hget.c 24 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186118711881189119011911192119311941195119611971198119912001201120212031204120512061207120812091210121112121213121412151216121712181219122012211222122312241225122612271228122912301231123212331234123512361237123812391240124112421243124412451246124712481249125012511252125312541255125612571258125912601261126212631264126512661267126812691270127112721273127412751276127712781279128012811282128312841285128612871288128912901291129212931294129512961297129812991300130113021303130413051306130713081309131013111312131313141315131613171318131913201321132213231324132513261327132813291330133113321333133413351336133713381339134013411342134313441345134613471348134913501351135213531354135513561357135813591360136113621363136413651366136713681369137013711372137313741375137613771378137913801381138213831384138513861387138813891390139113921393139413951396139713981399140014011402140314041405140614071408140914101411141214131414141514161417141814191420142114221423
  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. typedef struct Out Out;
  26. struct Out
  27. {
  28. int fd;
  29. int offset; /* notional current offset in output */
  30. int written; /* number of bytes successfully transferred to output */
  31. DigestState *curr; /* digest state up to offset (if known) */
  32. DigestState *hiwat; /* digest state of all bytes written */
  33. };
  34. enum
  35. {
  36. Http,
  37. Https,
  38. Ftp,
  39. Other
  40. };
  41. enum
  42. {
  43. Eof = 0,
  44. Error = -1,
  45. Server = -2,
  46. Changed = -3,
  47. };
  48. int debug;
  49. char *ofile;
  50. int doftp(URL*, URL*, Range*, Out*, long);
  51. int dohttp(URL*, URL*, Range*, Out*, long);
  52. int crackurl(URL*, char*);
  53. Range* crackrange(char*);
  54. int getheader(int, char*, int);
  55. int httpheaders(int, int, URL*, Range*);
  56. int httprcode(int);
  57. int cistrncmp(char*, char*, int);
  58. int cistrcmp(char*, char*);
  59. void initibuf(void);
  60. int readline(int, char*, int);
  61. int readibuf(int, char*, int);
  62. int dfprint(int, char*, ...);
  63. void unreadline(char*);
  64. int output(Out*, char*, int);
  65. void setoffset(Out*, int);
  66. int verbose;
  67. char *net;
  68. char tcpdir[NETPATHLEN];
  69. int headerprint;
  70. struct {
  71. char *name;
  72. int (*f)(URL*, URL*, Range*, Out*, long);
  73. } method[] = {
  74. [Http] { "http", dohttp },
  75. [Https] { "https", dohttp },
  76. [Ftp] { "ftp", doftp },
  77. [Other] { "_______", nil },
  78. };
  79. void
  80. usage(void)
  81. {
  82. fprint(2, "usage: %s [-hv] [-o outfile] [-p body] [-x netmtpt] url\n", argv0);
  83. exits("usage");
  84. }
  85. void
  86. main(int argc, char **argv)
  87. {
  88. URL u;
  89. Range r;
  90. int errs, n;
  91. ulong mtime;
  92. Dir *d;
  93. char postbody[4096], *p, *e, *t, *hpx;
  94. URL px; // Proxy
  95. Out out;
  96. ofile = nil;
  97. p = postbody;
  98. e = p + sizeof(postbody);
  99. r.start = 0;
  100. r.end = -1;
  101. mtime = 0;
  102. memset(&u, 0, sizeof(u));
  103. memset(&px, 0, sizeof(px));
  104. hpx = getenv("httpproxy");
  105. ARGBEGIN {
  106. case 'o':
  107. ofile = ARGF();
  108. break;
  109. case 'd':
  110. debug = 1;
  111. break;
  112. case 'h':
  113. headerprint = 1;
  114. break;
  115. case 'v':
  116. verbose = 1;
  117. break;
  118. case 'x':
  119. net = ARGF();
  120. if(net == nil)
  121. usage();
  122. break;
  123. case 'p':
  124. t = ARGF();
  125. if(t == nil)
  126. usage();
  127. if(p != postbody)
  128. p = seprint(p, e, "&%s", t);
  129. else
  130. p = seprint(p, e, "%s", t);
  131. u.postbody = postbody;
  132. break;
  133. default:
  134. usage();
  135. } ARGEND;
  136. if(net != nil){
  137. if(strlen(net) > sizeof(tcpdir)-5)
  138. sysfatal("network mount point too long");
  139. snprint(tcpdir, sizeof(tcpdir), "%s/tcp", net);
  140. } else
  141. snprint(tcpdir, sizeof(tcpdir), "tcp");
  142. if(argc != 1)
  143. usage();
  144. out.fd = 1;
  145. out.written = 0;
  146. out.offset = 0;
