Annotation of src/usr.bin/mandoc/cgi.c, Revision 1.72
1.72 ! schwarze 1: /* $OpenBSD: cgi.c,v 1.71 2016/05/28 13:40:44 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[] = {
114: "amd64", "alpha", "armish", "armv7",
1.64 schwarze 115: "hppa", "hppa64", "i386", "landisk",
116: "loongson", "luna88k", "macppc", "mips64",
117: "octeon", "sgi", "socppc", "sparc",
118: "sparc64", "zaurus",
1.8 schwarze 119: "amiga", "arc", "arm32", "atari",
1.64 schwarze 120: "aviion", "beagle", "cats", "hp300",
121: "ia64", "mac68k", "mvme68k", "mvme88k",
122: "mvmeppc", "palm", "pc532", "pegasos",
123: "pmax", "powerpc", "solbourne", "sun3",
124: "vax", "wgrisc", "x68k"
1.8 schwarze 125: };
126: static const int arch_MAX = sizeof(arch_names) / sizeof(char *);
127:
1.1 schwarze 128: /*
129: * Print a character, escaping HTML along the way.
130: * This will pass non-ASCII straight to output: be warned!
131: */
132: static void
133: html_putchar(char c)
134: {
135:
136: switch (c) {
137: case ('"'):
138: printf(""e;");
139: break;
140: case ('&'):
141: printf("&");
142: break;
143: case ('>'):
144: printf(">");
145: break;
146: case ('<'):
147: printf("<");
148: break;
149: default:
150: putchar((unsigned char)c);
151: break;
152: }
153: }
154:
155: /*
156: * Call through to html_putchar().
157: * Accepts NULL strings.
158: */
159: static void
160: html_print(const char *p)
161: {
1.43 schwarze 162:
1.1 schwarze 163: if (NULL == p)
164: return;
165: while ('\0' != *p)
166: html_putchar(*p++);
167: }
168:
169: /*
1.23 schwarze 170: * Transfer the responsibility for the allocated string *val
171: * to the query structure.
1.1 schwarze 172: */
173: static void
1.23 schwarze 174: set_query_attr(char **attr, char **val)
1.1 schwarze 175: {
176:
1.23 schwarze 177: free(*attr);
178: if (**val == '\0') {
179: *attr = NULL;
180: free(*val);
181: } else
182: *attr = *val;
183: *val = NULL;
184: }
185:
186: /*
187: * Parse the QUERY_STRING for key-value pairs
188: * and store the values into the query structure.
189: */
190: static void
1.68 schwarze 191: parse_query_string(struct req *req, const char *qs)
1.23 schwarze 192: {
193: char *key, *val;
194: size_t keysz, valsz;
195:
1.60 schwarze 196: req->isquery = 1;
1.23 schwarze 197: req->q.manpath = NULL;
198: req->q.arch = NULL;
199: req->q.sec = NULL;
1.25 schwarze 200: req->q.query = NULL;
1.23 schwarze 201: req->q.equal = 1;
202:
203: key = val = NULL;
204: while (*qs != '\0') {
1.1 schwarze 205:
1.23 schwarze 206: /* Parse one key. */
207:
208: keysz = strcspn(qs, "=;&");
209: key = mandoc_strndup(qs, keysz);
210: qs += keysz;
211: if (*qs != '=')
212: goto next;
213:
214: /* Parse one value. */
215:
216: valsz = strcspn(++qs, ";&");
217: val = mandoc_strndup(qs, valsz);
218: qs += valsz;
219:
220: /* Decode and catch encoding errors. */
1.1 schwarze 221:
1.23 schwarze 222: if ( ! (http_decode(key) && http_decode(val)))
223: goto next;
1.1 schwarze 224:
1.23 schwarze 225: /* Handle key-value pairs. */
1.1 schwarze 226:
1.23 schwarze 227: if ( ! strcmp(key, "query"))
1.25 schwarze 228: set_query_attr(&req->q.query, &val);
1.1 schwarze 229:
1.23 schwarze 230: else if ( ! strcmp(key, "apropos"))
231: req->q.equal = !strcmp(val, "0");
232:
233: else if ( ! strcmp(key, "manpath")) {
1.13 schwarze 234: #ifdef COMPAT_OLDURI
1.23 schwarze 235: if ( ! strncmp(val, "OpenBSD ", 8)) {
1.13 schwarze 236: val[7] = '-';
237: if ('C' == val[8])
238: val[8] = 'c';
239: }
240: #endif
1.23 schwarze 241: set_query_attr(&req->q.manpath, &val);
242: }
243:
244: else if ( ! (strcmp(key, "sec")
1.13 schwarze 245: #ifdef COMPAT_OLDURI
1.23 schwarze 246: && strcmp(key, "sektion")
1.13 schwarze 247: #endif
1.23 schwarze 248: )) {
249: if ( ! strcmp(val, "0"))
250: *val = '\0';
251: set_query_attr(&req->q.sec, &val);
1.5 schwarze 252: }
1.23 schwarze 253:
254: else if ( ! strcmp(key, "arch")) {
255: if ( ! strcmp(val, "default"))
256: *val = '\0';
257: set_query_attr(&req->q.arch, &val);
258: }
259:
260: /*
261: * The key must be freed in any case.
