Annotation of libwww/Library/src/HTMLPDTD.c, revision 2.13
2.12 frystyk 1: /* HTMLPDTD.c
2: ** STATIC DTD FOR HTML
3: **
4: ** (c) COPYRIGHT CERN 1994.
5: ** Please first read the full copyright statement in the file COPYRIGH.
2.1 timbl 6: **
7: ** 6 Nov 93 MD Increased size of img_attr array to make space
8: ** for terminator.
9: */
10:
2.13 ! roeber 11: #include "sysdep.h"
2.1 timbl 12: #include "HTMLPDTD.h"
13:
14: /* Entity Names
15: ** ------------
16: **
17: ** This table must be matched exactly with ALL the translation tables
18: */
2.11 timbl 19:
2.10 timbl 20: static CONST char* entities[HTML_ENTITIES] = {
2.1 timbl 21: "AElig", /* capital AE diphthong (ligature) */
22: "Aacute", /* capital A, acute accent */
23: "Acirc", /* capital A, circumflex accent */
24: "Agrave", /* capital A, grave accent */
25: "Aring", /* capital A, ring */
26: "Atilde", /* capital A, tilde */
27: "Auml", /* capital A, dieresis or umlaut mark */
28: "Ccedil", /* capital C, cedilla */
29: "ETH", /* capital Eth, Icelandic */
30: "Eacute", /* capital E, acute accent */
31: "Ecirc", /* capital E, circumflex accent */
32: "Egrave", /* capital E, grave accent */
33: "Euml", /* capital E, dieresis or umlaut mark */
34: "Iacute", /* capital I, acute accent */
35: "Icirc", /* capital I, circumflex accent */
36: "Igrave", /* capital I, grave accent */
37: "Iuml", /* capital I, dieresis or umlaut mark */
38: "Ntilde", /* capital N, tilde */
39: "Oacute", /* capital O, acute accent */
40: "Ocirc", /* capital O, circumflex accent */
41: "Ograve", /* capital O, grave accent */
42: "Oslash", /* capital O, slash */
43: "Otilde", /* capital O, tilde */
44: "Ouml", /* capital O, dieresis or umlaut mark */
45: "THORN", /* capital THORN, Icelandic */
46: "Uacute", /* capital U, acute accent */
47: "Ucirc", /* capital U, circumflex accent */
48: "Ugrave", /* capital U, grave accent */
49: "Uuml", /* capital U, dieresis or umlaut mark */
50: "Yacute", /* capital Y, acute accent */
51: "aacute", /* small a, acute accent */
52: "acirc", /* small a, circumflex accent */
53: "aelig", /* small ae diphthong (ligature) */
54: "agrave", /* small a, grave accent */
55: "amp", /* ampersand */
56: "aring", /* small a, ring */
57: "atilde", /* small a, tilde */
58: "auml", /* small a, dieresis or umlaut mark */
59: "ccedil", /* small c, cedilla */
60: "eacute", /* small e, acute accent */
61: "ecirc", /* small e, circumflex accent */
62: "egrave", /* small e, grave accent */
63: "eth", /* small eth, Icelandic */
64: "euml", /* small e, dieresis or umlaut mark */
65: "gt", /* greater than */
66: "iacute", /* small i, acute accent */
67: "icirc", /* small i, circumflex accent */
68: "igrave", /* small i, grave accent */
69: "iuml", /* small i, dieresis or umlaut mark */
70: "lt", /* less than */
71: "ntilde", /* small n, tilde */
72: "oacute", /* small o, acute accent */
73: "ocirc", /* small o, circumflex accent */
