1 #ifndef _HTML_H_
2 #define _HTML_H_ 1
3 #ifdef __cplusplus
4 extern "C" {
5 #endif
6 
7 AUTOLIB(html)
8 /*
9  #pragma lib "libhtml.a"
10  #pragma src "/sys/src/libhtml"
11 */
12 
13 // UTILS
14 extern uchar*	fromStr(Rune* buf, int n, int chset);
15 extern Rune*	toStr(uchar* buf, int n, int chset);
16 
17 // Common LEX and BUILD enums
18 
19 // Media types
20 enum
21 {
22 	ApplMsword,
23 	ApplOctets,
24 	ApplPdf,
25 	ApplPostscript,
26 	ApplRtf,
27 	ApplFramemaker,
28 	ApplMsexcel,
29 	ApplMspowerpoint,
30 	UnknownType,
31 	Audio32kadpcm,
32 	AudioBasic,
33 	ImageCgm,
34 	ImageG3fax,
35 	ImageGif,
36 	ImageIef,
37 	ImageJpeg,
38 	ImagePng,
39 	ImageTiff,
40 	ImageXBit,
41 	ImageXBit2,
42 	ImageXBitmulti,
43 	ImageXXBitmap,
44 	ModelVrml,
45 	MultiDigest,
46 	MultiMixed,
47 	TextCss,
48 	TextEnriched,
49 	TextHtml,
50 	TextJavascript,
51 	TextPlain,
52 	TextRichtext,
53 	TextSgml,
54 	TextTabSeparatedValues,
55 	TextXml,
56 	VideoMpeg,
57 	VideoQuicktime,
58 	NMEDIATYPES
59 };
60 
61 // HTTP methods
62 enum
63 {
64 	HGet,
65 	HPost
66 };
67 
68 // Charsets
69 enum
70 {
71 	UnknownCharset,
72 	US_Ascii,
73 	ISO_8859_1,
74 	UTF_8,
75 	Unicode,
76 	NCHARSETS
77 };
78 
79 // Frame Target IDs
80 enum {
81 	FTtop,
82 	FTself,
83 	FTparent,
84 	FTblank
85 };
86 
87 // LEX
88 typedef struct Token Token;
89 typedef struct Attr Attr;
90 
91 // BUILD
92 
93 typedef struct Item Item;
94 typedef struct Itext Itext;
95 typedef struct Irule Irule;
96 typedef struct Iimage Iimage;
97 typedef struct Iformfield Iformfield;
98 typedef struct Itable Itable;
99 typedef struct Ifloat Ifloat;
100 typedef struct Ispacer Ispacer;
101 typedef struct Genattr Genattr;
102 typedef struct SEvent SEvent;
103 typedef struct Formfield Formfield;
104 typedef struct Option Option;
105 typedef struct Form Form;
106 typedef struct Table Table;
107 typedef struct Tablecol Tablecol;
108 typedef struct Tablerow Tablerow;
109 typedef struct Tablecell Tablecell;
110 typedef struct Align Align;
111 typedef struct Dimen Dimen;
112 typedef struct Anchor Anchor;
113 typedef struct DestAnchor DestAnchor;
114 typedef struct Map Map;
115 typedef struct Area Area;
116 typedef struct Background Background;
117 typedef struct Kidinfo Kidinfo;
118 typedef struct Docinfo Docinfo;
119 typedef struct Stack Stack;
120 typedef struct Pstate Pstate;
121 typedef struct ItemSource ItemSource;
122 typedef struct Lay Lay;	// defined in Layout module
123 
124 // Alignment types
125 enum {
126 	ALnone = 0, ALleft, ALcenter, ALright, ALjustify,
127 	ALchar, ALtop, ALmiddle, ALbottom, ALbaseline
128 };
129 
130 struct Align
131 {
132 	uchar	halign;	// one of ALnone, ALleft, etc.
133 	uchar	valign;	// one of ALnone, ALtop, etc.
134 };
135 
136 // A Dimen holds a dimension specification, especially for those
137 // cases when a number can be followed by a % or a * to indicate
138 // percentage of total or relative weight.
