Annotation of libwww/Library/src/HTML.c, revision 1.10
1.2 timbl 1: /* Structured stream to Rich hypertext converter
2: ** ============================================
1.1 timbl 3: **
1.2 timbl 4: ** This generates of a hypertext object. It converts from the
5: ** structured stream interface fro HTMl events into the style-
6: ** oriented iunterface of the HText.h interface. This module is
7: ** only used in clients and shouldnot be linked into servers.
1.1 timbl 8: **
1.6 timbl 9: ** Override this module if making a new GUI browser.
1.1 timbl 10: **
11: */
12: #include "HTML.h"
13:
1.6 timbl 14: /* #define CAREFUL Check nesting here notreally necessary */
1.2 timbl 15:
1.1 timbl 16: #include <ctype.h>
17: #include <stdio.h>
18:
19: #include "HTAtom.h"
20: #include "HTChunk.h"
21: #include "HText.h"
22: #include "HTStyle.h"
23:
1.3 timbl 24: #include "HTAlert.h"
1.4 timbl 25: #include "HTMLGen.h"
1.8 timbl 26: #include "HTParse.h"
1.1 timbl 27:
28: extern HTStyleSheet * styleSheet; /* Application-wide */
29:
30: /* Module-wide style cache
31: */
32: PRIVATE int got_styles = 0;
1.2 timbl 33: PRIVATE HTStyle *styles[HTML_ELEMENTS];
34: PRIVATE HTStyle *default_style;
1.1 timbl 35:
36:
37: /* HTML Object
38: ** -----------
39: */
1.2 timbl 40: #define MAX_NESTING 20 /* Should be checked by parser */
41:
42: typedef struct _stack_element {
43: HTStyle * style;
44: int tag_number;
45: } stack_element;
46:
47: struct _HTStructured {
48: CONST HTStructuredClass * isa;
49: HTParentAnchor * node_anchor;
50: HText * text;
51:
52: HTStream* target; /* Output stream */
53: HTStreamClass targetClass; /* Output routines */
54:
55: HTChunk title; /* Grow by 128 */
56:
57: char * comment_start; /* for literate programming */
58: char * comment_end;
59:
60: HTTag * current_tag;
61: BOOL style_change;
62: HTStyle * new_style;
63: HTStyle * old_style;
64: BOOL in_word; /* Have just had a non-white char */
65: stack_element stack[MAX_NESTING];
66: stack_element *sp; /* Style stack pointer */
1.1 timbl 67: };
68:
1.2 timbl 69: struct _HTStream {
70: CONST HTStreamClass * isa;
71: /* .... */
72: };
1.1 timbl 73:
74: /* Forward declarations of routines
75: */
76: PRIVATE void get_styles NOPARAMS;
77:
78:
1.4 timbl 79: PRIVATE void actually_set_style PARAMS((HTStructured * me));
80: PRIVATE void change_style PARAMS((HTStructured * me, HTStyle * style));
1.1 timbl 81:
82: /* Style buffering avoids dummy paragraph begin/ends.
83: */
1.4 timbl 84: #define UPDATE_STYLE if (me->style_change) { actually_set_style(me); }
1.1 timbl 85:
86:
1.2 timbl 87: #ifdef OLD_CODE
1.1 timbl 88: /* The following accented characters are from peter Flynn, curia project */
89:
90: /* these ifdefs don't solve the problem of a simple terminal emulator
91: ** with a different character set to the client machine. But nothing does,
92: ** except looking at the TERM setting */
93:
1.2 timbl 94:
1.1 timbl 95: { "ocus" , "&" }, /* for CURIA */
96: #ifdef IBMPC
97: { "aacute" , "\240" }, /* For PC display */
98: { "eacute" , "\202" },
99: { "iacute" , "\241" },
100: { "oacute" , "\242" },
101: { "uacute" , "\243" },
102: { "Aacute" , "\101" },
103: { "Eacute" , "\220" },
104: { "Iacute" , "\111" },
105: { "Oacute" , "\117" },
106: { "Uacute" , "\125" },
107: #else
108: { "aacute" , "\341" }, /* Works for openwindows -- Peter Flynn */
109: { "eacute" , "\351" },
110: { "iacute" , "\355" },
111: { "oacute" , "\363" },
112: { "uacute" , "\372" },
113: { "Aacute" , "\301" },
114: { "Eacute" , "\310" },
115: { "Iacute" , "\315" },
116: { "Oacute" , "\323" },
117: { "Uacute" , "\332" },
118: #endif
119: { 0, 0 } /* Terminate list */
120: };
1.2 timbl 121: #endif
1.1 timbl 122:
123:
1.2 timbl 124: /* Entity values -- for ISO Latin 1 local representation
125: **
126: ** This MUST match exactly the table referred to in the DTD!
