internal void Intersect(Bitset bs, bool inverse)
{
Subtract(bs,!inverse);
}
internal void Subtract(Bitset bs)
{
Subtract(bs,false);
}
internal void Add(Bitset bs, bool inverse)
{
weight += AddImpl(this,bs,!bs.positive^inverse);
}
internal void Intersect(Bitset bs)
{
Intersect(bs,false);
}
internal static int ParseClass(char[] data, int i, int _out, Term term, bool icase, bool skipspaces, bool unicode, bool xml)
{
Bitset bs = new Bitset();
i = ParseClass(data,i,_out,bs,icase,skipspaces,unicode,xml);
Bitset.Unify(bs,term);
return i;
}
internal void Add(Bitset bs)
{
Add(bs,false);
}
private static void InitPosixClasses()
{
Bitset lower = new Bitset();
lower.SetRange('a','z');
RegisterClass("Lower", lower, posixClasses);
Bitset upper = new Bitset();
upper.SetRange('A','Z');
RegisterClass("Upper",upper,posixClasses);
Bitset ascii = new Bitset();
ascii.SetRange((char)0,(char)0x7f);
RegisterClass("ASCII",ascii,posixClasses);
Bitset alpha = new Bitset();
alpha.Add(lower);
alpha.Add(upper);
RegisterClass("Alpha",alpha,posixClasses);
Bitset digit = new Bitset();
digit.SetRange('0','9');
RegisterClass("Digit",digit,posixClasses);
Bitset alnum = new Bitset();
alnum.Add(alpha);
alnum.Add(digit);
RegisterClass("Alnum",alnum,posixClasses);
Bitset punct = new Bitset();
punct.SetChars("!\"#$%&'()*+,-./:;<=>?@[\\]^_`{|}~");
RegisterClass("Punct",punct,posixClasses);
Bitset graph = new Bitset();
graph.Add(alnum);
graph.Add(punct);
RegisterClass("Graph",graph,posixClasses);
RegisterClass("Print",graph,posixClasses);
Bitset blank = new Bitset();
blank.SetChars(" \t");
RegisterClass("Blank",blank,posixClasses);
Bitset cntrl = new Bitset();
cntrl.SetRange((char)0,(char)0x1f);
cntrl.SetChar((char)0x7f);
RegisterClass("Cntrl",cntrl,posixClasses);
Bitset xdigit = new Bitset();
xdigit.SetRange('0','9');
xdigit.SetRange('a','f');
xdigit.SetRange('A','F');
RegisterClass("XDigit",xdigit,posixClasses);
Bitset space = new Bitset();
space.SetChars(" \t\n\r\f\u000b");
RegisterClass("Space",space,posixClasses);
}
private static int SubtractImpl(Bitset bs1, Bitset bs2, bool inv)
{
int s=0;
if(!bs1.isLarge && !bs2.isLarge && !inv){
bool[] bits1,bits2;
if((bits2=bs2.block0)!=null){
bits1=bs1.block0;
if(bits1==null) return 0;
s += Subtract(bits1,bits2,0,BLOCK_SIZE-1,false);
}
}
else {
if(!bs1.isLarge) bs1.EnableLargeMode();
if(!bs2.isLarge) bs2.EnableLargeMode();
s += Block.Subtract(bs1.blocks,bs2.blocks,0,BLOCK_COUNT-1,inv);
}
return s;
}
private static void InitNamedCategory(string name, int[] cats)
{
Bitset bs = new Bitset();
foreach(int cat in cats){
bs.SetCategory(cat);
}
namedClasses[name] = bs;
}
private static void InitNames()
{
InitNamedCategory("C",new int[]{UnicodeConstants.Cn,UnicodeConstants.Cc,UnicodeConstants.Cf,UnicodeConstants.Co,UnicodeConstants.Cs});
InitNamedCategory("Cn",UnicodeConstants.Cn);
InitNamedCategory("Cc",UnicodeConstants.Cc);
InitNamedCategory("Cf",UnicodeConstants.Cf);
InitNamedCategory("Co",UnicodeConstants.Co);
InitNamedCategory("Cs",UnicodeConstants.Cs);
InitNamedCategory("L",new int[]{UnicodeConstants.Lu,UnicodeConstants.Ll,UnicodeConstants.Lt,UnicodeConstants.Lm,UnicodeConstants.Lo});
