Annotation of mandoc/cgi.c, Revision 1.30
1.30 ! schwarze 1: /* $Id: cgi.c,v 1.29 2011/12/11 00:22:47 kristaps Exp $ */
1.6 kristaps 2: /*
3: * Copyright (c) 2011 Kristaps Dzonsons <kristaps@bsd.lv>
4: *
5: * Permission to use, copy, modify, and distribute this software for any
6: * purpose with or without fee is hereby granted, provided that the above
7: * copyright notice and this permission notice appear in all copies.
8: *
9: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
10: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
11: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
12: * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
13: * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
14: * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
15: * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
16: */
17: #ifdef HAVE_CONFIG_H
18: #include "config.h"
19: #endif
20:
21: #include <sys/param.h>
22: #include <sys/wait.h>
23:
1.1 kristaps 24: #include <assert.h>
1.6 kristaps 25: #include <ctype.h>
26: #include <errno.h>
1.24 kristaps 27: #include <dirent.h>
1.1 kristaps 28: #include <fcntl.h>
1.6 kristaps 29: #include <limits.h>
1.1 kristaps 30: #include <regex.h>
31: #include <stdio.h>
32: #include <stdarg.h>
1.5 kristaps 33: #include <stdint.h>
1.1 kristaps 34: #include <stdlib.h>
35: #include <string.h>
1.6 kristaps 36: #include <unistd.h>
1.1 kristaps 37:
1.6 kristaps 38: #include "apropos_db.h"
1.4 schwarze 39: #include "mandoc.h"
1.8 kristaps 40: #include "mdoc.h"
41: #include "man.h"
42: #include "main.h"
1.6 kristaps 43: #include "manpath.h"
44:
45: #ifdef __linux__
46: # include <db_185.h>
47: #else
48: # include <db.h>
49: #endif
1.1 kristaps 50:
51: enum page {
52: PAGE_INDEX,
53: PAGE_SEARCH,
1.6 kristaps 54: PAGE_SHOW,
1.1 kristaps 55: PAGE__MAX
56: };
57:
1.24 kristaps 58: struct paths {
59: char *name;
60: char *path;
61: };
62:
1.20 kristaps 63: /*
64: * A query as passed to the search function.
65: */
66: struct query {
67: const char *arch; /* architecture */
68: const char *sec; /* manual section */
69: const char *expr; /* unparsed expression string */
1.26 kristaps 70: int manroot; /* manroot index (or -1)*/
1.20 kristaps 71: int whatis; /* whether whatis mode */
72: int legacy; /* whether legacy mode */
73: };
74:
1.1 kristaps 75: struct req {
1.24 kristaps 76: struct query q;
77: struct paths *p;
78: size_t psz;
1.1 kristaps 79: enum page page;
80: };
81:
1.6 kristaps 82: static int atou(const char *, unsigned *);
1.9 kristaps 83: static void catman(const char *);
1.15 kristaps 84: static int cmp(const void *, const void *);
1.8 kristaps 85: static void format(const char *);
1.6 kristaps 86: static void html_print(const char *);
1.10 kristaps 87: static void html_putchar(char);
1.24 kristaps 88: static int http_decode(char *);
1.26 kristaps 89: static void http_parse(struct req *, char *);
1.24 kristaps 90: static int pathstop(DIR *);
91: static void pathgen(DIR *, char *, struct req *);
92: static void pg_index(const struct req *, char *);
93: static void pg_search(const struct req *, char *);
94: static void pg_show(const struct req *, char *);
1.7 kristaps 95: static void resp_bad(void);
1.6 kristaps 96: static void resp_baddb(void);
1.10 kristaps 97: static void resp_error400(void);
98: static void resp_error404(const char *);
1.6 kristaps 99: static void resp_begin_html(int, const char *);
100: static void resp_begin_http(int, const char *);
101: static void resp_end_html(void);
102: static void resp_index(const struct req *);
103: static void resp_search(struct res *, size_t, void *);
104: static void resp_searchform(const struct req *);
105:
1.29 kristaps 106: static const char *progname; /* cgi script name */
107: static const char *cache; /* cache directory */
108: static const char *css; /* css directory */
109: static const char *host; /* hostname */
1.1 kristaps 110:
111: static const char * const pages[PAGE__MAX] = {
112: "index", /* PAGE_INDEX */
113: "search", /* PAGE_SEARCH */
1.6 kristaps 114: "show", /* PAGE_SHOW */
1.1 kristaps 115: };
116:
1.6 kristaps 117: /*
118: * This is just OpenBSD's strtol(3) suggestion.
119: * I use it instead of strtonum(3) for portability's sake.
120: */
121: static int
122: atou(const char *buf, unsigned *v)
123: {
124: char *ep;
125: long lval;
126:
127: errno = 0;
128: lval = strtol(buf, &ep, 10);
129: if (buf[0] == '\0' || *ep != '\0')
130: return(0);
131: if ((errno == ERANGE && (lval == LONG_MAX ||
132: lval == LONG_MIN)) ||
133: (lval > UINT_MAX || lval < 0))
134: return(0);
135:
136: *v = (unsigned int)lval;
137: return(1);
138: }
1.1 kristaps 139:
1.20 kristaps 140: /*
141: * Print a character, escaping HTML along the way.
142: * This will pass non-ASCII straight to output: be warned!