  147. out.curr = nil;
  148. out.hiwat = nil;
  149. if(ofile != nil){
  150. d = dirstat(ofile);
  151. if(d == nil){
  152. out.fd = create(ofile, OWRITE, 0664);
  153. if(out.fd < 0)
  154. sysfatal("creating %s: %r", ofile);
  155. } else {
  156. out.fd = open(ofile, OWRITE);
  157. if(out.fd < 0)
  158. sysfatal("can't open %s: %r", ofile);
  159. r.start = d->length;
  160. mtime = d->mtime;
  161. free(d);
  162. }
  163. }
  164. errs = 0;
  165. if(crackurl(&u, argv[0]) < 0)
  166. sysfatal("%r");
  167. if(hpx && crackurl(&px, hpx) < 0)
  168. sysfatal("%r");
  169. for(;;){
  170. setoffset(&out, 0);
  171. /* transfer data */
  172. werrstr("");
  173. n = (*method[u.method].f)(&u, &px, &r, &out, mtime);
  174. switch(n){
  175. case Eof:
  176. exits(0);
  177. break;
  178. case Error:
  179. if(errs++ < 10)
  180. continue;
  181. sysfatal("too many errors with no progress %r");
  182. break;
  183. case Server:
  184. sysfatal("server returned: %r");
  185. break;
  186. }
  187. /* forward progress */
  188. errs = 0;
  189. r.start += n;
  190. if(r.start >= r.end)
  191. break;
  192. }
  193. exits(0);
  194. }
  195. int
  196. crackurl(URL *u, char *s)
  197. {
  198. char *p;
  199. int i;
  200. if(u->host != nil){
  201. free(u->host);
  202. u->host = nil;
  203. }
  204. if(u->page != nil){
  205. free(u->page);
  206. u->page = nil;
  207. }
  208. /* get type */
  209. u->method = Other;
  210. for(p = s; *p; p++){
  211. if(*p == '/'){
  212. u->method = Http;
  213. p = s;
  214. break;
  215. }
  216. if(*p == ':' && *(p+1)=='/' && *(p+2)=='/'){
  217. *p = 0;
  218. p += 3;
  219. for(i = 0; i < nelem(method); i++){
  220. if(cistrcmp(s, method[i].name) == 0){
  221. u->method = i;
  222. break;
  223. }
  224. }
  225. break;
  226. }
  227. }
  228. if(u->method == Other){
  229. werrstr("unsupported URL type %s", s);
  230. return -1;
  231. }
  232. /* get system */
  233. s = p;
  234. p = strchr(s, '/');
  235. if(p == nil){
  236. u->host = strdup(s);
  237. u->page = strdup("/");
  238. } else {
  239. u->page = strdup(p);
  240. *p = 0;
  241. u->host = strdup(s);
  242. *p = '/';
  243. }
  244. if(p = strchr(u->host, ':')) {
  245. *p++ = 0;
  246. u->port = p;
  247. } else
  248. u->port = method[u->method].name;
  249. if(*(u->host) == 0){
  250. werrstr("bad url, null host");
  251. return -1;
  252. }
  253. return 0;
  254. }
  255. char *day[] = {
  256. "Sun", "Mon", "Tue", "Wed", "Thu", "Fri", "Sat"
  257. };
  258. char *month[] = {
  259. "Jan", "Feb", "Mar", "Apr", "May", "Jun", "Jul", "Aug", "Sep", "Oct", "Nov", "Dec"
  260. };
  261. struct
  262. {
  263. int fd;
  264. long mtime;
  265. } note;
  266. void
  267. catch(void*, char*)
  268. {
  269. Dir d;
  270. nulldir(&d);
  271. d.mtime = note.mtime;
  272. if(dirfwstat(note.fd, &d) < 0)
  273. sysfatal("catch: can't dirfwstat: %r");
  274. noted(NDFLT);
  275. }
  276. int
  277. dohttp(URL *u, URL *px, Range *r, Out *out, long mtime)
  278. {
  279. int fd, cfd;
  280. int redirect, loop;
  281. int n, rv, code;
  282. long tot, vtime;
  283. Tm *tm;
  284. char buf[1024];
  285. char err[ERRMAX];
  286. /* always move back to a previous 512 byte bound because some
  287. * servers can't seem to deal with requests that start at the
  288. * end of the file
  289. */
  290. if(r->start)
  291. r->start = ((r->start-1)/512)*512;
  292. /* loop for redirects, requires reading both response code and headers */
  293. fd = -1;
  294. for(loop = 0; loop < 32; loop++){
  295. if(px->host == nil){
  296. fd = dial(netmkaddr(u->host, tcpdir, u->port), 0, 0, 0);
  297. } else {
  298. fd = dial(netmkaddr(px->host, tcpdir, px->port), 0, 0, 0);
  299. }
  300. if(fd < 0)
  301. return Error;
  302. if(u->method == Https){
  303. int tfd;
  304. TLSconn conn;
  305. memset(&conn, 0, sizeof conn);
  306. tfd = tlsClient(fd, &conn);
  307. if(tfd < 0){
  308. fprint(2, "tlsClient: %r\n");
  309. close(fd);
  310. return Error;
  311. }
  312. /* BUG: check cert here? */
  313. if(conn.cert)
