Annotation of mandoc/cgi.c, Revision 1.148
1.148 ! schwarze 1: /* $Id: cgi.c,v 1.147 2017/02/08 13:34:27 schwarze Exp $ */
1.6 kristaps 2: /*
1.42 kristaps 3: * Copyright (c) 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
1.142 schwarze 4: * Copyright (c) 2014, 2015, 2016, 2017 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>
1.147 schwarze 24: #if HAVE_ERR
1.128 schwarze 25: #include <err.h>
1.147 schwarze 26: #endif
1.58 schwarze 27: #include <errno.h>
1.1 kristaps 28: #include <fcntl.h>
1.6 kristaps 29: #include <limits.h>
1.92 schwarze 30: #include <stdint.h>
1.1 kristaps 31: #include <stdio.h>
32: #include <stdlib.h>
33: #include <string.h>
1.6 kristaps 34: #include <unistd.h>
1.1 kristaps 35:
1.108 schwarze 36: #include "mandoc_aux.h"
1.4 schwarze 37: #include "mandoc.h"
1.108 schwarze 38: #include "roff.h"
1.111 schwarze 39: #include "mdoc.h"
1.112 schwarze 40: #include "man.h"
1.8 kristaps 41: #include "main.h"
1.105 schwarze 42: #include "manconf.h"
1.52 schwarze 43: #include "mansearch.h"
1.67 schwarze 44: #include "cgi.h"
1.1 kristaps 45:
1.20 kristaps 46: /*
47: * A query as passed to the search function.
48: */
49: struct query {
1.83 schwarze 50: char *manpath; /* desired manual directory */
51: char *arch; /* architecture */
52: char *sec; /* manual section */
1.85 schwarze 53: char *query; /* unparsed query expression */
1.65 schwarze 54: int equal; /* match whole names, not substrings */
1.20 kristaps 55: };
56:
1.1 kristaps 57: struct req {
1.58 schwarze 58: struct query q;
59: char **p; /* array of available manpaths */
60: size_t psz; /* number of available manpaths */
1.121 schwarze 61: int isquery; /* QUERY_STRING used, not PATH_INFO */
1.1 kristaps 62: };
63:
1.131 schwarze 64: enum focus {
65: FOCUS_NONE = 0,
66: FOCUS_QUERY
67: };
68:
1.6 kristaps 69: static void html_print(const char *);
1.10 kristaps 70: static void html_putchar(char);
1.104 schwarze 71: static int http_decode(char *);
1.129 schwarze 72: static void parse_manpath_conf(struct req *);
73: static void parse_path_info(struct req *req, const char *path);
74: static void parse_query_string(struct req *, const char *);
1.72 schwarze 75: static void pg_error_badrequest(const char *);
76: static void pg_error_internal(void);
77: static void pg_index(const struct req *);
78: static void pg_noresult(const struct req *, const char *);
1.66 schwarze 79: static void pg_search(const struct req *);
1.72 schwarze 80: static void pg_searchres(const struct req *,
81: struct manpage *, size_t);
1.79 schwarze 82: static void pg_show(struct req *, const char *);
1.6 kristaps 83: static void resp_begin_html(int, const char *);
84: static void resp_begin_http(int, const char *);
1.129 schwarze 85: static void resp_catman(const struct req *, const char *);
1.114 schwarze 86: static void resp_copy(const char *);
1.6 kristaps 87: static void resp_end_html(void);
1.129 schwarze 88: static void resp_format(const struct req *, const char *);
1.131 schwarze 89: static void resp_searchform(const struct req *, enum focus);
1.70 schwarze 90: static void resp_show(const struct req *, const char *);
1.85 schwarze 91: static void set_query_attr(char **, char **);
92: static int validate_filename(const char *);
93: static int validate_manpath(const struct req *, const char *);
94: static int validate_urifrag(const char *);
1.6 kristaps 95:
1.119 schwarze 96: static const char *scriptname = SCRIPT_NAME;
1.1 kristaps 97:
1.70 schwarze 98: static const int sec_prios[] = {1, 4, 5, 8, 6, 3, 7, 2, 9};
1.68 schwarze 99: static const char *const sec_numbers[] = {
100: "0", "1", "2", "3", "3p", "4", "5", "6", "7", "8", "9"
101: };
102: static const char *const sec_names[] = {
103: "All Sections",
104: "1 - General Commands",
105: "2 - System Calls",
1.96 schwarze 106: "3 - Library Functions",
107: "3p - Perl Library",
108: "4 - Device Drivers",
1.68 schwarze 109: "5 - File Formats",
110: "6 - Games",
1.96 schwarze 111: "7 - Miscellaneous Information",
112: "8 - System Manager\'s Manual",
113: "9 - Kernel Developer\'s Manual"
1.68 schwarze 114: };
115: static const int sec_MAX = sizeof(sec_names) / sizeof(char *);
116:
117: static const char *const arch_names[] = {
1.146 schwarze 118: "amd64", "alpha", "armv7", "arm64",
1.137 schwarze 119: "hppa", "i386", "landisk",
1.125 schwarze 120: "loongson", "luna88k", "macppc", "mips64",
1.140 schwarze 121: "octeon", "sgi", "socppc", "sparc64",
1.137 schwarze 122: "amiga", "arc", "armish", "arm32",
123: "atari", "aviion", "beagle", "cats",
124: "hppa64", "hp300",
1.125 schwarze 125: "ia64", "mac68k", "mvme68k", "mvme88k",
126: "mvmeppc", "palm", "pc532", "pegasos",
1.139 schwarze 127: "pmax", "powerpc", "solbourne", "sparc",
1.140 schwarze 128: "sun3", "vax", "wgrisc", "x68k",
129: "zaurus"
1.68 schwarze 130: };
131: static const int arch_MAX = sizeof(arch_names) / sizeof(char *);
132:
1.6 kristaps 133: /*
1.20 kristaps 134: * Print a character, escaping HTML along the way.
