Annotation of mandoc/cgi.c, Revision 1.133
1.133 ! schwarze 1: /* $Id: cgi.c,v 1.132 2016/05/28 13:40:48 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[] = {
116: "amd64", "alpha", "armish", "armv7",
1.125 schwarze 117: "hppa", "hppa64", "i386", "landisk",
118: "loongson", "luna88k", "macppc", "mips64",
119: "octeon", "sgi", "socppc", "sparc",
120: "sparc64", "zaurus",
1.68 schwarze 121: "amiga", "arc", "arm32", "atari",
1.125 schwarze 122: "aviion", "beagle", "cats", "hp300",
123: "ia64", "mac68k", "mvme68k", "mvme88k",
124: "mvmeppc", "palm", "pc532", "pegasos",
125: "pmax", "powerpc", "solbourne", "sun3",
126: "vax", "wgrisc", "x68k"
1.68 schwarze 127: };
128: static const int arch_MAX = sizeof(arch_names) / sizeof(char *);
129:
1.6 kristaps 130: /*
1.20 kristaps 131: * Print a character, escaping HTML along the way.
132: * This will pass non-ASCII straight to output: be warned!
133: */
1.10 kristaps 134: static void
135: html_putchar(char c)
136: {
137:
138: switch (c) {
139: case ('"'):
140: printf(""e;");
141: break;
142: case ('&'):
143: printf("&");
144: break;
145: case ('>'):
146: printf(">");
147: break;
148: case ('<'):
149: printf("<");
150: break;
151: default:
152: putchar((unsigned char)c);
153: break;
154: }
155: }
1.57 schwarze 156:
1.6 kristaps 157: /*
1.20 kristaps 158: * Call through to html_putchar().
159: * Accepts NULL strings.
1.6 kristaps 160: */
1.1 kristaps 161: static void
1.6 kristaps 162: html_print(const char *p)
1.1 kristaps 163: {
1.104 schwarze 164:
1.6 kristaps 165: if (NULL == p)
166: return;
1.1 kristaps 167: while ('\0' != *p)
1.10 kristaps 168: html_putchar(*p++);
1.1 kristaps 169: }
170:
171: /*
1.83 schwarze 172: * Transfer the responsibility for the allocated string *val
173: * to the query structure.
1.1 kristaps 174: */
175: static void
1.83 schwarze 176: set_query_attr(char **attr, char **val)
1.1 kristaps 177: {
178:
1.83 schwarze 179: free(*attr);
180: if (**val == '\0') {
181: *attr = NULL;
182: free(*val);
183: } else
184: *attr = *val;
185: *val = NULL;
186: }
187:
188: /*
189: * Parse the QUERY_STRING for key-value pairs
190: * and store the values into the query structure.
191: */
192: static void
1.129 schwarze 193: parse_query_string(struct req *req, const char *qs)
1.83 schwarze 194: {
195: char *key, *val;
196: size_t keysz, valsz;
197:
1.121 schwarze 198: req->isquery = 1;
1.83 schwarze 199: req->q.manpath = NULL;
200: req->q.arch = NULL;
201: req->q.sec = NULL;
1.85 schwarze 202: req->q.query = NULL;
1.83 schwarze 203: req->q.equal = 1;
204:
205: key = val = NULL;
206: while (*qs != '\0') {
1.24 kristaps 207:
1.83 schwarze 208: /* Parse one key. */
209:
210: keysz = strcspn(qs, "=;&");
211: key = mandoc_strndup(qs, keysz);
212: qs += keysz;
213: if (*qs != '=')
214: goto next;
215:
216: /* Parse one value. */
217:
218: valsz = strcspn(++qs, ";&");
219: val = mandoc_strndup(qs, valsz);
220: qs += valsz;
221:
222: /* Decode and catch encoding errors. */
1.1 kristaps 223:
1.83 schwarze 224: if ( ! (http_decode(key) && http_decode(val)))
225: goto next;
1.1 kristaps 226:
1.83 schwarze 227: /* Handle key-value pairs. */
1.1 kristaps 228:
1.83 schwarze 229: if ( ! strcmp(key, "query"))
1.85 schwarze 230: set_query_attr(&req->q.query, &val);
1.1 kristaps 231:
1.83 schwarze 232: else if ( ! strcmp(key, "apropos"))
233: req->q.equal = !strcmp(val, "0");
234:
235: else if ( ! strcmp(key, "manpath")) {
1.73 schwarze 236: #ifdef COMPAT_OLDURI
1.83 schwarze 237: if ( ! strncmp(val, "OpenBSD ", 8)) {
1.73 schwarze 238: val[7] = '-';
239: if ('C' == val[8])
240: val[8] = 'c';
241: }
242: #endif
1.83 schwarze 243: set_query_attr(&req->q.manpath, &val);
244: }
245:
246: else if ( ! (strcmp(key, "sec")
1.73 schwarze 247: #ifdef COMPAT_OLDURI
1.83 schwarze 248: && strcmp(key, "sektion")
1.73 schwarze 249: #endif
1.83 schwarze 250: )) {
251: if ( ! strcmp(val, "0"))
252: *val = '\0';
253: set_query_attr(&req->q.sec, &val);
1.65 schwarze 254: }
1.83 schwarze 255:
256: else if ( ! strcmp(key, "arch")) {
257: if ( ! strcmp(val, "default"))
258: *val = '\0';
259: set_query_attr(&req->q.arch, &val);
260: }
261:
262: /*
263: * The key must be freed in any case.
