hget.c 25 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962963964965966967968969970971972973974975976977978979980981982983984985986987988989990991992993994995996997998999100010011002100310041005100610071008100910101011101210131014101510161017101810191020102110221023102410251026102710281029103010311032103310341035103610371038103910401041104210431044104510461047104810491050105110521053105410551056105710581059106010611062106310641065106610671068106910701071107210731074107510761077107810791080108110821083108410851086108710881089109010911092109310941095109610971098109911001101110211031104110511061107110811091110111111121113111411151116111711181119112011211122112311241125112611271128112911301131113211331134113511361137113811391140114111421143114411451146114711481149115011511152115311541155115611571158115911601161116211631164116511661167116811691170117111721173117411751176117711781179118011811182118311841185118611871188118911901191119211931194119511961197119811991200120112021203120412051206120712081209121012111212121312141215121612171218121912201221122212231224122512261227122812291230123112321233123412351236123712381239124012411242124312441245124612471248124912501251125212531254125512561257125812591260126112621263126412651266126712681269127012711272127312741275127612771278127912801281128212831284128512861287128812891290129112921293129412951296129712981299130013011302130313041305130613071308130913101311131213131314131513161317131813191320132113221323132413251326132713281329133013311332133313341335133613371338133913401341134213431344134513461347134813491350135113521353135413551356135713581359136013611362136313641365136613671368136913701371137213731374137513761377137813791380138113821383138413851386138713881389139013911392139313941395139613971398139914001401140214031404140514061407140814091410141114121413141414151416141714181419142014211422142314241425142614271428142914301431143214331434143514361437143814391440144114421443144414451446144714481449145014511452145314541455145614571458145914601461146214631464146514661467146814691470147114721473147414751476
  1. #include <u.h>
  2. #include <libc.h>
  3. #include <ctype.h>
  4. #include <bio.h>
  5. #include <ip.h>
  6. #include <libsec.h>
  7. #include <auth.h>
  8. typedef struct URL URL;
  9. struct URL
  10. {
  11. int method;
  12. char *host;
  13. char *port;
  14. char *page;
  15. char *etag;
  16. char *redirect;
  17. char *postbody;
  18. char *cred;
  19. long mtime;
  20. };
  21. typedef struct Range Range;
  22. struct Range
  23. {
  24. long start; /* only 2 gig supported, tdb */
  25. long end;
  26. };
  27. typedef struct Out Out;
  28. struct Out
  29. {
  30. int fd;
  31. int offset; /* notional current offset in output */
  32. int written; /* number of bytes successfully transferred to output */
  33. DigestState *curr; /* digest state up to offset (if known) */
  34. DigestState *hiwat; /* digest state of all bytes written */
  35. };
  36. enum
  37. {
  38. Other,
  39. Http,
  40. Https,
  41. Ftp,
  42. };
  43. enum
  44. {
  45. Eof = 0,
  46. Error = -1,
  47. Server = -2,
  48. Changed = -3,
  49. };
  50. int debug;
  51. char *ofile;
  52. int doftp(URL*, URL*, Range*, Out*, long);
  53. int dohttp(URL*, URL*, Range*, Out*, long);
  54. int crackurl(URL*, char*);
  55. Range* crackrange(char*);
  56. int getheader(int, char*, int);
  57. int httpheaders(int, int, URL*, Range*);
  58. int httprcode(int);
  59. int cistrncmp(char*, char*, int);
  60. int cistrcmp(char*, char*);
  61. void initibuf(void);
  62. int readline(int, char*, int);
  63. int readibuf(int, char*, int);
  64. int dfprint(int, char*, ...);
  65. void unreadline(char*);
  66. int output(Out*, char*, int);
  67. void setoffset(Out*, int);
  68. int verbose;
  69. char *net;
  70. char tcpdir[NETPATHLEN];
  71. int headerprint;
  72. struct {
  73. char *name;
  74. int (*f)(URL*, URL*, Range*, Out*, long);
  75. } method[] = {
  76. [Http] { "http", dohttp },
  77. [Https] { "https", dohttp },
  78. [Ftp] { "ftp", doftp },
  79. [Other] { "_______", nil },
  80. };
  81. void
  82. usage(void)
  83. {
  84. fprint(2, "usage: %s [-dhv] [-o outfile] [-p body] [-x netmtpt] url\n", argv0);
  85. exits("usage");
  86. }
  87. void
  88. main(int argc, char **argv)
  89. {
  90. URL u;
  91. Range r;
  92. int errs, n;
  93. ulong mtime;
  94. Dir *d;
  95. char postbody[4096], *p, *e, *t, *hpx;
  96. URL px; // Proxy
  97. Out out;
  98. ofile = nil;
  99. p = postbody;
  100. e = p + sizeof(postbody);
  101. r.start = 0;
  102. r.end = -1;
  103. mtime = 0;
  104. memset(&u, 0, sizeof(u));
  105. memset(&px, 0, sizeof(px));
