Annotation of src/usr.bin/mandoc/cgi.c, Revision 1.59
1.59 ! schwarze 1: /* $OpenBSD: cgi.c,v 1.58 2016/03/18 13:21:07 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>
22: #include <errno.h>
23: #include <fcntl.h>
24: #include <limits.h>
1.32 schwarze 25: #include <stdint.h>
1.1 schwarze 26: #include <stdio.h>
27: #include <stdlib.h>
28: #include <string.h>
29: #include <unistd.h>
30:
1.47 schwarze 31: #include "mandoc_aux.h"
1.1 schwarze 32: #include "mandoc.h"
1.47 schwarze 33: #include "roff.h"
1.50 schwarze 34: #include "mdoc.h"
1.51 schwarze 35: #include "man.h"
1.1 schwarze 36: #include "main.h"
1.44 schwarze 37: #include "manconf.h"
1.1 schwarze 38: #include "mansearch.h"
1.7 schwarze 39: #include "cgi.h"
1.1 schwarze 40:
41: /*
42: * A query as passed to the search function.
43: */
44: struct query {
1.23 schwarze 45: char *manpath; /* desired manual directory */
46: char *arch; /* architecture */
47: char *sec; /* manual section */
1.25 schwarze 48: char *query; /* unparsed query expression */
1.5 schwarze 49: int equal; /* match whole names, not substrings */
1.1 schwarze 50: };
51:
52: struct req {
53: struct query q;
54: char **p; /* array of available manpaths */
55: size_t psz; /* number of available manpaths */
56: };
57:
58: static void catman(const struct req *, const char *);
59: static void format(const struct req *, const char *);
60: static void html_print(const char *);
61: static void html_putchar(char);
1.43 schwarze 62: static int http_decode(char *);
1.23 schwarze 63: static void http_parse(struct req *, const char *);
1.1 schwarze 64: static void pathgen(struct req *);
1.56 schwarze 65: static void path_parse(struct req *req, const char *path);
1.12 schwarze 66: static void pg_error_badrequest(const char *);
67: static void pg_error_internal(void);
68: static void pg_index(const struct req *);
69: static void pg_noresult(const struct req *, const char *);
1.6 schwarze 70: static void pg_search(const struct req *);
1.12 schwarze 71: static void pg_searchres(const struct req *,
72: struct manpage *, size_t);
1.19 schwarze 73: static void pg_show(struct req *, const char *);
1.1 schwarze 74: static void resp_begin_html(int, const char *);
75: static void resp_begin_http(int, const char *);
1.53 schwarze 76: static void resp_copy(const char *);
1.1 schwarze 77: static void resp_end_html(void);
78: static void resp_searchform(const struct req *);
1.10 schwarze 79: static void resp_show(const struct req *, const char *);
1.25 schwarze 80: static void set_query_attr(char **, char **);
81: static int validate_filename(const char *);
82: static int validate_manpath(const struct req *, const char *);
83: static int validate_urifrag(const char *);
1.1 schwarze 84:
1.58 schwarze 85: static const char *scriptname = SCRIPT_NAME;
1.1 schwarze 86:
1.10 schwarze 87: static const int sec_prios[] = {1, 4, 5, 8, 6, 3, 7, 2, 9};
1.8 schwarze 88: static const char *const sec_numbers[] = {
89: "0", "1", "2", "3", "3p", "4", "5", "6", "7", "8", "9"
90: };
91: static const char *const sec_names[] = {
92: "All Sections",
93: "1 - General Commands",
94: "2 - System Calls",
1.34 schwarze 95: "3 - Library Functions",
96: "3p - Perl Library",
97: "4 - Device Drivers",
1.8 schwarze 98: "5 - File Formats",
99: "6 - Games",
1.34 schwarze 100: "7 - Miscellaneous Information",
101: "8 - System Manager\'s Manual",
102: "9 - Kernel Developer\'s Manual"
1.8 schwarze 103: };
104: static const int sec_MAX = sizeof(sec_names) / sizeof(char *);
105:
106: static const char *const arch_names[] = {
107: "amd64", "alpha", "armish", "armv7",
108: "aviion", "hppa", "hppa64", "i386",
109: "ia64", "landisk", "loongson", "luna88k",
110: "macppc", "mips64", "octeon", "sgi",
111: "socppc", "solbourne", "sparc", "sparc64",
112: "vax", "zaurus",
113: "amiga", "arc", "arm32", "atari",
114: "beagle", "cats", "hp300", "mac68k",
115: "mvme68k", "mvme88k", "mvmeppc", "palm",
116: "pc532", "pegasos", "pmax", "powerpc",
117: "sun3", "wgrisc", "x68k"
118: };
119: static const int arch_MAX = sizeof(arch_names) / sizeof(char *);
120:
1.1 schwarze 121: /*
122: * Print a character, escaping HTML along the way.
123: * This will pass non-ASCII straight to output: be warned!
124: */
125: static void
126: html_putchar(char c)
127: {
128:
129: switch (c) {
130: case ('"'):
131: printf(""e;");
132: break;
133: case ('&'):
134: printf("&");
135: break;
136: case ('>'):
137: printf(">");
138: break;
139: case ('<'):
140: printf("<");
141: break;
142: default:
143: putchar((unsigned char)c);
144: break;
145: }
146: }
147:
148: /*
149: * Call through to html_putchar().
150: * Accepts NULL strings.
151: */
152: static void
153: html_print(const char *p)
154: {
1.43 schwarze 155:
1.1 schwarze 156: if (NULL == p)
157: return;
158: while ('\0' != *p)
159: html_putchar(*p++);
160: }
161:
162: /*
1.23 schwarze 163: * Transfer the responsibility for the allocated string *val
164: * to the query structure.