135: * This will pass non-ASCII straight to output: be warned!
136: */
1.10 kristaps 137: static void
138: html_putchar(char c)
139: {
140:
141: switch (c) {
142: case ('"'):
1.141 schwarze 143: printf(""");
1.10 kristaps 144: break;
145: case ('&'):
146: printf("&");
147: break;
148: case ('>'):
149: printf(">");
150: break;
151: case ('<'):
152: printf("<");
153: break;
154: default:
155: putchar((unsigned char)c);
156: break;
157: }
158: }
1.57 schwarze 159:
1.6 kristaps 160: /*
1.20 kristaps 161: * Call through to html_putchar().
162: * Accepts NULL strings.
1.6 kristaps 163: */
1.1 kristaps 164: static void
1.6 kristaps 165: html_print(const char *p)
1.1 kristaps 166: {
1.104 schwarze 167:
1.6 kristaps 168: if (NULL == p)
169: return;
1.1 kristaps 170: while ('\0' != *p)
1.10 kristaps 171: html_putchar(*p++);
1.1 kristaps 172: }
173:
174: /*
1.83 schwarze 175: * Transfer the responsibility for the allocated string *val
176: * to the query structure.
1.1 kristaps 177: */
178: static void
1.83 schwarze 179: set_query_attr(char **attr, char **val)
1.1 kristaps 180: {
181:
1.83 schwarze 182: free(*attr);
183: if (**val == '\0') {
184: *attr = NULL;
185: free(*val);
186: } else
187: *attr = *val;
188: *val = NULL;
189: }
190:
191: /*
192: * Parse the QUERY_STRING for key-value pairs
193: * and store the values into the query structure.
194: */
195: static void
1.129 schwarze 196: parse_query_string(struct req *req, const char *qs)
1.83 schwarze 197: {
198: char *key, *val;
199: size_t keysz, valsz;
200:
1.121 schwarze 201: req->isquery = 1;
1.83 schwarze 202: req->q.manpath = NULL;
203: req->q.arch = NULL;
204: req->q.sec = NULL;
1.85 schwarze 205: req->q.query = NULL;
1.83 schwarze 206: req->q.equal = 1;
207:
208: key = val = NULL;
209: while (*qs != '\0') {
1.24 kristaps 210:
1.83 schwarze 211: /* Parse one key. */
212:
213: keysz = strcspn(qs, "=;&");
214: key = mandoc_strndup(qs, keysz);
215: qs += keysz;
216: if (*qs != '=')
217: goto next;
218:
219: /* Parse one value. */
220:
221: valsz = strcspn(++qs, ";&");
222: val = mandoc_strndup(qs, valsz);
223: qs += valsz;
224:
225: /* Decode and catch encoding errors. */
1.1 kristaps 226:
1.83 schwarze 227: if ( ! (http_decode(key) && http_decode(val)))
228: goto next;
1.1 kristaps 229:
1.83 schwarze 230: /* Handle key-value pairs. */
1.1 kristaps 231:
1.83 schwarze 232: if ( ! strcmp(key, "query"))
1.85 schwarze 233: set_query_attr(&req->q.query, &val);
1.1 kristaps 234:
1.83 schwarze 235: else if ( ! strcmp(key, "apropos"))
236: req->q.equal = !strcmp(val, "0");
237:
238: else if ( ! strcmp(key, "manpath")) {
1.73 schwarze 239: #ifdef COMPAT_OLDURI
1.83 schwarze 240: if ( ! strncmp(val, "OpenBSD ", 8)) {
1.73 schwarze 241: val[7] = '-';
242: if ('C' == val[8])
243: val[8] = 'c';
244: }
245: #endif
1.83 schwarze 246: set_query_attr(&req->q.manpath, &val);
247: }
248:
249: else if ( ! (strcmp(key, "sec")
1.73 schwarze 250: #ifdef COMPAT_OLDURI
1.83 schwarze 251: && strcmp(key, "sektion")
1.73 schwarze 252: #endif
1.83 schwarze 253: )) {
254: if ( ! strcmp(val, "0"))
255: *val = '\0';
256: set_query_attr(&req->q.sec, &val);
1.65 schwarze 257: }
1.83 schwarze 258:
259: else if ( ! strcmp(key, "arch")) {
260: if ( ! strcmp(val, "default"))
261: *val = '\0';
262: set_query_attr(&req->q.arch, &val);
263: }
264:
265: /*
266: * The key must be freed in any case.
267: * The val may have been handed over to the query
268: * structure, in which case it is now NULL.
269: */
270: next:
271: free(key);
272: key = NULL;
273: free(val);
274: val = NULL;
275:
276: if (*qs != '\0')
277: qs++;
1.24 kristaps 278: }
1.1 kristaps 279: }
280:
281: /*
1.6 kristaps 282: * HTTP-decode a string. The standard explanation is that this turns
283: * "%4e+foo" into "n foo" in the regular way. This is done in-place
284: * over the allocated string.
