Annotation of libwww/Library/src/HTML.c, revision 1.73
1.39 frystyk 1: /* HTML.c
2: ** STRUCTURED STREAM TO RICH HYPERTEXT CONVERTER
3: **
1.43 frystyk 4: ** (c) COPYRIGHT MIT 1995.
1.39 frystyk 5: ** Please first read the full copyright statement in the file COPYRIGH.
1.73 ! frystyk 6: ** @(#) $Id: HTML.c,v 1.72 1998/05/04 19:36:55 frystyk Exp $
1.1 timbl 7: **
1.2 timbl 8: ** This generates of a hypertext object. It converts from the
9: ** structured stream interface fro HTMl events into the style-
1.47 frystyk 10: ** oriented iunterface of the HText interface. This module is
1.2 timbl 11: ** only used in clients and shouldnot be linked into servers.
1.1 timbl 12: **
1.6 timbl 13: ** Override this module if making a new GUI browser.
1.1 timbl 14: **
1.35 duns 15: ** HISTORY:
16: ** 8 Jul 94 FM Insulate free() from _free structure element.
17: **
1.1 timbl 18: */
1.16 timbl 19:
1.41 frystyk 20: /* Library include files */
1.72 frystyk 21: #include "wwwsys.h"
1.63 frystyk 22: #include "WWWUtil.h"
23: #include "WWWCore.h"
24: #include "WWWHTML.h"
1.1 timbl 25: #include "HText.h"
1.73 ! frystyk 26: #include "HTML.h"
1.1 timbl 27: #include "HTStyle.h"
1.73 ! frystyk 28:
1.1 timbl 29:
30: extern HTStyleSheet * styleSheet; /* Application-wide */
31:
32: /* Module-wide style cache
33: */
34: PRIVATE int got_styles = 0;
1.16 timbl 35: PRIVATE HTStyle *styles[HTMLP_ELEMENTS];
1.2 timbl 36: PRIVATE HTStyle *default_style;
1.1 timbl 37:
1.71 frystyk 38: #define HTTAB '\0'
1.1 timbl 39:
40: /* HTML Object
41: ** -----------
42: */
1.2 timbl 43: #define MAX_NESTING 20 /* Should be checked by parser */
44:
45: typedef struct _stack_element {
46: HTStyle * style;
47: int tag_number;
48: } stack_element;
49:
50: struct _HTStructured {
1.60 frystyk 51: const HTStructuredClass * isa;
1.54 frystyk 52: HTRequest * request;
1.2 timbl 53: HTParentAnchor * node_anchor;
54: HText * text;
55:
56: HTStream* target; /* Output stream */
57: HTStreamClass targetClass; /* Output routines */
58:
1.56 frystyk 59: HTChunk * title; /* Grow by 128 */
1.2 timbl 60:
61: char * comment_start; /* for literate programming */
62: char * comment_end;
1.16 timbl 63:
1.60 frystyk 64: const SGML_dtd* dtd;
1.16 timbl 65:
1.2 timbl 66: HTTag * current_tag;
67: BOOL style_change;
68: HTStyle * new_style;
69: HTStyle * old_style;
70: BOOL in_word; /* Have just had a non-white char */
1.44 frystyk 71:
72: stack_element stack[MAX_NESTING];
73: stack_element *sp; /* Style stack pointer */
74: int overflow; /* Keep track of overflow nesting */
1.1 timbl 75: };
76:
1.2 timbl 77: struct _HTStream {
1.60 frystyk 78: const HTStreamClass * isa;
1.2 timbl 79: /* .... */
80: };
1.1 timbl 81:
82: /* Forward declarations of routines
83: */
1.52 frystyk 84: PRIVATE void get_styles (void);
1.1 timbl 85:
1.52 frystyk 86: PRIVATE void actually_set_style (HTStructured * me);
87: PRIVATE void change_paragraph_style (HTStructured * me, HTStyle * style);
1.1 timbl 88:
89: /* Style buffering avoids dummy paragraph begin/ends.
90: */
1.4 timbl 91: #define UPDATE_STYLE if (me->style_change) { actually_set_style(me); }
1.1 timbl 92:
1.2 timbl 93: /* Entity values -- for ISO Latin 1 local representation
94: **
95: ** This MUST match exactly the table referred to in the DTD!
