Annotation of libwww/Library/src/HTML.c, revision 1.39.2.1
1.39 frystyk 1: /* HTML.c
2: ** STRUCTURED STREAM TO RICH HYPERTEXT CONVERTER
3: **
4: ** (c) COPYRIGHT CERN 1994.
5: ** Please first read the full copyright statement in the file COPYRIGH.
1.1 timbl 6: **
1.2 timbl 7: ** This generates of a hypertext object. It converts from the
8: ** structured stream interface fro HTMl events into the style-
9: ** oriented iunterface of the HText.h interface. This module is
10: ** only used in clients and shouldnot be linked into servers.
1.1 timbl 11: **
1.6 timbl 12: ** Override this module if making a new GUI browser.
1.1 timbl 13: **
1.35 duns 14: ** HISTORY:
15: ** 8 Jul 94 FM Insulate free() from _free structure element.
16: **
1.1 timbl 17: */
1.16 timbl 18:
1.39.2.1! frystyk 19: /* Library include files */
! 20: #include "tcp.h"
! 21: #include "HTUtils.h"
! 22: #include "HTString.h"
1.1 timbl 23: #include "HTAtom.h"
24: #include "HTChunk.h"
25: #include "HText.h"
26: #include "HTStyle.h"
1.3 timbl 27: #include "HTAlert.h"
1.4 timbl 28: #include "HTMLGen.h"
1.8 timbl 29: #include "HTParse.h"
1.39.2.1! frystyk 30: #include "HTML.h"
1.1 timbl 31:
32: extern HTStyleSheet * styleSheet; /* Application-wide */
33:
34: /* Module-wide style cache
35: */
36: PRIVATE int got_styles = 0;
1.16 timbl 37: PRIVATE HTStyle *styles[HTMLP_ELEMENTS];
1.2 timbl 38: PRIVATE HTStyle *default_style;
1.1 timbl 39:
40:
41: /* HTML Object
42: ** -----------
43: */
1.2 timbl 44: #define MAX_NESTING 20 /* Should be checked by parser */
45:
46: typedef struct _stack_element {
47: HTStyle * style;
48: int tag_number;
49: } stack_element;
50:
51: struct _HTStructured {
52: CONST HTStructuredClass * isa;
53: HTParentAnchor * node_anchor;
54: HText * text;
55:
56: HTStream* target; /* Output stream */
57: HTStreamClass targetClass; /* Output routines */
58:
59: HTChunk title; /* Grow by 128 */
60:
61: char * comment_start; /* for literate programming */
62: char * comment_end;
1.16 timbl 63:
64: CONST SGML_dtd* dtd;
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 */
71: stack_element stack[MAX_NESTING];
72: stack_element *sp; /* Style stack pointer */
1.1 timbl 73: };
74:
1.2 timbl 75: struct _HTStream {
76: CONST HTStreamClass * isa;
77: /* .... */
78: };
1.1 timbl 79:
80: /* Forward declarations of routines
81: */
82: PRIVATE void get_styles NOPARAMS;
83:
84:
1.4 timbl 85: PRIVATE void actually_set_style PARAMS((HTStructured * me));
1.11 timbl 86: PRIVATE void change_paragraph_style PARAMS((HTStructured * me, HTStyle * style));
1.1 timbl 87:
88: /* Style buffering avoids dummy paragraph begin/ends.
89: */
1.4 timbl 90: #define UPDATE_STYLE if (me->style_change) { actually_set_style(me); }
1.1 timbl 91:
92:
1.2 timbl 93: #ifdef OLD_CODE
1.1 timbl 94: /* The following accented characters are from peter Flynn, curia project */
95:
96: /* these ifdefs don't solve the problem of a simple terminal emulator
97: ** with a different character set to the client machine. But nothing does,
98: ** except looking at the TERM setting */
99:
1.2 timbl 100:
1.1 timbl 101: { "ocus" , "&" }, /* for CURIA */
102: #ifdef IBMPC
103: { "aacute" , "\240" }, /* For PC display */
104: { "eacute" , "\202" },
105: { "iacute" , "\241" },
106: { "oacute" , "\242" },
107: { "uacute" , "\243" },
108: { "Aacute" , "\101" },
109: { "Eacute" , "\220" },
110: { "Iacute" , "\111" },
111: { "Oacute" , "\117" },
112: { "Uacute" , "\125" },
113: #else
114: { "aacute" , "\341" }, /* Works for openwindows -- Peter Flynn */
115: { "eacute" , "\351" },
116: { "iacute" , "\355" },
117: { "oacute" , "\363" },
118: { "uacute" , "\372" },
119: { "Aacute" , "\301" },
120: { "Eacute" , "\310" },
121: { "Iacute" , "\315" },
122: { "Oacute" , "\323" },
123: { "Uacute" , "\332" },
124: #endif
125: { 0, 0 } /* Terminate list */
126: };
1.2 timbl 127: #endif
1.1 timbl 128:
129:
1.2 timbl 130: /* Entity values -- for ISO Latin 1 local representation
131: **
132: ** This MUST match exactly the table referred to in the DTD!
