blob: 2406e7e2e43e3aa9b6df06de2d498e647073f6f2 [file] [log] [blame]
Owen Taylor3473f882001-02-23 17:55:21 +00001/*
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 *
Daniel Veillardc5d64342001-06-24 12:13:24 +000011 * daniel@veillard.com
Owen Taylor3473f882001-02-23 17:55:21 +000012 */
13
Daniel Veillardf3afa7d2001-06-09 13:52:58 +000014#define NEED_SOCKETS
Daniel Veillard34ce8be2002-03-18 19:37:11 +000015#define IN_LIBXML
Bjorn Reese70a9da52001-04-21 16:57:29 +000016#include "libxml.h"
Owen Taylor3473f882001-02-23 17:55:21 +000017
18#ifdef LIBXML_HTTP_ENABLED
Owen Taylor3473f882001-02-23 17:55:21 +000019#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
Daniel Veillard75eb1ad2003-07-07 14:42:44 +000027#ifdef HAVE_SYS_TYPES_H
28#include <sys/types.h>
29#endif
Owen Taylor3473f882001-02-23 17:55:21 +000030#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
Daniel Veillardd85f4f42002-03-25 10:48:46 +000042#ifdef HAVE_RESOLV_H
Daniel Veillard9b731d72002-04-14 12:56:08 +000043#ifdef HAVE_ARPA_NAMESER_H
44#include <arpa/nameser.h>
45#endif
Daniel Veillardd85f4f42002-03-25 10:48:46 +000046#include <resolv.h>
47#endif
Owen Taylor3473f882001-02-23 17:55:21 +000048#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#ifdef HAVE_SYS_SELECT_H
58#include <sys/select.h>
59#endif
60#ifdef HAVE_STRINGS_H
61#include <strings.h>
62#endif
63#ifdef SUPPORT_IP6
64#include <resolv.h>
65#endif
Daniel Veillard9a2724d2005-12-15 11:12:26 +000066#ifdef HAVE_ZLIB_H
67#include <zlib.h>
68#endif
69
Owen Taylor3473f882001-02-23 17:55:21 +000070
71#ifdef VMS
72#include <stropts>
Daniel Veillardc284c642005-03-31 10:24:24 +000073#define XML_SOCKLEN_T unsigned int
Owen Taylor3473f882001-02-23 17:55:21 +000074#define SOCKET int
75#endif
76
Daniel Veillard59d3ed82007-04-17 12:44:58 +000077#if defined(__MINGW32__) || defined(_WIN32_WCE)
Daniel Veillard1638a472003-08-14 01:23:25 +000078#define _WINSOCKAPI_
79#include <wsockcompat.h>
80#include <winsock2.h>
Daniel Veillardc284c642005-03-31 10:24:24 +000081#undef XML_SOCKLEN_T
82#define XML_SOCKLEN_T unsigned int
Daniel Veillard1638a472003-08-14 01:23:25 +000083#endif
84
85
Daniel Veillardd0463562001-10-13 09:15:48 +000086#include <libxml/globals.h>
Daniel Veillardf012a642001-07-23 19:10:52 +000087#include <libxml/xmlerror.h>
Owen Taylor3473f882001-02-23 17:55:21 +000088#include <libxml/xmlmemory.h>
89#include <libxml/parser.h> /* for xmlStr(n)casecmp() */
90#include <libxml/nanohttp.h>
Daniel Veillard3c01b1d2001-10-17 15:58:35 +000091#include <libxml/globals.h>
Daniel Veillard8efff672002-12-04 11:44:48 +000092#include <libxml/uri.h>
Owen Taylor3473f882001-02-23 17:55:21 +000093
94/**
95 * A couple portability macros
96 */
97#ifndef _WINSOCKAPI_
Daniel Veillarda9cce9c2003-09-29 13:20:24 +000098#ifndef __BEOS__
Owen Taylor3473f882001-02-23 17:55:21 +000099#define closesocket(s) close(s)
Daniel Veillarda9cce9c2003-09-29 13:20:24 +0000100#endif
Owen Taylor3473f882001-02-23 17:55:21 +0000101#define SOCKET int
102#endif
103
Daniel Veillard89f7f272003-09-29 13:29:09 +0000104#ifdef __BEOS__
105#ifndef PF_INET
106#define PF_INET AF_INET
107#endif
108#endif
109
Daniel Veillardc284c642005-03-31 10:24:24 +0000110#ifndef XML_SOCKLEN_T
111#define XML_SOCKLEN_T unsigned int
Daniel Veillard75be0132002-03-13 10:03:35 +0000112#endif
113#ifndef SOCKET
114#define SOCKET int
115#endif
Daniel Veillardf012a642001-07-23 19:10:52 +0000116
Owen Taylor3473f882001-02-23 17:55:21 +0000117#ifdef STANDALONE
118#define DEBUG_HTTP
119#define xmlStrncasecmp(a, b, n) strncasecmp((char *)a, (char *)b, n)
120#define xmlStrcasecmpi(a, b) strcasecmp((char *)a, (char *)b)
121#endif
122
123#define XML_NANO_HTTP_MAX_REDIR 10
124
125#define XML_NANO_HTTP_CHUNK 4096
126
127#define XML_NANO_HTTP_CLOSED 0
128#define XML_NANO_HTTP_WRITE 1
129#define XML_NANO_HTTP_READ 2
130#define XML_NANO_HTTP_NONE 4
131
132typedef struct xmlNanoHTTPCtxt {
133 char *protocol; /* the protocol name */
134 char *hostname; /* the host name */
135 int port; /* the port */
136 char *path; /* the path within the URL */
Daniel Veillard351f2d62005-04-13 02:55:12 +0000137 char *query; /* the query string */
Owen Taylor3473f882001-02-23 17:55:21 +0000138 SOCKET fd; /* the file descriptor for the socket */
139 int state; /* WRITE / READ / CLOSED */
140 char *out; /* buffer sent (zero terminated) */
141 char *outptr; /* index within the buffer sent */
142 char *in; /* the receiving buffer */
143 char *content; /* the start of the content */
144 char *inptr; /* the next byte to read from network */
145 char *inrptr; /* the next byte to give back to the client */
146 int inlen; /* len of the input buffer */
147 int last; /* return code for last operation */
148 int returnValue; /* the protocol return value */
Daniel Veillardf012a642001-07-23 19:10:52 +0000149 int ContentLength; /* specified content length from HTTP header */
Owen Taylor3473f882001-02-23 17:55:21 +0000150 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 */
Daniel Veillard847332a2003-10-18 11:29:40 +0000153 char *encoding; /* encoding extracted from the contentType */
Daniel Veillarda840b692003-10-19 13:35:37 +0000154 char *mimeType; /* Mime-Type extracted from the contentType */
Daniel Veillard9a2724d2005-12-15 11:12:26 +0000155#ifdef HAVE_ZLIB_H
156 z_stream *strm; /* Zlib stream object */
157 int usesGzip; /* "Content-Encoding: gzip" was detected */
158#endif
Owen Taylor3473f882001-02-23 17:55:21 +0000159} xmlNanoHTTPCtxt, *xmlNanoHTTPCtxtPtr;
160
161static int initialized = 0;
162static char *proxy = NULL; /* the proxy name if any */
163static int proxyPort; /* the proxy port if any */
164static unsigned int timeout = 60;/* the select() timeout in seconds */
165
Daniel Veillarda2351322004-06-27 12:08:10 +0000166static int xmlNanoHTTPFetchContent( void * ctx, char ** ptr, int * len );
Daniel Veillardf012a642001-07-23 19:10:52 +0000167
Owen Taylor3473f882001-02-23 17:55:21 +0000168/**
Daniel Veillard2b0f8792003-10-10 19:36:36 +0000169 * xmlHTTPErrMemory:
170 * @extra: extra informations
171 *
172 * Handle an out of memory condition
173 */
174static void
175xmlHTTPErrMemory(const char *extra)
176{
177 __xmlSimpleError(XML_FROM_HTTP, XML_ERR_NO_MEMORY, NULL, NULL, extra);
178}
179
180/**
Owen Taylor3473f882001-02-23 17:55:21 +0000181 * A portability function
182 */
Daniel Veillard56a4cb82001-03-24 17:00:36 +0000183static int socket_errno(void) {
Owen Taylor3473f882001-02-23 17:55:21 +0000184#ifdef _WINSOCKAPI_
185 return(WSAGetLastError());
186#else
187 return(errno);
188#endif
189}
190
Daniel Veillardde2a67b2003-06-21 14:20:04 +0000191#ifdef SUPPORT_IP6
Daniel Veillard2db8c122003-07-08 12:16:59 +0000192static
193int have_ipv6(void) {
Daniel Veillardde2a67b2003-06-21 14:20:04 +0000194 int s;
195
196 s = socket (AF_INET6, SOCK_STREAM, 0);
197 if (s != -1) {
198 close (s);
199 return (1);
200 }
201 return (0);
202}
203#endif
204
Owen Taylor3473f882001-02-23 17:55:21 +0000205/**
206 * xmlNanoHTTPInit:
207 *
208 * Initialize the HTTP protocol layer.
209 * Currently it just checks for proxy informations
210 */
211
212void
213xmlNanoHTTPInit(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");
Daniel Veillard29b17482004-08-16 00:39:03 +0000230 if (env && ((env[0] == '*') && (env[1] == 0)))
Owen Taylor3473f882001-02-23 17:55:21 +0000231 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 }
243done:
244 initialized = 1;
245}
246
247/**
Daniel Veillard5e2dace2001-07-18 19:30:27 +0000248 * xmlNanoHTTPCleanup:
Owen Taylor3473f882001-02-23 17:55:21 +0000249 *
250 * Cleanup the HTTP protocol layer.
251 */
252
253void
254xmlNanoHTTPCleanup(void) {
Daniel Veillard744acff2005-07-12 15:09:53 +0000255 if (proxy != NULL) {
Owen Taylor3473f882001-02-23 17:55:21 +0000256 xmlFree(proxy);
Daniel Veillard744acff2005-07-12 15:09:53 +0000257 proxy = NULL;
258 }
Owen Taylor3473f882001-02-23 17:55:21 +0000259#ifdef _WINSOCKAPI_
260 if (initialized)
261 WSACleanup();
262#endif
263 initialized = 0;
264 return;
265}
266
267/**
Owen Taylor3473f882001-02-23 17:55:21 +0000268 * 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
276static void
277xmlNanoHTTPScanURL(xmlNanoHTTPCtxtPtr ctxt, const char *URL) {
William M. Brack015ccb22005-02-13 08:18:52 +0000278 xmlURIPtr uri;
279 /*
280 * Clear any existing data from the context
281 */
Owen Taylor3473f882001-02-23 17:55:21 +0000282 if (ctxt->protocol != NULL) {
283 xmlFree(ctxt->protocol);
284 ctxt->protocol = NULL;
285 }
286 if (ctxt->hostname != NULL) {
287 xmlFree(ctxt->hostname);
288 ctxt->hostname = NULL;
289 }
290 if (ctxt->path != NULL) {
291 xmlFree(ctxt->path);
292 ctxt->path = NULL;
293 }
Daniel Veillard351f2d62005-04-13 02:55:12 +0000294 if (ctxt->query != NULL) {
295 xmlFree(ctxt->query);
296 ctxt->query = NULL;
297 }
Owen Taylor3473f882001-02-23 17:55:21 +0000298 if (URL == NULL) return;
William M. Brack015ccb22005-02-13 08:18:52 +0000299
Daniel Veillard336a8e12005-08-07 10:46:19 +0000300 uri = xmlParseURIRaw(URL, 1);
William M. Brack015ccb22005-02-13 08:18:52 +0000301 if (uri == NULL)
302 return;
303
304 if ((uri->scheme == NULL) || (uri->server == NULL)) {
305 xmlFreeURI(uri);
306 return;
Owen Taylor3473f882001-02-23 17:55:21 +0000307 }
Daniel Veillardde2a67b2003-06-21 14:20:04 +0000308
William M. Brack015ccb22005-02-13 08:18:52 +0000309 ctxt->protocol = xmlMemStrdup(uri->scheme);
310 ctxt->hostname = xmlMemStrdup(uri->server);
311 if (uri->path != NULL)
312 ctxt->path = xmlMemStrdup(uri->path);
313 else
314 ctxt->path = xmlMemStrdup("/");
Daniel Veillard351f2d62005-04-13 02:55:12 +0000315 if (uri->query != NULL)
316 ctxt->query = xmlMemStrdup(uri->query);
William M. Brack015ccb22005-02-13 08:18:52 +0000317 if (uri->port != 0)
318 ctxt->port = uri->port;
Daniel Veillardde2a67b2003-06-21 14:20:04 +0000319
William M. Brack015ccb22005-02-13 08:18:52 +0000320 xmlFreeURI(uri);
Owen Taylor3473f882001-02-23 17:55:21 +0000321}
322
323/**
324 * xmlNanoHTTPScanProxy:
325 * @URL: The proxy URL used to initialize the proxy context
326 *
327 * (Re)Initialize the HTTP Proxy context by parsing the URL and finding
328 * the protocol host port it indicates.
