Annotation of mandoc/cgi.c, Revision 1.117
1.117 ! schwarze 1: /* $Id: cgi.c,v 1.116 2016/01/04 12:36:26 schwarze Exp $ */
1.6 kristaps 2: /*
1.42 kristaps 3: * Copyright (c) 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
1.106 schwarze 4: * Copyright (c) 2014, 2015 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>
1.58 schwarze 24: #include <errno.h>
1.1 kristaps 25: #include <fcntl.h>
1.6 kristaps 26: #include <limits.h>
1.92 schwarze 27: #include <stdint.h>
1.1 kristaps 28: #include <stdio.h>
29: #include <stdlib.h>
30: #include <string.h>
1.6 kristaps 31: #include <unistd.h>
1.1 kristaps 32:
1.108 schwarze 33: #include "mandoc_aux.h"
1.4 schwarze 34: #include "mandoc.h"
1.108 schwarze 35: #include "roff.h"
1.111 schwarze 36: #include "mdoc.h"
1.112 schwarze 37: #include "man.h"
1.8 kristaps 38: #include "main.h"
1.105 schwarze 39: #include "manconf.h"
1.52 schwarze 40: #include "mansearch.h"
1.67 schwarze 41: #include "cgi.h"
1.1 kristaps 42:
1.20 kristaps 43: /*
44: * A query as passed to the search function.
45: */
46: struct query {
1.83 schwarze 47: char *manpath; /* desired manual directory */
48: char *arch; /* architecture */
49: char *sec; /* manual section */
1.85 schwarze 50: char *query; /* unparsed query expression */
1.65 schwarze 51: int equal; /* match whole names, not substrings */
1.20 kristaps 52: };
53:
1.1 kristaps 54: struct req {
1.58 schwarze 55: struct query q;
56: char **p; /* array of available manpaths */
57: size_t psz; /* number of available manpaths */
1.1 kristaps 58: };
59:
1.32 kristaps 60: static void catman(const struct req *, const char *);
61: static void format(const struct req *, const char *);
1.6 kristaps 62: static void html_print(const char *);
1.10 kristaps 63: static void html_putchar(char);
1.104 schwarze 64: static int http_decode(char *);
1.83 schwarze 65: static void http_parse(struct req *, const char *);
1.54 schwarze 66: static void pathgen(struct req *);
1.117 ! schwarze 67: static void path_parse(struct req *req, const char *path);
1.72 schwarze 68: static void pg_error_badrequest(const char *);
69: static void pg_error_internal(void);
70: static void pg_index(const struct req *);
71: static void pg_noresult(const struct req *, const char *);
1.66 schwarze 72: static void pg_search(const struct req *);
1.72 schwarze 73: static void pg_searchres(const struct req *,
74: struct manpage *, size_t);
1.79 schwarze 75: static void pg_show(struct req *, const char *);
1.6 kristaps 76: static void resp_begin_html(int, const char *);
77: static void resp_begin_http(int, const char *);
1.114 schwarze 78: static void resp_copy(const char *);
1.6 kristaps 79: static void resp_end_html(void);
80: static void resp_searchform(const struct req *);
1.70 schwarze 81: static void resp_show(const struct req *, const char *);
1.85 schwarze 82: static void set_query_attr(char **, char **);
83: static int validate_filename(const char *);
84: static int validate_manpath(const struct req *, const char *);
85: static int validate_urifrag(const char *);
1.6 kristaps 86:
1.58 schwarze 87: static const char *scriptname; /* CGI script name */
1.1 kristaps 88:
1.70 schwarze 89: static const int sec_prios[] = {1, 4, 5, 8, 6, 3, 7, 2, 9};
1.68 schwarze 90: static const char *const sec_numbers[] = {
91: "0", "1", "2", "3", "3p", "4", "5", "6", "7", "8", "9"
92: };
93: static const char *const sec_names[] = {
94: "All Sections",
95: "1 - General Commands",
96: "2 - System Calls",
1.96 schwarze 97: "3 - Library Functions",
98: "3p - Perl Library",
99: "4 - Device Drivers",
1.68 schwarze 100: "5 - File Formats",
101: "6 - Games",
1.96 schwarze 102: "7 - Miscellaneous Information",
103: "8 - System Manager\'s Manual",
104: "9 - Kernel Developer\'s Manual"
1.68 schwarze 105: };
106: static const int sec_MAX = sizeof(sec_names) / sizeof(char *);
107:
108: static const char *const arch_names[] = {
109: "amd64", "alpha", "armish", "armv7",
110: "aviion", "hppa", "hppa64", "i386",
111: "ia64", "landisk", "loongson", "luna88k",
112: "macppc", "mips64", "octeon", "sgi",
113: "socppc", "solbourne", "sparc", "sparc64",
114: "vax", "zaurus",
115: "amiga", "arc", "arm32", "atari",
116: "beagle", "cats", "hp300", "mac68k",
117: "mvme68k", "mvme88k", "mvmeppc", "palm",
118: "pc532", "pegasos", "pmax", "powerpc",
119: "sun3", "wgrisc", "x68k"
120: };
121: static const int arch_MAX = sizeof(arch_names) / sizeof(char *);
122:
1.6 kristaps 123: /*
1.20 kristaps 124: * Print a character, escaping HTML along the way.
125: * This will pass non-ASCII straight to output: be warned!
126: */
1.10 kristaps 127: static void
128: html_putchar(char c)
129: {
130:
131: switch (c) {
132: case ('"'):
133: printf(""e;");
134: break;
135: case ('&'):
136: printf("&");
137: break;
138: case ('>'):
139: printf(">");
140: break;
141: case ('<'):
142: printf("<");
143: break;
144: default:
145: putchar((unsigned char)c);
146: break;
147: }
148: }
1.57 schwarze 149:
1.6 kristaps 150: /*
1.20 kristaps 151: * Call through to html_putchar().
