hget.c 25 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186118711881189119011911192119311941195119611971198119912001201120212031204120512061207120812091210121112121213121412151216121712181219122012211222122312241225122612271228122912301231123212331234123512361237123812391240124112421243124412451246124712481249125012511252125312541255125612571258125912601261126212631264126512661267126812691270127112721273127412751276127712781279128012811282128312841285128612871288128912901291129212931294129512961297129812991300130113021303130413051306130713081309131013111312131313141315131613171318131913201321132213231324132513261327132813291330133113321333133413351336133713381339134013411342134313441345134613471348134913501351135213531354135513561357135813591360136113621363136413651366136713681369137013711372137313741375137613771378137913801381138213831384138513861387138813891390139113921393139413951396139713981399140014011402140314041405140614071408140914101411141214131414141514161417141814191420142114221423142414251426142714281429143014311432143314341435143614371438143914401441144214431444144514461447144814491450145114521453145414551456145714581459146014611462146314641465146614671468146914701471147214731474147514761477147814791480
  1. #include <u.h>
  2. #include <libc.h>
  3. #include <ctype.h>
  4. #include <bio.h>
  5. #include <ip.h>
  6. #include <libsec.h>
  7. #include <auth.h>
  8. typedef struct URL URL;
  9. struct URL
  10. {
  11. int method;
  12. char *host;
  13. char *port;
  14. char *page;
  15. char *etag;
  16. char *redirect;
  17. char *postbody;
  18. char *cred;
  19. char *rhead;
  20. long mtime;
  21. };
  22. typedef struct Range Range;
  23. struct Range
  24. {
  25. long start; /* only 2 gig supported, tdb */
  26. long end;
  27. };
  28. typedef struct Out Out;
  29. struct Out
  30. {
  31. int fd;
  32. int offset; /* notional current offset in output */
  33. int written; /* number of bytes successfully transferred to output */
  34. DigestState *curr; /* digest state up to offset (if known) */
  35. DigestState *hiwat; /* digest state of all bytes written */
  36. };
  37. enum
  38. {
  39. Other,
  40. Http,
  41. Https,
  42. Ftp,
  43. };
  44. enum
  45. {
  46. Eof = 0,
  47. Error = -1,
  48. Server = -2,
  49. Changed = -3,
  50. };
  51. int debug;
  52. char *ofile;
  53. int doftp(URL*, URL*, Range*, Out*, long);
  54. int dohttp(URL*, URL*, Range*, Out*, long);
  55. int crackurl(URL*, char*);
  56. Range* crackrange(char*);
  57. int getheader(int, char*, int);
  58. int httpheaders(int, int, URL*, Range*);
  59. int httprcode(int);
  60. int cistrncmp(char*, char*, int);
  61. int cistrcmp(char*, char*);
  62. void initibuf(void);
  63. int readline(int, char*, int);
  64. int readibuf(int, char*, int);
  65. int dfprint(int, char*, ...);
  66. void unreadline(char*);
  67. int output(Out*, char*, int);
  68. void setoffset(Out*, int);
  69. int verbose;
  70. char *net;
  71. char tcpdir[NETPATHLEN];
  72. int headerprint;
  73. struct {
  74. char *name;
  75. int (*f)(URL*, URL*, Range*, Out*, long);
  76. } method[] = {
  77. [Http] { "http", dohttp },
  78. [Https] { "https", dohttp },
  79. [Ftp] { "ftp", doftp },
  80. [Other] { "_______", nil },
  81. };
  82. void
  83. usage(void)
  84. {
  85. fprint(2, "usage: %s [-dhv] [-o outfile] [-p body] [-x netmtpt] [-r header] url\n", argv0);
  86. exits("usage");
  87. }
  88. void
  89. main(int argc, char **argv)
  90. {
  91. URL u;
  92. Range r;
  93. int errs, n;
  94. ulong mtime;
  95. Dir *d;
  96. char postbody[4096], *p, *e, *t, *hpx;
  97. URL px; // Proxy
  98. Out out;
  99. ofile = nil;
  100. p = postbody;
  101. e = p + sizeof(postbody);
  102. r.start = 0;
  103. r.end = -1;
  104. mtime = 0;
  105. memset(&u, 0, sizeof(u));
  106. memset(&px, 0, sizeof(px));
  107. hpx = getenv("httpproxy");
  108. ARGBEGIN {
  109. case 'o':
  110. ofile = EARGF(usage());
  111. break;
  112. case 'd':
  113. debug = 1;
  114. break;
  115. case 'h':
  116. headerprint = 1;
  117. break;
  118. case 'v':
  119. verbose = 1;
  120. break;
  121. case 'x':
  122. net = EARGF(usage());
  123. break;
  124. case 'r':
  125. u.rhead = EARGF(usage());
  126. break;
  127. case 'p':
  128. t = EARGF(usage());
  129. if(p != postbody)
  130. p = seprint(p, e, "&%s", t);
  131. else
  132. p = seprint(p, e, "%s", t);
  133. u.postbody = postbody;
  134. break;
  135. default:
  136. usage();
  137. } ARGEND;
  138. if(net != nil){
