build.c

来自「这是一个同样来自贝尔实验室的和UNIX有着渊源的操作系统, 其简洁的设计和实现易」· C语言 代码 · 共 2,670 行 · 第 1/5 页

#include <u.h>
#include <libc.h>
#include <draw.h>
#include <ctype.h>
#include <html.h>
#include "impl.h"

// A stack for holding integer values
enum {
	Nestmax = 40	// max nesting level of lists, font styles, etc.
};

struct Stack {
	int		n;				// next available slot (top of stack is stack[n-1])
	int		slots[Nestmax];	// stack entries
};

// Parsing state
struct Pstate
{
	Pstate*	next;			// in stack of Pstates
	int		skipping;		// true when we shouldn't add items
	int		skipwhite;		// true when we should strip leading space
	int		curfont;		// font index for current font
	int		curfg;		// current foreground color
	Background	curbg;	// current background
	int		curvoff;		// current baseline offset
	uchar	curul;		// current underline/strike state
	uchar	curjust;		// current justify state
	int		curanchor;	// current (href) anchor id (if in one), or 0
	int		curstate;		// current value of item state
	int		literal;		// current literal state
	int		inpar;		// true when in a paragraph-like construct
	int		adjsize;		// current font size adjustment
	Item*	items;		// dummy head of item list we're building
	Item*	lastit;		// tail of item list we're building
	Item*	prelastit;		// item before lastit
	Stack	fntstylestk;	// style stack
	Stack	fntsizestk;		// size stack
	Stack	fgstk;		// text color stack
	Stack	ulstk;		// underline stack
	Stack	voffstk;		// vertical offset stack
	Stack	listtypestk;	// list type stack
	Stack	listcntstk;		// list counter stack
	Stack	juststk;		// justification stack
	Stack	hangstk;		// hanging stack
};

struct ItemSource
{
	Docinfo*		doc;
	Pstate*		psstk;
	int			nforms;
	int			ntables;
	int			nanchors;
	int			nframes;
	Form*		curform;
	Map*		curmap;
	Table*		tabstk;
	Kidinfo*		kidstk;
};

// Some layout parameters
enum {
	FRKIDMARGIN = 6,	// default margin around kid frames
	IMGHSPACE = 0,	// default hspace for images (0 matches IE, Netscape)
	IMGVSPACE = 0,	// default vspace for images
	FLTIMGHSPACE = 2,	// default hspace for float images
	TABSP = 5,		// default cellspacing for tables
	TABPAD = 1,		// default cell padding for tables
	LISTTAB = 1,		// number of tabs to indent lists
	BQTAB = 1,		// number of tabs to indent blockquotes
	HRSZ = 2,			// thickness of horizontal rules
	SUBOFF = 4,		// vertical offset for subscripts
	SUPOFF = 6,		// vertical offset for superscripts
	NBSP = 160		// non-breaking space character
};

// These tables must be sorted
static StringInt align_tab[] = {
	{L"baseline",	ALbaseline},
	{L"bottom",	ALbottom},
	{L"center",	ALcenter},
	{L"char",		ALchar},
	{L"justify",	ALjustify},
	{L"left",		ALleft},
	{L"middle",	ALmiddle},
	{L"right",		ALright},
	{L"top",		ALtop}
};
#define NALIGNTAB (sizeof(align_tab)/sizeof(StringInt))

static StringInt input_tab[] = {
	{L"button",	Fbutton},
	{L"checkbox",	Fcheckbox},
	{L"file",		Ffile},
	{L"hidden",	Fhidden},
	{L"image",	Fimage},
	{L"password",	Fpassword},
	{L"radio",		Fradio},
	{L"reset",		Freset},
	{L"submit",	Fsubmit},
	{L"text",		Ftext}
};
#define NINPUTTAB (sizeof(input_tab)/sizeof(StringInt))

static StringInt clear_tab[] = {
	{L"all",	IFcleft|IFcright},
	{L"left",	IFcleft},
	{L"right",	IFcright}
};
#define NCLEARTAB (sizeof(clear_tab)/sizeof(StringInt))

