Annotation of mandoc/cgi.c, Revision 1.101
1.101 ! schwarze 1: /* $Id: cgi.c,v 1.100 2014/10/28 17:36:19 schwarze Exp $ */
1.6 kristaps 2: /*
1.42 kristaps 3: * Copyright (c) 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
1.52 schwarze 4: * Copyright (c) 2014 Ingo Schwarze <schwarze@usta.de>
1.6 kristaps 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: *
10: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
11: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
12: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
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: */
18: #include "config.h"
1.93 schwarze 19:
20: #include <sys/types.h>
1.95 schwarze 21: #include <sys/time.h>
1.6 kristaps 22:
23: #include <ctype.h>
1.58 schwarze 24: #include <errno.h>
1.1 kristaps 25: #include <fcntl.h>
1.6 kristaps 26: #include <limits.h>
1.92 schwarze 27: #include <stdint.h>
1.1 kristaps 28: #include <stdio.h>
29: #include <stdlib.h>
30: #include <string.h>
1.6 kristaps 31: #include <unistd.h>
1.1 kristaps 32:
1.4 schwarze 33: #include "mandoc.h"
1.50 schwarze 34: #include "mandoc_aux.h"
1.8 kristaps 35: #include "main.h"
1.6 kristaps 36: #include "manpath.h"
1.52 schwarze 37: #include "mansearch.h"
1.67 schwarze 38: #include "cgi.h"
1.1 kristaps 39:
1.20 kristaps 40: /*
41: * A query as passed to the search function.
42: */
43: struct query {
1.83 schwarze 44: char *manpath; /* desired manual directory */
45: char *arch; /* architecture */
46: char *sec; /* manual section */
1.85 schwarze 47: char *query; /* unparsed query expression */
1.65 schwarze 48: int equal; /* match whole names, not substrings */
1.20 kristaps 49: };
50:
1.1 kristaps 51: struct req {
1.58 schwarze 52: struct query q;
53: char **p; /* array of available manpaths */
54: size_t psz; /* number of available manpaths */
1.1 kristaps 55: };
56:
1.32 kristaps 57: static void catman(const struct req *, const char *);
58: static void format(const struct req *, const char *);
1.6 kristaps 59: static void html_print(const char *);
1.10 kristaps 60: static void html_putchar(char);
1.24 kristaps 61: static int http_decode(char *);
1.83 schwarze 62: static void http_parse(struct req *, const char *);
1.36 kristaps 63: static void http_print(const char *);
64: static void http_putchar(char);
1.91 schwarze 65: static void http_printquery(const struct req *, const char *);
1.54 schwarze 66: static void pathgen(struct req *);
1.72 schwarze 67: static void pg_error_badrequest(const char *);
68: static void pg_error_internal(void);
69: static void pg_index(const struct req *);
70: static void pg_noresult(const struct req *, const char *);
1.66 schwarze 71: static void pg_search(const struct req *);
1.72 schwarze 72: static void pg_searchres(const struct req *,
73: struct manpage *, size_t);
1.79 schwarze 74: static void pg_show(struct req *, const char *);
1.6 kristaps 75: static void resp_begin_html(int, const char *);
76: static void resp_begin_http(int, const char *);
77: static void resp_end_html(void);
78: static void resp_searchform(const struct req *);
1.70 schwarze 79: static void resp_show(const struct req *, const char *);
1.85 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 *);
1.6 kristaps 84:
1.58 schwarze 85: static const char *scriptname; /* CGI script name */
1.1 kristaps 86:
1.70 schwarze 87: static const int sec_prios[] = {1, 4, 5, 8, 6, 3, 7, 2, 9};
1.68 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",
1.96 schwarze 95: "3 - Library Functions",
96: "3p - Perl Library",
97: "4 - Device Drivers",
1.68 schwarze 98: "5 - File Formats",
99: "6 - Games",
1.96 schwarze 100: "7 - Miscellaneous Information",
101: "8 - System Manager\'s Manual",
102: "9 - Kernel Developer\'s Manual"
1.68 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:
1.6 kristaps 121: /*
1.20 kristaps 122: * Print a character, escaping HTML along the way.
123: * This will pass non-ASCII straight to output: be warned!
124: */
1.10 kristaps 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: }
1.57 schwarze 147:
1.36 kristaps 148: static void
1.91 schwarze 149: http_printquery(const struct req *req, const char *sep)
1.36 kristaps 150: {
151:
1.88 schwarze 152: if (NULL != req->q.query) {
153: printf("query=");
154: http_print(req->q.query);
1.53 schwarze 155: }
1.88 schwarze 156: if (0 == req->q.equal)
1.91 schwarze 157: printf("%sapropos=1", sep);
1.53 schwarze 158: if (NULL != req->q.sec) {
1.91 schwarze 159: printf("%ssec=", sep);
1.53 schwarze 160: http_print(req->q.sec);
161: }
162: if (NULL != req->q.arch) {
1.91 schwarze 163: printf("%sarch=", sep);
1.53 schwarze 164: http_print(req->q.arch);
165: }
1.88 schwarze 166: if (NULL != req->q.manpath &&
167: strcmp(req->q.manpath, req->p[0])) {
1.91 schwarze 168: printf("%smanpath=", sep);
1.88 schwarze 169: http_print(req->q.manpath);
1.53 schwarze 170: }
1.36 kristaps 171: }
172:
173: static void
174: http_print(const char *p)
175: {
176:
177: if (NULL == p)
178: return;
179: while ('\0' != *p)
180: http_putchar(*p++);
181: }
1.10 kristaps 182:
1.6 kristaps 183: /*
1.20 kristaps 184: * Call through to html_putchar().
