920: }
921:
922: search.arch = req->q.arch;
923: search.sec = req->q.sec;
1.35 schwarze 924: search.outkey = "Nd";
925: search.argmode = req->q.equal ? ARG_NAME : ARG_EXPR;
1.40 schwarze 926: search.firstmatch = 1;
1.1 schwarze 927:
928: paths.sz = 1;
929: paths.paths = mandoc_malloc(sizeof(char *));
930: paths.paths[0] = mandoc_strdup(".");
931:
932: /*
1.36 schwarze 933: * Break apart at spaces with backslash-escaping.
1.1 schwarze 934: */
935:
1.36 schwarze 936: argc = 0;
937: argv = NULL;
938: rp = query = mandoc_strdup(req->q.query);
939: for (;;) {
940: while (isspace((unsigned char)*rp))
941: rp++;
942: if (*rp == '\0')
943: break;
944: argv = mandoc_reallocarray(argv, argc + 1, sizeof(char *));
945: argv[argc++] = wp = rp;
946: for (;;) {
947: if (isspace((unsigned char)*rp)) {
948: *wp = '\0';
949: rp++;
950: break;
951: }
952: if (rp[0] == '\\' && rp[1] != '\0')
953: rp++;
954: if (wp != rp)
955: *wp = *rp;
956: if (*rp == '\0')
957: break;
958: wp++;
959: rp++;
960: }
1.1 schwarze 961: }
962:
1.36 schwarze 963: if (0 == mansearch(&search, &paths, argc, argv, &res, &ressz))
1.12 schwarze 964: pg_noresult(req, "You entered an invalid query.");
1.1 schwarze 965: else if (0 == ressz)
1.12 schwarze 966: pg_noresult(req, "No results found.");
1.1 schwarze 967: else
1.12 schwarze 968: pg_searchres(req, res, ressz);
1.1 schwarze 969:
1.36 schwarze 970: free(query);
971: mansearch_free(res, ressz);
1.1 schwarze 972: free(paths.paths[0]);
973: free(paths.paths);
974: }
975:
976: int
977: main(void)
978: {
1.6 schwarze 979: struct req req;
1.33 schwarze 980: struct itimerval itimer;
1.6 schwarze 981: const char *path;
1.23 schwarze 982: const char *querystring;
1.1 schwarze 983: int i;
1.33 schwarze 984:
985: /* Poor man's ReDoS mitigation. */
986:
1.38 schwarze 987: itimer.it_value.tv_sec = 2;
1.33 schwarze 988: itimer.it_value.tv_usec = 0;
1.38 schwarze 989: itimer.it_interval.tv_sec = 2;
1.33 schwarze 990: itimer.it_interval.tv_usec = 0;
991: if (setitimer(ITIMER_VIRTUAL, &itimer, NULL) == -1) {
1.67 schwarze 992: warn("setitimer");
1.20 schwarze 993: pg_error_internal();
1.48 schwarze 994: return EXIT_FAILURE;
1.20 schwarze 995: }
996:
1.1 schwarze 997: /*
1.7 schwarze 998: * First we change directory into the MAN_DIR so that
1.1 schwarze 999: * subsequent scanning for manpath directories is rooted
1000: * relative to the same position.