  106. hpx = getenv("httpproxy");
  107. ARGBEGIN {
  108. case 'o':
  109. ofile = EARGF(usage());
  110. break;
  111. case 'd':
  112. debug = 1;
  113. break;
  114. case 'h':
  115. headerprint = 1;
  116. break;
  117. case 'v':
  118. verbose = 1;
  119. break;
  120. case 'x':
  121. net = EARGF(usage());
  122. break;
  123. case 'p':
  124. t = EARGF(usage());
  125. if(p != postbody)
  126. p = seprint(p, e, "&%s", t);
  127. else
  128. p = seprint(p, e, "%s", t);
  129. u.postbody = postbody;
  130. break;
  131. default:
  132. usage();
  133. } ARGEND;
  134. if(net != nil){
  135. if(strlen(net) > sizeof(tcpdir)-5)
  136. sysfatal("network mount point too long");
  137. snprint(tcpdir, sizeof(tcpdir), "%s/tcp", net);
  138. } else
  139. snprint(tcpdir, sizeof(tcpdir), "tcp");
  140. if(argc != 1)
  141. usage();
  142. out.fd = 1;
  143. out.written = 0;
  144. out.offset = 0;
  145. out.curr = nil;
  146. out.hiwat = nil;
  147. if(ofile != nil){
  148. d = dirstat(ofile);
  149. if(d == nil){
  150. out.fd = create(ofile, OWRITE, 0664);
  151. if(out.fd < 0)
  152. sysfatal("creating %s: %r", ofile);
  153. } else {
  154. out.fd = open(ofile, OWRITE);
  155. if(out.fd < 0)
  156. sysfatal("can't open %s: %r", ofile);
  157. r.start = d->length;
  158. mtime = d->mtime;
  159. free(d);
  160. }
  161. }
  162. errs = 0;
  163. if(crackurl(&u, argv[0]) < 0)
  164. sysfatal("%r");
  165. if(hpx && crackurl(&px, hpx) < 0)
  166. sysfatal("%r");
  167. for(;;){
  168. setoffset(&out, 0);
  169. /* transfer data */
  170. werrstr("");
  171. n = (*method[u.method].f)(&u, &px, &r, &out, mtime);
  172. switch(n){
  173. case Eof:
  174. exits(0);
  175. break;
  176. case Error:
  177. if(errs++ < 10)
  178. continue;
  179. sysfatal("too many errors with no progress %r");
  180. break;
  181. case Server:
  182. sysfatal("server returned: %r");
  183. break;
  184. }
  185. /* forward progress */
  186. errs = 0;
  187. r.start += n;
  188. if(r.start >= r.end)
  189. break;
  190. }
  191. exits(0);
  192. }
  193. int
  194. crackurl(URL *u, char *s)
  195. {
  196. char *p;
  197. int i;
  198. if(u->page != nil){
  199. free(u->page);
  200. u->page = nil;
  201. }
  202. /* get type */
  203. for(p = s; *p; p++){
  204. if(*p == '/'){
  205. p = s;
  206. if(u->method == Other){
  207. werrstr("missing method");
  208. return -1;
  209. }
  210. if(u->host == nil){
  211. werrstr("missing host");
  212. return -1;
  213. }
  214. u->page = strdup(p);
  215. return 0;
  216. }
  217. if(*p == ':' && *(p+1)=='/' && *(p+2)=='/'){
  218. *p = 0;
  219. p += 3;
  220. for(i = 0; i < nelem(method); i++){
  221. if(cistrcmp(s, method[i].name) == 0){
  222. u->method = i;
  223. break;
  224. }
  225. }
  226. break;
  227. }
  228. }
  229. if(u->method == Other){
  230. werrstr("unsupported URL type %s", s);
  231. return -1;
  232. }
  233. /* get system */
  234. free(u->host);
  235. s = p;
  236. p = strchr(s, '/');
  237. if(p == nil){
  238. u->host = strdup(s);
  239. u->page = strdup("/");
  240. } else {
  241. u->page = strdup(p);
  242. *p = 0;
  243. u->host = strdup(s);
  244. *p = '/';
  245. }
  246. if(p = strchr(u->host, ':')) {
  247. *p++ = 0;
  248. u->port = p;
  249. } else
  250. u->port = method[u->method].name;
  251. if(*(u->host) == 0){
  252. werrstr("bad url, null host");
  253. return -1;
  254. }
  255. return 0;
  256. }
  257. char *day[] = {
  258. "Sun", "Mon", "Tue", "Wed", "Thu", "Fri", "Sat"
  259. };
  260. char *month[] = {
  261. "Jan", "Feb", "Mar", "Apr", "May", "Jun", "Jul", "Aug", "Sep", "Oct", "Nov", "Dec"
  262. };
  263. struct
  264. {
  265. int fd;
  266. long mtime;
  267. } note;
  268. void
  269. catch(void*, char*)
  270. {
  271. Dir d;
  272. nulldir(&d);
  273. d.mtime = note.mtime;
  274. if(dirfwstat(note.fd, &d) < 0)
  275. sysfatal("catch: can't dirfwstat: %r");
  276. noted(NDFLT);
  277. }
  278. int
  279. dohttp(URL *u, URL *px, Range *r, Out *out, long mtime)
  280. {
  281. int fd, cfd;
  282. int redirect, auth, loop;
  283. int n, rv, code;
  284. long tot, vtime;
  285. Tm *tm;
  286. char buf[1024];
  287. char err[ERRMAX];
  288. /* always move back to a previous 512 byte bound because some
  289. * servers can't seem to deal with requests that start at the
  290. * end of the file
  291. */
  292. if(r->start)
  293. r->start = ((r->start-1)/512)*512;
  294. /* loop for redirects, requires reading both response code and headers */
  295. fd = -1;
  296. for(loop = 0; loop < 32; loop++){
  297. if(px->host == nil){
  298. fd = dial(netmkaddr(u->host, tcpdir, u->port), 0, 0, 0);
