hget.c 21 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186118711881189119011911192119311941195119611971198119912001201120212031204120512061207120812091210121112121213121412151216121712181219122012211222122312241225122612271228122912301231123212331234123512361237123812391240124112421243124412451246124712481249125012511252125312541255125612571258125912601261126212631264126512661267126812691270127112721273127412751276127712781279128012811282128312841285128612871288128912901291
  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. typedef struct URL URL;
  8. struct URL
  9. {
  10. int method;
  11. char *host;
  12. char *port;
  13. char *page;
  14. char *etag;
  15. char *redirect;
  16. char *postbody;
  17. long mtime;
  18. };
  19. typedef struct Range Range;
  20. struct Range
  21. {
  22. long start; /* only 2 gig supported, tdb */
  23. long end;
  24. };
  25. enum
  26. {
  27. Http,
  28. Https,
  29. Ftp,
  30. Other
  31. };
  32. enum
  33. {
  34. Eof = 0,
  35. Error = -1,
  36. Server = -2,
  37. Changed = -3,
  38. };
  39. int debug;
  40. char *ofile;
  41. int doftp(URL*, Range*, int, long);
  42. int dohttp(URL*, Range*, int, long);
  43. int crackurl(URL*, char*);
  44. Range* crackrange(char*);
  45. int getheader(int, char*, int);
  46. int httpheaders(int, int, URL*, Range*);
  47. int httprcode(int);
  48. int cistrncmp(char*, char*, int);
  49. int cistrcmp(char*, char*);
  50. void initibuf(void);
  51. int readline(int, char*, int);
  52. int readibuf(int, char*, int);
  53. int dfprint(int, char*, ...);
  54. void unreadline(char*);
  55. int verbose;
  56. char *net;
  57. char tcpdir[64];
  58. struct {
  59. char *name;
  60. int (*f)(URL*, Range*, int, long);
  61. } method[] = {
  62. [Http] { "http", dohttp },
  63. [Https] { "https", dohttp },
  64. [Ftp] { "ftp", doftp },
  65. [Other] { "_______", nil },
  66. };
  67. void
  68. usage(void)
  69. {
  70. fprint(2, "usage: %s [-v] [-o outfile] [-p body] [-x netmtpt] url\n", argv0);
  71. exits("usage");
  72. }
  73. void
  74. main(int argc, char **argv)
  75. {
  76. URL u;
  77. Range r;
  78. int fd, errs, n;
  79. ulong mtime;
  80. Dir *d;
  81. char postbody[4096], *p, *e, *t;
  82. ofile = nil;
  83. p = postbody;
  84. e = p + sizeof(postbody);
  85. r.start = 0;
  86. r.end = -1;
  87. mtime = 0;
  88. memset(&u, 0, sizeof(u));
  89. ARGBEGIN {
  90. case 'o':
  91. ofile = ARGF();
  92. break;
  93. case 'd':
  94. debug = 1;
  95. break;
  96. case 'v':
  97. verbose = 1;
  98. break;
  99. case 'x':
  100. net = ARGF();
  101. if(net == nil)
  102. usage();
  103. break;
  104. case 'p':
  105. t = ARGF();
  106. if(t == nil)
  107. usage();
  108. if(p != postbody)
  109. p = seprint(p, e, "&%s", t);
  110. else
  111. p = seprint(p, e, "%s", t);
  112. u.postbody = postbody;
  113. break;
  114. default:
  115. usage();
  116. } ARGEND;
  117. if(net != nil){
  118. if(strlen(net) > sizeof(tcpdir)-5)
  119. sysfatal("network mount point too long");
  120. snprint(tcpdir, sizeof(tcpdir), "%s/tcp", net);
  121. } else
  122. snprint(tcpdir, sizeof(tcpdir), "tcp");
  123. if(argc != 1)
  124. usage();
  125. fd = 1;
  126. if(ofile != nil){
  127. d = dirstat(ofile);
  128. if(d == nil){
  129. fd = create(ofile, OWRITE, 0664);
  130. if(fd < 0)
  131. sysfatal("creating %s: %r", ofile);
  132. } else {
  133. fd = open(ofile, OWRITE);
  134. if(fd < 0)
  135. sysfatal("can't open %s: %r", ofile);
  136. r.start = d->length;
  137. mtime = d->mtime;
  138. free(d);
  139. }
  140. }
  141. errs = 0;
  142. if(crackurl(&u, argv[0]) < 0)
  143. sysfatal("%r");
  144. for(;;){
  145. /* transfer data */
  146. werrstr("");
  147. seek(fd, 0, 0);