1001: */
1002:
1.67 schwarze 1003: if (chdir(MAN_DIR) == -1) {
1004: warn("MAN_DIR: %s", MAN_DIR);
1.12 schwarze 1005: pg_error_internal();
1.48 schwarze 1006: return EXIT_FAILURE;
1.43 schwarze 1007: }
1.1 schwarze 1008:
1009: memset(&req, 0, sizeof(struct req));
1.57 schwarze 1010: req.q.equal = 1;
1.68 schwarze 1011: parse_manpath_conf(&req);
1.1 schwarze 1012:
1.56 schwarze 1013: /* Parse the path info and the query string. */
1.1 schwarze 1014:
1.56 schwarze 1015: if ((path = getenv("PATH_INFO")) == NULL)
1016: path = "";
1017: else if (*path == '/')
1018: path++;
1019:
1.63 schwarze 1020: if (*path != '\0') {
1.68 schwarze 1021: parse_path_info(&req, path);
1.72 schwarze 1022: if (req.q.manpath == NULL || access(path, F_OK) == -1)
1.63 schwarze 1023: path = "";
1.56 schwarze 1024: } else if ((querystring = getenv("QUERY_STRING")) != NULL)
1.68 schwarze 1025: parse_query_string(&req, querystring);
1.17 schwarze 1026:
1.56 schwarze 1027: /* Validate parsed data and add defaults. */
1028:
1.41 schwarze 1029: if (req.q.manpath == NULL)
1030: req.q.manpath = mandoc_strdup(req.p[0]);
1031: else if ( ! validate_manpath(&req, req.q.manpath)) {
1.17 schwarze 1032: pg_error_badrequest(
1033: "You specified an invalid manpath.");
1.48 schwarze 1034: return EXIT_FAILURE;
1.17 schwarze 1035: }
1.1 schwarze 1036:
1.20 schwarze 1037: if ( ! (NULL == req.q.arch || validate_urifrag(req.q.arch))) {
1038: pg_error_badrequest(
1039: "You specified an invalid architecture.");
1.48 schwarze 1040: return EXIT_FAILURE;
1.20 schwarze 1041: }
1042:
1.6 schwarze 1043: /* Dispatch to the three different pages. */
1.1 schwarze 1044:
1.6 schwarze 1045: if ('\0' != *path)
1046: pg_show(&req, path);
1.25 schwarze 1047: else if (NULL != req.q.query)
1.6 schwarze 1048: pg_search(&req);
1049: else
1.12 schwarze 1050: pg_index(&req);
1.1 schwarze 1051:
1.23 schwarze 1052: free(req.q.manpath);
1053: free(req.q.arch);
1054: free(req.q.sec);
1.25 schwarze 1055: free(req.q.query);
1.1 schwarze 1056: for (i = 0; i < (int)req.psz; i++)
1057: free(req.p[i]);
1058: free(req.p);
1.48 schwarze 1059: return EXIT_SUCCESS;
1.56 schwarze 1060: }
1061:
1062: /*
1063: * If PATH_INFO is not a file name, translate it to a query.
1064: */
1065: static void
1.68 schwarze 1066: parse_path_info(struct req *req, const char *path)
1.56 schwarze 1067: {
1.74 schwarze 1068: char *dir[4];
1069: int i;
1.56 schwarze 1070:
1.60 schwarze 1071: req->isquery = 0;
1.56 schwarze 1072: req->q.equal = 1;
1073: req->q.manpath = mandoc_strdup(path);
1.74 schwarze 1074: req->q.arch = NULL;
1.56 schwarze 1075:
1076: /* Mandatory manual page name. */
1077: if ((req->q.query = strrchr(req->q.manpath, '/')) == NULL) {
1078: req->q.query = req->q.manpath;
1079: req->q.manpath = NULL;
1080: } else
1081: *req->q.query++ = '\0';
1082:
1083: /* Optional trailing section. */
1084: if ((req->q.sec = strrchr(req->q.query, '.')) != NULL) {
1085: if(isdigit((unsigned char)req->q.sec[1])) {
1086: *req->q.sec++ = '\0';
1087: req->q.sec = mandoc_strdup(req->q.sec);
1088: } else
1089: req->q.sec = NULL;
1090: }
