Cheap isFill check and add --skip-filled-tiles option (#234)

planetiler-core/src/main/java/com/onthegomap/planetiler/VectorTile.java

@@ -495,6 +495,34 @@ public byte[] encode() {
     return tile.build().toByteArray();
   }
 
+  /**
+   * Returns true if this tile contains only polygon fills.
+   */
+  public boolean containsOnlyFills() {
+    return containsOnlyFillsOrEdges(false);
+  }
+
+  /**
+   * Returns true if this tile contains only polygon fills or horizontal/vertical edges that are likely to be repeated
+   * across tiles.
+   */
+  public boolean containsOnlyFillsOrEdges() {
+    return containsOnlyFillsOrEdges(true);
+  }
+
+  private boolean containsOnlyFillsOrEdges(boolean allowEdges) {
+    boolean empty = true;
+    for (var layer : layers.values()) {
+      for (var feature : layer.encodedFeatures) {
+        empty = false;
+        if (!feature.geometry.isFillOrEdge(allowEdges)) {
+          return false;
+        }
+      }
+    }
+    return !empty;
+  }
+
   private enum Command {
     MOVE_TO(1),
     LINE_TO(2),
@@ -519,12 +547,63 @@ private enum Command {
    */
   public record VectorGeometry(int[] commands, GeometryType geomType, int scale) {
 
+    private static final int LEFT = 1;
+    private static final int RIGHT = 1 << 1;
+    private static final int TOP = 1 << 2;
+    private static final int BOTTOM = 1 << 3;
+    private static final int INSIDE = 0;
+    private static final int ALL = TOP | LEFT | RIGHT | BOTTOM;
+
     public VectorGeometry {
       if (scale < 0) {
         throw new IllegalArgumentException("scale can not be less than 0, got: " + scale);
       }
     }
 
+    private static int getSide(int x, int y, int extent) {
+      int result = INSIDE;
+      if (x < 0) {
+        result |= LEFT;
+      } else if (x > extent) {
+        result |= RIGHT;
+      }
+      if (y < 0) {
+        result |= TOP;
+      } else if (y > extent) {
+        result |= BOTTOM;
+      }
+      return result;
+    }
+
+    private static boolean slanted(int x1, int y1, int x2, int y2) {
+      return x1 != x2 && y1 != y2;
+    }
+
+    private static boolean segmentCrossesTile(int x1, int y1, int x2, int y2, int extent) {
+      return (y1 >= 0 || y2 >= 0) &&
+        (y1 <= extent || y2 <= extent) &&
+        (x1 >= 0 || x2 >= 0) &&
+        (x1 <= extent || x2 <= extent);
+    }
+
+    private static boolean isSegmentInvalid(boolean allowEdges, int x1, int y1, int x2, int y2, int extent) {
+      boolean crossesTile = segmentCrossesTile(x1, y1, x2, y2, extent);
+      if (allowEdges) {
+        return crossesTile && slanted(x1, y1, x2, y2);
+      } else {
+        return crossesTile;
+      }
+    }
+
+
+    private static boolean visitedEnoughSides(boolean allowEdges, int sides) {
+      if (allowEdges) {
+        return ((sides & LEFT) > 0 && (sides & RIGHT) > 0) || ((sides & TOP) > 0 && (sides & BOTTOM) > 0);
+      } else {
+        return sides == ALL;
+      }
+    }
+
     /** Converts an encoded geometry back to a JTS geometry. */
     public Geometry decode() throws GeometryException {
       return decodeCommands(geomType, commands, scale);
@@ -566,6 +645,95 @@ public String toString() {
         "], geomType=" + geomType +
         " (" + geomType.asByte() + ")]";
     }
+
+    /** Returns true if the encoded geometry is a polygon fill. */
+    public boolean isFill() {
+      return isFillOrEdge(false);
+    }
+
+    /**
+     * Returns true if the encoded geometry is a polygon fill, rectangle edge, or part of a horizontal/vertical line
+     * that is likely to be repeated across tiles.
+     */
+    public boolean isFillOrEdge() {
+      return isFillOrEdge(true);
+    }
+
+    /**
+     * Returns true if the encoded geometry is a polygon fill, or if {@code allowEdges == true} then also a rectangle
+     * edge, or part of a horizontal/vertical line that is likely to be repeated across tiles.
+     */
+    public boolean isFillOrEdge(boolean allowEdges) {
+      if (geomType != GeometryType.POLYGON && (!allowEdges || geomType != GeometryType.LINE)) {
+        return false;
+      }
+
+      boolean isLine = geomType == GeometryType.LINE;
+
+      int extent = EXTENT << scale;
+      int visited = INSIDE;
+      int firstX = 0;
+      int firstY = 0;
+      int x = 0;
+      int y = 0;
+
+      int geometryCount = commands.length;
+      int length = 0;
+      int command = 0;
+      int i = 0;
+      while (i < geometryCount) {
+
+        if (length <= 0) {
+          length = commands[i++];
+          command = length & ((1 << 3) - 1);
+          length = length >> 3;
+          if (isLine && length > 2) {
+            return false;
+          }
+        }
+
+        if (length > 0) {
+          if (command == Command.CLOSE_PATH.value) {
+            if (isSegmentInvalid(allowEdges, x, y, firstX, firstY, extent) ||
+              !visitedEnoughSides(allowEdges, visited)) {
+              return false;
+            }
+            length--;
+            continue;
+          }
+
+          int dx = commands[i++];
+          int dy = commands[i++];
+
+          length--;
+
+          dx = zigZagDecode(dx);
+          dy = zigZagDecode(dy);
+
+          int nextX = x + dx;
+          int nextY = y + dy;
+
+          if (command == Command.MOVE_TO.value) {
+            firstX = nextX;
+            firstY = nextY;
+            if ((visited = getSide(firstX, firstY, extent)) == INSIDE) {
+              return false;
+            }
+          } else {
+            if (isSegmentInvalid(allowEdges, x, y, nextX, nextY, extent)) {
+              return false;
+            }
+            visited |= getSide(nextX, nextY, extent);
+          }
+          y = nextY;
+          x = nextX;
+        }
+
+      }
+
+      return visitedEnoughSides(allowEdges, visited);
+    }
+
   }
 