1.1 kristaps 285: */
286: static int
1.24 kristaps 287: http_decode(char *p)
1.1 kristaps 288: {
289: char hex[3];
1.63 schwarze 290: char *q;
1.1 kristaps 291: int c;
292:
293: hex[2] = '\0';
294:
1.63 schwarze 295: q = p;
296: for ( ; '\0' != *p; p++, q++) {
1.1 kristaps 297: if ('%' == *p) {
298: if ('\0' == (hex[0] = *(p + 1)))
1.109 schwarze 299: return 0;
1.1 kristaps 300: if ('\0' == (hex[1] = *(p + 2)))
1.109 schwarze 301: return 0;
1.1 kristaps 302: if (1 != sscanf(hex, "%x", &c))
1.109 schwarze 303: return 0;
1.1 kristaps 304: if ('\0' == c)
1.109 schwarze 305: return 0;
1.1 kristaps 306:
1.63 schwarze 307: *q = (char)c;
308: p += 2;
1.1 kristaps 309: } else
1.63 schwarze 310: *q = '+' == *p ? ' ' : *p;
1.1 kristaps 311: }
312:
1.63 schwarze 313: *q = '\0';
1.109 schwarze 314: return 1;
1.1 kristaps 315: }
316:
1.6 kristaps 317: static void
318: resp_begin_http(int code, const char *msg)
319: {
320:
321: if (200 != code)
1.62 schwarze 322: printf("Status: %d %s\r\n", code, msg);
1.6 kristaps 323:
1.62 schwarze 324: printf("Content-Type: text/html; charset=utf-8\r\n"
325: "Cache-Control: no-cache\r\n"
326: "Pragma: no-cache\r\n"
327: "\r\n");
1.6 kristaps 328:
329: fflush(stdout);
330: }
331:
332: static void
1.114 schwarze 333: resp_copy(const char *filename)
334: {
335: char buf[4096];
336: ssize_t sz;
337: int fd;
338:
339: if ((fd = open(filename, O_RDONLY)) != -1) {
340: fflush(stdout);
341: while ((sz = read(fd, buf, sizeof(buf))) > 0)
342: write(STDOUT_FILENO, buf, sz);
1.138 schwarze 343: close(fd);
1.114 schwarze 344: }
345: }
346:
347: static void
1.6 kristaps 348: resp_begin_html(int code, const char *msg)
349: {
350:
351: resp_begin_http(code, msg);
352:
1.98 kristaps 353: printf("<!DOCTYPE html>\n"
1.126 schwarze 354: "<html>\n"
355: "<head>\n"
1.143 schwarze 356: " <meta charset=\"UTF-8\"/>\n"
357: " <link rel=\"stylesheet\" href=\"%s/mandoc.css\""
1.126 schwarze 358: " type=\"text/css\" media=\"all\">\n"
1.143 schwarze 359: " <title>%s</title>\n"
1.126 schwarze 360: "</head>\n"
1.142 schwarze 361: "<body>\n",
1.113 schwarze 362: CSS_DIR, CUSTOMIZE_TITLE);
1.114 schwarze 363:
364: resp_copy(MAN_DIR "/header.html");
1.6 kristaps 365: }
366:
367: static void
368: resp_end_html(void)
369: {
370:
1.114 schwarze 371: resp_copy(MAN_DIR "/footer.html");
372:
1.126 schwarze 373: puts("</body>\n"
374: "</html>");
1.6 kristaps 375: }
376:
377: static void
1.131 schwarze 378: resp_searchform(const struct req *req, enum focus focus)
1.6 kristaps 379: {
1.27 kristaps 380: int i;
1.13 kristaps 381:
1.142 schwarze 382: printf("<form action=\"/%s\" method=\"get\">\n"
1.143 schwarze 383: " <fieldset>\n"
384: " <legend>Manual Page Search Parameters</legend>\n",
1.58 schwarze 385: scriptname);
1.68 schwarze 386:
387: /* Write query input box. */
388:
1.143 schwarze 389: printf(" <input type=\"text\" name=\"query\" value=\"");
1.131 schwarze 390: if (req->q.query != NULL)
1.85 schwarze 391: html_print(req->q.query);
1.131 schwarze 392: printf( "\" size=\"40\"");
393: if (focus == FOCUS_QUERY)
394: printf(" autofocus");
395: puts(">");
1.68 schwarze 396:
1.132 schwarze 397: /* Write submission buttons. */
1.68 schwarze 398:
1.143 schwarze 399: printf( " <button type=\"submit\" name=\"apropos\" value=\"0\">"
1.132 schwarze 400: "man</button>\n"
1.143 schwarze 401: " <button type=\"submit\" name=\"apropos\" value=\"1\">"
402: "apropos</button>\n"
403: " <br/>\n");
1.68 schwarze 404:
405: /* Write section selector. */
406:
1.143 schwarze 407: puts(" <select name=\"sec\">");
1.68 schwarze 408: for (i = 0; i < sec_MAX; i++) {
1.143 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: }
1.143 schwarze 415: puts(" </select>");
1.68 schwarze 416:
417: /* Write architecture selector. */
418:
1.143 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++) {
1.143 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: }
1.143 schwarze 431: puts(" </select>");
1.68 schwarze 432:
433: /* Write manpath selector. */
434:
1.27 kristaps 435: if (req->psz > 1) {
1.143 schwarze 436: puts(" <select name=\"manpath\">");
1.27 kristaps 437: for (i = 0; i < (int)req->psz; i++) {
1.143 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: }
1.143 schwarze 447: puts(" </select>");
1.27 kristaps 448: }
1.68 schwarze 449:
1.143 schwarze 450: puts(" </fieldset>\n"
1.142 schwarze 451: "</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: for (i = 0; i < req->psz; i++)
474: if ( ! strcmp(manpath, req->p[i]))
1.109 schwarze 475: return 1;
1.77 schwarze 476:
1.109 schwarze 477: return 0;
1.77 schwarze 478: }
479:
480: static int
1.76 schwarze 481: validate_filename(const char *file)
482: {
483:
484: if ('.' == file[0] && '/' == file[1])
485: file += 2;
486:
1.109 schwarze 487: return ! (strstr(file, "../") || strstr(file, "/..") ||
488: (strncmp(file, "man", 3) && strncmp(file, "cat", 3)));
1.76 schwarze 489: }
490:
1.6 kristaps 491: static void
1.72 schwarze 492: pg_index(const struct req *req)
1.6 kristaps 493: {
494:
495: resp_begin_html(200, NULL);
1.131 schwarze 496: resp_searchform(req, FOCUS_QUERY);
1.126 schwarze 497: printf("<p>\n"
1.86 schwarze 498: "This web interface is documented in the\n"
1.144 schwarze 499: "<a class=\"Xr\" href=\"/%s%sman.cgi.8\">man.cgi(8)</a>\n"
1.86 schwarze 500: "manual, and the\n"
