/*****************************************************************************
* Copyright (C) Codehaus.org *
* ------------------------------------------------------------------------- *
* Licensed under the Apache License, Version 2.0 (the "License"); *
* you may not use this file except in compliance with the License. *
* You may obtain a copy of the License at *
* *
* http://www.apache.org/licenses/LICENSE-2.0 *
* *
* Unless required by applicable law or agreed to in writing, software *
* distributed under the License is distributed on an "AS IS" BASIS, *
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. *
* See the License for the specific language governing permissions and *
* limitations under the License. *
*****************************************************************************/
package net.ion.rosetta;
import static net.ion.rosetta.util.Checks.checkArgument;
import java.io.IOException;
import java.nio.CharBuffer;
import java.util.Collection;
import java.util.List;
import java.util.concurrent.atomic.AtomicReference;
import net.ion.rosetta.annotations.Private;
import net.ion.rosetta.error.ParserException;
import net.ion.rosetta.functors.Map;
import net.ion.rosetta.functors.Map2;
import net.ion.rosetta.functors.Maps;
import net.ion.rosetta.util.Checks;
/**
* Defines grammar and encapsulates parsing logic. A {@link Parser} takes as input a {@link CharSequence} source and parses it when the {@link #parse(CharSequence)} method is called. A value of type {@code T} will be returned if parsing succeeds, or a {@link ParserException} is thrown to indicate parsing error. For example:
*
* <pre>
* Parser<String> scanner = Scanners.IDENTIFIER;
* assertEquals("foo", scanner.parse("foo"));
* </pre>
*
* <p>
* {@code Parser}s are immutable and inherently covariant on the type parameter {@code T}. Because Java generics has no native support for covariant type parameter, a workaround is to use the {@link Parser#cast()} method to explicitly force covariance whenever needed.
*
* <p>
* {@code Parser}s run either on character level to scan the source, or on token level to parse a list of {@link Token} objects returned from another parser. This other parser that returns the list of tokens for token level parsing is hooked up via the {@link #from(Parser)} or {@link #from(Parser, Parser)} method.
*
* <p>
* The following are important naming conventions used throughout the library:
* <ul>
* <li>A character level parser object that recognizes a single lexical word is called a scanner.
* <li>A scanner that translates the recognized lexical word into a token is called a tokenizer.
* <li >A character level parser object that does lexical analysis and returns a list of {@link Token} is called a lexer.
* <li>All {@code index} parameters are 0-based indexes in the original source.
* </ul>
*
* @author Ben Yu
*/
public abstract class Parser<T> {
Parser() {
}
/**
* An atomic mutable reference to {@link Parser}. Is useful to work around circular dependency between parser objects.
*
* <p>
* Example usage:
*
* <pre>
* Parser.Reference<Foo> ref = Parser.newReference();
* ...
* Parser<Bar> barParser = barParser(ref.lazy());
* Parser<Foo> fooParser = fooParser(barParser);
* ref.set(fooParser);
* </pre>
*/
public static final class Reference<T> extends AtomicReference<Parser<T>> {
private static final long serialVersionUID = -8778697271614979497L;
private final Parser<T> lazy = new LazyParser<T>(this);
/**
* A {@link Parser} that delegates to the parser object referenced by {@code this} during parsing time.
*/
public Parser<T> lazy() {
return lazy;
}
}
/** Creates a new instance of {@link Reference}. */
public static <T> Reference<T> newReference() {
return new Reference<T>();
}
/**
* A {@link Parser} that executes {@code this}, and returns {@code value} if succeeds.
*/
public final <R> Parser<R> retn(R value) {
return next(Parsers.constant(value));
}
/**
* A {@link Parser} that sequentially executes {@code this} and then {@code parser}. The return value of {@code parser} is preserved.
*/
public final <R> Parser<R> next(Parser<R> parser) {
return Parsers.sequence(this, parser);
}
/**
* A {@link Parser} that executes {@code this}, maps the result using {@code map} to another {@code Parser} object to be executed as the next step.