264: * The val may have been handed over to the query
265: * structure, in which case it is now NULL.
266: */
267: next:
268: free(key);
269: key = NULL;
270: free(val);
271: val = NULL;
272:
273: if (*qs != '\0')
274: qs++;
1.24 kristaps 275: }
1.1 kristaps 276: }
277:
278: /*
1.6 kristaps 279: * HTTP-decode a string. The standard explanation is that this turns
280: * "%4e+foo" into "n foo" in the regular way. This is done in-place
281: * over the allocated string.
1.1 kristaps 282: */
283: static int
1.24 kristaps 284: http_decode(char *p)
1.1 kristaps 285: {
286: char hex[3];
1.63 schwarze 287: char *q;
1.1 kristaps 288: int c;
289:
290: hex[2] = '\0';
291:
1.63 schwarze 292: q = p;
293: for ( ; '\0' != *p; p++, q++) {
1.1 kristaps 294: if ('%' == *p) {
295: if ('\0' == (hex[0] = *(p + 1)))
1.109 schwarze 296: return 0;
1.1 kristaps 297: if ('\0' == (hex[1] = *(p + 2)))
1.109 schwarze 298: return 0;
1.1 kristaps 299: if (1 != sscanf(hex, "%x", &c))
1.109 schwarze 300: return 0;
1.1 kristaps 301: if ('\0' == c)
1.109 schwarze 302: return 0;
1.1 kristaps 303:
1.63 schwarze 304: *q = (char)c;
305: p += 2;
1.1 kristaps 306: } else
1.63 schwarze 307: *q = '+' == *p ? ' ' : *p;
1.1 kristaps 308: }
309:
1.63 schwarze 310: *q = '\0';
1.109 schwarze 311: return 1;
1.1 kristaps 312: }
313:
1.6 kristaps 314: static void
315: resp_begin_http(int code, const char *msg)
316: {
317:
318: if (200 != code)
1.62 schwarze 319: printf("Status: %d %s\r\n", code, msg);
1.6 kristaps 320:
1.62 schwarze 321: printf("Content-Type: text/html; charset=utf-8\r\n"
322: "Cache-Control: no-cache\r\n"
323: "Pragma: no-cache\r\n"
324: "\r\n");
1.6 kristaps 325:
326: fflush(stdout);
327: }
328:
329: static void
1.114 schwarze 330: resp_copy(const char *filename)
331: {
332: char buf[4096];
333: ssize_t sz;
334: int fd;
335:
336: if ((fd = open(filename, O_RDONLY)) != -1) {
337: fflush(stdout);
338: while ((sz = read(fd, buf, sizeof(buf))) > 0)
339: write(STDOUT_FILENO, buf, sz);
340: }
341: }
342:
343: static void
1.6 kristaps 344: resp_begin_html(int code, const char *msg)
345: {
346:
347: resp_begin_http(code, msg);
348:
1.98 kristaps 349: printf("<!DOCTYPE html>\n"
1.126 schwarze 350: "<html>\n"
351: "<head>\n"
352: "<meta charset=\"UTF-8\"/>\n"
353: "<link rel=\"stylesheet\" href=\"%s/mandoc.css\""
354: " type=\"text/css\" media=\"all\">\n"
355: "<title>%s</title>\n"
356: "</head>\n"
357: "<body>\n"
1.58 schwarze 358: "<!-- Begin page content. //-->\n",
1.113 schwarze 359: CSS_DIR, CUSTOMIZE_TITLE);
1.114 schwarze 360:
361: resp_copy(MAN_DIR "/header.html");
1.6 kristaps 362: }
363:
364: static void
365: resp_end_html(void)
366: {
367:
1.114 schwarze 368: resp_copy(MAN_DIR "/footer.html");
369:
1.126 schwarze 370: puts("</body>\n"
371: "</html>");
1.6 kristaps 372: }
373:
374: static void
1.131 schwarze 375: resp_searchform(const struct req *req, enum focus focus)
1.6 kristaps 376: {
1.27 kristaps 377: int i;
1.13 kristaps 378:
1.6 kristaps 379: puts("<!-- Begin search form. //-->");
1.126 schwarze 380: printf("<div id=\"mancgi\">\n"
381: "<form action=\"/%s\" method=\"get\">\n"
382: "<fieldset>\n"
383: "<legend>Manual Page Search Parameters</legend>\n",
1.58 schwarze 384: scriptname);
1.68 schwarze 385:
386: /* Write query input box. */
387:
1.132 schwarze 388: printf("<input type=\"text\" name=\"query\" value=\"");
1.131 schwarze 389: if (req->q.query != NULL)
1.85 schwarze 390: html_print(req->q.query);
1.131 schwarze 391: printf( "\" size=\"40\"");
392: if (focus == FOCUS_QUERY)
393: printf(" autofocus");
394: puts(">");
