Annotation of XML/nanohttp.c, revision 1.33
1.1 daniel 1: /*
1.5 daniel 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/
1.1 daniel 8: *
9: * See Copyright for the status of this software.
10: *
11: * Daniel.Veillard@w3.org
12: */
13:
1.5 daniel 14: /* TODO add compression support, Send the Accept- , and decompress on the
15: fly with ZLIB if found at compile-time */
16:
1.9 daniel 17: #ifdef WIN32
1.11 daniel 18: #define INCLUDE_WINSOCK
1.9 daniel 19: #include "win32config.h"
20: #else
1.4 daniel 21: #include "config.h"
22: #endif
1.9 daniel 23:
1.25 veillard 24: #include <libxml/xmlversion.h>
1.4 daniel 25:
1.17 daniel 26: #ifdef LIBXML_HTTP_ENABLED
1.1 daniel 27: #include <stdio.h>
28: #include <string.h>
1.4 daniel 29:
30: #ifdef HAVE_STDLIB_H
1.1 daniel 31: #include <stdlib.h>
1.4 daniel 32: #endif
33: #ifdef HAVE_UNISTD_H
1.1 daniel 34: #include <unistd.h>
1.4 daniel 35: #endif
36: #ifdef HAVE_SYS_SOCKET_H
1.1 daniel 37: #include <sys/socket.h>
1.4 daniel 38: #endif
39: #ifdef HAVE_NETINET_IN_H
1.1 daniel 40: #include <netinet/in.h>
1.4 daniel 41: #endif
42: #ifdef HAVE_ARPA_INET_H
1.1 daniel 43: #include <arpa/inet.h>
1.4 daniel 44: #endif
45: #ifdef HAVE_NETDB_H
1.1 daniel 46: #include <netdb.h>
1.4 daniel 47: #endif
48: #ifdef HAVE_FCNTL_H
1.1 daniel 49: #include <fcntl.h>
1.4 daniel 50: #endif
51: #ifdef HAVE_ERRNO_H
1.1 daniel 52: #include <errno.h>
1.4 daniel 53: #endif
54: #ifdef HAVE_SYS_TIME_H
1.1 daniel 55: #include <sys/time.h>
1.4 daniel 56: #endif
57: #ifdef HAVE_SYS_SELECT_H
1.1 daniel 58: #include <sys/select.h>
1.4 daniel 59: #endif
1.15 daniel 60: #ifdef HAVE_STRINGS_H
61: #include <strings.h>
62: #endif
1.1 daniel 63:
1.17 daniel 64: #include <libxml/xmlmemory.h>
1.23 veillard 65: #include <libxml/parser.h> /* for xmlStr(n)casecmp() */
1.17 daniel 66: #include <libxml/nanohttp.h>
1.7 daniel 67:
1.30 veillard 68: /**
69: * A couple portability macros
70: */
71: #ifndef _WINSOCKAPI_
72: #define closesocket(s) close(s)
73: #define SOCKET int
74: #endif
75:
1.5 daniel 76: #ifdef STANDALONE
77: #define DEBUG_HTTP
1.23 veillard 78: #define xmlStrncasecmp(a, b, n) strncasecmp((char *)a, (char *)b, n)
79: #define xmlStrcasecmpi(a, b) strcasecmp((char *)a, (char *)b)
1.5 daniel 80: #endif
81:
1.1 daniel 82: #define XML_NANO_HTTP_MAX_REDIR 10
83:
84: #define XML_NANO_HTTP_CHUNK 4096
85:
86: #define XML_NANO_HTTP_CLOSED 0
87: #define XML_NANO_HTTP_WRITE 1
88: #define XML_NANO_HTTP_READ 2
89: #define XML_NANO_HTTP_NONE 4
90:
91: typedef struct xmlNanoHTTPCtxt {
92: char *protocol; /* the protocol name */
93: char *hostname; /* the host name */
94: int port; /* the port */
95: char *path; /* the path within the URL */
1.30 veillard 96: SOCKET fd; /* the file descriptor for the socket */
1.1 daniel 97: int state; /* WRITE / READ / CLOSED */
98: char *out; /* buffer sent (zero terminated) */
99: char *outptr; /* index within the buffer sent */
100: char *in; /* the receiving buffer */
101: char *content; /* the start of the content */
102: char *inptr; /* the next byte to read from network */
103: char *inrptr; /* the next byte to give back to the client */
104: int inlen; /* len of the input buffer */
105: int last; /* return code for last operation */
106: int returnValue; /* the protocol return value */
107: char *contentType; /* the MIME type for the input */
108: char *location; /* the new URL in case of redirect */
109: } xmlNanoHTTPCtxt, *xmlNanoHTTPCtxtPtr;
110:
1.12 daniel 111: static int initialized = 0;
1.28 veillard 112: static char *proxy = NULL; /* the proxy name if any */
1.12 daniel 113: static int proxyPort; /* the proxy port if any */
1.28 veillard 114: static unsigned int timeout = 60;/* the select() timeout in seconds */
1.12 daniel 115:
116: /**
1.30 veillard 117: * A portability function
1.26 veillard 118: */
119: int socket_errno(void) {
120: #ifdef _WINSOCKAPI_
121: return(WSAGetLastError());
122: #else
123: return(errno);
124: #endif
125: }
126:
127: /**
1.12 daniel 128: * xmlNanoHTTPInit:
129: *
130: * Initialize the HTTP protocol layer.
