Annotation of libwww/Library/src/HTML.c, revision 1.68
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.
1.68 ! frystyk 6: ** @(#) $Id: HTML.c,v 1.67 1997/03/21 19:33:15 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:
371: {
1.8 timbl 372: HTChildAnchor * source;
1.9 timbl 373: char * href = NULL;
1.42 frystyk 374: if (present[HTML_A_HREF])
1.9 timbl 375: StrAllocCopy(href, value[HTML_A_HREF]);
1.8 timbl 376: source = HTAnchor_findChildAndLink(
1.4 timbl 377: me->node_anchor, /* parent */
1.2 timbl 378: present[HTML_A_NAME] ? value[HTML_A_NAME] : 0, /* Tag */
1.9 timbl 379: present[HTML_A_HREF] ? href : 0, /* Addresss */
1.16 timbl 380: present[HTML_A_REL] && value[HTML_A_REL] ?
1.54 frystyk 381: (HTLinkType) HTAtom_for(value[HTML_A_REL])
382: : 0);
1.2 timbl 383:
1.63 frystyk 384: if (present[HTML_A_TITLE] && value[HTML_A_TITLE]) {
385: HTLink * link = HTAnchor_mainLink((HTAnchor *) source);
386: HTParentAnchor *dest=HTAnchor_parent(HTLink_destination(link));
1.2 timbl 387: if (!HTAnchor_title(dest))
388: HTAnchor_setTitle(dest, value[HTML_A_TITLE]);
389: }
390: UPDATE_STYLE;
1.4 timbl 391: HText_beginAnchor(me->text, source);
1.58 frystyk 392: HT_FREE(href); /* Leak fix Henrik 17/02-94 */
1.2 timbl 393: }
394: break;
395:
1.63 frystyk 396: case HTML_LINK:
397:
398: /* MORE */
399:
400: break;
401:
1.2 timbl 402: case HTML_TITLE:
1.56 frystyk 403: HTChunk_clear(me->title);
1.2 timbl 404: break;
405:
406: case HTML_NEXTID:
407: /* if (present[NEXTID_N] && value[NEXTID_N])
1.4 timbl 408: HText_setNextId(me->text, atoi(value[NEXTID_N])); */
1.2 timbl 409: break;
410:
411: case HTML_ISINDEX:
1.4 timbl 412: HTAnchor_setIndex(me->node_anchor);
1.2 timbl 413: break;
414:
1.15 timbl 415: case HTML_BR:
416: UPDATE_STYLE;
417: HText_appendCharacter(me->text, '\n');
418: me->in_word = NO;
419: break;
420:
421: case HTML_HR:
422: UPDATE_STYLE;
423: HText_appendCharacter(me->text, '\n');
1.16 timbl 424: HText_appendText(me->text, "___________________________________");
1.15 timbl 425: HText_appendCharacter(me->text, '\n');
426: me->in_word = NO;
427: break;
428:
1.2 timbl 429: case HTML_P:
430: UPDATE_STYLE;
1.4 timbl 431: HText_appendParagraph(me->text);
432: me->in_word = NO;
1.2 timbl 433: break;
434:
435: case HTML_DL:
1.11 timbl 436: change_paragraph_style(me, present && present[DL_COMPACT]
1.16 timbl 437: ? styles[HTML_DL]
1.2 timbl 438: : styles[HTML_DL]);
439: break;
440:
441: case HTML_DT:
1.4 timbl 442: if (!me->style_change) {
443: HText_appendParagraph(me->text);
444: me->in_word = NO;
1.2 timbl 445: }
446: break;
447:
448: case HTML_DD:
449: UPDATE_STYLE;
1.62 frystyk 450: HTML_put_character(me, TAB); /* Just tab out one stop */
1.4 timbl 451: me->in_word = NO;
452: break;
1.2 timbl 453:
454: case HTML_UL:
455: case HTML_OL:
456: case HTML_MENU:
457: case HTML_DIR:
1.11 timbl 458: change_paragraph_style(me, styles[element_number]);
1.2 timbl 459: break;
460:
461: case HTML_LI:
462: UPDATE_STYLE;
1.7 timbl 463: if (me->sp[0].tag_number != HTML_DIR)
1.4 timbl 464: HText_appendParagraph(me->text);
1.2 timbl 465: else
1.62 frystyk 466: HText_appendCharacter(me->text, TAB);
1.4 timbl 467: me->in_word = NO;
1.2 timbl 468: break;
469:
470: case HTML_LISTING: /* Litteral text */
471: case HTML_XMP:
472: case HTML_PLAINTEXT:
473: case HTML_PRE:
1.11 timbl 474: change_paragraph_style(me, styles[element_number]);
1.2 timbl 475: UPDATE_STYLE;
1.4 timbl 476: if (me->comment_end)
477: HText_appendText(me->text, me->comment_end);
1.2 timbl 478: break;
1.11 timbl 479:
1.23 frystyk 480: case HTML_IMG: /* Images */
481: {
482: HTChildAnchor *source;
483: char *src = NULL;
1.49 frystyk 484: if (present[HTML_IMG_SRC])
1.23 frystyk 485: StrAllocCopy(src, value[HTML_IMG_SRC]);
486: source = HTAnchor_findChildAndLink(
487: me->node_anchor, /* parent */
488: 0, /* Tag */
489: src ? src : 0, /* Addresss */
490: 0);
491: UPDATE_STYLE;
492: HText_appendImage(me->text, source,
1.24 frystyk 493: present[HTML_IMG_ALT] ? value[HTML_IMG_ALT] : NULL,
494: present[HTML_IMG_ALIGN] ? value[HTML_IMG_ALIGN] : NULL,
495: present[HTML_IMG_ISMAP] ? YES : NO);
1.58 frystyk 496: HT_FREE(src);
1.24 frystyk 497: }
498: break;
499:
1.63 frystyk 500: case HTML_BASE: /* Base header */
501: if (present[HTML_BASE_HREF]) {
502: char * base = (char *) value[HTML_BASE_HREF];
503: if (base) {
504: HTAnchor_setBase(me->node_anchor, base);
505: if (SGML_TRACE) HTTrace("HTML Parser. New base `%s\'\n", base);
506: } else {
507: if (SGML_TRACE) HTTrace("HTML Parser. No base found\n");
508: }
509: }
510: break;
511:
1.24 frystyk 512: case HTML_HTML: /* Ignore these altogether */
513: case HTML_HEAD:
514: case HTML_BODY:
1.62 frystyk 515: break;
1.24 frystyk 516:
1.10 timbl 517: case HTML_TT: /* Physical character highlighting */
518: case HTML_B: /* Currently ignored */
519: case HTML_I:
520: case HTML_U:
1.62 frystyk 521: UPDATE_STYLE;
1.65 frystyk 522: #if 0
1.62 frystyk 523: HText_appendCharacter(me->text, '_');
1.65 frystyk 524: #endif
1.62 frystyk 525: me->in_word = NO;
526: break;
1.10 timbl 527:
528: case HTML_EM: /* Logical character highlighting */
529: case HTML_STRONG: /* Currently ignored */
530: case HTML_CODE:
531: case HTML_SAMP:
532: case HTML_KBD:
533: case HTML_VAR:
534: case HTML_DFN:
535: case HTML_CITE:
536: break;
537:
1.11 timbl 538: case HTML_H1: /* paragraph styles */
539: case HTML_H2:
540: case HTML_H3:
541: case HTML_H4:
542: case HTML_H5:
543: case HTML_H6:
544: case HTML_H7:
545: case HTML_ADDRESS:
546: case HTML_BLOCKQUOTE:
547: change_paragraph_style(me, styles[element_number]); /* May be postponed */
1.2 timbl 548: break;
549:
550: } /* end switch */
551:
1.16 timbl 552: if (me->dtd->tags[element_number].contents!= SGML_EMPTY) {
1.13 timbl 553: if (me->sp == me->stack) {
1.44 frystyk 554: if (SGML_TRACE)
1.63 frystyk 555: HTTrace("HTML Parser. Maximum nesting of %d exceded!\n",
1.44 frystyk 556: MAX_NESTING);
557: me->overflow++;
1.12 timbl 558: return;
559: }
1.4 timbl 560: --(me->sp);
561: me->sp[0].style = me->new_style; /* Stack new style */
562: me->sp[0].tag_number = element_number;
1.10 timbl 563: }
1.1 timbl 564: }
1.10 timbl 565:
1.2 timbl 566:
1.1 timbl 567: /* End Element
1.2 timbl 568: ** -----------
1.1 timbl 569: **
1.2 timbl 570: */
571: /* When we end an element, the style must be returned to that
1.1 timbl 572: ** in effect before that element. Note that anchors (etc?)
573: ** don't have an associated style, so that we must scan down the
574: ** stack for an element with a defined style. (In fact, the styles
575: ** should be linked to the whole stack not just the top one.)
