hget.c 25 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962963964965966967968969970971972973974975976977978979980981982983984985986987988989990991992993994995996997998999100010011002100310041005100610071008100910101011101210131014101510161017101810191020102110221023102410251026102710281029103010311032103310341035103610371038103910401041104210431044104510461047104810491050105110521053105410551056105710581059106010611062106310641065106610671068106910701071107210731074107510761077107810791080108110821083108410851086108710881089109010911092109310941095109610971098109911001101110211031104110511061107110811091110111111121113111411151116111711181119112011211122112311241125112611271128112911301131113211331134113511361137113811391140114111421143114411451146114711481149115011511152115311541155115611571158115911601161116211631164116511661167116811691170117111721173117411751176117711781179118011811182118311841185118611871188118911901191119211931194119511961197119811991200120112021203120412051206120712081209121012111212121312141215121612171218121912201221122212231224122512261227122812291230123112321233123412351236123712381239124012411242124312441245124612471248124912501251125212531254125512561257125812591260126112621263126412651266126712681269127012711272127312741275127612771278127912801281128212831284128512861287128812891290129112921293129412951296129712981299130013011302130313041305130613071308130913101311131213131314131513161317131813191320132113221323132413251326132713281329133013311332133313341335133613371338133913401341134213431344134513461347134813491350135113521353135413551356135713581359136013611362136313641365136613671368136913701371137213731374137513761377137813791380138113821383138413851386138713881389139013911392139313941395139613971398139914001401140214031404140514061407140814091410141114121413141414151416141714181419142014211422142314241425142614271428142914301431143214331434143514361437143814391440144114421443144414451446144714481449145014511452145314541455145614571458145914601461146214631464146514661467146814691470147114721473147414751476147714781479
  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 = ARGF();
  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 = ARGF();
  122. if(net == nil)
  123. usage();
  124. break;
  125. case 'p':
  126. t = ARGF();
  127. if(t == nil)
  128. usage();
  129. if(p != postbody)
  130. p = seprint(p, e, "&%s", t);
  131. else
  132. p = seprint(p, e, "%s", t);
  133. u.postbody = postbody;
  134. break;
  135. default:
  136. usage();
  137. } ARGEND;
  138. if(net != nil){
  139. if(strlen(net) > sizeof(tcpdir)-5)
  140. sysfatal("network mount point too long");
  141. snprint(tcpdir, sizeof(tcpdir), "%s/tcp", net);
  142. } else
  143. snprint(tcpdir, sizeof(tcpdir), "tcp");
  144. if(argc != 1)
  145. usage();
  146. out.fd = 1;
  147. out.written = 0;
  148. out.offset = 0;
  149. out.curr = nil;
  150. out.hiwat = nil;
  151. if(ofile != nil){
  152. d = dirstat(ofile);
  153. if(d == nil){
  154. out.fd = create(ofile, OWRITE, 0664);
  155. if(out.fd < 0)
  156. sysfatal("creating %s: %r", ofile);
  157. } else {
  158. out.fd = open(ofile, OWRITE);
  159. if(out.fd < 0)
  160. sysfatal("can't open %s: %r", ofile);
  161. r.start = d->length;
  162. mtime = d->mtime;
  163. free(d);
  164. }
  165. }
  166. errs = 0;
  167. if(crackurl(&u, argv[0]) < 0)
  168. sysfatal("%r");
  169. if(hpx && crackurl(&px, hpx) < 0)
  170. sysfatal("%r");
  171. for(;;){
  172. setoffset(&out, 0);
  173. /* transfer data */
  174. werrstr("");
  175. n = (*method[u.method].f)(&u, &px, &r, &out, mtime);
  176. switch(n){
  177. case Eof:
  178. exits(0);
  179. break;
  180. case Error:
  181. if(errs++ < 10)
  182. continue;
  183. sysfatal("too many errors with no progress %r");
  184. break;
  185. case Server:
  186. sysfatal("server returned: %r");
  187. break;
  188. }
  189. /* forward progress */
  190. errs = 0;
  191. r.start += n;
  192. if(r.start >= r.end)
  193. break;
  194. }
  195. exits(0);
  196. }
  197. int
  198. crackurl(URL *u, char *s)
  199. {
  200. char *p;
  201. int i;
  202. if(u->page != nil){
  203. free(u->page);
  204. u->page = nil;
  205. }
  206. /* get type */
  207. for(p = s; *p; p++){
  208. if(*p == '/'){
  209. p = s;
  210. if(u->method == Other){
  211. werrstr("missing method");
  212. return -1;
  213. }
  214. if(u->host == nil){
  215. werrstr("missing host");
  216. return -1;
  217. }
  218. u->page = strdup(p);
  219. return 0;
  220. }
  221. if(*p == ':' && *(p+1)=='/' && *(p+2)=='/'){