133: */
134: static char * ISO_Latin1[] = {
135: "\306", /* capital AE diphthong (ligature) */
136: "\301", /* capital A, acute accent */
137: "\302", /* capital A, circumflex accent */
138: "\300", /* capital A, grave accent */
139: "\305", /* capital A, ring */
140: "\303", /* capital A, tilde */
141: "\304", /* capital A, dieresis or umlaut mark */
142: "\307", /* capital C, cedilla */
143: "\320", /* capital Eth, Icelandic */
144: "\311", /* capital E, acute accent */
145: "\312", /* capital E, circumflex accent */
146: "\310", /* capital E, grave accent */
147: "\313", /* capital E, dieresis or umlaut mark */
148: "\315", /* capital I, acute accent */
149: "\316", /* capital I, circumflex accent */
150: "\314", /* capital I, grave accent */
151: "\317", /* capital I, dieresis or umlaut mark */
152: "\321", /* capital N, tilde */
153: "\323", /* capital O, acute accent */
154: "\324", /* capital O, circumflex accent */
155: "\322", /* capital O, grave accent */
156: "\330", /* capital O, slash */
157: "\325", /* capital O, tilde */
158: "\326", /* capital O, dieresis or umlaut mark */
159: "\336", /* capital THORN, Icelandic */
160: "\332", /* capital U, acute accent */
161: "\333", /* capital U, circumflex accent */
162: "\331", /* capital U, grave accent */
163: "\334", /* capital U, dieresis or umlaut mark */
164: "\335", /* capital Y, acute accent */
165: "\341", /* small a, acute accent */
166: "\342", /* small a, circumflex accent */
167: "\346", /* small ae diphthong (ligature) */
168: "\340", /* small a, grave accent */
169: "\046", /* ampersand */
170: "\345", /* small a, ring */
171: "\343", /* small a, tilde */
172: "\344", /* small a, dieresis or umlaut mark */
173: "\347", /* small c, cedilla */
174: "\351", /* small e, acute accent */
175: "\352", /* small e, circumflex accent */
176: "\350", /* small e, grave accent */
177: "\360", /* small eth, Icelandic */
178: "\353", /* small e, dieresis or umlaut mark */
179: "\076", /* greater than */
180: "\355", /* small i, acute accent */
181: "\356", /* small i, circumflex accent */
182: "\354", /* small i, grave accent */
183: "\357", /* small i, dieresis or umlaut mark */
184: "\074", /* less than */
185: "\361", /* small n, tilde */
186: "\363", /* small o, acute accent */
187: "\364", /* small o, circumflex accent */
188: "\362", /* small o, grave accent */
189: "\370", /* small o, slash */
190: "\365", /* small o, tilde */
191: "\366", /* small o, dieresis or umlaut mark */
1.36 frystyk 192: "\042", /* double quote sign - June 94 */
1.2 timbl 193: "\337", /* small sharp s, German (sz ligature) */
194: "\376", /* small thorn, Icelandic */
195: "\372", /* small u, acute accent */
196: "\373", /* small u, circumflex accent */
197: "\371", /* small u, grave accent */
198: "\374", /* small u, dieresis or umlaut mark */
199: "\375", /* small y, acute accent */
200: "\377", /* small y, dieresis or umlaut mark */
1.1 timbl 201: };
202:
1.2 timbl 203:
204: /* Entity values -- for NeXT local representation
205: **
206: ** This MUST match exactly the table referred to in the DTD!
