1 /*
2 * nanohttp.c: minimalist HTTP GET implementation to fetch external subsets.
3 * focuses on size, streamability, reentrancy and portability
4 *
5 * This is clearly not a general purpose HTTP implementation
6 * If you look for one, check:
7 * http://www.w3.org/Library/
8 *
9 * See Copyright for the status of this software.
10 *
11 * daniel@veillard.com
12 */
13
14 #define IN_LIBXML
15 #include "libxml.h"
16
17 #ifdef LIBXML_HTTP_ENABLED
18 #include <string.h>
19
20 #ifdef HAVE_STDLIB_H
21 #include <stdlib.h>
22 #endif
23 #ifdef HAVE_UNISTD_H
24 #include <unistd.h>
25 #endif
26 #ifdef HAVE_SYS_TYPES_H
27 #include <sys/types.h>
28 #endif
29 #ifdef HAVE_SYS_SOCKET_H
30 #include <sys/socket.h>
31 #endif
32 #ifdef HAVE_NETINET_IN_H
33 #include <netinet/in.h>
34 #endif
35 #ifdef HAVE_ARPA_INET_H
36 #include <arpa/inet.h>
37 #endif
38 #ifdef HAVE_NETDB_H
39 #include <netdb.h>
40 #endif
41 #ifdef HAVE_RESOLV_H
42 #ifdef HAVE_ARPA_NAMESER_H
43 #include <arpa/nameser.h>
44 #endif
45 #include <resolv.h>
46 #endif
47 #ifdef HAVE_FCNTL_H
48 #include <fcntl.h>
49 #endif
50 #ifdef HAVE_ERRNO_H
51 #include <errno.h>
52 #endif
53 #ifdef HAVE_SYS_TIME_H
54 #include <sys/time.h>
55 #endif
56 #ifndef HAVE_POLL_H
57 #ifdef HAVE_SYS_SELECT_H
58 #include <sys/select.h>
59 #endif
60 #else
61 #include <poll.h>
62 #endif
63 #ifdef HAVE_STRINGS_H
64 #include <strings.h>
65 #endif
66 #ifdef LIBXML_ZLIB_ENABLED
67 #include <zlib.h>
68 #endif
69
70
71 #ifdef VMS
72 #include <stropts>
73 #define XML_SOCKLEN_T unsigned int
74 #endif
75
76 #if defined(_WIN32) && !defined(__CYGWIN__)
77 #include <wsockcompat.h>
78 #endif
79
80 #include <libxml/globals.h>
81 #include <libxml/xmlerror.h>
82 #include <libxml/xmlmemory.h>
83 #include <libxml/parser.h> /* for xmlStr(n)casecmp() */
84 #include <libxml/nanohttp.h>
85 #include <libxml/globals.h>
86 #include <libxml/uri.h>
87
88 /**
89 * A couple portability macros
90 */
91 #ifndef _WINSOCKAPI_
92 #if !defined(__BEOS__) || defined(__HAIKU__)
93 #define closesocket(s) close(s)
94 #endif
95 #define SOCKET int
96 #define INVALID_SOCKET (-1)
97 #endif
98
99 #ifdef __BEOS__
100 #ifndef PF_INET
101 #define PF_INET AF_INET
102 #endif
103 #endif
104
105 #ifndef XML_SOCKLEN_T
106 #define XML_SOCKLEN_T unsigned int
107 #endif
108
109 #ifdef STANDALONE
110 #define DEBUG_HTTP
111 #define xmlStrncasecmp(a, b, n) strncasecmp((char *)a, (char *)b, n)
112 #define xmlStrcasecmpi(a, b) strcasecmp((char *)a, (char *)b)
113 #endif
114
115 #define XML_NANO_HTTP_MAX_REDIR 10
116
117 #define XML_NANO_HTTP_CHUNK 4096
118
119 #define XML_NANO_HTTP_CLOSED 0
120 #define XML_NANO_HTTP_WRITE 1
121 #define XML_NANO_HTTP_READ 2
122 #define XML_NANO_HTTP_NONE 4
123
124 typedef struct xmlNanoHTTPCtxt {
125 char *protocol; /* the protocol name */
126 char *hostname; /* the host name */
127 int port; /* the port */
128 char *path; /* the path within the URL */
129 char *query; /* the query string */
130 SOCKET fd; /* the file descriptor for the socket */
131 int state; /* WRITE / READ / CLOSED */
132 char *out; /* buffer sent (zero terminated) */
133 char *outptr; /* index within the buffer sent */
134 char *in; /* the receiving buffer */
135 char *content; /* the start of the content */
136 char *inptr; /* the next byte to read from network */
137 char *inrptr; /* the next byte to give back to the client */
138 int inlen; /* len of the input buffer */
139 int last; /* return code for last operation */
140 int returnValue; /* the protocol return value */
141 int version; /* the protocol version */
142 int ContentLength; /* specified content length from HTTP header */
143 char *contentType; /* the MIME type for the input */
144 char *location; /* the new URL in case of redirect */
145 char *authHeader; /* contents of {WWW,Proxy}-Authenticate header */
146 char *encoding; /* encoding extracted from the contentType */
147 char *mimeType; /* Mime-Type extracted from the contentType */
148 #ifdef LIBXML_ZLIB_ENABLED
149 z_stream *strm; /* Zlib stream object */
150 int usesGzip; /* "Content-Encoding: gzip" was detected */
151 #endif
152 } xmlNanoHTTPCtxt, *xmlNanoHTTPCtxtPtr;
153
154 static int initialized = 0;
155 static char *proxy = NULL; /* the proxy name if any */
156 static int proxyPort; /* the proxy port if any */
157 static unsigned int timeout = 60;/* the select() timeout in seconds */
158
159 static int xmlNanoHTTPFetchContent( void * ctx, char ** ptr, int * len );
160
161 /**
162 * xmlHTTPErrMemory:
163 * @extra: extra information
164 *
165 * Handle an out of memory condition
166 */
167 static void
xmlHTTPErrMemory(const char * extra)168 xmlHTTPErrMemory(const char *extra)
169 {
170 __xmlSimpleError(XML_FROM_HTTP, XML_ERR_NO_MEMORY, NULL, NULL, extra);
171 }
172
173 /**
174 * A portability function
175 */
socket_errno(void)176 static int socket_errno(void) {
177 #ifdef _WINSOCKAPI_
178 int err = WSAGetLastError();
179 switch(err) {
180 case WSAECONNRESET:
181 return(ECONNRESET);
182 case WSAEINPROGRESS:
183 return(EINPROGRESS);
184 case WSAEINTR:
185 return(EINTR);
186 case WSAESHUTDOWN:
187 return(ESHUTDOWN);
188 case WSAEWOULDBLOCK:
189 return(EWOULDBLOCK);
190 default:
191 return(err);
192 }
193 #else
194 return(errno);
195 #endif
196 }
197
198 #ifdef SUPPORT_IP6
199 static
have_ipv6(void)200 int have_ipv6(void) {
201 SOCKET s;
202
203 s = socket (AF_INET6, SOCK_STREAM, 0);
204 if (s != INVALID_SOCKET) {
205 close (s);
206 return (1);
207 }
208 return (0);
209 }
210 #endif
211
212 /**
213 * xmlNanoHTTPInit:
214 *
215 * Initialize the HTTP protocol layer.
216 * Currently it just checks for proxy information
217 */
218
219 void
xmlNanoHTTPInit(void)220 xmlNanoHTTPInit(void) {
221 const char *env;
222 #ifdef _WINSOCKAPI_
223 WSADATA wsaData;
224 #endif
225
226 if (initialized)
227 return;
228
229 #ifdef _WINSOCKAPI_
230 if (WSAStartup(MAKEWORD(1, 1), &wsaData) != 0)
231 return;
232 #endif
233
234 if (proxy == NULL) {
235 proxyPort = 80;
236 env = getenv("no_proxy");
237 if (env && ((env[0] == '*') && (env[1] == 0)))
238 goto done;
239 env = getenv("http_proxy");
240 if (env != NULL) {
241 xmlNanoHTTPScanProxy(env);
242 goto done;
243 }
244 env = getenv("HTTP_PROXY");
245 if (env != NULL) {
246 xmlNanoHTTPScanProxy(env);
247 goto done;
248 }
249 }
250 done:
251 initialized = 1;
252 }
253
254 /**
255 * xmlNanoHTTPCleanup:
256 *
257 * Cleanup the HTTP protocol layer.
258 */
259
260 void
xmlNanoHTTPCleanup(void)261 xmlNanoHTTPCleanup(void) {
262 if (proxy != NULL) {
263 xmlFree(proxy);
264 proxy = NULL;
265 }
266 #ifdef _WINSOCKAPI_
267 if (initialized)
268 WSACleanup();
269 #endif
270 initialized = 0;
271 return;
272 }
273
274 /**
275 * xmlNanoHTTPScanURL:
276 * @ctxt: an HTTP context
277 * @URL: The URL used to initialize the context
278 *
279 * (Re)Initialize an HTTP context by parsing the URL and finding
280 * the protocol host port and path it indicates.
