Annotation of libwww/Library/src/HTML.c, revision 1.69
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.
1.69 ! frystyk 6: ** @(#) $Id: HTML.c,v 1.68 1998/02/01 19:04:14 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 */
1.60 frystyk 21: #include "sysdep.h"
1.63 frystyk 22: #include "WWWUtil.h"
23: #include "WWWCore.h"
24: #include "WWWHTML.h"
1.1 timbl 25: #include "HText.h"
26: #include "HTStyle.h"
1.41 frystyk 27: #include "HTML.h"
1.1 timbl 28:
29: extern HTStyleSheet * styleSheet; /* Application-wide */
30:
31: /* Module-wide style cache
32: */
33: PRIVATE int got_styles = 0;
1.16 timbl 34: PRIVATE HTStyle *styles[HTMLP_ELEMENTS];
1.2 timbl 35: PRIVATE HTStyle *default_style;
1.1 timbl 36:
1.62 frystyk 37: #define TAB '\0'
1.1 timbl 38:
39: /* HTML Object
40: ** -----------
41: */
1.2 timbl 42: #define MAX_NESTING 20 /* Should be checked by parser */
43:
44: typedef struct _stack_element {
45: HTStyle * style;
46: int tag_number;
47: } stack_element;
48:
49: struct _HTStructured {
1.60 frystyk 50: const HTStructuredClass * isa;
1.54 frystyk 51: HTRequest * request;
1.2 timbl 52: HTParentAnchor * node_anchor;
53: HText * text;
54:
55: HTStream* target; /* Output stream */
56: HTStreamClass targetClass; /* Output routines */
57:
1.56 frystyk 58: HTChunk * title; /* Grow by 128 */
1.2 timbl 59:
60: char * comment_start; /* for literate programming */
61: char * comment_end;
1.16 timbl 62:
1.60 frystyk 63: const SGML_dtd* dtd;
1.16 timbl 64:
1.2 timbl 65: HTTag * current_tag;
66: BOOL style_change;
67: HTStyle * new_style;
68: HTStyle * old_style;
69: BOOL in_word; /* Have just had a non-white char */
1.44 frystyk 70:
71: stack_element stack[MAX_NESTING];
72: stack_element *sp; /* Style stack pointer */
73: int overflow; /* Keep track of overflow nesting */
1.1 timbl 74: };
75:
1.2 timbl 76: struct _HTStream {
1.60 frystyk 77: const HTStreamClass * isa;
1.2 timbl 78: /* .... */
79: };
1.1 timbl 80:
81: /* Forward declarations of routines
82: */
1.52 frystyk 83: PRIVATE void get_styles (void);
1.1 timbl 84:
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:
1.64 frystyk 307: PRIVATE int HTML_put_character (HTStructured * me, char c)
1.1 timbl 308: {
1.4 timbl 309: switch (me->sp[0].tag_number) {
1.2 timbl 310: case HTML_COMMENT:
311: break; /* Do Nothing */
312:
313: case HTML_TITLE:
1.56 frystyk 314: HTChunk_putb(me->title, &c, 1);
1.2 timbl 315: break;
316:
317:
318: case HTML_LISTING: /* Litteral text */
319: case HTML_XMP:
320: case HTML_PLAINTEXT:
321: case HTML_PRE:
322: /* We guarrantee that the style is up-to-date in begin_litteral
323: */
1.4 timbl 324: HText_appendCharacter(me->text, c);
1.2 timbl 325: break;
326:
327: default: /* Free format text */
1.4 timbl 328: if (me->style_change) {
1.42 frystyk 329: if ((c=='\n') || (c==' ')) return HT_OK; /* Ignore it */
1.2 timbl 330: UPDATE_STYLE;
331: }
1.62 frystyk 332: if (c == TAB)
333: HText_appendCharacter(me->text, '\t');
1.68 frystyk 334: else if (isspace((int) c)) {
1.4 timbl 335: if (me->in_word) {