207: **
208: */
209: static char * NeXTCharacters[] = {
210: "\341", /* capital AE diphthong (ligature) */
211: "\202", /* capital A, acute accent */
212: "\203", /* capital A, circumflex accent */
213: "\201", /* capital A, grave accent */
214: "\206", /* capital A, ring */
215: "\204", /* capital A, tilde */
216: "\205", /* capital A, dieresis or umlaut mark */
217: "\207", /* capital C, cedilla */
218: "\220", /* capital Eth, Icelandic */
219: "\211", /* capital E, acute accent */
220: "\212", /* capital E, circumflex accent */
221: "\210", /* capital E, grave accent */
222: "\213", /* capital E, dieresis or umlaut mark */
223: "\215", /* capital I, acute accent */
224: "\216", /* capital I, circumflex accent these are */
225: "\214", /* capital I, grave accent ISO -100 hex */
226: "\217", /* capital I, dieresis or umlaut mark */
227: "\221", /* capital N, tilde */
228: "\223", /* capital O, acute accent */
229: "\224", /* capital O, circumflex accent */
230: "\222", /* capital O, grave accent */
231: "\351", /* capital O, slash 'cept this */
232: "\225", /* capital O, tilde */
233: "\226", /* capital O, dieresis or umlaut mark */
234: "\234", /* capital THORN, Icelandic */
235: "\230", /* capital U, acute accent */
236: "\231", /* capital U, circumflex accent */
237: "\227", /* capital U, grave accent */
238: "\232", /* capital U, dieresis or umlaut mark */
239: "\233", /* capital Y, acute accent */
240: "\326", /* small a, acute accent */
241: "\327", /* small a, circumflex accent */
242: "\361", /* small ae diphthong (ligature) */
243: "\325", /* small a, grave accent */
244: "\046", /* ampersand */
245: "\332", /* small a, ring */
246: "\330", /* small a, tilde */
247: "\331", /* small a, dieresis or umlaut mark */
248: "\333", /* small c, cedilla */
249: "\335", /* small e, acute accent */
250: "\336", /* small e, circumflex accent */
251: "\334", /* small e, grave accent */
252: "\346", /* small eth, Icelandic */
253: "\337", /* small e, dieresis or umlaut mark */
254: "\076", /* greater than */
255: "\342", /* small i, acute accent */
256: "\344", /* small i, circumflex accent */
257: "\340", /* small i, grave accent */
258: "\345", /* small i, dieresis or umlaut mark */
259: "\074", /* less than */
260: "\347", /* small n, tilde */
261: "\355", /* small o, acute accent */
262: "\356", /* small o, circumflex accent */
263: "\354", /* small o, grave accent */
264: "\371", /* small o, slash */
265: "\357", /* small o, tilde */
266: "\360", /* small o, dieresis or umlaut mark */
1.36 frystyk 267: "\042", /* double quote sign - June 94 */
1.2 timbl 268: "\373", /* small sharp s, German (sz ligature) */
269: "\374", /* small thorn, Icelandic */
270: "\363", /* small u, acute accent */
271: "\364", /* small u, circumflex accent */
272: "\362", /* small u, grave accent */
273: "\366", /* small u, dieresis or umlaut mark */
274: "\367", /* small y, acute accent */
275: "\375", /* small y, dieresis or umlaut mark */
1.1 timbl 276: };
277:
1.2 timbl 278: /* Entity values -- for IBM/PC Code Page 850 (International)
279: **
280: ** This MUST match exactly the table referred to in the DTD!
281: **
282: */
283: /* @@@@@@@@@@@@@@@@@ TBD */
284:
285:
286:
287: /* Set character set
288: ** ----------------
289: */
290:
291: PRIVATE char** p_entity_values = ISO_Latin1; /* Pointer to translation */
1.1 timbl 292:
1.2 timbl 293: PUBLIC void HTMLUseCharacterSet ARGS1(HTMLCharacterSet, i)
294: {
295: p_entity_values = (i == HTML_NEXT_CHARS) ? NeXTCharacters
296: : ISO_Latin1;
297: }
1.1 timbl 298:
299:
300: /* Flattening the style structure
301: ** ------------------------------
302: **
303: On the NeXT, and on any read-only browser, it is simpler for the text to have
304: a sequence of styles, rather than a nested tree of styles. In this
305: case we have to flatten the structure as it arrives from SGML tags into
306: a sequence of styles.
307: */
308:
309: /* If style really needs to be set, call this
310: */
1.4 timbl 311: PRIVATE void actually_set_style ARGS1(HTStructured *, me)
1.1 timbl 312: {
1.4 timbl 313: if (!me->text) { /* First time through */
314: me->text = HText_new2(me->node_anchor, me->target);
315: HText_beginAppend(me->text);
316: HText_setStyle(me->text, me->new_style);
317: me->in_word = NO;
1.1 timbl 318: } else {
1.4 timbl 319: HText_setStyle(me->text, me->new_style);
1.1 timbl 320: }
1.4 timbl 321: me->old_style = me->new_style;
322: me->style_change = NO;
1.1 timbl 323: }
324:
325: /* If you THINK you need to change style, call this
326: */
327:
1.11 timbl 328: PRIVATE void change_paragraph_style ARGS2(HTStructured *, me, HTStyle *,style)
1.1 timbl 329: {
1.4 timbl 330: if (me->new_style!=style) {
331: me->style_change = YES;
332: me->new_style = style;
1.1 timbl 333: }
1.11 timbl 334: me->in_word = NO;
1.1 timbl 335: }
336:
1.2 timbl 337: /*_________________________________________________________________________
338: **
339: ** A C T I O N R O U T I N E S
340: */
341:
342: /* Character handling
343: ** ------------------
1.1 timbl 344: */
1.4 timbl 345: PRIVATE void HTML_put_character ARGS2(HTStructured *, me, char, c)
1.1 timbl 346: {
1.2 timbl 347:
1.4 timbl 348: switch (me->sp[0].tag_number) {
1.2 timbl 349: case HTML_COMMENT:
350: break; /* Do Nothing */
351:
352: case HTML_TITLE:
1.4 timbl 353: HTChunkPutc(&me->title, c);
1.2 timbl 354: break;
355:
356:
357: case HTML_LISTING: /* Litteral text */
358: case HTML_XMP:
359: case HTML_PLAINTEXT:
360: case HTML_PRE:
361: /* We guarrantee that the style is up-to-date in begin_litteral
362: */
1.4 timbl 363: HText_appendCharacter(me->text, c);
1.2 timbl 364: break;
365:
366: default: /* Free format text */
1.4 timbl 367: if (me->style_change) {
1.2 timbl 368: if ((c=='\n') || (c==' ')) return; /* Ignore it */
369: UPDATE_STYLE;
370: }
371: if (c=='\n') {
1.4 timbl 372: if (me->in_word) {
373: HText_appendCharacter(me->text, ' ');
374: me->in_word = NO;
1.2 timbl 375: }
376: } else {
1.4 timbl 377: HText_appendCharacter(me->text, c);
378: me->in_word = YES;
1.2 timbl 379: }
380: } /* end switch */
1.1 timbl 381: }
382:
1.2 timbl 383:
384:
385: /* String handling
386: ** ---------------
387: **
388: ** This is written separately from put_character becuase the loop can
1.11 timbl 389: ** in some cases be promoted to a higher function call level for speed.
