Annotation of libwww/Library/src/HTTP.c, revision 1.8
1.1 timbl 1: /* HyperText Tranfer Protocol - Client implementation HTTP.c
2: ** ==========================
1.2 timbl 3: **
4: ** Bugs:
5: ** Not implemented:
6: ** Forward
7: ** Redirection
8: ** Error handling
1.1 timbl 9: */
10:
11: /* Module parameters:
12: ** -----------------
13: **
14: ** These may be undefined and redefined by syspec.h
15: */
1.2 timbl 16:
17: /* Implements:
18: */
19: #include "HTTP.h"
20:
21: #define HTTP_VERSION "HTTP/1.0"
22: #define HTTP2 /* Version is greater than 0.9 */
23:
1.3 timbl 24: #define CR FROMASCII('015円') /* Must be converted to ^M for transmission */
25: #define LF FROMASCII('012円') /* Must be converted to ^J for transmission */
26:
1.2 timbl 27: #define INIT_LINE_SIZE 1024 /* Start with line buffer this big */
28: #define LINE_EXTEND_THRESH 256 /* Minimum read size */
29: #define VERSION_LENGTH 20 /* for returned protocol version */
30:
31: /* Uses:
32: */
1.1 timbl 33: #include "HTParse.h"
34: #include "HTUtils.h"
35: #include "tcp.h"
36: #include "HTTCP.h"
37: #include "HTFormat.h"
1.2 timbl 38: #include <ctype.h>
39: #include "HTAlert.h"
40: #include "HTMIME.h"
1.5 timbl 41: #include "HTML.h" /* SCW */
42: #include "HTInit.h" /* SCW */
1.1 timbl 43:
1.2 timbl 44: struct _HTStream {
45: HTStreamClass * isa; /* all we need to know */
46: };
47:
48:
1.6 timbl 49: extern char * HTAppName; /* Application name: please supply */
50: extern char * HTAppVersion; /* Application version: please supply */
51:
1.1 timbl 52: /* Load Document from HTTP Server HTLoadHTTP()
53: ** ==============================
54: **
55: ** Given a hypertext address, this routine loads a document.
56: **
57: **
58: ** On entry,
59: ** arg is the hypertext reference of the article to be loaded.
60: ** gate is nill if no gateway, else the gateway address.
61: **
62: ** On exit,
63: ** returns >=0 If no error, a good socket number
64: ** <0 Error.
65: **
66: ** The socket must be closed by the caller after the document has been
67: ** read.
68: **
69: */
1.2 timbl 70: PUBLIC int HTLoadHTTP ARGS4 (
71: CONST char *, arg,
72: /* CONST char *, gate, */
73: HTParentAnchor *, anAnchor,
74: HTFormat, format_out,
75: HTStream*, sink)
1.1 timbl 76: {
77: int s; /* Socket number for returned data */
78: char *command; /* The whole command */
1.3 timbl 79: char * eol = 0; /* End of line if found */
1.7 timbl 80: char * start_of_data; /* Start of body of reply */
1.1 timbl 81: int status; /* tcp return */
1.3 timbl 82: HTStream * target = NULL; /* Unconverted data */
83: HTFormat format_in; /* Format arriving in the message */
84:
1.2 timbl 85: CONST char* gate = 0; /* disable this feature */
1.1 timbl 86: SockA soc_address; /* Binary network address */
87: SockA * sin = &soc_address;
1.2 timbl 88: BOOL had_header = NO; /* Have we had at least one header? */
89: char * line_buffer = NULL;
90: BOOL extensions = YES; /* Assume good HTTP server */
1.1 timbl 91: if (!arg) return -3; /* Bad if no name sepcified */
92: if (!*arg) return -2; /* Bad if name had zero length */
93:
94: /* Set up defaults:
95: */
96: #ifdef DECNET
1.2 timbl 97: sin->sdn_family = AF_DECnet; /* Family = DECnet, host order */
98: sin->sdn_objnum = DNP_OBJ; /* Default: http object number */
