Switch to iterative version of WKT format parser (opensearch-project#…

…14086) Signed-off-by: Heemin Kim <heemin@amazon.com>
wdongyu · Aug 22, 2024 · 51c9927 · 51c9927
1 parent 6890f4e
commit 51c9927
Show file tree

Hide file tree

Showing 3 changed files with 94 additions and 6 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -28,6 +28,7 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 
 ### Fixed
 - Fix handling of Short and Byte data types in ScriptProcessor ingest pipeline ([#14379](https://github.com/opensearch-project/OpenSearch/issues/14379))
+- Switch to iterative version of WKT format parser ([#14086](https://github.com/opensearch-project/OpenSearch/pull/14086))
 
 ### Security
 

diff --git a/libs/geo/src/main/java/org/opensearch/geometry/utils/WellKnownText.java b/libs/geo/src/main/java/org/opensearch/geometry/utils/WellKnownText.java
@@ -49,8 +49,10 @@
 import java.io.StreamTokenizer;
 import java.io.StringReader;
 import java.text.ParseException;
+import java.util.ArrayDeque;
 import java.util.ArrayList;
 import java.util.Collections;
+import java.util.Deque;
 import java.util.List;
 import java.util.Locale;
 
@@ -67,6 +69,7 @@ public class WellKnownText {
     public static final String RPAREN = ")";
     public static final String COMMA = ",";
     public static final String NAN = "NaN";
+    public static final int MAX_DEPTH_OF_GEO_COLLECTION = 1000;
 
     private final String NUMBER = "<NUMBER>";
     private final String EOF = "END-OF-STREAM";
@@ -278,6 +281,16 @@ public Geometry fromWKT(String wkt) throws IOException, ParseException {
      */
     private Geometry parseGeometry(StreamTokenizer stream) throws IOException, ParseException {
         final String type = nextWord(stream).toLowerCase(Locale.ROOT);
+        switch (type) {
+            case "geometrycollection":
+                return parseGeometryCollection(stream);
+            default:
+                return parseSimpleGeometry(stream, type);
+        }
+    }
+
+    private Geometry parseSimpleGeometry(StreamTokenizer stream, String type) throws IOException, ParseException {
+        assert "geometrycollection".equals(type) == false;
         switch (type) {
             case "point":
                 return parsePoint(stream);
@@ -294,7 +307,7 @@ private Geometry parseGeometry(StreamTokenizer stream) throws IOException, Parse
             case "bbox":
                 return parseBBox(stream);
             case "geometrycollection":
-                return parseGeometryCollection(stream);
+                throw new IllegalStateException("Unexpected type: geometrycollection");
             case "circle": // Not part of the standard, but we need it for internal serialization
                 return parseCircle(stream);
         }
@@ -305,12 +318,56 @@ private GeometryCollection<Geometry> parseGeometryCollection(StreamTokenizer str
         if (nextEmptyOrOpen(stream).equals(EMPTY)) {
             return GeometryCollection.EMPTY;
         }
-        List<Geometry> shapes = new ArrayList<>();
-        shapes.add(parseGeometry(stream));
-        while (nextCloserOrComma(stream).equals(COMMA)) {
-            shapes.add(parseGeometry(stream));
+
+        List<Geometry> topLevelShapes = new ArrayList<>();
+        Deque<List<Geometry>> deque = new ArrayDeque<>();
+        deque.push(topLevelShapes);
+        boolean isFirstIteration = true;
+        List<Geometry> currentLevelShapes = null;
+        while (!deque.isEmpty()) {
+            List<Geometry> previousShapes = deque.pop();
+            if (currentLevelShapes != null) {
+                previousShapes.add(new GeometryCollection<>(currentLevelShapes));
+            }
+            currentLevelShapes = previousShapes;
+
+            if (isFirstIteration == true) {
+                isFirstIteration = false;
+            } else {
+                if (nextCloserOrComma(stream).equals(COMMA) == false) {
+                    // Done with current level, continue with parent level
+                    continue;
+                }
+            }
+            while (true) {
+                final String type = nextWord(stream).toLowerCase(Locale.ROOT);
+                if (type.equals("geometrycollection")) {
+                    if (nextEmptyOrOpen(stream).equals(EMPTY) == false) {
+                        // GEOMETRYCOLLECTION() -> 1 depth, GEOMETRYCOLLECTION(GEOMETRYCOLLECTION()) -> 2 depth
+                        // When parsing the top level geometry collection, the queue size is zero.
+                        // When max depth is 1, we don't want to push any sub geometry collection in the queue.
+                        // Therefore, we subtract 2 from max depth.
+                        if (deque.size() >= MAX_DEPTH_OF_GEO_COLLECTION - 2) {
+                            throw new IllegalArgumentException(
+                                "a geometry collection with a depth greater than " + MAX_DEPTH_OF_GEO_COLLECTION + " is not supported"
+                            );
+                        }
+                        deque.push(currentLevelShapes);
+                        currentLevelShapes = new ArrayList<>();
+                        continue;
+                    }
+                    currentLevelShapes.add(GeometryCollection.EMPTY);
+                } else {
+                    currentLevelShapes.add(parseSimpleGeometry(stream, type));
+                }
+
+                if (nextCloserOrComma(stream).equals(COMMA) == false) {
+                    break;
+                }
+            }
         }
-        return new GeometryCollection<>(shapes);
+
+        return new GeometryCollection<>(topLevelShapes);
     }
 
