Annotation of src/usr.bin/mandoc/cgi.c, Revision 1.76
1.76 ! schwarze 1: /* $OpenBSD: cgi.c,v 1.75 2016/07/31 23:35:26 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",
117: "octeon", "sgi", "socppc", "sparc",
118: "sparc64", "zaurus",
1.76 ! schwarze 119: "amiga", "arc", "armish", "arm32",
! 120: "atari", "aviion", "beagle", "cats",
! 121: "hppa64", "hp300",
1.64 schwarze 122: "ia64", "mac68k", "mvme68k", "mvme88k",
123: "mvmeppc", "palm", "pc532", "pegasos",
124: "pmax", "powerpc", "solbourne", "sun3",
125: "vax", "wgrisc", "x68k"
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);
339: }
340: }
341:
342: static void
1.1 schwarze 343: resp_begin_html(int code, const char *msg)
344: {
345:
346: resp_begin_http(code, msg);
347:
1.37 schwarze 348: printf("<!DOCTYPE html>\n"
1.65 schwarze 349: "<html>\n"
350: "<head>\n"
351: "<meta charset=\"UTF-8\"/>\n"
352: "<link rel=\"stylesheet\" href=\"%s/mandoc.css\""
353: " type=\"text/css\" media=\"all\">\n"
354: "<title>%s</title>\n"
355: "</head>\n"
356: "<body>\n"
1.1 schwarze 357: "<!-- Begin page content. //-->\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:
378: puts("<!-- Begin search form. //-->");
1.65 schwarze 379: printf("<div id=\"mancgi\">\n"
380: "<form action=\"/%s\" method=\"get\">\n"
381: "<fieldset>\n"
382: "<legend>Manual Page Search Parameters</legend>\n",
1.1 schwarze 383: scriptname);
1.8 schwarze 384:
385: /* Write query input box. */
386:
1.71 schwarze 387: printf("<input type=\"text\" name=\"query\" value=\"");
1.70 schwarze 388: if (req->q.query != NULL)
1.25 schwarze 389: html_print(req->q.query);
1.70 schwarze 390: printf( "\" size=\"40\"");
391: if (focus == FOCUS_QUERY)
392: printf(" autofocus");
393: puts(">");
1.8 schwarze 394:
1.71 schwarze 395: /* Write submission buttons. */
1.8 schwarze 396:
1.71 schwarze 397: printf( "<button type=\"submit\" name=\"apropos\" value=\"0\">"
398: "man</button>\n"
399: "<button type=\"submit\" name=\"apropos\" value=\"1\">"
400: "apropos</button>\n<br/>\n");
1.8 schwarze 401:
402: /* Write section selector. */
403:
1.71 schwarze 404: puts("<select name=\"sec\">");
1.8 schwarze 405: for (i = 0; i < sec_MAX; i++) {
1.65 schwarze 406: printf("<option value=\"%s\"", sec_numbers[i]);
1.8 schwarze 407: if (NULL != req->q.sec &&
408: 0 == strcmp(sec_numbers[i], req->q.sec))
1.65 schwarze 409: printf(" selected=\"selected\"");
410: printf(">%s</option>\n", sec_names[i]);
1.8 schwarze 411: }
1.65 schwarze 412: puts("</select>");
1.8 schwarze 413:
414: /* Write architecture selector. */
415:
1.65 schwarze 416: printf( "<select name=\"arch\">\n"
417: "<option value=\"default\"");
1.21 schwarze 418: if (NULL == req->q.arch)
1.65 schwarze 419: printf(" selected=\"selected\"");
420: puts(">All Architectures</option>");
1.8 schwarze 421: for (i = 0; i < arch_MAX; i++) {
1.65 schwarze 422: printf("<option value=\"%s\"", arch_names[i]);
1.8 schwarze 423: if (NULL != req->q.arch &&
424: 0 == strcmp(arch_names[i], req->q.arch))
1.65 schwarze 425: printf(" selected=\"selected\"");
426: printf(">%s</option>\n", arch_names[i]);
1.8 schwarze 427: }
1.65 schwarze 428: puts("</select>");
1.8 schwarze 429:
430: /* Write manpath selector. */
431:
1.1 schwarze 432: if (req->psz > 1) {
1.65 schwarze 433: puts("<select name=\"manpath\">");
1.1 schwarze 434: for (i = 0; i < (int)req->psz; i++) {
1.65 schwarze 435: printf("<option ");
1.41 schwarze 436: if (strcmp(req->q.manpath, req->p[i]) == 0)
1.65 schwarze 437: printf("selected=\"selected\" ");
438: printf("value=\"");
1.1 schwarze 439: html_print(req->p[i]);
440: printf("\">");
441: html_print(req->p[i]);
1.65 schwarze 442: puts("</option>");
