Annotation of libwww/Library/src/SGML.c, revision 1.29
1.23 frystyk 1: /* SGML.c
2: ** GENERAL SGML PARSER CODE
3: **
1.27 frystyk 4: ** (c) COPYRIGHT MIT 1995.
1.23 frystyk 5: ** Please first read the full copyright statement in the file COPYRIGH.
1.1 timbl 6: **
1.2 timbl 7: ** This module implements an HTStream object. To parse an
1.1 timbl 8: ** SGML file, create this object which is a parser. The object
1.2 timbl 9: ** is (currently) created by being passed a DTD structure,
10: ** and a target HTStructured oject at which to throw the parsed stuff.
1.1 timbl 11: **
1.19 duns 12: ** 6 Feb 93 Binary seraches used. Intreface modified.
13: ** 8 Jul 94 FM Insulate free() from _free structure element.
1.1 timbl 14: */
15:
1.25 frystyk 16: /* Library include files */
17: #include "tcp.h"
1.1 timbl 18: #include "HTUtils.h"
1.25 frystyk 19: #include "HTString.h"
1.1 timbl 20: #include "HTChunk.h"
1.20 frystyk 21: #include "SGML.h"
1.1 timbl 22:
1.2 timbl 23: #define INVALID (-1)
24:
1.1 timbl 25: /* The State (context) of the parser
26: **
1.2 timbl 27: ** This is passed with each call to make the parser reentrant
1.1 timbl 28: **
29: */
30:
1.16 frystyk 31:
1.2 timbl 32:
33:
34: /* Element Stack
35: ** -------------
36: ** This allows us to return down the stack reselcting styles.
37: ** As we return, attribute values will be garbage in general.
38: */
39: typedef struct _HTElement HTElement;
40: struct _HTElement {
41: HTElement * next; /* Previously nested element or 0 */
42: HTTag* tag; /* The tag at this level */
43: };
44:
45:
1.21 frystyk 46: typedef enum _sgml_state {
47: S_text, S_literal, S_tag, S_tag_gap,
48: S_attr, S_attr_gap, S_equals, S_value, S_after_open,
49: S_nl, S_nl_tago,
50: S_ero, S_cro,
51: #ifdef ISO_2022_JP
52: S_esc, S_dollar, S_paren, S_nonascii_text,
53: #endif
54: S_squoted, S_dquoted, S_end, S_entity, S_junk_tag
55: } sgml_state;
56:
57:
1.2 timbl 58: /* Internal Context Data Structure
59: ** -------------------------------
60: */
61: struct _HTStream {
62:
63: CONST HTStreamClass * isa; /* inherited from HTStream */
64:
65: CONST SGML_dtd *dtd;
66: HTStructuredClass *actions; /* target class */
67: HTStructured *target; /* target object */
68:
1.1 timbl 69: HTTag *current_tag;
1.2 timbl 70: int current_attribute_number;
1.1 timbl 71: HTChunk *string;
72: HTElement *element_stack;
1.21 frystyk 73: sgml_state state;
1.2 timbl 74: #ifdef CALLERDATA
1.1 timbl 75: void * callerData;
1.2 timbl 76: #endif
77: BOOL present[MAX_ATTRIBUTES]; /* Flags: attribute is present? */
78: char * value[MAX_ATTRIBUTES]; /* malloc'd strings or NULL if none */
79: } ;
80:
81:
82: #define PUTC(ch) ((*context->actions->put_character)(context->target, ch))
83:
1.1 timbl 84:
1.17 timbl 85: /* Find Attribute Number
86: ** ---------------------
87: */
88:
89: PUBLIC int SGMLFindAttribute ARGS2 (HTTag*, tag, CONST char *, s)
90: {
91: attr* attributes = tag->attributes;
92:
93: int high, low, i, diff; /* Binary search for attribute name */
94: for(low=0, high=tag->number_of_attributes;
95: high > low ;
96: diff < 0 ? (low = i+1) : (high = i) ) {
97: i = (low + (high-low)/2);
98: diff = strcasecomp(attributes[i].name, s);
99: if (diff==0) return i; /* success: found it */
100: } /* for */
101:
102: return -1;
103: }
104:
1.1 timbl 105:
106: /* Handle Attribute
107: ** ----------------
108: */
109: /* PUBLIC CONST char * SGML_default = ""; ?? */
110:
1.21 frystyk 111: PRIVATE void handle_attribute_name ARGS2(HTStream *, context, CONST char *, s)
1.1 timbl 112: {
1.2 timbl 113:
114: HTTag * tag = context->current_tag;
115:
1.17 timbl 116: int i = SGMLFindAttribute(tag, s);
117: if (i>=0) {
118: context->current_attribute_number = i;
119: context->present[i] = YES;
120: if (context->value[i]) {
121: free(context->value[i]);
122: context->value[i] = NULL;
123: }
124: return;
125: } /* if */
1.2 timbl 126:
1.20 frystyk 127: if (SGML_TRACE)
1.29 ! frystyk 128: TTYPrint(TDEST, "SGML: Unknown attribute %s for tag %s\n",
1.2 timbl 129: s, context->current_tag->name);
130: context->current_attribute_number = INVALID; /* Invalid */
1.1 timbl 131: }
132:
133:
134: /* Handle attribute value
135: ** ----------------------
136: */
1.21 frystyk 137: PRIVATE void handle_attribute_value ARGS2(HTStream *, context, CONST char *, s)
1.1 timbl 138: {
1.2 timbl 139: if (context->current_attribute_number != INVALID) {
140: StrAllocCopy(context->value[context->current_attribute_number], s);
1.1 timbl 141: } else {
1.29 ! frystyk 142: if (SGML_TRACE) TTYPrint(TDEST, "SGML: Attribute value %s ignored\n", s);
1.1 timbl 143: }
1.2 timbl 144: context->current_attribute_number = INVALID; /* can't have two assignments! */
1.1 timbl 145: }
146:
1.2 timbl 147:
1.1 timbl 148: /* Handle entity
149: ** -------------
150: **
151: ** On entry,
152: ** s contains the entity name zero terminated
153: ** Bugs:
154: ** If the entity name is unknown, the terminator is treated as
155: ** a printable non-special character in all cases, even if it is '<'
156: */
1.21 frystyk 157: PRIVATE void handle_entity ARGS2(HTStream *, context, char, term)
1.1 timbl 158: {
1.2 timbl 159:
1.3 timbl 160: CONST char ** entities = context->dtd->entity_names;
1.1 timbl 161: CONST char *s = context->string->data;
1.2 timbl 162:
163: int high, low, i, diff;
164: for(low=0, high = context->dtd->number_of_entities;
165: high > low ;
166: diff < 0 ? (low = i+1) : (high = i)) { /* Binary serach */
167: i = (low + (high-low)/2);
168: diff = strcmp(entities[i], s); /* Csse sensitive! */
169: if (diff==0) { /* success: found it */
170: (*context->actions->put_entity)(context->target, i);
171: return;
1.1 timbl 172: }
173: }
174: /* If entity string not found, display as text */
1.20 frystyk 175: if (SGML_TRACE)
1.29 ! frystyk 176: TTYPrint(TDEST, "SGML: Unknown entity %s\n", s);
1.2 timbl 177: PUTC('&');
1.1 timbl 178: {
179: CONST char *p;
180: for (p=s; *p; p++) {
1.2 timbl 181: PUTC(*p);
1.1 timbl 182: }
183: }
1.2 timbl 184: PUTC(term);
1.1 timbl 185: }
186:
1.2 timbl 187:
1.1 timbl 188: /* End element
1.2 timbl 189: ** -----------
1.1 timbl 190: */
1.21 frystyk 191: PRIVATE void end_element ARGS2(HTStream *, context, HTTag *, old_tag)
1.1 timbl 192: {
1.29 ! frystyk 193: if (SGML_TRACE) TTYPrint(TDEST, "SGML: End </%s>\n", old_tag->name);
1.2 timbl 194: if (old_tag->contents == SGML_EMPTY) {
1.29 ! frystyk 195: if (SGML_TRACE) TTYPrint(TDEST,"SGML: Illegal end tag </%s> found.\n",
1.1 timbl 196: old_tag->name);
197: return;
198: }
199: while (context->element_stack) {/* Loop is error path only */
200: HTElement * N = context->element_stack;
201: HTTag * t = N->tag;
202:
203: if (old_tag != t) { /* Mismatch: syntax error */
204: if (context->element_stack->next) { /* This is not the last level */
1.29 ! frystyk 205: if (SGML_TRACE) TTYPrint(TDEST,
1.1 timbl 206: "SGML: Found </%s> when expecting </%s>. </%s> assumed.\n",