131: * Currently it just checks for proxy informations
132: */
133:
134: void
135: xmlNanoHTTPInit(void) {
136: const char *env;
1.30 veillard 137: #ifdef _WINSOCKAPI_
138: WSADATA wsaData;
139: #endif
1.12 daniel 140:
141: if (initialized)
142: return;
143:
1.26 veillard 144: #ifdef _WINSOCKAPI_
1.30 veillard 145: if (WSAStartup(MAKEWORD(1, 1), &wsaData) != 0)
146: return;
1.26 veillard 147: #endif
148:
1.13 daniel 149: if (proxy == NULL) {
150: proxyPort = 80;
151: env = getenv("no_proxy");
152: if (env != NULL)
153: goto done;
154: env = getenv("http_proxy");
155: if (env != NULL) {
156: xmlNanoHTTPScanProxy(env);
157: goto done;
158: }
159: env = getenv("HTTP_PROXY");
160: if (env != NULL) {
161: xmlNanoHTTPScanProxy(env);
162: goto done;
163: }
1.12 daniel 164: }
1.13 daniel 165: done:
1.12 daniel 166: initialized = 1;
167: }
168:
169: /**
170: * xmlNanoHTTPClenup:
171: *
172: * Cleanup the HTTP protocol layer.
173: */
174:
175: void
176: xmlNanoHTTPCleanup(void) {
177: if (proxy != NULL)
178: xmlFree(proxy);
1.26 veillard 179: #ifdef _WINSOCKAPI_
1.30 veillard 180: if (initialized)
181: WSACleanup();
1.26 veillard 182: #endif
1.30 veillard 183: initialized = 0;
1.12 daniel 184: return;
185: }
186:
1.5 daniel 187: /**
1.28 veillard 188: * xmlNanoHTTPTimeout:
189: * @delay: the delay in seconds
190: *
191: * Set the HTTP timeout, (default is 60secs). 0 means immediate
192: * return, while -1 infinite.
193: */
194:
195: void
196: xmlNanoHTTPTimeout(int delay) {
197: timeout = (unsigned int) delay;
198: }
199:
200: /**
1.5 daniel 201: * xmlNanoHTTPScanURL:
202: * @ctxt: an HTTP context
203: * @URL: The URL used to initialize the context
204: *
205: * (Re)Initialize an HTTP context by parsing the URL and finding
206: * the protocol host port and path it indicates.
207: */
208:
209: static void
210: xmlNanoHTTPScanURL(xmlNanoHTTPCtxtPtr ctxt, const char *URL) {
1.1 daniel 211: const char *cur = URL;
212: char buf[4096];
213: int index = 0;
214: int port = 0;
215:
216: if (ctxt->protocol != NULL) {
1.7 daniel 217: xmlFree(ctxt->protocol);
1.1 daniel 218: ctxt->protocol = NULL;
219: }
220: if (ctxt->hostname != NULL) {
1.7 daniel 221: xmlFree(ctxt->hostname);
1.1 daniel 222: ctxt->hostname = NULL;
223: }
224: if (ctxt->path != NULL) {
1.7 daniel 225: xmlFree(ctxt->path);
1.1 daniel 226: ctxt->path = NULL;
227: }
1.12 daniel 228: if (URL == NULL) return;
1.1 daniel 229: buf[index] = 0;
230: while (*cur != 0) {
231: if ((cur[0] == ':') && (cur[1] == '/') && (cur[2] == '/')) {
232: buf[index] = 0;
1.7 daniel 233: ctxt->protocol = xmlMemStrdup(buf);
1.1 daniel 234: index = 0;
235: cur += 3;
236: break;
237: }
238: buf[index++] = *cur++;
239: }
240: if (*cur == 0) return;
241:
242: buf[index] = 0;
243: while (1) {
244: if (cur[0] == ':') {
245: buf[index] = 0;
1.7 daniel 246: ctxt->hostname = xmlMemStrdup(buf);
1.1 daniel 247: index = 0;
248: cur += 1;
249: while ((*cur >= '0') && (*cur <= '9')) {
250: port *= 10;
251: port += *cur - '0';
252: cur++;
253: }
254: if (port != 0) ctxt->port = port;
255: while ((cur[0] != '/') && (*cur != 0))
256: cur++;
257: break;
258: }
259: if ((*cur == '/') || (*cur == 0)) {
260: buf[index] = 0;
1.7 daniel 261: ctxt->hostname = xmlMemStrdup(buf);
1.1 daniel 262: index = 0;
263: break;
264: }
265: buf[index++] = *cur++;
266: }
267: if (*cur == 0)
1.7 daniel 268: ctxt->path = xmlMemStrdup("/");
1.5 daniel 269: else {
1.14 daniel 270: index = 0;
1.5 daniel 271: buf[index] = 0;
1.14 daniel 272: while (*cur != 0)
1.5 daniel 273: buf[index++] = *cur++;
274: buf[index] = 0;
1.7 daniel 275: ctxt->path = xmlMemStrdup(buf);
1.5 daniel 276: }
1.1 daniel 277: }
278:
1.5 daniel 279: /**
1.12 daniel 280: * xmlNanoHTTPScanProxy:
281: * @URL: The proxy URL used to initialize the proxy context
282: *
283: * (Re)Initialize the HTTP Proxy context by parsing the URL and finding
284: * the protocol host port it indicates.
285: * Should be like http://myproxy/ or http://myproxy:3128/
286: * A NULL URL cleans up proxy informations.
