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 	IFbrksp =			0x40000000,	// add 1 line space to break (IFbrk set too)
264 	IFnobrk =			0x20000000,	// break not allowed before this item
265 	IFcleft =			0x10000000,	// clear left floats (IFbrk set too)
266 	IFcright =			0x08000000,	// clear right floats (IFbrk set too)
267 	IFwrap =			0x04000000,	// in a wrapping (non-pre) line
268 	IFhang =			0x02000000,	// in a hanging (into left indent) item
269 	IFrjust =			0x01000000,	// right justify current line
270 	IFcjust =			0x00800000,	// center justify current line
271 	IFsmap =			0x00400000,	// image is server-side map
272 	IFindentshift =		8,
273 	IFindentmask =		(255<<IFindentshift),	// current indent, in tab stops
274 	IFhangmask =		255			// current hang into left indent, in 1/10th tabstops
275 };
276 
277 // Bias added to Itext's voff field
278 enum { Voffbias = 128 };
279 
280 // Spacer kinds
281 enum {
282 	ISPnull,			// 0 height and width
283 	ISPvline,			// height and ascent of current font
284 	ISPhspace,		// width of space in current font
285 	ISPgeneral		// other purposes (e.g., between markers and list)
286 };
287 
288 // Generic attributes and events (not many elements will have any of these set)
289 struct Genattr
290 {
291 	Rune*	id;
292 	Rune*	class;
293 	Rune*	style;
294 	Rune*	title;
295 	SEvent*	events;
296 };
297 
298 struct SEvent
299 {
300 	SEvent*	next;		// in list of events
301 	int		type;		// SEonblur, etc.
302 	Rune*	script;
303 };
304 
305 enum {
306 	SEonblur, SEonchange, SEonclick, SEondblclick,
307 	SEonfocus, SEonkeypress, SEonkeyup, SEonload,
308 	SEonmousedown, SEonmousemove, SEonmouseout,
309 	SEonmouseover, SEonmouseup, SEonreset, SEonselect,
310 	SEonsubmit, SEonunload,
311 	Numscriptev
312 };
313 
314 // Form field types
315 enum {
316 	Ftext,
317 	Fpassword,
318 	Fcheckbox,
319 	Fradio,
320 	Fsubmit,
321 	Fhidden,
322 	Fimage,
323 	Freset,
324 	Ffile,
325 	Fbutton,
326 	Fselect,
327 	Ftextarea
328 };
329 
330 // Information about a field in a form
331 struct Formfield
332 {
333 	Formfield*	next;		// in list of fields for a form
334 	int			ftype;	// Ftext, Fpassword, etc.
335 	int			fieldid;	// serial no. of field within its form
336 	Form*		form;	// containing form
337 	Rune*		name;	// name attr
338 	Rune*		value;	// value attr
339 	int			size;		// size attr
340 	int			maxlength;	// maxlength attr
341 	int			rows;	// rows attr
342 	int			cols;		// cols attr
343 	uchar		flags;	// FFchecked, etc.
344 	Option*		options;	// for Fselect fields
345 	Item*		image;	// image item, for Fimage fields
346 	int			ctlid;		// identifies control for this field in layout
347 	SEvent*		events;	// same as genattr->events of containing item
348 };
349 
350 enum {
351 	FFchecked =	(1<<7),
352 	FFmultiple =	(1<<6)
353 };
354 
355 // Option holds info about an option in a "select" form field
356 struct Option
357 {
358 	Option*	next;			// next in list of options for a field
359 	int		selected;		// true if selected initially
360 	Rune*	value;		// value attr
361 	Rune*	display;		// display string
362 };
363 
364 // Form holds info about a form
365 struct Form
366 {
367 	Form*		next;		// in list of forms for document
368 	int			formid;	// serial no. of form within its doc
369 	Rune*		name;	// name or id attr (netscape uses name, HTML 4.0 uses id)
370 	Rune*		action;	// action attr
371 	int			target;	// target attr as targetid
372 	int			method;	// HGet or HPost
373 	int			nfields;	// number of fields
374 	Formfield*	fields;	// field's forms, in input order
375 };
376 
377 // Flags used in various table structures
378 enum {
379 	TFparsing =	(1<<7),
380 	TFnowrap =	(1<<6),
381 	TFisth =		(1<<5)
382 };
383 
384 
385 // Information about a table