139 // Dnone means no dimension was specified
140 
141 // To fit in a word, use top bits to identify kind, rest for value
142 enum {
143 	Dnone =		0,
144 	Dpixels =		(1<<29),
145 	Dpercent =	(2<<29),
146 	Drelative =	(3<<29),
147 	Dkindmask =	(3<<29),
148 	Dspecmask =	(~Dkindmask)
149 };
150 
151 struct Dimen
152 {
153 	int	kindspec;		// kind | spec
154 };
155 
156 // Background is either an image or a color.
157 // If both are set, the image has precedence.
158 struct Background
159 {
160 	Rune*	image;	// url
161 	int		color;
162 };
163 
164 
165 // There are about a half dozen Item variants.
166 // The all look like this at the start (using Plan 9 C's
167 // anonymous structure member mechanism),
168 // and then the tag field dictates what extra fields there are.
169 struct Item
170 {
171 	Item*	next;		// successor in list of items
172 	int		width;	// width in pixels (0 for floating items)
173 	int		height;	// height in pixels
174 	Rectangle	r;
175 	int		ascent;	// ascent (from top to baseline) in pixels
176 	int		anchorid;	// if nonzero, which anchor we're in
177 	int		state;	// flags and values (see below)
178 	Genattr*	genattr;	// generic attributes and events
179 	int		tag;		// variant discriminator: Itexttag, etc.
180 };
181 
182 // Item variant tags
183 enum {
184 	Itexttag,
185 	Iruletag,
186 	Iimagetag,
187 	Iformfieldtag,
188 	Itabletag,
189 	Ifloattag,
190 	Ispacertag
191 };
192 
193 struct Itext
194 {
195 	Item item;				// (with tag ==Itexttag)
196 	Rune*	s;			// the characters
197 	int		fnt;			// style*NumSize+size (see font stuff, below)
198 	int		fg;			// Pixel (color) for text
199 	uchar	voff;			// Voffbias+vertical offset from baseline, in pixels (+ve == down)
200 	uchar	ul;			// ULnone, ULunder, or ULmid
201 };
202 
203 struct Irule
204 {
205 	Item item;				// (with tag ==Iruletag)
206 	uchar	align;		// alignment spec
207 	uchar	noshade;		// if true, don't shade
208 	int		size;			// size attr (rule height)
209 	Dimen	wspec;		// width spec
210 };
211 
212 
213 struct Iimage
214 {
215 	Item item;				// (with tag ==Iimagetag)
216 	Rune*	imsrc;		// image src url
217 	int		imwidth;		// spec width (actual, if no spec)
218 	int		imheight;		// spec height (actual, if no spec)
219 	Rune*	altrep;		// alternate representation, in absence of image
220 	Map*	map;			// if non-nil, client side map
221 	int		ctlid;			// if animated
222 	uchar	align;		// vertical alignment
223 	uchar	hspace;		// in pixels; buffer space on each side
224 	uchar	vspace;		// in pixels; buffer space on top and bottom
225 	uchar	border;		// in pixels: border width to draw around image
226 	Iimage*	nextimage;	// next in list of document's images
227 	void *aux;
228 };
229 
230 
231 struct Iformfield
232 {
233 	Item item;				// (with tag ==Iformfieldtag)
234 	Formfield*	formfield;
235 	void *aux;
236 };
237 
238 
239 struct Itable
240 {
241 	Item item;				// (with tag ==Itabletag)
242 	Table*	table;
243 };
244 
245 
246 struct Ifloat
247 {
248 	Item _item;				// (with tag ==Ifloattag)
249 	Item*	item;			// table or image item that floats
250 	int		x;			// x coord of top (from right, if ALright)
251 	int		y;			// y coord of top
252 	uchar	side;			// margin it floats to: ALleft or ALright
253 	uchar	infloats;		// true if this has been added to a lay.floats
254 	Ifloat*	nextfloat;		// in list of floats
255 };
256 
257 
258 struct Ispacer
259 {
260 	Item item;				// (with tag ==Ispacertag)
261 	int		spkind;		// ISPnull, etc.
