sys/src/cmd/webcookies.c

   1 /*
   2  * Cookie file system.  Allows hget and multiple webfs's to collaborate.
   3  * Conventionally mounted on /mnt/webcookies.
   4  */
   5
   6 #include <u.h>
   7 #include <libc.h>
   8 #include <bio.h>
   9 #include <ndb.h>
  10 #include <fcall.h>
  11 #include <thread.h>
  12 #include <9p.h>
  13 #include <ctype.h>
  14
  15 int debug = 0;
  16
  17 typedef struct Cookie Cookie;
  18 typedef struct Jar Jar;
  19
  20 struct Cookie
  21 {
  22         /* external info */
  23         char*   name;
  24         char*   value;
  25         char*   dom;            /* starts with . */
  26         char*   path;
  27         char*   version;
  28         char*   comment;        /* optional, may be nil */
  29
  30         uint    expire;         /* time of expiration: ~0 means when webcookies dies */
  31         int     secure;
  32         int     explicitdom;    /* dom was explicitly set */
  33         int     explicitpath;   /* path was explicitly set */
  34         int     netscapestyle;
  35
  36         /* internal info */
  37         int     deleted;
  38         int     mark;
  39         int     ondisk;
  40 };
  41
  42 struct Jar
  43 {
  44         Cookie  *c;
  45         int     nc;
  46         int     mc;
  47
  48         Qid     qid;
  49         int     dirty;
  50         char    *file;
  51         char    *lockfile;
  52 };
  53
  54 struct {
  55         char    *s;
  56         int     offset;
  57         int     ishttp;
  58 } stab[] = {
  59         "domain",               offsetof(Cookie, dom),          1,
  60         "path",                 offsetof(Cookie, path),         1,
  61         "name",                 offsetof(Cookie, name),         0,
  62         "value",                offsetof(Cookie, value),        0,
  63         "comment",              offsetof(Cookie, comment),      1,
  64         "version",              offsetof(Cookie, version),      1,
  65 };
  66
  67 struct {
  68         char *s;
  69         int     offset;
  70 } itab[] = {
  71         "expire",               offsetof(Cookie, expire),
  72         "secure",               offsetof(Cookie, secure),
  73         "explicitdomain",       offsetof(Cookie, explicitdom),
  74         "explicitpath",         offsetof(Cookie, explicitpath),
  75         "netscapestyle",        offsetof(Cookie, netscapestyle),
  76 };
  77
  78 #pragma varargck type "J"       Jar*
  79 #pragma varargck type "K"       Cookie*
  80
  81 /* HTTP format */
  82 int
  83 jarfmt(Fmt *fmt)
  84 {
  85         int i;
  86         Jar *jar;
  87
  88         jar = va_arg(fmt->args, Jar*);
  89         if(jar == nil || jar->nc == 0)
  90                 return fmtstrcpy(fmt, "");
  91
  92         fmtprint(fmt, "Cookie: ");
  93         if(jar->c[0].version)
  94                 fmtprint(fmt, "$Version=%s; ", jar->c[0].version);
  95         for(i=0; i<jar->nc; i++)
  96                 fmtprint(fmt, "%s%s=%s", i ? "; ":"", jar->c[i].name, jar->c[i].value);
  97         fmtprint(fmt, "\r\n");
  98         return 0;
  99 }
 100
 101 /* individual cookie */
 102 int
 103 cookiefmt(Fmt *fmt)
 104 {
 105         int j, k, first;
 106         char *t;
 107         Cookie *c;
 108
 109         c = va_arg(fmt->args, Cookie*);
 110
 111         first = 1;
 112         for(j=0; j<nelem(stab); j++){
 113                 t = *(char**)((char*)c+stab[j].offset);
 114                 if(t == nil)
 115                         continue;
 116                 if(first)
 117                         first = 0;
 118                 else
 119                         fmtprint(fmt, " ");
 120                 fmtprint(fmt, "%s=%q", stab[j].s, t);
 121         }
 122         for(j=0; j<nelem(itab); j++){
 123                 k = *(int*)((char*)c+itab[j].offset);
 124                 if(k == 0)
 125                         continue;
 126                 if(first)
 127                         first = 0;
 128                 else
 129                         fmtprint(fmt, " ");
 130                 fmtprint(fmt, "%s=%ud", itab[j].s, k);
 131         }
 132         return 0;
 133 }
 134
 135 /*
 136  * sort cookies:
 137  *      - alpha by name
 138  *      - alpha by domain
 139  *      - longer paths first, then alpha by path (RFC2109 4.3.4)
 140  */
 141 int
 142 cookiecmp(Cookie *a, Cookie *b)
 143 {
 144         int i;
 145
 146         if((i = strcmp(a->name, b->name)) != 0)
 147                 return i;
 148         if((i = cistrcmp(a->dom, b->dom)) != 0)
 149                 return i;
 150         if((i = strlen(b->path) - strlen(a->path)) != 0)
 151                 return i;
 152         if((i = strcmp(a->path, b->path)) != 0)
 153                 return i;
 154         return 0;
 155 }
 156
 157 int
 158 exactcookiecmp(Cookie *a, Cookie *b)
 159 {
 160         int i;
 161
 162         if((i = cookiecmp(a, b)) != 0)
 163                 return i;
 164         if((i = strcmp(a->value, b->value)) != 0)
 165                 return i;
 166         if(a->version || b->version){
 167                 if(!a->version)
 168                         return -1;