1.1 schwarze 165: */
166: static void
1.23 schwarze 167: set_query_attr(char **attr, char **val)
1.1 schwarze 168: {
169:
1.23 schwarze 170: free(*attr);
171: if (**val == '\0') {
172: *attr = NULL;
173: free(*val);
174: } else
175: *attr = *val;
176: *val = NULL;
177: }
178:
179: /*
180: * Parse the QUERY_STRING for key-value pairs
181: * and store the values into the query structure.
182: */
183: static void
184: http_parse(struct req *req, const char *qs)
185: {
186: char *key, *val;
187: size_t keysz, valsz;
188:
189: req->q.manpath = NULL;
190: req->q.arch = NULL;
191: req->q.sec = NULL;
1.25 schwarze 192: req->q.query = NULL;
1.23 schwarze 193: req->q.equal = 1;
194:
195: key = val = NULL;
196: while (*qs != '\0') {
1.1 schwarze 197:
1.23 schwarze 198: /* Parse one key. */
199:
200: keysz = strcspn(qs, "=;&");
201: key = mandoc_strndup(qs, keysz);
202: qs += keysz;
203: if (*qs != '=')
204: goto next;
205:
206: /* Parse one value. */
207:
208: valsz = strcspn(++qs, ";&");
209: val = mandoc_strndup(qs, valsz);
210: qs += valsz;
211:
212: /* Decode and catch encoding errors. */
1.1 schwarze 213:
1.23 schwarze 214: if ( ! (http_decode(key) && http_decode(val)))
215: goto next;
1.1 schwarze 216:
1.23 schwarze 217: /* Handle key-value pairs. */
1.1 schwarze 218:
1.23 schwarze 219: if ( ! strcmp(key, "query"))
1.25 schwarze 220: set_query_attr(&req->q.query, &val);
1.1 schwarze 221:
1.23 schwarze 222: else if ( ! strcmp(key, "apropos"))
223: req->q.equal = !strcmp(val, "0");
224:
225: else if ( ! strcmp(key, "manpath")) {
1.13 schwarze 226: #ifdef COMPAT_OLDURI
1.23 schwarze 227: if ( ! strncmp(val, "OpenBSD ", 8)) {
1.13 schwarze 228: val[7] = '-';
229: if ('C' == val[8])
230: val[8] = 'c';
231: }
232: #endif
1.23 schwarze 233: set_query_attr(&req->q.manpath, &val);
234: }
235:
236: else if ( ! (strcmp(key, "sec")
1.13 schwarze 237: #ifdef COMPAT_OLDURI
1.23 schwarze 238: && strcmp(key, "sektion")
1.13 schwarze 239: #endif
1.23 schwarze 240: )) {
241: if ( ! strcmp(val, "0"))
242: *val = '\0';
243: set_query_attr(&req->q.sec, &val);
1.5 schwarze 244: }
1.23 schwarze 245:
246: else if ( ! strcmp(key, "arch")) {
247: if ( ! strcmp(val, "default"))
248: *val = '\0';
249: set_query_attr(&req->q.arch, &val);
250: }
251:
252: /*
253: * The key must be freed in any case.
254: * The val may have been handed over to the query
255: * structure, in which case it is now NULL.
256: */
257: next:
258: free(key);
259: key = NULL;
260: free(val);
261: val = NULL;
262:
263: if (*qs != '\0')
264: qs++;
1.1 schwarze 265: }
266: }
267:
268: /*
269: * HTTP-decode a string. The standard explanation is that this turns
270: * "%4e+foo" into "n foo" in the regular way. This is done in-place
271: * over the allocated string.
272: */
273: static int
274: http_decode(char *p)
275: {
276: char hex[3];
1.3 tedu 277: char *q;
1.1 schwarze 278: int c;
279:
280: hex[2] = '\0';
281:
1.3 tedu 282: q = p;
283: for ( ; '\0' != *p; p++, q++) {
1.1 schwarze 284: if ('%' == *p) {
285: if ('\0' == (hex[0] = *(p + 1)))
1.48 schwarze 286: return 0;
1.1 schwarze 287: if ('\0' == (hex[1] = *(p + 2)))
1.48 schwarze 288: return 0;
1.1 schwarze 289: if (1 != sscanf(hex, "%x", &c))
1.48 schwarze 290: return 0;
1.1 schwarze 291: if ('\0' == c)
1.48 schwarze 292: return 0;
1.1 schwarze 293:
1.3 tedu 294: *q = (char)c;
295: p += 2;
1.1 schwarze 296: } else
1.3 tedu 297: *q = '+' == *p ? ' ' : *p;
1.1 schwarze 298: }
299:
1.3 tedu 300: *q = '\0';
1.48 schwarze 301: return 1;
1.1 schwarze 302: }
303:
304: static void
305: resp_begin_http(int code, const char *msg)
306: {
307:
308: if (200 != code)
1.2 tedu 309: printf("Status: %d %s\r\n", code, msg);
1.1 schwarze 310:
1.2 tedu 311: printf("Content-Type: text/html; charset=utf-8\r\n"
312: "Cache-Control: no-cache\r\n"
313: "Pragma: no-cache\r\n"
314: "\r\n");
1.1 schwarze 315:
316: fflush(stdout);
317: }
318:
319: static void
1.53 schwarze 320: resp_copy(const char *filename)
321: {
322: char buf[4096];
323: ssize_t sz;
324: int fd;
325:
326: if ((fd = open(filename, O_RDONLY)) != -1) {
327: fflush(stdout);
328: while ((sz = read(fd, buf, sizeof(buf))) > 0)
329: write(STDOUT_FILENO, buf, sz);