1091:
1092: /* Handle the case of name[.section] only. */
1.74 schwarze 1093: if (req->q.manpath == NULL)
1.56 schwarze 1094: return;
1.74 schwarze 1095: req->q.query = mandoc_strdup(req->q.query);
1096:
1097: /* Split directory components. */
1098: dir[i = 0] = req->q.manpath;
1099: while ((dir[i + 1] = strchr(dir[i], '/')) != NULL) {
1100: if (++i == 3) {
1101: pg_error_badrequest(
1102: "You specified too many directory components.");
1103: exit(EXIT_FAILURE);
1104: }
1105: *dir[i]++ = '\0';
1.56 schwarze 1106: }
1107:
1.74 schwarze 1108: /* Optional manpath. */
1109: if ((i = validate_manpath(req, req->q.manpath)) == 0)
1110: req->q.manpath = NULL;
1111: else if (dir[1] == NULL)
1112: return;
1.56 schwarze 1113:
1.74 schwarze 1114: /* Optional section. */
1115: if (strncmp(dir[i], "man", 3) == 0) {
1.66 schwarze 1116: free(req->q.sec);
1.74 schwarze 1117: req->q.sec = mandoc_strdup(dir[i++] + 3);
1.56 schwarze 1118: }
1.74 schwarze 1119: if (dir[i] == NULL) {
1120: if (req->q.manpath == NULL)
1121: free(dir[0]);
1122: return;
1123: }
1124: if (dir[i + 1] != NULL) {
1125: pg_error_badrequest(
1126: "You specified an invalid directory component.");
1127: exit(EXIT_FAILURE);
1128: }
1129:
1130: /* Optional architecture. */
1131: if (i) {
1132: req->q.arch = mandoc_strdup(dir[i]);
1133: if (req->q.manpath == NULL)
1134: free(dir[0]);
1135: } else
1136: req->q.arch = dir[0];
1.1 schwarze 1137: }
1138:
1139: /*
1140: * Scan for indexable paths.
1141: */
1142: static void
1.68 schwarze 1143: parse_manpath_conf(struct req *req)
1.1 schwarze 1144: {
1145: FILE *fp;
1146: char *dp;
1147: size_t dpsz;
1.54 schwarze 1148: ssize_t len;
1.1 schwarze 1149:
1.67 schwarze 1150: if ((fp = fopen("manpath.conf", "r")) == NULL) {
1151: warn("%s/manpath.conf", MAN_DIR);
1.14 schwarze 1152: pg_error_internal();
1153: exit(EXIT_FAILURE);
1154: }
1.1 schwarze 1155:
1.54 schwarze 1156: dp = NULL;
1157: dpsz = 0;
1158:
1159: while ((len = getline(&dp, &dpsz, fp)) != -1) {
1160: if (dp[len - 1] == '\n')
1161: dp[--len] = '\0';
1.1 schwarze 1162: req->p = mandoc_realloc(req->p,
1163: (req->psz + 1) * sizeof(char *));
1.20 schwarze 1164: if ( ! validate_urifrag(dp)) {
1.67 schwarze 1165: warnx("%s/manpath.conf contains "
1166: "unsafe path \"%s\"", MAN_DIR, dp);
1.20 schwarze 1167: pg_error_internal();
1168: exit(EXIT_FAILURE);
1169: }
1.67 schwarze 1170: if (strchr(dp, '/') != NULL) {
1171: warnx("%s/manpath.conf contains "
1172: "path with slash \"%s\"", MAN_DIR, dp);
1.20 schwarze 1173: pg_error_internal();
1174: exit(EXIT_FAILURE);
1175: }
1176: req->p[req->psz++] = dp;
1.54 schwarze 1177: dp = NULL;
1178: dpsz = 0;
1.14 schwarze 1179: }
1.54 schwarze 1180: free(dp);
1.14 schwarze 1181:
1.67 schwarze 1182: if (req->p == NULL) {
1183: warnx("%s/manpath.conf is empty", MAN_DIR);
1.14 schwarze 1184: pg_error_internal();
1185: exit(EXIT_FAILURE);
1.1 schwarze 1186: }
1187: }
![[BACK]](/icons/back.gif){kind=icon}
Return to cgi.c CVS log ![[TXT]](/icons/text.gif){kind=icon}
![[DIR]](/icons/dir.gif){kind=icon}
Up to [local] / src / usr.bin / mandoc