207: old_tag->name, t->name, t->name);
208: } else { /* last level */
1.29 ! frystyk 209: if (SGML_TRACE) TTYPrint(TDEST,
1.1 timbl 210: "SGML: Found </%s> when expecting </%s>. </%s> Ignored.\n",
211: old_tag->name, t->name, old_tag->name);
212: return; /* Ignore */
213: }
214: }
215:
216: context->element_stack = N->next; /* Remove from stack */
217: free(N);
1.2 timbl 218: (*context->actions->end_element)(context->target,
219: t - context->dtd->tags);
1.1 timbl 220: if (old_tag == t) return; /* Correct sequence */
221:
222: /* Syntax error path only */
223:
224: }
1.29 ! frystyk 225: if (SGML_TRACE) TTYPrint(TDEST,
1.1 timbl 226: "SGML: Extra end tag </%s> found and ignored.\n", old_tag->name);
227: }
228:
229:
1.17 timbl 230: /* Start an element
231: ** ----------------
1.1 timbl 232: */
1.21 frystyk 233: PRIVATE void start_element ARGS1(HTStream *, context)
1.1 timbl 234: {
235: HTTag * new_tag = context->current_tag;
236:
1.29 ! frystyk 237: if (SGML_TRACE) TTYPrint(TDEST, "SGML: Start <%s>\n", new_tag->name);
1.2 timbl 238: (*context->actions->start_element)(
239: context->target,
240: new_tag - context->dtd->tags,
241: context->present,
1.3 timbl 242: (CONST char**) context->value); /* coerce type for think c */
1.2 timbl 243: if (new_tag->contents != SGML_EMPTY) { /* i.e. tag not empty */
1.1 timbl 244: HTElement * N = (HTElement *)malloc(sizeof(HTElement));
245: if (N == NULL) outofmem(__FILE__, "start_element");
246: N->next = context->element_stack;
247: N->tag = new_tag;
248: context->element_stack = N;
249: }
250: }
251:
252:
1.2 timbl 253: /* Find Tag in DTD tag list
254: ** ------------------------
1.1 timbl 255: **
256: ** On entry,
1.2 timbl 257: ** dtd points to dtd structire including valid tag list
258: ** string points to name of tag in question
1.1 timbl 259: **
1.2 timbl 260: ** On exit,
261: ** returns:
1.7 timbl 262: ** NULL tag not found
263: ** else address of tag structure in dtd
1.2 timbl 264: */
1.11 timbl 265: PUBLIC HTTag * SGMLFindTag ARGS2(CONST SGML_dtd*, dtd, CONST char *, string)
1.2 timbl 266: {
267: int high, low, i, diff;
268: for(low=0, high=dtd->number_of_tags;
269: high > low ;
270: diff < 0 ? (low = i+1) : (high = i)) { /* Binary serach */
271: i = (low + (high-low)/2);
1.3 timbl 272: diff = strcasecomp(dtd->tags[i].name, string); /* Case insensitive */
1.2 timbl 273: if (diff==0) { /* success: found it */
1.7 timbl 274: return &dtd->tags[i];
1.2 timbl 275: }
276: }
1.7 timbl 277: return NULL;
1.2 timbl 278: }
279:
280: /*________________________________________________________________________
281: ** Public Methods
1.1 timbl 282: */
283:
1.2 timbl 284:
285: /* Could check that we are back to bottom of stack! @@ */
1.26 frystyk 286: PUBLIC int SGML_flush ARGS1(HTStream *, context)
287: {
288: while (context->element_stack) {
289: HTElement *ptr = context->element_stack;
290: if (SGML_TRACE)
1.29 ! frystyk 291: TTYPrint(TDEST, "SGML........ Non-matched tag found: <%s>\n",
1.26 frystyk 292: context->element_stack->tag->name);
293: context->element_stack = ptr->next;
294: free(ptr);
295: }
296: return (*context->actions->flush)(context->target);
297: }
1.1 timbl 298:
1.22 frystyk 299: PUBLIC int SGML_free ARGS1(HTStream *, context)
1.8 timbl 300: {
1.26 frystyk 301: int status;
1.14 frystyk 302: int cnt;
1.15 frystyk 303: while (context->element_stack) { /* Make sure, that all tags are gone */
304: HTElement *ptr = context->element_stack;
