Annotation of libwww/Library/src/HTML.c, revision 1.17
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: */
1.16 timbl 12:
1.1 timbl 13: #include "HTML.h"
14:
1.16 timbl 15: /* #define CAREFUL Check nesting here not really necessary */
1.2 timbl 16:
1.1 timbl 17: #include <ctype.h>
18: #include <stdio.h>
19:
20: #include "HTAtom.h"
21: #include "HTChunk.h"
22: #include "HText.h"
23: #include "HTStyle.h"
24:
1.3 timbl 25: #include "HTAlert.h"
1.4 timbl 26: #include "HTMLGen.h"
1.8 timbl 27: #include "HTParse.h"
1.1 timbl 28:
29: extern HTStyleSheet * styleSheet; /* Application-wide */
30:
31: /* Module-wide style cache
32: */
33: PRIVATE int got_styles = 0;
1.16 timbl 34: PRIVATE HTStyle *styles[HTMLP_ELEMENTS];
1.2 timbl 35: PRIVATE HTStyle *default_style;
1.1 timbl 36:
37:
38: /* HTML Object
39: ** -----------
40: */
1.2 timbl 41: #define MAX_NESTING 20 /* Should be checked by parser */
42:
43: typedef struct _stack_element {
44: HTStyle * style;
45: int tag_number;
46: } stack_element;
47:
48: struct _HTStructured {
49: CONST HTStructuredClass * isa;
50: HTParentAnchor * node_anchor;
51: HText * text;
52:
53: HTStream* target; /* Output stream */
54: HTStreamClass targetClass; /* Output routines */
55:
56: HTChunk title; /* Grow by 128 */
57:
58: char * comment_start; /* for literate programming */
59: char * comment_end;
1.16 timbl 60:
61: CONST SGML_dtd* dtd;
62:
1.2 timbl 63: HTTag * current_tag;
64: BOOL style_change;
65: HTStyle * new_style;
66: HTStyle * old_style;
67: BOOL in_word; /* Have just had a non-white char */
68: stack_element stack[MAX_NESTING];
69: stack_element *sp; /* Style stack pointer */
1.1 timbl 70: };
71:
1.2 timbl 72: struct _HTStream {
73: CONST HTStreamClass * isa;
74: /* .... */
75: };
1.1 timbl 76:
77: /* Forward declarations of routines
78: */
79: PRIVATE void get_styles NOPARAMS;
80:
81:
1.4 timbl 82: PRIVATE void actually_set_style PARAMS((HTStructured * me));
1.11 timbl 83: PRIVATE void change_paragraph_style PARAMS((HTStructured * me, HTStyle * style));
1.1 timbl 84:
85: /* Style buffering avoids dummy paragraph begin/ends.
86: */
1.4 timbl 87: #define UPDATE_STYLE if (me->style_change) { actually_set_style(me); }
1.1 timbl 88:
89:
1.2 timbl 90: #ifdef OLD_CODE
1.1 timbl 91: /* The following accented characters are from peter Flynn, curia project */
92:
93: /* these ifdefs don't solve the problem of a simple terminal emulator
94: ** with a different character set to the client machine. But nothing does,
95: ** except looking at the TERM setting */
96:
1.2 timbl 97:
1.1 timbl 98: { "ocus" , "&" }, /* for CURIA */
99: #ifdef IBMPC
100: { "aacute" , "240円" }, /* For PC display */
101: { "eacute" , "202円" },
102: { "iacute" , "241円" },
103: { "oacute" , "242円" },
104: { "uacute" , "243円" },
105: { "Aacute" , "101円" },
106: { "Eacute" , "220円" },
107: { "Iacute" , "111円" },
108: { "Oacute" , "117円" },
109: { "Uacute" , "125円" },
110: #else
111: { "aacute" , "341円" }, /* Works for openwindows -- Peter Flynn */
112: { "eacute" , "351円" },
113: { "iacute" , "355円" },
114: { "oacute" , "363円" },
115: { "uacute" , "372円" },
116: { "Aacute" , "301円" },
117: { "Eacute" , "310円" },
118: { "Iacute" , "315円" },
119: { "Oacute" , "323円" },
120: { "Uacute" , "332円" },
121: #endif
122: { 0, 0 } /* Terminate list */
123: };
1.2 timbl 124: #endif
1.1 timbl 125:
126:
1.2 timbl 127: /* Entity values -- for ISO Latin 1 local representation
128: **
129: ** This MUST match exactly the table referred to in the DTD!