127: */
128: static char * ISO_Latin1[] = {
129: "\306", /* capital AE diphthong (ligature) */
130: "\301", /* capital A, acute accent */
131: "\302", /* capital A, circumflex accent */
132: "\300", /* capital A, grave accent */
133: "\305", /* capital A, ring */
134: "\303", /* capital A, tilde */
135: "\304", /* capital A, dieresis or umlaut mark */
136: "\307", /* capital C, cedilla */
137: "\320", /* capital Eth, Icelandic */
138: "\311", /* capital E, acute accent */
139: "\312", /* capital E, circumflex accent */
140: "\310", /* capital E, grave accent */
141: "\313", /* capital E, dieresis or umlaut mark */
142: "\315", /* capital I, acute accent */
143: "\316", /* capital I, circumflex accent */
144: "\314", /* capital I, grave accent */
145: "\317", /* capital I, dieresis or umlaut mark */
146: "\321", /* capital N, tilde */
147: "\323", /* capital O, acute accent */
148: "\324", /* capital O, circumflex accent */
149: "\322", /* capital O, grave accent */
150: "\330", /* capital O, slash */
151: "\325", /* capital O, tilde */
152: "\326", /* capital O, dieresis or umlaut mark */
153: "\336", /* capital THORN, Icelandic */
154: "\332", /* capital U, acute accent */
155: "\333", /* capital U, circumflex accent */
156: "\331", /* capital U, grave accent */
157: "\334", /* capital U, dieresis or umlaut mark */
158: "\335", /* capital Y, acute accent */
159: "\341", /* small a, acute accent */
160: "\342", /* small a, circumflex accent */
161: "\346", /* small ae diphthong (ligature) */
162: "\340", /* small a, grave accent */
163: "\046", /* ampersand */
164: "\345", /* small a, ring */
165: "\343", /* small a, tilde */
166: "\344", /* small a, dieresis or umlaut mark */
167: "\347", /* small c, cedilla */
168: "\351", /* small e, acute accent */
169: "\352", /* small e, circumflex accent */
170: "\350", /* small e, grave accent */
171: "\360", /* small eth, Icelandic */
172: "\353", /* small e, dieresis or umlaut mark */
173: "\076", /* greater than */
174: "\355", /* small i, acute accent */
175: "\356", /* small i, circumflex accent */
176: "\354", /* small i, grave accent */
177: "\357", /* small i, dieresis or umlaut mark */
178: "\074", /* less than */
179: "\361", /* small n, tilde */
180: "\363", /* small o, acute accent */
181: "\364", /* small o, circumflex accent */
182: "\362", /* small o, grave accent */
183: "\370", /* small o, slash */
184: "\365", /* small o, tilde */
185: "\366", /* small o, dieresis or umlaut mark */
186: "\337", /* small sharp s, German (sz ligature) */
187: "\376", /* small thorn, Icelandic */
188: "\372", /* small u, acute accent */
189: "\373", /* small u, circumflex accent */
190: "\371", /* small u, grave accent */
191: "\374", /* small u, dieresis or umlaut mark */
192: "\375", /* small y, acute accent */
193: "\377", /* small y, dieresis or umlaut mark */
1.1 timbl 194: };
195:
1.2 timbl 196:
197: /* Entity values -- for NeXT local representation
198: **
199: ** This MUST match exactly the table referred to in the DTD!