  314. free(conn.cert);
  315. close(fd);
  316. fd = tfd;
  317. }
  318. /* write request, use range if not start of file */
  319. if(u->postbody == nil){
  320. if(px->host == nil){
  321. dfprint(fd, "GET %s HTTP/1.0\r\n"
  322. "Host: %s\r\n"
  323. "User-agent: Plan9/hget\r\n"
  324. "Cache-Control: no-cache\r\n"
  325. "Pragma: no-cache\r\n",
  326. u->page, u->host);
  327. } else {
  328. dfprint(fd, "GET http://%s%s HTTP/1.0\r\n"
  329. "Host: %s\r\n"
  330. "User-agent: Plan9/hget\r\n"
  331. "Cache-Control: no-cache\r\n"
  332. "Pragma: no-cache\r\n",
  333. u->host, u->page, u->host);
  334. }
  335. } else {
  336. dfprint(fd, "POST %s HTTP/1.0\r\n"
  337. "Host: %s\r\n"
  338. "Content-type: application/x-www-form-urlencoded\r\n"
  339. "Content-length: %d\r\n"
  340. "User-agent: Plan9/hget\r\n"
  341. "\r\n",
  342. u->page, u->host, strlen(u->postbody));
  343. dfprint(fd, "%s", u->postbody);
  344. }
  345. if(r->start != 0){
  346. dfprint(fd, "Range: bytes=%d-\n", r->start);
  347. if(u->etag != nil){
  348. dfprint(fd, "If-range: %s\n", u->etag);
  349. } else {
  350. tm = gmtime(mtime);
  351. dfprint(fd, "If-range: %s, %d %s %d %2d:%2.2d:%2.2d GMT\n",
  352. day[tm->wday], tm->mday, month[tm->mon],
  353. tm->year+1900, tm->hour, tm->min, tm->sec);
  354. }
  355. }
  356. if((cfd = open("/mnt/webcookies/http", ORDWR)) >= 0){
  357. if(fprint(cfd, "http://%s%s", u->host, u->page) > 0){
  358. while((n = read(cfd, buf, sizeof buf)) > 0){
  359. if(debug)
  360. write(2, buf, n);
  361. write(fd, buf, n);
  362. }
  363. }else{
  364. close(cfd);
  365. cfd = -1;
  366. }
  367. }
  368. dfprint(fd, "\r\n", u->host);
  369. redirect = 0;
  370. initibuf();
  371. code = httprcode(fd);
  372. switch(code){
  373. case Error: /* connection timed out */
  374. case Eof:
  375. close(fd);
  376. close(cfd);
  377. return code;
  378. case 200: /* OK */
  379. case 201: /* Created */
  380. case 202: /* Accepted */
  381. if(ofile == nil && r->start != 0)
  382. sysfatal("page changed underfoot");
  383. break;
  384. case 204: /* No Content */
  385. sysfatal("No Content");
  386. case 206: /* Partial Content */
  387. setoffset(out, r->start);
  388. break;
  389. case 301: /* Moved Permanently */
  390. case 302: /* Moved Temporarily */
  391. redirect = 1;
  392. u->postbody = nil;
  393. break;
  394. case 304: /* Not Modified */
  395. break;
  396. case 400: /* Bad Request */
  397. sysfatal("Bad Request");
  398. case 401: /* Unauthorized */
  399. case 402: /* ??? */
  400. sysfatal("Unauthorized");
  401. case 403: /* Forbidden */
  402. sysfatal("Forbidden by server");
  403. case 404: /* Not Found */
  404. sysfatal("Not found on server");
  405. case 500: /* Internal server error */
  406. sysfatal("Server choked");
  407. case 501: /* Not implemented */
  408. sysfatal("Server can't do it!");
  409. case 502: /* Bad gateway */
  410. sysfatal("Bad gateway");
  411. case 503: /* Service unavailable */
  412. sysfatal("Service unavailable");
  413. default:
  414. sysfatal("Unknown response code %d", code);
  415. }
  416. if(u->redirect != nil){
  417. free(u->redirect);
  418. u->redirect = nil;
  419. }
  420. rv = httpheaders(fd, cfd, u, r);
  421. close(cfd);
  422. if(rv != 0){
  423. close(fd);
  424. return rv;
  425. }
  426. if(!redirect)
  427. break;
  428. if(u->redirect == nil)
  429. sysfatal("redirect: no URL");
  430. if(crackurl(u, u->redirect) < 0)
  431. sysfatal("redirect: %r");
  432. }
  433. /* transfer whatever you get */
  434. if(ofile != nil && u->mtime != 0){
  435. note.fd = out->fd;
  436. note.mtime = u->mtime;
  437. notify(catch);
  438. }
  439. tot = 0;
  440. vtime = 0;
  441. for(;;){
  442. n = readibuf(fd, buf, sizeof(buf));
  443. if(n <= 0)
  444. break;
  445. if(output(out, buf, n) != n)
  446. break;
  447. tot += n;
  448. if(verbose && vtime != time(0)) {
  449. vtime = time(0);
  450. fprint(2, "%ld %ld\n", r->start+tot, r->end);
  451. }
  452. }
  453. notify(nil);
  454. close(fd);
  455. if(ofile != nil && u->mtime != 0){
