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 NEED_SOCKETS
15 #define IN_LIBXML
16 #include "libxml.h"
17
18 #ifdef LIBXML_HTTP_ENABLED
19 #include <string.h>
20
21 #ifdef HAVE_STDLIB_H
22 #include <stdlib.h>
23 #endif
24 #ifdef HAVE_UNISTD_H
25 #include <unistd.h>
26 #endif
27 #ifdef HAVE_SYS_TYPES_H
28 #include <sys/types.h>
29 #endif
30 #ifdef HAVE_SYS_SOCKET_H
31 #include <sys/socket.h>
32 #endif
33 #ifdef HAVE_NETINET_IN_H
34 #include <netinet/in.h>
35 #endif
36 #ifdef HAVE_ARPA_INET_H
37 #include <arpa/inet.h>
38 #endif
39 #ifdef HAVE_NETDB_H
40 #include <netdb.h>
41 #endif
42 #ifdef HAVE_RESOLV_H
43 #ifdef HAVE_ARPA_NAMESER_H
44 #include <arpa/nameser.h>
45 #endif
46 #include <resolv.h>
47 #endif
48 #ifdef HAVE_FCNTL_H
49 #include <fcntl.h>
50 #endif
51 #ifdef HAVE_ERRNO_H
52 #include <errno.h>
53 #endif
54 #ifdef HAVE_SYS_TIME_H
55 #include <sys/time.h>
56 #endif
57 #ifndef HAVE_POLL_H
58 #ifdef HAVE_SYS_SELECT_H
59 #include <sys/select.h>
60 #endif
61 #else
62 #include <poll.h>
63 #endif
64 #ifdef HAVE_STRINGS_H
65 #include <strings.h>
66 #endif
67 #ifdef HAVE_ZLIB_H
68 #include <zlib.h>
69 #endif
70
71
72 #ifdef VMS
73 #include <stropts>
74 #define XML_SOCKLEN_T unsigned int
75 #endif
76
77 #if defined(__MINGW32__) || defined(_WIN32_WCE)
78 #ifndef _WINSOCKAPI_
79 #define _WINSOCKAPI_
80 #endif
81 #include <wsockcompat.h>
82 #include <winsock2.h>
83 #undef XML_SOCKLEN_T
84 #define XML_SOCKLEN_T unsigned int
85 #endif
86
87 #include <libxml/globals.h>
88 #include <libxml/xmlerror.h>
89 #include <libxml/xmlmemory.h>
90 #include <libxml/parser.h> /* for xmlStr(n)casecmp() */
91 #include <libxml/nanohttp.h>
92 #include <libxml/globals.h>
93 #include <libxml/uri.h>
94
95 /**
96 * A couple portability macros
97 */
98 #ifndef _WINSOCKAPI_
99 #if !defined(__BEOS__) || defined(__HAIKU__)
100 #define closesocket(s) close(s)
101 #endif
102 #define SOCKET int
103 #define INVALID_SOCKET (-1)
104 #endif
105
106 #ifdef __BEOS__
107 #ifndef PF_INET
108 #define PF_INET AF_INET
109 #endif
110 #endif
111
112 #ifndef XML_SOCKLEN_T
113 #define XML_SOCKLEN_T unsigned int
114 #endif
115
116 #ifdef STANDALONE
117 #define DEBUG_HTTP
118 #define xmlStrncasecmp(a, b, n) strncasecmp((char *)a, (char *)b, n)
119 #define xmlStrcasecmpi(a, b) strcasecmp((char *)a, (char *)b)
120 #endif
121
122 #define XML_NANO_HTTP_MAX_REDIR 10
123
124 #define XML_NANO_HTTP_CHUNK 4096
125
126 #define XML_NANO_HTTP_CLOSED 0
127 #define XML_NANO_HTTP_WRITE 1
128 #define XML_NANO_HTTP_READ 2
129 #define XML_NANO_HTTP_NONE 4
130
131 typedef struct xmlNanoHTTPCtxt {
132 char *protocol; /* the protocol name */
133 char *hostname; /* the host name */
134 int port; /* the port */
135 char *path; /* the path within the URL */
136 char *query; /* the query string */
137 SOCKET fd; /* the file descriptor for the socket */
138 int state; /* WRITE / READ / CLOSED */
139 char *out; /* buffer sent (zero terminated) */
140 char *outptr; /* index within the buffer sent */
141 char *in; /* the receiving buffer */
142 char *content; /* the start of the content */
143 char *inptr; /* the next byte to read from network */
144 char *inrptr; /* the next byte to give back to the client */
145 int inlen; /* len of the input buffer */
146 int last; /* return code for last operation */
147 int returnValue; /* the protocol return value */
148 int version; /* the protocol version */
149 int ContentLength; /* specified content length from HTTP header */
150 char *contentType; /* the MIME type for the input */
151 char *location; /* the new URL in case of redirect */
152 char *authHeader; /* contents of {WWW,Proxy}-Authenticate header */
153 char *encoding; /* encoding extracted from the contentType */
154 char *mimeType; /* Mime-Type extracted from the contentType */
155 #ifdef HAVE_ZLIB_H
156 z_stream *strm; /* Zlib stream object */
157 int usesGzip; /* "Content-Encoding: gzip" was detected */
158 #endif
159 } xmlNanoHTTPCtxt, *xmlNanoHTTPCtxtPtr;
160
161 static int initialized = 0;
162 static char *proxy = NULL; /* the proxy name if any */
163 static int proxyPort; /* the proxy port if any */
164 static unsigned int timeout = 60;/* the select() timeout in seconds */
165
166 static int xmlNanoHTTPFetchContent( void * ctx, char ** ptr, int * len );
167
168 /**
169 * xmlHTTPErrMemory:
170 * @extra: extra informations
171 *
172 * Handle an out of memory condition
173 */
174 static void
xmlHTTPErrMemory(const char * extra)175 xmlHTTPErrMemory(const char *extra)
176 {
177 __xmlSimpleError(XML_FROM_HTTP, XML_ERR_NO_MEMORY, NULL, NULL, extra);
178 }
179
180 /**
181 * A portability function
182 */
socket_errno(void)183 static int socket_errno(void) {
184 #ifdef _WINSOCKAPI_
185 return(WSAGetLastError());
186 #else
187 return(errno);
188 #endif
189 }
190
191 #ifdef SUPPORT_IP6
192 static
have_ipv6(void)193 int have_ipv6(void) {
194 SOCKET s;
195
196 s = socket (AF_INET6, SOCK_STREAM, 0);
197 if (s != INVALID_SOCKET) {
198 close (s);
199 return (1);
200 }
201 return (0);
202 }
203 #endif
204
205 /**
206 * xmlNanoHTTPInit:
207 *
208 * Initialize the HTTP protocol layer.
209 * Currently it just checks for proxy informations
210 */
211
212 void
xmlNanoHTTPInit(void)213 xmlNanoHTTPInit(void) {
214 const char *env;
215 #ifdef _WINSOCKAPI_
216 WSADATA wsaData;
217 #endif
218
219 if (initialized)
220 return;
221
222 #ifdef _WINSOCKAPI_
223 if (WSAStartup(MAKEWORD(1, 1), &wsaData) != 0)
224 return;
225 #endif
226
227 if (proxy == NULL) {
228 proxyPort = 80;
229 env = getenv("no_proxy");
230 if (env && ((env[0] == '*') && (env[1] == 0)))
231 goto done;
232 env = getenv("http_proxy");
233 if (env != NULL) {
234 xmlNanoHTTPScanProxy(env);
235 goto done;
236 }
237 env = getenv("HTTP_PROXY");
238 if (env != NULL) {
239 xmlNanoHTTPScanProxy(env);
240 goto done;
241 }
242 }
243 done:
244 initialized = 1;
245 }
246
247 /**
248 * xmlNanoHTTPCleanup:
249 *
250 * Cleanup the HTTP protocol layer.
251 */
252
253 void
xmlNanoHTTPCleanup(void)254 xmlNanoHTTPCleanup(void) {
255 if (proxy != NULL) {
256 xmlFree(proxy);
257 proxy = NULL;
258 }
259 #ifdef _WINSOCKAPI_
260 if (initialized)
261 WSACleanup();
262 #endif
263 initialized = 0;
264 return;
265 }
266
267 /**
268 * xmlNanoHTTPScanURL:
269 * @ctxt: an HTTP context
270 * @URL: The URL used to initialize the context
271 *
272 * (Re)Initialize an HTTP context by parsing the URL and finding
273 * the protocol host port and path it indicates.