1.2 timbl 390: */
1.4 timbl 391: PRIVATE void HTML_put_string ARGS2(HTStructured *, me, CONST char*, s)
1.1 timbl 392: {
1.2 timbl 393:
1.4 timbl 394: switch (me->sp[0].tag_number) {
1.2 timbl 395: case HTML_COMMENT:
396: break; /* Do Nothing */
397:
398: case HTML_TITLE:
1.4 timbl 399: HTChunkPuts(&me->title, s);
1.2 timbl 400: break;
401:
402:
403: case HTML_LISTING: /* Litteral text */
404: case HTML_XMP:
405: case HTML_PLAINTEXT:
406: case HTML_PRE:
407:
408: /* We guarrantee that the style is up-to-date in begin_litteral
409: */
1.4 timbl 410: HText_appendText(me->text, s);
1.2 timbl 411: break;
412:
413: default: /* Free format text */
414: {
415: CONST char *p = s;
1.4 timbl 416: if (me->style_change) {
1.2 timbl 417: for (; *p && ((*p=='\n') || (*p==' ')); p++) ; /* Ignore leaders */
418: if (!*p) return;
419: UPDATE_STYLE;
420: }
421: for(; *p; p++) {
1.4 timbl 422: if (me->style_change) {
1.2 timbl 423: if ((*p=='\n') || (*p==' ')) continue; /* Ignore it */
424: UPDATE_STYLE;
425: }
426: if (*p=='\n') {
1.4 timbl 427: if (me->in_word) {
428: HText_appendCharacter(me->text, ' ');
429: me->in_word = NO;
1.2 timbl 430: }
431: } else {
1.4 timbl 432: HText_appendCharacter(me->text, *p);
433: me->in_word = YES;
1.2 timbl 434: }
435: } /* for */
436: }
437: } /* end switch */
1.1 timbl 438: }
439:
440:
1.2 timbl 441: /* Buffer write
1.3 timbl 442: ** ------------
1.1 timbl 443: */
1.4 timbl 444: PRIVATE void HTML_write ARGS3(HTStructured *, me, CONST char*, s, int, l)
1.1 timbl 445: {
1.38 frystyk 446: while (l-- > 0)
447: HTML_put_character(me, *s++);
1.1 timbl 448: }
1.2 timbl 449:
450:
451: /* Start Element
452: ** -------------
453: */
454: PRIVATE void HTML_start_element ARGS4(
1.4 timbl 455: HTStructured *, me,
1.16 timbl 456: int, element_number,
1.3 timbl 457: CONST BOOL*, present,
1.16 timbl 458: CONST char **, value)
1.2 timbl 459: {
460: switch (element_number) {
461: case HTML_A:
462: {
1.8 timbl 463: HTChildAnchor * source;
1.9 timbl 464: char * href = NULL;
465: if (present[HTML_A_HREF]) {
466: StrAllocCopy(href, value[HTML_A_HREF]);
1.36 frystyk 467: #ifdef OLD_CODE
1.9 timbl 468: HTSimplify(href);
1.36 frystyk 469: #endif
1.9 timbl 470: }
1.8 timbl 471: source = HTAnchor_findChildAndLink(
1.4 timbl 472: me->node_anchor, /* parent */
1.2 timbl 473: present[HTML_A_NAME] ? value[HTML_A_NAME] : 0, /* Tag */
1.9 timbl 474: present[HTML_A_HREF] ? href : 0, /* Addresss */
1.16 timbl 475: present[HTML_A_REL] && value[HTML_A_REL] ?