1.1 schwarze 443: }
1.65 schwarze 444: puts("</select>");
1.1 schwarze 445: }
1.8 schwarze 446:
1.71 schwarze 447: puts("</fieldset>\n"
1.65 schwarze 448: "</form>\n"
449: "</div>");
1.1 schwarze 450: puts("<!-- End search form. //-->");
451: }
452:
1.16 schwarze 453: static int
1.20 schwarze 454: validate_urifrag(const char *frag)
455: {
456:
457: while ('\0' != *frag) {
458: if ( ! (isalnum((unsigned char)*frag) ||
459: '-' == *frag || '.' == *frag ||
460: '/' == *frag || '_' == *frag))
1.48 schwarze 461: return 0;
1.20 schwarze 462: frag++;
463: }
1.48 schwarze 464: return 1;
1.20 schwarze 465: }
466:
467: static int
1.17 schwarze 468: validate_manpath(const struct req *req, const char* manpath)
469: {
470: size_t i;
471:
472: for (i = 0; i < req->psz; i++)
473: if ( ! strcmp(manpath, req->p[i]))
1.48 schwarze 474: return 1;
1.17 schwarze 475:
1.48 schwarze 476: return 0;
1.17 schwarze 477: }
478:
479: static int
1.16 schwarze 480: validate_filename(const char *file)
481: {
482:
483: if ('.' == file[0] && '/' == file[1])
484: file += 2;
485:
1.48 schwarze 486: return ! (strstr(file, "../") || strstr(file, "/..") ||
487: (strncmp(file, "man", 3) && strncmp(file, "cat", 3)));
1.16 schwarze 488: }
489:
1.1 schwarze 490: static void
1.12 schwarze 491: pg_index(const struct req *req)
1.1 schwarze 492: {
493:
494: resp_begin_html(200, NULL);
1.70 schwarze 495: resp_searchform(req, FOCUS_QUERY);
1.65 schwarze 496: printf("<p>\n"
1.26 schwarze 497: "This web interface is documented in the\n"
1.73 schwarze 498: "<a href=\"/%s%sman.cgi.8\">man.cgi(8)</a>\n"
1.26 schwarze 499: "manual, and the\n"
1.73 schwarze 500: "<a href=\"/%s%sapropos.1\">apropos(1)</a>\n"
1.9 schwarze 501: "manual explains the query syntax.\n"
1.65 schwarze 502: "</p>\n",
1.58 schwarze 503: scriptname, *scriptname == '\0' ? "" : "/",
504: scriptname, *scriptname == '\0' ? "" : "/");
1.1 schwarze 505: resp_end_html();
506: }
507:
508: static void
1.12 schwarze 509: pg_noresult(const struct req *req, const char *msg)
1.1 schwarze 510: {
511: resp_begin_html(200, NULL);
1.70 schwarze 512: resp_searchform(req, FOCUS_QUERY);
1.65 schwarze 513: puts("<p>");
1.1 schwarze 514: puts(msg);
1.65 schwarze 515: puts("</p>");
1.1 schwarze 516: resp_end_html();
517: }
518:
519: static void
1.12 schwarze 520: pg_error_badrequest(const char *msg)
1.1 schwarze 521: {
522:
523: resp_begin_html(400, "Bad Request");
1.65 schwarze 524: puts("<h1>Bad Request</h1>\n"
525: "<p>\n");
1.1 schwarze 526: puts(msg);
527: printf("Try again from the\n"
1.65 schwarze 528: "<a href=\"/%s\">main page</a>.\n"
529: "</p>", scriptname);
1.1 schwarze 530: resp_end_html();
531: }
532:
533: static void
1.12 schwarze 534: pg_error_internal(void)
1.1 schwarze 535: {
536: resp_begin_html(500, "Internal Server Error");
1.65 schwarze 537: puts("<p>Internal Server Error</p>");
1.1 schwarze 538: resp_end_html();
539: }
540:
541: static void
1.12 schwarze 542: pg_searchres(const struct req *req, struct manpage *r, size_t sz)
1.1 schwarze 543: {
1.21 schwarze 544: char *arch, *archend;
1.59 schwarze 545: const char *sec;
546: size_t i, iuse;
1.21 schwarze 547: int archprio, archpriouse;
1.10 schwarze 548: int prio, priouse;
1.1 schwarze 549:
1.16 schwarze 550: for (i = 0; i < sz; i++) {
551: if (validate_filename(r[i].file))
552: continue;
1.67 schwarze 553: warnx("invalid filename %s in %s database",
1.16 schwarze 554: r[i].file, req->q.manpath);
555: pg_error_internal();
556: return;
557: }
558:
1.60 schwarze 559: if (req->isquery && sz == 1) {
1.1 schwarze 560: /*
561: * If we have just one result, then jump there now
562: * without any delay.