74: "ograve", /* small o, grave accent */
75: "oslash", /* small o, slash */
76: "otilde", /* small o, tilde */
2.10 timbl 77: "ouml", /* small o, dieresis or umlaut mark */
78: "quot", /* double quote sign - June 1994 */
2.1 timbl 79: "szlig", /* small sharp s, German (sz ligature) */
80: "thorn", /* small thorn, Icelandic */
81: "uacute", /* small u, acute accent */
82: "ucirc", /* small u, circumflex accent */
83: "ugrave", /* small u, grave accent */
84: "uuml", /* small u, dieresis or umlaut mark */
85: "yacute", /* small y, acute accent */
86: "yuml", /* small y, dieresis or umlaut mark */
87: };
88:
89:
90:
91: /* Attribute Lists
92: ** ---------------
93: **
94: ** Lists must be in alphatbetical order by attribute name
95: ** The tag elements contain the number of attributes
96: */
97: static attr no_attr[1] =
98: {{ 0 }};
99:
100: static attr a_attr[HTML_A_ATTRIBUTES+1] = { /* Anchor attributes */
101: { "EFFECT" },
102: { "HREF"},
103: { "ID" },
104: { "METHODS" },
105: { "NAME" }, /* Should be ID */
106: { "PRINT" },
107: { "REL" }, /* Relationship */
108: { "REV" }, /* Reverse relationship */
109: { "SHAPE" },
110: { "TITLE" },
111: { 0 } /* Terminate list */
112: };
113:
114: static attr base_attr[] = { /* BASE attributes */
115: { "HREF"},
116: { 0 } /* Terminate list */
117: };
118:
119:
120: static attr changed_attr[] = {
121: { "ID" },
122: { "IDREF" },
123: { 0 } /* terminate list */
124: };
125:
126: static attr fig_attr[] = { /* Figures */
127: { "ALIGN" },
128: { "ID" },
129: { "INDEX" },
130: { "ISMAP" }, /* @@@ NO! */
131: { "LANG" },
132: { "SRC" },
133: { 0 } /* terminate list */
134: };
135:
136: static attr form_attr[] = { /* General, for many things */
137: { "ACTION" },
138: { "ID" },
139: { "INDEX" },
140: { "LANG" },
141: { "METHOD" },
142: { 0 } /* terminate list */
143: };
144:
145: static attr gen_attr[] = { /* General, for many things */
146: { "ID" },
147: { "INDEX" },
148: { "LANG" },
149: { 0 } /* terminate list */
150: };
151:
152: static attr htmlplus_attr[] = { /* wrapper HTMLPLUS */
153: { "FORMS" },
154: { "VERSION" },
155: { 0 } /* terminate list */
156: };
157:
158: static attr id_attr[2] = {
159: { "ID" },
160: { 0 } /* terminate list */
161: };
162:
163: static attr image_attr[HTML_IMAGE_ATTRIBUTES+1] = { /* Image attributes */
164: { "ALIGN" },
165: { "ISMAP"}, /* Use HTTP SpaceJump instead */
166: { "LANG" },
167: { "SEETHRU"},
168: { "SRC"},
169: { 0 } /* Terminate list */
170: };
171:
172: static attr img_attr[HTML_IMG_ATTRIBUTES+1] = { /* IMG attributes */
173: { "ALIGN" },
174: { "ALT" },
175: { "ISMAP"}, /* Use HTTP SpaceJump instead */
176: { "SEETHRU"},
177: { "SRC"},
178: { 0 } /* Terminate list */
179: };
180:
181: static attr input_attr[HTML_INPUT_ATTRIBUTES+1] = {
182: { "ALIGN" },
183: { "CHECKED" },
184: { "DISABLED" },
185: { "ERROR" },
186: { "MAX" },