386 struct Table
387 {
388 	Table*		next;			// next in list of document's tables
389 	int			tableid;		// serial no. of table within its doc
390 	Tablerow*	rows;		// array of row specs (list during parsing)
391 	int			nrow;		// total number of rows
392 	Tablecol*		cols;			// array of column specs
393 	int			ncol;			// total number of columns
394 	Tablecell*		cells;			// list of unique cells
395 	int			ncell;		// total number of cells
396 	Tablecell***	grid;			// 2-D array of cells
397 	Align		align;		// alignment spec for whole table
398 	Dimen		width;		// width spec for whole table
399 	int			border;		// border attr
400 	int			cellspacing;	// cellspacing attr
401 	int			cellpadding;	// cellpadding attr
402 	Background	background;	// table background
403 	Item*		caption;		// linked list of Items, giving caption
404 	uchar		caption_place;	// ALtop or ALbottom
405 	Lay*			caption_lay;	// layout of caption
406 	int			totw;			// total width
407 	int			toth;			// total height
408 	int			caph;		// caption height
409 	int			availw;		// used for previous 3 sizes
410 	Token*		tabletok;		// token that started the table
411 	uchar		flags;		// Lchanged, perhaps
412 };
413 
414 
415 struct Tablecol
416 {
417 	int		width;
418 	Align	align;
419 	Point		pos;
420 };
421 
422 
423 struct Tablerow
424 {
425 	Tablerow*	next;			// Next in list of rows, during parsing
426 	Tablecell*		cells;			// Cells in row, linked through nextinrow
427 	int			height;
428 	int			ascent;
429 	Align		align;
430 	Background	background;
431 	Point			pos;
432 	uchar		flags;		// 0 or TFparsing
433 };
434 
435 
436 // A Tablecell is one cell of a table.
437 // It may span multiple rows and multiple columns.
438 // Cells are linked on two lists: the list for all the cells of
439 // a document (the next pointers), and the list of all the
440 // cells that start in a given row (the nextinrow pointers)
441 struct Tablecell
442 {
443 	Tablecell*		next;			// next in list of table's cells
444 	Tablecell*		nextinrow;	// next in list of row's cells
445 	int			cellid;		// serial no. of cell within table
446 	Item*		content;		// contents before layout
447 	Lay*			lay;			// layout of cell
448 	int			rowspan;		// number of rows spanned by this cell
449 	int			colspan;		// number of cols spanned by this cell
450 	Align		align;		// alignment spec
451 	uchar		flags;		// TFparsing, TFnowrap, TFisth
452 	Dimen		wspec;		// suggested width
453 	int			hspec;		// suggested height
454 	Background	background;	// cell background
455 	int			minw;		// minimum possible width
456 	int			maxw;		// maximum width
457 	int			ascent;		// cell's ascent
458 	int			row;			// row of upper left corner
459 	int			col;			// col of upper left corner
460 	Point			pos;			// nw corner of cell contents, in cell
461 };
462 
463 // Anchor is for info about hyperlinks that go somewhere
464 struct Anchor
465 {
466 	Anchor*		next;		// next in list of document's anchors
467 	int			index;	// serial no. of anchor within its doc
468 	Rune*		name;	// name attr
469 	Rune*		href;		// href attr
470 	int			target;	// target attr as targetid
471 };
472 
473 
474 // DestAnchor is for info about hyperlinks that are destinations
475 struct DestAnchor
476 {
477 	DestAnchor*	next;		// next in list of document's destanchors
478 	int			index;	// serial no. of anchor within its doc
479 	Rune*		name;	// name attr
480 	Item*		item;		// the destination
481 };
482 
483 
484 // Maps (client side)
485 struct Map
486 {
487 	Map*	next;			// next in list of document's maps
488 	Rune*	name;		// map name
489 	Area*	areas;		// list of map areas
490 };
491 
492 
493 struct Area
494 {
495 	Area*		next;		// next in list of a map's areas
496 	int			shape;	// SHrect, etc.