185: * Accepts NULL strings.
1.6 kristaps 186: */
1.1 kristaps 187: static void
1.6 kristaps 188: html_print(const char *p)
1.1 kristaps 189: {
1.6 kristaps 190:
191: if (NULL == p)
192: return;
1.1 kristaps 193: while ('\0' != *p)
1.10 kristaps 194: html_putchar(*p++);
1.1 kristaps 195: }
196:
197: /*
1.83 schwarze 198: * Transfer the responsibility for the allocated string *val
199: * to the query structure.
1.1 kristaps 200: */
201: static void
1.83 schwarze 202: set_query_attr(char **attr, char **val)
1.1 kristaps 203: {
204:
1.83 schwarze 205: free(*attr);
206: if (**val == '\0') {
207: *attr = NULL;
208: free(*val);
209: } else
210: *attr = *val;
211: *val = NULL;
212: }
213:
214: /*
215: * Parse the QUERY_STRING for key-value pairs
216: * and store the values into the query structure.
217: */
218: static void
219: http_parse(struct req *req, const char *qs)
220: {
221: char *key, *val;
222: size_t keysz, valsz;
223:
224: req->q.manpath = NULL;
225: req->q.arch = NULL;
226: req->q.sec = NULL;
1.85 schwarze 227: req->q.query = NULL;
1.83 schwarze 228: req->q.equal = 1;
229:
230: key = val = NULL;
231: while (*qs != '\0') {
1.24 kristaps 232:
1.83 schwarze 233: /* Parse one key. */
234:
235: keysz = strcspn(qs, "=;&");
236: key = mandoc_strndup(qs, keysz);
237: qs += keysz;
238: if (*qs != '=')
239: goto next;
240:
241: /* Parse one value. */
242:
243: valsz = strcspn(++qs, ";&");
244: val = mandoc_strndup(qs, valsz);
245: qs += valsz;
246:
247: /* Decode and catch encoding errors. */
1.1 kristaps 248:
1.83 schwarze 249: if ( ! (http_decode(key) && http_decode(val)))
250: goto next;
1.1 kristaps 251:
1.83 schwarze 252: /* Handle key-value pairs. */
1.1 kristaps 253:
1.83 schwarze 254: if ( ! strcmp(key, "query"))
1.85 schwarze 255: set_query_attr(&req->q.query, &val);
1.1 kristaps 256:
1.83 schwarze 257: else if ( ! strcmp(key, "apropos"))
258: req->q.equal = !strcmp(val, "0");
259:
260: else if ( ! strcmp(key, "manpath")) {
1.73 schwarze 261: #ifdef COMPAT_OLDURI
1.83 schwarze 262: if ( ! strncmp(val, "OpenBSD ", 8)) {
1.73 schwarze 263: val[7] = '-';
264: if ('C' == val[8])
265: val[8] = 'c';
266: }
267: #endif
1.83 schwarze 268: set_query_attr(&req->q.manpath, &val);
269: }
270:
271: else if ( ! (strcmp(key, "sec")
1.73 schwarze 272: #ifdef COMPAT_OLDURI
1.83 schwarze 273: && strcmp(key, "sektion")
1.73 schwarze 274: #endif
1.83 schwarze 275: )) {
276: if ( ! strcmp(val, "0"))
277: *val = '\0';
278: set_query_attr(&req->q.sec, &val);
1.65 schwarze 279: }
1.83 schwarze 280:
281: else if ( ! strcmp(key, "arch")) {
282: if ( ! strcmp(val, "default"))
283: *val = '\0';
284: set_query_attr(&req->q.arch, &val);
285: }
286:
287: /*
288: * The key must be freed in any case.
289: * The val may have been handed over to the query
290: * structure, in which case it is now NULL.
291: */
292: next:
293: free(key);
294: key = NULL;
295: free(val);
296: val = NULL;
297:
298: if (*qs != '\0')
299: qs++;
1.24 kristaps 300: }
1.83 schwarze 301:
302: /* Fall back to the default manpath. */
303:
304: if (req->q.manpath == NULL)
305: req->q.manpath = mandoc_strdup(req->p[0]);
1.1 kristaps 306: }
307:
1.36 kristaps 308: static void
309: http_putchar(char c)
310: {
311:
312: if (isalnum((unsigned char)c)) {
313: putchar((unsigned char)c);
314: return;
315: } else if (' ' == c) {
316: putchar('+');
317: return;
318: }
319: printf("%%%.2x", c);
320: }
321:
1.1 kristaps 322: /*
1.6 kristaps 323: * HTTP-decode a string. The standard explanation is that this turns
324: * "%4e+foo" into "n foo" in the regular way. This is done in-place
325: * over the allocated string.