262: * The val may have been handed over to the query
263: * structure, in which case it is now NULL.
264: */
265: next:
266: free(key);
267: key = NULL;
268: free(val);
269: val = NULL;
270:
271: if (*qs != '\0')
272: qs++;
1.1 schwarze 273: }
274: }
275:
276: /*
277: * HTTP-decode a string. The standard explanation is that this turns
278: * "%4e+foo" into "n foo" in the regular way. This is done in-place
279: * over the allocated string.
280: */
281: static int
282: http_decode(char *p)
283: {
284: char hex[3];
1.3 tedu 285: char *q;
1.1 schwarze 286: int c;
287:
288: hex[2] = '\0';
289:
1.3 tedu 290: q = p;
291: for ( ; '\0' != *p; p++, q++) {
1.1 schwarze 292: if ('%' == *p) {
293: if ('\0' == (hex[0] = *(p + 1)))
1.48 schwarze 294: return 0;
1.1 schwarze 295: if ('\0' == (hex[1] = *(p + 2)))
1.48 schwarze 296: return 0;
1.1 schwarze 297: if (1 != sscanf(hex, "%x", &c))
1.48 schwarze 298: return 0;
1.1 schwarze 299: if ('\0' == c)
1.48 schwarze 300: return 0;
1.1 schwarze 301:
1.3 tedu 302: *q = (char)c;
303: p += 2;
1.1 schwarze 304: } else
1.3 tedu 305: *q = '+' == *p ? ' ' : *p;
1.1 schwarze 306: }
307:
1.3 tedu 308: *q = '\0';
1.48 schwarze 309: return 1;
1.1 schwarze 310: }
311:
312: static void
313: resp_begin_http(int code, const char *msg)
314: {
315:
316: if (200 != code)
1.2 tedu 317: printf("Status: %d %s\r\n", code, msg);
1.1 schwarze 318:
1.2 tedu 319: printf("Content-Type: text/html; charset=utf-8\r\n"
320: "Cache-Control: no-cache\r\n"
321: "Pragma: no-cache\r\n"
322: "\r\n");
1.1 schwarze 323:
324: fflush(stdout);
325: }
326:
327: static void
1.53 schwarze 328: resp_copy(const char *filename)
329: {
330: char buf[4096];
331: ssize_t sz;
332: int fd;
333:
334: if ((fd = open(filename, O_RDONLY)) != -1) {
335: fflush(stdout);
336: while ((sz = read(fd, buf, sizeof(buf))) > 0)
337: write(STDOUT_FILENO, buf, sz);
338: }
339: }
340:
341: static void
1.1 schwarze 342: resp_begin_html(int code, const char *msg)
343: {
344:
345: resp_begin_http(code, msg);
346:
1.37 schwarze 347: printf("<!DOCTYPE html>\n"
1.65 schwarze 348: "<html>\n"
349: "<head>\n"
350: "<meta charset=\"UTF-8\"/>\n"
351: "<link rel=\"stylesheet\" href=\"%s/mandoc.css\""
352: " type=\"text/css\" media=\"all\">\n"
353: "<title>%s</title>\n"
354: "</head>\n"
355: "<body>\n"
1.1 schwarze 356: "<!-- Begin page content. //-->\n",
1.52 schwarze 357: CSS_DIR, CUSTOMIZE_TITLE);
1.53 schwarze 358:
359: resp_copy(MAN_DIR "/header.html");
1.1 schwarze 360: }
361:
362: static void
363: resp_end_html(void)
364: {
365:
1.53 schwarze 366: resp_copy(MAN_DIR "/footer.html");
367:
1.65 schwarze 368: puts("</body>\n"
369: "</html>");
1.1 schwarze 370: }
371:
372: static void
1.70 schwarze 373: resp_searchform(const struct req *req, enum focus focus)
1.1 schwarze 374: {
375: int i;
376:
377: puts("<!-- Begin search form. //-->");
1.65 schwarze 378: printf("<div id=\"mancgi\">\n"
379: "<form action=\"/%s\" method=\"get\">\n"
380: "<fieldset>\n"
381: "<legend>Manual Page Search Parameters</legend>\n",
1.1 schwarze 382: scriptname);
1.8 schwarze 383:
384: /* Write query input box. */
385:
1.71 schwarze 386: printf("<input type=\"text\" name=\"query\" value=\"");
1.70 schwarze 387: if (req->q.query != NULL)
1.25 schwarze 388: html_print(req->q.query);
1.70 schwarze 389: printf( "\" size=\"40\"");
390: if (focus == FOCUS_QUERY)
391: printf(" autofocus");
392: puts(">");
1.8 schwarze 393:
1.71 schwarze 394: /* Write submission buttons. */
1.8 schwarze 395:
1.71 schwarze 396: printf( "<button type=\"submit\" name=\"apropos\" value=\"0\">"
397: "man</button>\n"
398: "<button type=\"submit\" name=\"apropos\" value=\"1\">"
399: "apropos</button>\n<br/>\n");
1.8 schwarze 400:
401: /* Write section selector. */
402:
1.71 schwarze 403: puts("<select name=\"sec\">");
1.8 schwarze 404: for (i = 0; i < sec_MAX; i++) {
1.65 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: }
1.65 schwarze 411: puts("</select>");
1.8 schwarze 412:
413: /* Write architecture selector. */
414:
1.65 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++) {
1.65 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: }
1.65 schwarze 427: puts("</select>");
1.8 schwarze 428:
429: /* Write manpath selector. */
430:
1.1 schwarze 431: if (req->psz > 1) {
1.65 schwarze 432: puts("<select name=\"manpath\">");
1.1 schwarze 433: for (i = 0; i < (int)req->psz; i++) {
1.65 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: }
1.65 schwarze 443: puts("</select>");
1.1 schwarze 444: }
1.8 schwarze 445:
1.71 schwarze 446: puts("</fieldset>\n"
1.65 schwarze 447: "</form>\n"
448: "</div>");
1.1 schwarze 449: puts("<!-- End search form. //-->");
450: }
451:
1.16 schwarze 452: static int
1.20 schwarze 453: validate_urifrag(const char *frag)
454: {
455:
456: while ('\0' != *frag) {
457: if ( ! (isalnum((unsigned char)*frag) ||
458: '-' == *frag || '.' == *frag ||
459: '/' == *frag || '_' == *frag))
1.48 schwarze 460: return 0;
1.20 schwarze 461: frag++;
462: }
1.48 schwarze 463: return 1;
1.20 schwarze 464: }
465:
466: static int
1.17 schwarze 467: validate_manpath(const struct req *req, const char* manpath)
468: {
469: size_t i;
470:
471: if ( ! strcmp(manpath, "mandoc"))
1.48 schwarze 472: return 1;
1.17 schwarze 473:
474: for (i = 0; i < req->psz; i++)
475: if ( ! strcmp(manpath, req->p[i]))
1.48 schwarze 476: return 1;
1.17 schwarze 477:
1.48 schwarze 478: return 0;
1.17 schwarze 479: }
480:
481: static int
1.16 schwarze 482: validate_filename(const char *file)
483: {
484:
485: if ('.' == file[0] && '/' == file[1])
486: file += 2;
487:
1.48 schwarze 488: return ! (strstr(file, "../") || strstr(file, "/..") ||
489: (strncmp(file, "man", 3) && strncmp(file, "cat", 3)));
1.16 schwarze 490: }
491:
1.1 schwarze 492: static void
1.12 schwarze 493: pg_index(const struct req *req)
1.1 schwarze 494: {
495:
496: resp_begin_html(200, NULL);
1.70 schwarze 497: resp_searchform(req, FOCUS_QUERY);
1.65 schwarze 498: printf("<p>\n"
1.26 schwarze 499: "This web interface is documented in the\n"
1.65 schwarze 500: "<a href=\"/%s%smandoc/man8/man.cgi.8\">man.cgi</a>\n"
1.26 schwarze 501: "manual, and the\n"
1.65 schwarze 502: "<a href=\"/%s%smandoc/man1/apropos.1\">apropos</a>\n"
1.9 schwarze 503: "manual explains the query syntax.\n"
1.65 schwarze 504: "</p>\n",
