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