563: */
1.2 tedu 564: printf("Status: 303 See Other\r\n");
1.58 schwarze 565: printf("Location: http://%s/%s%s%s/%s",
566: HTTP_HOST, scriptname,
567: *scriptname == '\0' ? "" : "/",
568: req->q.manpath, r[0].file);
1.2 tedu 569: printf("\r\n"
570: "Content-Type: text/html; charset=utf-8\r\n"
571: "\r\n");
1.1 schwarze 572: return;
573: }
574:
575: resp_begin_html(200, NULL);
1.70 schwarze 576: resp_searchform(req,
577: req->q.equal || sz == 1 ? FOCUS_NONE : FOCUS_QUERY);
1.1 schwarze 578:
1.62 schwarze 579: if (sz > 1) {
1.65 schwarze 580: puts("<div class=\"results\">");
581: puts("<table>");
1.62 schwarze 582:
583: for (i = 0; i < sz; i++) {
1.65 schwarze 584: printf("<tr>\n"
585: "<td class=\"title\">\n"
586: "<a href=\"/%s%s%s/%s",
1.62 schwarze 587: scriptname, *scriptname == '\0' ? "" : "/",
588: req->q.manpath, r[i].file);
589: printf("\">");
590: html_print(r[i].names);
1.65 schwarze 591: printf("</a>\n"
592: "</td>\n"
593: "<td class=\"desc\">");
1.62 schwarze 594: html_print(r[i].output);
1.65 schwarze 595: puts("</td>\n"
596: "</tr>");
1.62 schwarze 597: }
598:
1.65 schwarze 599: puts("</table>\n"
600: "</div>");
1.1 schwarze 601: }
602:
1.10 schwarze 603: /*
604: * In man(1) mode, show one of the pages
605: * even if more than one is found.
606: */
607:
1.62 schwarze 608: if (req->q.equal || sz == 1) {
1.65 schwarze 609: puts("<hr>");
1.10 schwarze 610: iuse = 0;
1.59 schwarze 611: priouse = 20;
1.21 schwarze 612: archpriouse = 3;
1.10 schwarze 613: for (i = 0; i < sz; i++) {
1.59 schwarze 614: sec = r[i].file;
615: sec += strcspn(sec, "123456789");
616: if (sec[0] == '\0')
1.10 schwarze 617: continue;
1.59 schwarze 618: prio = sec_prios[sec[0] - '1'];
619: if (sec[1] != '/')
620: prio += 10;
621: if (req->q.arch == NULL) {
1.21 schwarze 622: archprio =
1.59 schwarze 623: ((arch = strchr(sec + 1, '/'))
624: == NULL) ? 3 :
625: ((archend = strchr(arch + 1, '/'))
626: == NULL) ? 0 :
1.21 schwarze 627: strncmp(arch, "amd64/",
628: archend - arch) ? 2 : 1;
629: if (archprio < archpriouse) {
630: archpriouse = archprio;
631: priouse = prio;
632: iuse = i;
633: continue;
634: }
635: if (archprio > archpriouse)
636: continue;
637: }
1.10 schwarze 638: if (prio >= priouse)
639: continue;
640: priouse = prio;
641: iuse = i;
642: }
643: resp_show(req, r[iuse].file);
644: }
645:
1.1 schwarze 646: resp_end_html();
647: }
648:
649: static void
1.68 schwarze 650: resp_catman(const struct req *req, const char *file)
1.1 schwarze 651: {
652: FILE *f;
1.54 schwarze 653: char *p;
654: size_t sz;
655: ssize_t len;
1.1 schwarze 656: int i;
657: int italic, bold;
658:
1.54 schwarze 659: if ((f = fopen(file, "r")) == NULL) {
1.65 schwarze 660: puts("<p>You specified an invalid manual file.</p>");
1.1 schwarze 661: return;
662: }
663:
1.65 schwarze 664: puts("<div class=\"catman\">\n"
665: "<pre>");
1.1 schwarze 666:
1.54 schwarze 667: p = NULL;
668: sz = 0;
669:
670: while ((len = getline(&p, &sz, f)) != -1) {
1.1 schwarze 671: bold = italic = 0;
1.54 schwarze 672: for (i = 0; i < len - 1; i++) {
1.43 schwarze 673: /*
1.1 schwarze 674: * This means that the catpage is out of state.
