hget.c 22 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186118711881189119011911192119311941195119611971198119912001201120212031204120512061207120812091210121112121213121412151216121712181219122012211222122312241225122612271228122912301231123212331234123512361237123812391240124112421243124412451246124712481249125012511252125312541255125612571258125912601261126212631264126512661267126812691270127112721273127412751276127712781279128012811282128312841285128612871288128912901291129212931294129512961297129812991300
  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. int headerprint;
  59. struct {
  60. char *name;
  61. int (*f)(URL*, Range*, int, long);
  62. } method[] = {
  63. [Http] { "http", dohttp },
  64. [Https] { "https", dohttp },
  65. [Ftp] { "ftp", doftp },
  66. [Other] { "_______", nil },
  67. };
  68. void
  69. usage(void)
  70. {
  71. fprint(2, "usage: %s [-hv] [-o outfile] [-p body] [-x netmtpt] url\n", argv0);
  72. exits("usage");
  73. }
  74. void
  75. main(int argc, char **argv)
  76. {
  77. URL u;
  78. Range r;
  79. int fd, errs, n;
  80. ulong mtime;
  81. Dir *d;
  82. char postbody[4096], *p, *e, *t;
  83. ofile = nil;
  84. p = postbody;
  85. e = p + sizeof(postbody);
  86. r.start = 0;
  87. r.end = -1;
  88. mtime = 0;
  89. memset(&u, 0, sizeof(u));
  90. ARGBEGIN {
  91. case 'o':
  92. ofile = ARGF();
  93. break;
  94. case 'd':
  95. debug = 1;
  96. break;
  97. case 'h':
  98. headerprint = 1;
  99. break;
  100. case 'v':
  101. verbose = 1;
  102. break;
  103. case 'x':
  104. net = ARGF();
  105. if(net == nil)
  106. usage();
  107. break;
  108. case 'p':
  109. t = ARGF();
  110. if(t == nil)
  111. usage();
  112. if(p != postbody)
  113. p = seprint(p, e, "&%s", t);
  114. else
  115. p = seprint(p, e, "%s", t);
  116. u.postbody = postbody;
  117. break;
  118. default:
  119. usage();
  120. } ARGEND;
  121. if(net != nil){
  122. if(strlen(net) > sizeof(tcpdir)-5)
  123. sysfatal("network mount point too long");
  124. snprint(tcpdir, sizeof(tcpdir), "%s/tcp", net);
  125. } else
  126. snprint(tcpdir, sizeof(tcpdir), "tcp");
  127. if(argc != 1)
  128. usage();
  129. fd = 1;
  130. if(ofile != nil){
  131. d = dirstat(ofile);
  132. if(d == nil){
  133. fd = create(ofile, OWRITE, 0664);
  134. if(fd < 0)
  135. sysfatal("creating %s: %r", ofile);
  136. } else {
  137. fd = open(ofile, OWRITE);
  138. if(fd < 0)
  139. sysfatal("can't open %s: %r", ofile);
  140. r.start = d->length;
  141. mtime = d->mtime;
  142. free(d);
  143. }
  144. }
  145. errs = 0;
  146. if(crackurl(&u, argv[0]) < 0)
  147. sysfatal("%r");
  148. for(;;){
  149. /* transfer data */
  150. werrstr("");
  151. seek(fd, 0, 0);
  152. n = (*method[u.method].f)(&u, &r, fd, mtime);
  153. switch(n){
  154. case Eof:
  155. exits(0);
  156. break;
  157. case Error:
  158. if(errs++ < 10)
  159. continue;
  160. sysfatal("too many errors with no progress %r");
  161. break;
  162. case Server:
  163. sysfatal("server returned: %r");
  164. break;
  165. }
  166. /* forward progress */
  167. errs = 0;
  168. r.start += n;
  169. if(r.start >= r.end)
  170. break;
  171. }
  172. exits(0);
  173. }
  174. int
  175. crackurl(URL *u, char *s)
  176. {
  177. char *p;
  178. int i;
  179. if(u->host != nil){
  180. free(u->host);
  181. u->host = nil;
  182. }
  183. if(u->page != nil){
  184. free(u->page);
  185. u->page = nil;
  186. }
  187. /* get type */
  188. u->method = Other;
  189. for(p = s; *p; p++){
  190. if(*p == '/'){