1.1 kristaps 326: */
327: static int
1.24 kristaps 328: http_decode(char *p)
1.1 kristaps 329: {
330: char hex[3];
1.63 schwarze 331: char *q;
1.1 kristaps 332: int c;
333:
334: hex[2] = '\0';
335:
1.63 schwarze 336: q = p;
337: for ( ; '\0' != *p; p++, q++) {
1.1 kristaps 338: if ('%' == *p) {
339: if ('\0' == (hex[0] = *(p + 1)))
340: return(0);
341: if ('\0' == (hex[1] = *(p + 2)))
342: return(0);
343: if (1 != sscanf(hex, "%x", &c))
344: return(0);
345: if ('\0' == c)
346: return(0);
347:
1.63 schwarze 348: *q = (char)c;
349: p += 2;
1.1 kristaps 350: } else
1.63 schwarze 351: *q = '+' == *p ? ' ' : *p;
1.1 kristaps 352: }
353:
1.63 schwarze 354: *q = '\0';
1.1 kristaps 355: return(1);
356: }
357:
1.6 kristaps 358: static void
359: resp_begin_http(int code, const char *msg)
360: {
361:
362: if (200 != code)
1.62 schwarze 363: printf("Status: %d %s\r\n", code, msg);
1.6 kristaps 364:
1.62 schwarze 365: printf("Content-Type: text/html; charset=utf-8\r\n"
366: "Cache-Control: no-cache\r\n"
367: "Pragma: no-cache\r\n"
368: "\r\n");
1.6 kristaps 369:
370: fflush(stdout);
371: }
372:
373: static void
374: resp_begin_html(int code, const char *msg)
375: {
376:
377: resp_begin_http(code, msg);
378:
1.98 kristaps 379: printf("<!DOCTYPE html>\n"
1.29 kristaps 380: "<HTML>\n"
381: "<HEAD>\n"
1.98 kristaps 382: "<META CHARSET=\"UTF-8\" />\n"
1.32 kristaps 383: "<LINK REL=\"stylesheet\" HREF=\"%s/man-cgi.css\""
384: " TYPE=\"text/css\" media=\"all\">\n"
385: "<LINK REL=\"stylesheet\" HREF=\"%s/man.css\""
1.29 kristaps 386: " TYPE=\"text/css\" media=\"all\">\n"
1.67 schwarze 387: "<TITLE>%s</TITLE>\n"
1.29 kristaps 388: "</HEAD>\n"
389: "<BODY>\n"
1.58 schwarze 390: "<!-- Begin page content. //-->\n",
1.67 schwarze 391: CSS_DIR, CSS_DIR, CUSTOMIZE_TITLE);
1.6 kristaps 392: }
393:
394: static void
395: resp_end_html(void)
396: {
397:
1.20 kristaps 398: puts("</BODY>\n"
399: "</HTML>");
1.6 kristaps 400: }
401:
402: static void
403: resp_searchform(const struct req *req)
404: {
1.27 kristaps 405: int i;
1.13 kristaps 406:
1.67 schwarze 407: puts(CUSTOMIZE_BEGIN);
1.6 kristaps 408: puts("<!-- Begin search form. //-->");
1.32 kristaps 409: printf("<DIV ID=\"mancgi\">\n"
1.66 schwarze 410: "<FORM ACTION=\"%s\" METHOD=\"get\">\n"
1.29 kristaps 411: "<FIELDSET>\n"
1.68 schwarze 412: "<LEGEND>Manual Page Search Parameters</LEGEND>\n",
1.58 schwarze 413: scriptname);
1.68 schwarze 414:
415: /* Write query input box. */
416:
417: printf( "<TABLE><TR><TD>\n"
418: "<INPUT TYPE=\"text\" NAME=\"query\" VALUE=\"");
1.85 schwarze 419: if (NULL != req->q.query)
420: html_print(req->q.query);
1.68 schwarze 421: puts("\" SIZE=\"40\">");
422:
423: /* Write submission and reset buttons. */
424:
425: printf( "<INPUT TYPE=\"submit\" VALUE=\"Submit\">\n"
426: "<INPUT TYPE=\"reset\" VALUE=\"Reset\">\n");
427:
428: /* Write show radio button */
429:
430: printf( "</TD><TD>\n"
431: "<INPUT TYPE=\"radio\" ");
1.65 schwarze 432: if (req->q.equal)
1.86 schwarze 433: printf("CHECKED=\"checked\" ");
1.68 schwarze 434: printf( "NAME=\"apropos\" ID=\"show\" VALUE=\"0\">\n"
435: "<LABEL FOR=\"show\">Show named manual page</LABEL>\n");
436:
437: /* Write section selector. */
438:
1.86 schwarze 439: puts( "</TD></TR><TR><TD>\n"
1.68 schwarze 440: "<SELECT NAME=\"sec\">");
441: for (i = 0; i < sec_MAX; i++) {
442: printf("<OPTION VALUE=\"%s\"", sec_numbers[i]);
443: if (NULL != req->q.sec &&
444: 0 == strcmp(sec_numbers[i], req->q.sec))
1.86 schwarze 445: printf(" SELECTED=\"selected\"");
1.68 schwarze 446: printf(">%s</OPTION>\n", sec_names[i]);
447: }
448: puts("</SELECT>");
449:
450: /* Write architecture selector. */
451:
1.81 schwarze 452: printf( "<SELECT NAME=\"arch\">\n"
453: "<OPTION VALUE=\"default\"");
454: if (NULL == req->q.arch)
1.86 schwarze 455: printf(" SELECTED=\"selected\"");
1.81 schwarze 456: puts(">All Architectures</OPTION>");
1.68 schwarze 457: for (i = 0; i < arch_MAX; i++) {
458: printf("<OPTION VALUE=\"%s\"", arch_names[i]);
459: if (NULL != req->q.arch &&
460: 0 == strcmp(arch_names[i], req->q.arch))
1.86 schwarze 461: printf(" SELECTED=\"selected\"");
1.68 schwarze 462: printf(">%s</OPTION>\n", arch_names[i]);
463: }
464: puts("</SELECT>");