200: **
201: */
202: static char * NeXTCharacters[] = {
203: "\341", /* capital AE diphthong (ligature) */
204: "\202", /* capital A, acute accent */
205: "\203", /* capital A, circumflex accent */
206: "\201", /* capital A, grave accent */
207: "\206", /* capital A, ring */
208: "\204", /* capital A, tilde */
209: "\205", /* capital A, dieresis or umlaut mark */
210: "\207", /* capital C, cedilla */
211: "\220", /* capital Eth, Icelandic */
212: "\211", /* capital E, acute accent */
213: "\212", /* capital E, circumflex accent */
214: "\210", /* capital E, grave accent */
215: "\213", /* capital E, dieresis or umlaut mark */
216: "\215", /* capital I, acute accent */
217: "\216", /* capital I, circumflex accent these are */
218: "\214", /* capital I, grave accent ISO -100 hex */
219: "\217", /* capital I, dieresis or umlaut mark */
220: "\221", /* capital N, tilde */
221: "\223", /* capital O, acute accent */
222: "\224", /* capital O, circumflex accent */
223: "\222", /* capital O, grave accent */
224: "\351", /* capital O, slash 'cept this */
225: "\225", /* capital O, tilde */
226: "\226", /* capital O, dieresis or umlaut mark */
227: "\234", /* capital THORN, Icelandic */
228: "\230", /* capital U, acute accent */
229: "\231", /* capital U, circumflex accent */
230: "\227", /* capital U, grave accent */
231: "\232", /* capital U, dieresis or umlaut mark */
232: "\233", /* capital Y, acute accent */
233: "\326", /* small a, acute accent */
234: "\327", /* small a, circumflex accent */
235: "\361", /* small ae diphthong (ligature) */
236: "\325", /* small a, grave accent */
237: "\046", /* ampersand */
238: "\332", /* small a, ring */
239: "\330", /* small a, tilde */
240: "\331", /* small a, dieresis or umlaut mark */
241: "\333", /* small c, cedilla */
242: "\335", /* small e, acute accent */
243: "\336", /* small e, circumflex accent */
244: "\334", /* small e, grave accent */
245: "\346", /* small eth, Icelandic */
246: "\337", /* small e, dieresis or umlaut mark */
247: "\076", /* greater than */
248: "\342", /* small i, acute accent */
249: "\344", /* small i, circumflex accent */
250: "\340", /* small i, grave accent */
251: "\345", /* small i, dieresis or umlaut mark */
252: "\074", /* less than */
253: "\347", /* small n, tilde */
254: "\355", /* small o, acute accent */
255: "\356", /* small o, circumflex accent */
256: "\354", /* small o, grave accent */
257: "\371", /* small o, slash */
258: "\357", /* small o, tilde */
259: "\360", /* small o, dieresis or umlaut mark */
260: "\373", /* small sharp s, German (sz ligature) */
261: "\374", /* small thorn, Icelandic */
262: "\363", /* small u, acute accent */
263: "\364", /* small u, circumflex accent */
264: "\362", /* small u, grave accent */
265: "\366", /* small u, dieresis or umlaut mark */
266: "\367", /* small y, acute accent */
267: "\375", /* small y, dieresis or umlaut mark */
1.1 timbl 268: };
269:
1.2 timbl 270: /* Entity values -- for IBM/PC Code Page 850 (International)
271: **
272: ** This MUST match exactly the table referred to in the DTD!
273: **
274: */
275: /* @@@@@@@@@@@@@@@@@ TBD */
276:
277:
278:
279: /* Set character set
280: ** ----------------
281: */
282:
283: PRIVATE char** p_entity_values = ISO_Latin1; /* Pointer to translation */
1.1 timbl 284:
1.2 timbl 285: PUBLIC void HTMLUseCharacterSet ARGS1(HTMLCharacterSet, i)
286: {
287: p_entity_values = (i == HTML_NEXT_CHARS) ? NeXTCharacters
288: : ISO_Latin1;
289: }
1.1 timbl 290:
291:
292: /* Flattening the style structure
293: ** ------------------------------
294: **
295: On the NeXT, and on any read-only browser, it is simpler for the text to have
296: a sequence of styles, rather than a nested tree of styles. In this
297: case we have to flatten the structure as it arrives from SGML tags into
298: a sequence of styles.