143: */
1.10 kristaps 144: static void
145: html_putchar(char c)
146: {
147:
148: switch (c) {
149: case ('"'):
150: printf(""e;");
151: break;
152: case ('&'):
153: printf("&");
154: break;
155: case ('>'):
156: printf(">");
157: break;
158: case ('<'):
159: printf("<");
160: break;
161: default:
162: putchar((unsigned char)c);
163: break;
164: }
165: }
166:
1.6 kristaps 167: /*
1.20 kristaps 168: * Call through to html_putchar().
169: * Accepts NULL strings.
1.6 kristaps 170: */
1.1 kristaps 171: static void
1.6 kristaps 172: html_print(const char *p)
1.1 kristaps 173: {
1.6 kristaps 174:
175: if (NULL == p)
176: return;
1.1 kristaps 177: while ('\0' != *p)
1.10 kristaps 178: html_putchar(*p++);
1.1 kristaps 179: }
180:
181: /*
182: * Parse out key-value pairs from an HTTP request variable.
1.6 kristaps 183: * This can be either a cookie or a POST/GET string, although man.cgi
184: * uses only GET for simplicity.
1.1 kristaps 185: */
186: static void
1.26 kristaps 187: http_parse(struct req *req, char *p)
1.1 kristaps 188: {
1.26 kristaps 189: char *key, *val, *manroot;
1.24 kristaps 190: size_t sz;
1.26 kristaps 191: int i, legacy;
1.1 kristaps 192:
1.26 kristaps 193: memset(&req->q, 0, sizeof(struct query));
1.24 kristaps 194:
1.26 kristaps 195: req->q.whatis = 1;
1.24 kristaps 196: legacy = -1;
1.26 kristaps 197: manroot = NULL;
1.1 kristaps 198:
199: while (p && '\0' != *p) {
200: while (' ' == *p)
201: p++;
202:
203: key = p;
204: val = NULL;
205:
206: if (NULL != (p = strchr(p, '='))) {
207: *p++ = '\0';
208: val = p;
209:
210: sz = strcspn(p, ";&");
211: /* LINTED */
212: p += sz;
213:
214: if ('\0' != *p)
215: *p++ = '\0';
216: } else {
217: p = key;
218: sz = strcspn(p, ";&");
219: /* LINTED */
220: p += sz;
221:
222: if ('\0' != *p)
223: p++;
224: continue;
225: }
226:
227: if ('\0' == *key || '\0' == *val)
228: continue;
229:
230: /* Just abort handling. */
231:
1.24 kristaps 232: if ( ! http_decode(key))
233: break;
234: if ( ! http_decode(val))
235: break;
236:
237: if (0 == strcmp(key, "expr"))
1.26 kristaps 238: req->q.expr = val;
1.24 kristaps 239: else if (0 == strcmp(key, "query"))
1.26 kristaps 240: req->q.expr = val;
1.24 kristaps 241: else if (0 == strcmp(key, "sec"))
1.26 kristaps 242: req->q.sec = val;
1.24 kristaps 243: else if (0 == strcmp(key, "sektion"))
1.26 kristaps 244: req->q.sec = val;
1.24 kristaps 245: else if (0 == strcmp(key, "arch"))
1.26 kristaps 246: req->q.arch = val;
247: else if (0 == strcmp(key, "manpath"))
248: manroot = val;
1.24 kristaps 249: else if (0 == strcmp(key, "apropos"))
250: legacy = 0 == strcmp(val, "0");
251: else if (0 == strcmp(key, "op"))
1.26 kristaps 252: req->q.whatis = 0 == strcasecmp(val, "whatis");
1.24 kristaps 253: }
1.1 kristaps 254:
1.24 kristaps 255: /* Test for old man.cgi compatibility mode. */
1.1 kristaps 256:
1.24 kristaps 257: if (legacy == 0) {
1.26 kristaps 258: req->q.whatis = 0;
259: req->q.legacy = 1;
1.24 kristaps 260: } else if (legacy > 0) {
1.26 kristaps 261: req->q.legacy = 1;
262: req->q.whatis = 1;
1.1 kristaps 263: }
1.24 kristaps 264:
265: /*
266: * Section "0" means no section when in legacy mode.
267: * For some man.cgi scripts, "default" arch is none.
268: */
269:
1.26 kristaps 270: if (req->q.legacy && NULL != req->q.sec)
271: if (0 == strcmp(req->q.sec, "0"))
272: req->q.sec = NULL;
273: if (req->q.legacy && NULL != req->q.arch)
274: if (0 == strcmp(req->q.arch, "default"))
275: req->q.arch = NULL;
276:
277: /* Default to first manroot. */
278:
279: if (NULL != manroot) {
280: for (i = 0; i < (int)req->psz; i++)
281: if (0 == strcmp(req->p[i].name, manroot))
282: break;
283: req->q.manroot = i < (int)req->psz ? i : -1;
284: }
1.1 kristaps 285: }
286:
287: /*
1.6 kristaps 288: * HTTP-decode a string. The standard explanation is that this turns
289: * "%4e+foo" into "n foo" in the regular way. This is done in-place
290: * over the allocated string.