329 * Should be like http://myproxy/ or http://myproxy:3128/
330 * A NULL URL cleans up proxy informations.
331 */
332
333void
334xmlNanoHTTPScanProxy(const char *URL) {
William M. Brack015ccb22005-02-13 08:18:52 +0000335 xmlURIPtr uri;
Owen Taylor3473f882001-02-23 17:55:21 +0000336
337 if (proxy != NULL) {
338 xmlFree(proxy);
339 proxy = NULL;
340 }
William M. Brack015ccb22005-02-13 08:18:52 +0000341 proxyPort = 0;
342
Owen Taylor3473f882001-02-23 17:55:21 +0000343#ifdef DEBUG_HTTP
344 if (URL == NULL)
345 xmlGenericError(xmlGenericErrorContext,
346 "Removing HTTP proxy info\n");
347 else
348 xmlGenericError(xmlGenericErrorContext,
349 "Using HTTP proxy %s\n", URL);
350#endif
351 if (URL == NULL) return;
William M. Brack015ccb22005-02-13 08:18:52 +0000352
Daniel Veillard336a8e12005-08-07 10:46:19 +0000353 uri = xmlParseURIRaw(URL, 1);
William M. Brack015ccb22005-02-13 08:18:52 +0000354 if ((uri == NULL) || (uri->scheme == NULL) ||
355 (strcmp(uri->scheme, "http")) || (uri->server == NULL)) {
356 __xmlIOErr(XML_FROM_HTTP, XML_HTTP_URL_SYNTAX, "Syntax Error\n");
357 if (uri != NULL)
358 xmlFreeURI(uri);
359 return;
Owen Taylor3473f882001-02-23 17:55:21 +0000360 }
William M. Brack015ccb22005-02-13 08:18:52 +0000361
362 proxy = xmlMemStrdup(uri->server);
363 if (uri->port != 0)
364 proxyPort = uri->port;
Owen Taylor3473f882001-02-23 17:55:21 +0000365
William M. Brack015ccb22005-02-13 08:18:52 +0000366 xmlFreeURI(uri);
Owen Taylor3473f882001-02-23 17:55:21 +0000367}
368
369/**
370 * xmlNanoHTTPNewCtxt:
371 * @URL: The URL used to initialize the context
372 *
373 * Allocate and initialize a new HTTP context.
374 *
375 * Returns an HTTP context or NULL in case of error.
376 */
377
378static xmlNanoHTTPCtxtPtr
379xmlNanoHTTPNewCtxt(const char *URL) {
380 xmlNanoHTTPCtxtPtr ret;
381
382 ret = (xmlNanoHTTPCtxtPtr) xmlMalloc(sizeof(xmlNanoHTTPCtxt));
Daniel Veillard2b0f8792003-10-10 19:36:36 +0000383 if (ret == NULL) {
384 xmlHTTPErrMemory("allocating context");
385 return(NULL);
386 }
Owen Taylor3473f882001-02-23 17:55:21 +0000387
388 memset(ret, 0, sizeof(xmlNanoHTTPCtxt));
389 ret->port = 80;
390 ret->returnValue = 0;
391 ret->fd = -1;
Daniel Veillardf012a642001-07-23 19:10:52 +0000392 ret->ContentLength = -1;
Owen Taylor3473f882001-02-23 17:55:21 +0000393
Daniel Veillardcacbe5d2003-01-10 16:09:51 +0000394 xmlNanoHTTPScanURL(ret, URL);
Owen Taylor3473f882001-02-23 17:55:21 +0000395
396 return(ret);
397}
398
399/**
400 * xmlNanoHTTPFreeCtxt:
401 * @ctxt: an HTTP context
402 *
403 * Frees the context after closing the connection.
404 */
405
406static void
407xmlNanoHTTPFreeCtxt(xmlNanoHTTPCtxtPtr ctxt) {
408 if (ctxt == NULL) return;
409 if (ctxt->hostname != NULL) xmlFree(ctxt->hostname);
410 if (ctxt->protocol != NULL) xmlFree(ctxt->protocol);
411 if (ctxt->path != NULL) xmlFree(ctxt->path);
Daniel Veillard351f2d62005-04-13 02:55:12 +0000412 if (ctxt->query != NULL) xmlFree(ctxt->query);
Owen Taylor3473f882001-02-23 17:55:21 +0000413 if (ctxt->out != NULL) xmlFree(ctxt->out);
414 if (ctxt->in != NULL) xmlFree(ctxt->in);
415 if (ctxt->contentType != NULL) xmlFree(ctxt->contentType);
Daniel Veillard847332a2003-10-18 11:29:40 +0000416 if (ctxt->encoding != NULL) xmlFree(ctxt->encoding);
Daniel Veillarda840b692003-10-19 13:35:37 +0000417 if (ctxt->mimeType != NULL) xmlFree(ctxt->mimeType);
Owen Taylor3473f882001-02-23 17:55:21 +0000418 if (ctxt->location != NULL) xmlFree(ctxt->location);
419 if (ctxt->authHeader != NULL) xmlFree(ctxt->authHeader);
Daniel Veillard9a2724d2005-12-15 11:12:26 +0000420#ifdef HAVE_ZLIB_H
421 if (ctxt->strm != NULL) {
422 inflateEnd(ctxt->strm);
423 xmlFree(ctxt->strm);
424 }
425#endif
426
Owen Taylor3473f882001-02-23 17:55:21 +0000427 ctxt->state = XML_NANO_HTTP_NONE;
428 if (ctxt->fd >= 0) closesocket(ctxt->fd);
429 ctxt->fd = -1;
430 xmlFree(ctxt);
431}
432
433/**
434 * xmlNanoHTTPSend:
435 * @ctxt: an HTTP context
436 *
437 * Send the input needed to initiate the processing on the server side
Daniel Veillardf012a642001-07-23 19:10:52 +0000438 * Returns number of bytes sent or -1 on error.
Owen Taylor3473f882001-02-23 17:55:21 +0000439 */
440
Daniel Veillardf012a642001-07-23 19:10:52 +0000441static int
442xmlNanoHTTPSend(xmlNanoHTTPCtxtPtr ctxt, const char * xmt_ptr, int outlen) {
443
444 int total_sent = 0;
445
446 if ( (ctxt->state & XML_NANO_HTTP_WRITE) && (xmt_ptr != NULL ) ) {
447 while (total_sent < outlen) {
448 int nsent = send(ctxt->fd, xmt_ptr + total_sent,
449 outlen - total_sent, 0);
Owen Taylor3473f882001-02-23 17:55:21 +0000450 if (nsent>0)
451 total_sent += nsent;
Daniel Veillardf012a642001-07-23 19:10:52 +0000452 else if ( ( nsent == -1 ) &&
Daniel Veillardba6db032001-07-31 16:25:45 +0000453#if defined(EAGAIN) && EAGAIN != EWOULDBLOCK
Daniel Veillardf012a642001-07-23 19:10:52 +0000454 ( socket_errno( ) != EAGAIN ) &&
Daniel Veillardba6db032001-07-31 16:25:45 +0000455#endif
Daniel Veillardf012a642001-07-23 19:10:52 +0000456 ( socket_errno( ) != EWOULDBLOCK ) ) {
Daniel Veillard2b0f8792003-10-10 19:36:36 +0000457 __xmlIOErr(XML_FROM_HTTP, 0, "send failed\n");
Daniel Veillardf012a642001-07-23 19:10:52 +0000458 if ( total_sent == 0 )
459 total_sent = -1;
460 break;
461 }
462 else {
463 /*
464 ** No data sent
465 ** Since non-blocking sockets are used, wait for
466 ** socket to be writable or default timeout prior
467 ** to retrying.
468 */
469
470 struct timeval tv;
471 fd_set wfd;
472
473 tv.tv_sec = timeout;
474 tv.tv_usec = 0;
475 FD_ZERO( &wfd );
Daniel Veillard9e2110b2005-08-08 20:33:54 +0000476#ifdef _MSC_VER
477#pragma warning(push)
478#pragma warning(disable: 4018)
479#endif
Daniel Veillardf012a642001-07-23 19:10:52 +0000480 FD_SET( ctxt->fd, &wfd );
Daniel Veillard9e2110b2005-08-08 20:33:54 +0000481#ifdef _MSC_VER
482#pragma warning(pop)
483#endif
Daniel Veillardf012a642001-07-23 19:10:52 +0000484 (void)select( ctxt->fd + 1, NULL, &wfd, NULL, &tv );
485 }
Daniel Veillard56a4cb82001-03-24 17:00:36 +0000486 }
Owen Taylor3473f882001-02-23 17:55:21 +0000487 }
Daniel Veillardf012a642001-07-23 19:10:52 +0000488
489 return total_sent;
Owen Taylor3473f882001-02-23 17:55:21 +0000490}
491
492/**
493 * xmlNanoHTTPRecv:
494 * @ctxt: an HTTP context
495 *
496 * Read information coming from the HTTP connection.
497 * This is a blocking call (but it blocks in select(), not read()).
498 *
499 * Returns the number of byte read or -1 in case of error.