  456. Dir d;
  457. rerrstr(err, sizeof err);
  458. nulldir(&d);
  459. d.mtime = u->mtime;
  460. if(dirfwstat(out->fd, &d) < 0)
  461. fprint(2, "couldn't set mtime: %r\n");
  462. errstr(err, sizeof err);
  463. }
  464. return tot;
  465. }
  466. /* get the http response code */
  467. int
  468. httprcode(int fd)
  469. {
  470. int n;
  471. char *p;
  472. char buf[256];
  473. n = readline(fd, buf, sizeof(buf)-1);
  474. if(n <= 0)
  475. return n;
  476. if(debug)
  477. fprint(2, "%d <- %s\n", fd, buf);
  478. p = strchr(buf, ' ');
  479. if(strncmp(buf, "HTTP/", 5) != 0 || p == nil){
  480. werrstr("bad response from server");
  481. return -1;
  482. }
  483. buf[n] = 0;
  484. return atoi(p+1);
  485. }
  486. /* read in and crack the http headers, update u and r */
  487. void hhetag(char*, URL*, Range*);
  488. void hhmtime(char*, URL*, Range*);
  489. void hhclen(char*, URL*, Range*);
  490. void hhcrange(char*, URL*, Range*);
  491. void hhuri(char*, URL*, Range*);
  492. void hhlocation(char*, URL*, Range*);
  493. struct {
  494. char *name;
  495. void (*f)(char*, URL*, Range*);
  496. } headers[] = {
  497. { "etag:", hhetag },
  498. { "last-modified:", hhmtime },
  499. { "content-length:", hhclen },
  500. { "content-range:", hhcrange },
  501. { "uri:", hhuri },
  502. { "location:", hhlocation },
  503. };
  504. int
  505. httpheaders(int fd, int cfd, URL *u, Range *r)
  506. {
  507. char buf[2048];
  508. char *p;
  509. int i, n;
  510. for(;;){
  511. n = getheader(fd, buf, sizeof(buf));
  512. if(n <= 0)
  513. break;
  514. if(cfd >= 0)
  515. fprint(cfd, "%s\n", buf);
  516. for(i = 0; i < nelem(headers); i++){
  517. n = strlen(headers[i].name);
  518. if(cistrncmp(buf, headers[i].name, n) == 0){
  519. /* skip field name and leading white */
  520. p = buf + n;
  521. while(*p == ' ' || *p == '\t')
  522. p++;
  523. (*headers[i].f)(p, u, r);
  524. break;
  525. }
  526. }
  527. }
  528. return n;
  529. }
  530. /*
  531. * read a single mime header, collect continuations.
  532. *
  533. * this routine assumes that there is a blank line twixt
  534. * the header and the message body, otherwise bytes will
  535. * be lost.
  536. */
  537. int
  538. getheader(int fd, char *buf, int n)
  539. {
  540. char *p, *e;
  541. int i;
  542. n--;
  543. p = buf;
  544. for(e = p + n; ; p += i){
  545. i = readline(fd, p, e-p);
  546. if(i < 0)
  547. return i;
  548. if(p == buf){
  549. /* first line */
  550. if(strchr(buf, ':') == nil)
  551. break; /* end of headers */
  552. } else {
  553. /* continuation line */
  554. if(*p != ' ' && *p != '\t'){
  555. unreadline(p);
  556. *p = 0;
  557. break; /* end of this header */
  558. }
  559. }
  560. }
  561. if(headerprint)
  562. print("%s\n", buf);
  563. if(debug)
  564. fprint(2, "%d <- %s\n", fd, buf);
  565. return p-buf;
  566. }
  567. void
  568. hhetag(char *p, URL *u, Range*)
  569. {
  570. if(u->etag != nil){
  571. if(strcmp(u->etag, p) != 0)
  572. sysfatal("file changed underfoot");
  573. } else
  574. u->etag = strdup(p);
  575. }
  576. char* monthchars = "janfebmaraprmayjunjulaugsepoctnovdec";
  577. void
  578. hhmtime(char *p, URL *u, Range*)
  579. {
  580. char *month, *day, *yr, *hms;
  581. char *fields[6];
  582. Tm tm, now;
  583. int i;
  584. i = getfields(p, fields, 6, 1, " \t");
  585. if(i < 5)
  586. return;
  587. day = fields[1];
  588. month = fields[2];
  589. yr = fields[3];
  590. hms = fields[4];
  591. /* default time */
  592. now = *gmtime(time(0));
  593. tm = now;
  594. tm.yday = 0;
  595. /* convert ascii month to a number twixt 1 and 12 */
  596. if(*month >= '0' && *month <= '9'){
  597. tm.mon = atoi(month) - 1;
  598. if(tm.mon < 0 || tm.mon > 11)
  599. tm.mon = 5;
  600. } else {
  601. for(p = month; *p; p++)
  602. *p = tolower(*p);
  603. for(i = 0; i < 12; i++)
  604. if(strncmp(&monthchars[i*3], month, 3) == 0){
  605. tm.mon = i;
  606. break;
  607. }
  608. }
  609. tm.mday = atoi(day);
  610. if(hms) {
  611. tm.hour = strtoul(hms, &p, 10);