336: HText_appendCharacter(me->text, ' ');
337: me->in_word = NO;
1.2 timbl 338: }
339: } else {
1.4 timbl 340: HText_appendCharacter(me->text, c);
341: me->in_word = YES;
1.2 timbl 342: }
343: } /* end switch */
1.42 frystyk 344: return HT_OK;
1.1 timbl 345: }
346:
1.64 frystyk 347:
348: PRIVATE int HTML_write (HTStructured * me, const char * b, int l)
1.1 timbl 349: {
1.64 frystyk 350: while (l-- > 0) HTML_put_character(me, *b++);
351: return HT_OK;
1.1 timbl 352: }
353:
1.64 frystyk 354: PRIVATE int HTML_put_string (HTStructured * me, const char* s)
1.1 timbl 355: {
1.64 frystyk 356: while (*s) HTML_put_character(me, *s++);
357: return HT_OK;
1.1 timbl 358: }
1.2 timbl 359:
360: /* Start Element
361: ** -------------
362: */
1.53 frystyk 363: PRIVATE void HTML_start_element (
364: HTStructured * me,
365: int element_number,
1.60 frystyk 366: const BOOL* present,
367: const char ** value)
1.2 timbl 368: {
369: switch (element_number) {
370: case HTML_A:
1.69 ! frystyk 371: {
! 372: HTChildAnchor * source = HTAnchor_findChildAndLink(
! 373: me->node_anchor, /* parent */
! 374: present[HTML_A_NAME] ? value[HTML_A_NAME] : NULL, /* Tag */
! 375: present[HTML_A_HREF] ? value[HTML_A_HREF] : NULL, /* Addresss */
! 376: present[HTML_A_REL] && value[HTML_A_REL] ?
! 377: (HTLinkType) HTAtom_caseFor(value[HTML_A_REL]) : NULL);
1.2 timbl 378:
1.69 ! frystyk 379: if (present[HTML_A_TITLE] && value[HTML_A_TITLE]) {
! 380: HTLink * link = HTAnchor_mainLink((HTAnchor *) source);
! 381: HTParentAnchor * dest = HTAnchor_parent(HTLink_destination(link));
! 382: if (!HTAnchor_title(dest)) HTAnchor_setTitle(dest, value[HTML_A_TITLE]);
1.2 timbl 383: }
1.69 ! frystyk 384: UPDATE_STYLE;
! 385: HText_beginAnchor(me->text, source);
! 386: }
! 387: break;
1.2 timbl 388:
1.63 frystyk 389: case HTML_LINK:
1.69 ! frystyk 390: {
! 391: if (present[HTML_LINK_HREF] && value[HTML_LINK_HREF]) {
! 392: char * relative_to = HTAnchor_expandedAddress((HTAnchor *) me->node_anchor);
! 393: char * dest_addr = HTParse(value[HTML_LINK_HREF], relative_to, PARSE_ALL);
! 394: HTParentAnchor * dest = HTAnchor_parent(HTAnchor_findAddress(dest_addr));
! 395:
! 396: /* If forward reference */
! 397: if ((present[HTML_LINK_REL] && value[HTML_LINK_REL])) {
! 398: char * strval = NULL;
! 399: char * ptr = NULL;
! 400: char * relation = NULL;
! 401: StrAllocCopy(strval, value[HTML_LINK_REL]);
! 402: ptr = strval;
! 403: while ((relation = HTNextLWSToken(&ptr)) != NULL) {
! 404: HTLink_add((HTAnchor *) me->node_anchor, (HTAnchor *) dest,
! 405: (HTLinkType) HTAtom_caseFor(relation),
! 406: METHOD_INVALID);
! 407: }
! 408: HT_FREE(strval);
! 409: }
! 410:
! 411: /* If reverse reference */
! 412: if ((present[HTML_LINK_REV] && value[HTML_LINK_REV])) {
! 413: char * strval = NULL;
! 414: char * ptr = NULL;
! 415: char * relation = NULL;
! 416: StrAllocCopy(strval, value[HTML_LINK_REV]);
! 417: ptr = strval;
! 418: while ((relation = HTNextLWSToken(&ptr)) != NULL) {
! 419: HTLink_add((HTAnchor *) dest, (HTAnchor *) me->node_anchor,