  191. u->method = Http;
  192. p = s;
  193. break;
  194. }
  195. if(*p == ':' && *(p+1)=='/' && *(p+2)=='/'){
  196. *p = 0;
  197. p += 3;
  198. for(i = 0; i < nelem(method); i++){
  199. if(cistrcmp(s, method[i].name) == 0){
  200. u->method = i;
  201. break;
  202. }
  203. }
  204. break;
  205. }
  206. }
  207. if(u->method == Other){
  208. werrstr("unsupported URL type %s", s);
  209. return -1;
  210. }
  211. /* get system */
  212. s = p;
  213. p = strchr(s, '/');
  214. if(p == nil){
  215. u->host = strdup(s);
  216. u->page = strdup("/");
  217. } else {
  218. u->page = strdup(p);
  219. *p = 0;
  220. u->host = strdup(s);
  221. *p = '/';
  222. }
  223. if(p = strchr(u->host, ':')) {
  224. *p++ = 0;
  225. u->port = p;
  226. } else
  227. u->port = method[u->method].name;
  228. if(*(u->host) == 0){
  229. werrstr("bad url, null host");
  230. return -1;
  231. }
  232. return 0;
  233. }
  234. char *day[] = {
  235. "Sun", "Mon", "Tue", "Wed", "Thu", "Fri", "Sat"
  236. };
  237. char *month[] = {
  238. "Jan", "Feb", "Mar", "Apr", "May", "Jun", "Jul", "Aug", "Sep", "Oct", "Nov", "Dec"
  239. };
  240. struct
  241. {
  242. int fd;
  243. long mtime;
  244. } note;
  245. void
  246. catch(void*, char*)
  247. {
  248. Dir d;
  249. nulldir(&d);
  250. d.mtime = note.mtime;
  251. if(dirfwstat(note.fd, &d) < 0)
  252. sysfatal("catch: can't dirfwstat: %r");
  253. noted(NDFLT);
  254. }
  255. int
  256. dohttp(URL *u, Range *r, int out, long mtime)
  257. {
  258. int fd, cfd;
  259. int redirect, loop;
  260. int n, rv, code;
  261. long tot, vtime;
  262. Tm *tm;
  263. char buf[1024];
  264. char err[ERRMAX];
  265. /* always move back to a previous 512 byte bound because some
  266. * servers can't seem to deal with requests that start at the
  267. * end of the file
  268. */
  269. if(r->start)
  270. r->start = ((r->start-1)/512)*512;
  271. /* loop for redirects, requires reading both response code and headers */
  272. fd = -1;
  273. for(loop = 0; loop < 32; loop++){
  274. fd = dial(netmkaddr(u->host, tcpdir, u->port), 0, 0, 0);
  275. if(fd < 0)
  276. return Error;
  277. if(u->method == Https){
  278. int tfd;
  279. TLSconn conn;
  280. memset(&conn, 0, sizeof conn);
  281. tfd = tlsClient(fd, &conn);
  282. if(tfd < 0){
  283. fprint(2, "tlsClient: %r\n");
  284. close(fd);
  285. return Error;
  286. }
  287. /* BUG: check cert here? */
  288. if(conn.cert)
  289. free(conn.cert);
  290. close(fd);
  291. fd = tfd;
  292. }
  293. /* write request, use range if not start of file */
  294. if(u->postbody == nil){
  295. dfprint(fd, "GET %s HTTP/1.0\r\n"
  296. "Host: %s\r\n"
  297. "User-agent: Plan9/hget\r\n"
  298. "Cache-Control: no-cache\r\n"
  299. "Pragma: no-cache\r\n",
  300. u->page, u->host);
  301. } else {
  302. dfprint(fd, "POST %s HTTP/1.0\r\n"
  303. "Host: %s\r\n"
  304. "Content-type: application/x-www-form-urlencoded\r\n"
  305. "Content-length: %d\r\n"
  306. "User-agent: Plan9/hget\r\n"
  307. "\r\n",
  308. u->page, u->host, strlen(u->postbody));
  309. dfprint(fd, "%s", u->postbody);
  310. }
  311. if(r->start != 0){
  312. dfprint(fd, "Range: bytes=%d-\n", r->start);
  313. if(u->etag != nil){
  314. dfprint(fd, "If-range: %s\n", u->etag);
  315. } else {
  316. tm = gmtime(mtime);
  317. dfprint(fd, "If-range: %s, %d %s %d %2d:%2.2d:%2.2d GMT\n",
  318. day[tm->wday], tm->mday, month[tm->mon],
  319. tm->year+1900, tm->hour, tm->min, tm->sec);
  320. }
  321. }
  322. if((cfd = open("/mnt/webcookies/http", ORDWR)) >= 0){