  222. *p = 0;
  223. p += 3;
  224. for(i = 0; i < nelem(method); i++){
  225. if(cistrcmp(s, method[i].name) == 0){
  226. u->method = i;
  227. break;
  228. }
  229. }
  230. break;
  231. }
  232. }
  233. if(u->method == Other){
  234. werrstr("unsupported URL type %s", s);
  235. return -1;
  236. }
  237. /* get system */
  238. free(u->host);
  239. s = p;
  240. p = strchr(s, '/');
  241. if(p == nil){
  242. u->host = strdup(s);
  243. u->page = strdup("/");
  244. } else {
  245. u->page = strdup(p);
  246. *p = 0;
  247. u->host = strdup(s);
  248. *p = '/';
  249. }
  250. if(p = strchr(u->host, ':')) {
  251. *p++ = 0;
  252. u->port = p;
  253. } else
  254. u->port = method[u->method].name;
  255. if(*(u->host) == 0){
  256. werrstr("bad url, null host");
  257. return -1;
  258. }
  259. return 0;
  260. }
  261. char *day[] = {
  262. "Sun", "Mon", "Tue", "Wed", "Thu", "Fri", "Sat"
  263. };
  264. char *month[] = {
  265. "Jan", "Feb", "Mar", "Apr", "May", "Jun", "Jul", "Aug", "Sep", "Oct", "Nov", "Dec"
  266. };
  267. struct
  268. {
  269. int fd;
  270. long mtime;
  271. } note;
  272. void
  273. catch(void*, char*)
  274. {
  275. Dir d;
  276. nulldir(&d);
  277. d.mtime = note.mtime;
  278. if(dirfwstat(note.fd, &d) < 0)
  279. sysfatal("catch: can't dirfwstat: %r");
  280. noted(NDFLT);
  281. }
  282. int
  283. dohttp(URL *u, URL *px, Range *r, Out *out, long mtime)
  284. {
  285. int fd, cfd;
  286. int redirect, auth, loop;
  287. int n, rv, code;
  288. long tot, vtime;
  289. Tm *tm;
  290. char buf[1024];
  291. char err[ERRMAX];
  292. /* always move back to a previous 512 byte bound because some
  293. * servers can't seem to deal with requests that start at the
  294. * end of the file
  295. */
  296. if(r->start)
  297. r->start = ((r->start-1)/512)*512;
  298. /* loop for redirects, requires reading both response code and headers */
  299. fd = -1;
  300. for(loop = 0; loop < 32; loop++){
  301. if(px->host == nil){
  302. fd = dial(netmkaddr(u->host, tcpdir, u->port), 0, 0, 0);
  303. } else {
  304. fd = dial(netmkaddr(px->host, tcpdir, px->port), 0, 0, 0);
  305. }
  306. if(fd < 0)
  307. return Error;
  308. if(u->method == Https){
  309. int tfd;
  310. TLSconn conn;
  311. memset(&conn, 0, sizeof conn);
  312. tfd = tlsClient(fd, &conn);
  313. if(tfd < 0){
  314. fprint(2, "tlsClient: %r\n");
  315. close(fd);
  316. return Error;
  317. }
  318. /* BUG: check cert here? */
  319. if(conn.cert)
  320. free(conn.cert);
  321. close(fd);
  322. fd = tfd;
  323. }
  324. /* write request, use range if not start of file */
  325. if(u->postbody == nil){
  326. if(px->host == nil){
  327. dfprint(fd, "GET %s HTTP/1.0\r\n"
  328. "Host: %s\r\n"
  329. "User-agent: Plan9/hget\r\n"
  330. "Cache-Control: no-cache\r\n"
  331. "Pragma: no-cache\r\n",
  332. u->page, u->host);
  333. } else {
  334. dfprint(fd, "GET http://%s%s HTTP/1.0\r\n"
  335. "Host: %s\r\n"
  336. "User-agent: Plan9/hget\r\n"
  337. "Cache-Control: no-cache\r\n"
  338. "Pragma: no-cache\r\n",
  339. u->host, u->page, u->host);
  340. }
  341. if(u->cred)
  342. dfprint(fd, "Authorization: Basic %s\r\n",
  343. u->cred);
  344. } else {
  345. dfprint(fd, "POST %s HTTP/1.0\r\n"
  346. "Host: %s\r\n"
  347. "Content-type: application/x-www-form-urlencoded\r\n"
  348. "Content-length: %d\r\n"
  349. "User-agent: Plan9/hget\r\n",
  350. u->page, u->host, strlen(u->postbody));
  351. if(u->cred)
  352. dfprint(fd, "Authorization: Basic %s\r\n", u->cred);
  353. }
  354. if(r->start != 0){
  355. dfprint(fd, "Range: bytes=%d-\n", r->start);
  356. if(u->etag != nil){
  357. dfprint(fd, "If-range: %s\n", u->etag);
  358. } else {
  359. tm = gmtime(mtime);
  360. dfprint(fd, "If-range: %s, %d %s %d %2d:%2.2d:%2.2d GMT\n",
  361. day[tm->wday], tm->mday, month[tm->mon],
  362. tm->year+1900, tm->hour, tm->min, tm->sec);
  363. }
  364. }
  365. if((cfd = open("/mnt/webcookies/http", ORDWR)) >= 0){
  366. if(fprint(cfd, "http://%s%s", u->host, u->page) > 0){
  367. while((n = read(cfd, buf, sizeof buf)) > 0){
  368. if(debug)
  369. write(2, buf, n);
  370. write(fd, buf, n);
  371. }
  372. }else{
  373. close(cfd);
  374. cfd = -1;
  375. }
  376. }
  377. dfprint(fd, "\r\n", u->host);
  378. if(u->postbody)
  379. dfprint(fd, "%s", u->postbody);
  380. auth = 0;
  381. redirect = 0;
  382. initibuf();
  383. code = httprcode(fd);
  384. switch(code){
  385. case Error: /* connection timed out */
  386. case Eof:
  387. close(fd);
  388. close(cfd);
  389. return code;