*/
public final <To> Parser<To> next(Map<? super T, ? extends Parser<? extends To>> map) {
return new BindNextParser<T, To>(this, map);
}
/**
* A {@link Parser} that sequentially executes {@code this} and then {@code parser}, whose return value is ignored.
*/
public final Parser<T> followedBy(Parser<?> parser) {
return Parsers.sequence(this, parser, InternalFunctors.<T, Object> firstOfTwo());
}
/**
* A {@link Parser} that succeeds if {@code this} succeeds and the pattern recognized by {@code parser} isn't following.
*/
public final Parser<T> notFollowedBy(Parser<?> parser) {
return followedBy(parser.not());
}
/**
* {@code p.many()} is equivalent to {@code p*} in EBNF. The return values are collected and returned in a {@link List}.
*/
public final Parser<List<T>> many() {
return atLeast(0);
}
/**
* {@code p.skipMany()} is equivalent to {@code p*} in EBNF. The return values are discarded.
*/
public final Parser<Void> skipMany() {
return skipAtLeast(0);
}
/**
* {@code p.many1()} is equivalent to {@code p+} in EBNF. The return values are collected and returned in a {@link List}.
*/
public final Parser<List<T>> many1() {
return atLeast(1);
}
/**
* {@code p.skipMany1()} is equivalent to {@code p+} in EBNF. The return values are discarded.
*/
public final Parser<Void> skipMany1() {
return skipAtLeast(1);
}
/**
* A {@link Parser} that runs {@code this} parser greedily for at least {@code min} times. The return values are collected and returned in a {@link List}.
*/
public final Parser<List<T>> atLeast(int min) {
return new RepeatAtLeastParser<T>(this, Checks.checkMin(min));
}
/**
* A {@link Parser} that runs {@code this} parser greedily for at least {@code min} times and ignores the return values.
*/
public final Parser<Void> skipAtLeast(int min) {
return new SkipAtLeastParser(this, Checks.checkMin(min));
}
/**
* A {@link Parser} that sequentially runs {@code this} for {@code n} times and ignores the return values.
*/
public final Parser<Void> skipTimes(int n) {
return skipTimes(n, n);
}
/**
* A {@link Parser} that runs {@code this} for {@code n} times and collects the return values in a {@link List}.
*/
public final Parser<List<T>> times(int n) {
return times(n, n);
}
/**
* A {@link Parser} that runs {@code this} parser for at least {@code min} times and up to {@code max} times. The return values are collected and returned in {@link List}.
*/
public final Parser<List<T>> times(int min, int max) {
Checks.checkMinMax(min, max);
return new RepeatTimesParser<T>(this, min, max);
}
/**
* A {@link Parser} that runs {@code this} parser for at least {@code min} times and up to {@code max} times, with all the return values ignored.
*/
public final Parser<Void> skipTimes(int min, int max) {
Checks.checkMinMax(min, max);
return new SkipTimesParser(this, min, max);
}
/**
* A {@link Parser} that runs {@code this} parser and transforms the return value using {@code map}.
*/
public final <R> Parser<R> map(Map<? super T, ? extends R> map) {
return new MapParser<T, R>(this, map);
}
/**
* {@code p1.or(p2)} is equivalent to {@code p1 | p2} in EBNF.
*
* @param alternative
* the alternative parser to run if this fails.
*/
@SuppressWarnings("unchecked")
public final Parser<T> or(Parser<? extends T> alternative) {
return Parsers.or(this, alternative);
}
/**
* {@code p.optional()} is equivalent to {@code p?} in EBNF. {@code null} is the result when {@code this} fails with no partial match.
*/
public final Parser<T> optional() {
return Parsers.plus(this, Parsers.<T> always());
}
/**
* A {@link Parser} that returns {@code defaultValue} if {@code this} fails with no partial match.
*/
public final Parser<T> optional(T defaultValue) {
return Parsers.plus(this, Parsers.constant(defaultValue));
}
/**
* A {@link Parser} that fails if {@code this} succeeds. Any input consumption is undone.
*/
public final Parser<?> not() {
return not(toString());
}
/**
* A {@link Parser} that fails if {@code this} succeeds. Any input consumption is undone.