576: ** TBL 921119
1.6 timbl 577: **
578: ** We don't turn on "CAREFUL" check because the parser produces
579: ** (internal code errors apart) good nesting. The parser checks
580: ** incoming code errors, not this module.
1.1 timbl 581: */
1.53 frystyk 582: PRIVATE void HTML_end_element (HTStructured * me, int element_number)
1.1 timbl 583: {
1.2 timbl 584: #ifdef CAREFUL /* parser assumed to produce good nesting */
1.4 timbl 585: if (element_number != me->sp[0].tag_number) {
1.59 eric 586: HTTrace("HTMLText: end of element %s when expecting end of %s\n",
1.16 timbl 587: me->dtd->tags[element_number].name,
588: me->dtd->tags[me->sp->tag_number].name);
1.6 timbl 589: /* panic */
1.1 timbl 590: }
1.2 timbl 591: #endif
1.44 frystyk 592:
593: /* HFN, If overflow of nestings, we need to get back to reality */
594: if (me->overflow > 0) {
595: me->overflow--;
596: return;
597: }
598:
1.4 timbl 599: me->sp++; /* Pop state off stack */
1.67 frystyk 600: if (me->sp > me->stack + MAX_NESTING - 1) {
601: if (SGML_TRACE) HTTrace("HTML Parser. Bottom of style stack reached\n");
602: me->sp = me->stack + MAX_NESTING - 1;
603: }
1.44 frystyk 604:
1.2 timbl 605: switch(element_number) {
606:
607: case HTML_A:
608: UPDATE_STYLE;
1.4 timbl 609: HText_endAnchor(me->text);
1.2 timbl 610: break;
611:
612: case HTML_TITLE:
1.56 frystyk 613: HTAnchor_setTitle(me->node_anchor, HTChunk_data(me->title));
1.2 timbl 614: break;
615:
1.62 frystyk 616: case HTML_TT: /* Physical character highlighting */
617: case HTML_B: /* Currently ignored */
618: case HTML_I:
619: case HTML_U:
620: UPDATE_STYLE;
1.66 frystyk 621: #if 0
1.62 frystyk 622: HText_appendCharacter(me->text, '_');
1.66 frystyk 623: #endif
1.62 frystyk 624: break;
1.66 frystyk 625:
626: case HTML_EM: /* Logical character highlighting */
627: case HTML_STRONG: /* Currently ignored */
628: case HTML_CODE:
629: case HTML_SAMP:
630: case HTML_KBD:
631: case HTML_VAR:
632: case HTML_DFN:
633: case HTML_CITE:
634: break;
1.62 frystyk 635:
1.2 timbl 636: case HTML_LISTING: /* Litteral text */
637: case HTML_XMP:
638: case HTML_PLAINTEXT:
639: case HTML_PRE:
1.4 timbl 640: if (me->comment_start)
641: HText_appendText(me->text, me->comment_start);
1.2 timbl 642: /* Fall through */
643:
644: default:
1.44 frystyk 645:
646: /* Often won't really change */
647: change_paragraph_style(me, me->sp->style);
1.2 timbl 648: break;
649:
650: } /* switch */
1.1 timbl 651: }
652:
1.2 timbl 653:
654: /* Expanding entities
655: ** ------------------
656: */
657: /* (In fact, they all shrink!)
1.1 timbl 658: */
1.2 timbl 659:
1.53 frystyk 660: PRIVATE void HTML_put_entity (HTStructured * me, int entity_number)
1.1 timbl 661: {
1.4 timbl 662: HTML_put_string(me, ISO_Latin1[entity_number]); /* @@ Other representations */
1.1 timbl 663: }
1.2 timbl 664:
1.42 frystyk 665: /* Flush an HTML object
666: ** --------------------
667: */
1.53 frystyk 668: PUBLIC int HTML_flush (HTStructured * me)
1.42 frystyk 669: {
670: UPDATE_STYLE; /* Creates empty document here! */
1.57 frystyk 671: if (me->comment_end) HTML_put_string(me,me->comment_end);
672: return me->target ? (*me->targetClass.flush)(me->target) : HT_OK;
1.42 frystyk 673: }
1.2 timbl 674:
675: /* Free an HTML object
676: ** -------------------
677: **
1.4 timbl 678: ** If the document is empty, the text object will not yet exist.
679: So we could in fact abandon creating the document and return
680: an error code. In fact an empty document is an important type
681: of document, so we don't.
682: **
683: ** If non-interactive, everything is freed off. No: crashes -listrefs
1.2 timbl 684: ** Otherwise, the interactive object is left.