  612. if(*p == ':') {
  613. p++;
  614. tm.min = strtoul(p, &p, 10);
  615. if(*p == ':') {
  616. p++;
  617. tm.sec = strtoul(p, &p, 10);
  618. }
  619. }
  620. if(tolower(*p) == 'p')
  621. tm.hour += 12;
  622. }
  623. if(yr) {
  624. tm.year = atoi(yr);
  625. if(tm.year >= 1900)
  626. tm.year -= 1900;
  627. } else {
  628. if(tm.mon > now.mon || (tm.mon == now.mon && tm.mday > now.mday+1))
  629. tm.year--;
  630. }
  631. strcpy(tm.zone, "GMT");
  632. /* convert to epoch seconds */
  633. u->mtime = tm2sec(&tm);
  634. }
  635. void
  636. hhclen(char *p, URL*, Range *r)
  637. {
  638. r->end = atoi(p);
  639. }
  640. void
  641. hhcrange(char *p, URL*, Range *r)
  642. {
  643. char *x;
  644. vlong l;
  645. l = 0;
  646. x = strchr(p, '/');
  647. if(x)
  648. l = atoll(x+1);
  649. if(l == 0)
  650. x = strchr(p, '-');
  651. if(x)
  652. l = atoll(x+1);
  653. if(l)
  654. r->end = l;
  655. }
  656. void
  657. hhuri(char *p, URL *u, Range*)
  658. {
  659. if(*p != '<')
  660. return;
  661. u->redirect = strdup(p+1);
  662. p = strchr(u->redirect, '>');
  663. if(p != nil)
  664. *p = 0;
  665. }
  666. void
  667. hhlocation(char *p, URL *u, Range*)
  668. {
  669. u->redirect = strdup(p);
  670. }
  671. enum
  672. {
  673. /* ftp return codes */
  674. Extra= 1,
  675. Success= 2,
  676. Incomplete= 3,
  677. TempFail= 4,
  678. PermFail= 5,
  679. Nnetdir= 64, /* max length of network directory paths */
  680. Ndialstr= 64, /* max length of dial strings */
  681. };
  682. int ftpcmd(int, char*, ...);
  683. int ftprcode(int, char*, int);
  684. int hello(int);
  685. int logon(int);
  686. int xfertype(int, char*);
  687. int passive(int, URL*);
  688. int active(int, URL*);
  689. int ftpxfer(int, Out*, Range*);
  690. int terminateftp(int, int);
  691. int getaddrport(char*, uchar*, uchar*);
  692. int ftprestart(int, Out*, URL*, Range*, long);
  693. int
  694. doftp(URL *u, URL *px, Range *r, Out *out, long mtime)
  695. {
  696. int pid, ctl, data, rv;
  697. Waitmsg *w;
  698. char msg[64];
  699. char conndir[NETPATHLEN];
  700. char *p;
  701. /* untested, proxy dosn't work with ftp (I think) */
  702. if(px->host == nil){
  703. ctl = dial(netmkaddr(u->host, tcpdir, u->port), 0, conndir, 0);
  704. } else {
  705. ctl = dial(netmkaddr(px->host, tcpdir, px->port), 0, conndir, 0);
  706. }
  707. if(ctl < 0)
  708. return Error;
  709. if(net == nil){
  710. p = strrchr(conndir, '/');
  711. *p = 0;
  712. snprint(tcpdir, sizeof(tcpdir), conndir);
  713. }
  714. initibuf();
  715. rv = hello(ctl);
  716. if(rv < 0)
  717. return terminateftp(ctl, rv);
  718. rv = logon(ctl);
  719. if(rv < 0)
  720. return terminateftp(ctl, rv);
  721. rv = xfertype(ctl, "I");
  722. if(rv < 0)
  723. return terminateftp(ctl, rv);
  724. /* if file is up to date and the right size, stop */
  725. if(ftprestart(ctl, out, u, r, mtime) > 0){
  726. close(ctl);
  727. return Eof;
  728. }
  729. /* first try passive mode, then active */
  730. data = passive(ctl, u);
  731. if(data < 0){
  732. data = active(ctl, u);
  733. if(data < 0)
  734. return Error;
  735. }
  736. /* fork */
  737. switch(pid = rfork(RFPROC|RFFDG|RFMEM)){
  738. case -1:
  739. close(data);
  740. return terminateftp(ctl, Error);
  741. case 0:
  742. ftpxfer(data, out, r);
  743. close(data);
  744. _exits(0);
  745. default:
  746. close(data);
  747. break;
  748. }
  749. /* wait for reply message */
  750. rv = ftprcode(ctl, msg, sizeof(msg));
  751. close(ctl);
  752. /* wait for process to terminate */
  753. w = nil;
  754. for(;;){
  755. free(w);
  756. w = wait();
  757. if(w == nil)
  758. return Error;
  759. if(w->pid == pid){
  760. if(w->msg[0] == 0){
  761. free(w);
  762. break;
  763. }
  764. werrstr("xfer: %s", w->msg);
  765. free(w);
  766. return Error;
  767. }
  768. }
  769. switch(rv){
  770. case Success:
  771. return Eof;
  772. case TempFail:
  773. return Server;
  774. default:
  775. return Error;
  776. }
  777. }
  778. int
  779. ftpcmd(int ctl, char *fmt, ...)