  148. n = (*method[u.method].f)(&u, &r, fd, mtime);
  149. switch(n){
  150. case Eof:
  151. exits(0);
  152. break;
  153. case Error:
  154. if(errs++ < 10)
  155. continue;
  156. sysfatal("too many errors with no progress %r");
  157. break;
  158. case Server:
  159. sysfatal("server returned: %r");
  160. break;
  161. }
  162. /* forward progress */
  163. errs = 0;
  164. r.start += n;
  165. if(r.start >= r.end)
  166. break;
  167. }
  168. exits(0);
  169. }
  170. int
  171. crackurl(URL *u, char *s)
  172. {
  173. char *p;
  174. int i;
  175. if(u->host != nil){
  176. free(u->host);
  177. u->host = nil;
  178. }
  179. if(u->page != nil){
  180. free(u->page);
  181. u->page = nil;
  182. }
  183. /* get type */
  184. u->method = Other;
  185. for(p = s; *p; p++){
  186. if(*p == '/'){
  187. u->method = Http;
  188. p = s;
  189. break;
  190. }
  191. if(*p == ':' && *(p+1)=='/' && *(p+2)=='/'){
  192. *p = 0;
  193. p += 3;
  194. for(i = 0; i < nelem(method); i++){
  195. if(cistrcmp(s, method[i].name) == 0){
  196. u->method = i;
  197. break;
  198. }
  199. }
  200. break;
  201. }
  202. }
  203. if(u->method == Other){
  204. werrstr("unsupported URL type %s", s);
  205. return -1;
  206. }
  207. /* get system */
  208. s = p;
  209. p = strchr(s, '/');
  210. if(p == nil){
  211. u->host = strdup(s);
  212. u->page = strdup("/");
  213. } else {
  214. u->page = strdup(p);
  215. *p = 0;
  216. u->host = strdup(s);
  217. *p = '/';
  218. }
  219. if(p = strchr(u->host, ':')) {
  220. *p++ = 0;
  221. u->port = p;
  222. } else
  223. u->port = method[u->method].name;
  224. if(*(u->host) == 0){
  225. werrstr("bad url, null host");
  226. return -1;
  227. }
  228. return 0;
  229. }
  230. char *day[] = {
  231. "Sun", "Mon", "Tue", "Wed", "Thu", "Fri", "Sat"
  232. };
  233. char *month[] = {
  234. "Jan", "Feb", "Mar", "Apr", "May", "Jun", "Jul", "Aug", "Sep", "Oct", "Nov", "Dec"
  235. };
  236. struct
  237. {
  238. int fd;
  239. long mtime;
  240. } note;
  241. void
  242. catch(void*, char*)
  243. {
  244. Dir d;
  245. nulldir(&d);
  246. d.mtime = note.mtime;
  247. if(dirfwstat(note.fd, &d) < 0)
  248. sysfatal("catch: can't dirfwstat: %r");
  249. noted(NDFLT);
  250. }
  251. int
  252. dohttp(URL *u, Range *r, int out, long mtime)
  253. {
  254. int fd, cfd;
  255. int redirect, loop;
  256. int n, rv, code;
  257. long tot, vtime;
  258. Tm *tm;
  259. char buf[1024];
  260. char err[ERRMAX];
  261. /* always move back to a previous 512 byte bound because some
  262. * servers can't seem to deal with requests that start at the
  263. * end of the file
  264. */
  265. if(r->start)
  266. r->start = ((r->start-1)/512)*512;
  267. /* loop for redirects, requires reading both response code and headers */
  268. fd = -1;
  269. for(loop = 0; loop < 32; loop++){
  270. fd = dial(netmkaddr(u->host, tcpdir, u->port), 0, 0, 0);
  271. if(fd < 0)
  272. return Error;
  273. if(u->method == Https){
  274. int tfd;
  275. TLSconn conn;
  276. memset(&conn, 0, sizeof conn);
  277. tfd = tlsClient(fd, &conn);
  278. if(tfd < 0){
  279. fprint(2, "tlsClient: %r\n");
  280. close(fd);
  281. return Error;
  282. }
  283. /* BUG: check cert here? */
  284. if(conn.cert)
  285. free(conn.cert);
  286. close(fd);
  287. fd = tfd;
  288. }
  289. /* write request, use range if not start of file */
  290. if(u->postbody == nil){
  291. dfprint(fd, "GET %s HTTP/1.0\r\n"
  292. "Host: %s\r\n"
  293. "User-agent: Plan9/hget\r\n",
  294. u->page, u->host);
  295. } else {
  296. dfprint(fd, "POST %s HTTP/1.0\r\n"
  297. "Host: %s\r\n"
  298. "Content-type: application/x-www-form-urlencoded\r\n"
  299. "Content-length: %d\r\n"