281 */
282
283 static void
xmlNanoHTTPScanURL(xmlNanoHTTPCtxtPtr ctxt,const char * URL)284 xmlNanoHTTPScanURL(xmlNanoHTTPCtxtPtr ctxt, const char *URL) {
285 xmlURIPtr uri;
286 int len;
287
288 /*
289 * Clear any existing data from the context
290 */
291 if (ctxt->protocol != NULL) {
292 xmlFree(ctxt->protocol);
293 ctxt->protocol = NULL;
294 }
295 if (ctxt->hostname != NULL) {
296 xmlFree(ctxt->hostname);
297 ctxt->hostname = NULL;
298 }
299 if (ctxt->path != NULL) {
300 xmlFree(ctxt->path);
301 ctxt->path = NULL;
302 }
303 if (ctxt->query != NULL) {
304 xmlFree(ctxt->query);
305 ctxt->query = NULL;
306 }
307 if (URL == NULL) return;
308
309 uri = xmlParseURIRaw(URL, 1);
310 if (uri == NULL)
311 return;
312
313 if ((uri->scheme == NULL) || (uri->server == NULL)) {
314 xmlFreeURI(uri);
315 return;
316 }
317
318 ctxt->protocol = xmlMemStrdup(uri->scheme);
319 /* special case of IPv6 addresses, the [] need to be removed */
320 if ((uri->server != NULL) && (*uri->server == '[')) {
321 len = strlen(uri->server);
322 if ((len > 2) && (uri->server[len - 1] == ']')) {
323 ctxt->hostname = (char *) xmlCharStrndup(uri->server + 1, len -2);
324 } else
325 ctxt->hostname = xmlMemStrdup(uri->server);
326 } else
327 ctxt->hostname = xmlMemStrdup(uri->server);
328 if (uri->path != NULL)
329 ctxt->path = xmlMemStrdup(uri->path);
330 else
331 ctxt->path = xmlMemStrdup("/");
332 if (uri->query != NULL)
333 ctxt->query = xmlMemStrdup(uri->query);
334 if (uri->port != 0)
335 ctxt->port = uri->port;
336
337 xmlFreeURI(uri);
338 }
339
340 /**
341 * xmlNanoHTTPScanProxy:
342 * @URL: The proxy URL used to initialize the proxy context
343 *
344 * (Re)Initialize the HTTP Proxy context by parsing the URL and finding
345 * the protocol host port it indicates.
346 * Should be like http://myproxy/ or http://myproxy:3128/
347 * A NULL URL cleans up proxy information.
348 */
349
350 void
xmlNanoHTTPScanProxy(const char * URL)351 xmlNanoHTTPScanProxy(const char *URL) {
352 xmlURIPtr uri;
353
354 if (proxy != NULL) {
355 xmlFree(proxy);
356 proxy = NULL;
357 }
358 proxyPort = 0;
359
360 #ifdef DEBUG_HTTP
361 if (URL == NULL)
362 xmlGenericError(xmlGenericErrorContext,
363 "Removing HTTP proxy info\n");
364 else
365 xmlGenericError(xmlGenericErrorContext,
366 "Using HTTP proxy %s\n", URL);
367 #endif
368 if (URL == NULL) return;
369
370 uri = xmlParseURIRaw(URL, 1);
371 if ((uri == NULL) || (uri->scheme == NULL) ||
372 (strcmp(uri->scheme, "http")) || (uri->server == NULL)) {
373 __xmlIOErr(XML_FROM_HTTP, XML_HTTP_URL_SYNTAX, "Syntax Error\n");
374 if (uri != NULL)
375 xmlFreeURI(uri);
376 return;
377 }
378
379 proxy = xmlMemStrdup(uri->server);
380 if (uri->port != 0)
381 proxyPort = uri->port;
382
383 xmlFreeURI(uri);
384 }
385
386 /**
387 * xmlNanoHTTPNewCtxt:
388 * @URL: The URL used to initialize the context
389 *
390 * Allocate and initialize a new HTTP context.
391 *
392 * Returns an HTTP context or NULL in case of error.
393 */
394
395 static xmlNanoHTTPCtxtPtr
xmlNanoHTTPNewCtxt(const char * URL)396 xmlNanoHTTPNewCtxt(const char *URL) {
397 xmlNanoHTTPCtxtPtr ret;
398
399 ret = (xmlNanoHTTPCtxtPtr) xmlMalloc(sizeof(xmlNanoHTTPCtxt));
400 if (ret == NULL) {
401 xmlHTTPErrMemory("allocating context");
402 return(NULL);
403 }
404
405 memset(ret, 0, sizeof(xmlNanoHTTPCtxt));
406 ret->port = 80;
407 ret->returnValue = 0;
408 ret->fd = INVALID_SOCKET;
409 ret->ContentLength = -1;
410
411 xmlNanoHTTPScanURL(ret, URL);
412
413 return(ret);
414 }
415
416 /**
417 * xmlNanoHTTPFreeCtxt:
418 * @ctxt: an HTTP context
419 *
420 * Frees the context after closing the connection.
421 */
422
423 static void
xmlNanoHTTPFreeCtxt(xmlNanoHTTPCtxtPtr ctxt)424 xmlNanoHTTPFreeCtxt(xmlNanoHTTPCtxtPtr ctxt) {
425 if (ctxt == NULL) return;
426 if (ctxt->hostname != NULL) xmlFree(ctxt->hostname);
427 if (ctxt->protocol != NULL) xmlFree(ctxt->protocol);
428 if (ctxt->path != NULL) xmlFree(ctxt->path);
429 if (ctxt->query != NULL) xmlFree(ctxt->query);
430 if (ctxt->out != NULL) xmlFree(ctxt->out);
431 if (ctxt->in != NULL) xmlFree(ctxt->in);
432 if (ctxt->contentType != NULL) xmlFree(ctxt->contentType);
433 if (ctxt->encoding != NULL) xmlFree(ctxt->encoding);
434 if (ctxt->mimeType != NULL) xmlFree(ctxt->mimeType);
435 if (ctxt->location != NULL) xmlFree(ctxt->location);
436 if (ctxt->authHeader != NULL) xmlFree(ctxt->authHeader);
437 #ifdef LIBXML_ZLIB_ENABLED
438 if (ctxt->strm != NULL) {
439 inflateEnd(ctxt->strm);
440 xmlFree(ctxt->strm);
441 }
442 #endif
443
444 ctxt->state = XML_NANO_HTTP_NONE;
445 if (ctxt->fd != INVALID_SOCKET) closesocket(ctxt->fd);
446 ctxt->fd = INVALID_SOCKET;
447 xmlFree(ctxt);
448 }
449
450 /**
451 * xmlNanoHTTPSend:
452 * @ctxt: an HTTP context
453 *
454 * Send the input needed to initiate the processing on the server side
455 * Returns number of bytes sent or -1 on error.
456 */
457
458 static int
xmlNanoHTTPSend(xmlNanoHTTPCtxtPtr ctxt,const char * xmt_ptr,int outlen)459 xmlNanoHTTPSend(xmlNanoHTTPCtxtPtr ctxt, const char *xmt_ptr, int outlen)
460 {
461 int total_sent = 0;
462 #ifdef HAVE_POLL_H
463 struct pollfd p;
464 #else
465 struct timeval tv;
466 fd_set wfd;
467 #endif
468
469 if ((ctxt->state & XML_NANO_HTTP_WRITE) && (xmt_ptr != NULL)) {
470 while (total_sent < outlen) {
471 int nsent = send(ctxt->fd, SEND_ARG2_CAST (xmt_ptr + total_sent),
472 outlen - total_sent, 0);
473
474 if (nsent > 0)
475 total_sent += nsent;
476 else if ((nsent == -1) &&
477 #if defined(EAGAIN) && EAGAIN != EWOULDBLOCK
478 (socket_errno() != EAGAIN) &&
479 #endif
480 (socket_errno() != EWOULDBLOCK)) {
481 __xmlIOErr(XML_FROM_HTTP, 0, "send failed\n");
482 if (total_sent == 0)
483 total_sent = -1;
484 break;
485 } else {
486 /*
487 * No data sent
488 * Since non-blocking sockets are used, wait for
489 * socket to be writable or default timeout prior
490 * to retrying.
491 */
492 #ifndef HAVE_POLL_H
493 #ifndef _WINSOCKAPI_
494 if (ctxt->fd > FD_SETSIZE)
495 return -1;
496 #endif
497
498 tv.tv_sec = timeout;
499 tv.tv_usec = 0;
500 FD_ZERO(&wfd);
501 #ifdef _MSC_VER
502 #pragma warning(push)
503 #pragma warning(disable: 4018)
504 #endif
505 FD_SET(ctxt->fd, &wfd);
506 #ifdef _MSC_VER
507 #pragma warning(pop)
508 #endif
509 (void) select(ctxt->fd + 1, NULL, &wfd, NULL, &tv);
510 #else
511 p.fd = ctxt->fd;
512 p.events = POLLOUT;
513 (void) poll(&p, 1, timeout * 1000);
514 #endif /* !HAVE_POLL_H */
515 }
516 }
517 }
518
519 return total_sent;
520 }
521
522 /**
523 * xmlNanoHTTPRecv:
524 * @ctxt: an HTTP context
525 *
526 * Read information coming from the HTTP connection.