1.1 kristaps 291: */
292: static int
1.24 kristaps 293: http_decode(char *p)
1.1 kristaps 294: {
295: char hex[3];
296: int c;
297:
298: hex[2] = '\0';
299:
300: for ( ; '\0' != *p; p++) {
301: if ('%' == *p) {
302: if ('\0' == (hex[0] = *(p + 1)))
303: return(0);
304: if ('\0' == (hex[1] = *(p + 2)))
305: return(0);
306: if (1 != sscanf(hex, "%x", &c))
307: return(0);
308: if ('\0' == c)
309: return(0);
310:
311: *p = (char)c;
312: memmove(p + 1, p + 3, strlen(p + 3) + 1);
313: } else
314: *p = '+' == *p ? ' ' : *p;
315: }
316:
317: *p = '\0';
318: return(1);
319: }
320:
1.6 kristaps 321: static void
322: resp_begin_http(int code, const char *msg)
323: {
324:
325: if (200 != code)
326: printf("Status: %d %s\n", code, msg);
327:
1.20 kristaps 328: puts("Content-Type: text/html; charset=utf-8\n"
329: "Cache-Control: no-cache\n"
330: "Pragma: no-cache\n"
1.6 kristaps 331: "");
332:
333: fflush(stdout);
334: }
335:
336: static void
337: resp_begin_html(int code, const char *msg)
338: {
339:
340: resp_begin_http(code, msg);
341:
1.29 kristaps 342: printf("<!DOCTYPE HTML PUBLIC "
343: " \"-//W3C//DTD HTML 4.01//EN\""
344: " \"http://www.w3.org/TR/html4/strict.dtd\">\n"
345: "<HTML>\n"
346: "<HEAD>\n"
347: "<META HTTP-EQUIV=\"Content-Type\""
348: " CONTENT=\"text/html; charset=utf-8\">\n"
349: "<LINK REL=\"stylesheet\" HREF=\"%s/man.cgi.css\""
350: " TYPE=\"text/css\" media=\"all\">\n"
351: "<TITLE>System Manpage Reference</TITLE>\n"
352: "</HEAD>\n"
353: "<BODY>\n"
354: "<!-- Begin page content. //-->\n", css);
1.6 kristaps 355: }
356:
357: static void
358: resp_end_html(void)
359: {
360:
1.20 kristaps 361: puts("</BODY>\n"
362: "</HTML>");
1.6 kristaps 363: }
364:
365: static void
366: resp_searchform(const struct req *req)
367: {
1.27 kristaps 368: int i;
1.13 kristaps 369:
1.6 kristaps 370: puts("<!-- Begin search form. //-->");
1.29 kristaps 371: printf("<FORM ACTION=\"%s/search.html\" METHOD=\"get\">\n"
372: "<FIELDSET>\n"
1.16 kristaps 373: "<LEGEND>Search Parameters</LEGEND>\n"
1.20 kristaps 374: "<INPUT TYPE=\"submit\" NAME=\"op\""
375: " VALUE=\"Whatis\"> or \n"
376: "<INPUT TYPE=\"submit\" NAME=\"op\""
377: " VALUE=\"apropos\"> for manuals satisfying \n"
1.29 kristaps 378: "<INPUT TYPE=\"text\" NAME=\"expr\" VALUE=\"",
379: progname);
1.24 kristaps 380: html_print(req->q.expr ? req->q.expr : "");
1.14 kristaps 381: printf("\">, section "
1.20 kristaps 382: "<INPUT TYPE=\"text\""
383: " SIZE=\"4\" NAME=\"sec\" VALUE=\"");
1.24 kristaps 384: html_print(req->q.sec ? req->q.sec : "");
1.14 kristaps 385: printf("\">, arch "
1.20 kristaps 386: "<INPUT TYPE=\"text\""
387: " SIZE=\"8\" NAME=\"arch\" VALUE=\"");
1.24 kristaps 388: html_print(req->q.arch ? req->q.arch : "");
1.27 kristaps 389: printf("\">");
390: if (req->psz > 1) {
391: puts(", <SELECT NAME=\"manpath\">");
392: for (i = 0; i < (int)req->psz; i++) {
393: printf("<OPTION %s VALUE=\"",
394: (i == req->q.manroot) ||
395: (0 == i && -1 == req->q.manroot) ?