  299. } else {
  300. fd = dial(netmkaddr(px->host, tcpdir, px->port), 0, 0, 0);
  301. }
  302. if(fd < 0)
  303. return Error;
  304. if(u->method == Https){
  305. int tfd;
  306. TLSconn conn;
  307. memset(&conn, 0, sizeof conn);
  308. tfd = tlsClient(fd, &conn);
  309. if(tfd < 0){
  310. fprint(2, "tlsClient: %r\n");
  311. close(fd);
  312. return Error;
  313. }
  314. /* BUG: check cert here? */
  315. if(conn.cert)
  316. free(conn.cert);
  317. close(fd);
  318. fd = tfd;
  319. }
  320. /* write request, use range if not start of file */
  321. if(u->postbody == nil){
  322. if(px->host == nil){
  323. dfprint(fd, "GET %s HTTP/1.0\r\n"
  324. "Host: %s\r\n"
  325. "User-agent: Plan9/hget\r\n"
  326. "Cache-Control: no-cache\r\n"
  327. "Pragma: no-cache\r\n",
  328. u->page, u->host);
  329. } else {
  330. dfprint(fd, "GET http://%s%s HTTP/1.0\r\n"
  331. "Host: %s\r\n"
  332. "User-agent: Plan9/hget\r\n"
  333. "Cache-Control: no-cache\r\n"
  334. "Pragma: no-cache\r\n",
  335. u->host, u->page, u->host);
  336. }
  337. if(u->cred)
  338. dfprint(fd, "Authorization: Basic %s\r\n",
  339. u->cred);
  340. } else {
  341. dfprint(fd, "POST %s HTTP/1.0\r\n"
  342. "Host: %s\r\n"
  343. "Content-type: application/x-www-form-urlencoded\r\n"
  344. "Content-length: %d\r\n"
  345. "User-agent: Plan9/hget\r\n",
  346. u->page, u->host, strlen(u->postbody));
  347. if(u->cred)
  348. dfprint(fd, "Authorization: Basic %s\r\n", u->cred);
  349. }
  350. if(r->start != 0){
  351. dfprint(fd, "Range: bytes=%d-\n", r->start);
  352. if(u->etag != nil){
  353. dfprint(fd, "If-range: %s\n", u->etag);
  354. } else {
  355. tm = gmtime(mtime);
  356. dfprint(fd, "If-range: %s, %d %s %d %2d:%2.2d:%2.2d GMT\n",
  357. day[tm->wday], tm->mday, month[tm->mon],
  358. tm->year+1900, tm->hour, tm->min, tm->sec);
  359. }
  360. }
  361. if((cfd = open("/mnt/webcookies/http", ORDWR)) >= 0){
  362. if(fprint(cfd, "http://%s%s", u->host, u->page) > 0){
  363. while((n = read(cfd, buf, sizeof buf)) > 0){
  364. if(debug)
  365. write(2, buf, n);
  366. write(fd, buf, n);
  367. }
  368. }else{
  369. close(cfd);
  370. cfd = -1;
  371. }
  372. }
  373. dfprint(fd, "\r\n", u->host);
  374. if(u->postbody)
  375. dfprint(fd, "%s", u->postbody);
  376. auth = 0;
  377. redirect = 0;
  378. initibuf();
  379. code = httprcode(fd);
  380. switch(code){
  381. case Error: /* connection timed out */
  382. case Eof:
  383. close(fd);
  384. close(cfd);
  385. return code;
  386. case 200: /* OK */
  387. case 201: /* Created */
  388. case 202: /* Accepted */
  389. if(ofile == nil && r->start != 0)
  390. sysfatal("page changed underfoot");
  391. break;
  392. case 204: /* No Content */
  393. sysfatal("No Content");
  394. case 206: /* Partial Content */
  395. setoffset(out, r->start);
  396. break;
  397. case 301: /* Moved Permanently */
  398. case 302: /* Moved Temporarily */
  399. case 303: /* See Other */
  400. redirect = 1;
  401. u->postbody = nil;
  402. break;
  403. case 304: /* Not Modified */
  404. break;
  405. case 400: /* Bad Request */
  406. sysfatal("Bad Request");
  407. case 401: /* Unauthorized */
  408. if (auth)
  409. sysfatal("Authentication failed");
  410. auth = 1;
  411. break;
  412. case 402: /* ??? */
  413. sysfatal("Unauthorized");
  414. case 403: /* Forbidden */
  415. sysfatal("Forbidden by server");
  416. case 404: /* Not Found */
  417. sysfatal("Not found on server");
  418. case 407: /* Proxy Authentication */
  419. sysfatal("Proxy authentication required");
  420. case 500: /* Internal server error */
  421. sysfatal("Server choked");
  422. case 501: /* Not implemented */
  423. sysfatal("Server can't do it!");
  424. case 502: /* Bad gateway */
  425. sysfatal("Bad gateway");
  426. case 503: /* Service unavailable */
  427. sysfatal("Service unavailable");
  428. default:
  429. sysfatal("Unknown response code %d", code);
  430. }
  431. if(u->redirect != nil){
  432. free(u->redirect);
  433. u->redirect = nil;
  434. }
  435. rv = httpheaders(fd, cfd, u, r);
  436. close(cfd);
  437. if(rv != 0){
  438. close(fd);
  439. return rv;
  440. }
  441. if(!redirect && !auth)
  442. break;
  443. if (redirect){
  444. if(u->redirect == nil)
  445. sysfatal("redirect: no URL");
  446. if(crackurl(u, u->redirect) < 0)
  447. sysfatal("redirect: %r");
  448. }
  449. }
  450. /* transfer whatever you get */
  451. if(ofile != nil && u->mtime != 0){
  452. note.fd = out->fd;
  453. note.mtime = u->mtime;
  454. notify(catch);
  455. }
  456. tot = 0;
  457. vtime = 0;
  458. for(;;){
