Annotation of mandoc/cgi.c, Revision 1.140
1.140 ! schwarze 1: /* $Id: cgi.c,v 1.139 2016/09/01 15:07:24 schwarze Exp $ */
1.6 kristaps 2: /*
1.42 kristaps 3: * Copyright (c) 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
1.118 schwarze 4: * Copyright (c) 2014, 2015, 2016 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: *
1.105 schwarze 10: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1.6 kristaps 11: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1.105 schwarze 12: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1.6 kristaps 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>
1.128 schwarze 24: #include <err.h>
1.58 schwarze 25: #include <errno.h>
1.1 kristaps 26: #include <fcntl.h>
1.6 kristaps 27: #include <limits.h>
1.92 schwarze 28: #include <stdint.h>
1.1 kristaps 29: #include <stdio.h>
30: #include <stdlib.h>
31: #include <string.h>
1.6 kristaps 32: #include <unistd.h>
1.1 kristaps 33:
1.108 schwarze 34: #include "mandoc_aux.h"
1.4 schwarze 35: #include "mandoc.h"
1.108 schwarze 36: #include "roff.h"
1.111 schwarze 37: #include "mdoc.h"
1.112 schwarze 38: #include "man.h"
1.8 kristaps 39: #include "main.h"
1.105 schwarze 40: #include "manconf.h"
1.52 schwarze 41: #include "mansearch.h"
1.67 schwarze 42: #include "cgi.h"
1.1 kristaps 43:
1.20 kristaps 44: /*
45: * A query as passed to the search function.
46: */
47: struct query {
1.83 schwarze 48: char *manpath; /* desired manual directory */
49: char *arch; /* architecture */
50: char *sec; /* manual section */
1.85 schwarze 51: char *query; /* unparsed query expression */
1.65 schwarze 52: int equal; /* match whole names, not substrings */
1.20 kristaps 53: };
54:
1.1 kristaps 55: struct req {
1.58 schwarze 56: struct query q;
57: char **p; /* array of available manpaths */
58: size_t psz; /* number of available manpaths */
1.121 schwarze 59: int isquery; /* QUERY_STRING used, not PATH_INFO */
1.1 kristaps 60: };
61:
1.131 schwarze 62: enum focus {
63: FOCUS_NONE = 0,
64: FOCUS_QUERY
65: };
66:
1.6 kristaps 67: static void html_print(const char *);
1.10 kristaps 68: static void html_putchar(char);
1.104 schwarze 69: static int http_decode(char *);
1.129 schwarze 70: static void parse_manpath_conf(struct req *);
71: static void parse_path_info(struct req *req, const char *path);
72: static void parse_query_string(struct req *, const char *);
1.72 schwarze 73: static void pg_error_badrequest(const char *);
74: static void pg_error_internal(void);
75: static void pg_index(const struct req *);
76: static void pg_noresult(const struct req *, const char *);
1.66 schwarze 77: static void pg_search(const struct req *);
1.72 schwarze 78: static void pg_searchres(const struct req *,
79: struct manpage *, size_t);
1.79 schwarze 80: static void pg_show(struct req *, const char *);
1.6 kristaps 81: static void resp_begin_html(int, const char *);
82: static void resp_begin_http(int, const char *);
1.129 schwarze 83: static void resp_catman(const struct req *, const char *);
1.114 schwarze 84: static void resp_copy(const char *);
1.6 kristaps 85: static void resp_end_html(void);
1.129 schwarze 86: static void resp_format(const struct req *, const char *);
1.131 schwarze 87: static void resp_searchform(const struct req *, enum focus);
1.70 schwarze 88: static void resp_show(const struct req *, const char *);
1.85 schwarze 89: static void set_query_attr(char **, char **);
90: static int validate_filename(const char *);
91: static int validate_manpath(const struct req *, const char *);
92: static int validate_urifrag(const char *);
1.6 kristaps 93:
1.119 schwarze 94: static const char *scriptname = SCRIPT_NAME;
1.1 kristaps 95:
1.70 schwarze 96: static const int sec_prios[] = {1, 4, 5, 8, 6, 3, 7, 2, 9};
1.68 schwarze 97: static const char *const sec_numbers[] = {
98: "0", "1", "2", "3", "3p", "4", "5", "6", "7", "8", "9"
99: };
100: static const char *const sec_names[] = {
101: "All Sections",
102: "1 - General Commands",
103: "2 - System Calls",
1.96 schwarze 104: "3 - Library Functions",
105: "3p - Perl Library",
106: "4 - Device Drivers",
1.68 schwarze 107: "5 - File Formats",
108: "6 - Games",
1.96 schwarze 109: "7 - Miscellaneous Information",
110: "8 - System Manager\'s Manual",
111: "9 - Kernel Developer\'s Manual"
1.68 schwarze 112: };
113: static const int sec_MAX = sizeof(sec_names) / sizeof(char *);
114:
115: static const char *const arch_names[] = {
1.137 schwarze 116: "amd64", "alpha", "armv7",
117: "hppa", "i386", "landisk",
1.125 schwarze 118: "loongson", "luna88k", "macppc", "mips64",
1.140 ! schwarze 119: "octeon", "sgi", "socppc", "sparc64",
1.137 schwarze 120: "amiga", "arc", "armish", "arm32",
121: "atari", "aviion", "beagle", "cats",
122: "hppa64", "hp300",
1.125 schwarze 123: "ia64", "mac68k", "mvme68k", "mvme88k",
124: "mvmeppc", "palm", "pc532", "pegasos",
1.139 schwarze 125: "pmax", "powerpc", "solbourne", "sparc",
1.140 ! schwarze 126: "sun3", "vax", "wgrisc", "x68k",
! 127: "zaurus"
1.68 schwarze 128: };
129: static const int arch_MAX = sizeof(arch_names) / sizeof(char *);
130:
1.6 kristaps 131: /*
1.20 kristaps 132: * Print a character, escaping HTML along the way.
133: * This will pass non-ASCII straight to output: be warned!