287: */
288:
289: void
290: xmlNanoHTTPScanProxy(const char *URL) {
291: const char *cur = URL;
292: char buf[4096];
293: int index = 0;
294: int port = 0;
295:
296: if (proxy != NULL) {
297: xmlFree(proxy);
298: proxy = NULL;
299: }
300: if (proxyPort != 0) {
301: proxyPort = 0;
302: }
303: #ifdef DEBUG_HTTP
304: if (URL == NULL)
305: printf("Removing HTTP proxy info\n");
306: else
307: printf("Using HTTP proxy %s\n", URL);
308: #endif
309: if (URL == NULL) return;
310: buf[index] = 0;
311: while (*cur != 0) {
312: if ((cur[0] == ':') && (cur[1] == '/') && (cur[2] == '/')) {
313: buf[index] = 0;
314: index = 0;
315: cur += 3;
316: break;
317: }
318: buf[index++] = *cur++;
319: }
320: if (*cur == 0) return;
321:
322: buf[index] = 0;
323: while (1) {
324: if (cur[0] == ':') {
325: buf[index] = 0;
326: proxy = xmlMemStrdup(buf);
327: index = 0;
328: cur += 1;
329: while ((*cur >= '0') && (*cur <= '9')) {
330: port *= 10;
331: port += *cur - '0';
332: cur++;
333: }
334: if (port != 0) proxyPort = port;
335: while ((cur[0] != '/') && (*cur != 0))
336: cur++;
337: break;
338: }
339: if ((*cur == '/') || (*cur == 0)) {
340: buf[index] = 0;
341: proxy = xmlMemStrdup(buf);
342: index = 0;
343: break;
344: }
345: buf[index++] = *cur++;
346: }
347: }
348:
349: /**
1.5 daniel 350: * xmlNanoHTTPNewCtxt:
351: * @URL: The URL used to initialize the context
352: *
353: * Allocate and initialize a new HTTP context.
354: *
355: * Returns an HTTP context or NULL in case of error.
356: */
357:
358: static xmlNanoHTTPCtxtPtr
359: xmlNanoHTTPNewCtxt(const char *URL) {
1.1 daniel 360: xmlNanoHTTPCtxtPtr ret;
361:
1.7 daniel 362: ret = (xmlNanoHTTPCtxtPtr) xmlMalloc(sizeof(xmlNanoHTTPCtxt));
1.1 daniel 363: if (ret == NULL) return(NULL);
364:
365: memset(ret, 0, sizeof(xmlNanoHTTPCtxt));
366: ret->port = 80;
367: ret->returnValue = 0;
368:
369: xmlNanoHTTPScanURL(ret, URL);
370:
371: return(ret);
372: }
373:
1.5 daniel 374: /**
375: * xmlNanoHTTPFreeCtxt:
376: * @ctxt: an HTTP context
377: *
378: * Frees the context after closing the connection.
379: */
380:
381: static void
382: xmlNanoHTTPFreeCtxt(xmlNanoHTTPCtxtPtr ctxt) {
383: if (ctxt == NULL) return;
1.7 daniel 384: if (ctxt->hostname != NULL) xmlFree(ctxt->hostname);
385: if (ctxt->protocol != NULL) xmlFree(ctxt->protocol);
386: if (ctxt->path != NULL) xmlFree(ctxt->path);
387: if (ctxt->out != NULL) xmlFree(ctxt->out);
388: if (ctxt->in != NULL) xmlFree(ctxt->in);
389: if (ctxt->contentType != NULL) xmlFree(ctxt->contentType);
390: if (ctxt->location != NULL) xmlFree(ctxt->location);
1.1 daniel 391: ctxt->state = XML_NANO_HTTP_NONE;
1.26 veillard 392: if (ctxt->fd >= 0) closesocket(ctxt->fd);
1.1 daniel 393: ctxt->fd = -1;
1.7 daniel 394: xmlFree(ctxt);
1.1 daniel 395: }
396:
1.5 daniel 397: /**
398: * xmlNanoHTTPSend:
399: * @ctxt: an HTTP context
400: *
401: * Send the input needed to initiate the processing on the server side
402: */
403:
404: static void
405: xmlNanoHTTPSend(xmlNanoHTTPCtxtPtr ctxt) {
1.31 veillard 406: if (ctxt->state & XML_NANO_HTTP_WRITE) {
407: int total_sent = 0;
408: while (total_sent <strlen(ctxt->outptr)) {
409: int nsent = send(ctxt->fd, ctxt->outptr+total_sent,
410: strlen(ctxt->outptr)-total_sent, 0);
411: if (nsent>0)
412: total_sent += nsent;
413: }
414:
415: ctxt->last = total_sent;
416: }
1.1 daniel 417: }
418:
1.5 daniel 419: /**
420: * xmlNanoHTTPRecv:
421: * @ctxt: an HTTP context
422: *
423: * Read information coming from the HTTP connection.
424: * This is a blocking call (but it blocks in select(), not read()).
425: *
426: * Returns the number of byte read or -1 in case of error.