396: "SELECTED=\"selected\"" : "");
397: html_print(req->p[i].name);
398: printf("\">");
399: html_print(req->p[i].name);
400: puts("</OPTION>");
401: }
402: puts("</SELECT>");
403: }
404: puts(".\n"
1.12 kristaps 405: "<INPUT TYPE=\"reset\" VALUE=\"Reset\">\n"
406: "</FIELDSET>\n"
1.20 kristaps 407: "</FORM>");
408: puts("<!-- End search form. //-->");
1.6 kristaps 409: }
410:
411: static void
412: resp_index(const struct req *req)
413: {
414:
415: resp_begin_html(200, NULL);
416: resp_searchform(req);
417: resp_end_html();
418: }
419:
420: static void
1.10 kristaps 421: resp_error400(void)
1.9 kristaps 422: {
423:
1.10 kristaps 424: resp_begin_html(400, "Query Malformed");
1.12 kristaps 425: printf("<H1>Malformed Query</H1>\n"
426: "<P>\n"
1.20 kristaps 427: "The query your entered was malformed.\n"
428: "Try again from the\n"
1.22 kristaps 429: "<A HREF=\"%s/index.html\">main page</A>.\n"
1.12 kristaps 430: "</P>", progname);
1.9 kristaps 431: resp_end_html();
432: }
433:
434: static void
1.10 kristaps 435: resp_error404(const char *page)
1.6 kristaps 436: {
437:
438: resp_begin_html(404, "Not Found");
1.10 kristaps 439: puts("<H1>Page Not Found</H1>\n"
440: "<P>\n"
1.20 kristaps 441: "The page you're looking for, ");
442: printf("<B>");
1.10 kristaps 443: html_print(page);
1.12 kristaps 444: printf("</B>,\n"
1.20 kristaps 445: "could not be found.\n"
446: "Try searching from the\n"
1.22 kristaps 447: "<A HREF=\"%s/index.html\">main page</A>.\n"
1.12 kristaps 448: "</P>", progname);
1.6 kristaps 449: resp_end_html();
450: }
1.1 kristaps 451:
452: static void
1.7 kristaps 453: resp_bad(void)
454: {
455: resp_begin_html(500, "Internal Server Error");
456: puts("<P>Generic badness happened.</P>");
457: resp_end_html();
458: }
459:
460: static void
1.6 kristaps 461: resp_baddb(void)
1.1 kristaps 462: {
463:
1.6 kristaps 464: resp_begin_html(500, "Internal Server Error");
465: puts("<P>Your database is broken.</P>");
466: resp_end_html();
1.1 kristaps 467: }
468:
469: static void
1.6 kristaps 470: resp_search(struct res *r, size_t sz, void *arg)
1.1 kristaps 471: {
1.20 kristaps 472: int i;
1.19 kristaps 473: const struct req *req;
474:
1.29 kristaps 475: req = (const struct req *)arg;
476: assert(req->q.manroot >= 0);
477:
1.6 kristaps 478: if (1 == sz) {
479: /*
480: * If we have just one result, then jump there now
481: * without any delay.
482: */
483: puts("Status: 303 See Other");
1.29 kristaps 484: printf("Location: http://%s%s/show/%d/%u/%u.html\n",
485: host, progname, req->q.manroot,
1.6 kristaps 486: r[0].volume, r[0].rec);
487: puts("Content-Type: text/html; charset=utf-8\n");
488: return;
489: }
490:
1.15 kristaps 491: qsort(r, sz, sizeof(struct res), cmp);
492:
1.12 kristaps 493: resp_begin_html(200, NULL);
1.19 kristaps 494: resp_searchform(req);
1.6 kristaps 495:
1.14 kristaps 496: if (0 == sz) {
1.21 kristaps 497: printf("<P>\n"
1.24 kristaps 498: "No %s results found.\n",
499: req->q.whatis ? "whatis" : "apropos");
500: if (req->q.whatis) {
1.29 kristaps 501: printf("(Try <A HREF=\"%s/search.html?"
502: "op=apropos&expr=", progname);
1.24 kristaps 503: html_print(req->q.expr ? req->q.expr : "");
1.19 kristaps 504: printf("&sec=");
1.24 kristaps 505: html_print(req->q.sec ? req->q.sec : "");
1.19 kristaps 506: printf("&arch=");
1.24 kristaps 507: html_print(req->q.arch ? req->q.arch : "");
1.19 kristaps 508: puts("\">apropos</A>?)");
509: }
510: puts("</P>");
1.14 kristaps 511: resp_end_html();
512: return;
513: }
514:
515: puts("<P></P>\n"
516: "<TABLE>");
1.1 kristaps 517:
518: for (i = 0; i < (int)sz; i++) {
1.20 kristaps 519: printf("<TR>\n"
520: "<TD CLASS=\"title\">\n"
1.29 kristaps 521: "<A HREF=\"%s/show/%d/%u/%u.html\">",
522: progname, req->q.manroot,
1.26 kristaps 523: r[i].volume, r[i].rec);
1.15 kristaps 524: html_print(r[i].title);
1.1 kristaps 525: putchar('(');
1.6 kristaps 526: html_print(r[i].cat);
527: if (r[i].arch && '\0' != *r[i].arch) {
528: putchar('/');
529: html_print(r[i].arch);
530: }
1.20 kristaps 531: printf(")</A>\n"
532: "</TD>\n"
533: "<TD CLASS=\"desc\">");
1.6 kristaps 534: html_print(r[i].desc);
1.20 kristaps 535: puts("</TD>\n"
536: "</TR>");
1.1 kristaps 537: }
1.16 kristaps 538:
539: puts("</TABLE>");
1.6 kristaps 540: resp_end_html();
541: }
542:
543: /* ARGSUSED */
544: static void
1.24 kristaps 545: pg_index(const struct req *req, char *path)
1.6 kristaps 546: {
547:
548: resp_index(req);
1.1 kristaps 549: }
550:
551: static void
1.9 kristaps 552: catman(const char *file)
553: {
1.10 kristaps 554: FILE *f;
555: size_t len;
556: int i;
557: char *p;
558: int italic, bold;
1.9 kristaps 559:
1.10 kristaps 560: if (NULL == (f = fopen(file, "r"))) {
1.9 kristaps 561: resp_baddb();
562: return;
563: }
564:
1.12 kristaps 565: resp_begin_http(200, NULL);
1.29 kristaps 566: printf("<!DOCTYPE HTML PUBLIC "
567: " \"-//W3C//DTD HTML 4.01//EN\""
568: " \"http://www.w3.org/TR/html4/strict.dtd\">\n"
569: "<HTML>\n"
570: "<HEAD>\n"
571: "<META HTTP-EQUIV=\"Content-Type\""
572: " CONTENT=\"text/html; charset=utf-8\">\n"
573: "<LINK REL=\"stylesheet\" HREF=\"%s/catman.css\""
574: " TYPE=\"text/css\" media=\"all\">\n"
575: "<TITLE>System Manpage Reference</TITLE>\n"
576: "</HEAD>\n"
577: "<BODY>\n"
578: "<!-- Begin page content. //-->\n"
579: "<PRE>\n", css);
1.10 kristaps 580:
581: while (NULL != (p = fgetln(f, &len))) {
582: bold = italic = 0;
583: for (i = 0; i < (int)len - 1; i++) {
584: /*
585: * This means that the catpage is out of state.