 169                 if(!b->version)
 170                         return 1;
 171                 if((i = strcmp(a->version, b->version)) != 0)
 172                         return i;
 173         }
 174         if(a->comment || b->comment){
 175                 if(!a->comment)
 176                         return -1;
 177                 if(!b->comment)
 178                         return 1;
 179                 if((i = strcmp(a->comment, b->comment)) != 0)
 180                         return i;
 181         }
 182         if((i = b->expire - a->expire) != 0)
 183                 return i;
 184         if((i = b->secure - a->secure) != 0)
 185                 return i;
 186         if((i = b->explicitdom - a->explicitdom) != 0)
 187                 return i;
 188         if((i = b->explicitpath - a->explicitpath) != 0)
 189                 return i;
 190         if((i = b->netscapestyle - a->netscapestyle) != 0)
 191                 return i;
 192
 193         return 0;
 194 }
 195
 196 void
 197 freecookie(Cookie *c)
 198 {
 199         int i;
 200
 201         for(i=0; i<nelem(stab); i++)
 202                 free(*(char**)((char*)c+stab[i].offset));
 203 }
 204
 205 void
 206 copycookie(Cookie *c)
 207 {
 208         int i;
 209         char **ps;
 210
 211         for(i=0; i<nelem(stab); i++){
 212                 ps = (char**)((char*)c+stab[i].offset);
 213                 if(*ps)
 214                         *ps = estrdup9p(*ps);
 215         }
 216 }
 217
 218 void
 219 delcookie(Jar *j, Cookie *c)
 220 {
 221         int i;
 222
 223         j->dirty = 1;
 224         i = c - j->c;
 225         if(i < 0 || i >= j->nc)
 226                 abort();
 227         c->deleted = 1;
 228 }
 229
 230 void
 231 addcookie(Jar *j, Cookie *c)
 232 {
 233         int i;
 234
 235         if(!c->name || !c->value || !c->path || !c->dom){
 236                 fprint(2, "not adding incomplete cookie\n");
 237                 return;
 238         }
 239
 240         if(debug)
 241                 fprint(2, "add %K\n", c);
 242
 243         for(i=0; i<j->nc; i++)
 244                 if(cookiecmp(&j->c[i], c) == 0){
 245                         if(debug)
 246                                 fprint(2, "cookie %K matches %K\n", &j->c[i], c);
 247                         if(exactcookiecmp(&j->c[i], c) == 0){
 248                                 if(debug)
 249                                         fprint(2, "\texactly\n");
 250                                 j->c[i].mark = 0;
 251                                 return;
 252                         }
 253                         delcookie(j, &j->c[i]);
 254                 }
 255
 256         j->dirty = 1;
 257         if(j->nc == j->mc){
 258                 j->mc += 16;
 259                 j->c = erealloc9p(j->c, j->mc*sizeof(Cookie));
 260         }
 261         j->c[j->nc] = *c;
 262         copycookie(&j->c[j->nc]);
 263         j->nc++;
 264 }
 265
 266 void
 267 purgejar(Jar *j)
 268 {
 269         int i;
 270
 271         for(i=j->nc-1; i>=0; i--){
 272                 if(!j->c[i].deleted)
 273                         continue;
 274                 freecookie(&j->c[i]);
 275                 --j->nc;
 276                 j->c[i] = j->c[j->nc];
 277         }
 278 }
 279
 280 void
 281 addtojar(Jar *jar, char *line, int ondisk)
 282 {
 283         Cookie c;
 284         int i, j, nf, *pint;
 285         char *f[20], *attr, *val, **pstr;
 286
 287         memset(&c, 0, sizeof c);
 288         c.expire = ~0;
 289         c.ondisk = ondisk;
 290         nf = tokenize(line, f, nelem(f));
 291         for(i=0; i<nf; i++){
 292                 attr = f[i];
 293                 if((val = strchr(attr, '=')) != nil)
 294                         *val++ = '\0';
 295                 else
 296                         val = "";
 297                 /* string attributes */
 298                 for(j=0; j<nelem(stab); j++){
 299                         if(strcmp(stab[j].s, attr) == 0){
 300                                 pstr = (char**)((char*)&c+stab[j].offset);
 301                                 *pstr = val;
 302                         }
 303                 }
 304                 /* integer attributes */
 305                 for(j=0; j<nelem(itab); j++){
 306                         if(strcmp(itab[j].s, attr) == 0){
 307                                 pint = (int*)((char*)&c+itab[j].offset);
 308                                 if(val[0]=='\0')
 309                                         *pint = 1;
 310                                 else
 311                                         *pint = strtoul(val, 0, 0);
 312                         }
 313                 }
 314         }
 315         if(c.name==nil || c.value==nil || c.dom==nil || c.path==nil){
 316                 if(debug)
 317                         fprint(2, "ignoring fractional cookie %K\n", &c);
 318                 return;
 319         }
 320         addcookie(jar, &c);
 321 }
 322
 323 Jar*
 324 newjar(void)
 325 {
 326         Jar *jar;
 327
 328         jar = emalloc9p(sizeof(Jar));
 329         return jar;
 330 }
 331
 332 int
 333 expirejar(Jar *jar, int exiting)