427: */
428:
429: static int
430: xmlNanoHTTPRecv(xmlNanoHTTPCtxtPtr ctxt) {
1.1 daniel 431: fd_set rfd;
432: struct timeval tv;
433:
434:
435: while (ctxt->state & XML_NANO_HTTP_READ) {
436: if (ctxt->in == NULL) {
1.7 daniel 437: ctxt->in = (char *) xmlMalloc(65000 * sizeof(char));
1.1 daniel 438: if (ctxt->in == NULL) {
439: ctxt->last = -1;
440: return(-1);
441: }
442: ctxt->inlen = 65000;
443: ctxt->inptr = ctxt->content = ctxt->inrptr = ctxt->in;
444: }
445: if (ctxt->inrptr > ctxt->in + XML_NANO_HTTP_CHUNK) {
446: int delta = ctxt->inrptr - ctxt->in;
447: int len = ctxt->inptr - ctxt->inrptr;
448:
449: memmove(ctxt->in, ctxt->inrptr, len);
450: ctxt->inrptr -= delta;
451: ctxt->content -= delta;
452: ctxt->inptr -= delta;
453: }
454: if ((ctxt->in + ctxt->inlen) < (ctxt->inptr + XML_NANO_HTTP_CHUNK)) {
455: int d_inptr = ctxt->inptr - ctxt->in;
456: int d_content = ctxt->content - ctxt->in;
457: int d_inrptr = ctxt->inrptr - ctxt->in;
458:
459: ctxt->inlen *= 2;
1.7 daniel 460: ctxt->in = (char *) xmlRealloc(ctxt->in, ctxt->inlen);
1.1 daniel 461: if (ctxt->in == NULL) {
462: ctxt->last = -1;
463: return(-1);
464: }
465: ctxt->inptr = ctxt->in + d_inptr;
466: ctxt->content = ctxt->in + d_content;
467: ctxt->inrptr = ctxt->in + d_inrptr;
468: }
1.26 veillard 469: ctxt->last = recv(ctxt->fd, ctxt->inptr, XML_NANO_HTTP_CHUNK, 0);
1.1 daniel 470: if (ctxt->last > 0) {
471: ctxt->inptr += ctxt->last;
472: return(ctxt->last);
473: }
474: if (ctxt->last == 0) {
475: return(0);
476: }
1.26 veillard 477: if (ctxt->last == -1) {
478: switch (socket_errno()) {
479: case EINPROGRESS:
480: case EWOULDBLOCK:
481: #if defined(EAGAIN) && EAGAIN != EWOULDBLOCK
482: case EAGAIN:
483: #endif
484: break;
485: default:
486: return(0);
487: }
1.1 daniel 488: }
1.26 veillard 489:
1.28 veillard 490: tv.tv_sec = timeout;
1.26 veillard 491: tv.tv_usec = 0;
1.1 daniel 492: FD_ZERO(&rfd);
493: FD_SET(ctxt->fd, &rfd);
494:
1.26 veillard 495: if (select(ctxt->fd+1, &rfd, NULL, NULL, &tv)<1)
1.5 daniel 496: return(0);
1.1 daniel 497: }
498: return(0);
499: }
500:
1.5 daniel 501: /**
502: * xmlNanoHTTPReadLine:
503: * @ctxt: an HTTP context
504: *
505: * Read one line in the HTTP server output, usually for extracting
506: * the HTTP protocol informations from the answer header.
507: *
508: * Returns a newly allocated string with a copy of the line, or NULL
509: * which indicate the end of the input.
510: */
511:
512: static char *
513: xmlNanoHTTPReadLine(xmlNanoHTTPCtxtPtr ctxt) {
514: char buf[4096];
1.26 veillard 515: char *bp = buf;
1.1 daniel 516:
1.26 veillard 517: while (bp - buf < 4095) {
518: if (ctxt->inrptr == ctxt->inptr) {
1.1 daniel 519: if (xmlNanoHTTPRecv(ctxt) == 0) {
520: if (bp == buf)
1.5 daniel 521: return(NULL);
1.1 daniel 522: else
523: *bp = 0;
1.7 daniel 524: return(xmlMemStrdup(buf));
1.1 daniel 525: }
526: }
527: *bp = *ctxt->inrptr++;
1.26 veillard 528: if (*bp == '\n') {
1.1 daniel 529: *bp = 0;
1.7 daniel 530: return(xmlMemStrdup(buf));
1.1 daniel 531: }
1.26 veillard 532: if (*bp != '\r')
1.1 daniel 533: bp++;
534: }
535: buf[4095] = 0;
1.7 daniel 536: return(xmlMemStrdup(buf));
1.1 daniel 537: }
538:
1.5 daniel 539:
540: /**
541: * xmlNanoHTTPScanAnswer:
542: * @ctxt: an HTTP context
543: * @line: an HTTP header line
544: *
545: * Try to extract useful informations from the server answer.
546: * We currently parse and process:
547: * - The HTTP revision/ return code
548: * - The Content-Type
549: * - The Location for redirrect processing.