305:
1.26 frystyk 306: if (SGML_TRACE)
1.29 ! frystyk 307: TTYPrint(TDEST, "SGML........ Non-matched tag found: <%s>\n",
1.26 frystyk 308: context->element_stack->tag->name);
1.15 frystyk 309: context->element_stack = ptr->next;
310: free(ptr);
311: }
1.26 frystyk 312: if ((status = (*context->actions->_free)(context->target)) != HT_OK)
313: return status;
1.8 timbl 314: HTChunkFree(context->string);
1.15 frystyk 315: for(cnt=0; cnt<MAX_ATTRIBUTES; cnt++) /* Leak fix Henrik 18/02-94 */
1.14 frystyk 316: if(context->value[cnt])
317: free(context->value[cnt]);
1.8 timbl 318: free(context);
1.26 frystyk 319: return HT_OK;
1.1 timbl 320: }
321:
1.22 frystyk 322: PUBLIC int SGML_abort ARGS2(HTStream *, context, HTError, e)
1.1 timbl 323: {
1.14 frystyk 324: int cnt;
1.15 frystyk 325: while (context->element_stack) { /* Make sure, that all tags are gone */
326: HTElement *ptr = context->element_stack;
1.26 frystyk 327: if (SGML_TRACE)
1.29 ! frystyk 328: TTYPrint(TDEST, "SGML........ Non-matched tag found: <%s>\n",
1.26 frystyk 329: context->element_stack->tag->name);
1.15 frystyk 330: context->element_stack = ptr->next;
331: free(ptr);
332: }
1.8 timbl 333: (*context->actions->abort)(context->target, e);
1.1 timbl 334: HTChunkFree(context->string);
1.14 frystyk 335: for(cnt=0; cnt<MAX_ATTRIBUTES; cnt++) /* Leak fix Henrik 18/02-94 */
336: if(context->value[cnt])
337: free(context->value[cnt]);
1.1 timbl 338: free(context);
1.26 frystyk 339: return HT_ERROR;
1.1 timbl 340: }
341:
1.2 timbl 342:
1.1 timbl 343: /* Read and write user callback handle
344: ** -----------------------------------
345: **
346: ** The callbacks from the SGML parser have an SGML context parameter.
347: ** These calls allow the caller to associate his own context with a
348: ** particular SGML context.
349: */
350:
1.2 timbl 351: #ifdef CALLERDATA
352: PUBLIC void* SGML_callerData ARGS1(HTStream *, context)
1.1 timbl 353: {
354: return context->callerData;
355: }
356:
1.2 timbl 357: PUBLIC void SGML_setCallerData ARGS2(HTStream *, context, void*, data)
1.1 timbl 358: {
359: context->callerData = data;
360: }
1.2 timbl 361: #endif
1.1 timbl 362:
1.26 frystyk 363: PUBLIC int SGML_character ARGS2(HTStream *, context, char,c)
1.1 timbl 364:
365: {
1.2 timbl 366: CONST SGML_dtd *dtd = context->dtd;
1.1 timbl 367: HTChunk *string = context->string;
368:
369: switch(context->state) {
1.18 timbl 370:
371: case S_after_open: /* Strip one trainling newline
372: only after opening nonempty element. - SGML:Ugh! */
373: if (c=='\n' && (context->current_tag->contents != SGML_EMPTY)) {
374: break;
375: }
376: context->state = S_text;
377: goto normal_text;
378: /* (***falls through***) */
379:
1.1 timbl 380: case S_text:
1.18 timbl 381: normal_text:
382:
1.13 timbl 383: #ifdef ISO_2022_JP
384: if (c=='\033') {
385: context->state = S_esc;
386: PUTC(c);
387: break;
388: }
389: #endif /* ISO_2022_JP */
1.6 timbl 390: if (c=='&' && (!context->element_stack || (
391: context->element_stack->tag &&
392: ( context->element_stack->tag->contents == SGML_MIXED
393: || context->element_stack->tag->contents ==
394: SGML_RCDATA)
395: ))) {
1.1 timbl 396: string->size = 0;
397: context->state = S_ero;
398:
399: } else if (c=='<') {
400: string->size = 0;
401: context->state = (context->element_stack &&
1.13 timbl 402: context->element_stack->tag &&
403: context->element_stack->tag->contents == SGML_LITERAL) ?