static StringInt fscroll_tab[] = {
	{L"auto",	FRhscrollauto|FRvscrollauto},
	{L"no",	FRnoscroll},
	{L"yes",	FRhscroll|FRvscroll},
};
#define NFSCROLLTAB (sizeof(fscroll_tab)/sizeof(StringInt))

static StringInt shape_tab[] = {
	{L"circ",		SHcircle},
	{L"circle",		SHcircle},
	{L"poly",		SHpoly},
	{L"polygon",	SHpoly},
	{L"rect",		SHrect},
	{L"rectangle",	SHrect}
};
#define NSHAPETAB (sizeof(shape_tab)/sizeof(StringInt))

static StringInt method_tab[] = {
	{L"get",		HGet},
	{L"post",		HPost}
};
#define NMETHODTAB (sizeof(method_tab)/sizeof(StringInt))

static Rune* roman[15]= {
	L"I", L"II", L"III", L"IV", L"V", L"VI", L"VII", L"VIII", L"IX", L"X",
	L"XI", L"XII", L"XIII", L"XIV", L"XV"
};
#define NROMAN 15

// List number types
enum {
	LTdisc, LTsquare, LTcircle, LT1, LTa, LTA, LTi, LTI
};

enum {
	SPBefore = 2,
	SPAfter = 4,
	BL = 1,
	BLBA = (BL|SPBefore|SPAfter)
};

// blockbrk[tag] is break info for a block level element, or one
// of a few others that get the same treatment re ending open paragraphs
// and requiring a line break / vertical space before them.
// If we want a line of space before the given element, SPBefore is OR'd in.
// If we want a line of space after the given element, SPAfter is OR'd in.

static uchar blockbrk[Numtags]= {
	[Taddress] BLBA, [Tblockquote] BLBA, [Tcenter] BL,
	[Tdir] BLBA, [Tdiv] BL, [Tdd] BL, [Tdl] BLBA,
	[Tdt] BL, [Tform] BLBA,
	// headings and tables get breaks added manually
	[Th1] BL, [Th2] BL, [Th3] BL,
	[Th4] BL, [Th5] BL, [Th6] BL,
	[Thr] BL, [Tisindex] BLBA, [Tli] BL, [Tmenu] BLBA,
	[Tol] BLBA, [Tp] BLBA, [Tpre] BLBA,
	[Tul] BLBA
};

enum {
	AGEN = 1
};

// attrinfo is information about attributes.
// The AGEN value means that the attribute is generic (applies to almost all elements)
static uchar attrinfo[Numattrs]= {
	[Aid] AGEN, [Aclass] AGEN, [Astyle] AGEN, [Atitle] AGEN,
	[Aonblur] AGEN, [Aonchange] AGEN, [Aonclick] AGEN,
	[Aondblclick] AGEN, [Aonfocus] AGEN, [Aonkeypress] AGEN,
	[Aonkeyup] AGEN, [Aonload] AGEN, [Aonmousedown] AGEN,
	[Aonmousemove] AGEN, [Aonmouseout] AGEN, [Aonmouseover] AGEN,
	[Aonmouseup] AGEN, [Aonreset] AGEN, [Aonselect] AGEN,
	[Aonsubmit] AGEN, [Aonunload] AGEN
};

static uchar scriptev[Numattrs]= {
	[Aonblur] SEonblur, [Aonchange] SEonchange, [Aonclick] SEonclick,
	[Aondblclick] SEondblclick, [Aonfocus] SEonfocus, [Aonkeypress] SEonkeypress,
	[Aonkeyup] SEonkeyup, [Aonload] SEonload, [Aonmousedown] SEonmousedown,
	[Aonmousemove] SEonmousemove, [Aonmouseout] SEonmouseout, [Aonmouseover] SEonmouseover,
	[Aonmouseup] SEonmouseup, [Aonreset] SEonreset, [Aonselect] SEonselect,
	[Aonsubmit] SEonsubmit, [Aonunload] SEonunload
};