274 */
275
276 static void
xmlNanoHTTPScanURL(xmlNanoHTTPCtxtPtr ctxt,const char * URL)277 xmlNanoHTTPScanURL(xmlNanoHTTPCtxtPtr ctxt, const char *URL) {
278 xmlURIPtr uri;
279 int len;
280
281 /*
282 * Clear any existing data from the context
283 */
284 if (ctxt->protocol != NULL) {
285 xmlFree(ctxt->protocol);
286 ctxt->protocol = NULL;
287 }
288 if (ctxt->hostname != NULL) {
289 xmlFree(ctxt->hostname);
290 ctxt->hostname = NULL;
291 }
292 if (ctxt->path != NULL) {
293 xmlFree(ctxt->path);
294 ctxt->path = NULL;
295 }
296 if (ctxt->query != NULL) {
297 xmlFree(ctxt->query);
298 ctxt->query = NULL;
299 }
300 if (URL == NULL) return;
301
302 uri = xmlParseURIRaw(URL, 1);
303 if (uri == NULL)
304 return;
305
306 if ((uri->scheme == NULL) || (uri->server == NULL)) {
307 xmlFreeURI(uri);
308 return;
309 }
310
311 ctxt->protocol = xmlMemStrdup(uri->scheme);
312 /* special case of IPv6 addresses, the [] need to be removed */
313 if ((uri->server != NULL) && (*uri->server == '[')) {
314 len = strlen(uri->server);
315 if ((len > 2) && (uri->server[len - 1] == ']')) {
316 ctxt->hostname = (char *) xmlCharStrndup(uri->server + 1, len -2);
317 } else
318 ctxt->hostname = xmlMemStrdup(uri->server);
319 } else
320 ctxt->hostname = xmlMemStrdup(uri->server);
321 if (uri->path != NULL)
322 ctxt->path = xmlMemStrdup(uri->path);
323 else
324 ctxt->path = xmlMemStrdup("/");
325 if (uri->query != NULL)
326 ctxt->query = xmlMemStrdup(uri->query);
327 if (uri->port != 0)
328 ctxt->port = uri->port;
329
330 xmlFreeURI(uri);
331 }
332
333 /**
334 * xmlNanoHTTPScanProxy:
335 * @URL: The proxy URL used to initialize the proxy context
336 *
337 * (Re)Initialize the HTTP Proxy context by parsing the URL and finding
338 * the protocol host port it indicates.
339 * Should be like http://myproxy/ or http://myproxy:3128/
340 * A NULL URL cleans up proxy informations.
341 */
342
343 void
xmlNanoHTTPScanProxy(const char * URL)344 xmlNanoHTTPScanProxy(const char *URL) {
345 xmlURIPtr uri;
346
347 if (proxy != NULL) {
348 xmlFree(proxy);
349 proxy = NULL;
350 }
351 proxyPort = 0;
352
353 #ifdef DEBUG_HTTP
354 if (URL == NULL)
355 xmlGenericError(xmlGenericErrorContext,
356 "Removing HTTP proxy info\n");
357 else
358 xmlGenericError(xmlGenericErrorContext,
359 "Using HTTP proxy %s\n", URL);
360 #endif
361 if (URL == NULL) return;
362
363 uri = xmlParseURIRaw(URL, 1);
364 if ((uri == NULL) || (uri->scheme == NULL) ||
365 (strcmp(uri->scheme, "http")) || (uri->server == NULL)) {
366 __xmlIOErr(XML_FROM_HTTP, XML_HTTP_URL_SYNTAX, "Syntax Error\n");
367 if (uri != NULL)
368 xmlFreeURI(uri);
369 return;
370 }
371
372 proxy = xmlMemStrdup(uri->server);
373 if (uri->port != 0)
374 proxyPort = uri->port;
375
376 xmlFreeURI(uri);
377 }
378
379 /**
380 * xmlNanoHTTPNewCtxt:
381 * @URL: The URL used to initialize the context
382 *
383 * Allocate and initialize a new HTTP context.
384 *
385 * Returns an HTTP context or NULL in case of error.
386 */
387
388 static xmlNanoHTTPCtxtPtr
xmlNanoHTTPNewCtxt(const char * URL)389 xmlNanoHTTPNewCtxt(const char *URL) {
390 xmlNanoHTTPCtxtPtr ret;
391
392 ret = (xmlNanoHTTPCtxtPtr) xmlMalloc(sizeof(xmlNanoHTTPCtxt));
393 if (ret == NULL) {
394 xmlHTTPErrMemory("allocating context");
395 return(NULL);
396 }
397
398 memset(ret, 0, sizeof(xmlNanoHTTPCtxt));
399 ret->port = 80;
400 ret->returnValue = 0;
401 ret->fd = INVALID_SOCKET;
402 ret->ContentLength = -1;
403
404 xmlNanoHTTPScanURL(ret, URL);
405
406 return(ret);
407 }
408
409 /**
410 * xmlNanoHTTPFreeCtxt:
411 * @ctxt: an HTTP context
412 *
413 * Frees the context after closing the connection.
414 */
415
416 static void
xmlNanoHTTPFreeCtxt(xmlNanoHTTPCtxtPtr ctxt)417 xmlNanoHTTPFreeCtxt(xmlNanoHTTPCtxtPtr ctxt) {
418 if (ctxt == NULL) return;
419 if (ctxt->hostname != NULL) xmlFree(ctxt->hostname);
420 if (ctxt->protocol != NULL) xmlFree(ctxt->protocol);
421 if (ctxt->path != NULL) xmlFree(ctxt->path);
422 if (ctxt->query != NULL) xmlFree(ctxt->query);
423 if (ctxt->out != NULL) xmlFree(ctxt->out);
424 if (ctxt->in != NULL) xmlFree(ctxt->in);
425 if (ctxt->contentType != NULL) xmlFree(ctxt->contentType);
426 if (ctxt->encoding != NULL) xmlFree(ctxt->encoding);
427 if (ctxt->mimeType != NULL) xmlFree(ctxt->mimeType);
428 if (ctxt->location != NULL) xmlFree(ctxt->location);
429 if (ctxt->authHeader != NULL) xmlFree(ctxt->authHeader);
430 #ifdef HAVE_ZLIB_H
431 if (ctxt->strm != NULL) {
432 inflateEnd(ctxt->strm);
433 xmlFree(ctxt->strm);
434 }
435 #endif
436
437 ctxt->state = XML_NANO_HTTP_NONE;
438 if (ctxt->fd != INVALID_SOCKET) closesocket(ctxt->fd);
439 ctxt->fd = INVALID_SOCKET;
440 xmlFree(ctxt);
441 }
442
443 /**
444 * xmlNanoHTTPSend:
445 * @ctxt: an HTTP context
446 *
447 * Send the input needed to initiate the processing on the server side
448 * Returns number of bytes sent or -1 on error.
449 */
450
451 static int
xmlNanoHTTPSend(xmlNanoHTTPCtxtPtr ctxt,const char * xmt_ptr,int outlen)452 xmlNanoHTTPSend(xmlNanoHTTPCtxtPtr ctxt, const char *xmt_ptr, int outlen)
453 {
454 int total_sent = 0;
455 #ifdef HAVE_POLL_H
456 struct pollfd p;
457 #else
458 struct timeval tv;
459 fd_set wfd;
460 #endif
461
462 if ((ctxt->state & XML_NANO_HTTP_WRITE) && (xmt_ptr != NULL)) {
463 while (total_sent < outlen) {
464 int nsent = send(ctxt->fd, SEND_ARG2_CAST (xmt_ptr + total_sent),
465 outlen - total_sent, 0);
466
467 if (nsent > 0)
468 total_sent += nsent;
469 else if ((nsent == -1) &&
470 #if defined(EAGAIN) && EAGAIN != EWOULDBLOCK
471 (socket_errno() != EAGAIN) &&
472 #endif
473 (socket_errno() != EWOULDBLOCK)) {
474 __xmlIOErr(XML_FROM_HTTP, 0, "send failed\n");
475 if (total_sent == 0)
476 total_sent = -1;
477 break;
478 } else {
479 /*
480 * No data sent
481 * Since non-blocking sockets are used, wait for
482 * socket to be writable or default timeout prior
483 * to retrying.
484 */
485 #ifndef HAVE_POLL_H
486 #ifndef _WINSOCKAPI_
487 if (ctxt->fd > FD_SETSIZE)
488 return -1;
489 #endif
490
491 tv.tv_sec = timeout;
492 tv.tv_usec = 0;
493 FD_ZERO(&wfd);
494 #ifdef _MSC_VER
495 #pragma warning(push)
496 #pragma warning(disable: 4018)
497 #endif
498 FD_SET(ctxt->fd, &wfd);
499 #ifdef _MSC_VER
500 #pragma warning(pop)
501 #endif
502 (void) select(ctxt->fd + 1, NULL, &wfd, NULL, &tv);
503 #else
504 p.fd = ctxt->fd;
505 p.events = POLLOUT;
506 (void) poll(&p, 1, timeout * 1000);
507 #endif /* !HAVE_POLL_H */
508 }
509 }
510 }
511
512 return total_sent;
513 }
514
515 /**
516 * xmlNanoHTTPRecv:
517 * @ctxt: an HTTP context
518 *
519 * Read information coming from the HTTP connection.
520 * This is a blocking call (but it blocks in select(), not read()).