500 */
501
502static int
503xmlNanoHTTPRecv(xmlNanoHTTPCtxtPtr ctxt) {
504 fd_set rfd;
505 struct timeval tv;
506
507
508 while (ctxt->state & XML_NANO_HTTP_READ) {
509 if (ctxt->in == NULL) {
Daniel Veillard3c908dc2003-04-19 00:07:51 +0000510 ctxt->in = (char *) xmlMallocAtomic(65000 * sizeof(char));
Owen Taylor3473f882001-02-23 17:55:21 +0000511 if (ctxt->in == NULL) {
Daniel Veillard2b0f8792003-10-10 19:36:36 +0000512 xmlHTTPErrMemory("allocating input");
Owen Taylor3473f882001-02-23 17:55:21 +0000513 ctxt->last = -1;
514 return(-1);
515 }
516 ctxt->inlen = 65000;
517 ctxt->inptr = ctxt->content = ctxt->inrptr = ctxt->in;
518 }
519 if (ctxt->inrptr > ctxt->in + XML_NANO_HTTP_CHUNK) {
520 int delta = ctxt->inrptr - ctxt->in;
521 int len = ctxt->inptr - ctxt->inrptr;
522
523 memmove(ctxt->in, ctxt->inrptr, len);
524 ctxt->inrptr -= delta;
525 ctxt->content -= delta;
526 ctxt->inptr -= delta;
527 }
528 if ((ctxt->in + ctxt->inlen) < (ctxt->inptr + XML_NANO_HTTP_CHUNK)) {
529 int d_inptr = ctxt->inptr - ctxt->in;
530 int d_content = ctxt->content - ctxt->in;
531 int d_inrptr = ctxt->inrptr - ctxt->in;
Daniel Veillardf012a642001-07-23 19:10:52 +0000532 char * tmp_ptr = ctxt->in;
Owen Taylor3473f882001-02-23 17:55:21 +0000533
534 ctxt->inlen *= 2;
Daniel Veillardf012a642001-07-23 19:10:52 +0000535 ctxt->in = (char *) xmlRealloc(tmp_ptr, ctxt->inlen);
Owen Taylor3473f882001-02-23 17:55:21 +0000536 if (ctxt->in == NULL) {
Daniel Veillard2b0f8792003-10-10 19:36:36 +0000537 xmlHTTPErrMemory("allocating input buffer");
Daniel Veillardf012a642001-07-23 19:10:52 +0000538 xmlFree( tmp_ptr );
Owen Taylor3473f882001-02-23 17:55:21 +0000539 ctxt->last = -1;
540 return(-1);
541 }
542 ctxt->inptr = ctxt->in + d_inptr;
543 ctxt->content = ctxt->in + d_content;
544 ctxt->inrptr = ctxt->in + d_inrptr;
545 }
546 ctxt->last = recv(ctxt->fd, ctxt->inptr, XML_NANO_HTTP_CHUNK, 0);
547 if (ctxt->last > 0) {
548 ctxt->inptr += ctxt->last;
549 return(ctxt->last);
550 }
551 if (ctxt->last == 0) {
552 return(0);
553 }
554 if (ctxt->last == -1) {
555 switch (socket_errno()) {
556 case EINPROGRESS:
557 case EWOULDBLOCK:
558#if defined(EAGAIN) && EAGAIN != EWOULDBLOCK
559 case EAGAIN:
560#endif
561 break;
Daniel Veillardf012a642001-07-23 19:10:52 +0000562
563 case ECONNRESET:
564 case ESHUTDOWN:
565 return ( 0 );
566
Owen Taylor3473f882001-02-23 17:55:21 +0000567 default:
Daniel Veillard2b0f8792003-10-10 19:36:36 +0000568 __xmlIOErr(XML_FROM_HTTP, 0, "recv failed\n");
Daniel Veillardf012a642001-07-23 19:10:52 +0000569 return(-1);
Owen Taylor3473f882001-02-23 17:55:21 +0000570 }
571 }
572
573 tv.tv_sec = timeout;
574 tv.tv_usec = 0;
575 FD_ZERO(&rfd);
Daniel Veillard9e2110b2005-08-08 20:33:54 +0000576#ifdef _MSC_VER
577#pragma warning(push)
578#pragma warning(disable: 4018)
579#endif
Owen Taylor3473f882001-02-23 17:55:21 +0000580 FD_SET(ctxt->fd, &rfd);
Daniel Veillard9e2110b2005-08-08 20:33:54 +0000581#ifdef _MSC_VER
582#pragma warning(pop)
583#endif
Owen Taylor3473f882001-02-23 17:55:21 +0000584
Daniel Veillard50f34372001-08-03 12:06:36 +0000585 if ( (select(ctxt->fd+1, &rfd, NULL, NULL, &tv)<1)
586#if defined(EINTR)
587 && (errno != EINTR)
588#endif
589 )
Owen Taylor3473f882001-02-23 17:55:21 +0000590 return(0);
591 }
592 return(0);
593}
594
595/**
596 * xmlNanoHTTPReadLine:
597 * @ctxt: an HTTP context
598 *
599 * Read one line in the HTTP server output, usually for extracting
600 * the HTTP protocol informations from the answer header.
601 *
602 * Returns a newly allocated string with a copy of the line, or NULL
603 * which indicate the end of the input.
604 */
605
606static char *
607xmlNanoHTTPReadLine(xmlNanoHTTPCtxtPtr ctxt) {
608 char buf[4096];
609 char *bp = buf;
Daniel Veillardf012a642001-07-23 19:10:52 +0000610 int rc;
Owen Taylor3473f882001-02-23 17:55:21 +0000611
612 while (bp - buf < 4095) {
613 if (ctxt->inrptr == ctxt->inptr) {
Daniel Veillardf012a642001-07-23 19:10:52 +0000614 if ( (rc = xmlNanoHTTPRecv(ctxt)) == 0) {
Owen Taylor3473f882001-02-23 17:55:21 +0000615 if (bp == buf)
616 return(NULL);
617 else
618 *bp = 0;
619 return(xmlMemStrdup(buf));
620 }
Daniel Veillardf012a642001-07-23 19:10:52 +0000621 else if ( rc == -1 ) {
622 return ( NULL );
623 }
Owen Taylor3473f882001-02-23 17:55:21 +0000624 }
625 *bp = *ctxt->inrptr++;
626 if (*bp == '\n') {
627 *bp = 0;
628 return(xmlMemStrdup(buf));
629 }
630 if (*bp != '\r')
631 bp++;
632 }
633 buf[4095] = 0;
634 return(xmlMemStrdup(buf));
635}
636
637
638/**
639 * xmlNanoHTTPScanAnswer:
640 * @ctxt: an HTTP context
641 * @line: an HTTP header line
642 *
643 * Try to extract useful informations from the server answer.
644 * We currently parse and process:
645 * - The HTTP revision/ return code
Daniel Veillarda840b692003-10-19 13:35:37 +0000646 * - The Content-Type, Mime-Type and charset used
Daniel Veillardcbaf3992001-12-31 16:16:02 +0000647 * - The Location for redirect processing.
Owen Taylor3473f882001-02-23 17:55:21 +0000648 *
649 * Returns -1 in case of failure, the file descriptor number otherwise
650 */
651
652static void
653xmlNanoHTTPScanAnswer(xmlNanoHTTPCtxtPtr ctxt, const char *line) {
654 const char *cur = line;
655
656 if (line == NULL) return;
657
658 if (!strncmp(line, "HTTP/", 5)) {
659 int version = 0;
660 int ret = 0;
661
662 cur += 5;
663 while ((*cur >= '0') && (*cur <= '9')) {
664 version *= 10;
665 version += *cur - '0';
666 cur++;
667 }
668 if (*cur == '.') {
669 cur++;
670 if ((*cur >= '0') && (*cur <= '9')) {
671 version *= 10;
672 version += *cur - '0';
673 cur++;
674 }
675 while ((*cur >= '0') && (*cur <= '9'))
676 cur++;
677 } else
678 version *= 10;
679 if ((*cur != ' ') && (*cur != '\t')) return;
680 while ((*cur == ' ') || (*cur == '\t')) cur++;
681 if ((*cur < '0') || (*cur > '9')) return;
682 while ((*cur >= '0') && (*cur <= '9')) {
683 ret *= 10;
684 ret += *cur - '0';
685 cur++;
686 }
687 if ((*cur != 0) && (*cur != ' ') && (*cur != '\t')) return;
688 ctxt->returnValue = ret;
689 } else if (!xmlStrncasecmp(BAD_CAST line, BAD_CAST"Content-Type:", 13)) {
Daniel Veillarda840b692003-10-19 13:35:37 +0000690 const xmlChar *charset, *last, *mime;
Owen Taylor3473f882001-02-23 17:55:21 +0000691 cur += 13;
692 while ((*cur == ' ') || (*cur == '\t')) cur++;
693 if (ctxt->contentType != NULL)
694 xmlFree(ctxt->contentType);
695 ctxt->contentType = xmlMemStrdup(cur);
Daniel Veillarda840b692003-10-19 13:35:37 +0000696 mime = (const xmlChar *) cur;
697 last = mime;
698 while ((*last != 0) && (*last != ' ') && (*last != '\t') &&
699 (*last != ';') && (*last != ','))
700 last++;
701 if (ctxt->mimeType != NULL)
702 xmlFree(ctxt->mimeType);
703 ctxt->mimeType = (char *) xmlStrndup(mime, last - mime);
704 charset = xmlStrstr(BAD_CAST ctxt->contentType, BAD_CAST "charset=");
705 if (charset != NULL) {
706 charset += 8;
707 last = charset;
708 while ((*last != 0) && (*last != ' ') && (*last != '\t') &&
709 (*last != ';') && (*last != ','))
710 last++;
711 if (ctxt->encoding != NULL)
712 xmlFree(ctxt->encoding);
713 ctxt->encoding = (char *) xmlStrndup(charset, last - charset);
714 }
Owen Taylor3473f882001-02-23 17:55:21 +0000715 } else if (!xmlStrncasecmp(BAD_CAST line, BAD_CAST"ContentType:", 12)) {
Daniel Veillarda840b692003-10-19 13:35:37 +0000716 const xmlChar *charset, *last, *mime;
Owen Taylor3473f882001-02-23 17:55:21 +0000717 cur += 12;
718 if (ctxt->contentType != NULL) return;
719 while ((*cur == ' ') || (*cur == '\t')) cur++;
720 ctxt->contentType = xmlMemStrdup(cur);
Daniel Veillarda840b692003-10-19 13:35:37 +0000721 mime = (const xmlChar *) cur;
722 last = mime;
723 while ((*last != 0) && (*last != ' ') && (*last != '\t') &&
724 (*last != ';') && (*last != ','))
725 last++;
726 if (ctxt->mimeType != NULL)
727 xmlFree(ctxt->mimeType);
728 ctxt->mimeType = (char *) xmlStrndup(mime, last - mime);
729 charset = xmlStrstr(BAD_CAST ctxt->contentType, BAD_CAST "charset=");
730 if (charset != NULL) {
731 charset += 8;
732 last = charset;
733 while ((*last != 0) && (*last != ' ') && (*last != '\t') &&
734 (*last != ';') && (*last != ','))
735 last++;
736 if (ctxt->encoding != NULL)
737 xmlFree(ctxt->encoding);
738 ctxt->encoding = (char *) xmlStrndup(charset, last - charset);
739 }
Owen Taylor3473f882001-02-23 17:55:21 +0000740 } else if (!xmlStrncasecmp(BAD_CAST line, BAD_CAST"Location:", 9)) {
741 cur += 9;
742 while ((*cur == ' ') || (*cur == '\t')) cur++;
743 if (ctxt->location != NULL)
744 xmlFree(ctxt->location);
William M. Brack7e29c0a2004-04-02 09:07:22 +0000745 if (*cur == '/') {
746 xmlChar *tmp_http = xmlStrdup(BAD_CAST "http://");
747 xmlChar *tmp_loc =
748 xmlStrcat(tmp_http, (const xmlChar *) ctxt->hostname);
749 ctxt->location =
750 (char *) xmlStrcat (tmp_loc, (const xmlChar *) cur);
751 } else {
752 ctxt->location = xmlMemStrdup(cur);
753 }
Owen Taylor3473f882001-02-23 17:55:21 +0000754 } else if (!xmlStrncasecmp(BAD_CAST line, BAD_CAST"WWW-Authenticate:", 17)) {
755 cur += 17;
756 while ((*cur == ' ') || (*cur == '\t')) cur++;
757 if (ctxt->authHeader != NULL)
758 xmlFree(ctxt->authHeader);
759 ctxt->authHeader = xmlMemStrdup(cur);
760 } else if (!xmlStrncasecmp(BAD_CAST line, BAD_CAST"Proxy-Authenticate:", 19)) {
761 cur += 19;
762 while ((*cur == ' ') || (*cur == '\t')) cur++;
763 if (ctxt->authHeader != NULL)
764 xmlFree(ctxt->authHeader);
765 ctxt->authHeader = xmlMemStrdup(cur);
Daniel Veillard9a2724d2005-12-15 11:12:26 +0000766#ifdef HAVE_ZLIB_H
767 } else if ( !xmlStrncasecmp( BAD_CAST line, BAD_CAST"Content-Encoding:", 17) ) {
768 cur += 17;
769 while ((*cur == ' ') || (*cur == '\t')) cur++;
770 if ( !xmlStrncasecmp( BAD_CAST cur, BAD_CAST"gzip", 4) ) {
771 ctxt->usesGzip = 1;
772
773 ctxt->strm = xmlMalloc(sizeof(z_stream));
774
775 if (ctxt->strm != NULL) {
776 ctxt->strm->zalloc = Z_NULL;
777 ctxt->strm->zfree = Z_NULL;
778 ctxt->strm->opaque = Z_NULL;
779 ctxt->strm->avail_in = 0;
780 ctxt->strm->next_in = Z_NULL;
781
782 inflateInit2( ctxt->strm, 31 );
783 }
784 }
785#endif
Daniel Veillardf012a642001-07-23 19:10:52 +0000786 } else if ( !xmlStrncasecmp( BAD_CAST line, BAD_CAST"Content-Length:", 15) ) {
787 cur += 15;
788 ctxt->ContentLength = strtol( cur, NULL, 10 );
Owen Taylor3473f882001-02-23 17:55:21 +0000789 }
790}
791
792/**
793 * xmlNanoHTTPConnectAttempt:
Daniel Veillardcbaf3992001-12-31 16:16:02 +0000794 * @addr: a socket address structure
Owen Taylor3473f882001-02-23 17:55:21 +0000795 *
796 * Attempt a connection to the given IP:port endpoint. It forces
797 * non-blocking semantic on the socket, and allow 60 seconds for
798 * the host to answer.
