From 41b441a8114956a62553f80c3cee2fd75652d9f5 Mon Sep 17 00:00:00 2001
From: Jey Puget Gil <LinkJerome@gmail.com>
Date: Wed, 24 Jul 2024 13:52:16 +0200
Subject: [PATCH] Add support for generating multiple versions of BSBM dataset

---
 README.md     |  2 ++
 entrypoint.sh | 19 +++++++++++++++++--
 generate-n    | 26 ++++++++++++++++++++++++++
 3 files changed, 45 insertions(+), 2 deletions(-)
 create mode 100755 generate-n

diff --git a/README.md b/README.md
index 0dad878..9ba9af6 100644
--- a/README.md
+++ b/README.md
@@ -10,6 +10,7 @@ This is the Dockerized version of the Berlin SPARQL Benchmark.
 ## Usage
 ```bash
 docker run -v "$PWD:/data" vcity/bsbm generate [args]
+docker run -v "$PWD:/data" vcity/bsbm generate-n [args]
 docker run -v "$PWD:/data" vcity/bsbm qualification [args]
 docker run -v "$PWD:/data" vcity/bsbm testdriver [args]
 ```
@@ -18,6 +19,7 @@ If you want more information about the different arguments, please refer to the
 
 ```bash
 docker run vcity/bsbm generate -help
+docker run vcity/bsbm generate-n -help
 docker run vcity/bsbm qualification -help
 docker run vcity/bsbm testdriver -help
 ```
diff --git a/entrypoint.sh b/entrypoint.sh
index 3ee5098..300970c 100755
--- a/entrypoint.sh
+++ b/entrypoint.sh
@@ -1,8 +1,9 @@
 #!/bin/bash
 
 if [ $# == 0 ]; then
-    echo "Usage: $0 [generate|testdriver|qualification] ?[args]"
+    echo "Usage: $0 [generate|generate-n|testdriver|qualification] ?[args]"
     ./generate -help
+    ./generate-n -help
     ./testdriver -help
     ./qualification -help
     exit 1
@@ -14,7 +15,21 @@ command=$1
 mkdir -p /data
 chmod 777 /data
 
-if [ "$command" = "generate" ]; then
+if [ "$command" = "generate" ] || [ "$command" = "generate-n" ]; then
+    ./$@
+
+    # get the -fn parameter value, if it exists else set it to "dataset"
+    fn=$(echo $@ | grep -oP '(?<=-fn )[^ ]+' || echo "dataset")
+    # get the -s parameter value, if it exists else set it to "nt"
+    s=$(echo $@ | grep -oP '(?<=-s )[^ ]+' || echo "nt")
+    mv $fn*.$s /data
+
+    # check if the -ud parameter exists
+    if [[ $@ == *"-ud"* ]]; then
+        udf=$(echo $@ | grep -oP '(?<=-udf )[^ ]+' || echo "dataset_update")
+        mv $udf*.nt /data
+    fi
+elif [ "$command" = "generate" ]; then
     ./$@
 
     # get the -fn parameter value, if it exists else set it to "dataset"
diff --git a/generate-n b/generate-n
new file mode 100755
index 0000000..df370cd
--- /dev/null
+++ b/generate-n
@@ -0,0 +1,26 @@
+#!/bin/bash
+if [ ! -e "./generate-n" ]
+then
+  echo "Please run this script from the exec directory"
+  exit 1
+fi
+if [ $# == 0 ]
+then
+  echo "For generator options type ./generate-n -help"
+fi
+# check that $1 is a integer
+if ! [[ $1 =~ ^[0-9]+$ ]]
+then
+  echo "Please provide an integer as the first argument"
+  exit 1
+fi
+
+number_of_versions=$1
+echo "Generating $number_of_versions versions"
+
+# loop through the number of versions and call ./generate
+for i in $(seq 1 $number_of_versions)
+do
+  echo "Generating version $i"
+  ./generate "${@:2}"
+done