  300. "User-agent: Plan9/hget\r\n"
  301. "\r\n",
  302. u->page, u->host, strlen(u->postbody));
  303. dfprint(fd, "%s", u->postbody);
  304. }
  305. if(r->start != 0){
  306. dfprint(fd, "Range: bytes=%d-\n", r->start);
  307. if(u->etag != nil){
  308. dfprint(fd, "If-range: %s\n", u->etag);
  309. } else {
  310. tm = gmtime(mtime);
  311. dfprint(fd, "If-range: %s, %d %s %d %2d:%2.2d:%2.2d GMT\n",
  312. day[tm->wday], tm->mday, month[tm->mon],
  313. tm->year+1900, tm->hour, tm->min, tm->sec);
  314. }
  315. }
  316. if((cfd = open("/mnt/webcookies/http", ORDWR)) >= 0){
  317. if(fprint(cfd, "http://%s%s", u->host, u->page) > 0){
  318. while((n = read(cfd, buf, sizeof buf)) > 0){
  319. if(debug)
  320. write(2, buf, n);
  321. write(fd, buf, n);
  322. }
  323. }else{
  324. close(cfd);
  325. cfd = -1;
  326. }
  327. }
  328. dfprint(fd, "\r\n", u->host);
  329. redirect = 0;
  330. initibuf();
  331. code = httprcode(fd);
  332. switch(code){
  333. case Error: /* connection timed out */
  334. case Eof:
  335. close(fd);
  336. close(cfd);
  337. return code;
  338. case 200: /* OK */
  339. case 201: /* Created */
  340. case 202: /* Accepted */
  341. if(ofile == nil && r->start != 0)
  342. sysfatal("page changed underfoot");
  343. break;
  344. case 204: /* No Content */
  345. sysfatal("No Content");
  346. case 206: /* Partial Content */
  347. seek(out, r->start, 0);
  348. break;
  349. case 301: /* Moved Permanently */
  350. case 302: /* Moved Temporarily */
  351. redirect = 1;
  352. u->postbody = nil;
  353. break;
  354. case 304: /* Not Modified */
  355. break;
  356. case 400: /* Bad Request */
  357. sysfatal("Bad Request");
  358. case 401: /* Unauthorized */
  359. case 402: /* ??? */
  360. sysfatal("Unauthorized");
  361. case 403: /* Forbidden */
  362. sysfatal("Forbidden by server");
  363. case 404: /* Not Found */
  364. sysfatal("Not found on server");
  365. case 500: /* Internal server error */
  366. sysfatal("Server choked");
  367. case 501: /* Not implemented */
  368. sysfatal("Server can't do it!");
  369. case 502: /* Bad gateway */
  370. sysfatal("Bad gateway");
  371. case 503: /* Service unavailable */
  372. sysfatal("Service unavailable");
  373. default:
  374. sysfatal("Unknown response code %d", code);
  375. }
  376. if(u->redirect != nil){
  377. free(u->redirect);
  378. u->redirect = nil;
  379. }
  380. rv = httpheaders(fd, cfd, u, r);
  381. close(cfd);
  382. if(rv != 0){
  383. close(fd);
  384. return rv;
  385. }
  386. if(!redirect)
  387. break;
  388. if(u->redirect == nil)
  389. sysfatal("redirect: no URL");
  390. if(crackurl(u, u->redirect) < 0)
  391. sysfatal("redirect: %r");
  392. }
  393. /* transfer whatever you get */
  394. if(ofile != nil && u->mtime != 0){
  395. note.fd = out;
  396. note.mtime = u->mtime;
  397. notify(catch);
  398. }
  399. tot = 0;
  400. vtime = 0;
  401. for(;;){
  402. n = readibuf(fd, buf, sizeof(buf));
  403. if(n <= 0)
  404. break;
  405. if(write(out, buf, n) != n)
  406. break;
  407. tot += n;
  408. if(verbose && vtime != time(0)) {
  409. vtime = time(0);
  410. fprint(2, "%ld %ld\n", r->start+tot, r->end);
  411. }
  412. }
  413. notify(nil);
  414. close(fd);
  415. if(ofile != nil && u->mtime != 0){
  416. Dir d;
  417. rerrstr(err, sizeof err);
  418. nulldir(&d);
  419. d.mtime = u->mtime;
  420. if(dirfwstat(out, &d) < 0)
  421. fprint(2, "couldn't set mtime: %r\n");
  422. errstr(err, sizeof err);
  423. }
  424. return tot;
  425. }
  426. /* get the http response code */
  427. int
  428. httprcode(int fd)
  429. {
  430. int n;
  431. char *p;
  432. char buf[256];
  433. n = readline(fd, buf, sizeof(buf)-1);