187: { "MIN" },
188: { "NAME" },
189: { "SIZE" },
190: { "SRC" },
191: { "TYPE" },
192: { "VALUE" },
2.2 timbl 193: { "0" }
2.1 timbl 194: };
195:
196: static attr l_attr[] = {
197: { "ALIGN"},
198: { "ID" },
199: { "LANG" },
200: { "INDEX" },
201: { 0 } /* Terminate list */
202: };
203:
204: static attr li_attr[] = {
205: { "ID" },
206: { "LANG" },
207: { "INDEX" },
208: { "SRC"},
209: { 0 } /* Terminate list */
210: };
211:
212: static attr link_attr[HTML_LINK_ATTRIBUTES+1] = { /* link attributes */
213: { "HREF"},
214: { "IDREF" },
215: { "METHODS" },
216: { "REL" }, /* Relationship */
217: { "REV" }, /* Reverse relationship */
218: { 0 } /* Terminate list */
219: };
220:
221: static attr list_attr[] = {
222: { "COMPACT"},
223: { "ID" },
224: { "LANG" },
225: { "INDEX" },
226: { 0 } /* Terminate list */
227: };
228:
2.3 frystyk 229: static attr glossary_attr[HTML_DL_ATTRIBUTES+1] = {
230: { "ID" },
231: { "COMPACT " },
232: { "INDEX" },
2.1 timbl 233: { 0 } /* Terminate list */
234: };
235:
236: static attr nextid_attr[HTML_NEXTID_ATTRIBUTES+1] = {
237: { "N" },
238: { 0 } /* Terminate list */
239: };
240:
241: static attr note_attr[HTML_NOTE_ATTRIBUTES+1] = { /* Footnotes etc etc */
242: { "ID" },
243: { "INDEX" },
244: { "LANG" },
245: { "ROLE" },
246: { 0 } /* terminate list */
247: };
248:
249: static attr option_attr[HTML_OPTION_ATTRIBUTES+1] = {
250: { "DISABLED" },
251: { "LANG" },
252: { "SELECTED" },
253: { 0 }
254: };
255:
256: #ifdef OLD_CODE
257: static attr pre_attr[HTML_PRE_ATTRIBUTES+1] = {
258: { "WIDTH" },
259: { 0 } /* Terminate list */
260: };
261: #endif
262:
263: static attr render_attr[HTML_RENDER_ATTRIBUTES+1] = {
264: { "STYLE" },
265: { "TAG" },
266: { 0 } /* Terminate list */
267: };
268:
269: static attr select_attr[HTML_SELECT_ATTRIBUTES+1] = {
270: { "ERROR" },
271: { "LANG" },
272: { "NAME" },
273: { "SEVERAL" },
274: { 0 },
275: };
276:
277: static attr tab_attr[HTML_TAB_ATTRIBUTES+1] = {
278: { "ALIGN" },
279: { "AT" },
280: { 0 }
281: };
282:
283: static attr table_attr[HTML_TABLE_ATTRIBUTES+1] = {
284: { "BORDER" },
285: { "ID" },
286: { "INDEX" },
287: { "LANG" },
288: { 0 }
289: };
290:
291: static attr td_attr[HTML_TD_ATTRIBUTES+1] = {
292: { "ALIGN" },
293: { "COLSPAN" },
294: { "ROWSPAN" },
295: { "TAG" },
296: { 0 }
297: };
298:
299: static attr textarea_attr[HTML_TEXTAREA_ATTRIBUTES+1] = {
300: { "COLS" },
301: { "DISABLED" },
302: { "ERROR" },
303: { "LANG" },
304: { "NAME" },
305: { "ROWS" },
306: { 0 }
307: };
308:
309: static attr ul_attr[HTML_UL_ATTRIBUTES+1] = {
310: { "COMPACT" },
311: { "ID" },
312: { "INDEX" },
313: { "LANG" },
314: { "PLAIN" },
315: { "WRAP" },
316: { 0 }
317: };
318:
319:
320: /* Elements
321: ** --------
322: **
323: ** Must match definitions in HTMLPDTD.html!
324: ** Must be in alphabetical order.
325: **
326: ** HTML is included to allow HTML documents to be parsed as a subset.