! 420: (HTLinkType) HTAtom_caseFor(relation),
! 421: METHOD_INVALID);
! 422: }
! 423: HT_FREE(strval);
! 424: }
1.63 frystyk 425:
1.69 ! frystyk 426: /* If we got any type information as well */
! 427: if (present[HTML_LINK_TYPE] && value[HTML_LINK_TYPE]) {
! 428: if (HTAnchor_format(dest) == WWW_UNKNOWN)
! 429: HTAnchor_setFormat(dest,
! 430: (HTFormat) HTAtom_caseFor(value[HTML_LINK_TYPE]));
! 431: }
1.63 frystyk 432:
1.69 ! frystyk 433: HT_FREE(dest_addr);
! 434: HT_FREE(relative_to);
! 435: }
! 436: }
! 437: break;
1.63 frystyk 438:
1.2 timbl 439: case HTML_TITLE:
1.56 frystyk 440: HTChunk_clear(me->title);
1.2 timbl 441: break;
442:
443: case HTML_NEXTID:
444: /* if (present[NEXTID_N] && value[NEXTID_N])
1.4 timbl 445: HText_setNextId(me->text, atoi(value[NEXTID_N])); */
1.2 timbl 446: break;
447:
448: case HTML_ISINDEX:
1.4 timbl 449: HTAnchor_setIndex(me->node_anchor);
1.2 timbl 450: break;
451:
1.15 timbl 452: case HTML_BR:
453: UPDATE_STYLE;
454: HText_appendCharacter(me->text, '\n');
455: me->in_word = NO;
456: break;
457:
458: case HTML_HR:
459: UPDATE_STYLE;
460: HText_appendCharacter(me->text, '\n');
1.16 timbl 461: HText_appendText(me->text, "___________________________________");
1.15 timbl 462: HText_appendCharacter(me->text, '\n');
463: me->in_word = NO;
464: break;
465:
1.2 timbl 466: case HTML_P:
467: UPDATE_STYLE;
1.4 timbl 468: HText_appendParagraph(me->text);
469: me->in_word = NO;
1.2 timbl 470: break;
471:
472: case HTML_DL:
1.11 timbl 473: change_paragraph_style(me, present && present[DL_COMPACT]
1.16 timbl 474: ? styles[HTML_DL]
1.2 timbl 475: : styles[HTML_DL]);
476: break;
477:
478: case HTML_DT:
1.4 timbl 479: if (!me->style_change) {
480: HText_appendParagraph(me->text);
481: me->in_word = NO;
1.2 timbl 482: }
483: break;
484:
485: case HTML_DD:
486: UPDATE_STYLE;
1.62 frystyk 487: HTML_put_character(me, TAB); /* Just tab out one stop */
1.4 timbl 488: me->in_word = NO;
489: break;
1.2 timbl 490:
491: case HTML_UL:
492: case HTML_OL:
493: case HTML_MENU:
494: case HTML_DIR:
1.11 timbl 495: change_paragraph_style(me, styles[element_number]);
1.2 timbl 496: break;
497:
498: case HTML_LI:
499: UPDATE_STYLE;
1.7 timbl 500: if (me->sp[0].tag_number != HTML_DIR)
1.4 timbl 501: HText_appendParagraph(me->text);
1.2 timbl 502: else
1.62 frystyk 503: HText_appendCharacter(me->text, TAB);
1.4 timbl 504: me->in_word = NO;
1.2 timbl 505: break;
506:
507: case HTML_LISTING: /* Litteral text */
508: case HTML_XMP:
509: case HTML_PLAINTEXT:
510: case HTML_PRE:
1.11 timbl 511: change_paragraph_style(me, styles[element_number]);
1.2 timbl 512: UPDATE_STYLE;
1.4 timbl 513: if (me->comment_end)
514: HText_appendText(me->text, me->comment_end);
1.2 timbl 515: break;
1.11 timbl 516:
1.23 frystyk 517: case HTML_IMG: /* Images */
518: {
519: HTChildAnchor *source;
520: char *src = NULL;
1.49 frystyk 521: if (present[HTML_IMG_SRC])
1.23 frystyk 522: StrAllocCopy(src, value[HTML_IMG_SRC]);
523: source = HTAnchor_findChildAndLink(
524: me->node_anchor, /* parent */
