package mir.util;
-import gnu.regexp.RE;
-import gnu.regexp.REException;
-import gnu.regexp.REMatch;
-import multex.Exc;
import multex.Failure;
+import org.apache.oro.text.regex.*;
/**
* Simple parser. Can be used to parse a <code>String</code> based using
* regular epxression.
*/
public class SimpleParser {
- private String data;
private int position;
+ private Perl5Compiler compiler = new Perl5Compiler();
+ private Perl5Matcher matcher = new Perl5Matcher();
+ private PatternMatcherInput input;
+
+
/**
- * Initialize a new <code>SimpleParser</code>, with <code>aData</code>
+ * Initialize a new <code>SimpleParser</code>, with <code>someData</code>
* as the text to parse.
+ *
+ * @param someData the text to parse
*/
- public SimpleParser(String aData) {
- data=aData;
- position=0;
+ public SimpleParser(String someData) {
+ position = 0;
+ input = new PatternMatcherInput(someData);
}
/**
* Parses a regular expression. Uses suppled <code>aMessage</code> parameter
* in case of an error.
+ *
+ * @param aRegularExpression the expression to find at the cursor position
+ * @param aMessage the error message to apply if the regular expression couldn't be found
+ *
+ * @return the text that matched the regular expression
+ *
+ * @throws mir.util.SimpleParser.UnExpectedTokenExc
+ * if the token wasn't found. <code>aMessage</code> will be used in the
+ * exception's message
*/
- public String parse(RE aRegularExpression, String aMessage) throws SimpleParserExc {
- REMatch match = aRegularExpression.getMatch(data, position);
+ public String parse(Pattern aRegularExpression, String aMessage) throws UnExpectedTokenExc {
+ input.setCurrentOffset(position);
+ if ( matcher.contains(input, aRegularExpression) && input.getMatchBeginOffset() == position) {
+ position = input.getMatchEndOffset();
- if (match==null || match.getStartIndex()!=position) {
- throw new SimpleParserExc(aMessage+" at position "+position+" in '"+data+"'");
+ return input.match();
+ }
+ else {
+ throw new UnExpectedTokenExc(aMessage + " at position " +
+ position + " in '" + input.toString() + "'");
}
-
- position=match.getEndIndex();
-
- return match.toString();
}
/**
* Attempts to match the data right after the parsing position to
* a regular expression
*
- * @throws SimpleParserExc if no match could be found for the given regular
+ * @throws UnExpectedTokenExc if no match could be found for the given regular
* expression.
*/
- public String parse(RE aRegularExpression) throws SimpleParserExc {
- return parse( aRegularExpression, "No match found for '"+aRegularExpression.toString()+"'");
+ public String parse(Pattern aRegularExpression) throws UnExpectedTokenExc {
+ return parse( aRegularExpression, "No match found for '" + aRegularExpression.getPattern() + "'");
}
/**
- * Attempts to skip data at the parsing position matching the supplied regular expression
+ * Attempts to skip data at the parsing position matching the supplied regular expression.
+ * If no match is found, the method will simply return
*
- * @throws SimpleParserExc if no match could be found for the given regular
- * expression.
+ * @param aRegularExpression the expression to find and skip at the parsing position.
*/
- public void skip(RE aRegularExpression) throws SimpleParserExc {
- REMatch match = aRegularExpression.getMatch(data, position);
-
- if (match!=null && match.getStartIndex()==position)
- position=match.getEndIndex();
+ public void skip(Pattern aRegularExpression) {
+ input.setCurrentOffset(position);
+
+ if (matcher.contains(input, aRegularExpression) &&
+ input.getMatchBeginOffset() == position) {
+ position = input.getMatchEndOffset();
+ }
}
/**
* Returns <code>true</code> if the data at the parsing position matches the given
* regular expression.
*/
- public boolean parses(String anExpression) throws SimpleParserExc {
+ public boolean parses(String anExpression) throws MalformedRegularExpressionExc {
try {
- return parses(new RE(anExpression));
- }
- catch (SimpleParserExc e) {
- throw e;
- }
- catch (REException e) {
- throw new SimpleParserFailure( "Error compiling regular expression '" + anExpression + "': " + e.getMessage(), e);
+ return parses(compiler.compile(anExpression));
}
- catch (Throwable t) {
- throw new SimpleParserFailure( t );
+ catch (MalformedPatternException e) {
+ throw new MalformedRegularExpressionExc(e);
}
}
/**
- * Returns <code>true</code> if the data at the parsing position matches the given
+ * Tests whether the data at the parsing position matches a given
* regular expression.
