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: recEnd = pos; recKind = 1; if (ch >= '0' && ch <= '9' || ch >= 'A' && ch <= 'Z' || ch == '_' || ch >= 'a' && ch <= 'z') {AddCh(); goto case 1;} else {t.kind = 1; t.val = new String(tval, 0, tlen); CheckLiteral(); return t;} case 2: { tlen -= apx; SetScannerBehindT(); t.kind = 2; break;} case 3: if (ch >= '0' && ch <= '9' || ch >= 'A' && ch <= 'F' || ch >= 'a' && ch <= 'f') {AddCh(); goto case 4;} else {goto case 0;} case 4: recEnd = pos; recKind = 2; if (ch >= '0' && ch <= '9' || ch >= 'A' && ch <= 'F' || ch >= 'a' && ch <= 'f') {AddCh(); goto case 4;} else if (ch == 'U') {AddCh(); goto case 80;} else if (ch == 'u') {AddCh(); goto case 81;} else if (ch == 'L') {AddCh(); goto case 82;} else if (ch == 'l') {AddCh(); goto case 83;} else {t.kind = 2; break;} case 5: {t.kind = 2; break;} case 6: recEnd = pos; recKind = 3; if (ch >= '0' && ch <= '9') {AddCh(); goto case 6;} else if (ch == 'D' || ch == 'F' || ch == 'M' || ch == 'd' || ch == 'f' || ch == 'm') {AddCh(); goto case 17;} else if (ch == 'E' || ch == 'e') {AddCh(); goto case 7;} else {t.kind = 3; break;} case 7: if (ch >= '0' && ch <= '9') {AddCh(); goto case 9;} else if (ch == '+' || ch == '-') {AddCh(); goto case 8;} else {goto case 0;} case 8: if (ch >= '0' && ch <= '9') {AddCh(); goto case 9;} else {goto case 0;} case 9: recEnd = pos; recKind = 3; if (ch >= '0' && ch <= '9') {AddCh(); goto case 9;} else if (ch == 'D' || ch == 'F' || ch == 'M' || ch == 'd' || ch == 'f' || ch == 'm') {AddCh(); goto case 17;} else {t.kind = 3; break;} case 10: recEnd = pos; recKind = 3; if (ch >= '0' && ch <= '9') {AddCh(); goto case 10;} else if (ch == 'D' || ch == 'F' || ch == 'M' || ch == 'd' || ch == 'f' || ch == 'm') {AddCh(); goto case 17;} else if (ch == 'E' || ch == 'e') {AddCh(); goto case 11;} else {t.kind = 3; break;} case 11: if (ch >= '0' && ch <= '9') {AddCh(); goto case 13;} else if (ch == '+' || ch == '-') {AddCh(); goto case 12;} else {goto case 0;} case 12: if (ch >= '0' && ch <= '9') {AddCh(); goto case 13;} else {goto case 0;} case 13: recEnd = pos; recKind = 3; if (ch >= '0' && ch <= '9') {AddCh(); goto case 13;} else if (ch == 'D' || ch == 'F' || ch == 'M' || ch == 'd' || ch == 'f' || ch == 'm') {AddCh(); goto case 17;} else {t.kind = 3; break;} case 14: if (ch >= '0' && ch <= '9') {AddCh(); goto case 16;} else if (ch == '+' || ch == '-') {AddCh(); goto case 15;} else {goto case 0;} case 15: if (ch >= '0' && ch <= '9') {AddCh(); goto case 16;} else {goto case 0;} case 16: recEnd = pos; recKind = 3; if (ch >= '0' && ch <= '9') {AddCh(); goto case 16;} else if (ch == 'D' || ch == 'F' || ch == 'M' || ch == 'd' || ch == 'f' || ch == 'm') {AddCh(); goto case 17;} else {t.kind = 3; break;} case 17: {t.kind = 3; break;} case 18: if (ch <= 9 || ch >= 11 && ch <= 12 || ch >= 14 && ch <= '&' || ch >= '(' && ch <= '[' || ch >= ']' && ch <= 65535) {AddCh(); goto case 19;} else if (ch == 92) {AddCh(); goto case 84;} else {goto case 0;} case 19: if (ch == 39) {AddCh(); goto case 34;} else {goto case 0;} case 20: if (ch >= '0' && ch <= '9' || ch >= 'A' && ch <= 'F' || ch >= 'a' && ch <= 'f') {AddCh(); goto case 21;} else {goto case 0;} case 21: if (ch >= '0' && ch <= '9' || ch >= 'A' && ch <= 'F' || ch >= 'a' && ch <= 'f') {AddCh(); goto case 85;} else if (ch == 39) {AddCh(); goto case 34;} else {goto case 0;} case 22: if (ch >= '0' && ch <= '9' || ch >= 'A' && ch <= 'F' || ch >= 'a' && ch <= 'f') {AddCh(); goto case 23;} else {goto case 0;} case 23: if (ch >= '0' && ch <= '9' || ch >= 'A' && ch <= 'F' || ch >= 