525: 0, /* Tag */
526: src ? src : 0, /* Addresss */
527: 0);
528: UPDATE_STYLE;
529: HText_appendImage(me->text, source,
1.24 frystyk 530: present[HTML_IMG_ALT] ? value[HTML_IMG_ALT] : NULL,
531: present[HTML_IMG_ALIGN] ? value[HTML_IMG_ALIGN] : NULL,
532: present[HTML_IMG_ISMAP] ? YES : NO);
1.58 frystyk 533: HT_FREE(src);
1.24 frystyk 534: }
535: break;
536:
1.63 frystyk 537: case HTML_BASE: /* Base header */
538: if (present[HTML_BASE_HREF]) {
539: char * base = (char *) value[HTML_BASE_HREF];
540: if (base) {
541: HTAnchor_setBase(me->node_anchor, base);
542: if (SGML_TRACE) HTTrace("HTML Parser. New base `%s\'\n", base);
543: } else {
544: if (SGML_TRACE) HTTrace("HTML Parser. No base found\n");
545: }
546: }
547: break;
548:
1.24 frystyk 549: case HTML_HTML: /* Ignore these altogether */
550: case HTML_HEAD:
551: case HTML_BODY:
1.62 frystyk 552: break;
1.24 frystyk 553:
1.10 timbl 554: case HTML_TT: /* Physical character highlighting */
555: case HTML_B: /* Currently ignored */
556: case HTML_I:
557: case HTML_U:
1.62 frystyk 558: UPDATE_STYLE;
1.65 frystyk 559: #if 0
1.62 frystyk 560: HText_appendCharacter(me->text, '_');
1.65 frystyk 561: #endif
1.62 frystyk 562: me->in_word = NO;
563: break;
1.10 timbl 564:
565: case HTML_EM: /* Logical character highlighting */
566: case HTML_STRONG: /* Currently ignored */
567: case HTML_CODE:
568: case HTML_SAMP:
569: case HTML_KBD:
570: case HTML_VAR:
571: case HTML_DFN:
572: case HTML_CITE:
573: break;
574:
1.11 timbl 575: case HTML_H1: /* paragraph styles */
576: case HTML_H2:
577: case HTML_H3:
578: case HTML_H4:
579: case HTML_H5:
580: case HTML_H6:
581: case HTML_H7:
582: case HTML_ADDRESS:
583: case HTML_BLOCKQUOTE:
584: change_paragraph_style(me, styles[element_number]); /* May be postponed */
1.2 timbl 585: break;
586:
587: } /* end switch */
588:
1.16 timbl 589: if (me->dtd->tags[element_number].contents!= SGML_EMPTY) {
1.13 timbl 590: if (me->sp == me->stack) {
1.44 frystyk 591: if (SGML_TRACE)
1.63 frystyk 592: HTTrace("HTML Parser. Maximum nesting of %d exceded!\n",
1.44 frystyk 593: MAX_NESTING);
594: me->overflow++;
1.12 timbl 595: return;
596: }
1.4 timbl 597: --(me->sp);
598: me->sp[0].style = me->new_style; /* Stack new style */
599: me->sp[0].tag_number = element_number;
1.10 timbl 600: }
1.1 timbl 601: }
1.10 timbl 602:
1.2 timbl 603:
1.1 timbl 604: /* End Element
1.2 timbl 605: ** -----------
1.1 timbl 606: **
1.2 timbl 607: */
608: /* When we end an element, the style must be returned to that
1.1 timbl 609: ** in effect before that element. Note that anchors (etc?)
610: ** don't have an associated style, so that we must scan down the
611: ** stack for an element with a defined style. (In fact, the styles
612: ** should be linked to the whole stack not just the top one.)
613: ** TBL 921119
1.6 timbl 614: **
615: ** We don't turn on "CAREFUL" check because the parser produces
616: ** (internal code errors apart) good nesting. The parser checks
617: ** incoming code errors, not this module.