  434. if(n <= 0)
  435. return n;
  436. if(debug)
  437. fprint(2, "%d <- %s\n", fd, buf);
  438. p = strchr(buf, ' ');
  439. if(strncmp(buf, "HTTP/", 5) != 0 || p == nil){
  440. werrstr("bad response from server");
  441. return -1;
  442. }
  443. buf[n] = 0;
  444. return atoi(p+1);
  445. }
  446. /* read in and crack the http headers, update u and r */
  447. void hhetag(char*, URL*, Range*);
  448. void hhmtime(char*, URL*, Range*);
  449. void hhclen(char*, URL*, Range*);
  450. void hhcrange(char*, URL*, Range*);
  451. void hhuri(char*, URL*, Range*);
  452. void hhlocation(char*, URL*, Range*);
  453. struct {
  454. char *name;
  455. void (*f)(char*, URL*, Range*);
  456. } headers[] = {
  457. { "etag:", hhetag },
  458. { "last-modified:", hhmtime },
  459. { "content-length:", hhclen },
  460. { "content-range:", hhcrange },
  461. { "uri:", hhuri },
  462. { "location:", hhlocation },
  463. };
  464. int
  465. httpheaders(int fd, int cfd, URL *u, Range *r)
  466. {
  467. char buf[2048];
  468. char *p;
  469. int i, n;
  470. for(;;){
  471. n = getheader(fd, buf, sizeof(buf));
  472. if(n <= 0)
  473. break;
  474. if(cfd >= 0)
  475. fprint(cfd, "%s\n", buf);
  476. for(i = 0; i < nelem(headers); i++){
  477. n = strlen(headers[i].name);
  478. if(cistrncmp(buf, headers[i].name, n) == 0){
  479. /* skip field name and leading white */
  480. p = buf + n;
  481. while(*p == ' ' || *p == '\t')
  482. p++;
  483. (*headers[i].f)(p, u, r);
  484. break;
  485. }
  486. }
  487. }
  488. return n;
  489. }
  490. /*
  491. * read a single mime header, collect continuations.
  492. *
  493. * this routine assumes that there is a blank line twixt
  494. * the header and the message body, otherwise bytes will
  495. * be lost.
  496. */
  497. int
  498. getheader(int fd, char *buf, int n)
  499. {
  500. char *p, *e;
  501. int i;
  502. n--;
  503. p = buf;
  504. for(e = p + n; ; p += i){
  505. i = readline(fd, p, e-p);
  506. if(i < 0)
  507. return i;
  508. if(p == buf){
  509. /* first line */
  510. if(strchr(buf, ':') == nil)
  511. break; /* end of headers */
  512. } else {
  513. /* continuation line */
  514. if(*p != ' ' && *p != '\t'){
  515. unreadline(p);
  516. *p = 0;
  517. break; /* end of this header */
  518. }
  519. }
  520. }
  521. if(debug)
  522. fprint(2, "%d <- %s\n", fd, buf);
  523. return p-buf;
  524. }
  525. void
  526. hhetag(char *p, URL *u, Range*)
  527. {
  528. if(u->etag != nil){
  529. if(strcmp(u->etag, p) != 0)
  530. sysfatal("file changed underfoot");
  531. } else
  532. u->etag = strdup(p);
  533. }
  534. char* monthchars = "janfebmaraprmayjunjulaugsepoctnovdec";
  535. void
  536. hhmtime(char *p, URL *u, Range*)
  537. {
  538. char *month, *day, *yr, *hms;
  539. char *fields[6];
  540. Tm tm, now;
  541. int i;
  542. i = getfields(p, fields, 6, 1, " \t");
  543. if(i < 5)
  544. return;
  545. day = fields[1];
  546. month = fields[2];
  547. yr = fields[3];
  548. hms = fields[4];
  549. /* default time */
  550. now = *gmtime(time(0));
  551. tm = now;
  552. /* convert ascii month to a number twixt 1 and 12 */
  553. if(*month >= '0' && *month <= '9'){
  554. tm.mon = atoi(month) - 1;
  555. if(tm.mon < 0 || tm.mon > 11)
  556. tm.mon = 5;
  557. } else {
  558. for(p = month; *p; p++)
  559. *p = tolower(*p);
  560. for(i = 0; i < 12; i++)
  561. if(strncmp(&monthchars[i*3], month, 3) == 0){
  562. tm.mon = i;
  563. break;
  564. }
  565. }
  566. tm.mday = atoi(day);
  567. if(hms) {
  568. tm.hour = strtoul(hms, &p, 10);
  569. if(*p == ':') {
  570. p++;
  571. tm.min = strtoul(p, &p, 10);
  572. if(*p == ':') {
  573. p++;
  574. tm.sec = strtoul(p, &p, 10);
  575. }