'a' && ch <= 'f') {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 19;} 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: if (ch >= '0' && ch <= '9' || ch >= 'A' && ch <= 'F' || ch >= 'a' && ch <= 'f') {AddCh(); goto case 29;} else {goto case 0;} 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: if (ch >= '0' && ch <= '9' || ch >= 'A' && ch <= 'F' || ch >= 'a' && ch <= 'f') {AddCh(); goto case 31;} else {goto case 0;} case 31: if (ch >= '0' && ch <= '9' || ch >= 'A' && ch <= 'F' || ch >= 'a' && ch <= 'f') {AddCh(); goto case 32;} else {goto case 0;} case 32: if (ch >= '0' && ch <= '9' || ch >= 'A' && ch <= 'F' || ch >= 'a' && ch <= 'f') {AddCh(); goto case 33;} else {goto case 0;} case 33: if (ch >= '0' && ch <= '9' || ch >= 'A' && ch <= 'F' || ch >= 'a' && ch <= 'f') {AddCh(); goto case 19;} else {goto case 0;} case 34: {t.kind = 4; break;} case 35: if (ch <= 9 || ch >= 11 && ch <= 12 || ch >= 14 && ch <= '!' || ch >= '#' && ch <= '[' || ch >= ']' && ch <= 65535) {AddCh(); goto case 35;} else if (ch == '"') {AddCh(); goto case 51;} else if (ch == 92) {AddCh(); goto case 87;} 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 <= 9 || ch >= 11 && ch <= 12 || ch >= 14 && ch <= '!' || ch >= '#' && ch <= '/' || ch >= ':' && ch <= '@' || ch >= 'G' && ch <= '[' || ch >= ']' && ch <= '`' || ch >= 'g' && ch <= 65535) {AddCh(); goto case 35;} else if (ch >= '0' && ch <= '9' || ch >= 'A' && ch <= 'F' || ch >= 'a' && ch <= 'f') {AddCh(); goto case 88;} else if (ch == '"') {AddCh(); goto case 51;} else if (ch == 92) {AddCh(); goto case 87;} else {goto case 0;} case 38: if (ch >= '0' && ch <= '9' || ch >= 'A' && ch <= 'F' || ch >= 'a' && ch <= 'f') {AddCh(); goto case 39;} else {goto case 0;} case 39: if (ch >= '0' && ch <= '9' || ch >= 'A' && ch <= 'F' || ch >= 'a' && ch <= 'f') {AddCh(); goto case 40;} else {goto case 0;} case 40: if (ch >= '0' && ch <= '9' || ch >= 'A' && ch <= 'F' || ch >= 'a' && ch <= 'f') {AddCh(); goto case 41;} else {goto case 0;} case 41: if (ch >= '0' && ch <= '9' || ch >= 'A' && ch <= 'F' || ch >= 'a' && ch <= 'f') {AddCh(); goto case 35;} else {goto case 0;} case 42: if (ch >= '0' && ch <= '9' || ch >= 'A' && ch <= 'F' || ch >= 'a' && ch <= 'f') {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 >= '0' && ch <= '9' || ch >= 'A' && ch <= 'F' || ch >= 'a' && ch <= 'f') {AddCh(); goto case 48;} else {goto case 0;} case 48: if (ch >= '0' && ch <= '9' || ch >= 'A' && ch <= 'F' || ch >= 'a' && ch <= 'f') {AddCh(); goto case 49;} else {goto case 0;} case 49: if (ch >= '0' && ch <= '9' || ch >= 'A' && ch <= 'F' || ch >= 'a' && ch <= 'f') {AddCh(); goto case 35;} else {goto case 0;} case 50: if (ch <= '!' || ch >= '#' && ch <= 65535) {AddCh(); goto case 50;} else if (ch == '"') {AddCh(); goto case 90;} else {goto case 0;} case 51: {t.kind = 5; break;} case 52: {t.kind = 41; break;} case 53: {t.kind = 42; break;} case 54: {t.kind = 43; break;} case 55: {t.kind = 47; break;} case 56: {t.kind = 49; break;} case 57: {t.kind = 50; break;} case 58: {t.kind = 51; break;} case 59: {t.kind = 52; break;} case 60: {t.kind = 53; break;} case 61: {t.kind = 55; break;} case 62: {t.kind = 58; break;} case 63: {t.kind = 62; break;} case 64: {t.kind = 63; break;} case 65: {t.kind = 64; break;} case 66: {t.kind = 65; break;} case 67: {t.kind = 66; break;} case 68: recEnd = pos; recKind = 2; if (ch >= '0' && ch <= '9') {AddCh(); goto case 68;} else if (ch == 'U') {AddCh(); goto case 76;} else if (ch == 'u') {AddCh(); goto case 77;} else if (ch == 'L') {AddCh(); goto case 78;} else if (ch == 'l') {AddCh(); goto