  323. if(fprint(cfd, "http://%s%s", u->host, u->page) > 0){
  324. while((n = read(cfd, buf, sizeof buf)) > 0){
  325. if(debug)
  326. write(2, buf, n);
  327. write(fd, buf, n);
  328. }
  329. }else{
  330. close(cfd);
  331. cfd = -1;
  332. }
  333. }
  334. dfprint(fd, "\r\n", u->host);
  335. redirect = 0;
  336. initibuf();
  337. code = httprcode(fd);
  338. switch(code){
  339. case Error: /* connection timed out */
  340. case Eof:
  341. close(fd);
  342. close(cfd);
  343. return code;
  344. case 200: /* OK */
  345. case 201: /* Created */
  346. case 202: /* Accepted */
  347. if(ofile == nil && r->start != 0)
  348. sysfatal("page changed underfoot");
  349. break;
  350. case 204: /* No Content */
  351. sysfatal("No Content");
  352. case 206: /* Partial Content */
  353. seek(out, r->start, 0);
  354. break;
  355. case 301: /* Moved Permanently */
  356. case 302: /* Moved Temporarily */
  357. redirect = 1;
  358. u->postbody = nil;
  359. break;
  360. case 304: /* Not Modified */
  361. break;
  362. case 400: /* Bad Request */
  363. sysfatal("Bad Request");
  364. case 401: /* Unauthorized */
  365. case 402: /* ??? */
  366. sysfatal("Unauthorized");
  367. case 403: /* Forbidden */
  368. sysfatal("Forbidden by server");
  369. case 404: /* Not Found */
  370. sysfatal("Not found on server");
  371. case 500: /* Internal server error */
  372. sysfatal("Server choked");
  373. case 501: /* Not implemented */
  374. sysfatal("Server can't do it!");
  375. case 502: /* Bad gateway */
  376. sysfatal("Bad gateway");
  377. case 503: /* Service unavailable */
  378. sysfatal("Service unavailable");
  379. default:
  380. sysfatal("Unknown response code %d", code);
  381. }
  382. if(u->redirect != nil){
  383. free(u->redirect);
  384. u->redirect = nil;
  385. }
  386. rv = httpheaders(fd, cfd, u, r);
  387. close(cfd);
  388. if(rv != 0){
  389. close(fd);
  390. return rv;
  391. }
  392. if(!redirect)
  393. break;
  394. if(u->redirect == nil)
  395. sysfatal("redirect: no URL");
  396. if(crackurl(u, u->redirect) < 0)
  397. sysfatal("redirect: %r");
  398. }
  399. /* transfer whatever you get */
  400. if(ofile != nil && u->mtime != 0){
  401. note.fd = out;
  402. note.mtime = u->mtime;
  403. notify(catch);
  404. }
  405. tot = 0;
  406. vtime = 0;
  407. for(;;){
  408. n = readibuf(fd, buf, sizeof(buf));
  409. if(n <= 0)
  410. break;
  411. if(write(out, buf, n) != n)
  412. break;
  413. tot += n;
  414. if(verbose && vtime != time(0)) {
  415. vtime = time(0);
  416. fprint(2, "%ld %ld\n", r->start+tot, r->end);
  417. }
  418. }
  419. notify(nil);
  420. close(fd);
  421. if(ofile != nil && u->mtime != 0){
  422. Dir d;
  423. rerrstr(err, sizeof err);
  424. nulldir(&d);
  425. d.mtime = u->mtime;
  426. if(dirfwstat(out, &d) < 0)
  427. fprint(2, "couldn't set mtime: %r\n");
  428. errstr(err, sizeof err);
  429. }
  430. return tot;
  431. }
  432. /* get the http response code */
  433. int
  434. httprcode(int fd)
  435. {
  436. int n;
  437. char *p;
  438. char buf[256];
  439. n = readline(fd, buf, sizeof(buf)-1);
  440. if(n <= 0)
  441. return n;
  442. if(debug)
  443. fprint(2, "%d <- %s\n", fd, buf);
  444. p = strchr(buf, ' ');
  445. if(strncmp(buf, "HTTP/", 5) != 0 || p == nil){
  446. werrstr("bad response from server");
  447. return -1;
  448. }
  449. buf[n] = 0;
  450. return atoi(p+1);
  451. }
  452. /* read in and crack the http headers, update u and r */
  453. void hhetag(char*, URL*, Range*);