  576. }
  577. if(tolower(*p) == 'p')
  578. tm.hour += 12;
  579. }
  580. if(yr) {
  581. tm.year = atoi(yr);
  582. if(tm.year >= 1900)
  583. tm.year -= 1900;
  584. } else {
  585. if(tm.mon > now.mon || (tm.mon == now.mon && tm.mday > now.mday+1))
  586. tm.year--;
  587. }
  588. strcpy(tm.zone, "GMT");
  589. /* convert to epoch seconds */
  590. u->mtime = tm2sec(&tm);
  591. }
  592. void
  593. hhclen(char *p, URL*, Range *r)
  594. {
  595. r->end = atoi(p);
  596. }
  597. void
  598. hhcrange(char *p, URL*, Range *r)
  599. {
  600. char *x;
  601. vlong l;
  602. l = 0;
  603. x = strchr(p, '/');
  604. if(x)
  605. l = atoll(x+1);
  606. if(l == 0)
  607. x = strchr(p, '-');
  608. if(x)
  609. l = atoll(x+1);
  610. if(l)
  611. r->end = l;
  612. }
  613. void
  614. hhuri(char *p, URL *u, Range*)
  615. {
  616. if(*p != '<')
  617. return;
  618. u->redirect = strdup(p+1);
  619. p = strchr(u->redirect, '>');
  620. if(p != nil)
  621. *p = 0;
  622. }
  623. void
  624. hhlocation(char *p, URL *u, Range*)
  625. {
  626. u->redirect = strdup(p);
  627. }
  628. enum
  629. {
  630. /* ftp return codes */
  631. Extra= 1,
  632. Success= 2,
  633. Incomplete= 3,
  634. TempFail= 4,
  635. PermFail= 5,
  636. Nnetdir= 64, /* max length of network directory paths */
  637. Ndialstr= 64, /* max length of dial strings */
  638. };
  639. int ftpcmd(int, char*, ...);
  640. int ftprcode(int, char*, int);
  641. int hello(int);
  642. int logon(int);
  643. int xfertype(int, char*);
  644. int passive(int, URL*);
  645. int active(int, URL*);
  646. int ftpxfer(int, int, Range*);
  647. int terminateftp(int, int);
  648. int getaddrport(char*, uchar*, uchar*);
  649. int ftprestart(int, int, URL*, Range*, long);
  650. int
  651. doftp(URL *u, Range *r, int out, long mtime)
  652. {
  653. int pid, ctl, data, rv;
  654. Waitmsg *w;
  655. char msg[64];
  656. char conndir[NETPATHLEN];
  657. char *p;
  658. ctl = dial(netmkaddr(u->host, tcpdir, u->port), 0, conndir, 0);
  659. if(ctl < 0)
  660. return Error;
  661. if(net == nil){
  662. p = strrchr(conndir, '/');
  663. *p = 0;
  664. snprint(tcpdir, sizeof(tcpdir), conndir);
  665. }
  666. initibuf();
  667. rv = hello(ctl);
  668. if(rv < 0)
  669. return terminateftp(ctl, rv);
  670. rv = logon(ctl);
  671. if(rv < 0)
  672. return terminateftp(ctl, rv);
  673. rv = xfertype(ctl, "I");
  674. if(rv < 0)
  675. return terminateftp(ctl, rv);
  676. /* if file is up to date and the right size, stop */
  677. if(ftprestart(ctl, out, u, r, mtime) > 0){
  678. close(ctl);
  679. return Eof;
  680. }
  681. /* first try passive mode, then active */
  682. data = passive(ctl, u);
  683. if(data < 0){
  684. data = active(ctl, u);
  685. if(data < 0)
  686. return Error;
  687. }
  688. /* fork */
  689. switch(pid = rfork(RFPROC|RFFDG|RFMEM)){
  690. case -1:
  691. close(data);
  692. return terminateftp(ctl, Error);
  693. case 0:
  694. ftpxfer(data, out, r);
  695. close(data);
  696. _exits(0);
  697. default:
  698. close(data);
  699. break;
  700. }
  701. /* wait for reply message */
  702. rv = ftprcode(ctl, msg, sizeof(msg));
  703. close(ctl);
  704. /* wait for process to terminate */
  705. w = nil;
  706. for(;;){
  707. free(w);
  708. w = wait();
  709. if(w == nil)
  710. return Error;
  711. if(w->pid == pid){
  712. if(w->msg[0] == 0){
  713. free(w);
  714. break;
  715. }
  716. werrstr("xfer: %s", w->msg);
  717. free(w);
  718. return Error;
  719. }
  720. }
  721. switch(rv){
  722. case Success:
  723. return Eof;
  724. case TempFail:
  725. return Server;
  726. default:
  727. return Error;
  728. }
  729. }
  730. int
  731. ftpcmd(int ctl, char *fmt, ...)