1.12 timbl 404: S_literal : S_tag;
1.18 timbl 405: } else if (c=='\n') { /* Newline - ignore if before tag end! */
406: context->state = S_nl;
1.2 timbl 407: } else PUTC(c);
1.1 timbl 408: break;
1.13 timbl 409:
1.18 timbl 410: case S_nl:
411: if (c=='<') {
412: string->size = 0;
413: context->state = (context->element_stack &&
414: context->element_stack->tag &&
415: context->element_stack->tag->contents == SGML_LITERAL) ?
416: S_literal : S_nl_tago;
417: } else {
418: PUTC('\n');
419: context->state = S_text;
420: goto normal_text;
421: }
422: break;
423:
424: case S_nl_tago: /* Had newline and tag opener */
425: if (c != '/') {
426: PUTC('\n'); /* Only ignore newline before </ */
427: }
428: context->state = S_tag;
429: goto handle_S_tag;
430:
1.13 timbl 431: #ifdef ISO_2022_JP
432: case S_esc:
433: if (c=='$') {
434: context->state = S_dollar;
435: } else if (c=='(') {
436: context->state = S_paren;
437: } else {
438: context->state = S_text;
439: }
440: PUTC(c);
441: break;
442: case S_dollar:
443: if (c=='@' || c=='B') {
444: context->state = S_nonascii_text;
445: } else {
446: context->state = S_text;
447: }
448: PUTC(c);
449: break;
450: case S_paren:
451: if (c=='B' || c=='J') {
452: context->state = S_text;
453: } else {
454: context->state = S_text;
455: }
456: PUTC(c);
457: break;
458: case S_nonascii_text:
459: if (c=='\033') {
460: context->state = S_esc;
461: PUTC(c);
462: } else {
463: PUTC(c);
464: }
465: break;
466: #endif /* ISO_2022_JP */
1.1 timbl 467:
1.12 timbl 468: /* In literal mode, waits only for specific end tag!
1.2 timbl 469: ** Only foir compatibility with old servers.
1.1 timbl 470: */
1.12 timbl 471: case S_literal :
1.1 timbl 472: HTChunkPutc(string, c);
473: if ( TOUPPER(c) != ((string->size ==1) ? '/'
474: : context->element_stack->tag->name[string->size-2])) {
475: int i;
476:
1.12 timbl 477: /* If complete match, end literal */
1.1 timbl 478: if ((c=='>') && (!context->element_stack->tag->name[string->size-2])) {
479: end_element(context, context->element_stack->tag);
480: string->size = 0;
1.2 timbl 481: context->current_attribute_number = INVALID;
1.1 timbl 482: context->state = S_text;
483: break;
484: } /* If Mismatch: recover string. */
1.2 timbl 485: PUTC( '<');
1.1 timbl 486: for (i=0; i<string->size; i++) /* recover */
1.2 timbl 487: PUTC(
1.1 timbl 488: string->data[i]);
489: context->state = S_text;
490: }
491:
492: break;
493:
494: /* Character reference or Entity
495: */
496: case S_ero:
497: if (c=='#') {
498: context->state = S_cro; /* &# is Char Ref Open */
499: break;
500: }
501: context->state = S_entity; /* Fall through! */
502:
503: /* Handle Entities
504: */
505: case S_entity:
506: if (isalnum(c))
507: HTChunkPutc(string, c);
508: else {
509: HTChunkTerminate(string);
510: handle_entity(context, c);
511: context->state = S_text;
512: }
513: break;
514:
515: /* Character reference
516: */
517: case S_cro:
518: if (isalnum(c))
519: HTChunkPutc(string, c); /* accumulate a character NUMBER */
520: else {
521: int value;
522: HTChunkTerminate(string);
523: if (sscanf(string->data, "%d", &value)==1)
1.28 frystyk 524: PUTC((char) value);
1.1 timbl 525: context->state = S_text;
526: }
527: break;
528:
529: /* Tag
530: */
531: case S_tag: /* new tag */
1.18 timbl 532: handle_S_tag:
533:
1.1 timbl 534: if (isalnum(c))
535: HTChunkPutc(string, c);
536: else { /* End of tag name */
1.7 timbl 537: HTTag * t;
1.1 timbl 538: if (c=='/') {
1.20 frystyk 539: if (SGML_TRACE) if (string->size!=0)