476: (HTLinkType*)HTAtom_for(value[HTML_A_REL])
1.2 timbl 477: : 0);
478:
479: if (present[HTML_A_TITLE] && value[HTML_A_TITLE]) {
480: HTParentAnchor * dest =
481: HTAnchor_parent(
482: HTAnchor_followMainLink((HTAnchor*)source)
483: );
484: if (!HTAnchor_title(dest))
485: HTAnchor_setTitle(dest, value[HTML_A_TITLE]);
486: }
487: UPDATE_STYLE;
1.4 timbl 488: HText_beginAnchor(me->text, source);
1.18 frystyk 489: free(href); /* Leak fix Henrik 17/02-94 */
1.2 timbl 490: }
491: break;
492:
493: case HTML_TITLE:
1.4 timbl 494: HTChunkClear(&me->title);
1.2 timbl 495: break;
496:
497: case HTML_NEXTID:
498: /* if (present[NEXTID_N] && value[NEXTID_N])
1.4 timbl 499: HText_setNextId(me->text, atoi(value[NEXTID_N])); */
1.2 timbl 500: break;
501:
502: case HTML_ISINDEX:
1.4 timbl 503: HTAnchor_setIndex(me->node_anchor);
1.2 timbl 504: break;
505:
1.15 timbl 506: case HTML_BR:
507: UPDATE_STYLE;
508: HText_appendCharacter(me->text, '\n');
509: me->in_word = NO;
510: break;
511:
512: case HTML_HR:
513: UPDATE_STYLE;
514: HText_appendCharacter(me->text, '\n');
1.16 timbl 515: HText_appendText(me->text, "___________________________________");
1.15 timbl 516: HText_appendCharacter(me->text, '\n');
517: me->in_word = NO;
518: break;
519:
1.2 timbl 520: case HTML_P:
521: UPDATE_STYLE;
1.4 timbl 522: HText_appendParagraph(me->text);
523: me->in_word = NO;
1.2 timbl 524: break;
525:
526: case HTML_DL:
1.11 timbl 527: change_paragraph_style(me, present && present[DL_COMPACT]
1.16 timbl 528: ? styles[HTML_DL]
1.2 timbl 529: : styles[HTML_DL]);
530: break;
531:
532: case HTML_DT:
1.4 timbl 533: if (!me->style_change) {
534: HText_appendParagraph(me->text);
535: me->in_word = NO;
1.2 timbl 536: }
537: break;
538:
539: case HTML_DD:
540: UPDATE_STYLE;
1.4 timbl 541: HTML_put_character(me, '\t'); /* Just tab out one stop */
542: me->in_word = NO;
543: break;
1.2 timbl 544:
545: case HTML_UL:
546: case HTML_OL:
547: case HTML_MENU:
548: case HTML_DIR:
1.11 timbl 549: change_paragraph_style(me, styles[element_number]);
1.2 timbl 550: break;
551:
552: case HTML_LI:
553: UPDATE_STYLE;
1.7 timbl 554: if (me->sp[0].tag_number != HTML_DIR)
1.4 timbl 555: HText_appendParagraph(me->text);
1.2 timbl 556: else
1.4 timbl 557: HText_appendCharacter(me->text, '\t'); /* Tab @@ nl for UL? */
558: me->in_word = NO;
1.2 timbl 559: break;
560:
561: case HTML_LISTING: /* Litteral text */
562: case HTML_XMP:
563: case HTML_PLAINTEXT:
564: case HTML_PRE:
1.11 timbl 565: change_paragraph_style(me, styles[element_number]);
1.2 timbl 566: UPDATE_STYLE;
1.4 timbl 567: if (me->comment_end)
568: HText_appendText(me->text, me->comment_end);
1.2 timbl 569: break;
1.11 timbl 570:
1.23 frystyk 571: case HTML_IMG: /* Images */
572: {
573: HTChildAnchor *source;
574: char *src = NULL;
575: if (present[HTML_IMG_SRC]) {
576: StrAllocCopy(src, value[HTML_IMG_SRC]);
1.36 frystyk 577: #ifdef OLD_CODE
1.23 frystyk 578: HTSimplify(src);
1.36 frystyk 579: #endif
1.23 frystyk 580: }
581: source = HTAnchor_findChildAndLink(
582: me->node_anchor, /* parent */
583: 0, /* Tag */
584: src ? src : 0, /* Addresss */
585: 0);
586: UPDATE_STYLE;
587: HText_appendImage(me->text, source,
1.24 frystyk 588: present[HTML_IMG_ALT] ? value[HTML_IMG_ALT] : NULL,
589: present[HTML_IMG_ALIGN] ? value[HTML_IMG_ALIGN] : NULL,
590: present[HTML_IMG_ISMAP] ? YES : NO);
1.23 frystyk 591: free(src);
1.24 frystyk 592: }
593: break;
594:
595: case HTML_HTML: /* Ignore these altogether */
596: case HTML_HEAD:
597: case HTML_BODY:
598:
1.10 timbl 599: case HTML_TT: /* Physical character highlighting */
600: case HTML_B: /* Currently ignored */
601: case HTML_I:
602: case HTML_U:
603:
604: case HTML_EM: /* Logical character highlighting */
605: case HTML_STRONG: /* Currently ignored */