134: */
1.10 kristaps 135: static void
136: html_putchar(char c)
137: {
138:
139: switch (c) {
140: case ('"'):
141: printf(""e;");
142: break;
143: case ('&'):
144: printf("&");
145: break;
146: case ('>'):
147: printf(">");
148: break;
149: case ('<'):
150: printf("<");
151: break;
152: default:
153: putchar((unsigned char)c);
154: break;
155: }
156: }
1.57 schwarze 157:
1.6 kristaps 158: /*
1.20 kristaps 159: * Call through to html_putchar().
160: * Accepts NULL strings.
1.6 kristaps 161: */
1.1 kristaps 162: static void
1.6 kristaps 163: html_print(const char *p)
1.1 kristaps 164: {
1.104 schwarze 165:
1.6 kristaps 166: if (NULL == p)
167: return;
1.1 kristaps 168: while ('\0' != *p)
1.10 kristaps 169: html_putchar(*p++);
1.1 kristaps 170: }
171:
172: /*
1.83 schwarze 173: * Transfer the responsibility for the allocated string *val
174: * to the query structure.
1.1 kristaps 175: */
176: static void
1.83 schwarze 177: set_query_attr(char **attr, char **val)
1.1 kristaps 178: {
179:
1.83 schwarze 180: free(*attr);
181: if (**val == '\0') {
182: *attr = NULL;
183: free(*val);
184: } else
185: *attr = *val;
186: *val = NULL;
187: }
188:
189: /*
190: * Parse the QUERY_STRING for key-value pairs
191: * and store the values into the query structure.
192: */
193: static void
1.129 schwarze 194: parse_query_string(struct req *req, const char *qs)
1.83 schwarze 195: {
196: char *key, *val;
197: size_t keysz, valsz;
198:
1.121 schwarze 199: req->isquery = 1;
1.83 schwarze 200: req->q.manpath = NULL;
201: req->q.arch = NULL;
202: req->q.sec = NULL;
1.85 schwarze 203: req->q.query = NULL;
1.83 schwarze 204: req->q.equal = 1;
205:
206: key = val = NULL;
207: while (*qs != '\0') {
1.24 kristaps 208:
1.83 schwarze 209: /* Parse one key. */
210:
211: keysz = strcspn(qs, "=;&");
212: key = mandoc_strndup(qs, keysz);
213: qs += keysz;
214: if (*qs != '=')
215: goto next;
216:
217: /* Parse one value. */
218:
219: valsz = strcspn(++qs, ";&");
220: val = mandoc_strndup(qs, valsz);
221: qs += valsz;
222:
223: /* Decode and catch encoding errors. */
1.1 kristaps 224:
1.83 schwarze 225: if ( ! (http_decode(key) && http_decode(val)))
226: goto next;
1.1 kristaps 227:
1.83 schwarze 228: /* Handle key-value pairs. */
1.1 kristaps 229:
1.83 schwarze 230: if ( ! strcmp(key, "query"))
1.85 schwarze 231: set_query_attr(&req->q.query, &val);
1.1 kristaps 232:
1.83 schwarze 233: else if ( ! strcmp(key, "apropos"))
234: req->q.equal = !strcmp(val, "0");
235:
236: else if ( ! strcmp(key, "manpath")) {
1.73 schwarze 237: #ifdef COMPAT_OLDURI
1.83 schwarze 238: if ( ! strncmp(val, "OpenBSD ", 8)) {
1.73 schwarze 239: val[7] = '-';
240: if ('C' == val[8])
241: val[8] = 'c';
242: }
243: #endif
1.83 schwarze 244: set_query_attr(&req->q.manpath, &val);
245: }
246:
247: else if ( ! (strcmp(key, "sec")
1.73 schwarze 248: #ifdef COMPAT_OLDURI
1.83 schwarze 249: && strcmp(key, "sektion")
1.73 schwarze 250: #endif
1.83 schwarze 251: )) {
252: if ( ! strcmp(val, "0"))
253: *val = '\0';
254: set_query_attr(&req->q.sec, &val);
1.65 schwarze 255: }
1.83 schwarze 256:
257: else if ( ! strcmp(key, "arch")) {
258: if ( ! strcmp(val, "default"))
259: *val = '\0';
260: set_query_attr(&req->q.arch, &val);
261: }
262:
263: /*
264: * The key must be freed in any case.
265: * The val may have been handed over to the query
266: * structure, in which case it is now NULL.
267: */
268: next:
269: free(key);
270: key = NULL;
271: free(val);
272: val = NULL;
273:
274: if (*qs != '\0')
275: qs++;
1.24 kristaps 276: }
1.1 kristaps 277: }
278:
279: /*
1.6 kristaps 280: * HTTP-decode a string. The standard explanation is that this turns
281: * "%4e+foo" into "n foo" in the regular way. This is done in-place
282: * over the allocated string.
