Annotation of libwww/Library/src/HTML.c, revision 1.44
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.
1.1 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:
1.41 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"
1.41 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 */
1.44 ! frystyk 71:
! 72: stack_element stack[MAX_NESTING];
! 73: stack_element *sp; /* Style stack pointer */
! 74: int overflow; /* Keep track of overflow nesting */
1.1 timbl 75: };
76:
1.2 timbl 77: struct _HTStream {
78: CONST HTStreamClass * isa;
79: /* .... */
80: };
1.1 timbl 81:
82: /* Forward declarations of routines
83: */
84: PRIVATE void get_styles NOPARAMS;
85:
86:
1.4 timbl 87: PRIVATE void actually_set_style PARAMS((HTStructured * me));
1.11 timbl 88: PRIVATE void change_paragraph_style PARAMS((HTStructured * me, HTStyle * style));
1.1 timbl 89:
90: /* Style buffering avoids dummy paragraph begin/ends.
91: */
1.4 timbl 92: #define UPDATE_STYLE if (me->style_change) { actually_set_style(me); }
1.1 timbl 93:
94:
1.2 timbl 95: #ifdef OLD_CODE
1.1 timbl 96: /* The following accented characters are from peter Flynn, curia project */
97:
98: /* these ifdefs don't solve the problem of a simple terminal emulator
99: ** with a different character set to the client machine. But nothing does,
100: ** except looking at the TERM setting */
101:
1.2 timbl 102:
1.1 timbl 103: { "ocus" , "&" }, /* for CURIA */
104: #ifdef IBMPC
105: { "aacute" , "\240" }, /* For PC display */
106: { "eacute" , "\202" },
107: { "iacute" , "\241" },
108: { "oacute" , "\242" },
109: { "uacute" , "\243" },
110: { "Aacute" , "\101" },
111: { "Eacute" , "\220" },
112: { "Iacute" , "\111" },
113: { "Oacute" , "\117" },
114: { "Uacute" , "\125" },
115: #else
116: { "aacute" , "\341" }, /* Works for openwindows -- Peter Flynn */
117: { "eacute" , "\351" },
118: { "iacute" , "\355" },
119: { "oacute" , "\363" },
120: { "uacute" , "\372" },
121: { "Aacute" , "\301" },
122: { "Eacute" , "\310" },
123: { "Iacute" , "\315" },
124: { "Oacute" , "\323" },
125: { "Uacute" , "\332" },
126: #endif
127: { 0, 0 } /* Terminate list */
128: };
1.2 timbl 129: #endif
1.1 timbl 130:
131:
1.2 timbl 132: /* Entity values -- for ISO Latin 1 local representation
133: **
134: ** This MUST match exactly the table referred to in the DTD!
135: */
136: static char * ISO_Latin1[] = {
137: "\306", /* capital AE diphthong (ligature) */
138: "\301", /* capital A, acute accent */
139: "\302", /* capital A, circumflex accent */
140: "\300", /* capital A, grave accent */
141: "\305", /* capital A, ring */
142: "\303", /* capital A, tilde */
143: "\304", /* capital A, dieresis or umlaut mark */
144: "\307", /* capital C, cedilla */
145: "\320", /* capital Eth, Icelandic */
146: "\311", /* capital E, acute accent */
147: "\312", /* capital E, circumflex accent */
148: "\310", /* capital E, grave accent */
149: "\313", /* capital E, dieresis or umlaut mark */
150: "\315", /* capital I, acute accent */
151: "\316", /* capital I, circumflex accent */
152: "\314", /* capital I, grave accent */
153: "\317", /* capital I, dieresis or umlaut mark */
154: "\321", /* capital N, tilde */
155: "\323", /* capital O, acute accent */
156: "\324", /* capital O, circumflex accent */
157: "\322", /* capital O, grave accent */
158: "\330", /* capital O, slash */
159: "\325", /* capital O, tilde */
160: "\326", /* capital O, dieresis or umlaut mark */
161: "\336", /* capital THORN, Icelandic */
162: "\332", /* capital U, acute accent */
163: "\333", /* capital U, circumflex accent */
164: "\331", /* capital U, grave accent */
165: "\334", /* capital U, dieresis or umlaut mark */
166: "\335", /* capital Y, acute accent */
167: "\341", /* small a, acute accent */
168: "\342", /* small a, circumflex accent */
169: "\346", /* small ae diphthong (ligature) */
170: "\340", /* small a, grave accent */
171: "\046", /* ampersand */
172: "\345", /* small a, ring */
173: "\343", /* small a, tilde */
174: "\344", /* small a, dieresis or umlaut mark */
175: "\347", /* small c, cedilla */
176: "\351", /* small e, acute accent */
177: "\352", /* small e, circumflex accent */
178: "\350", /* small e, grave accent */
179: "\360", /* small eth, Icelandic */
180: "\353", /* small e, dieresis or umlaut mark */
181: "\076", /* greater than */
182: "\355", /* small i, acute accent */
183: "\356", /* small i, circumflex accent */
184: "\354", /* small i, grave accent */
185: "\357", /* small i, dieresis or umlaut mark */
186: "\074", /* less than */
187: "\361", /* small n, tilde */
188: "\363", /* small o, acute accent */
189: "\364", /* small o, circumflex accent */
190: "\362", /* small o, grave accent */
191: "\370", /* small o, slash */
192: "\365", /* small o, tilde */
193: "\366", /* small o, dieresis or umlaut mark */
1.36 frystyk 194: "\042", /* double quote sign - June 94 */
1.2 timbl 195: "\337", /* small sharp s, German (sz ligature) */
196: "\376", /* small thorn, Icelandic */
197: "\372", /* small u, acute accent */
198: "\373", /* small u, circumflex accent */
199: "\371", /* small u, grave accent */
200: "\374", /* small u, dieresis or umlaut mark */
201: "\375", /* small y, acute accent */
202: "\377", /* small y, dieresis or umlaut mark */
1.1 timbl 203: };
204:
1.2 timbl 205:
206: /* Entity values -- for NeXT local representation
207: **
208: ** This MUST match exactly the table referred to in the DTD!