  390. case 200: /* OK */
  391. case 201: /* Created */
  392. case 202: /* Accepted */
  393. if(ofile == nil && r->start != 0)
  394. sysfatal("page changed underfoot");
  395. break;
  396. case 204: /* No Content */
  397. sysfatal("No Content");
  398. case 206: /* Partial Content */
  399. setoffset(out, r->start);
  400. break;
  401. case 301: /* Moved Permanently */
  402. case 302: /* Moved Temporarily */
  403. redirect = 1;
  404. u->postbody = nil;
  405. break;
  406. case 304: /* Not Modified */
  407. break;
  408. case 400: /* Bad Request */
  409. sysfatal("Bad Request");
  410. case 401: /* Unauthorized */
  411. if (auth)
  412. sysfatal("Authentication failed");
  413. auth = 1;
  414. break;
  415. case 402: /* ??? */
  416. sysfatal("Unauthorized");
  417. case 403: /* Forbidden */
  418. sysfatal("Forbidden by server");
  419. case 404: /* Not Found */
  420. sysfatal("Not found on server");
  421. case 407: /* Proxy Authentication */
  422. sysfatal("Proxy authentication required");
  423. case 500: /* Internal server error */
  424. sysfatal("Server choked");
  425. case 501: /* Not implemented */
  426. sysfatal("Server can't do it!");
  427. case 502: /* Bad gateway */
  428. sysfatal("Bad gateway");
  429. case 503: /* Service unavailable */
  430. sysfatal("Service unavailable");
  431. default:
  432. sysfatal("Unknown response code %d", code);
  433. }
  434. if(u->redirect != nil){
  435. free(u->redirect);
  436. u->redirect = nil;
  437. }
  438. rv = httpheaders(fd, cfd, u, r);
  439. close(cfd);
  440. if(rv != 0){
  441. close(fd);
  442. return rv;
  443. }
  444. if(!redirect && !auth)
  445. break;
  446. if (redirect){
  447. if(u->redirect == nil)
  448. sysfatal("redirect: no URL");
  449. if(crackurl(u, u->redirect) < 0)
  450. sysfatal("redirect: %r");
  451. }
  452. }
  453. /* transfer whatever you get */
  454. if(ofile != nil && u->mtime != 0){
  455. note.fd = out->fd;
  456. note.mtime = u->mtime;
  457. notify(catch);
  458. }
  459. tot = 0;
  460. vtime = 0;
  461. for(;;){
  462. n = readibuf(fd, buf, sizeof(buf));
  463. if(n <= 0)
  464. break;
  465. if(output(out, buf, n) != n)
  466. break;
  467. tot += n;
  468. if(verbose && vtime != time(0)) {
  469. vtime = time(0);
  470. fprint(2, "%ld %ld\n", r->start+tot, r->end);
  471. }
  472. }
  473. notify(nil);
  474. close(fd);
  475. if(ofile != nil && u->mtime != 0){
  476. Dir d;
  477. rerrstr(err, sizeof err);
  478. nulldir(&d);
  479. d.mtime = u->mtime;
  480. if(dirfwstat(out->fd, &d) < 0)
  481. fprint(2, "couldn't set mtime: %r\n");
  482. errstr(err, sizeof err);
  483. }
  484. return tot;
  485. }
  486. /* get the http response code */
  487. int
  488. httprcode(int fd)
  489. {
  490. int n;
  491. char *p;
  492. char buf[256];
  493. n = readline(fd, buf, sizeof(buf)-1);
  494. if(n <= 0)
  495. return n;
  496. if(debug)
  497. fprint(2, "%d <- %s\n", fd, buf);
  498. p = strchr(buf, ' ');
  499. if(strncmp(buf, "HTTP/", 5) != 0 || p == nil){
  500. werrstr("bad response from server");
  501. return -1;
  502. }
  503. buf[n] = 0;
  504. return atoi(p+1);
  505. }
  506. /* read in and crack the http headers, update u and r */
  507. void hhetag(char*, URL*, Range*);
  508. void hhmtime(char*, URL*, Range*);
  509. void hhclen(char*, URL*, Range*);
  510. void hhcrange(char*, URL*, Range*);
  511. void hhuri(char*, URL*, Range*);
  512. void hhlocation(char*, URL*, Range*);
  513. void hhauth(char*, URL*, Range*);
  514. struct {
  515. char *name;
  516. void (*f)(char*, URL*, Range*);
  517. } headers[] = {
  518. { "etag:", hhetag },
  519. { "last-modified:", hhmtime },
  520. { "content-length:", hhclen },
  521. { "content-range:", hhcrange },
  522. { "uri:", hhuri },
  523. { "location:", hhlocation },
  524. { "WWW-Authenticate:", hhauth },
  525. };
  526. int
  527. httpheaders(int fd, int cfd, URL *u, Range *r)
  528. {
  529. char buf[2048];
  530. char *p;
  531. int i, n;
  532. for(;;){
  533. n = getheader(fd, buf, sizeof(buf));
  534. if(n <= 0)
  535. break;
  536. if(cfd >= 0)
  537. fprint(cfd, "%s\n", buf);
  538. for(i = 0; i < nelem(headers); i++){
  539. n = strlen(headers[i].name);
  540. if(cistrncmp(buf, headers[i].name, n) == 0){
  541. /* skip field name and leading white */
  542. p = buf + n;
  543. while(*p == ' ' || *p == '\t')
  544. p++;
  545. (*headers[i].f)(p, u, r);
  546. break;
  547. }
  548. }
  549. }
  550. return n;
  551. }
  552. /*
  553. * read a single mime header, collect continuations.