799 *
800 * Returns -1 in case of failure, the file descriptor number otherwise
801 */
802
803static int
Daniel Veillard56a4cb82001-03-24 17:00:36 +0000804xmlNanoHTTPConnectAttempt(struct sockaddr *addr)
Owen Taylor3473f882001-02-23 17:55:21 +0000805{
Owen Taylor3473f882001-02-23 17:55:21 +0000806 fd_set wfd;
Daniel Veillard5bb9ccd2004-02-09 12:39:02 +0000807#ifdef _WINSOCKAPI_
808 fd_set xfd;
809#endif
Owen Taylor3473f882001-02-23 17:55:21 +0000810 struct timeval tv;
811 int status;
Daniel Veillardde2a67b2003-06-21 14:20:04 +0000812 int addrlen;
813 SOCKET s;
Owen Taylor3473f882001-02-23 17:55:21 +0000814
Daniel Veillardde2a67b2003-06-21 14:20:04 +0000815#ifdef SUPPORT_IP6
816 if (addr->sa_family == AF_INET6) {
817 s = socket (PF_INET6, SOCK_STREAM, IPPROTO_TCP);
818 addrlen = sizeof (struct sockaddr_in6);
819 }
820 else
821#endif
822 {
823 s = socket (PF_INET, SOCK_STREAM, IPPROTO_TCP);
824 addrlen = sizeof (struct sockaddr_in);
825 }
Owen Taylor3473f882001-02-23 17:55:21 +0000826 if (s==-1) {
827#ifdef DEBUG_HTTP
828 perror("socket");
829#endif
Daniel Veillard2b0f8792003-10-10 19:36:36 +0000830 __xmlIOErr(XML_FROM_HTTP, 0, "socket failed\n");
Owen Taylor3473f882001-02-23 17:55:21 +0000831 return(-1);
832 }
833
834#ifdef _WINSOCKAPI_
835 {
836 u_long one = 1;
837
838 status = ioctlsocket(s, FIONBIO, &one) == SOCKET_ERROR ? -1 : 0;
839 }
840#else /* _WINSOCKAPI_ */
841#if defined(VMS)
842 {
843 int enable = 1;
844 status = ioctl(s, FIONBIO, &enable);
845 }
846#else /* VMS */
Daniel Veillard254b1262003-11-01 17:04:58 +0000847#if defined(__BEOS__)
848 {
849 bool noblock = true;
850 status = setsockopt(s, SOL_SOCKET, SO_NONBLOCK, &noblock, sizeof(noblock));
851 }
852#else /* __BEOS__ */
Owen Taylor3473f882001-02-23 17:55:21 +0000853 if ((status = fcntl(s, F_GETFL, 0)) != -1) {
854#ifdef O_NONBLOCK
855 status |= O_NONBLOCK;
856#else /* O_NONBLOCK */
857#ifdef F_NDELAY
858 status |= F_NDELAY;
859#endif /* F_NDELAY */
860#endif /* !O_NONBLOCK */
861 status = fcntl(s, F_SETFL, status);
862 }
863 if (status < 0) {
864#ifdef DEBUG_HTTP
865 perror("nonblocking");
866#endif
Daniel Veillard2b0f8792003-10-10 19:36:36 +0000867 __xmlIOErr(XML_FROM_HTTP, 0, "error setting non-blocking IO\n");
Owen Taylor3473f882001-02-23 17:55:21 +0000868 closesocket(s);
869 return(-1);
870 }
Daniel Veillard254b1262003-11-01 17:04:58 +0000871#endif /* !__BEOS__ */
Owen Taylor3473f882001-02-23 17:55:21 +0000872#endif /* !VMS */
873#endif /* !_WINSOCKAPI_ */
874
Daniel Veillardde2a67b2003-06-21 14:20:04 +0000875 if (connect (s, addr, addrlen) == -1) {
Owen Taylor3473f882001-02-23 17:55:21 +0000876 switch (socket_errno()) {
877 case EINPROGRESS:
878 case EWOULDBLOCK:
879 break;
880 default:
Daniel Veillard2b0f8792003-10-10 19:36:36 +0000881 __xmlIOErr(XML_FROM_HTTP, 0, "error connecting to HTTP server");
Owen Taylor3473f882001-02-23 17:55:21 +0000882 closesocket(s);
883 return(-1);
884 }
885 }
886
887 tv.tv_sec = timeout;
888 tv.tv_usec = 0;
Daniel Veillard9e2110b2005-08-08 20:33:54 +0000889
890#ifdef _MSC_VER
891#pragma warning(push)
892#pragma warning(disable: 4018)
893#endif
Owen Taylor3473f882001-02-23 17:55:21 +0000894 FD_ZERO(&wfd);
895 FD_SET(s, &wfd);
Daniel Veillard5bb9ccd2004-02-09 12:39:02 +0000896
897#ifdef _WINSOCKAPI_
898 FD_ZERO(&xfd);
899 FD_SET(s, &xfd);
Owen Taylor3473f882001-02-23 17:55:21 +0000900
Daniel Veillard5bb9ccd2004-02-09 12:39:02 +0000901 switch(select(s+1, NULL, &wfd, &xfd, &tv))
902#else
Owen Taylor3473f882001-02-23 17:55:21 +0000903 switch(select(s+1, NULL, &wfd, NULL, &tv))
Daniel Veillard5bb9ccd2004-02-09 12:39:02 +0000904#endif
Daniel Veillard9e2110b2005-08-08 20:33:54 +0000905#ifdef _MSC_VER
906#pragma warning(pop)
907#endif
Owen Taylor3473f882001-02-23 17:55:21 +0000908 {
909 case 0:
910 /* Time out */
Daniel Veillard2b0f8792003-10-10 19:36:36 +0000911 __xmlIOErr(XML_FROM_HTTP, 0, "Connect attempt timed out");
Owen Taylor3473f882001-02-23 17:55:21 +0000912 closesocket(s);
913 return(-1);
914 case -1:
915 /* Ermm.. ?? */
Daniel Veillard2b0f8792003-10-10 19:36:36 +0000916 __xmlIOErr(XML_FROM_HTTP, 0, "Connect failed");
Owen Taylor3473f882001-02-23 17:55:21 +0000917 closesocket(s);
918 return(-1);
919 }
920
Daniel Veillard5bb9ccd2004-02-09 12:39:02 +0000921 if ( FD_ISSET(s, &wfd)
922#ifdef _WINSOCKAPI_
923 || FD_ISSET(s, &xfd)
924#endif
925 ) {
Daniel Veillardc284c642005-03-31 10:24:24 +0000926 XML_SOCKLEN_T len;
Owen Taylor3473f882001-02-23 17:55:21 +0000927 len = sizeof(status);
Daniel Veillarda9cce9c2003-09-29 13:20:24 +0000928#ifdef SO_ERROR
Owen Taylor3473f882001-02-23 17:55:21 +0000929 if (getsockopt(s, SOL_SOCKET, SO_ERROR, (char*)&status, &len) < 0 ) {
930 /* Solaris error code */
Daniel Veillard2b0f8792003-10-10 19:36:36 +0000931 __xmlIOErr(XML_FROM_HTTP, 0, "getsockopt failed\n");
Owen Taylor3473f882001-02-23 17:55:21 +0000932 return (-1);
933 }
Daniel Veillarda9cce9c2003-09-29 13:20:24 +0000934#endif
Owen Taylor3473f882001-02-23 17:55:21 +0000935 if ( status ) {
Daniel Veillard2b0f8792003-10-10 19:36:36 +0000936 __xmlIOErr(XML_FROM_HTTP, 0, "Error connecting to remote host");
Owen Taylor3473f882001-02-23 17:55:21 +0000937 closesocket(s);
938 errno = status;
939 return (-1);
940 }
941 } else {
942 /* pbm */
Daniel Veillard2b0f8792003-10-10 19:36:36 +0000943 __xmlIOErr(XML_FROM_HTTP, 0, "select failed\n");
Daniel Veillardf012a642001-07-23 19:10:52 +0000944 closesocket(s);
Owen Taylor3473f882001-02-23 17:55:21 +0000945 return (-1);
946 }
947
948 return(s);
949}
950
951/**
952 * xmlNanoHTTPConnectHost:
953 * @host: the host name
954 * @port: the port number
955 *
956 * Attempt a connection to the given host:port endpoint. It tries
957 * the multiple IP provided by the DNS if available.
