hget.c 25 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186118711881189119011911192119311941195119611971198119912001201120212031204120512061207120812091210121112121213121412151216121712181219122012211222122312241225122612271228122912301231123212331234123512361237123812391240124112421243124412451246124712481249125012511252125312541255125612571258125912601261126212631264126512661267126812691270127112721273127412751276127712781279128012811282128312841285128612871288128912901291129212931294129512961297129812991300130113021303130413051306130713081309131013111312131313141315131613171318131913201321132213231324132513261327132813291330133113321333133413351336133713381339134013411342134313441345134613471348134913501351135213531354135513561357135813591360136113621363136413651366136713681369137013711372137313741375137613771378137913801381138213831384138513861387138813891390139113921393139413951396139713981399140014011402140314041405140614071408140914101411141214131414141514161417141814191420142114221423142414251426142714281429143014311432143314341435143614371438143914401441144214431444144514461447144814491450145114521453145414551456145714581459146014611462146314641465146614671468146914701471147214731474147514761477
  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. long mtime;
  20. };
  21. typedef struct Range Range;
  22. struct Range
  23. {
  24. long start; /* only 2 gig supported, tdb */
  25. long end;
  26. };
  27. typedef struct Out Out;
  28. struct Out
  29. {
  30. int fd;
  31. int offset; /* notional current offset in output */
  32. int written; /* number of bytes successfully transferred to output */
  33. DigestState *curr; /* digest state up to offset (if known) */
  34. DigestState *hiwat; /* digest state of all bytes written */
  35. };
  36. enum
  37. {
  38. Other,
  39. Http,
  40. Https,
  41. Ftp,
  42. };
  43. enum
  44. {
  45. Eof = 0,
  46. Error = -1,
  47. Server = -2,
  48. Changed = -3,
  49. };
  50. int debug;
  51. char *ofile;
  52. int doftp(URL*, URL*, Range*, Out*, long);
  53. int dohttp(URL*, URL*, Range*, Out*, long);
  54. int crackurl(URL*, char*);
  55. Range* crackrange(char*);
  56. int getheader(int, char*, int);
  57. int httpheaders(int, int, URL*, Range*);
  58. int httprcode(int);
  59. int cistrncmp(char*, char*, int);
  60. int cistrcmp(char*, char*);
  61. void initibuf(void);
  62. int readline(int, char*, int);
  63. int readibuf(int, char*, int);
  64. int dfprint(int, char*, ...);
  65. void unreadline(char*);
  66. int output(Out*, char*, int);
  67. void setoffset(Out*, int);
  68. int verbose;
  69. char *net;
  70. char tcpdir[NETPATHLEN];
  71. int headerprint;
  72. struct {
  73. char *name;
  74. int (*f)(URL*, URL*, Range*, Out*, long);
  75. } method[] = {
  76. [Http] { "http", dohttp },
  77. [Https] { "https", dohttp },
  78. [Ftp] { "ftp", doftp },
  79. [Other] { "_______", nil },
  80. };
  81. void
  82. usage(void)
  83. {
  84. fprint(2, "usage: %s [-dhv] [-o outfile] [-p body] [-x netmtpt] url\n", argv0);
  85. exits("usage");
  86. }
  87. void
  88. main(int argc, char **argv)
  89. {
  90. URL u;
  91. Range r;
  92. int errs, n;
  93. ulong mtime;
  94. Dir *d;
  95. char postbody[4096], *p, *e, *t, *hpx;
  96. URL px; // Proxy
  97. Out out;
  98. ofile = nil;
  99. p = postbody;
  100. e = p + sizeof(postbody);
  101. r.start = 0;
  102. r.end = -1;
  103. mtime = 0;
  104. memset(&u, 0, sizeof(u));
  105. memset(&px, 0, sizeof(px));
  106. hpx = getenv("httpproxy");
  107. ARGBEGIN {
  108. case 'o':
  109. ofile = EARGF(usage());
  110. break;
  111. case 'd':
  112. debug = 1;
  113. break;
  114. case 'h':
  115. headerprint = 1;
  116. break;
  117. case 'v':
  118. verbose = 1;
  119. break;
  120. case 'x':
  121. net = EARGF(usage());
  122. break;
  123. case 'p':
  124. t = EARGF(usage());
  125. if(p != postbody)
  126. p = seprint(p, e, "&%s", t);
  127. else
  128. p = seprint(p, e, "%s", t);
  129. u.postbody = postbody;
  130. break;
  131. default:
  132. usage();
  133. } ARGEND;
  134. if(net != nil){
  135. if(strlen(net) > sizeof(tcpdir)-5)
  136. sysfatal("network mount point too long");
  137. snprint(tcpdir, sizeof(tcpdir), "%s/tcp", net);
  138. } else
  139. snprint(tcpdir, sizeof(tcpdir), "tcp");
  140. if(argc != 1)
  141. usage();
  142. out.fd = 1;
  143. out.written = 0;
  144. out.offset = 0;
  145. out.curr = nil;