262 };
263 
264 // Item state flags and value fields
265 enum {
266 //	IFbrk =			0x80000000,	// forced break before this item
267 #define	IFbrk		0x80000000 /* too big for sun */
268 	IFbrksp =			0x40000000,	// add 1 line space to break (IFbrk set too)
269 	IFnobrk =			0x20000000,	// break not allowed before this item
270 	IFcleft =			0x10000000,	// clear left floats (IFbrk set too)
271 	IFcright =			0x08000000,	// clear right floats (IFbrk set too)
272 	IFwrap =			0x04000000,	// in a wrapping (non-pre) line
273 	IFhang =			0x02000000,	// in a hanging (into left indent) item
274 	IFrjust =			0x01000000,	// right justify current line
275 	IFcjust =			0x00800000,	// center justify current line
276 	IFsmap =			0x00400000,	// image is server-side map
277 	IFindentshift =		8,
278 	IFindentmask =		(255<<IFindentshift),	// current indent, in tab stops
279 	IFhangmask =		255			// current hang into left indent, in 1/10th tabstops
280 };
281 
282 // Bias added to Itext's voff field
283 enum { Voffbias = 128 };
284 
285 // Spacer kinds
286 enum {
287 	ISPnull,			// 0 height and width
288 	ISPvline,			// height and ascent of current font
289 	ISPhspace,		// width of space in current font
290 	ISPgeneral		// other purposes (e.g., between markers and list)
291 };
292 
293 // Generic attributes and events (not many elements will have any of these set)
294 struct Genattr
295 {
296 	Rune*	id;
297 	Rune*	class;
298 	Rune*	style;
299 	Rune*	title;
300 	SEvent*	events;
301 };
302 
303 struct SEvent
304 {
305 	SEvent*	next;		// in list of events
306 	int		type;		// SEonblur, etc.
307 	Rune*	script;
308 };
309 
310 enum {
311 	SEonblur, SEonchange, SEonclick, SEondblclick,
312 	SEonfocus, SEonkeypress, SEonkeyup, SEonload,
313 	SEonmousedown, SEonmousemove, SEonmouseout,
314 	SEonmouseover, SEonmouseup, SEonreset, SEonselect,
315 	SEonsubmit, SEonunload,
316 	Numscriptev
317 };
318 
319 // Form field types
320 enum {
321 	Ftext,
322 	Fpassword,
323 	Fcheckbox,
324 	Fradio,
325 	Fsubmit,
326 	Fhidden,
327 	Fimage,
328 	Freset,
329 	Ffile,
330 	Fbutton,
331 	Fselect,
332 	Ftextarea
333 };
334 
335 // Information about a field in a form
336 struct Formfield
337 {
338 	Formfield*	next;		// in list of fields for a form
339 	int			ftype;	// Ftext, Fpassword, etc.
340 	int			fieldid;	// serial no. of field within its form
341 	Form*		form;	// containing form
342 	Rune*		name;	// name attr
343 	Rune*		value;	// value attr
344 	int			size;		// size attr
345 	int			maxlength;	// maxlength attr
346 	int			rows;	// rows attr
347 	int			cols;		// cols attr
348 	uchar		flags;	// FFchecked, etc.
349 	Option*		options;	// for Fselect fields
350 	Item*		image;	// image item, for Fimage fields
351 	int			ctlid;		// identifies control for this field in layout
352 	SEvent*		events;	// same as genattr->events of containing item
353 };
354 
355 enum {
356 	FFchecked =	(1<<7),
357 	FFmultiple =	(1<<6)
358 };
359 
360 // Option holds info about an option in a "select" form field
361 struct Option
362 {
363 	Option*	next;			// next in list of options for a field
364 	int		selected;		// true if selected initially
365 	Rune*	value;		// value attr
366 	Rune*	display;		// display string
367 };
368 
369 // Form holds info about a form
370 struct Form
371 {
372 	Form*		next;		// in list of forms for document
373 	int			formid;	// serial no. of form within its doc
374 	Rune*		name;	// name or id attr (netscape uses name, HTML 4.0 uses id)
375 	Rune*		action;	// action attr
376 	int			target;	// target attr as targetid
377 	int			method;	// HGet or HPost
378 	int			nfields;	// number of fields
379 	Formfield*	fields;	// field's forms, in input order
380 };
381 
382 // Flags used in various table structures
383 enum {
384 	TFparsing =	(1<<7),
385 	TFnowrap =	(1<<6),