  554. *
  555. * this routine assumes that there is a blank line twixt
  556. * the header and the message body, otherwise bytes will
  557. * be lost.
  558. */
  559. int
  560. getheader(int fd, char *buf, int n)
  561. {
  562. char *p, *e;
  563. int i;
  564. n--;
  565. p = buf;
  566. for(e = p + n; ; p += i){
  567. i = readline(fd, p, e-p);
  568. if(i < 0)
  569. return i;
  570. if(p == buf){
  571. /* first line */
  572. if(strchr(buf, ':') == nil)
  573. break; /* end of headers */
  574. } else {
  575. /* continuation line */
  576. if(*p != ' ' && *p != '\t'){
  577. unreadline(p);
  578. *p = 0;
  579. break; /* end of this header */
  580. }
  581. }
  582. }
  583. if(headerprint)
  584. print("%s\n", buf);
  585. if(debug)
  586. fprint(2, "%d <- %s\n", fd, buf);
  587. return p-buf;
  588. }
  589. void
  590. hhetag(char *p, URL *u, Range*)
  591. {
  592. if(u->etag != nil){
  593. if(strcmp(u->etag, p) != 0)
  594. sysfatal("file changed underfoot");
  595. } else
  596. u->etag = strdup(p);
  597. }
  598. char* monthchars = "janfebmaraprmayjunjulaugsepoctnovdec";
  599. void
  600. hhmtime(char *p, URL *u, Range*)
  601. {
  602. char *month, *day, *yr, *hms;
  603. char *fields[6];
  604. Tm tm, now;
  605. int i;
  606. i = getfields(p, fields, 6, 1, " \t");
  607. if(i < 5)
  608. return;
  609. day = fields[1];
  610. month = fields[2];
  611. yr = fields[3];
  612. hms = fields[4];
  613. /* default time */
  614. now = *gmtime(time(0));
  615. tm = now;
  616. tm.yday = 0;
  617. /* convert ascii month to a number twixt 1 and 12 */
  618. if(*month >= '0' && *month <= '9'){
  619. tm.mon = atoi(month) - 1;
  620. if(tm.mon < 0 || tm.mon > 11)
  621. tm.mon = 5;
  622. } else {
  623. for(p = month; *p; p++)
  624. *p = tolower(*p);
  625. for(i = 0; i < 12; i++)
  626. if(strncmp(&monthchars[i*3], month, 3) == 0){
  627. tm.mon = i;
  628. break;
  629. }
  630. }
  631. tm.mday = atoi(day);
  632. if(hms) {
  633. tm.hour = strtoul(hms, &p, 10);
  634. if(*p == ':') {
  635. p++;
  636. tm.min = strtoul(p, &p, 10);
  637. if(*p == ':') {
  638. p++;
  639. tm.sec = strtoul(p, &p, 10);
  640. }
  641. }
  642. if(tolower(*p) == 'p')
  643. tm.hour += 12;
  644. }
  645. if(yr) {
  646. tm.year = atoi(yr);
  647. if(tm.year >= 1900)
  648. tm.year -= 1900;
  649. } else {
  650. if(tm.mon > now.mon || (tm.mon == now.mon && tm.mday > now.mday+1))
  651. tm.year--;
  652. }
  653. strcpy(tm.zone, "GMT");
  654. /* convert to epoch seconds */
  655. u->mtime = tm2sec(&tm);
  656. }
  657. void
  658. hhclen(char *p, URL*, Range *r)
  659. {
  660. r->end = atoi(p);
  661. }
  662. void
  663. hhcrange(char *p, URL*, Range *r)
  664. {
  665. char *x;
  666. vlong l;
  667. l = 0;
  668. x = strchr(p, '/');
  669. if(x)
  670. l = atoll(x+1);
  671. if(l == 0) {
  672. x = strchr(p, '-');
  673. if(x)
  674. l = atoll(x+1);
  675. }
  676. if(l)
  677. r->end = l;
  678. }
  679. void
  680. hhuri(char *p, URL *u, Range*)
  681. {
  682. if(*p != '<')
  683. return;
  684. u->redirect = strdup(p+1);
  685. p = strchr(u->redirect, '>');
  686. if(p != nil)
  687. *p = 0;
  688. }
  689. void
  690. hhlocation(char *p, URL *u, Range*)
  691. {
  692. u->redirect = strdup(p);
  693. }
  694. void
  695. hhauth(char *p, URL *u, Range*)
  696. {
  697. char *f[4];
  698. UserPasswd *up;
  699. char *s, cred[64];