958 *
959 * Returns -1 in case of failure, the file descriptor number otherwise
960 */
961
962static int
963xmlNanoHTTPConnectHost(const char *host, int port)
964{
965 struct hostent *h;
Daniel Veillard2db8c122003-07-08 12:16:59 +0000966 struct sockaddr *addr = NULL;
Owen Taylor3473f882001-02-23 17:55:21 +0000967 struct in_addr ia;
Daniel Veillard56a4cb82001-03-24 17:00:36 +0000968 struct sockaddr_in sockin;
Daniel Veillard5c396542002-03-15 07:57:50 +0000969
Owen Taylor3473f882001-02-23 17:55:21 +0000970#ifdef SUPPORT_IP6
971 struct in6_addr ia6;
Daniel Veillard56a4cb82001-03-24 17:00:36 +0000972 struct sockaddr_in6 sockin6;
Owen Taylor3473f882001-02-23 17:55:21 +0000973#endif
974 int i;
975 int s;
Daniel Veillard5c396542002-03-15 07:57:50 +0000976
Daniel Veillardde2a67b2003-06-21 14:20:04 +0000977 memset (&sockin, 0, sizeof(sockin));
978#ifdef SUPPORT_IP6
979 memset (&sockin6, 0, sizeof(sockin6));
Rob Richardscb418de2005-10-13 23:12:42 +0000980#endif
981
982#if !defined(HAVE_GETADDRINFO) && defined(SUPPORT_IP6) && defined(RES_USE_INET6)
Daniel Veillardde2a67b2003-06-21 14:20:04 +0000983 if (have_ipv6 ())
Daniel Veillard560c2a42003-07-06 21:13:49 +0000984 {
Daniel Veillardde2a67b2003-06-21 14:20:04 +0000985 if (!(_res.options & RES_INIT))
986 res_init();
987 _res.options |= RES_USE_INET6;
988 }
Rob Richardscb418de2005-10-13 23:12:42 +0000989#endif
990
991#if defined(HAVE_GETADDRINFO) && defined(SUPPORT_IP6) && !defined(_WIN32)
992 if (have_ipv6 ())
993#endif
994#if defined(HAVE_GETADDRINFO) && (defined(SUPPORT_IP6) || defined(_WIN32))
Daniel Veillard560c2a42003-07-06 21:13:49 +0000995 {
Daniel Veillardde2a67b2003-06-21 14:20:04 +0000996 int status;
997 struct addrinfo hints, *res, *result;
998
999 result = NULL;
1000 memset (&hints, 0,sizeof(hints));
1001 hints.ai_socktype = SOCK_STREAM;
1002
1003 status = getaddrinfo (host, NULL, &hints, &result);
1004 if (status) {
Daniel Veillard2b0f8792003-10-10 19:36:36 +00001005 __xmlIOErr(XML_FROM_HTTP, 0, "getaddrinfo failed\n");
Daniel Veillardde2a67b2003-06-21 14:20:04 +00001006 return (-1);
1007 }
1008
1009 for (res = result; res; res = res->ai_next) {
Rob Richardscb418de2005-10-13 23:12:42 +00001010 if (res->ai_family == AF_INET) {
1011 if (res->ai_addrlen > sizeof(sockin)) {
1012 __xmlIOErr(XML_FROM_HTTP, 0, "address size mismatch\n");
Daniel Veillardde2a67b2003-06-21 14:20:04 +00001013 freeaddrinfo (result);
Rob Richardscb418de2005-10-13 23:12:42 +00001014 return (-1);
Daniel Veillardde2a67b2003-06-21 14:20:04 +00001015 }
Rob Richardscb418de2005-10-13 23:12:42 +00001016 memcpy (&sockin, res->ai_addr, res->ai_addrlen);
1017 sockin.sin_port = htons (port);
1018 addr = (struct sockaddr *)&sockin;
1019#ifdef SUPPORT_IP6
1020 } else if (have_ipv6 () && (res->ai_family == AF_INET6)) {
1021 if (res->ai_addrlen > sizeof(sockin6)) {
1022 __xmlIOErr(XML_FROM_HTTP, 0, "address size mismatch\n");
1023 freeaddrinfo (result);
1024 return (-1);
1025 }
1026 memcpy (&sockin6, res->ai_addr, res->ai_addrlen);
1027 sockin6.sin6_port = htons (port);
1028 addr = (struct sockaddr *)&sockin6;
1029#endif
1030 } else
1031 continue; /* for */
1032
1033 s = xmlNanoHTTPConnectAttempt (addr);
1034 if (s != -1) {
1035 freeaddrinfo (result);
1036 return (s);
Daniel Veillardde2a67b2003-06-21 14:20:04 +00001037 }
Daniel Veillardde2a67b2003-06-21 14:20:04 +00001038 }
Rob Richardscb418de2005-10-13 23:12:42 +00001039
Daniel Veillard3dc93a42003-07-10 14:04:33 +00001040 if (result)
1041 freeaddrinfo (result);
Rob Richardscb418de2005-10-13 23:12:42 +00001042 }
Owen Taylor3473f882001-02-23 17:55:21 +00001043#endif
Rob Richardscb418de2005-10-13 23:12:42 +00001044#if defined(HAVE_GETADDRINFO) && defined(SUPPORT_IP6) && !defined(_WIN32)
1045 else
Daniel Veillardde2a67b2003-06-21 14:20:04 +00001046#endif
Rob Richardscb418de2005-10-13 23:12:42 +00001047#if !defined(HAVE_GETADDRINFO) || !defined(_WIN32)
1048 {
Daniel Veillardde2a67b2003-06-21 14:20:04 +00001049 h = gethostbyname (host);
1050 if (h == NULL) {
Daniel Veillard56b2db72002-03-25 16:35:28 +00001051
1052/*
1053 * Okay, I got fed up by the non-portability of this error message
1054 * extraction code. it work on Linux, if it work on your platform
1055 * and one want to enable it, send me the defined(foobar) needed
1056 */
1057#if defined(HAVE_NETDB_H) && defined(HOST_NOT_FOUND) && defined(linux)
Daniel Veillardde2a67b2003-06-21 14:20:04 +00001058 const char *h_err_txt = "";
Daniel Veillardf012a642001-07-23 19:10:52 +00001059
Daniel Veillardde2a67b2003-06-21 14:20:04 +00001060 switch (h_errno) {
1061 case HOST_NOT_FOUND:
1062 h_err_txt = "Authoritive host not found";
1063 break;
Daniel Veillardf012a642001-07-23 19:10:52 +00001064
Daniel Veillardde2a67b2003-06-21 14:20:04 +00001065 case TRY_AGAIN:
1066 h_err_txt =
1067 "Non-authoritive host not found or server failure.";
1068 break;
Daniel Veillardf012a642001-07-23 19:10:52 +00001069
Daniel Veillardde2a67b2003-06-21 14:20:04 +00001070 case NO_RECOVERY:
1071 h_err_txt =
1072 "Non-recoverable errors: FORMERR, REFUSED, or NOTIMP.";
1073 break;
Daniel Veillard5c396542002-03-15 07:57:50 +00001074
Daniel Veillardde2a67b2003-06-21 14:20:04 +00001075 case NO_ADDRESS:
1076 h_err_txt =
1077 "Valid name, no data record of requested type.";
1078 break;
Daniel Veillard5c396542002-03-15 07:57:50 +00001079
Daniel Veillardde2a67b2003-06-21 14:20:04 +00001080 default:
1081 h_err_txt = "No error text defined.";
1082 break;
1083 }
Daniel Veillard2b0f8792003-10-10 19:36:36 +00001084 __xmlIOErr(XML_FROM_HTTP, 0, h_err_txt);
Daniel Veillard5c396542002-03-15 07:57:50 +00001085#else
Daniel Veillard2b0f8792003-10-10 19:36:36 +00001086 __xmlIOErr(XML_FROM_HTTP, 0, "Failed to resolve host");
Owen Taylor3473f882001-02-23 17:55:21 +00001087#endif
Daniel Veillardde2a67b2003-06-21 14:20:04 +00001088 return (-1);
1089 }
Daniel Veillard5c396542002-03-15 07:57:50 +00001090
Daniel Veillardde2a67b2003-06-21 14:20:04 +00001091 for (i = 0; h->h_addr_list[i]; i++) {
1092 if (h->h_addrtype == AF_INET) {
1093 /* A records (IPv4) */
Daniel Veillard8e2c9792004-10-27 09:39:50 +00001094 if ((unsigned int) h->h_length > sizeof(ia)) {
1095 __xmlIOErr(XML_FROM_HTTP, 0, "address size mismatch\n");
1096 return (-1);
1097 }
Daniel Veillardde2a67b2003-06-21 14:20:04 +00001098 memcpy (&ia, h->h_addr_list[i], h->h_length);
1099 sockin.sin_family = h->h_addrtype;
1100 sockin.sin_addr = ia;
Daniel Veillard9e2110b2005-08-08 20:33:54 +00001101 sockin.sin_port = (u_short)htons ((unsigned short)port);
Daniel Veillardde2a67b2003-06-21 14:20:04 +00001102 addr = (struct sockaddr *) &sockin;
Daniel Veillard5c396542002-03-15 07:57:50 +00001103#ifdef SUPPORT_IP6
Daniel Veillardde2a67b2003-06-21 14:20:04 +00001104 } else if (have_ipv6 () && (h->h_addrtype == AF_INET6)) {
1105 /* AAAA records (IPv6) */
Daniel Veillard8e2c9792004-10-27 09:39:50 +00001106 if ((unsigned int) h->h_length > sizeof(ia6)) {
1107 __xmlIOErr(XML_FROM_HTTP, 0, "address size mismatch\n");
1108 return (-1);
1109 }
Daniel Veillardde2a67b2003-06-21 14:20:04 +00001110 memcpy (&ia6, h->h_addr_list[i], h->h_length);
1111 sockin6.sin6_family = h->h_addrtype;
1112 sockin6.sin6_addr = ia6;
1113 sockin6.sin6_port = htons (port);
1114 addr = (struct sockaddr *) &sockin6;
Daniel Veillard5c396542002-03-15 07:57:50 +00001115#endif
Daniel Veillardde2a67b2003-06-21 14:20:04 +00001116 } else
1117 break; /* for */
Daniel Veillard5c396542002-03-15 07:57:50 +00001118
Daniel Veillardde2a67b2003-06-21 14:20:04 +00001119 s = xmlNanoHTTPConnectAttempt (addr);
1120 if (s != -1)
1121 return (s);
1122 }
Owen Taylor3473f882001-02-23 17:55:21 +00001123 }
Rob Richardscb418de2005-10-13 23:12:42 +00001124#endif
1125
Owen Taylor3473f882001-02-23 17:55:21 +00001126#ifdef DEBUG_HTTP
1127 xmlGenericError(xmlGenericErrorContext,
Daniel Veillard5c396542002-03-15 07:57:50 +00001128 "xmlNanoHTTPConnectHost: unable to connect to '%s'.\n",
1129 host);
Owen Taylor3473f882001-02-23 17:55:21 +00001130#endif
Daniel Veillard5c396542002-03-15 07:57:50 +00001131 return (-1);
Owen Taylor3473f882001-02-23 17:55:21 +00001132}
1133
1134
1135/**
1136 * xmlNanoHTTPOpen:
1137 * @URL: The URL to load
1138 * @contentType: if available the Content-Type information will be
1139 * returned at that location
1140 *
1141 * This function try to open a connection to the indicated resource
1142 * via HTTP GET.
1143 *
1144 * Returns NULL in case of failure, otherwise a request handler.
1145 * The contentType, if provided must be freed by the caller
1146 */
1147
1148void*
1149xmlNanoHTTPOpen(const char *URL, char **contentType) {
1150 if (contentType != NULL) *contentType = NULL;
Daniel Veillardf012a642001-07-23 19:10:52 +00001151 return(xmlNanoHTTPMethod(URL, NULL, NULL, contentType, NULL, 0));
Daniel Veillard9403a042001-05-28 11:00:53 +00001152}
1153
1154/**
1155 * xmlNanoHTTPOpenRedir:
1156 * @URL: The URL to load
1157 * @contentType: if available the Content-Type information will be
1158 * returned at that location
Daniel Veillardcbaf3992001-12-31 16:16:02 +00001159 * @redir: if available the redirected URL will be returned
Daniel Veillard9403a042001-05-28 11:00:53 +00001160 *
1161 * This function try to open a connection to the indicated resource
1162 * via HTTP GET.