*
* @param unexpected
* the name of what we don't expect.
*/
public final Parser<?> not(String unexpected) {
return peek().ifelse(Parsers.unexpected(unexpected), Parsers.always());
}
/**
* A {@link Parser} that runs {@code this} and undoes any input consumption if succeeds.
*/
public final Parser<T> peek() {
return new PeekParser<T>(this);
}
/** A {@link Parser} that undoes any partial match if {@code this} fails. */
public final Parser<T> atomic() {
return new AtomicParser<T>(this);
}
/**
* A {@link Parser} that runs {@code this} parser and sets the number of logical steps explicitly to {@code n}.
*/
final Parser<T> step(int n) {
checkArgument(n >= 0, "step < 0");
return new StepParser<T>(this, n);
}
/**
* A {@link Parser} that returns {@code true} if {@code this} succeeds, {@code false} otherwise.
*/
public final Parser<Boolean> succeeds() {
return ifelse(Parsers.TRUE, Parsers.FALSE);
}
/**
* A {@link Parser} that returns {@code true} if {@code this} fails, {@code false} otherwise.
*/
public final Parser<Boolean> fails() {
return ifelse(Parsers.FALSE, Parsers.TRUE);
}
/**
* A {@link Parser} that runs {@code consequence} if {@code this} succeeds, or {@code alternative} otherwise.
*/
public final <R> Parser<R> ifelse(Parser<? extends R> consequence, Parser<? extends R> alternative) {
return ifelse(Maps.constant(consequence), alternative);
}
/**
* A {@link Parser} that runs {@code consequence} if {@code this} succeeds, or {@code alternative} otherwise.
*/
public final <R> Parser<R> ifelse(Map<? super T, ? extends Parser<? extends R>> consequence, Parser<? extends R> alternative) {
return new IfElseParser<R, T>(this, consequence, alternative);
}
/**
* A {@link Parser} that reports reports an error about {@code name} expected, if {@code this} fails with no partial match.
*/
public final Parser<T> label(String name) {
return Parsers.plus(this, Parsers.<T> expect(name));
}
/**
* Casts {@code this} to a {@link Parser} of type {@code R}. Use it only if you know the parser actually returns value of type {@code R}.
*/
@SuppressWarnings("unchecked")
public final <R> Parser<R> cast() {
return (Parser<R>) this;
}
/**
* A {@link Parser} that runs {@code this} between {@code before} and {@code after}. The return value of {@code this} is preserved.
*
* <p>
* Equivalent to {@link Parsers#between(Parser, Parser, Parser)}, which preserves the natural order of the parsers in the argument list, but is a bit more verbose.
*/
public final Parser<T> between(Parser<?> before, Parser<?> after) {
return before.next(followedBy(after));
}
/**
* A {@link Parser} that runs {@code this} 1 or more times separated by {@code delim}.
*
* <p>
* The return values are collected in a {@link List}.
*/
public final Parser<List<T>> sepBy1(Parser<?> delim) {
final Parser<T> afterFirst = delim.step(0).next(this);
Map<T, Parser<List<T>>> binder = new Map<T, Parser<List<T>>>() {
public Parser<List<T>> map(T firstValue) {
return new RepeatAtLeastParser<T>(afterFirst, 0, ListFactories.arrayListFactoryWithFirstElement(firstValue));
}
};
return next(binder);
}
/**
* A {@link Parser} that runs {@code this} 0 or more times separated by {@code delim}.
*
* <p>
* The return values are collected in a {@link List}.
*/
public final Parser<List<T>> sepBy(Parser<?> delim) {
return Parsers.plus(sepBy1(delim), EmptyListParser.<T> instance());
}
/**
* A {@link Parser} that runs {@code this} for 0 or more times delimited and terminated by {@code delim}.
*
* <p>
* The return values are collected in a {@link List}.
*/
public final Parser<List<T>> endBy(Parser<?> delim) {
return followedBy(delim).many();
}
/**
* A {@link Parser} that runs {@code this} for 1 or more times delimited and terminated by {@code delim}.
*
* <p>
* The return values are collected in a {@link List}.