  139. if(strlen(net) > sizeof(tcpdir)-5)
  140. sysfatal("network mount point too long");
  141. snprint(tcpdir, sizeof(tcpdir), "%s/tcp", net);
  142. } else
  143. snprint(tcpdir, sizeof(tcpdir), "tcp");
  144. if(argc != 1)
  145. usage();
  146. out.fd = 1;
  147. out.written = 0;
  148. out.offset = 0;
  149. out.curr = nil;
  150. out.hiwat = nil;
  151. if(ofile != nil){
  152. d = dirstat(ofile);
  153. if(d == nil){
  154. out.fd = create(ofile, OWRITE, 0664);
  155. if(out.fd < 0)
  156. sysfatal("creating %s: %r", ofile);
  157. } else {
  158. out.fd = open(ofile, OWRITE);
  159. if(out.fd < 0)
  160. sysfatal("can't open %s: %r", ofile);
  161. r.start = d->length;
  162. mtime = d->mtime;
  163. free(d);
  164. }
  165. }
  166. errs = 0;
  167. if(crackurl(&u, argv[0]) < 0)
  168. sysfatal("%r");
  169. if(hpx && crackurl(&px, hpx) < 0)
  170. sysfatal("%r");
  171. for(;;){
  172. setoffset(&out, 0);
  173. /* transfer data */
  174. werrstr("");
  175. n = (*method[u.method].f)(&u, &px, &r, &out, mtime);
  176. switch(n){
  177. case Eof:
  178. exits(0);
  179. break;
  180. case Error:
  181. if(errs++ < 10)
  182. continue;
  183. sysfatal("too many errors with no progress %r");
  184. break;
  185. case Server:
  186. sysfatal("server returned: %r");
  187. break;
  188. }
  189. /* forward progress */
  190. errs = 0;
  191. r.start += n;
  192. if(r.start >= r.end)
  193. break;
  194. }
  195. exits(0);
  196. }
  197. int
  198. crackurl(URL *u, char *s)
  199. {
  200. char *p;
  201. int i;
  202. if(u->page != nil){
  203. free(u->page);
  204. u->page = nil;
  205. }
  206. /* get type */
  207. for(p = s; *p; p++){
  208. if(*p == '/'){
  209. p = s;
  210. if(u->method == Other){
  211. werrstr("missing method");
  212. return -1;
  213. }
  214. if(u->host == nil){
  215. werrstr("missing host");
  216. return -1;
  217. }
  218. u->page = strdup(p);
  219. return 0;
  220. }
  221. if(*p == ':' && *(p+1)=='/' && *(p+2)=='/'){
  222. *p = 0;
  223. p += 3;
  224. for(i = 0; i < nelem(method); i++){
  225. if(cistrcmp(s, method[i].name) == 0){
  226. u->method = i;
  227. break;
  228. }
  229. }
  230. break;
  231. }
  232. }
  233. if(u->method == Other){
  234. werrstr("unsupported URL type %s", s);
  235. return -1;
  236. }
  237. /* get system */
  238. free(u->host);
  239. s = p;
  240. p = strchr(s, '/');
  241. if(p == nil){
  242. u->host = strdup(s);
  243. u->page = strdup("/");
  244. } else {
  245. u->page = strdup(p);
  246. *p = 0;
  247. u->host = strdup(s);
  248. *p = '/';
  249. }
  250. if(p = strchr(u->host, ':')) {
  251. *p++ = 0;
  252. u->port = p;
  253. } else
  254. u->port = method[u->method].name;
  255. if(*(u->host) == 0){
  256. werrstr("bad url, null host");
  257. return -1;
  258. }
  259. return 0;
  260. }
  261. char *day[] = {
  262. "Sun", "Mon", "Tue", "Wed", "Thu", "Fri", "Sat"
  263. };
  264. char *month[] = {
  265. "Jan", "Feb", "Mar", "Apr", "May", "Jun", "Jul", "Aug", "Sep", "Oct", "Nov", "Dec"
  266. };
  267. struct
  268. {
  269. int fd;
  270. long mtime;
  271. } note;
  272. void
  273. catch(void*, char*)
  274. {
  275. Dir d;
  276. nulldir(&d);
  277. d.mtime = note.mtime;
  278. if(dirfwstat(note.fd, &d) < 0)
  279. sysfatal("catch: can't dirfwstat: %r");
  280. noted(NDFLT);
  281. }
  282. int
  283. dohttp(URL *u, URL *px, Range *r, Out *out, long mtime)
  284. {
  285. int fd, cfd;
  286. int redirect, auth, loop;
  287. int n, rv, code;
  288. long tot, vtime;
  289. Tm *tm;
  290. char buf[1024];
  291. char err[ERRMAX];
  292. /* always move back to a previous 512 byte bound because some
  293. * servers can't seem to deal with requests that start at the
  294. * end of the file
  295. */
  296. if(r->start)
  297. r->start = ((r->start-1)/512)*512;
  298. /* loop for redirects, requires reading both response code and headers */
  299. fd = -1;
  300. for(loop = 0; loop < 32; loop++){
  301. if(px->host == nil){
  302. fd = dial(netmkaddr(u->host, tcpdir, u->port), 0, 0, 0);
  303. } else {
  304. fd = dial(netmkaddr(px->host, tcpdir, px->port), 0, 0, 0);
  305. }
  306. if(fd < 0)
  307. return Error;
  308. if(u->method == Https){
  309. int tfd;
  310. TLSconn conn;
  311. memset(&conn, 0, sizeof conn);
  312. tfd = tlsClient(fd, &conn);
  313. if(tfd < 0){
  314. fprint(2, "tlsClient: %r\n");
  315. close(fd);
  316. return Error;
  317. }
  318. /* BUG: check cert here? */
  319. if(conn.cert)