1.68 schwarze 395:
1.132 schwarze 396: /* Write submission buttons. */
1.68 schwarze 397:
1.132 schwarze 398: printf( "<button type=\"submit\" name=\"apropos\" value=\"0\">"
399: "man</button>\n"
400: "<button type=\"submit\" name=\"apropos\" value=\"1\">"
401: "apropos</button>\n<br/>\n");
1.68 schwarze 402:
403: /* Write section selector. */
404:
1.132 schwarze 405: puts("<select name=\"sec\">");
1.68 schwarze 406: for (i = 0; i < sec_MAX; i++) {
1.126 schwarze 407: printf("<option value=\"%s\"", sec_numbers[i]);
1.68 schwarze 408: if (NULL != req->q.sec &&
409: 0 == strcmp(sec_numbers[i], req->q.sec))
1.126 schwarze 410: printf(" selected=\"selected\"");
411: printf(">%s</option>\n", sec_names[i]);
1.68 schwarze 412: }
1.126 schwarze 413: puts("</select>");
1.68 schwarze 414:
415: /* Write architecture selector. */
416:
1.126 schwarze 417: printf( "<select name=\"arch\">\n"
418: "<option value=\"default\"");
1.81 schwarze 419: if (NULL == req->q.arch)
1.126 schwarze 420: printf(" selected=\"selected\"");
421: puts(">All Architectures</option>");
1.68 schwarze 422: for (i = 0; i < arch_MAX; i++) {
1.126 schwarze 423: printf("<option value=\"%s\"", arch_names[i]);
1.68 schwarze 424: if (NULL != req->q.arch &&
425: 0 == strcmp(arch_names[i], req->q.arch))
1.126 schwarze 426: printf(" selected=\"selected\"");
427: printf(">%s</option>\n", arch_names[i]);
1.68 schwarze 428: }
1.126 schwarze 429: puts("</select>");
1.68 schwarze 430:
431: /* Write manpath selector. */
432:
1.27 kristaps 433: if (req->psz > 1) {
1.126 schwarze 434: puts("<select name=\"manpath\">");
1.27 kristaps 435: for (i = 0; i < (int)req->psz; i++) {
1.126 schwarze 436: printf("<option ");
1.102 schwarze 437: if (strcmp(req->q.manpath, req->p[i]) == 0)
1.126 schwarze 438: printf("selected=\"selected\" ");
439: printf("value=\"");
1.52 schwarze 440: html_print(req->p[i]);
1.27 kristaps 441: printf("\">");
1.52 schwarze 442: html_print(req->p[i]);
1.126 schwarze 443: puts("</option>");
1.27 kristaps 444: }
1.126 schwarze 445: puts("</select>");
1.27 kristaps 446: }
1.68 schwarze 447:
1.132 schwarze 448: puts("</fieldset>\n"
1.126 schwarze 449: "</form>\n"
450: "</div>");
1.20 kristaps 451: puts("<!-- End search form. //-->");
1.6 kristaps 452: }
453:
1.76 schwarze 454: static int
1.80 schwarze 455: validate_urifrag(const char *frag)
456: {
457:
458: while ('\0' != *frag) {
459: if ( ! (isalnum((unsigned char)*frag) ||
460: '-' == *frag || '.' == *frag ||
461: '/' == *frag || '_' == *frag))
1.109 schwarze 462: return 0;
1.80 schwarze 463: frag++;
464: }
1.109 schwarze 465: return 1;
1.80 schwarze 466: }
467:
468: static int
1.77 schwarze 469: validate_manpath(const struct req *req, const char* manpath)
470: {
471: size_t i;
472:
473: if ( ! strcmp(manpath, "mandoc"))
1.109 schwarze 474: return 1;
1.77 schwarze 475:
476: for (i = 0; i < req->psz; i++)
477: if ( ! strcmp(manpath, req->p[i]))
1.109 schwarze 478: return 1;
1.77 schwarze 479:
1.109 schwarze 480: return 0;
1.77 schwarze 481: }
482:
483: static int
1.76 schwarze 484: validate_filename(const char *file)
485: {
486:
487: if ('.' == file[0] && '/' == file[1])
488: file += 2;
489:
1.109 schwarze 490: return ! (strstr(file, "../") || strstr(file, "/..") ||
491: (strncmp(file, "man", 3) && strncmp(file, "cat", 3)));
1.76 schwarze 492: }
493:
1.6 kristaps 494: static void
1.72 schwarze 495: pg_index(const struct req *req)
1.6 kristaps 496: {
497:
498: resp_begin_html(200, NULL);
1.131 schwarze 499: resp_searchform(req, FOCUS_QUERY);
1.126 schwarze 500: printf("<p>\n"