1.144 schwarze 501: "<a class=\"Xr\" href=\"/%s%sapropos.1\">apropos(1)</a>\n"
1.69 schwarze 502: "manual explains the query syntax.\n"
1.126 schwarze 503: "</p>\n",
1.119 schwarze 504: scriptname, *scriptname == '\0' ? "" : "/",
505: scriptname, *scriptname == '\0' ? "" : "/");
1.6 kristaps 506: resp_end_html();
507: }
508:
509: static void
1.72 schwarze 510: pg_noresult(const struct req *req, const char *msg)
1.59 schwarze 511: {
512: resp_begin_html(200, NULL);
1.131 schwarze 513: resp_searchform(req, FOCUS_QUERY);
1.126 schwarze 514: puts("<p>");
1.59 schwarze 515: puts(msg);
1.126 schwarze 516: puts("</p>");
1.59 schwarze 517: resp_end_html();
518: }
519:
520: static void
1.72 schwarze 521: pg_error_badrequest(const char *msg)
1.9 kristaps 522: {
523:
1.59 schwarze 524: resp_begin_html(400, "Bad Request");
1.126 schwarze 525: puts("<h1>Bad Request</h1>\n"
526: "<p>\n");
1.59 schwarze 527: puts(msg);
528: printf("Try again from the\n"
1.126 schwarze 529: "<a href=\"/%s\">main page</a>.\n"
530: "</p>", scriptname);
1.9 kristaps 531: resp_end_html();
532: }
533:
534: static void
1.72 schwarze 535: pg_error_internal(void)
1.7 kristaps 536: {
537: resp_begin_html(500, "Internal Server Error");
1.126 schwarze 538: puts("<p>Internal Server Error</p>");
1.7 kristaps 539: resp_end_html();
540: }
541:
542: static void
1.72 schwarze 543: pg_searchres(const struct req *req, struct manpage *r, size_t sz)
1.1 kristaps 544: {
1.81 schwarze 545: char *arch, *archend;
1.120 schwarze 546: const char *sec;
547: size_t i, iuse;
1.81 schwarze 548: int archprio, archpriouse;
1.70 schwarze 549: int prio, priouse;
1.19 kristaps 550:
1.76 schwarze 551: for (i = 0; i < sz; i++) {
552: if (validate_filename(r[i].file))
553: continue;
1.128 schwarze 554: warnx("invalid filename %s in %s database",
1.76 schwarze 555: r[i].file, req->q.manpath);
556: pg_error_internal();
557: return;
558: }
559:
1.121 schwarze 560: if (req->isquery && sz == 1) {
1.6 kristaps 561: /*
562: * If we have just one result, then jump there now
563: * without any delay.
564: */
1.62 schwarze 565: printf("Status: 303 See Other\r\n");
1.119 schwarze 566: printf("Location: http://%s/%s%s%s/%s",
567: HTTP_HOST, scriptname,
568: *scriptname == '\0' ? "" : "/",
569: req->q.manpath, r[0].file);
1.62 schwarze 570: printf("\r\n"
571: "Content-Type: text/html; charset=utf-8\r\n"
572: "\r\n");
1.6 kristaps 573: return;
574: }
575:
1.12 kristaps 576: resp_begin_html(200, NULL);
1.131 schwarze 577: resp_searchform(req,
578: req->q.equal || sz == 1 ? FOCUS_NONE : FOCUS_QUERY);
1.1 kristaps 579:
1.123 schwarze 580: if (sz > 1) {
1.144 schwarze 581: puts("<table class=\"results\">");
1.123 schwarze 582: for (i = 0; i < sz; i++) {
1.143 schwarze 583: printf(" <tr>\n"
1.144 schwarze 584: " <td>"
585: "<a class=\"Xr\" href=\"/%s%s%s/%s\">",
1.123 schwarze 586: scriptname, *scriptname == '\0' ? "" : "/",
587: req->q.manpath, r[i].file);
588: html_print(r[i].names);
1.143 schwarze 589: printf("</a></td>\n"
1.144 schwarze 590: " <td><span class=\"Nd\">");
1.123 schwarze 591: html_print(r[i].output);
1.144 schwarze 592: puts("</span></td>\n"
1.143 schwarze 593: " </tr>");
1.123 schwarze 594: }
1.144 schwarze 595: puts("</table>");
1.1 kristaps 596: }
1.16 kristaps 597:
1.70 schwarze 598: /*
599: * In man(1) mode, show one of the pages
600: * even if more than one is found.
601: */
602:
1.123 schwarze 603: if (req->q.equal || sz == 1) {
1.126 schwarze 604: puts("<hr>");
1.70 schwarze 605: iuse = 0;
1.120 schwarze 606: priouse = 20;
1.81 schwarze 607: archpriouse = 3;
1.70 schwarze 608: for (i = 0; i < sz; i++) {
1.120 schwarze 609: sec = r[i].file;
610: sec += strcspn(sec, "123456789");
611: if (sec[0] == '\0')
1.70 schwarze 612: continue;
1.120 schwarze 613: prio = sec_prios[sec[0] - '1'];
614: if (sec[1] != '/')
615: prio += 10;
616: if (req->q.arch == NULL) {
1.81 schwarze 617: archprio =
1.120 schwarze 618: ((arch = strchr(sec + 1, '/'))
619: == NULL) ? 3 :
620: ((archend = strchr(arch + 1, '/'))
621: == NULL) ? 0 :
1.81 schwarze 622: strncmp(arch, "amd64/",
623: archend - arch) ? 2 : 1;
624: if (archprio < archpriouse) {
625: archpriouse = archprio;
626: priouse = prio;
627: iuse = i;
628: continue;
629: }
630: if (archprio > archpriouse)
631: continue;
632: }
1.70 schwarze 633: if (prio >= priouse)
634: continue;
635: priouse = prio;
636: iuse = i;
637: }
638: resp_show(req, r[iuse].file);
639: }
640:
1.6 kristaps 641: resp_end_html();
642: }
643:
1.1 kristaps 644: static void
1.129 schwarze 645: resp_catman(const struct req *req, const char *file)
1.9 kristaps 646: {
1.10 kristaps 647: FILE *f;
1.115 schwarze 648: char *p;
649: size_t sz;
650: ssize_t len;
1.10 kristaps 651: int i;
652: int italic, bold;
1.9 kristaps 653:
1.115 schwarze 654: if ((f = fopen(file, "r")) == NULL) {
1.126 schwarze 655: puts("<p>You specified an invalid manual file.</p>");
1.9 kristaps 656: return;
657: }
658:
1.126 schwarze 659: puts("<div class=\"catman\">\n"
660: "<pre>");
1.10 kristaps 661:
1.115 schwarze 662: p = NULL;
663: sz = 0;
664:
665: while ((len = getline(&p, &sz, f)) != -1) {
1.10 kristaps 666: bold = italic = 0;
1.115 schwarze 667: for (i = 0; i < len - 1; i++) {
1.104 schwarze 668: /*
1.10 kristaps 669: * This means that the catpage is out of state.
