X-Git-Url: http://www2.svjatoslav.eu/gitweb/?a=blobdiff_plain;f=src%2Fmain%2Fjava%2Feu%2Fsvjatoslav%2Fcommons%2Fstring%2Ftokenizer%2FTokenizer.java;h=4eb9dde36ff4679a5f2280f486ec893ed55c1d4b;hb=b91077678c2a0af17a2f852c0b3901813837117a;hp=dc0cb9c1ea390fb869c0ac7d6e43499af73e40c2;hpb=cf965fda534cc562368c9f2a3f34475e2519fcdc;p=svjatoslav_commons.git diff --git a/src/main/java/eu/svjatoslav/commons/string/tokenizer/Tokenizer.java b/src/main/java/eu/svjatoslav/commons/string/tokenizer/Tokenizer.java old mode 100644 new mode 100755 index dc0cb9c..4eb9dde --- a/src/main/java/eu/svjatoslav/commons/string/tokenizer/Tokenizer.java +++ b/src/main/java/eu/svjatoslav/commons/string/tokenizer/Tokenizer.java @@ -1,129 +1,201 @@ /* - * Svjatoslav Commons - shared library of common functionality. - * Copyright ©2012-2013, Svjatoslav Agejenko, svjatoslav@svjatoslav.eu - * - * This program is free software; you can redistribute it and/or - * modify it under the terms of version 2 of the GNU General Public License - * as published by the Free Software Foundation. + * Svjatoslav Commons - shared library of common functionality. Author: Svjatoslav Agejenko. + * This project is released under Creative Commons Zero (CC0) license. */ - package eu.svjatoslav.commons.string.tokenizer; import java.util.ArrayList; import java.util.List; import java.util.Stack; +import java.util.stream.Stream; + +import static eu.svjatoslav.commons.string.tokenizer.Terminator.TerminationStrategy.DROP; +import static eu.svjatoslav.commons.string.tokenizer.Terminator.TerminationStrategy.PRESERVE; public class Tokenizer { - private final List terminators = new ArrayList(); - private final String source; - - Stack tokenIndexes = new Stack(); - - private int currentIndex = 0; - - public Tokenizer(final String source) { - this.source = source; - } - - public void addTerminator(final String startSequence, - final boolean ignoreTerminator) { - terminators.add(new Terminator(startSequence, ignoreTerminator)); - } - - public void addTerminator(final String startSequence, - final String endSequence, final boolean ignoreTerminator) { - terminators.add(new Terminator(startSequence, endSequence, - ignoreTerminator)); - } - - public void expectNextToken(final String value) - throws InvalidSyntaxException { - final TokenizerMatch match = getNextToken(); - if (!value.equals(match.token)) - throw new InvalidSyntaxException("Expected \"" + value - + "\" but got \"" + match.token + "\" instead."); - } - - public TokenizerMatch getNextToken() { - tokenIndexes.push(currentIndex); - final StringBuffer result = new StringBuffer(); - - while (true) { - if (currentIndex >= source.length()) - return null; - - boolean accumulateCurrentChar = true; - - findTerminator: for (final Terminator terminator : terminators) - if (sequenceMatches(terminator.startSequence)) - - if (terminator.ignoreTerminator) { - currentIndex += terminator.startSequence.length(); - - if (terminator.endSequence != null) - skipUntilSequence(terminator.endSequence); - - if (result.length() > 0) - return new TokenizerMatch(result.toString(), - terminator); - else { - accumulateCurrentChar = false; - break findTerminator; - } - } else if (result.length() > 0) - return new TokenizerMatch(result.toString(), terminator); - else { - currentIndex += terminator.startSequence.length(); - return new TokenizerMatch(terminator.startSequence, - terminator); - } - - if (accumulateCurrentChar) { - result.append(source.charAt(currentIndex)); - currentIndex++; - } - } - - } - - public boolean probeNextToken(final String token) { - if (token.equals(getNextToken().token)) - return true; - - unreadToken(); - return false; - } - - public boolean sequenceMatches(final String sequence) { - if ((currentIndex + sequence.length()) > source.length()) - return false; - - for (int i = 0; i < sequence.length(); i++) - if (sequence.charAt(i) != source.charAt(i + currentIndex)) - return false; - - return true; - } - - public void skipUntilDataEnd() { - tokenIndexes.push(currentIndex); - currentIndex = source.length(); - } - - public void skipUntilSequence(final String sequence) { - while (currentIndex < source.length()) { - if (sequenceMatches(sequence)) { - currentIndex += sequence.length(); - return; - } - - currentIndex++; - } - } - - public void unreadToken() { - currentIndex = tokenIndexes.pop(); - } + private final Stack tokenIndexes = new Stack<>(); + private final List terminators = new ArrayList<>(); + private String source; + private int currentIndex = 0; + + private int cachedTerminatorIndex = -1; + private Terminator cachedTerminator; + + public Tokenizer(final String source) { + this.source = source; + } + + public Tokenizer() { + } + + public Tokenizer setSource(String source) { + this.source = source; + currentIndex = 0; + tokenIndexes.clear(); + + cachedTerminatorIndex = -1; + cachedTerminator = null; + return this; + } + + public Tokenizer addTerminator(final String startSequence, + final