527 * This is a blocking call (but it blocks in select(), not read()).
528 *
529 * Returns the number of byte read or -1 in case of error.
530 */
531
532 static int
xmlNanoHTTPRecv(xmlNanoHTTPCtxtPtr ctxt)533 xmlNanoHTTPRecv(xmlNanoHTTPCtxtPtr ctxt)
534 {
535 #ifdef HAVE_POLL_H
536 struct pollfd p;
537 #else
538 fd_set rfd;
539 struct timeval tv;
540 #endif
541
542
543 while (ctxt->state & XML_NANO_HTTP_READ) {
544 if (ctxt->in == NULL) {
545 ctxt->in = (char *) xmlMallocAtomic(65000 * sizeof(char));
546 if (ctxt->in == NULL) {
547 xmlHTTPErrMemory("allocating input");
548 ctxt->last = -1;
549 return (-1);
550 }
551 ctxt->inlen = 65000;
552 ctxt->inptr = ctxt->content = ctxt->inrptr = ctxt->in;
553 }
554 if (ctxt->inrptr > ctxt->in + XML_NANO_HTTP_CHUNK) {
555 int delta = ctxt->inrptr - ctxt->in;
556 int len = ctxt->inptr - ctxt->inrptr;
557
558 memmove(ctxt->in, ctxt->inrptr, len);
559 ctxt->inrptr -= delta;
560 ctxt->content -= delta;
561 ctxt->inptr -= delta;
562 }
563 if ((ctxt->in + ctxt->inlen) < (ctxt->inptr + XML_NANO_HTTP_CHUNK)) {
564 int d_inptr = ctxt->inptr - ctxt->in;
565 int d_content = ctxt->content - ctxt->in;
566 int d_inrptr = ctxt->inrptr - ctxt->in;
567 char *tmp_ptr = ctxt->in;
568
569 ctxt->inlen *= 2;
570 ctxt->in = (char *) xmlRealloc(tmp_ptr, ctxt->inlen);
571 if (ctxt->in == NULL) {
572 xmlHTTPErrMemory("allocating input buffer");
573 xmlFree(tmp_ptr);
574 ctxt->last = -1;
575 return (-1);
576 }
577 ctxt->inptr = ctxt->in + d_inptr;
578 ctxt->content = ctxt->in + d_content;
579 ctxt->inrptr = ctxt->in + d_inrptr;
580 }
581 ctxt->last = recv(ctxt->fd, ctxt->inptr, XML_NANO_HTTP_CHUNK, 0);
582 if (ctxt->last > 0) {
583 ctxt->inptr += ctxt->last;
584 return (ctxt->last);
585 }
586 if (ctxt->last == 0) {
587 return (0);
588 }
589 if (ctxt->last == -1) {
590 switch (socket_errno()) {
591 case EINPROGRESS:
592 case EWOULDBLOCK:
593 #if defined(EAGAIN) && EAGAIN != EWOULDBLOCK
594 case EAGAIN:
595 #endif
596 break;
597
598 case ECONNRESET:
599 case ESHUTDOWN:
600 return (0);
601
602 default:
603 __xmlIOErr(XML_FROM_HTTP, 0, "recv failed\n");
604 return (-1);
605 }
606 }
607 #ifdef HAVE_POLL_H
608 p.fd = ctxt->fd;
609 p.events = POLLIN;
610 if ((poll(&p, 1, timeout * 1000) < 1)
611 #if defined(EINTR)
612 && (errno != EINTR)
613 #endif
614 )
615 return (0);
616 #else /* !HAVE_POLL_H */
617 #ifndef _WINSOCKAPI_
618 if (ctxt->fd > FD_SETSIZE)
619 return 0;
620 #endif
621
622 tv.tv_sec = timeout;
623 tv.tv_usec = 0;
624 FD_ZERO(&rfd);
625
626 #ifdef _MSC_VER
627 #pragma warning(push)
628 #pragma warning(disable: 4018)
629 #endif
630
631 FD_SET(ctxt->fd, &rfd);
632
633 #ifdef _MSC_VER
634 #pragma warning(pop)
635 #endif
636
637 if ((select(ctxt->fd + 1, &rfd, NULL, NULL, &tv) < 1)
638 #if defined(EINTR)
639 && (socket_errno() != EINTR)
640 #endif
641 )
642 return (0);
643 #endif /* !HAVE_POLL_H */
644 }
645 return (0);
646 }
647
648 /**
649 * xmlNanoHTTPReadLine:
650 * @ctxt: an HTTP context
651 *
652 * Read one line in the HTTP server output, usually for extracting
653 * the HTTP protocol information from the answer header.
654 *
655 * Returns a newly allocated string with a copy of the line, or NULL
656 * which indicate the end of the input.
657 */
658
659 static char *
xmlNanoHTTPReadLine(xmlNanoHTTPCtxtPtr ctxt)660 xmlNanoHTTPReadLine(xmlNanoHTTPCtxtPtr ctxt) {
661 char buf[4096];
662 char *bp = buf;
663 int rc;
664
665 while (bp - buf < 4095) {
666 if (ctxt->inrptr == ctxt->inptr) {
667 if ( (rc = xmlNanoHTTPRecv(ctxt)) == 0) {
668 if (bp == buf)
669 return(NULL);
670 else
671 *bp = 0;
672 return(xmlMemStrdup(buf));
673 }
674 else if ( rc == -1 ) {
675 return ( NULL );
676 }
677 }
678 *bp = *ctxt->inrptr++;
679 if (*bp == '\n') {
680 *bp = 0;
681 return(xmlMemStrdup(buf));
682 }
683 if (*bp != '\r')
684 bp++;
685 }
686 buf[4095] = 0;
687 return(xmlMemStrdup(buf));
688 }
689
690
691 /**
692 * xmlNanoHTTPScanAnswer:
693 * @ctxt: an HTTP context
694 * @line: an HTTP header line
695 *
696 * Try to extract useful information from the server answer.
697 * We currently parse and process:
698 * - The HTTP revision/ return code
699 * - The Content-Type, Mime-Type and charset used
700 * - The Location for redirect processing.