  146. out.hiwat = nil;
  147. if(ofile != nil){
  148. d = dirstat(ofile);
  149. if(d == nil){
  150. out.fd = create(ofile, OWRITE, 0664);
  151. if(out.fd < 0)
  152. sysfatal("creating %s: %r", ofile);
  153. } else {
  154. out.fd = open(ofile, OWRITE);
  155. if(out.fd < 0)
  156. sysfatal("can't open %s: %r", ofile);
  157. r.start = d->length;
  158. mtime = d->mtime;
  159. free(d);
  160. }
  161. }
  162. errs = 0;
  163. if(crackurl(&u, argv[0]) < 0)
  164. sysfatal("%r");
  165. if(hpx && crackurl(&px, hpx) < 0)
  166. sysfatal("%r");
  167. for(;;){
  168. setoffset(&out, 0);
  169. /* transfer data */
  170. werrstr("");
  171. n = (*method[u.method].f)(&u, &px, &r, &out, mtime);
  172. switch(n){
  173. case Eof:
  174. exits(0);
  175. break;
  176. case Error:
  177. if(errs++ < 10)
  178. continue;
  179. sysfatal("too many errors with no progress %r");
  180. break;
  181. case Server:
  182. sysfatal("server returned: %r");
  183. break;
  184. }
  185. /* forward progress */
  186. errs = 0;
  187. r.start += n;
  188. if(r.start >= r.end)
  189. break;
  190. }
  191. exits(0);
  192. }
  193. int
  194. crackurl(URL *u, char *s)
  195. {
  196. char *p;
  197. int i;
  198. if(u->page != nil){
  199. free(u->page);
  200. u->page = nil;
  201. }
  202. /* get type */
  203. for(p = s; *p; p++){
  204. if(*p == '/'){
  205. p = s;
  206. if(u->method == Other){
  207. werrstr("missing method");
  208. return -1;
  209. }
  210. if(u->host == nil){
  211. werrstr("missing host");
  212. return -1;
  213. }
  214. u->page = strdup(p);
  215. return 0;
  216. }
  217. if(*p == ':' && *(p+1)=='/' && *(p+2)=='/'){
  218. *p = 0;
  219. p += 3;
  220. for(i = 0; i < nelem(method); i++){
  221. if(cistrcmp(s, method[i].name) == 0){
  222. u->method = i;
  223. break;
  224. }
  225. }
  226. break;
  227. }
  228. }
  229. if(u->method == Other){
  230. werrstr("unsupported URL type %s", s);
  231. return -1;
  232. }
  233. /* get system */
  234. free(u->host);
  235. s = p;
  236. p = strchr(s, '/');
  237. if(p == nil){
  238. u->host = strdup(s);
  239. u->page = strdup("/");
  240. } else {
  241. u->page = strdup(p);
  242. *p = 0;
  243. u->host = strdup(s);
  244. *p = '/';
  245. }
  246. if(p = strchr(u->host, ':')) {
  247. *p++ = 0;
  248. u->port = p;
  249. } else
  250. u->port = method[u->method].name;
  251. if(*(u->host) == 0){
  252. werrstr("bad url, null host");
  253. return -1;
  254. }
  255. return 0;
  256. }
  257. char *day[] = {
  258. "Sun", "Mon", "Tue", "Wed", "Thu", "Fri", "Sat"
  259. };
  260. char *month[] = {
  261. "Jan", "Feb", "Mar", "Apr", "May", "Jun", "Jul", "Aug", "Sep", "Oct", "Nov", "Dec"
  262. };
  263. struct
  264. {
  265. int fd;
  266. long mtime;
  267. } note;
  268. void
  269. catch(void*, char*)
  270. {
  271. Dir d;
  272. nulldir(&d);
  273. d.mtime = note.mtime;
  274. if(dirfwstat(note.fd, &d) < 0)
  275. sysfatal("catch: can't dirfwstat: %r");
  276. noted(NDFLT);
  277. }
  278. int
  279. dohttp(URL *u, URL *px, Range *r, Out *out, long mtime)
  280. {
  281. int fd, cfd;
  282. int redirect, auth, loop;
  283. int n, rv, code;
  284. long tot, vtime;
  285. Tm *tm;
  286. char buf[1024];
  287. char err[ERRMAX];
  288. /* always move back to a previous 512 byte bound because some
  289. * servers can't seem to deal with requests that start at the
  290. * end of the file
  291. */
  292. if(r->start)
  293. r->start = ((r->start-1)/512)*512;
  294. /* loop for redirects, requires reading both response code and headers */
  295. fd = -1;
  296. for(loop = 0; loop < 32; loop++){
  297. if(px->host == nil){
  298. fd = dial(netmkaddr(u->host, tcpdir, u->port), 0, 0, 0);
  299. } else {
  300. fd = dial(netmkaddr(px->host, tcpdir, px->port), 0, 0, 0);
  301. }
  302. if(fd < 0)
  303. return Error;
  304. if(u->method == Https){
  305. int tfd;
  306. TLSconn conn;
  307. memset(&conn, 0, sizeof conn);
  308. tfd = tlsClient(fd, &conn);
  309. if(tfd < 0){
  310. fprint(2, "tlsClient: %r\n");
  311. close(fd);
  312. return Error;
  313. }
  314. /* BUG: check cert here? */
  315. if(conn.cert)
  316. free(conn.cert);
  317. close(fd);
  318. fd = tfd;
  319. }
  320. /* write request, use range if not start of file */
  321. if(u->postbody == nil){
  322. if(px->host == nil){
  323. dfprint(fd, "GET %s HTTP/1.0\r\n"
  324. "Host: %s\r\n"
  325. "User-agent: Plan9/hget\r\n"