  700. if (cistrncmp(p, "basic ", 6) != 0)
  701. sysfatal("only Basic authentication supported");
  702. if (gettokens(p, f, nelem(f), "\"") < 2)
  703. sysfatal("garbled auth data");
  704. if ((up = auth_getuserpasswd(auth_getkey, "proto=pass service=http server=%q realm=%q",
  705. u->host, f[1])) == nil)
  706. sysfatal("cannot authenticate");
  707. s = smprint("%s:%s", up->user, up->passwd);
  708. if(enc64(cred, sizeof(cred), (uchar *)s, strlen(s)) == -1)
  709. sysfatal("enc64");
  710. free(s);
  711. assert(u->cred = strdup(cred));
  712. }
  713. enum
  714. {
  715. /* ftp return codes */
  716. Extra= 1,
  717. Success= 2,
  718. Incomplete= 3,
  719. TempFail= 4,
  720. PermFail= 5,
  721. Nnetdir= 64, /* max length of network directory paths */
  722. Ndialstr= 64, /* max length of dial strings */
  723. };
  724. int ftpcmd(int, char*, ...);
  725. int ftprcode(int, char*, int);
  726. int hello(int);
  727. int logon(int);
  728. int xfertype(int, char*);
  729. int passive(int, URL*);
  730. int active(int, URL*);
  731. int ftpxfer(int, Out*, Range*);
  732. int terminateftp(int, int);
  733. int getaddrport(char*, uchar*, uchar*);
  734. int ftprestart(int, Out*, URL*, Range*, long);
  735. int
  736. doftp(URL *u, URL *px, Range *r, Out *out, long mtime)
  737. {
  738. int pid, ctl, data, rv;
  739. Waitmsg *w;
  740. char msg[64];
  741. char conndir[NETPATHLEN];
  742. char *p;
  743. /* untested, proxy doesn't work with ftp (I think) */
  744. if(px->host == nil){
  745. ctl = dial(netmkaddr(u->host, tcpdir, u->port), 0, conndir, 0);
  746. } else {
  747. ctl = dial(netmkaddr(px->host, tcpdir, px->port), 0, conndir, 0);
  748. }
  749. if(ctl < 0)
  750. return Error;
  751. if(net == nil){
  752. p = strrchr(conndir, '/');
  753. *p = 0;
  754. snprint(tcpdir, sizeof(tcpdir), conndir);
  755. }
  756. initibuf();
  757. rv = hello(ctl);
  758. if(rv < 0)
  759. return terminateftp(ctl, rv);
  760. rv = logon(ctl);
  761. if(rv < 0)
  762. return terminateftp(ctl, rv);
  763. rv = xfertype(ctl, "I");
  764. if(rv < 0)
  765. return terminateftp(ctl, rv);
  766. /* if file is up to date and the right size, stop */
  767. if(ftprestart(ctl, out, u, r, mtime) > 0){
  768. close(ctl);
  769. return Eof;
  770. }
  771. /* first try passive mode, then active */
  772. data = passive(ctl, u);
  773. if(data < 0){
  774. data = active(ctl, u);
  775. if(data < 0)
  776. return Error;
  777. }
  778. /* fork */
  779. switch(pid = rfork(RFPROC|RFFDG|RFMEM)){
  780. case -1:
  781. close(data);
  782. return terminateftp(ctl, Error);
  783. case 0:
  784. ftpxfer(data, out, r);
  785. close(data);
  786. _exits(0);
  787. default:
  788. close(data);
  789. break;
  790. }
  791. /* wait for reply message */
  792. rv = ftprcode(ctl, msg, sizeof(msg));
  793. close(ctl);
  794. /* wait for process to terminate */
  795. w = nil;
  796. for(;;){
  797. free(w);
  798. w = wait();
  799. if(w == nil)
  800. return Error;
  801. if(w->pid == pid){
  802. if(w->msg[0] == 0){
  803. free(w);
  804. break;
  805. }
  806. werrstr("xfer: %s", w->msg);
  807. free(w);
  808. return Error;
  809. }
  810. }
  811. switch(rv){
  812. case Success:
  813. return Eof;
  814. case TempFail:
  815. return Server;
  816. default:
  817. return Error;
  818. }
  819. }
  820. int
  821. ftpcmd(int ctl, char *fmt, ...)