465:
466: /* Write manpath selector. */
467:
1.27 kristaps 468: if (req->psz > 1) {
1.68 schwarze 469: puts("<SELECT NAME=\"manpath\">");
1.27 kristaps 470: for (i = 0; i < (int)req->psz; i++) {
1.52 schwarze 471: printf("<OPTION ");
1.58 schwarze 472: if (NULL == req->q.manpath ? 0 == i :
473: 0 == strcmp(req->q.manpath, req->p[i]))
1.86 schwarze 474: printf("SELECTED=\"selected\" ");
1.52 schwarze 475: printf("VALUE=\"");
476: html_print(req->p[i]);
1.27 kristaps 477: printf("\">");
1.52 schwarze 478: html_print(req->p[i]);
1.27 kristaps 479: puts("</OPTION>");
480: }
481: puts("</SELECT>");
482: }
1.68 schwarze 483:
484: /* Write search radio button */
485:
486: printf( "</TD><TD>\n"
487: "<INPUT TYPE=\"radio\" ");
488: if (0 == req->q.equal)
1.86 schwarze 489: printf("CHECKED=\"checked\" ");
1.68 schwarze 490: printf( "NAME=\"apropos\" ID=\"search\" VALUE=\"1\">\n"
491: "<LABEL FOR=\"search\">Search with apropos query</LABEL>\n");
492:
493: puts("</TD></TR></TABLE>\n"
1.12 kristaps 494: "</FIELDSET>\n"
1.32 kristaps 495: "</FORM>\n"
496: "</DIV>");
1.20 kristaps 497: puts("<!-- End search form. //-->");
1.6 kristaps 498: }
499:
1.76 schwarze 500: static int
1.80 schwarze 501: validate_urifrag(const char *frag)
502: {
503:
504: while ('\0' != *frag) {
505: if ( ! (isalnum((unsigned char)*frag) ||
506: '-' == *frag || '.' == *frag ||
507: '/' == *frag || '_' == *frag))
508: return(0);
509: frag++;
510: }
511: return(1);
512: }
513:
514: static int
1.77 schwarze 515: validate_manpath(const struct req *req, const char* manpath)
516: {
517: size_t i;
518:
519: if ( ! strcmp(manpath, "mandoc"))
520: return(1);
521:
522: for (i = 0; i < req->psz; i++)
523: if ( ! strcmp(manpath, req->p[i]))
524: return(1);
525:
526: return(0);
527: }
528:
529: static int
1.76 schwarze 530: validate_filename(const char *file)
531: {
532:
533: if ('.' == file[0] && '/' == file[1])
534: file += 2;
535:
536: return ( ! (strstr(file, "../") || strstr(file, "/..") ||
537: (strncmp(file, "man", 3) && strncmp(file, "cat", 3))));
538: }
539:
1.6 kristaps 540: static void
1.72 schwarze 541: pg_index(const struct req *req)
1.6 kristaps 542: {
543:
544: resp_begin_html(200, NULL);
545: resp_searchform(req);
1.64 schwarze 546: printf("<P>\n"
1.86 schwarze 547: "This web interface is documented in the\n"
548: "<A HREF=\"%s/mandoc/man8/man.cgi.8\">man.cgi</A>\n"
549: "manual, and the\n"
550: "<A HREF=\"%s/mandoc/man1/apropos.1\">apropos</A>\n"
1.69 schwarze 551: "manual explains the query syntax.\n"
1.64 schwarze 552: "</P>\n",
553: scriptname, scriptname);
1.6 kristaps 554: resp_end_html();
555: }
556:
557: static void
1.72 schwarze 558: pg_noresult(const struct req *req, const char *msg)
1.59 schwarze 559: {
560: resp_begin_html(200, NULL);
561: resp_searchform(req);
562: puts("<P>");
563: puts(msg);
564: puts("</P>");
565: resp_end_html();
566: }
567:
568: static void
1.72 schwarze 569: pg_error_badrequest(const char *msg)
1.9 kristaps 570: {
571:
1.59 schwarze 572: resp_begin_html(400, "Bad Request");
573: puts("<H1>Bad Request</H1>\n"
574: "<P>\n");
575: puts(msg);
576: printf("Try again from the\n"
577: "<A HREF=\"%s\">main page</A>.\n"
1.58 schwarze 578: "</P>", scriptname);
1.9 kristaps 579: resp_end_html();
580: }
581:
582: static void
1.72 schwarze 583: pg_error_internal(void)
1.7 kristaps 584: {
585: resp_begin_html(500, "Internal Server Error");
1.58 schwarze 586: puts("<P>Internal Server Error</P>");
1.7 kristaps 587: resp_end_html();
588: }
589:
590: static void
1.72 schwarze 591: pg_searchres(const struct req *req, struct manpage *r, size_t sz)
1.1 kristaps 592: {
1.81 schwarze 593: char *arch, *archend;
1.70 schwarze 594: size_t i, iuse, isec;
1.81 schwarze 595: int archprio, archpriouse;
1.70 schwarze 596: int prio, priouse;
597: char sec;
1.19 kristaps 598:
1.76 schwarze 599: for (i = 0; i < sz; i++) {
600: if (validate_filename(r[i].file))
601: continue;
602: fprintf(stderr, "invalid filename %s in %s database\n",
603: r[i].file, req->q.manpath);
604: pg_error_internal();
605: return;
606: }
607:
1.52 schwarze 608: if (1 == sz) {
1.6 kristaps 609: /*
610: * If we have just one result, then jump there now
611: * without any delay.