701 *
702 * Returns -1 in case of failure, the file descriptor number otherwise
703 */
704
705 static void
xmlNanoHTTPScanAnswer(xmlNanoHTTPCtxtPtr ctxt,const char * line)706 xmlNanoHTTPScanAnswer(xmlNanoHTTPCtxtPtr ctxt, const char *line) {
707 const char *cur = line;
708
709 if (line == NULL) return;
710
711 if (!strncmp(line, "HTTP/", 5)) {
712 int version = 0;
713 int ret = 0;
714
715 cur += 5;
716 while ((*cur >= '0') && (*cur <= '9')) {
717 version *= 10;
718 version += *cur - '0';
719 cur++;
720 }
721 if (*cur == '.') {
722 cur++;
723 if ((*cur >= '0') && (*cur <= '9')) {
724 version *= 10;
725 version += *cur - '0';
726 cur++;
727 }
728 while ((*cur >= '0') && (*cur <= '9'))
729 cur++;
730 } else
731 version *= 10;
732 if ((*cur != ' ') && (*cur != '\t')) return;
733 while ((*cur == ' ') || (*cur == '\t')) cur++;
734 if ((*cur < '0') || (*cur > '9')) return;
735 while ((*cur >= '0') && (*cur <= '9')) {
736 ret *= 10;
737 ret += *cur - '0';
738 cur++;
739 }
740 if ((*cur != 0) && (*cur != ' ') && (*cur != '\t')) return;
741 ctxt->returnValue = ret;
742 ctxt->version = version;
743 } else if (!xmlStrncasecmp(BAD_CAST line, BAD_CAST"Content-Type:", 13)) {
744 const xmlChar *charset, *last, *mime;
745 cur += 13;
746 while ((*cur == ' ') || (*cur == '\t')) cur++;
747 if (ctxt->contentType != NULL)
748 xmlFree(ctxt->contentType);
749 ctxt->contentType = xmlMemStrdup(cur);
750 mime = (const xmlChar *) cur;
751 last = mime;
752 while ((*last != 0) && (*last != ' ') && (*last != '\t') &&
753 (*last != ';') && (*last != ','))
754 last++;
755 if (ctxt->mimeType != NULL)
756 xmlFree(ctxt->mimeType);
757 ctxt->mimeType = (char *) xmlStrndup(mime, last - mime);
758 charset = xmlStrstr(BAD_CAST ctxt->contentType, BAD_CAST "charset=");
759 if (charset != NULL) {
760 charset += 8;
761 last = charset;
762 while ((*last != 0) && (*last != ' ') && (*last != '\t') &&
763 (*last != ';') && (*last != ','))
764 last++;
765 if (ctxt->encoding != NULL)
766 xmlFree(ctxt->encoding);
767 ctxt->encoding = (char *) xmlStrndup(charset, last - charset);
768 }
769 } else if (!xmlStrncasecmp(BAD_CAST line, BAD_CAST"ContentType:", 12)) {
770 const xmlChar *charset, *last, *mime;
771 cur += 12;
772 if (ctxt->contentType != NULL) return;
773 while ((*cur == ' ') || (*cur == '\t')) cur++;
774 ctxt->contentType = xmlMemStrdup(cur);
775 mime = (const xmlChar *) cur;
776 last = mime;
777 while ((*last != 0) && (*last != ' ') && (*last != '\t') &&
778 (*last != ';') && (*last != ','))
779 last++;
780 if (ctxt->mimeType != NULL)
781 xmlFree(ctxt->mimeType);
782 ctxt->mimeType = (char *) xmlStrndup(mime, last - mime);
783 charset = xmlStrstr(BAD_CAST ctxt->contentType, BAD_CAST "charset=");
784 if (charset != NULL) {
785 charset += 8;
786 last = charset;
787 while ((*last != 0) && (*last != ' ') && (*last != '\t') &&
788 (*last != ';') && (*last != ','))
789 last++;
790 if (ctxt->encoding != NULL)
791 xmlFree(ctxt->encoding);
792 ctxt->encoding = (char *) xmlStrndup(charset, last - charset);
793 }
794 } else if (!xmlStrncasecmp(BAD_CAST line, BAD_CAST"Location:", 9)) {
795 cur += 9;
796 while ((*cur == ' ') || (*cur == '\t')) cur++;
797 if (ctxt->location != NULL)
798 xmlFree(ctxt->location);
799 if (*cur == '/') {
800 xmlChar *tmp_http = xmlStrdup(BAD_CAST "http://");
801 xmlChar *tmp_loc =
802 xmlStrcat(tmp_http, (const xmlChar *) ctxt->hostname);
803 ctxt->location =
804 (char *) xmlStrcat (tmp_loc, (const xmlChar *) cur);
805 } else {
806 ctxt->location = xmlMemStrdup(cur);
807 }
808 } else if (!xmlStrncasecmp(BAD_CAST line, BAD_CAST"WWW-Authenticate:", 17)) {
809 cur += 17;
810 while ((*cur == ' ') || (*cur == '\t')) cur++;
811 if (ctxt->authHeader != NULL)
812 xmlFree(ctxt->authHeader);
813 ctxt->authHeader = xmlMemStrdup(cur);
814 } else if (!xmlStrncasecmp(BAD_CAST line, BAD_CAST"Proxy-Authenticate:", 19)) {
815 cur += 19;
816 while ((*cur == ' ') || (*cur == '\t')) cur++;
817 if (ctxt->authHeader != NULL)
818 xmlFree(ctxt->authHeader);
819 ctxt->authHeader = xmlMemStrdup(cur);
820 #ifdef LIBXML_ZLIB_ENABLED
821 } else if ( !xmlStrncasecmp( BAD_CAST line, BAD_CAST"Content-Encoding:", 17) ) {
822 cur += 17;
823 while ((*cur == ' ') || (*cur == '\t')) cur++;
824 if ( !xmlStrncasecmp( BAD_CAST cur, BAD_CAST"gzip", 4) ) {
825 ctxt->usesGzip = 1;
826
827 ctxt->strm = xmlMalloc(sizeof(z_stream));
828
829 if (ctxt->strm != NULL) {
830 ctxt->strm->zalloc = Z_NULL;
831 ctxt->strm->zfree = Z_NULL;
832 ctxt->strm->opaque = Z_NULL;
833 ctxt->strm->avail_in = 0;
834 ctxt->strm->next_in = Z_NULL;
835
836 inflateInit2( ctxt->strm, 31 );
837 }
838 }
839 #endif
840 } else if ( !xmlStrncasecmp( BAD_CAST line, BAD_CAST"Content-Length:", 15) ) {
841 cur += 15;
842 ctxt->ContentLength = strtol( cur, NULL, 10 );
843 }
844 }
845
846 /**
847 * xmlNanoHTTPConnectAttempt:
848 * @addr: a socket address structure
849 *
850 * Attempt a connection to the given IP:port endpoint. It forces
851 * non-blocking semantic on the socket, and allow 60 seconds for
852 * the host to answer.
853 *
854 * Returns -1 in case of failure, the file descriptor number otherwise
855 */
856
857 static SOCKET
xmlNanoHTTPConnectAttempt(struct sockaddr * addr)858 xmlNanoHTTPConnectAttempt(struct sockaddr *addr)
859 {
860 #ifndef HAVE_POLL_H
861 fd_set wfd;
862 #ifdef _WINSOCKAPI_
863 fd_set xfd;
864 #endif
865 struct timeval tv;
866 #else /* !HAVE_POLL_H */
867 struct pollfd p;
868 #endif /* !HAVE_POLL_H */
869 int status;
870
871 int addrlen;
872
873 SOCKET s;
874
875 #ifdef SUPPORT_IP6
876 if (addr->sa_family == AF_INET6) {
877 s = socket(PF_INET6, SOCK_STREAM, IPPROTO_TCP);
878 addrlen = sizeof(struct sockaddr_in6);
879 } else
880 #endif
881 {
882 s = socket(PF_INET, SOCK_STREAM, IPPROTO_TCP);
883 addrlen = sizeof(struct sockaddr_in);
884 }
885 if (s == INVALID_SOCKET) {
886 #ifdef DEBUG_HTTP
887 perror("socket");
888 #endif
889 __xmlIOErr(XML_FROM_HTTP, 0, "socket failed\n");
890 return INVALID_SOCKET;
891 }
892 #ifdef _WINSOCKAPI_
893 {
894 u_long one = 1;
895
896 status = ioctlsocket(s, FIONBIO, &one) == SOCKET_ERROR ? -1 : 0;
897 }
898 #else /* _WINSOCKAPI_ */
899 #if defined(VMS)
900 {
901 int enable = 1;
902
903 status = ioctl(s, FIONBIO, &enable);
904 }
905 #else /* VMS */
906 #if defined(__BEOS__) && !defined(__HAIKU__)
907 {
908 bool noblock = true;
909
910 status =
911 setsockopt(s, SOL_SOCKET, SO_NONBLOCK, &noblock,
912 sizeof(noblock));
913 }
914 #else /* __BEOS__ */
915 if ((status = fcntl(s, F_GETFL, 0)) != -1) {
916 #ifdef O_NONBLOCK
917 status |= O_NONBLOCK;
918 #else /* O_NONBLOCK */
919 #ifdef F_NDELAY
920 status |= F_NDELAY;
921 #endif /* F_NDELAY */
922 #endif /* !O_NONBLOCK */
923 status = fcntl(s, F_SETFL, status);
924 }
925 if (status < 0) {
926 #ifdef DEBUG_HTTP
927 perror("nonblocking");
928 #endif
929 __xmlIOErr(XML_FROM_HTTP, 0, "error setting non-blocking IO\n");
930 closesocket(s);
931 return INVALID_SOCKET;
932 }
933 #endif /* !__BEOS__ */
934 #endif /* !VMS */
935 #endif /* !_WINSOCKAPI_ */
936
937 if (connect(s, addr, addrlen) == -1) {
938 switch (socket_errno()) {
939 case EINPROGRESS:
940 case EWOULDBLOCK:
941 break;
942 default:
943 __xmlIOErr(XML_FROM_HTTP, 0,
944 "error connecting to HTTP server");
945 closesocket(s);
946 return INVALID_SOCKET;
947 }
948 }
949 #ifndef HAVE_POLL_H
950 tv.tv_sec = timeout;
951 tv.tv_usec = 0;
952
953 #ifdef _MSC_VER
954 #pragma warning(push)
955 #pragma warning(disable: 4018)
956 #endif
957 #ifndef _WINSOCKAPI_
958 if (s > FD_SETSIZE)
959 return INVALID_SOCKET;
960 #endif
961 FD_ZERO(&wfd);
962 FD_SET(s, &wfd);
963
964 #ifdef _WINSOCKAPI_
965 FD_ZERO(&xfd);
966 FD_SET(s, &xfd);
967
968 switch (select(s + 1, NULL, &wfd, &xfd, &tv))
969 #else
970 switch (select(s + 1, NULL, &wfd, NULL, &tv))
971 #endif
972 #ifdef _MSC_VER
973 #pragma warning(pop)
974 #endif
975
976 #else /* !HAVE_POLL_H */
977 p.fd = s;
978 p.events = POLLOUT;
979 switch (poll(&p, 1, timeout * 1000))
980 #endif /* !HAVE_POLL_H */
981
982 {
983 case 0:
984 /* Time out */
985 __xmlIOErr(XML_FROM_HTTP, 0, "Connect attempt timed out");
986 closesocket(s);
987 return INVALID_SOCKET;
988 case -1:
989 /* Ermm.. ?? */
990 __xmlIOErr(XML_FROM_HTTP, 0, "Connect failed");
991 closesocket(s);
992 return INVALID_SOCKET;
993 }
994
995 #ifndef HAVE_POLL_H
996 if (FD_ISSET(s, &wfd)
997 #ifdef _WINSOCKAPI_
998 || FD_ISSET(s, &xfd)
999 #endif
1000 )
1001 #else /* !HAVE_POLL_H */
1002 if (p.revents == POLLOUT)
1003 #endif /* !HAVE_POLL_H */
1004 {
1005 XML_SOCKLEN_T len;
1006
1007 len = sizeof(status);
1008 #ifdef SO_ERROR
1009 if (getsockopt(s, SOL_SOCKET, SO_ERROR, (char *) &status, &len) <
1010 0) {
1011 /* Solaris error code */
1012 __xmlIOErr(XML_FROM_HTTP, 0, "getsockopt failed\n");
1013 closesocket(s);
1014 return INVALID_SOCKET;
1015 }
1016 #endif
1017 if (status) {
1018 __xmlIOErr(XML_FROM_HTTP, 0,
1019 "Error connecting to remote host");
1020 closesocket(s);
1021 errno = status;
1022 return INVALID_SOCKET;
1023 }
1024 } else {
1025 /* pbm */
1026 __xmlIOErr(XML_FROM_HTTP, 0, "select failed\n");
1027 closesocket(s);
1028 return INVALID_SOCKET;
1029 }
1030
1031 return (s);
1032 }
1033
1034 /**
1035 * xmlNanoHTTPConnectHost:
1036 * @host: the host name
1037 * @port: the port number
1038 *
1039 * Attempt a connection to the given host:port endpoint. It tries
1040 * the multiple IP provided by the DNS if available.