96: */
97: static char * ISO_Latin1[] = {
98: "\306", /* capital AE diphthong (ligature) */
99: "\301", /* capital A, acute accent */
100: "\302", /* capital A, circumflex accent */
101: "\300", /* capital A, grave accent */
102: "\305", /* capital A, ring */
103: "\303", /* capital A, tilde */
104: "\304", /* capital A, dieresis or umlaut mark */
105: "\307", /* capital C, cedilla */
106: "\320", /* capital Eth, Icelandic */
107: "\311", /* capital E, acute accent */
108: "\312", /* capital E, circumflex accent */
109: "\310", /* capital E, grave accent */
110: "\313", /* capital E, dieresis or umlaut mark */
111: "\315", /* capital I, acute accent */
112: "\316", /* capital I, circumflex accent */
113: "\314", /* capital I, grave accent */
114: "\317", /* capital I, dieresis or umlaut mark */
115: "\321", /* capital N, tilde */
116: "\323", /* capital O, acute accent */
117: "\324", /* capital O, circumflex accent */
118: "\322", /* capital O, grave accent */
119: "\330", /* capital O, slash */
120: "\325", /* capital O, tilde */
121: "\326", /* capital O, dieresis or umlaut mark */
122: "\336", /* capital THORN, Icelandic */
123: "\332", /* capital U, acute accent */
124: "\333", /* capital U, circumflex accent */
125: "\331", /* capital U, grave accent */
126: "\334", /* capital U, dieresis or umlaut mark */
127: "\335", /* capital Y, acute accent */
128: "\341", /* small a, acute accent */
129: "\342", /* small a, circumflex accent */
130: "\346", /* small ae diphthong (ligature) */
131: "\340", /* small a, grave accent */
132: "\046", /* ampersand */
133: "\345", /* small a, ring */
134: "\343", /* small a, tilde */
135: "\344", /* small a, dieresis or umlaut mark */
136: "\347", /* small c, cedilla */
137: "\351", /* small e, acute accent */
138: "\352", /* small e, circumflex accent */
139: "\350", /* small e, grave accent */
140: "\360", /* small eth, Icelandic */
141: "\353", /* small e, dieresis or umlaut mark */
142: "\076", /* greater than */
143: "\355", /* small i, acute accent */
144: "\356", /* small i, circumflex accent */
145: "\354", /* small i, grave accent */
146: "\357", /* small i, dieresis or umlaut mark */
147: "\074", /* less than */
1.62 frystyk 148: "\040", /* non-breaking space */
1.2 timbl 149: "\361", /* small n, tilde */
150: "\363", /* small o, acute accent */
151: "\364", /* small o, circumflex accent */
152: "\362", /* small o, grave accent */
153: "\370", /* small o, slash */
154: "\365", /* small o, tilde */
155: "\366", /* small o, dieresis or umlaut mark */
1.36 frystyk 156: "\042", /* double quote sign - June 94 */
1.2 timbl 157: "\337", /* small sharp s, German (sz ligature) */
158: "\376", /* small thorn, Icelandic */
159: "\372", /* small u, acute accent */
160: "\373", /* small u, circumflex accent */
161: "\371", /* small u, grave accent */
162: "\374", /* small u, dieresis or umlaut mark */
163: "\375", /* small y, acute accent */
164: "\377", /* small y, dieresis or umlaut mark */
1.1 timbl 165: };
166:
1.2 timbl 167:
168: /* Entity values -- for NeXT local representation
169: **
170: ** This MUST match exactly the table referred to in the DTD!
171: **
172: */
173: static char * NeXTCharacters[] = {
174: "\341", /* capital AE diphthong (ligature) */
175: "\202", /* capital A, acute accent */
176: "\203", /* capital A, circumflex accent */
177: "\201", /* capital A, grave accent */
178: "\206", /* capital A, ring */
179: "\204", /* capital A, tilde */
180: "\205", /* capital A, dieresis or umlaut mark */
181: "\207", /* capital C, cedilla */
182: "\220", /* capital Eth, Icelandic */
183: "\211", /* capital E, acute accent */
184: "\212", /* capital E, circumflex accent */
185: "\210", /* capital E, grave accent */
186: "\213", /* capital E, dieresis or umlaut mark */
187: "\215", /* capital I, acute accent */
188: "\216", /* capital I, circumflex accent these are */
189: "\214", /* capital I, grave accent ISO -100 hex */
190: "\217", /* capital I, dieresis or umlaut mark */
191: "\221", /* capital N, tilde */
192: "\223", /* capital O, acute accent */
193: "\224", /* capital O, circumflex accent */
194: "\222", /* capital O, grave accent */
195: "\351", /* capital O, slash 'cept this */
196: "\225", /* capital O, tilde */
197: "\226", /* capital O, dieresis or umlaut mark */
198: "\234", /* capital THORN, Icelandic */
199: "\230", /* capital U, acute accent */
200: "\231", /* capital U, circumflex accent */
201: "\227", /* capital U, grave accent */
202: "\232", /* capital U, dieresis or umlaut mark */
203: "\233", /* capital Y, acute accent */
204: "\326", /* small a, acute accent */
205: "\327", /* small a, circumflex accent */
206: "\361", /* small ae diphthong (ligature) */
207: "\325", /* small a, grave accent */
208: "\046", /* ampersand */
209: "\332", /* small a, ring */
210: "\330", /* small a, tilde */
211: "\331", /* small a, dieresis or umlaut mark */
212: "\333", /* small c, cedilla */
213: "\335", /* small e, acute accent */
214: "\336", /* small e, circumflex accent */
215: "\334", /* small e, grave accent */
216: "\346", /* small eth, Icelandic */
217: "\337", /* small e, dieresis or umlaut mark */
218: "\076", /* greater than */
219: "\342", /* small i, acute accent */
220: "\344", /* small i, circumflex accent */
221: "\340", /* small i, grave accent */
222: "\345", /* small i, dieresis or umlaut mark */
223: "\074", /* less than */
1.62 frystyk 224: "\040", /* non-breaking space */
1.2 timbl 225: "\347", /* small n, tilde */
226: "\355", /* small o, acute accent */
227: "\356", /* small o, circumflex accent */
228: "\354", /* small o, grave accent */
229: "\371", /* small o, slash */
230: "\357", /* small o, tilde */
231: "\360", /* small o, dieresis or umlaut mark */
1.36 frystyk 232: "\042", /* double quote sign - June 94 */
1.2 timbl 233: "\373", /* small sharp s, German (sz ligature) */
234: "\374", /* small thorn, Icelandic */
235: "\363", /* small u, acute accent */
236: "\364", /* small u, circumflex accent */
237: "\362", /* small u, grave accent */
238: "\366", /* small u, dieresis or umlaut mark */
239: "\367", /* small y, acute accent */
240: "\375", /* small y, dieresis or umlaut mark */
1.1 timbl 241: };
242:
1.2 timbl 243: /* Entity values -- for IBM/PC Code Page 850 (International)
244: **
245: ** This MUST match exactly the table referred to in the DTD!