1.1 timbl 99: #else /* Internet */
1.2 timbl 100: sin->sin_family = AF_INET; /* Family = internet, host order */
101: sin->sin_port = htons(TCP_PORT); /* Default: http port */
1.1 timbl 102: #endif
103:
104: if (TRACE) {
105: if (gate) fprintf(stderr,
106: "HTTPAccess: Using gateway %s for %s\n", gate, arg);
107: else fprintf(stderr, "HTTPAccess: Direct access for %s\n", arg);
108: }
109:
110: /* Get node name and optional port number:
111: */
112: {
113: char *p1 = HTParse(gate ? gate : arg, "", PARSE_HOST);
114: int status = HTParseInet(sin, p1); /* TBL 920622 */
115: free(p1);
116: if (status) return status; /* No such host for example */
117: }
118:
1.2 timbl 119: retry:
1.1 timbl 120:
121: /* Now, let's get a socket set up from the server for the sgml data:
122: */
123: #ifdef DECNET
124: s = socket(AF_DECnet, SOCK_STREAM, 0);
125: #else
126: s = socket(AF_INET, SOCK_STREAM, IPPROTO_TCP);
127: #endif
128: status = connect(s, (struct sockaddr*)&soc_address, sizeof(soc_address));
129: if (status < 0) {
130: #ifndef DECNET
131: /* This code is temporary backward-compatibility. It should
132: go away when no server runs on port 2784 alone */
133: if (sin->sin_port == htons(TCP_PORT)) { /* Try the old one */
134: if (TRACE) printf (
135: "HTTP: Port %d doesn't answer (errno = %d). Trying good old port %d...\n",
136: TCP_PORT, errno, OLD_TCP_PORT);
137: sin->sin_port = htons(OLD_TCP_PORT);
138: /* First close current socket and open a clean one */
139: status = NETCLOSE (s);
140: s = socket(AF_INET, SOCK_STREAM, IPPROTO_TCP);
141: status = connect(s, (struct sockaddr*)&soc_address,
142: sizeof(soc_address));
143: }
144: if (status < 0)
145: #endif
146: {
147: if (TRACE) fprintf(stderr,
148: "HTTP: Unable to connect to remote host for `%s' (errno = %d).\n", arg, errno);
149: /* free(command); BUG OUT TBL 921121 */
150: return HTInetStatus("connect");
151: }
152: }
153:
154: if (TRACE) fprintf(stderr, "HTTP connected, socket %d\n", s);
155:
156: /* Ask that node for the document,
157: ** omitting the host name & anchor if not gatewayed.
158: */
159: if (gate) {
1.2 timbl 160: command = malloc(4 + strlen(arg)+ 2 + 31);
1.1 timbl 161: if (command == NULL) outofmem(__FILE__, "HTLoadHTTP");
162: strcpy(command, "GET ");
163: strcat(command, arg);
164: } else { /* not gatewayed */
165: char * p1 = HTParse(arg, "", PARSE_PATH|PARSE_PUNCTUATION);
1.2 timbl 166: command = malloc(4 + strlen(p1)+ 2 + 31);
1.1 timbl 167: if (command == NULL) outofmem(__FILE__, "HTLoadHTTP");
168: strcpy(command, "GET ");
169: strcat(command, p1);
170: free(p1);
171: }
1.2 timbl 172: #ifdef HTTP2
173: if (extensions) {
174: strcat(command, " ");
175: strcat(command, HTTP_VERSION);
176: }
177: #endif
1.3 timbl 178: {
179: char * p = command + strlen(command);
180: *p++ = CR; /* Macros to be correct on Mac */
181: *p++ = LF;
182: *p++ = 0;
183: /* strcat(command, "\r\n"); */ /* CR LF, as in rfc 977 */
184: }
1.1 timbl 185:
1.2 timbl 186: if (extensions) {
187:
188: int n;
189: int i;
190: HTAtom * present = WWW_PRESENT;
191: char line[256]; /*@@@@ */
192:
193: if (!HTPresentations) HTFormatInit();
194: n = HTList_count(HTPresentations);
195:
196: for(i=0; i<n; i++) {
197: HTPresentation * pres = HTList_objectAt(HTPresentations, i);
198: if (pres->rep_out == present) {