550: *
551: * Returns -1 in case of failure, the file descriptor number otherwise
552: */
553:
554: static void
555: xmlNanoHTTPScanAnswer(xmlNanoHTTPCtxtPtr ctxt, const char *line) {
1.1 daniel 556: const char *cur = line;
557:
558: if (line == NULL) return;
559:
560: if (!strncmp(line, "HTTP/", 5)) {
561: int version = 0;
562: int ret = 0;
563:
564: cur += 5;
565: while ((*cur >= '0') && (*cur <= '9')) {
566: version *= 10;
567: version += *cur - '0';
568: cur++;
569: }
570: if (*cur == '.') {
571: cur++;
572: if ((*cur >= '0') && (*cur <= '9')) {
573: version *= 10;
574: version += *cur - '0';
575: cur++;
576: }
577: while ((*cur >= '0') && (*cur <= '9'))
578: cur++;
579: } else
580: version *= 10;
581: if ((*cur != ' ') && (*cur != '\t')) return;
582: while ((*cur == ' ') || (*cur == '\t')) cur++;
583: if ((*cur < '0') || (*cur > '9')) return;
584: while ((*cur >= '0') && (*cur <= '9')) {
585: ret *= 10;
586: ret += *cur - '0';
587: cur++;
588: }
589: if ((*cur != 0) && (*cur != ' ') && (*cur != '\t')) return;
590: ctxt->returnValue = ret;
1.23 veillard 591: } else if (!xmlStrncasecmp(BAD_CAST line, BAD_CAST"Content-Type:", 13)) {
1.1 daniel 592: cur += 13;
593: while ((*cur == ' ') || (*cur == '\t')) cur++;
594: if (ctxt->contentType != NULL)
1.7 daniel 595: xmlFree(ctxt->contentType);
596: ctxt->contentType = xmlMemStrdup(cur);
1.23 veillard 597: } else if (!xmlStrncasecmp(BAD_CAST line, BAD_CAST"ContentType:", 12)) {
1.1 daniel 598: cur += 12;
599: if (ctxt->contentType != NULL) return;
600: while ((*cur == ' ') || (*cur == '\t')) cur++;
1.7 daniel 601: ctxt->contentType = xmlMemStrdup(cur);
1.23 veillard 602: } else if (!xmlStrncasecmp(BAD_CAST line, BAD_CAST"Location:", 9)) {
1.1 daniel 603: cur += 9;
604: while ((*cur == ' ') || (*cur == '\t')) cur++;
605: if (ctxt->location != NULL)
1.7 daniel 606: xmlFree(ctxt->location);
607: ctxt->location = xmlMemStrdup(cur);
1.1 daniel 608: }
609: }
610:
1.5 daniel 611: /**
612: * xmlNanoHTTPConnectAttempt:
613: * @ia: an internet adress structure
614: * @port: the port number
615: *
616: * Attempt a connection to the given IP:port endpoint. It forces
617: * non-blocking semantic on the socket, and allow 60 seconds for
618: * the host to answer.
619: *
620: * Returns -1 in case of failure, the file descriptor number otherwise
621: */
622:
623: static int
624: xmlNanoHTTPConnectAttempt(struct in_addr ia, int port)
1.1 daniel 625: {
1.30 veillard 626: SOCKET s = socket(PF_INET, SOCK_STREAM, IPPROTO_TCP);
1.1 daniel 627: struct sockaddr_in sin;
628: fd_set wfd;
629: struct timeval tv;
1.2 daniel 630: int status;
1.1 daniel 631:
1.26 veillard 632: if (s==-1) {
1.5 daniel 633: #ifdef DEBUG_HTTP
1.1 daniel 634: perror("socket");
1.5 daniel 635: #endif
1.1 daniel 636: return(-1);
637: }
638:
1.2 daniel 639: #ifdef _WINSOCKAPI_
640: {
641: u_long one = 1;
642:
1.3 daniel 643: status = ioctlsocket(s, FIONBIO, &one) == SOCKET_ERROR ? -1 : 0;
1.2 daniel 644: }
645: #else /* _WINSOCKAPI_ */
646: #if defined(VMS)
647: {
648: int enable = 1;
1.3 daniel 649: status = IOCTL(s, FIONBIO, &enable);
1.2 daniel 650: }
651: #else /* VMS */
1.26 veillard 652: if ((status = fcntl(s, F_GETFL, 0)) != -1) {
1.2 daniel 653: #ifdef O_NONBLOCK
654: status |= O_NONBLOCK;
655: #else /* O_NONBLOCK */
656: #ifdef F_NDELAY
657: status |= F_NDELAY;
658: #endif /* F_NDELAY */
659: #endif /* !O_NONBLOCK */
1.3 daniel 660: status = fcntl(s, F_SETFL, status);
1.2 daniel 661: }
1.26 veillard 662: if (status < 0) {
1.5 daniel 663: #ifdef DEBUG_HTTP
1.1 daniel 664: perror("nonblocking");
1.5 daniel 665: #endif
1.26 veillard 666: closesocket(s);
1.1 daniel 667: return(-1);
668: }
1.2 daniel 669: #endif /* !VMS */
670: #endif /* !_WINSOCKAPI_ */
671:
1.1 daniel 672:
673: sin.sin_family = AF_INET;
674: sin.sin_addr = ia;
675: sin.sin_port = htons(port);
676:
1.30 veillard 677: if ((connect(s, (struct sockaddr *)&sin, sizeof(sin))==-1)) {
678: switch (socket_errno()) {
679: case EINPROGRESS:
680: case EWOULDBLOCK:
681: break;
682: default:
683: perror("connect");
684: closesocket(s);
685: return(-1);
686: }
1.1 daniel 687: }
688:
1.28 veillard 689: tv.tv_sec = timeout;
1.1 daniel 690: tv.tv_usec = 0;
691:
692: FD_ZERO(&wfd);
693: FD_SET(s, &wfd);
694:
695: switch(select(s+1, NULL, &wfd, NULL, &tv))
696: {
697: case 0:
698: /* Time out */
1.26 veillard 699: closesocket(s);
1.1 daniel 700: return(-1);
701: case -1:
702: /* Ermm.. ?? */
1.5 daniel 703: #ifdef DEBUG_HTTP
1.1 daniel 704: perror("select");
1.5 daniel 705: #endif
1.26 veillard 706: closesocket(s);
1.1 daniel 707: return(-1);
708: }
1.19 daniel 709:
710: if ( FD_ISSET(s, &wfd) ) {
1.27 veillard 711: SOCKLEN_T len;
1.19 daniel 712: len = sizeof(status);
1.30 veillard 713: if (getsockopt(s, SOL_SOCKET, SO_ERROR, (char*)&status, &len) < 0 ) {
1.19 daniel 714: /* Solaris error code */
715: return (-1);
716: }
717: if ( status ) {
1.26 veillard 718: closesocket(s);
1.19 daniel 719: errno = status;
720: return (-1);
721: }
722: } else {
723: /* pbm */
724: return (-1);
725: }
1.1 daniel 726:
1.5 daniel 727: return(s);
1.1 daniel 728: }
729:
1.5 daniel 730: /**
731: * xmlNanoHTTPConnectHost:
732: * @host: the host name
733: * @port: the port number
734: *
735: * Attempt a connection to the given host:port endpoint. It tries
736: * the multiple IP provided by the DNS if available.