299: */
300:
301: /* If style really needs to be set, call this
302: */
1.4 timbl 303: PRIVATE void actually_set_style ARGS1(HTStructured *, me)
1.1 timbl 304: {
1.4 timbl 305: if (!me->text) { /* First time through */
306: me->text = HText_new2(me->node_anchor, me->target);
307: HText_beginAppend(me->text);
308: HText_setStyle(me->text, me->new_style);
309: me->in_word = NO;
1.1 timbl 310: } else {
1.4 timbl 311: HText_setStyle(me->text, me->new_style);
1.1 timbl 312: }
1.4 timbl 313: me->old_style = me->new_style;
314: me->style_change = NO;
1.1 timbl 315: }
316:
317: /* If you THINK you need to change style, call this
318: */
319:
1.4 timbl 320: PRIVATE void change_style ARGS2(HTStructured *, me, HTStyle *,style)
1.1 timbl 321: {
1.4 timbl 322: if (me->new_style!=style) {
323: me->style_change = YES;
324: me->new_style = style;
1.1 timbl 325: }
326: }
327:
1.2 timbl 328: /*_________________________________________________________________________
329: **
330: ** A C T I O N R O U T I N E S
331: */
332:
333: /* Character handling
334: ** ------------------
1.1 timbl 335: */
1.4 timbl 336: PRIVATE void HTML_put_character ARGS2(HTStructured *, me, char, c)
1.1 timbl 337: {
1.2 timbl 338:
1.4 timbl 339: switch (me->sp[0].tag_number) {
1.2 timbl 340: case HTML_COMMENT:
341: break; /* Do Nothing */
342:
343: case HTML_TITLE:
1.4 timbl 344: HTChunkPutc(&me->title, c);
1.2 timbl 345: break;
346:
347:
348: case HTML_LISTING: /* Litteral text */
349: case HTML_XMP:
350: case HTML_PLAINTEXT:
351: case HTML_PRE:
352: /* We guarrantee that the style is up-to-date in begin_litteral
353: */
1.4 timbl 354: HText_appendCharacter(me->text, c);
1.2 timbl 355: break;
356:
357: default: /* Free format text */
1.4 timbl 358: if (me->style_change) {
1.2 timbl 359: if ((c=='\n') || (c==' ')) return; /* Ignore it */
360: UPDATE_STYLE;
361: }
362: if (c=='\n') {
1.4 timbl 363: if (me->in_word) {
364: HText_appendCharacter(me->text, ' ');
365: me->in_word = NO;
1.2 timbl 366: }
367: } else {
1.4 timbl 368: HText_appendCharacter(me->text, c);
369: me->in_word = YES;
1.2 timbl 370: }
371: } /* end switch */
1.1 timbl 372: }
373:
1.2 timbl 374:
375:
376: /* String handling
377: ** ---------------
378: **
379: ** This is written separately from put_character becuase the loop can
380: ** in some cases be postponed to a lower level for speed.
381: */
1.4 timbl 382: PRIVATE void HTML_put_string ARGS2(HTStructured *, me, CONST char*, s)
1.1 timbl 383: {
1.2 timbl 384:
1.4 timbl 385: switch (me->sp[0].tag_number) {
1.2 timbl 386: case HTML_COMMENT:
387: break; /* Do Nothing */
388:
389: case HTML_TITLE:
1.4 timbl 390: HTChunkPuts(&me->title, s);
1.2 timbl 391: break;
392:
393:
394: case HTML_LISTING: /* Litteral text */
395: case HTML_XMP:
396: case HTML_PLAINTEXT:
397: case HTML_PRE:
398:
399: /* We guarrantee that the style is up-to-date in begin_litteral
400: */
1.4 timbl 401: HText_appendText(me->text, s);
1.2 timbl 402: break;
403:
404: default: /* Free format text */
405: {
406: CONST char *p = s;
1.4 timbl 407: if (me->style_change) {
1.2 timbl 408: for (; *p && ((*p=='\n') || (*p==' ')); p++) ; /* Ignore leaders */
409: if (!*p) return;
410: UPDATE_STYLE;
411: }
412: for(; *p; p++) {
1.4 timbl 413: if (me->style_change) {
1.2 timbl 414: if ((*p=='\n') || (*p==' ')) continue; /* Ignore it */
415: UPDATE_STYLE;
416: }
417: if (*p=='\n') {
1.4 timbl 418: if (me->in_word) {
419: HText_appendCharacter(me->text, ' ');
420: me->in_word = NO;
1.2 timbl 421: }
422: } else {
1.4 timbl 423: HText_appendCharacter(me->text, *p);
424: me->in_word = YES;
1.2 timbl 425: }
426: } /* for */
427: }
428: } /* end switch */
1.1 timbl 429: }
430:
431:
1.2 timbl 432: /* Buffer write
1.3 timbl 433: ** ------------
1.1 timbl 434: */
1.4 timbl 435: PRIVATE void HTML_write ARGS3(HTStructured *, me, CONST char*, s, int, l)
1.1 timbl 436: {
1.2 timbl 437: CONST char* p;
438: CONST char* e = s+l;
1.4 timbl 439: for (p=s; s<e; p++) HTML_put_character(me, *p);
1.1 timbl 440: }
1.2 timbl 441:
442:
443: /* Start Element
444: ** -------------
445: */
446: PRIVATE void HTML_start_element ARGS4(
1.4 timbl 447: HTStructured *, me,
1.2 timbl 448: int, element_number,
1.3 timbl 449: CONST BOOL*, present,
450: CONST char **, value)
1.2 timbl 451: {
452: switch (element_number) {
453: case HTML_A:
454: {
1.8 timbl 455: HTChildAnchor * source;
1.9 timbl 456: char * href = NULL;
457: if (present[HTML_A_HREF]) {
458: StrAllocCopy(href, value[HTML_A_HREF]);
459: HTSimplify(href);
460: }
1.8 timbl 461: source = HTAnchor_findChildAndLink(
1.4 timbl 462: me->node_anchor, /* parent */
1.2 timbl 463: present[HTML_A_NAME] ? value[HTML_A_NAME] : 0, /* Tag */
1.9 timbl 464: present[HTML_A_HREF] ? href : 0, /* Addresss */
1.2 timbl 465: present[HTML_A_TYPE] && value[HTML_A_TYPE] ?
