Annotation of src/usr.bin/mandoc/cgi.c, Revision 1.57
1.57 ! schwarze 1: /* $OpenBSD: cgi.c,v 1.56 2016/03/17 21:22:43 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:
85: static const char *scriptname; /* CGI script name */
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"
1.6 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"
508: "<A HREF=\"%s/mandoc/man8/man.cgi.8\">man.cgi</A>\n"
509: "manual, and the\n"
510: "<A HREF=\"%s/mandoc/man1/apropos.1\">apropos</A>\n"
1.9 schwarze 511: "manual explains the query syntax.\n"
1.4 schwarze 512: "</P>\n",
513: scriptname, scriptname);
1.1 schwarze 514: resp_end_html();
515: }
516:
517: static void
1.12 schwarze 518: pg_noresult(const struct req *req, const char *msg)
1.1 schwarze 519: {
520: resp_begin_html(200, NULL);
521: resp_searchform(req);
522: puts("<P>");
523: puts(msg);
524: puts("</P>");
525: resp_end_html();
526: }
527:
528: static void
1.12 schwarze 529: pg_error_badrequest(const char *msg)
1.1 schwarze 530: {
531:
532: resp_begin_html(400, "Bad Request");
533: puts("<H1>Bad Request</H1>\n"
534: "<P>\n");
535: puts(msg);
536: printf("Try again from the\n"
537: "<A HREF=\"%s\">main page</A>.\n"
538: "</P>", scriptname);
539: resp_end_html();
540: }
541:
542: static void
1.12 schwarze 543: pg_error_internal(void)
1.1 schwarze 544: {
545: resp_begin_html(500, "Internal Server Error");
546: puts("<P>Internal Server Error</P>");
547: resp_end_html();
548: }
549:
550: static void
1.12 schwarze 551: pg_searchres(const struct req *req, struct manpage *r, size_t sz)
1.1 schwarze 552: {
1.21 schwarze 553: char *arch, *archend;
1.10 schwarze 554: size_t i, iuse, isec;
1.21 schwarze 555: int archprio, archpriouse;
1.10 schwarze 556: int prio, priouse;
557: char sec;
1.1 schwarze 558:
1.16 schwarze 559: for (i = 0; i < sz; i++) {
560: if (validate_filename(r[i].file))
561: continue;
562: fprintf(stderr, "invalid filename %s in %s database\n",
563: r[i].file, req->q.manpath);
564: pg_error_internal();
565: return;
566: }
567:
1.1 schwarze 568: if (1 == sz) {
569: /*
570: * If we have just one result, then jump there now
571: * without any delay.
572: */
1.2 tedu 573: printf("Status: 303 See Other\r\n");
1.55 bentley 574: printf("Location: http://%s%s/%s/%s",
1.18 schwarze 575: HTTP_HOST, scriptname, req->q.manpath, r[0].file);
1.2 tedu 576: printf("\r\n"
577: "Content-Type: text/html; charset=utf-8\r\n"
578: "\r\n");
1.1 schwarze 579: return;
580: }
581:
582: resp_begin_html(200, NULL);
583: resp_searchform(req);
584: puts("<DIV CLASS=\"results\">");
585: puts("<TABLE>");
586:
587: for (i = 0; i < sz; i++) {
588: printf("<TR>\n"
589: "<TD CLASS=\"title\">\n"
1.55 bentley 590: "<A HREF=\"%s/%s/%s",
1.1 schwarze 591: scriptname, req->q.manpath, r[i].file);
592: printf("\">");
593: html_print(r[i].names);
594: printf("</A>\n"
595: "</TD>\n"
596: "<TD CLASS=\"desc\">");
597: html_print(r[i].output);
598: puts("</TD>\n"
599: "</TR>");
600: }
601:
602: puts("</TABLE>\n"
603: "</DIV>");
1.10 schwarze 604:
605: /*
606: * In man(1) mode, show one of the pages
607: * even if more than one is found.