1.86 schwarze 501: "This web interface is documented in the\n"
1.126 schwarze 502: "<a href=\"/%s%smandoc/man8/man.cgi.8\">man.cgi</a>\n"
1.86 schwarze 503: "manual, and the\n"
1.126 schwarze 504: "<a href=\"/%s%smandoc/man1/apropos.1\">apropos</a>\n"
1.69 schwarze 505: "manual explains the query syntax.\n"
1.126 schwarze 506: "</p>\n",
1.119 schwarze 507: scriptname, *scriptname == '\0' ? "" : "/",
508: scriptname, *scriptname == '\0' ? "" : "/");
1.6 kristaps 509: resp_end_html();
510: }
511:
512: static void
1.72 schwarze 513: pg_noresult(const struct req *req, const char *msg)
1.59 schwarze 514: {
515: resp_begin_html(200, NULL);
1.131 schwarze 516: resp_searchform(req, FOCUS_QUERY);
1.126 schwarze 517: puts("<p>");
1.59 schwarze 518: puts(msg);
1.126 schwarze 519: puts("</p>");
1.59 schwarze 520: resp_end_html();
521: }
522:
523: static void
1.72 schwarze 524: pg_error_badrequest(const char *msg)
1.9 kristaps 525: {
526:
1.59 schwarze 527: resp_begin_html(400, "Bad Request");
1.126 schwarze 528: puts("<h1>Bad Request</h1>\n"
529: "<p>\n");
1.59 schwarze 530: puts(msg);
531: printf("Try again from the\n"
1.126 schwarze 532: "<a href=\"/%s\">main page</a>.\n"
533: "</p>", scriptname);
1.9 kristaps 534: resp_end_html();
535: }
536:
537: static void
1.72 schwarze 538: pg_error_internal(void)
1.7 kristaps 539: {
540: resp_begin_html(500, "Internal Server Error");
1.126 schwarze 541: puts("<p>Internal Server Error</p>");
1.7 kristaps 542: resp_end_html();
543: }
544:
545: static void
1.72 schwarze 546: pg_searchres(const struct req *req, struct manpage *r, size_t sz)
1.1 kristaps 547: {
1.81 schwarze 548: char *arch, *archend;
1.120 schwarze 549: const char *sec;
550: size_t i, iuse;
1.81 schwarze 551: int archprio, archpriouse;
1.70 schwarze 552: int prio, priouse;
1.19 kristaps 553:
1.76 schwarze 554: for (i = 0; i < sz; i++) {
555: if (validate_filename(r[i].file))
556: continue;
1.128 schwarze 557: warnx("invalid filename %s in %s database",
1.76 schwarze 558: r[i].file, req->q.manpath);
559: pg_error_internal();
560: return;
561: }
562:
1.121 schwarze 563: if (req->isquery && sz == 1) {
1.6 kristaps 564: /*
565: * If we have just one result, then jump there now
566: * without any delay.
567: */
1.62 schwarze 568: printf("Status: 303 See Other\r\n");
1.119 schwarze 569: printf("Location: http://%s/%s%s%s/%s",
570: HTTP_HOST, scriptname,
571: *scriptname == '\0' ? "" : "/",
572: req->q.manpath, r[0].file);
1.62 schwarze 573: printf("\r\n"
574: "Content-Type: text/html; charset=utf-8\r\n"
575: "\r\n");
1.6 kristaps 576: return;
577: }
578:
1.12 kristaps 579: resp_begin_html(200, NULL);
1.131 schwarze 580: resp_searchform(req,
581: req->q.equal || sz == 1 ? FOCUS_NONE : FOCUS_QUERY);
1.1 kristaps 582:
1.123 schwarze 583: if (sz > 1) {
1.126 schwarze 584: puts("<div class=\"results\">");
585: puts("<table>");
1.123 schwarze 586:
587: for (i = 0; i < sz; i++) {
1.126 schwarze 588: printf("<tr>\n"
589: "<td class=\"title\">\n"
590: "<a href=\"/%s%s%s/%s",
1.123 schwarze 591: scriptname, *scriptname == '\0' ? "" : "/",
592: req->q.manpath, r[i].file);
593: printf("\">");
594: html_print(r[i].names);
1.126 schwarze 595: printf("</a>\n"
596: "</td>\n"
597: "<td class=\"desc\">");
1.123 schwarze 598: html_print(r[i].output);
1.126 schwarze 599: puts("</td>\n"
600: "</tr>");
1.123 schwarze 601: }
602:
1.126 schwarze 603: puts("</table>\n"
604: "</div>");
1.1 kristaps 605: }
1.16 kristaps 606:
1.70 schwarze 607: /*
608: * In man(1) mode, show one of the pages
609: * even if more than one is found.