521 *
522 * Returns the number of byte read or -1 in case of error.
523 */
524
525 static int
xmlNanoHTTPRecv(xmlNanoHTTPCtxtPtr ctxt)526 xmlNanoHTTPRecv(xmlNanoHTTPCtxtPtr ctxt)
527 {
528 #ifdef HAVE_POLL_H
529 struct pollfd p;
530 #else
531 fd_set rfd;
532 struct timeval tv;
533 #endif
534
535
536 while (ctxt->state & XML_NANO_HTTP_READ) {
537 if (ctxt->in == NULL) {
538 ctxt->in = (char *) xmlMallocAtomic(65000 * sizeof(char));
539 if (ctxt->in == NULL) {
540 xmlHTTPErrMemory("allocating input");
541 ctxt->last = -1;
542 return (-1);
543 }
544 ctxt->inlen = 65000;
545 ctxt->inptr = ctxt->content = ctxt->inrptr = ctxt->in;
546 }
547 if (ctxt->inrptr > ctxt->in + XML_NANO_HTTP_CHUNK) {
548 int delta = ctxt->inrptr - ctxt->in;
549 int len = ctxt->inptr - ctxt->inrptr;
550
551 memmove(ctxt->in, ctxt->inrptr, len);
552 ctxt->inrptr -= delta;
553 ctxt->content -= delta;
554 ctxt->inptr -= delta;
555 }
556 if ((ctxt->in + ctxt->inlen) < (ctxt->inptr + XML_NANO_HTTP_CHUNK)) {
557 int d_inptr = ctxt->inptr - ctxt->in;
558 int d_content = ctxt->content - ctxt->in;
559 int d_inrptr = ctxt->inrptr - ctxt->in;
560 char *tmp_ptr = ctxt->in;
561
562 ctxt->inlen *= 2;
563 ctxt->in = (char *) xmlRealloc(tmp_ptr, ctxt->inlen);
564 if (ctxt->in == NULL) {
565 xmlHTTPErrMemory("allocating input buffer");
566 xmlFree(tmp_ptr);
567 ctxt->last = -1;
568 return (-1);
569 }
570 ctxt->inptr = ctxt->in + d_inptr;
571 ctxt->content = ctxt->in + d_content;
572 ctxt->inrptr = ctxt->in + d_inrptr;
573 }
574 ctxt->last = recv(ctxt->fd, ctxt->inptr, XML_NANO_HTTP_CHUNK, 0);
575 if (ctxt->last > 0) {
576 ctxt->inptr += ctxt->last;
577 return (ctxt->last);
578 }
579 if (ctxt->last == 0) {
580 return (0);
581 }
582 if (ctxt->last == -1) {
583 switch (socket_errno()) {
584 case EINPROGRESS:
585 case EWOULDBLOCK:
586 #if defined(EAGAIN) && EAGAIN != EWOULDBLOCK
587 case EAGAIN:
588 #endif
589 break;
590
591 case ECONNRESET:
592 case ESHUTDOWN:
593 return (0);
594
595 default:
596 __xmlIOErr(XML_FROM_HTTP, 0, "recv failed\n");
597 return (-1);
598 }
599 }
600 #ifdef HAVE_POLL_H
601 p.fd = ctxt->fd;
602 p.events = POLLIN;
603 if ((poll(&p, 1, timeout * 1000) < 1)
604 #if defined(EINTR)
605 && (errno != EINTR)
606 #endif
607 )
608 return (0);
609 #else /* !HAVE_POLL_H */
610 #ifndef _WINSOCKAPI_
611 if (ctxt->fd > FD_SETSIZE)
612 return 0;
613 #endif
614
615 tv.tv_sec = timeout;
616 tv.tv_usec = 0;
617 FD_ZERO(&rfd);
618
619 #ifdef _MSC_VER
620 #pragma warning(push)
621 #pragma warning(disable: 4018)
622 #endif
623
624 FD_SET(ctxt->fd, &rfd);
625
626 #ifdef _MSC_VER
627 #pragma warning(pop)
628 #endif
629
630 if ((select(ctxt->fd + 1, &rfd, NULL, NULL, &tv) < 1)
631 #if defined(EINTR)
632 && (errno != EINTR)
633 #endif
634 )
635 return (0);
636 #endif /* !HAVE_POLL_H */
637 }
638 return (0);
639 }
640
641 /**
642 * xmlNanoHTTPReadLine:
643 * @ctxt: an HTTP context
644 *
645 * Read one line in the HTTP server output, usually for extracting
646 * the HTTP protocol informations from the answer header.
647 *
648 * Returns a newly allocated string with a copy of the line, or NULL
649 * which indicate the end of the input.
650 */
651
652 static char *
xmlNanoHTTPReadLine(xmlNanoHTTPCtxtPtr ctxt)653 xmlNanoHTTPReadLine(xmlNanoHTTPCtxtPtr ctxt) {
654 char buf[4096];
655 char *bp = buf;
656 int rc;
657
658 while (bp - buf < 4095) {
659 if (ctxt->inrptr == ctxt->inptr) {
660 if ( (rc = xmlNanoHTTPRecv(ctxt)) == 0) {
661 if (bp == buf)
662 return(NULL);
663 else
664 *bp = 0;
665 return(xmlMemStrdup(buf));
666 }
667 else if ( rc == -1 ) {
668 return ( NULL );
669 }
670 }
671 *bp = *ctxt->inrptr++;
672 if (*bp == '\n') {
673 *bp = 0;
674 return(xmlMemStrdup(buf));
675 }
676 if (*bp != '\r')
677 bp++;
678 }
679 buf[4095] = 0;
680 return(xmlMemStrdup(buf));
681 }
682
683
684 /**
685 * xmlNanoHTTPScanAnswer:
686 * @ctxt: an HTTP context
687 * @line: an HTTP header line
688 *
689 * Try to extract useful informations from the server answer.
690 * We currently parse and process:
691 * - The HTTP revision/ return code
692 * - The Content-Type, Mime-Type and charset used
693 * - The Location for redirect processing.
694 *
695 * Returns -1 in case of failure, the file descriptor number otherwise
696 */
697
698 static void
xmlNanoHTTPScanAnswer(xmlNanoHTTPCtxtPtr ctxt,const char * line)699 xmlNanoHTTPScanAnswer(xmlNanoHTTPCtxtPtr ctxt, const char *line) {
700 const char *cur = line;
701
702 if (line == NULL) return;
703
704 if (!strncmp(line, "HTTP/", 5)) {
705 int version = 0;
706 int ret = 0;
707
708 cur += 5;
709 while ((*cur >= '0') && (*cur <= '9')) {
710 version *= 10;
711 version += *cur - '0';
712 cur++;
713 }
714 if (*cur == '.') {
715 cur++;
716 if ((*cur >= '0') && (*cur <= '9')) {
717 version *= 10;
718 version += *cur - '0';
719 cur++;
720 }
721 while ((*cur >= '0') && (*cur <= '9'))
722 cur++;
723 } else
724 version *= 10;
725 if ((*cur != ' ') && (*cur != '\t')) return;
726 while ((*cur == ' ') || (*cur == '\t')) cur++;
727 if ((*cur < '0') || (*cur > '9')) return;
728 while ((*cur >= '0') && (*cur <= '9')) {
729 ret *= 10;
730 ret += *cur - '0';
731 cur++;
732 }
733 if ((*cur != 0) && (*cur != ' ') && (*cur != '\t')) return;
734 ctxt->returnValue = ret;
735 ctxt->version = version;
736 } else if (!xmlStrncasecmp(BAD_CAST line, BAD_CAST"Content-Type:", 13)) {
737 const xmlChar *charset, *last, *mime;
738 cur += 13;
739 while ((*cur == ' ') || (*cur == '\t')) cur++;
740 if (ctxt->contentType != NULL)
741 xmlFree(ctxt->contentType);
742 ctxt->contentType = xmlMemStrdup(cur);
743 mime = (const xmlChar *) cur;
744 last = mime;
745 while ((*last != 0) && (*last != ' ') && (*last != '\t') &&
746 (*last != ';') && (*last != ','))
747 last++;
748 if (ctxt->mimeType != NULL)
749 xmlFree(ctxt->mimeType);
750 ctxt->mimeType = (char *) xmlStrndup(mime, last - mime);
751 charset = xmlStrstr(BAD_CAST ctxt->contentType, BAD_CAST "charset=");
752 if (charset != NULL) {
753 charset += 8;
754 last = charset;
755 while ((*last != 0) && (*last != ' ') && (*last != '\t') &&
756 (*last != ';') && (*last != ','))
757 last++;
758 if (ctxt->encoding != NULL)
759 xmlFree(ctxt->encoding);
760 ctxt->encoding = (char *) xmlStrndup(charset, last - charset);
761 }
762 } else if (!xmlStrncasecmp(BAD_CAST line, BAD_CAST"ContentType:", 12)) {
763 const xmlChar *charset, *last, *mime;
764 cur += 12;
765 if (ctxt->contentType != NULL) return;
766 while ((*cur == ' ') || (*cur == '\t')) cur++;
767 ctxt->contentType = xmlMemStrdup(cur);
768 mime = (const xmlChar *) cur;
769 last = mime;
770 while ((*last != 0) && (*last != ' ') && (*last != '\t') &&
771 (*last != ';') && (*last != ','))
772 last++;
773 if (ctxt->mimeType != NULL)
774 xmlFree(ctxt->mimeType);