608: */
609:
610: if (req->q.equal) {
611: puts("<HR>");
612: iuse = 0;
613: priouse = 10;
1.21 schwarze 614: archpriouse = 3;
1.10 schwarze 615: for (i = 0; i < sz; i++) {
616: isec = strcspn(r[i].file, "123456789");
617: sec = r[i].file[isec];
618: if ('\0' == sec)
619: continue;
620: prio = sec_prios[sec - '1'];
1.21 schwarze 621: if (NULL == req->q.arch) {
622: archprio =
623: (NULL == (arch = strchr(
624: r[i].file + isec, '/'))) ? 3 :
625: (NULL == (archend = strchr(
626: arch + 1, '/'))) ? 0 :
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
650: catman(const struct req *req, const char *file)
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) {
1.10 schwarze 660: puts("<P>You specified an invalid manual file.</P>");
1.1 schwarze 661: return;
662: }
663:
664: puts("<DIV CLASS=\"catman\">\n"
665: "<PRE>");
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)
691: printf("</I>");
692: if (bold)
693: printf("</B>");
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)
704: printf("</B>");
705: if ( ! italic)
706: printf("<I>");
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)
728: printf("</I>");
729: if (bold)
730: printf("</B>");
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)
742: printf("</I>");
743: if (bold)
744: printf("</B>");
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)
754: printf("</I>");
755: if ( ! bold)
756: printf("<B>");
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)
769: printf("</I>");
770: if (bold)
771: printf("</B>");
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:
780: puts("</PRE>\n"
1.10 schwarze 781: "</DIV>");
1.1 schwarze 782:
783: fclose(f);
784: }
785:
786: static void
787: format(const struct req *req, const char *file)
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))) {
1.10 schwarze 797: puts("<P>You specified an invalid manual file.</P>");
1.1 schwarze 798: return;
799: }
800:
1.49 schwarze 801: mchars_alloc();
802: mp = mparse_alloc(MPARSE_SO, MANDOCLEVEL_BADARG, NULL, req->q.manpath);
1.42 schwarze 803: mparse_readfd(mp, fd, file);
1.1 schwarze 804: close(fd);
805:
1.45 schwarze 806: memset(&conf, 0, sizeof(conf));
807: conf.fragment = 1;
1.30 schwarze 808: usepath = strcmp(req->q.manpath, req->p[0]);
1.45 schwarze 809: mandoc_asprintf(&conf.man, "%s?query=%%N&sec=%%S%s%s%s%s",
1.30 schwarze 810: scriptname,
811: req->q.arch ? "&arch=" : "",
812: req->q.arch ? req->q.arch : "",
813: usepath ? "&manpath=" : "",
814: usepath ? req->q.manpath : "");
1.1 schwarze 815:
1.47 schwarze 816: mparse_result(mp, &man, NULL);
817: if (man == NULL) {
1.1 schwarze 818: fprintf(stderr, "fatal mandoc error: %s/%s\n",
819: req->q.manpath, file);
1.12 schwarze 820: pg_error_internal();
1.1 schwarze 821: mparse_free(mp);
1.49 schwarze 822: mchars_free();
1.1 schwarze 823: return;
824: }
825:
1.49 schwarze 826: vp = html_alloc(&conf);
1.1 schwarze 827:
1.50 schwarze 828: if (man->macroset == MACROSET_MDOC) {
829: mdoc_validate(man);
1.47 schwarze 830: html_mdoc(vp, man);
1.51 schwarze 831: } else {
832: man_validate(man);
1.1 schwarze 833: html_man(vp, man);
1.51 schwarze 834: }
1.1 schwarze 835:
836: html_free(vp);
837: mparse_free(mp);
1.49 schwarze 838: mchars_free();
1.45 schwarze 839: free(conf.man);
1.1 schwarze 840: }
841:
842: static void
1.10 schwarze 843: resp_show(const struct req *req, const char *file)
844: {
1.16 schwarze 845:
846: if ('.' == file[0] && '/' == file[1])
1.11 schwarze 847: file += 2;
1.10 schwarze 848:
849: if ('c' == *file)
850: catman(req, file);
851: else
852: format(req, file);
853: }
854:
855: static void
1.24 schwarze 856: pg_show(struct req *req, const char *fullpath)
1.1 schwarze 857: {
1.24 schwarze 858: char *manpath;
859: const char *file;
1.1 schwarze 860:
1.24 schwarze 861: if ((file = strchr(fullpath, '/')) == NULL) {
1.12 schwarze 862: pg_error_badrequest(
1.1 schwarze 863: "You did not specify a page to show.");
864: return;
1.43 schwarze 865: }
1.24 schwarze 866: manpath = mandoc_strndup(fullpath, file - fullpath);
867: file++;
1.1 schwarze 868:
1.24 schwarze 869: if ( ! validate_manpath(req, manpath)) {
1.17 schwarze 870: pg_error_badrequest(
871: "You specified an invalid manpath.");
1.24 schwarze 872: free(manpath);
1.17 schwarze 873: return;
874: }
875:
1.1 schwarze 876: /*
877: * Begin by chdir()ing into the manpath.