1.58 schwarze 505: scriptname, *scriptname == '\0' ? "" : "/",
506: scriptname, *scriptname == '\0' ? "" : "/");
1.1 schwarze 507: resp_end_html();
508: }
509:
510: static void
1.12 schwarze 511: pg_noresult(const struct req *req, const char *msg)
1.1 schwarze 512: {
513: resp_begin_html(200, NULL);
1.70 schwarze 514: resp_searchform(req, FOCUS_QUERY);
1.65 schwarze 515: puts("<p>");
1.1 schwarze 516: puts(msg);
1.65 schwarze 517: puts("</p>");
1.1 schwarze 518: resp_end_html();
519: }
520:
521: static void
1.12 schwarze 522: pg_error_badrequest(const char *msg)
1.1 schwarze 523: {
524:
525: resp_begin_html(400, "Bad Request");
1.65 schwarze 526: puts("<h1>Bad Request</h1>\n"
527: "<p>\n");
1.1 schwarze 528: puts(msg);
529: printf("Try again from the\n"
1.65 schwarze 530: "<a href=\"/%s\">main page</a>.\n"
531: "</p>", scriptname);
1.1 schwarze 532: resp_end_html();
533: }
534:
535: static void
1.12 schwarze 536: pg_error_internal(void)
1.1 schwarze 537: {
538: resp_begin_html(500, "Internal Server Error");
1.65 schwarze 539: puts("<p>Internal Server Error</p>");
1.1 schwarze 540: resp_end_html();
541: }
542:
543: static void
1.12 schwarze 544: pg_searchres(const struct req *req, struct manpage *r, size_t sz)
1.1 schwarze 545: {
1.21 schwarze 546: char *arch, *archend;
1.59 schwarze 547: const char *sec;
548: size_t i, iuse;
1.21 schwarze 549: int archprio, archpriouse;
1.10 schwarze 550: int prio, priouse;
1.1 schwarze 551:
1.16 schwarze 552: for (i = 0; i < sz; i++) {
553: if (validate_filename(r[i].file))
554: continue;
1.67 schwarze 555: warnx("invalid filename %s in %s database",
1.16 schwarze 556: r[i].file, req->q.manpath);
557: pg_error_internal();
558: return;
559: }
560:
1.60 schwarze 561: if (req->isquery && sz == 1) {
1.1 schwarze 562: /*
563: * If we have just one result, then jump there now
564: * without any delay.
565: */
1.2 tedu 566: printf("Status: 303 See Other\r\n");
1.58 schwarze 567: printf("Location: http://%s/%s%s%s/%s",
568: HTTP_HOST, scriptname,
569: *scriptname == '\0' ? "" : "/",
570: req->q.manpath, r[0].file);
1.2 tedu 571: printf("\r\n"
572: "Content-Type: text/html; charset=utf-8\r\n"
573: "\r\n");
1.1 schwarze 574: return;
575: }
576:
577: resp_begin_html(200, NULL);
1.70 schwarze 578: resp_searchform(req,
579: req->q.equal || sz == 1 ? FOCUS_NONE : FOCUS_QUERY);
1.1 schwarze 580:
1.62 schwarze 581: if (sz > 1) {
1.65 schwarze 582: puts("<div class=\"results\">");
583: puts("<table>");
1.62 schwarze 584:
585: for (i = 0; i < sz; i++) {
1.65 schwarze 586: printf("<tr>\n"
587: "<td class=\"title\">\n"
588: "<a href=\"/%s%s%s/%s",
1.62 schwarze 589: scriptname, *scriptname == '\0' ? "" : "/",
590: req->q.manpath, r[i].file);
591: printf("\">");
592: html_print(r[i].names);
1.65 schwarze 593: printf("</a>\n"
594: "</td>\n"
595: "<td class=\"desc\">");
1.62 schwarze 596: html_print(r[i].output);
1.65 schwarze 597: puts("</td>\n"
598: "</tr>");
1.62 schwarze 599: }
600:
1.65 schwarze 601: puts("</table>\n"
602: "</div>");
1.1 schwarze 603: }
604:
1.10 schwarze 605: /*
606: * In man(1) mode, show one of the pages
607: * even if more than one is found.