246: **
247: */
248: /* @@@@@@@@@@@@@@@@@ TBD */
249:
250:
251:
252: /* Set character set
253: ** ----------------
254: */
255:
256: PRIVATE char** p_entity_values = ISO_Latin1; /* Pointer to translation */
1.1 timbl 257:
1.53 frystyk 258: PUBLIC void HTMLUseCharacterSet (HTMLCharacterSet i)
1.2 timbl 259: {
260: p_entity_values = (i == HTML_NEXT_CHARS) ? NeXTCharacters
261: : ISO_Latin1;
262: }
1.1 timbl 263:
264:
265: /* Flattening the style structure
266: ** ------------------------------
267: **
268: On the NeXT, and on any read-only browser, it is simpler for the text to have
269: a sequence of styles, rather than a nested tree of styles. In this
270: case we have to flatten the structure as it arrives from SGML tags into
271: a sequence of styles.
272: */
273:
274: /* If style really needs to be set, call this
275: */
1.53 frystyk 276: PRIVATE void actually_set_style (HTStructured * me)
1.1 timbl 277: {
1.4 timbl 278: if (!me->text) { /* First time through */
1.54 frystyk 279: me->text = HText_new2(me->request, me->node_anchor, me->target);
1.4 timbl 280: HText_beginAppend(me->text);
281: HText_setStyle(me->text, me->new_style);
282: me->in_word = NO;
1.1 timbl 283: } else {
1.4 timbl 284: HText_setStyle(me->text, me->new_style);
1.1 timbl 285: }
1.4 timbl 286: me->old_style = me->new_style;
287: me->style_change = NO;
1.1 timbl 288: }
289:
290: /* If you THINK you need to change style, call this
291: */
292:
1.53 frystyk 293: PRIVATE void change_paragraph_style (HTStructured * me, HTStyle *style)
1.1 timbl 294: {
1.4 timbl 295: if (me->new_style!=style) {
296: me->style_change = YES;
297: me->new_style = style;
1.1 timbl 298: }
1.11 timbl 299: me->in_word = NO;
1.1 timbl 300: }
301:
1.2 timbl 302: /*_________________________________________________________________________
303: **
304: ** A C T I O N R O U T I N E S
305: */
306:
1.71 frystyk 307: PRIVATE int HTML_write (HTStructured * me, const char * b, int l)
1.1 timbl 308: {
1.71 frystyk 309: while (l-- > 0) {
310: const char c = *b++;
311: switch (me->sp[0].tag_number) {
312: case HTML_COMMENT:
313: break; /* Do Nothing */
314:
315: case HTML_TITLE:
316: HTChunk_putb(me->title, &c, 1);
317: break;
318:
319: case HTML_LISTING: /* Litteral text */
320: case HTML_XMP:
321: case HTML_PLAINTEXT:
322: case HTML_PRE:
323: /* We guarrantee that the style is up-to-date in begin_litteral */
324: HText_appendCharacter(me->text, c);
325: break;
1.2 timbl 326:
1.71 frystyk 327: default: /* Free format text */
328: if (me->style_change) {
329: if ((c=='\n') || (c==' ')) return HT_OK; /* Ignore it */
330: UPDATE_STYLE;
331: }
332: if (c == HTTAB)
333: HText_appendCharacter(me->text, '\t');
334: else if (isspace((int) c)) {
335: if (me->in_word) {
336: HText_appendCharacter(me->text, ' ');
337: me->in_word = NO;
338: }
339: } else {
340: HText_appendCharacter(me->text, c);
341: me->in_word = YES;
1.2 timbl 342: }
343: }
1.71 frystyk 344: }
1.42 frystyk 345: return HT_OK;
1.1 timbl 346: }
347:
1.71 frystyk 348: PRIVATE int HTML_put_character (HTStructured * me, char c)
1.1 timbl 349: {
1.71 frystyk 350: return HTML_write(me, &c, sizeof(char));
1.1 timbl 351: }
352:
1.71 frystyk 353:
1.64 frystyk 354: PRIVATE int HTML_put_string (HTStructured * me, const char* s)
1.1 timbl 355: {
1.71 frystyk 356: return HTML_write(me, s, (int) strlen(s));
1.1 timbl 357: }
1.2 timbl 358:
359: /* Start Element
360: ** -------------
361: */
1.53 frystyk 362: PRIVATE void HTML_start_element (
363: HTStructured * me,
364: int element_number,
1.60 frystyk 365: const BOOL* present,
366: const char ** value)
1.2 timbl 367: {
368: switch (element_number) {
1.73 ! frystyk 369:
! 370: case HTML_FRAME:
! 371: {
! 372: UPDATE_STYLE;
! 373: HText_appendObject(me->text, HTML_FRAME, present, value);
! 374: }
! 375: break;
! 376:
1.2 timbl 377: case HTML_A:
1.69 frystyk 378: {
379: HTChildAnchor * source = HTAnchor_findChildAndLink(
380: me->node_anchor, /* parent */
381: present[HTML_A_NAME] ? value[HTML_A_NAME] : NULL, /* Tag */
382: present[HTML_A_HREF] ? value[HTML_A_HREF] : NULL, /* Addresss */
383: present[HTML_A_REL] && value[HTML_A_REL] ?