*/
public final Parser<List<T>> endBy1(Parser<?> delim) {
return followedBy(delim).many1();
}
/**
* A {@link Parser} that runs {@code this} for 1 ore more times separated and optionally terminated by {@code delim}. For example: {@code "foo;foo;foo"} and {@code "foo;foo;"} both matches {@code foo.sepEndBy1(semicolon)}.
*
* <p>
* The return values are collected in a {@link List}.
*/
public final Parser<List<T>> sepEndBy1(final Parser<?> delim) {
return next(new Map<T, Parser<List<T>>>() {
public Parser<List<T>> map(T first) {
return new DelimitedListParser<T>(Parser.this, delim, ListFactories.arrayListFactoryWithFirstElement(first));
}
});
}
/**
* A {@link Parser} that runs {@code this} for 0 ore more times separated and optionally terminated by {@code delim}. For example: {@code "foo;foo;foo"} and {@code "foo;foo;"} both matches {@code foo.sepEndBy(semicolon)}.
*
* <p>
* The return values are collected in a {@link List}.
*/
public final Parser<List<T>> sepEndBy(Parser<?> delim) {
return Parsers.plus(sepEndBy1(delim), EmptyListParser.<T> instance());
}
/**
* A {@link Parser} that runs {@code op} for 0 or more times greedily, then runs {@code this}. The {@link Map} objects returned from {@code op} are applied from right to left to the return value of {@code p}.
*
* <p>
* {@code p.prefix(op)} is equivalent to {@code op* p} in EBNF.
*/
@SuppressWarnings("unchecked")
public final Parser<T> prefix(Parser<? extends Map<? super T, ? extends T>> op) {
return Parsers.sequence(op.many(), this, Parsers.PREFIX_OPERATOR_MAP2);
}
/**
* A {@link Parser} that runs {@code this} and then runs {@code op} for 0 or more times greedily. The {@link Map} objects returned from {@code op} are applied from left to right to the return value of p.
*
* <p>
* {@code p.postfix(op)} is equivalent to {@code p op*} in EBNF.
*/
@SuppressWarnings("unchecked")
public final Parser<T> postfix(Parser<? extends Map<? super T, ? extends T>> op) {
return Parsers.sequence(this, op.many(), Parsers.POSTFIX_OPERATOR_MAP2);
}
/**
* A {@link Parser} that parses non-associative infix operator. Runs {@code this} for the left operand, and then runs {@code op} and {@code this} for the operator and the right operand optionally. The {@link Map2} objects returned from {@code op} are applied to the return values of the two operands, if any.
*
* <p>
* {@code p.infixn(op)} is equivalent to {@code p (op p)?} in EBNF.
*/
public final Parser<T> infixn(Parser<? extends Map2<? super T, ? super T, ? extends T>> op) {
return Parsers.infixn(this, op);
}
/**
* A {@link Parser} for left-associative infix operator. Runs {@code this} for the left operand, and then runs {@code op} and {@code this} for the operator and the right operand for 0 or more times greedily. The {@link Map2} objects returned from {@code op} are applied from left to right to the return values of {@code this}, if any. For example: {@code a + b + c + d} is evaluated as
* {@code (((a + b)+c)+d)}.
*
* <p>
* {@code p.infixl(op)} is equivalent to {@code p (op p)*} in EBNF.
*/
public final Parser<T> infixl(Parser<? extends Map2<? super T, ? super T, ? extends T>> op) {
// somehow generics doesn't work if we inline the code here.
return Parsers.infixl(this, op);
}
/**
* A {@link Parser} for right-associative infix operator. Runs {@code this} for the left operand, and then runs {@code op} and {@code this} for the operator and the right operand for 0 or more times greedily. The {@link Map2} objects returned from {@code op} are applied from right to left to the return values of {@code this}, if any. For example: {@code a + b + c + d} is evaluated as
* {@code a + (b + (c + d))}.
*
* <p>
* {@code p.infixr(op)} is equivalent to {@code p (op p)*} in EBNF.