  326. "Cache-Control: no-cache\r\n"
  327. "Pragma: no-cache\r\n",
  328. u->page, u->host);
  329. } else {
  330. dfprint(fd, "GET http://%s%s HTTP/1.0\r\n"
  331. "Host: %s\r\n"
  332. "User-agent: Plan9/hget\r\n"
  333. "Cache-Control: no-cache\r\n"
  334. "Pragma: no-cache\r\n",
  335. u->host, u->page, u->host);
  336. }
  337. if(u->cred)
  338. dfprint(fd, "Authorization: Basic %s\r\n",
  339. u->cred);
  340. } else {
  341. dfprint(fd, "POST %s HTTP/1.0\r\n"
  342. "Host: %s\r\n"
  343. "Content-type: application/x-www-form-urlencoded\r\n"
  344. "Content-length: %d\r\n"
  345. "User-agent: Plan9/hget\r\n",
  346. u->page, u->host, strlen(u->postbody));
  347. if(u->cred)
  348. dfprint(fd, "Authorization: Basic %s\r\n", u->cred);
  349. }
  350. if(r->start != 0){
  351. dfprint(fd, "Range: bytes=%d-\n", r->start);
  352. if(u->etag != nil){
  353. dfprint(fd, "If-range: %s\n", u->etag);
  354. } else {
  355. tm = gmtime(mtime);
  356. dfprint(fd, "If-range: %s, %d %s %d %2d:%2.2d:%2.2d GMT\n",
  357. day[tm->wday], tm->mday, month[tm->mon],
  358. tm->year+1900, tm->hour, tm->min, tm->sec);
  359. }
  360. }
  361. if((cfd = open("/mnt/webcookies/http", ORDWR)) >= 0){
  362. if(fprint(cfd, "http://%s%s", u->host, u->page) > 0){
  363. while((n = read(cfd, buf, sizeof buf)) > 0){
  364. if(debug)
  365. write(2, buf, n);
  366. write(fd, buf, n);
  367. }
  368. }else{
  369. close(cfd);
  370. cfd = -1;
  371. }
  372. }
  373. dfprint(fd, "\r\n", u->host);
  374. if(u->postbody)
  375. dfprint(fd, "%s", u->postbody);
  376. auth = 0;
  377. redirect = 0;
  378. initibuf();
  379. code = httprcode(fd);
  380. switch(code){
  381. case Error: /* connection timed out */
  382. case Eof:
  383. close(fd);
  384. close(cfd);
  385. return code;
  386. case 200: /* OK */
  387. case 201: /* Created */
  388. case 202: /* Accepted */
  389. if(ofile == nil && r->start != 0)
  390. sysfatal("page changed underfoot");
  391. break;
  392. case 204: /* No Content */
  393. sysfatal("No Content");
  394. case 206: /* Partial Content */
  395. setoffset(out, r->start);
  396. break;
  397. case 301: /* Moved Permanently */
  398. case 302: /* Moved Temporarily (actually Found) */
  399. case 303: /* See Other */
  400. case 307: /* Temporary Redirect (HTTP/1.1) */
  401. redirect = 1;
  402. u->postbody = nil;
  403. break;
  404. case 304: /* Not Modified */
  405. break;
  406. case 400: /* Bad Request */
  407. sysfatal("Bad Request");
  408. case 401: /* Unauthorized */
  409. if (auth)
  410. sysfatal("Authentication failed");
  411. auth = 1;
  412. break;
  413. case 402: /* ??? */
  414. sysfatal("Unauthorized");
  415. case 403: /* Forbidden */
  416. sysfatal("Forbidden by server");
  417. case 404: /* Not Found */
  418. sysfatal("Not found on server");
  419. case 407: /* Proxy Authentication */
  420. sysfatal("Proxy authentication required");
  421. case 500: /* Internal server error */
  422. sysfatal("Server choked");
  423. case 501: /* Not implemented */
  424. sysfatal("Server can't do it!");
  425. case 502: /* Bad gateway */
  426. sysfatal("Bad gateway");
  427. case 503: /* Service unavailable */
  428. sysfatal("Service unavailable");
  429. default:
  430. sysfatal("Unknown response code %d", code);
  431. }
  432. if(u->redirect != nil){
  433. free(u->redirect);
  434. u->redirect = nil;
  435. }
  436. rv = httpheaders(fd, cfd, u, r);
  437. close(cfd);
  438. if(rv != 0){
  439. close(fd);
  440. return rv;
  441. }
  442. if(!redirect && !auth)
  443. break;
  444. if (redirect){
  445. if(u->redirect == nil)
  446. sysfatal("redirect: no URL");
  447. if(crackurl(u, u->redirect) < 0)
  448. sysfatal("redirect: %r");
  449. }
  450. }
  451. /* transfer whatever you get */
  452. if(ofile != nil && u->mtime != 0){
  453. note.fd = out->fd;
  454. note.mtime = u->mtime;
  455. notify(catch);
  456. }
  457. tot = 0;
  458. vtime = 0;
  459. for(;;){
  460. n = readibuf(fd, buf, sizeof(buf));
  461. if(n <= 0)
  462. break;
  463. if(output(out, buf, n) != n)
  464. break;
  465. tot += n;
  466. if(verbose && (vtime != time(0) || r->start == r->end)) {
  467. vtime = time(0);
  468. fprint(2, "%ld %ld\n", r->start+tot, r->end);
  469. }
  470. }
  471. notify(nil);
  472. close(fd);
  473. if(ofile != nil && u->mtime != 0){
  474. Dir d;
  475. rerrstr(err, sizeof err);
  476. nulldir(&d);
