ana.java

用DFA方法写得词法分析器 和大家交流一下

import java.io.BufferedReader;
import java.io.IOException;
import java.io.StringReader;
import java.util.HashMap;

/*
 * this class is used for word ananalyze.
 * author:贺静
 */
public class Ana {
	public static char charToken;

	public static int lineNo, c;

	public static String token;

	private String outputText;

	private int state;

	private BufferedReader input;

	private HashMap table;

	private boolean isOper(char a) {
		if (table.containsKey(String.valueOf(a))) {
			return true;
		}
		return false;
	}

	/*
	 * i is the ASCII of the char if i is between 97 and 122, i is a-z; if i is
	 * between 65 and 90, i is A-Z;
	 */
	private boolean isLetter(char a) {
		int i = (int) a;
		if ((97 <= i & i <= 122) || (65 <= i & i <= 90)) {
			return true;
		}
		return false;
	}

	/*
	 * the ASCII of digits 0-9 is between 48 and 57
	 */
	private boolean isDigit(char a) {
		int i = (int) a;
		if (48 <= i & i <= 57) {
			return true;
		}
		return false;
	}

	// switch the state
	private void reSet() throws IOException {
		switch (state) {
		case 0:
			state = 8;
			break;
		case 8:
			state = 11;
			break;
		case 11:
			state = 14;
			break;
		default:
		}
	}

	// refresh the input,"@" 终止符
	public void setBufferInput(String s) {
		input = new BufferedReader(new StringReader(s + "@"));
	}

	/*
	 * the main method of word annalyze 0:operator; 8:key word or identify;
	 * 11:number
	 */
	public void analysis() throws IOException {
		state = 0;
		outputText = "";
		Error error = new Error();
		table = new HashMap(new Hashmap().getHashMap());
		c = input.read();
		lineNo = 1;
		token = " ";
		int lineNO = 0;
		charToken = (char) c;
		String token2 = " ";
		while (c != -1) {
			switch (state) {
			case 0:
				if (charToken == ' ')
					;
				else if (charToken == '\n')
					lineNo++;
				else if (charToken == '+' || charToken == '>'
						|| charToken == '[' || charToken == ']'
						|| charToken == '(' || charToken == ')'
						|| charToken == '{' || charToken == '}'
						|| charToken == ';') {
					token = String.valueOf(charToken);
					outputText += "LINE " + lineNo + ":" + "   " + token + "\n";
				} else if (charToken == '<')
					state = 1;
				else if (charToken == '=')
					state = 2;
				else if (charToken == '-')
					state = 3;
				else if (charToken == '/')
					state = 4;
				else if (charToken == '*')
					state = 7;
				else {
					reSet();
					break;
				}
				c = input.read();
				charToken = (char) c;
				break;
			case 1:
				if (charToken == '>') {
					token = "<>";
					outputText += "LINE " + lineNo + ":" + "   " + token + "\n";
					c = input.read();
					charToken = (char) c;
				} else {
					token = "<";
					outputText += "LINE " + lineNo + ":" + "   " + token + "\n";
				}
				state = 0;
				break;
			case 2:
				if (charToken == '=') {
					token = "==";
					c = input.read();
					charToken = (char) c;
					outputText += "LINE " + lineNo + ":" + "   " + token + "\n";
				} else {
					token = "=";
					outputText += "LINE " + lineNo + ":" + "   " + token + "\n";
				}
				state = 0;
				break;
			case 3:
				if (isDigit(charToken) & !isLetter(token.charAt(0))
						& !isDigit(token.charAt(0))) {
					token = "<->";
					state = 11;
				} else {
					token = "-";
					outputText += "LINE " + lineNo + ":" + "   " + token + "\n";
					state = 0;
				}
				break;
			case 4:
				if (charToken == '*') {
					state = 5;
					c = input.read();
					charToken = (char) c;
				} else {
					token = "/";
					outputText += "LINE " + lineNo + ":" + "   " + token + "\n";
					state = 0;
				}
				break;
			case 5:
				if (charToken == '*')
					state = 6;
				else {
					if (charToken == '\n')
						lineNO++;
					state = 5;
				}
				c = input.read();
				charToken = (char) c;
				break;
			case 6:
				if (charToken == '/') {
					lineNo = lineNo + lineNO;
					lineNO = 0;
					state = 0;
				} else if (charToken == '*')
					state = 6;
				else {
					if (charToken == '\n')
						lineNO++;
					state = 5;
				}
				c = input.read();
				charToken = (char) c;
				break;
			case 7:
				if (charToken == '/') {
					outputText += error.getError(input, lineNo, token, c,
							charToken, 0);
				} else {
					token = "*";
					outputText += "LINE " + lineNo + ":" + "   " + token + "\n";
				}
				state = 0;
				break;
			case 8:
				if (isLetter(charToken)) {
					state = 9;
					token2 = String.valueOf(charToken);
					c = input.read();
					charToken = (char) c;
				} else
					reSet();
				break;
			case 9:
				if (isLetter(charToken) || isDigit(charToken)
						|| charToken == '_') {
					token2 += String.valueOf(charToken);
					state = 9;
					c = input.read();
					charToken = (char) c;
				} else
					state = 10;
				break;
			case 10:
				token = token2;
				if (token2.endsWith("_")) {
					outputText += error.getError(input, lineNo, token, c,
							charToken, 1);
				} else {
					if (!table.containsKey(token2))
						table.put(token, "ID");
					outputText += "LINE " + lineNo + ":  " + table.get(token)
							+ ": " + token + "\n";
				}
				state = 0;
				break;
			case 11:
				if (isDigit(charToken)) {
					state = 12;
					token2 = String.valueOf(charToken);
					c = input.read();
					charToken = (char) c;
				} else
					reSet();
				break;
			case 12:
				if (isDigit(charToken) || charToken == '.') {
					token2 += String.valueOf(charToken);
					state = 12;
					c = input.read();
					charToken = (char) c;
				} else
					state = 13;
				break;
			case 13:
				if (isLetter(charToken)) {
					token = token2 + charToken;
					outputText += error.getError(input, lineNo, token, c,
							charToken, 2);
				} else {
					try {
						if (token2.contains(".")) {
							if (token2.endsWith(".")) {
								token = token2;
								outputText += error.getError(input, lineNo,
										token, c, charToken, 3);
								state = 0;
								break;
							} else {
								double num = Double.parseDouble(token2);
								token2 = "" + num;
							}
						} else {
							int num = Integer.parseInt(token2);
							token2 = "" + num;
						}
						if (token.equals("<->")) {
							token = "-" + token2;
						} else
							token = token2;
						if (!table.containsKey(token))
							table.put(token, "NUM");
						outputText += "LINE " + lineNo + ":  "
								+ table.get(token) + ": " + token + "\n";
					} catch (NumberFormatException nfe) {
						token = token2;
						outputText += error.getError(input, lineNo, token, c,
								charToken, 6);
						state = 0;
						break;
					}
				}
				state = 0;
				break;
			case 14:
				if (isOper(charToken)) {
					state = 0;
				} else if (isLetter(charToken)) {
					state = 8;
				} else if (isDigit(charToken))
					state = 11;
				else
					state = 15;
				break;
			case 15:// 16：error
				outputText += error.getError(input, lineNo, token, c,
						charToken, 4);
				state = 0;
				break;
			}
		}
		if (state == 5 || state == 6) {
			outputText += error.getError(input, lineNo, token, c, charToken, 5)
					+ "  END";
		}
	}

	// return the output text
	public String getOutput() {
		return outputText;
	}

}