384: (HTLinkType) HTAtom_caseFor(value[HTML_A_REL]) : NULL);
1.2 timbl 385:
1.69 frystyk 386: if (present[HTML_A_TITLE] && value[HTML_A_TITLE]) {
387: HTLink * link = HTAnchor_mainLink((HTAnchor *) source);
388: HTParentAnchor * dest = HTAnchor_parent(HTLink_destination(link));
389: if (!HTAnchor_title(dest)) HTAnchor_setTitle(dest, value[HTML_A_TITLE]);
1.2 timbl 390: }
1.69 frystyk 391: UPDATE_STYLE;
392: HText_beginAnchor(me->text, source);
393: }
394: break;
1.2 timbl 395:
1.63 frystyk 396: case HTML_LINK:
1.69 frystyk 397: {
398: if (present[HTML_LINK_HREF] && value[HTML_LINK_HREF]) {
1.73 ! frystyk 399: HTChildAnchor * source = HTAnchor_findChildAndLink(
! 400: me->node_anchor, /* parent */
! 401: present[HTML_A_NAME] ? value[HTML_A_NAME] : NULL, /* Tag */
! 402: present[HTML_A_HREF] ? value[HTML_A_HREF] : NULL, /* Addresss */
! 403: NULL); /* Rels */
! 404: HTParentAnchor * dest =
! 405: HTAnchor_parent(HTAnchor_followMainLink((HTAnchor *) source));
1.69 frystyk 406:
407: /* If forward reference */
408: if ((present[HTML_LINK_REL] && value[HTML_LINK_REL])) {
409: char * strval = NULL;
410: char * ptr = NULL;
411: char * relation = NULL;
412: StrAllocCopy(strval, value[HTML_LINK_REL]);
413: ptr = strval;
414: while ((relation = HTNextLWSToken(&ptr)) != NULL) {
415: HTLink_add((HTAnchor *) me->node_anchor, (HTAnchor *) dest,
416: (HTLinkType) HTAtom_caseFor(relation),
417: METHOD_INVALID);
418: }
419: HT_FREE(strval);
420: }
421:
422: /* If reverse reference */
423: if ((present[HTML_LINK_REV] && value[HTML_LINK_REV])) {
424: char * strval = NULL;
425: char * ptr = NULL;
426: char * relation = NULL;
427: StrAllocCopy(strval, value[HTML_LINK_REV]);
428: ptr = strval;
429: while ((relation = HTNextLWSToken(&ptr)) != NULL) {
430: HTLink_add((HTAnchor *) dest, (HTAnchor *) me->node_anchor,
431: (HTLinkType) HTAtom_caseFor(relation),
432: METHOD_INVALID);
433: }
434: HT_FREE(strval);
435: }
1.63 frystyk 436:
1.69 frystyk 437: /* If we got any type information as well */
438: if (present[HTML_LINK_TYPE] && value[HTML_LINK_TYPE]) {
439: if (HTAnchor_format(dest) == WWW_UNKNOWN)
440: HTAnchor_setFormat(dest,
441: (HTFormat) HTAtom_caseFor(value[HTML_LINK_TYPE]));
442: }
1.63 frystyk 443:
1.73 ! frystyk 444: UPDATE_STYLE;
! 445: HText_appendLink(me->text, source, present, value);
1.70 frystyk 446: }
1.73 ! frystyk 447:
1.70 frystyk 448: }
449: break;
450:
451: case HTML_META:
452: {
453: /*
454: ** We don't handle HTTP-EQUIV here - only "NAME". It shouldn't be
455: ** a problem, though :)
456: */
457: if (present[HTML_META_NAME] && value[HTML_META_NAME]) {
458: HTAnchor_addMeta (me->node_anchor,
459: value[HTML_META_NAME],
460: (present[HTML_META_CONTENT] && value[HTML_META_CONTENT]) ?