 334 {
 335         int i, n;
 336         uint now;
 337
 338         now = time(0);
 339         n = 0;
 340         for(i=0; i<jar->nc; i++){
 341                 if(jar->c[i].expire < now || (exiting && jar->c[i].expire==~0)){
 342                         delcookie(jar, &jar->c[i]);
 343                         n++;
 344                 }
 345         }
 346         return n;
 347 }
 348
 349 int
 350 syncjar(Jar *jar)
 351 {
 352         int i, fd;
 353         char *line;
 354         Dir *d;
 355         Biobuf *b;
 356         Qid q;
 357
 358         if(jar->file==nil)
 359                 return 0;
 360
 361         memset(&q, 0, sizeof q);
 362         if((d = dirstat(jar->file)) != nil){
 363                 q = d->qid;
 364                 if(d->qid.path != jar->qid.path || d->qid.vers != jar->qid.vers)
 365                         jar->dirty = 1;
 366                 free(d);
 367         }
 368
 369         if(jar->dirty == 0)
 370                 return 0;
 371
 372         fd = -1;
 373         for(i=0; i<50; i++){
 374                 if((fd = create(jar->lockfile, OWRITE, DMEXCL|0666)) < 0){
 375                         sleep(100);
 376                         continue;
 377                 }
 378                 break;
 379         }
 380         if(fd < 0){
 381                 if(debug)
 382                         fprint(2, "open %s: %r", jar->lockfile);
 383                 werrstr("cannot acquire jar lock: %r");
 384                 return -1;
 385         }
 386
 387         for(i=0; i<jar->nc; i++)        /* mark is cleared by addcookie */
 388                 jar->c[i].mark = jar->c[i].ondisk;
 389
 390         if((b = Bopen(jar->file, OREAD)) == nil){
 391                 if(debug)
 392                         fprint(2, "Bopen %s: %r", jar->file);
 393                 werrstr("cannot read cookie file %s: %r", jar->file);
 394                 close(fd);
 395                 return -1;
 396         }
 397         for(; (line = Brdstr(b, '\n', 1)) != nil; free(line)){
 398                 if(*line == '#')
 399                         continue;
 400                 addtojar(jar, line, 1);
 401         }
 402         Bterm(b);
 403
 404         for(i=0; i<jar->nc; i++)
 405                 if(jar->c[i].mark)
 406                         delcookie(jar, &jar->c[i]);
 407
 408         purgejar(jar);
 409
 410         b = Bopen(jar->file, OTRUNC|OWRITE);
 411         if(b == nil){
 412                 if(debug)
 413                         fprint(2, "Bopen write %s: %r", jar->file);
 414                 close(fd);
 415                 return -1;
 416         }
 417         Bprint(b, "# webcookies cookie jar\n");
 418         Bprint(b, "# comments and non-standard fields will be lost\n");
 419         for(i=0; i<jar->nc; i++){
 420                 if(jar->c[i].expire == ~0)
 421                         continue;
 422                 Bprint(b, "%K\n", &jar->c[i]);
 423                 jar->c[i].ondisk = 1;
 424         }
 425         Bterm(b);
 426
 427         jar->dirty = 0;
 428         close(fd);
 429         if((d = dirstat(jar->file)) != nil){
 430                 jar->qid = d->qid;
 431                 free(d);
 432         }
 433         return 0;
 434 }
 435
 436 Jar*
 437 readjar(char *file)
 438 {
 439         char *lock, *p;
 440         Jar *jar;
 441
 442         jar = newjar();
 443         lock = emalloc9p(strlen(file)+10);
 444         strcpy(lock, file);
 445         if((p = strrchr(lock, '/')) != nil)
 446                 p++;
 447         else
 448                 p = lock;
 449         memmove(p+2, p, strlen(p)+1);
 450         p[0] = 'L';
 451         p[1] = '.';
 452         jar->lockfile = lock;
 453         jar->file = file;
 454         jar->dirty = 1;
 455
 456         if(syncjar(jar) < 0){
 457                 free(jar->file);
 458                 free(jar->lockfile);
 459                 free(jar);
 460                 return nil;
 461         }
 462         return jar;
 463 }
 464
 465 void
 466 closejar(Jar *jar)
 467 {
 468         int i;
 469
 470         if(jar == nil)
 471                 return;
 472         expirejar(jar, 0);
 473         if(syncjar(jar) < 0)
 474                 fprint(2, "warning: cannot rewrite cookie jar: %r\n");
 475
 476         for(i=0; i<jar->nc; i++)
 477                 freecookie(&jar->c[i]);
 478
 479         free(jar->file);
 480         free(jar->c);
 481         free(jar);
 482 }
 483
 484 /*
 485  * Domain name matching is per RFC2109, section 2:
 486  *
 487  * Hosts names can be specified either as an IP address or a FQHN
 488  * string.  Sometimes we compare one host name with another.  Host A's
 489  * name domain-matches host B's if
 490  *
 491  * * both host names are IP addresses and their host name strings match
 492  *   exactly; or
 493  *
 494  * * both host names are FQDN strings and their host name strings match
 495  *   exactly; or
 496  *
 497  * * A is a FQDN string and has the form NB, where N is a non-empty name
 498  *   string, B has the form .B', and B' is a FQDN string.  (So, x.y.com
 499  *   domain-matches .y.com but not y.com.)