  732. {
  733. va_list arg;
  734. char buf[2*1024], *s;
  735. va_start(arg, fmt);
  736. s = vseprint(buf, buf + (sizeof(buf)-4) / sizeof(*buf), fmt, arg);
  737. va_end(arg);
  738. if(debug)
  739. fprint(2, "%d -> %s\n", ctl, buf);
  740. *s++ = '\r';
  741. *s++ = '\n';
  742. if(write(ctl, buf, s - buf) != s - buf)
  743. return -1;
  744. return 0;
  745. }
  746. int
  747. ftprcode(int ctl, char *msg, int len)
  748. {
  749. int rv;
  750. int i;
  751. char *p;
  752. len--; /* room for terminating null */
  753. for(;;){
  754. *msg = 0;
  755. i = readline(ctl, msg, len);
  756. if(i < 0)
  757. break;
  758. if(debug)
  759. fprint(2, "%d <- %s\n", ctl, msg);
  760. /* stop if not a continuation */
  761. rv = strtol(msg, &p, 10);
  762. if(rv >= 100 && rv < 600 && p==msg+3 && *p == ' ')
  763. return rv/100;
  764. }
  765. *msg = 0;
  766. return -1;
  767. }
  768. int
  769. hello(int ctl)
  770. {
  771. char msg[1024];
  772. /* wait for hello from other side */
  773. if(ftprcode(ctl, msg, sizeof(msg)) != Success){
  774. werrstr("HELLO: %s", msg);
  775. return Server;
  776. }
  777. return 0;
  778. }
  779. int
  780. getdec(char *p, int n)
  781. {
  782. int x = 0;
  783. int i;
  784. for(i = 0; i < n; i++)
  785. x = x*10 + (*p++ - '0');
  786. return x;
  787. }
  788. int
  789. ftprestart(int ctl, int out, URL *u, Range *r, long mtime)
  790. {
  791. Tm tm;
  792. char msg[1024];
  793. long x, rmtime;
  794. ftpcmd(ctl, "MDTM %s", u->page);
  795. if(ftprcode(ctl, msg, sizeof(msg)) != Success){
  796. r->start = 0;
  797. return 0; /* need to do something */
  798. }
  799. /* decode modification time */
  800. if(strlen(msg) < 4 + 4 + 2 + 2 + 2 + 2 + 2){
  801. r->start = 0;
  802. return 0; /* need to do something */
  803. }
  804. memset(&tm, 0, sizeof(tm));
  805. tm.year = getdec(msg+4, 4) - 1900;
  806. tm.mon = getdec(msg+4+4, 2) - 1;
  807. tm.mday = getdec(msg+4+4+2, 2);
  808. tm.hour = getdec(msg+4+4+2+2, 2);
  809. tm.min = getdec(msg+4+4+2+2+2, 2);
  810. tm.sec = getdec(msg+4+4+2+2+2+2, 2);
  811. strcpy(tm.zone, "GMT");
  812. rmtime = tm2sec(&tm);
  813. if(rmtime > mtime)
  814. r->start = 0;
  815. /* get size */
  816. ftpcmd(ctl, "SIZE %s", u->page);
  817. if(ftprcode(ctl, msg, sizeof(msg)) == Success){
  818. x = atol(msg+4);
  819. if(r->start == x)
  820. return 1; /* we're up to date */
  821. r->end = x;
  822. }
  823. /* seek to restart point */
  824. if(r->start > 0){
  825. ftpcmd(ctl, "REST %lud", r->start);
  826. if(ftprcode(ctl, msg, sizeof(msg)) == Incomplete)
  827. seek(out, r->start, 0);
  828. else
  829. r->start = 0;
  830. }
  831. return 0; /* need to do something */
  832. }
  833. int
  834. logon(int ctl)
  835. {
  836. char msg[1024];
  837. /* login anonymous */
  838. ftpcmd(ctl, "USER anonymous");
  839. switch(ftprcode(ctl, msg, sizeof(msg))){
  840. case Success:
  841. return 0;
  842. case Incomplete:
  843. break; /* need password */
  844. default:
  845. werrstr("USER: %s", msg);
  846. return Server;
  847. }
  848. /* send user id as password */
  849. sprint(msg, "%s@closedmind.org", getuser());
  850. ftpcmd(ctl, "PASS %s", msg);
  851. if(ftprcode(ctl, msg, sizeof(msg)) != Success){
  852. werrstr("PASS: %s", msg);
  853. return Server;
  854. }
  855. return 0;
  856. }
  857. int
  858. xfertype(int ctl, char *t)
  859. {
  860. char msg[1024];
