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