610: */
611:
1.123 schwarze 612: if (req->q.equal || sz == 1) {
1.126 schwarze 613: puts("<hr>");
1.70 schwarze 614: iuse = 0;
1.120 schwarze 615: priouse = 20;
1.81 schwarze 616: archpriouse = 3;
1.70 schwarze 617: for (i = 0; i < sz; i++) {
1.120 schwarze 618: sec = r[i].file;
619: sec += strcspn(sec, "123456789");
620: if (sec[0] == '\0')
1.70 schwarze 621: continue;
1.120 schwarze 622: prio = sec_prios[sec[0] - '1'];
623: if (sec[1] != '/')
624: prio += 10;
625: if (req->q.arch == NULL) {
1.81 schwarze 626: archprio =
1.120 schwarze 627: ((arch = strchr(sec + 1, '/'))
628: == NULL) ? 3 :
629: ((archend = strchr(arch + 1, '/'))
630: == NULL) ? 0 :
1.81 schwarze 631: strncmp(arch, "amd64/",
632: archend - arch) ? 2 : 1;
633: if (archprio < archpriouse) {
634: archpriouse = archprio;
635: priouse = prio;
636: iuse = i;
637: continue;
638: }
639: if (archprio > archpriouse)
640: continue;
641: }
1.70 schwarze 642: if (prio >= priouse)
643: continue;
644: priouse = prio;
645: iuse = i;
646: }
647: resp_show(req, r[iuse].file);
648: }
649:
1.6 kristaps 650: resp_end_html();
651: }
652:
1.1 kristaps 653: static void
1.129 schwarze 654: resp_catman(const struct req *req, const char *file)
1.9 kristaps 655: {
1.10 kristaps 656: FILE *f;
1.115 schwarze 657: char *p;
658: size_t sz;
659: ssize_t len;
1.10 kristaps 660: int i;
661: int italic, bold;
1.9 kristaps 662:
1.115 schwarze 663: if ((f = fopen(file, "r")) == NULL) {
1.126 schwarze 664: puts("<p>You specified an invalid manual file.</p>");
1.9 kristaps 665: return;
666: }
667:
1.126 schwarze 668: puts("<div class=\"catman\">\n"
669: "<pre>");
1.10 kristaps 670:
1.115 schwarze 671: p = NULL;
672: sz = 0;
673:
674: while ((len = getline(&p, &sz, f)) != -1) {
1.10 kristaps 675: bold = italic = 0;
1.115 schwarze 676: for (i = 0; i < len - 1; i++) {
1.104 schwarze 677: /*
1.10 kristaps 678: * This means that the catpage is out of state.
679: * Ignore it and keep going (although the
680: * catpage is bogus).
681: */
682:
683: if ('\b' == p[i] || '\n' == p[i])
684: continue;
685:
686: /*
687: * Print a regular character.
688: * Close out any bold/italic scopes.
689: * If we're in back-space mode, make sure we'll
690: * have something to enter when we backspace.
691: */
692:
693: if ('\b' != p[i + 1]) {
694: if (italic)
1.126 schwarze 695: printf("</i>");
1.10 kristaps 696: if (bold)
1.126 schwarze 697: printf("</b>");
1.10 kristaps 698: italic = bold = 0;
699: html_putchar(p[i]);
700: continue;
1.115 schwarze 701: } else if (i + 2 >= len)
1.10 kristaps 702: continue;
703:
704: /* Italic mode. */
705:
706: if ('_' == p[i]) {
707: if (bold)
1.126 schwarze 708: printf("</b>");
1.10 kristaps 709: if ( ! italic)
1.126 schwarze 710: printf("<i>");
1.10 kristaps 711: bold = 0;
712: italic = 1;
713: i += 2;
714: html_putchar(p[i]);
715: continue;
716: }
717:
1.104 schwarze 718: /*
1.10 kristaps 719: * Handle funny behaviour troff-isms.
720: * These grok'd from the original man2html.c.