2.4 frystyk 327: ** TITLE is changed to type SGML_MIXED so it get parsed as well
328: ** Henrik 08/03-94
2.1 timbl 329: **
330: ** Name, Attributes, content
331: */
332: static HTTag tags[HTMLP_ELEMENTS] = {
333: { "A" , a_attr, HTML_A_ATTRIBUTES, SGML_MIXED },
334: { "ABBREV" , gen_attr, HTML_GEN_ATTRIBUTES, SGML_MIXED },
335: { "ABSTRACT" , gen_attr, HTML_GEN_ATTRIBUTES, SGML_MIXED },
336: { "ACRONYM" , gen_attr, HTML_GEN_ATTRIBUTES, SGML_MIXED },
337: { "ADDED" , gen_attr, HTML_GEN_ATTRIBUTES, SGML_MIXED },
338: { "ADDRESS" , no_attr, 0, SGML_MIXED },
339: { "ARG" , gen_attr, HTML_GEN_ATTRIBUTES, SGML_MIXED },
340: { "B" , no_attr, 0, SGML_MIXED },
341: { "BASE" , base_attr, HTML_BASE_ATTRIBUTES, SGML_MIXED },
342: { "BLOCKQUOTE", no_attr, 0, SGML_MIXED },
343: { "BODY" , no_attr, 0, SGML_MIXED },
344: { "BOX" , no_attr, 0, SGML_MIXED },
345: { "BR" , gen_attr, HTML_GEN_ATTRIBUTES, SGML_EMPTY },
346: { "BYLINE" , gen_attr, HTML_GEN_ATTRIBUTES, SGML_MIXED },
347: { "CAPTION" , gen_attr, HTML_GEN_ATTRIBUTES, SGML_MIXED },
348: { "CHANGED" , changed_attr, HTML_CHANGED_ATTRIBUTES, SGML_MIXED },
349: { "CITE" , gen_attr, HTML_GEN_ATTRIBUTES, SGML_MIXED },
350: { "CMD" , gen_attr, HTML_GEN_ATTRIBUTES, SGML_MIXED },
351: { "CODE" , gen_attr, HTML_GEN_ATTRIBUTES, SGML_MIXED },
352: { "COMMENT" , no_attr, 0, SGML_MIXED },
353: { "DD" , gen_attr, HTML_GEN_ATTRIBUTES, SGML_EMPTY },
354: { "DFN" , gen_attr, HTML_GEN_ATTRIBUTES, SGML_MIXED },
355: { "DIR" , gen_attr, HTML_GEN_ATTRIBUTES, SGML_MIXED },
356: { "DL" , glossary_attr,HTML_DL_ATTRIBUTES, SGML_MIXED },
357: { "DT" , gen_attr, HTML_GEN_ATTRIBUTES, SGML_EMPTY },
358: { "EM" , gen_attr, HTML_GEN_ATTRIBUTES, SGML_MIXED },
359: { "FIG" , fig_attr, HTML_FIG_ATTRIBUTES, SGML_MIXED },
360: { "FOOTNOTE" , gen_attr, HTML_GEN_ATTRIBUTES, SGML_MIXED },
361: { "FORM" , form_attr, HTML_FORM_ATTRIBUTES, SGML_MIXED },
362: { "H1" , gen_attr, HTML_GEN_ATTRIBUTES, SGML_MIXED },
363: { "H2" , gen_attr, HTML_GEN_ATTRIBUTES, SGML_MIXED },
364: { "H3" , gen_attr, HTML_GEN_ATTRIBUTES, SGML_MIXED },
365: { "H4" , gen_attr, HTML_GEN_ATTRIBUTES, SGML_MIXED },
366: { "H5" , gen_attr, HTML_GEN_ATTRIBUTES, SGML_MIXED },
367: { "H6" , gen_attr, HTML_GEN_ATTRIBUTES, SGML_MIXED },
368: { "H7" , gen_attr, HTML_GEN_ATTRIBUTES, SGML_MIXED },
369: { "HEAD" , no_attr, 0, SGML_MIXED },
370: { "HR" , gen_attr, HTML_GEN_ATTRIBUTES, SGML_EMPTY },
371: { "HTML" , no_attr, 0, SGML_MIXED }, /* */
2.2 timbl 372: { "HTMLPLUS", htmlplus_attr,HTML_HTMLPLUS_ATTRIBUTES, SGML_MIXED },
2.1 timbl 373: { "I" , gen_attr, HTML_GEN_ATTRIBUTES, SGML_MIXED },
374: { "IMAGE" , image_attr, HTML_IMAGE_ATTRIBUTES, SGML_EMPTY },
2.7 frystyk 375: { "IMG" , img_attr, HTML_IMG_ATTRIBUTES, SGML_EMPTY },
2.1 timbl 376: { "INPUT" , input_attr, HTML_INPUT_ATTRIBUTES, SGML_EMPTY },
377: { "ISINDEX" , no_attr, 0, SGML_EMPTY },
378: { "KBD" , gen_attr, HTML_GEN_ATTRIBUTES, SGML_MIXED },
379: { "L" , l_attr, HTML_L_ATTRIBUTES, SGML_MIXED },