685: */
1.53 frystyk 686: PUBLIC int HTML_free (HTStructured * me)
1.1 timbl 687: {
1.4 timbl 688: UPDATE_STYLE; /* Creates empty document here! */
689: if (me->comment_end)
690: HTML_put_string(me,me->comment_end);
691: HText_endAppend(me->text);
692:
693: if (me->target) {
1.35 duns 694: (*me->targetClass._free)(me->target);
1.2 timbl 695: }
1.56 frystyk 696: HTChunk_delete(me->title);
1.58 frystyk 697: HT_FREE(me);
1.42 frystyk 698: return HT_OK;
1.1 timbl 699: }
700:
701:
1.53 frystyk 702: PRIVATE int HTML_abort (HTStructured * me, HTList * e)
1.1 timbl 703:
1.14 timbl 704: {
705: if (me->target) {
706: (*me->targetClass.abort)(me->target, e);
707: }
1.56 frystyk 708: HTChunk_delete(me->title);
1.58 frystyk 709: HT_FREE(me);
1.42 frystyk 710: return HT_ERROR;
1.1 timbl 711: }
712:
1.2 timbl 713:
714: /* Get Styles from style sheet
715: ** ---------------------------
716: */
1.53 frystyk 717: PRIVATE void get_styles (void)
1.1 timbl 718: {
1.2 timbl 719: got_styles = YES;
720:
721: default_style = HTStyleNamed(styleSheet, "Normal");
1.1 timbl 722:
1.2 timbl 723: styles[HTML_H1] = HTStyleNamed(styleSheet, "Heading1");
724: styles[HTML_H2] = HTStyleNamed(styleSheet, "Heading2");
725: styles[HTML_H3] = HTStyleNamed(styleSheet, "Heading3");
726: styles[HTML_H4] = HTStyleNamed(styleSheet, "Heading4");
727: styles[HTML_H5] = HTStyleNamed(styleSheet, "Heading5");
728: styles[HTML_H6] = HTStyleNamed(styleSheet, "Heading6");
729: styles[HTML_H7] = HTStyleNamed(styleSheet, "Heading7");
730:
731: styles[HTML_DL] = HTStyleNamed(styleSheet, "Glossary");
732: styles[HTML_UL] =
733: styles[HTML_OL] = HTStyleNamed(styleSheet, "List");
734: styles[HTML_MENU] = HTStyleNamed(styleSheet, "Menu");
735: styles[HTML_DIR] = HTStyleNamed(styleSheet, "Dir");
1.16 timbl 736: /* styles[HTML_DLC] = HTStyleNamed(styleSheet, "GlossaryCompact"); */
1.2 timbl 737: styles[HTML_ADDRESS]= HTStyleNamed(styleSheet, "Address");
738: styles[HTML_BLOCKQUOTE]= HTStyleNamed(styleSheet, "BlockQuote");
739: styles[HTML_PLAINTEXT] =
740: styles[HTML_XMP] = HTStyleNamed(styleSheet, "Example");
741: styles[HTML_PRE] = HTStyleNamed(styleSheet, "Preformatted");
742: styles[HTML_LISTING] = HTStyleNamed(styleSheet, "Listing");
743: }
744: /* P U B L I C
745: */
746:
747: /* Structured Object Class
748: ** -----------------------
749: */
1.60 frystyk 750: PRIVATE const HTStructuredClass HTMLPresentation = /* As opposed to print etc */
1.2 timbl 751: {
752: "text/html",
1.42 frystyk 753: HTML_flush,
1.2 timbl 754: HTML_free,
1.14 timbl 755: HTML_abort,
1.2 timbl 756: HTML_put_character, HTML_put_string, HTML_write,
757: HTML_start_element, HTML_end_element,
758: HTML_put_entity
759: };
1.1 timbl 760:
1.4 timbl 761:
1.2 timbl 762: /* New Structured Text object
763: ** --------------------------
764: **
1.16 timbl 765: ** The structured stream can generate either presentation,
1.4 timbl 766: ** or plain text, or HTML.
