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 	int		ascent;	// ascent (from top to baseline) in pixels
175 	int		anchorid;	// if nonzero, which anchor we're in
176 	int		state;	// flags and values (see below)
177 	Genattr*	genattr;	// generic attributes and events
178 	int		tag;		// variant discriminator: Itexttag, etc.
179 };
180 
181 // Item variant tags
182 enum {
183 	Itexttag,
184 	Iruletag,
185 	Iimagetag,
186 	Iformfieldtag,
187 	Itabletag,
188 	Ifloattag,
189 	Ispacertag
190 };
191 
192 struct Itext
193 {
194 	Item item;				// (with tag ==Itexttag)
195 	Rune*	s;			// the characters
196 	int		fnt;			// style*NumSize+size (see font stuff, below)
197 	int		fg;			// Pixel (color) for text
198 	uchar	voff;			// Voffbias+vertical offset from baseline, in pixels (+ve == down)
199 	uchar	ul;			// ULnone, ULunder, or ULmid
200 };
201 
202 struct Irule
203 {
204 	Item item;				// (with tag ==Iruletag)
205 	uchar	align;		// alignment spec
206 	uchar	noshade;		// if true, don't shade
207 	int		size;			// size attr (rule height)
208 	Dimen	wspec;		// width spec
209 };
210 
211 
212 struct Iimage
213 {
214 	Item item;				// (with tag ==Iimagetag)
215 	Rune*	imsrc;		// image src url
216 	int		imwidth;		// spec width (actual, if no spec)
217 	int		imheight;		// spec height (actual, if no spec)
218 	Rune*	altrep;		// alternate representation, in absence of image
219 	Map*	map;			// if non-nil, client side map
220 	int		ctlid;			// if animated
221 	uchar	align;		// vertical alignment
222 	uchar	hspace;		// in pixels; buffer space on each side
223 	uchar	vspace;		// in pixels; buffer space on top and bottom
224 	uchar	border;		// in pixels: border width to draw around image
225 	Iimage*	nextimage;	// next in list of document's images
226 };
227 
228 
229 struct Iformfield
230 {
231 	Item item;				// (with tag ==Iformfieldtag)
232 	Formfield*	formfield;
233 };
234 
235 
236 struct Itable
237 {
238 	Item item;				// (with tag ==Itabletag)
239 	Table*	table;
240 };
241 
242 
243 struct Ifloat
244 {
245 	Item _item;				// (with tag ==Ifloattag)
246 	Item*	item;			// table or image item that floats
247 	int		x;			// x coord of top (from right, if ALright)
248 	int		y;			// y coord of top
249 	uchar	side;			// margin it floats to: ALleft or ALright
250 	uchar	infloats;		// true if this has been added to a lay.floats
251 	Ifloat*	nextfloat;		// in list of floats
252 };
253 
254 
255 struct Ispacer
256 {
257 	Item item;				// (with tag ==Ispacertag)
258 	int		spkind;		// ISPnull, etc.
259 };
260 
261 // Item state flags and value fields
262 enum {
263 //	IFbrk =			0x80000000,	// forced break before this item
264 #define	IFbrk		0x80000000 /* too big for sun */
265 	IFbrksp =			0x40000000,	// add 1 line space to break (IFbrk set too)
266 	IFnobrk =			0x20000000,	// break not allowed before this item
267 	IFcleft =			0x10000000,	// clear left floats (IFbrk set too)
268 	IFcright =			0x08000000,	// clear right floats (IFbrk set too)
269 	IFwrap =			0x04000000,	// in a wrapping (non-pre) line
270 	IFhang =			0x02000000,	// in a hanging (into left indent) item
271 	IFrjust =			0x01000000,	// right justify current line
272 	IFcjust =			0x00800000,	// center justify current line
273 	IFsmap =			0x00400000,	// image is server-side map
274 	IFindentshift =		8,
275 	IFindentmask =		(255<<IFindentshift),	// current indent, in tab stops
276 	IFhangmask =		255			// current hang into left indent, in 1/10th tabstops
277 };
278 
279 // Bias added to Itext's voff field
280 enum { Voffbias = 128 };
281 
282 // Spacer kinds
283 enum {
284 	ISPnull,			// 0 height and width
285 	ISPvline,			// height and ascent of current font
286 	ISPhspace,		// width of space in current font
287 	ISPgeneral		// other purposes (e.g., between markers and list)
288 };
289 
290 // Generic attributes and events (not many elements will have any of these set)
291 struct Genattr
292 {
293 	Rune*	id;
294 	Rune*	class;
295 	Rune*	style;
296 	Rune*	title;
297 	SEvent*	events;
298 };
299 
300 struct SEvent
301 {
302 	SEvent*	next;		// in list of events
303 	int		type;		// SEonblur, etc.