 500  *
 501  * Note that domain-match is not a commutative operation: a.b.c.com
 502  * domain-matches .c.com, but not the reverse.
 503  *
 504  * (This does not verify that IP addresses and FQDN's are well-formed.)
 505  */
 506 int
 507 isdomainmatch(char *name, char *pattern)
 508 {
 509         int lname, lpattern;
 510
 511         if(cistrcmp(name, pattern)==0)
 512                 return 1;
 513
 514         if(strcmp(ipattr(name), "dom")==0 && pattern[0]=='.'){
 515                 lname = strlen(name);
 516                 lpattern = strlen(pattern);
 517                 if(lname >= lpattern && cistrcmp(name+lname-lpattern, pattern)==0)
 518                         return 1;
 519         }
 520
 521         return 0;
 522 }
 523
 524 /*
 525  * RFC2109 4.3.4:
 526  *      - domain must match
 527  *      - path in cookie must be a prefix of request path
 528  *      - cookie must not have expired
 529  */
 530 int
 531 iscookiematch(Cookie *c, char *dom, char *path, uint now)
 532 {
 533         return isdomainmatch(dom, c->dom)
 534                 && strncmp(c->path, path, strlen(c->path))==0
 535                 && c->expire >= now;
 536 }
 537
 538 /*
 539  * Produce a subjar of matching cookies.
 540  * Secure cookies are only included if secure is set.
 541  */
 542 Jar*
 543 cookiesearch(Jar *jar, char *dom, char *path, int issecure)
 544 {
 545         int i;
 546         Jar *j;
 547         uint now;
 548
 549         now = time(0);
 550         j = newjar();
 551         for(i=0; i<jar->nc; i++)
 552                 if((issecure || !jar->c[i].secure) && iscookiematch(&jar->c[i], dom, path, now))
 553                         addcookie(j, &jar->c[i]);
 554         if(j->nc == 0){
 555                 closejar(j);
 556                 werrstr("no cookies found");
 557                 return nil;
 558         }
 559         qsort(j->c, j->nc, sizeof(j->c[0]), (int(*)(const void*, const void*))cookiecmp);
 560         return j;
 561 }
 562
 563 /*
 564  * RFC2109 4.3.2 security checks
 565  */
 566 char*
 567 isbadcookie(Cookie *c, char *dom, char *path)
 568 {
 569         if(strncmp(c->path, path, strlen(c->path)) != 0)
 570                 return "cookie path is not a prefix of the request path";
 571
 572         if(c->explicitdom && c->dom[0] != '.')
 573                 return "cookie domain doesn't start with dot";
 574
 575         if(memchr(c->dom+1, '.', strlen(c->dom)-1-1) == nil)
 576                 return "cookie domain doesn't have embedded dots";
 577
 578         if(!isdomainmatch(dom, c->dom))
 579                 return "request host does not match cookie domain";
 580
 581         if(strcmp(ipattr(dom), "dom")==0
 582         && memchr(dom, '.', strlen(dom)-strlen(c->dom)) != nil)
 583                 return "request host contains dots before cookie domain";
 584
 585         return 0;
 586 }
 587
 588 /*
 589  * Sunday, 25-Jan-2002 12:24:36 GMT
 590  * Sunday, 25 Jan 2002 12:24:36 GMT
 591  * Sun, 25 Jan 02 12:24:36 GMT
 592  */
 593 int
 594 isleap(int year)
 595 {
 596         return year%4==0 && (year%100!=0 || year%400==0);
 597 }
 598
 599 uint
 600 strtotime(char *s)
 601 {
 602         char *os;
 603         int i;
 604         Tm tm;
 605
 606         static int mday[2][12] = {
 607                 31, 28, 31, 30, 31, 30, 31, 31, 30, 31, 30, 31,
 608                 31, 29, 31, 30, 31, 30, 31, 31, 30, 31, 30, 31,
 609         };
 610         static char *wday[] = {
 611                 "Sunday", "Monday", "Tuesday", "Wednesday",
 612                 "Thursday", "Friday", "Saturday",
 613         };
 614         static char *mon[] = {
 615                 "Jan", "Feb", "Mar", "Apr", "May", "Jun",
 616                 "Jul", "Aug", "Sep", "Oct", "Nov", "Dec",
 617         };
 618
 619         os = s;
 620         /* Sunday, */
 621         for(i=0; i<nelem(wday); i++){
 622                 if(cistrncmp(s, wday[i], strlen(wday[i])) == 0){
 623                         s += strlen(wday[i]);
 624                         break;
 625                 }
 626                 if(cistrncmp(s, wday[i], 3) == 0){
 627                         s += 3;
 628                         break;
 629                 }
 630         }
 631         if(i==nelem(wday)){
 632                 if(debug)
 633                         fprint(2, "bad wday (%s)\n", os);