209: **
210: */
211: static char * NeXTCharacters[] = {
212: "\341", /* capital AE diphthong (ligature) */
213: "\202", /* capital A, acute accent */
214: "\203", /* capital A, circumflex accent */
215: "\201", /* capital A, grave accent */
216: "\206", /* capital A, ring */
217: "\204", /* capital A, tilde */
218: "\205", /* capital A, dieresis or umlaut mark */
219: "\207", /* capital C, cedilla */
220: "\220", /* capital Eth, Icelandic */
221: "\211", /* capital E, acute accent */
222: "\212", /* capital E, circumflex accent */
223: "\210", /* capital E, grave accent */
224: "\213", /* capital E, dieresis or umlaut mark */
225: "\215", /* capital I, acute accent */
226: "\216", /* capital I, circumflex accent these are */
227: "\214", /* capital I, grave accent ISO -100 hex */
228: "\217", /* capital I, dieresis or umlaut mark */
229: "\221", /* capital N, tilde */
230: "\223", /* capital O, acute accent */
231: "\224", /* capital O, circumflex accent */
232: "\222", /* capital O, grave accent */
233: "\351", /* capital O, slash 'cept this */
234: "\225", /* capital O, tilde */
235: "\226", /* capital O, dieresis or umlaut mark */
236: "\234", /* capital THORN, Icelandic */
237: "\230", /* capital U, acute accent */
238: "\231", /* capital U, circumflex accent */
239: "\227", /* capital U, grave accent */
240: "\232", /* capital U, dieresis or umlaut mark */
241: "\233", /* capital Y, acute accent */
242: "\326", /* small a, acute accent */
243: "\327", /* small a, circumflex accent */
244: "\361", /* small ae diphthong (ligature) */
245: "\325", /* small a, grave accent */
246: "\046", /* ampersand */
247: "\332", /* small a, ring */
248: "\330", /* small a, tilde */
249: "\331", /* small a, dieresis or umlaut mark */
250: "\333", /* small c, cedilla */
251: "\335", /* small e, acute accent */
252: "\336", /* small e, circumflex accent */
253: "\334", /* small e, grave accent */
254: "\346", /* small eth, Icelandic */
255: "\337", /* small e, dieresis or umlaut mark */
256: "\076", /* greater than */
257: "\342", /* small i, acute accent */
258: "\344", /* small i, circumflex accent */
259: "\340", /* small i, grave accent */
260: "\345", /* small i, dieresis or umlaut mark */
261: "\074", /* less than */
262: "\347", /* small n, tilde */
263: "\355", /* small o, acute accent */
264: "\356", /* small o, circumflex accent */
265: "\354", /* small o, grave accent */
266: "\371", /* small o, slash */
267: "\357", /* small o, tilde */
268: "\360", /* small o, dieresis or umlaut mark */
1.36 frystyk 269: "\042", /* double quote sign - June 94 */
1.2 timbl 270: "\373", /* small sharp s, German (sz ligature) */
271: "\374", /* small thorn, Icelandic */
272: "\363", /* small u, acute accent */
273: "\364", /* small u, circumflex accent */
274: "\362", /* small u, grave accent */
275: "\366", /* small u, dieresis or umlaut mark */
276: "\367", /* small y, acute accent */
277: "\375", /* small y, dieresis or umlaut mark */
1.1 timbl 278: };
279:
1.2 timbl 280: /* Entity values -- for IBM/PC Code Page 850 (International)
281: **
282: ** This MUST match exactly the table referred to in the DTD!