586: * Ignore it and keep going (although the
587: * catpage is bogus).
588: */
589:
590: if ('\b' == p[i] || '\n' == p[i])
591: continue;
592:
593: /*
594: * Print a regular character.
595: * Close out any bold/italic scopes.
596: * If we're in back-space mode, make sure we'll
597: * have something to enter when we backspace.
598: */
599:
600: if ('\b' != p[i + 1]) {
601: if (italic)
602: printf("</I>");
603: if (bold)
604: printf("</B>");
605: italic = bold = 0;
606: html_putchar(p[i]);
607: continue;
608: } else if (i + 2 >= (int)len)
609: continue;
610:
611: /* Italic mode. */
612:
613: if ('_' == p[i]) {
614: if (bold)
615: printf("</B>");
616: if ( ! italic)
617: printf("<I>");
618: bold = 0;
619: italic = 1;
620: i += 2;
621: html_putchar(p[i]);
622: continue;
623: }
624:
625: /*
626: * Handle funny behaviour troff-isms.
627: * These grok'd from the original man2html.c.
628: */
629:
630: if (('+' == p[i] && 'o' == p[i + 2]) ||
631: ('o' == p[i] && '+' == p[i + 2]) ||
632: ('|' == p[i] && '=' == p[i + 2]) ||
633: ('=' == p[i] && '|' == p[i + 2]) ||
634: ('*' == p[i] && '=' == p[i + 2]) ||
635: ('=' == p[i] && '*' == p[i + 2]) ||
636: ('*' == p[i] && '|' == p[i + 2]) ||
637: ('|' == p[i] && '*' == p[i + 2])) {
638: if (italic)
639: printf("</I>");
640: if (bold)
641: printf("</B>");
642: italic = bold = 0;
643: putchar('*');
644: i += 2;
645: continue;
646: } else if (('|' == p[i] && '-' == p[i + 2]) ||
647: ('-' == p[i] && '|' == p[i + 1]) ||
648: ('+' == p[i] && '-' == p[i + 1]) ||
649: ('-' == p[i] && '+' == p[i + 1]) ||
650: ('+' == p[i] && '|' == p[i + 1]) ||
651: ('|' == p[i] && '+' == p[i + 1])) {
652: if (italic)
653: printf("</I>");
654: if (bold)
655: printf("</B>");
656: italic = bold = 0;
657: putchar('+');
658: i += 2;
659: continue;
660: }
661:
662: /* Bold mode. */
663:
664: if (italic)
665: printf("</I>");
666: if ( ! bold)
667: printf("<B>");
668: bold = 1;
669: italic = 0;
670: i += 2;
671: html_putchar(p[i]);
672: }
673:
674: /*
675: * Clean up the last character.
676: * We can get to a newline; don't print that.
677: */
1.9 kristaps 678:
1.10 kristaps 679: if (italic)
680: printf("</I>");
681: if (bold)
682: printf("</B>");
1.9 kristaps 683:
1.10 kristaps 684: if (i == (int)len - 1 && '\n' != p[i])
685: html_putchar(p[i]);
1.9 kristaps 686:
1.10 kristaps 687: putchar('\n');
688: }
689:
690: puts("</PRE>\n"
691: "</BODY>\n"
692: "</HTML>");
693:
694: fclose(f);
1.9 kristaps 695: }
696:
697: static void
1.8 kristaps 698: format(const char *file)
1.7 kristaps 699: {
1.8 kristaps 700: struct mparse *mp;
701: int fd;
702: struct mdoc *mdoc;
703: struct man *man;
704: void *vp;
705: enum mandoclevel rc;
1.10 kristaps 706: char opts[MAXPATHLEN + 128];
1.7 kristaps 707:
1.8 kristaps 708: if (-1 == (fd = open(file, O_RDONLY, 0))) {
709: resp_baddb();
1.7 kristaps 710: return;
711: }
712:
1.8 kristaps 713: mp = mparse_alloc(MPARSE_AUTO, MANDOCLEVEL_FATAL, NULL, NULL);
714: rc = mparse_readfd(mp, fd, file);
715: close(fd);
1.7 kristaps 716:
1.8 kristaps 717: if (rc >= MANDOCLEVEL_FATAL) {
1.7 kristaps 718: resp_baddb();
719: return;
720: }
721:
1.29 kristaps 722: snprintf(opts, sizeof(opts), "style=%s/man.css,"
1.10 kristaps 723: "man=%s/search.html?sec=%%S&expr=%%N,"
1.11 kristaps 724: /*"includes=/cgi-bin/man.cgi/usr/include/%%I"*/,
1.29 kristaps 725: css, progname);
1.10 kristaps 726:
1.8 kristaps 727: mparse_result(mp, &mdoc, &man);
1.10 kristaps 728: vp = html_alloc(opts);
1.7 kristaps 729:
1.8 kristaps 730: if (NULL != mdoc) {
731: resp_begin_http(200, NULL);
732: html_mdoc(vp, mdoc);
733: } else if (NULL != man) {
734: resp_begin_http(200, NULL);
735: html_man(vp, man);
736: } else
737: resp_baddb();
1.7 kristaps 738:
1.8 kristaps 739: html_free(vp);