606: case HTML_CODE:
607: case HTML_SAMP:
608: case HTML_KBD:
609: case HTML_VAR:
610: case HTML_DFN:
611: case HTML_CITE:
612: break;
613:
1.11 timbl 614: case HTML_H1: /* paragraph styles */
615: case HTML_H2:
616: case HTML_H3:
617: case HTML_H4:
618: case HTML_H5:
619: case HTML_H6:
620: case HTML_H7:
621: case HTML_ADDRESS:
622: case HTML_BLOCKQUOTE:
623: change_paragraph_style(me, styles[element_number]); /* May be postponed */
1.2 timbl 624: break;
625:
626: } /* end switch */
627:
1.16 timbl 628: if (me->dtd->tags[element_number].contents!= SGML_EMPTY) {
1.13 timbl 629: if (me->sp == me->stack) {
1.39.2.1! frystyk 630: fprintf(TDEST, "HTML: ****** Maximum nesting of %d exceded!\n",
1.12 timbl 631: MAX_NESTING);
632: return;
633: }
1.4 timbl 634: --(me->sp);
635: me->sp[0].style = me->new_style; /* Stack new style */
636: me->sp[0].tag_number = element_number;
1.10 timbl 637: }
1.1 timbl 638: }
1.10 timbl 639:
1.2 timbl 640:
1.1 timbl 641: /* End Element
1.2 timbl 642: ** -----------
1.1 timbl 643: **
1.2 timbl 644: */
645: /* When we end an element, the style must be returned to that
1.1 timbl 646: ** in effect before that element. Note that anchors (etc?)
647: ** don't have an associated style, so that we must scan down the
648: ** stack for an element with a defined style. (In fact, the styles
649: ** should be linked to the whole stack not just the top one.)
650: ** TBL 921119
1.6 timbl 651: **
652: ** We don't turn on "CAREFUL" check because the parser produces
653: ** (internal code errors apart) good nesting. The parser checks
654: ** incoming code errors, not this module.
1.1 timbl 655: */
1.4 timbl 656: PRIVATE void HTML_end_element ARGS2(HTStructured *, me, int , element_number)
1.1 timbl 657: {
1.2 timbl 658: #ifdef CAREFUL /* parser assumed to produce good nesting */
1.4 timbl 659: if (element_number != me->sp[0].tag_number) {
1.39.2.1! frystyk 660: fprintf(TDEST, "HTMLText: end of element %s when expecting end of %s\n",
1.16 timbl 661: me->dtd->tags[element_number].name,
662: me->dtd->tags[me->sp->tag_number].name);
1.6 timbl 663: /* panic */
1.1 timbl 664: }
1.2 timbl 665: #endif
666:
1.4 timbl 667: me->sp++; /* Pop state off stack */
1.2 timbl 668:
669: switch(element_number) {
670:
671: case HTML_A:
672: UPDATE_STYLE;
1.4 timbl 673: HText_endAnchor(me->text);
1.2 timbl 674: break;
675:
676: case HTML_TITLE:
1.4 timbl 677: HTChunkTerminate(&me->title);
678: HTAnchor_setTitle(me->node_anchor, me->title.data);
1.2 timbl 679: break;
680:
681: case HTML_LISTING: /* Litteral text */
682: case HTML_XMP:
683: case HTML_PLAINTEXT:
684: case HTML_PRE:
1.4 timbl 685: if (me->comment_start)
686: HText_appendText(me->text, me->comment_start);
1.2 timbl 687: /* Fall through */
688:
689: default:
690:
1.11 timbl 691: change_paragraph_style(me, me->sp->style); /* Often won't really change */
1.2 timbl 692: break;
693:
694: } /* switch */
1.1 timbl 695: }
696:
1.2 timbl 697:
698: /* Expanding entities
699: ** ------------------
700: */
701: /* (In fact, they all shrink!)
1.1 timbl 702: */
1.2 timbl 703:
1.4 timbl 704: PRIVATE void HTML_put_entity ARGS2(HTStructured *, me, int, entity_number)
1.1 timbl 705: {
1.4 timbl 706: HTML_put_string(me, ISO_Latin1[entity_number]); /* @@ Other representations */
1.1 timbl 707: }
1.2 timbl 708:
709:
710: /* Free an HTML object
711: ** -------------------
712: **
1.4 timbl 713: ** If the document is empty, the text object will not yet exist.
714: So we could in fact abandon creating the document and return
715: an error code. In fact an empty document is an important type
716: of document, so we don't.
717: **
718: ** If non-interactive, everything is freed off. No: crashes -listrefs
1.2 timbl 719: ** Otherwise, the interactive object is left.