199: if (pres->quality != 1.0) {
1.3 timbl 200: sprintf(line, "Accept: %s q=%.3f%c%c",
201: HTAtom_name(pres->rep), pres->quality, CR, LF);
1.2 timbl 202: } else {
1.3 timbl 203: sprintf(line, "Accept: %s%c%c",
204: HTAtom_name(pres->rep), CR, LF);
1.2 timbl 205: }
206: StrAllocCat(command, line);
207:
208: }
209: }
1.6 timbl 210:
211: sprintf(line, "User-Agent: %s/%s libwww/%s%c%c",
212: HTAppName ? HTAppName : "unknown",
213: HTAppVersion ? HTAppVersion : "0.0",
214: HTLibraryVersion, CR, LF);
215: StrAllocCat(command, line);
1.2 timbl 216: }
1.3 timbl 217:
1.4 timbl 218: StrAllocCat(command, "015円012円"); /* Blank line means "end" */
219: #ifdef NOT_ASCII
1.1 timbl 220: {
221: char * p;
222: for(p = command; *p; p++) {
223: *p = TOASCII(*p);
224: }
1.4 timbl 225: }
1.3 timbl 226: #endif
1.1 timbl 227:
1.2 timbl 228: if (TRACE) fprintf(stderr, "HTTP Tx: %s\n", command);
1.1 timbl 229: status = NETWRITE(s, command, (int)strlen(command));
230: free(command);
231: if (status<0) {
232: if (TRACE) fprintf(stderr, "HTTPAccess: Unable to send command.\n");
233: return HTInetStatus("send");
234: }
235:
1.2 timbl 236:
1.7 timbl 237: /* Read the first line of the response
238: ** -----------------------------------
1.2 timbl 239: */
1.3 timbl 240:
1.2 timbl 241: {
242:
243: /* Get numeric status etc */
244:
245: int status;
246: int length = 0;
247: BOOL end_of_file = NO;
248: HTAtom * encoding = HTAtom_for("7bit");
249: int buffer_length = INIT_LINE_SIZE; /* Why not? */
250:
251: line_buffer = (char *) malloc(buffer_length * sizeof(char));
252: if (!line_buffer) outofmem(__FILE__, "HTLoadHTTP");
253:
1.7 timbl 254: do { /* Loop to read in the first line */
1.2 timbl 255:
256: /* Extend line buffer if necessary for those crazy WAIS URLs ;-) */
257:
258: if (buffer_length - length < LINE_EXTEND_THRESH) {
259: buffer_length = buffer_length + buffer_length;
260: line_buffer = (char *) realloc(
261: line_buffer, buffer_length * sizeof(char));
262: if (!line_buffer) outofmem(__FILE__, "HTLoadHTTP");
263: }
264: status = NETREAD(s, line_buffer + length,
265: buffer_length - length -1);
266: if (status < 0) {
267: HTAlert("Unexpected network read error on response");
268: return status;
269: }
270: if (status == 0) {
271: end_of_file = YES;
272: break;
273: }
274: line_buffer[length+status] = 0;
275: #ifdef NOT_ASCII
276: {
277: char * p;
278: for(p = line_buffer+length; *p; p++) {
279: *p = FROMASCII(*p);
280: }
281: }
282: #endif
1.3 timbl 283: eol = strchr(line_buffer + length, LF);
284: if (eol && *(eol-1) == CR) *(eol-1) = ' ';
1.2 timbl 285:
286: length = length + status;
287:
1.7 timbl 288: if (eol) *eol = 0; /* Terminate the line */
1.2 timbl 289:
1.7 timbl 290: } while (!eol && !end_of_file); /* No LF */
291:
292: } /* Scope of loop variables */
1.2 timbl 293:
1.7 timbl 294:
295: /* We now have a terminated unfolded line. Parse it.
296: ** -------------------------------------------------
1.2 timbl 297: */
298:
1.7 timbl 299: if (TRACE)fprintf(stderr, "HTTP: Rx: %s\n", line_buffer);
300:
301: {
302: int fields;
303: char server_version [VERSION_LENGTH+1];
304: int server_status;
305:
1.2 timbl 306:
307: /* Kludge to work with old buggy servers. They can't handle the third word
308: ** so we try again without it.