466: (HTLinkType*)HTAtom_for(value[HTML_A_TYPE])
467: : 0);
468:
469: if (present[HTML_A_TITLE] && value[HTML_A_TITLE]) {
470: HTParentAnchor * dest =
471: HTAnchor_parent(
472: HTAnchor_followMainLink((HTAnchor*)source)
473: );
474: if (!HTAnchor_title(dest))
475: HTAnchor_setTitle(dest, value[HTML_A_TITLE]);
476: }
477: UPDATE_STYLE;
1.4 timbl 478: HText_beginAnchor(me->text, source);
1.2 timbl 479: }
480: break;
481:
482: case HTML_TITLE:
1.4 timbl 483: HTChunkClear(&me->title);
1.2 timbl 484: break;
485:
486: case HTML_NEXTID:
487: /* if (present[NEXTID_N] && value[NEXTID_N])
1.4 timbl 488: HText_setNextId(me->text, atoi(value[NEXTID_N])); */
1.2 timbl 489: break;
490:
491: case HTML_ISINDEX:
1.4 timbl 492: HTAnchor_setIndex(me->node_anchor);
1.2 timbl 493: break;
494:
495: case HTML_P:
496: UPDATE_STYLE;
1.4 timbl 497: HText_appendParagraph(me->text);
498: me->in_word = NO;
1.2 timbl 499: break;
500:
501: case HTML_DL:
1.4 timbl 502: change_style(me, present && present[DL_COMPACT]
1.2 timbl 503: ? styles[HTML_DLC]
504: : styles[HTML_DL]);
1.4 timbl 505: me->in_word = NO;
1.2 timbl 506: break;
507:
508: case HTML_DT:
1.4 timbl 509: if (!me->style_change) {
510: HText_appendParagraph(me->text);
511: me->in_word = NO;
1.2 timbl 512: }
513: break;
514:
515: case HTML_DD:
516: UPDATE_STYLE;
1.4 timbl 517: HTML_put_character(me, '\t'); /* Just tab out one stop */
518: me->in_word = NO;
519: break;
1.2 timbl 520:
521: case HTML_UL:
522: case HTML_OL:
523: case HTML_MENU:
524: case HTML_DIR:
1.4 timbl 525: change_style(me, styles[element_number]);
526: me->in_word = NO;
1.2 timbl 527: break;
528:
529: case HTML_LI:
530: UPDATE_STYLE;
1.7 timbl 531: if (me->sp[0].tag_number != HTML_DIR)
1.4 timbl 532: HText_appendParagraph(me->text);
1.2 timbl 533: else
1.4 timbl 534: HText_appendCharacter(me->text, '\t'); /* Tab @@ nl for UL? */
535: me->in_word = NO;
1.2 timbl 536: break;
537:
538: case HTML_LISTING: /* Litteral text */
539: case HTML_XMP:
540: case HTML_PLAINTEXT:
541: case HTML_PRE:
1.4 timbl 542: change_style(me, styles[element_number]);
1.2 timbl 543: UPDATE_STYLE;
1.4 timbl 544: if (me->comment_end)
545: HText_appendText(me->text, me->comment_end);
1.2 timbl 546: break;
547:
1.10 ! timbl 548: case HTML_IMG: /* Images -- ignore */
! 549:
! 550: case HTML_TT: /* Physical character highlighting */
! 551: case HTML_B: /* Currently ignored */
! 552: case HTML_I:
! 553: case HTML_U:
! 554:
! 555: case HTML_EM: /* Logical character highlighting */
! 556: case HTML_STRONG: /* Currently ignored */
! 557: case HTML_CODE:
! 558: case HTML_SAMP:
! 559: case HTML_KBD:
! 560: case HTML_VAR:
! 561: case HTML_DFN:
! 562: case HTML_CITE:
! 563: break;
! 564:
1.2 timbl 565: default:
1.4 timbl 566: change_style(me, styles[element_number]); /* May be postponed */
1.2 timbl 567: break;
568:
569: } /* end switch */
570:
571: if (HTML_dtd.tags[element_number].contents!= SGML_EMPTY) {
1.4 timbl 572: --(me->sp);
573: me->sp[0].style = me->new_style; /* Stack new style */
574: me->sp[0].tag_number = element_number;
1.10 ! timbl 575: }
1.1 timbl 576: }
1.10 ! timbl 577:
1.2 timbl 578:
1.1 timbl 579: /* End Element
1.2 timbl 580: ** -----------
1.1 timbl 581: **
1.2 timbl 582: */
583: /* When we end an element, the style must be returned to that
1.1 timbl 584: ** in effect before that element. Note that anchors (etc?)
