Added a way to download SkyServer dataset and queries

README.md

@@ -76,3 +76,19 @@ Example runs:
     ./run.sh 100000000.data crack 100000 Random 1e-2 NOUP 30
     ./run.sh skyserver.data dd1r 200000 SkyServer 1e-7 NOUP 60
     ./run.sh skyserver.data dd1r 200000 SkyServer 1e-7 HFLV 60
+
+Download SkyServer dataset and queries
+======
+
+The SkyServer dataset consists of a sequence of 585634221 integers which represents the degree of ascension
+in the Photoobjall table. Originally the degree is a floating point between 0 to 360, but in this dataset,
+it has been multiplied by 1 million and converted to integers. Run the following command to get the dataset
+(it may take very long to download).
+
+    ./run.sh get-skyserver-dataset
+
+Similarly, the SkyServer queries consist of a sequence of 158325 point queries on the ascension column
+(similarly formatted by multiplying it by 1 million and converted to integers).
+
+    ./run.sh get-skyserver-queries
+

run.sh

@@ -79,6 +79,14 @@ if [ "$1" == "batch" ]; then
 	./run.sh 100000000.data aics 1000 Random 1e-2 NOUP $T
 	./run.sh 100000000.data aics1r 1000 Random 1e-2 NOUP $T
 	./run.sh 100000000.data aiss 1000 Random 1e-2 NOUP $T
+elif [ "$1" == "get-skyserver-dataset" ]; then
+	make -s data/skyserver.data
+	g++ src/data.cpp -o bin/data
+	bin/data data/skyserver.data
+elif [ "$1" == "get-skyserver-queries" ]; then
+	make -s data/skyserver.data
+	g++ src/query.cpp -o bin/query
+	bin/query < data/skyserver.queries
 else
 	run $1 $2 $3 $4 $5 $6 $7
 fi

src/data.cpp

@@ -0,0 +1,15 @@
+#include <stdio.h>
+
+#define MAXN (1 << 25)
+
+int arr[MAXN];
+
+int main(int argc, char *argv[]){
+	FILE *in = fopen(argv[1],"rb");
+	if (!in){ fprintf(stderr,"Error opening file %s\n",argv[1]); return 1; }
+	while (true) {
+		int N = fread(arr, sizeof(int), MAXN, in);
+		for (int i = 0; i < N; i++) printf("%d\n", arr[i]);
+		if (N < MAXN) break;
+	}
+}

src/query.cpp

@@ -0,0 +1,10 @@
+#include <stdio.h>
+
+int nth;
+double ascension;
+
+int main(int argc, char *argv[]){
+	while (scanf("%d %lf", &nth, &ascension) != EOF) {
+		printf("%.0lf\n", ascension * 1e6);
+	}
+}