130: */
131: static char * ISO_Latin1[] = {
132: "306円", /* capital AE diphthong (ligature) */
133: "301円", /* capital A, acute accent */
134: "302円", /* capital A, circumflex accent */
135: "300円", /* capital A, grave accent */
136: "305円", /* capital A, ring */
137: "303円", /* capital A, tilde */
138: "304円", /* capital A, dieresis or umlaut mark */
139: "307円", /* capital C, cedilla */
140: "320円", /* capital Eth, Icelandic */
141: "311円", /* capital E, acute accent */
142: "312円", /* capital E, circumflex accent */
143: "310円", /* capital E, grave accent */
144: "313円", /* capital E, dieresis or umlaut mark */
145: "315円", /* capital I, acute accent */
146: "316円", /* capital I, circumflex accent */
147: "314円", /* capital I, grave accent */
148: "317円", /* capital I, dieresis or umlaut mark */
149: "321円", /* capital N, tilde */
150: "323円", /* capital O, acute accent */
151: "324円", /* capital O, circumflex accent */
152: "322円", /* capital O, grave accent */
153: "330円", /* capital O, slash */
154: "325円", /* capital O, tilde */
155: "326円", /* capital O, dieresis or umlaut mark */
156: "336円", /* capital THORN, Icelandic */
157: "332円", /* capital U, acute accent */
158: "333円", /* capital U, circumflex accent */
159: "331円", /* capital U, grave accent */
160: "334円", /* capital U, dieresis or umlaut mark */
161: "335円", /* capital Y, acute accent */
162: "341円", /* small a, acute accent */
163: "342円", /* small a, circumflex accent */
164: "346円", /* small ae diphthong (ligature) */
165: "340円", /* small a, grave accent */
166: "046円", /* ampersand */
167: "345円", /* small a, ring */
168: "343円", /* small a, tilde */
169: "344円", /* small a, dieresis or umlaut mark */
170: "347円", /* small c, cedilla */
171: "351円", /* small e, acute accent */
172: "352円", /* small e, circumflex accent */
173: "350円", /* small e, grave accent */
174: "360円", /* small eth, Icelandic */
175: "353円", /* small e, dieresis or umlaut mark */
176: "076円", /* greater than */
177: "355円", /* small i, acute accent */
178: "356円", /* small i, circumflex accent */
179: "354円", /* small i, grave accent */
180: "357円", /* small i, dieresis or umlaut mark */
181: "074円", /* less than */
182: "361円", /* small n, tilde */
183: "363円", /* small o, acute accent */
184: "364円", /* small o, circumflex accent */
185: "362円", /* small o, grave accent */
186: "370円", /* small o, slash */
187: "365円", /* small o, tilde */
188: "366円", /* small o, dieresis or umlaut mark */
189: "337円", /* small sharp s, German (sz ligature) */
190: "376円", /* small thorn, Icelandic */
191: "372円", /* small u, acute accent */
192: "373円", /* small u, circumflex accent */
193: "371円", /* small u, grave accent */
194: "374円", /* small u, dieresis or umlaut mark */
195: "375円", /* small y, acute accent */
196: "377円", /* small y, dieresis or umlaut mark */
1.1 timbl 197: };
198:
1.2 timbl 199:
200: /* Entity values -- for NeXT local representation
201: **
202: ** This MUST match exactly the table referred to in the DTD!