1.1 timbl 618: */
1.53 frystyk 619: PRIVATE void HTML_end_element (HTStructured * me, int element_number)
1.1 timbl 620: {
1.2 timbl 621: #ifdef CAREFUL /* parser assumed to produce good nesting */
1.4 timbl 622: if (element_number != me->sp[0].tag_number) {
1.59 eric 623: HTTrace("HTMLText: end of element %s when expecting end of %s\n",
1.16 timbl 624: me->dtd->tags[element_number].name,
625: me->dtd->tags[me->sp->tag_number].name);
1.6 timbl 626: /* panic */
1.1 timbl 627: }
1.2 timbl 628: #endif
1.44 frystyk 629:
630: /* HFN, If overflow of nestings, we need to get back to reality */
631: if (me->overflow > 0) {
632: me->overflow--;
633: return;
634: }
635:
1.4 timbl 636: me->sp++; /* Pop state off stack */
1.67 frystyk 637: if (me->sp > me->stack + MAX_NESTING - 1) {
638: if (SGML_TRACE) HTTrace("HTML Parser. Bottom of style stack reached\n");
639: me->sp = me->stack + MAX_NESTING - 1;
640: }
1.44 frystyk 641:
1.2 timbl 642: switch(element_number) {
643:
644: case HTML_A:
645: UPDATE_STYLE;
1.4 timbl 646: HText_endAnchor(me->text);
1.2 timbl 647: break;
648:
649: case HTML_TITLE:
1.56 frystyk 650: HTAnchor_setTitle(me->node_anchor, HTChunk_data(me->title));
1.2 timbl 651: break;
652:
1.62 frystyk 653: case HTML_TT: /* Physical character highlighting */
654: case HTML_B: /* Currently ignored */
655: case HTML_I:
656: case HTML_U:
657: UPDATE_STYLE;
1.66 frystyk 658: #if 0
1.62 frystyk 659: HText_appendCharacter(me->text, '_');
1.66 frystyk 660: #endif
1.62 frystyk 661: break;
1.66 frystyk 662:
663: case HTML_EM: /* Logical character highlighting */
664: case HTML_STRONG: /* Currently ignored */
665: case HTML_CODE:
666: case HTML_SAMP:
667: case HTML_KBD:
668: case HTML_VAR:
669: case HTML_DFN:
670: case HTML_CITE:
671: break;
1.62 frystyk 672:
1.2 timbl 673: case HTML_LISTING: /* Litteral text */
674: case HTML_XMP:
675: case HTML_PLAINTEXT:
676: case HTML_PRE:
1.4 timbl 677: if (me->comment_start)
678: HText_appendText(me->text, me->comment_start);
1.2 timbl 679: /* Fall through */
680:
681: default:
1.44 frystyk 682:
683: /* Often won't really change */
684: change_paragraph_style(me, me->sp->style);
1.2 timbl 685: break;
686:
687: } /* switch */
1.1 timbl 688: }
689:
1.2 timbl 690:
691: /* Expanding entities
692: ** ------------------
693: */
694: /* (In fact, they all shrink!)
1.1 timbl 695: */
1.2 timbl 696:
1.53 frystyk 697: PRIVATE void HTML_put_entity (HTStructured * me, int entity_number)
1.1 timbl 698: {
1.4 timbl 699: HTML_put_string(me, ISO_Latin1[entity_number]); /* @@ Other representations */
1.1 timbl 700: }
1.2 timbl 701:
1.42 frystyk 702: /* Flush an HTML object
703: ** --------------------
704: */
1.53 frystyk 705: PUBLIC int HTML_flush (HTStructured * me)
1.42 frystyk 706: {
707: UPDATE_STYLE; /* Creates empty document here! */
1.57 frystyk 708: if (me->comment_end) HTML_put_string(me,me->comment_end);
709: return me->target ? (*me->targetClass.flush)(me->target) : HT_OK;
1.42 frystyk 710: }
1.2 timbl 711:
712: /* Free an HTML object
713: ** -------------------
714: **
1.4 timbl 715: ** If the document is empty, the text object will not yet exist.
716: So we could in fact abandon creating the document and return
717: an error code. In fact an empty document is an important type
718: of document, so we don't.
719: **
720: ** If non-interactive, everything is freed off. No: crashes -listrefs
1.2 timbl 721: ** Otherwise, the interactive object is left.