     private Point parsePoint(StreamTokenizer stream) throws IOException, ParseException {

diff --git a/libs/geo/src/test/java/org/opensearch/geometry/GeometryCollectionTests.java b/libs/geo/src/test/java/org/opensearch/geometry/GeometryCollectionTests.java
@@ -62,6 +62,11 @@ public void testBasicSerialization() throws IOException, ParseException {
 
         assertEquals("GEOMETRYCOLLECTION EMPTY", wkt.toWKT(GeometryCollection.EMPTY));
         assertEquals(GeometryCollection.EMPTY, wkt.fromWKT("GEOMETRYCOLLECTION EMPTY)"));
+
+        assertEquals(
+            new GeometryCollection<Geometry>(Arrays.asList(GeometryCollection.EMPTY)),
+            wkt.fromWKT("GEOMETRYCOLLECTION (GEOMETRYCOLLECTION EMPTY)")
+        );
     }
 
     @SuppressWarnings("ConstantConditions")
@@ -86,4 +91,29 @@ public void testInitValidation() {
 
         new StandardValidator(true).validate(new GeometryCollection<Geometry>(Collections.singletonList(new Point(20, 10, 30))));
     }
+
+    public void testDeeplyNestedGeometryCollection() throws IOException, ParseException {
+        WellKnownText wkt = new WellKnownText(true, new GeographyValidator(true));
+        StringBuilder validGeometryCollectionHead = new StringBuilder("GEOMETRYCOLLECTION");
+        StringBuilder validGeometryCollectionTail = new StringBuilder(" EMPTY");
+        for (int i = 0; i < WellKnownText.MAX_DEPTH_OF_GEO_COLLECTION - 1; i++) {
+            validGeometryCollectionHead.append(" (GEOMETRYCOLLECTION");
+            validGeometryCollectionTail.append(")");
+        }
+        // Expect no exception
+        wkt.fromWKT(validGeometryCollectionHead.append(validGeometryCollectionTail).toString());
+
+        StringBuilder invalidGeometryCollectionHead = new StringBuilder("GEOMETRYCOLLECTION");
+        StringBuilder invalidGeometryCollectionTail = new StringBuilder(" EMPTY");
+        for (int i = 0; i < WellKnownText.MAX_DEPTH_OF_GEO_COLLECTION; i++) {
+            invalidGeometryCollectionHead.append(" (GEOMETRYCOLLECTION");
+            invalidGeometryCollectionTail.append(")");
+        }
+
+        IllegalArgumentException ex = expectThrows(
+            IllegalArgumentException.class,
+            () -> wkt.fromWKT(invalidGeometryCollectionHead.append(invalidGeometryCollectionTail).toString())
+        );
+        assertEquals("a geometry collection with a depth greater than 1000 is not supported", ex.getMessage());
+    }
 }