585: ** don't have an associated style, so that we must scan down the
586: ** stack for an element with a defined style. (In fact, the styles
587: ** should be linked to the whole stack not just the top one.)
588: ** TBL 921119
1.6 timbl 589: **
590: ** We don't turn on "CAREFUL" check because the parser produces
591: ** (internal code errors apart) good nesting. The parser checks
592: ** incoming code errors, not this module.
1.1 timbl 593: */
1.4 timbl 594: PRIVATE void HTML_end_element ARGS2(HTStructured *, me, int , element_number)
1.1 timbl 595: {
1.2 timbl 596: #ifdef CAREFUL /* parser assumed to produce good nesting */
1.4 timbl 597: if (element_number != me->sp[0].tag_number) {
1.2 timbl 598: fprintf(stderr, "HTMLText: end of element %s when expecting end of %s\n",
599: HTML_dtd.tags[element_number].name,
1.4 timbl 600: HTML_dtd.tags[me->sp->tag_number].name);
1.6 timbl 601: /* panic */
1.1 timbl 602: }
1.2 timbl 603: #endif
604:
1.4 timbl 605: me->sp++; /* Pop state off stack */
1.2 timbl 606:
607: switch(element_number) {
608:
609: case HTML_A:
610: UPDATE_STYLE;
1.4 timbl 611: HText_endAnchor(me->text);
1.2 timbl 612: break;
613:
614: case HTML_TITLE:
1.4 timbl 615: HTChunkTerminate(&me->title);
616: HTAnchor_setTitle(me->node_anchor, me->title.data);
1.2 timbl 617: break;
618:
619: case HTML_LISTING: /* Litteral text */
620: case HTML_XMP:
621: case HTML_PLAINTEXT:
622: case HTML_PRE:
1.4 timbl 623: if (me->comment_start)
624: HText_appendText(me->text, me->comment_start);
1.2 timbl 625: /* Fall through */
626:
627: default:
628:
1.4 timbl 629: change_style(me, me->sp->style); /* Often won't really change */
630: me->in_word = NO; /* Paragraph styles only @@ */
1.2 timbl 631: break;
632:
633: } /* switch */
1.1 timbl 634: }
635:
1.2 timbl 636:
637: /* Expanding entities
638: ** ------------------
639: */
640: /* (In fact, they all shrink!)
1.1 timbl 641: */
1.2 timbl 642:
1.4 timbl 643: PRIVATE void HTML_put_entity ARGS2(HTStructured *, me, int, entity_number)
1.1 timbl 644: {
1.4 timbl 645: HTML_put_string(me, ISO_Latin1[entity_number]); /* @@ Other representations */
1.1 timbl 646: }
647:
1.2 timbl 648:
649:
650: /* Free an HTML object
651: ** -------------------
652: **
1.4 timbl 653: ** If the document is empty, the text object will not yet exist.
654: So we could in fact abandon creating the document and return
655: an error code. In fact an empty document is an important type
656: of document, so we don't.
657: **
658: ** If non-interactive, everything is freed off. No: crashes -listrefs
1.2 timbl 659: ** Otherwise, the interactive object is left.
