Rewrote squeeze in rascal

usethesource · Jul 2, 2024 · 90d5afe · 90d5afe
1 parent 2dc75e6
commit 90d5afe
Show file tree

Hide file tree

Showing 3 changed files with 16 additions and 34 deletions.
diff --git a/src/org/rascalmpl/library/Prelude.java b/src/org/rascalmpl/library/Prelude.java
@@ -3040,31 +3040,6 @@ public IString trim(IString s) {
 		return values.string(s.getValue().trim());
 	}
 
-	public IString squeeze(IString src, IString charSet) {
-		final Pattern isCharset = Pattern.compile("[" + charSet.getValue() + "]", Pattern.UNICODE_CHARACTER_CLASS);
-		StringBuilder result = new StringBuilder(src.length());
-		var chars = src.iterator();
-		int previousMatch = -1;
-		while (chars.hasNext()) {
-			int cp = chars.nextInt();
-			if (cp == previousMatch) {
-				// swallow
-				continue;
-			}
-
-			String c = Character.toString(cp);
-			if (isCharset.matcher(c).matches()) {
-				previousMatch = cp;
-				// swallow the next
-			}
-			else {
-				previousMatch = -1;
-			}
-			result.append(c);
-		}
-		return values.string(result.toString());
-	}
-
 	public IString capitalize(IString src) {
 		StringBuilder result = new StringBuilder(src.length());
 		boolean lastWhitespace= true;

diff --git a/src/org/rascalmpl/library/String.rsc b/src/org/rascalmpl/library/String.rsc
@@ -519,8 +519,15 @@ import String;
 squeeze("hello", "el");
 ```
 }
-@javaClass{org.rascalmpl.library.Prelude}
-public java str squeeze(str src, str charSet);
+public str squeeze(str src, str charSet) {
+  if (charSet == "") {
+    return src;
+  }
+  return visit(src) {
+    case /<c:.><c>+/ => c
+      when /[<charSet>]/ := c
+  }
+}
 
 
 

diff --git a/src/org/rascalmpl/library/lang/rascal/tests/basic/Strings1.rsc b/src/org/rascalmpl/library/lang/rascal/tests/basic/Strings1.rsc
@@ -249,15 +249,15 @@ test bool tstSplit(str S1, str S2) = areOverlapping(S1,S2) || isEmpty(S1) || isE
 
 // squeeze
 
-private str rascalSqueeze(str s, str charSet) {
-  return visit (s) {
-    case /<c:.>(<c>)+/ => "<c>"
-      when /[<charSet>]/ := c
-  };
+test bool tstSqueeze1(str S) = /<c:[a-zA-Z]><c>/ !:= squeeze(S, "a-zA-Z");
+test bool tstSqueeze2(str S) = squeeze(S, "") == S;
+test bool tstSqueeze3(str S) {
+  if (/<c:[a-z]><c>/ := S) {
+    return /<c><c>/ := squeeze(S, "0-9");
+  }
+  return true;
 }
 
-test bool tstSqueeze1(str S) = rascalSqueeze(S, "a-zA-Z") == squeeze(S, "a-zA-Z");
-
 
 test bool tstStartsWith(str S1, str S2) = startsWith(S1+S2, S1);