152: * Accepts NULL strings.
1.6 kristaps 153: */
1.1 kristaps 154: static void
1.6 kristaps 155: html_print(const char *p)
1.1 kristaps 156: {
1.104 schwarze 157:
1.6 kristaps 158: if (NULL == p)
159: return;
1.1 kristaps 160: while ('\0' != *p)
1.10 kristaps 161: html_putchar(*p++);
1.1 kristaps 162: }
163:
164: /*
1.83 schwarze 165: * Transfer the responsibility for the allocated string *val
166: * to the query structure.
1.1 kristaps 167: */
168: static void
1.83 schwarze 169: set_query_attr(char **attr, char **val)
1.1 kristaps 170: {
171:
1.83 schwarze 172: free(*attr);
173: if (**val == '\0') {
174: *attr = NULL;
175: free(*val);
176: } else
177: *attr = *val;
178: *val = NULL;
179: }
180:
181: /*
182: * Parse the QUERY_STRING for key-value pairs
183: * and store the values into the query structure.
184: */
185: static void
186: http_parse(struct req *req, const char *qs)
187: {
188: char *key, *val;
189: size_t keysz, valsz;
190:
191: req->q.manpath = NULL;
192: req->q.arch = NULL;
193: req->q.sec = NULL;
1.85 schwarze 194: req->q.query = NULL;
1.83 schwarze 195: req->q.equal = 1;
196:
197: key = val = NULL;
198: while (*qs != '\0') {
1.24 kristaps 199:
1.83 schwarze 200: /* Parse one key. */
201:
202: keysz = strcspn(qs, "=;&");
203: key = mandoc_strndup(qs, keysz);
204: qs += keysz;
205: if (*qs != '=')
206: goto next;
207:
208: /* Parse one value. */
209:
210: valsz = strcspn(++qs, ";&");
211: val = mandoc_strndup(qs, valsz);
212: qs += valsz;
213:
214: /* Decode and catch encoding errors. */
1.1 kristaps 215:
1.83 schwarze 216: if ( ! (http_decode(key) && http_decode(val)))
217: goto next;
1.1 kristaps 218:
1.83 schwarze 219: /* Handle key-value pairs. */
1.1 kristaps 220:
1.83 schwarze 221: if ( ! strcmp(key, "query"))
1.85 schwarze 222: set_query_attr(&req->q.query, &val);
1.1 kristaps 223:
1.83 schwarze 224: else if ( ! strcmp(key, "apropos"))
225: req->q.equal = !strcmp(val, "0");
226:
227: else if ( ! strcmp(key, "manpath")) {
1.73 schwarze 228: #ifdef COMPAT_OLDURI
1.83 schwarze 229: if ( ! strncmp(val, "OpenBSD ", 8)) {
1.73 schwarze 230: val[7] = '-';
231: if ('C' == val[8])
232: val[8] = 'c';
233: }
234: #endif
1.83 schwarze 235: set_query_attr(&req->q.manpath, &val);
236: }
237:
238: else if ( ! (strcmp(key, "sec")
1.73 schwarze 239: #ifdef COMPAT_OLDURI
1.83 schwarze 240: && strcmp(key, "sektion")
1.73 schwarze 241: #endif
1.83 schwarze 242: )) {
243: if ( ! strcmp(val, "0"))
244: *val = '\0';
245: set_query_attr(&req->q.sec, &val);
1.65 schwarze 246: }
1.83 schwarze 247:
248: else if ( ! strcmp(key, "arch")) {
249: if ( ! strcmp(val, "default"))
250: *val = '\0';
251: set_query_attr(&req->q.arch, &val);
252: }
253:
254: /*
255: * The key must be freed in any case.
256: * The val may have been handed over to the query
257: * structure, in which case it is now NULL.
258: */
259: next:
260: free(key);
261: key = NULL;
262: free(val);
263: val = NULL;
264:
265: if (*qs != '\0')
266: qs++;
1.24 kristaps 267: }
1.1 kristaps 268: }
269:
270: /*
1.6 kristaps 271: * HTTP-decode a string. The standard explanation is that this turns
272: * "%4e+foo" into "n foo" in the regular way. This is done in-place
273: * over the allocated string.