1041 *
1042 * Returns -1 in case of failure, the file descriptor number otherwise
1043 */
1044
1045 static SOCKET
xmlNanoHTTPConnectHost(const char * host,int port)1046 xmlNanoHTTPConnectHost(const char *host, int port)
1047 {
1048 struct sockaddr *addr = NULL;
1049 struct sockaddr_in sockin;
1050
1051 #ifdef SUPPORT_IP6
1052 struct in6_addr ia6;
1053 struct sockaddr_in6 sockin6;
1054 #endif
1055 SOCKET s;
1056
1057 memset (&sockin, 0, sizeof(sockin));
1058 #ifdef SUPPORT_IP6
1059 memset (&sockin6, 0, sizeof(sockin6));
1060 #endif
1061
1062 #if !defined(HAVE_GETADDRINFO) && defined(SUPPORT_IP6) && defined(RES_USE_INET6)
1063 if (have_ipv6 ())
1064 {
1065 if (!(_res.options & RES_INIT))
1066 res_init();
1067 _res.options |= RES_USE_INET6;
1068 }
1069 #endif
1070
1071 #if defined(HAVE_GETADDRINFO) && defined(SUPPORT_IP6) && !defined(_WIN32)
1072 if (have_ipv6 ())
1073 #endif
1074 #if defined(HAVE_GETADDRINFO) && (defined(SUPPORT_IP6) || defined(_WIN32))
1075 {
1076 int status;
1077 struct addrinfo hints, *res, *result;
1078
1079 result = NULL;
1080 memset (&hints, 0,sizeof(hints));
1081 hints.ai_socktype = SOCK_STREAM;
1082
1083 status = getaddrinfo (host, NULL, &hints, &result);
1084 if (status) {
1085 __xmlIOErr(XML_FROM_HTTP, 0, "getaddrinfo failed\n");
1086 return INVALID_SOCKET;
1087 }
1088
1089 for (res = result; res; res = res->ai_next) {
1090 if (res->ai_family == AF_INET) {
1091 if ((size_t)res->ai_addrlen > sizeof(sockin)) {
1092 __xmlIOErr(XML_FROM_HTTP, 0, "address size mismatch\n");
1093 freeaddrinfo (result);
1094 return INVALID_SOCKET;
1095 }
1096 memcpy (&sockin, res->ai_addr, res->ai_addrlen);
1097 sockin.sin_port = htons (port);
1098 addr = (struct sockaddr *)&sockin;
1099 #ifdef SUPPORT_IP6
1100 } else if (have_ipv6 () && (res->ai_family == AF_INET6)) {
1101 if ((size_t)res->ai_addrlen > sizeof(sockin6)) {
1102 __xmlIOErr(XML_FROM_HTTP, 0, "address size mismatch\n");
1103 freeaddrinfo (result);
1104 return INVALID_SOCKET;
1105 }
1106 memcpy (&sockin6, res->ai_addr, res->ai_addrlen);
1107 sockin6.sin6_port = htons (port);
1108 addr = (struct sockaddr *)&sockin6;
1109 #endif
1110 } else
1111 continue; /* for */
1112
1113 s = xmlNanoHTTPConnectAttempt (addr);
1114 if (s != INVALID_SOCKET) {
1115 freeaddrinfo (result);
1116 return (s);
1117 }
1118 }
1119
1120 if (result)
1121 freeaddrinfo (result);
1122 }
1123 #endif
1124 #if defined(HAVE_GETADDRINFO) && defined(SUPPORT_IP6) && !defined(_WIN32)
1125 else
1126 #endif
1127 #if !defined(HAVE_GETADDRINFO) || !defined(_WIN32)
1128 {
1129 struct hostent *h;
1130 struct in_addr ia;
1131 int i;
1132
1133 h = gethostbyname (GETHOSTBYNAME_ARG_CAST host);
1134 if (h == NULL) {
1135
1136 /*
1137 * Okay, I got fed up by the non-portability of this error message
1138 * extraction code. it work on Linux, if it work on your platform
1139 * and one want to enable it, send me the defined(foobar) needed
1140 */
1141 #if defined(HAVE_NETDB_H) && defined(HOST_NOT_FOUND) && defined(__linux__)
1142 const char *h_err_txt = "";
1143
1144 switch (h_errno) {
1145 case HOST_NOT_FOUND:
1146 h_err_txt = "Authoritative host not found";
1147 break;
1148
1149 case TRY_AGAIN:
1150 h_err_txt =
1151 "Non-authoritative host not found or server failure.";
1152 break;
1153
1154 case NO_RECOVERY:
1155 h_err_txt =
1156 "Non-recoverable errors: FORMERR, REFUSED, or NOTIMP.";
1157 break;
1158
1159 #ifdef NO_ADDRESS
1160 case NO_ADDRESS:
1161 h_err_txt =
1162 "Valid name, no data record of requested type.";
1163 break;
1164 #endif
1165
1166 default:
1167 h_err_txt = "No error text defined.";
1168 break;
1169 }
1170 __xmlIOErr(XML_FROM_HTTP, 0, h_err_txt);
1171 #else
1172 __xmlIOErr(XML_FROM_HTTP, 0, "Failed to resolve host");
1173 #endif
1174 return INVALID_SOCKET;
1175 }
1176
1177 for (i = 0; h->h_addr_list[i]; i++) {
1178 if (h->h_addrtype == AF_INET) {
1179 /* A records (IPv4) */
1180 if ((unsigned int) h->h_length > sizeof(ia)) {
1181 __xmlIOErr(XML_FROM_HTTP, 0, "address size mismatch\n");
1182 return INVALID_SOCKET;
1183 }
1184 memcpy (&ia, h->h_addr_list[i], h->h_length);
1185 sockin.sin_family = h->h_addrtype;
1186 sockin.sin_addr = ia;
1187 sockin.sin_port = (unsigned short)htons ((unsigned short)port);
1188 addr = (struct sockaddr *) &sockin;
1189 #ifdef SUPPORT_IP6
1190 } else if (have_ipv6 () && (h->h_addrtype == AF_INET6)) {
1191 /* AAAA records (IPv6) */
1192 if ((unsigned int) h->h_length > sizeof(ia6)) {
1193 __xmlIOErr(XML_FROM_HTTP, 0, "address size mismatch\n");
1194 return INVALID_SOCKET;
1195 }
1196 memcpy (&ia6, h->h_addr_list[i], h->h_length);
1197 sockin6.sin6_family = h->h_addrtype;
1198 sockin6.sin6_addr = ia6;
1199 sockin6.sin6_port = htons (port);
1200 addr = (struct sockaddr *) &sockin6;
1201 #endif
1202 } else
1203 break; /* for */
1204
1205 s = xmlNanoHTTPConnectAttempt (addr);
1206 if (s != INVALID_SOCKET)
1207 return (s);
1208 }
1209 }
1210 #endif
1211
1212 #ifdef DEBUG_HTTP
1213 xmlGenericError(xmlGenericErrorContext,
1214 "xmlNanoHTTPConnectHost: unable to connect to '%s'.\n",
1215 host);
1216 #endif
1217 return INVALID_SOCKET;
1218 }
1219
1220
1221 /**
1222 * xmlNanoHTTPOpen:
1223 * @URL: The URL to load
1224 * @contentType: if available the Content-Type information will be
1225 * returned at that location
1226 *
1227 * This function try to open a connection to the indicated resource
1228 * via HTTP GET.