330: }
331: }
332:
333: static void
1.1 schwarze 334: resp_begin_html(int code, const char *msg)
335: {
336:
337: resp_begin_http(code, msg);
338:
1.37 schwarze 339: printf("<!DOCTYPE html>\n"
1.1 schwarze 340: "<HTML>\n"
341: "<HEAD>\n"
1.37 schwarze 342: "<META CHARSET=\"UTF-8\" />\n"
1.52 schwarze 343: "<LINK REL=\"stylesheet\" HREF=\"%s/mandoc.css\""
1.1 schwarze 344: " TYPE=\"text/css\" media=\"all\">\n"
1.7 schwarze 345: "<TITLE>%s</TITLE>\n"
1.1 schwarze 346: "</HEAD>\n"
347: "<BODY>\n"
348: "<!-- Begin page content. //-->\n",
1.52 schwarze 349: CSS_DIR, CUSTOMIZE_TITLE);
1.53 schwarze 350:
351: resp_copy(MAN_DIR "/header.html");
1.1 schwarze 352: }
353:
354: static void
355: resp_end_html(void)
356: {
357:
1.53 schwarze 358: resp_copy(MAN_DIR "/footer.html");
359:
1.1 schwarze 360: puts("</BODY>\n"
361: "</HTML>");
362: }
363:
364: static void
365: resp_searchform(const struct req *req)
366: {
367: int i;
368:
369: puts("<!-- Begin search form. //-->");
370: printf("<DIV ID=\"mancgi\">\n"
1.58 schwarze 371: "<FORM ACTION=\"/%s\" METHOD=\"get\">\n"
1.1 schwarze 372: "<FIELDSET>\n"
1.8 schwarze 373: "<LEGEND>Manual Page Search Parameters</LEGEND>\n",
1.1 schwarze 374: scriptname);
1.8 schwarze 375:
376: /* Write query input box. */
377:
378: printf( "<TABLE><TR><TD>\n"
379: "<INPUT TYPE=\"text\" NAME=\"query\" VALUE=\"");
1.25 schwarze 380: if (NULL != req->q.query)
381: html_print(req->q.query);
1.8 schwarze 382: puts("\" SIZE=\"40\">");
383:
384: /* Write submission and reset buttons. */
385:
386: printf( "<INPUT TYPE=\"submit\" VALUE=\"Submit\">\n"
387: "<INPUT TYPE=\"reset\" VALUE=\"Reset\">\n");
388:
389: /* Write show radio button */
390:
391: printf( "</TD><TD>\n"
392: "<INPUT TYPE=\"radio\" ");
1.5 schwarze 393: if (req->q.equal)
1.26 schwarze 394: printf("CHECKED=\"checked\" ");
1.8 schwarze 395: printf( "NAME=\"apropos\" ID=\"show\" VALUE=\"0\">\n"
396: "<LABEL FOR=\"show\">Show named manual page</LABEL>\n");
397:
398: /* Write section selector. */
399:
1.26 schwarze 400: puts( "</TD></TR><TR><TD>\n"
1.8 schwarze 401: "<SELECT NAME=\"sec\">");
402: for (i = 0; i < sec_MAX; i++) {
403: printf("<OPTION VALUE=\"%s\"", sec_numbers[i]);
404: if (NULL != req->q.sec &&
405: 0 == strcmp(sec_numbers[i], req->q.sec))
1.26 schwarze 406: printf(" SELECTED=\"selected\"");
1.8 schwarze 407: printf(">%s</OPTION>\n", sec_names[i]);
408: }
409: puts("</SELECT>");
410:
411: /* Write architecture selector. */
412:
1.21 schwarze 413: printf( "<SELECT NAME=\"arch\">\n"
414: "<OPTION VALUE=\"default\"");
415: if (NULL == req->q.arch)
1.26 schwarze 416: printf(" SELECTED=\"selected\"");
1.21 schwarze 417: puts(">All Architectures</OPTION>");
1.8 schwarze 418: for (i = 0; i < arch_MAX; i++) {
419: printf("<OPTION VALUE=\"%s\"", arch_names[i]);
420: if (NULL != req->q.arch &&
421: 0 == strcmp(arch_names[i], req->q.arch))
1.26 schwarze 422: printf(" SELECTED=\"selected\"");
1.8 schwarze 423: printf(">%s</OPTION>\n", arch_names[i]);
424: }
425: puts("</SELECT>");
426:
427: /* Write manpath selector. */
428:
1.1 schwarze 429: if (req->psz > 1) {
1.8 schwarze 430: puts("<SELECT NAME=\"manpath\">");
1.1 schwarze 431: for (i = 0; i < (int)req->psz; i++) {
432: printf("<OPTION ");
1.41 schwarze 433: if (strcmp(req->q.manpath, req->p[i]) == 0)
1.26 schwarze 434: printf("SELECTED=\"selected\" ");
1.1 schwarze 435: printf("VALUE=\"");
436: html_print(req->p[i]);
437: printf("\">");
438: html_print(req->p[i]);
439: puts("</OPTION>");
440: }
441: puts("</SELECT>");
442: }
1.8 schwarze 443:
444: /* Write search radio button */
445:
446: printf( "</TD><TD>\n"
447: "<INPUT TYPE=\"radio\" ");
448: if (0 == req->q.equal)
1.26 schwarze 449: printf("CHECKED=\"checked\" ");
1.8 schwarze 450: printf( "NAME=\"apropos\" ID=\"search\" VALUE=\"1\">\n"
451: "<LABEL FOR=\"search\">Search with apropos query</LABEL>\n");
452:
453: puts("</TD></TR></TABLE>\n"
1.1 schwarze 454: "</FIELDSET>\n"
455: "</FORM>\n"
456: "</DIV>");
457: puts("<!-- End search form. //-->");
458: }
459:
1.16 schwarze 460: static int
1.20 schwarze 461: validate_urifrag(const char *frag)
462: {
463:
464: while ('\0' != *frag) {