1.1 kristaps 283: */
284: static int
1.24 kristaps 285: http_decode(char *p)
1.1 kristaps 286: {
287: char hex[3];
1.63 schwarze 288: char *q;
1.1 kristaps 289: int c;
290:
291: hex[2] = '\0';
292:
1.63 schwarze 293: q = p;
294: for ( ; '\0' != *p; p++, q++) {
1.1 kristaps 295: if ('%' == *p) {
296: if ('\0' == (hex[0] = *(p + 1)))
1.109 schwarze 297: return 0;
1.1 kristaps 298: if ('\0' == (hex[1] = *(p + 2)))
1.109 schwarze 299: return 0;
1.1 kristaps 300: if (1 != sscanf(hex, "%x", &c))
1.109 schwarze 301: return 0;
1.1 kristaps 302: if ('\0' == c)
1.109 schwarze 303: return 0;
1.1 kristaps 304:
1.63 schwarze 305: *q = (char)c;
306: p += 2;
1.1 kristaps 307: } else
1.63 schwarze 308: *q = '+' == *p ? ' ' : *p;
1.1 kristaps 309: }
310:
1.63 schwarze 311: *q = '\0';
1.109 schwarze 312: return 1;
1.1 kristaps 313: }
314:
1.6 kristaps 315: static void
316: resp_begin_http(int code, const char *msg)
317: {
318:
319: if (200 != code)
1.62 schwarze 320: printf("Status: %d %s\r\n", code, msg);
1.6 kristaps 321:
1.62 schwarze 322: printf("Content-Type: text/html; charset=utf-8\r\n"
323: "Cache-Control: no-cache\r\n"
324: "Pragma: no-cache\r\n"
325: "\r\n");
1.6 kristaps 326:
327: fflush(stdout);
328: }
329:
330: static void
1.114 schwarze 331: resp_copy(const char *filename)
332: {
333: char buf[4096];
334: ssize_t sz;
335: int fd;
336:
337: if ((fd = open(filename, O_RDONLY)) != -1) {
338: fflush(stdout);
339: while ((sz = read(fd, buf, sizeof(buf))) > 0)
340: write(STDOUT_FILENO, buf, sz);
1.138 schwarze 341: close(fd);
1.114 schwarze 342: }
343: }
344:
345: static void
1.6 kristaps 346: resp_begin_html(int code, const char *msg)
347: {
348:
349: resp_begin_http(code, msg);
350:
1.98 kristaps 351: printf("<!DOCTYPE html>\n"
1.126 schwarze 352: "<html>\n"
353: "<head>\n"
354: "<meta charset=\"UTF-8\"/>\n"
355: "<link rel=\"stylesheet\" href=\"%s/mandoc.css\""
356: " type=\"text/css\" media=\"all\">\n"
357: "<title>%s</title>\n"
358: "</head>\n"
359: "<body>\n"
1.58 schwarze 360: "<!-- Begin page content. //-->\n",
1.113 schwarze 361: CSS_DIR, CUSTOMIZE_TITLE);
1.114 schwarze 362:
363: resp_copy(MAN_DIR "/header.html");
1.6 kristaps 364: }
365:
366: static void
367: resp_end_html(void)
368: {
369:
1.114 schwarze 370: resp_copy(MAN_DIR "/footer.html");
371:
1.126 schwarze 372: puts("</body>\n"
373: "</html>");
1.6 kristaps 374: }
375:
376: static void
1.131 schwarze 377: resp_searchform(const struct req *req, enum focus focus)
1.6 kristaps 378: {
1.27 kristaps 379: int i;
1.13 kristaps 380:
1.6 kristaps 381: puts("<!-- Begin search form. //-->");
1.126 schwarze 382: printf("<div id=\"mancgi\">\n"
383: "<form action=\"/%s\" method=\"get\">\n"
384: "<fieldset>\n"
385: "<legend>Manual Page Search Parameters</legend>\n",
1.58 schwarze 386: scriptname);
1.68 schwarze 387:
388: /* Write query input box. */
389:
1.132 schwarze 390: printf("<input type=\"text\" name=\"query\" value=\"");
1.131 schwarze 391: if (req->q.query != NULL)
1.85 schwarze 392: html_print(req->q.query);
1.131 schwarze 393: printf( "\" size=\"40\"");
394: if (focus == FOCUS_QUERY)
395: printf(" autofocus");
396: puts(">");
1.68 schwarze 397:
1.132 schwarze 398: /* Write submission buttons. */
1.68 schwarze 399:
1.132 schwarze 400: printf( "<button type=\"submit\" name=\"apropos\" value=\"0\">"
401: "man</button>\n"
402: "<button type=\"submit\" name=\"apropos\" value=\"1\">"
403: "apropos</button>\n<br/>\n");
1.68 schwarze 404:
405: /* Write section selector. */
406:
1.132 schwarze 407: puts("<select name=\"sec\">");
1.68 schwarze 408: for (i = 0; i < sec_MAX; i++) {
1.126 schwarze 409: printf("<option value=\"%s\"", sec_numbers[i]);
1.68 schwarze 410: if (NULL != req->q.sec &&
411: 0 == strcmp(sec_numbers[i], req->q.sec))
1.126 schwarze 412: printf(" selected=\"selected\"");
413: printf(">%s</option>\n", sec_names[i]);
1.68 schwarze 414: }
1.126 schwarze 415: puts("</select>");
1.68 schwarze 416:
417: /* Write architecture selector. */
418:
1.126 schwarze 419: printf( "<select name=\"arch\">\n"
420: "<option value=\"default\"");
1.81 schwarze 421: if (NULL == req->q.arch)
1.126 schwarze 422: printf(" selected=\"selected\"");
423: puts(">All Architectures</option>");
1.68 schwarze 424: for (i = 0; i < arch_MAX; i++) {
1.126 schwarze 425: printf("<option value=\"%s\"", arch_names[i]);
1.68 schwarze 426: if (NULL != req->q.arch &&
427: 0 == strcmp(arch_names[i], req->q.arch))
1.126 schwarze 428: printf(" selected=\"selected\"");
429: printf(">%s</option>\n", arch_names[i]);
1.68 schwarze 430: }
1.126 schwarze 431: puts("</select>");
1.68 schwarze 432:
433: /* Write manpath selector. */
434:
1.27 kristaps 435: if (req->psz > 1) {
1.126 schwarze 436: puts("<select name=\"manpath\">");
1.27 kristaps 437: for (i = 0; i < (int)req->psz; i++) {
1.126 schwarze 438: printf("<option ");
1.102 schwarze 439: if (strcmp(req->q.manpath, req->p[i]) == 0)
1.126 schwarze 440: printf("selected=\"selected\" ");
441: printf("value=\"");
1.52 schwarze 442: html_print(req->p[i]);
1.27 kristaps 443: printf("\">");
1.52 schwarze 444: html_print(req->p[i]);