  459. n = readibuf(fd, buf, sizeof(buf));
  460. if(n <= 0)
  461. break;
  462. if(output(out, buf, n) != n)
  463. break;
  464. tot += n;
  465. if(verbose && (vtime != time(0) || r->start == r->end)) {
  466. vtime = time(0);
  467. fprint(2, "%ld %ld\n", r->start+tot, r->end);
  468. }
  469. }
  470. notify(nil);
  471. close(fd);
  472. if(ofile != nil && u->mtime != 0){
  473. Dir d;
  474. rerrstr(err, sizeof err);
  475. nulldir(&d);
  476. d.mtime = u->mtime;
  477. if(dirfwstat(out->fd, &d) < 0)
  478. fprint(2, "couldn't set mtime: %r\n");
  479. errstr(err, sizeof err);
  480. }
  481. return tot;
  482. }
  483. /* get the http response code */
  484. int
  485. httprcode(int fd)
  486. {
  487. int n;
  488. char *p;
  489. char buf[256];
  490. n = readline(fd, buf, sizeof(buf)-1);
  491. if(n <= 0)
  492. return n;
  493. if(debug)
  494. fprint(2, "%d <- %s\n", fd, buf);
  495. p = strchr(buf, ' ');
  496. if(strncmp(buf, "HTTP/", 5) != 0 || p == nil){
  497. werrstr("bad response from server");
  498. return -1;
  499. }
  500. buf[n] = 0;
  501. return atoi(p+1);
  502. }
  503. /* read in and crack the http headers, update u and r */
  504. void hhetag(char*, URL*, Range*);
  505. void hhmtime(char*, URL*, Range*);
  506. void hhclen(char*, URL*, Range*);
  507. void hhcrange(char*, URL*, Range*);
  508. void hhuri(char*, URL*, Range*);
  509. void hhlocation(char*, URL*, Range*);
  510. void hhauth(char*, URL*, Range*);
  511. struct {
  512. char *name;
  513. void (*f)(char*, URL*, Range*);
  514. } headers[] = {
  515. { "etag:", hhetag },
  516. { "last-modified:", hhmtime },
  517. { "content-length:", hhclen },
  518. { "content-range:", hhcrange },
  519. { "uri:", hhuri },
  520. { "location:", hhlocation },
  521. { "WWW-Authenticate:", hhauth },
  522. };
  523. int
  524. httpheaders(int fd, int cfd, URL *u, Range *r)
  525. {
  526. char buf[2048];
  527. char *p;
  528. int i, n;
  529. for(;;){
  530. n = getheader(fd, buf, sizeof(buf));
  531. if(n <= 0)
  532. break;
  533. if(cfd >= 0)
  534. fprint(cfd, "%s\n", buf);
  535. for(i = 0; i < nelem(headers); i++){
  536. n = strlen(headers[i].name);
  537. if(cistrncmp(buf, headers[i].name, n) == 0){
  538. /* skip field name and leading white */
  539. p = buf + n;
  540. while(*p == ' ' || *p == '\t')
  541. p++;
  542. (*headers[i].f)(p, u, r);
  543. break;
  544. }
  545. }
  546. }
  547. return n;
  548. }
  549. /*
  550. * read a single mime header, collect continuations.
  551. *
  552. * this routine assumes that there is a blank line twixt
  553. * the header and the message body, otherwise bytes will
  554. * be lost.
  555. */
  556. int
  557. getheader(int fd, char *buf, int n)
  558. {
  559. char *p, *e;
  560. int i;
  561. n--;
  562. p = buf;
  563. for(e = p + n; ; p += i){
  564. i = readline(fd, p, e-p);
  565. if(i < 0)
  566. return i;
  567. if(p == buf){
  568. /* first line */
  569. if(strchr(buf, ':') == nil)
  570. break; /* end of headers */
  571. } else {
  572. /* continuation line */
  573. if(*p != ' ' && *p != '\t'){
  574. unreadline(p);
  575. *p = 0;
  576. break; /* end of this header */
  577. }
  578. }
  579. }
  580. if(headerprint)
  581. print("%s\n", buf);
  582. if(debug)
  583. fprint(2, "%d <- %s\n", fd, buf);
  584. return p-buf;
  585. }
  586. void
  587. hhetag(char *p, URL *u, Range*)
  588. {
  589. if(u->etag != nil){
  590. if(strcmp(u->etag, p) != 0)
  591. sysfatal("file changed underfoot");
  592. } else
  593. u->etag = strdup(p);
  594. }
  595. char* monthchars = "janfebmaraprmayjunjulaugsepoctnovdec";
  596. void
  597. hhmtime(char *p, URL *u, Range*)
  598. {
  599. char *month, *day, *yr, *hms;
  600. char *fields[6];
  601. Tm tm, now;
  602. int i;
  603. i = getfields(p, fields, 6, 1, " \t");
  604. if(i < 5)
  605. return;
  606. day = fields[1];
  607. month = fields[2];
  608. yr = fields[3];
  609. hms = fields[4];
  610. /* default time */
  611. now = *gmtime(time(0));
  612. tm = now;
  613. tm.yday = 0;
  614. /* convert ascii month to a number twixt 1 and 12 */
  615. if(*month >= '0' && *month <= '9'){
  616. tm.mon = atoi(month) - 1;
  617. if(tm.mon < 0 || tm.mon > 11)
  618. tm.mon = 5;
  619. } else {
  620. for(p = month; *p; p++)
  621. *p = tolower(*p);
  622. for(i = 0; i < 12; i++)
  623. if(strncmp(&monthchars[i*3], month, 3) == 0){
  624. tm.mon = i;
  625. break;
  626. }
  627. }
  628. tm.mday = atoi(day);
  629. if(hms) {
  630. tm.hour = strtoul(hms, &p, 10);
  631. if(*p == ':') {
  632. p++;
  633. tm.min = strtoul(p, &p, 10);
  634. if(*p == ':') {
  635. p++;
  636. tm.sec = strtoul(p, &p, 10);