612: */
1.62 schwarze 613: printf("Status: 303 See Other\r\n");
1.78 schwarze 614: printf("Location: http://%s%s/%s/%s?",
615: HTTP_HOST, scriptname, req->q.manpath, r[0].file);
1.91 schwarze 616: http_printquery(req, "&");
1.62 schwarze 617: printf("\r\n"
618: "Content-Type: text/html; charset=utf-8\r\n"
619: "\r\n");
1.6 kristaps 620: return;
621: }
622:
1.12 kristaps 623: resp_begin_html(200, NULL);
1.19 kristaps 624: resp_searchform(req);
1.33 kristaps 625: puts("<DIV CLASS=\"results\">");
626: puts("<TABLE>");
1.1 kristaps 627:
1.41 kristaps 628: for (i = 0; i < sz; i++) {
1.20 kristaps 629: printf("<TR>\n"
630: "<TD CLASS=\"title\">\n"
1.66 schwarze 631: "<A HREF=\"%s/%s/%s?",
1.58 schwarze 632: scriptname, req->q.manpath, r[i].file);
1.91 schwarze 633: http_printquery(req, "&");
1.36 kristaps 634: printf("\">");
1.52 schwarze 635: html_print(r[i].names);
636: printf("</A>\n"
1.20 kristaps 637: "</TD>\n"
638: "<TD CLASS=\"desc\">");
1.52 schwarze 639: html_print(r[i].output);
1.20 kristaps 640: puts("</TD>\n"
641: "</TR>");
1.1 kristaps 642: }
1.16 kristaps 643:
1.33 kristaps 644: puts("</TABLE>\n"
645: "</DIV>");
1.70 schwarze 646:
647: /*
648: * In man(1) mode, show one of the pages
649: * even if more than one is found.
650: */
651:
652: if (req->q.equal) {
653: puts("<HR>");
654: iuse = 0;
655: priouse = 10;
1.81 schwarze 656: archpriouse = 3;
1.70 schwarze 657: for (i = 0; i < sz; i++) {
658: isec = strcspn(r[i].file, "123456789");
659: sec = r[i].file[isec];
660: if ('\0' == sec)
661: continue;
662: prio = sec_prios[sec - '1'];
1.81 schwarze 663: if (NULL == req->q.arch) {
664: archprio =
665: (NULL == (arch = strchr(
666: r[i].file + isec, '/'))) ? 3 :
667: (NULL == (archend = strchr(
668: arch + 1, '/'))) ? 0 :
669: strncmp(arch, "amd64/",
670: archend - arch) ? 2 : 1;
671: if (archprio < archpriouse) {
672: archpriouse = archprio;
673: priouse = prio;
674: iuse = i;
675: continue;
676: }
677: if (archprio > archpriouse)
678: continue;
679: }
1.70 schwarze 680: if (prio >= priouse)
681: continue;
682: priouse = prio;
683: iuse = i;
684: }
685: resp_show(req, r[iuse].file);
686: }
687:
1.6 kristaps 688: resp_end_html();
689: }
690:
1.1 kristaps 691: static void
1.32 kristaps 692: catman(const struct req *req, const char *file)
1.9 kristaps 693: {
1.10 kristaps 694: FILE *f;
695: size_t len;
696: int i;
697: char *p;
698: int italic, bold;
1.9 kristaps 699:
1.10 kristaps 700: if (NULL == (f = fopen(file, "r"))) {
1.70 schwarze 701: puts("<P>You specified an invalid manual file.</P>");
1.9 kristaps 702: return;
703: }
704:
1.32 kristaps 705: puts("<DIV CLASS=\"catman\">\n"
706: "<PRE>");
1.10 kristaps 707:
708: while (NULL != (p = fgetln(f, &len))) {
709: bold = italic = 0;
710: for (i = 0; i < (int)len - 1; i++) {
711: /*
712: * This means that the catpage is out of state.
713: * Ignore it and keep going (although the
714: * catpage is bogus).
715: */
716:
717: if ('\b' == p[i] || '\n' == p[i])
718: continue;
719:
720: /*
721: * Print a regular character.
722: * Close out any bold/italic scopes.
723: * If we're in back-space mode, make sure we'll
724: * have something to enter when we backspace.
725: */
726:
727: if ('\b' != p[i + 1]) {
728: if (italic)
729: printf("</I>");
730: if (bold)
731: printf("</B>");
732: italic = bold = 0;
733: html_putchar(p[i]);
734: continue;
735: } else if (i + 2 >= (int)len)
736: continue;
737:
738: /* Italic mode. */
739:
740: if ('_' == p[i]) {
741: if (bold)
742: printf("</B>");
743: if ( ! italic)
744: printf("<I>");
745: bold = 0;
746: italic = 1;
747: i += 2;
748: html_putchar(p[i]);
749: continue;
750: }
751:
752: /*
753: * Handle funny behaviour troff-isms.
754: * These grok'd from the original man2html.c.