608: */
609:
1.62 schwarze 610: if (req->q.equal || sz == 1) {
1.65 schwarze 611: puts("<hr>");
1.10 schwarze 612: iuse = 0;
1.59 schwarze 613: priouse = 20;
1.21 schwarze 614: archpriouse = 3;
1.10 schwarze 615: for (i = 0; i < sz; i++) {
1.59 schwarze 616: sec = r[i].file;
617: sec += strcspn(sec, "123456789");
618: if (sec[0] == '\0')
1.10 schwarze 619: continue;
1.59 schwarze 620: prio = sec_prios[sec[0] - '1'];
621: if (sec[1] != '/')
622: prio += 10;
623: if (req->q.arch == NULL) {
1.21 schwarze 624: archprio =
1.59 schwarze 625: ((arch = strchr(sec + 1, '/'))
626: == NULL) ? 3 :
627: ((archend = strchr(arch + 1, '/'))
628: == NULL) ? 0 :
1.21 schwarze 629: strncmp(arch, "amd64/",
630: archend - arch) ? 2 : 1;
631: if (archprio < archpriouse) {
632: archpriouse = archprio;
633: priouse = prio;
634: iuse = i;
635: continue;
636: }
637: if (archprio > archpriouse)
638: continue;
639: }
1.10 schwarze 640: if (prio >= priouse)
641: continue;
642: priouse = prio;
643: iuse = i;
644: }
645: resp_show(req, r[iuse].file);
646: }
647:
1.1 schwarze 648: resp_end_html();
649: }
650:
651: static void
1.68 schwarze 652: resp_catman(const struct req *req, const char *file)
1.1 schwarze 653: {
654: FILE *f;
1.54 schwarze 655: char *p;
656: size_t sz;
657: ssize_t len;
1.1 schwarze 658: int i;
659: int italic, bold;
660:
1.54 schwarze 661: if ((f = fopen(file, "r")) == NULL) {
1.65 schwarze 662: puts("<p>You specified an invalid manual file.</p>");
1.1 schwarze 663: return;
664: }
665:
1.65 schwarze 666: puts("<div class=\"catman\">\n"
667: "<pre>");
1.1 schwarze 668:
1.54 schwarze 669: p = NULL;
670: sz = 0;
671:
672: while ((len = getline(&p, &sz, f)) != -1) {
1.1 schwarze 673: bold = italic = 0;
1.54 schwarze 674: for (i = 0; i < len - 1; i++) {
1.43 schwarze 675: /*
1.1 schwarze 676: * This means that the catpage is out of state.
677: * Ignore it and keep going (although the
678: * catpage is bogus).
679: */
680:
681: if ('\b' == p[i] || '\n' == p[i])
682: continue;
683:
684: /*
685: * Print a regular character.
686: * Close out any bold/italic scopes.
687: * If we're in back-space mode, make sure we'll
688: * have something to enter when we backspace.
689: */
690:
691: if ('\b' != p[i + 1]) {
692: if (italic)
1.65 schwarze 693: printf("</i>");
1.1 schwarze 694: if (bold)
1.65 schwarze 695: printf("</b>");
1.1 schwarze 696: italic = bold = 0;
697: html_putchar(p[i]);
698: continue;
1.54 schwarze 699: } else if (i + 2 >= len)
1.1 schwarze 700: continue;
701:
702: /* Italic mode. */
703:
704: if ('_' == p[i]) {
705: if (bold)
1.65 schwarze 706: printf("</b>");
1.1 schwarze 707: if ( ! italic)
1.65 schwarze 708: printf("<i>");
1.1 schwarze 709: bold = 0;
710: italic = 1;
711: i += 2;
712: html_putchar(p[i]);
713: continue;
714: }
715:
1.43 schwarze 716: /*
1.1 schwarze 717: * Handle funny behaviour troff-isms.
718: * These grok'd from the original man2html.c.
719: */
720:
721: if (('+' == p[i] && 'o' == p[i + 2]) ||
722: ('o' == 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: ('|' == p[i] && '*' == p[i + 2])) {
729: if (italic)
1.65 schwarze 730: printf("</i>");
1.1 schwarze 731: if (bold)
1.65 schwarze 732: printf("</b>");
1.1 schwarze 733: italic = bold = 0;
734: putchar('*');
735: i += 2;
736: continue;
737: } else if (('|' == p[i] && '-' == p[i + 2]) ||
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: ('|' == p[i] && '+' == p[i + 1])) {
743: if (italic)
1.65 schwarze 744: printf("</i>");
1.1 schwarze 745: if (bold)
1.65 schwarze 746: printf("</b>");
1.1 schwarze 747: italic = bold = 0;
748: putchar('+');
749: i += 2;
750: continue;
751: }
752:
753: /* Bold mode. */
1.43 schwarze 754:
1.1 schwarze 755: if (italic)
1.65 schwarze 756: printf("</i>");
1.1 schwarze 757: if ( ! bold)
1.65 schwarze 758: printf("<b>");
1.1 schwarze 759: bold = 1;
760: italic = 0;
761: i += 2;
762: html_putchar(p[i]);
763: }
764:
1.43 schwarze 765: /*
1.1 schwarze 766: * Clean up the last character.