465: if ( ! (isalnum((unsigned char)*frag) ||
466: '-' == *frag || '.' == *frag ||
467: '/' == *frag || '_' == *frag))
1.48 schwarze 468: return 0;
1.20 schwarze 469: frag++;
470: }
1.48 schwarze 471: return 1;
1.20 schwarze 472: }
473:
474: static int
1.17 schwarze 475: validate_manpath(const struct req *req, const char* manpath)
476: {
477: size_t i;
478:
479: if ( ! strcmp(manpath, "mandoc"))
1.48 schwarze 480: return 1;
1.17 schwarze 481:
482: for (i = 0; i < req->psz; i++)
483: if ( ! strcmp(manpath, req->p[i]))
1.48 schwarze 484: return 1;
1.17 schwarze 485:
1.48 schwarze 486: return 0;
1.17 schwarze 487: }
488:
489: static int
1.16 schwarze 490: validate_filename(const char *file)
491: {
492:
493: if ('.' == file[0] && '/' == file[1])
494: file += 2;
495:
1.48 schwarze 496: return ! (strstr(file, "../") || strstr(file, "/..") ||
497: (strncmp(file, "man", 3) && strncmp(file, "cat", 3)));
1.16 schwarze 498: }
499:
1.1 schwarze 500: static void
1.12 schwarze 501: pg_index(const struct req *req)
1.1 schwarze 502: {
503:
504: resp_begin_html(200, NULL);
505: resp_searchform(req);
1.4 schwarze 506: printf("<P>\n"
1.26 schwarze 507: "This web interface is documented in the\n"
1.58 schwarze 508: "<A HREF=\"/%s%smandoc/man8/man.cgi.8\">man.cgi</A>\n"
1.26 schwarze 509: "manual, and the\n"
1.58 schwarze 510: "<A HREF=\"/%s%smandoc/man1/apropos.1\">apropos</A>\n"
1.9 schwarze 511: "manual explains the query syntax.\n"
1.4 schwarze 512: "</P>\n",
1.58 schwarze 513: scriptname, *scriptname == '\0' ? "" : "/",
514: scriptname, *scriptname == '\0' ? "" : "/");
1.1 schwarze 515: resp_end_html();
516: }
517:
518: static void
1.12 schwarze 519: pg_noresult(const struct req *req, const char *msg)
1.1 schwarze 520: {
521: resp_begin_html(200, NULL);
522: resp_searchform(req);
523: puts("<P>");
524: puts(msg);
525: puts("</P>");
526: resp_end_html();
527: }
528:
529: static void
1.12 schwarze 530: pg_error_badrequest(const char *msg)
1.1 schwarze 531: {
532:
533: resp_begin_html(400, "Bad Request");
534: puts("<H1>Bad Request</H1>\n"
535: "<P>\n");
536: puts(msg);
537: printf("Try again from the\n"
1.58 schwarze 538: "<A HREF=\"/%s\">main page</A>.\n"
1.1 schwarze 539: "</P>", scriptname);
540: resp_end_html();
541: }
542:
543: static void
1.12 schwarze 544: pg_error_internal(void)
1.1 schwarze 545: {
546: resp_begin_html(500, "Internal Server Error");
547: puts("<P>Internal Server Error</P>");
548: resp_end_html();
549: }
550:
551: static void
1.12 schwarze 552: pg_searchres(const struct req *req, struct manpage *r, size_t sz)
1.1 schwarze 553: {
1.21 schwarze 554: char *arch, *archend;
1.59 ! schwarze 555: const char *sec;
! 556: size_t i, iuse;
1.21 schwarze 557: int archprio, archpriouse;
1.10 schwarze 558: int prio, priouse;
1.1 schwarze 559:
1.16 schwarze 560: for (i = 0; i < sz; i++) {
561: if (validate_filename(r[i].file))
562: continue;
563: fprintf(stderr, "invalid filename %s in %s database\n",
564: r[i].file, req->q.manpath);
565: pg_error_internal();
566: return;
567: }
568:
1.1 schwarze 569: if (1 == sz) {
570: /*
571: * If we have just one result, then jump there now
572: * without any delay.
573: */
1.2 tedu 574: printf("Status: 303 See Other\r\n");
1.58 schwarze 575: printf("Location: http://%s/%s%s%s/%s",
576: HTTP_HOST, scriptname,
577: *scriptname == '\0' ? "" : "/",
578: req->q.manpath, r[0].file);
1.2 tedu 579: printf("\r\n"
580: "Content-Type: text/html; charset=utf-8\r\n"
581: "\r\n");
1.1 schwarze 582: return;
583: }
584:
585: resp_begin_html(200, NULL);
586: resp_searchform(req);
587: puts("<DIV CLASS=\"results\">");
588: puts("<TABLE>");
589:
590: for (i = 0; i < sz; i++) {
591: printf("<TR>\n"
592: "<TD CLASS=\"title\">\n"
1.58 schwarze 593: "<A HREF=\"/%s%s%s/%s",
594: scriptname, *scriptname == '\0' ? "" : "/",
595: req->q.manpath, r[i].file);
1.1 schwarze 596: printf("\">");
597: html_print(r[i].names);
598: printf("</A>\n"
599: "</TD>\n"
600: "<TD CLASS=\"desc\">");
601: html_print(r[i].output);
602: puts("</TD>\n"
603: "</TR>");
604: }
605:
606: puts("</TABLE>\n"
607: "</DIV>");
1.10 schwarze 608:
609: /*
610: * In man(1) mode, show one of the pages
611: * even if more than one is found.