775 ctxt->mimeType = (char *) xmlStrndup(mime, last - mime);
776 charset = xmlStrstr(BAD_CAST ctxt->contentType, BAD_CAST "charset=");
777 if (charset != NULL) {
778 charset += 8;
779 last = charset;
780 while ((*last != 0) && (*last != ' ') && (*last != '\t') &&
781 (*last != ';') && (*last != ','))
782 last++;
783 if (ctxt->encoding != NULL)
784 xmlFree(ctxt->encoding);
785 ctxt->encoding = (char *) xmlStrndup(charset, last - charset);
786 }
787 } else if (!xmlStrncasecmp(BAD_CAST line, BAD_CAST"Location:", 9)) {
788 cur += 9;
789 while ((*cur == ' ') || (*cur == '\t')) cur++;
790 if (ctxt->location != NULL)
791 xmlFree(ctxt->location);
792 if (*cur == '/') {
793 xmlChar *tmp_http = xmlStrdup(BAD_CAST "http://");
794 xmlChar *tmp_loc =
795 xmlStrcat(tmp_http, (const xmlChar *) ctxt->hostname);
796 ctxt->location =
797 (char *) xmlStrcat (tmp_loc, (const xmlChar *) cur);
798 } else {
799 ctxt->location = xmlMemStrdup(cur);
800 }
801 } else if (!xmlStrncasecmp(BAD_CAST line, BAD_CAST"WWW-Authenticate:", 17)) {
802 cur += 17;
803 while ((*cur == ' ') || (*cur == '\t')) cur++;
804 if (ctxt->authHeader != NULL)
805 xmlFree(ctxt->authHeader);
806 ctxt->authHeader = xmlMemStrdup(cur);
807 } else if (!xmlStrncasecmp(BAD_CAST line, BAD_CAST"Proxy-Authenticate:", 19)) {
808 cur += 19;
809 while ((*cur == ' ') || (*cur == '\t')) cur++;
810 if (ctxt->authHeader != NULL)
811 xmlFree(ctxt->authHeader);
812 ctxt->authHeader = xmlMemStrdup(cur);
813 #ifdef HAVE_ZLIB_H
814 } else if ( !xmlStrncasecmp( BAD_CAST line, BAD_CAST"Content-Encoding:", 17) ) {
815 cur += 17;
816 while ((*cur == ' ') || (*cur == '\t')) cur++;
817 if ( !xmlStrncasecmp( BAD_CAST cur, BAD_CAST"gzip", 4) ) {
818 ctxt->usesGzip = 1;
819
820 ctxt->strm = xmlMalloc(sizeof(z_stream));
821
822 if (ctxt->strm != NULL) {
823 ctxt->strm->zalloc = Z_NULL;
824 ctxt->strm->zfree = Z_NULL;
825 ctxt->strm->opaque = Z_NULL;
826 ctxt->strm->avail_in = 0;
827 ctxt->strm->next_in = Z_NULL;
828
829 inflateInit2( ctxt->strm, 31 );
830 }
831 }
832 #endif
833 } else if ( !xmlStrncasecmp( BAD_CAST line, BAD_CAST"Content-Length:", 15) ) {
834 cur += 15;
835 ctxt->ContentLength = strtol( cur, NULL, 10 );
836 }
837 }
838
839 /**
840 * xmlNanoHTTPConnectAttempt:
841 * @addr: a socket address structure
842 *
843 * Attempt a connection to the given IP:port endpoint. It forces
844 * non-blocking semantic on the socket, and allow 60 seconds for
845 * the host to answer.
846 *
847 * Returns -1 in case of failure, the file descriptor number otherwise
848 */
849
850 static SOCKET
xmlNanoHTTPConnectAttempt(struct sockaddr * addr)851 xmlNanoHTTPConnectAttempt(struct sockaddr *addr)
852 {
853 #ifndef HAVE_POLL_H
854 fd_set wfd;
855 #ifdef _WINSOCKAPI_
856 fd_set xfd;
857 #endif
858 struct timeval tv;
859 #else /* !HAVE_POLL_H */
860 struct pollfd p;
861 #endif /* !HAVE_POLL_H */
862 int status;
863
864 int addrlen;
865
866 SOCKET s;
867
868 #ifdef SUPPORT_IP6
869 if (addr->sa_family == AF_INET6) {
870 s = socket(PF_INET6, SOCK_STREAM, IPPROTO_TCP);
871 addrlen = sizeof(struct sockaddr_in6);
872 } else
873 #endif
874 {
875 s = socket(PF_INET, SOCK_STREAM, IPPROTO_TCP);
876 addrlen = sizeof(struct sockaddr_in);
877 }
878 if (s == INVALID_SOCKET) {
879 #ifdef DEBUG_HTTP
880 perror("socket");
881 #endif
882 __xmlIOErr(XML_FROM_HTTP, 0, "socket failed\n");
883 return INVALID_SOCKET;
884 }
885 #ifdef _WINSOCKAPI_
886 {
887 u_long one = 1;
888
889 status = ioctlsocket(s, FIONBIO, &one) == SOCKET_ERROR ? -1 : 0;
890 }
891 #else /* _WINSOCKAPI_ */
892 #if defined(VMS)
893 {
894 int enable = 1;
895
896 status = ioctl(s, FIONBIO, &enable);
897 }
898 #else /* VMS */
899 #if defined(__BEOS__) && !defined(__HAIKU__)
900 {
901 bool noblock = true;
902
903 status =
904 setsockopt(s, SOL_SOCKET, SO_NONBLOCK, &noblock,
905 sizeof(noblock));
906 }
907 #else /* __BEOS__ */
908 if ((status = fcntl(s, F_GETFL, 0)) != -1) {
909 #ifdef O_NONBLOCK
910 status |= O_NONBLOCK;
911 #else /* O_NONBLOCK */
912 #ifdef F_NDELAY
913 status |= F_NDELAY;
914 #endif /* F_NDELAY */
915 #endif /* !O_NONBLOCK */
916 status = fcntl(s, F_SETFL, status);
917 }
918 if (status < 0) {
919 #ifdef DEBUG_HTTP
920 perror("nonblocking");
921 #endif
922 __xmlIOErr(XML_FROM_HTTP, 0, "error setting non-blocking IO\n");
923 closesocket(s);
924 return INVALID_SOCKET;
925 }
926 #endif /* !__BEOS__ */
927 #endif /* !VMS */
928 #endif /* !_WINSOCKAPI_ */
929
930 if (connect(s, addr, addrlen) == -1) {
931 switch (socket_errno()) {
932 case EINPROGRESS:
933 case EWOULDBLOCK:
934 break;
935 default:
936 __xmlIOErr(XML_FROM_HTTP, 0,
937 "error connecting to HTTP server");
938 closesocket(s);
939 return INVALID_SOCKET;
940 }
941 }
942 #ifndef HAVE_POLL_H
943 tv.tv_sec = timeout;
944 tv.tv_usec = 0;
945
946 #ifdef _MSC_VER
947 #pragma warning(push)
948 #pragma warning(disable: 4018)
949 #endif
950 #ifndef _WINSOCKAPI_
951 if (s > FD_SETSIZE)
952 return INVALID_SOCKET;
953 #endif
954 FD_ZERO(&wfd);
955 FD_SET(s, &wfd);
956
957 #ifdef _WINSOCKAPI_
958 FD_ZERO(&xfd);
959 FD_SET(s, &xfd);
960
961 switch (select(s + 1, NULL, &wfd, &xfd, &tv))
962 #else
963 switch (select(s + 1, NULL, &wfd, NULL, &tv))
964 #endif
965 #ifdef _MSC_VER
966 #pragma warning(pop)
967 #endif
968
969 #else /* !HAVE_POLL_H */
970 p.fd = s;
971 p.events = POLLOUT;
972 switch (poll(&p, 1, timeout * 1000))
973 #endif /* !HAVE_POLL_H */
974
975 {
976 case 0:
977 /* Time out */
978 __xmlIOErr(XML_FROM_HTTP, 0, "Connect attempt timed out");
979 closesocket(s);
980 return INVALID_SOCKET;
981 case -1:
982 /* Ermm.. ?? */
983 __xmlIOErr(XML_FROM_HTTP, 0, "Connect failed");
984 closesocket(s);
985 return INVALID_SOCKET;
986 }
987
988 #ifndef HAVE_POLL_H
989 if (FD_ISSET(s, &wfd)
990 #ifdef _WINSOCKAPI_
991 || FD_ISSET(s, &xfd)
992 #endif
993 )
994 #else /* !HAVE_POLL_H */
995 if (p.revents == POLLOUT)
996 #endif /* !HAVE_POLL_H */
997 {
998 XML_SOCKLEN_T len;
999
1000 len = sizeof(status);
1001 #ifdef SO_ERROR
1002 if (getsockopt(s, SOL_SOCKET, SO_ERROR, (char *) &status, &len) <
1003 0) {
1004 /* Solaris error code */
1005 __xmlIOErr(XML_FROM_HTTP, 0, "getsockopt failed\n");
1006 closesocket(s);
1007 return INVALID_SOCKET;
1008 }
1009 #endif
1010 if (status) {
1011 __xmlIOErr(XML_FROM_HTTP, 0,
1012 "Error connecting to remote host");
1013 closesocket(s);
1014 errno = status;
1015 return INVALID_SOCKET;
1016 }
1017 } else {
1018 /* pbm */
1019 __xmlIOErr(XML_FROM_HTTP, 0, "select failed\n");
1020 closesocket(s);
1021 return INVALID_SOCKET;
1022 }
1023
1024 return (s);
1025 }
1026
1027 /**
1028 * xmlNanoHTTPConnectHost:
1029 * @host: the host name
1030 * @port: the port number
1031 *
1032 * Attempt a connection to the given host:port endpoint. It tries
1033 * the multiple IP provided by the DNS if available.