  477. d.mtime = u->mtime;
  478. if(dirfwstat(out->fd, &d) < 0)
  479. fprint(2, "couldn't set mtime: %r\n");
  480. errstr(err, sizeof err);
  481. }
  482. return tot;
  483. }
  484. /* get the http response code */
  485. int
  486. httprcode(int fd)
  487. {
  488. int n;
  489. char *p;
  490. char buf[256];
  491. n = readline(fd, buf, sizeof(buf)-1);
  492. if(n <= 0)
  493. return n;
  494. if(debug)
  495. fprint(2, "%d <- %s\n", fd, buf);
  496. p = strchr(buf, ' ');
  497. if(strncmp(buf, "HTTP/", 5) != 0 || p == nil){
  498. werrstr("bad response from server");
  499. return -1;
  500. }
  501. buf[n] = 0;
  502. return atoi(p+1);
  503. }
  504. /* read in and crack the http headers, update u and r */
  505. void hhetag(char*, URL*, Range*);
  506. void hhmtime(char*, URL*, Range*);
  507. void hhclen(char*, URL*, Range*);
  508. void hhcrange(char*, URL*, Range*);
  509. void hhuri(char*, URL*, Range*);
  510. void hhlocation(char*, URL*, Range*);
  511. void hhauth(char*, URL*, Range*);
  512. struct {
  513. char *name;
  514. void (*f)(char*, URL*, Range*);
  515. } headers[] = {
  516. { "etag:", hhetag },
  517. { "last-modified:", hhmtime },
  518. { "content-length:", hhclen },
  519. { "content-range:", hhcrange },
  520. { "uri:", hhuri },
  521. { "location:", hhlocation },
  522. { "WWW-Authenticate:", hhauth },
  523. };
  524. int
  525. httpheaders(int fd, int cfd, URL *u, Range *r)
  526. {
  527. char buf[2048];
  528. char *p;
  529. int i, n;
  530. for(;;){
  531. n = getheader(fd, buf, sizeof(buf));
  532. if(n <= 0)
  533. break;
  534. if(cfd >= 0)
  535. fprint(cfd, "%s\n", buf);
  536. for(i = 0; i < nelem(headers); i++){
  537. n = strlen(headers[i].name);
  538. if(cistrncmp(buf, headers[i].name, n) == 0){
  539. /* skip field name and leading white */
  540. p = buf + n;
  541. while(*p == ' ' || *p == '\t')
  542. p++;
  543. (*headers[i].f)(p, u, r);
  544. break;
  545. }
  546. }
  547. }
  548. return n;
  549. }
  550. /*
  551. * read a single mime header, collect continuations.
  552. *
  553. * this routine assumes that there is a blank line twixt
  554. * the header and the message body, otherwise bytes will
  555. * be lost.
  556. */
  557. int
  558. getheader(int fd, char *buf, int n)
  559. {
  560. char *p, *e;
  561. int i;
  562. n--;
  563. p = buf;
  564. for(e = p + n; ; p += i){
  565. i = readline(fd, p, e-p);
  566. if(i < 0)
  567. return i;
  568. if(p == buf){
  569. /* first line */
  570. if(strchr(buf, ':') == nil)
  571. break; /* end of headers */
  572. } else {
  573. /* continuation line */
  574. if(*p != ' ' && *p != '\t'){
  575. unreadline(p);
  576. *p = 0;
  577. break; /* end of this header */
  578. }
  579. }
  580. }
  581. if(headerprint)
  582. print("%s\n", buf);
  583. if(debug)
  584. fprint(2, "%d <- %s\n", fd, buf);
  585. return p-buf;
  586. }
  587. void
  588. hhetag(char *p, URL *u, Range*)
  589. {
  590. if(u->etag != nil){
  591. if(strcmp(u->etag, p) != 0)
  592. sysfatal("file changed underfoot");
  593. } else
  594. u->etag = strdup(p);
  595. }
  596. char* monthchars = "janfebmaraprmayjunjulaugsepoctnovdec";
  597. void
  598. hhmtime(char *p, URL *u, Range*)
  599. {
  600. char *month, *day, *yr, *hms;
  601. char *fields[6];
  602. Tm tm, now;
  603. int i;
  604. i = getfields(p, fields, 6, 1, " \t");
  605. if(i < 5)
  606. return;
  607. day = fields[1];
  608. month = fields[2];
  609. yr = fields[3];
  610. hms = fields[4];
  611. /* default time */
  612. now = *gmtime(time(0));
  613. tm = now;
  614. tm.yday = 0;
  615. /* convert ascii month to a number twixt 1 and 12 */
  616. if(*month >= '0' && *month <= '9'){
  617. tm.mon = atoi(month) - 1;
  618. if(tm.mon < 0 || tm.mon > 11)
  619. tm.mon = 5;
  620. } else {
  621. for(p = month; *p; p++)
  622. *p = tolower(*p);
  623. for(i = 0; i < 12; i++)
  624. if(strncmp(&monthchars[i*3], month, 3) == 0){
  625. tm.mon = i;
  626. break;
  627. }
  628. }
  629. tm.mday = atoi(day);
  630. if(hms) {
  631. tm.hour = strtoul(hms, &p, 10);
  632. if(*p == ':') {
  633. p++;
  634. tm.min = strtoul(p, &p, 10);
  635. if(*p == ':') {
  636. p++;
  637. tm.sec = strtoul(p, &p, 10);
  638. }
  639. }
  640. if(tolower(*p) == 'p')
  641. tm.hour += 12;
  642. }
  643. if(yr) {
  644. tm.year = atoi(yr);
  645. if(tm.year >= 1900)
  646. tm.year -= 1900;
  647. } else {