461: value[HTML_META_CONTENT] : "");
1.69 frystyk 462: }
463: }
464: break;
1.63 frystyk 465:
1.2 timbl 466: case HTML_TITLE:
1.56 frystyk 467: HTChunk_clear(me->title);
1.2 timbl 468: break;
469:
470: case HTML_NEXTID:
471: /* if (present[NEXTID_N] && value[NEXTID_N])
1.4 timbl 472: HText_setNextId(me->text, atoi(value[NEXTID_N])); */
1.2 timbl 473: break;
474:
475: case HTML_ISINDEX:
1.4 timbl 476: HTAnchor_setIndex(me->node_anchor);
1.2 timbl 477: break;
478:
1.15 timbl 479: case HTML_BR:
480: UPDATE_STYLE;
481: HText_appendCharacter(me->text, '\n');
482: me->in_word = NO;
483: break;
484:
485: case HTML_HR:
486: UPDATE_STYLE;
487: HText_appendCharacter(me->text, '\n');
1.16 timbl 488: HText_appendText(me->text, "___________________________________");
1.15 timbl 489: HText_appendCharacter(me->text, '\n');
490: me->in_word = NO;
491: break;
492:
1.2 timbl 493: case HTML_P:
494: UPDATE_STYLE;
1.4 timbl 495: HText_appendParagraph(me->text);
496: me->in_word = NO;
1.2 timbl 497: break;
498:
499: case HTML_DL:
1.11 timbl 500: change_paragraph_style(me, present && present[DL_COMPACT]
1.16 timbl 501: ? styles[HTML_DL]
1.2 timbl 502: : styles[HTML_DL]);
503: break;
504:
505: case HTML_DT:
1.4 timbl 506: if (!me->style_change) {
507: HText_appendParagraph(me->text);
508: me->in_word = NO;
1.2 timbl 509: }
510: break;
511:
512: case HTML_DD:
513: UPDATE_STYLE;
1.71 frystyk 514: HTML_put_character(me, HTTAB); /* Just tab out one stop */
1.4 timbl 515: me->in_word = NO;
516: break;
1.2 timbl 517:
518: case HTML_UL:
519: case HTML_OL:
520: case HTML_MENU:
521: case HTML_DIR:
1.11 timbl 522: change_paragraph_style(me, styles[element_number]);
1.2 timbl 523: break;
524:
525: case HTML_LI:
526: UPDATE_STYLE;
1.7 timbl 527: if (me->sp[0].tag_number != HTML_DIR)
1.4 timbl 528: HText_appendParagraph(me->text);
1.2 timbl 529: else
1.71 frystyk 530: HText_appendCharacter(me->text, HTTAB);
1.4 timbl 531: me->in_word = NO;
1.2 timbl 532: break;
533:
534: case HTML_LISTING: /* Litteral text */
535: case HTML_XMP:
536: case HTML_PLAINTEXT:
537: case HTML_PRE:
1.11 timbl 538: change_paragraph_style(me, styles[element_number]);
1.2 timbl 539: UPDATE_STYLE;
1.4 timbl 540: if (me->comment_end)
541: HText_appendText(me->text, me->comment_end);
1.2 timbl 542: break;
1.11 timbl 543:
1.23 frystyk 544: case HTML_IMG: /* Images */
545: {
546: HTChildAnchor *source;
547: char *src = NULL;
1.49 frystyk 548: if (present[HTML_IMG_SRC])
1.23 frystyk 549: StrAllocCopy(src, value[HTML_IMG_SRC]);
550: source = HTAnchor_findChildAndLink(
551: me->node_anchor, /* parent */
552: 0, /* Tag */
553: src ? src : 0, /* Addresss */
554: 0);
555: UPDATE_STYLE;
556: HText_appendImage(me->text, source,
1.24 frystyk 557: present[HTML_IMG_ALT] ? value[HTML_IMG_ALT] : NULL,
558: present[HTML_IMG_ALIGN] ? value[HTML_IMG_ALIGN] : NULL,
559: present[HTML_IMG_ISMAP] ? YES : NO);
1.58 frystyk 560: HT_FREE(src);
1.24 frystyk 561: }
562: break;
563:
1.73 ! frystyk 564: case HTML_BODY: /* Images in body defn */
! 565:
! 566: UPDATE_STYLE;
! 567: HText_appendObject(me->text, HTML_BODY, present, value);
! 568: break;
! 569:
1.63 frystyk 570: case HTML_BASE: /* Base header */
571: if (present[HTML_BASE_HREF]) {
572: char * base = (char *) value[HTML_BASE_HREF];
573: if (base) {
574: HTAnchor_setBase(me->node_anchor, base);
575: if (SGML_TRACE) HTTrace("HTML Parser. New base `%s\'\n", base);
576: } else {
577: if (SGML_TRACE) HTTrace("HTML Parser. No base found\n");
578: }
579: }
580: break;
581:
1.24 frystyk 582: case HTML_HTML: /* Ignore these altogether */
583: case HTML_HEAD:
1.62 frystyk 584: break;
1.24 frystyk 585:
1.10 timbl 586: case HTML_TT: /* Physical character highlighting */
587: case HTML_B: /* Currently ignored */
588: case HTML_I:
589: case HTML_U:
1.62 frystyk 590: UPDATE_STYLE;
1.65 frystyk 591: #if 0
1.62 frystyk 592: HText_appendCharacter(me->text, '_');
1.65 frystyk 593: #endif
1.62 frystyk 594: me->in_word = NO;
595: break;
1.10 timbl 596:
597: case HTML_EM: /* Logical character highlighting */