Terminator.TerminationStrategy terminationStrategy) { + terminators.add(new Terminator(startSequence, terminationStrategy)); + return this; + } + + public Tokenizer addTerminator(final String startSequence, + final String endSequence, final Terminator.TerminationStrategy terminationStrategy) { + terminators.add(new Terminator(startSequence, endSequence, terminationStrategy)); + return this; + } + + public void expectAndConsumeNextToken(final String value) + throws InvalidSyntaxException { + final TokenizerMatch match = getNextToken(); + if (!value.equals(match.token)) + throw new InvalidSyntaxException("Expected \"" + value + + "\" but got \"" + match.token + "\" instead."); + } + + + public TokenizerMatch getNextToken() throws InvalidSyntaxException { + tokenIndexes.push(currentIndex); + + StringBuilder tokenAccumulator = new StringBuilder(); + + while (true) { + + if (currentIndex >= source.length()) { // reached end of input + if (hasAccumulatedToken(tokenAccumulator)) + return new TokenizerMatch(tokenAccumulator.toString(), null, null); + else + return null; + } + + if (isOngoingToken()) { + tokenAccumulator.append(source.charAt(currentIndex)); + currentIndex++; + continue; + } + + Terminator terminator = getOrFindTokenTerminator(); + + if (terminator.termination == PRESERVE) + return buildPreservedToken(tokenAccumulator, terminator); + else if (terminator.termination == DROP) { + skipUntilTerminatorEnd(terminator); + + if (hasAccumulatedToken(tokenAccumulator)) + return new TokenizerMatch(tokenAccumulator.toString(), null, terminator); + } + } + + } + + private void skipUntilTerminatorEnd(Terminator terminator) throws InvalidSyntaxException { + if (terminator.hasEndSequence()) + currentIndex = getEndSequenceIndex(terminator) + terminator.endSequence.length(); + else + currentIndex += terminator.startSequence.length(); + } + + private TokenizerMatch buildPreservedToken(StringBuilder token, Terminator terminator) throws InvalidSyntaxException { + if (hasAccumulatedToken(token)) + return new TokenizerMatch(token.toString(), null, terminator); + + if (terminator.hasEndSequence()) + return buildComplexPreservedToken(terminator); + else + return buildSimplePreservedToken(terminator); + } + + private TokenizerMatch buildSimplePreservedToken(Terminator terminator) { + currentIndex += terminator.startSequence.length(); + return new TokenizerMatch(terminator.startSequence, null, terminator); + } + + private TokenizerMatch buildComplexPreservedToken(Terminator terminator) throws InvalidSyntaxException { + int endSequenceIndex = getEndSequenceIndex(terminator); + String reminder = source.substring(currentIndex + terminator.startSequence.length(), endSequenceIndex); + currentIndex = endSequenceIndex + terminator.endSequence.length(); + + return new TokenizerMatch(terminator.startSequence, reminder, terminator); + } + + private int getEndSequenceIndex(Terminator terminator) throws InvalidSyntaxException { + int endSequenceIndex = source.indexOf(terminator.endSequence, + currentIndex + terminator.startSequence.length()); + + if (endSequenceIndex < 0) + throw new InvalidSyntaxException("Expected \"" + terminator.endSequence + "\" but not found."); + + return endSequenceIndex; + } + + private boolean hasAccumulatedToken(StringBuilder token) { + return token.length() > 0; + } + + private boolean isOngoingToken() { + return getOrFindTokenTerminator() == null; + } + + public boolean hasMoreTokens() { + return currentIndex < source.length(); + } + + /** + * Attempts to cache terminator search result. + */ + public Terminator getOrFindTokenTerminator() { + if (currentIndex == cachedTerminatorIndex) + return cachedTerminator; + + cachedTerminatorIndex = currentIndex; + cachedTerminator = findTokenTerminator(); + return cachedTerminator; + } + + private Terminator findTokenTerminator() { + for (Terminator terminator : terminators) + if (terminator.matches(source, currentIndex)) + return terminator; + return null; + } + + public boolean consumeIfNextToken(final String token) throws InvalidSyntaxException { + if (token.equals(getNextToken().token)) + return true; + + unreadToken(); + return false; + } + + public TokenizerMatch peekNextToken() throws InvalidSyntaxException { + TokenizerMatch result = getNextToken(); + unreadToken(); + return result; + } + + public boolean peekIsOneOf(String... possibilities) throws InvalidSyntaxException { + String nextToken = peekNextToken().token; + return Stream.of(possibilities).anyMatch(possibility -> possibility.equals(nextToken)); + } + + public void peekExpectNoneOf(String... possibilities) throws InvalidSyntaxException { + if (peekIsOneOf(possibilities)) + throw new InvalidSyntaxException("Not expected \"" + peekNextToken().token + "\" here."); + } + + public void unreadToken() { + currentIndex = tokenIndexes.pop(); + } + + public void skipUntilDataEnd() { + tokenIndexes.push(currentIndex); + currentIndex = source.length(); + } }