670: * Ignore it and keep going (although the
671: * catpage is bogus).
672: */
673:
674: if ('\b' == p[i] || '\n' == p[i])
675: continue;
676:
677: /*
678: * Print a regular character.
679: * Close out any bold/italic scopes.
680: * If we're in back-space mode, make sure we'll
681: * have something to enter when we backspace.
682: */
683:
684: if ('\b' != p[i + 1]) {
685: if (italic)
1.126 schwarze 686: printf("</i>");
1.10 kristaps 687: if (bold)
1.126 schwarze 688: printf("</b>");
1.10 kristaps 689: italic = bold = 0;
690: html_putchar(p[i]);
691: continue;
1.115 schwarze 692: } else if (i + 2 >= len)
1.10 kristaps 693: continue;
694:
695: /* Italic mode. */
696:
697: if ('_' == p[i]) {
698: if (bold)
1.126 schwarze 699: printf("</b>");
1.10 kristaps 700: if ( ! italic)
1.126 schwarze 701: printf("<i>");
1.10 kristaps 702: bold = 0;
703: italic = 1;
704: i += 2;
705: html_putchar(p[i]);
706: continue;
707: }
708:
1.104 schwarze 709: /*
1.10 kristaps 710: * Handle funny behaviour troff-isms.
711: * These grok'd from the original man2html.c.
712: */
713:
714: if (('+' == p[i] && 'o' == p[i + 2]) ||
715: ('o' == p[i] && '+' == p[i + 2]) ||
716: ('|' == p[i] && '=' == p[i + 2]) ||
717: ('=' == p[i] && '|' == p[i + 2]) ||
718: ('*' == p[i] && '=' == p[i + 2]) ||
719: ('=' == p[i] && '*' == p[i + 2]) ||
720: ('*' == p[i] && '|' == p[i + 2]) ||
721: ('|' == p[i] && '*' == p[i + 2])) {
722: if (italic)
1.126 schwarze 723: printf("</i>");
1.10 kristaps 724: if (bold)
1.126 schwarze 725: printf("</b>");
1.10 kristaps 726: italic = bold = 0;
727: putchar('*');
728: i += 2;
729: continue;
730: } else if (('|' == p[i] && '-' == p[i + 2]) ||
731: ('-' == p[i] && '|' == p[i + 1]) ||
732: ('+' == p[i] && '-' == p[i + 1]) ||
733: ('-' == p[i] && '+' == p[i + 1]) ||
734: ('+' == p[i] && '|' == p[i + 1]) ||
735: ('|' == p[i] && '+' == p[i + 1])) {
736: if (italic)
1.126 schwarze 737: printf("</i>");
1.10 kristaps 738: if (bold)
1.126 schwarze 739: printf("</b>");
1.10 kristaps 740: italic = bold = 0;
741: putchar('+');
742: i += 2;
743: continue;
744: }
745:
746: /* Bold mode. */
1.104 schwarze 747:
1.10 kristaps 748: if (italic)
1.126 schwarze 749: printf("</i>");
1.10 kristaps 750: if ( ! bold)
1.126 schwarze 751: printf("<b>");
1.10 kristaps 752: bold = 1;
753: italic = 0;
754: i += 2;
755: html_putchar(p[i]);
756: }
757:
1.104 schwarze 758: /*
1.10 kristaps 759: * Clean up the last character.
1.104 schwarze 760: * We can get to a newline; don't print that.