386 	TFisth =		(1<<5)
387 };
388 
389 
390 // Information about a table
391 struct Table
392 {
393 	Table*		next;			// next in list of document's tables
394 	int			tableid;		// serial no. of table within its doc
395 	Tablerow*	rows;		// array of row specs (list during parsing)
396 	int			nrow;		// total number of rows
397 	Tablecol*		cols;			// array of column specs
398 	int			ncol;			// total number of columns
399 	Tablecell*		cells;			// list of unique cells
400 	int			ncell;		// total number of cells
401 	Tablecell***	grid;			// 2-D array of cells
402 	Align		align;		// alignment spec for whole table
403 	Dimen		width;		// width spec for whole table
404 	int			border;		// border attr
405 	int			cellspacing;	// cellspacing attr
406 	int			cellpadding;	// cellpadding attr
407 	Background	background;	// table background
408 	Item*		caption;		// linked list of Items, giving caption
409 	uchar		caption_place;	// ALtop or ALbottom
410 	Lay*			caption_lay;	// layout of caption
411 	int			totw;			// total width
412 	int			toth;			// total height
413 	int			caph;		// caption height
414 	int			availw;		// used for previous 3 sizes
415 	Token*		tabletok;		// token that started the table
416 	uchar		flags;		// Lchanged, perhaps
417 };
418 
419 
420 struct Tablecol
421 {
422 	int		width;
423 	Align	align;
424 	Point		pos;
425 };
426 
427 
428 struct Tablerow
429 {
430 	Tablerow*	next;			// Next in list of rows, during parsing
431 	Tablecell*		cells;			// Cells in row, linked through nextinrow
432 	int			height;
433 	int			ascent;
434 	Align		align;
435 	Background	background;
436 	Point			pos;
437 	uchar		flags;		// 0 or TFparsing
438 };
439 
440 
441 // A Tablecell is one cell of a table.
442 // It may span multiple rows and multiple columns.
443 // Cells are linked on two lists: the list for all the cells of
444 // a document (the next pointers), and the list of all the
445 // cells that start in a given row (the nextinrow pointers)
446 struct Tablecell
447 {
448 	Tablecell*		next;			// next in list of table's cells
449 	Tablecell*		nextinrow;	// next in list of row's cells
450 	int			cellid;		// serial no. of cell within table
451 	Item*		content;		// contents before layout
452 	Lay*			lay;			// layout of cell
453 	int			rowspan;		// number of rows spanned by this cell
454 	int			colspan;		// number of cols spanned by this cell
455 	Align		align;		// alignment spec
456 	uchar		flags;		// TFparsing, TFnowrap, TFisth
457 	Dimen		wspec;		// suggested width
458 	int			hspec;		// suggested height
459 	Background	background;	// cell background
460 	int			minw;		// minimum possible width
461 	int			maxw;		// maximum width
462 	int			ascent;		// cell's ascent
463 	int			row;			// row of upper left corner
464 	int			col;			// col of upper left corner
465 	Point			pos;			// nw corner of cell contents, in cell
466 };
467 
468 // Anchor is for info about hyperlinks that go somewhere
469 struct Anchor
470 {
471 	Anchor*		next;		// next in list of document's anchors
472 	int			index;	// serial no. of anchor within its doc
473 	Rune*		name;	// name attr
474 	Rune*		href;		// href attr
475 	int			target;	// target attr as targetid
476 };
477 
478 
479 // DestAnchor is for info about hyperlinks that are destinations
480 struct DestAnchor
481 {
482 	DestAnchor*	next;		// next in list of document's destanchors
483 	int			index;	// serial no. of anchor within its doc
484 	Rune*		name;	// name attr
485 	Item*		item;		// the destination
486 };
487 
488 
489 // Maps (client side)
490 struct Map
491 {
492 	Map*	next;			// next in list of document's maps
493 	Rune*	name;		// map name
494 	Area*	areas;		// list of map areas
495 };
496 
497 
498 struct Area
499 {
500 	Area*		next;		// next in list of a map's areas
501 	int			shape;	// SHrect, etc.