  822. {
  823. va_list arg;
  824. char buf[2*1024], *s;
  825. va_start(arg, fmt);
  826. s = vseprint(buf, buf + (sizeof(buf)-4) / sizeof(*buf), fmt, arg);
  827. va_end(arg);
  828. if(debug)
  829. fprint(2, "%d -> %s\n", ctl, buf);
  830. *s++ = '\r';
  831. *s++ = '\n';
  832. if(write(ctl, buf, s - buf) != s - buf)
  833. return -1;
  834. return 0;
  835. }
  836. int
  837. ftprcode(int ctl, char *msg, int len)
  838. {
  839. int rv;
  840. int i;
  841. char *p;
  842. len--; /* room for terminating null */
  843. for(;;){
  844. *msg = 0;
  845. i = readline(ctl, msg, len);
  846. if(i < 0)
  847. break;
  848. if(debug)
  849. fprint(2, "%d <- %s\n", ctl, msg);
  850. /* stop if not a continuation */
  851. rv = strtol(msg, &p, 10);
  852. if(rv >= 100 && rv < 600 && p==msg+3 && *p == ' ')
  853. return rv/100;
  854. }
  855. *msg = 0;
  856. return -1;
  857. }
  858. int
  859. hello(int ctl)
  860. {
  861. char msg[1024];
  862. /* wait for hello from other side */
  863. if(ftprcode(ctl, msg, sizeof(msg)) != Success){
  864. werrstr("HELLO: %s", msg);
  865. return Server;
  866. }
  867. return 0;
  868. }
  869. int
  870. getdec(char *p, int n)
  871. {
  872. int x = 0;
  873. int i;
  874. for(i = 0; i < n; i++)
  875. x = x*10 + (*p++ - '0');
  876. return x;
  877. }
  878. int
  879. ftprestart(int ctl, Out *out, URL *u, Range *r, long mtime)
  880. {
  881. Tm tm;
  882. char msg[1024];
  883. long x, rmtime;
  884. ftpcmd(ctl, "MDTM %s", u->page);
  885. if(ftprcode(ctl, msg, sizeof(msg)) != Success){
  886. r->start = 0;
  887. return 0; /* need to do something */
  888. }
  889. /* decode modification time */
  890. if(strlen(msg) < 4 + 4 + 2 + 2 + 2 + 2 + 2){
  891. r->start = 0;
  892. return 0; /* need to do something */
  893. }
  894. memset(&tm, 0, sizeof(tm));
  895. tm.year = getdec(msg+4, 4) - 1900;
  896. tm.mon = getdec(msg+4+4, 2) - 1;
  897. tm.mday = getdec(msg+4+4+2, 2);
  898. tm.hour = getdec(msg+4+4+2+2, 2);
  899. tm.min = getdec(msg+4+4+2+2+2, 2);
  900. tm.sec = getdec(msg+4+4+2+2+2+2, 2);
  901. strcpy(tm.zone, "GMT");
  902. rmtime = tm2sec(&tm);
  903. if(rmtime > mtime)
  904. r->start = 0;
  905. /* get size */
  906. ftpcmd(ctl, "SIZE %s", u->page);
  907. if(ftprcode(ctl, msg, sizeof(msg)) == Success){
  908. x = atol(msg+4);
  909. if(r->start == x)
  910. return 1; /* we're up to date */
  911. r->end = x;
  912. }
  913. /* seek to restart point */
  914. if(r->start > 0){
  915. ftpcmd(ctl, "REST %lud", r->start);
  916. if(ftprcode(ctl, msg, sizeof(msg)) == Incomplete){
  917. setoffset(out, r->start);
  918. }else
  919. r->start = 0;
  920. }
  921. return 0; /* need to do something */
  922. }
  923. int
  924. logon(int ctl)
  925. {
  926. char msg[1024];
  927. /* login anonymous */
  928. ftpcmd(ctl, "USER anonymous");
  929. switch(ftprcode(ctl, msg, sizeof(msg))){
  930. case Success:
  931. return 0;
  932. case Incomplete:
  933. break; /* need password */
  934. default:
  935. werrstr("USER: %s", msg);
  936. return Server;
  937. }
  938. /* send user id as password */
  939. sprint(msg, "%s@closedmind.org", getuser());
  940. ftpcmd(ctl, "PASS %s", msg);
  941. if(ftprcode(ctl, msg, sizeof(msg)) != Success){
  942. werrstr("PASS: %s", msg);
  943. return Server;
  944. }
  945. return 0;
  946. }
  947. int
  948. xfertype(int ctl, char *t)
  949. {
  950. char msg[1024];
  951. ftpcmd(ctl, "TYPE %s", t);
  952. if(ftprcode(ctl, msg, sizeof(msg)) != Success){
  953. werrstr("TYPE %s: %s", t, msg);
  954. return Server;
  955. }
  956. return 0;
  957. }
  958. int
  959. passive(int ctl, URL *u)
  960. {
  961. char msg[1024];
  962. char ipaddr[32];
  963. char *f[6];
  964. char *p;
  965. int fd;
  966. int port;
  967. char aport[12];
  968. ftpcmd(ctl, "PASV");
  969. if(ftprcode(ctl, msg, sizeof(msg)) != Success)
  970. return Error;
  971. /* get address and port number from reply, this is AI */
  972. p = strchr(msg, '(');
  973. if(p == nil){
  974. for(p = msg+3; *p; p++)
  975. if(isdigit(*p))
  976. break;
  977. } else
  978. p++;
  979. if(getfields(p, f, 6, 0, ",)") < 6){
  980. werrstr("ftp protocol botch");
  981. return Server;
  982. }
  983. snprint(ipaddr, sizeof(ipaddr), "%s.%s.%s.%s",
  984. f[0], f[1], f[2], f[3]);
  985. port = ((atoi(f[4])&0xff)<<8) + (atoi(f[5])&0xff);
  986. sprint(aport, "%d", port);
  987. /* open data connection */
  988. fd = dial(netmkaddr(ipaddr, tcpdir, aport), 0, 0, 0);
  989. if(fd < 0){