1.1 kristaps 274: */
275: static int
1.24 kristaps 276: http_decode(char *p)
1.1 kristaps 277: {
278: char hex[3];
1.63 schwarze 279: char *q;
1.1 kristaps 280: int c;
281:
282: hex[2] = '\0';
283:
1.63 schwarze 284: q = p;
285: for ( ; '\0' != *p; p++, q++) {
1.1 kristaps 286: if ('%' == *p) {
287: if ('\0' == (hex[0] = *(p + 1)))
1.109 schwarze 288: return 0;
1.1 kristaps 289: if ('\0' == (hex[1] = *(p + 2)))
1.109 schwarze 290: return 0;
1.1 kristaps 291: if (1 != sscanf(hex, "%x", &c))
1.109 schwarze 292: return 0;
1.1 kristaps 293: if ('\0' == c)
1.109 schwarze 294: return 0;
1.1 kristaps 295:
1.63 schwarze 296: *q = (char)c;
297: p += 2;
1.1 kristaps 298: } else
1.63 schwarze 299: *q = '+' == *p ? ' ' : *p;
1.1 kristaps 300: }
301:
1.63 schwarze 302: *q = '\0';
1.109 schwarze 303: return 1;
1.1 kristaps 304: }
305:
1.6 kristaps 306: static void
307: resp_begin_http(int code, const char *msg)
308: {
309:
310: if (200 != code)
1.62 schwarze 311: printf("Status: %d %s\r\n", code, msg);
1.6 kristaps 312:
1.62 schwarze 313: printf("Content-Type: text/html; charset=utf-8\r\n"
314: "Cache-Control: no-cache\r\n"
315: "Pragma: no-cache\r\n"
316: "\r\n");
1.6 kristaps 317:
318: fflush(stdout);
319: }
320:
321: static void
1.114 schwarze 322: resp_copy(const char *filename)
323: {
324: char buf[4096];
325: ssize_t sz;
326: int fd;
327:
328: if ((fd = open(filename, O_RDONLY)) != -1) {
329: fflush(stdout);
330: while ((sz = read(fd, buf, sizeof(buf))) > 0)
331: write(STDOUT_FILENO, buf, sz);
332: }
333: }
334:
335: static void
1.6 kristaps 336: resp_begin_html(int code, const char *msg)
337: {
338:
339: resp_begin_http(code, msg);
340:
1.98 kristaps 341: printf("<!DOCTYPE html>\n"
1.29 kristaps 342: "<HTML>\n"
343: "<HEAD>\n"
1.98 kristaps 344: "<META CHARSET=\"UTF-8\" />\n"
1.113 schwarze 345: "<LINK REL=\"stylesheet\" HREF=\"%s/mandoc.css\""
1.29 kristaps 346: " TYPE=\"text/css\" media=\"all\">\n"
1.67 schwarze 347: "<TITLE>%s</TITLE>\n"
1.29 kristaps 348: "</HEAD>\n"
349: "<BODY>\n"
1.58 schwarze 350: "<!-- Begin page content. //-->\n",
1.113 schwarze 351: CSS_DIR, CUSTOMIZE_TITLE);
1.114 schwarze 352:
353: resp_copy(MAN_DIR "/header.html");
1.6 kristaps 354: }
355:
356: static void
357: resp_end_html(void)
358: {
359:
1.114 schwarze 360: resp_copy(MAN_DIR "/footer.html");
361:
1.20 kristaps 362: puts("</BODY>\n"
363: "</HTML>");
1.6 kristaps 364: }
365:
366: static void
367: resp_searchform(const struct req *req)
368: {
1.27 kristaps 369: int i;
1.13 kristaps 370:
1.6 kristaps 371: puts("<!-- Begin search form. //-->");
1.32 kristaps 372: printf("<DIV ID=\"mancgi\">\n"
1.66 schwarze 373: "<FORM ACTION=\"%s\" METHOD=\"get\">\n"
1.29 kristaps 374: "<FIELDSET>\n"
1.68 schwarze 375: "<LEGEND>Manual Page Search Parameters</LEGEND>\n",
1.58 schwarze 376: scriptname);
1.68 schwarze 377:
378: /* Write query input box. */
379:
380: printf( "<TABLE><TR><TD>\n"
381: "<INPUT TYPE=\"text\" NAME=\"query\" VALUE=\"");
1.85 schwarze 382: if (NULL != req->q.query)
383: html_print(req->q.query);
1.68 schwarze 384: puts("\" SIZE=\"40\">");
385:
386: /* Write submission and reset buttons. */
387:
388: printf( "<INPUT TYPE=\"submit\" VALUE=\"Submit\">\n"
389: "<INPUT TYPE=\"reset\" VALUE=\"Reset\">\n");
390:
391: /* Write show radio button */
392:
393: printf( "</TD><TD>\n"
394: "<INPUT TYPE=\"radio\" ");
1.65 schwarze 395: if (req->q.equal)
1.86 schwarze 396: printf("CHECKED=\"checked\" ");
1.68 schwarze 397: printf( "NAME=\"apropos\" ID=\"show\" VALUE=\"0\">\n"
398: "<LABEL FOR=\"show\">Show named manual page</LABEL>\n");
399:
400: /* Write section selector. */
401:
1.86 schwarze 402: puts( "</TD></TR><TR><TD>\n"
1.68 schwarze 403: "<SELECT NAME=\"sec\">");
404: for (i = 0; i < sec_MAX; i++) {
405: printf("<OPTION VALUE=\"%s\"", sec_numbers[i]);
406: if (NULL != req->q.sec &&
407: 0 == strcmp(sec_numbers[i], req->q.sec))
1.86 schwarze 408: printf(" SELECTED=\"selected\"");
1.68 schwarze 409: printf(">%s</OPTION>\n", sec_names[i]);
410: }
411: puts("</SELECT>");
412:
413: /* Write architecture selector. */
414:
1.81 schwarze 415: printf( "<SELECT NAME=\"arch\">\n"
416: "<OPTION VALUE=\"default\"");
417: if (NULL == req->q.arch)
1.86 schwarze 418: printf(" SELECTED=\"selected\"");
1.81 schwarze 419: puts(">All Architectures</OPTION>");
1.68 schwarze 420: for (i = 0; i < arch_MAX; i++) {
421: printf("<OPTION VALUE=\"%s\"", arch_names[i]);
422: if (NULL != req->q.arch &&
423: 0 == strcmp(arch_names[i], req->q.arch))