  320. free(conn.cert);
  321. close(fd);
  322. fd = tfd;
  323. }
  324. /* write request, use range if not start of file */
  325. if(u->postbody == nil){
  326. if(px->host == nil){
  327. dfprint(fd, "GET %s HTTP/1.0\r\n"
  328. "Host: %s\r\n"
  329. "User-agent: Plan9/hget\r\n"
  330. "Cache-Control: no-cache\r\n"
  331. "Pragma: no-cache\r\n",
  332. u->page, u->host);
  333. } else {
  334. dfprint(fd, "GET http://%s%s HTTP/1.0\r\n"
  335. "Host: %s\r\n"
  336. "User-agent: Plan9/hget\r\n"
  337. "Cache-Control: no-cache\r\n"
  338. "Pragma: no-cache\r\n",
  339. u->host, u->page, u->host);
  340. }
  341. } else {
  342. dfprint(fd, "POST %s HTTP/1.0\r\n"
  343. "Host: %s\r\n"
  344. "Content-type: application/x-www-form-urlencoded\r\n"
  345. "Content-length: %d\r\n"
  346. "User-agent: Plan9/hget\r\n",
  347. u->page, u->host, strlen(u->postbody));
  348. }
  349. if(u->cred)
  350. dfprint(fd, "Authorization: Basic %s\r\n", u->cred);
  351. if(u->rhead)
  352. dfprint(fd, "%s\r\n", u->rhead);
  353. if(r->start != 0){
  354. dfprint(fd, "Range: bytes=%d-\n", r->start);
  355. if(u->etag != nil){
  356. dfprint(fd, "If-range: %s\n", u->etag);
  357. } else {
  358. tm = gmtime(mtime);
  359. dfprint(fd, "If-range: %s, %d %s %d %2d:%2.2d:%2.2d GMT\n",
  360. day[tm->wday], tm->mday, month[tm->mon],
  361. tm->year+1900, tm->hour, tm->min, tm->sec);
  362. }
  363. }
  364. if((cfd = open("/mnt/webcookies/http", ORDWR)) >= 0){
  365. if(fprint(cfd, "http://%s%s", u->host, u->page) > 0){
  366. while((n = read(cfd, buf, sizeof buf)) > 0){
  367. if(debug)
  368. write(2, buf, n);
  369. write(fd, buf, n);
  370. }
  371. }else{
  372. close(cfd);
  373. cfd = -1;
  374. }
  375. }
  376. dfprint(fd, "\r\n", u->host);
  377. if(u->postbody)
  378. dfprint(fd, "%s", u->postbody);
  379. auth = 0;
  380. redirect = 0;
  381. initibuf();
  382. code = httprcode(fd);
  383. switch(code){
  384. case Error: /* connection timed out */
  385. case Eof:
  386. close(fd);
  387. close(cfd);
  388. return code;
  389. case 200: /* OK */
  390. case 201: /* Created */
  391. case 202: /* Accepted */
  392. if(ofile == nil && r->start != 0)
  393. sysfatal("page changed underfoot");
  394. break;
  395. case 204: /* No Content */
  396. sysfatal("No Content");
  397. case 206: /* Partial Content */
  398. setoffset(out, r->start);
  399. break;
  400. case 301: /* Moved Permanently */
  401. case 302: /* Moved Temporarily (actually Found) */
  402. case 303: /* See Other */
  403. case 307: /* Temporary Redirect (HTTP/1.1) */
  404. redirect = 1;
  405. u->postbody = nil;
  406. break;
  407. case 304: /* Not Modified */
  408. break;
  409. case 400: /* Bad Request */
  410. sysfatal("Bad Request");
  411. case 401: /* Unauthorized */
  412. if (auth)
  413. sysfatal("Authentication failed");
  414. auth = 1;
  415. break;
  416. case 402: /* ??? */
  417. sysfatal("Unauthorized");
  418. case 403: /* Forbidden */
  419. sysfatal("Forbidden by server");
  420. case 404: /* Not Found */
  421. sysfatal("Not found on server");
  422. case 407: /* Proxy Authentication */
  423. sysfatal("Proxy authentication required");
  424. case 500: /* Internal server error */
  425. sysfatal("Server choked");
  426. case 501: /* Not implemented */
  427. sysfatal("Server can't do it!");
  428. case 502: /* Bad gateway */
  429. sysfatal("Bad gateway");
  430. case 503: /* Service unavailable */
  431. sysfatal("Service unavailable");
  432. default:
  433. sysfatal("Unknown response code %d", code);
  434. }
  435. if(u->redirect != nil){
  436. free(u->redirect);
  437. u->redirect = nil;
  438. }
  439. rv = httpheaders(fd, cfd, u, r);
  440. close(cfd);
  441. if(rv != 0){
  442. close(fd);
  443. return rv;
  444. }
  445. if(!redirect && !auth)
  446. break;
  447. if (redirect){
  448. if(u->redirect == nil)
  449. sysfatal("redirect: no URL");
  450. if(crackurl(u, u->redirect) < 0)
  451. sysfatal("redirect: %r");
  452. }
  453. }
  454. /* transfer whatever you get */
  455. if(ofile != nil && u->mtime != 0){
  456. note.fd = out->fd;
  457. note.mtime = u->mtime;
  458. notify(catch);
  459. }
  460. tot = 0;
  461. vtime = 0;
  462. for(;;){
  463. n = readibuf(fd, buf, sizeof(buf));
  464. if(n <= 0)
  465. break;
  466. if(output(out, buf, n) != n)
  467. break;
  468. tot += n;
  469. if(verbose && (vtime != time(0) || r->start == r->end)) {
  470. vtime = time(0);
  471. fprint(2, "%ld %ld\n", r->start+tot, r->end);
  472. }