309: */
1.7 timbl 310: if (extensions &&
311: 0==strcmp(line_buffer, /* Old buggy server? */
312: "Document address invalid or access not authorised")) {
313: extensions = NO;
314: if (line_buffer) free(line_buffer);
315: if (TRACE) fprintf(stderr,
316: "HTTP: close socket %d to retry with HTTP0\n", s);
317: NETCLOSE(s);
318: goto retry; /* @@@@@@@@@@ */
319: }
1.2 timbl 320:
1.7 timbl 321: fields = sscanf(line_buffer, "%20s%d",
322: server_version,
323: &server_status);
324:
325: if (fields < 2) { /* HTTP0 reply */
326: format_in = WWW_HTML;
327: start_of_data = line_buffer; /* reread whole reply */
1.2 timbl 328:
1.7 timbl 329: } else { /* Ful HTTP reply */
330:
331: /* Decode full HTTP response */
332:
1.3 timbl 333: format_in = HTAtom_for("www/mime");
1.7 timbl 334: start_of_data = eol ? eol + 1 : "";
1.3 timbl 335:
1.2 timbl 336: switch (server_status / 100) {
337:
1.3 timbl 338: default: /* bad number */
339: HTAlert("Unknown status reply from server!");
340: break;
341:
1.2 timbl 342: case 3: /* Various forms of redirection */
1.7 timbl 343: HTAlert(
1.3 timbl 344: "Redirection response from server is not handled by this client");
345: break;
346:
1.2 timbl 347: case 4: /* "I think I goofed" */
348: case 5: /* I think you goofed */
1.6 timbl 349: {
350: char *p1 = HTParse(gate ? gate : arg, "", PARSE_HOST);
351: char * message = (char*)malloc(
1.7 timbl 352: strlen(line_buffer)+strlen(p1) + 100);
1.8 ! timbl 353: int status;
1.6 timbl 354: if (!message) outofmem(__FILE__, "HTTP 5xx status");
355: sprintf(message,
356: "HTTP server at %s replies:\n%s", p1, line_buffer);
1.8 ! timbl 357: status = HTLoadError(sink, server_status, message);
1.6 timbl 358: free(message);
359: free(p1);
360: goto clean_up;
361: }
1.3 timbl 362: break;
1.2 timbl 363:
364: case 2: /* Good: Got MIME object */
365: break;
366:
1.7 timbl 367: } /* switch on response code */
368:
369: } /* Full HTTP reply */
370:
371: } /* scope of fields */
1.2 timbl 372:
1.3 timbl 373: /* Set up the stream stack to handle the body of the message
374: */
375:
376: target = HTStreamStack(format_in,
377: format_out,
378: sink , anAnchor);
379:
380: if (!target) {
381: char buffer[1024]; /* @@@@@@@@ */
382: if (line_buffer) free(line_buffer);
383: sprintf(buffer, "Sorry, no known way of converting %s to %s.",
384: HTAtom_name(format_in), HTAtom_name(format_out));
385: fprintf(stderr, "HTTP: %s", buffer);
1.6 timbl 386: status = HTLoadError(sink, 501, buffer);
387: goto clean_up;
1.3 timbl 388: }
389:
390:
391: /* Push the data, maybe ignoring CR, down the stream
392: ** We have to remember the end of the first buffer we just read
1.2 timbl 393: */
1.3 timbl 394: if (format_in != WWW_HTML) {
1.7 timbl 395: (*target->isa->put_string)(target, start_of_data);
1.3 timbl 396: HTCopy(s, target);
397:
398: } else { /* ascii text with CRLFs :-( */
1.7 timbl 399: {
1.3 timbl 400: char * p;
1.7 timbl 401: for (p = start_of_data; *p; p++)
1.3 timbl 402: if (*p != '\r') (*target->isa->put_character)(target, *p);
403: }
404: HTCopyNoCR(s, target);
1.2 timbl 405: }
1.3 timbl 406: (*target->isa->end_document)(target);
407: (*target->isa->free)(target);
1.8 ! timbl 408: status = HT_LOADED;
1.2 timbl 409:
410: /* Clean up
1.1 timbl 411: */
1.3 timbl 412:
1.6 timbl 413: clean_up:
1.2 timbl 414: if (line_buffer) free(line_buffer);
1.3 timbl 415:
1.1 timbl 416: if (TRACE) fprintf(stderr, "HTTP: close socket %d.\n", s);
1.6 timbl 417: (void) NETCLOSE(s);
1.1 timbl 418:
1.8 ! timbl 419: return status; /* Good return */
1.3 timbl 420:
1.1 timbl 421: }
1.7 timbl 422:
1.1 timbl 423:
424: /* Protocol descriptor
425: */
426:
1.2 timbl 427: PUBLIC HTProtocol HTTP = { "http", HTLoadHTTP, 0 };
Webmaster