660: */
1.4 timbl 661: PUBLIC void HTML_free ARGS1(HTStructured *, me)
1.1 timbl 662: {
1.4 timbl 663: UPDATE_STYLE; /* Creates empty document here! */
664: if (me->comment_end)
665: HTML_put_string(me,me->comment_end);
666: HText_endAppend(me->text);
667:
668: if (me->target) {
669: (*me->targetClass.end_document)(me->target);
670: (*me->targetClass.free)(me->target);
671: /* HText_free(me->text); */ /* @@@@@@@@@@@@@@@ */
1.2 timbl 672: }
1.4 timbl 673: free(me);
1.1 timbl 674: }
675:
676:
1.4 timbl 677: PRIVATE void HTML_end_document ARGS1(HTStructured *, me)
1.1 timbl 678:
1.4 timbl 679: { /* Obsolete */
1.1 timbl 680: }
681:
1.2 timbl 682:
683: /* Get Styles from style sheet
684: ** ---------------------------
685: */
686: PRIVATE void get_styles NOARGS
1.1 timbl 687: {
1.2 timbl 688: got_styles = YES;
689:
690: default_style = HTStyleNamed(styleSheet, "Normal");
1.1 timbl 691:
1.2 timbl 692: styles[HTML_H1] = HTStyleNamed(styleSheet, "Heading1");
693: styles[HTML_H2] = HTStyleNamed(styleSheet, "Heading2");
694: styles[HTML_H3] = HTStyleNamed(styleSheet, "Heading3");
695: styles[HTML_H4] = HTStyleNamed(styleSheet, "Heading4");
696: styles[HTML_H5] = HTStyleNamed(styleSheet, "Heading5");
697: styles[HTML_H6] = HTStyleNamed(styleSheet, "Heading6");
698: styles[HTML_H7] = HTStyleNamed(styleSheet, "Heading7");
699:
700: styles[HTML_DL] = HTStyleNamed(styleSheet, "Glossary");
701: styles[HTML_UL] =
702: styles[HTML_OL] = HTStyleNamed(styleSheet, "List");
703: styles[HTML_MENU] = HTStyleNamed(styleSheet, "Menu");
704: styles[HTML_DIR] = HTStyleNamed(styleSheet, "Dir");
705: styles[HTML_DLC] = HTStyleNamed(styleSheet, "GlossaryCompact");
706: styles[HTML_ADDRESS]= HTStyleNamed(styleSheet, "Address");
707: styles[HTML_BLOCKQUOTE]= HTStyleNamed(styleSheet, "BlockQuote");
708: styles[HTML_PLAINTEXT] =
709: styles[HTML_XMP] = HTStyleNamed(styleSheet, "Example");
710: styles[HTML_PRE] = HTStyleNamed(styleSheet, "Preformatted");
711: styles[HTML_LISTING] = HTStyleNamed(styleSheet, "Listing");
712: }
713: /* P U B L I C
714: */
715:
716: /* Structured Object Class
717: ** -----------------------
718: */
719: PUBLIC CONST HTStructuredClass HTMLPresentation = /* As opposed to print etc */
720: {
721: "text/html",
722: HTML_free,
723: HTML_end_document,
724: HTML_put_character, HTML_put_string, HTML_write,
725: HTML_start_element, HTML_end_element,
726: HTML_put_entity
727: };
1.1 timbl 728:
1.4 timbl 729:
1.2 timbl 730: /* New Structured Text object
731: ** --------------------------
732: **
1.4 timbl 733: ** The strutcured stream can generate either presentation,
734: ** or plain text, or HTML.