1.126 schwarze 445: puts("</option>");
1.27 kristaps 446: }
1.126 schwarze 447: puts("</select>");
1.27 kristaps 448: }
1.68 schwarze 449:
1.132 schwarze 450: puts("</fieldset>\n"
1.126 schwarze 451: "</form>\n"
452: "</div>");
1.20 kristaps 453: puts("<!-- End search form. //-->");
1.6 kristaps 454: }
455:
1.76 schwarze 456: static int
1.80 schwarze 457: validate_urifrag(const char *frag)
458: {
459:
460: while ('\0' != *frag) {
461: if ( ! (isalnum((unsigned char)*frag) ||
462: '-' == *frag || '.' == *frag ||
463: '/' == *frag || '_' == *frag))
1.109 schwarze 464: return 0;
1.80 schwarze 465: frag++;
466: }
1.109 schwarze 467: return 1;
1.80 schwarze 468: }
469:
470: static int
1.77 schwarze 471: validate_manpath(const struct req *req, const char* manpath)
472: {
473: size_t i;
474:
475: for (i = 0; i < req->psz; i++)
476: if ( ! strcmp(manpath, req->p[i]))
1.109 schwarze 477: return 1;
1.77 schwarze 478:
1.109 schwarze 479: return 0;
1.77 schwarze 480: }
481:
482: static int
1.76 schwarze 483: validate_filename(const char *file)
484: {
485:
486: if ('.' == file[0] && '/' == file[1])
487: file += 2;
488:
1.109 schwarze 489: return ! (strstr(file, "../") || strstr(file, "/..") ||
490: (strncmp(file, "man", 3) && strncmp(file, "cat", 3)));
1.76 schwarze 491: }
492:
1.6 kristaps 493: static void
1.72 schwarze 494: pg_index(const struct req *req)
1.6 kristaps 495: {
496:
497: resp_begin_html(200, NULL);
1.131 schwarze 498: resp_searchform(req, FOCUS_QUERY);
1.126 schwarze 499: printf("<p>\n"
1.86 schwarze 500: "This web interface is documented in the\n"
1.134 schwarze 501: "<a href=\"/%s%sman.cgi.8\">man.cgi(8)</a>\n"
1.86 schwarze 502: "manual, and the\n"
1.134 schwarze 503: "<a href=\"/%s%sapropos.1\">apropos(1)</a>\n"
1.69 schwarze 504: "manual explains the query syntax.\n"
1.126 schwarze 505: "</p>\n",
1.119 schwarze 506: scriptname, *scriptname == '\0' ? "" : "/",
507: scriptname, *scriptname == '\0' ? "" : "/");
1.6 kristaps 508: resp_end_html();
509: }
510:
511: static void
1.72 schwarze 512: pg_noresult(const struct req *req, const char *msg)
1.59 schwarze 513: {
514: resp_begin_html(200, NULL);
1.131 schwarze 515: resp_searchform(req, FOCUS_QUERY);
1.126 schwarze 516: puts("<p>");
1.59 schwarze 517: puts(msg);
1.126 schwarze 518: puts("</p>");
1.59 schwarze 519: resp_end_html();
520: }
521:
522: static void
1.72 schwarze 523: pg_error_badrequest(const char *msg)
1.9 kristaps 524: {
525:
1.59 schwarze 526: resp_begin_html(400, "Bad Request");
1.126 schwarze 527: puts("<h1>Bad Request</h1>\n"
528: "<p>\n");
1.59 schwarze 529: puts(msg);
530: printf("Try again from the\n"
1.126 schwarze 531: "<a href=\"/%s\">main page</a>.\n"
532: "</p>", scriptname);
1.9 kristaps 533: resp_end_html();
534: }
535:
536: static void
1.72 schwarze 537: pg_error_internal(void)
1.7 kristaps 538: {
539: resp_begin_html(500, "Internal Server Error");
1.126 schwarze 540: puts("<p>Internal Server Error</p>");
1.7 kristaps 541: resp_end_html();
542: }
543:
544: static void
1.72 schwarze 545: pg_searchres(const struct req *req, struct manpage *r, size_t sz)
1.1 kristaps 546: {
1.81 schwarze 547: char *arch, *archend;
1.120 schwarze 548: const char *sec;
549: size_t i, iuse;
1.81 schwarze 550: int archprio, archpriouse;
1.70 schwarze 551: int prio, priouse;
1.19 kristaps 552:
1.76 schwarze 553: for (i = 0; i < sz; i++) {
554: if (validate_filename(r[i].file))
555: continue;
1.128 schwarze 556: warnx("invalid filename %s in %s database",
1.76 schwarze 557: r[i].file, req->q.manpath);
558: pg_error_internal();
559: return;
560: }
561:
1.121 schwarze 562: if (req->isquery && sz == 1) {
1.6 kristaps 563: /*
564: * If we have just one result, then jump there now
565: * without any delay.
566: */
1.62 schwarze 567: printf("Status: 303 See Other\r\n");
1.119 schwarze 568: printf("Location: http://%s/%s%s%s/%s",
569: HTTP_HOST, scriptname,
570: *scriptname == '\0' ? "" : "/",
571: req->q.manpath, r[0].file);
1.62 schwarze 572: printf("\r\n"
573: "Content-Type: text/html; charset=utf-8\r\n"
574: "\r\n");
1.6 kristaps 575: return;
576: }
577:
1.12 kristaps 578: resp_begin_html(200, NULL);
1.131 schwarze 579: resp_searchform(req,
580: req->q.equal || sz == 1 ? FOCUS_NONE : FOCUS_QUERY);
1.1 kristaps 581:
1.123 schwarze 582: if (sz > 1) {
1.126 schwarze 583: puts("<div class=\"results\">");
584: puts("<table>");
1.123 schwarze 585:
586: for (i = 0; i < sz; i++) {
1.126 schwarze 587: printf("<tr>\n"
588: "<td class=\"title\">\n"
589: "<a href=\"/%s%s%s/%s",
1.123 schwarze 590: scriptname, *scriptname == '\0' ? "" : "/",
591: req->q.manpath, r[i].file);
592: printf("\">");
593: html_print(r[i].names);
1.126 schwarze 594: printf("</a>\n"
595: "</td>\n"
596: "<td class=\"desc\">");
1.123 schwarze 597: html_print(r[i].output);
1.126 schwarze 598: puts("</td>\n"
599: "</tr>");
1.123 schwarze 600: }
601:
1.126 schwarze 602: puts("</table>\n"
603: "</div>");
1.1 kristaps 604: }
1.16 kristaps 605:
1.70 schwarze 606: /*
607: * In man(1) mode, show one of the pages
608: * even if more than one is found.