203: **
204: */
205: static char * NeXTCharacters[] = {
206: "341円", /* capital AE diphthong (ligature) */
207: "202円", /* capital A, acute accent */
208: "203円", /* capital A, circumflex accent */
209: "201円", /* capital A, grave accent */
210: "206円", /* capital A, ring */
211: "204円", /* capital A, tilde */
212: "205円", /* capital A, dieresis or umlaut mark */
213: "207円", /* capital C, cedilla */
214: "220円", /* capital Eth, Icelandic */
215: "211円", /* capital E, acute accent */
216: "212円", /* capital E, circumflex accent */
217: "210円", /* capital E, grave accent */
218: "213円", /* capital E, dieresis or umlaut mark */
219: "215円", /* capital I, acute accent */
220: "216円", /* capital I, circumflex accent these are */
221: "214円", /* capital I, grave accent ISO -100 hex */
222: "217円", /* capital I, dieresis or umlaut mark */
223: "221円", /* capital N, tilde */
224: "223円", /* capital O, acute accent */
225: "224円", /* capital O, circumflex accent */
226: "222円", /* capital O, grave accent */
227: "351円", /* capital O, slash 'cept this */
228: "225円", /* capital O, tilde */
229: "226円", /* capital O, dieresis or umlaut mark */
230: "234円", /* capital THORN, Icelandic */
231: "230円", /* capital U, acute accent */
232: "231円", /* capital U, circumflex accent */
233: "227円", /* capital U, grave accent */
234: "232円", /* capital U, dieresis or umlaut mark */
235: "233円", /* capital Y, acute accent */
236: "326円", /* small a, acute accent */
237: "327円", /* small a, circumflex accent */
238: "361円", /* small ae diphthong (ligature) */
239: "325円", /* small a, grave accent */
240: "046円", /* ampersand */
241: "332円", /* small a, ring */
242: "330円", /* small a, tilde */
243: "331円", /* small a, dieresis or umlaut mark */
244: "333円", /* small c, cedilla */
245: "335円", /* small e, acute accent */
246: "336円", /* small e, circumflex accent */
247: "334円", /* small e, grave accent */
248: "346円", /* small eth, Icelandic */
249: "337円", /* small e, dieresis or umlaut mark */
250: "076円", /* greater than */
251: "342円", /* small i, acute accent */
252: "344円", /* small i, circumflex accent */
253: "340円", /* small i, grave accent */
254: "345円", /* small i, dieresis or umlaut mark */
255: "074円", /* less than */
256: "347円", /* small n, tilde */
257: "355円", /* small o, acute accent */
258: "356円", /* small o, circumflex accent */
259: "354円", /* small o, grave accent */
260: "371円", /* small o, slash */
261: "357円", /* small o, tilde */
262: "360円", /* small o, dieresis or umlaut mark */
263: "373円", /* small sharp s, German (sz ligature) */
264: "374円", /* small thorn, Icelandic */
265: "363円", /* small u, acute accent */
266: "364円", /* small u, circumflex accent */
267: "362円", /* small u, grave accent */
268: "366円", /* small u, dieresis or umlaut mark */
269: "367円", /* small y, acute accent */
270: "375円", /* small y, dieresis or umlaut mark */
1.1 timbl 271: };
272:
1.2 timbl 273: /* Entity values -- for IBM/PC Code Page 850 (International)
274: **
275: ** This MUST match exactly the table referred to in the DTD!
276: **
277: */
278: /* @@@@@@@@@@@@@@@@@ TBD */
279:
280:
281:
282: /* Set character set
283: ** ----------------
284: */
285:
286: PRIVATE char** p_entity_values = ISO_Latin1; /* Pointer to translation */
1.1 timbl 287:
1.2 timbl 288: PUBLIC void HTMLUseCharacterSet ARGS1(HTMLCharacterSet, i)
289: {
290: p_entity_values = (i == HTML_NEXT_CHARS) ? NeXTCharacters
291: : ISO_Latin1;
292: }
1.1 timbl 293:
294:
295: /* Flattening the style structure
296: ** ------------------------------
297: **
298: On the NeXT, and on any read-only browser, it is simpler for the text to have
299: a sequence of styles, rather than a nested tree of styles. In this
300: case we have to flatten the structure as it arrives from SGML tags into
301: a sequence of styles.
302: */
303:
304: /* If style really needs to be set, call this
305: */
1.4 timbl 306: PRIVATE void actually_set_style ARGS1(HTStructured *, me)
1.1 timbl 307: {
1.4 timbl 308: if (!me->text) { /* First time through */
309: me->text = HText_new2(me->node_anchor, me->target);
310: HText_beginAppend(me->text);
311: HText_setStyle(me->text, me->new_style);
312: me->in_word = NO;
1.1 timbl 313: } else {
1.4 timbl 314: HText_setStyle(me->text, me->new_style);
1.1 timbl 315: }
1.4 timbl 316: me->old_style = me->new_style;
317: me->style_change = NO;
1.1 timbl 318: }
319:
320: /* If you THINK you need to change style, call this
321: */
322:
1.11 timbl 323: PRIVATE void change_paragraph_style ARGS2(HTStructured *, me, HTStyle *,style)
1.1 timbl 324: {
1.4 timbl 325: if (me->new_style!=style) {
326: me->style_change = YES;
327: me->new_style = style;
1.1 timbl 328: }
1.11 timbl 329: me->in_word = NO;
1.1 timbl 330: }
331:
1.2 timbl 332: /*_________________________________________________________________________
333: **
334: ** A C T I O N R O U T I N E S
335: */
336:
337: /* Character handling
338: ** ------------------
1.1 timbl 339: */
1.4 timbl 340: PRIVATE void HTML_put_character ARGS2(HTStructured *, me, char, c)
1.1 timbl 341: {
1.2 timbl 342:
1.4 timbl 343: switch (me->sp[0].tag_number) {
1.2 timbl 344: case HTML_COMMENT:
345: break; /* Do Nothing */
346:
347: case HTML_TITLE:
1.4 timbl 348: HTChunkPutc(&me->title, c);
1.2 timbl 349: break;
350:
351:
352: case HTML_LISTING: /* Litteral text */
353: case HTML_XMP:
354: case HTML_PLAINTEXT:
355: case HTML_PRE:
356: /* We guarrantee that the style is up-to-date in begin_litteral
357: */
1.4 timbl 358: HText_appendCharacter(me->text, c);
1.2 timbl 359: break;
360:
361: default: /* Free format text */
1.4 timbl 362: if (me->style_change) {
1.2 timbl 363: if ((c=='\n') || (c==' ')) return; /* Ignore it */
364: UPDATE_STYLE;
365: }
366: if (c=='\n') {
1.4 timbl 367: if (me->in_word) {
368: HText_appendCharacter(me->text, ' ');
369: me->in_word = NO;
1.2 timbl 370: }
371: } else {
1.4 timbl 372: HText_appendCharacter(me->text, c);
373: me->in_word = YES;
1.2 timbl 374: }
375: } /* end switch */
1.1 timbl 376: }
377:
1.2 timbl 378:
379:
380: /* String handling
381: ** ---------------
382: **
383: ** This is written separately from put_character becuase the loop can
1.11 timbl 384: ** in some cases be promoted to a higher function call level for speed.