1.10 kristaps 761: */
1.9 kristaps 762:
1.10 kristaps 763: if (italic)
1.126 schwarze 764: printf("</i>");
1.10 kristaps 765: if (bold)
1.126 schwarze 766: printf("</b>");
1.9 kristaps 767:
1.115 schwarze 768: if (i == len - 1 && p[i] != '\n')
1.10 kristaps 769: html_putchar(p[i]);
1.9 kristaps 770:
1.10 kristaps 771: putchar('\n');
772: }
1.115 schwarze 773: free(p);
1.10 kristaps 774:
1.126 schwarze 775: puts("</pre>\n"
776: "</div>");
1.10 kristaps 777:
778: fclose(f);
1.9 kristaps 779: }
780:
781: static void
1.129 schwarze 782: resp_format(const struct req *req, const char *file)
1.7 kristaps 783: {
1.106 schwarze 784: struct manoutput conf;
1.8 kristaps 785: struct mparse *mp;
1.107 schwarze 786: struct roff_man *man;
1.8 kristaps 787: void *vp;
1.90 schwarze 788: int fd;
789: int usepath;
1.7 kristaps 790:
1.8 kristaps 791: if (-1 == (fd = open(file, O_RDONLY, 0))) {
1.126 schwarze 792: puts("<p>You specified an invalid manual file.</p>");
1.7 kristaps 793: return;
794: }
795:
1.110 schwarze 796: mchars_alloc();
1.136 schwarze 797: mp = mparse_alloc(MPARSE_SO | MPARSE_UTF8 | MPARSE_LATIN1,
798: MANDOCLEVEL_BADARG, NULL, req->q.manpath);
1.103 schwarze 799: mparse_readfd(mp, fd, file);
1.8 kristaps 800: close(fd);
1.7 kristaps 801:
1.106 schwarze 802: memset(&conf, 0, sizeof(conf));
803: conf.fragment = 1;
1.145 schwarze 804: conf.style = mandoc_strdup(CSS_DIR "/mandoc.css");
1.90 schwarze 805: usepath = strcmp(req->q.manpath, req->p[0]);
1.122 schwarze 806: mandoc_asprintf(&conf.man, "/%s%s%%N.%%S",
807: usepath ? req->q.manpath : "", usepath ? "/" : "");
1.10 kristaps 808:
1.108 schwarze 809: mparse_result(mp, &man, NULL);
810: if (man == NULL) {
1.128 schwarze 811: warnx("fatal mandoc error: %s/%s", req->q.manpath, file);
1.72 schwarze 812: pg_error_internal();
1.32 kristaps 813: mparse_free(mp);
1.110 schwarze 814: mchars_free();
1.32 kristaps 815: return;
816: }
817:
1.110 schwarze 818: vp = html_alloc(&conf);
1.7 kristaps 819:
1.111 schwarze 820: if (man->macroset == MACROSET_MDOC) {
821: mdoc_validate(man);
1.108 schwarze 822: html_mdoc(vp, man);
1.112 schwarze 823: } else {
824: man_validate(man);
1.8 kristaps 825: html_man(vp, man);
1.112 schwarze 826: }
1.32 kristaps 827:
1.8 kristaps 828: html_free(vp);
829: mparse_free(mp);
1.110 schwarze 830: mchars_free();
1.106 schwarze 831: free(conf.man);
1.145 schwarze 832: free(conf.style);
1.7 kristaps 833: }
834:
835: static void
1.70 schwarze 836: resp_show(const struct req *req, const char *file)
837: {
1.76 schwarze 838:
839: if ('.' == file[0] && '/' == file[1])
1.71 schwarze 840: file += 2;
1.70 schwarze 841:
842: if ('c' == *file)
1.129 schwarze 843: resp_catman(req, file);
1.70 schwarze 844: else
1.129 schwarze 845: resp_format(req, file);
1.70 schwarze 846: }
847:
848: static void
1.84 schwarze 849: pg_show(struct req *req, const char *fullpath)
1.1 kristaps 850: {
1.84 schwarze 851: char *manpath;
852: const char *file;
1.25 kristaps 853:
1.84 schwarze 854: if ((file = strchr(fullpath, '/')) == NULL) {
1.72 schwarze 855: pg_error_badrequest(
1.59 schwarze 856: "You did not specify a page to show.");
1.25 kristaps 857: return;
1.104 schwarze 858: }
1.84 schwarze 859: manpath = mandoc_strndup(fullpath, file - fullpath);
860: file++;
1.6 kristaps 861:
1.84 schwarze 862: if ( ! validate_manpath(req, manpath)) {
1.77 schwarze 863: pg_error_badrequest(
864: "You specified an invalid manpath.");
1.84 schwarze 865: free(manpath);
1.77 schwarze 866: return;
867: }
868:
1.24 kristaps 869: /*
1.58 schwarze 870: * Begin by chdir()ing into the manpath.
1.24 kristaps 871: * This way we can pick up the database files, which are
872: * relative to the manpath root.
873: */
874:
1.84 schwarze 875: if (chdir(manpath) == -1) {
1.128 schwarze 876: warn("chdir %s", manpath);
1.77 schwarze 877: pg_error_internal();
1.84 schwarze 878: free(manpath);
1.76 schwarze 879: return;
880: }
1.134 schwarze 881: free(manpath);
1.84 schwarze 882:
883: if ( ! validate_filename(file)) {
1.76 schwarze 884: pg_error_badrequest(
885: "You specified an invalid manual file.");
1.24 kristaps 886: return;
887: }
1.79 schwarze 888:
1.70 schwarze 889: resp_begin_html(200, NULL);
1.131 schwarze 890: resp_searchform(req, FOCUS_NONE);
1.84 schwarze 891: resp_show(req, file);
1.70 schwarze 892: resp_end_html();
1.6 kristaps 893: }
894:
895: static void
1.66 schwarze 896: pg_search(const struct req *req)
1.6 kristaps 897: {
1.52 schwarze 898: struct mansearch search;
899: struct manpaths paths;
900: struct manpage *res;
1.97 schwarze 901: char **argv;
902: char *query, *rp, *wp;
1.52 schwarze 903: size_t ressz;
1.97 schwarze 904: int argc;
1.6 kristaps 905:
906: /*
1.24 kristaps 907: * Begin by chdir()ing into the root of the manpath.
908: * This way we can pick up the database files, which are
909: * relative to the manpath root.
910: */
911:
1.128 schwarze 912: if (chdir(req->q.manpath) == -1) {
913: warn("chdir %s", req->q.manpath);
1.77 schwarze 914: pg_error_internal();
1.24 kristaps 915: return;
916: }
917:
1.52 schwarze 918: search.arch = req->q.arch;
919: search.sec = req->q.sec;
1.94 schwarze 920: search.outkey = "Nd";
921: search.argmode = req->q.equal ? ARG_NAME : ARG_EXPR;
1.101 schwarze 922: search.firstmatch = 1;
1.52 schwarze 923:
924: paths.sz = 1;
925: paths.paths = mandoc_malloc(sizeof(char *));
926: paths.paths[0] = mandoc_strdup(".");
1.24 kristaps 927:
928: /*
1.97 schwarze 929: * Break apart at spaces with backslash-escaping.