612: */
613:
614: if (req->q.equal) {
615: puts("<HR>");
616: iuse = 0;
1.59 ! schwarze 617: priouse = 20;
1.21 schwarze 618: archpriouse = 3;
1.10 schwarze 619: for (i = 0; i < sz; i++) {
1.59 ! schwarze 620: sec = r[i].file;
! 621: sec += strcspn(sec, "123456789");
! 622: if (sec[0] == '\0')
1.10 schwarze 623: continue;
1.59 ! schwarze 624: prio = sec_prios[sec[0] - '1'];
! 625: if (sec[1] != '/')
! 626: prio += 10;
! 627: if (req->q.arch == NULL) {
1.21 schwarze 628: archprio =
1.59 ! schwarze 629: ((arch = strchr(sec + 1, '/'))
! 630: == NULL) ? 3 :
! 631: ((archend = strchr(arch + 1, '/'))
! 632: == NULL) ? 0 :
1.21 schwarze 633: strncmp(arch, "amd64/",
634: archend - arch) ? 2 : 1;
635: if (archprio < archpriouse) {
636: archpriouse = archprio;
637: priouse = prio;
638: iuse = i;
639: continue;
640: }
641: if (archprio > archpriouse)
642: continue;
643: }
1.10 schwarze 644: if (prio >= priouse)
645: continue;
646: priouse = prio;
647: iuse = i;
648: }
649: resp_show(req, r[iuse].file);
650: }
651:
1.1 schwarze 652: resp_end_html();
653: }
654:
655: static void
656: catman(const struct req *req, const char *file)
657: {
658: FILE *f;
1.54 schwarze 659: char *p;
660: size_t sz;
661: ssize_t len;
1.1 schwarze 662: int i;
663: int italic, bold;
664:
1.54 schwarze 665: if ((f = fopen(file, "r")) == NULL) {
1.10 schwarze 666: puts("<P>You specified an invalid manual file.</P>");
1.1 schwarze 667: return;
668: }
669:
670: puts("<DIV CLASS=\"catman\">\n"
671: "<PRE>");
672:
1.54 schwarze 673: p = NULL;
674: sz = 0;
675:
676: while ((len = getline(&p, &sz, f)) != -1) {
1.1 schwarze 677: bold = italic = 0;
1.54 schwarze 678: for (i = 0; i < len - 1; i++) {
1.43 schwarze 679: /*
1.1 schwarze 680: * This means that the catpage is out of state.
681: * Ignore it and keep going (although the
682: * catpage is bogus).
683: */
684:
685: if ('\b' == p[i] || '\n' == p[i])
686: continue;
687:
688: /*
689: * Print a regular character.
690: * Close out any bold/italic scopes.
691: * If we're in back-space mode, make sure we'll
692: * have something to enter when we backspace.
693: */
694:
695: if ('\b' != p[i + 1]) {
696: if (italic)
697: printf("</I>");
698: if (bold)
699: printf("</B>");
700: italic = bold = 0;
701: html_putchar(p[i]);
702: continue;
1.54 schwarze 703: } else if (i + 2 >= len)
1.1 schwarze 704: continue;
705:
706: /* Italic mode. */
707:
708: if ('_' == p[i]) {
709: if (bold)
710: printf("</B>");
711: if ( ! italic)
712: printf("<I>");
713: bold = 0;
714: italic = 1;
715: i += 2;
716: html_putchar(p[i]);
717: continue;
718: }
719:
1.43 schwarze 720: /*
1.1 schwarze 721: * Handle funny behaviour troff-isms.
722: * These grok'd from the original man2html.c.
723: */
724:
725: if (('+' == p[i] && 'o' == p[i + 2]) ||
726: ('o' == p[i] && '+' == p[i + 2]) ||
727: ('|' == p[i] && '=' == p[i + 2]) ||
728: ('=' == p[i] && '|' == p[i + 2]) ||
729: ('*' == p[i] && '=' == p[i + 2]) ||
730: ('=' == p[i] && '*' == p[i + 2]) ||
731: ('*' == p[i] && '|' == p[i + 2]) ||
732: ('|' == p[i] && '*' == p[i + 2])) {
733: if (italic)
734: printf("</I>");
735: if (bold)
736: printf("</B>");
737: italic = bold = 0;
738: putchar('*');
739: i += 2;
740: continue;
741: } else if (('|' == p[i] && '-' == p[i + 2]) ||
742: ('-' == p[i] && '|' == p[i + 1]) ||
743: ('+' == p[i] && '-' == p[i + 1]) ||
744: ('-' == p[i] && '+' == p[i + 1]) ||
745: ('+' == p[i] && '|' == p[i + 1]) ||
746: ('|' == p[i] && '+' == p[i + 1])) {
747: if (italic)
748: printf("</I>");
749: if (bold)
750: printf("</B>");
751: italic = bold = 0;
752: putchar('+');
753: i += 2;
754: continue;
755: }
756:
757: /* Bold mode. */
1.43 schwarze 758:
1.1 schwarze 759: if (italic)
760: printf("</I>");
761: if ( ! bold)
762: printf("<B>");
763: bold = 1;
764: italic = 0;
765: i += 2;
766: html_putchar(p[i]);
767: }
768:
1.43 schwarze 769: /*
1.1 schwarze 770: * Clean up the last character.