737: *
738: * Returns -1 in case of failure, the file descriptor number otherwise
739: */
740:
741: static int
742: xmlNanoHTTPConnectHost(const char *host, int port)
1.1 daniel 743: {
744: struct hostent *h;
745: int i;
746: int s;
747:
748: h=gethostbyname(host);
1.26 veillard 749: if (h==NULL)
1.1 daniel 750: {
1.5 daniel 751: #ifdef DEBUG_HTTP
1.1 daniel 752: fprintf(stderr,"unable to resolve '%s'.\n", host);
1.5 daniel 753: #endif
1.1 daniel 754: return(-1);
755: }
756:
757: for(i=0; h->h_addr_list[i]; i++)
758: {
759: struct in_addr ia;
760: memcpy(&ia, h->h_addr_list[i],4);
761: s = xmlNanoHTTPConnectAttempt(ia, port);
1.26 veillard 762: if (s != -1)
1.5 daniel 763: return(s);
1.1 daniel 764: }
1.5 daniel 765:
766: #ifdef DEBUG_HTTP
1.1 daniel 767: fprintf(stderr, "unable to connect to '%s'.\n", host);
1.5 daniel 768: #endif
1.1 daniel 769: return(-1);
770: }
771:
772:
1.5 daniel 773: /**
774: * xmlNanoHTTPOpen:
775: * @URL: The URL to load
776: * @contentType: if available the Content-Type information will be
777: * returned at that location
778: *
779: * This function try to open a connection to the indicated resource
780: * via HTTP GET.
781: *
1.6 daniel 782: * Returns NULL in case of failure, otherwise a request handler.
783: * The contentType, if provided must be freed by the caller
1.5 daniel 784: */
1.1 daniel 785:
1.18 daniel 786: void*
1.1 daniel 787: xmlNanoHTTPOpen(const char *URL, char **contentType) {
1.5 daniel 788: if (contentType != NULL) *contentType = NULL;
1.33 ! veillard 789: return xmlNanoHTTPMethod(URL, NULL, NULL, contentType, NULL);
1.1 daniel 790: }
791:
1.5 daniel 792: /**
793: * xmlNanoHTTPRead:
794: * @ctx: the HTTP context
795: * @dest: a buffer
796: * @len: the buffer length
797: *
798: * This function tries to read @len bytes from the existing HTTP connection
799: * and saves them in @dest. This is a blocking call.
800: *
801: * Returns the number of byte read. 0 is an indication of an end of connection.
802: * -1 indicates a parameter error.
803: */
1.1 daniel 804: int
805: xmlNanoHTTPRead(void *ctx, void *dest, int len) {
806: xmlNanoHTTPCtxtPtr ctxt = (xmlNanoHTTPCtxtPtr) ctx;
807:
808: if (ctx == NULL) return(-1);
809: if (dest == NULL) return(-1);
810: if (len <= 0) return(0);
811:
812: while (ctxt->inptr - ctxt->inrptr < len) {
813: if (xmlNanoHTTPRecv(ctxt) == 0) break;
814: }
815: if (ctxt->inptr - ctxt->inrptr < len)
816: len = ctxt->inptr - ctxt->inrptr;
817: memcpy(dest, ctxt->inrptr, len);
818: ctxt->inrptr += len;
819: return(len);
820: }
821:
1.5 daniel 822: /**
823: * xmlNanoHTTPClose:
824: * @ctx: the HTTP context
825: *
826: * This function closes an HTTP context, it ends up the connection and
827: * free all data related to it.
828: */
1.1 daniel 829: void
830: xmlNanoHTTPClose(void *ctx) {
831: xmlNanoHTTPCtxtPtr ctxt = (xmlNanoHTTPCtxtPtr) ctx;
832:
833: if (ctx == NULL) return;
834:
835: xmlNanoHTTPFreeCtxt(ctxt);
836: }
837:
1.5 daniel 838: /**
1.6 daniel 839: * xmlNanoHTTPMethod:
840: * @URL: The URL to load
841: * @method: the HTTP method to use
842: * @input: the input string if any
843: * @contentType: the Content-Type information IN and OUT
844: * @headers: the extra headers
845: *
846: * This function try to open a connection to the indicated resource
847: * via HTTP using the given @method, adding the given extra headers
848: * and the input buffer for the request content.
849: *
850: * Returns NULL in case of failure, otherwise a request handler.