502 	Rune*		href;		// associated hypertext link
503 	int			target;	// associated target frame
504 	Dimen*		coords;	// array of coords for shape
505 	int			ncoords;	// size of coords array
506 };
507 
508 // Area shapes
509 enum {
510 	SHrect, SHcircle, SHpoly
511 };
512 
513 // Fonts are represented by integers: style*NumSize + size
514 
515 // Font styles
516 enum {
517 	FntR,			// roman
518 	FntI,			// italic
519 	FntB,			// bold
520 	FntT,			// typewriter
521 	NumStyle
522 };
523 
524 // Font sizes
525 enum {
526 	Tiny,
527 	Small,
528 	Normal,
529 	Large,
530 	Verylarge,
531 	NumSize
532 };
533 
534 enum {
535 	NumFnt = (NumStyle*NumSize),
536 	DefFnt = (FntR*NumSize+Normal)
537 };
538 
539 // Lines are needed through some text items, for underlining or strikethrough
540 enum {
541 	ULnone, ULunder, ULmid
542 };
543 
544 // Kidinfo flags
545 enum {
546 	FRnoresize =	(1<<0),
547 	FRnoscroll =	(1<<1),
548 	FRhscroll = 	(1<<2),
549 	FRvscroll =	(1<<3),
550 	FRhscrollauto = (1<<4),
551 	FRvscrollauto =	(1<<5)
552 };
553 
554 // Information about child frame or frameset
555 struct Kidinfo
556 {
557 	Kidinfo*		next;		// in list of kidinfos for a frameset
558 	int			isframeset;
559 
560 	// fields for "frame"
561 	Rune*		src;		// only nil if a "dummy" frame or this is frameset
562 	Rune*		name;	// always non-empty if this isn't frameset
563 	int			marginw;
564 	int			marginh;
565 	int			framebd;
566 	int			flags;
567 
568 	// fields for "frameset"
569 	Dimen*		rows;	// array of row dimensions
570 	int			nrows;	// length of rows
571 	Dimen*		cols;		// array of col dimensions
572 	int			ncols;	// length of cols
573 	Kidinfo*		kidinfos;
574 	Kidinfo*		nextframeset;	// parsing stack
575 };
576 
577 
578 // Document info (global information about HTML page)
579 struct Docinfo
580 {
581 	// stuff from HTTP headers, doc head, and body tag
582 	Rune*		src;				// original source of doc
583 	Rune*		base;			// base URL of doc
584 	Rune*		doctitle;			// from <title> element
585 	Background	background;		// background specification
586 	Iimage*		backgrounditem;	// Image Item for doc background image, or nil
587 	int			text;				// doc foreground (text) color
588 	int			link;				// unvisited hyperlink color
589 	int			vlink;			// visited hyperlink color
590 	int			alink;			// highlighting hyperlink color
591 	int			target;			// target frame default
592 	int			chset;			// ISO_8859, etc.
593 	int			mediatype;		// TextHtml, etc.
594 	int			scripttype;		// TextJavascript, etc.
595 	int			hasscripts;		// true if scripts used
596 	Rune*		refresh;			// content of <http-equiv=Refresh ...>
597 	Kidinfo*		kidinfo;			// if a frameset
598 	int			frameid;			// id of document frame
599 
600 	// info needed to respond to user actions
601 	Anchor*		anchors;			// list of href anchors
602 	DestAnchor*	dests;			// list of destination anchors
603 	Form*		forms;			// list of forms
604 	Table*		tables;			// list of tables
605 	Map*		maps;			// list of maps
606 	Iimage*		images;			// list of image items (through nextimage links)
607 };
608 
609 extern int			dimenkind(Dimen d);
610 extern int			dimenspec(Dimen d);
611 extern void		freedocinfo(Docinfo* d);
612 extern void		freeitems(Item* ithead);
613 extern Item*		parsehtml(uchar* data, int datalen, Rune* src, int mtype, int chset, Docinfo** pdi);
614 extern void		printitems(Item* items, char* msg);
615 extern int			targetid(Rune* s);
616 extern Rune*		targetname(int targid);
617 extern int			validitems(Item* i);
618 
619 /* #pragma varargck	type "I"	Item* */
620 
621 // Control print output
622 extern int			warn;
623 extern int			dbglex;
624 extern int			dbgbuild;
625 
626 // To be provided by caller
627 // emalloc and erealloc should not return if can't get memory.
628 // emalloc should zero its memory.
629 extern void*	emalloc(ulong);
630 extern void*	erealloc(void* p, ulong size);
631 #ifdef __cpluspplus
632 }
633 #endif
634 #endif