  473. }
  474. notify(nil);
  475. close(fd);
  476. if(ofile != nil && u->mtime != 0){
  477. Dir d;
  478. rerrstr(err, sizeof err);
  479. nulldir(&d);
  480. d.mtime = u->mtime;
  481. if(dirfwstat(out->fd, &d) < 0)
  482. fprint(2, "couldn't set mtime: %r\n");
  483. errstr(err, sizeof err);
  484. }
  485. return tot;
  486. }
  487. /* get the http response code */
  488. int
  489. httprcode(int fd)
  490. {
  491. int n;
  492. char *p;
  493. char buf[256];
  494. n = readline(fd, buf, sizeof(buf)-1);
  495. if(n <= 0)
  496. return n;
  497. if(debug)
  498. fprint(2, "%d <- %s\n", fd, buf);
  499. p = strchr(buf, ' ');
  500. if(strncmp(buf, "HTTP/", 5) != 0 || p == nil){
  501. werrstr("bad response from server");
  502. return -1;
  503. }
  504. buf[n] = 0;
  505. return atoi(p+1);
  506. }
  507. /* read in and crack the http headers, update u and r */
  508. void hhetag(char*, URL*, Range*);
  509. void hhmtime(char*, URL*, Range*);
  510. void hhclen(char*, URL*, Range*);
  511. void hhcrange(char*, URL*, Range*);
  512. void hhuri(char*, URL*, Range*);
  513. void hhlocation(char*, URL*, Range*);
  514. void hhauth(char*, URL*, Range*);
  515. struct {
  516. char *name;
  517. void (*f)(char*, URL*, Range*);
  518. } headers[] = {
  519. { "etag:", hhetag },
  520. { "last-modified:", hhmtime },
  521. { "content-length:", hhclen },
  522. { "content-range:", hhcrange },
  523. { "uri:", hhuri },
  524. { "location:", hhlocation },
  525. { "WWW-Authenticate:", hhauth },
  526. };
  527. int
  528. httpheaders(int fd, int cfd, URL *u, Range *r)
  529. {
  530. char buf[2048];
  531. char *p;
  532. int i, n;
  533. for(;;){
  534. n = getheader(fd, buf, sizeof(buf));
  535. if(n <= 0)
  536. break;
  537. if(cfd >= 0)
  538. fprint(cfd, "%s\n", buf);
  539. for(i = 0; i < nelem(headers); i++){
  540. n = strlen(headers[i].name);
  541. if(cistrncmp(buf, headers[i].name, n) == 0){
  542. /* skip field name and leading white */
  543. p = buf + n;
  544. while(*p == ' ' || *p == '\t')
  545. p++;
  546. (*headers[i].f)(p, u, r);
  547. break;
  548. }
  549. }
  550. }
  551. return n;
  552. }
  553. /*
  554. * read a single mime header, collect continuations.
  555. *
  556. * this routine assumes that there is a blank line twixt
  557. * the header and the message body, otherwise bytes will
  558. * be lost.
  559. */
  560. int
  561. getheader(int fd, char *buf, int n)
  562. {
  563. char *p, *e;
  564. int i;
  565. n--;
  566. p = buf;
  567. for(e = p + n; ; p += i){
  568. i = readline(fd, p, e-p);
  569. if(i < 0)
  570. return i;
  571. if(p == buf){
  572. /* first line */
  573. if(strchr(buf, ':') == nil)
  574. break; /* end of headers */
  575. } else {
  576. /* continuation line */
  577. if(*p != ' ' && *p != '\t'){
  578. unreadline(p);
  579. *p = 0;
  580. break; /* end of this header */
  581. }
  582. }
  583. }
  584. if(headerprint)
  585. print("%s\n", buf);
  586. if(debug)
  587. fprint(2, "%d <- %s\n", fd, buf);
  588. return p-buf;
  589. }
  590. void
  591. hhetag(char *p, URL *u, Range*)
  592. {
  593. if(u->etag != nil){
  594. if(strcmp(u->etag, p) != 0)
  595. sysfatal("file changed underfoot");
  596. } else
  597. u->etag = strdup(p);
  598. }
  599. char* monthchars = "janfebmaraprmayjunjulaugsepoctnovdec";
  600. void
  601. hhmtime(char *p, URL *u, Range*)
  602. {
  603. char *month, *day, *yr, *hms;
  604. char *fields[6];
  605. Tm tm, now;
  606. int i;
  607. i = getfields(p, fields, 6, 1, " \t");
  608. if(i < 5)
  609. return;
  610. day = fields[1];
  611. month = fields[2];
  612. yr = fields[3];
  613. hms = fields[4];
  614. /* default time */
  615. now = *gmtime(time(0));
  616. tm = now;
  617. tm.yday = 0;
  618. /* convert ascii month to a number twixt 1 and 12 */
  619. if(*month >= '0' && *month <= '9'){
  620. tm.mon = atoi(month) - 1;
  621. if(tm.mon < 0 || tm.mon > 11)
  622. tm.mon = 5;
  623. } else {
  624. for(p = month; *p; p++)
  625. *p = tolower(*p);
  626. for(i = 0; i < 12; i++)
  627. if(strncmp(&monthchars[i*3], month, 3) == 0){
  628. tm.mon = i;
  629. break;
  630. }
  631. }
  632. tm.mday = atoi(day);
  633. if(hms) {
  634. tm.hour = strtoul(hms, &p, 10);
  635. if(*p == ':') {
  636. p++;
  637. tm.min = strtoul(p, &p, 10);
  638. if(*p == ':') {
  639. p++;
  640. tm.sec = strtoul(p, &p, 10);
  641. }
  642. }
  643. if(tolower(*p) == 'p')
  644. tm.hour += 12;
  645. }
  646. if(yr) {
  647. tm.year = atoi(yr);