1.2 timbl 385: */
1.4 timbl 386: PRIVATE void HTML_put_string ARGS2(HTStructured *, me, CONST char*, s)
1.1 timbl 387: {
1.2 timbl 388:
1.4 timbl 389: switch (me->sp[0].tag_number) {
1.2 timbl 390: case HTML_COMMENT:
391: break; /* Do Nothing */
392:
393: case HTML_TITLE:
1.4 timbl 394: HTChunkPuts(&me->title, s);
1.2 timbl 395: break;
396:
397:
398: case HTML_LISTING: /* Litteral text */
399: case HTML_XMP:
400: case HTML_PLAINTEXT:
401: case HTML_PRE:
402:
403: /* We guarrantee that the style is up-to-date in begin_litteral
404: */
1.4 timbl 405: HText_appendText(me->text, s);
1.2 timbl 406: break;
407:
408: default: /* Free format text */
409: {
410: CONST char *p = s;
1.4 timbl 411: if (me->style_change) {
1.2 timbl 412: for (; *p && ((*p=='\n') || (*p==' ')); p++) ; /* Ignore leaders */
413: if (!*p) return;
414: UPDATE_STYLE;
415: }
416: for(; *p; p++) {
1.4 timbl 417: if (me->style_change) {
1.2 timbl 418: if ((*p=='\n') || (*p==' ')) continue; /* Ignore it */
419: UPDATE_STYLE;
420: }
421: if (*p=='\n') {
1.4 timbl 422: if (me->in_word) {
423: HText_appendCharacter(me->text, ' ');
424: me->in_word = NO;
1.2 timbl 425: }
426: } else {
1.4 timbl 427: HText_appendCharacter(me->text, *p);
428: me->in_word = YES;
1.2 timbl 429: }
430: } /* for */
431: }
432: } /* end switch */
1.1 timbl 433: }
434:
435:
1.2 timbl 436: /* Buffer write
1.3 timbl 437: ** ------------
1.1 timbl 438: */
1.4 timbl 439: PRIVATE void HTML_write ARGS3(HTStructured *, me, CONST char*, s, int, l)
1.1 timbl 440: {
1.2 timbl 441: CONST char* p;
442: CONST char* e = s+l;
1.4 timbl 443: for (p=s; s<e; p++) HTML_put_character(me, *p);
1.1 timbl 444: }
1.2 timbl 445:
446:
447: /* Start Element
448: ** -------------
449: */
450: PRIVATE void HTML_start_element ARGS4(
1.4 timbl 451: HTStructured *, me,
1.16 timbl 452: int, element_number,
1.3 timbl 453: CONST BOOL*, present,
1.16 timbl 454: CONST char **, value)
1.2 timbl 455: {
456: switch (element_number) {
457: case HTML_A:
458: {
1.8 timbl 459: HTChildAnchor * source;
1.9 timbl 460: char * href = NULL;
461: if (present[HTML_A_HREF]) {
462: StrAllocCopy(href, value[HTML_A_HREF]);
463: HTSimplify(href);
464: }
1.8 timbl 465: source = HTAnchor_findChildAndLink(
1.4 timbl 466: me->node_anchor, /* parent */
1.2 timbl 467: present[HTML_A_NAME] ? value[HTML_A_NAME] : 0, /* Tag */
1.9 timbl 468: present[HTML_A_HREF] ? href : 0, /* Addresss */
1.16 timbl 469: present[HTML_A_REL] && value[HTML_A_REL] ?