   /**

planetiler-core/src/main/java/com/onthegomap/planetiler/config/PlanetilerConfig.java

@@ -41,7 +41,8 @@ public record PlanetilerConfig(
   double simplifyToleranceAtMaxZoom,
   double simplifyToleranceBelowMaxZoom,
   boolean osmLazyReads,
-  boolean compactDb
+  boolean compactDb,
+  boolean skipFilledTiles
 ) {
 
   public static final int MIN_MINZOOM = 0;
@@ -142,7 +143,10 @@ public static PlanetilerConfig from(Arguments arguments) {
         false),
       arguments.getBoolean("compact_db",
         "Reduce the DB size by separating and deduping the tile data",
-        true)
+        true),
+      arguments.getBoolean("skip_filled_tiles",
+        "Skip writing tiles containing only polygon fills to the output",
+        false)
     );
   }
 

planetiler-core/src/main/java/com/onthegomap/planetiler/mbtiles/MbtilesWriter.java

@@ -49,7 +49,6 @@ public class MbtilesWriter {
   private static final Logger LOGGER = LoggerFactory.getLogger(MbtilesWriter.class);
   private static final long MAX_FEATURES_PER_BATCH = 10_000;
   private static final long MAX_TILES_PER_BATCH = 1_000;
-  private static final int MAX_FEATURES_HASHING_THRESHOLD = 5;
   private final Counter.Readable featuresProcessed;
   private final Counter memoizedTiles;
   private final Mbtiles db;
@@ -258,7 +257,9 @@ private void tileEncoder(Iterable<TileBatch> prev, Consumer<TileBatch> next) thr
      */
     byte[] lastBytes = null, lastEncoded = null;
     Long lastTileDataHash = null;
+    boolean lastIsFill = false;
     boolean compactDb = config.compactDb();
+    boolean skipFilled = config.skipFilledTiles();
 
     for (TileBatch batch : prev) {
       Queue<TileEncodingResult> result = new ArrayDeque<>(batch.size());
@@ -270,23 +271,30 @@ private void tileEncoder(Iterable<TileBatch> prev, Consumer<TileBatch> next) thr
         byte[] bytes, encoded;
         Long tileDataHash;
         if (tileFeatures.hasSameContents(last)) {
+          if (skipFilled && lastIsFill) {
+            continue;
+          }
           bytes = lastBytes;
           encoded = lastEncoded;
           tileDataHash = lastTileDataHash;
           memoizedTiles.inc();
         } else {
           VectorTile en = tileFeatures.getVectorTileEncoder();
-          encoded = en.encode();
-          bytes = gzip(encoded);
+          if (skipFilled) {
+            lastIsFill = en.containsOnlyFills();
+            if (lastIsFill) {
+              continue;
+            }
+          }
+          lastEncoded = encoded = en.encode();
+          lastBytes = bytes = gzip(encoded);
           last = tileFeatures;
-          lastEncoded = encoded;
-          lastBytes = bytes;
           if (encoded.length > 1_000_000) {
             LOGGER.warn("{} {}kb uncompressed",
               tileFeatures.tileCoord(),
               encoded.length / 1024);
           }
-          if (compactDb && tileFeatures.getNumFeaturesToEmit() < MAX_FEATURES_HASHING_THRESHOLD) {
+          if (compactDb && en.containsOnlyFillsOrEdges()) {
             tileDataHash = tileFeatures.generateContentHash();
           } else {
             tileDataHash = null;

planetiler-core/src/test/java/com/onthegomap/planetiler/PlanetilerTests.java

@@ -675,6 +675,25 @@ void testFullWorldPolygon() throws Exception {
     )).stream().map(d -> d.geometry().geom().norm()).toList());
   }
 
+  @Test
+  void testSkipFill() throws Exception {
+    var results = runWithReaderFeatures(
+      Map.of("threads", "1", "skip-filled-tiles", "true"),
+      List.of(
+        newReaderFeature(WORLD_POLYGON, Map.of())
+      ),
+      (in, features) -> features.polygon("layer")
+        .setZoomRange(0, 6)
+        .setBufferPixels(4)
+    );
+
+    assertEquals(481, results.tiles.size());
+    // spot-check one filled tile does not exist
+    assertNull(results.tiles.get(TileCoord.ofXYZ(
+      Z4_TILES / 2, Z4_TILES / 2, 4
+    )));
+  }
+
   @ParameterizedTest
   @CsvSource({
     "chesapeake.wkb, 4076",