  861. ftpcmd(ctl, "TYPE %s", t);
  862. if(ftprcode(ctl, msg, sizeof(msg)) != Success){
  863. werrstr("TYPE %s: %s", t, msg);
  864. return Server;
  865. }
  866. return 0;
  867. }
  868. int
  869. passive(int ctl, URL *u)
  870. {
  871. char msg[1024];
  872. char ipaddr[32];
  873. char *f[6];
  874. char *p;
  875. int fd;
  876. int port;
  877. char aport[12];
  878. ftpcmd(ctl, "PASV");
  879. if(ftprcode(ctl, msg, sizeof(msg)) != Success)
  880. return Error;
  881. /* get address and port number from reply, this is AI */
  882. p = strchr(msg, '(');
  883. if(p == nil){
  884. for(p = msg+3; *p; p++)
  885. if(isdigit(*p))
  886. break;
  887. } else
  888. p++;
  889. if(getfields(p, f, 6, 0, ",)") < 6){
  890. werrstr("ftp protocol botch");
  891. return Server;
  892. }
  893. snprint(ipaddr, sizeof(ipaddr), "%s.%s.%s.%s",
  894. f[0], f[1], f[2], f[3]);
  895. port = ((atoi(f[4])&0xff)<<8) + (atoi(f[5])&0xff);
  896. sprint(aport, "%d", port);
  897. /* open data connection */
  898. fd = dial(netmkaddr(ipaddr, tcpdir, aport), 0, 0, 0);
  899. if(fd < 0){
  900. werrstr("passive mode failed: %r");
  901. return Error;
  902. }
  903. /* tell remote to send a file */
  904. ftpcmd(ctl, "RETR %s", u->page);
  905. if(ftprcode(ctl, msg, sizeof(msg)) != Extra){
  906. werrstr("RETR %s: %s", u->page, msg);
  907. return Error;
  908. }
  909. return fd;
  910. }
  911. int
  912. active(int ctl, URL *u)
  913. {
  914. char msg[1024];
  915. char dir[40], ldir[40];
  916. uchar ipaddr[4];
  917. uchar port[2];
  918. int lcfd, dfd, afd;
  919. /* announce a port for the call back */
  920. snprint(msg, sizeof(msg), "%s!*!0", tcpdir);
  921. afd = announce(msg, dir);
  922. if(afd < 0)
  923. return Error;
  924. /* get a local address/port of the annoucement */
  925. if(getaddrport(dir, ipaddr, port) < 0){
  926. close(afd);
  927. return Error;
  928. }
  929. /* tell remote side address and port*/
  930. ftpcmd(ctl, "PORT %d,%d,%d,%d,%d,%d", ipaddr[0], ipaddr[1], ipaddr[2],
  931. ipaddr[3], port[0], port[1]);
  932. if(ftprcode(ctl, msg, sizeof(msg)) != Success){
  933. close(afd);
  934. werrstr("active: %s", msg);
  935. return Error;
  936. }
  937. /* tell remote to send a file */
  938. ftpcmd(ctl, "RETR %s", u->page);
  939. if(ftprcode(ctl, msg, sizeof(msg)) != Extra){
  940. close(afd);
  941. werrstr("RETR: %s", msg);
  942. return Server;
  943. }
  944. /* wait for a connection */
  945. lcfd = listen(dir, ldir);
  946. if(lcfd < 0){
  947. close(afd);
  948. return Error;
  949. }
  950. dfd = accept(lcfd, ldir);
  951. if(dfd < 0){
  952. close(afd);
  953. close(lcfd);
  954. return Error;
  955. }
  956. close(afd);
  957. close(lcfd);
  958. return dfd;
  959. }
  960. int
  961. ftpxfer(int in, int out, Range *r)
  962. {
  963. char buf[1024];
  964. long vtime;
  965. int i, n;
  966. vtime = 0;
  967. for(n = 0;;n += i){
  968. i = read(in, buf, sizeof(buf));
  969. if(i == 0)
  970. break;
  971. if(i < 0)
  972. return Error;
  973. if(write(out, buf, i) != i)
  974. return Error;
  975. r->start += i;
  976. if(verbose && vtime != time(0)) {
  977. vtime = time(0);
  978. fprint(2, "%ld %ld\n", r->start, r->end);
  979. }
  980. }
  981. return n;
  982. }
  983. int
  984. terminateftp(int ctl, int rv)
  985. {
  986. close(ctl);
  987. return rv;
  988. }
  989. /*
  990. * case insensitive strcmp (why aren't these in libc?)
