javascanner.java

MSP IM Source code 2008 update

package se.sics.mspsim.extutil.highlight;
// Public domain, no restrictions, Ian Holyer, University of Bristol.

/**
 * <p>
 * Provide a hand-written scanner for the Java language.
 */

public class JavaScanner extends Scanner {

  // The version of Java supported.
  private int version = 15;

  private boolean debug = false;

  /** Create a Java scanner, for Java version 1.5 by default. */
  public JavaScanner() {
    super();
    initKind();
    initUniKind();
  }

  /** Create a Java scanner, for a given version between "1.1" and "1.5". */
  public JavaScanner(String version) {
    super();
    initKind();
    initUniKind();
    if (version.equals("1.1"))
      this.version = 11;
    else if (version.equals("1.2"))
      this.version = 12;
    else if (version.equals("1.3"))
      this.version = 13;
    else if (version.equals("1.4"))
      this.version = 14;
    else if (version.equals("1.5"))
      this.version = 15;
    else
      throw new Error("Unknown version of Java: " + version);
  }

  /** Override the read method from the Scanner class. */
  protected int read() {
    int type, saveStart = 0;
    if (debug)
      saveStart = start;

    if (start >= end)
      return WHITESPACE;

    switch (state) {
    case MID_COMMENT:
    case END_COMMENT:
      type = readComment(MID_COMMENT);
      if (type == END_COMMENT)
        state = WHITESPACE;
      else
        state = MID_COMMENT;
      return type;
    default:
      char c = buffer[start];
      if (c == '\\')
        c = next();
      if (c < 128)
        type = kind[c];
      else
        type = unikind[Character.getType(c)];
      switch (type) {
      case WHITESPACE:
        start = start + charlength;
        charlength = 1;
        while (start < end) {
          c = buffer[start];
          if (c == '\\')
            c = next();
          int k;
          if (c < 128)
            k = kind[c];
          else
            k = unikind[Character.getType(c)];
          if (k != WHITESPACE)
            break;
          start = start + charlength;
          charlength = 1;
        }
        break;
      case UNRECOGNIZED:
      case BRACKET:
      case SEPARATOR:
        start = start + charlength;
        charlength = 1;
        break;
      case OPERATOR:
        start = start + charlength;
        charlength = 1;
        type = readOperator(c);
        break;
      case CHARACTER:
        start = start + charlength;
        charlength = 1;
        type = readCharLiteral();
        break;
      case STRING:
        start = start + charlength;
        charlength = 1;
        type = readStringLiteral();
        break;
      case IDENTIFIER:
        start = start + charlength;
        charlength = 1;
        while (start < end) {
          c = buffer[start];
          if (c == '\\')
            c = next();
          int k;
          if (c < 128)
            k = kind[c];
          else
            k = unikind[Character.getType(c)];
          if (k != IDENTIFIER && k != NUMBER)
            break;
          start = start + charlength;
          charlength = 1;
        }
        break;
      case NUMBER:
        start = start + charlength;
        charlength = 1;
        type = readNumber(c);
        break;
      case PUNCTUATION:
        start = start + charlength;
        charlength = 1;
        type = readDot();
        break;
      case COMMENT:
        start = start + charlength;
        charlength = 1;
        type = readSlash();
        if (type == START_COMMENT)
          state = MID_COMMENT;
        break;
      }
    }
    if (debug) {
      System.out.print(TokenTypes.typeNames[type]);
      System.out.println(" " + saveStart + "," + end + "("
          + (start - saveStart) + ")");
    }
    return type;
  }

  private int readOperator(char c) {
    if (start >= end)
      return OPERATOR;
    char c2;

    switch (c) {
    case '~':
    case '?':
    case ':':
      break;
    case '+':
    case '-':
    case '&':
    case '|':
      c2 = buffer[start];
      if (c2 == '\\')
        c2 = next();
      if (c2 != c && c2 != '=')
        break;
      start = start + charlength;
      charlength = 1;
      break;
    case '=':
    case '*':
    case '!':
    case '^':
    case '%':
    case '/':
      c2 = buffer[start];
      if (c2 == '\\')
        c2 = next();
      if (c2 != '=')
        break;
      start = start + charlength;
      charlength = 1;
      break;
    case '<':
    case '>':
      c2 = buffer[start];
      if (c2 == '\\')
        c2 = next();
      if (c2 == '=') {
        start = start + charlength;
        charlength = 1;
      } else if (c2 == c) {
        start = start + charlength;
        charlength = 1;
        if (start >= end)
          break;
        char c3 = buffer[start];
        if (c3 == '\\')
          c3 = next();
        if (c3 == '=') {
          start = start + charlength;
          charlength = 1;
        } else if (c == '>' && c3 == '>') // >>>
        {
          start = start + charlength;
          charlength = 1;
          if (start >= end)
            break;
          char c4 = buffer[start];
          if (c4 == '\\')
            c4 = next();
          if (c4 != '=')
            break;
          start = start + charlength;
          charlength = 1;
        }
      }
      break;
    }
    return OPERATOR;
  }

  private int readCharLiteral() {
    if (start >= end)
      return bad(CHARACTER);
    char c2 = buffer[start];
    if (c2 == '\\')
      c2 = next();

