Annotation of libwww/Library/src/HTMIME.c, revision 2.64
2.15 frystyk 1: /* HTMIME.c
2: ** MIME MESSAGE PARSE
3: **
2.22 frystyk 4: ** (c) COPYRIGHT MIT 1995.
2.15 frystyk 5: ** Please first read the full copyright statement in the file COPYRIGH.
2.64 ! eric 6: ** @(#) $Id: HTMIME.c,v 2.63 1996/05/20 15:06:54 frystyk Exp $
2.1 timbl 7: **
8: ** This is RFC 1341-specific code.
9: ** The input stream pushed into this parser is assumed to be
10: ** stripped on CRs, ie lines end with LF, not CR LF.
11: ** (It is easy to change this except for the body part where
12: ** conversion can be slow.)
13: **
14: ** History:
15: ** Feb 92 Written Tim Berners-Lee, CERN
2.13 duns 16: ** 8 Jul 94 FM Insulate free() from _free structure element.
2.18 frystyk 17: ** 14 Mar 95 HFN Now using anchor for storing data. No more `\n',
18: ** static buffers etc.
2.1 timbl 19: */
2.17 frystyk 20:
21: /* Library include files */
2.57 frystyk 22: #include "sysdep.h"
2.60 frystyk 23: #include "WWWUtil.h"
2.61 frystyk 24: #include "WWWCore.h"
25: #include "HTReqMan.h"
26: #include "HTNetMan.h"
2.36 frystyk 27: #include "HTHeader.h"
2.64 ! eric 28: #include "HTWWWStr.h"
2.14 frystyk 29: #include "HTMIME.h" /* Implemented here */
2.1 timbl 30:
2.64 ! eric 31: #define MIME_HASH_SIZE 101
! 32:
2.1 timbl 33: /* MIME Object
34: ** -----------
35: */
36: struct _HTStream {
2.57 frystyk 37: const HTStreamClass * isa;
2.18 frystyk 38: HTRequest * request;
2.32 frystyk 39: HTNet * net;
40: HTParentAnchor * anchor;
2.18 frystyk 41: HTStream * target;
42: HTFormat target_format;
2.64 ! eric 43: HTChunk * token;
! 44: HTChunk * value;
! 45: int hash;
2.59 frystyk 46: HTEOLState EOLstate;
2.18 frystyk 47: BOOL transparent;
2.48 frystyk 48: BOOL head_only;
2.35 frystyk 49: BOOL nntp;
2.62 frystyk 50: BOOL footer;
2.64 ! eric 51: BOOL haveToken;
2.1 timbl 52: };
53:
2.18 frystyk 54: /* ------------------------------------------------------------------------- */
2.1 timbl 55:
2.64 ! eric 56: PRIVATE int pumpData (HTStream * me)
2.18 frystyk 57: {
2.64 ! eric 58: HTRequest * request = me->request;
! 59: HTParentAnchor * anchor = me->anchor;
2.48 frystyk 60: me->transparent = YES; /* Pump rest of data right through */
2.27 frystyk 61:
2.48 frystyk 62: /* If this request us a source in PostWeb then pause here */
63: if (me->head_only || HTRequest_isSource(request)) return HT_PAUSE;
2.47 frystyk 64:
2.48 frystyk 65: /* If HEAD method then we just stop here */
2.62 frystyk 66: if (me->footer || request->method == METHOD_HEAD) return HT_LOADED;
2.43 frystyk 67:
2.60 frystyk 68: /*
69: ** Handle any Content Type
70: ** News server almost never send content type or content length
71: */
2.61 frystyk 72: {
73: HTFormat format = HTAnchor_format(anchor);
74: if (format != WWW_UNKNOWN || me->nntp) {
75: if (STREAM_TRACE) HTTrace("Building.... C-T stack from %s to %s\n",
76: HTAtom_name(format),
77: HTAtom_name(me->target_format));
78: me->target = HTStreamStack(format, me->target_format,
79: me->target, request, YES);
80: }
2.18 frystyk 81: }
2.60 frystyk 82:
83: /* Handle any Content Encoding */
2.61 frystyk 84: {
85: HTList * cc = HTAnchor_encoding(anchor);
