new scripts for setup

Author: Ashkan Vedadi Gargary

paper_reproduced/scripts/README.md

@@ -12,6 +12,13 @@ Two sample datasets are included in the `dataset` folder. Large datasets (used i
 ## Setup:
 **RapidJSON**, **simdjson**, **cuJSON**, and **pison** are already available in the related_works directory. No need for further installation. For **cudf** and **GPJSON**, it requires to install their library and create their enviroment to be used which are also explained in the `related_works/gpjson` and `related_works/cudf`'s `readme.md` section. 
 
+We provide a script to setup gpjson and cudf too. You can use:
+```
+./setup.sh
+```
+
+It contains three scripts, `./setup/install_gpjson.sh` will download GraalVM (a dependency for gpjson) as well as clone and build gpjson. It's output also advises on what environment variables need to be set. `./setup/setup_cudf_env.sh` will setup cudf within a conda environment following steps from their documentation (https://github.com/rapidsai/cudf). `./setup/clean.sh` will just clean up the installation of gpjson, graalvm, and remove the cudf_env conda environment.
+
 ## 📊 Run all experiments
 We have added a new script called `run_experiments.sh`, which executes all experiments and generates the final figures and tables presented in the paper. For other methods, their libraries are already included in the `related_works` folder. If you need for further manually scritps checking and runnign per each figure you can follow the following subsections.
 

paper_reproduced/scripts/setup.sh

@@ -0,0 +1,9 @@
+#!/bin/bash
+
+# Run all figure scripts sequentially
+./setup/clean.sh
+./setup/setup_cudf_env.sh
+./setip/install_gpjson.sh
+
+
+echo "✅ Setup Successfully!"

paper_reproduced/scripts/setup/clean.sh

@@ -0,0 +1,4 @@
+#!/bin/bash -l
+
+rm -rf graalvm-ce-java8-21.0.0.2/ gpjson/
+conda env remove -n cudf_env -y

paper_reproduced/scripts/setup/install_gpjson.sh

@@ -0,0 +1,38 @@
+#!/bin/bash -l
+
+# Notes:
+# Must have cuda installed (tested with toolkit 12.8)
+# Must have GPU enabled machine (tested on H100)
+
+module purge
+module load cuda/12.8
+
+# Download GraalVM
+if [ ! -d "graalvm-ce-java8-21.0.0.2" ]; then
+ wget https://github.com/graalvm/graalvm-ce-builds/releases/download/vm-21.0.0.2/graalvm-ce-java8-linux-amd64-21.0.0.2.tar.gz
+ tar xvf graalvm-ce-java8-linux-amd64-21.0.0.2.tar.gz
+ rm graalvm-ce-java8-linux-amd64-21.0.0.2.tar.gz 
+fi
+
+export JAVA_HOME=$PWD/graalvm-ce-java8-21.0.0.2
+export GRAALVM_HOME=$PWD/graalvm-ce-java8-21.0.0.2
+export PATH=$PWD/graalvm-ce-java8-21.0.0.2/bin:$PATH
+
+which java
+java -version
+
+if [ ! -d "gpjson" ]; then
+ git clone https://github.com/koesie10/gpjson
+fi
+cd gpjson
+./gradlew copyToGraalVM -PgraalVMDirectory=$GRAALVM_HOME
+cd ..
+
+echo -e "\n\n\n"
+echo "gpjson was installed"
+echo "Due to GraalVM being a dependency, modify the following environment variables:"
+echo "export JAVA_HOME=$PWD/graalvm-ce-java8-21.0.0.2"
+echo "export GRAALVM_HOME=$PWD/graalvm-ce-java8-21.0.0.2"
+echo "export PATH=$PWD/graalvm-ce-java8-21.0.0.2/bin:\$PATH"
+echo -e "\n\n"
+echo "This was tested on a machine using CUDA 12.8 and an H100 GPU"

paper_reproduced/scripts/setup/setup_cudf_env.sh

@@ -0,0 +1,11 @@
+#!/bin/bash -l
+
+module purge
+module load miniconda3
+module load cuda/12.8
+
+conda create -n cudf_env -c rapidsai -c conda-forge cudf=25.08
+
+echo "Conda environment for cudf created"
+echo "Run 'conda activate cudf_env' to activate"
+