case 79;} else if (ch == '.') {apx++; AddCh(); goto case 91;} else if (ch == 'E' || ch == 'e') {AddCh(); goto case 14;} else if (ch == 'D' || ch == 'F' || ch == 'M' || ch == 'd' || ch == 'f' || ch == 'm') {AddCh(); goto case 17;} else {t.kind = 2; break;} case 69: if (ch >= 'A' && ch <= 'Z' || ch == '_' || ch >= 'a' && ch <= 'z') {AddCh(); goto case 1;} else if (ch == '"') {AddCh(); goto case 50;} else {goto case 0;} case 70: recEnd = pos; recKind = 2; if (ch >= '0' && ch <= '9') {AddCh(); goto case 68;} else if (ch == 'U') {AddCh(); goto case 76;} else if (ch == 'u') {AddCh(); goto case 77;} else if (ch == 'L') {AddCh(); goto case 78;} else if (ch == 'l') {AddCh(); goto case 79;} else if (ch == '.') {apx++; AddCh(); goto case 91;} else if (ch == 'X' || ch == 'x') {AddCh(); goto case 3;} else if (ch == 'E' || ch == 'e') {AddCh(); goto case 14;} else if (ch == 'D' || ch == 'F' || ch == 'M' || ch == 'd' || ch == 'f' || ch == 'm') {AddCh(); goto case 17;} else {t.kind = 2; break;} case 71: recEnd = pos; recKind = 45; if (ch >= '0' && ch <= '9') {AddCh(); goto case 6;} else {t.kind = 45; break;} case 72: recEnd = pos; recKind = 40; if (ch == '=') {AddCh(); goto case 92;} else {t.kind = 40; break;} case 73: recEnd = pos; recKind = 48; if (ch == '=') {AddCh(); goto case 56;} else {t.kind = 48; break;} case 74: recEnd = pos; recKind = 54; if (ch == '=') {AddCh(); goto case 61;} else {t.kind = 54; break;} case 75: recEnd = pos; recKind = 59; if (ch == '=') {AddCh(); goto case 62;} else {t.kind = 59; break;} case 76: recEnd = pos; recKind = 2; if (ch == 'L' || ch == 'l') {AddCh(); goto case 5;} else {t.kind = 2; break;} case 77: recEnd = pos; recKind = 2; if (ch == 'L' || ch == 'l') {AddCh(); goto case 5;} else {t.kind = 2; break;} case 78: recEnd = pos; recKind = 2; if (ch == 'U' || ch == 'u') {AddCh(); goto case 5;} else {t.kind = 2; break;} case 79: recEnd = pos; recKind = 2; if (ch == 'U' || ch == 'u') {AddCh(); goto case 5;} else {t.kind = 2; break;} case 80: recEnd = pos; recKind = 2; if (ch == 'L' || ch == 'l') {AddCh(); goto case 5;} else {t.kind = 2; break;} case 81: recEnd = pos; recKind = 2; if (ch == 'L' || ch == 'l') {AddCh(); goto case 5;} else {t.kind = 2; break;} case 82: recEnd = pos; recKind = 2; if (ch == 'U' || ch == 'u') {AddCh(); goto case 5;} else {t.kind = 2; break;} case 83: recEnd = pos; recKind = 2; if (ch == 'U' || ch == 'u') {AddCh(); goto case 5;} else {t.kind = 2; break;} case 84: if (ch == '"' || ch == 39 || ch == '0' || ch == 92 || ch >= 'a' && ch <= 'b' || ch == 'f' || ch == 'n' || ch == 'r' || ch == 't' || ch == 'v') {AddCh(); goto case 19;} else if (ch == 'x') {AddCh(); goto case 20;} else if (ch == 'u') {AddCh(); goto case 22;} else if (ch == 'U') {AddCh(); goto case 26;} else {goto case 0;} case 85: if (ch >= '0' && ch <= '9' || ch >= 'A' && ch <= 'F' || ch >= 'a' && ch <= 'f') {AddCh(); goto case 86;} else if (ch == 39) {AddCh(); goto case 34;} else {goto case 0;} case 86: if (ch >= '0' && ch <= '9' || ch >= 'A' && ch <= 'F' || ch >= 'a' && ch <= 'f') {AddCh(); goto case 19;} else if (ch == 39) {AddCh(); goto case 34;} else {goto case 0;} case 87: if (ch == '"' || ch == 39 || ch == '0' || ch == 92 || ch >= 'a' && ch <= 'b' || ch == 'f' || ch == 'n' || ch == 'r' || ch == 't' || ch == 'v') {AddCh(); goto case 35;} else if (ch == 'x') {AddCh(); goto case 36;} else if (ch == 'u') {AddCh(); goto case 38;} else if (ch == 'U') {AddCh(); goto case 42;} else {goto case 0;} case 88: if (ch >= '0' && ch <= '9' || ch >= 'A' && ch <= 'F' || ch >= 'a' && ch <= 'f') {AddCh(); goto case 89;} else