878: * This way we can pick up the database files, which are
879: * relative to the manpath root.
880: */
881:
1.24 schwarze 882: if (chdir(manpath) == -1) {
1.17 schwarze 883: fprintf(stderr, "chdir %s: %s\n",
1.24 schwarze 884: manpath, strerror(errno));
1.17 schwarze 885: pg_error_internal();
1.24 schwarze 886: free(manpath);
1.16 schwarze 887: return;
888: }
889:
1.24 schwarze 890: if (strcmp(manpath, "mandoc")) {
891: free(req->q.manpath);
892: req->q.manpath = manpath;
893: } else
894: free(manpath);
895:
896: if ( ! validate_filename(file)) {
1.16 schwarze 897: pg_error_badrequest(
898: "You specified an invalid manual file.");
1.1 schwarze 899: return;
900: }
1.19 schwarze 901:
1.10 schwarze 902: resp_begin_html(200, NULL);
903: resp_searchform(req);
1.24 schwarze 904: resp_show(req, file);
1.10 schwarze 905: resp_end_html();
1.1 schwarze 906: }
907:
908: static void
1.6 schwarze 909: pg_search(const struct req *req)
1.1 schwarze 910: {
911: struct mansearch search;
912: struct manpaths paths;
913: struct manpage *res;
1.36 schwarze 914: char **argv;
915: char *query, *rp, *wp;
1.1 schwarze 916: size_t ressz;
1.36 schwarze 917: int argc;
1.1 schwarze 918:
919: /*
920: * Begin by chdir()ing into the root of the manpath.
921: * This way we can pick up the database files, which are
922: * relative to the manpath root.
923: */
924:
925: if (-1 == (chdir(req->q.manpath))) {
1.17 schwarze 926: fprintf(stderr, "chdir %s: %s\n",
927: req->q.manpath, strerror(errno));
928: pg_error_internal();
1.1 schwarze 929: return;
930: }
931:
932: search.arch = req->q.arch;
933: search.sec = req->q.sec;
1.35 schwarze 934: search.outkey = "Nd";
935: search.argmode = req->q.equal ? ARG_NAME : ARG_EXPR;
1.40 schwarze 936: search.firstmatch = 1;
1.1 schwarze 937:
938: paths.sz = 1;
939: paths.paths = mandoc_malloc(sizeof(char *));
940: paths.paths[0] = mandoc_strdup(".");
941:
942: /*
1.36 schwarze 943: * Break apart at spaces with backslash-escaping.
1.1 schwarze 944: */
945:
1.36 schwarze 946: argc = 0;
947: argv = NULL;
948: rp = query = mandoc_strdup(req->q.query);
949: for (;;) {
950: while (isspace((unsigned char)*rp))
951: rp++;
952: if (*rp == '\0')
953: break;
954: argv = mandoc_reallocarray(argv, argc + 1, sizeof(char *));
955: argv[argc++] = wp = rp;
956: for (;;) {
957: if (isspace((unsigned char)*rp)) {
958: *wp = '\0';
959: rp++;
960: break;
961: }
962: if (rp[0] == '\\' && rp[1] != '\0')
963: rp++;
964: if (wp != rp)
965: *wp = *rp;
966: if (*rp == '\0')
967: break;
968: wp++;
969: rp++;
970: }
1.1 schwarze 971: }
972:
1.36 schwarze 973: if (0 == mansearch(&search, &paths, argc, argv, &res, &ressz))
1.12 schwarze 974: pg_noresult(req, "You entered an invalid query.");
1.1 schwarze 975: else if (0 == ressz)
1.12 schwarze 976: pg_noresult(req, "No results found.");
1.1 schwarze 977: else
1.12 schwarze 978: pg_searchres(req, res, ressz);
1.1 schwarze 979:
1.36 schwarze 980: free(query);
981: mansearch_free(res, ressz);
1.1 schwarze 982: free(paths.paths[0]);
983: free(paths.paths);
984: }
985:
986: int
987: main(void)
988: {
1.6 schwarze 989: struct req req;
1.33 schwarze 990: struct itimerval itimer;
1.6 schwarze 991: const char *path;
1.23 schwarze 992: const char *querystring;