740: mparse_free(mp);
1.7 kristaps 741: }
742:
743: static void
1.24 kristaps 744: pg_show(const struct req *req, char *path)
1.1 kristaps 745: {
1.24 kristaps 746: struct manpaths ps;
1.6 kristaps 747: char *sub;
1.7 kristaps 748: char file[MAXPATHLEN];
1.9 kristaps 749: const char *fn, *cp;
1.6 kristaps 750: int rc;
1.25 kristaps 751: unsigned int vol, rec, mr;
1.9 kristaps 752: DB *idx;
1.6 kristaps 753: DBT key, val;
754:
1.24 kristaps 755: idx = NULL;
756:
1.25 kristaps 757: /* Parse out mroot, volume, and record from the path. */
758:
759: if (NULL == path || NULL == (sub = strchr(path, '/'))) {
760: resp_error400();
761: return;
762: }
763: *sub++ = '\0';
764: if ( ! atou(path, &mr)) {
765: resp_error400();
766: return;
767: }
768: path = sub;
769: if (NULL == (sub = strchr(path, '/'))) {
770: resp_error400();
771: return;
772: }
773: *sub++ = '\0';
774: if ( ! atou(path, &vol) || ! atou(sub, &rec)) {
1.10 kristaps 775: resp_error400();
1.6 kristaps 776: return;
1.25 kristaps 777: } else if (mr >= (unsigned int)req->psz) {
1.10 kristaps 778: resp_error400();
1.6 kristaps 779: return;
1.25 kristaps 780: }
1.6 kristaps 781:
1.24 kristaps 782: /*
1.26 kristaps 783: * Begin by chdir()ing into the manroot.
1.24 kristaps 784: * This way we can pick up the database files, which are
785: * relative to the manpath root.
786: */
787:
1.25 kristaps 788: if (-1 == chdir(req->p[(int)mr].path)) {
789: perror(req->p[(int)mr].path);
790: resp_baddb();
1.24 kristaps 791: return;
792: }
793:
794: memset(&ps, 0, sizeof(struct manpaths));
1.30 ! schwarze 795: manpath_manconf(&ps, "etc/catman.conf");
1.24 kristaps 796:
1.25 kristaps 797: if (vol >= (unsigned int)ps.sz) {
1.10 kristaps 798: resp_error400();
1.24 kristaps 799: goto out;
1.6 kristaps 800: }
801:
1.24 kristaps 802: strlcpy(file, ps.paths[vol], MAXPATHLEN);
1.6 kristaps 803: strlcat(file, "/mandoc.index", MAXPATHLEN);
804:
805: /* Open the index recno(3) database. */
806:
1.9 kristaps 807: idx = dbopen(file, O_RDONLY, 0, DB_RECNO, NULL);
808: if (NULL == idx) {
1.24 kristaps 809: perror(file);
1.6 kristaps 810: resp_baddb();
1.24 kristaps 811: goto out;
1.6 kristaps 812: }
813:
814: key.data = &rec;
815: key.size = 4;
816:
1.9 kristaps 817: if (0 != (rc = (*idx->get)(idx, &key, &val, 0))) {
1.10 kristaps 818: rc < 0 ? resp_baddb() : resp_error400();
1.9 kristaps 819: goto out;
1.6 kristaps 820: }
821:
1.9 kristaps 822: cp = (char *)val.data;
1.6 kristaps 823:
1.9 kristaps 824: if (NULL == (fn = memchr(cp, '\0', val.size)))
825: resp_baddb();
826: else if (++fn - cp >= (int)val.size)
827: resp_baddb();
828: else if (NULL == memchr(fn, '\0', val.size - (fn - cp)))
829: resp_baddb();
830: else {
831: if (0 == strcmp(cp, "cat"))
1.24 kristaps 832: catman(fn + 1);
1.9 kristaps 833: else
1.24 kristaps 834: format(fn + 1);
1.9 kristaps 835: }
836: out:
1.24 kristaps 837: if (idx)
838: (*idx->close)(idx);
839: manpath_free(&ps);
1.6 kristaps 840: }
841:
842: static void
1.24 kristaps 843: pg_search(const struct req *req, char *path)
1.6 kristaps 844: {
845: size_t tt;
1.24 kristaps 846: struct manpaths ps;
1.20 kristaps 847: int i, sz, rc;
1.6 kristaps 848: const char *ep, *start;
849: char **cp;
850: struct opts opt;
851: struct expr *expr;
852:
1.28 kristaps 853: if (req->q.manroot < 0 || 0 == req->psz) {
1.24 kristaps 854: resp_search(NULL, 0, (void *)req);
855: return;
856: }
857:
1.1 kristaps 858: memset(&opt, 0, sizeof(struct opts));
1.6 kristaps 859:
1.24 kristaps 860: ep = req->q.expr;
861: opt.arch = req->q.arch;
862: opt.cat = req->q.sec;
1.20 kristaps 863: rc = -1;
864: sz = 0;
865: cp = NULL;
1.6 kristaps 866:
867: /*
1.24 kristaps 868: * Begin by chdir()ing into the root of the manpath.
