New splitter implementation

5 files changed, 310 insertions, 0 deletions

diff --git a/BJC-Utils2/src/main/java/bjc/utils/parserutils/splitterv2/ChainTokenSplitter.java b/BJC-Utils2/src/main/java/bjc/utils/parserutils/splitterv2/ChainTokenSplitter.java
new file mode 100644
index 0000000..2ecadaf
--- /dev/null
+++ b/BJC-Utils2/src/main/java/bjc/utils/parserutils/splitterv2/ChainTokenSplitter.java
@@ -0,0 +1,51 @@
+package bjc.utils.parserutils.splitterv2;
+
+import bjc.utils.funcdata.FunctionalList;
+import bjc.utils.funcdata.IList;
+import bjc.utils.functypes.ID;
+
+/**
+ * A token splitter that chains several other splitters together.
+ * 
+ * @author EVE
+ *
+ */
+public class ChainTokenSplitter implements TokenSplitter {
+	private IList<TokenSplitter> spliters;
+
+	/**
+	 * Create a new chain token splitter.
+	 */
+	public ChainTokenSplitter() {
+		spliters = new FunctionalList<>();
+	}
+
+	/**
+	 * Append a series of splitters to the chain.
+	 * 
+	 * @param splitters
+	 *                The splitters to append to the chain.
+	 */
+	public void appendSplitters(TokenSplitter... splitters) {
+		spliters.addAll(splitters);
+	}
+
+	/**
+	 * Prepend a series of splitters to the chain.
+	 * 
+	 * @param splitters
+	 *                The splitters to append to the chain.
+	 */
+	public void prependSplitters(TokenSplitter... splitters) {
+		spliters.prependAll(splitters);
+	}
+
+	@Override
+	public IList<String> split(String input) {
+		IList<String> initList = new FunctionalList<>(input);
+
+		return spliters.reduceAux(initList, (splitter, strangs) -> {
+			return strangs.flatMap(splitter::split);
+		}, ID.id());
+	}
+}
\ No newline at end of file
diff --git a/BJC-Utils2/src/main/java/bjc/utils/parserutils/splitterv2/ConfigurableTokenSplitter.java b/BJC-Utils2/src/main/java/bjc/utils/parserutils/splitterv2/ConfigurableTokenSplitter.java
new file mode 100644
index 0000000..021821a
--- /dev/null
+++ b/BJC-Utils2/src/main/java/bjc/utils/parserutils/splitterv2/ConfigurableTokenSplitter.java
@@ -0,0 +1,124 @@
+package bjc.utils.parserutils.splitterv2;
+
+import bjc.utils.funcdata.IList;
+
+import java.util.LinkedHashSet;
+import java.util.Set;
+import java.util.regex.Pattern;
+
+import static bjc.utils.PropertyDB.applyFormat;
+
+/**
+ * Split a string into pieces around a regular expression, and offer an easy way
+ * to configure the regular expression.
+ * 
+ * @author EVE
+ *
+ */
+public class ConfigurableTokenSplitter extends SimpleTokenSplitter {
+	private Set<String>	simpleDelimiters;
+	private Set<String>	multipleDelimiters;
+	private Set<String>	rRawDelimiters;
+
+	/**
+	 * Create a new token splitter with blank configuration.
+	 * 
+	 * @param keepDelims
+	 *                Whether or not to keep delimiters.
+	 */
+	public ConfigurableTokenSplitter(boolean keepDelims) {
+		super(null, keepDelims);
+
+		/*
+		 * Use linked hash-sets to keep items in insertion order.
+		 */
+		simpleDelimiters = new LinkedHashSet<>();
+		multipleDelimiters = new LinkedHashSet<>();
+		rRawDelimiters = new LinkedHashSet<>();
+	}
+
+	/**
+	 * Add a set of simple delimiters to this splitter.
+	 * 
+	 * Simple delimiters match one occurrence of themselves as literals.