721: */
722:
723: if (('+' == p[i] && 'o' == p[i + 2]) ||
724: ('o' == 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: ('|' == p[i] && '*' == p[i + 2])) {
731: if (italic)
1.126 schwarze 732: printf("</i>");
1.10 kristaps 733: if (bold)
1.126 schwarze 734: printf("</b>");
1.10 kristaps 735: italic = bold = 0;
736: putchar('*');
737: i += 2;
738: continue;
739: } else if (('|' == p[i] && '-' == p[i + 2]) ||
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: ('|' == p[i] && '+' == p[i + 1])) {
745: if (italic)
1.126 schwarze 746: printf("</i>");
1.10 kristaps 747: if (bold)
1.126 schwarze 748: printf("</b>");
1.10 kristaps 749: italic = bold = 0;
750: putchar('+');
751: i += 2;
752: continue;
753: }
754:
755: /* Bold mode. */
1.104 schwarze 756:
1.10 kristaps 757: if (italic)
1.126 schwarze 758: printf("</i>");
1.10 kristaps 759: if ( ! bold)
1.126 schwarze 760: printf("<b>");
1.10 kristaps 761: bold = 1;
762: italic = 0;
763: i += 2;
764: html_putchar(p[i]);
765: }
766:
1.104 schwarze 767: /*
1.10 kristaps 768: * Clean up the last character.
1.104 schwarze 769: * We can get to a newline; don't print that.
1.10 kristaps 770: */
1.9 kristaps 771:
1.10 kristaps 772: if (italic)
1.126 schwarze 773: printf("</i>");
1.10 kristaps 774: if (bold)
1.126 schwarze 775: printf("</b>");
1.9 kristaps 776:
1.115 schwarze 777: if (i == len - 1 && p[i] != '\n')
1.10 kristaps 778: html_putchar(p[i]);
1.9 kristaps 779:
1.10 kristaps 780: putchar('\n');
781: }
1.115 schwarze 782: free(p);
1.10 kristaps 783:
1.126 schwarze 784: puts("</pre>\n"
785: "</div>");
1.10 kristaps 786:
787: fclose(f);
1.9 kristaps 788: }
789:
790: static void
1.129 schwarze 791: resp_format(const struct req *req, const char *file)
1.7 kristaps 792: {
1.106 schwarze 793: struct manoutput conf;
1.8 kristaps 794: struct mparse *mp;
1.107 schwarze 795: struct roff_man *man;
1.8 kristaps 796: void *vp;
1.90 schwarze 797: int fd;
798: int usepath;
1.7 kristaps 799:
1.8 kristaps 800: if (-1 == (fd = open(file, O_RDONLY, 0))) {
1.126 schwarze 801: puts("<p>You specified an invalid manual file.</p>");
1.7 kristaps 802: return;
803: }
804:
1.110 schwarze 805: mchars_alloc();
806: mp = mparse_alloc(MPARSE_SO, 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: }
887:
1.84 schwarze 888: if (strcmp(manpath, "mandoc")) {
889: free(req->q.manpath);
890: req->q.manpath = manpath;
891: } else
892: free(manpath);
893:
894: if ( ! validate_filename(file)) {
1.76 schwarze 895: pg_error_badrequest(
896: "You specified an invalid manual file.");
1.24 kristaps 897: return;
898: }
1.79 schwarze 899:
1.70 schwarze 900: resp_begin_html(200, NULL);
1.131 schwarze 901: resp_searchform(req, FOCUS_NONE);
1.84 schwarze 902: resp_show(req, file);
1.70 schwarze 903: resp_end_html();
1.6 kristaps 904: }
905:
906: static void
1.66 schwarze 907: pg_search(const struct req *req)
1.6 kristaps 908: {
1.52 schwarze 909: struct mansearch search;
910: struct manpaths paths;
911: struct manpage *res;
1.97 schwarze 912: char **argv;
913: char *query, *rp, *wp;
1.52 schwarze 914: size_t ressz;
1.97 schwarze 915: int argc;
1.6 kristaps 916:
917: /*
1.24 kristaps 918: * Begin by chdir()ing into the root of the manpath.
919: * This way we can pick up the database files, which are
920: * relative to the manpath root.
921: */
922:
1.128 schwarze 923: if (chdir(req->q.manpath) == -1) {
924: warn("chdir %s", req->q.manpath);
1.77 schwarze 925: pg_error_internal();
1.24 kristaps 926: return;
927: }
928:
1.52 schwarze 929: search.arch = req->q.arch;
930: search.sec = req->q.sec;
1.94 schwarze 931: search.outkey = "Nd";
932: search.argmode = req->q.equal ? ARG_NAME : ARG_EXPR;
1.101 schwarze 933: search.firstmatch = 1;
1.52 schwarze 934:
935: paths.sz = 1;
936: paths.paths = mandoc_malloc(sizeof(char *));
937: paths.paths[0] = mandoc_strdup(".");
1.24 kristaps 938:
939: /*
1.97 schwarze 940: * Break apart at spaces with backslash-escaping.