1229 *
1230 * Returns NULL in case of failure, otherwise a request handler.
1231 * The contentType, if provided must be freed by the caller
1232 */
1233
1234 void*
xmlNanoHTTPOpen(const char * URL,char ** contentType)1235 xmlNanoHTTPOpen(const char *URL, char **contentType) {
1236 if (contentType != NULL) *contentType = NULL;
1237 return(xmlNanoHTTPMethod(URL, NULL, NULL, contentType, NULL, 0));
1238 }
1239
1240 /**
1241 * xmlNanoHTTPOpenRedir:
1242 * @URL: The URL to load
1243 * @contentType: if available the Content-Type information will be
1244 * returned at that location
1245 * @redir: if available the redirected URL will be returned
1246 *
1247 * This function try to open a connection to the indicated resource
1248 * via HTTP GET.
1249 *
1250 * Returns NULL in case of failure, otherwise a request handler.
1251 * The contentType, if provided must be freed by the caller
1252 */
1253
1254 void*
xmlNanoHTTPOpenRedir(const char * URL,char ** contentType,char ** redir)1255 xmlNanoHTTPOpenRedir(const char *URL, char **contentType, char **redir) {
1256 if (contentType != NULL) *contentType = NULL;
1257 if (redir != NULL) *redir = NULL;
1258 return(xmlNanoHTTPMethodRedir(URL, NULL, NULL, contentType, redir, NULL,0));
1259 }
1260
1261 /**
1262 * xmlNanoHTTPRead:
1263 * @ctx: the HTTP context
1264 * @dest: a buffer
1265 * @len: the buffer length
1266 *
1267 * This function tries to read @len bytes from the existing HTTP connection
1268 * and saves them in @dest. This is a blocking call.
1269 *
1270 * Returns the number of byte read. 0 is an indication of an end of connection.
1271 * -1 indicates a parameter error.
1272 */
1273 int
xmlNanoHTTPRead(void * ctx,void * dest,int len)1274 xmlNanoHTTPRead(void *ctx, void *dest, int len) {
1275 xmlNanoHTTPCtxtPtr ctxt = (xmlNanoHTTPCtxtPtr) ctx;
1276 #ifdef LIBXML_ZLIB_ENABLED
1277 int bytes_read = 0;
1278 int orig_avail_in;
1279 int z_ret;
1280 #endif
1281
1282 if (ctx == NULL) return(-1);
1283 if (dest == NULL) return(-1);
1284 if (len <= 0) return(0);
1285
1286 #ifdef LIBXML_ZLIB_ENABLED
1287 if (ctxt->usesGzip == 1) {
1288 if (ctxt->strm == NULL) return(0);
1289
1290 ctxt->strm->next_out = dest;
1291 ctxt->strm->avail_out = len;
1292 ctxt->strm->avail_in = ctxt->inptr - ctxt->inrptr;
1293
1294 while (ctxt->strm->avail_out > 0 &&
1295 (ctxt->strm->avail_in > 0 || xmlNanoHTTPRecv(ctxt) > 0)) {
1296 orig_avail_in = ctxt->strm->avail_in =
1297 ctxt->inptr - ctxt->inrptr - bytes_read;
1298 ctxt->strm->next_in = BAD_CAST (ctxt->inrptr + bytes_read);
1299
1300 z_ret = inflate(ctxt->strm, Z_NO_FLUSH);
1301 bytes_read += orig_avail_in - ctxt->strm->avail_in;
1302
1303 if (z_ret != Z_OK) break;
1304 }
1305
1306 ctxt->inrptr += bytes_read;
1307 return(len - ctxt->strm->avail_out);
1308 }
1309 #endif
1310
1311 while (ctxt->inptr - ctxt->inrptr < len) {
1312 if (xmlNanoHTTPRecv(ctxt) <= 0) break;
1313 }
1314 if (ctxt->inptr - ctxt->inrptr < len)
1315 len = ctxt->inptr - ctxt->inrptr;
1316 memcpy(dest, ctxt->inrptr, len);
1317 ctxt->inrptr += len;
1318 return(len);
1319 }
1320
1321 /**
1322 * xmlNanoHTTPClose:
1323 * @ctx: the HTTP context
1324 *
1325 * This function closes an HTTP context, it ends up the connection and
1326 * free all data related to it.
1327 */
1328 void
xmlNanoHTTPClose(void * ctx)1329 xmlNanoHTTPClose(void *ctx) {
1330 xmlNanoHTTPCtxtPtr ctxt = (xmlNanoHTTPCtxtPtr) ctx;
1331
1332 if (ctx == NULL) return;
1333
1334 xmlNanoHTTPFreeCtxt(ctxt);
1335 }
1336
1337 /**
1338 * xmlNanoHTTPMethodRedir:
1339 * @URL: The URL to load
1340 * @method: the HTTP method to use
1341 * @input: the input string if any
1342 * @contentType: the Content-Type information IN and OUT
1343 * @redir: the redirected URL OUT
1344 * @headers: the extra headers
1345 * @ilen: input length
1346 *
1347 * This function try to open a connection to the indicated resource
1348 * via HTTP using the given @method, adding the given extra headers
1349 * and the input buffer for the request content.
1350 *
1351 * Returns NULL in case of failure, otherwise a request handler.
1352 * The contentType, or redir, if provided must be freed by the caller
1353 */
1354
1355 void*
xmlNanoHTTPMethodRedir(const char * URL,const char * method,const char * input,char ** contentType,char ** redir,const char * headers,int ilen)1356 xmlNanoHTTPMethodRedir(const char *URL, const char *method, const char *input,
1357 char **contentType, char **redir,
1358 const char *headers, int ilen ) {
1359 xmlNanoHTTPCtxtPtr ctxt;
1360 char *bp, *p;
1361 int blen;
1362 SOCKET ret;
1363 int nbRedirects = 0;
1364 char *redirURL = NULL;
1365 #ifdef DEBUG_HTTP
1366 int xmt_bytes;
1367 #endif
1368
1369 if (URL == NULL) return(NULL);
1370 if (method == NULL) method = "GET";
1371 xmlNanoHTTPInit();
1372
1373 retry:
1374 if (redirURL == NULL) {
1375 ctxt = xmlNanoHTTPNewCtxt(URL);
1376 if (ctxt == NULL)
1377 return(NULL);
1378 } else {
1379 ctxt = xmlNanoHTTPNewCtxt(redirURL);
1380 if (ctxt == NULL)
1381 return(NULL);
1382 ctxt->location = xmlMemStrdup(redirURL);
1383 }
1384
1385 if ((ctxt->protocol == NULL) || (strcmp(ctxt->protocol, "http"))) {
1386 __xmlIOErr(XML_FROM_HTTP, XML_HTTP_URL_SYNTAX, "Not a valid HTTP URI");
1387 xmlNanoHTTPFreeCtxt(ctxt);
1388 if (redirURL != NULL) xmlFree(redirURL);
1389 return(NULL);
1390 }
1391 if (ctxt->hostname == NULL) {
1392 __xmlIOErr(XML_FROM_HTTP, XML_HTTP_UNKNOWN_HOST,
1393 "Failed to identify host in URI");
1394 xmlNanoHTTPFreeCtxt(ctxt);
1395 if (redirURL != NULL) xmlFree(redirURL);
1396 return(NULL);
1397 }
1398 if (proxy) {
1399 blen = strlen(ctxt->hostname) * 2 + 16;
1400 ret = xmlNanoHTTPConnectHost(proxy, proxyPort);
1401 }
1402 else {
1403 blen = strlen(ctxt->hostname);
1404 ret = xmlNanoHTTPConnectHost(ctxt->hostname, ctxt->port);
1405 }
1406 if (ret == INVALID_SOCKET) {
1407 xmlNanoHTTPFreeCtxt(ctxt);
1408 if (redirURL != NULL) xmlFree(redirURL);
1409 return(NULL);
1410 }
1411 ctxt->fd = ret;
1412
1413 if (input == NULL)
1414 ilen = 0;
1415 else
1416 blen += 36;
1417
1418 if (headers != NULL)
1419 blen += strlen(headers) + 2;
1420 if (contentType && *contentType)
1421 /* reserve for string plus 'Content-Type: \r\n" */
1422 blen += strlen(*contentType) + 16;
1423 if (ctxt->query != NULL)
1424 /* 1 for '?' */
1425 blen += strlen(ctxt->query) + 1;
1426 blen += strlen(method) + strlen(ctxt->path) + 24;
1427 #ifdef LIBXML_ZLIB_ENABLED
1428 /* reserve for possible 'Accept-Encoding: gzip' string */
1429 blen += 23;
1430 #endif
1431 if (ctxt->port != 80) {
1432 /* reserve space for ':xxxxx', incl. potential proxy */
1433 if (proxy)
1434 blen += 17;
1435 else
1436 blen += 11;
1437 }
1438 bp = (char*)xmlMallocAtomic(blen);
1439 if ( bp == NULL ) {
1440 xmlNanoHTTPFreeCtxt( ctxt );
1441 xmlHTTPErrMemory("allocating header buffer");
1442 return ( NULL );
1443 }
1444
1445 p = bp;
1446
1447 if (proxy) {
1448 if (ctxt->port != 80) {
1449 p += snprintf( p, blen - (p - bp), "%s http://%s:%d%s",