  780. {
  781. va_list arg;
  782. char buf[2*1024], *s;
  783. va_start(arg, fmt);
  784. s = vseprint(buf, buf + (sizeof(buf)-4) / sizeof(*buf), fmt, arg);
  785. va_end(arg);
  786. if(debug)
  787. fprint(2, "%d -> %s\n", ctl, buf);
  788. *s++ = '\r';
  789. *s++ = '\n';
  790. if(write(ctl, buf, s - buf) != s - buf)
  791. return -1;
  792. return 0;
  793. }
  794. int
  795. ftprcode(int ctl, char *msg, int len)
  796. {
  797. int rv;
  798. int i;
  799. char *p;
  800. len--; /* room for terminating null */
  801. for(;;){
  802. *msg = 0;
  803. i = readline(ctl, msg, len);
  804. if(i < 0)
  805. break;
  806. if(debug)
  807. fprint(2, "%d <- %s\n", ctl, msg);
  808. /* stop if not a continuation */
  809. rv = strtol(msg, &p, 10);
  810. if(rv >= 100 && rv < 600 && p==msg+3 && *p == ' ')
  811. return rv/100;
  812. }
  813. *msg = 0;
  814. return -1;
  815. }
  816. int
  817. hello(int ctl)
  818. {
  819. char msg[1024];
  820. /* wait for hello from other side */
  821. if(ftprcode(ctl, msg, sizeof(msg)) != Success){
  822. werrstr("HELLO: %s", msg);
  823. return Server;
  824. }
  825. return 0;
  826. }
  827. int
  828. getdec(char *p, int n)
  829. {
  830. int x = 0;
  831. int i;
  832. for(i = 0; i < n; i++)
  833. x = x*10 + (*p++ - '0');
  834. return x;
  835. }
  836. int
  837. ftprestart(int ctl, Out *out, URL *u, Range *r, long mtime)
  838. {
  839. Tm tm;
  840. char msg[1024];
  841. long x, rmtime;
  842. ftpcmd(ctl, "MDTM %s", u->page);
  843. if(ftprcode(ctl, msg, sizeof(msg)) != Success){
  844. r->start = 0;
  845. return 0; /* need to do something */
  846. }
  847. /* decode modification time */
  848. if(strlen(msg) < 4 + 4 + 2 + 2 + 2 + 2 + 2){
  849. r->start = 0;
  850. return 0; /* need to do something */
  851. }
  852. memset(&tm, 0, sizeof(tm));
  853. tm.year = getdec(msg+4, 4) - 1900;
  854. tm.mon = getdec(msg+4+4, 2) - 1;
  855. tm.mday = getdec(msg+4+4+2, 2);
  856. tm.hour = getdec(msg+4+4+2+2, 2);
  857. tm.min = getdec(msg+4+4+2+2+2, 2);
  858. tm.sec = getdec(msg+4+4+2+2+2+2, 2);
  859. strcpy(tm.zone, "GMT");
  860. rmtime = tm2sec(&tm);
  861. if(rmtime > mtime)
  862. r->start = 0;
  863. /* get size */
  864. ftpcmd(ctl, "SIZE %s", u->page);
  865. if(ftprcode(ctl, msg, sizeof(msg)) == Success){
  866. x = atol(msg+4);
  867. if(r->start == x)
  868. return 1; /* we're up to date */
  869. r->end = x;
  870. }
  871. /* seek to restart point */
  872. if(r->start > 0){
  873. ftpcmd(ctl, "REST %lud", r->start);
  874. if(ftprcode(ctl, msg, sizeof(msg)) == Incomplete){
  875. setoffset(out, r->start);
  876. }else
  877. r->start = 0;
  878. }
  879. return 0; /* need to do something */
  880. }
  881. int
  882. logon(int ctl)
  883. {
  884. char msg[1024];
  885. /* login anonymous */
  886. ftpcmd(ctl, "USER anonymous");
  887. switch(ftprcode(ctl, msg, sizeof(msg))){
  888. case Success:
  889. return 0;
  890. case Incomplete:
  891. break; /* need password */
  892. default:
  893. werrstr("USER: %s", msg);
  894. return Server;
  895. }
  896. /* send user id as password */
  897. sprint(msg, "%s@closedmind.org", getuser());
  898. ftpcmd(ctl, "PASS %s", msg);
  899. if(ftprcode(ctl, msg, sizeof(msg)) != Success){
  900. werrstr("PASS: %s", msg);
  901. return Server;
  902. }
  903. return 0;
  904. }
  905. int
  906. xfertype(int ctl, char *t)
  907. {
  908. char msg[1024];
  909. ftpcmd(ctl, "TYPE %s", t);