497 	Rune*		href;		// associated hypertext link
498 	int			target;	// associated target frame
499 	Dimen*		coords;	// array of coords for shape
500 	int			ncoords;	// size of coords array
501 };
502 
503 // Area shapes
504 enum {
505 	SHrect, SHcircle, SHpoly
506 };
507 
508 // Fonts are represented by integers: style*NumSize + size
509 
510 // Font styles
511 enum {
512 	FntR,			// roman
513 	FntI,			// italic
514 	FntB,			// bold
515 	FntT,			// typewriter
516 	NumStyle
517 };
518 
519 // Font sizes
520 enum {
521 	Tiny,
522 	Small,
523 	Normal,
524 	Large,
525 	Verylarge,
526 	NumSize
527 };
528 
529 enum {
530 	NumFnt = (NumStyle*NumSize),
531 	DefFnt = (FntR*NumSize+Normal)
532 };
533 
534 // Lines are needed through some text items, for underlining or strikethrough
535 enum {
536 	ULnone, ULunder, ULmid
537 };
538 
539 // Kidinfo flags
540 enum {
541 	FRnoresize =	(1<<0),
542 	FRnoscroll =	(1<<1),
543 	FRhscroll = 	(1<<2),
544 	FRvscroll =	(1<<3),
545 	FRhscrollauto = (1<<4),
546 	FRvscrollauto =	(1<<5)
547 };
548 
549 // Information about child frame or frameset
550 struct Kidinfo
551 {
552 	Kidinfo*		next;		// in list of kidinfos for a frameset
553 	int			isframeset;
554 
555 	// fields for "frame"
556 	Rune*		src;		// only nil if a "dummy" frame or this is frameset
557 	Rune*		name;	// always non-empty if this isn't frameset
558 	int			marginw;
559 	int			marginh;
560 	int			framebd;
561 	int			flags;
562 
563 	// fields for "frameset"
564 	Dimen*		rows;	// array of row dimensions
565 	int			nrows;	// length of rows
566 	Dimen*		cols;		// array of col dimensions
567 	int			ncols;	// length of cols
568 	Kidinfo*		kidinfos;
569 	Kidinfo*		nextframeset;	// parsing stack
570 };
571 
572 
573 // Document info (global information about HTML page)
574 struct Docinfo
575 {
576 	// stuff from HTTP headers, doc head, and body tag
577 	Rune*		src;				// original source of doc
578 	Rune*		base;			// base URL of doc
579 	Rune*		doctitle;			// from <title> element
580 	Background	background;		// background specification
581 	Iimage*		backgrounditem;	// Image Item for doc background image, or nil
582 	int			text;				// doc foreground (text) color
583 	int			link;				// unvisited hyperlink color
584 	int			vlink;			// visited hyperlink color
585 	int			alink;			// highlighting hyperlink color
586 	int			target;			// target frame default
587 	int			chset;			// ISO_8859, etc.
588 	int			mediatype;		// TextHtml, etc.
589 	int			scripttype;		// TextJavascript, etc.
590 	int			hasscripts;		// true if scripts used
591 	Rune*		refresh;			// content of <http-equiv=Refresh ...>
592 	Kidinfo*		kidinfo;			// if a frameset
593 	int			frameid;			// id of document frame
594 
595 	// info needed to respond to user actions
596 	Anchor*		anchors;			// list of href anchors
597 	DestAnchor*	dests;			// list of destination anchors
598 	Form*		forms;			// list of forms
599 	Table*		tables;			// list of tables
600 	Map*		maps;			// list of maps
601 	Iimage*		images;			// list of image items (through nextimage links)
602 };
603 
604 extern int			dimenkind(Dimen d);
605 extern int			dimenspec(Dimen d);
606 extern void		freedocinfo(Docinfo* d);
607 extern void		freeitems(Item* ithead);
608 extern Item*		parsehtml(uchar* data, int datalen, Rune* src, int mtype, int chset, Docinfo** pdi);
609 extern void		printitems(Item* items, char* msg);
610 extern int			targetid(Rune* s);
611 extern Rune*		targetname(int targid);
612 extern int			validitems(Item* i);
613 
614 #pragma varargck	type "I"	Item*
615 
616 // Control print output
617 extern int			warn;
618 extern int			dbglex;
619 extern int			dbgbuild;
620 
621 // To be provided by caller
622 // emalloc and erealloc should not return if can't get memory.
623 // emalloc should zero its memory.
624 extern void*	emalloc(ulong);
625 extern void*	erealloc(void* p, ulong size);
626 #ifdef __cpluspplus
627 }
628 #endif
629 #endif