1450 method, ctxt->hostname,
1451 ctxt->port, ctxt->path );
1452 }
1453 else
1454 p += snprintf( p, blen - (p - bp), "%s http://%s%s", method,
1455 ctxt->hostname, ctxt->path);
1456 }
1457 else
1458 p += snprintf( p, blen - (p - bp), "%s %s", method, ctxt->path);
1459
1460 if (ctxt->query != NULL)
1461 p += snprintf( p, blen - (p - bp), "?%s", ctxt->query);
1462
1463 if (ctxt->port == 80) {
1464 p += snprintf( p, blen - (p - bp), " HTTP/1.0\r\nHost: %s\r\n",
1465 ctxt->hostname);
1466 } else {
1467 p += snprintf( p, blen - (p - bp), " HTTP/1.0\r\nHost: %s:%d\r\n",
1468 ctxt->hostname, ctxt->port);
1469 }
1470
1471 #ifdef LIBXML_ZLIB_ENABLED
1472 p += snprintf(p, blen - (p - bp), "Accept-Encoding: gzip\r\n");
1473 #endif
1474
1475 if (contentType != NULL && *contentType)
1476 p += snprintf(p, blen - (p - bp), "Content-Type: %s\r\n", *contentType);
1477
1478 if (headers != NULL)
1479 p += snprintf( p, blen - (p - bp), "%s", headers );
1480
1481 if (input != NULL)
1482 snprintf(p, blen - (p - bp), "Content-Length: %d\r\n\r\n", ilen );
1483 else
1484 snprintf(p, blen - (p - bp), "\r\n");
1485
1486 #ifdef DEBUG_HTTP
1487 xmlGenericError(xmlGenericErrorContext,
1488 "-> %s%s", proxy? "(Proxy) " : "", bp);
1489 if ((blen -= strlen(bp)+1) < 0)
1490 xmlGenericError(xmlGenericErrorContext,
1491 "ERROR: overflowed buffer by %d bytes\n", -blen);
1492 #endif
1493 ctxt->outptr = ctxt->out = bp;
1494 ctxt->state = XML_NANO_HTTP_WRITE;
1495 blen = strlen( ctxt->out );
1496 #ifdef DEBUG_HTTP
1497 xmt_bytes = xmlNanoHTTPSend(ctxt, ctxt->out, blen );
1498 if ( xmt_bytes != blen )
1499 xmlGenericError( xmlGenericErrorContext,
1500 "xmlNanoHTTPMethodRedir: Only %d of %d %s %s\n",
1501 xmt_bytes, blen,
1502 "bytes of HTTP headers sent to host",
1503 ctxt->hostname );
1504 #else
1505 xmlNanoHTTPSend(ctxt, ctxt->out, blen );
1506 #endif
1507
1508 if ( input != NULL ) {
1509 #ifdef DEBUG_HTTP
1510 xmt_bytes = xmlNanoHTTPSend( ctxt, input, ilen );
1511
1512 if ( xmt_bytes != ilen )
1513 xmlGenericError( xmlGenericErrorContext,
1514 "xmlNanoHTTPMethodRedir: Only %d of %d %s %s\n",
1515 xmt_bytes, ilen,
1516 "bytes of HTTP content sent to host",
1517 ctxt->hostname );
1518 #else
1519 xmlNanoHTTPSend( ctxt, input, ilen );
1520 #endif
1521 }
1522
1523 ctxt->state = XML_NANO_HTTP_READ;
1524
1525 while ((p = xmlNanoHTTPReadLine(ctxt)) != NULL) {
1526 if (*p == 0) {
1527 ctxt->content = ctxt->inrptr;
1528 xmlFree(p);
1529 break;
1530 }
1531 xmlNanoHTTPScanAnswer(ctxt, p);
1532
1533 #ifdef DEBUG_HTTP
1534 xmlGenericError(xmlGenericErrorContext, "<- %s\n", p);
1535 #endif
1536 xmlFree(p);
1537 }
1538
1539 if ((ctxt->location != NULL) && (ctxt->returnValue >= 300) &&
1540 (ctxt->returnValue < 400)) {
1541 #ifdef DEBUG_HTTP
1542 xmlGenericError(xmlGenericErrorContext,
1543 "\nRedirect to: %s\n", ctxt->location);
1544 #endif
1545 while ( xmlNanoHTTPRecv(ctxt) > 0 )
1546 ;
1547 if (nbRedirects < XML_NANO_HTTP_MAX_REDIR) {
1548 nbRedirects++;
1549 if (redirURL != NULL)
1550 xmlFree(redirURL);
1551 redirURL = xmlMemStrdup(ctxt->location);
1552 xmlNanoHTTPFreeCtxt(ctxt);
1553 goto retry;
1554 }
1555 xmlNanoHTTPFreeCtxt(ctxt);
1556 if (redirURL != NULL) xmlFree(redirURL);
1557 #ifdef DEBUG_HTTP
1558 xmlGenericError(xmlGenericErrorContext,
1559 "xmlNanoHTTPMethodRedir: Too many redirects, aborting ...\n");
1560 #endif
1561 return(NULL);
1562 }
1563
1564 if (contentType != NULL) {
1565 if (ctxt->contentType != NULL)
1566 *contentType = xmlMemStrdup(ctxt->contentType);
1567 else
1568 *contentType = NULL;
1569 }
1570
1571 if ((redir != NULL) && (redirURL != NULL)) {
1572 *redir = redirURL;
1573 } else {
1574 if (redirURL != NULL)
1575 xmlFree(redirURL);
1576 if (redir != NULL)
1577 *redir = NULL;
1578 }
1579
1580 #ifdef DEBUG_HTTP
1581 if (ctxt->contentType != NULL)
1582 xmlGenericError(xmlGenericErrorContext,
1583 "\nCode %d, content-type '%s'\n\n",
1584 ctxt->returnValue, ctxt->contentType);
1585 else
1586 xmlGenericError(xmlGenericErrorContext,
1587 "\nCode %d, no content-type\n\n",
1588 ctxt->returnValue);
1589 #endif
1590
1591 return((void *) ctxt);
1592 }
1593
1594 /**
1595 * xmlNanoHTTPMethod:
1596 * @URL: The URL to load
1597 * @method: the HTTP method to use
1598 * @input: the input string if any
1599 * @contentType: the Content-Type information IN and OUT
1600 * @headers: the extra headers
1601 * @ilen: input length
1602 *
1603 * This function try to open a connection to the indicated resource
1604 * via HTTP using the given @method, adding the given extra headers
1605 * and the input buffer for the request content.
1606 *
1607 * Returns NULL in case of failure, otherwise a request handler.
1608 * The contentType, if provided must be freed by the caller
1609 */
1610
1611 void*
xmlNanoHTTPMethod(const char * URL,const char * method,const char * input,char ** contentType,const char * headers,int ilen)1612 xmlNanoHTTPMethod(const char *URL, const char *method, const char *input,
1613 char **contentType, const char *headers, int ilen) {
1614 return(xmlNanoHTTPMethodRedir(URL, method, input, contentType,
1615 NULL, headers, ilen));
1616 }
1617
1618 /**
1619 * xmlNanoHTTPFetch:
1620 * @URL: The URL to load
1621 * @filename: the filename where the content should be saved
1622 * @contentType: if available the Content-Type information will be
1623 * returned at that location
1624 *
1625 * This function try to fetch the indicated resource via HTTP GET
1626 * and save it's content in the file.
1627 *
1628 * Returns -1 in case of failure, 0 in case of success. The contentType,
1629 * if provided must be freed by the caller
1630 */
1631 int
xmlNanoHTTPFetch(const char * URL,const char * filename,char ** contentType)1632 xmlNanoHTTPFetch(const char *URL, const char *filename, char **contentType) {
1633 void *ctxt = NULL;
1634 char *buf = NULL;
1635 int fd;
1636 int len;
1637 int ret = 0;
1638
1639 if (filename == NULL) return(-1);
1640 ctxt = xmlNanoHTTPOpen(URL, contentType);
1641 if (ctxt == NULL) return(-1);
1642
1643 if (!strcmp(filename, "-"))
1644 fd = 0;
1645 else {
1646 fd = open(filename, O_CREAT | O_WRONLY, 00644);
1647 if (fd < 0) {
1648 xmlNanoHTTPClose(ctxt);
1649 if ((contentType != NULL) && (*contentType != NULL)) {
1650 xmlFree(*contentType);
1651 *contentType = NULL;
1652 }
1653 return(-1);
1654 }
1655 }
1656
1657 xmlNanoHTTPFetchContent( ctxt, &buf, &len );
1658 if ( len > 0 ) {
1659 if (write(fd, buf, len) == -1) {
1660 ret = -1;
1661 }
1662 }
1663
1664 xmlNanoHTTPClose(ctxt);
1665 close(fd);
1666 return(ret);
1667 }
1668
1669 #ifdef LIBXML_OUTPUT_ENABLED
1670 /**
1671 * xmlNanoHTTPSave:
1672 * @ctxt: the HTTP context
1673 * @filename: the filename where the content should be saved
1674 *
1675 * This function saves the output of the HTTP transaction to a file
1676 * It closes and free the context at the end
1677 *
1678 * Returns -1 in case of failure, 0 in case of success.