1.86 schwarze 424: printf(" SELECTED=\"selected\"");
1.68 schwarze 425: printf(">%s</OPTION>\n", arch_names[i]);
426: }
427: puts("</SELECT>");
428:
429: /* Write manpath selector. */
430:
1.27 kristaps 431: if (req->psz > 1) {
1.68 schwarze 432: puts("<SELECT NAME=\"manpath\">");
1.27 kristaps 433: for (i = 0; i < (int)req->psz; i++) {
1.52 schwarze 434: printf("<OPTION ");
1.102 schwarze 435: if (strcmp(req->q.manpath, req->p[i]) == 0)
1.86 schwarze 436: printf("SELECTED=\"selected\" ");
1.52 schwarze 437: printf("VALUE=\"");
438: html_print(req->p[i]);
1.27 kristaps 439: printf("\">");
1.52 schwarze 440: html_print(req->p[i]);
1.27 kristaps 441: puts("</OPTION>");
442: }
443: puts("</SELECT>");
444: }
1.68 schwarze 445:
446: /* Write search radio button */
447:
448: printf( "</TD><TD>\n"
449: "<INPUT TYPE=\"radio\" ");
450: if (0 == req->q.equal)
1.86 schwarze 451: printf("CHECKED=\"checked\" ");
1.68 schwarze 452: printf( "NAME=\"apropos\" ID=\"search\" VALUE=\"1\">\n"
453: "<LABEL FOR=\"search\">Search with apropos query</LABEL>\n");
454:
455: puts("</TD></TR></TABLE>\n"
1.12 kristaps 456: "</FIELDSET>\n"
1.32 kristaps 457: "</FORM>\n"
458: "</DIV>");
1.20 kristaps 459: puts("<!-- End search form. //-->");
1.6 kristaps 460: }
461:
1.76 schwarze 462: static int
1.80 schwarze 463: validate_urifrag(const char *frag)
464: {
465:
466: while ('\0' != *frag) {
467: if ( ! (isalnum((unsigned char)*frag) ||
468: '-' == *frag || '.' == *frag ||
469: '/' == *frag || '_' == *frag))
1.109 schwarze 470: return 0;
1.80 schwarze 471: frag++;
472: }
1.109 schwarze 473: return 1;
1.80 schwarze 474: }
475:
476: static int
1.77 schwarze 477: validate_manpath(const struct req *req, const char* manpath)
478: {
479: size_t i;
480:
481: if ( ! strcmp(manpath, "mandoc"))
1.109 schwarze 482: return 1;
1.77 schwarze 483:
484: for (i = 0; i < req->psz; i++)
485: if ( ! strcmp(manpath, req->p[i]))
1.109 schwarze 486: return 1;
1.77 schwarze 487:
1.109 schwarze 488: return 0;
1.77 schwarze 489: }
490:
491: static int
1.76 schwarze 492: validate_filename(const char *file)
493: {
494:
495: if ('.' == file[0] && '/' == file[1])
496: file += 2;
497:
1.109 schwarze 498: return ! (strstr(file, "../") || strstr(file, "/..") ||
499: (strncmp(file, "man", 3) && strncmp(file, "cat", 3)));
1.76 schwarze 500: }
501:
1.6 kristaps 502: static void
1.72 schwarze 503: pg_index(const struct req *req)
1.6 kristaps 504: {
505:
506: resp_begin_html(200, NULL);
507: resp_searchform(req);
1.64 schwarze 508: printf("<P>\n"
1.86 schwarze 509: "This web interface is documented in the\n"
510: "<A HREF=\"%s/mandoc/man8/man.cgi.8\">man.cgi</A>\n"
511: "manual, and the\n"
512: "<A HREF=\"%s/mandoc/man1/apropos.1\">apropos</A>\n"
1.69 schwarze 513: "manual explains the query syntax.\n"
1.64 schwarze 514: "</P>\n",
515: scriptname, scriptname);
1.6 kristaps 516: resp_end_html();
517: }
518:
519: static void
1.72 schwarze 520: pg_noresult(const struct req *req, const char *msg)
1.59 schwarze 521: {
522: resp_begin_html(200, NULL);
523: resp_searchform(req);
524: puts("<P>");
525: puts(msg);
526: puts("</P>");
527: resp_end_html();
528: }
529:
530: static void
1.72 schwarze 531: pg_error_badrequest(const char *msg)
1.9 kristaps 532: {
533:
1.59 schwarze 534: resp_begin_html(400, "Bad Request");
535: puts("<H1>Bad Request</H1>\n"
536: "<P>\n");
537: puts(msg);
538: printf("Try again from the\n"
539: "<A HREF=\"%s\">main page</A>.\n"
1.58 schwarze 540: "</P>", scriptname);
1.9 kristaps 541: resp_end_html();
542: }
543:
544: static void
1.72 schwarze 545: pg_error_internal(void)
1.7 kristaps 546: {
547: resp_begin_html(500, "Internal Server Error");
1.58 schwarze 548: puts("<P>Internal Server Error</P>");
1.7 kristaps 549: resp_end_html();
550: }
551:
552: static void
1.72 schwarze 553: pg_searchres(const struct req *req, struct manpage *r, size_t sz)
1.1 kristaps 554: {
1.81 schwarze 555: char *arch, *archend;
1.70 schwarze 556: size_t i, iuse, isec;
1.81 schwarze 557: int archprio, archpriouse;
1.70 schwarze 558: int prio, priouse;
559: char sec;
1.19 kristaps 560:
1.76 schwarze 561: for (i = 0; i < sz; i++) {
562: if (validate_filename(r[i].file))
563: continue;
564: fprintf(stderr, "invalid filename %s in %s database\n",
565: r[i].file, req->q.manpath);
566: pg_error_internal();
567: return;
568: }
569:
1.52 schwarze 570: if (1 == sz) {
1.6 kristaps 571: /*
572: * If we have just one result, then jump there now
573: * without any delay.