  454. void hhmtime(char*, URL*, Range*);
  455. void hhclen(char*, URL*, Range*);
  456. void hhcrange(char*, URL*, Range*);
  457. void hhuri(char*, URL*, Range*);
  458. void hhlocation(char*, URL*, Range*);
  459. struct {
  460. char *name;
  461. void (*f)(char*, URL*, Range*);
  462. } headers[] = {
  463. { "etag:", hhetag },
  464. { "last-modified:", hhmtime },
  465. { "content-length:", hhclen },
  466. { "content-range:", hhcrange },
  467. { "uri:", hhuri },
  468. { "location:", hhlocation },
  469. };
  470. int
  471. httpheaders(int fd, int cfd, URL *u, Range *r)
  472. {
  473. char buf[2048];
  474. char *p;
  475. int i, n;
  476. for(;;){
  477. n = getheader(fd, buf, sizeof(buf));
  478. if(n <= 0)
  479. break;
  480. if(cfd >= 0)
  481. fprint(cfd, "%s\n", buf);
  482. for(i = 0; i < nelem(headers); i++){
  483. n = strlen(headers[i].name);
  484. if(cistrncmp(buf, headers[i].name, n) == 0){
  485. /* skip field name and leading white */
  486. p = buf + n;
  487. while(*p == ' ' || *p == '\t')
  488. p++;
  489. (*headers[i].f)(p, u, r);
  490. break;
  491. }
  492. }
  493. }
  494. return n;
  495. }
  496. /*
  497. * read a single mime header, collect continuations.
  498. *
  499. * this routine assumes that there is a blank line twixt
  500. * the header and the message body, otherwise bytes will
  501. * be lost.
  502. */
  503. int
  504. getheader(int fd, char *buf, int n)
  505. {
  506. char *p, *e;
  507. int i;
  508. n--;
  509. p = buf;
  510. for(e = p + n; ; p += i){
  511. i = readline(fd, p, e-p);
  512. if(i < 0)
  513. return i;
  514. if(p == buf){
  515. /* first line */
  516. if(strchr(buf, ':') == nil)
  517. break; /* end of headers */
  518. } else {
  519. /* continuation line */
  520. if(*p != ' ' && *p != '\t'){
  521. unreadline(p);
  522. *p = 0;
  523. break; /* end of this header */
  524. }
  525. }
  526. }
  527. if(headerprint)
  528. print("%s\n", buf);
  529. if(debug)
  530. fprint(2, "%d <- %s\n", fd, buf);
  531. return p-buf;
  532. }
  533. void
  534. hhetag(char *p, URL *u, Range*)
  535. {
  536. if(u->etag != nil){
  537. if(strcmp(u->etag, p) != 0)
  538. sysfatal("file changed underfoot");
  539. } else
  540. u->etag = strdup(p);
  541. }
  542. char* monthchars = "janfebmaraprmayjunjulaugsepoctnovdec";
  543. void
  544. hhmtime(char *p, URL *u, Range*)
  545. {
  546. char *month, *day, *yr, *hms;
  547. char *fields[6];
  548. Tm tm, now;
  549. int i;
  550. i = getfields(p, fields, 6, 1, " \t");
  551. if(i < 5)
  552. return;
  553. day = fields[1];
  554. month = fields[2];
  555. yr = fields[3];
  556. hms = fields[4];
  557. /* default time */
  558. now = *gmtime(time(0));
  559. tm = now;
  560. tm.yday = 0;
  561. /* convert ascii month to a number twixt 1 and 12 */
  562. if(*month >= '0' && *month <= '9'){
  563. tm.mon = atoi(month) - 1;
  564. if(tm.mon < 0 || tm.mon > 11)
  565. tm.mon = 5;
  566. } else {
  567. for(p = month; *p; p++)
  568. *p = tolower(*p);
  569. for(i = 0; i < 12; i++)
  570. if(strncmp(&monthchars[i*3], month, 3) == 0){
  571. tm.mon = i;
  572. break;
  573. }
  574. }
  575. tm.mday = atoi(day);
  576. if(hms) {
  577. tm.hour = strtoul(hms, &p, 10);
  578. if(*p == ':') {
  579. p++;
  580. tm.min = strtoul(p, &p, 10);
  581. if(*p == ':') {
  582. p++;
  583. tm.sec = strtoul(p, &p, 10);
  584. }
  585. }
  586. if(tolower(*p) == 'p')
  587. tm.hour += 12;
  588. }
  589. if(yr) {
  590. tm.year = atoi(yr);
  591. if(tm.year >= 1900)
  592. tm.year -= 1900;
  593. } else {