1163 *
1164 * Returns NULL in case of failure, otherwise a request handler.
1165 * The contentType, if provided must be freed by the caller
1166 */
1167
1168void*
1169xmlNanoHTTPOpenRedir(const char *URL, char **contentType, char **redir) {
1170 if (contentType != NULL) *contentType = NULL;
1171 if (redir != NULL) *redir = NULL;
Daniel Veillardf012a642001-07-23 19:10:52 +00001172 return(xmlNanoHTTPMethodRedir(URL, NULL, NULL, contentType, redir, NULL,0));
Owen Taylor3473f882001-02-23 17:55:21 +00001173}
1174
1175/**
1176 * xmlNanoHTTPRead:
1177 * @ctx: the HTTP context
1178 * @dest: a buffer
1179 * @len: the buffer length
1180 *
1181 * This function tries to read @len bytes from the existing HTTP connection
1182 * and saves them in @dest. This is a blocking call.
1183 *
1184 * Returns the number of byte read. 0 is an indication of an end of connection.
1185 * -1 indicates a parameter error.
1186 */
1187int
1188xmlNanoHTTPRead(void *ctx, void *dest, int len) {
1189 xmlNanoHTTPCtxtPtr ctxt = (xmlNanoHTTPCtxtPtr) ctx;
Daniel Veillard9a2724d2005-12-15 11:12:26 +00001190#ifdef HAVE_ZLIB_H
1191 int bytes_read = 0;
1192 int orig_avail_in;
1193 int z_ret;
1194#endif
Owen Taylor3473f882001-02-23 17:55:21 +00001195
1196 if (ctx == NULL) return(-1);
1197 if (dest == NULL) return(-1);
1198 if (len <= 0) return(0);
1199
Daniel Veillard9a2724d2005-12-15 11:12:26 +00001200#ifdef HAVE_ZLIB_H
1201 if (ctxt->usesGzip == 1) {
1202 if (ctxt->strm == NULL) return(0);
1203
1204 ctxt->strm->next_out = dest;
1205 ctxt->strm->avail_out = len;
William M. Bracke8827652007-05-16 05:19:13 +00001206 ctxt->strm->avail_in = ctxt->inptr - ctxt->inrptr;
Daniel Veillard9a2724d2005-12-15 11:12:26 +00001207
William M. Bracke8827652007-05-16 05:19:13 +00001208 while (ctxt->strm->avail_out > 0 &&
1209 (ctxt->strm->avail_in > 0 || xmlNanoHTTPRecv(ctxt) > 0)) {
William M. Brackd2f682a2007-05-15 19:42:08 +00001210 orig_avail_in = ctxt->strm->avail_in =
1211 ctxt->inptr - ctxt->inrptr - bytes_read;
Daniel Veillard9a2724d2005-12-15 11:12:26 +00001212 ctxt->strm->next_in = BAD_CAST (ctxt->inrptr + bytes_read);
1213
1214 z_ret = inflate(ctxt->strm, Z_NO_FLUSH);
1215 bytes_read += orig_avail_in - ctxt->strm->avail_in;
1216
1217 if (z_ret != Z_OK) break;
William M. Brackd2f682a2007-05-15 19:42:08 +00001218 }
Daniel Veillard9a2724d2005-12-15 11:12:26 +00001219
1220 ctxt->inrptr += bytes_read;
1221 return(len - ctxt->strm->avail_out);
1222 }
1223#endif
1224
Owen Taylor3473f882001-02-23 17:55:21 +00001225 while (ctxt->inptr - ctxt->inrptr < len) {
Daniel Veillardf012a642001-07-23 19:10:52 +00001226 if (xmlNanoHTTPRecv(ctxt) <= 0) break;
Owen Taylor3473f882001-02-23 17:55:21 +00001227 }
1228 if (ctxt->inptr - ctxt->inrptr < len)
1229 len = ctxt->inptr - ctxt->inrptr;
1230 memcpy(dest, ctxt->inrptr, len);
1231 ctxt->inrptr += len;
1232 return(len);
1233}
1234
1235/**
1236 * xmlNanoHTTPClose:
1237 * @ctx: the HTTP context
1238 *
1239 * This function closes an HTTP context, it ends up the connection and
1240 * free all data related to it.
1241 */
1242void
1243xmlNanoHTTPClose(void *ctx) {
1244 xmlNanoHTTPCtxtPtr ctxt = (xmlNanoHTTPCtxtPtr) ctx;
1245
1246 if (ctx == NULL) return;
1247
1248 xmlNanoHTTPFreeCtxt(ctxt);
1249}
1250
1251/**
Daniel Veillard9403a042001-05-28 11:00:53 +00001252 * xmlNanoHTTPMethodRedir:
Owen Taylor3473f882001-02-23 17:55:21 +00001253 * @URL: The URL to load
1254 * @method: the HTTP method to use
1255 * @input: the input string if any
1256 * @contentType: the Content-Type information IN and OUT
Daniel Veillard9403a042001-05-28 11:00:53 +00001257 * @redir: the redirected URL OUT
Owen Taylor3473f882001-02-23 17:55:21 +00001258 * @headers: the extra headers
Daniel Veillardcbaf3992001-12-31 16:16:02 +00001259 * @ilen: input length
Owen Taylor3473f882001-02-23 17:55:21 +00001260 *
1261 * This function try to open a connection to the indicated resource
1262 * via HTTP using the given @method, adding the given extra headers
1263 * and the input buffer for the request content.
1264 *
1265 * Returns NULL in case of failure, otherwise a request handler.
Daniel Veillard9403a042001-05-28 11:00:53 +00001266 * The contentType, or redir, if provided must be freed by the caller
Owen Taylor3473f882001-02-23 17:55:21 +00001267 */
1268
1269void*
Daniel Veillard9403a042001-05-28 11:00:53 +00001270xmlNanoHTTPMethodRedir(const char *URL, const char *method, const char *input,
Daniel Veillardf012a642001-07-23 19:10:52 +00001271 char **contentType, char **redir,
1272 const char *headers, int ilen ) {
Owen Taylor3473f882001-02-23 17:55:21 +00001273 xmlNanoHTTPCtxtPtr ctxt;
1274 char *bp, *p;
Daniel Veillardf012a642001-07-23 19:10:52 +00001275 int blen, ret;
Owen Taylor3473f882001-02-23 17:55:21 +00001276 int head;
1277 int nbRedirects = 0;
1278 char *redirURL = NULL;
William M. Brack78637da2003-07-31 14:47:38 +00001279#ifdef DEBUG_HTTP
1280 int xmt_bytes;
1281#endif
Owen Taylor3473f882001-02-23 17:55:21 +00001282
1283 if (URL == NULL) return(NULL);
1284 if (method == NULL) method = "GET";
1285 xmlNanoHTTPInit();
1286
1287retry:
1288 if (redirURL == NULL)
1289 ctxt = xmlNanoHTTPNewCtxt(URL);
1290 else {
1291 ctxt = xmlNanoHTTPNewCtxt(redirURL);
Daniel Veillarda840b692003-10-19 13:35:37 +00001292 ctxt->location = xmlMemStrdup(redirURL);
Owen Taylor3473f882001-02-23 17:55:21 +00001293 }
1294
Daniel Veillardf012a642001-07-23 19:10:52 +00001295 if ( ctxt == NULL ) {
Daniel Veillardf012a642001-07-23 19:10:52 +00001296 return ( NULL );
1297 }
1298
Owen Taylor3473f882001-02-23 17:55:21 +00001299 if ((ctxt->protocol == NULL) || (strcmp(ctxt->protocol, "http"))) {
Daniel Veillard2b0f8792003-10-10 19:36:36 +00001300 __xmlIOErr(XML_FROM_HTTP, XML_HTTP_URL_SYNTAX, "Not a valid HTTP URI");
Owen Taylor3473f882001-02-23 17:55:21 +00001301 xmlNanoHTTPFreeCtxt(ctxt);
1302 if (redirURL != NULL) xmlFree(redirURL);
1303 return(NULL);
1304 }
1305 if (ctxt->hostname == NULL) {
Daniel Veillard2b0f8792003-10-10 19:36:36 +00001306 __xmlIOErr(XML_FROM_HTTP, XML_HTTP_UNKNOWN_HOST,
1307 "Failed to identify host in URI");
Owen Taylor3473f882001-02-23 17:55:21 +00001308 xmlNanoHTTPFreeCtxt(ctxt);
Daniel Veillard9403a042001-05-28 11:00:53 +00001309 if (redirURL != NULL) xmlFree(redirURL);
Owen Taylor3473f882001-02-23 17:55:21 +00001310 return(NULL);
1311 }
1312 if (proxy) {
1313 blen = strlen(ctxt->hostname) * 2 + 16;
1314 ret = xmlNanoHTTPConnectHost(proxy, proxyPort);
1315 }
1316 else {
1317 blen = strlen(ctxt->hostname);
1318 ret = xmlNanoHTTPConnectHost(ctxt->hostname, ctxt->port);
1319 }
1320 if (ret < 0) {
1321 xmlNanoHTTPFreeCtxt(ctxt);
Daniel Veillard9403a042001-05-28 11:00:53 +00001322 if (redirURL != NULL) xmlFree(redirURL);
Owen Taylor3473f882001-02-23 17:55:21 +00001323 return(NULL);
1324 }
1325 ctxt->fd = ret;
1326
Daniel Veillardf012a642001-07-23 19:10:52 +00001327 if (input == NULL)
Owen Taylor3473f882001-02-23 17:55:21 +00001328 ilen = 0;
Daniel Veillardf012a642001-07-23 19:10:52 +00001329 else
1330 blen += 36;
1331
Owen Taylor3473f882001-02-23 17:55:21 +00001332 if (headers != NULL)
Daniel Veillardf012a642001-07-23 19:10:52 +00001333 blen += strlen(headers) + 2;
Owen Taylor3473f882001-02-23 17:55:21 +00001334 if (contentType && *contentType)
1335 blen += strlen(*contentType) + 16;
Daniel Veillard351f2d62005-04-13 02:55:12 +00001336 if (ctxt->query != NULL)
1337 blen += strlen(ctxt->query) + 1;
Daniel Veillardf012a642001-07-23 19:10:52 +00001338 blen += strlen(method) + strlen(ctxt->path) + 24;
Daniel Veillard9a2724d2005-12-15 11:12:26 +00001339#ifdef HAVE_ZLIB_H
1340 blen += 23;
1341#endif
Daniel Veillard82cb3192003-10-29 13:39:15 +00001342 bp = (char*)xmlMallocAtomic(blen);
Daniel Veillardf012a642001-07-23 19:10:52 +00001343 if ( bp == NULL ) {
1344 xmlNanoHTTPFreeCtxt( ctxt );
Daniel Veillard2b0f8792003-10-10 19:36:36 +00001345 xmlHTTPErrMemory("allocating header buffer");