1.6 kristaps 930: */
931:
1.97 schwarze 932: argc = 0;
933: argv = NULL;
934: rp = query = mandoc_strdup(req->q.query);
935: for (;;) {
936: while (isspace((unsigned char)*rp))
937: rp++;
938: if (*rp == '\0')
939: break;
940: argv = mandoc_reallocarray(argv, argc + 1, sizeof(char *));
941: argv[argc++] = wp = rp;
942: for (;;) {
943: if (isspace((unsigned char)*rp)) {
944: *wp = '\0';
945: rp++;
946: break;
947: }
948: if (rp[0] == '\\' && rp[1] != '\0')
949: rp++;
950: if (wp != rp)
951: *wp = *rp;
952: if (*rp == '\0')
953: break;
954: wp++;
955: rp++;
956: }
1.6 kristaps 957: }
958:
1.97 schwarze 959: if (0 == mansearch(&search, &paths, argc, argv, &res, &ressz))
1.72 schwarze 960: pg_noresult(req, "You entered an invalid query.");
1.59 schwarze 961: else if (0 == ressz)
1.72 schwarze 962: pg_noresult(req, "No results found.");
1.59 schwarze 963: else
1.72 schwarze 964: pg_searchres(req, res, ressz);
1.6 kristaps 965:
1.97 schwarze 966: free(query);
967: mansearch_free(res, ressz);
1.52 schwarze 968: free(paths.paths[0]);
969: free(paths.paths);
1.1 kristaps 970: }
971:
972: int
973: main(void)
974: {
1.66 schwarze 975: struct req req;
1.95 schwarze 976: struct itimerval itimer;
1.66 schwarze 977: const char *path;
1.83 schwarze 978: const char *querystring;
1.1 kristaps 979: int i;
1.148 ! schwarze 980:
! 981: #if HAVE_PLEDGE
! 982: /*
! 983: * The "rpath" pledge could be revoked after mparse_readfd()
! 984: * if the file desciptor to "/footer.html" would be opened
! 985: * up front, but it's probably not worth the complication
! 986: * of the code it would cause: it would require scattering
! 987: * pledge() calls in multiple low-level resp_*() functions.
! 988: */
! 989:
! 990: if (pledge("stdio rpath", NULL) == -1) {
! 991: warn("pledge");
! 992: pg_error_internal();
! 993: return EXIT_FAILURE;
! 994: }
! 995: #endif
1.95 schwarze 996:
997: /* Poor man's ReDoS mitigation. */
998:
1.99 schwarze 999: itimer.it_value.tv_sec = 2;
1.95 schwarze 1000: itimer.it_value.tv_usec = 0;
1.99 schwarze 1001: itimer.it_interval.tv_sec = 2;
1.95 schwarze 1002: itimer.it_interval.tv_usec = 0;
1003: if (setitimer(ITIMER_VIRTUAL, &itimer, NULL) == -1) {
1.128 schwarze 1004: warn("setitimer");
1.80 schwarze 1005: pg_error_internal();
1.109 schwarze 1006: return EXIT_FAILURE;
1.80 schwarze 1007: }
1008:
1.24 kristaps 1009: /*
1.67 schwarze 1010: * First we change directory into the MAN_DIR so that
1.24 kristaps 1011: * subsequent scanning for manpath directories is rooted
1012: * relative to the same position.
1013: */
1014:
1.128 schwarze 1015: if (chdir(MAN_DIR) == -1) {
1016: warn("MAN_DIR: %s", MAN_DIR);
1.72 schwarze 1017: pg_error_internal();
1.109 schwarze 1018: return EXIT_FAILURE;
1.104 schwarze 1019: }
1.24 kristaps 1020:
1021: memset(&req, 0, sizeof(struct req));
1.118 schwarze 1022: req.q.equal = 1;
1.129 schwarze 1023: parse_manpath_conf(&req);
1.1 kristaps 1024:
1.117 schwarze 1025: /* Parse the path info and the query string. */
1.1 kristaps 1026:
1.117 schwarze 1027: if ((path = getenv("PATH_INFO")) == NULL)
1028: path = "";
1029: else if (*path == '/')
1030: path++;
1031:
1.124 schwarze 1032: if (*path != '\0') {
1.129 schwarze 1033: parse_path_info(&req, path);
1.133 schwarze 1034: if (req.q.manpath == NULL || access(path, F_OK) == -1)
1.124 schwarze 1035: path = "";
1.117 schwarze 1036: } else if ((querystring = getenv("QUERY_STRING")) != NULL)
1.129 schwarze 1037: parse_query_string(&req, querystring);
1.77 schwarze 1038:
1.117 schwarze 1039: /* Validate parsed data and add defaults. */
1040:
1.102 schwarze 1041: if (req.q.manpath == NULL)
1042: req.q.manpath = mandoc_strdup(req.p[0]);
1043: else if ( ! validate_manpath(&req, req.q.manpath)) {
1.77 schwarze 1044: pg_error_badrequest(
1045: "You specified an invalid manpath.");
1.109 schwarze 1046: return EXIT_FAILURE;
1.77 schwarze 1047: }
1.1 kristaps 1048:
1.80 schwarze 1049: if ( ! (NULL == req.q.arch || validate_urifrag(req.q.arch))) {
1050: pg_error_badrequest(
1051: "You specified an invalid architecture.");
1.109 schwarze 1052: return EXIT_FAILURE;
1.80 schwarze 1053: }
1054:
1.66 schwarze 1055: /* Dispatch to the three different pages. */
1.1 kristaps 1056:
1.66 schwarze 1057: if ('\0' != *path)
1058: pg_show(&req, path);
1.85 schwarze 1059: else if (NULL != req.q.query)
1.66 schwarze 1060: pg_search(&req);
1061: else
1.72 schwarze 1062: pg_index(&req);
1.1 kristaps 1063:
1.83 schwarze 1064: free(req.q.manpath);
1065: free(req.q.arch);
1066: free(req.q.sec);
1.85 schwarze 1067: free(req.q.query);
1.52 schwarze 1068: for (i = 0; i < (int)req.psz; i++)
1069: free(req.p[i]);
1.24 kristaps 1070: free(req.p);
1.109 schwarze 1071: return EXIT_SUCCESS;
1.117 schwarze 1072: }
1073:
1074: /*
1075: * If PATH_INFO is not a file name, translate it to a query.