675: * Ignore it and keep going (although the
676: * catpage is bogus).
677: */
678:
679: if ('\b' == p[i] || '\n' == p[i])
680: continue;
681:
682: /*
683: * Print a regular character.
684: * Close out any bold/italic scopes.
685: * If we're in back-space mode, make sure we'll
686: * have something to enter when we backspace.
687: */
688:
689: if ('\b' != p[i + 1]) {
690: if (italic)
1.65 schwarze 691: printf("</i>");
1.1 schwarze 692: if (bold)
1.65 schwarze 693: printf("</b>");
1.1 schwarze 694: italic = bold = 0;
695: html_putchar(p[i]);
696: continue;
1.54 schwarze 697: } else if (i + 2 >= len)
1.1 schwarze 698: continue;
699:
700: /* Italic mode. */
701:
702: if ('_' == p[i]) {
703: if (bold)
1.65 schwarze 704: printf("</b>");
1.1 schwarze 705: if ( ! italic)
1.65 schwarze 706: printf("<i>");
1.1 schwarze 707: bold = 0;
708: italic = 1;
709: i += 2;
710: html_putchar(p[i]);
711: continue;
712: }
713:
1.43 schwarze 714: /*
1.1 schwarze 715: * Handle funny behaviour troff-isms.
716: * These grok'd from the original man2html.c.
717: */
718:
719: if (('+' == p[i] && 'o' == p[i + 2]) ||
720: ('o' == p[i] && '+' == p[i + 2]) ||
721: ('|' == 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: if (italic)
1.65 schwarze 728: printf("</i>");
1.1 schwarze 729: if (bold)
1.65 schwarze 730: printf("</b>");
1.1 schwarze 731: italic = bold = 0;
732: putchar('*');
733: i += 2;
734: continue;
735: } else if (('|' == p[i] && '-' == p[i + 2]) ||
736: ('-' == p[i] && '|' == p[i + 1]) ||
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: if (italic)
1.65 schwarze 742: printf("</i>");
1.1 schwarze 743: if (bold)
1.65 schwarze 744: printf("</b>");
1.1 schwarze 745: italic = bold = 0;
746: putchar('+');
747: i += 2;
748: continue;
749: }
750:
751: /* Bold mode. */
1.43 schwarze 752:
1.1 schwarze 753: if (italic)
1.65 schwarze 754: printf("</i>");
1.1 schwarze 755: if ( ! bold)
1.65 schwarze 756: printf("<b>");
1.1 schwarze 757: bold = 1;
758: italic = 0;
759: i += 2;
760: html_putchar(p[i]);
761: }
762:
1.43 schwarze 763: /*
1.1 schwarze 764: * Clean up the last character.
1.43 schwarze 765: * We can get to a newline; don't print that.