  990. werrstr("passive mode failed: %r");
  991. return Error;
  992. }
  993. /* tell remote to send a file */
  994. ftpcmd(ctl, "RETR %s", u->page);
  995. if(ftprcode(ctl, msg, sizeof(msg)) != Extra){
  996. werrstr("RETR %s: %s", u->page, msg);
  997. return Error;
  998. }
  999. return fd;
  1000. }
  1001. int
  1002. active(int ctl, URL *u)
  1003. {
  1004. char msg[1024];
  1005. char dir[40], ldir[40];
  1006. uchar ipaddr[4];
  1007. uchar port[2];
  1008. int lcfd, dfd, afd;
  1009. /* announce a port for the call back */
  1010. snprint(msg, sizeof(msg), "%s!*!0", tcpdir);
  1011. afd = announce(msg, dir);
  1012. if(afd < 0)
  1013. return Error;
  1014. /* get a local address/port of the annoucement */
  1015. if(getaddrport(dir, ipaddr, port) < 0){
  1016. close(afd);
  1017. return Error;
  1018. }
  1019. /* tell remote side address and port*/
  1020. ftpcmd(ctl, "PORT %d,%d,%d,%d,%d,%d", ipaddr[0], ipaddr[1], ipaddr[2],
  1021. ipaddr[3], port[0], port[1]);
  1022. if(ftprcode(ctl, msg, sizeof(msg)) != Success){
  1023. close(afd);
  1024. werrstr("active: %s", msg);
  1025. return Error;
  1026. }
  1027. /* tell remote to send a file */
  1028. ftpcmd(ctl, "RETR %s", u->page);
  1029. if(ftprcode(ctl, msg, sizeof(msg)) != Extra){
  1030. close(afd);
  1031. werrstr("RETR: %s", msg);
  1032. return Server;
  1033. }
  1034. /* wait for a connection */
  1035. lcfd = listen(dir, ldir);
  1036. if(lcfd < 0){
  1037. close(afd);
  1038. return Error;
  1039. }
  1040. dfd = accept(lcfd, ldir);
  1041. if(dfd < 0){
  1042. close(afd);
  1043. close(lcfd);
  1044. return Error;
  1045. }
  1046. close(afd);
  1047. close(lcfd);
  1048. return dfd;
  1049. }
  1050. int
  1051. ftpxfer(int in, Out *out, Range *r)
  1052. {
  1053. char buf[1024];
  1054. long vtime;
  1055. int i, n;
  1056. vtime = 0;
  1057. for(n = 0;;n += i){
  1058. i = read(in, buf, sizeof(buf));
  1059. if(i == 0)
  1060. break;
  1061. if(i < 0)
  1062. return Error;
  1063. if(output(out, buf, i) != i)
  1064. return Error;
  1065. r->start += i;
  1066. if(verbose && vtime != time(0)) {
  1067. vtime = time(0);
  1068. fprint(2, "%ld %ld\n", r->start, r->end);
  1069. }
  1070. }
  1071. return n;
  1072. }
  1073. int
  1074. terminateftp(int ctl, int rv)
  1075. {
  1076. close(ctl);
  1077. return rv;
  1078. }
  1079. /*
  1080. * case insensitive strcmp (why aren't these in libc?)
  1081. */
  1082. int
  1083. cistrncmp(char *a, char *b, int n)
  1084. {
  1085. while(n-- > 0){
  1086. if(tolower(*a++) != tolower(*b++))
  1087. return -1;
  1088. }
  1089. return 0;
  1090. }
  1091. int
  1092. cistrcmp(char *a, char *b)
  1093. {
  1094. while(*a || *b)
  1095. if(tolower(*a++) != tolower(*b++))
  1096. return -1;
  1097. return 0;
  1098. }
  1099. /*
  1100. * buffered io
  1101. */
  1102. struct
  1103. {
  1104. char *rp;
  1105. char *wp;
  1106. char buf[4*1024];
  1107. } b;
  1108. void
  1109. initibuf(void)
  1110. {
  1111. b.rp = b.wp = b.buf;
  1112. }
  1113. /*
  1114. * read a possibly buffered line, strip off trailing while
  1115. */
  1116. int
  1117. readline(int fd, char *buf, int len)
  1118. {
  1119. int n;
  1120. char *p;
  1121. int eof = 0;
  1122. len--;
  1123. for(p = buf;;){
  1124. if(b.rp >= b.wp){
  1125. n = read(fd, b.wp, sizeof(b.buf)/2);
  1126. if(n < 0)
  1127. return -1;
  1128. if(n == 0){
  1129. eof = 1;
  1130. break;
  1131. }
  1132. b.wp += n;
  1133. }
  1134. n = *b.rp++;
  1135. if(len > 0){
  1136. *p++ = n;
  1137. len--;
  1138. }
  1139. if(n == '\n')
  1140. break;
  1141. }
  1142. /* drop trailing white */
  1143. for(;;){
  1144. if(p <= buf)
  1145. break;
  1146. n = *(p-1);
  1147. if(n != ' ' && n != '\t' && n != '\r' && n != '\n')
  1148. break;
  1149. p--;
  1150. }
  1151. *p = 0;
  1152. if(eof && p == buf)
  1153. return -1;
  1154. return p-buf;
  1155. }
  1156. void
  1157. unreadline(char *line)
  1158. {
  1159. int i, n;
  1160. i = strlen(line);
  1161. n = b.wp-b.rp;
  1162. memmove(&b.buf[i+1], b.rp, n);
  1163. memmove(b.buf, line, i);
  1164. b.buf[i] = '\n';
  1165. b.rp = b.buf;
  1166. b.wp = b.rp + i + 1 + n;
  1167. }
  1168. int
  1169. readibuf(int fd, char *buf, int len)
  1170. {
  1171. int n;
  1172. n = b.wp-b.rp;
  1173. if(n > 0){
  1174. if(n > len)
  1175. n = len;
  1176. memmove(buf, b.rp, n);
  1177. b.rp += n;
  1178. return n;
  1179. }
  1180. return read(fd, buf, len);
  1181. }
  1182. int
  1183. dfprint(int fd, char *fmt, ...)