1.6 kristaps 941: */
942:
1.97 schwarze 943: argc = 0;
944: argv = NULL;
945: rp = query = mandoc_strdup(req->q.query);
946: for (;;) {
947: while (isspace((unsigned char)*rp))
948: rp++;
949: if (*rp == '\0')
950: break;
951: argv = mandoc_reallocarray(argv, argc + 1, sizeof(char *));
952: argv[argc++] = wp = rp;
953: for (;;) {
954: if (isspace((unsigned char)*rp)) {
955: *wp = '\0';
956: rp++;
957: break;
958: }
959: if (rp[0] == '\\' && rp[1] != '\0')
960: rp++;
961: if (wp != rp)
962: *wp = *rp;
963: if (*rp == '\0')
964: break;
965: wp++;
966: rp++;
967: }
1.6 kristaps 968: }
969:
1.97 schwarze 970: if (0 == mansearch(&search, &paths, argc, argv, &res, &ressz))
1.72 schwarze 971: pg_noresult(req, "You entered an invalid query.");
1.59 schwarze 972: else if (0 == ressz)
1.72 schwarze 973: pg_noresult(req, "No results found.");
1.59 schwarze 974: else
1.72 schwarze 975: pg_searchres(req, res, ressz);
1.6 kristaps 976:
1.97 schwarze 977: free(query);
978: mansearch_free(res, ressz);
1.52 schwarze 979: free(paths.paths[0]);
980: free(paths.paths);
1.1 kristaps 981: }
982:
983: int
984: main(void)
985: {
1.66 schwarze 986: struct req req;
1.95 schwarze 987: struct itimerval itimer;
1.66 schwarze 988: const char *path;
1.83 schwarze 989: const char *querystring;
1.1 kristaps 990: int i;
1.95 schwarze 991:
992: /* Poor man's ReDoS mitigation. */
993:
1.99 schwarze 994: itimer.it_value.tv_sec = 2;
1.95 schwarze 995: itimer.it_value.tv_usec = 0;
1.99 schwarze 996: itimer.it_interval.tv_sec = 2;
1.95 schwarze 997: itimer.it_interval.tv_usec = 0;
998: if (setitimer(ITIMER_VIRTUAL, &itimer, NULL) == -1) {
1.128 schwarze 999: warn("setitimer");
1.80 schwarze 1000: pg_error_internal();
1.109 schwarze 1001: return EXIT_FAILURE;
1.80 schwarze 1002: }
1003:
1.24 kristaps 1004: /*
1.67 schwarze 1005: * First we change directory into the MAN_DIR so that
1.24 kristaps 1006: * subsequent scanning for manpath directories is rooted
1007: * relative to the same position.
1008: */
1009:
1.128 schwarze 1010: if (chdir(MAN_DIR) == -1) {
1011: warn("MAN_DIR: %s", MAN_DIR);
1.72 schwarze 1012: pg_error_internal();
1.109 schwarze 1013: return EXIT_FAILURE;
1.104 schwarze 1014: }
1.24 kristaps 1015:
1016: memset(&req, 0, sizeof(struct req));
1.118 schwarze 1017: req.q.equal = 1;
1.129 schwarze 1018: parse_manpath_conf(&req);
1.1 kristaps 1019:
1.117 schwarze 1020: /* Parse the path info and the query string. */
1.1 kristaps 1021:
1.117 schwarze 1022: if ((path = getenv("PATH_INFO")) == NULL)
1023: path = "";
1024: else if (*path == '/')
1025: path++;
1026:
1.124 schwarze 1027: if (*path != '\0') {
1.129 schwarze 1028: parse_path_info(&req, path);
1.133 ! schwarze 1029: if (req.q.manpath == NULL || access(path, F_OK) == -1)
1.124 schwarze 1030: path = "";
1.117 schwarze 1031: } else if ((querystring = getenv("QUERY_STRING")) != NULL)
1.129 schwarze 1032: parse_query_string(&req, querystring);
1.77 schwarze 1033:
1.117 schwarze 1034: /* Validate parsed data and add defaults. */
1035:
1.102 schwarze 1036: if (req.q.manpath == NULL)
1037: req.q.manpath = mandoc_strdup(req.p[0]);
1038: else if ( ! validate_manpath(&req, req.q.manpath)) {
1.77 schwarze 1039: pg_error_badrequest(
1040: "You specified an invalid manpath.");
1.109 schwarze 1041: return EXIT_FAILURE;
1.77 schwarze 1042: }
1.1 kristaps 1043:
1.80 schwarze 1044: if ( ! (NULL == req.q.arch || validate_urifrag(req.q.arch))) {
1045: pg_error_badrequest(
1046: "You specified an invalid architecture.");
1.109 schwarze 1047: return EXIT_FAILURE;
1.80 schwarze 1048: }
1049:
1.66 schwarze 1050: /* Dispatch to the three different pages. */
1.1 kristaps 1051:
1.66 schwarze 1052: if ('\0' != *path)
1053: pg_show(&req, path);
1.85 schwarze 1054: else if (NULL != req.q.query)
1.66 schwarze 1055: pg_search(&req);
1056: else
1.72 schwarze 1057: pg_index(&req);
1.1 kristaps 1058:
1.83 schwarze 1059: free(req.q.manpath);
1060: free(req.q.arch);
1061: free(req.q.sec);
1.85 schwarze 1062: free(req.q.query);
1.52 schwarze 1063: for (i = 0; i < (int)req.psz; i++)
1064: free(req.p[i]);
1.24 kristaps 1065: free(req.p);
1.109 schwarze 1066: return EXIT_SUCCESS;
1.117 schwarze 1067: }
1068:
1069: /*
1070: * If PATH_INFO is not a file name, translate it to a query.