1.1 schwarze 766: */
767:
768: if (italic)
1.65 schwarze 769: printf("</i>");
1.1 schwarze 770: if (bold)
1.65 schwarze 771: printf("</b>");
1.1 schwarze 772:
1.54 schwarze 773: if (i == len - 1 && p[i] != '\n')
1.1 schwarze 774: html_putchar(p[i]);
775:
776: putchar('\n');
777: }
1.54 schwarze 778: free(p);
1.1 schwarze 779:
1.65 schwarze 780: puts("</pre>\n"
781: "</div>");
1.1 schwarze 782:
783: fclose(f);
784: }
785:
786: static void
1.68 schwarze 787: resp_format(const struct req *req, const char *file)
1.1 schwarze 788: {
1.45 schwarze 789: struct manoutput conf;
1.1 schwarze 790: struct mparse *mp;
1.46 schwarze 791: struct roff_man *man;
1.1 schwarze 792: void *vp;
1.30 schwarze 793: int fd;
794: int usepath;
1.1 schwarze 795:
796: if (-1 == (fd = open(file, O_RDONLY, 0))) {
1.65 schwarze 797: puts("<p>You specified an invalid manual file.</p>");
1.1 schwarze 798: return;
799: }
800:
1.49 schwarze 801: mchars_alloc();
1.75 schwarze 802: mp = mparse_alloc(MPARSE_SO | MPARSE_UTF8 | MPARSE_LATIN1,
803: MANDOCLEVEL_BADARG, NULL, req->q.manpath);
1.42 schwarze 804: mparse_readfd(mp, fd, file);
1.1 schwarze 805: close(fd);
806:
1.45 schwarze 807: memset(&conf, 0, sizeof(conf));
808: conf.fragment = 1;
1.30 schwarze 809: usepath = strcmp(req->q.manpath, req->p[0]);
1.61 schwarze 810: mandoc_asprintf(&conf.man, "/%s%s%%N.%%S",
811: usepath ? req->q.manpath : "", usepath ? "/" : "");
1.1 schwarze 812:
1.47 schwarze 813: mparse_result(mp, &man, NULL);
814: if (man == NULL) {
1.67 schwarze 815: warnx("fatal mandoc error: %s/%s", req->q.manpath, file);
1.12 schwarze 816: pg_error_internal();
1.1 schwarze 817: mparse_free(mp);
1.49 schwarze 818: mchars_free();
1.1 schwarze 819: return;
820: }
821:
1.49 schwarze 822: vp = html_alloc(&conf);
1.1 schwarze 823:
1.50 schwarze 824: if (man->macroset == MACROSET_MDOC) {
825: mdoc_validate(man);
1.47 schwarze 826: html_mdoc(vp, man);
1.51 schwarze 827: } else {
828: man_validate(man);
1.1 schwarze 829: html_man(vp, man);
1.51 schwarze 830: }
1.1 schwarze 831:
832: html_free(vp);
833: mparse_free(mp);
1.49 schwarze 834: mchars_free();
1.45 schwarze 835: free(conf.man);
1.1 schwarze 836: }
837:
838: static void
1.10 schwarze 839: resp_show(const struct req *req, const char *file)
840: {
1.16 schwarze 841:
842: if ('.' == file[0] && '/' == file[1])
1.11 schwarze 843: file += 2;
1.10 schwarze 844:
845: if ('c' == *file)
1.68 schwarze 846: resp_catman(req, file);
1.10 schwarze 847: else
1.68 schwarze 848: resp_format(req, file);
1.10 schwarze 849: }
850:
851: static void
1.24 schwarze 852: pg_show(struct req *req, const char *fullpath)
1.1 schwarze 853: {
1.24 schwarze 854: char *manpath;
855: const char *file;
1.1 schwarze 856:
1.24 schwarze 857: if ((file = strchr(fullpath, '/')) == NULL) {
1.12 schwarze 858: pg_error_badrequest(
1.1 schwarze 859: "You did not specify a page to show.");
860: return;
1.43 schwarze 861: }
1.24 schwarze 862: manpath = mandoc_strndup(fullpath, file - fullpath);
863: file++;
1.1 schwarze 864:
1.24 schwarze 865: if ( ! validate_manpath(req, manpath)) {
1.17 schwarze 866: pg_error_badrequest(
867: "You specified an invalid manpath.");
1.24 schwarze 868: free(manpath);
1.17 schwarze 869: return;
870: }
871:
1.1 schwarze 872: /*
873: * Begin by chdir()ing into the manpath.
874: * This way we can pick up the database files, which are
875: * relative to the manpath root.