470: (HTLinkType*)HTAtom_for(value[HTML_A_REL])
1.2 timbl 471: : 0);
472:
473: if (present[HTML_A_TITLE] && value[HTML_A_TITLE]) {
474: HTParentAnchor * dest =
475: HTAnchor_parent(
476: HTAnchor_followMainLink((HTAnchor*)source)
477: );
478: if (!HTAnchor_title(dest))
479: HTAnchor_setTitle(dest, value[HTML_A_TITLE]);
480: }
481: UPDATE_STYLE;
1.4 timbl 482: HText_beginAnchor(me->text, source);
1.2 timbl 483: }
484: break;
485:
486: case HTML_TITLE:
1.4 timbl 487: HTChunkClear(&me->title);
1.2 timbl 488: break;
489:
490: case HTML_NEXTID:
491: /* if (present[NEXTID_N] && value[NEXTID_N])
1.4 timbl 492: HText_setNextId(me->text, atoi(value[NEXTID_N])); */
1.2 timbl 493: break;
494:
495: case HTML_ISINDEX:
1.4 timbl 496: HTAnchor_setIndex(me->node_anchor);
1.2 timbl 497: break;
498:
1.15 timbl 499: case HTML_BR:
500: UPDATE_STYLE;
501: HText_appendCharacter(me->text, '\n');
502: me->in_word = NO;
503: break;
504:
505: case HTML_HR:
506: UPDATE_STYLE;
507: HText_appendCharacter(me->text, '\n');
1.16 timbl 508: HText_appendText(me->text, "___________________________________");
1.15 timbl 509: HText_appendCharacter(me->text, '\n');
510: me->in_word = NO;
511: break;
512:
1.2 timbl 513: case HTML_P:
514: UPDATE_STYLE;
1.4 timbl 515: HText_appendParagraph(me->text);
516: me->in_word = NO;
1.2 timbl 517: break;
518:
519: case HTML_DL:
1.11 timbl 520: change_paragraph_style(me, present && present[DL_COMPACT]
1.16 timbl 521: ? styles[HTML_DL]
1.2 timbl 522: : styles[HTML_DL]);
523: break;
524:
525: case HTML_DT:
1.4 timbl 526: if (!me->style_change) {
527: HText_appendParagraph(me->text);
528: me->in_word = NO;
1.2 timbl 529: }
530: break;
531:
532: case HTML_DD:
533: UPDATE_STYLE;
1.4 timbl 534: HTML_put_character(me, '\t'); /* Just tab out one stop */
535: me->in_word = NO;
536: break;
1.2 timbl 537:
538: case HTML_UL:
539: case HTML_OL:
540: case HTML_MENU:
541: case HTML_DIR:
1.11 timbl 542: change_paragraph_style(me, styles[element_number]);
1.2 timbl 543: break;
544:
545: case HTML_LI:
546: UPDATE_STYLE;
1.7 timbl 547: if (me->sp[0].tag_number != HTML_DIR)
1.4 timbl 548: HText_appendParagraph(me->text);
1.2 timbl 549: else
1.4 timbl 550: HText_appendCharacter(me->text, '\t'); /* Tab @@ nl for UL? */
551: me->in_word = NO;
1.2 timbl 552: break;
553:
554: case HTML_LISTING: /* Litteral text */
555: case HTML_XMP:
556: case HTML_PLAINTEXT:
557: case HTML_PRE:
1.11 timbl 558: change_paragraph_style(me, styles[element_number]);
1.2 timbl 559: UPDATE_STYLE;
1.4 timbl 560: if (me->comment_end)
561: HText_appendText(me->text, me->comment_end);
1.2 timbl 562: break;
1.11 timbl 563:
564: case HTML_HTML: /* Ignore these altogether */
565: case HTML_HEAD:
566: case HTML_BODY:
567:
1.10 timbl 568: case HTML_IMG: /* Images -- ignore */
569:
570: case HTML_TT: /* Physical character highlighting */
571: case HTML_B: /* Currently ignored */
572: case HTML_I:
573: case HTML_U:
574:
575: case HTML_EM: /* Logical character highlighting */
576: case HTML_STRONG: /* Currently ignored */
577: case HTML_CODE:
578: case HTML_SAMP:
579: case HTML_KBD:
580: case HTML_VAR:
581: case HTML_DFN:
582: case HTML_CITE:
583: break;
584:
1.11 timbl 585: case HTML_H1: /* paragraph styles */
586: case HTML_H2:
587: case HTML_H3:
588: case HTML_H4:
589: case HTML_H5:
590: case HTML_H6:
591: case HTML_H7:
592: case HTML_ADDRESS:
593: case HTML_BLOCKQUOTE:
594: change_paragraph_style(me, styles[element_number]); /* May be postponed */
1.2 timbl 595: break;
596:
597: } /* end switch */
598:
1.16 timbl 599: if (me->dtd->tags[element_number].contents!= SGML_EMPTY) {
1.13 timbl 600: if (me->sp == me->stack) {
1.12 timbl 601: fprintf(stderr, "HTML: ****** Maximum nesting of %d exceded!\n",
602: MAX_NESTING);
603: return;
604: }
1.4 timbl 605: --(me->sp);
606: me->sp[0].style = me->new_style; /* Stack new style */
607: me->sp[0].tag_number = element_number;
1.10 timbl 608: }
1.1 timbl 609: }
1.10 timbl 610:
1.2 timbl 611:
1.1 timbl 612: /* End Element
1.2 timbl 613: ** -----------
1.1 timbl 614: **
1.2 timbl 615: */
616: /* When we end an element, the style must be returned to that
1.1 timbl 617: ** in effect before that element. Note that anchors (etc?)