380: { "LI" , li_attr, HTML_LI_ATTRIBUTES, SGML_EMPTY },
381: { "LINK" , link_attr, HTML_LINK_ATTRIBUTES, SGML_EMPTY },
382: { "LISTING" , no_attr, 0, SGML_LITERAL },
383: { "LIT" , gen_attr, HTML_GEN_ATTRIBUTES, SGML_MIXED },
384: { "MARGIN" , gen_attr, HTML_GEN_ATTRIBUTES, SGML_MIXED },
385: { "MATH" , id_attr, HTML_ID_ATTRIBUTE, SGML_MIXED },
386: { "MENU" , gen_attr, HTML_GEN_ATTRIBUTES, SGML_MIXED },
387: { "NEXTID" , nextid_attr, 1, SGML_EMPTY },
388: { "NOTE" , note_attr, HTML_NOTE_ATTRIBUTES, SGML_EMPTY },
389: { "OL" , list_attr, HTML_LIST_ATTRIBUTES, SGML_MIXED },
2.8 timbl 390: { "OPTION" , option_attr, HTML_OPTION_ATTRIBUTES, SGML_EMPTY },/*Mixed?*/
2.1 timbl 391: { "OVER" , no_attr, 0, SGML_MIXED },
392: { "P" , l_attr, HTML_L_ATTRIBUTES, SGML_EMPTY },
393: { "PERSON" , gen_attr, HTML_GEN_ATTRIBUTES, SGML_MIXED },
394: { "PLAINTEXT", no_attr, 0, SGML_LITERAL },
395: { "PRE" , gen_attr, HTML_GEN_ATTRIBUTES, SGML_MIXED },
396: { "Q" , gen_attr, HTML_GEN_ATTRIBUTES, SGML_MIXED },
397: { "QUOTE" , gen_attr, HTML_GEN_ATTRIBUTES, SGML_MIXED },
398: { "RENDER" , render_attr, HTML_RENDER_ATTRIBUTES, SGML_MIXED },
399: { "REMOVED" , gen_attr, HTML_GEN_ATTRIBUTES, SGML_MIXED },
400: { "S" , gen_attr, HTML_GEN_ATTRIBUTES, SGML_MIXED },
401: { "SAMP" , gen_attr, HTML_GEN_ATTRIBUTES, SGML_MIXED },
402: { "SELECT" , select_attr, HTML_SELECT_ATTRIBUTES, SGML_MIXED },
403: { "STRONG" , gen_attr, HTML_GEN_ATTRIBUTES, SGML_MIXED },
404: { "SUB" , gen_attr, HTML_GEN_ATTRIBUTES, SGML_MIXED },
405: { "SUP" , gen_attr, HTML_GEN_ATTRIBUTES, SGML_MIXED },
406: { "TAB" , tab_attr, HTML_TAB_ATTRIBUTES, SGML_EMPTY },
407: { "TABLE" , table_attr, HTML_TABLE_ATTRIBUTES, SGML_MIXED },
408: { "TD" , td_attr, HTML_TABLE_ATTRIBUTES, SGML_MIXED },
409: { "TEXTAREA", textarea_attr,HTML_TEXTAREA_ATTRIBUTES,SGML_MIXED},
410: { "TH" , td_attr, HTML_TD_ATTRIBUTES, SGML_MIXED },
2.4 frystyk 411: { "TITLE" , gen_attr, HTML_GEN_ATTRIBUTES, SGML_MIXED },
2.1 timbl 412: { "TR" , id_attr, HTML_ID_ATTRIBUTE, SGML_MIXED },
413: { "TT" , gen_attr, HTML_GEN_ATTRIBUTES, SGML_MIXED },
414: { "U" , gen_attr, HTML_GEN_ATTRIBUTES, SGML_MIXED },
415: { "UL" , ul_attr, HTML_UL_ATTRIBUTES, SGML_MIXED },
416: { "VAR" , gen_attr, HTML_GEN_ATTRIBUTES, SGML_MIXED },
417: { "XMP" , no_attr, 0, SGML_LITERAL }
418: };
419:
420:
421: #ifdef OLD_CODE
422: PUBLIC CONST SGML_dtd HTML_dtd = {
423: tags,
424: HTML_ELEMENTS,
425: entities,
426: sizeof(entities)/sizeof(char**)
427: };
428: #endif
429:
430: PUBLIC CONST SGML_dtd HTMLP_dtd = {
431: tags,
432: HTMLP_ELEMENTS,
433: entities,
434: sizeof(entities)/sizeof(char**)
435: };
436:
437: /* Utility Routine: useful for people building HTML objects */
438:
439: /* Start anchor element
440: ** --------------------
441: **
442: ** It is kinda convenient to have a particulr routine for
443: ** starting an anchor element, as everything else for HTML is
444: ** simple anyway.