722: */
1.53 frystyk 723: PUBLIC int HTML_free (HTStructured * me)
1.1 timbl 724: {
1.4 timbl 725: UPDATE_STYLE; /* Creates empty document here! */
726: if (me->comment_end)
727: HTML_put_string(me,me->comment_end);
728: HText_endAppend(me->text);
729:
730: if (me->target) {
1.35 duns 731: (*me->targetClass._free)(me->target);
1.2 timbl 732: }
1.56 frystyk 733: HTChunk_delete(me->title);
1.58 frystyk 734: HT_FREE(me);
1.42 frystyk 735: return HT_OK;
1.1 timbl 736: }
737:
738:
1.53 frystyk 739: PRIVATE int HTML_abort (HTStructured * me, HTList * e)
1.1 timbl 740:
1.14 timbl 741: {
742: if (me->target) {
743: (*me->targetClass.abort)(me->target, e);
744: }
1.56 frystyk 745: HTChunk_delete(me->title);
1.58 frystyk 746: HT_FREE(me);
1.42 frystyk 747: return HT_ERROR;
1.1 timbl 748: }
749:
1.2 timbl 750:
751: /* Get Styles from style sheet
752: ** ---------------------------
753: */
1.53 frystyk 754: PRIVATE void get_styles (void)
1.1 timbl 755: {
1.2 timbl 756: got_styles = YES;
757:
758: default_style = HTStyleNamed(styleSheet, "Normal");
1.1 timbl 759:
1.2 timbl 760: styles[HTML_H1] = HTStyleNamed(styleSheet, "Heading1");
761: styles[HTML_H2] = HTStyleNamed(styleSheet, "Heading2");
762: styles[HTML_H3] = HTStyleNamed(styleSheet, "Heading3");
763: styles[HTML_H4] = HTStyleNamed(styleSheet, "Heading4");
764: styles[HTML_H5] = HTStyleNamed(styleSheet, "Heading5");
765: styles[HTML_H6] = HTStyleNamed(styleSheet, "Heading6");
766: styles[HTML_H7] = HTStyleNamed(styleSheet, "Heading7");
767:
768: styles[HTML_DL] = HTStyleNamed(styleSheet, "Glossary");
769: styles[HTML_UL] =
770: styles[HTML_OL] = HTStyleNamed(styleSheet, "List");
771: styles[HTML_MENU] = HTStyleNamed(styleSheet, "Menu");
772: styles[HTML_DIR] = HTStyleNamed(styleSheet, "Dir");
1.16 timbl 773: /* styles[HTML_DLC] = HTStyleNamed(styleSheet, "GlossaryCompact"); */
1.2 timbl 774: styles[HTML_ADDRESS]= HTStyleNamed(styleSheet, "Address");
775: styles[HTML_BLOCKQUOTE]= HTStyleNamed(styleSheet, "BlockQuote");
776: styles[HTML_PLAINTEXT] =
777: styles[HTML_XMP] = HTStyleNamed(styleSheet, "Example");
778: styles[HTML_PRE] = HTStyleNamed(styleSheet, "Preformatted");
779: styles[HTML_LISTING] = HTStyleNamed(styleSheet, "Listing");
780: }
781: /* P U B L I C
782: */
783:
784: /* Structured Object Class
785: ** -----------------------
786: */
1.60 frystyk 787: PRIVATE const HTStructuredClass HTMLPresentation = /* As opposed to print etc */
1.2 timbl 788: {
789: "text/html",
1.42 frystyk 790: HTML_flush,
1.2 timbl 791: HTML_free,
1.14 timbl 792: HTML_abort,
1.2 timbl 793: HTML_put_character, HTML_put_string, HTML_write,
794: HTML_start_element, HTML_end_element,
795: HTML_put_entity
796: };
1.1 timbl 797:
1.4 timbl 798:
1.2 timbl 799: /* New Structured Text object
800: ** --------------------------
801: **
1.16 timbl 802: ** The structured stream can generate either presentation,
1.4 timbl 803: ** or plain text, or HTML.