86: if (cc) {
87: if (STREAM_TRACE) HTTrace("Building.... C-E stack\n");
88: me->target = HTContentDecodingStack(cc, me->target, request, NULL);
89: }
2.60 frystyk 90: }
91:
92: /* Handle any Transfer encoding */
2.61 frystyk 93: {
94: HTEncoding transfer = HTAnchor_transfer(anchor);
95: if (!HTFormat_isUnityTransfer(transfer)) {
96: if (STREAM_TRACE) HTTrace("Building.... C-T-E stack\n");
97: me->target = HTTransferCodingStack(transfer, me->target,
98: request, NULL, NO);
99: }
100: }
2.27 frystyk 101: return HT_OK;
2.1 timbl 102: }
103:
2.18 frystyk 104: /*
105: ** Header is terminated by CRCR, LFLF, CRLFLF, CRLFCRLF
106: ** Folding is either of CF LWS, LF LWS, CRLF LWS
107: */
2.57 frystyk 108: PRIVATE int HTMIME_put_block (HTStream * me, const char * b, int l)
2.18 frystyk 109: {
2.57 frystyk 110: const char * start = b;
111: const char * end = start;
2.64 ! eric 112: const char * value = me->value->size ? b : NULL;
! 113: long cl;
! 114: int status;
! 115: /* enum {Line_CHAR, Line_END, Line_FOLD, Line_LINE} line = Line_CHAR; */
! 116:
! 117: while (!me->transparent) {
2.18 frystyk 118: if (me->EOLstate == EOL_FCR) {
2.64 ! eric 119: if (*b == CR) /* End of header */
! 120: me->EOLstate = EOL_END;
! 121: else if (*b == LF) /* CRLF */
2.18 frystyk 122: me->EOLstate = EOL_FLF;
2.64 ! eric 123: else if (WHITE(*b)) /* Folding: CR SP */
! 124: me->EOLstate = EOL_FOLD;
! 125: else /* New line */
! 126: me->EOLstate = EOL_LINE;
2.18 frystyk 127: } else if (me->EOLstate == EOL_FLF) {
128: if (*b == CR) /* LF CR or CR LF CR */
129: me->EOLstate = EOL_SCR;
2.64 ! eric 130: else if (*b == LF) /* End of header */
! 131: me->EOLstate = EOL_END;
! 132: else if (WHITE(*b)) /* Folding: LF SP or CR LF SP */
! 133: me->EOLstate = EOL_FOLD;
! 134: else /* New line */
! 135: me->EOLstate = EOL_LINE;
! 136: } else if (me->EOLstate == EOL_SCR) {
! 137: if (*b==CR || *b==LF) /* End of header */
! 138: me->EOLstate = EOL_END;
! 139: else if (WHITE(*b)) /* Folding: LF CR SP or CR LF CR SP */
! 140: me->EOLstate = EOL_FOLD;
! 141: else /* New line */
! 142: me->EOLstate = EOL_LINE;
! 143: } else if (*b == CR)
! 144: me->EOLstate = EOL_FCR;
! 145: else if (*b == LF)
! 146: me->EOLstate = EOL_FLF; /* Line found */
! 147: else {
! 148: if (!me->haveToken) {
! 149: if (*b == ':' || isspace(*b)) {
! 150: HTChunk_putb(me->token, start, end-start);
! 151: HTChunk_putc(me->token, '\0');
! 152: me->haveToken = YES;
! 153: } else {
! 154: unsigned char ch = *(unsigned char *) b;
! 155: tolower(ch);
! 156: /* if (ch >= 'A' && ch <= 'Z')
! 157: ch += ('a' - 'A'); */
! 158: me->hash = (me->hash * 3 + ch) % MIME_HASH_SIZE;
! 159: }
! 160: } else if (value == NULL && *b != ':' && !isspace(*b))
! 161: value = b;
! 162: end++;
! 163: }
! 164: switch (me->EOLstate) {
! 165: case EOL_LINE:
! 166: case EOL_END: {
! 167: int status;
! 168: HTChunk_putb(me->value, value, end-value);
! 169: HTChunk_putc(me->value, '\0');
! 170: start=b, end=b;
! 171: status = HTRequest_dispatchMIMEParse(me->request,
! 172: HTChunk_data(me->token),
! 173: HTChunk_data(me->value));
! 174: if (me->EOLstate == EOL_END) { /* EOL_END */
! 175: if (status == HT_OK)