869: * This way we can pick up the database files, which are
870: * relative to the manpath root.
871: */
872:
1.26 kristaps 873: assert(req->q.manroot < (int)req->psz);
874: if (-1 == (chdir(req->p[req->q.manroot].path))) {
875: perror(req->p[req->q.manroot].path);
1.24 kristaps 876: resp_search(NULL, 0, (void *)req);
877: return;
878: }
879:
880: memset(&ps, 0, sizeof(struct manpaths));
1.30 ! schwarze 881: manpath_manconf(&ps, "etc/catman.conf");
1.24 kristaps 882:
883: /*
884: * Poor man's tokenisation: just break apart by spaces.
1.6 kristaps 885: * Yes, this is half-ass. But it works for now.
886: */
887:
888: while (ep && isspace((unsigned char)*ep))
889: ep++;
890:
891: while (ep && '\0' != *ep) {
892: cp = mandoc_realloc(cp, (sz + 1) * sizeof(char *));
893: start = ep;
894: while ('\0' != *ep && ! isspace((unsigned char)*ep))
895: ep++;
896: cp[sz] = mandoc_malloc((ep - start) + 1);
897: memcpy(cp[sz], start, ep - start);
898: cp[sz++][ep - start] = '\0';
899: while (isspace((unsigned char)*ep))
900: ep++;
901: }
902:
903: /*
904: * Pump down into apropos backend.
905: * The resp_search() function is called with the results.
906: */
907:
1.24 kristaps 908: expr = req->q.whatis ?
909: termcomp(sz, cp, &tt) : exprcomp(sz, cp, &tt);
1.12 kristaps 910:
911: if (NULL != expr)
1.6 kristaps 912: rc = apropos_search
1.24 kristaps 913: (ps.sz, ps.paths, &opt,
1.6 kristaps 914: expr, tt, (void *)req, resp_search);
915:
916: /* ...unless errors occured. */
917:
918: if (0 == rc)
919: resp_baddb();
920: else if (-1 == rc)
1.10 kristaps 921: resp_search(NULL, 0, (void *)req);
1.6 kristaps 922:
923: for (i = 0; i < sz; i++)
924: free(cp[i]);
925:
926: free(cp);
927: exprfree(expr);
1.24 kristaps 928: manpath_free(&ps);
1.1 kristaps 929: }
930:
931: int
932: main(void)
933: {
934: int i;
1.24 kristaps 935: char buf[MAXPATHLEN];
936: DIR *cwd;
1.1 kristaps 937: struct req req;
1.6 kristaps 938: char *p, *path, *subpath;
939:
1.24 kristaps 940: /* Scan our run-time environment. */
1.6 kristaps 941:
1.29 kristaps 942: if (NULL == (cache = getenv("CACHE_DIR")))
943: cache = "/cache/man.cgi";
944:
945: if (NULL == (progname = getenv("SCRIPT_NAME")))
1.6 kristaps 946: progname = "";
947:
1.29 kristaps 948: if (NULL == (css = getenv("CSS_DIR")))
949: css = "/";
1.7 kristaps 950:
1.29 kristaps 951: if (NULL == (host = getenv("HTTP_HOST")))
1.24 kristaps 952: host = "localhost";
953:
954: /*
955: * First we change directory into the cache directory so that
956: * subsequent scanning for manpath directories is rooted
957: * relative to the same position.
958: */
959:
1.8 kristaps 960: if (-1 == chdir(cache)) {
1.24 kristaps 961: perror(cache);
962: resp_bad();
963: return(EXIT_FAILURE);
964: } else if (NULL == (cwd = opendir(cache))) {
965: perror(cache);
1.8 kristaps 966: resp_bad();
967: return(EXIT_FAILURE);
1.24 kristaps 968: }
969:
970: memset(&req, 0, sizeof(struct req));
1.7 kristaps 971:
1.24 kristaps 972: strlcpy(buf, ".", MAXPATHLEN);
973: pathgen(cwd, buf, &req);
974: closedir(cwd);
1.1 kristaps 975:
1.24 kristaps 976: /* Next parse out the query string. */
1.1 kristaps 977:
978: if (NULL != (p = getenv("QUERY_STRING")))
1.26 kristaps 979: http_parse(&req, p);
1.1 kristaps 980:
1.24 kristaps 981: /*
982: * Now juggle paths to extract information.
983: * We want to extract our filetype (the file suffix), the
984: * initial path component, then the trailing component(s).
985: * Start with leading subpath component.
986: */
1.1 kristaps 987:
1.6 kristaps 988: subpath = path = NULL;
1.1 kristaps 989: req.page = PAGE__MAX;
990:
991: if (NULL == (path = getenv("PATH_INFO")) || '\0' == *path)
992: req.page = PAGE_INDEX;
1.6 kristaps 993:
1.1 kristaps 994: if (NULL != path && '/' == *path && '\0' == *++path)
995: req.page = PAGE_INDEX;
996:
1.6 kristaps 997: /* Strip file suffix. */
998:
999: if (NULL != path && NULL != (p = strrchr(path, '.')))