  594. if(tm.mon > now.mon || (tm.mon == now.mon && tm.mday > now.mday+1))
  595. tm.year--;
  596. }
  597. strcpy(tm.zone, "GMT");
  598. /* convert to epoch seconds */
  599. u->mtime = tm2sec(&tm);
  600. }
  601. void
  602. hhclen(char *p, URL*, Range *r)
  603. {
  604. r->end = atoi(p);
  605. }
  606. void
  607. hhcrange(char *p, URL*, Range *r)
  608. {
  609. char *x;
  610. vlong l;
  611. l = 0;
  612. x = strchr(p, '/');
  613. if(x)
  614. l = atoll(x+1);
  615. if(l == 0)
  616. x = strchr(p, '-');
  617. if(x)
  618. l = atoll(x+1);
  619. if(l)
  620. r->end = l;
  621. }
  622. void
  623. hhuri(char *p, URL *u, Range*)
  624. {
  625. if(*p != '<')
  626. return;
  627. u->redirect = strdup(p+1);
  628. p = strchr(u->redirect, '>');
  629. if(p != nil)
  630. *p = 0;
  631. }
  632. void
  633. hhlocation(char *p, URL *u, Range*)
  634. {
  635. u->redirect = strdup(p);
  636. }
  637. enum
  638. {
  639. /* ftp return codes */
  640. Extra= 1,
  641. Success= 2,
  642. Incomplete= 3,
  643. TempFail= 4,
  644. PermFail= 5,
  645. Nnetdir= 64, /* max length of network directory paths */
  646. Ndialstr= 64, /* max length of dial strings */
  647. };
  648. int ftpcmd(int, char*, ...);
  649. int ftprcode(int, char*, int);
  650. int hello(int);
  651. int logon(int);
  652. int xfertype(int, char*);
  653. int passive(int, URL*);
  654. int active(int, URL*);
  655. int ftpxfer(int, int, Range*);
  656. int terminateftp(int, int);
  657. int getaddrport(char*, uchar*, uchar*);
  658. int ftprestart(int, int, URL*, Range*, long);
  659. int
  660. doftp(URL *u, Range *r, int out, long mtime)
  661. {
  662. int pid, ctl, data, rv;
  663. Waitmsg *w;
  664. char msg[64];
  665. char conndir[NETPATHLEN];
  666. char *p;
  667. ctl = dial(netmkaddr(u->host, tcpdir, u->port), 0, conndir, 0);
  668. if(ctl < 0)
  669. return Error;
  670. if(net == nil){
  671. p = strrchr(conndir, '/');
  672. *p = 0;
  673. snprint(tcpdir, sizeof(tcpdir), conndir);
  674. }
  675. initibuf();
  676. rv = hello(ctl);
  677. if(rv < 0)
  678. return terminateftp(ctl, rv);
  679. rv = logon(ctl);
  680. if(rv < 0)
  681. return terminateftp(ctl, rv);
  682. rv = xfertype(ctl, "I");
  683. if(rv < 0)
  684. return terminateftp(ctl, rv);
  685. /* if file is up to date and the right size, stop */
  686. if(ftprestart(ctl, out, u, r, mtime) > 0){
  687. close(ctl);
  688. return Eof;
  689. }
  690. /* first try passive mode, then active */
  691. data = passive(ctl, u);
  692. if(data < 0){
  693. data = active(ctl, u);
  694. if(data < 0)
  695. return Error;
  696. }
  697. /* fork */
  698. switch(pid = rfork(RFPROC|RFFDG|RFMEM)){
  699. case -1:
  700. close(data);
  701. return terminateftp(ctl, Error);
  702. case 0:
  703. ftpxfer(data, out, r);
  704. close(data);
  705. _exits(0);
  706. default:
  707. close(data);
  708. break;
  709. }
  710. /* wait for reply message */
  711. rv = ftprcode(ctl, msg, sizeof(msg));
  712. close(ctl);
  713. /* wait for process to terminate */
  714. w = nil;
  715. for(;;){
  716. free(w);
  717. w = wait();
  718. if(w == nil)
  719. return Error;
  720. if(w->pid == pid){
  721. if(w->msg[0] == 0){
  722. free(w);
  723. break;
  724. }
  725. werrstr("xfer: %s", w->msg);
  726. free(w);
  727. return Error;
  728. }
  729. }
  730. switch(rv){
  731. case Success:
  732. return Eof;
  733. case TempFail:
  734. return Server;
  735. default:
  736. return Error;
  737. }
  738. }
  739. int
  740. ftpcmd(int ctl, char *fmt, ...)