1.43 schwarze 771: * We can get to a newline; don't print that.
1.1 schwarze 772: */
773:
774: if (italic)
775: printf("</I>");
776: if (bold)
777: printf("</B>");
778:
1.54 schwarze 779: if (i == len - 1 && p[i] != '\n')
1.1 schwarze 780: html_putchar(p[i]);
781:
782: putchar('\n');
783: }
1.54 schwarze 784: free(p);
1.1 schwarze 785:
786: puts("</PRE>\n"
1.10 schwarze 787: "</DIV>");
1.1 schwarze 788:
789: fclose(f);
790: }
791:
792: static void
793: format(const struct req *req, const char *file)
794: {
1.45 schwarze 795: struct manoutput conf;
1.1 schwarze 796: struct mparse *mp;
1.46 schwarze 797: struct roff_man *man;
1.1 schwarze 798: void *vp;
1.30 schwarze 799: int fd;
800: int usepath;
1.1 schwarze 801:
802: if (-1 == (fd = open(file, O_RDONLY, 0))) {
1.10 schwarze 803: puts("<P>You specified an invalid manual file.</P>");
1.1 schwarze 804: return;
805: }
806:
1.49 schwarze 807: mchars_alloc();
808: mp = mparse_alloc(MPARSE_SO, MANDOCLEVEL_BADARG, NULL, req->q.manpath);
1.42 schwarze 809: mparse_readfd(mp, fd, file);
1.1 schwarze 810: close(fd);
811:
1.45 schwarze 812: memset(&conf, 0, sizeof(conf));
813: conf.fragment = 1;
1.30 schwarze 814: usepath = strcmp(req->q.manpath, req->p[0]);
1.58 schwarze 815: mandoc_asprintf(&conf.man, "/%s?query=%%N&sec=%%S%s%s%s%s",
1.30 schwarze 816: scriptname,
817: req->q.arch ? "&arch=" : "",
818: req->q.arch ? req->q.arch : "",
819: usepath ? "&manpath=" : "",
820: usepath ? req->q.manpath : "");
1.1 schwarze 821:
1.47 schwarze 822: mparse_result(mp, &man, NULL);
823: if (man == NULL) {
1.1 schwarze 824: fprintf(stderr, "fatal mandoc error: %s/%s\n",
825: req->q.manpath, file);
1.12 schwarze 826: pg_error_internal();
1.1 schwarze 827: mparse_free(mp);
1.49 schwarze 828: mchars_free();
1.1 schwarze 829: return;
830: }
831:
1.49 schwarze 832: vp = html_alloc(&conf);
1.1 schwarze 833:
1.50 schwarze 834: if (man->macroset == MACROSET_MDOC) {
835: mdoc_validate(man);
1.47 schwarze 836: html_mdoc(vp, man);
1.51 schwarze 837: } else {
838: man_validate(man);
1.1 schwarze 839: html_man(vp, man);
1.51 schwarze 840: }
1.1 schwarze 841:
842: html_free(vp);
843: mparse_free(mp);
1.49 schwarze 844: mchars_free();
1.45 schwarze 845: free(conf.man);
1.1 schwarze 846: }
847:
848: static void
1.10 schwarze 849: resp_show(const struct req *req, const char *file)
850: {
1.16 schwarze 851:
852: if ('.' == file[0] && '/' == file[1])
1.11 schwarze 853: file += 2;
1.10 schwarze 854:
855: if ('c' == *file)
856: catman(req, file);
857: else
858: format(req, file);
859: }
860:
861: static void
1.24 schwarze 862: pg_show(struct req *req, const char *fullpath)
1.1 schwarze 863: {
1.24 schwarze 864: char *manpath;
865: const char *file;
1.1 schwarze 866:
1.24 schwarze 867: if ((file = strchr(fullpath, '/')) == NULL) {
1.12 schwarze 868: pg_error_badrequest(
1.1 schwarze 869: "You did not specify a page to show.");
870: return;
1.43 schwarze 871: }
1.24 schwarze 872: manpath = mandoc_strndup(fullpath, file - fullpath);
873: file++;
1.1 schwarze 874:
1.24 schwarze 875: if ( ! validate_manpath(req, manpath)) {
1.17 schwarze 876: pg_error_badrequest(
877: "You specified an invalid manpath.");
1.24 schwarze 878: free(manpath);
1.17 schwarze 879: return;
880: }
881:
1.1 schwarze 882: /*
883: * Begin by chdir()ing into the manpath.
884: * This way we can pick up the database files, which are
885: * relative to the manpath root.
886: */
887:
1.24 schwarze 888: if (chdir(manpath) == -1) {
1.17 schwarze 889: fprintf(stderr, "chdir %s: %s\n",
1.24 schwarze 890: manpath, strerror(errno));
1.17 schwarze 891: pg_error_internal();
1.24 schwarze 892: free(manpath);
1.16 schwarze 893: return;
894: }
895:
1.24 schwarze 896: if (strcmp(manpath, "mandoc")) {
897: free(req->q.manpath);
898: req->q.manpath = manpath;
899: } else
900: free(manpath);
901:
902: if ( ! validate_filename(file)) {
1.16 schwarze 903: pg_error_badrequest(
904: "You specified an invalid manual file.");
1.1 schwarze 905: return;
906: }
1.19 schwarze 907:
1.10 schwarze 908: resp_begin_html(200, NULL);
909: resp_searchform(req);
1.24 schwarze 910: resp_show(req, file);
1.10 schwarze 911: resp_end_html();
1.1 schwarze 912: }
913:
914: static void
1.6 schwarze 915: pg_search(const struct req *req)
1.1 schwarze 916: {
917: struct mansearch search;
918: struct manpaths paths;
919: struct manpage *res;
1.36 schwarze 920: char **argv;
921: char *query, *rp, *wp;
1.1 schwarze 922: size_t ressz;
1.36 schwarze 923: int argc;
1.1 schwarze 924:
925: /*
926: * Begin by chdir()ing into the root of the manpath.