574: */
1.62 schwarze 575: printf("Status: 303 See Other\r\n");
1.116 schwarze 576: printf("Location: http://%s%s/%s/%s",
1.78 schwarze 577: HTTP_HOST, scriptname, req->q.manpath, r[0].file);
1.62 schwarze 578: printf("\r\n"
579: "Content-Type: text/html; charset=utf-8\r\n"
580: "\r\n");
1.6 kristaps 581: return;
582: }
583:
1.12 kristaps 584: resp_begin_html(200, NULL);
1.19 kristaps 585: resp_searchform(req);
1.33 kristaps 586: puts("<DIV CLASS=\"results\">");
587: puts("<TABLE>");
1.1 kristaps 588:
1.41 kristaps 589: for (i = 0; i < sz; i++) {
1.20 kristaps 590: printf("<TR>\n"
591: "<TD CLASS=\"title\">\n"
1.116 schwarze 592: "<A HREF=\"%s/%s/%s",
1.58 schwarze 593: scriptname, req->q.manpath, r[i].file);
1.36 kristaps 594: printf("\">");
1.52 schwarze 595: html_print(r[i].names);
596: printf("</A>\n"
1.20 kristaps 597: "</TD>\n"
598: "<TD CLASS=\"desc\">");
1.52 schwarze 599: html_print(r[i].output);
1.20 kristaps 600: puts("</TD>\n"
601: "</TR>");
1.1 kristaps 602: }
1.16 kristaps 603:
1.33 kristaps 604: puts("</TABLE>\n"
605: "</DIV>");
1.70 schwarze 606:
607: /*
608: * In man(1) mode, show one of the pages
609: * even if more than one is found.
610: */
611:
612: if (req->q.equal) {
613: puts("<HR>");
614: iuse = 0;
615: priouse = 10;
1.81 schwarze 616: archpriouse = 3;
1.70 schwarze 617: for (i = 0; i < sz; i++) {
618: isec = strcspn(r[i].file, "123456789");
619: sec = r[i].file[isec];
620: if ('\0' == sec)
621: continue;
622: prio = sec_prios[sec - '1'];
1.81 schwarze 623: if (NULL == req->q.arch) {
624: archprio =
625: (NULL == (arch = strchr(
626: r[i].file + isec, '/'))) ? 3 :
627: (NULL == (archend = strchr(
628: arch + 1, '/'))) ? 0 :
629: strncmp(arch, "amd64/",
630: archend - arch) ? 2 : 1;
631: if (archprio < archpriouse) {
632: archpriouse = archprio;
633: priouse = prio;
634: iuse = i;
635: continue;
636: }
637: if (archprio > archpriouse)
638: continue;
639: }
1.70 schwarze 640: if (prio >= priouse)
641: continue;
642: priouse = prio;
643: iuse = i;
644: }
645: resp_show(req, r[iuse].file);
646: }
647:
1.6 kristaps 648: resp_end_html();
649: }
650:
1.1 kristaps 651: static void
1.32 kristaps 652: catman(const struct req *req, const char *file)
1.9 kristaps 653: {
1.10 kristaps 654: FILE *f;
1.115 schwarze 655: char *p;
656: size_t sz;
657: ssize_t len;
1.10 kristaps 658: int i;
659: int italic, bold;
1.9 kristaps 660:
1.115 schwarze 661: if ((f = fopen(file, "r")) == NULL) {
1.70 schwarze 662: puts("<P>You specified an invalid manual file.</P>");
1.9 kristaps 663: return;
664: }
665:
1.32 kristaps 666: puts("<DIV CLASS=\"catman\">\n"
667: "<PRE>");
1.10 kristaps 668:
1.115 schwarze 669: p = NULL;
670: sz = 0;
671:
672: while ((len = getline(&p, &sz, f)) != -1) {
1.10 kristaps 673: bold = italic = 0;
1.115 schwarze 674: for (i = 0; i < len - 1; i++) {
1.104 schwarze 675: /*
1.10 kristaps 676: * This means that the catpage is out of state.
677: * Ignore it and keep going (although the
678: * catpage is bogus).
679: */
680:
681: if ('\b' == p[i] || '\n' == p[i])
682: continue;
683:
684: /*
685: * Print a regular character.
686: * Close out any bold/italic scopes.
687: * If we're in back-space mode, make sure we'll
688: * have something to enter when we backspace.
689: */
690:
691: if ('\b' != p[i + 1]) {
692: if (italic)
693: printf("</I>");
694: if (bold)
695: printf("</B>");
696: italic = bold = 0;
697: html_putchar(p[i]);
698: continue;
1.115 schwarze 699: } else if (i + 2 >= len)
1.10 kristaps 700: continue;
701:
702: /* Italic mode. */
703:
704: if ('_' == p[i]) {
705: if (bold)
706: printf("</B>");
707: if ( ! italic)
708: printf("<I>");
709: bold = 0;
710: italic = 1;
711: i += 2;
712: html_putchar(p[i]);
713: continue;
714: }
715:
1.104 schwarze 716: /*
1.10 kristaps 717: * Handle funny behaviour troff-isms.
718: * These grok'd from the original man2html.c.
719: */
720:
721: if (('+' == p[i] && 'o' == p[i + 2]) ||
722: ('o' == p[i] && '+' == p[i + 2]) ||
723: ('|' == p[i] && '=' == p[i + 2]) ||
724: ('=' == p[i] && '|' == p[i + 2]) ||
725: ('*' == p[i] && '=' == p[i + 2]) ||
726: ('=' == p[i] && '*' == p[i + 2]) ||
727: ('*' == p[i] && '|' == p[i + 2]) ||
728: ('|' == p[i] && '*' == p[i + 2])) {
729: if (italic)
730: printf("</I>");
731: if (bold)
732: printf("</B>");
733: italic = bold = 0;
734: putchar('*');
735: i += 2;
736: continue;
737: } else if (('|' == p[i] && '-' == p[i + 2]) ||
738: ('-' == p[i] && '|' == p[i + 1]) ||
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: if (italic)
744: printf("</I>");
745: if (bold)
746: printf("</B>");
747: italic = bold = 0;
748: putchar('+');
749: i += 2;
750: continue;
751: }
752:
753: /* Bold mode. */
1.104 schwarze 754:
1.10 kristaps 755: if (italic)
756: printf("</I>");
757: if ( ! bold)
758: printf("<B>");
759: bold = 1;
760: italic = 0;
761: i += 2;
762: html_putchar(p[i]);
763: }
764:
1.104 schwarze 765: /*
1.10 kristaps 766: * Clean up the last character.