720: */
1.37 frystyk 721: PUBLIC int HTML_free ARGS1(HTStructured *, me)
1.1 timbl 722: {
1.4 timbl 723: UPDATE_STYLE; /* Creates empty document here! */
724: if (me->comment_end)
725: HTML_put_string(me,me->comment_end);
726: HText_endAppend(me->text);
727:
728: if (me->target) {
1.35 duns 729: (*me->targetClass._free)(me->target);
1.2 timbl 730: }
1.19 frystyk 731: HTChunkClear(&me->title); /* Henrik 18/02-94 */
1.4 timbl 732: free(me);
1.37 frystyk 733: return 0;
1.1 timbl 734: }
735:
736:
1.37 frystyk 737: PRIVATE int HTML_abort ARGS2(HTStructured *, me, HTError, e)
1.1 timbl 738:
1.14 timbl 739: {
740: if (me->target) {
741: (*me->targetClass.abort)(me->target, e);
742: }
1.19 frystyk 743: HTChunkClear(&me->title); /* Henrik 18/02-94 */
1.14 timbl 744: free(me);
1.37 frystyk 745: return EOF;
1.1 timbl 746: }
747:
1.2 timbl 748:
749: /* Get Styles from style sheet
750: ** ---------------------------
751: */
752: PRIVATE void get_styles NOARGS
1.1 timbl 753: {
1.2 timbl 754: got_styles = YES;
755:
756: default_style = HTStyleNamed(styleSheet, "Normal");
1.1 timbl 757:
1.2 timbl 758: styles[HTML_H1] = HTStyleNamed(styleSheet, "Heading1");
759: styles[HTML_H2] = HTStyleNamed(styleSheet, "Heading2");
760: styles[HTML_H3] = HTStyleNamed(styleSheet, "Heading3");
761: styles[HTML_H4] = HTStyleNamed(styleSheet, "Heading4");
762: styles[HTML_H5] = HTStyleNamed(styleSheet, "Heading5");
763: styles[HTML_H6] = HTStyleNamed(styleSheet, "Heading6");
764: styles[HTML_H7] = HTStyleNamed(styleSheet, "Heading7");
765:
766: styles[HTML_DL] = HTStyleNamed(styleSheet, "Glossary");
767: styles[HTML_UL] =
768: styles[HTML_OL] = HTStyleNamed(styleSheet, "List");
769: styles[HTML_MENU] = HTStyleNamed(styleSheet, "Menu");
770: styles[HTML_DIR] = HTStyleNamed(styleSheet, "Dir");
1.16 timbl 771: /* styles[HTML_DLC] = HTStyleNamed(styleSheet, "GlossaryCompact"); */
1.2 timbl 772: styles[HTML_ADDRESS]= HTStyleNamed(styleSheet, "Address");
773: styles[HTML_BLOCKQUOTE]= HTStyleNamed(styleSheet, "BlockQuote");
774: styles[HTML_PLAINTEXT] =
775: styles[HTML_XMP] = HTStyleNamed(styleSheet, "Example");
776: styles[HTML_PRE] = HTStyleNamed(styleSheet, "Preformatted");
777: styles[HTML_LISTING] = HTStyleNamed(styleSheet, "Listing");
778: }
779: /* P U B L I C
780: */
781:
782: /* Structured Object Class
783: ** -----------------------
784: */
785: PUBLIC CONST HTStructuredClass HTMLPresentation = /* As opposed to print etc */
786: {
787: "text/html",
788: HTML_free,
1.14 timbl 789: HTML_abort,
1.2 timbl 790: HTML_put_character, HTML_put_string, HTML_write,
791: HTML_start_element, HTML_end_element,
792: HTML_put_entity
793: };
1.1 timbl 794:
1.4 timbl 795:
1.2 timbl 796: /* New Structured Text object
797: ** --------------------------
798: **
1.16 timbl 799: ** The structured stream can generate either presentation,
1.4 timbl 800: ** or plain text, or HTML.
1.1 timbl 801: */
1.16 timbl 802: PUBLIC HTStructured* HTML_new ARGS5(
803: HTRequest *, request,
804: void *, param,
805: HTFormat, input_format,
806: HTFormat, output_format,
807: HTStream *, output_stream)
1.1 timbl 808: {
809:
1.4 timbl 810: HTStructured * me;
811:
1.16 timbl 812: if (output_format != WWW_PLAINTEXT
813: && output_format != WWW_PRESENT
814: && output_format != HTAtom_for("text/x-c")) {
1.37 frystyk 815: HTStream * intermediate = HTStreamStack(WWW_HTML, output_format,
816: output_stream, request, NO);
1.6 timbl 817: if (intermediate) return HTMLGenerator(intermediate);
1.39.2.1! frystyk 818: fprintf(TDEST, "** Internal error: can't parse HTML to %s\n",
1.16 timbl 819: HTAtom_name(output_format));
1.4 timbl 820: exit (-99);
821: }
822:
823: me = (HTStructured*) malloc(sizeof(*me));
824: if (me == NULL) outofmem(__FILE__, "HTML_new");
1.1 timbl 825:
826: if (!got_styles) get_styles();
827:
1.4 timbl 828: me->isa = &HTMLPresentation;
1.16 timbl 829: me->dtd = &DTD;
830: me->node_anchor = request->anchor;
1.4 timbl 831: me->title.size = 0;
832: me->title.growby = 128;
833: me->title.allocated = 0;
834: me->title.data = 0;
835: me->text = 0;
836: me->style_change = YES; /* Force check leading to text creation */
837: me->new_style = default_style;
838: me->old_style = 0;
839: me->sp = me->stack + MAX_NESTING - 1;
840: me->sp->tag_number = -1; /* INVALID */
841: me->sp->style = default_style; /* INVALID */
1.1 timbl 842:
1.4 timbl 843: me->comment_start = NULL;
844: me->comment_end = NULL;
1.16 timbl 845: me->target = output_stream;
846: if (output_stream) me->targetClass = *output_stream->isa; /* Copy pointers */
1.1 timbl 847:
1.4 timbl 848: return (HTStructured*) me;
1.1 timbl 849: }
850:
851:
1.2 timbl 852: /* HTConverter for HTML to plain text
853: ** ----------------------------------
1.1 timbl 854: **
1.2 timbl 855: ** This will convert from HTML to presentation or plain text.