  741. {
  742. va_list arg;
  743. char buf[2*1024], *s;
  744. va_start(arg, fmt);
  745. s = vseprint(buf, buf + (sizeof(buf)-4) / sizeof(*buf), fmt, arg);
  746. va_end(arg);
  747. if(debug)
  748. fprint(2, "%d -> %s\n", ctl, buf);
  749. *s++ = '\r';
  750. *s++ = '\n';
  751. if(write(ctl, buf, s - buf) != s - buf)
  752. return -1;
  753. return 0;
  754. }
  755. int
  756. ftprcode(int ctl, char *msg, int len)
  757. {
  758. int rv;
  759. int i;
  760. char *p;
  761. len--; /* room for terminating null */
  762. for(;;){
  763. *msg = 0;
  764. i = readline(ctl, msg, len);
  765. if(i < 0)
  766. break;
  767. if(debug)
  768. fprint(2, "%d <- %s\n", ctl, msg);
  769. /* stop if not a continuation */
  770. rv = strtol(msg, &p, 10);
  771. if(rv >= 100 && rv < 600 && p==msg+3 && *p == ' ')
  772. return rv/100;
  773. }
  774. *msg = 0;
  775. return -1;
  776. }
  777. int
  778. hello(int ctl)
  779. {
  780. char msg[1024];
  781. /* wait for hello from other side */
  782. if(ftprcode(ctl, msg, sizeof(msg)) != Success){
  783. werrstr("HELLO: %s", msg);
  784. return Server;
  785. }
  786. return 0;
  787. }
  788. int
  789. getdec(char *p, int n)
  790. {
  791. int x = 0;
  792. int i;
  793. for(i = 0; i < n; i++)
  794. x = x*10 + (*p++ - '0');
  795. return x;
  796. }
  797. int
  798. ftprestart(int ctl, int out, URL *u, Range *r, long mtime)
  799. {
  800. Tm tm;
  801. char msg[1024];
  802. long x, rmtime;
  803. ftpcmd(ctl, "MDTM %s", u->page);
  804. if(ftprcode(ctl, msg, sizeof(msg)) != Success){
  805. r->start = 0;
  806. return 0; /* need to do something */
  807. }
  808. /* decode modification time */
  809. if(strlen(msg) < 4 + 4 + 2 + 2 + 2 + 2 + 2){
  810. r->start = 0;
  811. return 0; /* need to do something */
  812. }
  813. memset(&tm, 0, sizeof(tm));
  814. tm.year = getdec(msg+4, 4) - 1900;
  815. tm.mon = getdec(msg+4+4, 2) - 1;
  816. tm.mday = getdec(msg+4+4+2, 2);
  817. tm.hour = getdec(msg+4+4+2+2, 2);
  818. tm.min = getdec(msg+4+4+2+2+2, 2);
  819. tm.sec = getdec(msg+4+4+2+2+2+2, 2);
  820. strcpy(tm.zone, "GMT");
  821. rmtime = tm2sec(&tm);
  822. if(rmtime > mtime)
  823. r->start = 0;
  824. /* get size */
  825. ftpcmd(ctl, "SIZE %s", u->page);
  826. if(ftprcode(ctl, msg, sizeof(msg)) == Success){
  827. x = atol(msg+4);
  828. if(r->start == x)
  829. return 1; /* we're up to date */
  830. r->end = x;
  831. }
  832. /* seek to restart point */
  833. if(r->start > 0){
  834. ftpcmd(ctl, "REST %lud", r->start);
  835. if(ftprcode(ctl, msg, sizeof(msg)) == Incomplete)
  836. seek(out, r->start, 0);
  837. else
  838. r->start = 0;
  839. }
  840. return 0; /* need to do something */
  841. }
  842. int
  843. logon(int ctl)
  844. {
  845. char msg[1024];
  846. /* login anonymous */
  847. ftpcmd(ctl, "USER anonymous");
  848. switch(ftprcode(ctl, msg, sizeof(msg))){
  849. case Success:
  850. return 0;
  851. case Incomplete:
  852. break; /* need password */
  853. default:
  854. werrstr("USER: %s", msg);
  855. return Server;
  856. }
  857. /* send user id as password */
  858. sprint(msg, "%s@closedmind.org", getuser());
  859. ftpcmd(ctl, "PASS %s", msg);
  860. if(ftprcode(ctl, msg, sizeof(msg)) != Success){
  861. werrstr("PASS: %s", msg);
  862. return Server;
  863. }
  864. return 0;
  865. }
  866. int
  867. xfertype(int ctl, char *t)
  868. {
  869. char msg[1024];