  648. if(tm.year >= 1900)
  649. tm.year -= 1900;
  650. } else {
  651. if(tm.mon > now.mon || (tm.mon == now.mon && tm.mday > now.mday+1))
  652. tm.year--;
  653. }
  654. strcpy(tm.zone, "GMT");
  655. /* convert to epoch seconds */
  656. u->mtime = tm2sec(&tm);
  657. }
  658. void
  659. hhclen(char *p, URL*, Range *r)
  660. {
  661. r->end = atoi(p);
  662. }
  663. void
  664. hhcrange(char *p, URL*, Range *r)
  665. {
  666. char *x;
  667. vlong l;
  668. l = 0;
  669. x = strchr(p, '/');
  670. if(x)
  671. l = atoll(x+1);
  672. if(l == 0) {
  673. x = strchr(p, '-');
  674. if(x)
  675. l = atoll(x+1);
  676. }
  677. if(l)
  678. r->end = l;
  679. }
  680. void
  681. hhuri(char *p, URL *u, Range*)
  682. {
  683. if(*p != '<')
  684. return;
  685. u->redirect = strdup(p+1);
  686. p = strchr(u->redirect, '>');
  687. if(p != nil)
  688. *p = 0;
  689. }
  690. void
  691. hhlocation(char *p, URL *u, Range*)
  692. {
  693. u->redirect = strdup(p);
  694. }
  695. void
  696. hhauth(char *p, URL *u, Range*)
  697. {
  698. char *f[4];
  699. UserPasswd *up;
  700. char *s, cred[64];
  701. if (cistrncmp(p, "basic ", 6) != 0)
  702. sysfatal("only Basic authentication supported");
  703. if (gettokens(p, f, nelem(f), "\"") < 2)
  704. sysfatal("garbled auth data");
  705. if ((up = auth_getuserpasswd(auth_getkey, "proto=pass service=http server=%q realm=%q",
  706. u->host, f[1])) == nil)
  707. sysfatal("cannot authenticate");
  708. s = smprint("%s:%s", up->user, up->passwd);
  709. if(enc64(cred, sizeof(cred), (uchar *)s, strlen(s)) == -1)
  710. sysfatal("enc64");
  711. free(s);
  712. assert(u->cred = strdup(cred));
  713. }
  714. enum
  715. {
  716. /* ftp return codes */
  717. Extra= 1,
  718. Success= 2,
  719. Incomplete= 3,
  720. TempFail= 4,
  721. PermFail= 5,
  722. Nnetdir= 64, /* max length of network directory paths */
  723. Ndialstr= 64, /* max length of dial strings */
  724. };
  725. int ftpcmd(int, char*, ...);
  726. int ftprcode(int, char*, int);
  727. int hello(int);
  728. int logon(int);
  729. int xfertype(int, char*);
  730. int passive(int, URL*);
  731. int active(int, URL*);
  732. int ftpxfer(int, Out*, Range*);
  733. int terminateftp(int, int);
  734. int getaddrport(char*, uchar*, uchar*);
  735. int ftprestart(int, Out*, URL*, Range*, long);
  736. int
  737. doftp(URL *u, URL *px, Range *r, Out *out, long mtime)
  738. {
  739. int pid, ctl, data, rv;
  740. Waitmsg *w;
  741. char msg[64];
  742. char conndir[NETPATHLEN];
  743. char *p;
  744. /* untested, proxy doesn't work with ftp (I think) */
  745. if(px->host == nil){
  746. ctl = dial(netmkaddr(u->host, tcpdir, u->port), 0, conndir, 0);
  747. } else {
  748. ctl = dial(netmkaddr(px->host, tcpdir, px->port), 0, conndir, 0);
  749. }
  750. if(ctl < 0)
  751. return Error;
  752. if(net == nil){
  753. p = strrchr(conndir, '/');
  754. *p = 0;
  755. snprint(tcpdir, sizeof(tcpdir), conndir);
  756. }
  757. initibuf();
  758. rv = hello(ctl);
  759. if(rv < 0)
  760. return terminateftp(ctl, rv);
  761. rv = logon(ctl);
  762. if(rv < 0)
  763. return terminateftp(ctl, rv);
  764. rv = xfertype(ctl, "I");
  765. if(rv < 0)
  766. return terminateftp(ctl, rv);
  767. /* if file is up to date and the right size, stop */
  768. if(ftprestart(ctl, out, u, r, mtime) > 0){
  769. close(ctl);
  770. return Eof;
  771. }
  772. /* first try passive mode, then active */
  773. data = passive(ctl, u);
  774. if(data < 0){
  775. data = active(ctl, u);
  776. if(data < 0)
  777. return Error;
  778. }
  779. /* fork */
  780. switch(pid = rfork(RFPROC|RFFDG|RFMEM)){
  781. case -1:
  782. close(data);
  783. return terminateftp(ctl, Error);
  784. case 0:
  785. ftpxfer(data, out, r);
  786. close(data);
  787. _exits(0);
  788. default:
  789. close(data);
  790. break;
  791. }
  792. /* wait for reply message */
  793. rv = ftprcode(ctl, msg, sizeof(msg));
  794. close(ctl);
  795. /* wait for process to terminate */
  796. w = nil;
  797. for(;;){
  798. free(w);
  799. w = wait();
  800. if(w == nil)
  801. return Error;
  802. if(w->pid == pid){
  803. if(w->msg[0] == 0){
  804. free(w);
  805. break;
  806. }
  807. werrstr("xfer: %s", w->msg);
  808. free(w);
  809. return Error;
  810. }
  811. }
  812. switch(rv){
  813. case Success:
  814. return Eof;
  815. case TempFail:
  816. return Server;
  817. default:
  818. return Error;
  819. }
  820. }
  821. int
  822. ftpcmd(int ctl, char *fmt, ...)
