Updated grid benchmark

SmartDataAnalytics · Mar 13, 2024 · e63e574 · e63e574
1 parent 98687b0
commit e63e574
Show file tree

Hide file tree

Showing 6 changed files with 140 additions and 50 deletions.
diff --git a/rdf-processing-toolkit-cli/src/main/java/org/aksw/bench/geo/cmd/Cell2D.java b/rdf-processing-toolkit-cli/src/main/java/org/aksw/bench/geo/cmd/Cell2D.java
@@ -2,4 +2,8 @@
 
 import org.locationtech.jts.geom.Envelope;
 
-public record Cell2D(Grid2D grid, int row, int col, Envelope envelope) {}
+public record Cell2D(Grid2D grid, int row, int col, Envelope envelope) {
+    public int id() {
+        return row * grid.sizeY + col;
+    }
+}
diff --git a/rdf-processing-toolkit-cli/src/main/java/org/aksw/bench/geo/cmd/CmdBenchGenGridDataGen.java b/rdf-processing-toolkit-cli/src/main/java/org/aksw/bench/geo/cmd/CmdBenchGenGridDataGen.java
@@ -38,20 +38,6 @@ public class CmdBenchGenGridDataGen
     @ArgGroup(exclusive = false)
     protected GridOffsets gridOffsets = new GridOffsets();
 
-    static class GridOffsets {
-        @Option(names = "--minX", required = true, defaultValue = "-180")
-        protected double minX;
-
-        @Option(names = "--maxX", required = true, defaultValue = "180")
-        protected double maxX;
-
-        @Option(names = "--minY", required = true, defaultValue = "-90")
-        protected double minY;
-
-        @Option(names = "--maxY", required = true, defaultValue = "90")
-        protected double maxY;
-    }
-
     @Option(names = "--rows", required = true, defaultValue = "2")
     protected int rows;
 
@@ -126,16 +112,16 @@ public static Node toNode(Geometry geom) {
 //    }
 
     public static String genGraphName(int id) {
-        return "https://www.example.org/graph/" + id;
+        return "http://www.example.org/graph/" + id;
     }
 
     @Override
     public Integer call() throws Exception {
         Grid2D grid = Grid2D.newBuilder()
             .setMinX(gridOffsets.minX)
             .setMaxX(gridOffsets.maxX)
-            .setMinX(gridOffsets.minY)
-            .setMaxX(gridOffsets.maxY)
+            .setMinY(gridOffsets.minY)
+            .setMaxY(gridOffsets.maxY)
             .setRowCount(rows)
             .setColCount(cols)
             .build();
@@ -146,9 +132,9 @@ public Integer call() throws Exception {
                 // Always make cells a bit smaller to avoid corner cases with touching envelopes
                 double ratio = 0.95 * ratioBase;
                 Node graph = NodeFactory.createURI(genGraphName(g));
-                return grid.stream().flatMap(cell -> {
-                    Node feature = NodeFactory.createURI("https://www.example.org/feature/" + g + "/" + cell.row() + "/" + cell.col());
-                    Node geom = NodeFactory.createURI("https://www.example.org/geometry/" + g + "/" + cell.row() + "/" + cell.col());
+                return grid.stream(true).flatMap(cell -> {
+                    Node feature = NodeFactory.createURI("http://www.example.org/feature/" + g + "/" + cell.row() + "/" + cell.col());
+                    Node geom = NodeFactory.createURI("http://www.example.org/geometry/" + g + "/" + cell.row() + "/" + cell.col());
 
                     Geometry cellGeom = toGeometry(cell.envelope());
                     Geometry scaledCellGeom = ratio != 1.0
@@ -168,7 +154,7 @@ public Integer call() throws Exception {
         try (OutputStream out = StdIo.openStdOutWithCloseShield()) {
             StreamRDF writer = StreamRDFWriter.getWriterStream(out, RDFFormat.NQUADS);
             writer.start();
-            quads.forEach(writer::quad);
+            quads.sequential().forEach(writer::quad);
             writer.finish();
             out.flush();
         }

diff --git a/rdf-processing-toolkit-cli/src/main/java/org/aksw/bench/geo/cmd/CmdBenchGenGridQueryGen.java b/rdf-processing-toolkit-cli/src/main/java/org/aksw/bench/geo/cmd/CmdBenchGenGridQueryGen.java
@@ -6,6 +6,8 @@
 