  637. }
  638. }
  639. if(tolower(*p) == 'p')
  640. tm.hour += 12;
  641. }
  642. if(yr) {
  643. tm.year = atoi(yr);
  644. if(tm.year >= 1900)
  645. tm.year -= 1900;
  646. } else {
  647. if(tm.mon > now.mon || (tm.mon == now.mon && tm.mday > now.mday+1))
  648. tm.year--;
  649. }
  650. strcpy(tm.zone, "GMT");
  651. /* convert to epoch seconds */
  652. u->mtime = tm2sec(&tm);
  653. }
  654. void
  655. hhclen(char *p, URL*, Range *r)
  656. {
  657. r->end = atoi(p);
  658. }
  659. void
  660. hhcrange(char *p, URL*, Range *r)
  661. {
  662. char *x;
  663. vlong l;
  664. l = 0;
  665. x = strchr(p, '/');
  666. if(x)
  667. l = atoll(x+1);
  668. if(l == 0) {
  669. x = strchr(p, '-');
  670. if(x)
  671. l = atoll(x+1);
  672. }
  673. if(l)
  674. r->end = l;
  675. }
  676. void
  677. hhuri(char *p, URL *u, Range*)
  678. {
  679. if(*p != '<')
  680. return;
  681. u->redirect = strdup(p+1);
  682. p = strchr(u->redirect, '>');
  683. if(p != nil)
  684. *p = 0;
  685. }
  686. void
  687. hhlocation(char *p, URL *u, Range*)
  688. {
  689. u->redirect = strdup(p);
  690. }
  691. void
  692. hhauth(char *p, URL *u, Range*)
  693. {
  694. char *f[4];
  695. UserPasswd *up;
  696. char *s, cred[64];
  697. if (cistrncmp(p, "basic ", 6) != 0)
  698. sysfatal("only Basic authentication supported");
  699. if (gettokens(p, f, nelem(f), "\"") < 2)
  700. sysfatal("garbled auth data");
  701. if ((up = auth_getuserpasswd(auth_getkey, "proto=pass service=http server=%q realm=%q",
  702. u->host, f[1])) == nil)
  703. sysfatal("cannot authenticate");
  704. s = smprint("%s:%s", up->user, up->passwd);
  705. if(enc64(cred, sizeof(cred), (uchar *)s, strlen(s)) == -1)
  706. sysfatal("enc64");
  707. free(s);
  708. assert(u->cred = strdup(cred));
  709. }
  710. enum
  711. {
  712. /* ftp return codes */
  713. Extra= 1,
  714. Success= 2,
  715. Incomplete= 3,
  716. TempFail= 4,
  717. PermFail= 5,
  718. Nnetdir= 64, /* max length of network directory paths */
  719. Ndialstr= 64, /* max length of dial strings */
  720. };
  721. int ftpcmd(int, char*, ...);
  722. int ftprcode(int, char*, int);
  723. int hello(int);
  724. int logon(int);
  725. int xfertype(int, char*);
  726. int passive(int, URL*);
  727. int active(int, URL*);
  728. int ftpxfer(int, Out*, Range*);
  729. int terminateftp(int, int);
  730. int getaddrport(char*, uchar*, uchar*);
  731. int ftprestart(int, Out*, URL*, Range*, long);
  732. int
  733. doftp(URL *u, URL *px, Range *r, Out *out, long mtime)
  734. {
  735. int pid, ctl, data, rv;
  736. Waitmsg *w;
  737. char msg[64];
  738. char conndir[NETPATHLEN];
  739. char *p;
  740. /* untested, proxy doesn't work with ftp (I think) */
  741. if(px->host == nil){
  742. ctl = dial(netmkaddr(u->host, tcpdir, u->port), 0, conndir, 0);
  743. } else {
  744. ctl = dial(netmkaddr(px->host, tcpdir, px->port), 0, conndir, 0);
  745. }
  746. if(ctl < 0)
  747. return Error;
  748. if(net == nil){
  749. p = strrchr(conndir, '/');
  750. *p = 0;
  751. snprint(tcpdir, sizeof(tcpdir), conndir);
  752. }
  753. initibuf();
  754. rv = hello(ctl);
  755. if(rv < 0)
  756. return terminateftp(ctl, rv);
  757. rv = logon(ctl);
  758. if(rv < 0)
  759. return terminateftp(ctl, rv);
  760. rv = xfertype(ctl, "I");
  761. if(rv < 0)
  762. return terminateftp(ctl, rv);
  763. /* if file is up to date and the right size, stop */
  764. if(ftprestart(ctl, out, u, r, mtime) > 0){
  765. close(ctl);
  766. return Eof;
  767. }
  768. /* first try passive mode, then active */
  769. data = passive(ctl, u);
  770. if(data < 0){
  771. data = active(ctl, u);
  772. if(data < 0)
  773. return Error;
  774. }
  775. /* fork */
  776. switch(pid = rfork(RFPROC|RFFDG|RFMEM)){
  777. case -1:
  778. close(data);
  779. return terminateftp(ctl, Error);
  780. case 0:
  781. ftpxfer(data, out, r);
  782. close(data);
  783. _exits(0);
  784. default:
  785. close(data);
  786. break;
  787. }
  788. /* wait for reply message */
  789. rv = ftprcode(ctl, msg, sizeof(msg));
  790. close(ctl);
  791. /* wait for process to terminate */
  792. w = nil;
  793. for(;;){
  794. free(w);
  795. w = wait();
  796. if(w == nil)
  797. return Error;
  798. if(w->pid == pid){
  799. if(w->msg[0] == 0){
  800. free(w);
  801. break;
  802. }
  803. werrstr("xfer: %s", w->msg);
  804. free(w);
  805. return Error;
  806. }
  807. }
  808. switch(rv){
  809. case Success:
  810. return Eof;
  811. case TempFail:
  812. return Server;
  813. default:
  814. return Error;
  815. }
  816. }
  817. int
  818. ftpcmd(int ctl, char *fmt, ...)