598: case HTML_STRONG: /* Currently ignored */
599: case HTML_CODE:
600: case HTML_SAMP:
601: case HTML_KBD:
602: case HTML_VAR:
603: case HTML_DFN:
604: case HTML_CITE:
605: break;
606:
1.11 timbl 607: case HTML_H1: /* paragraph styles */
608: case HTML_H2:
609: case HTML_H3:
610: case HTML_H4:
611: case HTML_H5:
612: case HTML_H6:
613: case HTML_H7:
614: case HTML_ADDRESS:
615: case HTML_BLOCKQUOTE:
616: change_paragraph_style(me, styles[element_number]); /* May be postponed */
1.2 timbl 617: break;
618:
619: } /* end switch */
620:
1.16 timbl 621: if (me->dtd->tags[element_number].contents!= SGML_EMPTY) {
1.13 timbl 622: if (me->sp == me->stack) {
1.44 frystyk 623: if (SGML_TRACE)
1.63 frystyk 624: HTTrace("HTML Parser. Maximum nesting of %d exceded!\n",
1.44 frystyk 625: MAX_NESTING);
626: me->overflow++;
1.12 timbl 627: return;
628: }
1.4 timbl 629: --(me->sp);
630: me->sp[0].style = me->new_style; /* Stack new style */
631: me->sp[0].tag_number = element_number;
1.10 timbl 632: }
1.1 timbl 633: }
1.10 timbl 634:
1.2 timbl 635:
1.1 timbl 636: /* End Element
1.2 timbl 637: ** -----------
1.1 timbl 638: **
1.2 timbl 639: */
640: /* When we end an element, the style must be returned to that
1.1 timbl 641: ** in effect before that element. Note that anchors (etc?)
642: ** don't have an associated style, so that we must scan down the
643: ** stack for an element with a defined style. (In fact, the styles
644: ** should be linked to the whole stack not just the top one.)
645: ** TBL 921119
1.6 timbl 646: **
647: ** We don't turn on "CAREFUL" check because the parser produces
648: ** (internal code errors apart) good nesting. The parser checks
649: ** incoming code errors, not this module.
1.1 timbl 650: */
1.53 frystyk 651: PRIVATE void HTML_end_element (HTStructured * me, int element_number)
1.1 timbl 652: {
1.2 timbl 653: #ifdef CAREFUL /* parser assumed to produce good nesting */
1.4 timbl 654: if (element_number != me->sp[0].tag_number) {
1.59 eric 655: HTTrace("HTMLText: end of element %s when expecting end of %s\n",
1.16 timbl 656: me->dtd->tags[element_number].name,
657: me->dtd->tags[me->sp->tag_number].name);
1.6 timbl 658: /* panic */
1.1 timbl 659: }
1.2 timbl 660: #endif
1.44 frystyk 661:
662: /* HFN, If overflow of nestings, we need to get back to reality */
663: if (me->overflow > 0) {
664: me->overflow--;
665: return;
666: }
667:
1.4 timbl 668: me->sp++; /* Pop state off stack */
1.67 frystyk 669: if (me->sp > me->stack + MAX_NESTING - 1) {
670: if (SGML_TRACE) HTTrace("HTML Parser. Bottom of style stack reached\n");
671: me->sp = me->stack + MAX_NESTING - 1;
672: }
1.44 frystyk 673:
1.2 timbl 674: switch(element_number) {
675:
676: case HTML_A:
677: UPDATE_STYLE;
1.4 timbl 678: HText_endAnchor(me->text);
1.2 timbl 679: break;
680:
681: case HTML_TITLE:
1.56 frystyk 682: HTAnchor_setTitle(me->node_anchor, HTChunk_data(me->title));
1.2 timbl 683: break;
684:
1.62 frystyk 685: case HTML_TT: /* Physical character highlighting */
686: case HTML_B: /* Currently ignored */
687: case HTML_I:
688: case HTML_U:
689: UPDATE_STYLE;
1.66 frystyk 690: #if 0
1.62 frystyk 691: HText_appendCharacter(me->text, '_');
1.66 frystyk 692: #endif
1.62 frystyk 693: break;
1.66 frystyk 694:
695: case HTML_EM: /* Logical character highlighting */
696: case HTML_STRONG: /* Currently ignored */
697: case HTML_CODE:
698: case HTML_SAMP:
699: case HTML_KBD:
700: case HTML_VAR:
701: case HTML_DFN:
702: case HTML_CITE:
703: break;
1.62 frystyk 704:
1.2 timbl 705: case HTML_LISTING: /* Litteral text */
706: case HTML_XMP:
707: case HTML_PLAINTEXT:
708: case HTML_PRE:
1.4 timbl 709: if (me->comment_start)
710: HText_appendText(me->text, me->comment_start);
1.2 timbl 711: /* Fall through */
712:
713: default:
1.44 frystyk 714:
715: /* Often won't really change */
716: change_paragraph_style(me, me->sp->style);
1.2 timbl 717: break;
718:
719: } /* switch */
1.1 timbl 720: }
721:
1.2 timbl 722:
723: /* Expanding entities
724: ** ------------------
725: */
726: /* (In fact, they all shrink!)