  910. if(ftprcode(ctl, msg, sizeof(msg)) != Success){
  911. werrstr("TYPE %s: %s", t, msg);
  912. return Server;
  913. }
  914. return 0;
  915. }
  916. int
  917. passive(int ctl, URL *u)
  918. {
  919. char msg[1024];
  920. char ipaddr[32];
  921. char *f[6];
  922. char *p;
  923. int fd;
  924. int port;
  925. char aport[12];
  926. ftpcmd(ctl, "PASV");
  927. if(ftprcode(ctl, msg, sizeof(msg)) != Success)
  928. return Error;
  929. /* get address and port number from reply, this is AI */
  930. p = strchr(msg, '(');
  931. if(p == nil){
  932. for(p = msg+3; *p; p++)
  933. if(isdigit(*p))
  934. break;
  935. } else
  936. p++;
  937. if(getfields(p, f, 6, 0, ",)") < 6){
  938. werrstr("ftp protocol botch");
  939. return Server;
  940. }
  941. snprint(ipaddr, sizeof(ipaddr), "%s.%s.%s.%s",
  942. f[0], f[1], f[2], f[3]);
  943. port = ((atoi(f[4])&0xff)<<8) + (atoi(f[5])&0xff);
  944. sprint(aport, "%d", port);
  945. /* open data connection */
  946. fd = dial(netmkaddr(ipaddr, tcpdir, aport), 0, 0, 0);
  947. if(fd < 0){
  948. werrstr("passive mode failed: %r");
  949. return Error;
  950. }
  951. /* tell remote to send a file */
  952. ftpcmd(ctl, "RETR %s", u->page);
  953. if(ftprcode(ctl, msg, sizeof(msg)) != Extra){
  954. werrstr("RETR %s: %s", u->page, msg);
  955. return Error;
  956. }
  957. return fd;
  958. }
  959. int
  960. active(int ctl, URL *u)
  961. {
  962. char msg[1024];
  963. char dir[40], ldir[40];
  964. uchar ipaddr[4];
  965. uchar port[2];
  966. int lcfd, dfd, afd;
  967. /* announce a port for the call back */
  968. snprint(msg, sizeof(msg), "%s!*!0", tcpdir);
  969. afd = announce(msg, dir);
  970. if(afd < 0)
  971. return Error;
  972. /* get a local address/port of the annoucement */
  973. if(getaddrport(dir, ipaddr, port) < 0){
  974. close(afd);
  975. return Error;
  976. }
  977. /* tell remote side address and port*/
  978. ftpcmd(ctl, "PORT %d,%d,%d,%d,%d,%d", ipaddr[0], ipaddr[1], ipaddr[2],
  979. ipaddr[3], port[0], port[1]);
  980. if(ftprcode(ctl, msg, sizeof(msg)) != Success){
  981. close(afd);
  982. werrstr("active: %s", msg);
  983. return Error;
  984. }
  985. /* tell remote to send a file */
  986. ftpcmd(ctl, "RETR %s", u->page);
  987. if(ftprcode(ctl, msg, sizeof(msg)) != Extra){
  988. close(afd);
  989. werrstr("RETR: %s", msg);
  990. return Server;
  991. }
  992. /* wait for a connection */
  993. lcfd = listen(dir, ldir);
  994. if(lcfd < 0){
  995. close(afd);
  996. return Error;
  997. }
  998. dfd = accept(lcfd, ldir);
  999. if(dfd < 0){
  1000. close(afd);
  1001. close(lcfd);
  1002. return Error;
  1003. }
  1004. close(afd);
  1005. close(lcfd);
  1006. return dfd;
  1007. }
  1008. int
  1009. ftpxfer(int in, Out *out, Range *r)
  1010. {
  1011. char buf[1024];
  1012. long vtime;
  1013. int i, n;
  1014. vtime = 0;
  1015. for(n = 0;;n += i){
  1016. i = read(in, buf, sizeof(buf));
  1017. if(i == 0)
  1018. break;
  1019. if(i < 0)
  1020. return Error;
  1021. if(output(out, buf, i) != i)
  1022. return Error;
  1023. r->start += i;
  1024. if(verbose && vtime != time(0)) {
  1025. vtime = time(0);
  1026. fprint(2, "%ld %ld\n", r->start, r->end);
  1027. }
  1028. }
  1029. return n;
  1030. }
  1031. int
  1032. terminateftp(int ctl, int rv)
  1033. {
  1034. close(ctl);
  1035. return rv;
  1036. }
  1037. /*
  1038. * case insensitive strcmp (why aren't these in libc?)