  819. {
  820. va_list arg;
  821. char buf[2*1024], *s;
  822. va_start(arg, fmt);
  823. s = vseprint(buf, buf + (sizeof(buf)-4) / sizeof(*buf), fmt, arg);
  824. va_end(arg);
  825. if(debug)
  826. fprint(2, "%d -> %s\n", ctl, buf);
  827. *s++ = '\r';
  828. *s++ = '\n';
  829. if(write(ctl, buf, s - buf) != s - buf)
  830. return -1;
  831. return 0;
  832. }
  833. int
  834. ftprcode(int ctl, char *msg, int len)
  835. {
  836. int rv;
  837. int i;
  838. char *p;
  839. len--; /* room for terminating null */
  840. for(;;){
  841. *msg = 0;
  842. i = readline(ctl, msg, len);
  843. if(i < 0)
  844. break;
  845. if(debug)
  846. fprint(2, "%d <- %s\n", ctl, msg);
  847. /* stop if not a continuation */
  848. rv = strtol(msg, &p, 10);
  849. if(rv >= 100 && rv < 600 && p==msg+3 && *p == ' ')
  850. return rv/100;
  851. }
  852. *msg = 0;
  853. return -1;
  854. }
  855. int
  856. hello(int ctl)
  857. {
  858. char msg[1024];
  859. /* wait for hello from other side */
  860. if(ftprcode(ctl, msg, sizeof(msg)) != Success){
  861. werrstr("HELLO: %s", msg);
  862. return Server;
  863. }
  864. return 0;
  865. }
  866. int
  867. getdec(char *p, int n)
  868. {
  869. int x = 0;
  870. int i;
  871. for(i = 0; i < n; i++)
  872. x = x*10 + (*p++ - '0');
  873. return x;
  874. }
  875. int
  876. ftprestart(int ctl, Out *out, URL *u, Range *r, long mtime)
  877. {
  878. Tm tm;
  879. char msg[1024];
  880. long x, rmtime;
  881. ftpcmd(ctl, "MDTM %s", u->page);
  882. if(ftprcode(ctl, msg, sizeof(msg)) != Success){
  883. r->start = 0;
  884. return 0; /* need to do something */
  885. }
  886. /* decode modification time */
  887. if(strlen(msg) < 4 + 4 + 2 + 2 + 2 + 2 + 2){
  888. r->start = 0;
  889. return 0; /* need to do something */
  890. }
  891. memset(&tm, 0, sizeof(tm));
  892. tm.year = getdec(msg+4, 4) - 1900;
  893. tm.mon = getdec(msg+4+4, 2) - 1;
  894. tm.mday = getdec(msg+4+4+2, 2);
  895. tm.hour = getdec(msg+4+4+2+2, 2);
  896. tm.min = getdec(msg+4+4+2+2+2, 2);
  897. tm.sec = getdec(msg+4+4+2+2+2+2, 2);
  898. strcpy(tm.zone, "GMT");
  899. rmtime = tm2sec(&tm);
  900. if(rmtime > mtime)
  901. r->start = 0;
  902. /* get size */
  903. ftpcmd(ctl, "SIZE %s", u->page);
  904. if(ftprcode(ctl, msg, sizeof(msg)) == Success){
  905. x = atol(msg+4);
  906. if(r->start == x)
  907. return 1; /* we're up to date */
  908. r->end = x;
  909. }
  910. /* seek to restart point */
  911. if(r->start > 0){
  912. ftpcmd(ctl, "REST %lud", r->start);
  913. if(ftprcode(ctl, msg, sizeof(msg)) == Incomplete){
  914. setoffset(out, r->start);
  915. }else
  916. r->start = 0;
  917. }
  918. return 0; /* need to do something */
  919. }
  920. int
  921. logon(int ctl)
  922. {
  923. char msg[1024];
  924. /* login anonymous */
  925. ftpcmd(ctl, "USER anonymous");
  926. switch(ftprcode(ctl, msg, sizeof(msg))){
  927. case Success:
  928. return 0;
  929. case Incomplete:
  930. break; /* need password */
  931. default:
  932. werrstr("USER: %s", msg);
  933. return Server;
  934. }
  935. /* send user id as password */
  936. sprint(msg, "%s@closedmind.org", getuser());
  937. ftpcmd(ctl, "PASS %s", msg);
  938. if(ftprcode(ctl, msg, sizeof(msg)) != Success){
  939. werrstr("PASS: %s", msg);
  940. return Server;
  941. }
  942. return 0;
  943. }
  944. int
  945. xfertype(int ctl, char *t)
  946. {
  947. char msg[1024];
  948. ftpcmd(ctl, "TYPE %s", t);