283: **
284: */
285: /* @@@@@@@@@@@@@@@@@ TBD */
286:
287:
288:
289: /* Set character set
290: ** ----------------
291: */
292:
293: PRIVATE char** p_entity_values = ISO_Latin1; /* Pointer to translation */
1.1 timbl 294:
1.2 timbl 295: PUBLIC void HTMLUseCharacterSet ARGS1(HTMLCharacterSet, i)
296: {
297: p_entity_values = (i == HTML_NEXT_CHARS) ? NeXTCharacters
298: : ISO_Latin1;
299: }
1.1 timbl 300:
301:
302: /* Flattening the style structure
303: ** ------------------------------
304: **
305: On the NeXT, and on any read-only browser, it is simpler for the text to have
306: a sequence of styles, rather than a nested tree of styles. In this
307: case we have to flatten the structure as it arrives from SGML tags into
308: a sequence of styles.
309: */
310:
311: /* If style really needs to be set, call this
312: */
1.4 timbl 313: PRIVATE void actually_set_style ARGS1(HTStructured *, me)
1.1 timbl 314: {
1.4 timbl 315: if (!me->text) { /* First time through */
316: me->text = HText_new2(me->node_anchor, me->target);
317: HText_beginAppend(me->text);
318: HText_setStyle(me->text, me->new_style);
319: me->in_word = NO;
1.1 timbl 320: } else {
1.4 timbl 321: HText_setStyle(me->text, me->new_style);
1.1 timbl 322: }
1.4 timbl 323: me->old_style = me->new_style;
324: me->style_change = NO;
1.1 timbl 325: }
326:
327: /* If you THINK you need to change style, call this
328: */
329:
1.11 timbl 330: PRIVATE void change_paragraph_style ARGS2(HTStructured *, me, HTStyle *,style)
1.1 timbl 331: {
1.4 timbl 332: if (me->new_style!=style) {
333: me->style_change = YES;
334: me->new_style = style;
1.1 timbl 335: }
1.11 timbl 336: me->in_word = NO;
1.1 timbl 337: }
338:
1.2 timbl 339: /*_________________________________________________________________________
340: **
341: ** A C T I O N R O U T I N E S
342: */
343:
344: /* Character handling
345: ** ------------------
1.1 timbl 346: */
1.42 frystyk 347: PRIVATE int HTML_put_character ARGS2(HTStructured *, me, char, c)
1.1 timbl 348: {
1.2 timbl 349:
1.4 timbl 350: switch (me->sp[0].tag_number) {
1.2 timbl 351: case HTML_COMMENT:
352: break; /* Do Nothing */
353:
354: case HTML_TITLE:
1.4 timbl 355: HTChunkPutc(&me->title, c);
1.2 timbl 356: break;
357:
358:
359: case HTML_LISTING: /* Litteral text */
360: case HTML_XMP:
361: case HTML_PLAINTEXT:
362: case HTML_PRE:
363: /* We guarrantee that the style is up-to-date in begin_litteral
364: */
1.4 timbl 365: HText_appendCharacter(me->text, c);
1.2 timbl 366: break;
367:
368: default: /* Free format text */
1.4 timbl 369: if (me->style_change) {
1.42 frystyk 370: if ((c=='\n') || (c==' ')) return HT_OK; /* Ignore it */
1.2 timbl 371: UPDATE_STYLE;
372: }
373: if (c=='\n') {
1.4 timbl 374: if (me->in_word) {
375: HText_appendCharacter(me->text, ' ');
376: me->in_word = NO;
1.2 timbl 377: }
378: } else {
1.4 timbl 379: HText_appendCharacter(me->text, c);
380: me->in_word = YES;
1.2 timbl 381: }
382: } /* end switch */
1.42 frystyk 383: return HT_OK;
1.1 timbl 384: }
385:
1.2 timbl 386:
387:
388: /* String handling
389: ** ---------------
390: **
391: ** This is written separately from put_character becuase the loop can
1.11 timbl 392: ** in some cases be promoted to a higher function call level for speed.