1.104 schwarze 767: * We can get to a newline; don't print that.
1.10 kristaps 768: */
1.9 kristaps 769:
1.10 kristaps 770: if (italic)
771: printf("</I>");
772: if (bold)
773: printf("</B>");
1.9 kristaps 774:
1.115 schwarze 775: if (i == len - 1 && p[i] != '\n')
1.10 kristaps 776: html_putchar(p[i]);
1.9 kristaps 777:
1.10 kristaps 778: putchar('\n');
779: }
1.115 schwarze 780: free(p);
1.10 kristaps 781:
782: puts("</PRE>\n"
1.70 schwarze 783: "</DIV>");
1.10 kristaps 784:
785: fclose(f);
1.9 kristaps 786: }
787:
788: static void
1.32 kristaps 789: format(const struct req *req, const char *file)
1.7 kristaps 790: {
1.106 schwarze 791: struct manoutput conf;
1.8 kristaps 792: struct mparse *mp;
1.107 schwarze 793: struct roff_man *man;
1.8 kristaps 794: void *vp;
1.90 schwarze 795: int fd;
796: int usepath;
1.7 kristaps 797:
1.8 kristaps 798: if (-1 == (fd = open(file, O_RDONLY, 0))) {
1.70 schwarze 799: puts("<P>You specified an invalid manual file.</P>");
1.7 kristaps 800: return;
801: }
802:
1.110 schwarze 803: mchars_alloc();
804: mp = mparse_alloc(MPARSE_SO, MANDOCLEVEL_BADARG, NULL, req->q.manpath);
1.103 schwarze 805: mparse_readfd(mp, fd, file);
1.8 kristaps 806: close(fd);
1.7 kristaps 807:
1.106 schwarze 808: memset(&conf, 0, sizeof(conf));
809: conf.fragment = 1;
1.90 schwarze 810: usepath = strcmp(req->q.manpath, req->p[0]);
1.106 schwarze 811: mandoc_asprintf(&conf.man, "%s?query=%%N&sec=%%S%s%s%s%s",
1.90 schwarze 812: scriptname,
813: req->q.arch ? "&arch=" : "",
814: req->q.arch ? req->q.arch : "",
815: usepath ? "&manpath=" : "",
816: usepath ? req->q.manpath : "");
1.10 kristaps 817:
1.108 schwarze 818: mparse_result(mp, &man, NULL);
819: if (man == NULL) {
1.59 schwarze 820: fprintf(stderr, "fatal mandoc error: %s/%s\n",
821: req->q.manpath, file);
1.72 schwarze 822: pg_error_internal();
1.32 kristaps 823: mparse_free(mp);
1.110 schwarze 824: mchars_free();
1.32 kristaps 825: return;
826: }
827:
1.110 schwarze 828: vp = html_alloc(&conf);
1.7 kristaps 829:
1.111 schwarze 830: if (man->macroset == MACROSET_MDOC) {
831: mdoc_validate(man);
1.108 schwarze 832: html_mdoc(vp, man);
1.112 schwarze 833: } else {
834: man_validate(man);
1.8 kristaps 835: html_man(vp, man);
1.112 schwarze 836: }
1.32 kristaps 837:
1.8 kristaps 838: html_free(vp);
839: mparse_free(mp);
1.110 schwarze 840: mchars_free();
1.106 schwarze 841: free(conf.man);
1.7 kristaps 842: }
843:
844: static void
1.70 schwarze 845: resp_show(const struct req *req, const char *file)
846: {
1.76 schwarze 847:
848: if ('.' == file[0] && '/' == file[1])
1.71 schwarze 849: file += 2;
1.70 schwarze 850:
851: if ('c' == *file)
852: catman(req, file);
853: else
854: format(req, file);
855: }
856:
857: static void
1.84 schwarze 858: pg_show(struct req *req, const char *fullpath)
1.1 kristaps 859: {
1.84 schwarze 860: char *manpath;
861: const char *file;
1.25 kristaps 862:
1.84 schwarze 863: if ((file = strchr(fullpath, '/')) == NULL) {
1.72 schwarze 864: pg_error_badrequest(
1.59 schwarze 865: "You did not specify a page to show.");
1.25 kristaps 866: return;
1.104 schwarze 867: }
1.84 schwarze 868: manpath = mandoc_strndup(fullpath, file - fullpath);
869: file++;
1.6 kristaps 870:
1.84 schwarze 871: if ( ! validate_manpath(req, manpath)) {
1.77 schwarze 872: pg_error_badrequest(
873: "You specified an invalid manpath.");
1.84 schwarze 874: free(manpath);
1.77 schwarze 875: return;
876: }
877:
1.24 kristaps 878: /*
1.58 schwarze 879: * Begin by chdir()ing into the manpath.
1.24 kristaps 880: * This way we can pick up the database files, which are
881: * relative to the manpath root.