445: */
446: struct _HTStructured {
447: HTStructuredClass * isa;
448: /* ... */
449: };
450:
451: PUBLIC void HTStartAnchor ARGS3(HTStructured *, obj,
452: CONST char *, name,
453: CONST char *, href)
454: {
455: BOOL present[HTML_A_ATTRIBUTES];
456: CONST char* value[HTML_A_ATTRIBUTES];
457:
458: {
459: int i;
460: for(i=0; i<HTML_A_ATTRIBUTES; i++)
461: present[i] = NO;
462: }
463: if (name) {
464: present[HTML_A_NAME] = YES;
465: value[HTML_A_NAME] = name;
466: }
467: if (href) {
468: present[HTML_A_HREF] = YES;
469: value[HTML_A_HREF] = href;
470: }
471:
472: (*obj->isa->start_element)(obj, HTML_A , present, value);
473:
474: }
475:
2.5 frystyk 476: /* Utility Routine: useful for people building HTML objects */
477:
478: /* Put image element
479: ** --------------------
480: **
481: ** Hopefully as usefull as HTStartAnchor. Henrik 23/03-94
482: */
483: PUBLIC void HTMLPutImg ARGS4(HTStructured *, obj,
484: CONST char *, src,
485: CONST char *, alt,
486: CONST char *, align)
487: {
488: BOOL present[HTML_IMG_ATTRIBUTES];
489: CONST char* value[HTML_IMG_ATTRIBUTES];
490:
491: {
492: int i;
493: for(i=0; i<HTML_IMG_ATTRIBUTES; i++)
494: present[i] = NO;
495: }
496: if (src) {
497: present[HTML_IMG_SRC] = YES;
498: value[HTML_IMG_SRC] = src;
499: }
500: if (alt) {
501: present[HTML_IMG_ALT] = YES;
502: value[HTML_IMG_ALT] = alt;
503: }
504: if (align) {
505: present[HTML_IMG_ALIGN] = YES;
506: value[HTML_IMG_ALIGN] = align;
507: }
508: (*obj->isa->start_element)(obj, HTML_IMG , present, value);
2.7 frystyk 509: /* (*obj->isa->end_element)(obj, HTML_IMG); */
2.5 frystyk 510: }
511:
512:
2.1 timbl 513: PUBLIC void HTNextID ARGS2(HTStructured *, obj,
2.9 timbl 514: CONST char *, next_one)
2.1 timbl 515: {
516: BOOL present[HTML_NEXTID_ATTRIBUTES];
517: CONST char* value[HTML_NEXTID_ATTRIBUTES];
518: {
519: int i;
520: for(i=0; i<HTML_NEXTID_ATTRIBUTES; i++)
521: present[i] = NO;
522: }
523: present[HTML_NEXTID_N] = YES;
2.9 timbl 524: value[HTML_NEXTID_N] = next_one;
2.1 timbl 525:
526: (*obj->isa->start_element)(obj, HTML_NEXTID , present, value);
527:
528: }
2.5 frystyk 529:
530:
2.1 timbl 531:
Webmaster