  648. if(tm.mon > now.mon || (tm.mon == now.mon && tm.mday > now.mday+1))
  649. tm.year--;
  650. }
  651. strcpy(tm.zone, "GMT");
  652. /* convert to epoch seconds */
  653. u->mtime = tm2sec(&tm);
  654. }
  655. void
  656. hhclen(char *p, URL*, Range *r)
  657. {
  658. r->end = atoi(p);
  659. }
  660. void
  661. hhcrange(char *p, URL*, Range *r)
  662. {
  663. char *x;
  664. vlong l;
  665. l = 0;
  666. x = strchr(p, '/');
  667. if(x)
  668. l = atoll(x+1);
  669. if(l == 0) {
  670. x = strchr(p, '-');
  671. if(x)
  672. l = atoll(x+1);
  673. }
  674. if(l)
  675. r->end = l;
  676. }
  677. void
  678. hhuri(char *p, URL *u, Range*)
  679. {
  680. if(*p != '<')
  681. return;
  682. u->redirect = strdup(p+1);
  683. p = strchr(u->redirect, '>');
  684. if(p != nil)
  685. *p = 0;
  686. }
  687. void
  688. hhlocation(char *p, URL *u, Range*)
  689. {
  690. u->redirect = strdup(p);
  691. }
  692. void
  693. hhauth(char *p, URL *u, Range*)
  694. {
  695. char *f[4];
  696. UserPasswd *up;
  697. char *s, cred[64];
  698. if (cistrncmp(p, "basic ", 6) != 0)
  699. sysfatal("only Basic authentication supported");
  700. if (gettokens(p, f, nelem(f), "\"") < 2)
  701. sysfatal("garbled auth data");
  702. if ((up = auth_getuserpasswd(auth_getkey, "proto=pass service=http server=%q realm=%q",
  703. u->host, f[1])) == nil)
  704. sysfatal("cannot authenticate");
  705. s = smprint("%s:%s", up->user, up->passwd);
  706. if(enc64(cred, sizeof(cred), (uchar *)s, strlen(s)) == -1)
  707. sysfatal("enc64");
  708. free(s);
  709. assert(u->cred = strdup(cred));
  710. }
  711. enum
  712. {
  713. /* ftp return codes */
  714. Extra= 1,
  715. Success= 2,
  716. Incomplete= 3,
  717. TempFail= 4,
  718. PermFail= 5,
  719. Nnetdir= 64, /* max length of network directory paths */
  720. Ndialstr= 64, /* max length of dial strings */
  721. };
  722. int ftpcmd(int, char*, ...);
  723. int ftprcode(int, char*, int);
  724. int hello(int);
  725. int logon(int);
  726. int xfertype(int, char*);
  727. int passive(int, URL*);
  728. int active(int, URL*);
  729. int ftpxfer(int, Out*, Range*);
  730. int terminateftp(int, int);
  731. int getaddrport(char*, uchar*, uchar*);
  732. int ftprestart(int, Out*, URL*, Range*, long);
  733. int
  734. doftp(URL *u, URL *px, Range *r, Out *out, long mtime)
  735. {
  736. int pid, ctl, data, rv;
  737. Waitmsg *w;
  738. char msg[64];
  739. char conndir[NETPATHLEN];
  740. char *p;
  741. /* untested, proxy doesn't work with ftp (I think) */
  742. if(px->host == nil){
  743. ctl = dial(netmkaddr(u->host, tcpdir, u->port), 0, conndir, 0);
  744. } else {
  745. ctl = dial(netmkaddr(px->host, tcpdir, px->port), 0, conndir, 0);
  746. }
  747. if(ctl < 0)
  748. return Error;
  749. if(net == nil){
  750. p = strrchr(conndir, '/');
  751. *p = 0;
  752. snprint(tcpdir, sizeof(tcpdir), conndir);
  753. }
  754. initibuf();
  755. rv = hello(ctl);
  756. if(rv < 0)
  757. return terminateftp(ctl, rv);
  758. rv = logon(ctl);
  759. if(rv < 0)
  760. return terminateftp(ctl, rv);
  761. rv = xfertype(ctl, "I");
  762. if(rv < 0)
  763. return terminateftp(ctl, rv);
  764. /* if file is up to date and the right size, stop */
  765. if(ftprestart(ctl, out, u, r, mtime) > 0){
  766. close(ctl);
  767. return Eof;
  768. }
  769. /* first try passive mode, then active */
  770. data = passive(ctl, u);
  771. if(data < 0){
  772. data = active(ctl, u);
  773. if(data < 0)
  774. return Error;
  775. }
  776. /* fork */
  777. switch(pid = rfork(RFPROC|RFFDG|RFMEM)){
  778. case -1:
  779. close(data);
  780. return terminateftp(ctl, Error);
  781. case 0:
  782. ftpxfer(data, out, r);
  783. close(data);
  784. _exits(0);
  785. default:
  786. close(data);
  787. break;
  788. }
  789. /* wait for reply message */
  790. rv = ftprcode(ctl, msg, sizeof(msg));
  791. close(ctl);
  792. /* wait for process to terminate */
  793. w = nil;
  794. for(;;){
  795. free(w);
  796. w = wait();
  797. if(w == nil)
  798. return Error;
  799. if(w->pid == pid){
  800. if(w->msg[0] == 0){
  801. free(w);
  802. break;
  803. }
  804. werrstr("xfer: %s", w->msg);
  805. free(w);
  806. return Error;
  807. }
  808. }
  809. switch(rv){
  810. case Success:
  811. return Eof;
  812. case TempFail:
  813. return Server;
  814. default:
  815. return Error;
  816. }
  817. }
  818. int
  819. ftpcmd(int ctl, char *fmt, ...)