// Color lookup table
static StringInt color_tab[] = {
	{L"aqua", 0x00FFFF},
	{L"black",  0x000000},
	{L"blue", 0x0000CC},
	{L"fuchsia", 0xFF00FF},
	{L"gray", 0x808080},
	{L"green", 0x008000},
	{L"lime", 0x00FF00},
	{L"maroon", 0x800000},
	{L"navy", 0x000080,},
	{L"olive", 0x808000},
	{L"purple", 0x800080},
	{L"red", 0xFF0000},
	{L"silver", 0xC0C0C0},
	{L"teal", 0x008080},
	{L"white", 0xFFFFFF},
	{L"yellow", 0xFFFF00}
};
#define NCOLORS (sizeof(color_tab)/sizeof(StringInt))

static StringInt 		*targetmap;
static int			targetmapsize;
static int			ntargets;

static int buildinited = 0;

#define SMALLBUFSIZE 240
#define BIGBUFSIZE 2000

int	dbgbuild = 0;
int	warn = 0;

static Align		aalign(Token* tok);
static int			acolorval(Token* tok, int attid, int dflt);
static void			addbrk(Pstate* ps, int sp, int clr);
static void			additem(Pstate* ps, Item* it, Token* tok);
static void			addlinebrk(Pstate* ps, int clr);
static void			addnbsp(Pstate* ps);
static void			addtext(Pstate* ps, Rune* s);
static Dimen		adimen(Token* tok, int attid);
static int			aflagval(Token* tok, int attid);
static int			aintval(Token* tok, int attid, int dflt);
static Rune*		astrval(Token* tok, int attid, Rune* dflt);
static int			atabval(Token* tok, int attid, StringInt* tab, int ntab, int dflt);
static int			atargval(Token* tok, int dflt);
static int			auintval(Token* tok, int attid, int dflt);
static Rune*		aurlval(Token* tok, int attid, Rune* dflt, Rune* base);
static Rune*		aval(Token* tok, int attid);
static void			buildinit(void);
static Pstate*		cell_pstate(Pstate* oldps, int ishead);
static void			changehang(Pstate* ps, int delta);
static void			changeindent(Pstate* ps, int delta);
static int			color(Rune* s, int dflt);
static void			copystack(Stack* tostk, Stack* fromstk);
static int			dimprint(char* buf, int nbuf, Dimen d);
static Pstate*		finishcell(Table* curtab, Pstate* psstk);
static void			finish_table(Table* t);
static void			freeanchor(Anchor* a);
static void			freedestanchor(DestAnchor* da);
static void			freeform(Form* f);
static void			freeformfield(Formfield* ff);
static void			freeitem(Item* it);
static void			freepstate(Pstate* p);
static void			freepstatestack(Pstate* pshead);
static void			freescriptevents(SEvent* ehead);
static void			freetable(Table* t);
static Map*		getmap(Docinfo* di, Rune* name);
static Rune*		getpcdata(Token* toks, int tokslen, int* ptoki);
static Pstate*		lastps(Pstate* psl);
static Rune*		listmark(uchar ty, int n);
static int			listtyval(Token* tok, int dflt);
static Align		makealign(int halign, int valign);
static Background	makebackground(Rune* imgurl, int color);
static Dimen		makedimen(int kind, int spec);
static Anchor*		newanchor(int index, Rune* name, Rune* href, int target, Anchor* link);
static Area*		newarea(int shape, Rune* href, int target, Area* link);
static DestAnchor*	newdestanchor(int index, Rune* name, Item* item, DestAnchor* link);
static Docinfo*		newdocinfo(void);
static Genattr*		newgenattr(Rune* id, Rune* class, Rune* style, Rune* title, Attr* events);
static Form*		newform(int formid, Rune* name, Rune* action,
					int target, int method, Form* link);
static Formfield*	newformfield(int ftype, int fieldid, Form* form, Rune* name,
					Rune* value, int size, int maxlength, Formfield* link);
static Item*		newifloat(Item* it, int side);
static Item*		newiformfield(Formfield* ff);
static Item*		newiimage(Rune* src, Rune* altrep, int align, int width, int height,
					int hspace, int vspace, int border, int ismap, Map* map);
static Item*		newirule(int align, int size, int noshade, Dimen wspec);
static Item*		newispacer(int spkind);
static Item*		newitable(Table* t);
static ItemSource*	newitemsource(Docinfo* di);
static Item*		newitext(Rune* s, int fnt, int fg, int voff, int ul);
static Kidinfo*		newkidinfo(int isframeset, Kidinfo* link);
static Option*		newoption(int selected, Rune* value, Rune* display, Option* link);
static Pstate*		newpstate(Pstate* link);
static SEvent*		newscriptevent(int type, Rune* script, SEvent* link);
static Table*		newtable(int tableid, Align align, Dimen width, int border,
					int cellspacing, int cellpadding, Background bg, Token* tok, Table* link);
static Tablecell*	newtablecell(int cellid, int rowspan, int colspan, Align align, Dimen wspec,
					int hspec, Background bg, int flags, Tablecell* link);
static Tablerow*	newtablerow(Align align, Background bg, int flags, Tablerow* link);
static Dimen		parsedim(Rune* s, int ns);
static void			pop(Stack* stk);
static void			popfontsize(Pstate* ps);
static void			popfontstyle(Pstate* ps);
static void			popjust(Pstate* ps);
static int			popretnewtop(Stack* stk, int dflt);
static int			push(Stack* stk, int val);
static void			pushfontsize(Pstate* ps, int sz);
static void			pushfontstyle(Pstate* ps, int sty);
static void			pushjust(Pstate* ps, int j);
static Item*		textit(Pstate* ps, Rune* s);
static Rune*		removeallwhite(Rune* s);
static void			resetdocinfo(Docinfo* d);
static void			setcurfont(Pstate* ps);
static void			setcurjust(Pstate* ps);
static void			setdimarray(Token* tok, int attid, Dimen** pans, int* panslen);
static Rune*		stringalign(int a);
static void			targetmapinit(void);
static int			toint(Rune* s);
static int			top(Stack* stk, int dflt);
static void			trim_cell(Tablecell* c);
static int			validalign(Align a);
static int			validdimen(Dimen d);
static int			validformfield(Formfield* f);
static int			validhalign(int a);
static int			validptr(void* p);
static int			validStr(Rune* s);
static int			validtable(Table* t);
static int			validtablerow(Tablerow* r);
static int			validtablecol(Tablecol* c);
static int			validtablecell(Tablecell* c);
static int			validvalign(int a);
static int			Iconv(Fmt *f);