618: ** don't have an associated style, so that we must scan down the
619: ** stack for an element with a defined style. (In fact, the styles
620: ** should be linked to the whole stack not just the top one.)
621: ** TBL 921119
1.6 timbl 622: **
623: ** We don't turn on "CAREFUL" check because the parser produces
624: ** (internal code errors apart) good nesting. The parser checks
625: ** incoming code errors, not this module.
1.1 timbl 626: */
1.4 timbl 627: PRIVATE void HTML_end_element ARGS2(HTStructured *, me, int , element_number)
1.1 timbl 628: {
1.2 timbl 629: #ifdef CAREFUL /* parser assumed to produce good nesting */
1.4 timbl 630: if (element_number != me->sp[0].tag_number) {
1.2 timbl 631: fprintf(stderr, "HTMLText: end of element %s when expecting end of %s\n",
1.16 timbl 632: me->dtd->tags[element_number].name,
633: me->dtd->tags[me->sp->tag_number].name);
1.6 timbl 634: /* panic */
1.1 timbl 635: }
1.2 timbl 636: #endif
637:
1.4 timbl 638: me->sp++; /* Pop state off stack */
1.2 timbl 639:
640: switch(element_number) {
641:
642: case HTML_A:
643: UPDATE_STYLE;
1.4 timbl 644: HText_endAnchor(me->text);
1.2 timbl 645: break;
646:
647: case HTML_TITLE:
1.4 timbl 648: HTChunkTerminate(&me->title);
649: HTAnchor_setTitle(me->node_anchor, me->title.data);
1.2 timbl 650: break;
651:
652: case HTML_LISTING: /* Litteral text */
653: case HTML_XMP:
654: case HTML_PLAINTEXT:
655: case HTML_PRE:
1.4 timbl 656: if (me->comment_start)
657: HText_appendText(me->text, me->comment_start);
1.2 timbl 658: /* Fall through */
659:
660: default:
661:
1.11 timbl 662: change_paragraph_style(me, me->sp->style); /* Often won't really change */
1.2 timbl 663: break;
664:
665: } /* switch */
1.1 timbl 666: }
667:
1.2 timbl 668:
669: /* Expanding entities
670: ** ------------------
671: */
672: /* (In fact, they all shrink!)
1.1 timbl 673: */
1.2 timbl 674:
1.4 timbl 675: PRIVATE void HTML_put_entity ARGS2(HTStructured *, me, int, entity_number)
1.1 timbl 676: {
1.4 timbl 677: HTML_put_string(me, ISO_Latin1[entity_number]); /* @@ Other representations */
1.1 timbl 678: }
1.2 timbl 679:
680:
681: /* Free an HTML object
682: ** -------------------
683: **
1.4 timbl 684: ** If the document is empty, the text object will not yet exist.
685: So we could in fact abandon creating the document and return
686: an error code. In fact an empty document is an important type
687: of document, so we don't.
688: **
689: ** If non-interactive, everything is freed off. No: crashes -listrefs
1.2 timbl 690: ** Otherwise, the interactive object is left.