  949. if(ftprcode(ctl, msg, sizeof(msg)) != Success){
  950. werrstr("TYPE %s: %s", t, msg);
  951. return Server;
  952. }
  953. return 0;
  954. }
  955. int
  956. passive(int ctl, URL *u)
  957. {
  958. char msg[1024];
  959. char ipaddr[32];
  960. char *f[6];
  961. char *p;
  962. int fd;
  963. int port;
  964. char aport[12];
  965. ftpcmd(ctl, "PASV");
  966. if(ftprcode(ctl, msg, sizeof(msg)) != Success)
  967. return Error;
  968. /* get address and port number from reply, this is AI */
  969. p = strchr(msg, '(');
  970. if(p == nil){
  971. for(p = msg+3; *p; p++)
  972. if(isdigit(*p))
  973. break;
  974. } else
  975. p++;
  976. if(getfields(p, f, 6, 0, ",)") < 6){
  977. werrstr("ftp protocol botch");
  978. return Server;
  979. }
  980. snprint(ipaddr, sizeof(ipaddr), "%s.%s.%s.%s",
  981. f[0], f[1], f[2], f[3]);
  982. port = ((atoi(f[4])&0xff)<<8) + (atoi(f[5])&0xff);
  983. sprint(aport, "%d", port);
  984. /* open data connection */
  985. fd = dial(netmkaddr(ipaddr, tcpdir, aport), 0, 0, 0);
  986. if(fd < 0){
  987. werrstr("passive mode failed: %r");
  988. return Error;
  989. }
  990. /* tell remote to send a file */
  991. ftpcmd(ctl, "RETR %s", u->page);
  992. if(ftprcode(ctl, msg, sizeof(msg)) != Extra){
  993. werrstr("RETR %s: %s", u->page, msg);
  994. return Error;
  995. }
  996. return fd;
  997. }
  998. int
  999. active(int ctl, URL *u)
  1000. {
  1001. char msg[1024];
  1002. char dir[40], ldir[40];
  1003. uchar ipaddr[4];
  1004. uchar port[2];
  1005. int lcfd, dfd, afd;
  1006. /* announce a port for the call back */
  1007. snprint(msg, sizeof(msg), "%s!*!0", tcpdir);
  1008. afd = announce(msg, dir);
  1009. if(afd < 0)
  1010. return Error;
  1011. /* get a local address/port of the annoucement */
  1012. if(getaddrport(dir, ipaddr, port) < 0){
  1013. close(afd);
  1014. return Error;
  1015. }
  1016. /* tell remote side address and port*/
  1017. ftpcmd(ctl, "PORT %d,%d,%d,%d,%d,%d", ipaddr[0], ipaddr[1], ipaddr[2],
  1018. ipaddr[3], port[0], port[1]);
  1019. if(ftprcode(ctl, msg, sizeof(msg)) != Success){
  1020. close(afd);
  1021. werrstr("active: %s", msg);
  1022. return Error;
  1023. }
  1024. /* tell remote to send a file */
  1025. ftpcmd(ctl, "RETR %s", u->page);
  1026. if(ftprcode(ctl, msg, sizeof(msg)) != Extra){
  1027. close(afd);
  1028. werrstr("RETR: %s", msg);
  1029. return Server;
  1030. }
  1031. /* wait for a connection */
  1032. lcfd = listen(dir, ldir);
  1033. if(lcfd < 0){
  1034. close(afd);
  1035. return Error;
  1036. }
  1037. dfd = accept(lcfd, ldir);
  1038. if(dfd < 0){
  1039. close(afd);
  1040. close(lcfd);
  1041. return Error;
  1042. }
  1043. close(afd);
  1044. close(lcfd);
  1045. return dfd;
  1046. }
  1047. int
  1048. ftpxfer(int in, Out *out, Range *r)
  1049. {
  1050. char buf[1024];
  1051. long vtime;
  1052. int i, n;
  1053. vtime = 0;
  1054. for(n = 0;;n += i){
  1055. i = read(in, buf, sizeof(buf));
  1056. if(i == 0)
  1057. break;
  1058. if(i < 0)
  1059. return Error;
  1060. if(output(out, buf, i) != i)
  1061. return Error;
  1062. r->start += i;
  1063. if(verbose && (vtime != time(0) || r->start == r->end)) {
  1064. vtime = time(0);
  1065. fprint(2, "%ld %ld\n", r->start, r->end);
  1066. }
  1067. }
  1068. return n;
  1069. }
  1070. int
  1071. terminateftp(int ctl, int rv)
  1072. {
  1073. close(ctl);
  1074. return rv;
  1075. }
  1076. /*
  1077. * case insensitive strcmp (why aren't these in libc?)