927: * This way we can pick up the database files, which are
928: * relative to the manpath root.
929: */
930:
931: if (-1 == (chdir(req->q.manpath))) {
1.17 schwarze 932: fprintf(stderr, "chdir %s: %s\n",
933: req->q.manpath, strerror(errno));
934: pg_error_internal();
1.1 schwarze 935: return;
936: }
937:
938: search.arch = req->q.arch;
939: search.sec = req->q.sec;
1.35 schwarze 940: search.outkey = "Nd";
941: search.argmode = req->q.equal ? ARG_NAME : ARG_EXPR;
1.40 schwarze 942: search.firstmatch = 1;
1.1 schwarze 943:
944: paths.sz = 1;
945: paths.paths = mandoc_malloc(sizeof(char *));
946: paths.paths[0] = mandoc_strdup(".");
947:
948: /*
1.36 schwarze 949: * Break apart at spaces with backslash-escaping.
1.1 schwarze 950: */
951:
1.36 schwarze 952: argc = 0;
953: argv = NULL;
954: rp = query = mandoc_strdup(req->q.query);
955: for (;;) {
956: while (isspace((unsigned char)*rp))
957: rp++;
958: if (*rp == '\0')
959: break;
960: argv = mandoc_reallocarray(argv, argc + 1, sizeof(char *));
961: argv[argc++] = wp = rp;
962: for (;;) {
963: if (isspace((unsigned char)*rp)) {
964: *wp = '\0';
965: rp++;
966: break;
967: }
968: if (rp[0] == '\\' && rp[1] != '\0')
969: rp++;
970: if (wp != rp)
971: *wp = *rp;
972: if (*rp == '\0')
973: break;
974: wp++;
975: rp++;
976: }
1.1 schwarze 977: }
978:
1.36 schwarze 979: if (0 == mansearch(&search, &paths, argc, argv, &res, &ressz))
1.12 schwarze 980: pg_noresult(req, "You entered an invalid query.");
1.1 schwarze 981: else if (0 == ressz)
1.12 schwarze 982: pg_noresult(req, "No results found.");
1.1 schwarze 983: else
1.12 schwarze 984: pg_searchres(req, res, ressz);
1.1 schwarze 985:
1.36 schwarze 986: free(query);
987: mansearch_free(res, ressz);
1.1 schwarze 988: free(paths.paths[0]);
989: free(paths.paths);
990: }
991:
992: int
993: main(void)
994: {
1.6 schwarze 995: struct req req;
1.33 schwarze 996: struct itimerval itimer;
1.6 schwarze 997: const char *path;
1.23 schwarze 998: const char *querystring;
1.1 schwarze 999: int i;
1.33 schwarze 1000:
1001: /* Poor man's ReDoS mitigation. */
1002:
1.38 schwarze 1003: itimer.it_value.tv_sec = 2;
1.33 schwarze 1004: itimer.it_value.tv_usec = 0;
1.38 schwarze 1005: itimer.it_interval.tv_sec = 2;
1.33 schwarze 1006: itimer.it_interval.tv_usec = 0;
1007: if (setitimer(ITIMER_VIRTUAL, &itimer, NULL) == -1) {
1008: fprintf(stderr, "setitimer: %s\n", strerror(errno));
1.20 schwarze 1009: pg_error_internal();
1.48 schwarze 1010: return EXIT_FAILURE;
1.20 schwarze 1011: }
1012:
1.1 schwarze 1013: /*
1.7 schwarze 1014: * First we change directory into the MAN_DIR so that
1.1 schwarze 1015: * subsequent scanning for manpath directories is rooted
1016: * relative to the same position.