1.1 timbl 856: */
1.16 timbl 857: PUBLIC HTStream* HTMLToPlain ARGS5(
858: HTRequest *, request,
859: void *, param,
860: HTFormat, input_format,
861: HTFormat, output_format,
862: HTStream *, output_stream)
1.1 timbl 863: {
1.16 timbl 864: return SGML_new(&DTD, HTML_new(
865: request, NULL, input_format, output_format, output_stream));
1.1 timbl 866: }
867:
868:
1.2 timbl 869: /* HTConverter for HTML to C code
870: ** ------------------------------
871: **
1.36 frystyk 872: ** C code is like plain text but all non-preformatted code
1.2 timbl 873: ** is commented out.
874: ** This will convert from HTML to presentation or plain text.
875: */
1.16 timbl 876: PUBLIC HTStream* HTMLToC ARGS5(
877: HTRequest *, request,
878: void *, param,
879: HTFormat, input_format,
880: HTFormat, output_format,
881: HTStream *, output_stream)
1.1 timbl 882: {
1.4 timbl 883:
884: HTStructured * html;
885:
1.36 frystyk 886: (*output_stream->isa->put_string)(output_stream, "/* "); /* Before title */
1.16 timbl 887: html = HTML_new(request, NULL, input_format, output_format, output_stream);
1.2 timbl 888: html->comment_start = "/* ";
1.16 timbl 889: html->dtd = &DTD;
1.2 timbl 890: html->comment_end = " */\n"; /* Must start in col 1 for cpp */
1.16 timbl 891: return SGML_new(&DTD, html);
1.1 timbl 892: }
893:
894:
1.2 timbl 895: /* Presenter for HTML
896: ** ------------------
897: **
898: ** This will convert from HTML to presentation or plain text.
899: **
900: ** Override this if you have a windows version
1.1 timbl 901: */
1.2 timbl 902: #ifndef GUI
1.16 timbl 903: PUBLIC HTStream* HTMLPresent ARGS5(
904: HTRequest *, request,
905: void *, param,
906: HTFormat, input_format,
907: HTFormat, output_format,
908: HTStream *, output_stream)
1.1 timbl 909: {
1.16 timbl 910: return SGML_new(&DTD, HTML_new(
911: request, NULL, input_format, output_format, output_stream));
1.1 timbl 912: }
1.2 timbl 913: #endif
1.1 timbl 914:
915:
1.2 timbl 916: /* Record error message as a hypertext object
917: ** ------------------------------------------
918: **
919: ** The error message should be marked as an error so that
920: ** it can be reloaded later.
921: ** This implementation just throws up an error message
922: ** and leaves the document unloaded.
1.9 timbl 923: ** A smarter implementation would load an error document,
924: ** marking at such so that it is retried on reload.
1.1 timbl 925: **
1.2 timbl 926: ** On entry,
927: ** sink is a stream to the output device if any
928: ** number is the HTTP error number
929: ** message is the human readable message.
1.9 timbl 930: **
931: ** On exit,
932: ** returns a negative number to indicate lack of success in the load.
1.1 timbl 933: */
1.2 timbl 934:
935: PUBLIC int HTLoadError ARGS3(
1.17 luotonen 936: HTRequest *, req,
1.2 timbl 937: int, number,
938: CONST char *, message)
939: {
1.20 frystyk 940: char *err = "Oh I screwed up!"; /* Dummy pointer not used (I hope) */
1.2 timbl 941: HTAlert(message); /* @@@@@@@@@@@@@@@@@@@ */
1.20 frystyk 942: /* Clean up! Henrik 04/03-94 */
943: if (req && req->output_stream)
944: (*req->output_stream->isa->abort)(req->output_stream, err);
1.33 frystyk 945: #if OLD_CODE
1.25 luotonen 946: HTClearErrors(req);
1.33 frystyk 947: #endif
1.2 timbl 948: return -number;
949: }
1.29 frystyk 950:
Webmaster