1.1 timbl 735: */
1.4 timbl 736: PUBLIC HTStructured* HTML_new ARGS3(
1.2 timbl 737: HTParentAnchor *, anchor,
1.4 timbl 738: HTFormat, format_out,
1.2 timbl 739: HTStream*, stream)
1.1 timbl 740: {
741:
1.4 timbl 742: HTStructured * me;
743:
744: if (format_out != WWW_PLAINTEXT && format_out != WWW_PRESENT) {
1.6 timbl 745: HTStream * intermediate = HTStreamStack(WWW_HTML, format_out,
746: stream, anchor);
747: if (intermediate) return HTMLGenerator(intermediate);
1.4 timbl 748: fprintf(stderr, "** Internal error: can't parse HTML to %s\n",
749: HTAtom_name(format_out));
750: exit (-99);
751: }
752:
753: me = (HTStructured*) malloc(sizeof(*me));
754: if (me == NULL) outofmem(__FILE__, "HTML_new");
1.1 timbl 755:
756: if (!got_styles) get_styles();
757:
1.4 timbl 758: me->isa = &HTMLPresentation;
759: me->node_anchor = anchor;
760: me->title.size = 0;
761: me->title.growby = 128;
762: me->title.allocated = 0;
763: me->title.data = 0;
764: me->text = 0;
765: me->style_change = YES; /* Force check leading to text creation */
766: me->new_style = default_style;
767: me->old_style = 0;
768: me->sp = me->stack + MAX_NESTING - 1;
769: me->sp->tag_number = -1; /* INVALID */
770: me->sp->style = default_style; /* INVALID */
1.1 timbl 771:
1.4 timbl 772: me->comment_start = NULL;
773: me->comment_end = NULL;
774: me->target = stream;
775: if (stream) me->targetClass = *stream->isa; /* Copy pointers */
1.1 timbl 776:
1.4 timbl 777: return (HTStructured*) me;
1.1 timbl 778: }
779:
780:
1.2 timbl 781: /* HTConverter for HTML to plain text
782: ** ----------------------------------
1.1 timbl 783: **
1.2 timbl 784: ** This will convert from HTML to presentation or plain text.
1.1 timbl 785: */
1.2 timbl 786: PUBLIC HTStream* HTMLToPlain ARGS3(
787: HTPresentation *, pres,
788: HTParentAnchor *, anchor,
789: HTStream *, sink)
1.1 timbl 790: {
1.4 timbl 791: return SGML_new(&HTML_dtd, HTML_new(anchor, pres->rep_out, sink));
1.1 timbl 792: }
793:
794:
1.2 timbl 795: /* HTConverter for HTML to C code
796: ** ------------------------------
797: **
798: ** C copde is like plain text but all non-preformatted code
799: ** is commented out.
800: ** This will convert from HTML to presentation or plain text.
801: */
802: PUBLIC HTStream* HTMLToC ARGS3(
803: HTPresentation *, pres,
804: HTParentAnchor *, anchor,
805: HTStream *, sink)
1.1 timbl 806: {
1.4 timbl 807:
808: HTStructured * html;
809:
810: (*sink->isa->put_string)(sink, "/* "); /* Before even title */
811: html = HTML_new(anchor, WWW_PLAINTEXT, sink);
1.2 timbl 812: html->comment_start = "/* ";
813: html->comment_end = " */\n"; /* Must start in col 1 for cpp */
1.4 timbl 814: /* HTML_put_string(html,html->comment_start); */
1.2 timbl 815: return SGML_new(&HTML_dtd, html);
1.1 timbl 816: }
817:
818:
1.2 timbl 819: /* Presenter for HTML
820: ** ------------------
821: **
822: ** This will convert from HTML to presentation or plain text.
823: **
824: ** Override this if you have a windows version
1.1 timbl 825: */
1.2 timbl 826: #ifndef GUI
827: PUBLIC HTStream* HTMLPresent ARGS3(
828: HTPresentation *, pres,
829: HTParentAnchor *, anchor,
830: HTStream *, sink)
1.1 timbl 831: {
1.4 timbl 832: return SGML_new(&HTML_dtd, HTML_new(anchor, WWW_PRESENT, NULL));
1.1 timbl 833: }
1.2 timbl 834: #endif
1.1 timbl 835:
836:
1.2 timbl 837: /* Record error message as a hypertext object
838: ** ------------------------------------------
839: **
840: ** The error message should be marked as an error so that
841: ** it can be reloaded later.
842: ** This implementation just throws up an error message
843: ** and leaves the document unloaded.
1.9 timbl 844: ** A smarter implementation would load an error document,
845: ** marking at such so that it is retried on reload.
1.1 timbl 846: **
1.2 timbl 847: ** On entry,
848: ** sink is a stream to the output device if any
849: ** number is the HTTP error number
850: ** message is the human readable message.
1.9 timbl 851: **
852: ** On exit,
853: ** returns a negative number to indicate lack of success in the load.
1.1 timbl 854: */
1.2 timbl 855:
856: PUBLIC int HTLoadError ARGS3(
857: HTStream *, sink,
858: int, number,
859: CONST char *, message)
860: {
861: HTAlert(message); /* @@@@@@@@@@@@@@@@@@@ */
862: return -number;
863: }
864:
Webmaster