  870. ftpcmd(ctl, "TYPE %s", t);
  871. if(ftprcode(ctl, msg, sizeof(msg)) != Success){
  872. werrstr("TYPE %s: %s", t, msg);
  873. return Server;
  874. }
  875. return 0;
  876. }
  877. int
  878. passive(int ctl, URL *u)
  879. {
  880. char msg[1024];
  881. char ipaddr[32];
  882. char *f[6];
  883. char *p;
  884. int fd;
  885. int port;
  886. char aport[12];
  887. ftpcmd(ctl, "PASV");
  888. if(ftprcode(ctl, msg, sizeof(msg)) != Success)
  889. return Error;
  890. /* get address and port number from reply, this is AI */
  891. p = strchr(msg, '(');
  892. if(p == nil){
  893. for(p = msg+3; *p; p++)
  894. if(isdigit(*p))
  895. break;
  896. } else
  897. p++;
  898. if(getfields(p, f, 6, 0, ",)") < 6){
  899. werrstr("ftp protocol botch");
  900. return Server;
  901. }
  902. snprint(ipaddr, sizeof(ipaddr), "%s.%s.%s.%s",
  903. f[0], f[1], f[2], f[3]);
  904. port = ((atoi(f[4])&0xff)<<8) + (atoi(f[5])&0xff);
  905. sprint(aport, "%d", port);
  906. /* open data connection */
  907. fd = dial(netmkaddr(ipaddr, tcpdir, aport), 0, 0, 0);
  908. if(fd < 0){
  909. werrstr("passive mode failed: %r");
  910. return Error;
  911. }
  912. /* tell remote to send a file */
  913. ftpcmd(ctl, "RETR %s", u->page);
  914. if(ftprcode(ctl, msg, sizeof(msg)) != Extra){
  915. werrstr("RETR %s: %s", u->page, msg);
  916. return Error;
  917. }
  918. return fd;
  919. }
  920. int
  921. active(int ctl, URL *u)
  922. {
  923. char msg[1024];
  924. char dir[40], ldir[40];
  925. uchar ipaddr[4];
  926. uchar port[2];
  927. int lcfd, dfd, afd;
  928. /* announce a port for the call back */
  929. snprint(msg, sizeof(msg), "%s!*!0", tcpdir);
  930. afd = announce(msg, dir);
  931. if(afd < 0)
  932. return Error;
  933. /* get a local address/port of the annoucement */
  934. if(getaddrport(dir, ipaddr, port) < 0){
  935. close(afd);
  936. return Error;
  937. }
  938. /* tell remote side address and port*/
  939. ftpcmd(ctl, "PORT %d,%d,%d,%d,%d,%d", ipaddr[0], ipaddr[1], ipaddr[2],
  940. ipaddr[3], port[0], port[1]);
  941. if(ftprcode(ctl, msg, sizeof(msg)) != Success){
  942. close(afd);
  943. werrstr("active: %s", msg);
  944. return Error;
  945. }
  946. /* tell remote to send a file */
  947. ftpcmd(ctl, "RETR %s", u->page);
  948. if(ftprcode(ctl, msg, sizeof(msg)) != Extra){
  949. close(afd);
  950. werrstr("RETR: %s", msg);
  951. return Server;
  952. }
  953. /* wait for a connection */
  954. lcfd = listen(dir, ldir);
  955. if(lcfd < 0){
  956. close(afd);
  957. return Error;
  958. }
  959. dfd = accept(lcfd, ldir);
  960. if(dfd < 0){
  961. close(afd);
  962. close(lcfd);
  963. return Error;
  964. }
  965. close(afd);
  966. close(lcfd);
  967. return dfd;
  968. }
  969. int
  970. ftpxfer(int in, int out, Range *r)
  971. {
  972. char buf[1024];
  973. long vtime;
  974. int i, n;
  975. vtime = 0;
  976. for(n = 0;;n += i){
  977. i = read(in, buf, sizeof(buf));
  978. if(i == 0)
  979. break;
  980. if(i < 0)
  981. return Error;
  982. if(write(out, buf, i) != i)
  983. return Error;
  984. r->start += i;
  985. if(verbose && vtime != time(0)) {
  986. vtime = time(0);
  987. fprint(2, "%ld %ld\n", r->start, r->end);
  988. }
  989. }
  990. return n;
  991. }
  992. int
  993. terminateftp(int ctl, int rv)
  994. {
  995. close(ctl);
  996. return rv;
  997. }
  998. /*
  999. * case insensitive strcmp (why aren't these in libc?)