  1078. */
  1079. int
  1080. cistrncmp(char *a, char *b, int n)
  1081. {
  1082. while(n-- > 0){
  1083. if(tolower(*a++) != tolower(*b++))
  1084. return -1;
  1085. }
  1086. return 0;
  1087. }
  1088. int
  1089. cistrcmp(char *a, char *b)
  1090. {
  1091. while(*a || *b)
  1092. if(tolower(*a++) != tolower(*b++))
  1093. return -1;
  1094. return 0;
  1095. }
  1096. /*
  1097. * buffered io
  1098. */
  1099. struct
  1100. {
  1101. char *rp;
  1102. char *wp;
  1103. char buf[4*1024];
  1104. } b;
  1105. void
  1106. initibuf(void)
  1107. {
  1108. b.rp = b.wp = b.buf;
  1109. }
  1110. /*
  1111. * read a possibly buffered line, strip off trailing while
  1112. */
  1113. int
  1114. readline(int fd, char *buf, int len)
  1115. {
  1116. int n;
  1117. char *p;
  1118. int eof = 0;
  1119. len--;
  1120. for(p = buf;;){
  1121. if(b.rp >= b.wp){
  1122. n = read(fd, b.wp, sizeof(b.buf)/2);
  1123. if(n < 0)
  1124. return -1;
  1125. if(n == 0){
  1126. eof = 1;
  1127. break;
  1128. }
  1129. b.wp += n;
  1130. }
  1131. n = *b.rp++;
  1132. if(len > 0){
  1133. *p++ = n;
  1134. len--;
  1135. }
  1136. if(n == '\n')
  1137. break;
  1138. }
  1139. /* drop trailing white */
  1140. for(;;){
  1141. if(p <= buf)
  1142. break;
  1143. n = *(p-1);
  1144. if(n != ' ' && n != '\t' && n != '\r' && n != '\n')
  1145. break;
  1146. p--;
  1147. }
  1148. *p = 0;
  1149. if(eof && p == buf)
  1150. return -1;
  1151. return p-buf;
  1152. }
  1153. void
  1154. unreadline(char *line)
  1155. {
  1156. int i, n;
  1157. i = strlen(line);
  1158. n = b.wp-b.rp;
  1159. memmove(&b.buf[i+1], b.rp, n);
  1160. memmove(b.buf, line, i);
  1161. b.buf[i] = '\n';
  1162. b.rp = b.buf;
  1163. b.wp = b.rp + i + 1 + n;
  1164. }
  1165. int
  1166. readibuf(int fd, char *buf, int len)
  1167. {
  1168. int n;
  1169. n = b.wp-b.rp;
  1170. if(n > 0){
  1171. if(n > len)
  1172. n = len;
  1173. memmove(buf, b.rp, n);
  1174. b.rp += n;
  1175. return n;
  1176. }
  1177. return read(fd, buf, len);
  1178. }
  1179. int
  1180. dfprint(int fd, char *fmt, ...)
  1181. {
  1182. char buf[4*1024];
  1183. va_list arg;
  1184. va_start(arg, fmt);
  1185. vseprint(buf, buf+sizeof(buf), fmt, arg);
  1186. va_end(arg);
  1187. if(debug)
  1188. fprint(2, "%d -> %s", fd, buf);
  1189. return fprint(fd, "%s", buf);
  1190. }
  1191. int
  1192. getaddrport(char *dir, uchar *ipaddr, uchar *port)
  1193. {
  1194. char buf[256];
  1195. int fd, i;
  1196. char *p;
  1197. snprint(buf, sizeof(buf), "%s/local", dir);
  1198. fd = open(buf, OREAD);
  1199. if(fd < 0)
  1200. return -1;
  1201. i = read(fd, buf, sizeof(buf)-1);
  1202. close(fd);
  1203. if(i <= 0)
  1204. return -1;
  1205. buf[i] = 0;
  1206. p = strchr(buf, '!');
  1207. if(p != nil)
  1208. *p++ = 0;
  1209. v4parseip(ipaddr, buf);
  1210. i = atoi(p);
  1211. port[0] = i>>8;
  1212. port[1] = i;
  1213. return 0;
  1214. }
  1215. void
  1216. md5free(DigestState *state)
  1217. {
  1218. uchar x[MD5dlen];
  1219. md5(nil, 0, x, state);
  1220. }
  1221. DigestState*
  1222. md5dup(DigestState *state)
  1223. {
  1224. char *p;
  1225. p = md5pickle(state);
  1226. if(p == nil)
  1227. sysfatal("md5pickle: %r");
  1228. state = md5unpickle(p);
  1229. if(state == nil)
  1230. sysfatal("md5unpickle: %r");
  1231. free(p);
  1232. return state;
  1233. }
  1234. void
  1235. setoffset(Out *out, int offset)
  1236. {
  1237. md5free(out->curr);
  1238. if(offset == 0)
  1239. out->curr = md5(nil, 0, nil, nil);
  1240. else
  1241. out->curr = nil;
  1242. out->offset = offset;
  1243. out->written = offset;
  1244. if(ofile != nil)
  1245. if(seek(out->fd, offset, 0) != offset)
  1246. sysfatal("seek: %r");
  1247. }
  1248. /*
  1249. * write some output, discarding it (but keeping track)
  1250. * if we've already written it. if we've gone backwards,
  1251. * verify that everything previously written matches
  1252. * that which would have been written from the current
  1253. * output.
  1254. */
  1255. int
  1256. output(Out *out, char *buf, int nb)
  1257. {
  1258. int n, d;
  1259. uchar m0[MD5dlen], m1[MD5dlen];
  1260. n = nb;
  1261. d = out->written - out->offset;
  1262. assert(d >= 0);
  1263. if(d > 0){
  1264. if(n < d){
  1265. if(out->curr != nil)
  1266. md5((uchar*)buf, n, nil, out->curr);
  1267. out->offset += n;
  1268. return n;
  1269. }
  1270. if(out->curr != nil){
  1271. md5((uchar*)buf, d, m0, out->curr);
  1272. out->curr = nil;
  1273. md5(nil, 0, m1, md5dup(out->hiwat));
  1274. if(memcmp(m0, m1, MD5dlen) != 0){
  1275. fprint(2, "integrity check failure at offset %d\n", out->written);
  1276. return -1;
  1277. }
  1278. }
  1279. buf += d;
  1280. n -= d;
  1281. out->offset += d;
  1282. }
  1283. if(n > 0){
  1284. out->hiwat = md5((uchar*)buf, n, nil, out->hiwat);
  1285. n = write(out->fd, buf, n);
  1286. if(n > 0){
  1287. out->offset += n;
  1288. out->written += n;
  1289. }
  1290. }
  1291. return n + d;
  1292. }