882: */
883:
1.84 schwarze 884: if (chdir(manpath) == -1) {
1.77 schwarze 885: fprintf(stderr, "chdir %s: %s\n",
1.84 schwarze 886: manpath, strerror(errno));
1.77 schwarze 887: pg_error_internal();
1.84 schwarze 888: free(manpath);
1.76 schwarze 889: return;
890: }
891:
1.84 schwarze 892: if (strcmp(manpath, "mandoc")) {
893: free(req->q.manpath);
894: req->q.manpath = manpath;
895: } else
896: free(manpath);
897:
898: if ( ! validate_filename(file)) {
1.76 schwarze 899: pg_error_badrequest(
900: "You specified an invalid manual file.");
1.24 kristaps 901: return;
902: }
1.79 schwarze 903:
1.70 schwarze 904: resp_begin_html(200, NULL);
905: resp_searchform(req);
1.84 schwarze 906: resp_show(req, file);
1.70 schwarze 907: resp_end_html();
1.6 kristaps 908: }
909:
910: static void
1.66 schwarze 911: pg_search(const struct req *req)
1.6 kristaps 912: {
1.52 schwarze 913: struct mansearch search;
914: struct manpaths paths;
915: struct manpage *res;
1.97 schwarze 916: char **argv;
917: char *query, *rp, *wp;
1.52 schwarze 918: size_t ressz;
1.97 schwarze 919: int argc;
1.6 kristaps 920:
921: /*
1.24 kristaps 922: * Begin by chdir()ing into the root of the manpath.
923: * This way we can pick up the database files, which are
924: * relative to the manpath root.
925: */
926:
1.58 schwarze 927: if (-1 == (chdir(req->q.manpath))) {
1.77 schwarze 928: fprintf(stderr, "chdir %s: %s\n",
929: req->q.manpath, strerror(errno));
930: pg_error_internal();
1.24 kristaps 931: return;
932: }
933:
1.52 schwarze 934: search.arch = req->q.arch;
935: search.sec = req->q.sec;
1.94 schwarze 936: search.outkey = "Nd";
937: search.argmode = req->q.equal ? ARG_NAME : ARG_EXPR;
1.101 schwarze 938: search.firstmatch = 1;
1.52 schwarze 939:
940: paths.sz = 1;
941: paths.paths = mandoc_malloc(sizeof(char *));
942: paths.paths[0] = mandoc_strdup(".");
1.24 kristaps 943:
944: /*
1.97 schwarze 945: * Break apart at spaces with backslash-escaping.
1.6 kristaps 946: */
947:
1.97 schwarze 948: argc = 0;
949: argv = NULL;
950: rp = query = mandoc_strdup(req->q.query);
951: for (;;) {
952: while (isspace((unsigned char)*rp))
953: rp++;
954: if (*rp == '\0')
955: break;
956: argv = mandoc_reallocarray(argv, argc + 1, sizeof(char *));
957: argv[argc++] = wp = rp;
958: for (;;) {
959: if (isspace((unsigned char)*rp)) {
960: *wp = '\0';
961: rp++;
962: break;
963: }
964: if (rp[0] == '\\' && rp[1] != '\0')
965: rp++;
966: if (wp != rp)
967: *wp = *rp;
968: if (*rp == '\0')
969: break;
970: wp++;
971: rp++;
972: }
1.6 kristaps 973: }
974:
1.97 schwarze 975: if (0 == mansearch(&search, &paths, argc, argv, &res, &ressz))
1.72 schwarze 976: pg_noresult(req, "You entered an invalid query.");
1.59 schwarze 977: else if (0 == ressz)
1.72 schwarze 978: pg_noresult(req, "No results found.");
1.59 schwarze 979: else
1.72 schwarze 980: pg_searchres(req, res, ressz);
1.6 kristaps 981:
1.97 schwarze 982: free(query);
983: mansearch_free(res, ressz);
1.52 schwarze 984: free(paths.paths[0]);
985: free(paths.paths);
1.1 kristaps 986: }
987:
988: int
989: main(void)
990: {
1.66 schwarze 991: struct req req;
1.95 schwarze 992: struct itimerval itimer;
1.66 schwarze 993: const char *path;
1.83 schwarze 994: const char *querystring;
1.1 kristaps 995: int i;
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) {
1004: fprintf(stderr, "setitimer: %s\n", strerror(errno));
1005: pg_error_internal();
1.109 schwarze 1006: return EXIT_FAILURE;
1.95 schwarze 1007: }
1.6 kristaps 1008:
1.24 kristaps 1009: /* Scan our run-time environment. */
1.6 kristaps 1010:
1.58 schwarze 1011: if (NULL == (scriptname = getenv("SCRIPT_NAME")))
1012: scriptname = "";
1.24 kristaps 1013:
1.80 schwarze 1014: if ( ! validate_urifrag(scriptname)) {
1015: fprintf(stderr, "unsafe SCRIPT_NAME \"%s\"\n",
1016: scriptname);
1017: pg_error_internal();
1.109 schwarze 1018: return EXIT_FAILURE;
1.80 schwarze 1019: }
1020:
1.24 kristaps 1021: /*
1.67 schwarze 1022: * First we change directory into the MAN_DIR so that
1.24 kristaps 1023: * subsequent scanning for manpath directories is rooted
1024: * relative to the same position.