 634                 return -1;
 635         }
 636         if(*s++ != ',' || *s++ != ' '){
 637                 if(debug)
 638                         fprint(2, "bad wday separator (%s)\n", os);
 639                 return -1;
 640         }
 641
 642         /* 25- */
 643         if(!isdigit(s[0]) || !isdigit(s[1]) || (s[2]!='-' && s[2]!=' ')){
 644                 if(debug)
 645                         fprint(2, "bad day of month (%s)\n", os);
 646                 return -1;
 647         }
 648         tm.mday = strtol(s, 0, 10);
 649         s += 3;
 650
 651         /* Jan- */
 652         for(i=0; i<nelem(mon); i++)
 653                 if(cistrncmp(s, mon[i], 3) == 0){
 654                         tm.mon = i;
 655                         s += 3;
 656                         break;
 657                 }
 658         if(i==nelem(mon)){
 659                 if(debug)
 660                         fprint(2, "bad month (%s)\n", os);
 661                 return -1;
 662         }
 663         if(s[0] != '-' && s[0] != ' '){
 664                 if(debug)
 665                         fprint(2, "bad month separator (%s)\n", os);
 666                 return -1;
 667         }
 668         s++;
 669
 670         /* 2002 */
 671         if(!isdigit(s[0]) || !isdigit(s[1])){
 672                 if(debug)
 673                         fprint(2, "bad year (%s)\n", os);
 674                 return -1;
 675         }
 676         tm.year = strtol(s, 0, 10);
 677         s += 2;
 678         if(isdigit(s[0]) && isdigit(s[1]))
 679                 s += 2;
 680         else{
 681                 if(tm.year <= 68)
 682                         tm.year += 2000;
 683                 else
 684                         tm.year += 1900;
 685         }
 686         if(tm.mday==0 || tm.mday > mday[isleap(tm.year)][tm.mon]){
 687                 if(debug)
 688                         fprint(2, "invalid day of month (%s)\n", os);
 689                 return -1;
 690         }
 691         tm.year -= 1900;
 692         if(*s++ != ' '){
 693                 if(debug)
 694                         fprint(2, "bad year separator (%s)\n", os);
 695                 return -1;
 696         }
 697
 698         if(!isdigit(s[0]) || !isdigit(s[1]) || s[2]!=':'
 699         || !isdigit(s[3]) || !isdigit(s[4]) || s[5]!=':'
 700         || !isdigit(s[6]) || !isdigit(s[7]) || s[8]!=' '){
 701                 if(debug)
 702                         fprint(2, "bad time (%s)\n", os);
 703                 return -1;
 704         }
 705
 706         tm.hour = atoi(s);
 707         tm.min = atoi(s+3);
 708         tm.sec = atoi(s+6);
 709         if(tm.hour >= 24 || tm.min >= 60 || tm.sec >= 60){
 710                 if(debug)
 711                         fprint(2, "invalid time (%s)\n", os);
 712                 return -1;
 713         }
 714         s += 9;
 715
 716         if(cistrcmp(s, "GMT") != 0){
 717                 if(debug)
 718                         fprint(2, "time zone not GMT (%s)\n", os);
 719                 return -1;
 720         }
 721         strcpy(tm.zone, "GMT");
 722         tm.yday = 0;
 723         return tm2sec(&tm);
 724 }
 725
 726 /*
 727  * skip linear whitespace.  we're a bit more lenient than RFC2616 2.2.
 728  */
 729 char*
 730 skipspace(char *s)
 731 {
 732         while(*s=='\r' || *s=='\n' || *s==' ' || *s=='\t')
 733                 s++;
 734         return s;
 735 }
 736
 737 /*
 738  * Try to identify old netscape headers.
 739  * The old headers:
 740  *      - didn't allow spaces around the '='
 741  *      - used an 'Expires' attribute
 742  *      - had no 'Version' attribute
 743  *      - had no quotes
 744  *      - allowed whitespace in values
 745  *      - apparently separated attr/value pairs with ';' exclusively
 746  */
 747 int
 748 isnetscape(char *hdr)
 749 {
 750         char *s;
 751
 752         for(s=hdr; (s=strchr(s, '=')) != nil; s++){
 753                 if(isspace(s[1]) || (s > hdr && isspace(s[-1])))
 754                         return 0;
 755                 if(s[1]=='"')
 756                         return 0;
 757         }
 758         if(cistrstr(hdr, "version="))
 759                 return 0;
 760         return 1;
 761 }
 762
 763 /*
 764  * Parse HTTP response headers, adding cookies to jar.
 765  * Overwrites the headers.  May overwrite path.