691: */
1.4 timbl 692: PUBLIC void HTML_free ARGS1(HTStructured *, me)
1.1 timbl 693: {
1.4 timbl 694: UPDATE_STYLE; /* Creates empty document here! */
695: if (me->comment_end)
696: HTML_put_string(me,me->comment_end);
697: HText_endAppend(me->text);
698:
699: if (me->target) {
700: (*me->targetClass.free)(me->target);
1.2 timbl 701: }
1.4 timbl 702: free(me);
1.1 timbl 703: }
704:
705:
1.14 timbl 706: PRIVATE void HTML_abort ARGS2(HTStructured *, me, HTError, e)
1.1 timbl 707:
1.14 timbl 708: {
709: if (me->target) {
710: (*me->targetClass.abort)(me->target, e);
711: }
712: free(me);
713:
1.1 timbl 714: }
715:
1.2 timbl 716:
717: /* Get Styles from style sheet
718: ** ---------------------------
719: */
720: PRIVATE void get_styles NOARGS
1.1 timbl 721: {
1.2 timbl 722: got_styles = YES;
723:
724: default_style = HTStyleNamed(styleSheet, "Normal");
1.1 timbl 725:
1.2 timbl 726: styles[HTML_H1] = HTStyleNamed(styleSheet, "Heading1");
727: styles[HTML_H2] = HTStyleNamed(styleSheet, "Heading2");
728: styles[HTML_H3] = HTStyleNamed(styleSheet, "Heading3");
729: styles[HTML_H4] = HTStyleNamed(styleSheet, "Heading4");
730: styles[HTML_H5] = HTStyleNamed(styleSheet, "Heading5");
731: styles[HTML_H6] = HTStyleNamed(styleSheet, "Heading6");
732: styles[HTML_H7] = HTStyleNamed(styleSheet, "Heading7");
733:
734: styles[HTML_DL] = HTStyleNamed(styleSheet, "Glossary");
735: styles[HTML_UL] =
736: styles[HTML_OL] = HTStyleNamed(styleSheet, "List");
737: styles[HTML_MENU] = HTStyleNamed(styleSheet, "Menu");
738: styles[HTML_DIR] = HTStyleNamed(styleSheet, "Dir");
1.16 timbl 739: /* styles[HTML_DLC] = HTStyleNamed(styleSheet, "GlossaryCompact"); */
1.2 timbl 740: styles[HTML_ADDRESS]= HTStyleNamed(styleSheet, "Address");
741: styles[HTML_BLOCKQUOTE]= HTStyleNamed(styleSheet, "BlockQuote");
742: styles[HTML_PLAINTEXT] =
743: styles[HTML_XMP] = HTStyleNamed(styleSheet, "Example");
744: styles[HTML_PRE] = HTStyleNamed(styleSheet, "Preformatted");
745: styles[HTML_LISTING] = HTStyleNamed(styleSheet, "Listing");
746: }
747: /* P U B L I C
748: */
749:
750: /* Structured Object Class
751: ** -----------------------
752: */
753: PUBLIC CONST HTStructuredClass HTMLPresentation = /* As opposed to print etc */
754: {
755: "text/html",
756: HTML_free,
1.14 timbl 757: HTML_abort,
1.2 timbl 758: HTML_put_character, HTML_put_string, HTML_write,
759: HTML_start_element, HTML_end_element,
760: HTML_put_entity
761: };
1.1 timbl 762:
1.4 timbl 763:
1.2 timbl 764: /* New Structured Text object
765: ** --------------------------
766: **
1.16 timbl 767: ** The structured stream can generate either presentation,
1.4 timbl 768: ** or plain text, or HTML.
1.1 timbl 769: */
1.16 timbl 770: PUBLIC HTStructured* HTML_new ARGS5(
771: HTRequest *, request,
772: void *, param,
773: HTFormat, input_format,
774: HTFormat, output_format,
775: HTStream *, output_stream)
1.1 timbl 776: {
777:
1.4 timbl 778: HTStructured * me;
779:
1.16 timbl 780: if (output_format != WWW_PLAINTEXT
781: && output_format != WWW_PRESENT
782: && output_format != HTAtom_for("text/x-c")) {
783: HTStream * intermediate = HTStreamStack(WWW_HTML, request);
1.6 timbl 784: if (intermediate) return HTMLGenerator(intermediate);
1.4 timbl 785: fprintf(stderr, "** Internal error: can't parse HTML to %s\n",
1.16 timbl 786: HTAtom_name(output_format));
1.4 timbl 787: exit (-99);
788: }
789:
790: me = (HTStructured*) malloc(sizeof(*me));
791: if (me == NULL) outofmem(__FILE__, "HTML_new");
1.1 timbl 792:
793: if (!got_styles) get_styles();
794:
1.4 timbl 795: me->isa = &HTMLPresentation;
1.16 timbl 796: me->dtd = &DTD;
797: me->node_anchor = request->anchor;
1.4 timbl 798: me->title.size = 0;
799: me->title.growby = 128;
800: me->title.allocated = 0;
801: me->title.data = 0;
802: me->text = 0;
803: me->style_change = YES; /* Force check leading to text creation */
804: me->new_style = default_style;
805: me->old_style = 0;
806: me->sp = me->stack + MAX_NESTING - 1;
807: me->sp->tag_number = -1; /* INVALID */
808: me->sp->style = default_style; /* INVALID */
1.1 timbl 809:
1.4 timbl 810: me->comment_start = NULL;
811: me->comment_end = NULL;
1.16 timbl 812: me->target = output_stream;
813: if (output_stream) me->targetClass = *output_stream->isa; /* Copy pointers */
1.1 timbl 814:
1.4 timbl 815: return (HTStructured*) me;
1.1 timbl 816: }
817:
818:
1.2 timbl 819: /* HTConverter for HTML to plain text
820: ** ----------------------------------
1.1 timbl 821: **
1.2 timbl 822: ** This will convert from HTML to presentation or plain text.