1017: */
1018:
1.7 schwarze 1019: if (-1 == chdir(MAN_DIR)) {
1.1 schwarze 1020: fprintf(stderr, "MAN_DIR: %s: %s\n",
1.7 schwarze 1021: MAN_DIR, strerror(errno));
1.12 schwarze 1022: pg_error_internal();
1.48 schwarze 1023: return EXIT_FAILURE;
1.43 schwarze 1024: }
1.1 schwarze 1025:
1026: memset(&req, 0, sizeof(struct req));
1.57 schwarze 1027: req.q.equal = 1;
1.1 schwarze 1028: pathgen(&req);
1029:
1.56 schwarze 1030: /* Parse the path info and the query string. */
1.1 schwarze 1031:
1.56 schwarze 1032: if ((path = getenv("PATH_INFO")) == NULL)
1033: path = "";
1034: else if (*path == '/')
1035: path++;
1036:
1037: if (*path != '\0' && access(path, F_OK) == -1) {
1038: path_parse(&req, path);
1039: path = "";
1040: } else if ((querystring = getenv("QUERY_STRING")) != NULL)
1.1 schwarze 1041: http_parse(&req, querystring);
1.17 schwarze 1042:
1.56 schwarze 1043: /* Validate parsed data and add defaults. */
1044:
1.41 schwarze 1045: if (req.q.manpath == NULL)
1046: req.q.manpath = mandoc_strdup(req.p[0]);
1047: else if ( ! validate_manpath(&req, req.q.manpath)) {
1.17 schwarze 1048: pg_error_badrequest(
1049: "You specified an invalid manpath.");
1.48 schwarze 1050: return EXIT_FAILURE;
1.17 schwarze 1051: }
1.1 schwarze 1052:
1.20 schwarze 1053: if ( ! (NULL == req.q.arch || validate_urifrag(req.q.arch))) {
1054: pg_error_badrequest(
1055: "You specified an invalid architecture.");
1.48 schwarze 1056: return EXIT_FAILURE;
1.20 schwarze 1057: }
1058:
1.6 schwarze 1059: /* Dispatch to the three different pages. */
1.1 schwarze 1060:
1.6 schwarze 1061: if ('\0' != *path)
1062: pg_show(&req, path);
1.25 schwarze 1063: else if (NULL != req.q.query)
1.6 schwarze 1064: pg_search(&req);
1065: else
1.12 schwarze 1066: pg_index(&req);
1.1 schwarze 1067:
1.23 schwarze 1068: free(req.q.manpath);
1069: free(req.q.arch);
1070: free(req.q.sec);
1.25 schwarze 1071: free(req.q.query);
1.1 schwarze 1072: for (i = 0; i < (int)req.psz; i++)
1073: free(req.p[i]);
1074: free(req.p);
1.48 schwarze 1075: return EXIT_SUCCESS;
1.56 schwarze 1076: }
1077:
1078: /*
1079: * If PATH_INFO is not a file name, translate it to a query.
1080: */
1081: static void
1082: path_parse(struct req *req, const char *path)
1083: {
1084: int dir_done;
1085:
1086: req->q.equal = 1;
1087: req->q.manpath = mandoc_strdup(path);
1088:
1089: /* Mandatory manual page name. */
1090: if ((req->q.query = strrchr(req->q.manpath, '/')) == NULL) {
1091: req->q.query = req->q.manpath;
1092: req->q.manpath = NULL;
1093: } else
1094: *req->q.query++ = '\0';
1095:
1096: /* Optional trailing section. */
1097: if ((req->q.sec = strrchr(req->q.query, '.')) != NULL) {
1098: if(isdigit((unsigned char)req->q.sec[1])) {
1099: *req->q.sec++ = '\0';
1100: req->q.sec = mandoc_strdup(req->q.sec);
1101: } else
1102: req->q.sec = NULL;
1103: }
1104:
1105: /* Handle the case of name[.section] only. */
1106: if (req->q.manpath == NULL) {
1107: req->q.arch = NULL;
1108: return;
1109: }
1110: req->q.query = mandoc_strdup(req->q.query);
1111:
1112: /* Optional architecture. */
1113: dir_done = 0;
1114: for (;;) {
1115: if ((req->q.arch = strrchr(req->q.manpath, '/')) == NULL)
1116: break;
1117: *req->q.arch++ = '\0';
1118: if (dir_done || strncmp(req->q.arch, "man", 3)) {
1119: req->q.arch = mandoc_strdup(req->q.arch);
1120: break;
1121: }
1122:
1123: /* Optional directory name. */
1124: req->q.arch += 3;
1125: if (*req->q.arch != '\0') {
1126: free(req->q.sec);
1127: req->q.sec = mandoc_strdup(req->q.arch);
1128: }
1129: dir_done = 1;
1130: }
1.1 schwarze 1131: }
1132:
1133: /*
1134: * Scan for indexable paths.
1135: */
1136: static void
1137: pathgen(struct req *req)
1138: {
1139: FILE *fp;
1140: char *dp;
1141: size_t dpsz;
1.54 schwarze 1142: ssize_t len;
1.1 schwarze 1143:
1.14 schwarze 1144: if (NULL == (fp = fopen("manpath.conf", "r"))) {
1145: fprintf(stderr, "%s/manpath.conf: %s\n",
1146: MAN_DIR, strerror(errno));
1147: pg_error_internal();
1148: exit(EXIT_FAILURE);
1149: }
1.1 schwarze 1150:
1.54 schwarze 1151: dp = NULL;
1152: dpsz = 0;
1153:
1154: while ((len = getline(&dp, &dpsz, fp)) != -1) {
1155: if (dp[len - 1] == '\n')
1156: dp[--len] = '\0';
1.1 schwarze 1157: req->p = mandoc_realloc(req->p,
1158: (req->psz + 1) * sizeof(char *));
1.20 schwarze 1159: if ( ! validate_urifrag(dp)) {
1160: fprintf(stderr, "%s/manpath.conf contains "
1161: "unsafe path \"%s\"\n", MAN_DIR, dp);
1162: pg_error_internal();
1163: exit(EXIT_FAILURE);
1164: }
1165: if (NULL != strchr(dp, '/')) {
1166: fprintf(stderr, "%s/manpath.conf contains "
1167: "path with slash \"%s\"\n", MAN_DIR, dp);
1168: pg_error_internal();
1169: exit(EXIT_FAILURE);
1170: }
1171: req->p[req->psz++] = dp;
1.54 schwarze 1172: dp = NULL;
1173: dpsz = 0;
1.14 schwarze 1174: }
1.54 schwarze 1175: free(dp);
1.14 schwarze 1176:
1177: if ( req->p == NULL ) {
1178: fprintf(stderr, "%s/manpath.conf is empty\n", MAN_DIR);
1179: pg_error_internal();
1180: exit(EXIT_FAILURE);
1.1 schwarze 1181: }
1182: }