755: */
756:
757: if (('+' == p[i] && 'o' == p[i + 2]) ||
758: ('o' == p[i] && '+' == p[i + 2]) ||
759: ('|' == p[i] && '=' == p[i + 2]) ||
760: ('=' == p[i] && '|' == p[i + 2]) ||
761: ('*' == p[i] && '=' == p[i + 2]) ||
762: ('=' == p[i] && '*' == p[i + 2]) ||
763: ('*' == p[i] && '|' == p[i + 2]) ||
764: ('|' == p[i] && '*' == p[i + 2])) {
765: if (italic)
766: printf("</I>");
767: if (bold)
768: printf("</B>");
769: italic = bold = 0;
770: putchar('*');
771: i += 2;
772: continue;
773: } else if (('|' == p[i] && '-' == p[i + 2]) ||
774: ('-' == p[i] && '|' == p[i + 1]) ||
775: ('+' == p[i] && '-' == p[i + 1]) ||
776: ('-' == p[i] && '+' == p[i + 1]) ||
777: ('+' == p[i] && '|' == p[i + 1]) ||
778: ('|' == p[i] && '+' == p[i + 1])) {
779: if (italic)
780: printf("</I>");
781: if (bold)
782: printf("</B>");
783: italic = bold = 0;
784: putchar('+');
785: i += 2;
786: continue;
787: }
788:
789: /* Bold mode. */
790:
791: if (italic)
792: printf("</I>");
793: if ( ! bold)
794: printf("<B>");
795: bold = 1;
796: italic = 0;
797: i += 2;
798: html_putchar(p[i]);
799: }
800:
801: /*
802: * Clean up the last character.
803: * We can get to a newline; don't print that.
804: */
1.9 kristaps 805:
1.10 kristaps 806: if (italic)
807: printf("</I>");
808: if (bold)
809: printf("</B>");
1.9 kristaps 810:
1.10 kristaps 811: if (i == (int)len - 1 && '\n' != p[i])
812: html_putchar(p[i]);
1.9 kristaps 813:
1.10 kristaps 814: putchar('\n');
815: }
816:
817: puts("</PRE>\n"
1.70 schwarze 818: "</DIV>");
1.10 kristaps 819:
820: fclose(f);
1.9 kristaps 821: }
822:
823: static void
1.32 kristaps 824: format(const struct req *req, const char *file)
1.7 kristaps 825: {
1.8 kristaps 826: struct mparse *mp;
1.100 schwarze 827: struct mchars *mchars;
1.8 kristaps 828: struct mdoc *mdoc;
829: struct man *man;
830: void *vp;
1.90 schwarze 831: char *opts;
1.8 kristaps 832: enum mandoclevel rc;
1.90 schwarze 833: int fd;
834: int usepath;
1.7 kristaps 835:
1.8 kristaps 836: if (-1 == (fd = open(file, O_RDONLY, 0))) {
1.70 schwarze 837: puts("<P>You specified an invalid manual file.</P>");
1.7 kristaps 838: return;
839: }
840:
1.100 schwarze 841: mchars = mchars_alloc();
1.56 schwarze 842: mp = mparse_alloc(MPARSE_SO, MANDOCLEVEL_FATAL, NULL,
1.100 schwarze 843: mchars, req->q.manpath);
1.8 kristaps 844: rc = mparse_readfd(mp, fd, file);
845: close(fd);
1.7 kristaps 846:
1.8 kristaps 847: if (rc >= MANDOCLEVEL_FATAL) {
1.59 schwarze 848: fprintf(stderr, "fatal mandoc error: %s/%s\n",
849: req->q.manpath, file);
1.72 schwarze 850: pg_error_internal();
1.7 kristaps 851: return;
852: }
853:
1.90 schwarze 854: usepath = strcmp(req->q.manpath, req->p[0]);
855: mandoc_asprintf(&opts,
856: "fragment,man=%s?query=%%N&sec=%%S%s%s%s%s",
857: scriptname,
858: req->q.arch ? "&arch=" : "",
859: req->q.arch ? req->q.arch : "",
860: usepath ? "&manpath=" : "",
861: usepath ? req->q.manpath : "");
1.10 kristaps 862:
1.49 schwarze 863: mparse_result(mp, &mdoc, &man, NULL);
1.32 kristaps 864: if (NULL == man && NULL == mdoc) {
1.59 schwarze 865: fprintf(stderr, "fatal mandoc error: %s/%s\n",
866: req->q.manpath, file);
1.72 schwarze 867: pg_error_internal();
1.32 kristaps 868: mparse_free(mp);
1.100 schwarze 869: mchars_free(mchars);
1.32 kristaps 870: return;
871: }
872:
1.100 schwarze 873: vp = html_alloc(mchars, opts);
1.7 kristaps 874:
1.32 kristaps 875: if (NULL != mdoc)
1.8 kristaps 876: html_mdoc(vp, mdoc);
1.32 kristaps 877: else
1.8 kristaps 878: html_man(vp, man);
1.32 kristaps 879:
1.8 kristaps 880: html_free(vp);
881: mparse_free(mp);
1.100 schwarze 882: mchars_free(mchars);
1.90 schwarze 883: free(opts);
1.7 kristaps 884: }
885:
886: static void
1.70 schwarze 887: resp_show(const struct req *req, const char *file)
888: {
1.76 schwarze 889:
890: if ('.' == file[0] && '/' == file[1])
1.71 schwarze 891: file += 2;
1.70 schwarze 892:
893: if ('c' == *file)
894: catman(req, file);
895: else
896: format(req, file);
897: }
898:
899: static void
1.84 schwarze 900: pg_show(struct req *req, const char *fullpath)
1.1 kristaps 901: {
1.84 schwarze 902: char *manpath;
903: const char *file;
1.25 kristaps 904:
1.84 schwarze 905: if ((file = strchr(fullpath, '/')) == NULL) {
1.72 schwarze 906: pg_error_badrequest(
1.59 schwarze 907: "You did not specify a page to show.");
1.25 kristaps 908: return;
909: }
1.84 schwarze 910: manpath = mandoc_strndup(fullpath, file - fullpath);
911: file++;
1.6 kristaps 912:
1.84 schwarze 913: if ( ! validate_manpath(req, manpath)) {
1.77 schwarze 914: pg_error_badrequest(
915: "You specified an invalid manpath.");
1.84 schwarze 916: free(manpath);
1.77 schwarze 917: return;
918: }
919:
1.24 kristaps 920: /*
1.58 schwarze 921: * Begin by chdir()ing into the manpath.