1.29 ! frystyk 540: TTYPrint(TDEST,"SGML: `<%s/' found!\n", string->data);
1.1 timbl 541: context->state = S_end;
542: break;
543: }
544: HTChunkTerminate(string) ;
1.2 timbl 545:
1.10 timbl 546: t = SGMLFindTag(dtd, string->data);
1.7 timbl 547: if (!t) {
1.29 ! frystyk 548: if(SGML_TRACE) TTYPrint(TDEST, "SGML: *** Unknown element %s\n",
1.1 timbl 549: string->data);
550: context->state = (c=='>') ? S_text : S_junk_tag;
551: break;
552: }
1.7 timbl 553: context->current_tag = t;
1.2 timbl 554:
555: /* Clear out attributes
556: */
1.1 timbl 557:
1.2 timbl 558: {
559: int i;
560: for (i=0; i< context->current_tag->number_of_attributes; i++)
561: context->present[i] = NO;
1.1 timbl 562: }
563: string->size = 0;
1.2 timbl 564: context->current_attribute_number = INVALID;
1.1 timbl 565:
566: if (c=='>') {
567: if (context->current_tag->name) start_element(context);
1.18 timbl 568: context->state = S_after_open;
1.1 timbl 569: } else {
570: context->state = S_tag_gap;
571: }
572: }
573: break;
574:
575:
576: case S_tag_gap: /* Expecting attribute or > */
577: if (WHITE(c)) break; /* Gap between attributes */
578: if (c=='>') { /* End of tag */
579: if (context->current_tag->name) start_element(context);
1.18 timbl 580: context->state = S_after_open;
1.1 timbl 581: break;
582: }
583: HTChunkPutc(string, c);
584: context->state = S_attr; /* Get attribute */
585: break;
586:
587: /* accumulating value */
588: case S_attr:
589: if (WHITE(c) || (c=='>') || (c=='=')) { /* End of word */
590: HTChunkTerminate(string) ;
591: handle_attribute_name(context, string->data);
592: string->size = 0;
593: if (c=='>') { /* End of tag */
594: if (context->current_tag->name) start_element(context);
1.18 timbl 595: context->state = S_after_open;
1.1 timbl 596: break;
597: }
598: context->state = (c=='=' ? S_equals: S_attr_gap);
599: } else {
600: HTChunkPutc(string, c);
601: }
602: break;
603:
604: case S_attr_gap: /* Expecting attribute or = or > */
605: if (WHITE(c)) break; /* Gap after attribute */
606: if (c=='>') { /* End of tag */
607: if (context->current_tag->name) start_element(context);
1.18 timbl 608: context->state = S_after_open;
1.1 timbl 609: break;
610: } else if (c=='=') {
611: context->state = S_equals;
612: break;
613: }
614: HTChunkPutc(string, c);
615: context->state = S_attr; /* Get next attribute */
616: break;
617:
618: case S_equals: /* After attr = */
619: if (WHITE(c)) break; /* Before attribute value */
620: if (c=='>') { /* End of tag */
1.29 ! frystyk 621: if (SGML_TRACE) TTYPrint(TDEST, "SGML: found = but no value\n");
1.1 timbl 622: if (context->current_tag->name) start_element(context);
1.18 timbl 623: context->state = S_after_open;
1.1 timbl 624: break;
625:
626: } else if (c=='\'') {
627: context->state = S_squoted;
628: break;
629:
630: } else if (c=='"') {
631: context->state = S_dquoted;
632: break;
633: }
634: HTChunkPutc(string, c);
635: context->state = S_value;
636: break;
637:
638: case S_value:
639: if (WHITE(c) || (c=='>')) { /* End of word */
640: HTChunkTerminate(string) ;
641: handle_attribute_value(context, string->data);
642: string->size = 0;
643: if (c=='>') { /* End of tag */
644: if (context->current_tag->name) start_element(context);
1.18 timbl 645: context->state = S_after_open;
1.1 timbl 646: break;
647: }
648: else context->state = S_tag_gap;
649: } else {
650: HTChunkPutc(string, c);
651: }
652: break;
653:
654: case S_squoted: /* Quoted attribute value */
655: if (c=='\'') { /* End of attribute value */
