Merge with main

PRL-PRG · Feb 13, 2024 · 0dbe2c2 · 0dbe2c2
1 parent 7cdb25d
commit 0dbe2c2
Show file tree

Hide file tree

Showing 9 changed files with 69 additions and 29 deletions.
diff --git a/src/main/java/org/prlprg/bc/Compiler.java b/src/main/java/org/prlprg/bc/Compiler.java
@@ -1090,12 +1090,27 @@ private boolean inlineDotCall(LangSXP call) {
 
   private boolean inlineMultiColon(LangSXP call) {
     if (!dotsOrMissing(call.args()) && call.args().size() == 2) {
-      if (call.arg(0).value() instanceof StrOrRegSymSXP s1 &&
-              call.arg(1).value() instanceof StrOrRegSymSXP s2) {
-        var args = SEXPs.list(SEXPs.string(s1.reifyString()), SEXPs.string(s2.reifyString()));
-        compileCallSymFun((RegSymSXP) call.fun(), args, call);
-        return true;
+
+      // FIXME: ugly
+      String s1 = switch(call.arg(0).value()) {
+        case StrSXP s when s.size() == 1 -> s.get(0);
+        case RegSymSXP s -> s.name();
+        default -> null;
+      };
+
+      String s2 = switch(call.arg(1).value()) {
+        case StrSXP s when s.size() == 1 -> s.get(0);
+        case RegSymSXP s -> s.name();
+        default -> null;
+      };
+
+      if (s1 == null || s2 == null) {
+        return false;
       }
+
+      var args = SEXPs.list(SEXPs.string(s1), SEXPs.string(s2));
+      compileCallSymFun((RegSymSXP) call.fun(), args, call);
+      return true;
     }
 
     return false;

diff --git a/src/main/java/org/prlprg/bc/Context.java b/src/main/java/org/prlprg/bc/Context.java
@@ -103,30 +103,28 @@ public Set<String> findLocals(SEXP e) {
     // the code is following compiler:::findLocals1 from R
     while (!todo.isEmpty()) {
       var elem = todo.removeFirst();
-      if (elem instanceof LangSXP l && l.fun() instanceof RegSymSXP s) {
+      if (elem instanceof LangSXP l && l.fun() instanceof RegSymSXP fun) {
+        var args = l.args().values();
         var local =
-            switch (s.name()) {
+            switch (fun.name()) {
               case "=", "<-" -> {
-                var args = l.args().values();
                 todo.addAll(args.subList(1, args.size()));
                 yield getAssignedVar(l);
               }
               case "for" -> {
-                var args = l.args().values();
                 todo.addAll(args.subList(1, args.size()));
 
-                var sym = (RegSymSXP) l.arg(0).value();
+                var sym = (RegSymSXP) args.getFirst();
                 yield Optional.of(sym.name());
               }
               case "assign", "delayedAssign" -> {
                 // The variable in assign and delayedAssign expressions is considered
                 // local if it is an explicit
                 // character string and there is no environment argument.
-                var args = l.args().values();
                 todo.addAll(args.subList(1, args.size()));
 
