void Init() { pos = -1; line = 1; col = 0; charPos = -1; oldEols = 0; NextCh(); /*if (ch == 0xEF) { // check optional byte order mark for UTF-8 NextCh(); int ch1 = ch; NextCh(); int ch2 = ch; if (ch1 != 0xBB || ch2 != 0xBF) { throw new FatalError(String.Format("illegal byte order mark: EF {0,2:X} {1,2:X}", ch1, ch2)); } buffer = new UTF8Buffer(buffer); col = 0; charPos = -1; NextCh(); }*/ pt = tokens = new Token(); // first token is a dummy }
Token NextToken() { while (ch == ' ' || ch >= 9 && ch <= 10 || ch == 13 ) NextCh(); if (ch == '/' && Comment0() ||ch == '/' && Comment1()) return NextToken(); int apx = 0; int recKind = noSym; int recEnd = pos; t = new Token(); t.pos = pos; t.col = col; t.line = line; t.charPos = charPos; int state; if (start.ContainsKey(ch)) { state = (int) start[ch]; } else { state = 0; } tlen = 0; AddCh(); switch (state) { case -1: { t.kind = eofSym; break; } // NextCh already done case 0: { if (recKind != noSym) { tlen = recEnd - t.pos; SetScannerBehindT(); } t.kind = recKind; break; } // NextCh already done case 1: {t.kind = 2; break;} case 2: {t.kind = 3; break;} case 3: {t.kind = 5; break;} case 4: {t.kind = 6; break;} case 5: {t.kind = 7; break;} case 6: {t.kind = 9; break;} case 7: {t.kind = 10; break;} case 8: {t.kind = 11; break;} case 9: {t.kind = 12; break;} case 10: recEnd = pos; recKind = 14; if (ch >= '0' && ch <= '9' || ch >= 'A' && ch <= 'Z' || ch == '_' || ch >= 'a' && ch <= 'z') {AddCh(); goto case 10;} else {t.kind = 14; t.val = new String(tval, 0, tlen); CheckLiteral(); return t;} case 11: {t.kind = 15; break;} case 12: { tlen -= apx; SetScannerBehindT(); t.kind = 15; break;} case 13: if (ch >= '0' && ch <= '9') {AddCh(); goto case 15;} else if (ch == '+' || ch == '-') {AddCh(); goto case 14;} else {goto case 0;} case 14: if (ch >= '0' && ch <= '9') {AddCh(); goto case 15;} else {goto case 0;} case 15: recEnd = pos; recKind = 16; if (ch >= '0' && ch <= '9' || ch == '_') {AddCh(); goto case 15;} else if (ch == 'F' || ch == 'f') {AddCh(); goto case 20;} else {t.kind = 16; break;} case 16: recEnd = pos; recKind = 16; if (ch >= '0' && ch <= '9' || ch == '_') {AddCh(); goto case 16;} else if (ch == 'F' || ch == 'f') {AddCh(); goto case 20;} else if (ch == 'E' || ch == 'e') {AddCh(); goto case 17;} else {t.kind = 16; break;} case 17: if (ch >= '0' && ch <= '9') {AddCh(); goto case 19;} else if (ch == '+' || ch == '-') {AddCh(); goto case 18;} else {goto case 0;} case 18: if (ch >= '0' && ch <= '9') {AddCh(); goto case 19;} else {goto case 0;} case 19: recEnd = pos; recKind = 16; if (ch >= '0' && ch <= '9' || ch == '_') {AddCh(); goto case 19;} else if (ch == 'F' || ch == 'f') {AddCh(); goto case 20;} else {t.kind = 16; break;} case 20: {t.kind = 16; break;} case 21: if (ch >= '0' && ch <= '9' || ch >= 'A' && ch <= 'F' || ch >= 'a' && ch <= 'f') {AddCh(); goto case 22;} else {goto case 0;} case 22: recEnd = pos; recKind = 17; if (ch >= '0' && ch <= '9' || ch >= 'A' && ch <= 'F' || ch >= 'a' && ch <= 'f') {AddCh(); goto case 22;} else {t.kind = 17; break;} case 23: if (ch == 'U' || ch == 'u') {AddCh(); goto case 24;} else {goto case 0;} case 24: if (ch >= '0' && ch <= '9' || ch >= 'A' && ch <= 'F' || ch >= 'a' && ch <= 'f') {AddCh(); goto case 25;} else {goto case 0;} case 25: if (ch >= '0' && ch <= '9' || ch >= 'A' && ch <= 'F' || ch >= 'a' && ch <= 'f') {AddCh(); goto case 26;} else {goto case 0;} case 26: if (ch >= '0' && ch <= '9' || ch >= 'A' && ch <= 'F' || ch >= 'a' && ch <= 'f') {AddCh(); goto case 27;} else {goto case 0;} case 27: if (ch >= '0' && ch <= '9' || ch >= 'A' && ch <= 'F' || ch >= 'a' && ch <= 'f') {AddCh(); goto case 28;} else {goto case 0;} case 28: recEnd = pos; recKind = 18; if (ch >= '0' && ch <= '9' || ch >= 'A' && ch <= 'F' || ch >= 'a' && ch <= 'f') {AddCh(); goto case 29;} else {t.kind = 18; break;} case 29: if (ch >= '0' && ch <= '9' || ch >= 'A' && ch <= 'F' || ch >= 'a' && ch <= 'f') {AddCh(); goto case 30;} else {goto case 0;} case 30: {t.kind = 18; break;} case 31: if (ch <= '&' || ch >= '(' && ch <= '[' || ch >= ']' && ch <= 65535) {AddCh(); goto case 32;} else if (ch == 92) {AddCh(); goto case 33;} else if (ch == 39) {AddCh(); goto case 40;} else {goto case 0;} case 32: if (ch == 39) {AddCh(); goto case 40;} else {goto case 0;} case 33: if (ch == 92 || ch >= 'a' && ch <= 'b' || ch == 'f' || ch == 'n' || ch == 'r' || ch == 't') {AddCh(); goto case 32;} else if (ch == 'U' || ch == 'u') {AddCh(); goto case 34;} else {goto case 0;} case 34: if (ch >= '0' && ch <= '9' || ch >= 'A' && ch <= 'F' || ch >= 'a' && ch <= 'f') {AddCh(); goto case 35;} else {goto case 0;} case 35: if (ch >= '0' && ch <= '9' || ch >= 'A' && ch <= 'F' || ch >= 'a' && ch <= 'f') {AddCh(); goto case 36;} else {goto case 0;} case 36: if (ch >= '0' && ch <= '9' || ch >= 'A' && ch <= 'F' || ch >= 'a' && ch <= 'f') {AddCh(); goto case 37;} else {goto case 0;} case 37: if (ch >= '0' && ch <= '9' || ch >= 'A' && ch <= 'F' || ch >= 'a' && ch <= 'f') {AddCh(); goto case 38;} else {goto case 0;} case 38: if (ch >= '0' && ch <= '9' || ch >= 'A' && ch <= 'F' || ch >= 'a' && ch <= 'f') {AddCh(); goto case 39;} else if (ch == 39) {AddCh(); goto case 40;} else {goto case 0;} case 39: if (ch >= '0' && ch <= '9' || ch >= 'A' && ch <= 'F' || ch >= 'a' && ch <= 'f') {AddCh(); goto case 32;} else {goto case 0;} case 40: {t.kind = 19; break;} case 41: if (ch <= '!' || ch >= '#' && ch <= '[' || ch >= ']' && ch <= 65535) {AddCh(); goto case 41;} else if (ch == 92) {AddCh(); goto case 42;} else if (ch == '"') {AddCh(); goto case 48;} else {goto case 0;} case 42: if (ch == 92 || ch >= 'a' && ch <= 'b' || ch == 'f' || ch == 'n' || ch == 'r' || ch == 't') {AddCh(); goto case 41;} else if (ch == 'U' || ch == 'u') {AddCh(); goto case 43;} else {goto case 0;} case 43: if (ch >= '0' && ch <= '9' || ch >= 'A' && ch <= 'F' || ch >= 'a' && ch <= 'f') {AddCh(); goto case 44;} else {goto case 0;} case 44: if (ch >= '0' && ch <= '9' || ch >= 'A' && ch <= 'F' || ch >= 'a' && ch <= 'f') {AddCh(); goto case 45;} else {goto case 0;} case 45: if (ch >= '0' && ch <= '9' || ch >= 'A' && ch <= 'F' || ch >= 'a' && ch <= 'f') {AddCh(); goto case 46;} else {goto case 0;} case 46: if (ch >= '0' && ch <= '9' || ch >= 'A' && ch <= 'F' || ch >= 'a' && ch <= 'f') {AddCh(); goto case 47;} else {goto case 0;} case 47: if (ch <= '!' || ch >= '#' && ch <= '/' || ch >= ':' && ch <= '@' || ch >= 'G' && ch <= '[' || ch >= ']' && ch <= '`' || ch >= 'g' && ch <= 65535) {AddCh(); goto case 41;} else if (ch == 92) {AddCh(); goto case 42;} else if (ch >= '0' && ch <= '9' || ch >= 'A' && ch <= 'F' || ch >= 'a' && ch <= 'f') {AddCh(); goto case 58;} else if (ch == '"') {AddCh(); goto case 48;} else {goto case 0;} case 48: {t.kind = 20; break;} case 49: if (ch <= '!' || ch >= '#' && ch <= 65535) {AddCh(); goto case 49;} else if (ch == '"') {AddCh(); goto case 52;} else {goto case 0;} case 50: if (ch == '"') {AddCh(); goto case 51;} else if (ch == '#') {AddCh(); goto case 50;} else {goto case 0;} case 51: if (ch <= '!' || ch >= '#' && ch <= 65535) {AddCh(); goto case 51;} else if (ch == '"') {AddCh(); goto case 59;} else {goto case 0;} case 52: {t.kind = 21; break;} case 53: recEnd = pos; recKind = 15; if (ch >= '0' && ch <= '9' || ch == '_') {AddCh(); goto case 53;} else if (ch == '.') {apx++; AddCh(); goto case 60;} else if (ch == 'L' || ch == 'U' || ch == 'l' || ch == 'u') {AddCh(); goto case 11;} else {t.kind = 15; break;} case 54: recEnd = pos; recKind = 13; if (ch >= '0' && ch <= '9') {AddCh(); goto case 16;} else if (ch == '.') {AddCh(); goto case 61;} else {t.kind = 13; break;} case 55: recEnd = pos; recKind = 14; if (ch >= '0' && ch <= '9' || ch >= 'A' && ch <= 'Z' || ch == '_' || ch >= 'a' && ch <= 'z') {AddCh(); goto case 10;} else if (ch == '"') {AddCh(); goto case 49;} else if (ch == '#') {AddCh(); goto case 50;} else {t.kind = 14; t.val = new String(tval, 0, tlen); CheckLiteral(); return t;} case 56: recEnd = pos; recKind = 4; if (ch == ':') {AddCh(); goto case 3;} else {t.kind = 4; break;} case 57: recEnd = pos; recKind = 15; if (ch >= '0' && ch <= '9' || ch == '_') {AddCh(); goto case 53;} else if (ch == '.') {apx++; AddCh(); goto case 60;} else if (ch == 'L' || ch == 'U' || ch == 'l' || ch == 'u') {AddCh(); goto case 11;} else if (ch == 'X' || ch == 'x') {AddCh(); goto case 21;} else {t.kind = 15; break;} case 58: if (ch <= '!' || ch >= '#' && ch <= '[' || ch >= ']' && ch <= 65535) {AddCh(); goto case 41;} else if (ch == 92) {AddCh(); goto case 42;} else if (ch == '"') {AddCh(); goto case 48;} else {goto case 0;} case 59: if (ch <= '!' || ch >= '$' && ch <= 65535) {AddCh(); goto case 51;} else if (ch == '"') {AddCh(); goto case 59;} else if (ch == '#') {AddCh(); goto case 62;} else {goto case 0;} case 60: recEnd = pos; recKind = 16; if (ch >= '0' && ch <= '9') {apx = 0; AddCh(); goto case 16;} else if (ch == '.') {apx++; AddCh(); goto case 12;} else if (ch == 'F' || ch == 'f') {apx = 0; AddCh(); goto case 20;} else if (ch == 'E' || ch == 'e') {apx = 0; AddCh(); goto case 13;} else {t.kind = 16; break;} case 61: recEnd = pos; recKind = 1; if (ch == '.') {AddCh(); goto case 1;} else {t.kind = 1; break;} case 62: recEnd = pos; recKind = 21; if (ch <= '!' || ch >= '$' && ch <= 65535) {AddCh(); goto case 51;} else if (ch == '"') {AddCh(); goto case 59;} else if (ch == '#') {AddCh(); goto case 62;} else {t.kind = 21; break;} case 63: {t.kind = 39; break;} case 64: {t.kind = 41; break;} case 65: {t.kind = 64; break;} case 66: {t.kind = 