 766  */
 767 char* parsecookie(Cookie*, char*, char**, int, char*, char*);
 768 int
 769 parsehttp(Jar *jar, char *hdr, char *dom, char *path)
 770 {
 771         static char setcookie[] = "Set-Cookie:";
 772         char *e, *p, *nextp;
 773         Cookie c;
 774         int isns, n;
 775
 776         isns = isnetscape(hdr);
 777         n = 0;
 778         for(p=hdr; p; p=nextp){
 779                 p = skipspace(p);
 780                 if(*p == '\0')
 781                         break;
 782                 nextp = strchr(p, '\n');
 783                 if(nextp != nil)
 784                         *nextp++ = '\0';
 785                 if(debug)
 786                         fprint(2, "?%s\n", p);
 787                 if(cistrncmp(p, setcookie, strlen(setcookie)) != 0)
 788                         continue;
 789                 if(debug)
 790                         fprint(2, "%s\n", p);
 791                 p = skipspace(p+strlen(setcookie));
 792                 for(; *p; p=skipspace(p)){
 793                         if((e = parsecookie(&c, p, &p, isns, dom, path)) != nil){
 794                                 if(debug)
 795                                         fprint(2, "parse cookie: %s\n", e);
 796                                 break;
 797                         }
 798                         if((e = isbadcookie(&c, dom, path)) != nil){
 799                                 if(debug)
 800                                         fprint(2, "reject cookie; %s\n", e);
 801                                 continue;
 802                         }
 803                         addcookie(jar, &c);
 804                         n++;
 805                 }
 806         }
 807         return n;
 808 }
 809
 810 static char*
 811 skipquoted(char *s)
 812 {
 813         /*
 814          * Sec 2.2 of RFC2616 defines a "quoted-string" as:
 815          *
 816          *  quoted-string  = ( <"> *(qdtext | quoted-pair ) <"> )
 817          *  qdtext         = <any TEXT except <">>
 818          *  quoted-pair    = "\" CHAR
 819          *
 820          * TEXT is any octet except CTLs, but including LWS;
 821          * LWS is [CR LF] 1*(SP | HT);
 822          * CHARs are ASCII octets 0-127;  (NOTE: we reject 0's)
 823          * CTLs are octets 0-31 and 127;
 824          */
 825         if(*s != '"')
 826                 return s;
 827
 828         for(s++; 32 <= *s && *s < 127 && *s != '"'; s++)
 829                 if(*s == '\\' && *(s+1) != '\0')
 830                         s++;
 831         return s;
 832 }
 833
 834 static char*
 835 skiptoken(char *s)
 836 {
 837         /*
 838          * Sec 2.2 of RFC2616 defines a "token" as
 839          *  1*<any CHAR except CTLs or separators>;
 840          * CHARs are ASCII octets 0-127;
 841          * CTLs are octets 0-31 and 127;
 842          * separators are "()<>@,;:\/[]?={}", double-quote, SP (32), and HT (9)
 843          */
 844         while(32 <= *s && *s < 127 && strchr("()<>@,;:[]?={}\" \t\\", *s)==nil)
 845                 s++;
 846
 847         return s;
 848 }
 849
 850 static char*
 851 skipvalue(char *s, int isns)
 852 {
 853         char *t;
 854
 855         /*
 856          * An RFC2109 value is an HTTP token or an HTTP quoted string.
 857          * Netscape servers ignore the spec and rely on semicolons, apparently.
 858          */
 859         if(isns){
 860                 if((t = strchr(s, ';')) == nil)
 861                         t = s+strlen(s);
 862                 return t;
 863         }
 864         if(*s == '"')
 865                 return skipquoted(s);
 866         return skiptoken(s);
 867 }
 868
 869 /*
 870  * RMID=80b186bb64c03c65fab767f8; expires=Monday, 10-Feb-2003 04:44:39 GMT;
 871  *      path=/; domain=.nytimes.com
 872  */
 873 char*
 874 parsecookie(Cookie *c, char *p, char **e, int isns, char *dom, char *path)
 875 {
 876         int i, done;
 877         char *t, *u, *attr, *val;
 878
 879         memset(c, 0, sizeof *c);
 880         c->expire = ~0;
 881
 882         /* NAME=VALUE */
 883         t = skiptoken(p);
 884         c->name = p;
 885         p = skipspace(t);
 886         if(*p != '='){
 887         Badname:
 888                 return "malformed cookie: no NAME=VALUE";
 889         }
 890         *t = '\0';
 891         p = skipspace(p+1);
 892         t = skipvalue(p, isns);
 893         if(*t)
 894                 *t++ = '\0';
 895         c->value = p;
 896         p = skipspace(t);
 897         if(c->name[0]=='\0' || c->value[0]=='\0')
 898                 goto Badname;
 899
 900         done = 0;
 901         for(; *p && !done; p=skipspace(p)){
 902                 attr = p;
 903                 t = skiptoken(p);
 904                 u = skipspace(t);
 905                 switch(*u){
 906                 case '\0':
 907                         *t = '\0';
 908                         p = val = u;
 909                         break;
 910                 case ';':
 911                         *t = '\0';
 912                         val = "";
 913                         p = u+1;
 914                         break;
 915                 case '=':
 916                         *t = '\0';
 917                         val = skipspace(u+1);
 918                         p = skipvalue(val, isns);
 919                         if(*p==',')
 920                                 done = 1;
 921                         if(*p)
 922                                 *p++ = '\0';
 923                         break;
 924                 case ',':
 925                         if(!isns){
 926                                 val = "";
 927                                 p = u;
 928                                 *p++ = '\0';
 929                                 done = 1;
 930                                 break;
 931                         }
 932                 default:
 933                         if(debug)
 934                                 fprint(2, "syntax: %s\n", p);
 935                         return "syntax error";
 936                 }
 937                 for(i=0; i<nelem(stab); i++)