304 	Rune*	script;
305 };
306 
307 enum {
308 	SEonblur, SEonchange, SEonclick, SEondblclick,
309 	SEonfocus, SEonkeypress, SEonkeyup, SEonload,
310 	SEonmousedown, SEonmousemove, SEonmouseout,
311 	SEonmouseover, SEonmouseup, SEonreset, SEonselect,
312 	SEonsubmit, SEonunload,
313 	Numscriptev
314 };
315 
316 // Form field types
317 enum {
318 	Ftext,
319 	Fpassword,
320 	Fcheckbox,
321 	Fradio,
322 	Fsubmit,
323 	Fhidden,
324 	Fimage,
325 	Freset,
326 	Ffile,
327 	Fbutton,
328 	Fselect,
329 	Ftextarea
330 };
331 
332 // Information about a field in a form
333 struct Formfield
334 {
335 	Formfield*	next;		// in list of fields for a form
336 	int			ftype;	// Ftext, Fpassword, etc.
337 	int			fieldid;	// serial no. of field within its form
338 	Form*		form;	// containing form
339 	Rune*		name;	// name attr
340 	Rune*		value;	// value attr
341 	int			size;		// size attr
342 	int			maxlength;	// maxlength attr
343 	int			rows;	// rows attr
344 	int			cols;		// cols attr
345 	uchar		flags;	// FFchecked, etc.
346 	Option*		options;	// for Fselect fields
347 	Item*		image;	// image item, for Fimage fields
348 	int			ctlid;		// identifies control for this field in layout
349 	SEvent*		events;	// same as genattr->events of containing item
350 };
351 
352 enum {
353 	FFchecked =	(1<<7),
354 	FFmultiple =	(1<<6)
355 };
356 
357 // Option holds info about an option in a "select" form field
358 struct Option
359 {
360 	Option*	next;			// next in list of options for a field
361 	int		selected;		// true if selected initially
362 	Rune*	value;		// value attr
363 	Rune*	display;		// display string
364 };
365 
366 // Form holds info about a form
367 struct Form
368 {
369 	Form*		next;		// in list of forms for document
370 	int			formid;	// serial no. of form within its doc
371 	Rune*		name;	// name or id attr (netscape uses name, HTML 4.0 uses id)
372 	Rune*		action;	// action attr
373 	int			target;	// target attr as targetid
374 	int			method;	// HGet or HPost
375 	int			nfields;	// number of fields
376 	Formfield*	fields;	// field's forms, in input order
377 };
378 
379 // Flags used in various table structures
380 enum {
381 	TFparsing =	(1<<7),
382 	TFnowrap =	(1<<6),
383 	TFisth =		(1<<5)
384 };
385 
386 
387 // Information about a table
388 struct Table
389 {
390 	Table*		next;			// next in list of document's tables
391 	int			tableid;		// serial no. of table within its doc
392 	Tablerow*	rows;		// array of row specs (list during parsing)
393 	int			nrow;		// total number of rows
394 	Tablecol*		cols;			// array of column specs
395 	int			ncol;			// total number of columns
396 	Tablecell*		cells;			// list of unique cells
397 	int			ncell;		// total number of cells
398 	Tablecell***	grid;			// 2-D array of cells
399 	Align		align;		// alignment spec for whole table
400 	Dimen		width;		// width spec for whole table
401 	int			border;		// border attr
402 	int			cellspacing;	// cellspacing attr
403 	int			cellpadding;	// cellpadding attr