1.43 schwarze 767: * We can get to a newline; don't print that.
1.1 schwarze 768: */
769:
770: if (italic)
1.65 schwarze 771: printf("</i>");
1.1 schwarze 772: if (bold)
1.65 schwarze 773: printf("</b>");
1.1 schwarze 774:
1.54 schwarze 775: if (i == len - 1 && p[i] != '\n')
1.1 schwarze 776: html_putchar(p[i]);
777:
778: putchar('\n');
779: }
1.54 schwarze 780: free(p);
1.1 schwarze 781:
1.65 schwarze 782: puts("</pre>\n"
783: "</div>");
1.1 schwarze 784:
785: fclose(f);
786: }
787:
788: static void
1.68 schwarze 789: resp_format(const struct req *req, const char *file)
1.1 schwarze 790: {
1.45 schwarze 791: struct manoutput conf;
1.1 schwarze 792: struct mparse *mp;
1.46 schwarze 793: struct roff_man *man;
1.1 schwarze 794: void *vp;
1.30 schwarze 795: int fd;
796: int usepath;
1.1 schwarze 797:
798: if (-1 == (fd = open(file, O_RDONLY, 0))) {
1.65 schwarze 799: puts("<p>You specified an invalid manual file.</p>");
1.1 schwarze 800: return;
801: }
802:
1.49 schwarze 803: mchars_alloc();
804: mp = mparse_alloc(MPARSE_SO, 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: }
885:
1.24 schwarze 886: if (strcmp(manpath, "mandoc")) {
887: free(req->q.manpath);
888: req->q.manpath = manpath;
889: } else
890: free(manpath);
891:
892: if ( ! validate_filename(file)) {
1.16 schwarze 893: pg_error_badrequest(
894: "You specified an invalid manual file.");
1.1 schwarze 895: return;
896: }
1.19 schwarze 897:
1.10 schwarze 898: resp_begin_html(200, NULL);
1.70 schwarze 899: resp_searchform(req, FOCUS_NONE);
1.24 schwarze 900: resp_show(req, file);
1.10 schwarze 901: resp_end_html();
1.1 schwarze 902: }
903:
904: static void
1.6 schwarze 905: pg_search(const struct req *req)
1.1 schwarze 906: {
907: struct mansearch search;
908: struct manpaths paths;
909: struct manpage *res;
1.36 schwarze 910: char **argv;
911: char *query, *rp, *wp;
1.1 schwarze 912: size_t ressz;
1.36 schwarze 913: int argc;
1.1 schwarze 914:
915: /*
916: * Begin by chdir()ing into the root of the manpath.
917: * This way we can pick up the database files, which are
918: * relative to the manpath root.
919: */
920:
1.67 schwarze 921: if (chdir(req->q.manpath) == -1) {
922: warn("chdir %s", req->q.manpath);
1.17 schwarze 923: pg_error_internal();
1.1 schwarze 924: return;
925: }
926:
927: search.arch = req->q.arch;
928: search.sec = req->q.sec;
1.35 schwarze 929: search.outkey = "Nd";
930: search.argmode = req->q.equal ? ARG_NAME : ARG_EXPR;
1.40 schwarze 931: search.firstmatch = 1;
1.1 schwarze 932:
933: paths.sz = 1;
934: paths.paths = mandoc_malloc(sizeof(char *));
935: paths.paths[0] = mandoc_strdup(".");
936:
937: /*
1.36 schwarze 938: * Break apart at spaces with backslash-escaping.