1.1 timbl 823: */
1.16 timbl 824: PUBLIC HTStream* HTMLToPlain ARGS5(
825: HTRequest *, request,
826: void *, param,
827: HTFormat, input_format,
828: HTFormat, output_format,
829: HTStream *, output_stream)
1.1 timbl 830: {
1.16 timbl 831: return SGML_new(&DTD, HTML_new(
832: request, NULL, input_format, output_format, output_stream));
1.1 timbl 833: }
834:
835:
1.2 timbl 836: /* HTConverter for HTML to C code
837: ** ------------------------------
838: **
839: ** C copde is like plain text but all non-preformatted code
840: ** is commented out.
841: ** This will convert from HTML to presentation or plain text.
842: */
1.16 timbl 843: PUBLIC HTStream* HTMLToC ARGS5(
844: HTRequest *, request,
845: void *, param,
846: HTFormat, input_format,
847: HTFormat, output_format,
848: HTStream *, output_stream)
1.1 timbl 849: {
1.4 timbl 850:
851: HTStructured * html;
852:
1.16 timbl 853: (*output_stream->isa->put_string)(output_stream, "/* "); /* Before even title */
854: html = HTML_new(request, NULL, input_format, output_format, output_stream);
1.2 timbl 855: html->comment_start = "/* ";
1.16 timbl 856: html->dtd = &DTD;
1.2 timbl 857: html->comment_end = " */\n"; /* Must start in col 1 for cpp */
1.4 timbl 858: /* HTML_put_string(html,html->comment_start); */
1.16 timbl 859: return SGML_new(&DTD, html);
1.1 timbl 860: }
861:
862:
1.2 timbl 863: /* Presenter for HTML
864: ** ------------------
865: **
866: ** This will convert from HTML to presentation or plain text.
867: **
868: ** Override this if you have a windows version
1.1 timbl 869: */
1.2 timbl 870: #ifndef GUI
1.16 timbl 871: PUBLIC HTStream* HTMLPresent ARGS5(
872: HTRequest *, request,
873: void *, param,
874: HTFormat, input_format,
875: HTFormat, output_format,
876: HTStream *, output_stream)
1.1 timbl 877: {
1.16 timbl 878: return SGML_new(&DTD, HTML_new(
879: request, NULL, input_format, output_format, output_stream));
1.1 timbl 880: }
1.2 timbl 881: #endif
1.1 timbl 882:
883:
1.2 timbl 884: /* Record error message as a hypertext object
885: ** ------------------------------------------
886: **
887: ** The error message should be marked as an error so that
888: ** it can be reloaded later.
889: ** This implementation just throws up an error message
890: ** and leaves the document unloaded.
1.9 timbl 891: ** A smarter implementation would load an error document,
892: ** marking at such so that it is retried on reload.
1.1 timbl 893: **
1.2 timbl 894: ** On entry,
895: ** sink is a stream to the output device if any
896: ** number is the HTTP error number
897: ** message is the human readable message.
1.9 timbl 898: **
899: ** On exit,
900: ** returns a negative number to indicate lack of success in the load.
1.1 timbl 901: */
1.2 timbl 902:
903: PUBLIC int HTLoadError ARGS3(
1.17 ! luotonen 904: HTRequest *, req,
1.2 timbl 905: int, number,
906: CONST char *, message)
907: {
908: HTAlert(message); /* @@@@@@@@@@@@@@@@@@@ */
909: return -number;
910: }
911:
Webmaster