1034 *
1035 * Returns -1 in case of failure, the file descriptor number otherwise
1036 */
1037
1038 static SOCKET
xmlNanoHTTPConnectHost(const char * host,int port)1039 xmlNanoHTTPConnectHost(const char *host, int port)
1040 {
1041 struct hostent *h;
1042 struct sockaddr *addr = NULL;
1043 struct in_addr ia;
1044 struct sockaddr_in sockin;
1045
1046 #ifdef SUPPORT_IP6
1047 struct in6_addr ia6;
1048 struct sockaddr_in6 sockin6;
1049 #endif
1050 int i;
1051 SOCKET s;
1052
1053 memset (&sockin, 0, sizeof(sockin));
1054 #ifdef SUPPORT_IP6
1055 memset (&sockin6, 0, sizeof(sockin6));
1056 #endif
1057
1058 #if !defined(HAVE_GETADDRINFO) && defined(SUPPORT_IP6) && defined(RES_USE_INET6)
1059 if (have_ipv6 ())
1060 {
1061 if (!(_res.options & RES_INIT))
1062 res_init();
1063 _res.options |= RES_USE_INET6;
1064 }
1065 #endif
1066
1067 #if defined(HAVE_GETADDRINFO) && defined(SUPPORT_IP6) && !defined(_WIN32)
1068 if (have_ipv6 ())
1069 #endif
1070 #if defined(HAVE_GETADDRINFO) && (defined(SUPPORT_IP6) || defined(_WIN32))
1071 {
1072 int status;
1073 struct addrinfo hints, *res, *result;
1074
1075 result = NULL;
1076 memset (&hints, 0,sizeof(hints));
1077 hints.ai_socktype = SOCK_STREAM;
1078
1079 status = getaddrinfo (host, NULL, &hints, &result);
1080 if (status) {
1081 __xmlIOErr(XML_FROM_HTTP, 0, "getaddrinfo failed\n");
1082 return INVALID_SOCKET;
1083 }
1084
1085 for (res = result; res; res = res->ai_next) {
1086 if (res->ai_family == AF_INET) {
1087 if (res->ai_addrlen > sizeof(sockin)) {
1088 __xmlIOErr(XML_FROM_HTTP, 0, "address size mismatch\n");
1089 freeaddrinfo (result);
1090 return INVALID_SOCKET;
1091 }
1092 memcpy (&sockin, res->ai_addr, res->ai_addrlen);
1093 sockin.sin_port = htons (port);
1094 addr = (struct sockaddr *)&sockin;
1095 #ifdef SUPPORT_IP6
1096 } else if (have_ipv6 () && (res->ai_family == AF_INET6)) {
1097 if (res->ai_addrlen > sizeof(sockin6)) {
1098 __xmlIOErr(XML_FROM_HTTP, 0, "address size mismatch\n");
1099 freeaddrinfo (result);
1100 return INVALID_SOCKET;
1101 }
1102 memcpy (&sockin6, res->ai_addr, res->ai_addrlen);
1103 sockin6.sin6_port = htons (port);
1104 addr = (struct sockaddr *)&sockin6;
1105 #endif
1106 } else
1107 continue; /* for */
1108
1109 s = xmlNanoHTTPConnectAttempt (addr);
1110 if (s != INVALID_SOCKET) {
1111 freeaddrinfo (result);
1112 return (s);
1113 }
1114 }
1115
1116 if (result)
1117 freeaddrinfo (result);
1118 }
1119 #endif
1120 #if defined(HAVE_GETADDRINFO) && defined(SUPPORT_IP6) && !defined(_WIN32)
1121 else
1122 #endif
1123 #if !defined(HAVE_GETADDRINFO) || !defined(_WIN32)
1124 {
1125 h = gethostbyname (GETHOSTBYNAME_ARG_CAST host);
1126 if (h == NULL) {
1127
1128 /*
1129 * Okay, I got fed up by the non-portability of this error message
1130 * extraction code. it work on Linux, if it work on your platform
1131 * and one want to enable it, send me the defined(foobar) needed
1132 */
1133 #if defined(HAVE_NETDB_H) && defined(HOST_NOT_FOUND) && defined(linux)
1134 const char *h_err_txt = "";
1135
1136 switch (h_errno) {
1137 case HOST_NOT_FOUND:
1138 h_err_txt = "Authoritive host not found";
1139 break;
1140
1141 case TRY_AGAIN:
1142 h_err_txt =
1143 "Non-authoritive host not found or server failure.";
1144 break;
1145
1146 case NO_RECOVERY:
1147 h_err_txt =
1148 "Non-recoverable errors: FORMERR, REFUSED, or NOTIMP.";
1149 break;
1150
1151 #ifdef NO_ADDRESS
1152 case NO_ADDRESS:
1153 h_err_txt =
1154 "Valid name, no data record of requested type.";
1155 break;
1156 #endif
1157
1158 default:
1159 h_err_txt = "No error text defined.";
1160 break;
1161 }
1162 __xmlIOErr(XML_FROM_HTTP, 0, h_err_txt);
1163 #else
1164 __xmlIOErr(XML_FROM_HTTP, 0, "Failed to resolve host");
1165 #endif
1166 return INVALID_SOCKET;
1167 }
1168
1169 for (i = 0; h->h_addr_list[i]; i++) {
1170 if (h->h_addrtype == AF_INET) {
1171 /* A records (IPv4) */
1172 if ((unsigned int) h->h_length > sizeof(ia)) {
1173 __xmlIOErr(XML_FROM_HTTP, 0, "address size mismatch\n");
1174 return INVALID_SOCKET;
1175 }
1176 memcpy (&ia, h->h_addr_list[i], h->h_length);
1177 sockin.sin_family = h->h_addrtype;
1178 sockin.sin_addr = ia;
1179 sockin.sin_port = (unsigned short)htons ((unsigned short)port);
1180 addr = (struct sockaddr *) &sockin;
1181 #ifdef SUPPORT_IP6
1182 } else if (have_ipv6 () && (h->h_addrtype == AF_INET6)) {
1183 /* AAAA records (IPv6) */
1184 if ((unsigned int) h->h_length > sizeof(ia6)) {
1185 __xmlIOErr(XML_FROM_HTTP, 0, "address size mismatch\n");
1186 return INVALID_SOCKET;
1187 }
1188 memcpy (&ia6, h->h_addr_list[i], h->h_length);
1189 sockin6.sin6_family = h->h_addrtype;
1190 sockin6.sin6_addr = ia6;
1191 sockin6.sin6_port = htons (port);
1192 addr = (struct sockaddr *) &sockin6;
1193 #endif
1194 } else
1195 break; /* for */
1196
1197 s = xmlNanoHTTPConnectAttempt (addr);
1198 if (s != INVALID_SOCKET)
1199 return (s);
1200 }
1201 }
1202 #endif
1203
1204 #ifdef DEBUG_HTTP
1205 xmlGenericError(xmlGenericErrorContext,
1206 "xmlNanoHTTPConnectHost: unable to connect to '%s'.\n",
1207 host);
1208 #endif
1209 return INVALID_SOCKET;
1210 }
1211
1212
1213 /**
1214 * xmlNanoHTTPOpen:
1215 * @URL: The URL to load
1216 * @contentType: if available the Content-Type information will be
1217 * returned at that location
1218 *
1219 * This function try to open a connection to the indicated resource
1220 * via HTTP GET.
1221 *
1222 * Returns NULL in case of failure, otherwise a request handler.
