added regex scanner options: multiline, dotall

Nuix · Jan 22, 2021 · d1194b7 · d1194b7
1 parent d973ee2
commit d1194b7
Show file tree

Hide file tree

Showing 2 changed files with 29 additions and 10 deletions.
diff --git a/Java/src/main/java/com/nuix/superutilities/regex/PatternInfo.java b/Java/src/main/java/com/nuix/superutilities/regex/PatternInfo.java
@@ -26,12 +26,13 @@ public PatternInfo(String title, String expression){
 	 * Compiles the regular expression String provided into a Java Pattern object
 	 * @param caseSensitive Whether it should be case sensitive
 	 */
-	public void compile(boolean caseSensitive){
+	public void compile(boolean caseSensitive, boolean multiline, boolean dotAll){
 		if(pattern == null){
-			if(caseSensitive)
-				pattern = Pattern.compile(expression);
-			else
-				pattern = Pattern.compile(expression,Pattern.CASE_INSENSITIVE);
+			int bitflags = 0;
+			if(!caseSensitive) { bitflags |= Pattern.CASE_INSENSITIVE; }
+			if(multiline) { bitflags |= Pattern.MULTILINE; }
+			if(dotAll) { bitflags |= Pattern.DOTALL; }
+			pattern = Pattern.compile(expression,bitflags);
 		}
 	}
 

diff --git a/Java/src/main/java/com/nuix/superutilities/regex/RegexScanner.java b/Java/src/main/java/com/nuix/superutilities/regex/RegexScanner.java
@@ -50,6 +50,8 @@ public static void setMaxToStringLength(int maxLength){
 
 	private boolean scanContent = true;
 	private boolean caseSensitive = false;
+	private boolean multiline = false;
+	private boolean dotall = false;
 	private boolean captureContextualText = true;
 	private int contextSize = 100;
 
@@ -134,7 +136,7 @@ public List<ItemRegexMatchCollection> scanItems(Collection<Item> items){
 		List<ItemRegexMatchCollection> result = new ArrayList<ItemRegexMatchCollection>();
 
 		for (PatternInfo p : patterns) {
-			p.compile(caseSensitive);
+			p.compile(caseSensitive,multiline,dotall);
 		}
 
 		for (Item item : items) {
@@ -163,7 +165,7 @@ public void scanItems(Collection<Item> items, Consumer<ItemRegexMatchCollection>
 		abortWasRequested = false;
 
 		for (PatternInfo p : patterns) {
-			p.compile(caseSensitive);
+			p.compile(caseSensitive,multiline,dotall);
 		}
 
 		int itemIndex = 0;
@@ -196,7 +198,7 @@ public void scanItemsParallel(Collection<Item> items, Consumer<ItemRegexMatchCol
 		abortWasRequested = false;
 
 		for (PatternInfo p : patterns) {
-			p.compile(caseSensitive);
+			p.compile(caseSensitive,multiline,dotall);
 		}
 
 		AtomicInteger itemIndex = new AtomicInteger(0);
@@ -242,7 +244,7 @@ public void scanItemsParallel(Collection<Item> items, Consumer<ItemRegexMatchCol
 		abortWasRequested = false;
 
 		for (PatternInfo p : patterns) {
-			p.compile(caseSensitive);
+			p.compile(caseSensitive,multiline,dotall);
 		}
 
 		AtomicInteger itemIndex = new AtomicInteger(0);
@@ -306,7 +308,7 @@ protected ItemRegexMatchCollection scanItem(Item item) {
 					Set<String> entityValues = item.getEntities(namedEntityType);
 					for(String entityValue : entityValues) {
 						PatternInfo entityPattern = new PatternInfo(namedEntityType, "\\Q"+entityValue+"\\E");
-						entityPattern.compile(caseSensitive);
+						entityPattern.compile(caseSensitive,multiline,dotall);
 						patternsToScanFor.add(entityPattern);
 					}
 				} catch (IOException e) {
@@ -537,6 +539,22 @@ public void setCaseSensitive(boolean caseSensitive) {
 		this.caseSensitive = caseSensitive;
 	}
 
+	public boolean getMultiline() {
+		return multiline;
+	}
+
+	public void setMultiline(boolean multiline) {
+		this.multiline = multiline;
+	}
+
+	public boolean getDotall() {
+		return dotall;
+	}
+
+	public void setDotall(boolean dotall) {
+		this.dotall = dotall;
+	}
+
 	public boolean getCaptureContextualText() {
 		return captureContextualText;
 	}