1.1 schwarze 993: int i;
1.33 schwarze 994:
995: /* Poor man's ReDoS mitigation. */
996:
1.38 schwarze 997: itimer.it_value.tv_sec = 2;
1.33 schwarze 998: itimer.it_value.tv_usec = 0;
1.38 schwarze 999: itimer.it_interval.tv_sec = 2;
1.33 schwarze 1000: itimer.it_interval.tv_usec = 0;
1001: if (setitimer(ITIMER_VIRTUAL, &itimer, NULL) == -1) {
1002: fprintf(stderr, "setitimer: %s\n", strerror(errno));
1003: pg_error_internal();
1.48 schwarze 1004: return EXIT_FAILURE;
1.33 schwarze 1005: }
1.1 schwarze 1006:
1007: /* Scan our run-time environment. */
1008:
1009: if (NULL == (scriptname = getenv("SCRIPT_NAME")))
1010: scriptname = "";
1011:
1.20 schwarze 1012: if ( ! validate_urifrag(scriptname)) {
1013: fprintf(stderr, "unsafe SCRIPT_NAME \"%s\"\n",
1014: scriptname);
1015: pg_error_internal();
1.48 schwarze 1016: return EXIT_FAILURE;
1.20 schwarze 1017: }
1018:
1.1 schwarze 1019: /*
1.7 schwarze 1020: * First we change directory into the MAN_DIR so that
1.1 schwarze 1021: * subsequent scanning for manpath directories is rooted
1022: * relative to the same position.
1023: */
1024:
1.7 schwarze 1025: if (-1 == chdir(MAN_DIR)) {
1.1 schwarze 1026: fprintf(stderr, "MAN_DIR: %s: %s\n",
1.7 schwarze 1027: MAN_DIR, strerror(errno));
1.12 schwarze 1028: pg_error_internal();
1.48 schwarze 1029: return EXIT_FAILURE;
1.43 schwarze 1030: }
1.1 schwarze 1031:
1032: memset(&req, 0, sizeof(struct req));
1.57 ! schwarze 1033: req.q.equal = 1;
1.1 schwarze 1034: pathgen(&req);
1035:
1.56 schwarze 1036: /* Parse the path info and the query string. */
1.1 schwarze 1037:
1.56 schwarze 1038: if ((path = getenv("PATH_INFO")) == NULL)
1039: path = "";
1040: else if (*path == '/')
1041: path++;
1042:
1043: if (*path != '\0' && access(path, F_OK) == -1) {
1044: path_parse(&req, path);
1045: path = "";
1046: } else if ((querystring = getenv("QUERY_STRING")) != NULL)
1.1 schwarze 1047: http_parse(&req, querystring);
1.17 schwarze 1048:
1.56 schwarze 1049: /* Validate parsed data and add defaults. */
1050:
1.41 schwarze 1051: if (req.q.manpath == NULL)
1052: req.q.manpath = mandoc_strdup(req.p[0]);
1053: else if ( ! validate_manpath(&req, req.q.manpath)) {
1.17 schwarze 1054: pg_error_badrequest(
1055: "You specified an invalid manpath.");
1.48 schwarze 1056: return EXIT_FAILURE;
1.17 schwarze 1057: }
1.1 schwarze 1058:
1.20 schwarze 1059: if ( ! (NULL == req.q.arch || validate_urifrag(req.q.arch))) {
1060: pg_error_badrequest(
1061: "You specified an invalid architecture.");
1.48 schwarze 1062: return EXIT_FAILURE;
1.20 schwarze 1063: }
1064:
1.6 schwarze 1065: /* Dispatch to the three different pages. */
1.1 schwarze 1066:
1.6 schwarze 1067: if ('\0' != *path)
1068: pg_show(&req, path);
1.25 schwarze 1069: else if (NULL != req.q.query)
1.6 schwarze 1070: pg_search(&req);
1071: else
1.12 schwarze 1072: pg_index(&req);
1.1 schwarze 1073:
1.23 schwarze 1074: free(req.q.manpath);
1075: free(req.q.arch);
1076: free(req.q.sec);
1.25 schwarze 1077: free(req.q.query);
1.1 schwarze 1078: for (i = 0; i < (int)req.psz; i++)
1079: free(req.p[i]);
1080: free(req.p);
1.48 schwarze 1081: return EXIT_SUCCESS;
1.56 schwarze 1082: }
1083:
1084: /*
1085: * If PATH_INFO is not a file name, translate it to a query.