  823. {
  824. va_list arg;
  825. char buf[2*1024], *s;
  826. va_start(arg, fmt);
  827. s = vseprint(buf, buf + (sizeof(buf)-4) / sizeof(*buf), fmt, arg);
  828. va_end(arg);
  829. if(debug)
  830. fprint(2, "%d -> %s\n", ctl, buf);
  831. *s++ = '\r';
  832. *s++ = '\n';
  833. if(write(ctl, buf, s - buf) != s - buf)
  834. return -1;
  835. return 0;
  836. }
  837. int
  838. ftprcode(int ctl, char *msg, int len)
  839. {
  840. int rv;
  841. int i;
  842. char *p;
  843. len--; /* room for terminating null */
  844. for(;;){
  845. *msg = 0;
  846. i = readline(ctl, msg, len);
  847. if(i < 0)
  848. break;
  849. if(debug)
  850. fprint(2, "%d <- %s\n", ctl, msg);
  851. /* stop if not a continuation */
  852. rv = strtol(msg, &p, 10);
  853. if(rv >= 100 && rv < 600 && p==msg+3 && *p == ' ')
  854. return rv/100;
  855. }
  856. *msg = 0;
  857. return -1;
  858. }
  859. int
  860. hello(int ctl)
  861. {
  862. char msg[1024];
  863. /* wait for hello from other side */
  864. if(ftprcode(ctl, msg, sizeof(msg)) != Success){
  865. werrstr("HELLO: %s", msg);
  866. return Server;
  867. }
  868. return 0;
  869. }
  870. int
  871. getdec(char *p, int n)
  872. {
  873. int x = 0;
  874. int i;
  875. for(i = 0; i < n; i++)
  876. x = x*10 + (*p++ - '0');
  877. return x;
  878. }
  879. int
  880. ftprestart(int ctl, Out *out, URL *u, Range *r, long mtime)
  881. {
  882. Tm tm;
  883. char msg[1024];
  884. long x, rmtime;
  885. ftpcmd(ctl, "MDTM %s", u->page);
  886. if(ftprcode(ctl, msg, sizeof(msg)) != Success){
  887. r->start = 0;
  888. return 0; /* need to do something */
  889. }
  890. /* decode modification time */
  891. if(strlen(msg) < 4 + 4 + 2 + 2 + 2 + 2 + 2){
  892. r->start = 0;
  893. return 0; /* need to do something */
  894. }
  895. memset(&tm, 0, sizeof(tm));
  896. tm.year = getdec(msg+4, 4) - 1900;
  897. tm.mon = getdec(msg+4+4, 2) - 1;
  898. tm.mday = getdec(msg+4+4+2, 2);
  899. tm.hour = getdec(msg+4+4+2+2, 2);
  900. tm.min = getdec(msg+4+4+2+2+2, 2);
  901. tm.sec = getdec(msg+4+4+2+2+2+2, 2);
  902. strcpy(tm.zone, "GMT");
  903. rmtime = tm2sec(&tm);
  904. if(rmtime > mtime)
  905. r->start = 0;
  906. /* get size */
  907. ftpcmd(ctl, "SIZE %s", u->page);
  908. if(ftprcode(ctl, msg, sizeof(msg)) == Success){
  909. x = atol(msg+4);
  910. if(r->start == x)
  911. return 1; /* we're up to date */
  912. r->end = x;
  913. }
  914. /* seek to restart point */
  915. if(r->start > 0){
  916. ftpcmd(ctl, "REST %lud", r->start);
  917. if(ftprcode(ctl, msg, sizeof(msg)) == Incomplete){
  918. setoffset(out, r->start);
  919. }else
  920. r->start = 0;
  921. }
  922. return 0; /* need to do something */
  923. }
  924. int
  925. logon(int ctl)
  926. {
  927. char msg[1024];
  928. /* login anonymous */
  929. ftpcmd(ctl, "USER anonymous");
  930. switch(ftprcode(ctl, msg, sizeof(msg))){
  931. case Success:
  932. return 0;
  933. case Incomplete:
  934. break; /* need password */
  935. default:
  936. werrstr("USER: %s", msg);
  937. return Server;
  938. }
  939. /* send user id as password */
  940. sprint(msg, "%s@closedmind.org", getuser());
  941. ftpcmd(ctl, "PASS %s", msg);
  942. if(ftprcode(ctl, msg, sizeof(msg)) != Success){
  943. werrstr("PASS: %s", msg);
  944. return Server;
  945. }
  946. return 0;
  947. }
  948. int
  949. xfertype(int ctl, char *t)
  950. {
  951. char msg[1024];
  952. ftpcmd(ctl, "TYPE %s", t);
  953. if(ftprcode(ctl, msg, sizeof(msg)) != Success){