876: */
877:
1.24 schwarze 878: if (chdir(manpath) == -1) {
1.67 schwarze 879: warn("chdir %s", manpath);
1.17 schwarze 880: pg_error_internal();
1.24 schwarze 881: free(manpath);
1.16 schwarze 882: return;
883: }
1.73 schwarze 884: free(manpath);
1.24 schwarze 885:
886: if ( ! validate_filename(file)) {
1.16 schwarze 887: pg_error_badrequest(
888: "You specified an invalid manual file.");
1.1 schwarze 889: return;
890: }
1.19 schwarze 891:
1.10 schwarze 892: resp_begin_html(200, NULL);
1.70 schwarze 893: resp_searchform(req, FOCUS_NONE);
1.24 schwarze 894: resp_show(req, file);
1.10 schwarze 895: resp_end_html();
1.1 schwarze 896: }
897:
898: static void
1.6 schwarze 899: pg_search(const struct req *req)
1.1 schwarze 900: {
901: struct mansearch search;
902: struct manpaths paths;
903: struct manpage *res;
1.36 schwarze 904: char **argv;
905: char *query, *rp, *wp;
1.1 schwarze 906: size_t ressz;
1.36 schwarze 907: int argc;
1.1 schwarze 908:
909: /*
910: * Begin by chdir()ing into the root of the manpath.
911: * This way we can pick up the database files, which are
912: * relative to the manpath root.
913: */
914:
1.67 schwarze 915: if (chdir(req->q.manpath) == -1) {
916: warn("chdir %s", req->q.manpath);
1.17 schwarze 917: pg_error_internal();
1.1 schwarze 918: return;
919: }
920:
921: search.arch = req->q.arch;
922: search.sec = req->q.sec;
1.35 schwarze 923: search.outkey = "Nd";
924: search.argmode = req->q.equal ? ARG_NAME : ARG_EXPR;
1.40 schwarze 925: search.firstmatch = 1;
1.1 schwarze 926:
927: paths.sz = 1;
928: paths.paths = mandoc_malloc(sizeof(char *));
929: paths.paths[0] = mandoc_strdup(".");
930:
931: /*
1.36 schwarze 932: * Break apart at spaces with backslash-escaping.
1.1 schwarze 933: */
934:
1.36 schwarze 935: argc = 0;
936: argv = NULL;
937: rp = query = mandoc_strdup(req->q.query);
938: for (;;) {
939: while (isspace((unsigned char)*rp))
940: rp++;
941: if (*rp == '\0')
942: break;
943: argv = mandoc_reallocarray(argv, argc + 1, sizeof(char *));
944: argv[argc++] = wp = rp;
945: for (;;) {
946: if (isspace((unsigned char)*rp)) {
947: *wp = '\0';
948: rp++;
949: break;
950: }
951: if (rp[0] == '\\' && rp[1] != '\0')
952: rp++;
953: if (wp != rp)
954: *wp = *rp;
955: if (*rp == '\0')
956: break;
957: wp++;
958: rp++;
959: }
1.1 schwarze 960: }
961:
1.36 schwarze 962: if (0 == mansearch(&search, &paths, argc, argv, &res, &ressz))
1.12 schwarze 963: pg_noresult(req, "You entered an invalid query.");
1.1 schwarze 964: else if (0 == ressz)
1.12 schwarze 965: pg_noresult(req, "No results found.");
1.1 schwarze 966: else
1.12 schwarze 967: pg_searchres(req, res, ressz);
1.1 schwarze 968:
1.36 schwarze 969: free(query);
970: mansearch_free(res, ressz);
1.1 schwarze 971: free(paths.paths[0]);
972: free(paths.paths);
973: }
974:
975: int
976: main(void)
977: {
1.6 schwarze 978: struct req req;
1.33 schwarze 979: struct itimerval itimer;
1.6 schwarze 980: const char *path;
1.23 schwarze 981: const char *querystring;
1.1 schwarze 982: int i;
1.33 schwarze 983:
984: /* Poor man's ReDoS mitigation. */
985:
1.38 schwarze 986: itimer.it_value.tv_sec = 2;
1.33 schwarze 987: itimer.it_value.tv_usec = 0;
1.38 schwarze 988: itimer.it_interval.tv_sec = 2;
1.33 schwarze 989: itimer.it_interval.tv_usec = 0;
990: if (setitimer(ITIMER_VIRTUAL, &itimer, NULL) == -1) {
1.67 schwarze 991: warn("setitimer");
1.20 schwarze 992: pg_error_internal();
1.48 schwarze 993: return EXIT_FAILURE;
1.20 schwarze 994: }
995:
1.1 schwarze 996: /*
1.7 schwarze 997: * First we change directory into the MAN_DIR so that
1.1 schwarze 998: * subsequent scanning for manpath directories is rooted
999: * relative to the same position.
