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