1.1 schwarze 939: */
940:
1.36 schwarze 941: argc = 0;
942: argv = NULL;
943: rp = query = mandoc_strdup(req->q.query);
944: for (;;) {
945: while (isspace((unsigned char)*rp))
946: rp++;
947: if (*rp == '\0')
948: break;
949: argv = mandoc_reallocarray(argv, argc + 1, sizeof(char *));
950: argv[argc++] = wp = rp;
951: for (;;) {
952: if (isspace((unsigned char)*rp)) {
953: *wp = '\0';
954: rp++;
955: break;
956: }
957: if (rp[0] == '\\' && rp[1] != '\0')
958: rp++;
959: if (wp != rp)
960: *wp = *rp;
961: if (*rp == '\0')
962: break;
963: wp++;
964: rp++;
965: }
1.1 schwarze 966: }
967:
1.36 schwarze 968: if (0 == mansearch(&search, &paths, argc, argv, &res, &ressz))
1.12 schwarze 969: pg_noresult(req, "You entered an invalid query.");
1.1 schwarze 970: else if (0 == ressz)
1.12 schwarze 971: pg_noresult(req, "No results found.");
1.1 schwarze 972: else
1.12 schwarze 973: pg_searchres(req, res, ressz);
1.1 schwarze 974:
1.36 schwarze 975: free(query);
976: mansearch_free(res, ressz);
1.1 schwarze 977: free(paths.paths[0]);
978: free(paths.paths);
979: }
980:
981: int
982: main(void)
983: {
1.6 schwarze 984: struct req req;
1.33 schwarze 985: struct itimerval itimer;
1.6 schwarze 986: const char *path;
1.23 schwarze 987: const char *querystring;
1.1 schwarze 988: int i;
1.33 schwarze 989:
990: /* Poor man's ReDoS mitigation. */
991:
1.38 schwarze 992: itimer.it_value.tv_sec = 2;
1.33 schwarze 993: itimer.it_value.tv_usec = 0;
1.38 schwarze 994: itimer.it_interval.tv_sec = 2;
1.33 schwarze 995: itimer.it_interval.tv_usec = 0;
996: if (setitimer(ITIMER_VIRTUAL, &itimer, NULL) == -1) {
1.67 schwarze 997: warn("setitimer");
1.20 schwarze 998: pg_error_internal();
1.48 schwarze 999: return EXIT_FAILURE;
1.20 schwarze 1000: }
1001:
1.1 schwarze 1002: /*
1.7 schwarze 1003: * First we change directory into the MAN_DIR so that
1.1 schwarze 1004: * subsequent scanning for manpath directories is rooted
1005: * relative to the same position.
1006: */
1007:
1.67 schwarze 1008: if (chdir(MAN_DIR) == -1) {
1009: warn("MAN_DIR: %s", MAN_DIR);
1.12 schwarze 1010: pg_error_internal();
1.48 schwarze 1011: return EXIT_FAILURE;
1.43 schwarze 1012: }
1.1 schwarze 1013:
1014: memset(&req, 0, sizeof(struct req));
1.57 schwarze 1015: req.q.equal = 1;
1.68 schwarze 1016: parse_manpath_conf(&req);
1.1 schwarze 1017:
1.56 schwarze 1018: /* Parse the path info and the query string. */
1.1 schwarze 1019:
1.56 schwarze 1020: if ((path = getenv("PATH_INFO")) == NULL)
1021: path = "";
1022: else if (*path == '/')
1023: path++;
1024:
1.63 schwarze 1025: if (*path != '\0') {
1.68 schwarze 1026: parse_path_info(&req, path);
1.72 ! schwarze 1027: if (req.q.manpath == NULL || access(path, F_OK) == -1)
1.63 schwarze 1028: path = "";
1.56 schwarze 1029: } else if ((querystring = getenv("QUERY_STRING")) != NULL)
1.68 schwarze 1030: parse_query_string(&req, querystring);
1.17 schwarze 1031:
1.56 schwarze 1032: /* Validate parsed data and add defaults. */
1033:
1.41 schwarze 1034: if (req.q.manpath == NULL)
1035: req.q.manpath = mandoc_strdup(req.p[0]);
1036: else if ( ! validate_manpath(&req, req.q.manpath)) {
1.17 schwarze 1037: pg_error_badrequest(
1038: "You specified an invalid manpath.");
1.48 schwarze 1039: return EXIT_FAILURE;
1.17 schwarze 1040: }
1.1 schwarze 1041:
1.20 schwarze 1042: if ( ! (NULL == req.q.arch || validate_urifrag(req.q.arch))) {
1043: pg_error_badrequest(
1044: "You specified an invalid architecture.");
1.48 schwarze 1045: return EXIT_FAILURE;
1.20 schwarze 1046: }
1047:
1.6 schwarze 1048: /* Dispatch to the three different pages. */
1.1 schwarze 1049:
1.6 schwarze 1050: if ('\0' != *path)
1051: pg_show(&req, path);
1.25 schwarze 1052: else if (NULL != req.q.query)
1.6 schwarze 1053: pg_search(&req);
1054: else
1.12 schwarze 1055: pg_index(&req);
1.1 schwarze 1056:
1.23 schwarze 1057: free(req.q.manpath);
1058: free(req.q.arch);
1059: free(req.q.sec);
1.25 schwarze 1060: free(req.q.query);
1.1 schwarze 1061: for (i = 0; i < (int)req.psz; i++)
1062: free(req.p[i]);
1063: free(req.p);
1.48 schwarze 1064: return EXIT_SUCCESS;
1.56 schwarze 1065: }
1066:
1067: /*
1068: * If PATH_INFO is not a file name, translate it to a query.