1076: */
1077: static void
1.129 schwarze 1078: parse_path_info(struct req *req, const char *path)
1.117 schwarze 1079: {
1.135 schwarze 1080: char *dir[4];
1081: int i;
1.117 schwarze 1082:
1.121 schwarze 1083: req->isquery = 0;
1.117 schwarze 1084: req->q.equal = 1;
1085: req->q.manpath = mandoc_strdup(path);
1.135 schwarze 1086: req->q.arch = NULL;
1.117 schwarze 1087:
1088: /* Mandatory manual page name. */
1089: if ((req->q.query = strrchr(req->q.manpath, '/')) == NULL) {
1090: req->q.query = req->q.manpath;
1091: req->q.manpath = NULL;
1092: } else
1093: *req->q.query++ = '\0';
1094:
1095: /* Optional trailing section. */
1096: if ((req->q.sec = strrchr(req->q.query, '.')) != NULL) {
1097: if(isdigit((unsigned char)req->q.sec[1])) {
1098: *req->q.sec++ = '\0';
1099: req->q.sec = mandoc_strdup(req->q.sec);
1100: } else
1101: req->q.sec = NULL;
1102: }
1103:
1104: /* Handle the case of name[.section] only. */
1.135 schwarze 1105: if (req->q.manpath == NULL)
1.117 schwarze 1106: return;
1.135 schwarze 1107: req->q.query = mandoc_strdup(req->q.query);
1108:
1109: /* Split directory components. */
1110: dir[i = 0] = req->q.manpath;
1111: while ((dir[i + 1] = strchr(dir[i], '/')) != NULL) {
1112: if (++i == 3) {
1113: pg_error_badrequest(
1114: "You specified too many directory components.");
1115: exit(EXIT_FAILURE);
1116: }
1117: *dir[i]++ = '\0';
1.117 schwarze 1118: }
1119:
1.135 schwarze 1120: /* Optional manpath. */
1121: if ((i = validate_manpath(req, req->q.manpath)) == 0)
1122: req->q.manpath = NULL;
1123: else if (dir[1] == NULL)
1124: return;
1.117 schwarze 1125:
1.135 schwarze 1126: /* Optional section. */
1127: if (strncmp(dir[i], "man", 3) == 0) {
1.127 schwarze 1128: free(req->q.sec);
1.135 schwarze 1129: req->q.sec = mandoc_strdup(dir[i++] + 3);
1.117 schwarze 1130: }
1.135 schwarze 1131: if (dir[i] == NULL) {
1132: if (req->q.manpath == NULL)
1133: free(dir[0]);
1134: return;
1135: }
1136: if (dir[i + 1] != NULL) {
1137: pg_error_badrequest(
1138: "You specified an invalid directory component.");
1139: exit(EXIT_FAILURE);
1140: }
1141:
1142: /* Optional architecture. */
1143: if (i) {
1144: req->q.arch = mandoc_strdup(dir[i]);
1145: if (req->q.manpath == NULL)
1146: free(dir[0]);
1147: } else
1148: req->q.arch = dir[0];
1.24 kristaps 1149: }
1150:
1151: /*
1152: * Scan for indexable paths.
1153: */
1154: static void
1.129 schwarze 1155: parse_manpath_conf(struct req *req)
1.24 kristaps 1156: {
1.54 schwarze 1157: FILE *fp;
1158: char *dp;
1159: size_t dpsz;
1.115 schwarze 1160: ssize_t len;
1.54 schwarze 1161:
1.128 schwarze 1162: if ((fp = fopen("manpath.conf", "r")) == NULL) {
1163: warn("%s/manpath.conf", MAN_DIR);
1.74 schwarze 1164: pg_error_internal();
1165: exit(EXIT_FAILURE);
1166: }
1.24 kristaps 1167:
1.115 schwarze 1168: dp = NULL;
1169: dpsz = 0;
1170:
1171: while ((len = getline(&dp, &dpsz, fp)) != -1) {
1172: if (dp[len - 1] == '\n')
1173: dp[--len] = '\0';
1.54 schwarze 1174: req->p = mandoc_realloc(req->p,
1175: (req->psz + 1) * sizeof(char *));
1.80 schwarze 1176: if ( ! validate_urifrag(dp)) {
1.128 schwarze 1177: warnx("%s/manpath.conf contains "
1178: "unsafe path \"%s\"", MAN_DIR, dp);
1.80 schwarze 1179: pg_error_internal();
1180: exit(EXIT_FAILURE);
1181: }
1.128 schwarze 1182: if (strchr(dp, '/') != NULL) {
1183: warnx("%s/manpath.conf contains "
1184: "path with slash \"%s\"", MAN_DIR, dp);
1.80 schwarze 1185: pg_error_internal();
1186: exit(EXIT_FAILURE);
1187: }
1188: req->p[req->psz++] = dp;
1.115 schwarze 1189: dp = NULL;
1190: dpsz = 0;
1.74 schwarze 1191: }
1.115 schwarze 1192: free(dp);
1.74 schwarze 1193:
1.128 schwarze 1194: if (req->p == NULL) {
1195: warnx("%s/manpath.conf is empty", MAN_DIR);
1.74 schwarze 1196: pg_error_internal();
1197: exit(EXIT_FAILURE);
1.24 kristaps 1198: }
1199: }
CVSweb