  820. {
  821. va_list arg;
  822. char buf[2*1024], *s;
  823. va_start(arg, fmt);
  824. s = vseprint(buf, buf + (sizeof(buf)-4) / sizeof(*buf), fmt, arg);
  825. va_end(arg);
  826. if(debug)
  827. fprint(2, "%d -> %s\n", ctl, buf);
  828. *s++ = '\r';
  829. *s++ = '\n';
  830. if(write(ctl, buf, s - buf) != s - buf)
  831. return -1;
  832. return 0;
  833. }
  834. int
  835. ftprcode(int ctl, char *msg, int len)
  836. {
  837. int rv;
  838. int i;
  839. char *p;
  840. len--; /* room for terminating null */
  841. for(;;){
  842. *msg = 0;
  843. i = readline(ctl, msg, len);
  844. if(i < 0)
  845. break;
  846. if(debug)
  847. fprint(2, "%d <- %s\n", ctl, msg);
  848. /* stop if not a continuation */
  849. rv = strtol(msg, &p, 10);
  850. if(rv >= 100 && rv < 600 && p==msg+3 && *p == ' ')
  851. return rv/100;
  852. }
  853. *msg = 0;
  854. return -1;
  855. }
  856. int
  857. hello(int ctl)
  858. {
  859. char msg[1024];
  860. /* wait for hello from other side */
  861. if(ftprcode(ctl, msg, sizeof(msg)) != Success){
  862. werrstr("HELLO: %s", msg);
  863. return Server;
  864. }
  865. return 0;
  866. }
  867. int
  868. getdec(char *p, int n)
  869. {
  870. int x = 0;
  871. int i;
  872. for(i = 0; i < n; i++)
  873. x = x*10 + (*p++ - '0');
  874. return x;
  875. }
  876. int
  877. ftprestart(int ctl, Out *out, URL *u, Range *r, long mtime)
  878. {
  879. Tm tm;
  880. char msg[1024];
  881. long x, rmtime;
  882. ftpcmd(ctl, "MDTM %s", u->page);
  883. if(ftprcode(ctl, msg, sizeof(msg)) != Success){
  884. r->start = 0;
  885. return 0; /* need to do something */
  886. }
  887. /* decode modification time */
  888. if(strlen(msg) < 4 + 4 + 2 + 2 + 2 + 2 + 2){
  889. r->start = 0;
  890. return 0; /* need to do something */
  891. }
  892. memset(&tm, 0, sizeof(tm));
  893. tm.year = getdec(msg+4, 4) - 1900;
  894. tm.mon = getdec(msg+4+4, 2) - 1;
  895. tm.mday = getdec(msg+4+4+2, 2);
  896. tm.hour = getdec(msg+4+4+2+2, 2);
  897. tm.min = getdec(msg+4+4+2+2+2, 2);
  898. tm.sec = getdec(msg+4+4+2+2+2+2, 2);
  899. strcpy(tm.zone, "GMT");
  900. rmtime = tm2sec(&tm);
  901. if(rmtime > mtime)
  902. r->start = 0;
  903. /* get size */
  904. ftpcmd(ctl, "SIZE %s", u->page);
  905. if(ftprcode(ctl, msg, sizeof(msg)) == Success){
  906. x = atol(msg+4);
  907. if(r->start == x)
  908. return 1; /* we're up to date */
  909. r->end = x;
  910. }
  911. /* seek to restart point */
  912. if(r->start > 0){
  913. ftpcmd(ctl, "REST %lud", r->start);
  914. if(ftprcode(ctl, msg, sizeof(msg)) == Incomplete){
  915. setoffset(out, r->start);
  916. }else
  917. r->start = 0;
  918. }
  919. return 0; /* need to do something */
  920. }
  921. int
  922. logon(int ctl)
  923. {
  924. char msg[1024];
  925. /* login anonymous */
  926. ftpcmd(ctl, "USER anonymous");
  927. switch(ftprcode(ctl, msg, sizeof(msg))){
  928. case Success:
  929. return 0;
  930. case Incomplete:
  931. break; /* need password */
  932. default:
  933. werrstr("USER: %s", msg);
  934. return Server;
  935. }
  936. /* send user id as password */
  937. sprint(msg, "%s@closedmind.org", getuser());
  938. ftpcmd(ctl, "PASS %s", msg);
  939. if(ftprcode(ctl, msg, sizeof(msg)) != Success){
  940. werrstr("PASS: %s", msg);
  941. return Server;
  942. }
  943. return 0;
  944. }
  945. int
  946. xfertype(int ctl, char *t)
  947. {
  948. char msg[1024];
  949. ftpcmd(ctl, "TYPE %s", t);