1.1 timbl 804: */
1.53 frystyk 805: PRIVATE HTStructured* HTML_new (HTRequest * request,
806: void * param,
807: HTFormat input_format,
808: HTFormat output_format,
809: HTStream * output_stream)
1.1 timbl 810: {
811:
1.4 timbl 812: HTStructured * me;
813:
1.47 frystyk 814: #if 0
1.16 timbl 815: if (output_format != WWW_PLAINTEXT
816: && output_format != WWW_PRESENT
817: && output_format != HTAtom_for("text/x-c")) {
1.37 frystyk 818: HTStream * intermediate = HTStreamStack(WWW_HTML, output_format,
819: output_stream, request, NO);
1.6 timbl 820: if (intermediate) return HTMLGenerator(intermediate);
1.44 frystyk 821: if (SGML_TRACE)
1.63 frystyk 822: HTTrace("HTML Parser. Can't parse HTML to %s\n",
1.44 frystyk 823: HTAtom_name(output_format));
1.4 timbl 824: exit (-99);
825: }
1.47 frystyk 826: #endif
1.4 timbl 827:
1.58 frystyk 828: if ((me = (HTStructured *) HT_CALLOC(1, sizeof(*me))) == NULL)
829: HT_OUTOFMEM("HTML_new");
1.1 timbl 830:
831: if (!got_styles) get_styles();
832:
1.4 timbl 833: me->isa = &HTMLPresentation;
1.47 frystyk 834: me->dtd = &HTMLP_dtd;
1.54 frystyk 835: me->request = request;
1.48 frystyk 836: me->node_anchor = HTRequest_anchor(request);
1.56 frystyk 837: me->title = HTChunk_new(128);
1.4 timbl 838: me->text = 0;
839: me->style_change = YES; /* Force check leading to text creation */
840: me->new_style = default_style;
841: me->old_style = 0;
842: me->sp = me->stack + MAX_NESTING - 1;
843: me->sp->tag_number = -1; /* INVALID */
844: me->sp->style = default_style; /* INVALID */
1.1 timbl 845:
1.4 timbl 846: me->comment_start = NULL;
847: me->comment_end = NULL;
1.16 timbl 848: me->target = output_stream;
849: if (output_stream) me->targetClass = *output_stream->isa; /* Copy pointers */
1.1 timbl 850:
1.4 timbl 851: return (HTStructured*) me;
1.1 timbl 852: }
853:
854:
1.2 timbl 855: /* HTConverter for HTML to plain text
856: ** ----------------------------------
1.1 timbl 857: **
1.2 timbl 858: ** This will convert from HTML to presentation or plain text.
1.1 timbl 859: */
1.53 frystyk 860: PUBLIC HTStream* HTMLToPlain (
861: HTRequest * request,
862: void * param,
863: HTFormat input_format,
864: HTFormat output_format,
865: HTStream * output_stream)
1.1 timbl 866: {
1.47 frystyk 867: return SGML_new(&HTMLP_dtd, HTML_new(
1.16 timbl 868: request, NULL, input_format, output_format, output_stream));
1.1 timbl 869: }
870:
871:
1.2 timbl 872: /* HTConverter for HTML to C code
873: ** ------------------------------
874: **
1.36 frystyk 875: ** C code is like plain text but all non-preformatted code
1.2 timbl 876: ** is commented out.
877: ** This will convert from HTML to presentation or plain text.
878: */
1.53 frystyk 879: PUBLIC HTStream* HTMLToC (
880: HTRequest * request,
881: void * param,
882: HTFormat input_format,
883: HTFormat output_format,
884: HTStream * output_stream)
1.1 timbl 885: {
1.4 timbl 886:
887: HTStructured * html;
888:
1.36 frystyk 889: (*output_stream->isa->put_string)(output_stream, "/* "); /* Before title */
1.16 timbl 890: html = HTML_new(request, NULL, input_format, output_format, output_stream);
1.45 frystyk 891: html->comment_start = "\n/* ";
1.47 frystyk 892: html->dtd = &HTMLP_dtd;
1.2 timbl 893: html->comment_end = " */\n"; /* Must start in col 1 for cpp */
1.47 frystyk 894: return SGML_new(&HTMLP_dtd, html);
1.1 timbl 895: }
896:
897:
1.2 timbl 898: /* Presenter for HTML
899: ** ------------------
900: **
901: ** This will convert from HTML to presentation or plain text.
902: **
903: ** Override this if you have a windows version
1.1 timbl 904: */
1.2 timbl 905: #ifndef GUI
1.53 frystyk 906: PUBLIC HTStream* HTMLPresent (
907: HTRequest * request,
908: void * param,
909: HTFormat input_format,
910: HTFormat output_format,
911: HTStream * output_stream)
1.1 timbl 912: {
1.47 frystyk 913: return SGML_new(&HTMLP_dtd, HTML_new(
1.16 timbl 914: request, NULL, input_format, output_format, output_stream));
1.1 timbl 915: }
1.2 timbl 916: #endif
1.29 frystyk 917:
Webmaster