1.2 timbl 393: */
1.42 frystyk 394: PRIVATE int HTML_put_string ARGS2(HTStructured *, me, CONST char*, s)
1.1 timbl 395: {
1.2 timbl 396:
1.4 timbl 397: switch (me->sp[0].tag_number) {
1.2 timbl 398: case HTML_COMMENT:
399: break; /* Do Nothing */
400:
401: case HTML_TITLE:
1.4 timbl 402: HTChunkPuts(&me->title, s);
1.2 timbl 403: break;
404:
405:
406: case HTML_LISTING: /* Litteral text */
407: case HTML_XMP:
408: case HTML_PLAINTEXT:
409: case HTML_PRE:
410:
411: /* We guarrantee that the style is up-to-date in begin_litteral
412: */
1.4 timbl 413: HText_appendText(me->text, s);
1.2 timbl 414: break;
415:
416: default: /* Free format text */
417: {
418: CONST char *p = s;
1.4 timbl 419: if (me->style_change) {
1.2 timbl 420: for (; *p && ((*p=='\n') || (*p==' ')); p++) ; /* Ignore leaders */
1.42 frystyk 421: if (!*p) return HT_OK;
1.2 timbl 422: UPDATE_STYLE;
423: }
424: for(; *p; p++) {
1.4 timbl 425: if (me->style_change) {
1.2 timbl 426: if ((*p=='\n') || (*p==' ')) continue; /* Ignore it */
427: UPDATE_STYLE;
428: }
429: if (*p=='\n') {
1.4 timbl 430: if (me->in_word) {
431: HText_appendCharacter(me->text, ' ');
432: me->in_word = NO;
1.2 timbl 433: }
434: } else {
1.4 timbl 435: HText_appendCharacter(me->text, *p);
436: me->in_word = YES;
1.2 timbl 437: }
438: } /* for */
439: }
440: } /* end switch */
1.42 frystyk 441: return HT_OK;
1.1 timbl 442: }
443:
444:
1.2 timbl 445: /* Buffer write
1.3 timbl 446: ** ------------
1.1 timbl 447: */
1.42 frystyk 448: PRIVATE int HTML_write ARGS3(HTStructured *, me, CONST char*, s, int, l)
1.1 timbl 449: {
1.38 frystyk 450: while (l-- > 0)
451: HTML_put_character(me, *s++);
1.42 frystyk 452: return HT_OK;
1.1 timbl 453: }
1.2 timbl 454:
455:
456: /* Start Element
457: ** -------------
458: */
459: PRIVATE void HTML_start_element ARGS4(
1.4 timbl 460: HTStructured *, me,
1.16 timbl 461: int, element_number,
1.3 timbl 462: CONST BOOL*, present,
1.16 timbl 463: CONST char **, value)
1.2 timbl 464: {
465: switch (element_number) {
466: case HTML_A:
467: {
1.8 timbl 468: HTChildAnchor * source;
1.9 timbl 469: char * href = NULL;
1.42 frystyk 470: if (present[HTML_A_HREF])
1.9 timbl 471: StrAllocCopy(href, value[HTML_A_HREF]);
1.8 timbl 472: source = HTAnchor_findChildAndLink(
1.4 timbl 473: me->node_anchor, /* parent */
1.2 timbl 474: present[HTML_A_NAME] ? value[HTML_A_NAME] : 0, /* Tag */
1.9 timbl 475: present[HTML_A_HREF] ? href : 0, /* Addresss */
1.16 timbl 476: present[HTML_A_REL] && value[HTML_A_REL] ?
477: (HTLinkType*)HTAtom_for(value[HTML_A_REL])
1.2 timbl 478: : 0);
479:
480: if (present[HTML_A_TITLE] && value[HTML_A_TITLE]) {
481: HTParentAnchor * dest =
482: HTAnchor_parent(
483: HTAnchor_followMainLink((HTAnchor*)source)
484: );
485: if (!HTAnchor_title(dest))
486: HTAnchor_setTitle(dest, value[HTML_A_TITLE]);
487: }
488: UPDATE_STYLE;
1.4 timbl 489: HText_beginAnchor(me->text, source);
1.42 frystyk 490: FREE(href); /* Leak fix Henrik 17/02-94 */
1.2 timbl 491: }
492: break;
493:
494: case HTML_TITLE:
1.4 timbl 495: HTChunkClear(&me->title);
1.2 timbl 496: break;
497:
498: case HTML_NEXTID:
499: /* if (present[NEXTID_N] && value[NEXTID_N])
1.4 timbl 500: HText_setNextId(me->text, atoi(value[NEXTID_N])); */
1.2 timbl 501: break;
502:
503: case HTML_ISINDEX:
1.4 timbl 504: HTAnchor_setIndex(me->node_anchor);
1.2 timbl 505: break;
506:
1.15 timbl 507: case HTML_BR:
508: UPDATE_STYLE;
509: HText_appendCharacter(me->text, '\n');
510: me->in_word = NO;
511: break;
512:
513: case HTML_HR:
514: UPDATE_STYLE;
515: HText_appendCharacter(me->text, '\n');
1.16 timbl 516: HText_appendText(me->text, "___________________________________");
1.15 timbl 517: HText_appendCharacter(me->text, '\n');
518: me->in_word = NO;
519: break;
520:
1.2 timbl 521: case HTML_P:
522: UPDATE_STYLE;
1.4 timbl 523: HText_appendParagraph(me->text);
524: me->in_word = NO;
1.2 timbl 525: break;
526:
527: case HTML_DL:
1.11 timbl 528: change_paragraph_style(me, present && present[DL_COMPACT]
1.16 timbl 529: ? styles[HTML_DL]
1.2 timbl 530: : styles[HTML_DL]);
531: break;
532:
533: case HTML_DT:
1.4 timbl 534: if (!me->style_change) {
535: HText_appendParagraph(me->text);
536: me->in_word = NO;
1.2 timbl 537: }
538: break;
539:
540: case HTML_DD:
541: UPDATE_STYLE;
1.4 timbl 542: HTML_put_character(me, '\t'); /* Just tab out one stop */
543: me->in_word = NO;
544: break;
1.2 timbl 545:
546: case HTML_UL:
547: case HTML_OL:
548: case HTML_MENU:
549: case HTML_DIR:
1.11 timbl 550: change_paragraph_style(me, styles[element_number]);
1.2 timbl 551: break;
552:
553: case HTML_LI:
554: UPDATE_STYLE;
1.7 timbl 555: if (me->sp[0].tag_number != HTML_DIR)
1.4 timbl 556: HText_appendParagraph(me->text);
1.2 timbl 557: else
1.4 timbl 558: HText_appendCharacter(me->text, '\t'); /* Tab @@ nl for UL? */
559: me->in_word = NO;
1.2 timbl 560: break;
561:
562: case HTML_LISTING: /* Litteral text */
563: case HTML_XMP:
564: case HTML_PLAINTEXT:
565: case HTML_PRE:
1.11 timbl 566: change_paragraph_style(me, styles[element_number]);
1.2 timbl 567: UPDATE_STYLE;
1.4 timbl 568: if (me->comment_end)
569: HText_appendText(me->text, me->comment_end);
1.2 timbl 570: break;
1.11 timbl 571:
1.23 frystyk 572: case HTML_IMG: /* Images */
573: {
574: HTChildAnchor *source;
575: char *src = NULL;
576: if (present[HTML_IMG_SRC]) {
577: StrAllocCopy(src, value[HTML_IMG_SRC]);
1.36 frystyk 578: #ifdef OLD_CODE
1.23 frystyk 579: HTSimplify(src);
1.36 frystyk 580: #endif
1.23 frystyk 581: }
582: source = HTAnchor_findChildAndLink(
583: me->node_anchor, /* parent */
584: 0, /* Tag */
585: src ? src : 0, /* Addresss */
586: 0);
587: UPDATE_STYLE;
588: HText_appendImage(me->text, source,
1.24 frystyk 589: present[HTML_IMG_ALT] ? value[HTML_IMG_ALT] : NULL,
590: present[HTML_IMG_ALIGN] ? value[HTML_IMG_ALIGN] : NULL,
591: present[HTML_IMG_ISMAP] ? YES : NO);
1.23 frystyk 592: free(src);
1.24 frystyk 593: }
594: break;
595:
596: case HTML_HTML: /* Ignore these altogether */
597: case HTML_HEAD:
598: case HTML_BODY:
599:
1.10 timbl 600: case HTML_TT: /* Physical character highlighting */
601: case HTML_B: /* Currently ignored */
602: case HTML_I:
603: case HTML_U:
604:
605: case HTML_EM: /* Logical character highlighting */
606: case HTML_STRONG: /* Currently ignored */
607: case HTML_CODE:
608: case HTML_SAMP:
609: case HTML_KBD:
610: case HTML_VAR:
611: case HTML_DFN:
612: case HTML_CITE:
613: break;
614:
1.11 timbl 615: case HTML_H1: /* paragraph styles */
616: case HTML_H2:
617: case HTML_H3:
618: case HTML_H4:
619: case HTML_H5:
620: case HTML_H6:
621: case HTML_H7:
622: case HTML_ADDRESS:
623: case HTML_BLOCKQUOTE:
624: change_paragraph_style(me, styles[element_number]); /* May be postponed */
1.2 timbl 625: break;
626:
627: } /* end switch */
628:
1.16 timbl 629: if (me->dtd->tags[element_number].contents!= SGML_EMPTY) {
1.13 timbl 630: if (me->sp == me->stack) {
1.44 ! frystyk 631: if (SGML_TRACE)
! 632: fprintf(TDEST, "HTML........ Maximum nesting of %d exceded!\n",
! 633: MAX_NESTING);
! 634: me->overflow++;
1.12 timbl 635: return;
636: }
1.4 timbl 637: --(me->sp);
638: me->sp[0].style = me->new_style; /* Stack new style */
639: me->sp[0].tag_number = element_number;
1.10 timbl 640: }
1.1 timbl 641: }
1.10 timbl 642:
1.2 timbl 643:
1.1 timbl 644: /* End Element
1.2 timbl 645: ** -----------
1.1 timbl 646: **
1.2 timbl 647: */
648: /* When we end an element, the style must be returned to that
1.1 timbl 649: ** in effect before that element. Note that anchors (etc?)