static void
buildinit(void)
{
	fmtinstall('I', Iconv);
	targetmapinit();
	buildinited = 1;
}

static ItemSource*
newitemsource(Docinfo* di)
{
	ItemSource*	is;
	Pstate*	ps;

	ps = newpstate(nil);
	if(di->mediatype != TextHtml) {
		ps->curstate &= ~IFwrap;
		ps->literal = 1;
		pushfontstyle(ps, FntT);
	}
	is = (ItemSource*)emalloc(sizeof(ItemSource));
	is->doc = di;
	is->psstk = ps;
	is->nforms = 0;
	is->ntables = 0;
	is->nanchors = 0;
	is->nframes = 0;
	is->curform = nil;
	is->curmap = nil;
	is->tabstk = nil;
	is->kidstk = nil;
	return is;
}

static Item *getitems(ItemSource* is, uchar* data, int datalen);

// Parse an html document and create a list of layout items.
// Allocate and return document info in *pdi.
// When caller is done with the items, it should call
// freeitems on the returned result, and then
// freedocinfo(*pdi).
Item*
parsehtml(uchar* data, int datalen, Rune* pagesrc, int mtype, int chset, Docinfo** pdi)
{
	Item *it;
	Docinfo*	di;
	ItemSource*	is;

	di = newdocinfo();
	di->src = _Strdup(pagesrc);
	di->base = _Strdup(pagesrc);
	di->mediatype = mtype;
	di->chset = chset;
	*pdi = di;
	is = newitemsource(di);
	it = getitems(is, data, datalen);
	freepstatestack(is->psstk);
	free(is);
	return it;
}