851: * The contentType, if provided must be freed by the caller
852: */
853:
1.18 daniel 854: void*
1.6 daniel 855: xmlNanoHTTPMethod(const char *URL, const char *method, const char *input,
856: char **contentType, const char *headers) {
857: xmlNanoHTTPCtxtPtr ctxt;
1.33 ! veillard 858: char *bp, *p;
! 859: int blen, ilen, ret;
1.6 daniel 860: int head;
861: int nbRedirects = 0;
862: char *redirURL = NULL;
863:
864: if (URL == NULL) return(NULL);
865: if (method == NULL) method = "GET";
1.33 ! veillard 866: xmlNanoHTTPInit();
1.6 daniel 867:
868: retry:
869: if (redirURL == NULL)
870: ctxt = xmlNanoHTTPNewCtxt(URL);
871: else {
872: ctxt = xmlNanoHTTPNewCtxt(redirURL);
1.7 daniel 873: xmlFree(redirURL);
1.6 daniel 874: redirURL = NULL;
875: }
876:
877: if ((ctxt->protocol == NULL) || (strcmp(ctxt->protocol, "http"))) {
878: xmlNanoHTTPFreeCtxt(ctxt);
1.7 daniel 879: if (redirURL != NULL) xmlFree(redirURL);
1.6 daniel 880: return(NULL);
881: }
882: if (ctxt->hostname == NULL) {
883: xmlNanoHTTPFreeCtxt(ctxt);
884: return(NULL);
885: }
1.33 ! veillard 886: if (proxy) {
! 887: blen = strlen(ctxt->hostname) * 2 + 16;
! 888: ret = xmlNanoHTTPConnectHost(proxy, proxyPort);
! 889: }
! 890: else {
! 891: blen = strlen(ctxt->hostname);
! 892: ret = xmlNanoHTTPConnectHost(ctxt->hostname, ctxt->port);
! 893: }
1.6 daniel 894: if (ret < 0) {
895: xmlNanoHTTPFreeCtxt(ctxt);
896: return(NULL);
897: }
898: ctxt->fd = ret;
899:
1.33 ! veillard 900: if (input != NULL) {
! 901: ilen = strlen(input);
! 902: blen += ilen + 32;
! 903: }
! 904: else
! 905: ilen = 0;
! 906: if (headers != NULL)
! 907: blen += strlen(headers);
! 908: if (contentType && *contentType)
! 909: blen += strlen(*contentType) + 16;
! 910: blen += strlen(method) + strlen(ctxt->path) + 23;
! 911: bp = xmlMalloc(blen);
! 912: if (proxy) {
! 913: if (ctxt->port != 80) {
! 914: sprintf(bp, "%s http://%s:%d%s", method, ctxt->hostname,
! 915: ctxt->port, ctxt->path);
1.6 daniel 916: }
1.33 ! veillard 917: else
! 918: sprintf(bp, "%s http://%s%s", method, ctxt->hostname, ctxt->path);
! 919: }
! 920: else
! 921: sprintf(bp, "%s %s", method, ctxt->path);
! 922: p = bp + strlen(bp);
! 923: sprintf(p, " HTTP/1.0\r\nHost: %s\r\n", ctxt->hostname);
! 924: p += strlen(p);
! 925: if (contentType != NULL && *contentType) {
! 926: sprintf(p, "Content-Type: %s\r\n", *contentType);
! 927: p += strlen(p);
! 928: }
! 929: if (headers != NULL) {
! 930: strcpy(p, headers);
! 931: p += strlen(p);
1.6 daniel 932: }
1.33 ! veillard 933: if (input != NULL)
! 934: sprintf(p, "Content-Length: %d\r\n\r\n%s", ilen, input);
! 935: else
! 936: strcpy(p, "\r\n");
1.6 daniel 937: #ifdef DEBUG_HTTP
1.33 ! veillard 938: printf("-> %s%s", proxy? "(Proxy) " : "", bp);
! 939: if ((blen -= strlen(bp)+1) < 0)
! 940: printf("ERROR: overflowed buffer by %d bytes\n", -blen);
1.6 daniel 941: #endif
1.33 ! veillard 942: ctxt->outptr = ctxt->out = bp;
1.6 daniel 943: ctxt->state = XML_NANO_HTTP_WRITE;
944: xmlNanoHTTPSend(ctxt);
945: ctxt->state = XML_NANO_HTTP_READ;
946: head = 1;
947:
948: while ((p = xmlNanoHTTPReadLine(ctxt)) != NULL) {
949: if (head && (*p == 0)) {
950: head = 0;
951: ctxt->content = ctxt->inrptr;
1.33 ! veillard 952: xmlFree(p);
1.6 daniel 953: break;
954: }
955: xmlNanoHTTPScanAnswer(ctxt, p);
956:
957: #ifdef DEBUG_HTTP
1.33 ! veillard 958: printf("<- %s\n", p);
1.6 daniel 959: #endif
1.33 ! veillard 960: xmlFree(p);
1.6 daniel 961: }
962:
963: if ((ctxt->location != NULL) && (ctxt->returnValue >= 300) &&
964: (ctxt->returnValue < 400)) {
965: #ifdef DEBUG_HTTP
966: printf("\nRedirect to: %s\n", ctxt->location);
967: #endif
968: while (xmlNanoHTTPRecv(ctxt)) ;
969: if (nbRedirects < XML_NANO_HTTP_MAX_REDIR) {
970: nbRedirects++;
1.7 daniel 971: redirURL = xmlMemStrdup(ctxt->location);
1.6 daniel 972: xmlNanoHTTPFreeCtxt(ctxt);
973: goto retry;
974: }
975: xmlNanoHTTPFreeCtxt(ctxt);
976: #ifdef DEBUG_HTTP
1.33 ! veillard 977: printf("Too many redirects, aborting ...\n");
1.6 daniel 978: #endif
979: return(NULL);
980:
981: }
982:
1.33 ! veillard 983: if (contentType != NULL) {
! 984: if (ctxt->contentType != NULL)
! 985: *contentType = xmlMemStrdup(ctxt->contentType);
! 986: else
! 987: *contentType = NULL;
! 988: }
1.6 daniel 989:
990: #ifdef DEBUG_HTTP
991: if (ctxt->contentType != NULL)
992: printf("\nCode %d, content-type '%s'\n\n",
993: ctxt->returnValue, ctxt->contentType);
994: else
995: printf("\nCode %d, no content-type\n\n",
996: ctxt->returnValue);
997: #endif
998:
999: return((void *) ctxt);
1000: }
1001:
1002: /**
1.5 daniel 1003: * xmlNanoHTTPFetch:
1004: * @URL: The URL to load
1005: * @filename: the filename where the content should be saved
1006: * @contentType: if available the Content-Type information will be
1007: * returned at that location
1008: *
1009: * This function try to fetch the indicated resource via HTTP GET
1010: * and save it's content in the file.
