From 9d89261fedf23c11b684eb66cefdd86a9378ad20 Mon Sep 17 00:00:00 2001
From: bjculkin <bjculkin@mix.wvu.edu>
Date: Fri, 17 Mar 2017 08:34:57 -0400
Subject: Move parsing utilities.

Moved the parsing utilities SequenceDelimiter and TokenSplitter to the
parserutils package, instead of the funcutils package.
---
 .../java/bjc/utils/parserutils/TokenSplitter.java  | 161 +++++++++++++++++++++
 1 file changed, 161 insertions(+)
 create mode 100644 BJC-Utils2/src/main/java/bjc/utils/parserutils/TokenSplitter.java

(limited to 'BJC-Utils2/src/main/java/bjc/utils/parserutils/TokenSplitter.java')

diff --git a/BJC-Utils2/src/main/java/bjc/utils/parserutils/TokenSplitter.java b/BJC-Utils2/src/main/java/bjc/utils/parserutils/TokenSplitter.java
new file mode 100644
index 0000000..e6191b9
--- /dev/null
+++ b/BJC-Utils2/src/main/java/bjc/utils/parserutils/TokenSplitter.java
@@ -0,0 +1,161 @@
+package bjc.utils.parserutils;
+
+import java.util.regex.Pattern;
+
+/**
+ * Split a string and keep given delimiters.
+ *
+ * @author Ben Culkin
+ */
+public class TokenSplitter {
+	/*
+	 * This string is a format template for the delimiter matching regex
+	 *
+	 * It does two things:
+	 * 
+	 * <ol>
+	 * <li> Match to the left of the provided delimiter by positive lookahead </li>
+	 * <li> Match to the right of the provided delimiter by positive lookbehind </li>
+	 * </ol>
+	 *
+	 * Thus, it will only match in places where the delimiter is, but won't
+	 * actually match the delimiter, leaving split to put it into the stream
+	 */
+	private static String WITH_DELIM = "(?:(?<=%1$s)|(?=%1$s))";
+
+	/*
+	 * This string is a format template for the multi-delimiter matching
+	 * regex.
+	 *
+	 * It does the same thing as the single delimiter regex, but has to have
+	 * some negative lookahead/lookbehind assertions to avoid splitting a
+	 * delimiter into pieces.
+	 */
+	private static String WITH_MULTI_DELIM = "(?:(?<=%1$s+)(?!%1$s)|(?<!%1$s)(?=%1$s+))";
+
+	/*
+	 * These represent the internal state of the splitter.
+	 */
+	private StringBuilder	currPatt;
+	private StringBuilder	currExclusionPatt;
+
+	/*
+	 * These represent the external state of the splitter.
+	 *
+	 * Compilation causes internal to become external.
+	 */
+	private Pattern	compPatt;
+	private Pattern	exclusionPatt;
+
+	/**
+	 * Create a new token splitter.
+	 */
+	public TokenSplitter() {
+	}
+
+	/**
+	 * Split a provided string using configured delimiters, and keeping the
+	 * delimiters.
+	 *
+	 * <p>
+	 * The splitter must be compiled first.
+	 * </p>
+	 * 
+	 * @param inp
+	 *                The string to split.
+	 *
+	 * @return The split string, including delimiters.
+	 *
+	 * @throws IllegalStateException
+	 *                 If the splitter isn't compiled.
+	 */
+	public String[] split(String inp) {
+		if(compPatt == null) throw new IllegalStateException("Token splitter has not been compiled yet");
+
+		/*
+		 * Don't split something that we should exclude from being split.
+		 */
+		if(exclusionPatt.matcher(inp).matches()) return new String[] { inp };
+
+		return compPatt.split(inp);
+	}
+
+	/**
+	 * Adds one or more strings as matched delimiters to split on.
+	 *
+	 * Only works for fixed length delimiters.
+	 *
+	 * The provided strings are regex-escaped before being used.
+	 *
+	 * @param delims
+	 *                The delimiters to match on.
+	 */
+	public void addDelimiter(String... delims) {
+		for(String delim : delims) {
+			String quoteDelim = Pattern.quote(delim);
+			String delimPat   = String.format(WITH_DELIM, quoteDelim);
+	
+			if(currPatt == null) {
+				currPatt          = new StringBuilder();
+				currExclusionPatt = new StringBuilder();
+	
+				currPatt.append("(?:" + delimPat + ")");
+				currExclusionPatt.append("(?:" + quoteDelim + ")");
+			} else {
+				currPatt.append("|(?:" + delimPat + ")");
+				currExclusionPatt.append("|(?:" + quoteDelim + ")");
+			}
+		}
+	}
+
+	/**
+	 * Adds a character class as a matched delimiter to split on.
+	 *
+	 * The provided string should be a pattern to match one or more
+	 * occurances of.
+	 *
+	 * @param delim
+	 *                The delimiter to split on.
+	 */
+	public void addMultiDelimiter(String delim) {
+		String delimPat = String.format(WITH_MULTI_DELIM, "(?:" + delim + ")");
+
+		if(currPatt == null) {
+			currPatt          = new StringBuilder();
+			currExclusionPatt = new StringBuilder();
+
+			currPatt.append("(?:" + delimPat + ")");
+			currExclusionPatt.append("(?:(?:" + delim + ")+)");
+
+		} else {
+			currPatt.append("|(?:" + delimPat + ")");
+			currExclusionPatt.append("|(?:(?:" + delim + ")+)");
+		}
+	}
+
+	/**
+	 * Marks strings matching the pattern delim as non-splittable.
+	 *
+	 * @param delim
+	 * 		The regex to not splitting matching strings.
+	 */
+	public void addNonMatcher(String delim) {
+		if(currPatt == null) {
+			currPatt          = new StringBuilder();
+			currExclusionPatt = new StringBuilder();
+
+			currExclusionPatt.append("(?:" + delim + ")");
+		} else {
+			currExclusionPatt.append("|(?:" + delim + ")");
+		}
+	}
+	/**
+	 * Compiles the current set of delimiters to a pattern.
+	 *
+	 * Makes this splitter ready to use.
+	 */
+	public void compile() {
+		compPatt      = Pattern.compile(currPatt.toString());
+		exclusionPatt = Pattern.compile(currExclusionPatt.toString());
+	}
+}
-- 
cgit v1.2.3