  950. if(ftprcode(ctl, msg, sizeof(msg)) != Success){
  951. werrstr("TYPE %s: %s", t, msg);
  952. return Server;
  953. }
  954. return 0;
  955. }
  956. int
  957. passive(int ctl, URL *u)
  958. {
  959. char msg[1024];
  960. char ipaddr[32];
  961. char *f[6];
  962. char *p;
  963. int fd;
  964. int port;
  965. char aport[12];
  966. ftpcmd(ctl, "PASV");
  967. if(ftprcode(ctl, msg, sizeof(msg)) != Success)
  968. return Error;
  969. /* get address and port number from reply, this is AI */
  970. p = strchr(msg, '(');
  971. if(p == nil){
  972. for(p = msg+3; *p; p++)
  973. if(isdigit(*p))
  974. break;
  975. } else
  976. p++;
  977. if(getfields(p, f, 6, 0, ",)") < 6){
  978. werrstr("ftp protocol botch");
  979. return Server;
  980. }
  981. snprint(ipaddr, sizeof(ipaddr), "%s.%s.%s.%s",
  982. f[0], f[1], f[2], f[3]);
  983. port = ((atoi(f[4])&0xff)<<8) + (atoi(f[5])&0xff);
  984. sprint(aport, "%d", port);
  985. /* open data connection */
  986. fd = dial(netmkaddr(ipaddr, tcpdir, aport), 0, 0, 0);
  987. if(fd < 0){
  988. werrstr("passive mode failed: %r");
  989. return Error;
  990. }
  991. /* tell remote to send a file */
  992. ftpcmd(ctl, "RETR %s", u->page);
  993. if(ftprcode(ctl, msg, sizeof(msg)) != Extra){
  994. werrstr("RETR %s: %s", u->page, msg);
  995. return Error;
  996. }
  997. return fd;
  998. }
  999. int
  1000. active(int ctl, URL *u)
  1001. {
  1002. char msg[1024];
  1003. char dir[40], ldir[40];
  1004. uchar ipaddr[4];
  1005. uchar port[2];
  1006. int lcfd, dfd, afd;
  1007. /* announce a port for the call back */
  1008. snprint(msg, sizeof(msg), "%s!*!0", tcpdir);
  1009. afd = announce(msg, dir);
  1010. if(afd < 0)
  1011. return Error;
  1012. /* get a local address/port of the annoucement */
  1013. if(getaddrport(dir, ipaddr, port) < 0){
  1014. close(afd);
  1015. return Error;
  1016. }
  1017. /* tell remote side address and port*/
  1018. ftpcmd(ctl, "PORT %d,%d,%d,%d,%d,%d", ipaddr[0], ipaddr[1], ipaddr[2],
  1019. ipaddr[3], port[0], port[1]);
  1020. if(ftprcode(ctl, msg, sizeof(msg)) != Success){
  1021. close(afd);
  1022. werrstr("active: %s", msg);
  1023. return Error;
  1024. }
  1025. /* tell remote to send a file */
  1026. ftpcmd(ctl, "RETR %s", u->page);
  1027. if(ftprcode(ctl, msg, sizeof(msg)) != Extra){
  1028. close(afd);
  1029. werrstr("RETR: %s", msg);
  1030. return Server;
  1031. }
  1032. /* wait for a connection */
  1033. lcfd = listen(dir, ldir);
  1034. if(lcfd < 0){
  1035. close(afd);
  1036. return Error;
  1037. }
  1038. dfd = accept(lcfd, ldir);
  1039. if(dfd < 0){
  1040. close(afd);
  1041. close(lcfd);
  1042. return Error;
  1043. }
  1044. close(afd);
  1045. close(lcfd);
  1046. return dfd;
  1047. }
  1048. int
  1049. ftpxfer(int in, Out *out, Range *r)
  1050. {
  1051. char buf[1024];
  1052. long vtime;
  1053. int i, n;
  1054. vtime = 0;
  1055. for(n = 0;;n += i){
  1056. i = read(in, buf, sizeof(buf));
  1057. if(i == 0)
  1058. break;
  1059. if(i < 0)
  1060. return Error;
  1061. if(output(out, buf, i) != i)
  1062. return Error;
  1063. r->start += i;
  1064. if(verbose && (vtime != time(0) || r->start == r->end)) {
  1065. vtime = time(0);
  1066. fprint(2, "%ld %ld\n", r->start, r->end);
  1067. }
  1068. }
  1069. return n;
  1070. }
  1071. int
  1072. terminateftp(int ctl, int rv)
  1073. {
  1074. close(ctl);
  1075. return rv;
  1076. }
  1077. /*
  1078. * case insensitive strcmp (why aren't these in libc?)