1071: */
1072: static void
1.129 schwarze 1073: parse_path_info(struct req *req, const char *path)
1.117 schwarze 1074: {
1.127 schwarze 1075: char *dir;
1.117 schwarze 1076:
1.121 schwarze 1077: req->isquery = 0;
1.117 schwarze 1078: req->q.equal = 1;
1079: req->q.manpath = mandoc_strdup(path);
1080:
1081: /* Mandatory manual page name. */
1082: if ((req->q.query = strrchr(req->q.manpath, '/')) == NULL) {
1083: req->q.query = req->q.manpath;
1084: req->q.manpath = NULL;
1085: } else
1086: *req->q.query++ = '\0';
1087:
1088: /* Optional trailing section. */
1089: if ((req->q.sec = strrchr(req->q.query, '.')) != NULL) {
1090: if(isdigit((unsigned char)req->q.sec[1])) {
1091: *req->q.sec++ = '\0';
1092: req->q.sec = mandoc_strdup(req->q.sec);
1093: } else
1094: req->q.sec = NULL;
1095: }
1096:
1097: /* Handle the case of name[.section] only. */
1098: if (req->q.manpath == NULL) {
1099: req->q.arch = NULL;
1100: return;
1101: }
1102: req->q.query = mandoc_strdup(req->q.query);
1103:
1104: /* Optional architecture. */
1.127 schwarze 1105: dir = strrchr(req->q.manpath, '/');
1106: if (dir != NULL && strncmp(dir + 1, "man", 3) != 0) {
1107: *dir++ = '\0';
1108: req->q.arch = mandoc_strdup(dir);
1109: dir = strrchr(req->q.manpath, '/');
1110: } else
1111: req->q.arch = NULL;
1.117 schwarze 1112:
1.127 schwarze 1113: /* Optional directory name. */
1114: if (dir != NULL && strncmp(dir + 1, "man", 3) == 0) {
1115: *dir++ = '\0';
1116: free(req->q.sec);
1117: req->q.sec = mandoc_strdup(dir + 3);
1.117 schwarze 1118: }
1.24 kristaps 1119: }
1120:
1121: /*
1122: * Scan for indexable paths.
1123: */
1124: static void
1.129 schwarze 1125: parse_manpath_conf(struct req *req)
1.24 kristaps 1126: {
1.54 schwarze 1127: FILE *fp;
1128: char *dp;
1129: size_t dpsz;
1.115 schwarze 1130: ssize_t len;
1.54 schwarze 1131:
1.128 schwarze 1132: if ((fp = fopen("manpath.conf", "r")) == NULL) {
1133: warn("%s/manpath.conf", MAN_DIR);
1.74 schwarze 1134: pg_error_internal();
1135: exit(EXIT_FAILURE);
1136: }
1.24 kristaps 1137:
1.115 schwarze 1138: dp = NULL;
1139: dpsz = 0;
1140:
1141: while ((len = getline(&dp, &dpsz, fp)) != -1) {
1142: if (dp[len - 1] == '\n')
1143: dp[--len] = '\0';
1.54 schwarze 1144: req->p = mandoc_realloc(req->p,
1145: (req->psz + 1) * sizeof(char *));
1.80 schwarze 1146: if ( ! validate_urifrag(dp)) {
1.128 schwarze 1147: warnx("%s/manpath.conf contains "
1148: "unsafe path \"%s\"", MAN_DIR, dp);
1.80 schwarze 1149: pg_error_internal();
1150: exit(EXIT_FAILURE);
1151: }
1.128 schwarze 1152: if (strchr(dp, '/') != NULL) {
1153: warnx("%s/manpath.conf contains "
1154: "path with slash \"%s\"", MAN_DIR, dp);
1.80 schwarze 1155: pg_error_internal();
1156: exit(EXIT_FAILURE);
1157: }
1158: req->p[req->psz++] = dp;
1.115 schwarze 1159: dp = NULL;
1160: dpsz = 0;
1.74 schwarze 1161: }
1.115 schwarze 1162: free(dp);
1.74 schwarze 1163:
1.128 schwarze 1164: if (req->p == NULL) {
1165: warnx("%s/manpath.conf is empty", MAN_DIR);
1.74 schwarze 1166: pg_error_internal();
1167: exit(EXIT_FAILURE);
1.24 kristaps 1168: }
1169: }
CVSweb