1.24 kristaps 922: * This way we can pick up the database files, which are
923: * relative to the manpath root.
924: */
925:
1.84 schwarze 926: if (chdir(manpath) == -1) {
1.77 schwarze 927: fprintf(stderr, "chdir %s: %s\n",
1.84 schwarze 928: manpath, strerror(errno));
1.77 schwarze 929: pg_error_internal();
1.84 schwarze 930: free(manpath);
1.76 schwarze 931: return;
932: }
933:
1.84 schwarze 934: if (strcmp(manpath, "mandoc")) {
935: free(req->q.manpath);
936: req->q.manpath = manpath;
937: } else
938: free(manpath);
939:
940: if ( ! validate_filename(file)) {
1.76 schwarze 941: pg_error_badrequest(
942: "You specified an invalid manual file.");
1.24 kristaps 943: return;
944: }
1.79 schwarze 945:
1.70 schwarze 946: resp_begin_html(200, NULL);
947: resp_searchform(req);
1.84 schwarze 948: resp_show(req, file);
1.70 schwarze 949: resp_end_html();
1.6 kristaps 950: }
951:
952: static void
1.66 schwarze 953: pg_search(const struct req *req)
1.6 kristaps 954: {
1.52 schwarze 955: struct mansearch search;
956: struct manpaths paths;
957: struct manpage *res;
1.97 schwarze 958: char **argv;
959: char *query, *rp, *wp;
1.52 schwarze 960: size_t ressz;
1.97 schwarze 961: int argc;
1.6 kristaps 962:
963: /*
1.24 kristaps 964: * Begin by chdir()ing into the root of the manpath.
965: * This way we can pick up the database files, which are
966: * relative to the manpath root.
967: */
968:
1.58 schwarze 969: if (-1 == (chdir(req->q.manpath))) {
1.77 schwarze 970: fprintf(stderr, "chdir %s: %s\n",
971: req->q.manpath, strerror(errno));
972: pg_error_internal();
1.24 kristaps 973: return;
974: }
975:
1.52 schwarze 976: search.arch = req->q.arch;
977: search.sec = req->q.sec;
1.94 schwarze 978: search.outkey = "Nd";
979: search.argmode = req->q.equal ? ARG_NAME : ARG_EXPR;
1.101 ! schwarze 980: search.firstmatch = 1;
1.52 schwarze 981:
982: paths.sz = 1;
983: paths.paths = mandoc_malloc(sizeof(char *));
984: paths.paths[0] = mandoc_strdup(".");
1.24 kristaps 985:
986: /*
1.97 schwarze 987: * Break apart at spaces with backslash-escaping.
1.6 kristaps 988: */
989:
1.97 schwarze 990: argc = 0;
991: argv = NULL;
992: rp = query = mandoc_strdup(req->q.query);
993: for (;;) {
994: while (isspace((unsigned char)*rp))
995: rp++;
996: if (*rp == '\0')
997: break;
998: argv = mandoc_reallocarray(argv, argc + 1, sizeof(char *));
999: argv[argc++] = wp = rp;
1000: for (;;) {
1001: if (isspace((unsigned char)*rp)) {
1002: *wp = '\0';
1003: rp++;
1004: break;
1005: }
1006: if (rp[0] == '\\' && rp[1] != '\0')
1007: rp++;
1008: if (wp != rp)
1009: *wp = *rp;
1010: if (*rp == '\0')
1011: break;
1012: wp++;
1013: rp++;
1014: }
1.6 kristaps 1015: }
1016:
1.97 schwarze 1017: if (0 == mansearch(&search, &paths, argc, argv, &res, &ressz))
1.72 schwarze 1018: pg_noresult(req, "You entered an invalid query.");
1.59 schwarze 1019: else if (0 == ressz)
1.72 schwarze 1020: pg_noresult(req, "No results found.");
1.59 schwarze 1021: else
1.72 schwarze 1022: pg_searchres(req, res, ressz);
1.6 kristaps 1023:
1.97 schwarze 1024: free(query);
1025: mansearch_free(res, ressz);
1.52 schwarze 1026: free(paths.paths[0]);
1027: free(paths.paths);
1.1 kristaps 1028: }
1029:
1030: int
1031: main(void)
1032: {
1.66 schwarze 1033: struct req req;
1.95 schwarze 1034: struct itimerval itimer;
1.66 schwarze 1035: const char *path;
1.83 schwarze 1036: const char *querystring;
1.1 kristaps 1037: int i;
1.95 schwarze 1038:
1039: /* Poor man's ReDoS mitigation. */
1040:
1.99 schwarze 1041: itimer.it_value.tv_sec = 2;
1.95 schwarze 1042: itimer.it_value.tv_usec = 0;
1.99 schwarze 1043: itimer.it_interval.tv_sec = 2;
1.95 schwarze 1044: itimer.it_interval.tv_usec = 0;
1045: if (setitimer(ITIMER_VIRTUAL, &itimer, NULL) == -1) {
1046: fprintf(stderr, "setitimer: %s\n", strerror(errno));
1047: pg_error_internal();
1048: return(EXIT_FAILURE);
1049: }
1.6 kristaps 1050:
1.24 kristaps 1051: /* Scan our run-time environment. */
1.6 kristaps 1052:
1.58 schwarze 1053: if (NULL == (scriptname = getenv("SCRIPT_NAME")))
1054: scriptname = "";
1.24 kristaps 1055:
1.80 schwarze 1056: if ( ! validate_urifrag(scriptname)) {
1057: fprintf(stderr, "unsafe SCRIPT_NAME \"%s\"\n",
1058: scriptname);
1059: pg_error_internal();
1060: return(EXIT_FAILURE);
1061: }
1062:
1.24 kristaps 1063: /*
1.67 schwarze 1064: * First we change directory into the MAN_DIR so that
1.24 kristaps 1065: * subsequent scanning for manpath directories is rooted
1066: * relative to the same position.