InitNamedCategory("Lu",UnicodeConstants.Lu);
InitNamedCategory("Ll",UnicodeConstants.Ll);
InitNamedCategory("Lt",UnicodeConstants.Lt);
InitNamedCategory("Lm",UnicodeConstants.Lm);
InitNamedCategory("Lo",UnicodeConstants.Lo);
InitNamedCategory("M",new int[]{UnicodeConstants.Mn,UnicodeConstants.Me,UnicodeConstants.Mc});
InitNamedCategory("Mn",UnicodeConstants.Mn);
InitNamedCategory("Me",UnicodeConstants.Me);
InitNamedCategory("Mc",UnicodeConstants.Mc);
InitNamedCategory("N",new int[]{UnicodeConstants.Nd,UnicodeConstants.Nl,UnicodeConstants.No});
InitNamedCategory("Nd",UnicodeConstants.Nd);
InitNamedCategory("Nl",UnicodeConstants.Nl);
InitNamedCategory("No",UnicodeConstants.No);
InitNamedCategory("Z",new int[]{UnicodeConstants.Zs,UnicodeConstants.Zl,UnicodeConstants.Zp});
InitNamedCategory("Zs",UnicodeConstants.Zs);
InitNamedCategory("Zl",UnicodeConstants.Zl);
InitNamedCategory("Zp",UnicodeConstants.Zp);
InitNamedCategory("P",new int[]{UnicodeConstants.Pd,UnicodeConstants.Ps,UnicodeConstants.Pi,UnicodeConstants.Pe,UnicodeConstants.Pf,UnicodeConstants.Pc,UnicodeConstants.Po});
InitNamedCategory("Pd",UnicodeConstants.Pd);
InitNamedCategory("Ps",UnicodeConstants.Ps);
InitNamedCategory("Pi",UnicodeConstants.Pi);
InitNamedCategory("Pe",UnicodeConstants.Pe);
InitNamedCategory("Pf",UnicodeConstants.Pf);
InitNamedCategory("Pc",UnicodeConstants.Pc);
InitNamedCategory("Po",UnicodeConstants.Po);
InitNamedCategory("S",new int[]{UnicodeConstants.Sm,UnicodeConstants.Sc,UnicodeConstants.Sk,UnicodeConstants.So});
InitNamedCategory("Sm",UnicodeConstants.Sm);
InitNamedCategory("Sc",UnicodeConstants.Sc);
InitNamedCategory("Sk",UnicodeConstants.Sk);
InitNamedCategory("So",UnicodeConstants.So);
Bitset bs = new Bitset();
bs.SetCategory(UnicodeConstants.Cn);
RegisterClass("UNASSIGNED",bs,unicodeCategories);
bs = new Bitset();
bs.SetCategory(UnicodeConstants.Cn);
bs.SetPositive(false);
RegisterClass("ASSIGNED",bs,unicodeCategories);
string[] results = blockData.Split(new char[]{'.', ',', ':', ';'}, System.StringSplitOptions.RemoveEmptyEntries);
int ix = 0;
while(ix < results.Length) {
int first = Convert.ToInt32(results[ix++], 16);
int last = Convert.ToInt32(results[ix++], 16);
string name = results[ix++];
InitNamedBlock(name,first,last);
}
InitNamedBlock("ALL",0,0xffff);
namesInitialized=true;
}
private static void InitNamedCategory(string name, int cat)
{
Bitset bs = new Bitset();
bs.SetCategory(cat);
RegisterClass(name,bs,unicodeCategories);
}
private static void InitNamedBlock(string name, int first, int last)
{
if(first<char.MinValue || first>char.MaxValue) throw new ArgumentException("wrong start code ("+first+") in block "+name);
if(last<char.MinValue || last>char.MaxValue) throw new ArgumentException("wrong end code ("+last+") in block "+name);
if(last<first) throw new ArgumentException("end code < start code in block "+name);
Bitset bs;
if(namedClasses.ContainsKey(name)) {
bs = namedClasses[name];
} else {
bs = new Bitset();
RegisterClass(name,bs,unicodeBlocks);
}
bs.SetRange((char)first,(char)last);
}
internal static void ParseGroup(char[] data, int i, int _out, Term term, bool icase, bool skipspaces, bool unicode, bool xml)