                 if (args.size() == 2 && args.getFirst() instanceof StrOrRegSymSXP v) {
-                  yield Optional.of(v.reifyString());
+                  yield v.reifyString();
                 } else {
                   yield Optional.<String>empty();
                 }
@@ -142,27 +140,26 @@ public Set<String> findLocals(SEXP e) {
               case "~", "expression", "quote" -> {
                 // they do not evaluate their arguments and so do not contribute new local
                 // variables.
-                if (shadowed.contains(s.name()) || locals.contains(s.name())) {
-                  todo.addAll(l.args().values());
+                if (shadowed.contains(fun.name()) || locals.contains(fun.name())) {
+                  todo.addAll(args);
                 }
                 yield Optional.<String>empty();
               }
               case "local" -> {
-
                 // local calls without an environment argument create a new environment
                 // for evaluating their expression and do not add new local variables.
                 // If an environment argument is present then this might be the current
                 // environment and so assignments in the expression are considered to
                 // create possible local variables.
-                if (shadowed.contains(s.name())
-                    || locals.contains(s.name())
-                    || l.args().size() != 1) {
-                  todo.addAll(l.args().values());
+                if (shadowed.contains(fun.name())
+                    || locals.contains(fun.name())
+                    || args.size() != 1) {
+                  todo.addAll(args);
                 }
                 yield Optional.<String>empty();
               }
               default -> {
-                todo.addAll(l.args().values());
+                todo.addAll(args);
                 yield Optional.<String>empty();
               }
             };
@@ -180,7 +177,7 @@ private static Optional<String> getAssignedVar(LangSXP l) {
     if (v == SEXPs.MISSING_ARG) {
       throw new CompilerException("Bad assignment: " + l);
     } else if (v instanceof StrOrRegSymSXP s) {
-      return Optional.of(s.reifyString());
+      return s.reifyString();
     } else {
       if (l.args().isEmpty()) {
         throw new CompilerException("Bad assignment: " + l);
@@ -190,7 +187,7 @@ private static Optional<String> getAssignedVar(LangSXP l) {
           return getAssignedVar(ll);
         }
         case StrOrRegSymSXP s -> {
-          return Optional.of(s.reifyString());
+          return s.reifyString();
         }
         default -> {
           throw new CompilerException("Bad assignment: " + l);

diff --git a/src/main/java/org/prlprg/rds/RDSException.java b/src/main/java/org/prlprg/rds/RDSException.java
@@ -1,7 +1,7 @@
 package org.prlprg.rds;
 
 /** Thrown when deserializing R objects from malformed RDS. */
-public class RDSException extends Exception {
+public class RDSException extends RuntimeException {
   RDSException(String message) {
     super(message);
   }

diff --git a/src/main/java/org/prlprg/sexp/RegSymSXP.java b/src/main/java/org/prlprg/sexp/RegSymSXP.java
@@ -2,6 +2,7 @@
 
 import com.google.common.base.Objects;
 import com.google.common.collect.ImmutableList;
+import java.util.Optional;
 
 /** Symbol which isn't "unbound value" or "missing arg" */
 public final class RegSymSXP implements SymSXP, StrOrRegSymSXP {
@@ -27,6 +28,10 @@ public String name() {
     return name;
   }
 
+  public Optional<String> reifyString() {
+    return Optional.of(name);
+  }
+
   /** Whether this is a double-dot symbol ({@code ..0}, {@code ..1}, ...). */
   public boolean isDdSym() {
     return name.startsWith("..") && name.substring(2).chars().allMatch(Character::isDigit);

diff --git a/src/main/java/org/prlprg/sexp/SEXP.java b/src/main/java/org/prlprg/sexp/SEXP.java
@@ -1,5 +1,6 @@
 package org.prlprg.sexp;
 
+import java.util.Objects;
 import javax.annotation.Nullable;
 
 /**
@@ -10,7 +11,7 @@
  * suspect GNU-R SEXPs aren't actually S-expressions.
  */
 public sealed interface SEXP
-    permits StrOrRegSymSXP, SymOrLangSXP, ListOrVectorSXP, CloSXP, EnvSXP, BCodeSXP {
+    permits StrOrRegSymSXP, SymOrLangSXP, ListOrVectorSXP, CloSXP, EnvSXP, BCodeSXP, PromSXP {
   /**
    * SEXPTYPE. It's important to distinguish these from the SEXP's class, because there's a class
    * for every type but not vice versa due to subclasses (e.g. simple-scalar ints have the same
@@ -39,7 +40,7 @@ default SEXP withAttributes(Attributes attributes) {
   }
 
   default SEXP withClass(String name) {
-    var attrs = attributes().including("class", new SimpleStrSXPImpl(name));
+    var attrs = Objects.requireNonNull(attributes()).including("class", SEXPs.string(name));
     return withAttributes(attrs);
   }
 }
diff --git a/src/main/java/org/prlprg/sexp/SimpleStrSXP.java b/src/main/java/org/prlprg/sexp/SimpleStrSXP.java
@@ -1,5 +1,7 @@
 package org.prlprg.sexp;
 