    switch (c2) {
    case '\\':
      start = start + charlength;
      charlength = 1;
      boolean ok = readEscapeSequence();
      if (!ok)
        return bad(CHARACTER);
      break;
    case '\'':
    case '\n':
      return bad(CHARACTER);
    default:
      start = start + charlength;
      charlength = 1;
      break;
    }
    if (start >= end)
      return bad(CHARACTER);
    char c3 = buffer[start];
    if (c3 == '\\')
      c3 = next();
    if (c3 != '\'')
      return bad(CHARACTER);
    start = start + charlength;
    charlength = 1;
    return CHARACTER;
  }

  private int readStringLiteral() {
    if (start >= end)
      return bad(STRING);
    char c = buffer[start];
    if (c == '\\')
      c = next();

    while (c != '"') {
      switch (c) {
      case '\\':
        start = start + charlength;
        charlength = 1;
        boolean ok = readEscapeSequence();
        if (!ok)
          return bad(STRING);
        break;
      case '\n':
        return bad(STRING);
      default:
        start = start + charlength;
        charlength = 1;
        if (start >= end)
          return bad(STRING);
        break;
      }
      c = buffer[start];
      if (c == '\\')
        c = next();
    }
    if (c != '"')
      return bad(STRING);
    start = start + charlength;
    charlength = 1;
    return STRING;
  }

  private int readSlash() {
    if (start >= end)
      return OPERATOR;
    char c = buffer[start];
    if (c == '\\')
      c = next();
    if (c == '/') {
      while (c != '\n') {
        start = start + charlength;
        charlength = 1;
        if (start >= end)
          return COMMENT;
        c = buffer[start];
        if (c == '\\')
          c = next();
      }
      start = start + charlength;
      charlength = 1;
      return COMMENT;
    } else if (c == '*') {
      start = start + charlength;
      charlength = 1;
      return readComment(START_COMMENT);
    }
    return readOperator('/');
  }

  // Read one line of a /*...*/ comment, given the expected type
  int readComment(int type) {
    if (start >= end)
      return type;
    char c = buffer[start];
    if (c == '\\')
      c = next();

    while (true) {
      while (c != '*' && c != '\n') {
        start = start + charlength;
        charlength = 1;
        if (start >= end)
          return type;
        c = buffer[start];
        if (c == '\\')
          c = next();
      }
      start = start + charlength;
      charlength = 1;
      if (c == '\n')
        return type;
      if (start >= end)
        return type;
      c = buffer[start];
      if (c == '\\')
        c = next();
      if (c == '/') {
        start = start + charlength;
        charlength = 1;
        if (type == START_COMMENT) {
          return COMMENT;
        }
        return END_COMMENT;
      }
    }
  }

  // Read a number, without checking whether it is out of range
  // Doesn't deal with e.g. 0777.9 or 07779f
  private int readNumber(char c) {
    if (c == '0') {
      int saveStart = start, saveLength = charlength;
      start = start + charlength;
      charlength = 1;
      if (start >= end)
        return NUMBER;
      char c2 = buffer[start];
      if (c2 == '\\')
        c2 = next();
      switch (c2) {
      case 'x':
      case 'X':
        start = start + charlength;
        charlength = 1;
        boolean ok = readDigits(16);
        if (!ok)
          return bad(NUMBER);
        readSuffix();
        return NUMBER;
      case 0:
      case 1:
      case 2:
      case 3:
      case 4:
      case 5:
      case 6:
      case 7:
        readDigits(8);
        readSuffix();
        return NUMBER;
      case '.':
      case 'e':
      case 'E':
        start = saveStart;
        charlength = saveLength;
        break;
      case 'f':
      case 'F':
      case 'd':
      case 'D':
        start = start + charlength;
        charlength = 1;
        return NUMBER;
      case 'l':
      case 'L':
        start = start + charlength;
        charlength = 1;
        return NUMBER;
      }
    }
    boolean hasDigits = false;
    if ('0' <= c && c <= '9') {
      hasDigits = true;
      readDigits(10);
      if (start >= end)
        return NUMBER;
      c = buffer[start];
      if (c == '\\')
        c = next();
      if (c == 'l' || c == 'L') {
        start = start + charlength;
        charlength = 1;
        return NUMBER;
      }
    }
    if (c == '.') {
      start = start + charlength;
      charlength = 1;
      if (start >= end)
        return NUMBER;
      c = buffer[start];
      if (c == '\\')
        c = next();
      if ('0' <= c && c <= '9') {
        hasDigits = true;
        readDigits(10);
        if (start >= end)
          return NUMBER;
        c = buffer[start];
        if (c == '\\')
          c = next();
      }
    }
    if (!hasDigits)
      return bad(NUMBER);
    switch (c) {
    case 'e':
    case 'E':
      start = start + charlength;
      charlength = 1;
      if (start >= end)
        return bad(NUMBER);
      c = buffer[start];
      if (c == '\\')
        c = next();
      if (c == '+' || c == '-') {
        start = start + charlength;
        charlength = 1;
        if (start >= end)
          return bad(NUMBER);
        c = buffer[start];
        if (c == '\\')
          c = next();
      }
      readDigits(10);
      break;
    case 'f':
    case 'F':
    case 'd':
    case 'D':
      start = start + charlength;
      charlength = 1;
      return NUMBER;
    }
    return NUMBER;
  }