404 	Background	background;	// table background
405 	Item*		caption;		// linked list of Items, giving caption
406 	uchar		caption_place;	// ALtop or ALbottom
407 	Lay*			caption_lay;	// layout of caption
408 	int			totw;			// total width
409 	int			toth;			// total height
410 	int			caph;		// caption height
411 	int			availw;		// used for previous 3 sizes
412 	Token*		tabletok;		// token that started the table
413 	uchar		flags;		// Lchanged, perhaps
414 };
415 
416 
417 struct Tablecol
418 {
419 	int		width;
420 	Align	align;
421 	Point		pos;
422 };
423 
424 
425 struct Tablerow
426 {
427 	Tablerow*	next;			// Next in list of rows, during parsing
428 	Tablecell*		cells;			// Cells in row, linked through nextinrow
429 	int			height;
430 	int			ascent;
431 	Align		align;
432 	Background	background;
433 	Point			pos;
434 	uchar		flags;		// 0 or TFparsing
435 };
436 
437 
438 // A Tablecell is one cell of a table.
439 // It may span multiple rows and multiple columns.
440 // Cells are linked on two lists: the list for all the cells of
441 // a document (the next pointers), and the list of all the
442 // cells that start in a given row (the nextinrow pointers)
443 struct Tablecell
444 {
445 	Tablecell*		next;			// next in list of table's cells
446 	Tablecell*		nextinrow;	// next in list of row's cells
447 	int			cellid;		// serial no. of cell within table
448 	Item*		content;		// contents before layout
449 	Lay*			lay;			// layout of cell
450 	int			rowspan;		// number of rows spanned by this cell
451 	int			colspan;		// number of cols spanned by this cell
452 	Align		align;		// alignment spec
453 	uchar		flags;		// TFparsing, TFnowrap, TFisth
454 	Dimen		wspec;		// suggested width
455 	int			hspec;		// suggested height
456 	Background	background;	// cell background
457 	int			minw;		// minimum possible width
458 	int			maxw;		// maximum width
459 	int			ascent;		// cell's ascent
460 	int			row;			// row of upper left corner
461 	int			col;			// col of upper left corner
462 	Point			pos;			// nw corner of cell contents, in cell
463 };
464 
465 // Anchor is for info about hyperlinks that go somewhere
466 struct Anchor
467 {
468 	Anchor*		next;		// next in list of document's anchors
469 	int			index;	// serial no. of anchor within its doc
470 	Rune*		name;	// name attr
471 	Rune*		href;		// href attr
472 	int			target;	// target attr as targetid
473 };
474 
475 
476 // DestAnchor is for info about hyperlinks that are destinations
477 struct DestAnchor
478 {
479 	DestAnchor*	next;		// next in list of document's destanchors
480 	int			index;	// serial no. of anchor within its doc
481 	Rune*		name;	// name attr
482 	Item*		item;		// the destination
483 };
484 
485 
486 // Maps (client side)
487 struct Map
488 {
489 	Map*	next;			// next in list of document's maps
490 	Rune*	name;		// map name
491 	Area*	areas;		// list of map areas
492 };
493 
494 
495 struct Area
496 {
497 	Area*		next;		// next in list of a map's areas
498 	int			shape;	// SHrect, etc.