65; break;} case 67: {t.kind = 66; break;} case 68: {t.kind = 67; break;} case 69: {t.kind = 68; break;} case 70: {t.kind = 69; break;} case 71: {t.kind = 70; break;} case 72: {t.kind = 71; break;} case 73: {t.kind = 72; break;} case 74: {t.kind = 73; break;} case 75: {t.kind = 76; break;} case 76: {t.kind = 78; break;} case 77: {t.kind = 79; break;} case 78: {t.kind = 80; break;} case 79: {t.kind = 81; break;} case 80: {t.kind = 82; break;} case 81: {t.kind = 83; break;} case 82: {t.kind = 84; break;} case 83: {t.kind = 85; break;} case 84: {t.kind = 86; break;} case 85: recEnd = pos; recKind = 8; if (ch == '-') {AddCh(); goto case 64;} else {t.kind = 8; break;} case 86: recEnd = pos; recKind = 90; if (ch == '>') {AddCh(); goto case 63;} else if (ch == '=') {AddCh(); goto case 66;} else {t.kind = 90; break;} case 87: recEnd = pos; recKind = 40; if (ch == '>') {AddCh(); goto case 75;} else if (ch == '=') {AddCh(); goto case 80;} else {t.kind = 40; break;} case 88: recEnd = pos; recKind = 42; if (ch == '<') {AddCh(); goto case 97;} else if (ch == '=') {AddCh(); goto case 82;} else {t.kind = 42; break;} case 89: recEnd = pos; recKind = 43; if (ch == '>') {AddCh(); goto case 98;} else if (ch == '=') {AddCh(); goto case 83;} else {t.kind = 43; break;} case 90: recEnd = pos; recKind = 44; if (ch == '=') {AddCh(); goto case 71;} else if (ch == '&') {AddCh(); goto case 79;} else {t.kind = 44; break;} case 91: recEnd = pos; recKind = 89; if (ch == '=') {AddCh(); goto case 65;} else {t.kind = 89; break;} case 92: recEnd = pos; recKind = 91; if (ch == '=') {AddCh(); goto case 67;} else if (ch == '*') {AddCh(); goto case 99;} else {t.kind = 91; break;} case 93: recEnd = pos; recKind = 92; if (ch == '=') {AddCh(); goto case 68;} else {t.kind = 92; break;} case 94: recEnd = pos; recKind = 93; if (ch == '=') {AddCh(); goto case 70;} else {t.kind = 93; break;} case 95: recEnd = pos; recKind = 75; if (ch == '=') {AddCh(); goto case 72;} else if (ch == '|') {AddCh(); goto case 78;} else {t.kind = 75; break;} case 96: recEnd = pos; recKind = 95; if (ch == '=') {AddCh(); goto case 81;} else {t.kind = 95; break;} case 97: recEnd = pos; recKind = 87; if (ch == '=') {AddCh(); goto case 73;} else {t.kind = 87; break;} case 98: recEnd = pos; recKind = 88; if (ch == '=') {AddCh(); goto case 74;} else {t.kind = 88; break;} case 99: recEnd = pos; recKind = 94; if (ch == '=') {AddCh(); goto case 69;} else {t.kind = 94; break;} } t.val = new String(tval, 0, tlen); return t; }
// get the next token (possibly a token already seen during peeking) public Token Scan() { if (tokens.next == null) { return NextToken(); } else { pt = tokens = tokens.next; return tokens; } }
// make sure that peeking starts at the current scan position public void ResetPeek() { pt = tokens; }
// peek for the next token, ignore pragmas public Token Peek() { do { if (pt.next == null) { pt.next = NextToken(); } pt = pt.next; } while (pt.kind > maxT); // skip pragmas return pt; }
public void Parse() { la = new Token(); la.val = ""; Get(); Expresso(); Expect(0); }
void Get() { for (;;) { t = la; la = scanner.Scan(); if (la.kind <= maxT) { ++errDist; break; } la = t; } }