1067: */
1068:
1.67 schwarze 1069: if (-1 == chdir(MAN_DIR)) {
1.58 schwarze 1070: fprintf(stderr, "MAN_DIR: %s: %s\n",
1.67 schwarze 1071: MAN_DIR, strerror(errno));
1.72 schwarze 1072: pg_error_internal();
1.24 kristaps 1073: return(EXIT_FAILURE);
1074: }
1075:
1076: memset(&req, 0, sizeof(struct req));
1.54 schwarze 1077: pathgen(&req);
1.1 kristaps 1078:
1.24 kristaps 1079: /* Next parse out the query string. */
1.1 kristaps 1080:
1.58 schwarze 1081: if (NULL != (querystring = getenv("QUERY_STRING")))
1082: http_parse(&req, querystring);
1.77 schwarze 1083:
1.89 schwarze 1084: if ( ! (NULL == req.q.manpath ||
1085: validate_manpath(&req, req.q.manpath))) {
1.77 schwarze 1086: pg_error_badrequest(
1087: "You specified an invalid manpath.");
1088: return(EXIT_FAILURE);
1089: }
1.1 kristaps 1090:
1.80 schwarze 1091: if ( ! (NULL == req.q.arch || validate_urifrag(req.q.arch))) {
1092: pg_error_badrequest(
1093: "You specified an invalid architecture.");
1094: return(EXIT_FAILURE);
1095: }
1096:
1.66 schwarze 1097: /* Dispatch to the three different pages. */
1.1 kristaps 1098:
1.66 schwarze 1099: path = getenv("PATH_INFO");
1100: if (NULL == path)
1101: path = "";
1102: else if ('/' == *path)
1103: path++;
1104:
1105: if ('\0' != *path)
1106: pg_show(&req, path);
1.85 schwarze 1107: else if (NULL != req.q.query)
1.66 schwarze 1108: pg_search(&req);
1109: else
1.72 schwarze 1110: pg_index(&req);
1.1 kristaps 1111:
1.83 schwarze 1112: free(req.q.manpath);
1113: free(req.q.arch);
1114: free(req.q.sec);
1.85 schwarze 1115: free(req.q.query);
1.52 schwarze 1116: for (i = 0; i < (int)req.psz; i++)
1117: free(req.p[i]);
1.24 kristaps 1118: free(req.p);
1.1 kristaps 1119: return(EXIT_SUCCESS);
1.24 kristaps 1120: }
1121:
1122: /*
1123: * Scan for indexable paths.
1124: */
1125: static void
1.54 schwarze 1126: pathgen(struct req *req)
1.24 kristaps 1127: {
1.54 schwarze 1128: FILE *fp;
1129: char *dp;
1130: size_t dpsz;
1131:
1.74 schwarze 1132: if (NULL == (fp = fopen("manpath.conf", "r"))) {
1133: fprintf(stderr, "%s/manpath.conf: %s\n",
1134: MAN_DIR, strerror(errno));
1135: pg_error_internal();
1136: exit(EXIT_FAILURE);
1137: }
1.24 kristaps 1138:
1.54 schwarze 1139: while (NULL != (dp = fgetln(fp, &dpsz))) {
1.55 schwarze 1140: if ('\n' == dp[dpsz - 1])
1141: dpsz--;
1.54 schwarze 1142: req->p = mandoc_realloc(req->p,
1143: (req->psz + 1) * sizeof(char *));
1.80 schwarze 1144: dp = mandoc_strndup(dp, dpsz);
1145: if ( ! validate_urifrag(dp)) {
1146: fprintf(stderr, "%s/manpath.conf contains "
1147: "unsafe path \"%s\"\n", MAN_DIR, dp);
1148: pg_error_internal();
1149: exit(EXIT_FAILURE);
1150: }
1151: if (NULL != strchr(dp, '/')) {
1152: fprintf(stderr, "%s/manpath.conf contains "
1153: "path with slash \"%s\"\n", MAN_DIR, dp);
1154: pg_error_internal();
1155: exit(EXIT_FAILURE);
1156: }
1157: req->p[req->psz++] = dp;
1.74 schwarze 1158: }
1159:
1160: if ( req->p == NULL ) {
1161: fprintf(stderr, "%s/manpath.conf is empty\n", MAN_DIR);
1162: pg_error_internal();
1163: exit(EXIT_FAILURE);
1.24 kristaps 1164: }
1165: }
CVSweb