+	 * 
+	 * @param simpleDelims
+	 *                The simple delimiters to add.
+	 */
+	public void addSimpleDelimiters(String... simpleDelims) {
+		for(String simpleDelim : simpleDelims) {
+			simpleDelimiters.add(simpleDelim);
+		}
+	}
+
+	/**
+	 * Add a set of multiple delimiters to this splitter.
+	 * 
+	 * Multiple delimiters match one or more occurrences of themselves as
+	 * literals.
+	 * 
+	 * @param multiDelims
+	 *                The multiple delimiters to add.
+	 */
+	public void addMultiDelimiters(String... multiDelims) {
+		for(String multiDelim : multiDelims) {
+			multipleDelimiters.add(multiDelim);
+		}
+	}
+
+	/**
+	 * Add a set of raw delimiters to this splitter.
+	 * 
+	 * Raw delimiters match one occurrence of themselves as regular
+	 * expressions.
+	 * 
+	 * @param rRawDelims
+	 *                The raw delimiters to add.
+	 */
+	public void addRawDelimiters(String... rRawDelims) {
+		for(String rRawDelim : rRawDelims) {
+			rRawDelimiters.add(rRawDelim);
+		}
+	}
+
+	/**
+	 * Take the configuration and compile it into a regular expression to
+	 * use when splitting.
+	 */
+	public void compile() {
+		StringBuilder rPattern = new StringBuilder();
+
+		for(String rRawDelimiter : rRawDelimiters) {
+			rPattern.append(applyFormat("rawDelim", rRawDelimiter));
+		}
+
+		for(String multipleDelimiter : multipleDelimiters) {
+			rPattern.append(applyFormat("multipleDelim", multipleDelimiter));
+		}
+
+		for(String simpleDelimiter : simpleDelimiters) {
+			rPattern.append(applyFormat("simpleDelim", simpleDelimiter));
+		}
+
+		rPattern.deleteCharAt(rPattern.length() - 1);
+
+		spliter = Pattern.compile(rPattern.toString());
+	}
+
+	@Override
+	public IList<String> split(String input) {
+		if(spliter == null) {
+			throw new IllegalStateException("Must compile splitter before use");
+		}
+
+		return super.split(input);
+	}
+
+	@Override
+	public String toString() {
+		String fmt = "ConfigurableTokenSplitter [simpleDelimiters=%s, multipleDelimiters=%s,"
+				+ " rRawDelimiters=%s, spliter=%s]";
+
+		return String.format(fmt, simpleDelimiters, multipleDelimiters, rRawDelimiters, spliter);
+	}
+}
diff --git a/BJC-Utils2/src/main/java/bjc/utils/parserutils/splitterv2/ExcludingTokenSplitter.java b/BJC-Utils2/src/main/java/bjc/utils/parserutils/splitterv2/ExcludingTokenSplitter.java
new file mode 100644
index 0000000..0c93a25
--- /dev/null
+++ b/BJC-Utils2/src/main/java/bjc/utils/parserutils/splitterv2/ExcludingTokenSplitter.java
@@ -0,0 +1,66 @@
+package bjc.utils.parserutils.splitterv2;
+
+import bjc.utils.funcdata.FunctionalList;
+import bjc.utils.funcdata.IList;
+
+import java.util.HashSet;
+import java.util.Set;
+import java.util.function.Predicate;
+
+/**
+ * A token splitter that will not split certain tokens.
+ * 
+ * @author EVE
+ *
+ */
+public class ExcludingTokenSplitter implements TokenSplitter {
+	private Set<String> literalExclusions;
+
+	private IList<Predicate<String>> predExclusions;
+
+	private TokenSplitter spliter;
+
+	/**
+	 * Create a new excluding token splitter.
+	 * 
+	 * @param splitter
+	 *                The splitter to apply to non-excluded strings.