1000: */
1001:
1.67 schwarze 1002: if (chdir(MAN_DIR) == -1) {
1003: warn("MAN_DIR: %s", MAN_DIR);
1.12 schwarze 1004: pg_error_internal();
1.48 schwarze 1005: return EXIT_FAILURE;
1.43 schwarze 1006: }
1.1 schwarze 1007:
1008: memset(&req, 0, sizeof(struct req));
1.57 schwarze 1009: req.q.equal = 1;
1.68 schwarze 1010: parse_manpath_conf(&req);
1.1 schwarze 1011:
1.56 schwarze 1012: /* Parse the path info and the query string. */
1.1 schwarze 1013:
1.56 schwarze 1014: if ((path = getenv("PATH_INFO")) == NULL)
1015: path = "";
1016: else if (*path == '/')
1017: path++;
1018:
1.63 schwarze 1019: if (*path != '\0') {
1.68 schwarze 1020: parse_path_info(&req, path);
1.72 schwarze 1021: if (req.q.manpath == NULL || access(path, F_OK) == -1)
1.63 schwarze 1022: path = "";
1.56 schwarze 1023: } else if ((querystring = getenv("QUERY_STRING")) != NULL)
1.68 schwarze 1024: parse_query_string(&req, querystring);
1.17 schwarze 1025:
1.56 schwarze 1026: /* Validate parsed data and add defaults. */
1027:
1.41 schwarze 1028: if (req.q.manpath == NULL)
1029: req.q.manpath = mandoc_strdup(req.p[0]);
1030: else if ( ! validate_manpath(&req, req.q.manpath)) {
1.17 schwarze 1031: pg_error_badrequest(
1032: "You specified an invalid manpath.");
1.48 schwarze 1033: return EXIT_FAILURE;
1.17 schwarze 1034: }
1.1 schwarze 1035:
1.20 schwarze 1036: if ( ! (NULL == req.q.arch || validate_urifrag(req.q.arch))) {
1037: pg_error_badrequest(
1038: "You specified an invalid architecture.");
1.48 schwarze 1039: return EXIT_FAILURE;
1.20 schwarze 1040: }
1041:
1.6 schwarze 1042: /* Dispatch to the three different pages. */
1.1 schwarze 1043:
1.6 schwarze 1044: if ('\0' != *path)
1045: pg_show(&req, path);
1.25 schwarze 1046: else if (NULL != req.q.query)
1.6 schwarze 1047: pg_search(&req);
1048: else
1.12 schwarze 1049: pg_index(&req);
1.1 schwarze 1050:
1.23 schwarze 1051: free(req.q.manpath);
1052: free(req.q.arch);
1053: free(req.q.sec);
1.25 schwarze 1054: free(req.q.query);
1.1 schwarze 1055: for (i = 0; i < (int)req.psz; i++)
1056: free(req.p[i]);
1057: free(req.p);
1.48 schwarze 1058: return EXIT_SUCCESS;
1.56 schwarze 1059: }
1060:
1061: /*
1062: * If PATH_INFO is not a file name, translate it to a query.