1223 * The contentType, if provided must be freed by the caller
1224 */
1225
1226 void*
xmlNanoHTTPOpen(const char * URL,char ** contentType)1227 xmlNanoHTTPOpen(const char *URL, char **contentType) {
1228 if (contentType != NULL) *contentType = NULL;
1229 return(xmlNanoHTTPMethod(URL, NULL, NULL, contentType, NULL, 0));
1230 }
1231
1232 /**
1233 * xmlNanoHTTPOpenRedir:
1234 * @URL: The URL to load
1235 * @contentType: if available the Content-Type information will be
1236 * returned at that location
1237 * @redir: if available the redirected URL will be returned
1238 *
1239 * This function try to open a connection to the indicated resource
1240 * via HTTP GET.
1241 *
1242 * Returns NULL in case of failure, otherwise a request handler.
1243 * The contentType, if provided must be freed by the caller
1244 */
1245
1246 void*
xmlNanoHTTPOpenRedir(const char * URL,char ** contentType,char ** redir)1247 xmlNanoHTTPOpenRedir(const char *URL, char **contentType, char **redir) {
1248 if (contentType != NULL) *contentType = NULL;
1249 if (redir != NULL) *redir = NULL;
1250 return(xmlNanoHTTPMethodRedir(URL, NULL, NULL, contentType, redir, NULL,0));
1251 }
1252
1253 /**
1254 * xmlNanoHTTPRead:
1255 * @ctx: the HTTP context
1256 * @dest: a buffer
1257 * @len: the buffer length
1258 *
1259 * This function tries to read @len bytes from the existing HTTP connection
1260 * and saves them in @dest. This is a blocking call.
1261 *
1262 * Returns the number of byte read. 0 is an indication of an end of connection.
1263 * -1 indicates a parameter error.
1264 */
1265 int
xmlNanoHTTPRead(void * ctx,void * dest,int len)1266 xmlNanoHTTPRead(void *ctx, void *dest, int len) {
1267 xmlNanoHTTPCtxtPtr ctxt = (xmlNanoHTTPCtxtPtr) ctx;
1268 #ifdef HAVE_ZLIB_H
1269 int bytes_read = 0;
1270 int orig_avail_in;
1271 int z_ret;
1272 #endif
1273
1274 if (ctx == NULL) return(-1);
1275 if (dest == NULL) return(-1);
1276 if (len <= 0) return(0);
1277
1278 #ifdef HAVE_ZLIB_H
1279 if (ctxt->usesGzip == 1) {
1280 if (ctxt->strm == NULL) return(0);
1281
1282 ctxt->strm->next_out = dest;
1283 ctxt->strm->avail_out = len;
1284 ctxt->strm->avail_in = ctxt->inptr - ctxt->inrptr;
1285
1286 while (ctxt->strm->avail_out > 0 &&
1287 (ctxt->strm->avail_in > 0 || xmlNanoHTTPRecv(ctxt) > 0)) {
1288 orig_avail_in = ctxt->strm->avail_in =
1289 ctxt->inptr - ctxt->inrptr - bytes_read;
1290 ctxt->strm->next_in = BAD_CAST (ctxt->inrptr + bytes_read);
1291
1292 z_ret = inflate(ctxt->strm, Z_NO_FLUSH);
1293 bytes_read += orig_avail_in - ctxt->strm->avail_in;
1294
1295 if (z_ret != Z_OK) break;
1296 }
1297
1298 ctxt->inrptr += bytes_read;
1299 return(len - ctxt->strm->avail_out);
1300 }
1301 #endif
1302
1303 while (ctxt->inptr - ctxt->inrptr < len) {
1304 if (xmlNanoHTTPRecv(ctxt) <= 0) break;
1305 }
1306 if (ctxt->inptr - ctxt->inrptr < len)
1307 len = ctxt->inptr - ctxt->inrptr;
1308 memcpy(dest, ctxt->inrptr, len);
1309 ctxt->inrptr += len;
1310 return(len);
1311 }
1312
1313 /**
1314 * xmlNanoHTTPClose:
1315 * @ctx: the HTTP context
1316 *
1317 * This function closes an HTTP context, it ends up the connection and
1318 * free all data related to it.
1319 */
1320 void
xmlNanoHTTPClose(void * ctx)1321 xmlNanoHTTPClose(void *ctx) {
1322 xmlNanoHTTPCtxtPtr ctxt = (xmlNanoHTTPCtxtPtr) ctx;
1323
1324 if (ctx == NULL) return;
1325
1326 xmlNanoHTTPFreeCtxt(ctxt);
1327 }
1328
1329 /**
1330 * xmlNanoHTTPMethodRedir:
1331 * @URL: The URL to load
1332 * @method: the HTTP method to use
1333 * @input: the input string if any
1334 * @contentType: the Content-Type information IN and OUT
1335 * @redir: the redirected URL OUT
1336 * @headers: the extra headers
1337 * @ilen: input length
1338 *
1339 * This function try to open a connection to the indicated resource
1340 * via HTTP using the given @method, adding the given extra headers
1341 * and the input buffer for the request content.
1342 *
1343 * Returns NULL in case of failure, otherwise a request handler.
1344 * The contentType, or redir, if provided must be freed by the caller
1345 */
1346
1347 void*
xmlNanoHTTPMethodRedir(const char * URL,const char * method,const char * input,char ** contentType,char ** redir,const char * headers,int ilen)1348 xmlNanoHTTPMethodRedir(const char *URL, const char *method, const char *input,
1349 char **contentType, char **redir,
1350 const char *headers, int ilen ) {
1351 xmlNanoHTTPCtxtPtr ctxt;
1352 char *bp, *p;
1353 int blen;
1354 SOCKET ret;
1355 int nbRedirects = 0;
1356 char *redirURL = NULL;
1357 #ifdef DEBUG_HTTP
1358 int xmt_bytes;
1359 #endif
1360
1361 if (URL == NULL) return(NULL);
1362 if (method == NULL) method = "GET";
1363 xmlNanoHTTPInit();
1364
1365 retry:
1366 if (redirURL == NULL) {
1367 ctxt = xmlNanoHTTPNewCtxt(URL);
1368 if (ctxt == NULL)
1369 return(NULL);
1370 } else {
1371 ctxt = xmlNanoHTTPNewCtxt(redirURL);
1372 if (ctxt == NULL)
1373 return(NULL);
1374 ctxt->location = xmlMemStrdup(redirURL);
1375 }
1376
1377 if ((ctxt->protocol == NULL) || (strcmp(ctxt->protocol, "http"))) {
1378 __xmlIOErr(XML_FROM_HTTP, XML_HTTP_URL_SYNTAX, "Not a valid HTTP URI");
1379 xmlNanoHTTPFreeCtxt(ctxt);
1380 if (redirURL != NULL) xmlFree(redirURL);
1381 return(NULL);
1382 }
1383 if (ctxt->hostname == NULL) {
1384 __xmlIOErr(XML_FROM_HTTP, XML_HTTP_UNKNOWN_HOST,
1385 "Failed to identify host in URI");
1386 xmlNanoHTTPFreeCtxt(ctxt);
1387 if (redirURL != NULL) xmlFree(redirURL);
1388 return(NULL);
1389 }
1390 if (proxy) {
1391 blen = strlen(ctxt->hostname) * 2 + 16;
1392 ret = xmlNanoHTTPConnectHost(proxy, proxyPort);
1393 }
1394 else {
1395 blen = strlen(ctxt->hostname);
1396 ret = xmlNanoHTTPConnectHost(ctxt->hostname, ctxt->port);
1397 }
1398 if (ret == INVALID_SOCKET) {
1399 xmlNanoHTTPFreeCtxt(ctxt);
1400 if (redirURL != NULL) xmlFree(redirURL);
1401 return(NULL);
1402 }
1403 ctxt->fd = ret;
1404
1405 if (input == NULL)
1406 ilen = 0;
1407 else
1408 blen += 36;
1409
1410 if (headers != NULL)
1411 blen += strlen(headers) + 2;
1412 if (contentType && *contentType)
1413 /* reserve for string plus 'Content-Type: \r\n" */
1414 blen += strlen(*contentType) + 16;
1415 if (ctxt->query != NULL)
1416 /* 1 for '?' */
1417 blen += strlen(ctxt->query) + 1;
1418 blen += strlen(method) + strlen(ctxt->path) + 24;
1419 #ifdef HAVE_ZLIB_H
1420 /* reserve for possible 'Accept-Encoding: gzip' string */
1421 blen += 23;
1422 #endif
1423 if (ctxt->port != 80) {
1424 /* reserve space for ':xxxxx', incl. potential proxy */
1425 if (proxy)
1426 blen += 12;
1427 else
1428 blen += 6;
1429 }
1430 bp = (char*)xmlMallocAtomic(blen);
1431 if ( bp == NULL ) {
1432 xmlNanoHTTPFreeCtxt( ctxt );
1433 xmlHTTPErrMemory("allocating header buffer");
1434 return ( NULL );
1435 }
1436
1437 p = bp;
1438
1439 if (proxy) {
1440 if (ctxt->port != 80) {
1441 p += snprintf( p, blen - (p - bp), "%s http://%s:%d%s",
1442 method, ctxt->hostname,