656: HTChunkTerminate(string) ;
657: handle_attribute_value(context, string->data);
658: string->size = 0;
659: context->state = S_tag_gap;
660: } else {
661: HTChunkPutc(string, c);
662: }
663: break;
664:
665: case S_dquoted: /* Quoted attribute value */
666: if (c=='"') { /* End of attribute value */
667: HTChunkTerminate(string) ;
668: handle_attribute_value(context, string->data);
669: string->size = 0;
670: context->state = S_tag_gap;
671: } else {
672: HTChunkPutc(string, c);
673: }
674: break;
675:
676: case S_end: /* </ */
677: if (isalnum(c))
678: HTChunkPutc(string, c);
679: else { /* End of end tag name */
1.7 timbl 680: HTTag * t;
1.1 timbl 681: HTChunkTerminate(string) ;
1.7 timbl 682: if (!*string->data) { /* Empty end tag */
683: t = context->element_stack->tag;
684: } else {
1.10 timbl 685: t = SGMLFindTag(dtd, string->data);
1.1 timbl 686: }
1.7 timbl 687: if (!t) {
1.29 ! frystyk 688: if(SGML_TRACE) TTYPrint(TDEST,
1.1 timbl 689: "Unknown end tag </%s>\n", string->data);
1.2 timbl 690: } else {
1.7 timbl 691: context->current_tag = t;
1.2 timbl 692: end_element( context, context->current_tag);
1.1 timbl 693: }
1.2 timbl 694:
1.1 timbl 695: string->size = 0;
1.2 timbl 696: context->current_attribute_number = INVALID;
1.7 timbl 697: if (c!='>') {
1.20 frystyk 698: if (SGML_TRACE && !WHITE(c))
1.29 ! frystyk 699: TTYPrint(TDEST,"SGML: `</%s%c' found!\n",
1.7 timbl 700: string->data, c);
701: context->state = S_junk_tag;
702: } else {
703: context->state = S_text;
704: }
1.1 timbl 705: }
706: break;
707:
708:
709: case S_junk_tag:
710: if (c=='>') {
711: context->state = S_text;
712: }
713: } /* switch on context->state */
1.26 frystyk 714: return HT_OK;
715: }
1.2 timbl 716:
717:
1.26 frystyk 718: PUBLIC int SGML_string ARGS2(HTStream *, context, CONST char*, s)
1.2 timbl 719: {
1.26 frystyk 720: while (*s)
721: SGML_character(context, *s++);
722: return HT_OK;
1.2 timbl 723: }
724:
725:
1.26 frystyk 726: PUBLIC int SGML_write ARGS3(HTStream *, context, CONST char*, b, int, l)
1.2 timbl 727: {
1.26 frystyk 728: while (l-- > 0)
729: SGML_character(context, *b++);
730: return HT_OK;
1.2 timbl 731: }
732:
733: /*_______________________________________________________________________
734: */
735:
736: /* Structured Object Class
737: ** -----------------------
738: */
739: PUBLIC CONST HTStreamClass SGMLParser =
740: {
741: "SGMLParser",
1.26 frystyk 742: SGML_flush,
1.2 timbl 743: SGML_free,
1.8 timbl 744: SGML_abort,
1.9 timbl 745: SGML_character,
746: SGML_string,
747: SGML_write,
1.2 timbl 748: };
749:
750: /* Create SGML Engine
751: ** ------------------
752: **
753: ** On entry,
754: ** dtd represents the DTD, along with
755: ** actions is the sink for the data as a set of routines.
756: **
757: */
758:
759: PUBLIC HTStream* SGML_new ARGS2(
760: CONST SGML_dtd *, dtd,
761: HTStructured *, target)
762: {
763: int i;
764: HTStream* context = (HTStream *) malloc(sizeof(*context));
765: if (!context) outofmem(__FILE__, "SGML_begin");
766:
767: context->isa = &SGMLParser;
768: context->string = HTChunkCreate(128); /* Grow by this much */
769: context->dtd = dtd;
770: context->target = target;
771: context->actions = (HTStructuredClass*)(((HTStream*)target)->isa);
772: /* Ugh: no OO */
773: context->state = S_text;
774: context->element_stack = 0; /* empty */
775: #ifdef CALLERDATA
776: context->callerData = (void*) callerData;
777: #endif
778: for(i=0; i<MAX_ATTRIBUTES; i++) context->value[i] = 0;
779:
780: return context;
781: }
Webmaster