  954. werrstr("TYPE %s: %s", t, msg);
  955. return Server;
  956. }
  957. return 0;
  958. }
  959. int
  960. passive(int ctl, URL *u)
  961. {
  962. char msg[1024];
  963. char ipaddr[32];
  964. char *f[6];
  965. char *p;
  966. int fd;
  967. int port;
  968. char aport[12];
  969. ftpcmd(ctl, "PASV");
  970. if(ftprcode(ctl, msg, sizeof(msg)) != Success)
  971. return Error;
  972. /* get address and port number from reply, this is AI */
  973. p = strchr(msg, '(');
  974. if(p == nil){
  975. for(p = msg+3; *p; p++)
  976. if(isdigit(*p))
  977. break;
  978. } else
  979. p++;
  980. if(getfields(p, f, 6, 0, ",)") < 6){
  981. werrstr("ftp protocol botch");
  982. return Server;
  983. }
  984. snprint(ipaddr, sizeof(ipaddr), "%s.%s.%s.%s",
  985. f[0], f[1], f[2], f[3]);
  986. port = ((atoi(f[4])&0xff)<<8) + (atoi(f[5])&0xff);
  987. sprint(aport, "%d", port);
  988. /* open data connection */
  989. fd = dial(netmkaddr(ipaddr, tcpdir, aport), 0, 0, 0);
  990. if(fd < 0){
  991. werrstr("passive mode failed: %r");
  992. return Error;
  993. }
  994. /* tell remote to send a file */
  995. ftpcmd(ctl, "RETR %s", u->page);
  996. if(ftprcode(ctl, msg, sizeof(msg)) != Extra){
  997. werrstr("RETR %s: %s", u->page, msg);
  998. return Error;
  999. }
  1000. return fd;
  1001. }
  1002. int
  1003. active(int ctl, URL *u)
  1004. {
  1005. char msg[1024];
  1006. char dir[40], ldir[40];
  1007. uchar ipaddr[4];
  1008. uchar port[2];
  1009. int lcfd, dfd, afd;
  1010. /* announce a port for the call back */
  1011. snprint(msg, sizeof(msg), "%s!*!0", tcpdir);
  1012. afd = announce(msg, dir);
  1013. if(afd < 0)
  1014. return Error;
  1015. /* get a local address/port of the annoucement */
  1016. if(getaddrport(dir, ipaddr, port) < 0){
  1017. close(afd);
  1018. return Error;
  1019. }
  1020. /* tell remote side address and port*/
  1021. ftpcmd(ctl, "PORT %d,%d,%d,%d,%d,%d", ipaddr[0], ipaddr[1], ipaddr[2],
  1022. ipaddr[3], port[0], port[1]);
  1023. if(ftprcode(ctl, msg, sizeof(msg)) != Success){
  1024. close(afd);
  1025. werrstr("active: %s", msg);
  1026. return Error;
  1027. }
  1028. /* tell remote to send a file */
  1029. ftpcmd(ctl, "RETR %s", u->page);
  1030. if(ftprcode(ctl, msg, sizeof(msg)) != Extra){
  1031. close(afd);
  1032. werrstr("RETR: %s", msg);
  1033. return Server;
  1034. }
  1035. /* wait for a connection */
  1036. lcfd = listen(dir, ldir);
  1037. if(lcfd < 0){
  1038. close(afd);
  1039. return Error;
  1040. }
  1041. dfd = accept(lcfd, ldir);
  1042. if(dfd < 0){
  1043. close(afd);
  1044. close(lcfd);
  1045. return Error;
  1046. }
  1047. close(afd);
  1048. close(lcfd);
  1049. return dfd;
  1050. }
  1051. int
  1052. ftpxfer(int in, Out *out, Range *r)
  1053. {
  1054. char buf[1024];
  1055. long vtime;
  1056. int i, n;
  1057. vtime = 0;
  1058. for(n = 0;;n += i){
  1059. i = read(in, buf, sizeof(buf));
  1060. if(i == 0)
  1061. break;
  1062. if(i < 0)
  1063. return Error;
  1064. if(output(out, buf, i) != i)
  1065. return Error;
  1066. r->start += i;
  1067. if(verbose && (vtime != time(0) || r->start == r->end)) {
  1068. vtime = time(0);
  1069. fprint(2, "%ld %ld\n", r->start, r->end);
  1070. }
  1071. }
  1072. return n;
  1073. }
  1074. int
  1075. terminateftp(int ctl, int rv)
  1076. {
  1077. close(ctl);
  1078. return rv;
  1079. }
  1080. /*
  1081. * case insensitive strcmp (why aren't these in libc?)