  1039. */
  1040. int
  1041. cistrncmp(char *a, char *b, int n)
  1042. {
  1043. while(n-- > 0){
  1044. if(tolower(*a++) != tolower(*b++))
  1045. return -1;
  1046. }
  1047. return 0;
  1048. }
  1049. int
  1050. cistrcmp(char *a, char *b)
  1051. {
  1052. while(*a || *b)
  1053. if(tolower(*a++) != tolower(*b++))
  1054. return -1;
  1055. return 0;
  1056. }
  1057. /*
  1058. * buffered io
  1059. */
  1060. struct
  1061. {
  1062. char *rp;
  1063. char *wp;
  1064. char buf[4*1024];
  1065. } b;
  1066. void
  1067. initibuf(void)
  1068. {
  1069. b.rp = b.wp = b.buf;
  1070. }
  1071. /*
  1072. * read a possibly buffered line, strip off trailing while
  1073. */
  1074. int
  1075. readline(int fd, char *buf, int len)
  1076. {
  1077. int n;
  1078. char *p;
  1079. int eof = 0;
  1080. len--;
  1081. for(p = buf;;){
  1082. if(b.rp >= b.wp){
  1083. n = read(fd, b.wp, sizeof(b.buf)/2);
  1084. if(n < 0)
  1085. return -1;
  1086. if(n == 0){
  1087. eof = 1;
  1088. break;
  1089. }
  1090. b.wp += n;
  1091. }
  1092. n = *b.rp++;
  1093. if(len > 0){
  1094. *p++ = n;
  1095. len--;
  1096. }
  1097. if(n == '\n')
  1098. break;
  1099. }
  1100. /* drop trailing white */
  1101. for(;;){
  1102. if(p <= buf)
  1103. break;
  1104. n = *(p-1);
  1105. if(n != ' ' && n != '\t' && n != '\r' && n != '\n')
  1106. break;
  1107. p--;
  1108. }
  1109. *p = 0;
  1110. if(eof && p == buf)
  1111. return -1;
  1112. return p-buf;
  1113. }
  1114. void
  1115. unreadline(char *line)
  1116. {
  1117. int i, n;
  1118. i = strlen(line);
  1119. n = b.wp-b.rp;
  1120. memmove(&b.buf[i+1], b.rp, n);
  1121. memmove(b.buf, line, i);
  1122. b.buf[i] = '\n';
  1123. b.rp = b.buf;
  1124. b.wp = b.rp + i + 1 + n;
  1125. }
  1126. int
  1127. readibuf(int fd, char *buf, int len)
  1128. {
  1129. int n;
  1130. n = b.wp-b.rp;
  1131. if(n > 0){
  1132. if(n > len)
  1133. n = len;
  1134. memmove(buf, b.rp, n);
  1135. b.rp += n;
  1136. return n;
  1137. }
  1138. return read(fd, buf, len);
  1139. }
  1140. int
  1141. dfprint(int fd, char *fmt, ...)
  1142. {
  1143. char buf[4*1024];
  1144. va_list arg;
  1145. va_start(arg, fmt);
  1146. vseprint(buf, buf+sizeof(buf), fmt, arg);
  1147. va_end(arg);
  1148. if(debug)
  1149. fprint(2, "%d -> %s", fd, buf);
  1150. return fprint(fd, "%s", buf);
  1151. }
  1152. int
  1153. getaddrport(char *dir, uchar *ipaddr, uchar *port)
  1154. {
  1155. char buf[256];
  1156. int fd, i;
  1157. char *p;
  1158. snprint(buf, sizeof(buf), "%s/local", dir);
  1159. fd = open(buf, OREAD);
  1160. if(fd < 0)
  1161. return -1;
  1162. i = read(fd, buf, sizeof(buf)-1);
  1163. close(fd);
  1164. if(i <= 0)
  1165. return -1;
  1166. buf[i] = 0;
  1167. p = strchr(buf, '!');
  1168. if(p != nil)
  1169. *p++ = 0;
  1170. v4parseip(ipaddr, buf);
  1171. i = atoi(p);
  1172. port[0] = i>>8;
  1173. port[1] = i;
  1174. return 0;
  1175. }
  1176. void
  1177. md5free(DigestState *state)
  1178. {
  1179. uchar x[MD5dlen];
  1180. md5(nil, 0, x, state);
  1181. }
  1182. DigestState*
  1183. md5dup(DigestState *state)
  1184. {
  1185. char *p;
  1186. p = md5pickle(state);
  1187. if(p == nil)
  1188. sysfatal("md5pickle: %r");
  1189. state = md5unpickle(p);
  1190. if(state == nil)
  1191. sysfatal("md5unpickle: %r");
  1192. free(p);
  1193. return state;
  1194. }
  1195. void
  1196. setoffset(Out *out, int offset)
  1197. {
  1198. md5free(out->curr);
  1199. if(offset == 0)
  1200. out->curr = md5(nil, 0, nil, nil);
  1201. else
  1202. out->curr = nil;
  1203. out->offset = offset;
  1204. }
  1205. /*
  1206. * write some output, discarding it (but keeping track)
  1207. * if we've already written it. if we've gone backwards,
  1208. * verify that everything previously written matches
  1209. * that which would have been written from the current
  1210. * output.
  1211. */
  1212. int
  1213. output(Out *out, char *buf, int nb)
  1214. {
  1215. int n, d;
  1216. uchar m0[MD5dlen], m1[MD5dlen];
  1217. n = nb;
  1218. d = out->written - out->offset;
  1219. assert(d >= 0);
  1220. if(d > 0){
  1221. if(n < d){
  1222. if(out->curr != nil)
  1223. md5((uchar*)buf, n, nil, out->curr);
  1224. out->offset += n;
  1225. return n;
  1226. }
  1227. if(out->curr != nil){
  1228. md5((uchar*)buf, d, m0, out->curr);
  1229. out->curr = nil;
  1230. md5(nil, 0, m1, md5dup(out->hiwat));
  1231. if(memcmp(m0, m1, MD5dlen) != 0){
  1232. fprint(2, "integrity check failure at offset %d\n", out->written);
  1233. return -1;
  1234. }
  1235. }
  1236. buf += d;
  1237. n -= d;
  1238. out->offset += d;
  1239. }
  1240. if(n > 0){
  1241. out->hiwat = md5((uchar*)buf, n, nil, out->hiwat);
  1242. n = write(out->fd, buf, n);
  1243. if(n > 0){
  1244. out->offset += n;
  1245. out->written += n;
  1246. }
  1247. }
  1248. return n + d;
  1249. }