1063: */
1064: static void
1.68 schwarze 1065: parse_path_info(struct req *req, const char *path)
1.56 schwarze 1066: {
1.74 schwarze 1067: char *dir[4];
1068: int i;
1.56 schwarze 1069:
1.60 schwarze 1070: req->isquery = 0;
1.56 schwarze 1071: req->q.equal = 1;
1072: req->q.manpath = mandoc_strdup(path);
1.74 schwarze 1073: req->q.arch = NULL;
1.56 schwarze 1074:
1075: /* Mandatory manual page name. */
1076: if ((req->q.query = strrchr(req->q.manpath, '/')) == NULL) {
1077: req->q.query = req->q.manpath;
1078: req->q.manpath = NULL;
1079: } else
1080: *req->q.query++ = '\0';
1081:
1082: /* Optional trailing section. */
1083: if ((req->q.sec = strrchr(req->q.query, '.')) != NULL) {
1084: if(isdigit((unsigned char)req->q.sec[1])) {
1085: *req->q.sec++ = '\0';
1086: req->q.sec = mandoc_strdup(req->q.sec);
1087: } else
1088: req->q.sec = NULL;
1089: }
1090:
1091: /* Handle the case of name[.section] only. */
1.74 schwarze 1092: if (req->q.manpath == NULL)
1.56 schwarze 1093: return;
1.74 schwarze 1094: req->q.query = mandoc_strdup(req->q.query);
1095:
1096: /* Split directory components. */
1097: dir[i = 0] = req->q.manpath;
1098: while ((dir[i + 1] = strchr(dir[i], '/')) != NULL) {
1099: if (++i == 3) {
1100: pg_error_badrequest(
1101: "You specified too many directory components.");
1102: exit(EXIT_FAILURE);
1103: }
1104: *dir[i]++ = '\0';
1.56 schwarze 1105: }
1106:
1.74 schwarze 1107: /* Optional manpath. */
1108: if ((i = validate_manpath(req, req->q.manpath)) == 0)
1109: req->q.manpath = NULL;
1110: else if (dir[1] == NULL)
1111: return;
1.56 schwarze 1112:
1.74 schwarze 1113: /* Optional section. */
1114: if (strncmp(dir[i], "man", 3) == 0) {
1.66 schwarze 1115: free(req->q.sec);
1.74 schwarze 1116: req->q.sec = mandoc_strdup(dir[i++] + 3);
1.56 schwarze 1117: }
1.74 schwarze 1118: if (dir[i] == NULL) {
1119: if (req->q.manpath == NULL)
1120: free(dir[0]);
1121: return;
1122: }
1123: if (dir[i + 1] != NULL) {
1124: pg_error_badrequest(
1125: "You specified an invalid directory component.");
1126: exit(EXIT_FAILURE);
1127: }
1128:
1129: /* Optional architecture. */
1130: if (i) {
1131: req->q.arch = mandoc_strdup(dir[i]);
1132: if (req->q.manpath == NULL)
1133: free(dir[0]);
1134: } else
1135: req->q.arch = dir[0];
1.1 schwarze 1136: }
1137:
1138: /*
1139: * Scan for indexable paths.
1140: */
1141: static void
1.68 schwarze 1142: parse_manpath_conf(struct req *req)
1.1 schwarze 1143: {
1144: FILE *fp;
1145: char *dp;
1146: size_t dpsz;
1.54 schwarze 1147: ssize_t len;
1.1 schwarze 1148:
1.67 schwarze 1149: if ((fp = fopen("manpath.conf", "r")) == NULL) {
1150: warn("%s/manpath.conf", MAN_DIR);
1.14 schwarze 1151: pg_error_internal();
1152: exit(EXIT_FAILURE);
1153: }
1.1 schwarze 1154:
1.54 schwarze 1155: dp = NULL;
1156: dpsz = 0;
1157:
1158: while ((len = getline(&dp, &dpsz, fp)) != -1) {
1159: if (dp[len - 1] == '\n')
1160: dp[--len] = '\0';
1.1 schwarze 1161: req->p = mandoc_realloc(req->p,
1162: (req->psz + 1) * sizeof(char *));
1.20 schwarze 1163: if ( ! validate_urifrag(dp)) {
1.67 schwarze 1164: warnx("%s/manpath.conf contains "
1165: "unsafe path \"%s\"", MAN_DIR, dp);
1.20 schwarze 1166: pg_error_internal();
1167: exit(EXIT_FAILURE);
1168: }
1.67 schwarze 1169: if (strchr(dp, '/') != NULL) {
1170: warnx("%s/manpath.conf contains "
1171: "path with slash \"%s\"", MAN_DIR, dp);
1.20 schwarze 1172: pg_error_internal();
1173: exit(EXIT_FAILURE);
1174: }
1175: req->p[req->psz++] = dp;
1.54 schwarze 1176: dp = NULL;
1177: dpsz = 0;
1.14 schwarze 1178: }
1.54 schwarze 1179: free(dp);
1.14 schwarze 1180:
1.67 schwarze 1181: if (req->p == NULL) {
1182: warnx("%s/manpath.conf is empty", MAN_DIR);
1.14 schwarze 1183: pg_error_internal();
1184: exit(EXIT_FAILURE);
1.1 schwarze 1185: }
1186: }
![[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