! 176: status = pumpData(me);
! 177: HTNet_setBytesRead(me->net, l);
! 178: } else { /* EOL_LINE */
! 179: HTChunk_clear(me->token);
! 180: HTChunk_clear(me->value);
! 181: me->haveToken = NO;
! 182: me->hash = 0;
! 183: value = NULL;
! 184: }
2.18 frystyk 185: me->EOLstate = EOL_BEGIN;
2.27 frystyk 186: if (status != HT_OK)
187: return status;
2.64 ! eric 188: break;
! 189: }
! 190: case EOL_FOLD:
2.18 frystyk 191: me->EOLstate = EOL_BEGIN;
2.64 ! eric 192: if (!me->haveToken) {
! 193: HTChunk_putb(me->token, start, end-start);
! 194: HTChunk_putc(me->token, '\0');
! 195: me->haveToken = YES;
! 196: } else if (value) {
! 197: HTChunk_putb(me->value, value, end-value);
! 198: HTChunk_putc(me->value, ' ');
! 199: }
! 200: start=b, end=b;
! 201: break;
! 202: default:
! 203: b++;
! 204: l--;
! 205: if (!l) {
! 206: if (!me->haveToken)
! 207: HTChunk_putb(me->token, start, end-start);
! 208: else if (value)
! 209: HTChunk_putb(me->value, value, end-value);
! 210: return HT_OK;
! 211: }
! 212: }
2.18 frystyk 213: }
2.32 frystyk 214:
215: /*
216: ** Put the rest down the stream without touching the data but make sure
217: ** that we get the correct content length of data
218: */
2.64 ! eric 219: if ((status = (*me->target->isa->put_block)(me->target, b, l)) != HT_OK)
! 220: return status;
! 221: /* Check if CL at all - thanks to jwei@hal.com (John Wei) */
! 222: cl = HTAnchor_length(me->anchor);
! 223: return (cl>=0 && HTNet_bytesRead(me->net)>=cl) ? HT_LOADED : HT_OK;
2.18 frystyk 224: }
225:
226:
227: /* Character handling
228: ** ------------------
229: */
2.36 frystyk 230: PRIVATE int HTMIME_put_character (HTStream * me, char c)
2.18 frystyk 231: {
232: return HTMIME_put_block(me, &c, 1);
233: }
234:
2.1 timbl 235:
236: /* String handling
237: ** ---------------
238: */
2.57 frystyk 239: PRIVATE int HTMIME_put_string (HTStream * me, const char * s)
2.1 timbl 240: {
2.18 frystyk 241: return HTMIME_put_block(me, s, (int) strlen(s));
2.1 timbl 242: }
243:
244:
2.18 frystyk 245: /* Flush an stream object
246: ** ---------------------
2.1 timbl 247: */
2.36 frystyk 248: PRIVATE int HTMIME_flush (HTStream * me)
2.1 timbl 249: {
2.47 frystyk 250: return me->target ? (*me->target->isa->flush)(me->target) : HT_OK;
2.1 timbl 251: }
252:
2.18 frystyk 253: /* Free a stream object
254: ** --------------------
2.1 timbl 255: */
2.36 frystyk 256: PRIVATE int HTMIME_free (HTStream * me)
2.1 timbl 257: {
2.18 frystyk 258: int status = HT_OK;
2.64 ! eric 259: if (!me->transparent)
! 260: if (HTRequest_dispatchMIMEParse(me->request, HTChunk_data(me->token),
! 261: HTChunk_data(me->value)) == HT_OK)
! 262: pumpData(me);
2.25 frystyk 263: if (me->target) {
264: if ((status = (*me->target->isa->_free)(me->target))==HT_WOULD_BLOCK)
265: return HT_WOULD_BLOCK;
266: }
2.26 frystyk 267: if (PROT_TRACE)
2.55 eric 268: HTTrace("MIME........ FREEING....\n");
2.64 ! eric 269: HTChunk_delete(me->token);
! 270: HTChunk_delete(me->value);
2.52 frystyk 271: HT_FREE(me);
2.18 frystyk 272: return status;
2.1 timbl 273: }
274:
275: /* End writing
276: */
2.38 frystyk 277: PRIVATE int HTMIME_abort (HTStream * me, HTList * e)
2.1 timbl 278: {