609: */
610:
1.123 schwarze 611: if (req->q.equal || sz == 1) {
1.126 schwarze 612: puts("<hr>");
1.70 schwarze 613: iuse = 0;
1.120 schwarze 614: priouse = 20;
1.81 schwarze 615: archpriouse = 3;
1.70 schwarze 616: for (i = 0; i < sz; i++) {
1.120 schwarze 617: sec = r[i].file;
618: sec += strcspn(sec, "123456789");
619: if (sec[0] == '\0')
1.70 schwarze 620: continue;
1.120 schwarze 621: prio = sec_prios[sec[0] - '1'];
622: if (sec[1] != '/')
623: prio += 10;
624: if (req->q.arch == NULL) {
1.81 schwarze 625: archprio =
1.120 schwarze 626: ((arch = strchr(sec + 1, '/'))
627: == NULL) ? 3 :
628: ((archend = strchr(arch + 1, '/'))
629: == NULL) ? 0 :
1.81 schwarze 630: strncmp(arch, "amd64/",
631: archend - arch) ? 2 : 1;
632: if (archprio < archpriouse) {
633: archpriouse = archprio;
634: priouse = prio;
635: iuse = i;
636: continue;
637: }
638: if (archprio > archpriouse)
639: continue;
640: }
1.70 schwarze 641: if (prio >= priouse)
642: continue;
643: priouse = prio;
644: iuse = i;
645: }
646: resp_show(req, r[iuse].file);
647: }
648:
1.6 kristaps 649: resp_end_html();
650: }
651:
1.1 kristaps 652: static void
1.129 schwarze 653: resp_catman(const struct req *req, const char *file)
1.9 kristaps 654: {
1.10 kristaps 655: FILE *f;
1.115 schwarze 656: char *p;
657: size_t sz;
658: ssize_t len;
1.10 kristaps 659: int i;
660: int italic, bold;
1.9 kristaps 661:
1.115 schwarze 662: if ((f = fopen(file, "r")) == NULL) {
1.126 schwarze 663: puts("<p>You specified an invalid manual file.</p>");
1.9 kristaps 664: return;
665: }
666:
1.126 schwarze 667: puts("<div class=\"catman\">\n"
668: "<pre>");
1.10 kristaps 669:
1.115 schwarze 670: p = NULL;
671: sz = 0;
672:
673: while ((len = getline(&p, &sz, f)) != -1) {
1.10 kristaps 674: bold = italic = 0;
1.115 schwarze 675: for (i = 0; i < len - 1; i++) {
1.104 schwarze 676: /*
1.10 kristaps 677: * This means that the catpage is out of state.
678: * Ignore it and keep going (although the
679: * catpage is bogus).
680: */
681:
682: if ('\b' == p[i] || '\n' == p[i])
683: continue;
684:
685: /*
686: * Print a regular character.
687: * Close out any bold/italic scopes.
688: * If we're in back-space mode, make sure we'll
689: * have something to enter when we backspace.
690: */
691:
692: if ('\b' != p[i + 1]) {
693: if (italic)
1.126 schwarze 694: printf("</i>");
1.10 kristaps 695: if (bold)
1.126 schwarze 696: printf("</b>");
1.10 kristaps 697: italic = bold = 0;
698: html_putchar(p[i]);
699: continue;
1.115 schwarze 700: } else if (i + 2 >= len)
1.10 kristaps 701: continue;
702:
703: /* Italic mode. */
704:
705: if ('_' == p[i]) {
706: if (bold)
1.126 schwarze 707: printf("</b>");
1.10 kristaps 708: if ( ! italic)
1.126 schwarze 709: printf("<i>");
1.10 kristaps 710: bold = 0;
711: italic = 1;
712: i += 2;
713: html_putchar(p[i]);
714: continue;
715: }
716:
1.104 schwarze 717: /*
1.10 kristaps 718: * Handle funny behaviour troff-isms.
719: * These grok'd from the original man2html.c.
720: */
721:
722: if (('+' == p[i] && 'o' == p[i + 2]) ||
723: ('o' == p[i] && '+' == p[i + 2]) ||
724: ('|' == p[i] && '=' == p[i + 2]) ||
725: ('=' == p[i] && '|' == p[i + 2]) ||
726: ('*' == p[i] && '=' == p[i + 2]) ||
727: ('=' == p[i] && '*' == p[i + 2]) ||
728: ('*' == p[i] && '|' == p[i + 2]) ||
729: ('|' == p[i] && '*' == p[i + 2])) {
730: if (italic)
1.126 schwarze 731: printf("</i>");
1.10 kristaps 732: if (bold)
1.126 schwarze 733: printf("</b>");
1.10 kristaps 734: italic = bold = 0;
735: putchar('*');
736: i += 2;
737: continue;
738: } else if (('|' == p[i] && '-' == p[i + 2]) ||
739: ('-' == p[i] && '|' == p[i + 1]) ||
740: ('+' == p[i] && '-' == p[i + 1]) ||
741: ('-' == p[i] && '+' == p[i + 1]) ||
742: ('+' == p[i] && '|' == p[i + 1]) ||
743: ('|' == p[i] && '+' == p[i + 1])) {
744: if (italic)
1.126 schwarze 745: printf("</i>");
1.10 kristaps 746: if (bold)
1.126 schwarze 747: printf("</b>");
1.10 kristaps 748: italic = bold = 0;
749: putchar('+');
750: i += 2;
751: continue;
752: }
753:
754: /* Bold mode. */
1.104 schwarze 755:
1.10 kristaps 756: if (italic)
1.126 schwarze 757: printf("</i>");
1.10 kristaps 758: if ( ! bold)
1.126 schwarze 759: printf("<b>");
1.10 kristaps 760: bold = 1;
761: italic = 0;
762: i += 2;
763: html_putchar(p[i]);
764: }
765:
1.104 schwarze 766: /*
1.10 kristaps 767: * Clean up the last character.
