Improve WORD primitive (#201)

* Finish collecting characters if new line. * Use parentheses as comment indicator. * Fix processing whole lines of comment
bahmanm · Dec 15, 2024 · 6332d1a · 6332d1a
1 parent 3c62261
commit 6332d1a
Show file tree

Hide file tree

Showing 2 changed files with 60 additions and 3 deletions.
diff --git a/bjforth/src/main/java/bjforth/primitives/WORD.java b/bjforth/src/main/java/bjforth/primitives/WORD.java
@@ -18,7 +18,10 @@
  */
 package bjforth.primitives;
 
+import static bjforth.primitives.PrimitiveFactory.KEY;
+
 import bjforth.machine.Machine;
+import bjforth.machine.MachineException;
 
 class WORD implements Primitive {
 
@@ -32,15 +35,18 @@ private static enum State {
   @Override
   public void execute(Machine machine) {
     var state = State.BEGIN;
-    var keyWord = PrimitiveFactory.KEY();
     var result = new StringBuilder();
+    int parenthesesCount = 0;
     while (state != State.END) {
-      keyWord.execute(machine);
+      KEY().execute(machine);
       var ch = (int) machine.popFromParameterStack();
       switch (state) {
         case BEGIN:
           if (ch == '\\') {
             state = State.IN_COMMENT;
+          } else if (ch == '(') {
+            state = State.IN_COMMENT;
+            parenthesesCount += 1;
           } else if (ch != ' ' && ch != '\n') {
             result.appendCodePoint(ch);
             state = State.IN_WORD;
@@ -49,10 +55,19 @@ public void execute(Machine machine) {
         case IN_COMMENT:
           if (ch == '\n') {
             state = State.BEGIN;
+          } else if (ch == ')') {
+            if (parenthesesCount > 0) {
+              parenthesesCount -= 1;
+            } else {
+              throw new MachineException("Imbalanced parentheses.");
+            }
+            if (parenthesesCount == 0) {
+              state = State.BEGIN;
+            }
           }
           break;
         case IN_WORD:
-          if (ch == ' ') {
+          if (ch == ' ' || ch == '\t' || ch == '\n') {
             state = State.END;
           } else {
             result.appendCodePoint(ch);

diff --git a/bjforth/src/test/java/bjforth/primitives/WORDTest.java b/bjforth/src/test/java/bjforth/primitives/WORDTest.java
@@ -115,4 +115,46 @@ void skipsComments() {
     // THEN
     assertThat(actualState).hasParameterStackEqualTo(aParameterStack().with(wordStr).build());
   }
+
+  @DisplayName("Finish collecting characters if current character is new line.")
+  @Test
+  void endsWordWhenNewLine() {
+    // GIVEN
+    var wordStr = RandomStringUtils.randomAlphanumeric(10);
+    var str = wordStr + '\n';
+    var inputStream = new ByteArrayInputStream(str.getBytes());
+    System.setIn(inputStream);
+
+    var WORDaddr = getPrimitiveAddress("WORD");
+    var actualState = aMachineState().withInstrcutionPointer(WORDaddr).build();
+    var machine = aMachine().withState(actualState).build();
+    var referenceState = aMachineState().copyFrom(actualState).build();
+
+    // WHEN
+    machine.step();
+
+    // THEN
+    assertThat(actualState).hasParameterStackEqualTo(aParameterStack().with(wordStr).build());
+  }
+
+  @DisplayName("Treat balanced pairs of parentheses and the content between them as comments.")
+  @Test
+  void parensAsComments() {
+    // GIVEN
+    var wordStr = RandomStringUtils.randomAlphanumeric(10);
+    var str = wordStr + " ( this is a comment)" + '\n';
+    var inputStream = new ByteArrayInputStream(str.getBytes());
+    System.setIn(inputStream);
+
+    var WORDaddr = getPrimitiveAddress("WORD");
+    var actualState = aMachineState().withInstrcutionPointer(WORDaddr).build();
+    var machine = aMachine().withState(actualState).build();
+    var referenceState = aMachineState().copyFrom(actualState).build();
+
+    // WHEN
+    machine.step();
+
+    // THEN
+    assertThat(actualState).hasParameterStackEqualTo(aParameterStack().with(wordStr).build());
+  }
 }