  1000. */
  1001. int
  1002. cistrncmp(char *a, char *b, int n)
  1003. {
  1004. while(n-- > 0){
  1005. if(tolower(*a++) != tolower(*b++))
  1006. return -1;
  1007. }
  1008. return 0;
  1009. }
  1010. int
  1011. cistrcmp(char *a, char *b)
  1012. {
  1013. while(*a || *b)
  1014. if(tolower(*a++) != tolower(*b++))
  1015. return -1;
  1016. return 0;
  1017. }
  1018. /*
  1019. * buffered io
  1020. */
  1021. struct
  1022. {
  1023. char *rp;
  1024. char *wp;
  1025. char buf[4*1024];
  1026. } b;
  1027. void
  1028. initibuf(void)
  1029. {
  1030. b.rp = b.wp = b.buf;
  1031. }
  1032. /*
  1033. * read a possibly buffered line, strip off trailing while
  1034. */
  1035. int
  1036. readline(int fd, char *buf, int len)
  1037. {
  1038. int n;
  1039. char *p;
  1040. int eof = 0;
  1041. len--;
  1042. for(p = buf;;){
  1043. if(b.rp >= b.wp){
  1044. n = read(fd, b.wp, sizeof(b.buf)/2);
  1045. if(n < 0)
  1046. return -1;
  1047. if(n == 0){
  1048. eof = 1;
  1049. break;
  1050. }
  1051. b.wp += n;
  1052. }
  1053. n = *b.rp++;
  1054. if(len > 0){
  1055. *p++ = n;
  1056. len--;
  1057. }
  1058. if(n == '\n')
  1059. break;
  1060. }
  1061. /* drop trailing white */
  1062. for(;;){
  1063. if(p <= buf)
  1064. break;
  1065. n = *(p-1);
  1066. if(n != ' ' && n != '\t' && n != '\r' && n != '\n')
  1067. break;
  1068. p--;
  1069. }
  1070. *p = 0;
  1071. if(eof && p == buf)
  1072. return -1;
  1073. return p-buf;
  1074. }
  1075. void
  1076. unreadline(char *line)
  1077. {
  1078. int i, n;
  1079. i = strlen(line);
  1080. n = b.wp-b.rp;
  1081. memmove(&b.buf[i+1], b.rp, n);
  1082. memmove(b.buf, line, i);
  1083. b.buf[i] = '\n';
  1084. b.rp = b.buf;
  1085. b.wp = b.rp + i + 1 + n;
  1086. }
  1087. int
  1088. readibuf(int fd, char *buf, int len)
  1089. {
  1090. int n;
  1091. n = b.wp-b.rp;
  1092. if(n > 0){
  1093. if(n > len)
  1094. n = len;
  1095. memmove(buf, b.rp, n);
  1096. b.rp += n;
  1097. return n;
  1098. }
  1099. return read(fd, buf, len);
  1100. }
  1101. int
  1102. dfprint(int fd, char *fmt, ...)
  1103. {
  1104. char buf[4*1024];
  1105. va_list arg;
  1106. va_start(arg, fmt);
  1107. vseprint(buf, buf+sizeof(buf), fmt, arg);
  1108. va_end(arg);
  1109. if(debug)
  1110. fprint(2, "%d -> %s", fd, buf);
  1111. return fprint(fd, "%s", buf);
  1112. }
  1113. int
  1114. getaddrport(char *dir, uchar *ipaddr, uchar *port)
  1115. {
  1116. char buf[256];
  1117. int fd, i;
  1118. char *p;
  1119. snprint(buf, sizeof(buf), "%s/local", dir);
  1120. fd = open(buf, OREAD);
  1121. if(fd < 0)
  1122. return -1;
  1123. i = read(fd, buf, sizeof(buf)-1);
  1124. close(fd);
  1125. if(i <= 0)
  1126. return -1;
  1127. buf[i] = 0;
  1128. p = strchr(buf, '!');
  1129. if(p != nil)
  1130. *p++ = 0;
  1131. v4parseip(ipaddr, buf);
  1132. i = atoi(p);
  1133. port[0] = i>>8;
  1134. port[1] = i;
  1135. return 0;
  1136. }