+	 */
+	public ExcludingTokenSplitter(TokenSplitter splitter) {
+		spliter = splitter;
+
+		literalExclusions = new HashSet<>();
+
+		predExclusions = new FunctionalList<>();
+	}
+
+	/**
+	 * Exclude a literal string from splitting.
+	 * 
+	 * @param exclusion
+	 *                The string to exclude from splitting.
+	 */
+	public void addLiteralExclusion(String exclusion) {
+		literalExclusions.add(exclusion);
+	}
+
+	/**
+	 * Exclude all of the strings matching a predicate from splitting.
+	 * 
+	 * @param exclusion
+	 *                The predicate to use for exclusions.
+	 */
+	public void addPredicateExclusion(Predicate<String> exclusion) {
+		predExclusions.add(exclusion);
+	}
+
+	@Override
+	public IList<String> split(String input) {
+		if(literalExclusions.contains(input))
+			return new FunctionalList<>(input);
+		else if(predExclusions.anyMatch(pred -> pred.test(input))) return new FunctionalList<>(input);
+
+		return spliter.split(input);
+	}
+
+}
diff --git a/BJC-Utils2/src/main/java/bjc/utils/parserutils/splitterv2/SimpleTokenSplitter.java b/BJC-Utils2/src/main/java/bjc/utils/parserutils/splitterv2/SimpleTokenSplitter.java
new file mode 100644
index 0000000..b111ca3
--- /dev/null
+++ b/BJC-Utils2/src/main/java/bjc/utils/parserutils/splitterv2/SimpleTokenSplitter.java
@@ -0,0 +1,48 @@
+package bjc.utils.parserutils.splitterv2;
+
+import bjc.utils.funcdata.IList;
+import bjc.utils.functypes.ID;
+import bjc.utils.ioutils.RegexStringEditor;
+
+import java.util.regex.Pattern;
+
+/**
+ * Splits a string into pieces around a regular expression.
+ * 
+ * @author EVE
+ *
+ */
+public class SimpleTokenSplitter implements TokenSplitter {
+	protected Pattern spliter;
+
+	private boolean keepDelim;
+
+	/**
+	 * Create a new simple token splitter.
+	 * 
+	 * @param splitter
+	 *                The pattern to split around.
+	 * 
+	 * @param keepDelims
+	 *                Whether or not delimiters should be kept.
+	 */
+	public SimpleTokenSplitter(Pattern splitter, boolean keepDelims) {
+		spliter = splitter;
+
+		keepDelim = keepDelims;
+	}
+
+	@Override
+	public IList<String> split(String input) {
+		if(keepDelim) {
+			return RegexStringEditor.mapOccurances(input, spliter, ID.id(), ID.id());
+		} else {
+			return RegexStringEditor.mapOccurances(input, spliter, ID.id(), strang -> "");
+		}
+	}
+
+	@Override
+	public String toString() {
+		return String.format("SimpleTokenSplitter [spliter=%s, keepDelim=%s]", spliter, keepDelim);
+	}
+}
\ No newline at end of file
diff --git a/BJC-Utils2/src/main/java/bjc/utils/parserutils/splitterv2/TokenSplitter.java b/BJC-Utils2/src/main/java/bjc/utils/parserutils/splitterv2/TokenSplitter.java
new file mode 100644
index 0000000..5d510c1
--- /dev/null
+++ b/BJC-Utils2/src/main/java/bjc/utils/parserutils/splitterv2/TokenSplitter.java
@@ -0,0 +1,21 @@
+package bjc.utils.parserutils.splitterv2;
+
+import bjc.utils.funcdata.IList;
+
+/**
+ * Split a string into a list of pieces.
+ * 
+ * @author EVE
+ *
+ */
+public interface TokenSplitter {
+	/**
+	 * Split a string into a list of pieces.
+	 * 
+	 * @param input
+	 *                The string to split.
+	 * 
+	 * @return The pieces of the string.
+	 */
+	public IList<String> split(String input);
+}