650: ** don't have an associated style, so that we must scan down the
651: ** stack for an element with a defined style. (In fact, the styles
652: ** should be linked to the whole stack not just the top one.)
653: ** TBL 921119
1.6 timbl 654: **
655: ** We don't turn on "CAREFUL" check because the parser produces
656: ** (internal code errors apart) good nesting. The parser checks
657: ** incoming code errors, not this module.
1.1 timbl 658: */
1.4 timbl 659: PRIVATE void HTML_end_element ARGS2(HTStructured *, me, int , element_number)
1.1 timbl 660: {
1.2 timbl 661: #ifdef CAREFUL /* parser assumed to produce good nesting */
1.4 timbl 662: if (element_number != me->sp[0].tag_number) {
1.41 frystyk 663: fprintf(TDEST, "HTMLText: end of element %s when expecting end of %s\n",
1.16 timbl 664: me->dtd->tags[element_number].name,
665: me->dtd->tags[me->sp->tag_number].name);
1.6 timbl 666: /* panic */
1.1 timbl 667: }
1.2 timbl 668: #endif
1.44 ! frystyk 669:
! 670: /* HFN, If overflow of nestings, we need to get back to reality */
! 671: if (me->overflow > 0) {
! 672: me->overflow--;
! 673: return;
! 674: }
! 675:
1.4 timbl 676: me->sp++; /* Pop state off stack */
1.44 ! frystyk 677:
1.2 timbl 678: switch(element_number) {
679:
680: case HTML_A:
681: UPDATE_STYLE;
1.4 timbl 682: HText_endAnchor(me->text);
1.2 timbl 683: break;
684:
685: case HTML_TITLE:
1.4 timbl 686: HTChunkTerminate(&me->title);
687: HTAnchor_setTitle(me->node_anchor, me->title.data);
1.2 timbl 688: break;
689:
690: case HTML_LISTING: /* Litteral text */
691: case HTML_XMP:
692: case HTML_PLAINTEXT:
693: case HTML_PRE:
1.4 timbl 694: if (me->comment_start)
695: HText_appendText(me->text, me->comment_start);
1.2 timbl 696: /* Fall through */
697:
698: default:
1.44 ! frystyk 699:
! 700: /* Often won't really change */
! 701: change_paragraph_style(me, me->sp->style);
1.2 timbl 702: break;
703:
704: } /* switch */
1.1 timbl 705: }
706:
1.2 timbl 707:
708: /* Expanding entities
709: ** ------------------
710: */
711: /* (In fact, they all shrink!)
1.1 timbl 712: */
1.2 timbl 713:
1.4 timbl 714: PRIVATE void HTML_put_entity ARGS2(HTStructured *, me, int, entity_number)
1.1 timbl 715: {
1.4 timbl 716: HTML_put_string(me, ISO_Latin1[entity_number]); /* @@ Other representations */
1.1 timbl 717: }
1.2 timbl 718:
1.42 frystyk 719: /* Flush an HTML object
720: ** --------------------
721: */
722: PUBLIC int HTML_flush ARGS1(HTStructured *, me)
723: {
724: UPDATE_STYLE; /* Creates empty document here! */
725: if (me->comment_end)
726: HTML_put_string(me,me->comment_end);
727: HText_endAppend(me->text);
728: return (*me->targetClass.flush)(me->target);
729: }
1.2 timbl 730:
731: /* Free an HTML object
732: ** -------------------
733: **
1.4 timbl 734: ** If the document is empty, the text object will not yet exist.
735: So we could in fact abandon creating the document and return
736: an error code. In fact an empty document is an important type
737: of document, so we don't.
738: **
739: ** If non-interactive, everything is freed off. No: crashes -listrefs
1.2 timbl 740: ** Otherwise, the interactive object is left.