1.1 timbl 767: */
1.53 frystyk 768: PRIVATE HTStructured* HTML_new (HTRequest * request,
769: void * param,
770: HTFormat input_format,
771: HTFormat output_format,
772: HTStream * output_stream)
1.1 timbl 773: {
774:
1.4 timbl 775: HTStructured * me;
776:
1.47 frystyk 777: #if 0
1.16 timbl 778: if (output_format != WWW_PLAINTEXT
779: && output_format != WWW_PRESENT
780: && output_format != HTAtom_for("text/x-c")) {
1.37 frystyk 781: HTStream * intermediate = HTStreamStack(WWW_HTML, output_format,
782: output_stream, request, NO);
1.6 timbl 783: if (intermediate) return HTMLGenerator(intermediate);
1.44 frystyk 784: if (SGML_TRACE)
1.63 frystyk 785: HTTrace("HTML Parser. Can't parse HTML to %s\n",
1.44 frystyk 786: HTAtom_name(output_format));
1.4 timbl 787: exit (-99);
788: }
1.47 frystyk 789: #endif
1.4 timbl 790:
1.58 frystyk 791: if ((me = (HTStructured *) HT_CALLOC(1, sizeof(*me))) == NULL)
792: HT_OUTOFMEM("HTML_new");
1.1 timbl 793:
794: if (!got_styles) get_styles();
795:
1.4 timbl 796: me->isa = &HTMLPresentation;
1.47 frystyk 797: me->dtd = &HTMLP_dtd;
1.54 frystyk 798: me->request = request;
1.48 frystyk 799: me->node_anchor = HTRequest_anchor(request);
1.56 frystyk 800: me->title = HTChunk_new(128);
1.4 timbl 801: me->text = 0;
802: me->style_change = YES; /* Force check leading to text creation */
803: me->new_style = default_style;
804: me->old_style = 0;
805: me->sp = me->stack + MAX_NESTING - 1;
806: me->sp->tag_number = -1; /* INVALID */
807: me->sp->style = default_style; /* INVALID */
1.1 timbl 808:
1.4 timbl 809: me->comment_start = NULL;
810: me->comment_end = NULL;
1.16 timbl 811: me->target = output_stream;
812: if (output_stream) me->targetClass = *output_stream->isa; /* Copy pointers */
1.1 timbl 813:
1.4 timbl 814: return (HTStructured*) me;
1.1 timbl 815: }
816:
817:
1.2 timbl 818: /* HTConverter for HTML to plain text
819: ** ----------------------------------
1.1 timbl 820: **
1.2 timbl 821: ** This will convert from HTML to presentation or plain text.
1.1 timbl 822: */
1.53 frystyk 823: PUBLIC HTStream* HTMLToPlain (
824: HTRequest * request,
825: void * param,
826: HTFormat input_format,
827: HTFormat output_format,
828: HTStream * output_stream)
1.1 timbl 829: {
1.47 frystyk 830: return SGML_new(&HTMLP_dtd, HTML_new(
1.16 timbl 831: request, NULL, input_format, output_format, output_stream));
1.1 timbl 832: }
833:
834:
1.2 timbl 835: /* HTConverter for HTML to C code
836: ** ------------------------------
837: **
1.36 frystyk 838: ** C code is like plain text but all non-preformatted code
1.2 timbl 839: ** is commented out.
840: ** This will convert from HTML to presentation or plain text.
841: */
1.53 frystyk 842: PUBLIC HTStream* HTMLToC (
843: HTRequest * request,
844: void * param,
845: HTFormat input_format,
846: HTFormat output_format,
847: HTStream * output_stream)
1.1 timbl 848: {
1.4 timbl 849:
850: HTStructured * html;
851:
1.36 frystyk 852: (*output_stream->isa->put_string)(output_stream, "/* "); /* Before title */
1.16 timbl 853: html = HTML_new(request, NULL, input_format, output_format, output_stream);
1.45 frystyk 854: html->comment_start = "\n/* ";
1.47 frystyk 855: html->dtd = &HTMLP_dtd;
1.2 timbl 856: html->comment_end = " */\n"; /* Must start in col 1 for cpp */
1.47 frystyk 857: return SGML_new(&HTMLP_dtd, html);
1.1 timbl 858: }
859:
860:
1.2 timbl 861: /* Presenter for HTML
862: ** ------------------
863: **
864: ** This will convert from HTML to presentation or plain text.
865: **
866: ** Override this if you have a windows version
1.1 timbl 867: */
1.2 timbl 868: #ifndef GUI
1.53 frystyk 869: PUBLIC HTStream* HTMLPresent (
870: HTRequest * request,
871: void * param,
872: HTFormat input_format,
873: HTFormat output_format,
874: HTStream * output_stream)
1.1 timbl 875: {
1.47 frystyk 876: return SGML_new(&HTMLP_dtd, HTML_new(
1.16 timbl 877: request, NULL, input_format, output_format, output_stream));
1.1 timbl 878: }
1.2 timbl 879: #endif
1.29 frystyk 880:
Webmaster