1679 */
1680 int
xmlNanoHTTPSave(void * ctxt,const char * filename)1681 xmlNanoHTTPSave(void *ctxt, const char *filename) {
1682 char *buf = NULL;
1683 int fd;
1684 int len;
1685 int ret = 0;
1686
1687 if ((ctxt == NULL) || (filename == NULL)) return(-1);
1688
1689 if (!strcmp(filename, "-"))
1690 fd = 0;
1691 else {
1692 fd = open(filename, O_CREAT | O_WRONLY, 0666);
1693 if (fd < 0) {
1694 xmlNanoHTTPClose(ctxt);
1695 return(-1);
1696 }
1697 }
1698
1699 xmlNanoHTTPFetchContent( ctxt, &buf, &len );
1700 if ( len > 0 ) {
1701 if (write(fd, buf, len) == -1) {
1702 ret = -1;
1703 }
1704 }
1705
1706 xmlNanoHTTPClose(ctxt);
1707 close(fd);
1708 return(ret);
1709 }
1710 #endif /* LIBXML_OUTPUT_ENABLED */
1711
1712 /**
1713 * xmlNanoHTTPReturnCode:
1714 * @ctx: the HTTP context
1715 *
1716 * Get the latest HTTP return code received
1717 *
1718 * Returns the HTTP return code for the request.
1719 */
1720 int
xmlNanoHTTPReturnCode(void * ctx)1721 xmlNanoHTTPReturnCode(void *ctx) {
1722 xmlNanoHTTPCtxtPtr ctxt = (xmlNanoHTTPCtxtPtr) ctx;
1723
1724 if (ctxt == NULL) return(-1);
1725
1726 return(ctxt->returnValue);
1727 }
1728
1729 /**
1730 * xmlNanoHTTPAuthHeader:
1731 * @ctx: the HTTP context
1732 *
1733 * Get the authentication header of an HTTP context
1734 *
1735 * Returns the stashed value of the WWW-Authenticate or Proxy-Authenticate
1736 * header.
1737 */
1738 const char *
xmlNanoHTTPAuthHeader(void * ctx)1739 xmlNanoHTTPAuthHeader(void *ctx) {
1740 xmlNanoHTTPCtxtPtr ctxt = (xmlNanoHTTPCtxtPtr) ctx;
1741
1742 if (ctxt == NULL) return(NULL);
1743
1744 return(ctxt->authHeader);
1745 }
1746
1747 /**
1748 * xmlNanoHTTPContentLength:
1749 * @ctx: the HTTP context
1750 *
1751 * Provides the specified content length from the HTTP header.
1752 *
1753 * Return the specified content length from the HTTP header. Note that
1754 * a value of -1 indicates that the content length element was not included in
1755 * the response header.
1756 */
1757 int
xmlNanoHTTPContentLength(void * ctx)1758 xmlNanoHTTPContentLength( void * ctx ) {
1759 xmlNanoHTTPCtxtPtr ctxt = (xmlNanoHTTPCtxtPtr)ctx;
1760
1761 return ( ( ctxt == NULL ) ? -1 : ctxt->ContentLength );
1762 }
1763
1764 /**
1765 * xmlNanoHTTPRedir:
1766 * @ctx: the HTTP context
1767 *
1768 * Provides the specified redirection URL if available from the HTTP header.
1769 *
1770 * Return the specified redirection URL or NULL if not redirected.
1771 */
1772 const char *
xmlNanoHTTPRedir(void * ctx)1773 xmlNanoHTTPRedir( void * ctx ) {
1774 xmlNanoHTTPCtxtPtr ctxt = (xmlNanoHTTPCtxtPtr)ctx;
1775
1776 return ( ( ctxt == NULL ) ? NULL : ctxt->location );
1777 }
1778
1779 /**
1780 * xmlNanoHTTPEncoding:
1781 * @ctx: the HTTP context
1782 *
1783 * Provides the specified encoding if specified in the HTTP headers.
1784 *
1785 * Return the specified encoding or NULL if not available
1786 */
1787 const char *
xmlNanoHTTPEncoding(void * ctx)1788 xmlNanoHTTPEncoding( void * ctx ) {
1789 xmlNanoHTTPCtxtPtr ctxt = (xmlNanoHTTPCtxtPtr)ctx;
1790
1791 return ( ( ctxt == NULL ) ? NULL : ctxt->encoding );
1792 }
1793
1794 /**
1795 * xmlNanoHTTPMimeType:
1796 * @ctx: the HTTP context
1797 *
1798 * Provides the specified Mime-Type if specified in the HTTP headers.
1799 *
1800 * Return the specified Mime-Type or NULL if not available
1801 */
1802 const char *
xmlNanoHTTPMimeType(void * ctx)1803 xmlNanoHTTPMimeType( void * ctx ) {
1804 xmlNanoHTTPCtxtPtr ctxt = (xmlNanoHTTPCtxtPtr)ctx;
1805
1806 return ( ( ctxt == NULL ) ? NULL : ctxt->mimeType );
1807 }
1808
1809 /**
1810 * xmlNanoHTTPFetchContent:
1811 * @ctx: the HTTP context
1812 * @ptr: pointer to set to the content buffer.
1813 * @len: integer pointer to hold the length of the content
1814 *
1815 * Check if all the content was read
1816 *
1817 * Returns 0 if all the content was read and available, returns
1818 * -1 if received content length was less than specified or an error
1819 * occurred.
1820 */
1821 static int
xmlNanoHTTPFetchContent(void * ctx,char ** ptr,int * len)1822 xmlNanoHTTPFetchContent( void * ctx, char ** ptr, int * len ) {
1823 xmlNanoHTTPCtxtPtr ctxt = (xmlNanoHTTPCtxtPtr)ctx;
1824
1825 int rc = 0;
1826 int cur_lgth;
1827 int rcvd_lgth;
1828 int dummy_int;
1829 char * dummy_ptr = NULL;
1830
1831 /* Dummy up return input parameters if not provided */
1832
1833 if ( len == NULL )
1834 len = &dummy_int;
1835
1836 if ( ptr == NULL )
1837 ptr = &dummy_ptr;
1838
1839 /* But can't work without the context pointer */
1840
1841 if ( ( ctxt == NULL ) || ( ctxt->content == NULL ) ) {
1842 *len = 0;
1843 *ptr = NULL;
1844 return ( -1 );
1845 }
1846
1847 rcvd_lgth = ctxt->inptr - ctxt->content;
1848
1849 while ( (cur_lgth = xmlNanoHTTPRecv( ctxt )) > 0 ) {
1850
1851 rcvd_lgth += cur_lgth;
1852 if ( (ctxt->ContentLength > 0) && (rcvd_lgth >= ctxt->ContentLength) )
1853 break;
1854 }
1855
1856 *ptr = ctxt->content;
1857 *len = rcvd_lgth;
1858
1859 if ( ( ctxt->ContentLength > 0 ) && ( rcvd_lgth < ctxt->ContentLength ) )
1860 rc = -1;
1861 else if ( rcvd_lgth == 0 )
1862 rc = -1;
1863
1864 return ( rc );
1865 }
1866
1867 #ifdef STANDALONE
main(int argc,char ** argv)1868 int main(int argc, char **argv) {
1869 char *contentType = NULL;
1870
1871 if (argv[1] != NULL) {
1872 if (argv[2] != NULL)
1873 xmlNanoHTTPFetch(argv[1], argv[2], &contentType);
1874 else
1875 xmlNanoHTTPFetch(argv[1], "-", &contentType);
1876 if (contentType != NULL) xmlFree(contentType);
1877 } else {
1878 xmlGenericError(xmlGenericErrorContext,
1879 "%s: minimal HTTP GET implementation\n", argv[0]);
1880 xmlGenericError(xmlGenericErrorContext,
1881 "\tusage %s [ URL [ filename ] ]\n", argv[0]);
1882 }
1883 xmlNanoHTTPCleanup();
1884 xmlMemoryDump();
1885 return(0);
1886 }
1887 #endif /* STANDALONE */
1888 #else /* !LIBXML_HTTP_ENABLED */
1889 #ifdef STANDALONE
1890 #include <stdio.h>
main(int argc,char ** argv)1891 int main(int argc, char **argv) {
1892 xmlGenericError(xmlGenericErrorContext,
1893 "%s : HTTP support not compiled in\n", argv[0]);
1894 return(0);
1895 }
1896 #endif /* STANDALONE */
1897 #endif /* LIBXML_HTTP_ENABLED */
1898 #define bottom_nanohttp
1899 #include "elfgcchack.h"
1900