  1079. */
  1080. int
  1081. cistrncmp(char *a, char *b, int n)
  1082. {
  1083. while(n-- > 0){
  1084. if(tolower(*a++) != tolower(*b++))
  1085. return -1;
  1086. }
  1087. return 0;
  1088. }
  1089. int
  1090. cistrcmp(char *a, char *b)
  1091. {
  1092. while(*a || *b)
  1093. if(tolower(*a++) != tolower(*b++))
  1094. return -1;
  1095. return 0;
  1096. }
  1097. /*
  1098. * buffered io
  1099. */
  1100. struct
  1101. {
  1102. char *rp;
  1103. char *wp;
  1104. char buf[4*1024];
  1105. } b;
  1106. void
  1107. initibuf(void)
  1108. {
  1109. b.rp = b.wp = b.buf;
  1110. }
  1111. /*
  1112. * read a possibly buffered line, strip off trailing while
  1113. */
  1114. int
  1115. readline(int fd, char *buf, int len)
  1116. {
  1117. int n;
  1118. char *p;
  1119. int eof = 0;
  1120. len--;
  1121. for(p = buf;;){
  1122. if(b.rp >= b.wp){
  1123. n = read(fd, b.wp, sizeof(b.buf)/2);
  1124. if(n < 0)
  1125. return -1;
  1126. if(n == 0){
  1127. eof = 1;
  1128. break;
  1129. }
  1130. b.wp += n;
  1131. }
  1132. n = *b.rp++;
  1133. if(len > 0){
  1134. *p++ = n;
  1135. len--;
  1136. }
  1137. if(n == '\n')
  1138. break;
  1139. }
  1140. /* drop trailing white */
  1141. for(;;){
  1142. if(p <= buf)
  1143. break;
  1144. n = *(p-1);
  1145. if(n != ' ' && n != '\t' && n != '\r' && n != '\n')
  1146. break;
  1147. p--;
  1148. }
  1149. *p = 0;
  1150. if(eof && p == buf)
  1151. return -1;
  1152. return p-buf;
  1153. }
  1154. void
  1155. unreadline(char *line)
  1156. {
  1157. int i, n;
  1158. i = strlen(line);
  1159. n = b.wp-b.rp;
  1160. memmove(&b.buf[i+1], b.rp, n);
  1161. memmove(b.buf, line, i);
  1162. b.buf[i] = '\n';
  1163. b.rp = b.buf;
  1164. b.wp = b.rp + i + 1 + n;
  1165. }
  1166. int
  1167. readibuf(int fd, char *buf, int len)
  1168. {
  1169. int n;
  1170. n = b.wp-b.rp;
  1171. if(n > 0){
  1172. if(n > len)
  1173. n = len;
  1174. memmove(buf, b.rp, n);
  1175. b.rp += n;
  1176. return n;
  1177. }
  1178. return read(fd, buf, len);
  1179. }
  1180. int
  1181. dfprint(int fd, char *fmt, ...)
  1182. {
  1183. char buf[4*1024];
  1184. va_list arg;
  1185. va_start(arg, fmt);
  1186. vseprint(buf, buf+sizeof(buf), fmt, arg);
  1187. va_end(arg);
  1188. if(debug)
  1189. fprint(2, "%d -> %s", fd, buf);
  1190. return fprint(fd, "%s", buf);
  1191. }
  1192. int
  1193. getaddrport(char *dir, uchar *ipaddr, uchar *port)
  1194. {
  1195. char buf[256];
  1196. int fd, i;
  1197. char *p;
  1198. snprint(buf, sizeof(buf), "%s/local", dir);
  1199. fd = open(buf, OREAD);
  1200. if(fd < 0)
  1201. return -1;
  1202. i = read(fd, buf, sizeof(buf)-1);
  1203. close(fd);
  1204. if(i <= 0)
  1205. return -1;
  1206. buf[i] = 0;
  1207. p = strchr(buf, '!');
  1208. if(p != nil)
  1209. *p++ = 0;
  1210. v4parseip(ipaddr, buf);
  1211. i = atoi(p);
  1212. port[0] = i>>8;
  1213. port[1] = i;
  1214. return 0;
  1215. }
  1216. void
  1217. md5free(DigestState *state)
  1218. {
  1219. uchar x[MD5dlen];
  1220. md5(nil, 0, x, state);
  1221. }
  1222. DigestState*
  1223. md5dup(DigestState *state)
  1224. {
  1225. char *p;
  1226. p = md5pickle(state);
  1227. if(p == nil)
  1228. sysfatal("md5pickle: %r");
  1229. state = md5unpickle(p);
  1230. if(state == nil)
  1231. sysfatal("md5unpickle: %r");
  1232. free(p);
  1233. return state;
  1234. }
  1235. void
  1236. setoffset(Out *out, int offset)
  1237. {
  1238. md5free(out->curr);
  1239. if(offset == 0)
  1240. out->curr = md5(nil, 0, nil, nil);
  1241. else
  1242. out->curr = nil;
  1243. out->offset = offset;
  1244. out->written = offset;
  1245. if(ofile != nil)
  1246. if(seek(out->fd, offset, 0) != offset)
  1247. sysfatal("seek: %r");
  1248. }
  1249. /*
  1250. * write some output, discarding it (but keeping track)
  1251. * if we've already written it. if we've gone backwards,
  1252. * verify that everything previously written matches
  1253. * that which would have been written from the current
  1254. * output.
  1255. */
  1256. int
  1257. output(Out *out, char *buf, int nb)
  1258. {
  1259. int n, d;
  1260. uchar m0[MD5dlen], m1[MD5dlen];
  1261. n = nb;
  1262. d = out->written - out->offset;
  1263. assert(d >= 0);
  1264. if(d > 0){
  1265. if(n < d){
  1266. if(out->curr != nil)
  1267. md5((uchar*)buf, n, nil, out->curr);
  1268. out->offset += n;
  1269. return n;
  1270. }
  1271. if(out->curr != nil){
  1272. md5((uchar*)buf, d, m0, out->curr);
  1273. out->curr = nil;
  1274. md5(nil, 0, m1, md5dup(out->hiwat));
  1275. if(memcmp(m0, m1, MD5dlen) != 0){
  1276. fprint(2, "integrity check failure at offset %d\n", out->written);
  1277. return -1;
  1278. }
  1279. }
  1280. buf += d;
  1281. n -= d;
  1282. out->offset += d;
  1283. }
  1284. if(n > 0){
  1285. out->hiwat = md5((uchar*)buf, n, nil, out->hiwat);
  1286. n = write(out->fd, buf, n);
  1287. if(n > 0){
  1288. out->offset += n;
  1289. out->written += n;
  1290. }
  1291. }
  1292. return n + d;
  1293. }