1.104 schwarze 768: * We can get to a newline; don't print that.
1.10 kristaps 769: */
1.9 kristaps 770:
1.10 kristaps 771: if (italic)
1.126 schwarze 772: printf("</i>");
1.10 kristaps 773: if (bold)
1.126 schwarze 774: printf("</b>");
1.9 kristaps 775:
1.115 schwarze 776: if (i == len - 1 && p[i] != '\n')
1.10 kristaps 777: html_putchar(p[i]);
1.9 kristaps 778:
1.10 kristaps 779: putchar('\n');
780: }
1.115 schwarze 781: free(p);
1.10 kristaps 782:
1.126 schwarze 783: puts("</pre>\n"
784: "</div>");
1.10 kristaps 785:
786: fclose(f);
1.9 kristaps 787: }
788:
789: static void
1.129 schwarze 790: resp_format(const struct req *req, const char *file)
1.7 kristaps 791: {
1.106 schwarze 792: struct manoutput conf;
1.8 kristaps 793: struct mparse *mp;
1.107 schwarze 794: struct roff_man *man;
1.8 kristaps 795: void *vp;
1.90 schwarze 796: int fd;
797: int usepath;
1.7 kristaps 798:
1.8 kristaps 799: if (-1 == (fd = open(file, O_RDONLY, 0))) {
1.126 schwarze 800: puts("<p>You specified an invalid manual file.</p>");
1.7 kristaps 801: return;
802: }
803:
1.110 schwarze 804: mchars_alloc();
1.136 schwarze 805: mp = mparse_alloc(MPARSE_SO | MPARSE_UTF8 | MPARSE_LATIN1,
806: MANDOCLEVEL_BADARG, NULL, req->q.manpath);
1.103 schwarze 807: mparse_readfd(mp, fd, file);
1.8 kristaps 808: close(fd);
1.7 kristaps 809:
1.106 schwarze 810: memset(&conf, 0, sizeof(conf));
811: conf.fragment = 1;
1.90 schwarze 812: usepath = strcmp(req->q.manpath, req->p[0]);
1.122 schwarze 813: mandoc_asprintf(&conf.man, "/%s%s%%N.%%S",
814: usepath ? req->q.manpath : "", usepath ? "/" : "");
1.10 kristaps 815:
1.108 schwarze 816: mparse_result(mp, &man, NULL);
817: if (man == NULL) {
1.128 schwarze 818: warnx("fatal mandoc error: %s/%s", req->q.manpath, file);
1.72 schwarze 819: pg_error_internal();
1.32 kristaps 820: mparse_free(mp);
1.110 schwarze 821: mchars_free();
1.32 kristaps 822: return;
823: }
824:
1.110 schwarze 825: vp = html_alloc(&conf);
1.7 kristaps 826:
1.111 schwarze 827: if (man->macroset == MACROSET_MDOC) {
828: mdoc_validate(man);
1.108 schwarze 829: html_mdoc(vp, man);
1.112 schwarze 830: } else {
831: man_validate(man);
1.8 kristaps 832: html_man(vp, man);
1.112 schwarze 833: }
1.32 kristaps 834:
1.8 kristaps 835: html_free(vp);
836: mparse_free(mp);
1.110 schwarze 837: mchars_free();
1.106 schwarze 838: free(conf.man);
1.7 kristaps 839: }
840:
841: static void
1.70 schwarze 842: resp_show(const struct req *req, const char *file)
843: {
1.76 schwarze 844:
845: if ('.' == file[0] && '/' == file[1])
1.71 schwarze 846: file += 2;
1.70 schwarze 847:
848: if ('c' == *file)
1.129 schwarze 849: resp_catman(req, file);
1.70 schwarze 850: else
1.129 schwarze 851: resp_format(req, file);
1.70 schwarze 852: }
853:
854: static void
1.84 schwarze 855: pg_show(struct req *req, const char *fullpath)
1.1 kristaps 856: {
1.84 schwarze 857: char *manpath;
858: const char *file;
1.25 kristaps 859:
1.84 schwarze 860: if ((file = strchr(fullpath, '/')) == NULL) {
1.72 schwarze 861: pg_error_badrequest(
1.59 schwarze 862: "You did not specify a page to show.");
1.25 kristaps 863: return;
1.104 schwarze 864: }
1.84 schwarze 865: manpath = mandoc_strndup(fullpath, file - fullpath);
866: file++;
1.6 kristaps 867:
1.84 schwarze 868: if ( ! validate_manpath(req, manpath)) {
1.77 schwarze 869: pg_error_badrequest(
870: "You specified an invalid manpath.");
1.84 schwarze 871: free(manpath);
1.77 schwarze 872: return;
873: }
874:
1.24 kristaps 875: /*
1.58 schwarze 876: * Begin by chdir()ing into the manpath.
1.24 kristaps 877: * This way we can pick up the database files, which are
878: * relative to the manpath root.
879: */
880:
1.84 schwarze 881: if (chdir(manpath) == -1) {
1.128 schwarze 882: warn("chdir %s", manpath);
1.77 schwarze 883: pg_error_internal();
1.84 schwarze 884: free(manpath);
1.76 schwarze 885: return;
886: }
1.134 schwarze 887: free(manpath);
1.84 schwarze 888:
889: if ( ! validate_filename(file)) {
1.76 schwarze 890: pg_error_badrequest(
891: "You specified an invalid manual file.");
1.24 kristaps 892: return;
893: }
1.79 schwarze 894:
1.70 schwarze 895: resp_begin_html(200, NULL);
1.131 schwarze 896: resp_searchform(req, FOCUS_NONE);
1.84 schwarze 897: resp_show(req, file);
1.70 schwarze 898: resp_end_html();
1.6 kristaps 899: }
900:
901: static void
1.66 schwarze 902: pg_search(const struct req *req)
1.6 kristaps 903: {
1.52 schwarze 904: struct mansearch search;
905: struct manpaths paths;
906: struct manpage *res;
1.97 schwarze 907: char **argv;
908: char *query, *rp, *wp;
1.52 schwarze 909: size_t ressz;
1.97 schwarze 910: int argc;
1.6 kristaps 911:
912: /*
1.24 kristaps 913: * Begin by chdir()ing into the root of the manpath.