 938                         if(stab[i].ishttp && cistrcmp(stab[i].s, attr)==0)
 939                                 *(char**)((char*)c+stab[i].offset) = val;
 940                 if(cistrcmp(attr, "expires") == 0){
 941                         if(!isns)
 942                                 return "non-netscape cookie has Expires tag";
 943                         if(!val[0])
 944                                 return "bad expires tag";
 945                         c->expire = strtotime(val);
 946                         if(c->expire == ~0)
 947                                 return "cannot parse netscape expires tag";
 948                 }
 949                 if(cistrcmp(attr, "max-age") == 0)
 950                         c->expire = time(0)+atoi(val);
 951                 if(cistrcmp(attr, "secure") == 0)
 952                         c->secure = 1;
 953         }
 954         *e = p;
 955
 956         if(c->dom){
 957                 /* add leading dot for explicit domain */
 958                 if(c->dom[0] != '.' && strcmp(ipattr(c->dom), "dom") == 0){
 959                         static char *ddom = nil;
 960
 961                         ddom = realloc(ddom, strlen(c->dom)+2);
 962                         if(ddom != nil){
 963                                 ddom[0] = '.';
 964                                 strcpy(ddom+1, c->dom);
 965                                 c->dom = ddom;
 966                         }
 967                 }
 968                 c->explicitdom = 1;
 969         }else
 970                 c->dom = dom;
 971         if(c->path)
 972                 c->explicitpath = 1;
 973         else{
 974                 c->path = path;
 975                 if((t = strchr(c->path, '#')) != 0)
 976                         *t = '\0';
 977                 if((t = strchr(c->path, '?')) != 0)
 978                         *t = '\0';
 979                 if((t = strrchr(c->path, '/')) != 0)
 980                         *t = '\0';
 981         }
 982         c->netscapestyle = isns;
 983
 984         return nil;
 985 }
 986
 987 Jar *jar;
 988
 989 enum
 990 {
 991         Xhttp = 1,
 992         Xcookies,
 993
 994         NeedUrl = 0,
 995         HaveUrl,
 996 };
 997
 998 typedef struct Aux Aux;
 999 struct Aux
1000 {
1001         int state;
1002         char *dom;
1003         char *path;
1004         char *inhttp;
1005         char *outhttp;
1006         char *ctext;
1007         int rdoff;
1008 };
1009 enum
1010 {
1011         AuxBuf = 4096,
1012         MaxCtext = 16*1024*1024,
1013 };
1014
1015 void
1016 fsopen(Req *r)
1017 {
1018         char *s, *es;
1019         int i, sz;
1020         Aux *a;
1021
1022         switch((uintptr)r->fid->file->aux){
1023         case Xhttp:
1024                 syncjar(jar);
1025                 a = emalloc9p(sizeof(Aux));
1026                 r->fid->aux = a;
1027                 a->inhttp = emalloc9p(AuxBuf);
1028                 a->outhttp = emalloc9p(AuxBuf);
1029                 break;
1030
1031         case Xcookies:
1032                 syncjar(jar);
1033                 a = emalloc9p(sizeof(Aux));
1034                 r->fid->aux = a;
1035                 if(r->ifcall.mode&OTRUNC){
1036                         a->ctext = emalloc9p(1);
1037                         a->ctext[0] = '\0';
1038                 }else{
1039                         sz = 256*jar->nc+1024;  /* BUG should do better */
1040                         a->ctext = emalloc9p(sz);
1041                         a->ctext[0] = '\0';
1042                         s = a->ctext;
1043                         es = s+sz;
1044                         for(i=0; i<jar->nc; i++)
1045                                 s = seprint(s, es, "%K\n", &jar->c[i]);
1046                 }
1047                 break;
1048         }
1049         respond(r, nil);
1050 }
1051
1052 void
1053 fsread(Req *r)
1054 {
1055         Aux *a;
1056
1057         a = r->fid->aux;
1058         switch((uintptr)r->fid->file->aux){
1059         case Xhttp:
1060                 if(a->state == NeedUrl){
1061                         respond(r, "must write url before read");
1062                         return;
1063                 }
1064                 r->ifcall.offset = a->rdoff;
1065                 readstr(r, a->outhttp);
1066                 a->rdoff += r->ofcall.count;
1067                 respond(r, nil);
1068                 return;
1069
1070         case Xcookies:
1071                 readstr(r, a->ctext);
1072                 respond(r, nil);
1073                 return;
1074
1075         default:
1076                 respond(r, "bug in webcookies");
1077                 return;
1078         }
1079 }
1080
1081 void
1082 fswrite(Req *r)
1083 {
1084         Aux *a;
1085         int i, sz, hlen, issecure;
1086         char buf[1024], *p;
1087         Jar *j;
1088
1089         a = r->fid->aux;
1090         switch((uintptr)r->fid->file->aux){
1091         case Xhttp:
1092                 if(a->state == NeedUrl){
1093                         if(r->ifcall.count >= sizeof buf){
1094                                 respond(r, "url too long");
1095                                 return;
1096                         }
1097                         memmove(buf, r->ifcall.data, r->ifcall.count);
1098                         buf[r->ifcall.count] = '\0';
1099                         issecure = 0;
1100                         if(cistrncmp(buf, "http://", 7) == 0)
1101                                 hlen = 7;
1102                         else if(cistrncmp(buf, "https://", 8) == 0){
1103                                 hlen = 8;
1104                                 issecure = 1;
1105                         }else{
1106                                 respond(r, "url must begin http:// or https://");
1107                                 return;
1108                         }
1109                         if(buf[hlen]=='/'){
1110                                 respond(r, "url without host name");
1111                                 return;
1112                         }