499 	Rune*		href;		// associated hypertext link
500 	int			target;	// associated target frame
501 	Dimen*		coords;	// array of coords for shape
502 	int			ncoords;	// size of coords array
503 };
504 
505 // Area shapes
506 enum {
507 	SHrect, SHcircle, SHpoly
508 };
509 
510 // Fonts are represented by integers: style*NumSize + size
511 
512 // Font styles
513 enum {
514 	FntR,			// roman
515 	FntI,			// italic
516 	FntB,			// bold
517 	FntT,			// typewriter
518 	NumStyle
519 };
520 
521 // Font sizes
522 enum {
523 	Tiny,
524 	Small,
525 	Normal,
526 	Large,
527 	Verylarge,
528 	NumSize
529 };
530 
531 enum {
532 	NumFnt = (NumStyle*NumSize),
533 	DefFnt = (FntR*NumSize+Normal)
534 };
535 
536 // Lines are needed through some text items, for underlining or strikethrough
537 enum {
538 	ULnone, ULunder, ULmid
539 };
540 
541 // Kidinfo flags
542 enum {
543 	FRnoresize =	(1<<0),
544 	FRnoscroll =	(1<<1),
545 	FRhscroll = 	(1<<2),
546 	FRvscroll =	(1<<3),
547 	FRhscrollauto = (1<<4),
548 	FRvscrollauto =	(1<<5)
549 };
550 
551 // Information about child frame or frameset
552 struct Kidinfo
553 {
554 	Kidinfo*		next;		// in list of kidinfos for a frameset
555 	int			isframeset;
556 
557 	// fields for "frame"
558 	Rune*		src;		// only nil if a "dummy" frame or this is frameset
559 	Rune*		name;	// always non-empty if this isn't frameset
560 	int			marginw;
561 	int			marginh;
562 	int			framebd;
563 	int			flags;
564 
565 	// fields for "frameset"
566 	Dimen*		rows;	// array of row dimensions
567 	int			nrows;	// length of rows
568 	Dimen*		cols;		// array of col dimensions
569 	int			ncols;	// length of cols
570 	Kidinfo*		kidinfos;
571 	Kidinfo*		nextframeset;	// parsing stack
572 };
573 
574 
575 // Document info (global information about HTML page)
576 struct Docinfo
577 {
578 	// stuff from HTTP headers, doc head, and body tag
579 	Rune*		src;				// original source of doc
580 	Rune*		base;			// base URL of doc
581 	Rune*		doctitle;			// from <title> element
582 	Background	background;		// background specification
583 	Iimage*		backgrounditem;	// Image Item for doc background image, or nil
584 	int			text;				// doc foreground (text) color
585 	int			link;				// unvisited hyperlink color
586 	int			vlink;			// visited hyperlink color
587 	int			alink;			// highlighting hyperlink color
588 	int			target;			// target frame default
589 	int			chset;			// ISO_8859, etc.
590 	int			mediatype;		// TextHtml, etc.
591 	int			scripttype;		// TextJavascript, etc.
592 	int			hasscripts;		// true if scripts used
593 	Rune*		refresh;			// content of <http-equiv=Refresh ...>
594 	Kidinfo*		kidinfo;			// if a frameset
595 	int			frameid;			// id of document frame
596 
597 	// info needed to respond to user actions
598 	Anchor*		anchors;			// list of href anchors
599 	DestAnchor*	dests;			// list of destination anchors
600 	Form*		forms;			// list of forms
601 	Table*		tables;			// list of tables
602 	Map*		maps;			// list of maps
603 	Iimage*		images;			// list of image items (through nextimage links)
604 };
605 
606 extern int			dimenkind(Dimen d);
607 extern int			dimenspec(Dimen d);
608 extern void		freedocinfo(Docinfo* d);
609 extern void		freeitems(Item* ithead);
610 extern Item*		parsehtml(uchar* data, int datalen, Rune* src, int mtype, int chset, Docinfo** pdi);
611 extern void		printitems(Item* items, char* msg);
612 extern int			targetid(Rune* s);
613 extern Rune*		targetname(int targid);
614 extern int			validitems(Item* i);
615 
616 /* #pragma varargck	type "I"	Item* */
617 
618 // Control print output
619 extern int			warn;
620 extern int			dbglex;
621 extern int			dbgbuild;
622 
623 // To be provided by caller
624 // emalloc and erealloc should not return if can't get memory.
625 // emalloc should zero its memory.
626 extern void*	emalloc(ulong);
627 extern void*	erealloc(void* p, ulong size);
628 #ifdef __cpluspplus
629 }
630 #endif
631 #endif