 import org.aksw.commons.io.util.StdIo;
 import org.aksw.commons.picocli.CmdCommonBase;
+import org.aksw.jenax.arq.util.syntax.QueryUtils;
+import org.aksw.jenax.arq.util.var.Vars;
 import org.apache.jena.geosparql.implementation.GeometryWrapper;
 import org.apache.jena.geosparql.implementation.jts.CustomGeometryFactory;
 import org.apache.jena.geosparql.implementation.vocabulary.Geo;
@@ -14,6 +16,13 @@
 import org.apache.jena.query.Query;
 import org.apache.jena.query.QueryFactory;
 import org.apache.jena.riot.out.NodeFmtLib;
+import org.apache.jena.sparql.algebra.Algebra;
+import org.apache.jena.sparql.algebra.Op;
+import org.apache.jena.sparql.algebra.OpAsQuery;
+import org.apache.jena.sparql.algebra.Transform;
+import org.apache.jena.sparql.algebra.TransformCopy;
+import org.apache.jena.sparql.algebra.Transformer;
+import org.apache.jena.sparql.algebra.op.OpGraph;
 import org.locationtech.jts.geom.Envelope;
 
 import picocli.CommandLine.ArgGroup;
@@ -28,33 +37,43 @@ public class CmdBenchGenGridQueryGen
     @ArgGroup(exclusive = false)
     protected GridOffsets gridOffsets = new GridOffsets();
 
-    static class GridOffsets {
-        @Option(names = "--minX", required = true, defaultValue = "-180")
-        protected double minX;
-
-        @Option(names = "--maxX", required = true, defaultValue = "180")
-        protected double maxX;
-
-        @Option(names = "--minY", required = true, defaultValue = "-90")
-        protected double minY;
-
-        @Option(names = "--maxY", required = true, defaultValue = "90")
-        protected double maxY;
-    }
-
     @Option(names = "--rows", required = true, defaultValue = "2")
     protected int rows;
 
     @Option(names = "--cols", required = true, defaultValue = "2")
     protected int cols;
 
+    @Option(names = "--graphs", required = true, defaultValue = "2")
+    protected int graphs;
+
+//    @ArgGroup(exclusive = true, multiplicity = "0..1")
+//    public GraphSpec outputSpec;
+//
+//    public static class GraphSpec {
+//        /**
+//         * sparql-pattern file
+//         *
+//         */
+//        @Option(names = { "-o", "--out-file" }, description = "output file")
+//        public String outFile;
+//
+//        @Option(names = { "--io", },  description = "overwrites argument file on success with output; use with care")
+//        public String inOutFile = null;
+//    }
+
     @Option(names = "--allGraphs", required = true, defaultValue = "false")
     protected boolean allGraphs;
 
+    @Option(names = { "-u", "--union-default-graph" }, defaultValue = "false", fallbackValue = "true", description = "If --allGraphs is given. Use union default graph rather than ?g.")
+    protected boolean unionDefaultGraphMode;
 
-    public static Query genQuery(Cell2D cell, String fixedGraph) {
+    @Option(names = "--reverse", required = true, defaultValue = "false", fallbackValue = "true", description = "Generate queries by traversing grid cells from the last to the first.")
+    protected boolean reverse;
 
-        String g = fixedGraph == null ? "?g" : NodeFmtLib.strNT(NodeFactory.createURI(fixedGraph));
+    public static Query genQuery(Cell2D cell, Node graphNode) {
+
+        String gStr = graphNode == null ? "?dummy" : NodeFmtLib.strNT(graphNode);
+        // NodeFmtLib.strNT(NodeFactory.createURI(fixedGraph));
 
         String queryStr = """
             PREFIX geo: <http://www.opengis.net/ont/geosparql#>
@@ -71,11 +90,26 @@ public static Query genQuery(Cell2D cell, String fixedGraph) {
             }
             """
                 .replace("$CELLGEOM$", NodeFmtLib.strNT(toNode(cell.envelope())))
-                .replace("$GRAPH$", g)
+                .replace("$GRAPH$", gStr)
                 ;
 
-        System.out.println(queryStr);
         Query result = QueryFactory.create(queryStr);
+
+        // Drop GRAPH ?foo { } blocks if no graph is given
+        if (graphNode == null) {
+            Transform xform  = new TransformCopy() {
+                @Override
+                public Op transform(OpGraph opGraph, Op subOp) {
+                    return subOp;
+                }
+            };
+
+            Op op = Algebra.compile(result);
+            op = Transformer.transform(xform, op);
+            Query tmp = OpAsQuery.asQuery(op);
+            result = QueryUtils.restoreQueryForm(tmp, result);
+        }
+
         return result;
     }
 
@@ -88,18 +122,29 @@ public Integer call() throws Exception {
         Grid2D grid = Grid2D.newBuilder()
             .setMinX(gridOffsets.minX)
             .setMaxX(gridOffsets.maxX)
-            .setMinX(gridOffsets.minY)
-            .setMaxX(gridOffsets.maxY)
+            .setMinY(gridOffsets.minY)
+            .setMaxY(gridOffsets.maxY)
             .setRowCount(rows)
             .setColCount(cols)
             .build();
 