  1184. {
  1185. char buf[4*1024];
  1186. va_list arg;
  1187. va_start(arg, fmt);
  1188. vseprint(buf, buf+sizeof(buf), fmt, arg);
  1189. va_end(arg);
  1190. if(debug)
  1191. fprint(2, "%d -> %s", fd, buf);
  1192. return fprint(fd, "%s", buf);
  1193. }
  1194. int
  1195. getaddrport(char *dir, uchar *ipaddr, uchar *port)
  1196. {
  1197. char buf[256];
  1198. int fd, i;
  1199. char *p;
  1200. snprint(buf, sizeof(buf), "%s/local", dir);
  1201. fd = open(buf, OREAD);
  1202. if(fd < 0)
  1203. return -1;
  1204. i = read(fd, buf, sizeof(buf)-1);
  1205. close(fd);
  1206. if(i <= 0)
  1207. return -1;
  1208. buf[i] = 0;
  1209. p = strchr(buf, '!');
  1210. if(p != nil)
  1211. *p++ = 0;
  1212. v4parseip(ipaddr, buf);
  1213. i = atoi(p);
  1214. port[0] = i>>8;
  1215. port[1] = i;
  1216. return 0;
  1217. }
  1218. void
  1219. md5free(DigestState *state)
  1220. {
  1221. uchar x[MD5dlen];
  1222. md5(nil, 0, x, state);
  1223. }
  1224. DigestState*
  1225. md5dup(DigestState *state)
  1226. {
  1227. char *p;
  1228. p = md5pickle(state);
  1229. if(p == nil)
  1230. sysfatal("md5pickle: %r");
  1231. state = md5unpickle(p);
  1232. if(state == nil)
  1233. sysfatal("md5unpickle: %r");
  1234. free(p);
  1235. return state;
  1236. }
  1237. void
  1238. setoffset(Out *out, int offset)
  1239. {
  1240. md5free(out->curr);
  1241. if(offset == 0)
  1242. out->curr = md5(nil, 0, nil, nil);
  1243. else
  1244. out->curr = nil;
  1245. out->offset = offset;
  1246. out->written = offset;
  1247. if(ofile != nil)
  1248. if(seek(out->fd, offset, 0) != offset)
  1249. sysfatal("seek: %r");
  1250. }
  1251. /*
  1252. * write some output, discarding it (but keeping track)
  1253. * if we've already written it. if we've gone backwards,
  1254. * verify that everything previously written matches
  1255. * that which would have been written from the current
  1256. * output.
  1257. */
  1258. int
  1259. output(Out *out, char *buf, int nb)
  1260. {
  1261. int n, d;
  1262. uchar m0[MD5dlen], m1[MD5dlen];
  1263. n = nb;
  1264. d = out->written - out->offset;
  1265. assert(d >= 0);
  1266. if(d > 0){
  1267. if(n < d){
  1268. if(out->curr != nil)
  1269. md5((uchar*)buf, n, nil, out->curr);
  1270. out->offset += n;
  1271. return n;
  1272. }
  1273. if(out->curr != nil){
  1274. md5((uchar*)buf, d, m0, out->curr);
  1275. out->curr = nil;
  1276. md5(nil, 0, m1, md5dup(out->hiwat));
  1277. if(memcmp(m0, m1, MD5dlen) != 0){
  1278. fprint(2, "integrity check failure at offset %d\n", out->written);
  1279. return -1;
  1280. }
  1281. }
  1282. buf += d;
  1283. n -= d;
  1284. out->offset += d;
  1285. }
  1286. if(n > 0){
  1287. out->hiwat = md5((uchar*)buf, n, nil, out->hiwat);
  1288. n = write(out->fd, buf, n);
  1289. if(n > 0){
  1290. out->offset += n;
  1291. out->written += n;
  1292. }
  1293. }
  1294. return n + d;
  1295. }