if (ch <= 9 || ch >= 11 && ch <= 12 || ch >= 14 && ch <= '!' || ch >= '#' && ch <= '/' || ch >= ':' && ch <= '@' || ch >= 'G' && ch <= '[' || ch >= ']' && ch <= '`' || ch >= 'g' && ch <= 65535) {AddCh(); goto case 35;} else if (ch == '"') {AddCh(); goto case 51;} else if (ch == 92) {AddCh(); goto case 87;} else {goto case 0;} case 89: if (ch <= 9 || ch >= 11 && ch <= 12 || ch >= 14 && ch <= '!' || ch >= '#' && ch <= '[' || ch >= ']' && ch <= 65535) {AddCh(); goto case 35;} else if (ch == '"') {AddCh(); goto case 51;} else if (ch == 92) {AddCh(); goto case 87;} else {goto case 0;} case 90: recEnd = pos; recKind = 5; if (ch == '"') {AddCh(); goto case 50;} else {t.kind = 5; break;} case 91: if (ch <= '/' || ch >= ':' && ch <= 65535) {apx++; AddCh(); goto case 2;} else if (ch >= '0' && ch <= '9') {apx = 0; AddCh(); goto case 10;} else {goto case 0;} case 92: recEnd = pos; recKind = 46; if (ch == '=') {AddCh(); goto case 55;} else {t.kind = 46; break;} case 93: {t.kind = 71; break;} case 94: {t.kind = 72; break;} case 95: {t.kind = 73; break;} case 96: {t.kind = 74; break;} case 97: {t.kind = 75; break;} case 98: {t.kind = 76; break;} case 99: {t.kind = 77; break;} case 100: {t.kind = 78; break;} case 101: {t.kind = 79; break;} case 102: {t.kind = 80; break;} case 103: {t.kind = 81; break;} case 104: recEnd = pos; recKind = 39; if (ch == '=') {AddCh(); goto case 98;} else if (ch == '&') {AddCh(); goto case 103;} else {t.kind = 39; break;} case 105: recEnd = pos; recKind = 56; if (ch == '-') {AddCh(); goto case 54;} else if (ch == '=') {AddCh(); goto case 94;} else {t.kind = 56; break;} case 106: recEnd = pos; recKind = 44; if (ch == '=') {AddCh(); goto case 96;} else {t.kind = 44; break;} case 107: recEnd = pos; recKind = 61; if (ch == '+') {AddCh(); goto case 57;} else if (ch == '=') {AddCh(); goto case 93;} else {t.kind = 61; break;} case 108: recEnd = pos; recKind = 57; if (ch == '=') {AddCh(); goto case 97;} else {t.kind = 57; break;} case 109: recEnd = pos; recKind = 60; if (ch == '=') {AddCh(); goto case 99;} else if (ch == '|') {AddCh(); goto case 102;} else {t.kind = 60; break;} case 110: recEnd = pos; recKind = 67; if (ch == '=') {AddCh(); goto case 95;} else {t.kind = 67; break;} case 111: recEnd = pos; recKind = 68; if (ch == '=') {AddCh(); goto case 100;} else {t.kind = 68; break;} } t.val = new String(tval, 0, tlen); return t; }
// make sure that peeking starts at the current scan position public void ResetPeek() { pt = tokens; }
// 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; } }
// 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; }
/* Checks whether the next sequence of tokens is a qualident * * and returns the qualident string * * !!! Proceeds from current peek position !!! */ bool IsQualident(ref Token pt, out string qualident) { qualident = ""; if (pt.kind == _ident) { qualident = pt.val; pt = scanner.Peek(); while (pt.kind == _dot) { pt = scanner.Peek(); if (pt.kind != _ident) return false; qualident += "." + pt.val; pt = scanner.Peek(); } return true; } else return false; }
// skip: { "[" { "," } "]" | "*" } // !!! Proceeds from current peek position !!! bool IsPointerOrDims(ref Token pt) { for (;;) { if (pt.kind == _lbrack) { do pt = scanner.Peek(); while (pt.kind == _comma); if (pt.kind != _rbrack) return false; } else if (pt.kind != _times) break; pt = scanner.Peek(); } return true; }
void Get() { for (;;) { t = la; la = scanner.Scan(); if (la.kind <= maxT) { ++errDist; break; } la = t; } }
public void Parse() { la = new Token(); la.val = ""; Get(); JavaScript(); Expect(0); }