1069: */
1070: static void
1.68 schwarze 1071: parse_path_info(struct req *req, const char *path)
1.56 schwarze 1072: {
1.66 schwarze 1073: char *dir;
1.56 schwarze 1074:
1.60 schwarze 1075: req->isquery = 0;
1.56 schwarze 1076: req->q.equal = 1;
1077: req->q.manpath = mandoc_strdup(path);
1078:
1079: /* Mandatory manual page name. */
1080: if ((req->q.query = strrchr(req->q.manpath, '/')) == NULL) {
1081: req->q.query = req->q.manpath;
1082: req->q.manpath = NULL;
1083: } else
1084: *req->q.query++ = '\0';
1085:
1086: /* Optional trailing section. */
1087: if ((req->q.sec = strrchr(req->q.query, '.')) != NULL) {
1088: if(isdigit((unsigned char)req->q.sec[1])) {
1089: *req->q.sec++ = '\0';
1090: req->q.sec = mandoc_strdup(req->q.sec);
1091: } else
1092: req->q.sec = NULL;
1093: }
1094:
1095: /* Handle the case of name[.section] only. */
1096: if (req->q.manpath == NULL) {
1097: req->q.arch = NULL;
1098: return;
1099: }
1100: req->q.query = mandoc_strdup(req->q.query);
1101:
1102: /* Optional architecture. */
1.66 schwarze 1103: dir = strrchr(req->q.manpath, '/');
1104: if (dir != NULL && strncmp(dir + 1, "man", 3) != 0) {
1105: *dir++ = '\0';
1106: req->q.arch = mandoc_strdup(dir);
1107: dir = strrchr(req->q.manpath, '/');
1108: } else
1109: req->q.arch = NULL;
1.56 schwarze 1110:
1.66 schwarze 1111: /* Optional directory name. */
1112: if (dir != NULL && strncmp(dir + 1, "man", 3) == 0) {
1113: *dir++ = '\0';
1114: free(req->q.sec);
1115: req->q.sec = mandoc_strdup(dir + 3);
1.56 schwarze 1116: }
1.1 schwarze 1117: }
1118:
1119: /*
1120: * Scan for indexable paths.
1121: */
1122: static void
1.68 schwarze 1123: parse_manpath_conf(struct req *req)
1.1 schwarze 1124: {
1125: FILE *fp;
1126: char *dp;
1127: size_t dpsz;
1.54 schwarze 1128: ssize_t len;
1.1 schwarze 1129:
1.67 schwarze 1130: if ((fp = fopen("manpath.conf", "r")) == NULL) {
1131: warn("%s/manpath.conf", MAN_DIR);
1.14 schwarze 1132: pg_error_internal();
1133: exit(EXIT_FAILURE);
1134: }
1.1 schwarze 1135:
1.54 schwarze 1136: dp = NULL;
1137: dpsz = 0;
1138:
1139: while ((len = getline(&dp, &dpsz, fp)) != -1) {
1140: if (dp[len - 1] == '\n')
1141: dp[--len] = '\0';
1.1 schwarze 1142: req->p = mandoc_realloc(req->p,
1143: (req->psz + 1) * sizeof(char *));
1.20 schwarze 1144: if ( ! validate_urifrag(dp)) {
1.67 schwarze 1145: warnx("%s/manpath.conf contains "
1146: "unsafe path \"%s\"", MAN_DIR, dp);
1.20 schwarze 1147: pg_error_internal();
1148: exit(EXIT_FAILURE);
1149: }
1.67 schwarze 1150: if (strchr(dp, '/') != NULL) {
1151: warnx("%s/manpath.conf contains "
1152: "path with slash \"%s\"", MAN_DIR, dp);
1.20 schwarze 1153: pg_error_internal();
1154: exit(EXIT_FAILURE);
1155: }
1156: req->p[req->psz++] = dp;
1.54 schwarze 1157: dp = NULL;
1158: dpsz = 0;
1.14 schwarze 1159: }
1.54 schwarze 1160: free(dp);
1.14 schwarze 1161:
1.67 schwarze 1162: if (req->p == NULL) {
1163: warnx("%s/manpath.conf is empty", MAN_DIR);
1.14 schwarze 1164: pg_error_internal();
1165: exit(EXIT_FAILURE);
1.1 schwarze 1166: }
1167: }
![[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