1000: if (NULL != p && NULL == strchr(p, '/'))
1001: *p++ = '\0';
1002:
1003: /* Resolve subpath component. */
1.1 kristaps 1004:
1005: if (NULL != path && NULL != (subpath = strchr(path, '/')))
1.6 kristaps 1006: *subpath++ = '\0';
1.1 kristaps 1007:
1.6 kristaps 1008: /* Map path into one we recognise. */
1.1 kristaps 1009:
1010: if (NULL != path && '\0' != *path)
1011: for (i = 0; i < (int)PAGE__MAX; i++)
1012: if (0 == strcmp(pages[i], path)) {
1013: req.page = (enum page)i;
1014: break;
1015: }
1016:
1.6 kristaps 1017: /* Route pages. */
1018:
1.1 kristaps 1019: switch (req.page) {
1020: case (PAGE_INDEX):
1.24 kristaps 1021: pg_index(&req, subpath);
1.1 kristaps 1022: break;
1023: case (PAGE_SEARCH):
1.24 kristaps 1024: pg_search(&req, subpath);
1.6 kristaps 1025: break;
1026: case (PAGE_SHOW):
1.24 kristaps 1027: pg_show(&req, subpath);
1.1 kristaps 1028: break;
1029: default:
1.10 kristaps 1030: resp_error404(path);
1.1 kristaps 1031: break;
1032: }
1033:
1.24 kristaps 1034: for (i = 0; i < (int)req.psz; i++) {
1035: free(req.p[i].path);
1036: free(req.p[i].name);
1037: }
1.6 kristaps 1038:
1.24 kristaps 1039: free(req.p);
1.1 kristaps 1040: return(EXIT_SUCCESS);
1041: }
1.15 kristaps 1042:
1043: static int
1044: cmp(const void *p1, const void *p2)
1045: {
1046:
1047: return(strcasecmp(((const struct res *)p1)->title,
1048: ((const struct res *)p2)->title));
1049: }
1050:
1.24 kristaps 1051: /*
1052: * Check to see if an "etc" path consists of a catman.conf file. If it
1053: * does, that means that the path contains a tree created by catman(8)
1054: * and should be used for indexing.
1055: */
1056: static int
1057: pathstop(DIR *dir)
1058: {
1059: struct dirent *d;
1060:
1061: while (NULL != (d = readdir(dir)))
1062: if (DT_REG == d->d_type)
1063: if (0 == strcmp(d->d_name, "catman.conf"))
1064: return(1);
1065:
1066: return(0);
1067: }
1068:
1069: /*
1070: * Scan for indexable paths.
1071: * This adds all paths with "etc/catman.conf" to the buffer.
1072: */
1073: static void
1074: pathgen(DIR *dir, char *path, struct req *req)
1075: {
1076: struct dirent *d;
1077: char *cp;
1078: DIR *cd;
1079: int rc;
1080: size_t sz, ssz;
1081:
1082: sz = strlcat(path, "/", MAXPATHLEN);
1083: if (sz >= MAXPATHLEN) {
1084: fprintf(stderr, "%s: Path too long", path);
1085: return;
1086: }
1087:
1088: /*
1089: * First, scan for the "etc" directory.
1090: * If it's found, then see if it should cause us to stop. This
1091: * happens when a catman.conf is found in the directory.
1092: */
1093:
1094: rc = 0;
1095: while (0 == rc && NULL != (d = readdir(dir))) {
1096: if (DT_DIR != d->d_type || strcmp(d->d_name, "etc"))
1097: continue;
1098:
1099: path[(int)sz] = '\0';
1100: ssz = strlcat(path, d->d_name, MAXPATHLEN);
1101:
1102: if (ssz >= MAXPATHLEN) {
1103: fprintf(stderr, "%s: Path too long", path);
1104: return;
1105: } else if (NULL == (cd = opendir(path))) {
1106: perror(path);
1107: return;
1108: }
1109:
1110: rc = pathstop(cd);
1111: closedir(cd);
1112: }
1113:
1114: if (rc > 0) {
1115: /* This also strips the trailing slash. */
1.27 kristaps 1116: path[(int)--sz] = '\0';
1.24 kristaps 1117: req->p = mandoc_realloc
1118: (req->p,
1119: (req->psz + 1) * sizeof(struct paths));
1.27 kristaps 1120: /*
1121: * Strip out the leading "./" unless we're just a ".",
1122: * in which case use an empty string as our name.
1123: */
1.24 kristaps 1124: req->p[(int)req->psz].path = mandoc_strdup(path);
1125: req->p[(int)req->psz].name =
1.27 kristaps 1126: cp = mandoc_strdup(path + (1 == sz ? 1 : 2));
1.24 kristaps 1127: req->psz++;
1128: /*
1129: * The name is just the path with all the slashes taken
1130: * out of it. Simple but effective.
1131: */
1132: for ( ; '\0' != *cp; cp++)
1133: if ('/' == *cp)
1134: *cp = ' ';
1135: return;
1136: }
1137:
1138: /*
1139: * If no etc/catman.conf was found, recursively enter child
1140: * directory and continue scanning.
1141: */
1142:
1143: rewinddir(dir);
1144: while (NULL != (d = readdir(dir))) {
1145: if (DT_DIR != d->d_type || '.' == d->d_name[0])
1146: continue;
1147:
1148: path[(int)sz] = '\0';
1149: ssz = strlcat(path, d->d_name, MAXPATHLEN);
1150:
1151: if (ssz >= MAXPATHLEN) {
1152: fprintf(stderr, "%s: Path too long", path);
1153: return;
1154: } else if (NULL == (cd = opendir(path))) {
1155: perror(path);
1156: return;
1157: }
1158:
1159: pathgen(cd, path, req);
1160: closedir(cd);
1161: }
1162: }
CVSweb