1086: */
1087: static void
1088: path_parse(struct req *req, const char *path)
1089: {
1090: int dir_done;
1091:
1092: req->q.equal = 1;
1093: req->q.manpath = mandoc_strdup(path);
1094:
1095: /* Mandatory manual page name. */
1096: if ((req->q.query = strrchr(req->q.manpath, '/')) == NULL) {
1097: req->q.query = req->q.manpath;
1098: req->q.manpath = NULL;
1099: } else
1100: *req->q.query++ = '\0';
1101:
1102: /* Optional trailing section. */
1103: if ((req->q.sec = strrchr(req->q.query, '.')) != NULL) {
1104: if(isdigit((unsigned char)req->q.sec[1])) {
1105: *req->q.sec++ = '\0';
1106: req->q.sec = mandoc_strdup(req->q.sec);
1107: } else
1108: req->q.sec = NULL;
1109: }
1110:
1111: /* Handle the case of name[.section] only. */
1112: if (req->q.manpath == NULL) {
1113: req->q.arch = NULL;
1114: return;
1115: }
1116: req->q.query = mandoc_strdup(req->q.query);
1117:
1118: /* Optional architecture. */
1119: dir_done = 0;
1120: for (;;) {
1121: if ((req->q.arch = strrchr(req->q.manpath, '/')) == NULL)
1122: break;
1123: *req->q.arch++ = '\0';
1124: if (dir_done || strncmp(req->q.arch, "man", 3)) {
1125: req->q.arch = mandoc_strdup(req->q.arch);
1126: break;
1127: }
1128:
1129: /* Optional directory name. */
1130: req->q.arch += 3;
1131: if (*req->q.arch != '\0') {
1132: free(req->q.sec);
1133: req->q.sec = mandoc_strdup(req->q.arch);
1134: }
1135: dir_done = 1;
1136: }
1.1 schwarze 1137: }
1138:
1139: /*
1140: * Scan for indexable paths.
1141: */
1142: static void
1143: pathgen(struct req *req)
1144: {
1145: FILE *fp;
1146: char *dp;
1147: size_t dpsz;
1.54 schwarze 1148: ssize_t len;
1.1 schwarze 1149:
1.14 schwarze 1150: if (NULL == (fp = fopen("manpath.conf", "r"))) {
1151: fprintf(stderr, "%s/manpath.conf: %s\n",
1152: MAN_DIR, strerror(errno));
1153: pg_error_internal();
1154: exit(EXIT_FAILURE);
1155: }
1.1 schwarze 1156:
1.54 schwarze 1157: dp = NULL;
1158: dpsz = 0;
1159:
1160: while ((len = getline(&dp, &dpsz, fp)) != -1) {
1161: if (dp[len - 1] == '\n')
1162: dp[--len] = '\0';
1.1 schwarze 1163: req->p = mandoc_realloc(req->p,
1164: (req->psz + 1) * sizeof(char *));
1.20 schwarze 1165: if ( ! validate_urifrag(dp)) {
1166: fprintf(stderr, "%s/manpath.conf contains "
1167: "unsafe path \"%s\"\n", MAN_DIR, dp);
1168: pg_error_internal();
1169: exit(EXIT_FAILURE);
1170: }
1171: if (NULL != strchr(dp, '/')) {
1172: fprintf(stderr, "%s/manpath.conf contains "
1173: "path with slash \"%s\"\n", MAN_DIR, dp);
1174: pg_error_internal();
1175: exit(EXIT_FAILURE);
1176: }
1177: req->p[req->psz++] = dp;
1.54 schwarze 1178: dp = NULL;
1179: dpsz = 0;
1.14 schwarze 1180: }
1.54 schwarze 1181: free(dp);
1.14 schwarze 1182:
1183: if ( req->p == NULL ) {
1184: fprintf(stderr, "%s/manpath.conf is empty\n", MAN_DIR);
1185: pg_error_internal();
1186: exit(EXIT_FAILURE);
1.1 schwarze 1187: }
1188: }