1.1 timbl 727: */
1.2 timbl 728:
1.53 frystyk 729: PRIVATE void HTML_put_entity (HTStructured * me, int entity_number)
1.1 timbl 730: {
1.4 timbl 731: HTML_put_string(me, ISO_Latin1[entity_number]); /* @@ Other representations */
1.1 timbl 732: }
1.2 timbl 733:
1.42 frystyk 734: /* Flush an HTML object
735: ** --------------------
736: */
1.53 frystyk 737: PUBLIC int HTML_flush (HTStructured * me)
1.42 frystyk 738: {
739: UPDATE_STYLE; /* Creates empty document here! */
1.57 frystyk 740: if (me->comment_end) HTML_put_string(me,me->comment_end);
741: return me->target ? (*me->targetClass.flush)(me->target) : HT_OK;
1.42 frystyk 742: }
1.2 timbl 743:
744: /* Free an HTML object
745: ** -------------------
746: **
1.4 timbl 747: ** If the document is empty, the text object will not yet exist.
748: So we could in fact abandon creating the document and return
749: an error code. In fact an empty document is an important type
750: of document, so we don't.
751: **
752: ** If non-interactive, everything is freed off. No: crashes -listrefs
1.2 timbl 753: ** Otherwise, the interactive object is left.
754: */
1.53 frystyk 755: PUBLIC int HTML_free (HTStructured * me)
1.1 timbl 756: {
1.4 timbl 757: UPDATE_STYLE; /* Creates empty document here! */
758: if (me->comment_end)
759: HTML_put_string(me,me->comment_end);
760: HText_endAppend(me->text);
761:
762: if (me->target) {
1.35 duns 763: (*me->targetClass._free)(me->target);
1.2 timbl 764: }
1.56 frystyk 765: HTChunk_delete(me->title);
1.58 frystyk 766: HT_FREE(me);
1.42 frystyk 767: return HT_OK;
1.1 timbl 768: }
769:
770:
1.53 frystyk 771: PRIVATE int HTML_abort (HTStructured * me, HTList * e)
1.1 timbl 772:
1.14 timbl 773: {
774: if (me->target) {
775: (*me->targetClass.abort)(me->target, e);
776: }
1.56 frystyk 777: HTChunk_delete(me->title);
1.58 frystyk 778: HT_FREE(me);
1.42 frystyk 779: return HT_ERROR;
1.1 timbl 780: }
781:
1.2 timbl 782:
783: /* Get Styles from style sheet
784: ** ---------------------------
785: */
1.53 frystyk 786: PRIVATE void get_styles (void)
1.1 timbl 787: {
1.2 timbl 788: got_styles = YES;
789:
790: default_style = HTStyleNamed(styleSheet, "Normal");
1.1 timbl 791:
1.2 timbl 792: styles[HTML_H1] = HTStyleNamed(styleSheet, "Heading1");
793: styles[HTML_H2] = HTStyleNamed(styleSheet, "Heading2");
794: styles[HTML_H3] = HTStyleNamed(styleSheet, "Heading3");
795: styles[HTML_H4] = HTStyleNamed(styleSheet, "Heading4");
796: styles[HTML_H5] = HTStyleNamed(styleSheet, "Heading5");
797: styles[HTML_H6] = HTStyleNamed(styleSheet, "Heading6");
798: styles[HTML_H7] = HTStyleNamed(styleSheet, "Heading7");
799:
800: styles[HTML_DL] = HTStyleNamed(styleSheet, "Glossary");
801: styles[HTML_UL] =
802: styles[HTML_OL] = HTStyleNamed(styleSheet, "List");
803: styles[HTML_MENU] = HTStyleNamed(styleSheet, "Menu");
804: styles[HTML_DIR] = HTStyleNamed(styleSheet, "Dir");
1.16 timbl 805: /* styles[HTML_DLC] = HTStyleNamed(styleSheet, "GlossaryCompact"); */
1.2 timbl 806: styles[HTML_ADDRESS]= HTStyleNamed(styleSheet, "Address");
807: styles[HTML_BLOCKQUOTE]= HTStyleNamed(styleSheet, "BlockQuote");
808: styles[HTML_PLAINTEXT] =
809: styles[HTML_XMP] = HTStyleNamed(styleSheet, "Example");
810: styles[HTML_PRE] = HTStyleNamed(styleSheet, "Preformatted");
811: styles[HTML_LISTING] = HTStyleNamed(styleSheet, "Listing");
812: }
813: /* P U B L I C
814: */
815:
816: /* Structured Object Class
817: ** -----------------------
818: */
1.60 frystyk 819: PRIVATE const HTStructuredClass HTMLPresentation = /* As opposed to print etc */
1.2 timbl 820: {
821: "text/html",
1.42 frystyk 822: HTML_flush,
1.2 timbl 823: HTML_free,
1.14 timbl 824: HTML_abort,
1.2 timbl 825: HTML_put_character, HTML_put_string, HTML_write,
826: HTML_start_element, HTML_end_element,
827: HTML_put_entity
828: };
1.1 timbl 829:
1.4 timbl 830:
1.2 timbl 831: /* New Structured Text object
832: ** --------------------------
833: **
1.16 timbl 834: ** The structured stream can generate either presentation,
1.4 timbl 835: ** or plain text, or HTML.