2.18 frystyk 279: int status = HT_ERROR;
2.41 frystyk 280: if (me->target) status = (*me->target->isa->abort)(me->target, e);
2.26 frystyk 281: if (PROT_TRACE)
2.55 eric 282: HTTrace("MIME........ ABORTING...\n");
2.64 ! eric 283: HTChunk_delete(me->token);
! 284: HTChunk_delete(me->value);
2.52 frystyk 285: HT_FREE(me);
2.18 frystyk 286: return status;
2.1 timbl 287: }
288:
289:
290:
291: /* Structured Object Class
292: ** -----------------------
293: */
2.57 frystyk 294: PRIVATE const HTStreamClass HTMIME =
2.1 timbl 295: {
296: "MIMEParser",
2.18 frystyk 297: HTMIME_flush,
2.1 timbl 298: HTMIME_free,
2.6 timbl 299: HTMIME_abort,
300: HTMIME_put_character,
301: HTMIME_put_string,
2.18 frystyk 302: HTMIME_put_block
2.1 timbl 303: };
304:
305:
2.48 frystyk 306: /* MIME header parser stream.
2.1 timbl 307: ** -------------------------
2.48 frystyk 308: ** This stream parses a complete MIME header and if a content type header
309: ** is found then the stream stack is called. Any left over data is pumped
310: ** right through the stream
2.1 timbl 311: */
2.36 frystyk 312: PUBLIC HTStream* HTMIMEConvert (HTRequest * request,
313: void * param,
314: HTFormat input_format,
315: HTFormat output_format,
316: HTStream * output_stream)
2.1 timbl 317: {
2.62 frystyk 318: HTStream * me;
2.52 frystyk 319: if ((me = (HTStream *) HT_CALLOC(1, sizeof(* me))) == NULL)
320: HT_OUTOFMEM("HTMIMEConvert");
2.1 timbl 321: me->isa = &HTMIME;
2.18 frystyk 322: me->request = request;
2.32 frystyk 323: me->anchor = request->anchor;
324: me->net = request->net;
2.49 frystyk 325: me->target = output_stream;
2.18 frystyk 326: me->target_format = output_format;
2.64 ! eric 327: me->token = HTChunk_new(256);
! 328: me->value = HTChunk_new(256);
! 329: me->hash = 0;
2.18 frystyk 330: me->EOLstate = EOL_BEGIN;
2.64 ! eric 331: me->haveToken = NO;
2.1 timbl 332: return me;
333: }
2.32 frystyk 334:
2.48 frystyk 335: /* MIME header ONLY parser stream
336: ** ------------------------------
337: ** This stream parses a complete MIME header and then returnes HT_PAUSE.
338: ** It does not set up any streams and resting data stays in the buffer.
339: ** This can be used if you only want to parse the headers before you
340: ** decide what to do next. This is for example the case in a server app.
341: */
342: PUBLIC HTStream * HTMIMEHeader (HTRequest * request,
343: void * param,
344: HTFormat input_format,
345: HTFormat output_format,
346: HTStream * output_stream)
347: {
2.62 frystyk 348: HTStream * me = HTMIMEConvert(request, param, input_format,
349: output_format, output_stream);
350: me->head_only = YES;
2.48 frystyk 351: return me;
352: }
2.62 frystyk 353:
354: /* MIME footer ONLY parser stream
355: ** ------------------------------
356: ** Parse only a footer, for example after a chunked encoding.
357: */
358: PUBLIC HTStream * HTMIMEFooter (HTRequest * request,
359: void * param,
360: HTFormat input_format,
361: HTFormat output_format,
362: HTStream * output_stream)
363: {
364: HTStream * me = HTMIMEConvert(request, param, input_format,
365: output_format, output_stream);
366: me->footer = YES;
367: return me;
368: }
Webmaster