{
Bitset sum = new Bitset();
Bitset bs = new Bitset();
int mode = ADD;
for(;i<_out;){
switch(data[i++]){
case '+':
mode=ADD;
continue;
case '-':
mode=SUBTRACT;
continue;
case '&':
mode=INTERSECT;
continue;
case '[':
bs.Reset();
i=ParseClass(data,i,_out,bs,icase,skipspaces,unicode,xml);
switch(mode){
case ADD:
sum.Add(bs);
break;
case SUBTRACT:
sum.Subtract(bs);
break;
case INTERSECT:
sum.Intersect(bs);
break;
}
continue;
case ')':
throw new PatternSyntaxException("unbalanced class group");
}
}
Bitset.Unify(sum,term);
}
internal void Subtract(Bitset bs, bool inverse)
{
weight += SubtractImpl(this, bs, !bs.positive^inverse);
}
private static int ParseClass(char[] data, int i, int _out, Bitset bs, bool icase, bool skipspaces, bool unicode, bool xml)
{
char c;
int prev=-1;
bool isFirst=true, setFirst=false, inRange=false;
Bitset bs1=null;
StringBuilder sb=null;
for(;i<_out;isFirst=setFirst,setFirst=false){
switch(c=data[i++]){
case ']':
if(isFirst) break; //treat as normal char
if(inRange){
bs.SetChar('-');
}
if(prev>=0){
char c1=(char)prev;
if(icase){
bs.SetChar(char.ToLower(c1));
bs.SetChar(char.ToUpper(c1));
}
else bs.SetChar(c1);
}
return i;
case '-':
if(isFirst) break;
if(inRange) break;
inRange=true;
continue;
case '[':
if(inRange && xml) {
if(prev>=0) bs.SetChar((char)prev);
if(bs1==null) bs1 = new Bitset();
else bs1.Reset();
i=ParseClass(data,i,_out,bs1,icase,skipspaces,unicode,xml);
bs.Subtract(bs1);
inRange=false;
prev=-1;
continue;
}
else break;
case '^':
if(isFirst){
bs.SetPositive(false);
setFirst=true;
continue;
}
break;
case ' ':
case '\r':
case '\n':
case '\t':
case '\f':
if(skipspaces) continue;
else break;
case '\\':
Bitset negatigeClass = null;
bool inv = false;
bool handle_special = false;
switch(c=data[i++]){
case 'r':
c='\r';
handle_special = true;
break;
case 'n':
c='\n';
handle_special = true;
break;
case 'e':
c='\u001B';
handle_special = true;
break;
case 't':
c='\t';
handle_special = true;
break;
case 'f':
c='\f';
handle_special = true;
break;
case 'u':
if(i>=_out-4) throw new PatternSyntaxException("incomplete escape sequence \\uXXXX");
c=(char)((ToHexDigit(c)<<12)
+(ToHexDigit(data[i++])<<8)
+(ToHexDigit(data[i++])<<4)
+ToHexDigit(data[i++]));
handle_special = true;
break;
case 'v':
c=(char)((ToHexDigit(c)<<24)+
(ToHexDigit(data[i++])<<16)+
(ToHexDigit(data[i++])<<12)+
(ToHexDigit(data[i++])<<8)+
(ToHexDigit(data[i++])<<4)+
ToHexDigit(data[i++]));
handle_special = true;
break;
case 'b':
c=(char)8; // backspace
handle_special = true;
break;
case 'x':{ // hex 2-digit number
int hex=0;
char d;
if((d=data[i++])=='{'){
while((d=data[i++])!='}'){
hex=(hex<<4)+ToHexDigit(d);
}
if(hex>0xffff) throw new PatternSyntaxException("\\x{<out of range>}");
}
else{
hex=(ToHexDigit(d)<<4)+ToHexDigit(data[i++]);
}
c=(char)hex;
handle_special = true;
break;
}
case '0': // oct 2- or 3-digit number
case 'o': // oct 2- or 3-digit number
int oct=0;
for(;;){
char d=data[i++];
if(d>='0' && d<='7'){
oct*=8;
oct+=d-'0';
if(oct>0xffff) break;
}
else {
i--;
break;
}
}
c=(char)oct;
handle_special = true;
break;