914: * This way we can pick up the database files, which are
915: * relative to the manpath root.
916: */
917:
1.128 schwarze 918: if (chdir(req->q.manpath) == -1) {
919: warn("chdir %s", req->q.manpath);
1.77 schwarze 920: pg_error_internal();
1.24 kristaps 921: return;
922: }
923:
1.52 schwarze 924: search.arch = req->q.arch;
925: search.sec = req->q.sec;
1.94 schwarze 926: search.outkey = "Nd";
927: search.argmode = req->q.equal ? ARG_NAME : ARG_EXPR;
1.101 schwarze 928: search.firstmatch = 1;
1.52 schwarze 929:
930: paths.sz = 1;
931: paths.paths = mandoc_malloc(sizeof(char *));
932: paths.paths[0] = mandoc_strdup(".");
1.24 kristaps 933:
934: /*
1.97 schwarze 935: * Break apart at spaces with backslash-escaping.
1.6 kristaps 936: */
937:
1.97 schwarze 938: argc = 0;
939: argv = NULL;
940: rp = query = mandoc_strdup(req->q.query);
941: for (;;) {
942: while (isspace((unsigned char)*rp))
943: rp++;
944: if (*rp == '\0')
945: break;
946: argv = mandoc_reallocarray(argv, argc + 1, sizeof(char *));
947: argv[argc++] = wp = rp;
948: for (;;) {
949: if (isspace((unsigned char)*rp)) {
950: *wp = '\0';
951: rp++;
952: break;
953: }
954: if (rp[0] == '\\' && rp[1] != '\0')
955: rp++;
956: if (wp != rp)
957: *wp = *rp;
958: if (*rp == '\0')
959: break;
960: wp++;
961: rp++;
962: }
1.6 kristaps 963: }
964:
1.97 schwarze 965: if (0 == mansearch(&search, &paths, argc, argv, &res, &ressz))
1.72 schwarze 966: pg_noresult(req, "You entered an invalid query.");
1.59 schwarze 967: else if (0 == ressz)
1.72 schwarze 968: pg_noresult(req, "No results found.");
1.59 schwarze 969: else
1.72 schwarze 970: pg_searchres(req, res, ressz);
1.6 kristaps 971:
1.97 schwarze 972: free(query);
973: mansearch_free(res, ressz);
1.52 schwarze 974: free(paths.paths[0]);
975: free(paths.paths);
1.1 kristaps 976: }
977:
978: int
979: main(void)
980: {
1.66 schwarze 981: struct req req;
1.95 schwarze 982: struct itimerval itimer;
1.66 schwarze 983: const char *path;
1.83 schwarze 984: const char *querystring;
1.1 kristaps 985: int i;
1.95 schwarze 986:
987: /* Poor man's ReDoS mitigation. */
988:
1.99 schwarze 989: itimer.it_value.tv_sec = 2;
1.95 schwarze 990: itimer.it_value.tv_usec = 0;
1.99 schwarze 991: itimer.it_interval.tv_sec = 2;
1.95 schwarze 992: itimer.it_interval.tv_usec = 0;
993: if (setitimer(ITIMER_VIRTUAL, &itimer, NULL) == -1) {
1.128 schwarze 994: warn("setitimer");
1.80 schwarze 995: pg_error_internal();
1.109 schwarze 996: return EXIT_FAILURE;
1.80 schwarze 997: }
998:
1.24 kristaps 999: /*
1.67 schwarze 1000: * First we change directory into the MAN_DIR so that
1.24 kristaps 1001: * subsequent scanning for manpath directories is rooted
1002: * relative to the same position.
1003: */
1004:
1.128 schwarze 1005: if (chdir(MAN_DIR) == -1) {
1006: warn("MAN_DIR: %s", MAN_DIR);
1.72 schwarze 1007: pg_error_internal();
1.109 schwarze 1008: return EXIT_FAILURE;
1.104 schwarze 1009: }
1.24 kristaps 1010:
1011: memset(&req, 0, sizeof(struct req));
1.118 schwarze 1012: req.q.equal = 1;
1.129 schwarze 1013: parse_manpath_conf(&req);
1.1 kristaps 1014:
1.117 schwarze 1015: /* Parse the path info and the query string. */
1.1 kristaps 1016:
1.117 schwarze 1017: if ((path = getenv("PATH_INFO")) == NULL)
1018: path = "";
1019: else if (*path == '/')
1020: path++;
1021:
1.124 schwarze 1022: if (*path != '\0') {
1.129 schwarze 1023: parse_path_info(&req, path);
1.133 schwarze 1024: if (req.q.manpath == NULL || access(path, F_OK) == -1)
1.124 schwarze 1025: path = "";
1.117 schwarze 1026: } else if ((querystring = getenv("QUERY_STRING")) != NULL)
1.129 schwarze 1027: parse_query_string(&req, querystring);
1.77 schwarze 1028:
1.117 schwarze 1029: /* Validate parsed data and add defaults. */
1030:
1.102 schwarze 1031: if (req.q.manpath == NULL)
1032: req.q.manpath = mandoc_strdup(req.p[0]);
1033: else if ( ! validate_manpath(&req, req.q.manpath)) {
1.77 schwarze 1034: pg_error_badrequest(
1035: "You specified an invalid manpath.");
1.109 schwarze 1036: return EXIT_FAILURE;
1.77 schwarze 1037: }
1.1 kristaps 1038:
1.80 schwarze 1039: if ( ! (NULL == req.q.arch || validate_urifrag(req.q.arch))) {
1040: pg_error_badrequest(
1041: "You specified an invalid architecture.");
1.109 schwarze 1042: return EXIT_FAILURE;
1.80 schwarze 1043: }
1044:
1.66 schwarze 1045: /* Dispatch to the three different pages. */
1.1 kristaps 1046:
1.66 schwarze 1047: if ('\0' != *path)
1048: pg_show(&req, path);
1.85 schwarze 1049: else if (NULL != req.q.query)
1.66 schwarze 1050: pg_search(&req);
1051: else
1.72 schwarze 1052: pg_index(&req);
1.1 kristaps 1053:
1.83 schwarze 1054: free(req.q.manpath);
1055: free(req.q.arch);
1056: free(req.q.sec);
1.85 schwarze 1057: free(req.q.query);
1.52 schwarze 1058: for (i = 0; i < (int)req.psz; i++)
1059: free(req.p[i]);
1.24 kristaps 1060: free(req.p);
1.109 schwarze 1061: return EXIT_SUCCESS;
1.117 schwarze 1062: }
1063:
1064: /*
1065: * If PATH_INFO is not a file name, translate it to a query.