Daniel Veillardf012a642001-07-23 19:10:52 +00001346 return ( NULL );
1347 }
1348
1349 p = bp;
1350
Owen Taylor3473f882001-02-23 17:55:21 +00001351 if (proxy) {
1352 if (ctxt->port != 80) {
Aleksey Sanin49cc9752002-06-14 17:07:10 +00001353 p += snprintf( p, blen - (p - bp), "%s http://%s:%d%s",
1354 method, ctxt->hostname,
Daniel Veillardf012a642001-07-23 19:10:52 +00001355 ctxt->port, ctxt->path );
Owen Taylor3473f882001-02-23 17:55:21 +00001356 }
Aleksey Sanin49cc9752002-06-14 17:07:10 +00001357 else
1358 p += snprintf( p, blen - (p - bp), "%s http://%s%s", method,
Daniel Veillardf012a642001-07-23 19:10:52 +00001359 ctxt->hostname, ctxt->path);
Owen Taylor3473f882001-02-23 17:55:21 +00001360 }
1361 else
Aleksey Sanin49cc9752002-06-14 17:07:10 +00001362 p += snprintf( p, blen - (p - bp), "%s %s", method, ctxt->path);
Daniel Veillardf012a642001-07-23 19:10:52 +00001363
Daniel Veillard351f2d62005-04-13 02:55:12 +00001364 if (ctxt->query != NULL)
1365 p += snprintf( p, blen - (p - bp), "?%s", ctxt->query);
1366
Aleksey Sanin49cc9752002-06-14 17:07:10 +00001367 p += snprintf( p, blen - (p - bp), " HTTP/1.0\r\nHost: %s\r\n",
1368 ctxt->hostname);
Daniel Veillardf012a642001-07-23 19:10:52 +00001369
Daniel Veillard9a2724d2005-12-15 11:12:26 +00001370#ifdef HAVE_ZLIB_H
1371 p += snprintf(p, blen - (p - bp), "Accept-Encoding: gzip\r\n");
1372#endif
1373
Daniel Veillardf012a642001-07-23 19:10:52 +00001374 if (contentType != NULL && *contentType)
Aleksey Sanin49cc9752002-06-14 17:07:10 +00001375 p += snprintf(p, blen - (p - bp), "Content-Type: %s\r\n", *contentType);
Daniel Veillardf012a642001-07-23 19:10:52 +00001376
1377 if (headers != NULL)
Aleksey Sanin49cc9752002-06-14 17:07:10 +00001378 p += snprintf( p, blen - (p - bp), "%s", headers );
Daniel Veillardf012a642001-07-23 19:10:52 +00001379
Owen Taylor3473f882001-02-23 17:55:21 +00001380 if (input != NULL)
Aleksey Sanin49cc9752002-06-14 17:07:10 +00001381 snprintf(p, blen - (p - bp), "Content-Length: %d\r\n\r\n", ilen );
Owen Taylor3473f882001-02-23 17:55:21 +00001382 else
Aleksey Sanin49cc9752002-06-14 17:07:10 +00001383 snprintf(p, blen - (p - bp), "\r\n");
Daniel Veillardf012a642001-07-23 19:10:52 +00001384
Owen Taylor3473f882001-02-23 17:55:21 +00001385#ifdef DEBUG_HTTP
1386 xmlGenericError(xmlGenericErrorContext,
1387 "-> %s%s", proxy? "(Proxy) " : "", bp);
1388 if ((blen -= strlen(bp)+1) < 0)
1389 xmlGenericError(xmlGenericErrorContext,
1390 "ERROR: overflowed buffer by %d bytes\n", -blen);
1391#endif
1392 ctxt->outptr = ctxt->out = bp;
1393 ctxt->state = XML_NANO_HTTP_WRITE;
Daniel Veillardf012a642001-07-23 19:10:52 +00001394 blen = strlen( ctxt->out );
Daniel Veillardf012a642001-07-23 19:10:52 +00001395#ifdef DEBUG_HTTP
William M. Brack78637da2003-07-31 14:47:38 +00001396 xmt_bytes = xmlNanoHTTPSend(ctxt, ctxt->out, blen );
Daniel Veillardf012a642001-07-23 19:10:52 +00001397 if ( xmt_bytes != blen )
1398 xmlGenericError( xmlGenericErrorContext,
1399 "xmlNanoHTTPMethodRedir: Only %d of %d %s %s\n",
1400 xmt_bytes, blen,
1401 "bytes of HTTP headers sent to host",
1402 ctxt->hostname );
William M. Brack78637da2003-07-31 14:47:38 +00001403#else
1404 xmlNanoHTTPSend(ctxt, ctxt->out, blen );
Daniel Veillardf012a642001-07-23 19:10:52 +00001405#endif
1406
1407 if ( input != NULL ) {
William M. Brack78637da2003-07-31 14:47:38 +00001408#ifdef DEBUG_HTTP
Daniel Veillardf012a642001-07-23 19:10:52 +00001409 xmt_bytes = xmlNanoHTTPSend( ctxt, input, ilen );
1410
Daniel Veillardf012a642001-07-23 19:10:52 +00001411 if ( xmt_bytes != ilen )
1412 xmlGenericError( xmlGenericErrorContext,
1413 "xmlNanoHTTPMethodRedir: Only %d of %d %s %s\n",
1414 xmt_bytes, ilen,
1415 "bytes of HTTP content sent to host",
1416 ctxt->hostname );
William M. Brack78637da2003-07-31 14:47:38 +00001417#else
1418 xmlNanoHTTPSend( ctxt, input, ilen );
Daniel Veillardf012a642001-07-23 19:10:52 +00001419#endif
1420 }
1421
Owen Taylor3473f882001-02-23 17:55:21 +00001422 ctxt->state = XML_NANO_HTTP_READ;
1423 head = 1;
1424
1425 while ((p = xmlNanoHTTPReadLine(ctxt)) != NULL) {
1426 if (head && (*p == 0)) {
1427 head = 0;
1428 ctxt->content = ctxt->inrptr;
1429 xmlFree(p);
1430 break;
1431 }
1432 xmlNanoHTTPScanAnswer(ctxt, p);
1433
1434#ifdef DEBUG_HTTP
1435 xmlGenericError(xmlGenericErrorContext, "<- %s\n", p);
1436#endif
1437 xmlFree(p);
1438 }
1439
1440 if ((ctxt->location != NULL) && (ctxt->returnValue >= 300) &&
1441 (ctxt->returnValue < 400)) {
1442#ifdef DEBUG_HTTP
1443 xmlGenericError(xmlGenericErrorContext,
1444 "\nRedirect to: %s\n", ctxt->location);
1445#endif
Daniel Veillardf012a642001-07-23 19:10:52 +00001446 while ( xmlNanoHTTPRecv(ctxt) > 0 ) ;
Owen Taylor3473f882001-02-23 17:55:21 +00001447 if (nbRedirects < XML_NANO_HTTP_MAX_REDIR) {
1448 nbRedirects++;
Daniel Veillard9403a042001-05-28 11:00:53 +00001449 if (redirURL != NULL)
1450 xmlFree(redirURL);
Owen Taylor3473f882001-02-23 17:55:21 +00001451 redirURL = xmlMemStrdup(ctxt->location);
1452 xmlNanoHTTPFreeCtxt(ctxt);
1453 goto retry;
1454 }
1455 xmlNanoHTTPFreeCtxt(ctxt);
Daniel Veillard9403a042001-05-28 11:00:53 +00001456 if (redirURL != NULL) xmlFree(redirURL);
Owen Taylor3473f882001-02-23 17:55:21 +00001457#ifdef DEBUG_HTTP
1458 xmlGenericError(xmlGenericErrorContext,
Daniel Veillardf012a642001-07-23 19:10:52 +00001459 "xmlNanoHTTPMethodRedir: Too many redirects, aborting ...\n");
Owen Taylor3473f882001-02-23 17:55:21 +00001460#endif
1461 return(NULL);
Owen Taylor3473f882001-02-23 17:55:21 +00001462 }
1463
1464 if (contentType != NULL) {
1465 if (ctxt->contentType != NULL)
1466 *contentType = xmlMemStrdup(ctxt->contentType);
1467 else
1468 *contentType = NULL;
1469 }
1470
Daniel Veillard9403a042001-05-28 11:00:53 +00001471 if ((redir != NULL) && (redirURL != NULL)) {
1472 *redir = redirURL;
1473 } else {
1474 if (redirURL != NULL)
1475 xmlFree(redirURL);
1476 if (redir != NULL)
1477 *redir = NULL;
1478 }
1479
Owen Taylor3473f882001-02-23 17:55:21 +00001480#ifdef DEBUG_HTTP
1481 if (ctxt->contentType != NULL)
1482 xmlGenericError(xmlGenericErrorContext,
1483 "\nCode %d, content-type '%s'\n\n",
1484 ctxt->returnValue, ctxt->contentType);
1485 else
1486 xmlGenericError(xmlGenericErrorContext,
1487 "\nCode %d, no content-type\n\n",
1488 ctxt->returnValue);
1489#endif
1490
1491 return((void *) ctxt);
1492}
1493
1494/**
Daniel Veillard9403a042001-05-28 11:00:53 +00001495 * xmlNanoHTTPMethod:
1496 * @URL: The URL to load
1497 * @method: the HTTP method to use
1498 * @input: the input string if any
1499 * @contentType: the Content-Type information IN and OUT
1500 * @headers: the extra headers
Daniel Veillardcbaf3992001-12-31 16:16:02 +00001501 * @ilen: input length
Daniel Veillard9403a042001-05-28 11:00:53 +00001502 *
1503 * This function try to open a connection to the indicated resource
1504 * via HTTP using the given @method, adding the given extra headers
1505 * and the input buffer for the request content.
1506 *
1507 * Returns NULL in case of failure, otherwise a request handler.
1508 * The contentType, if provided must be freed by the caller
1509 */
1510
1511void*
1512xmlNanoHTTPMethod(const char *URL, const char *method, const char *input,
Daniel Veillardf012a642001-07-23 19:10:52 +00001513 char **contentType, const char *headers, int ilen) {
Daniel Veillard9403a042001-05-28 11:00:53 +00001514 return(xmlNanoHTTPMethodRedir(URL, method, input, contentType,
Daniel Veillardf012a642001-07-23 19:10:52 +00001515 NULL, headers, ilen));
Daniel Veillard9403a042001-05-28 11:00:53 +00001516}
1517
1518/**
Owen Taylor3473f882001-02-23 17:55:21 +00001519 * xmlNanoHTTPFetch:
1520 * @URL: The URL to load
1521 * @filename: the filename where the content should be saved
1522 * @contentType: if available the Content-Type information will be
1523 * returned at that location
1524 *
1525 * This function try to fetch the indicated resource via HTTP GET
1526 * and save it's content in the file.