case 'm': // decimal number -> char
int dec=0;
for(;;){
char d=data[i++];
if(d>='0' && d<='9'){
dec*=10;
dec+=d-'0';
if(dec>0xffff) break;
} else {
i--;
break;
}
}
c=(char)dec;
handle_special = true;
break;
case 'c': // ctrl-char
c=(char)(data[i++]&0x1f);
handle_special = true;
break;
case 'D': // non-digit
negatigeClass = unicode ? UNONDIGIT : NONDIGIT;
break;
case 'S': // space
negatigeClass =unicode ? UNONSPACE : NONSPACE;
break;
case 'W': // space
negatigeClass = unicode ? UNONWORDCHAR : NONWORDCHAR;
break;
case 'd': // digit
if(inRange) throw new PatternSyntaxException("illegal range: [..."+prev+"-\\d...]");
bs.SetDigit(unicode);
continue;
case 's': // digit
if(inRange) throw new PatternSyntaxException("illegal range: [..."+prev+"-\\s...]");
bs.SetSpace(unicode);
continue;
case 'w': // digit
if(inRange) throw new PatternSyntaxException("illegal range: [..."+prev+"-\\w...]");
bs.SetWordChar(unicode);
continue;
case 'P': // \\P{..}
inv=true;
goto case 'p';
case 'p': // \\p{..}
if(inRange) throw new PatternSyntaxException("illegal range: [..."+prev+"-\\w...]");
if(sb==null) sb = new StringBuilder();
else sb.Length = 0;
i=ParseName(data,i,_out,sb,skipspaces);
Bitset nc=GetNamedClass(sb.ToString());
if(nc==null) throw new PatternSyntaxException("unknown named class: {"+sb+"}");
bs.Add(nc,inv);
continue;
default:
handle_special = true;
break;
}
if(handle_special) break;
if(inRange) throw new PatternSyntaxException("illegal range: [..."+prev+"-\\"+c+"...]");
bs.Add(negatigeClass);
continue;
default:
break;
}
if(prev<0){
prev=c;
inRange=false;
continue;
}
if(!inRange){
char c1=(char)prev;
if(icase){
bs.SetChar(char.ToLower(c1));
bs.SetChar(char.ToUpper(c1));
}
else bs.SetChar(c1);
prev=c;
}
else {
if(prev>c) throw new PatternSyntaxException("illegal range: "+prev+">"+c);
char c0=(char)prev;
inRange=false;
prev=-1;
if(icase){
bs.SetRange(char.ToLower(c0),char.ToLower(c));
bs.SetRange(char.ToUpper(c0),char.ToUpper(c));
}
else bs.SetRange(c0,c);
}
}
throw new PatternSyntaxException("unbalanced brackets in a class def");
}
private static int AddImpl(Bitset bs1, Bitset bs2, bool inv)
{
int s=0;
if(!bs1.isLarge && !bs2.isLarge && !inv){
if(bs2.block0!=null){
bool[] bits = bs1.block0;
if(bits==null) bs1.block0 = bits = new bool[BLOCK_SIZE];
s += Add(bits,bs2.block0,0,BLOCK_SIZE-1,false);
}
}
else {
if(!bs1.isLarge) bs1.EnableLargeMode();
if(!bs2.isLarge) bs2.EnableLargeMode();
s+=Block.Add(bs1.blocks,bs2.blocks,0,BLOCK_COUNT-1,inv);
}
return s;
}
private static void RegisterClass(string name, Bitset cls, IList<string> realm)
{
namedClasses[name] = cls;
if(!realm.Contains(name)) realm.Add(name);
}
internal static void Unify(Bitset bs, Term term)
{
if(bs.isLarge){
term.type = Term.TermType.BITSET2;
term.bitset2 = Block.ToBitset2(bs.blocks);
}
else{
term.type = Term.TermType.BITSET;
term.bitset = bs.block0 == null ? emptyBlock0 : bs.block0;
}
term.inverse = !bs.positive;
term.weight = bs.positive ? bs.weight : MAX_WEIGHT-bs.weight;
}
internal static void MakeICase(Term term, char c)
{
Bitset bs = new Bitset();
bs.SetChar(char.ToLower(c));
bs.SetChar(char.ToUpper(c));
Bitset.Unify(bs,term);
}