1.1 timbl 836: */
1.53 frystyk 837: PRIVATE HTStructured* HTML_new (HTRequest * request,
838: void * param,
839: HTFormat input_format,
840: HTFormat output_format,
841: HTStream * output_stream)
1.1 timbl 842: {
843:
1.4 timbl 844: HTStructured * me;
845:
1.47 frystyk 846: #if 0
1.16 timbl 847: if (output_format != WWW_PLAINTEXT
848: && output_format != WWW_PRESENT
849: && output_format != HTAtom_for("text/x-c")) {
1.37 frystyk 850: HTStream * intermediate = HTStreamStack(WWW_HTML, output_format,
851: output_stream, request, NO);
1.6 timbl 852: if (intermediate) return HTMLGenerator(intermediate);
1.44 frystyk 853: if (SGML_TRACE)
1.63 frystyk 854: HTTrace("HTML Parser. Can't parse HTML to %s\n",
1.44 frystyk 855: HTAtom_name(output_format));
1.4 timbl 856: exit (-99);
857: }
1.47 frystyk 858: #endif
1.4 timbl 859:
1.58 frystyk 860: if ((me = (HTStructured *) HT_CALLOC(1, sizeof(*me))) == NULL)
861: HT_OUTOFMEM("HTML_new");
1.1 timbl 862:
863: if (!got_styles) get_styles();
864:
1.4 timbl 865: me->isa = &HTMLPresentation;
1.47 frystyk 866: me->dtd = &HTMLP_dtd;
1.54 frystyk 867: me->request = request;
1.48 frystyk 868: me->node_anchor = HTRequest_anchor(request);
1.56 frystyk 869: me->title = HTChunk_new(128);
1.4 timbl 870: me->text = 0;
871: me->style_change = YES; /* Force check leading to text creation */
872: me->new_style = default_style;
873: me->old_style = 0;
874: me->sp = me->stack + MAX_NESTING - 1;
875: me->sp->tag_number = -1; /* INVALID */
876: me->sp->style = default_style; /* INVALID */
1.1 timbl 877:
1.4 timbl 878: me->comment_start = NULL;
879: me->comment_end = NULL;
1.16 timbl 880: me->target = output_stream;
881: if (output_stream) me->targetClass = *output_stream->isa; /* Copy pointers */
1.1 timbl 882:
1.4 timbl 883: return (HTStructured*) me;
1.1 timbl 884: }
885:
886:
1.2 timbl 887: /* HTConverter for HTML to plain text
888: ** ----------------------------------
1.1 timbl 889: **
1.2 timbl 890: ** This will convert from HTML to presentation or plain text.
1.1 timbl 891: */
1.53 frystyk 892: PUBLIC HTStream* HTMLToPlain (
893: HTRequest * request,
894: void * param,
895: HTFormat input_format,
896: HTFormat output_format,
897: HTStream * output_stream)
1.1 timbl 898: {
1.47 frystyk 899: return SGML_new(&HTMLP_dtd, HTML_new(
1.16 timbl 900: request, NULL, input_format, output_format, output_stream));
1.1 timbl 901: }
902:
903:
1.2 timbl 904: /* HTConverter for HTML to C code
905: ** ------------------------------
906: **
1.36 frystyk 907: ** C code is like plain text but all non-preformatted code
1.2 timbl 908: ** is commented out.
909: ** This will convert from HTML to presentation or plain text.
910: */
1.53 frystyk 911: PUBLIC HTStream* HTMLToC (
912: HTRequest * request,
913: void * param,
914: HTFormat input_format,
915: HTFormat output_format,
916: HTStream * output_stream)
1.1 timbl 917: {
1.4 timbl 918:
919: HTStructured * html;
920:
1.36 frystyk 921: (*output_stream->isa->put_string)(output_stream, "/* "); /* Before title */
1.16 timbl 922: html = HTML_new(request, NULL, input_format, output_format, output_stream);
1.45 frystyk 923: html->comment_start = "\n/* ";
1.47 frystyk 924: html->dtd = &HTMLP_dtd;
1.2 timbl 925: html->comment_end = " */\n"; /* Must start in col 1 for cpp */
1.47 frystyk 926: return SGML_new(&HTMLP_dtd, html);
1.1 timbl 927: }
928:
929:
1.2 timbl 930: /* Presenter for HTML
931: ** ------------------
932: **
933: ** This will convert from HTML to presentation or plain text.
934: **
935: ** Override this if you have a windows version
1.1 timbl 936: */
1.2 timbl 937: #ifndef GUI
1.53 frystyk 938: PUBLIC HTStream* HTMLPresent (
939: HTRequest * request,
940: void * param,
941: HTFormat input_format,
942: HTFormat output_format,
943: HTStream * output_stream)
1.1 timbl 944: {
1.47 frystyk 945: return SGML_new(&HTMLP_dtd, HTML_new(
1.16 timbl 946: request, NULL, input_format, output_format, output_stream));
1.1 timbl 947: }
1.2 timbl 948: #endif
1.29 frystyk 949:
Webmaster