1113                         p = strchr(buf+hlen, '/');
1114                         if(p == nil)
1115                                 a->path = estrdup9p("/");
1116                         else{
1117                                 a->path = estrdup9p(p);
1118                                 *p = '\0';
1119                         }
1120                         a->dom = estrdup9p(buf+hlen);
1121                         a->state = HaveUrl;
1122                         j = cookiesearch(jar, a->dom, a->path, issecure);
1123                         if(debug){
1124                                 fprint(2, "search %s %s got %p\n", a->dom, a->path, j);
1125                                 if(j){
1126                                         fprint(2, "%d cookies\n", j->nc);
1127                                         for(i=0; i<j->nc; i++)
1128                                                 fprint(2, "%K\n", &j->c[i]);
1129                                 }
1130                         }
1131                         snprint(a->outhttp, AuxBuf, "%J", j);
1132                         if(j)
1133                                 closejar(j);
1134                 }else{
1135                         if(strlen(a->inhttp)+r->ifcall.count >= AuxBuf){
1136                                 respond(r, "http headers too large");
1137                                 return;
1138                         }
1139                         memmove(a->inhttp+strlen(a->inhttp), r->ifcall.data, r->ifcall.count);
1140                 }
1141                 r->ofcall.count = r->ifcall.count;
1142                 respond(r, nil);
1143                 return;
1144
1145         case Xcookies:
1146                 sz = r->ifcall.count+r->ifcall.offset;
1147                 if(sz > strlen(a->ctext)){
1148                         if(sz >= MaxCtext){
1149                                 respond(r, "cookie file too large");
1150                                 return;
1151                         }
1152                         a->ctext = erealloc9p(a->ctext, sz+1);
1153                         a->ctext[sz] = '\0';
1154                 }
1155                 memmove(a->ctext+r->ifcall.offset, r->ifcall.data, r->ifcall.count);
1156                 r->ofcall.count = r->ifcall.count;
1157                 respond(r, nil);
1158                 return;
1159
1160         default:
1161                 respond(r, "bug in webcookies");
1162                 return;
1163         }
1164 }
1165
1166 void
1167 fsdestroyfid(Fid *fid)
1168 {
1169         char *p, *nextp;
1170         Aux *a;
1171         int i;
1172
1173         a = fid->aux;
1174         if(a == nil)
1175                 return;
1176         switch((uintptr)fid->file->aux){
1177         case Xhttp:
1178                 parsehttp(jar, a->inhttp, a->dom, a->path);
1179                 break;
1180         case Xcookies:
1181                 for(i=0; i<jar->nc; i++)
1182                         jar->c[i].mark = 1;
1183                 for(p=a->ctext; *p; p=nextp){
1184                         if((nextp = strchr(p, '\n')) != nil)
1185                                 *nextp++ = '\0';
1186                         else
1187                                 nextp = "";
1188                         addtojar(jar, p, 0);
1189                 }
1190                 for(i=0; i<jar->nc; i++)
1191                         if(jar->c[i].mark)
1192                                 delcookie(jar, &jar->c[i]);
1193                 break;
1194         }
1195         syncjar(jar);
1196         free(a->dom);
1197         free(a->path);
1198         free(a->inhttp);
1199         free(a->outhttp);
1200         free(a->ctext);
1201         free(a);
1202 }
1203
1204 void
1205 fsend(Srv*)
1206 {
1207         closejar(jar);
1208 }
1209
1210 Srv fs =
1211 {
1212 .open=          fsopen,
1213 .read=          fsread,
1214 .write=         fswrite,
1215 .destroyfid=    fsdestroyfid,
1216 .end=           fsend,
1217 };
1218
1219 void
1220 usage(void)
1221 {
1222         fprint(2, "usage: webcookies [-f file] [-m mtpt] [-s service]\n");
1223         exits("usage");
1224 }
1225
1226 void
1227 main(int argc, char **argv)
1228 {
1229         char *file, *mtpt, *home, *srv;
1230         int fd;
1231
1232         file = nil;
1233         srv = nil;
1234         mtpt = "/mnt/webcookies";
1235         ARGBEGIN{
1236         case 'D':
1237                 chatty9p++;
1238                 break;
1239         case 'd':
1240                 debug = 1;
1241                 break;
1242         case 'f':
1243                 file = EARGF(usage());
1244                 break;
1245         case 's':
1246                 srv = EARGF(usage());
1247                 break;
1248         case 'm':
1249                 mtpt = EARGF(usage());
1250                 break;
1251         default:
1252                 usage();
1253         }ARGEND
1254
1255         if(argc != 0)
1256                 usage();
1257
1258         quotefmtinstall();
1259         fmtinstall('J', jarfmt);
1260         fmtinstall('K', cookiefmt);
1261
1262         if(file == nil){
1263                 home = getenv("home");
1264                 if(home == nil)
1265                         sysfatal("no cookie file specified and no $home");
1266                 file = emalloc9p(strlen(home)+30);
1267                 strcpy(file, home);
1268                 strcat(file, "/lib/webcookies");
1269         }
1270         if(access(file, AEXIST) < 0){
1271                 if((fd = create(file, OWRITE, 0600)) < 0)
1272                         sysfatal("create %s: %r", file);
1273                 close(fd);
1274         }
1275
1276         jar = readjar(file);
1277         if(jar == nil)
1278                 sysfatal("readjar: %r");
1279
1280         fs.tree = alloctree("cookie", "cookie", DMDIR|0555, nil);
1281         closefile(createfile(fs.tree->root, "http", "cookie", 0666, (void*)Xhttp));
1282         closefile(createfile(fs.tree->root, "cookies", "cookie", 0666, (void*)Xcookies));
1283
1284         postmountsrv(&fs, srv, mtpt, MREPL);
1285         exits(nil);
1286 }