-        String fixedGraph = allGraphs ? null : CmdBenchGenGridDataGen.genGraphName(0);
-
-        Stream<Query> queries = grid.stream().map(cell -> genQuery(cell, fixedGraph));
+        Stream<Query> queries = grid.stream(!reverse).map(cell -> {
+            Node graphNode;
+            if (allGraphs) {
+                graphNode = unionDefaultGraphMode
+                        ? null
+                        : Vars.g;
+            } else {
+                int cellId = cell.id();
+                int gid = cellId % graphs;
+                graphNode = NodeFactory.createURI(CmdBenchGenGridDataGen.genGraphName(gid));
+            }
+            Query r = genQuery(cell, graphNode);
+            return r;
+        });
 
         try (PrintWriter writer = new PrintWriter(StdIo.openStdOutWithCloseShield())) {
-            queries.forEach(query -> writer.println(query));
+            queries.sequential().forEach(query -> writer.println(query));
             writer.flush();
         }
         return 0;

diff --git a/rdf-processing-toolkit-cli/src/main/java/org/aksw/bench/geo/cmd/CmdGridBenchInit.java b/rdf-processing-toolkit-cli/src/main/java/org/aksw/bench/geo/cmd/CmdGridBenchInit.java
@@ -129,7 +129,7 @@ public Integer call() throws Exception {
         Stream<Quad> quads =
             IntStream.range(0, graphs).boxed().flatMap(g -> {
                 Node graph = NodeFactory.createURI("https://www.example.org/graph/" + g);
-                return grid.stream().flatMap(cell -> {
+                return grid.stream(true).flatMap(cell -> {
                     Node feature = NodeFactory.createURI("https://www.example.org/feature/" + graph + "/" + cell.row() + "/" + cell.col());
                     Node geom = NodeFactory.createURI("https://www.example.org/geometry/" + graph + "/" + cell.row() + "/" + cell.col());
                     Node wkt = toNode(cell.envelope());

diff --git a/rdf-processing-toolkit-cli/src/main/java/org/aksw/bench/geo/cmd/Grid2D.java b/rdf-processing-toolkit-cli/src/main/java/org/aksw/bench/geo/cmd/Grid2D.java
@@ -31,9 +31,15 @@ public Grid2D(double worldMinX, double worldMaxX, double worldMinY, double world
         this.deltaY = (worldMaxY - worldMinY) / sizeY;
     }
 
-    public Stream<Cell2D> stream() {
-        return IntStream.range(0, sizeY).mapToObj(row -> row).flatMap(row ->
-            IntStream.range(0, sizeX).mapToObj(col -> cellOf(row, col)));
+    public static IntStream range(int startInclusive, int endExclusive, boolean isForward) {
+        return isForward
+                ? IntStream.range(startInclusive, endExclusive)
+                : IntStream.iterate(endExclusive - 1, i -> i >= startInclusive, i -> i - 1);
+    }
+
+    public Stream<Cell2D> stream(boolean isForward) {
+        return range(0, sizeY, isForward).mapToObj(row -> row).flatMap(row ->
+            range(0, sizeX, isForward).mapToObj(col -> cellOf(row, col)));
     }
 
     public Cell2D cellOf(int row, int col) {

diff --git a/rdf-processing-toolkit-cli/src/main/java/org/aksw/bench/geo/cmd/GridOffsets.java b/rdf-processing-toolkit-cli/src/main/java/org/aksw/bench/geo/cmd/GridOffsets.java
@@ -0,0 +1,49 @@
+package org.aksw.bench.geo.cmd;
+
+import picocli.CommandLine.Option;
+
+public class GridOffsets {
+    @Option(names = "--minX", defaultValue = "-90") // -180
+    protected double minX;
+
+    @Option(names = "--maxX", defaultValue = "90") // 180
+    protected double maxX;
+
+    @Option(names = "--minY", defaultValue = "-90")
+    protected double minY;
+
+    @Option(names = "--maxY", defaultValue = "90")
+    protected double maxY;
+
+    public double getMinX() {
+        return minX;
+    }
+
+    public void setMinX(double minX) {
+        this.minX = minX;
+    }
+
+    public double getMaxX() {
+        return maxX;
+    }
+
+    public void setMaxX(double maxX) {
+        this.maxX = maxX;
+    }
+
+    public double getMinY() {
+        return minY;
+    }
+
+    public void setMinY(double minY) {
+        this.minY = minY;
+    }
+
+    public double getMaxY() {
+        return maxY;
+    }
+
+    public void setMaxY(double maxY) {
+        this.maxY = maxY;
+    }
+}