1527 *
1528 * Returns -1 in case of failure, 0 incase of success. The contentType,
1529 * if provided must be freed by the caller
1530 */
1531int
1532xmlNanoHTTPFetch(const char *URL, const char *filename, char **contentType) {
Daniel Veillardf012a642001-07-23 19:10:52 +00001533 void *ctxt = NULL;
1534 char *buf = NULL;
Owen Taylor3473f882001-02-23 17:55:21 +00001535 int fd;
1536 int len;
1537
William M. Brack015ccb22005-02-13 08:18:52 +00001538 if (filename == NULL) return(-1);
Owen Taylor3473f882001-02-23 17:55:21 +00001539 ctxt = xmlNanoHTTPOpen(URL, contentType);
1540 if (ctxt == NULL) return(-1);
1541
1542 if (!strcmp(filename, "-"))
1543 fd = 0;
1544 else {
1545 fd = open(filename, O_CREAT | O_WRONLY, 00644);
1546 if (fd < 0) {
1547 xmlNanoHTTPClose(ctxt);
1548 if ((contentType != NULL) && (*contentType != NULL)) {
1549 xmlFree(*contentType);
1550 *contentType = NULL;
1551 }
1552 return(-1);
1553 }
1554 }
1555
Daniel Veillardf012a642001-07-23 19:10:52 +00001556 xmlNanoHTTPFetchContent( ctxt, &buf, &len );
1557 if ( len > 0 ) {
Owen Taylor3473f882001-02-23 17:55:21 +00001558 write(fd, buf, len);
1559 }
1560
1561 xmlNanoHTTPClose(ctxt);
1562 close(fd);
1563 return(0);
1564}
1565
Daniel Veillarda9cce9c2003-09-29 13:20:24 +00001566#ifdef LIBXML_OUTPUT_ENABLED
Owen Taylor3473f882001-02-23 17:55:21 +00001567/**
1568 * xmlNanoHTTPSave:
1569 * @ctxt: the HTTP context
1570 * @filename: the filename where the content should be saved
1571 *
1572 * This function saves the output of the HTTP transaction to a file
1573 * It closes and free the context at the end
1574 *
1575 * Returns -1 in case of failure, 0 incase of success.
1576 */
1577int
1578xmlNanoHTTPSave(void *ctxt, const char *filename) {
Daniel Veillarde3924972001-07-25 20:25:21 +00001579 char *buf = NULL;
Owen Taylor3473f882001-02-23 17:55:21 +00001580 int fd;
1581 int len;
1582
William M. Brack015ccb22005-02-13 08:18:52 +00001583 if ((ctxt == NULL) || (filename == NULL)) return(-1);
Owen Taylor3473f882001-02-23 17:55:21 +00001584
1585 if (!strcmp(filename, "-"))
1586 fd = 0;
1587 else {
1588 fd = open(filename, O_CREAT | O_WRONLY);
1589 if (fd < 0) {
1590 xmlNanoHTTPClose(ctxt);
1591 return(-1);
1592 }
1593 }
1594
Daniel Veillardf012a642001-07-23 19:10:52 +00001595 xmlNanoHTTPFetchContent( ctxt, &buf, &len );
1596 if ( len > 0 ) {
Owen Taylor3473f882001-02-23 17:55:21 +00001597 write(fd, buf, len);
1598 }
1599
1600 xmlNanoHTTPClose(ctxt);
William M. Brack20d82362004-03-17 08:44:46 +00001601 close(fd);
Owen Taylor3473f882001-02-23 17:55:21 +00001602 return(0);
1603}
Daniel Veillarda9cce9c2003-09-29 13:20:24 +00001604#endif /* LIBXML_OUTPUT_ENABLED */
Owen Taylor3473f882001-02-23 17:55:21 +00001605
1606/**
1607 * xmlNanoHTTPReturnCode:
1608 * @ctx: the HTTP context
1609 *
Daniel Veillard5e2dace2001-07-18 19:30:27 +00001610 * Get the latest HTTP return code received
1611 *
Owen Taylor3473f882001-02-23 17:55:21 +00001612 * Returns the HTTP return code for the request.
1613 */
1614int
1615xmlNanoHTTPReturnCode(void *ctx) {
1616 xmlNanoHTTPCtxtPtr ctxt = (xmlNanoHTTPCtxtPtr) ctx;
1617
1618 if (ctxt == NULL) return(-1);
1619
1620 return(ctxt->returnValue);
1621}
1622
1623/**
1624 * xmlNanoHTTPAuthHeader:
1625 * @ctx: the HTTP context
1626 *
Daniel Veillard5e2dace2001-07-18 19:30:27 +00001627 * Get the authentication header of an HTTP context
1628 *
Owen Taylor3473f882001-02-23 17:55:21 +00001629 * Returns the stashed value of the WWW-Authenticate or Proxy-Authenticate
1630 * header.
1631 */
1632const char *
1633xmlNanoHTTPAuthHeader(void *ctx) {
1634 xmlNanoHTTPCtxtPtr ctxt = (xmlNanoHTTPCtxtPtr) ctx;
1635
1636 if (ctxt == NULL) return(NULL);
1637
1638 return(ctxt->authHeader);
1639}
1640
Daniel Veillardf012a642001-07-23 19:10:52 +00001641/**
Daniel Veillard01c13b52002-12-10 15:19:08 +00001642 * xmlNanoHTTPContentLength:
Daniel Veillardf012a642001-07-23 19:10:52 +00001643 * @ctx: the HTTP context
1644 *
Daniel Veillarda9b66d02002-12-11 14:23:49 +00001645 * Provides the specified content length from the HTTP header.
1646 *
Daniel Veillardf012a642001-07-23 19:10:52 +00001647 * Return the specified content length from the HTTP header. Note that
1648 * a value of -1 indicates that the content length element was not included in
1649 * the response header.
1650 */
1651int
1652xmlNanoHTTPContentLength( void * ctx ) {
Daniel Veillard82cb3192003-10-29 13:39:15 +00001653 xmlNanoHTTPCtxtPtr ctxt = (xmlNanoHTTPCtxtPtr)ctx;
Daniel Veillardf012a642001-07-23 19:10:52 +00001654
1655 return ( ( ctxt == NULL ) ? -1 : ctxt->ContentLength );
1656}
1657
1658/**
Daniel Veillard847332a2003-10-18 11:29:40 +00001659 * xmlNanoHTTPRedir:
1660 * @ctx: the HTTP context
1661 *
1662 * Provides the specified redirection URL if available from the HTTP header.
1663 *
1664 * Return the specified redirection URL or NULL if not redirected.
1665 */
1666const char *
1667xmlNanoHTTPRedir( void * ctx ) {
Daniel Veillard82cb3192003-10-29 13:39:15 +00001668 xmlNanoHTTPCtxtPtr ctxt = (xmlNanoHTTPCtxtPtr)ctx;
Daniel Veillard847332a2003-10-18 11:29:40 +00001669
1670 return ( ( ctxt == NULL ) ? NULL : ctxt->location );
1671}
1672
1673/**
1674 * xmlNanoHTTPEncoding:
1675 * @ctx: the HTTP context
1676 *
1677 * Provides the specified encoding if specified in the HTTP headers.
1678 *
1679 * Return the specified encoding or NULL if not available
1680 */
1681const char *
1682xmlNanoHTTPEncoding( void * ctx ) {
Daniel Veillard82cb3192003-10-29 13:39:15 +00001683 xmlNanoHTTPCtxtPtr ctxt = (xmlNanoHTTPCtxtPtr)ctx;
Daniel Veillard847332a2003-10-18 11:29:40 +00001684
1685 return ( ( ctxt == NULL ) ? NULL : ctxt->encoding );
1686}
1687
1688/**
Daniel Veillarda840b692003-10-19 13:35:37 +00001689 * xmlNanoHTTPMimeType:
1690 * @ctx: the HTTP context
1691 *
1692 * Provides the specified Mime-Type if specified in the HTTP headers.
1693 *
1694 * Return the specified Mime-Type or NULL if not available
1695 */
1696const char *
1697xmlNanoHTTPMimeType( void * ctx ) {
Daniel Veillard82cb3192003-10-29 13:39:15 +00001698 xmlNanoHTTPCtxtPtr ctxt = (xmlNanoHTTPCtxtPtr)ctx;
Daniel Veillarda840b692003-10-19 13:35:37 +00001699
1700 return ( ( ctxt == NULL ) ? NULL : ctxt->mimeType );
1701}
1702
1703/**
Daniel Veillard01c13b52002-12-10 15:19:08 +00001704 * xmlNanoHTTPFetchContent:
Daniel Veillardf012a642001-07-23 19:10:52 +00001705 * @ctx: the HTTP context
1706 * @ptr: pointer to set to the content buffer.
1707 * @len: integer pointer to hold the length of the content
1708 *
Daniel Veillarda9b66d02002-12-11 14:23:49 +00001709 * Check if all the content was read
1710 *
Daniel Veillardf012a642001-07-23 19:10:52 +00001711 * Returns 0 if all the content was read and available, returns
1712 * -1 if received content length was less than specified or an error
1713 * occurred.
1714 */
Daniel Veillarda2351322004-06-27 12:08:10 +00001715static int
Daniel Veillardf012a642001-07-23 19:10:52 +00001716xmlNanoHTTPFetchContent( void * ctx, char ** ptr, int * len ) {
Daniel Veillard82cb3192003-10-29 13:39:15 +00001717 xmlNanoHTTPCtxtPtr ctxt = (xmlNanoHTTPCtxtPtr)ctx;
Daniel Veillardf012a642001-07-23 19:10:52 +00001718
1719 int rc = 0;
1720 int cur_lgth;
1721 int rcvd_lgth;
1722 int dummy_int;
1723 char * dummy_ptr = NULL;
1724
1725 /* Dummy up return input parameters if not provided */
1726
1727 if ( len == NULL )
1728 len = &dummy_int;
1729
1730 if ( ptr == NULL )
1731 ptr = &dummy_ptr;
1732
1733 /* But can't work without the context pointer */
1734
1735 if ( ( ctxt == NULL ) || ( ctxt->content == NULL ) ) {
1736 *len = 0;
1737 *ptr = NULL;
1738 return ( -1 );
1739 }
1740
1741 rcvd_lgth = ctxt->inptr - ctxt->content;
1742
1743 while ( (cur_lgth = xmlNanoHTTPRecv( ctxt )) > 0 ) {
1744
1745 rcvd_lgth += cur_lgth;
1746 if ( (ctxt->ContentLength > 0) && (rcvd_lgth >= ctxt->ContentLength) )
1747 break;
1748 }
1749
1750 *ptr = ctxt->content;
1751 *len = rcvd_lgth;
1752
1753 if ( ( ctxt->ContentLength > 0 ) && ( rcvd_lgth < ctxt->ContentLength ) )
1754 rc = -1;
1755 else if ( rcvd_lgth == 0 )
1756 rc = -1;
1757
1758 return ( rc );
1759}
1760
Owen Taylor3473f882001-02-23 17:55:21 +00001761#ifdef STANDALONE
1762int main(int argc, char **argv) {
1763 char *contentType = NULL;
1764
1765 if (argv[1] != NULL) {
1766 if (argv[2] != NULL)
1767 xmlNanoHTTPFetch(argv[1], argv[2], &contentType);
1768 else
1769 xmlNanoHTTPFetch(argv[1], "-", &contentType);
1770 if (contentType != NULL) xmlFree(contentType);
1771 } else {
1772 xmlGenericError(xmlGenericErrorContext,
1773 "%s: minimal HTTP GET implementation\n", argv[0]);
1774 xmlGenericError(xmlGenericErrorContext,
1775 "\tusage %s [ URL [ filename ] ]\n", argv[0]);
1776 }
1777 xmlNanoHTTPCleanup();
1778 xmlMemoryDump();
1779 return(0);
1780}
1781#endif /* STANDALONE */
1782#else /* !LIBXML_HTTP_ENABLED */
1783#ifdef STANDALONE
1784#include <stdio.h>
1785int main(int argc, char **argv) {
1786 xmlGenericError(xmlGenericErrorContext,
1787 "%s : HTTP support not compiled in\n", argv[0]);
1788 return(0);
1789}
1790#endif /* STANDALONE */
1791#endif /* LIBXML_HTTP_ENABLED */
Daniel Veillard5d4644e2005-04-01 13:11:58 +00001792#define bottom_nanohttp
1793#include "elfgcchack.h"