  1082. */
  1083. int
  1084. cistrncmp(char *a, char *b, int n)
  1085. {
  1086. while(n-- > 0){
  1087. if(tolower(*a++) != tolower(*b++))
  1088. return -1;
  1089. }
  1090. return 0;
  1091. }
  1092. int
  1093. cistrcmp(char *a, char *b)
  1094. {
  1095. while(*a || *b)
  1096. if(tolower(*a++) != tolower(*b++))
  1097. return -1;
  1098. return 0;
  1099. }
  1100. /*
  1101. * buffered io
  1102. */
  1103. struct
  1104. {
  1105. char *rp;
  1106. char *wp;
  1107. char buf[4*1024];
  1108. } b;
  1109. void
  1110. initibuf(void)
  1111. {
  1112. b.rp = b.wp = b.buf;
  1113. }
  1114. /*
  1115. * read a possibly buffered line, strip off trailing while
  1116. */
  1117. int
  1118. readline(int fd, char *buf, int len)
  1119. {
  1120. int n;
  1121. char *p;
  1122. int eof = 0;
  1123. len--;
  1124. for(p = buf;;){
  1125. if(b.rp >= b.wp){
  1126. n = read(fd, b.wp, sizeof(b.buf)/2);
  1127. if(n < 0)
  1128. return -1;
  1129. if(n == 0){
  1130. eof = 1;
  1131. break;
  1132. }
  1133. b.wp += n;
  1134. }
  1135. n = *b.rp++;
  1136. if(len > 0){
  1137. *p++ = n;
  1138. len--;
  1139. }
  1140. if(n == '\n')
  1141. break;
  1142. }
  1143. /* drop trailing white */
  1144. for(;;){
  1145. if(p <= buf)
  1146. break;
  1147. n = *(p-1);
  1148. if(n != ' ' && n != '\t' && n != '\r' && n != '\n')
  1149. break;
  1150. p--;
  1151. }
  1152. *p = 0;
  1153. if(eof && p == buf)
  1154. return -1;
  1155. return p-buf;
  1156. }
  1157. void
  1158. unreadline(char *line)
  1159. {
  1160. int i, n;
  1161. i = strlen(line);
  1162. n = b.wp-b.rp;
  1163. memmove(&b.buf[i+1], b.rp, n);
  1164. memmove(b.buf, line, i);
  1165. b.buf[i] = '\n';
  1166. b.rp = b.buf;
  1167. b.wp = b.rp + i + 1 + n;
  1168. }
  1169. int
  1170. readibuf(int fd, char *buf, int len)
  1171. {
  1172. int n;
  1173. n = b.wp-b.rp;
  1174. if(n > 0){
  1175. if(n > len)
  1176. n = len;
  1177. memmove(buf, b.rp, n);
  1178. b.rp += n;
  1179. return n;
  1180. }
  1181. return read(fd, buf, len);
  1182. }
  1183. int
  1184. dfprint(int fd, char *fmt, ...)
  1185. {
  1186. char buf[4*1024];
  1187. va_list arg;
  1188. va_start(arg, fmt);
  1189. vseprint(buf, buf+sizeof(buf), fmt, arg);
  1190. va_end(arg);
  1191. if(debug)
  1192. fprint(2, "%d -> %s", fd, buf);
  1193. return fprint(fd, "%s", buf);
  1194. }
  1195. int
  1196. getaddrport(char *dir, uchar *ipaddr, uchar *port)
  1197. {
  1198. char buf[256];
  1199. int fd, i;
  1200. char *p;
  1201. snprint(buf, sizeof(buf), "%s/local", dir);
  1202. fd = open(buf, OREAD);
  1203. if(fd < 0)
  1204. return -1;
  1205. i = read(fd, buf, sizeof(buf)-1);
  1206. close(fd);
  1207. if(i <= 0)
  1208. return -1;
  1209. buf[i] = 0;
  1210. p = strchr(buf, '!');
  1211. if(p != nil)
  1212. *p++ = 0;
  1213. v4parseip(ipaddr, buf);
  1214. i = atoi(p);
  1215. port[0] = i>>8;
  1216. port[1] = i;
  1217. return 0;
  1218. }
  1219. void
  1220. md5free(DigestState *state)
  1221. {
  1222. uchar x[MD5dlen];
  1223. md5(nil, 0, x, state);
  1224. }
  1225. DigestState*
  1226. md5dup(DigestState *state)
  1227. {
  1228. char *p;
  1229. p = md5pickle(state);
  1230. if(p == nil)
  1231. sysfatal("md5pickle: %r");
  1232. state = md5unpickle(p);
  1233. if(state == nil)
  1234. sysfatal("md5unpickle: %r");
  1235. free(p);
  1236. return state;
  1237. }
  1238. void
  1239. setoffset(Out *out, int offset)
  1240. {
  1241. md5free(out->curr);
  1242. if(offset == 0)
  1243. out->curr = md5(nil, 0, nil, nil);
  1244. else
  1245. out->curr = nil;
  1246. out->offset = offset;
  1247. out->written = offset;
  1248. if(ofile != nil)
  1249. if(seek(out->fd, offset, 0) != offset)
  1250. sysfatal("seek: %r");
  1251. }
  1252. /*
  1253. * write some output, discarding it (but keeping track)
  1254. * if we've already written it. if we've gone backwards,
  1255. * verify that everything previously written matches
  1256. * that which would have been written from the current
  1257. * output.
  1258. */
  1259. int
  1260. output(Out *out, char *buf, int nb)
  1261. {
  1262. int n, d;
  1263. uchar m0[MD5dlen], m1[MD5dlen];
  1264. n = nb;
  1265. d = out->written - out->offset;
  1266. assert(d >= 0);
  1267. if(d > 0){
  1268. if(n < d){
  1269. if(out->curr != nil)
  1270. md5((uchar*)buf, n, nil, out->curr);
  1271. out->offset += n;
  1272. return n;
  1273. }
  1274. if(out->curr != nil){
  1275. md5((uchar*)buf, d, m0, out->curr);
  1276. out->curr = nil;
  1277. md5(nil, 0, m1, md5dup(out->hiwat));
  1278. if(memcmp(m0, m1, MD5dlen) != 0){
  1279. fprint(2, "integrity check failure at offset %d\n", out->written);
  1280. return -1;
  1281. }
  1282. }
  1283. buf += d;
  1284. n -= d;
  1285. out->offset += d;
  1286. }
  1287. if(n > 0){
  1288. out->hiwat = md5((uchar*)buf, n, nil, out->hiwat);
  1289. n = write(out->fd, buf, n);
  1290. if(n > 0){
  1291. out->offset += n;
  1292. out->written += n;
  1293. }
  1294. }
  1295. return n + d;
  1296. }