1443 ctxt->port, ctxt->path );
1444 }
1445 else
1446 p += snprintf( p, blen - (p - bp), "%s http://%s%s", method,
1447 ctxt->hostname, ctxt->path);
1448 }
1449 else
1450 p += snprintf( p, blen - (p - bp), "%s %s", method, ctxt->path);
1451
1452 if (ctxt->query != NULL)
1453 p += snprintf( p, blen - (p - bp), "?%s", ctxt->query);
1454
1455 if (ctxt->port == 80) {
1456 p += snprintf( p, blen - (p - bp), " HTTP/1.0\r\nHost: %s\r\n",
1457 ctxt->hostname);
1458 } else {
1459 p += snprintf( p, blen - (p - bp), " HTTP/1.0\r\nHost: %s:%d\r\n",
1460 ctxt->hostname, ctxt->port);
1461 }
1462
1463 #ifdef HAVE_ZLIB_H
1464 p += snprintf(p, blen - (p - bp), "Accept-Encoding: gzip\r\n");
1465 #endif
1466
1467 if (contentType != NULL && *contentType)
1468 p += snprintf(p, blen - (p - bp), "Content-Type: %s\r\n", *contentType);
1469
1470 if (headers != NULL)
1471 p += snprintf( p, blen - (p - bp), "%s", headers );
1472
1473 if (input != NULL)
1474 snprintf(p, blen - (p - bp), "Content-Length: %d\r\n\r\n", ilen );
1475 else
1476 snprintf(p, blen - (p - bp), "\r\n");
1477
1478 #ifdef DEBUG_HTTP
1479 xmlGenericError(xmlGenericErrorContext,
1480 "-> %s%s", proxy? "(Proxy) " : "", bp);
1481 if ((blen -= strlen(bp)+1) < 0)
1482 xmlGenericError(xmlGenericErrorContext,
1483 "ERROR: overflowed buffer by %d bytes\n", -blen);
1484 #endif
1485 ctxt->outptr = ctxt->out = bp;
1486 ctxt->state = XML_NANO_HTTP_WRITE;
1487 blen = strlen( ctxt->out );
1488 #ifdef DEBUG_HTTP
1489 xmt_bytes = xmlNanoHTTPSend(ctxt, ctxt->out, blen );
1490 if ( xmt_bytes != blen )
1491 xmlGenericError( xmlGenericErrorContext,
1492 "xmlNanoHTTPMethodRedir: Only %d of %d %s %s\n",
1493 xmt_bytes, blen,
1494 "bytes of HTTP headers sent to host",
1495 ctxt->hostname );
1496 #else
1497 xmlNanoHTTPSend(ctxt, ctxt->out, blen );
1498 #endif
1499
1500 if ( input != NULL ) {
1501 #ifdef DEBUG_HTTP
1502 xmt_bytes = xmlNanoHTTPSend( ctxt, input, ilen );
1503
1504 if ( xmt_bytes != ilen )
1505 xmlGenericError( xmlGenericErrorContext,
1506 "xmlNanoHTTPMethodRedir: Only %d of %d %s %s\n",
1507 xmt_bytes, ilen,
1508 "bytes of HTTP content sent to host",
1509 ctxt->hostname );
1510 #else
1511 xmlNanoHTTPSend( ctxt, input, ilen );
1512 #endif
1513 }
1514
1515 ctxt->state = XML_NANO_HTTP_READ;
1516
1517 while ((p = xmlNanoHTTPReadLine(ctxt)) != NULL) {
1518 if (*p == 0) {
1519 ctxt->content = ctxt->inrptr;
1520 xmlFree(p);
1521 break;
1522 }
1523 xmlNanoHTTPScanAnswer(ctxt, p);
1524
1525 #ifdef DEBUG_HTTP
1526 xmlGenericError(xmlGenericErrorContext, "<- %s\n", p);
1527 #endif
1528 xmlFree(p);
1529 }
1530
1531 if ((ctxt->location != NULL) && (ctxt->returnValue >= 300) &&
1532 (ctxt->returnValue < 400)) {
1533 #ifdef DEBUG_HTTP
1534 xmlGenericError(xmlGenericErrorContext,
1535 "\nRedirect to: %s\n", ctxt->location);
1536 #endif
1537 while ( xmlNanoHTTPRecv(ctxt) > 0 ) ;
1538 if (nbRedirects < XML_NANO_HTTP_MAX_REDIR) {
1539 nbRedirects++;
1540 if (redirURL != NULL)
1541 xmlFree(redirURL);
1542 redirURL = xmlMemStrdup(ctxt->location);
1543 xmlNanoHTTPFreeCtxt(ctxt);
1544 goto retry;
1545 }
1546 xmlNanoHTTPFreeCtxt(ctxt);
1547 if (redirURL != NULL) xmlFree(redirURL);
1548 #ifdef DEBUG_HTTP
1549 xmlGenericError(xmlGenericErrorContext,
1550 "xmlNanoHTTPMethodRedir: Too many redirects, aborting ...\n");
1551 #endif
1552 return(NULL);
1553 }
1554
1555 if (contentType != NULL) {
1556 if (ctxt->contentType != NULL)
1557 *contentType = xmlMemStrdup(ctxt->contentType);
1558 else
1559 *contentType = NULL;
1560 }
1561
1562 if ((redir != NULL) && (redirURL != NULL)) {
1563 *redir = redirURL;
1564 } else {
1565 if (redirURL != NULL)
1566 xmlFree(redirURL);
1567 if (redir != NULL)
1568 *redir = NULL;
1569 }
1570
1571 #ifdef DEBUG_HTTP
1572 if (ctxt->contentType != NULL)
1573 xmlGenericError(xmlGenericErrorContext,
1574 "\nCode %d, content-type '%s'\n\n",
1575 ctxt->returnValue, ctxt->contentType);
1576 else
1577 xmlGenericError(xmlGenericErrorContext,
1578 "\nCode %d, no content-type\n\n",
1579 ctxt->returnValue);
1580 #endif
1581
1582 return((void *) ctxt);
1583 }
1584
1585 /**
1586 * xmlNanoHTTPMethod:
1587 * @URL: The URL to load
1588 * @method: the HTTP method to use
1589 * @input: the input string if any
1590 * @contentType: the Content-Type information IN and OUT
1591 * @headers: the extra headers
1592 * @ilen: input length
1593 *
1594 * This function try to open a connection to the indicated resource
1595 * via HTTP using the given @method, adding the given extra headers
1596 * and the input buffer for the request content.
1597 *
1598 * Returns NULL in case of failure, otherwise a request handler.
1599 * The contentType, if provided must be freed by the caller
1600 */
1601
1602 void*
xmlNanoHTTPMethod(const char * URL,const char * method,const char * input,char ** contentType,const char * headers,int ilen)1603 xmlNanoHTTPMethod(const char *URL, const char *method, const char *input,
1604 char **contentType, const char *headers, int ilen) {
1605 return(xmlNanoHTTPMethodRedir(URL, method, input, contentType,
1606 NULL, headers, ilen));
1607 }
1608
1609 /**
1610 * xmlNanoHTTPFetch:
1611 * @URL: The URL to load
1612 * @filename: the filename where the content should be saved
1613 * @contentType: if available the Content-Type information will be
1614 * returned at that location
1615 *
1616 * This function try to fetch the indicated resource via HTTP GET
1617 * and save it's content in the file.
1618 *
1619 * Returns -1 in case of failure, 0 incase of success. The contentType,
1620 * if provided must be freed by the caller
1621 */
1622 int
xmlNanoHTTPFetch(const char * URL,const char * filename,char ** contentType)1623 xmlNanoHTTPFetch(const char *URL, const char *filename, char **contentType) {
1624 void *ctxt = NULL;
1625 char *buf = NULL;
1626 int fd;
1627 int len;
1628 int ret = 0;
1629
1630 if (filename == NULL) return(-1);
1631 ctxt = xmlNanoHTTPOpen(URL, contentType);
1632 if (ctxt == NULL) return(-1);
1633
1634 if (!strcmp(filename, "-"))
1635 fd = 0;
1636 else {
1637 fd = open(filename, O_CREAT | O_WRONLY, 00644);
1638 if (fd < 0) {
1639 xmlNanoHTTPClose(ctxt);
1640 if ((contentType != NULL) && (*contentType != NULL)) {
1641 xmlFree(*contentType);
1642 *contentType = NULL;
1643 }
1644 return(-1);
1645 }
1646 }
1647
1648 xmlNanoHTTPFetchContent( ctxt, &buf, &len );
1649 if ( len > 0 ) {
1650 if (write(fd, buf, len) == -1) {
1651 ret = -1;
1652 }
1653 }
1654
1655 xmlNanoHTTPClose(ctxt);
1656 close(fd);
1657 return(ret);
1658 }
1659
1660 #ifdef LIBXML_OUTPUT_ENABLED
1661 /**
1662 * xmlNanoHTTPSave:
1663 * @ctxt: the HTTP context
1664 * @filename: the filename where the content should be saved
1665 *
1666 * This function saves the output of the HTTP transaction to a file
1667 * It closes and free the context at the end
1668 *
1669 * Returns -1 in case of failure, 0 incase of success.