+ *
+ * @param aRegularExpression the pattern to look for
+ * @return returns <code>true</code> if the regular expression is found at the
+ * parsing position
*/
- public boolean parses(RE aRegularExpression) throws SimpleParserExc {
- REMatch match = aRegularExpression.getMatch(data, position);
+ public boolean parses(Pattern aRegularExpression) {
+ input.setCurrentOffset(position);
- return (match!=null && match.getStartIndex()==position) ;
+ return matcher.contains(input, aRegularExpression) && input.getMatchBeginOffset() == position;
}
/**
* @throws SimpleParserExc if no match could be found for the given regular
* expression.
*/
- public String parse(String anExpression) throws SimpleParserExc, SimpleParserFailure {
+ public String parse(String anExpression) throws MalformedRegularExpressionExc, UnExpectedTokenExc {
try {
- return parse(new RE(anExpression));
- }
- catch (SimpleParserExc e) {
- throw e;
- }
- catch (REException e) {
- throw new SimpleParserFailure( "Error compiling regular expression '" + anExpression + "': " + e.getMessage(), e);
+ return parse(compiler.compile(anExpression));
}
- catch (Throwable t) {
- throw new SimpleParserFailure( t );
+ catch (MalformedPatternException e) {
+ throw new MalformedRegularExpressionExc(e);
}
}
* Parses text at the current parsing position matching the given <code>anExpression</code>.
* Will apply the supplied message to an exception if no match could be found.
*/
- public String parse(String anExpression, String aMessage) throws SimpleParserExc, SimpleParserFailure {
+ public String parse(String anExpression, String aMessage) throws MalformedRegularExpressionExc, UnExpectedTokenExc {
try {
- return parse(new RE(anExpression), aMessage);
+ return parse(compiler.compile(anExpression), aMessage);
}
- catch (SimpleParserExc e) {
- throw e;
- }
- catch (REException e) {
- throw new SimpleParserFailure( "Error compiling regular expression '" + anExpression + "': " + e.getMessage(), e);
- }
- catch (Throwable t) {
- throw new SimpleParserFailure( t );
+ catch (MalformedPatternException e) {
+ throw new MalformedRegularExpressionExc(e);
}
}
* Skips (i.e. discards) text matching the supplied regular expression
*/
- public void skip(String anExpression) throws SimpleParserExc, SimpleParserFailure {
+ public void skip(String anExpression) throws MalformedRegularExpressionExc {
try {
- skip(new RE(anExpression));
- }
- catch (SimpleParserExc e) {
- throw e;
+ skip(compiler.compile(anExpression));
}
- catch (REException e) {
- throw new SimpleParserFailure( "Error compiling regular expression '" + anExpression + "': " + e.getMessage(), e);
- }
- catch (Throwable t) {
- throw new SimpleParserFailure( t );
+ catch (MalformedPatternException e) {
+ throw new MalformedRegularExpressionExc(e);
}
}
* returns <code>true</code> if the parser is at the end of the data
*/
public boolean isAtEnd() {
- return position>=data.length();
+ return position >=input.getEndOffset();
}
/**
* Returns the "remaining" data: the data after the current position
*/
public String remainingData() {
- return data.substring(position);
+ return input.toString().substring(input.getEndOffset());
}
/**
/**
* Checked exception thrown from <code>SimpleParser</code>
*/
- public static class SimpleParserExc extends Exc {
+ public static class SimpleParserExc extends Exception {
+ protected SimpleParserExc(String aMessage, Throwable aCause) {
+ super(aMessage, aCause);
+ }
+
+ public SimpleParserExc(Throwable aCause) {
+ super(aCause);
+ }
+
public SimpleParserExc(String aMessage) {
super(aMessage);
}
}
+
+ public static class MalformedRegularExpressionExc extends SimpleParserExc {
+ public MalformedRegularExpressionExc(String aMessage, Throwable aCause) {
+ super(aMessage, aCause);
+ }
+
+ public MalformedRegularExpressionExc(Throwable aCause) {
+ super(aCause);
+ }
+ }
+
+ public static class UnExpectedTokenExc extends SimpleParserExc {
+ public UnExpectedTokenExc(String aMessage) {
+ super(aMessage);
+ }
+ }
+
}
\ No newline at end of file