// Get a group of tokens for lexer, parse them, and create
// a list of layout items.
// When caller is done with the items, it should call
// freeitems on the returned result.
static Item*
getitems(ItemSource* is, uchar* data, int datalen)
{
	int	i;
	int	j;
	int	nt;
	int	pt;
	int	doscripts;
	int	tokslen;
	int	toki;
	int	h;
	int	sz;
	int	method;
	int	n;
	int	nblank;
	int	norsz;
	int	bramt;
	int	sty;
	int	nosh;
	int	oldcuranchor;
	int	dfltbd;
	int	v;
	int	hang;
	int	isempty;
	int	tag;
	int	brksp;
	int	target;
	uchar	brk;
	uchar	flags;
	uchar	align;
	uchar	al;
	uchar	ty;
	uchar	ty2;
	Pstate*	ps;
	Pstate*	nextps;
	Pstate*	outerps;
	Table*	curtab;
	Token*	tok;
	Token*	toks;
	Docinfo*	di;
	Item*	ans;
	Item*	img;
	Item*	ffit;
	Item*	tabitem;
	Rune*	s;
	Rune*	t;
	Rune*	name;
	Rune*	enctype;
	Rune*	usemap;
	Rune*	prompt;
	Rune*	equiv;
	Rune*	val;
	Rune*	nsz;
	Rune*	script;
	Map*	map;
	Form*	frm;
	Iimage*	ii;
	Kidinfo*	kd;
	Kidinfo*	ks;
	Kidinfo*	pks;
	Dimen	wd;
	Option*	option;
	Table*	tab;
	Tablecell*	c;
	Tablerow*	tr;
	Formfield*	field;
	Formfield*	ff;
	Rune*	href;
	Rune*	src;
	Rune*	scriptsrc;
	Rune*	bgurl;
	Rune*	action;
	Background	bg;

	if(!buildinited)
		buildinit();
	doscripts = 0;	// for now
	ps = is->psstk;
	curtab = is->tabstk;
	di = is->doc;
	toks = _gettoks(data, datalen, di->chset, di->mediatype, &tokslen);
	toki = 0;
	for(; toki < tokslen; toki++) {
		tok = &toks[toki];
		if(dbgbuild > 1)
			fprint(2, "build: curstate %ux, token %T\n", ps->curstate, tok);
		tag = tok->tag;
		brk = 0;
		brksp = 0;
		if(tag < Numtags) {
			brk = blockbrk[tag];
			if(brk&SPBefore)
				brksp = 1;
		}
		else if(tag < Numtags + RBRA) {
			brk = blockbrk[tag - RBRA];
			if(brk&SPAfter)
				brksp = 1;
		}
		if(brk) {
			addbrk(ps, brksp, 0);
			if(ps->inpar) {
				popjust(ps);
				ps->inpar = 0;
			}
		}
		// check common case first (Data), then switch statement on tag
		if(tag == Data) {
			// Lexing didn't pay attention to SGML record boundary rules:
			// \n after start tag or before end tag to be discarded.
			// (Lex has already discarded all \r's).
			// Some pages assume this doesn't happen in <PRE> text,
			// so we won't do it if literal is true.
			// BUG: won't discard \n before a start tag that begins
			// the next bufferful of tokens.
			s = tok->text;
			n = _Strlen(s);
			if(!ps->literal) {
				i = 0;
				j = n;
				if(toki > 0) {
					pt = toks[toki - 1].tag;
					// IE and Netscape both ignore this rule (contrary to spec)
					// if previous tag was img
					if(pt < Numtags && pt != Timg && j > 0 && s[0] == '\n')
						i++;
				}
				if(toki < tokslen - 1) {
					nt = toks[toki + 1].tag;
					if(nt >= RBRA && nt < Numtags + RBRA && j > i && s[j - 1] == '\n')
						j--;
				}
				if(i > 0 || j < n) {
					t = s;
					s = _Strsubstr(s, i, j);
					free(t);