+import java.util.Optional;
+
 /** Simple scalar string = vector of size 1 with no ALTERP, ATTRIB, or OBJECT. */
 public final class SimpleStrSXP extends SimpleScalarSXPImpl<String> implements StrSXP {
   SimpleStrSXP(String data) {
@@ -20,4 +22,9 @@ public String toString() {
   public StrSXP withAttributes(Attributes attributes) {
     return SEXPs.string(data, attributes);
   }
+
+  @Override
+  public Optional<String> reifyString() {
+    return Optional.of(data);
+  }
 }
diff --git a/src/main/java/org/prlprg/sexp/StrOrRegSymSXP.java b/src/main/java/org/prlprg/sexp/StrOrRegSymSXP.java
@@ -1,5 +1,6 @@
 package org.prlprg.sexp;
 
+import java.util.Optional;
 import javax.annotation.concurrent.Immutable;
 
 /**
@@ -11,4 +12,8 @@
  * org.prlprg.util.Either} instead...
  */
 @Immutable
-public sealed interface StrOrRegSymSXP extends SEXP permits StrSXP, RegSymSXP {}
+public sealed interface StrOrRegSymSXP extends SEXP permits StrSXP, RegSymSXP {
+  // FIXME: this is not great, but far the most often used pattern matching is
+  //  on either symbol or a scalar string
+  public Optional<String> reifyString();
+}
diff --git a/src/main/java/org/prlprg/sexp/StrSXP.java b/src/main/java/org/prlprg/sexp/StrSXP.java
@@ -4,6 +4,7 @@
 import com.google.common.collect.UnmodifiableIterator;
 import com.google.common.escape.Escaper;
 import com.google.common.escape.Escapers;
+import java.util.Optional;
 import javax.annotation.concurrent.Immutable;
 import org.prlprg.primitive.Constants;
 
@@ -49,6 +50,11 @@ public String toString() {
   public StrSXP withAttributes(Attributes attributes) {
     return SEXPs.string(data, attributes);
   }
+
+  @Override
+  public Optional<String> reifyString() {
+    return size() == 1 ? Optional.of(get(0)) : Optional.empty();
+  }
 }
 
 /** Empty string vector with no ALTREP, ATTRIB, or OBJECT. */
@@ -61,6 +67,11 @@ private EmptyStrSXPImpl() {}
   public StrSXP withAttributes(Attributes attributes) {
     return SEXPs.string(ImmutableList.of(), attributes);
   }
+
+  @Override
+  public Optional<String> reifyString() {
+    return Optional.empty();
+  }
 }
 
 final class StrSXPs {

diff --git a/src/test/java/org/prlprg/rds/RDSReaderTest.java b/src/test/java/org/prlprg/rds/RDSReaderTest.java
@@ -6,7 +6,6 @@
 
 import java.util.Objects;
 import org.junit.jupiter.api.Test;
-import org.prlprg.RPlatform;
 import org.prlprg.RSession;
 import org.prlprg.primitive.Constants;
 import org.prlprg.primitive.Logical;
@@ -27,12 +26,12 @@ public void testInts() throws Exception {
 
     if (sexp instanceof IntSXP ints) {
       assertEquals(6, ints.size());
-      assertEquals(-RPlatform.INT_MAX, ints.get(0));
+      assertEquals(Constants.INT_MIN, ints.get(0));
       assertEquals(-1, ints.get(1));
       assertEquals(0, ints.get(2));
       assertEquals(Constants.NA_INT, ints.get(3));
       assertEquals(1, ints.get(4));
-      assertEquals(RPlatform.INT_MAX, ints.get(5));
+      assertEquals(Integer.MAX_VALUE, ints.get(5));
     } else {
       fail("Expected IntSXP");
     }