*/
public final Parser<T> infixr(Parser<? extends Map2<? super T, ? super T, ? extends T>> op) {
return Parsers.infixr(this, op);
}
/**
* A {@link Parser} that runs {@code this} and wraps the return value in a {@link Token}.
*
* <p>
* It is normally not necessary to call this method explicitly. {@link #lexer(Parser)} and {@link #from(Parser, Parser)} both do the conversion automatically.
*/
public final Parser<Token> token() {
return new ToTokenParser(this);
}
/** A {@link Parser} that returns the matched string in the original source. */
public final Parser<String> source() {
return new ReturnSourceParser(this);
}
/**
* A {@link Parser} that takes as input the {@link Token} collection returned by {@code lexer}, and runs {@code this} to parse the tokens.
*
* <p>
* {@code this} must be a token level parser.
*/
public final Parser<T> from(Parser<? extends Collection<Token>> lexer) {
return Parsers.nested(Parsers.tokens(lexer), followedBy(Parsers.EOF));
}
/**
* A {@link Parser} that takes as input the tokens returned by {@code tokenizer} delimited by {@code delim}, and runs {@code this} to parse the tokens.
*
* <p>
* {@code this} must be a token level parser.
*/
public final Parser<T> from(Parser<?> tokenizer, Parser<Void> delim) {
return from(tokenizer.lexer(delim));
}
/**
* A {@link Parser} that greedily runs {@code this} repeatedly, and ignores the pattern recognized by {@code delim} before and after each occurrence. The result tokens are wrapped in {@link Token} and are collected and returned in a {@link List}.
*
* <p>
* It is normally not necessary to call this method explicitly. {@link #from(Parser, Parser)} is more convenient for simple uses that just need to connect a token level parser with a lexer that produces the tokens. When more flexible control over the token list is needed, for example, to parse indentation sensitive language, a pre-processor of the token list may be needed.
*
* <p>
* {@code this} must be a tokenizer that returns a token value.
*/
public Parser<List<Token>> lexer(Parser<?> delim) {
return delim.optional().next(token().sepEndBy(delim));
}
/**
* Parses a source string.
*
* @param source
* the source string
* @param moduleName
* the name of the module, this name appears in error message
* @param sourceLocator
* maps an index of char into line and column numbers
* @return the result
*/
final T parse(CharSequence source, String moduleName, SourceLocator sourceLocator) {
return Parsers.parse(source, followedBy(Parsers.EOF), sourceLocator, moduleName);
}
/**
* Parses {@code source}.
*
* @param source
* the source string
* @param moduleName
* the name of the module, this name appears in error message
* @return the result
*/
public final T parse(CharSequence source, String moduleName) {
return parse(source, moduleName, new DefaultSourceLocator(source));
}
/** Parses {@code source}. */
public final T parse(CharSequence source) {
return parse(source, null);
}
/** Parses source read from {@code readable}. */
public final T parse(Readable readable) throws IOException {
return parse(readable, null);
}
/**
* Parses source read from {@code readable}.
*
* @param readable
* where the source is read from
* @param moduleName
* the name of the module, this name appears in error message
* @return the result
*/
public final T parse(Readable readable, String moduleName) throws IOException {
StringBuilder builder = new StringBuilder();
copy(readable, builder);
return parse(builder, moduleName);
}
/** Copies all content from {@code from} to {@code to}. */
@Private
static void copy(Readable from, Appendable to) throws IOException {
CharBuffer buf = CharBuffer.allocate(2048);
for (;;) {
int r = from.read(buf);
if (r == -1)
break;
buf.flip();
to.append(buf, 0, r);
}
}
@SuppressWarnings("unchecked")
final T getReturn(ParseContext ctxt) {
return (T) ctxt.result;
}
private ParserException asParserException(Throwable e, ParseContext ctxt) {
if (e instanceof ParserException)
return (ParserException) e;
return new ParserException(e, null, ctxt.module, ctxt.locator.locate(ctxt.getIndex()));
}
final boolean run(ParseContext ctxt) {
try {
return apply(ctxt);
} catch (RuntimeException e) {
throw asParserException(e, ctxt);
}
}
abstract boolean apply(ParseContext ctxt);
}