1011: *
1012: * Returns -1 in case of failure, 0 incase of success. The contentType,
1013: * if provided must be freed by the caller
1014: */
1015: int
1016: xmlNanoHTTPFetch(const char *URL, const char *filename, char **contentType) {
1.1 daniel 1017: void *ctxt;
1018: char buf[4096];
1019: int fd;
1020: int len;
1021:
1022: ctxt = xmlNanoHTTPOpen(URL, contentType);
1023: if (ctxt == NULL) return(-1);
1024:
1025: if (!strcmp(filename, "-"))
1026: fd = 0;
1027: else {
1.13 daniel 1028: fd = open(filename, O_CREAT | O_WRONLY, 00644);
1.1 daniel 1029: if (fd < 0) {
1030: xmlNanoHTTPClose(ctxt);
1.5 daniel 1031: if ((contentType != NULL) && (*contentType != NULL)) {
1.7 daniel 1032: xmlFree(*contentType);
1.5 daniel 1033: *contentType = NULL;
1034: }
1.1 daniel 1035: return(-1);
1036: }
1037: }
1038:
1039: while ((len = xmlNanoHTTPRead(ctxt, buf, sizeof(buf))) > 0) {
1040: write(fd, buf, len);
1041: }
1042:
1043: xmlNanoHTTPClose(ctxt);
1.13 daniel 1044: close(fd);
1.1 daniel 1045: return(0);
1.6 daniel 1046: }
1047:
1048: /**
1049: * xmlNanoHTTPSave:
1.8 daniel 1050: * @ctxt: the HTTP context
1.6 daniel 1051: * @filename: the filename where the content should be saved
1052: *
1053: * This function saves the output of the HTTP transaction to a file
1054: * It closes and free the context at the end
1055: *
1056: * Returns -1 in case of failure, 0 incase of success.
1057: */
1058: int
1059: xmlNanoHTTPSave(void *ctxt, const char *filename) {
1060: char buf[4096];
1061: int fd;
1062: int len;
1063:
1064: if (ctxt == NULL) return(-1);
1065:
1066: if (!strcmp(filename, "-"))
1067: fd = 0;
1068: else {
1069: fd = open(filename, O_CREAT | O_WRONLY);
1070: if (fd < 0) {
1071: xmlNanoHTTPClose(ctxt);
1072: return(-1);
1073: }
1074: }
1075:
1076: while ((len = xmlNanoHTTPRead(ctxt, buf, sizeof(buf))) > 0) {
1077: write(fd, buf, len);
1078: }
1079:
1080: xmlNanoHTTPClose(ctxt);
1081: return(0);
1082: }
1083:
1084: /**
1085: * xmlNanoHTTPReturnCode:
1086: * @ctx: the HTTP context
1087: *
1088: * Returns the HTTP return code for the request.
1089: */
1090: int
1091: xmlNanoHTTPReturnCode(void *ctx) {
1092: xmlNanoHTTPCtxtPtr ctxt = (xmlNanoHTTPCtxtPtr) ctx;
1093:
1094: if (ctxt == NULL) return(-1);
1095:
1096: return(ctxt->returnValue);
1.1 daniel 1097: }
1098:
1099: #ifdef STANDALONE
1100: int main(int argc, char **argv) {
1101: char *contentType = NULL;
1102:
1103: if (argv[1] != NULL) {
1104: if (argv[2] != NULL)
1105: xmlNanoHTTPFetch(argv[1], argv[2], &contentType);
1106: else
1107: xmlNanoHTTPFetch(argv[1], "-", &contentType);
1.7 daniel 1108: if (contentType != NULL) xmlFree(contentType);
1.1 daniel 1109: } else {
1110: printf("%s: minimal HTTP GET implementation\n", argv[0]);
1111: printf("\tusage %s [ URL [ filename ] ]\n", argv[0]);
1112: }
1.12 daniel 1113: xmlNanoHTTPCleanup();
1114: xmlMemoryDump();
1.1 daniel 1115: return(0);
1116: }
1117: #endif /* STANDALONE */
1.17 daniel 1118: #else /* !LIBXML_HTTP_ENABLED */
1119: #ifdef STANDALONE
1120: #include <stdio.h>
1121: int main(int argc, char **argv) {
1122: printf("%s : HTTP support not compiled in\n", argv[0]);
1123: return(0);
1124: }
1125: #endif /* STANDALONE */
1126: #endif /* LIBXML_HTTP_ENABLED */
Webmaster