JS: Add OffsetTranslation table (preserving behavior)

javascript/extractor/src/com/semmle/js/extractor/LocationManager.java

@@ -65,7 +65,8 @@ public class LocationManager {
 
   /**
    * Emit location information for an AST node. The node's location is translated from the parser's
-   * 0-based column numbering scheme into our 1-based scheme and then emitted as a snippet location.
+   * 0-based column numbering scheme with exclusive offsets into our 1-based scheme with inclusive
+   * end-offsets and then emitted as a snippet location.
    */
   public void emitNodeLocation(SourceElement nd, Label lbl) {
     int sl = nd.getLoc().getStart().getLine(),
@@ -86,7 +87,15 @@ public class LocationManager {
     emitSnippetLocation(lbl, sl, sc, el, ec);
   }
 
-  /** Emit a relative location in the current snippet. */
+  /**
+   * Emit a relative location in the current snippet.
+   *
+   * @param lbl label to associate with the location
+   * @param sl start line (1-based)
+   * @param sc start column (1-based, inclusive)
+   * @param el end line (1-based)
+   * @param ec end column (1-based, inclusive)
+   */
   public void emitSnippetLocation(Label lbl, int sl, int sc, int el, int ec) {
     Position start = translatePosition(new Position(sl, sc, -1));
     Position end = translatePosition(new Position(el, ec, -1));

javascript/extractor/src/com/semmle/js/extractor/OffsetTranslation.java

@@ -0,0 +1,39 @@
+package com.semmle.js.extractor;
+
+import com.semmle.util.data.IntList;
+
+/**
+ * A mapping of some source range into a set of intervals in an output source range.
+ *
+ * <p>The mapping is constructed by adding "anchors": input/output pairs that correspond to the
+ * beginning of an interval, which is assumed to end at the next anchor.
+ */
+public class OffsetTranslation {
+  private IntList anchors = IntList.create();
+  private IntList deltas = IntList.create();
+
+  /** Returns the mapping of x. */
+  public int get(int x) {
+    int index = anchors.binarySearch(x);
+    if (index < 0) {
+      // The insertion point is -index - 1.
+      // Get the index immediately before that.
+      index = -index - 2;
+      if (index < 0) {
+        // If queried before the first anchor, use the first anchor anyway.
+        index = 0;
+      }
+    }
+    return x + deltas.get(index);
+  }
+
+  /**
+   * Maps the given input offset to the given output offset.
+   *
+   * <p>This is added as an anchor. Any offset is mapped based on its closest preceding anchor.
+   */
+  public void set(int from, int to) {
+    anchors.add(from);
+    deltas.add(to - from);
+  }
+}

javascript/extractor/src/com/semmle/js/extractor/OffsetTranslationBuilder.java

@@ -0,0 +1,37 @@
+package com.semmle.js.extractor;
+
+import com.semmle.util.data.IntList;
+
+/**
+ * A mapping from integers to integers, is encoded as a sequence of consecutive intervals and their
+ * corresponding output intervals.
+ */
+public class OffsetTranslationBuilder {
+  private IntList anchors = IntList.create();
+  private IntList deltas = IntList.create();
+
+  /** Returns the mapping of x. */
+  public int get(int x) {
+    int index = anchors.binarySearch(x);
+    if (index < 0) {
+      // The insertion point is -index - 1.
+      // Get the index immediately before that.
+      index = -index - 2;
+      if (index < 0) {
+        // If queried before the first anchor, use the first anchor anyway.
+        index = 0;
+      }
+    }
+    return x + deltas.get(index);
+  }
+
+  /**
+   * Maps the given input offset to the given output offset.
+   *
+   * <p>This is added as an anchor. Any offset is mapped based on its closest preceding anchor.
+   */
+  public void set(int from, int to) {
+    anchors.add(from);
+    deltas.add(to - from);
+  }
+}

javascript/extractor/src/com/semmle/js/extractor/RegExpExtractor.java

@@ -51,6 +51,7 @@ public class RegExpExtractor {
   private final LocationManager locationManager;
   private final RegExpParser parser = new RegExpParser();
   private Position literalStart;
+  private OffsetTranslation offsets;
 
   public RegExpExtractor(TrapWriter trapwriter, LocationManager locationManager) {
     this.trapwriter = trapwriter;
@@ -121,8 +122,12 @@ public class RegExpExtractor {
   public void emitLocation(SourceElement term, Label lbl) {
     int sl, sc, el, ec;
     sl = el = literalStart.getLine();
-    sc = literalStart.getColumn() + 2 + term.getLoc().getStart().getColumn();
-    ec = literalStart.getColumn() + 1 + term.getLoc().getEnd().getColumn();
+    // the offset table accounts for the position on the line and for skipping the initial '/'
+    sc = offsets.get(term.getLoc().getStart().getColumn());
+    ec = offsets.get(term.getLoc().getEnd().getColumn());
+    sc += 1; // convert to 1-based
+    ec += 1; // convert to 1-based
+    ec -= 1; // convert to inclusive
     locationManager.emitSnippetLocation(lbl, sl, sc, el, ec);
   }
 
@@ -349,6 +354,8 @@ public class RegExpExtractor {
     }
 
     this.literalStart = parent.getLoc().getStart();
+    offsets = new OffsetTranslation();
+    offsets.set(0, literalStart.getColumn() + 1); // add 1 to skip the leading '/' or quote
     RegExpTerm ast = res.getAST();
     new V().visit(ast, trapwriter.localID(parent), 0);
 

javascript/extractor/src/com/semmle/js/extractor/test/OffsetTranslationTest.java

@@ -0,0 +1,50 @@
+package com.semmle.js.extractor.test;
+
+import com.semmle.js.extractor.OffsetTranslation;
+import org.junit.Assert;
+import org.junit.Test;
+
+public class OffsetTranslationTest {
+  @Test
+  public void testBasic() {
+    OffsetTranslation table = new OffsetTranslation();
+    table.set(0, 10);
+    table.set(100, 250);
+    Assert.assertEquals(10, table.get(0));
+    Assert.assertEquals(15, table.get(5));
+    Assert.assertEquals(85, table.get(75));
+    Assert.assertEquals(109, table.get(99));
+    Assert.assertEquals(250, table.get(100));
+    Assert.assertEquals(251, table.get(101));
+  }
+
+  @Test
+  public void testLookupBefore() {
+    OffsetTranslation table = new OffsetTranslation();
+    table.set(0, 10);
+    table.set(100, 250);
+    Assert.assertEquals(9, table.get(-1));
+  }
+
+  @Test
+  public void testIdentity() {
+    OffsetTranslation table = new OffsetTranslation();
+    table.set(0, 0);
+    Assert.assertEquals(0, table.get(0));
+    Assert.assertEquals(75, table.get(75));
+  }
+
+  @Test
+  public void testDuplicateAnchor() {
+    OffsetTranslation table = new OffsetTranslation();
+    table.set(0, 0);
+    table.set(10, 100);
+    table.set(10, 100);
+    table.set(20, 150);
+    Assert.assertEquals(1, table.get(1));
+    Assert.assertEquals(100, table.get(10));
+    Assert.assertEquals(101, table.get(11));
+    Assert.assertEquals(150, table.get(20));
+    Assert.assertEquals(151, table.get(21));
+  }
+}