741: */
1.37 frystyk 742: PUBLIC int HTML_free ARGS1(HTStructured *, me)
1.1 timbl 743: {
1.4 timbl 744: UPDATE_STYLE; /* Creates empty document here! */
745: if (me->comment_end)
746: HTML_put_string(me,me->comment_end);
747: HText_endAppend(me->text);
748:
749: if (me->target) {
1.35 duns 750: (*me->targetClass._free)(me->target);
1.2 timbl 751: }
1.19 frystyk 752: HTChunkClear(&me->title); /* Henrik 18/02-94 */
1.4 timbl 753: free(me);
1.42 frystyk 754: return HT_OK;
1.1 timbl 755: }
756:
757:
1.37 frystyk 758: PRIVATE int HTML_abort ARGS2(HTStructured *, me, HTError, e)
1.1 timbl 759:
1.14 timbl 760: {
761: if (me->target) {
762: (*me->targetClass.abort)(me->target, e);
763: }
1.19 frystyk 764: HTChunkClear(&me->title); /* Henrik 18/02-94 */
1.14 timbl 765: free(me);
1.42 frystyk 766: return HT_ERROR;
1.1 timbl 767: }
768:
1.2 timbl 769:
770: /* Get Styles from style sheet
771: ** ---------------------------
772: */
773: PRIVATE void get_styles NOARGS
1.1 timbl 774: {
1.2 timbl 775: got_styles = YES;
776:
777: default_style = HTStyleNamed(styleSheet, "Normal");
1.1 timbl 778:
1.2 timbl 779: styles[HTML_H1] = HTStyleNamed(styleSheet, "Heading1");
780: styles[HTML_H2] = HTStyleNamed(styleSheet, "Heading2");
781: styles[HTML_H3] = HTStyleNamed(styleSheet, "Heading3");
782: styles[HTML_H4] = HTStyleNamed(styleSheet, "Heading4");
783: styles[HTML_H5] = HTStyleNamed(styleSheet, "Heading5");
784: styles[HTML_H6] = HTStyleNamed(styleSheet, "Heading6");
785: styles[HTML_H7] = HTStyleNamed(styleSheet, "Heading7");
786:
787: styles[HTML_DL] = HTStyleNamed(styleSheet, "Glossary");
788: styles[HTML_UL] =
789: styles[HTML_OL] = HTStyleNamed(styleSheet, "List");
790: styles[HTML_MENU] = HTStyleNamed(styleSheet, "Menu");
791: styles[HTML_DIR] = HTStyleNamed(styleSheet, "Dir");
1.16 timbl 792: /* styles[HTML_DLC] = HTStyleNamed(styleSheet, "GlossaryCompact"); */
1.2 timbl 793: styles[HTML_ADDRESS]= HTStyleNamed(styleSheet, "Address");
794: styles[HTML_BLOCKQUOTE]= HTStyleNamed(styleSheet, "BlockQuote");
795: styles[HTML_PLAINTEXT] =
796: styles[HTML_XMP] = HTStyleNamed(styleSheet, "Example");
797: styles[HTML_PRE] = HTStyleNamed(styleSheet, "Preformatted");
798: styles[HTML_LISTING] = HTStyleNamed(styleSheet, "Listing");
799: }
800: /* P U B L I C
801: */
802:
803: /* Structured Object Class
804: ** -----------------------
805: */
806: PUBLIC CONST HTStructuredClass HTMLPresentation = /* As opposed to print etc */
807: {
808: "text/html",
1.42 frystyk 809: HTML_flush,
1.2 timbl 810: HTML_free,
1.14 timbl 811: HTML_abort,
1.2 timbl 812: HTML_put_character, HTML_put_string, HTML_write,
813: HTML_start_element, HTML_end_element,
814: HTML_put_entity
815: };
1.1 timbl 816:
1.4 timbl 817:
1.2 timbl 818: /* New Structured Text object
819: ** --------------------------
820: **
1.16 timbl 821: ** The structured stream can generate either presentation,
1.4 timbl 822: ** or plain text, or HTML.
1.1 timbl 823: */
1.16 timbl 824: PUBLIC HTStructured* HTML_new ARGS5(
825: HTRequest *, request,
826: void *, param,
827: HTFormat, input_format,
828: HTFormat, output_format,
829: HTStream *, output_stream)
1.1 timbl 830: {
831:
1.4 timbl 832: HTStructured * me;
833:
1.16 timbl 834: if (output_format != WWW_PLAINTEXT
835: && output_format != WWW_PRESENT
836: && output_format != HTAtom_for("text/x-c")) {
1.37 frystyk 837: HTStream * intermediate = HTStreamStack(WWW_HTML, output_format,
838: output_stream, request, NO);
1.6 timbl 839: if (intermediate) return HTMLGenerator(intermediate);
1.44 ! frystyk 840: if (SGML_TRACE)
! 841: fprintf(TDEST, "HTML........ Can't parse HTML to %s\n",
! 842: HTAtom_name(output_format));
1.4 timbl 843: exit (-99);
844: }
845:
1.44 ! frystyk 846: if ((me = (HTStructured*) calloc(1, sizeof(*me))) == NULL)
! 847: outofmem(__FILE__, "HTML_new");
1.1 timbl 848:
849: if (!got_styles) get_styles();
850:
1.4 timbl 851: me->isa = &HTMLPresentation;
1.16 timbl 852: me->dtd = &DTD;
853: me->node_anchor = request->anchor;
1.4 timbl 854: me->title.size = 0;
855: me->title.growby = 128;
856: me->title.allocated = 0;
857: me->title.data = 0;
858: me->text = 0;
859: me->style_change = YES; /* Force check leading to text creation */
860: me->new_style = default_style;
861: me->old_style = 0;
862: me->sp = me->stack + MAX_NESTING - 1;
863: me->sp->tag_number = -1; /* INVALID */
864: me->sp->style = default_style; /* INVALID */
1.1 timbl 865:
1.4 timbl 866: me->comment_start = NULL;
867: me->comment_end = NULL;
1.16 timbl 868: me->target = output_stream;
869: if (output_stream) me->targetClass = *output_stream->isa; /* Copy pointers */
1.1 timbl 870:
1.4 timbl 871: return (HTStructured*) me;
1.1 timbl 872: }
873:
874:
1.2 timbl 875: /* HTConverter for HTML to plain text
876: ** ----------------------------------
1.1 timbl 877: **
1.2 timbl 878: ** This will convert from HTML to presentation or plain text.