  991. */
  992. int
  993. cistrncmp(char *a, char *b, int n)
  994. {
  995. while(n-- > 0){
  996. if(tolower(*a++) != tolower(*b++))
  997. return -1;
  998. }
  999. return 0;
  1000. }
  1001. int
  1002. cistrcmp(char *a, char *b)
  1003. {
  1004. while(*a || *b)
  1005. if(tolower(*a++) != tolower(*b++))
  1006. return -1;
  1007. return 0;
  1008. }
  1009. /*
  1010. * buffered io
  1011. */
  1012. struct
  1013. {
  1014. char *rp;
  1015. char *wp;
  1016. char buf[4*1024];
  1017. } b;
  1018. void
  1019. initibuf(void)
  1020. {
  1021. b.rp = b.wp = b.buf;
  1022. }
  1023. /*
  1024. * read a possibly buffered line, strip off trailing while
  1025. */
  1026. int
  1027. readline(int fd, char *buf, int len)
  1028. {
  1029. int n;
  1030. char *p;
  1031. int eof = 0;
  1032. len--;
  1033. for(p = buf;;){
  1034. if(b.rp >= b.wp){
  1035. n = read(fd, b.wp, sizeof(b.buf)/2);
  1036. if(n < 0)
  1037. return -1;
  1038. if(n == 0){
  1039. eof = 1;
  1040. break;
  1041. }
  1042. b.wp += n;
  1043. }
  1044. n = *b.rp++;
  1045. if(len > 0){
  1046. *p++ = n;
  1047. len--;
  1048. }
  1049. if(n == '\n')
  1050. break;
  1051. }
  1052. /* drop trailing white */
  1053. for(;;){
  1054. if(p <= buf)
  1055. break;
  1056. n = *(p-1);
  1057. if(n != ' ' && n != '\t' && n != '\r' && n != '\n')
  1058. break;
  1059. p--;
  1060. }
  1061. *p = 0;
  1062. if(eof && p == buf)
  1063. return -1;
  1064. return p-buf;
  1065. }
  1066. void
  1067. unreadline(char *line)
  1068. {
  1069. int i, n;
  1070. i = strlen(line);
  1071. n = b.wp-b.rp;
  1072. memmove(&b.buf[i+1], b.rp, n);
  1073. memmove(b.buf, line, i);
  1074. b.buf[i] = '\n';
  1075. b.rp = b.buf;
  1076. b.wp = b.rp + i + 1 + n;
  1077. }
  1078. int
  1079. readibuf(int fd, char *buf, int len)
  1080. {
  1081. int n;
  1082. n = b.wp-b.rp;
  1083. if(n > 0){
  1084. if(n > len)
  1085. n = len;
  1086. memmove(buf, b.rp, n);
  1087. b.rp += n;
  1088. return n;
  1089. }
  1090. return read(fd, buf, len);
  1091. }
  1092. int
  1093. dfprint(int fd, char *fmt, ...)
  1094. {
  1095. char buf[4*1024];
  1096. va_list arg;
  1097. va_start(arg, fmt);
  1098. vseprint(buf, buf+sizeof(buf), fmt, arg);
  1099. va_end(arg);
  1100. if(debug)
  1101. fprint(2, "%d -> %s", fd, buf);
  1102. return fprint(fd, "%s", buf);
  1103. }
  1104. int
  1105. getaddrport(char *dir, uchar *ipaddr, uchar *port)
  1106. {
  1107. char buf[256];
  1108. int fd, i;
  1109. char *p;
  1110. snprint(buf, sizeof(buf), "%s/local", dir);
  1111. fd = open(buf, OREAD);
  1112. if(fd < 0)
  1113. return -1;
  1114. i = read(fd, buf, sizeof(buf)-1);
  1115. close(fd);
  1116. if(i <= 0)
  1117. return -1;
  1118. buf[i] = 0;
  1119. p = strchr(buf, '!');
  1120. if(p != nil)
  1121. *p++ = 0;
  1122. v4parseip(ipaddr, buf);
  1123. i = atoi(p);
  1124. port[0] = i>>8;
  1125. port[1] = i;
  1126. return 0;
  1127. }