1670 */
1671 int
xmlNanoHTTPSave(void * ctxt,const char * filename)1672 xmlNanoHTTPSave(void *ctxt, const char *filename) {
1673 char *buf = NULL;
1674 int fd;
1675 int len;
1676 int ret = 0;
1677
1678 if ((ctxt == NULL) || (filename == NULL)) return(-1);
1679
1680 if (!strcmp(filename, "-"))
1681 fd = 0;
1682 else {
1683 fd = open(filename, O_CREAT | O_WRONLY, 0666);
1684 if (fd < 0) {
1685 xmlNanoHTTPClose(ctxt);
1686 return(-1);
1687 }
1688 }
1689
1690 xmlNanoHTTPFetchContent( ctxt, &buf, &len );
1691 if ( len > 0 ) {
1692 if (write(fd, buf, len) == -1) {
1693 ret = -1;
1694 }
1695 }
1696
1697 xmlNanoHTTPClose(ctxt);
1698 close(fd);
1699 return(ret);
1700 }
1701 #endif /* LIBXML_OUTPUT_ENABLED */
1702
1703 /**
1704 * xmlNanoHTTPReturnCode:
1705 * @ctx: the HTTP context
1706 *
1707 * Get the latest HTTP return code received
1708 *
1709 * Returns the HTTP return code for the request.
1710 */
1711 int
xmlNanoHTTPReturnCode(void * ctx)1712 xmlNanoHTTPReturnCode(void *ctx) {
1713 xmlNanoHTTPCtxtPtr ctxt = (xmlNanoHTTPCtxtPtr) ctx;
1714
1715 if (ctxt == NULL) return(-1);
1716
1717 return(ctxt->returnValue);
1718 }
1719
1720 /**
1721 * xmlNanoHTTPAuthHeader:
1722 * @ctx: the HTTP context
1723 *
1724 * Get the authentication header of an HTTP context
1725 *
1726 * Returns the stashed value of the WWW-Authenticate or Proxy-Authenticate
1727 * header.
1728 */
1729 const char *
xmlNanoHTTPAuthHeader(void * ctx)1730 xmlNanoHTTPAuthHeader(void *ctx) {
1731 xmlNanoHTTPCtxtPtr ctxt = (xmlNanoHTTPCtxtPtr) ctx;
1732
1733 if (ctxt == NULL) return(NULL);
1734
1735 return(ctxt->authHeader);
1736 }
1737
1738 /**
1739 * xmlNanoHTTPContentLength:
1740 * @ctx: the HTTP context
1741 *
1742 * Provides the specified content length from the HTTP header.
1743 *
1744 * Return the specified content length from the HTTP header. Note that
1745 * a value of -1 indicates that the content length element was not included in
1746 * the response header.
1747 */
1748 int
xmlNanoHTTPContentLength(void * ctx)1749 xmlNanoHTTPContentLength( void * ctx ) {
1750 xmlNanoHTTPCtxtPtr ctxt = (xmlNanoHTTPCtxtPtr)ctx;
1751
1752 return ( ( ctxt == NULL ) ? -1 : ctxt->ContentLength );
1753 }
1754
1755 /**
1756 * xmlNanoHTTPRedir:
1757 * @ctx: the HTTP context
1758 *
1759 * Provides the specified redirection URL if available from the HTTP header.
1760 *
1761 * Return the specified redirection URL or NULL if not redirected.
1762 */
1763 const char *
xmlNanoHTTPRedir(void * ctx)1764 xmlNanoHTTPRedir( void * ctx ) {
1765 xmlNanoHTTPCtxtPtr ctxt = (xmlNanoHTTPCtxtPtr)ctx;
1766
1767 return ( ( ctxt == NULL ) ? NULL : ctxt->location );
1768 }
1769
1770 /**
1771 * xmlNanoHTTPEncoding:
1772 * @ctx: the HTTP context
1773 *
1774 * Provides the specified encoding if specified in the HTTP headers.
1775 *
1776 * Return the specified encoding or NULL if not available
1777 */
1778 const char *
xmlNanoHTTPEncoding(void * ctx)1779 xmlNanoHTTPEncoding( void * ctx ) {
1780 xmlNanoHTTPCtxtPtr ctxt = (xmlNanoHTTPCtxtPtr)ctx;
1781
1782 return ( ( ctxt == NULL ) ? NULL : ctxt->encoding );
1783 }
1784
1785 /**
1786 * xmlNanoHTTPMimeType:
1787 * @ctx: the HTTP context
1788 *
1789 * Provides the specified Mime-Type if specified in the HTTP headers.
1790 *
1791 * Return the specified Mime-Type or NULL if not available
1792 */
1793 const char *
xmlNanoHTTPMimeType(void * ctx)1794 xmlNanoHTTPMimeType( void * ctx ) {
1795 xmlNanoHTTPCtxtPtr ctxt = (xmlNanoHTTPCtxtPtr)ctx;
1796
1797 return ( ( ctxt == NULL ) ? NULL : ctxt->mimeType );
1798 }
1799
1800 /**
1801 * xmlNanoHTTPFetchContent:
1802 * @ctx: the HTTP context
1803 * @ptr: pointer to set to the content buffer.
1804 * @len: integer pointer to hold the length of the content
1805 *
1806 * Check if all the content was read
1807 *
1808 * Returns 0 if all the content was read and available, returns
1809 * -1 if received content length was less than specified or an error
1810 * occurred.
1811 */
1812 static int
xmlNanoHTTPFetchContent(void * ctx,char ** ptr,int * len)1813 xmlNanoHTTPFetchContent( void * ctx, char ** ptr, int * len ) {
1814 xmlNanoHTTPCtxtPtr ctxt = (xmlNanoHTTPCtxtPtr)ctx;
1815
1816 int rc = 0;
1817 int cur_lgth;
1818 int rcvd_lgth;
1819 int dummy_int;
1820 char * dummy_ptr = NULL;
1821
1822 /* Dummy up return input parameters if not provided */
1823
1824 if ( len == NULL )
1825 len = &dummy_int;
1826
1827 if ( ptr == NULL )
1828 ptr = &dummy_ptr;
1829
1830 /* But can't work without the context pointer */
1831
1832 if ( ( ctxt == NULL ) || ( ctxt->content == NULL ) ) {
1833 *len = 0;
1834 *ptr = NULL;
1835 return ( -1 );
1836 }
1837
1838 rcvd_lgth = ctxt->inptr - ctxt->content;
1839
1840 while ( (cur_lgth = xmlNanoHTTPRecv( ctxt )) > 0 ) {
1841
1842 rcvd_lgth += cur_lgth;
1843 if ( (ctxt->ContentLength > 0) && (rcvd_lgth >= ctxt->ContentLength) )
1844 break;
1845 }
1846
1847 *ptr = ctxt->content;
1848 *len = rcvd_lgth;
1849
1850 if ( ( ctxt->ContentLength > 0 ) && ( rcvd_lgth < ctxt->ContentLength ) )
1851 rc = -1;
1852 else if ( rcvd_lgth == 0 )
1853 rc = -1;
1854
1855 return ( rc );
1856 }
1857
1858 #ifdef STANDALONE
main(int argc,char ** argv)1859 int main(int argc, char **argv) {
1860 char *contentType = NULL;
1861
1862 if (argv[1] != NULL) {
1863 if (argv[2] != NULL)
1864 xmlNanoHTTPFetch(argv[1], argv[2], &contentType);
1865 else
1866 xmlNanoHTTPFetch(argv[1], "-", &contentType);
1867 if (contentType != NULL) xmlFree(contentType);
1868 } else {
1869 xmlGenericError(xmlGenericErrorContext,
1870 "%s: minimal HTTP GET implementation\n", argv[0]);
1871 xmlGenericError(xmlGenericErrorContext,
1872 "\tusage %s [ URL [ filename ] ]\n", argv[0]);
1873 }
1874 xmlNanoHTTPCleanup();
1875 xmlMemoryDump();
1876 return(0);
1877 }
1878 #endif /* STANDALONE */
1879 #else /* !LIBXML_HTTP_ENABLED */
1880 #ifdef STANDALONE
1881 #include <stdio.h>
main(int argc,char ** argv)1882 int main(int argc, char **argv) {
1883 xmlGenericError(xmlGenericErrorContext,
1884 "%s : HTTP support not compiled in\n", argv[0]);
1885 return(0);
1886 }
1887 #endif /* STANDALONE */
1888 #endif /* LIBXML_HTTP_ENABLED */
1889 #define bottom_nanohttp
1890 #include "elfgcchack.h"
1891