1.1 timbl 879: */
1.16 timbl 880: PUBLIC HTStream* HTMLToPlain ARGS5(
881: HTRequest *, request,
882: void *, param,
883: HTFormat, input_format,
884: HTFormat, output_format,
885: HTStream *, output_stream)
1.1 timbl 886: {
1.16 timbl 887: return SGML_new(&DTD, HTML_new(
888: request, NULL, input_format, output_format, output_stream));
1.1 timbl 889: }
890:
891:
1.2 timbl 892: /* HTConverter for HTML to C code
893: ** ------------------------------
894: **
1.36 frystyk 895: ** C code is like plain text but all non-preformatted code
1.2 timbl 896: ** is commented out.
897: ** This will convert from HTML to presentation or plain text.
898: */
1.16 timbl 899: PUBLIC HTStream* HTMLToC ARGS5(
900: HTRequest *, request,
901: void *, param,
902: HTFormat, input_format,
903: HTFormat, output_format,
904: HTStream *, output_stream)
1.1 timbl 905: {
1.4 timbl 906:
907: HTStructured * html;
908:
1.36 frystyk 909: (*output_stream->isa->put_string)(output_stream, "/* "); /* Before title */
1.16 timbl 910: html = HTML_new(request, NULL, input_format, output_format, output_stream);
1.2 timbl 911: html->comment_start = "/* ";
1.16 timbl 912: html->dtd = &DTD;
1.2 timbl 913: html->comment_end = " */\n"; /* Must start in col 1 for cpp */
1.16 timbl 914: return SGML_new(&DTD, html);
1.1 timbl 915: }
916:
917:
1.2 timbl 918: /* Presenter for HTML
919: ** ------------------
920: **
921: ** This will convert from HTML to presentation or plain text.
922: **
923: ** Override this if you have a windows version
1.1 timbl 924: */
1.2 timbl 925: #ifndef GUI
1.16 timbl 926: PUBLIC HTStream* HTMLPresent ARGS5(
927: HTRequest *, request,
928: void *, param,
929: HTFormat, input_format,
930: HTFormat, output_format,
931: HTStream *, output_stream)
1.1 timbl 932: {
1.16 timbl 933: return SGML_new(&DTD, HTML_new(
934: request, NULL, input_format, output_format, output_stream));
1.1 timbl 935: }
1.2 timbl 936: #endif
1.1 timbl 937:
938:
1.2 timbl 939: /* Record error message as a hypertext object
940: ** ------------------------------------------
941: **
942: ** The error message should be marked as an error so that
943: ** it can be reloaded later.
944: ** This implementation just throws up an error message
945: ** and leaves the document unloaded.
1.9 timbl 946: ** A smarter implementation would load an error document,
947: ** marking at such so that it is retried on reload.
1.1 timbl 948: **
1.2 timbl 949: ** On entry,
950: ** sink is a stream to the output device if any
951: ** number is the HTTP error number
952: ** message is the human readable message.
1.9 timbl 953: **
954: ** On exit,
955: ** returns a negative number to indicate lack of success in the load.
1.1 timbl 956: */
1.2 timbl 957:
958: PUBLIC int HTLoadError ARGS3(
1.17 luotonen 959: HTRequest *, req,
1.2 timbl 960: int, number,
961: CONST char *, message)
962: {
1.20 frystyk 963: char *err = "Oh I screwed up!"; /* Dummy pointer not used (I hope) */
1.2 timbl 964: HTAlert(message); /* @@@@@@@@@@@@@@@@@@@ */
1.20 frystyk 965: /* Clean up! Henrik 04/03-94 */
966: if (req && req->output_stream)
967: (*req->output_stream->isa->abort)(req->output_stream, err);
1.33 frystyk 968: #if OLD_CODE
1.25 luotonen 969: HTClearErrors(req);
1.33 frystyk 970: #endif
1.2 timbl 971: return -number;
972: }
1.29 frystyk 973:
Webmaster