1066: */
1067: static void
1.129 schwarze 1068: parse_path_info(struct req *req, const char *path)
1.117 schwarze 1069: {
1.135 schwarze 1070: char *dir[4];
1071: int i;
1.117 schwarze 1072:
1.121 schwarze 1073: req->isquery = 0;
1.117 schwarze 1074: req->q.equal = 1;
1075: req->q.manpath = mandoc_strdup(path);
1.135 schwarze 1076: req->q.arch = NULL;
1.117 schwarze 1077:
1078: /* Mandatory manual page name. */
1079: if ((req->q.query = strrchr(req->q.manpath, '/')) == NULL) {
1080: req->q.query = req->q.manpath;
1081: req->q.manpath = NULL;
1082: } else
1083: *req->q.query++ = '\0';
1084:
1085: /* Optional trailing section. */
1086: if ((req->q.sec = strrchr(req->q.query, '.')) != NULL) {
1087: if(isdigit((unsigned char)req->q.sec[1])) {
1088: *req->q.sec++ = '\0';
1089: req->q.sec = mandoc_strdup(req->q.sec);
1090: } else
1091: req->q.sec = NULL;
1092: }
1093:
1094: /* Handle the case of name[.section] only. */
1.135 schwarze 1095: if (req->q.manpath == NULL)
1.117 schwarze 1096: return;
1.135 schwarze 1097: req->q.query = mandoc_strdup(req->q.query);
1098:
1099: /* Split directory components. */
1100: dir[i = 0] = req->q.manpath;
1101: while ((dir[i + 1] = strchr(dir[i], '/')) != NULL) {
1102: if (++i == 3) {
1103: pg_error_badrequest(
1104: "You specified too many directory components.");
1105: exit(EXIT_FAILURE);
1106: }
1107: *dir[i]++ = '\0';
1.117 schwarze 1108: }
1109:
1.135 schwarze 1110: /* Optional manpath. */
1111: if ((i = validate_manpath(req, req->q.manpath)) == 0)
1112: req->q.manpath = NULL;
1113: else if (dir[1] == NULL)
1114: return;
1.117 schwarze 1115:
1.135 schwarze 1116: /* Optional section. */
1117: if (strncmp(dir[i], "man", 3) == 0) {
1.127 schwarze 1118: free(req->q.sec);
1.135 schwarze 1119: req->q.sec = mandoc_strdup(dir[i++] + 3);
1.117 schwarze 1120: }
1.135 schwarze 1121: if (dir[i] == NULL) {
1122: if (req->q.manpath == NULL)
1123: free(dir[0]);
1124: return;
1125: }
1126: if (dir[i + 1] != NULL) {
1127: pg_error_badrequest(
1128: "You specified an invalid directory component.");
1129: exit(EXIT_FAILURE);
1130: }
1131:
1132: /* Optional architecture. */
1133: if (i) {
1134: req->q.arch = mandoc_strdup(dir[i]);
1135: if (req->q.manpath == NULL)
1136: free(dir[0]);
1137: } else
1138: req->q.arch = dir[0];
1.24 kristaps 1139: }
1140:
1141: /*
1142: * Scan for indexable paths.
1143: */
1144: static void
1.129 schwarze 1145: parse_manpath_conf(struct req *req)
1.24 kristaps 1146: {
1.54 schwarze 1147: FILE *fp;
1148: char *dp;
1149: size_t dpsz;
1.115 schwarze 1150: ssize_t len;
1.54 schwarze 1151:
1.128 schwarze 1152: if ((fp = fopen("manpath.conf", "r")) == NULL) {
1153: warn("%s/manpath.conf", MAN_DIR);
1.74 schwarze 1154: pg_error_internal();
1155: exit(EXIT_FAILURE);
1156: }
1.24 kristaps 1157:
1.115 schwarze 1158: dp = NULL;
1159: dpsz = 0;
1160:
1161: while ((len = getline(&dp, &dpsz, fp)) != -1) {
1162: if (dp[len - 1] == '\n')
1163: dp[--len] = '\0';
1.54 schwarze 1164: req->p = mandoc_realloc(req->p,
1165: (req->psz + 1) * sizeof(char *));
1.80 schwarze 1166: if ( ! validate_urifrag(dp)) {
1.128 schwarze 1167: warnx("%s/manpath.conf contains "
1168: "unsafe path \"%s\"", MAN_DIR, dp);
1.80 schwarze 1169: pg_error_internal();
1170: exit(EXIT_FAILURE);
1171: }
1.128 schwarze 1172: if (strchr(dp, '/') != NULL) {
1173: warnx("%s/manpath.conf contains "
1174: "path with slash \"%s\"", MAN_DIR, dp);
1.80 schwarze 1175: pg_error_internal();
1176: exit(EXIT_FAILURE);
1177: }
1178: req->p[req->psz++] = dp;
1.115 schwarze 1179: dp = NULL;
1180: dpsz = 0;
1.74 schwarze 1181: }
1.115 schwarze 1182: free(dp);
1.74 schwarze 1183:
1.128 schwarze 1184: if (req->p == NULL) {
1185: warnx("%s/manpath.conf is empty", MAN_DIR);
1.74 schwarze 1186: pg_error_internal();
1187: exit(EXIT_FAILURE);
1.24 kristaps 1188: }
1189: }
CVSweb