1025: */
1026:
1.67 schwarze 1027: if (-1 == chdir(MAN_DIR)) {
1.58 schwarze 1028: fprintf(stderr, "MAN_DIR: %s: %s\n",
1.67 schwarze 1029: MAN_DIR, strerror(errno));
1.72 schwarze 1030: pg_error_internal();
1.109 schwarze 1031: return EXIT_FAILURE;
1.104 schwarze 1032: }
1.24 kristaps 1033:
1034: memset(&req, 0, sizeof(struct req));
1.54 schwarze 1035: pathgen(&req);
1.1 kristaps 1036:
1.117 ! schwarze 1037: /* Parse the path info and the query string. */
1.1 kristaps 1038:
1.117 ! schwarze 1039: if ((path = getenv("PATH_INFO")) == NULL)
! 1040: path = "";
! 1041: else if (*path == '/')
! 1042: path++;
! 1043:
! 1044: if (*path != '\0' && access(path, F_OK) == -1) {
! 1045: path_parse(&req, path);
! 1046: path = "";
! 1047: } else if ((querystring = getenv("QUERY_STRING")) != NULL)
1.58 schwarze 1048: http_parse(&req, querystring);
1.77 schwarze 1049:
1.117 ! schwarze 1050: /* Validate parsed data and add defaults. */
! 1051:
1.102 schwarze 1052: if (req.q.manpath == NULL)
1053: req.q.manpath = mandoc_strdup(req.p[0]);
1054: else if ( ! validate_manpath(&req, req.q.manpath)) {
1.77 schwarze 1055: pg_error_badrequest(
1056: "You specified an invalid manpath.");
1.109 schwarze 1057: return EXIT_FAILURE;
1.77 schwarze 1058: }
1.1 kristaps 1059:
1.80 schwarze 1060: if ( ! (NULL == req.q.arch || validate_urifrag(req.q.arch))) {
1061: pg_error_badrequest(
1062: "You specified an invalid architecture.");
1.109 schwarze 1063: return EXIT_FAILURE;
1.80 schwarze 1064: }
1065:
1.66 schwarze 1066: /* Dispatch to the three different pages. */
1.1 kristaps 1067:
1.66 schwarze 1068: if ('\0' != *path)
1069: pg_show(&req, path);
1.85 schwarze 1070: else if (NULL != req.q.query)
1.66 schwarze 1071: pg_search(&req);
1072: else
1.72 schwarze 1073: pg_index(&req);
1.1 kristaps 1074:
1.83 schwarze 1075: free(req.q.manpath);
1076: free(req.q.arch);
1077: free(req.q.sec);
1.85 schwarze 1078: free(req.q.query);
1.52 schwarze 1079: for (i = 0; i < (int)req.psz; i++)
1080: free(req.p[i]);
1.24 kristaps 1081: free(req.p);
1.109 schwarze 1082: return EXIT_SUCCESS;
1.117 ! schwarze 1083: }
! 1084:
! 1085: /*
! 1086: * If PATH_INFO is not a file name, translate it to a query.
! 1087: */
! 1088: static void
! 1089: path_parse(struct req *req, const char *path)
! 1090: {
! 1091: int dir_done;
! 1092:
! 1093: req->q.equal = 1;
! 1094: req->q.manpath = mandoc_strdup(path);
! 1095:
! 1096: /* Mandatory manual page name. */
! 1097: if ((req->q.query = strrchr(req->q.manpath, '/')) == NULL) {
! 1098: req->q.query = req->q.manpath;
! 1099: req->q.manpath = NULL;
! 1100: } else
! 1101: *req->q.query++ = '\0';
! 1102:
! 1103: /* Optional trailing section. */
! 1104: if ((req->q.sec = strrchr(req->q.query, '.')) != NULL) {
! 1105: if(isdigit((unsigned char)req->q.sec[1])) {
! 1106: *req->q.sec++ = '\0';
! 1107: req->q.sec = mandoc_strdup(req->q.sec);
! 1108: } else
! 1109: req->q.sec = NULL;
! 1110: }
! 1111:
! 1112: /* Handle the case of name[.section] only. */
! 1113: if (req->q.manpath == NULL) {
! 1114: req->q.arch = NULL;
! 1115: return;
! 1116: }
! 1117: req->q.query = mandoc_strdup(req->q.query);
! 1118:
! 1119: /* Optional architecture. */
! 1120: dir_done = 0;
! 1121: for (;;) {
! 1122: if ((req->q.arch = strrchr(req->q.manpath, '/')) == NULL)
! 1123: break;
! 1124: *req->q.arch++ = '\0';
! 1125: if (dir_done || strncmp(req->q.arch, "man", 3)) {
! 1126: req->q.arch = mandoc_strdup(req->q.arch);
! 1127: break;
! 1128: }
! 1129:
! 1130: /* Optional directory name. */
! 1131: req->q.arch += 3;
! 1132: if (*req->q.arch != '\0') {
! 1133: free(req->q.sec);
! 1134: req->q.sec = mandoc_strdup(req->q.arch);
! 1135: }
! 1136: dir_done = 1;
! 1137: }
1.24 kristaps 1138: }
1139:
1140: /*
1141: * Scan for indexable paths.
1142: */
1143: static void
1.54 schwarze 1144: pathgen(struct req *req)
1.24 kristaps 1145: {
1.54 schwarze 1146: FILE *fp;
1147: char *dp;
1148: size_t dpsz;
1.115 schwarze 1149: ssize_t len;
1.54 schwarze 1150:
1.74 schwarze 1151: if (NULL == (fp = fopen("manpath.conf", "r"))) {
1152: fprintf(stderr, "%s/manpath.conf: %s\n",
1153: MAN_DIR, strerror(errno));
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)) {
1167: fprintf(stderr, "%s/manpath.conf contains "
1168: "unsafe path \"%s\"\n", MAN_DIR, dp);
1169: pg_error_internal();
1170: exit(EXIT_FAILURE);
1171: }
1172: if (NULL != strchr(dp, '/')) {
1173: fprintf(stderr, "%s/manpath.conf contains "
1174: "path with slash \"%s\"\n", MAN_DIR, dp);
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:
1184: if ( req->p == NULL ) {
1185: fprintf(stderr, "%s/manpath.conf is empty\n", MAN_DIR);
1186: pg_error_internal();
1187: exit(EXIT_FAILURE);
1.24 kristaps 1188: }
1189: }
CVSweb