From 16ddd5380918744dfbf065b27e19c59d9e74bba6 Mon Sep 17 00:00:00 2001
From: Ali Khan <akhanf@users.noreply.github.com>
Date: Wed, 25 Sep 2024 09:18:26 -0400
Subject: [PATCH] Bigstitcher spark (#1)

* changes to get bigstitcher spark working
* specify mem and threads from cli wrapper
---
 Dockerfile           |   8 +++-
 spark_custom_install | 100 +++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 106 insertions(+), 2 deletions(-)
 create mode 100755 spark_custom_install

diff --git a/Dockerfile b/Dockerfile
index 55613b9..07dab5a 100644
--- a/Dockerfile
+++ b/Dockerfile
@@ -36,7 +36,7 @@ ENV JAVA_HOME=/usr/lib/jvm/java-1.8.0-amazon-corretto
 
 
 #install maven hdf5-tools deps, 7zip (for zarr zipstore archiving)
-RUN apt-get update && apt-get install -y maven hdf5-tools libblosc-dev p7zip-full && mkdir -p /opt/bin
+RUN apt-get update && apt-get install -y maven hdf5-tools libblosc-dev p7zip-full libxtst-dev && mkdir -p /opt/bin
 
 ENV PATH $PATH:/opt/bin
 
@@ -44,7 +44,11 @@ ENV PATH $PATH:/opt/bin
 RUN cd /opt && git clone https://github.com/saalfeldlab/n5-utils && cd n5-utils && ./install /opt/bin
 
 #install bigstitcher-spark, and customize launcher to include args for mem and cpu 
-RUN cd /opt && git clone https://github.com/akhanf/BigStitcher-Spark.git && cd BigStitcher-Spark && ./install -t 32 -m 128 && cp -v affine-fusion /opt/bin && cp -v target/BigStitcher-Spark-0.0.2-SNAPSHOT.jar /opt/bin 
+RUN cd /opt && git clone https://github.com/akhanf/BigStitcher-Spark.git
+COPY spark_custom_install /opt/BigStitcher-Spark/install
+RUN cd /opt/BigStitcher-Spark && ./install -t 32 -m 128
+
+ENV PATH $PATH:/opt/BigStitcher-Spark
 
 # Install Fiji.
 RUN mkdir /opt/fiji \
diff --git a/spark_custom_install b/spark_custom_install
new file mode 100755
index 0000000..24ec763
--- /dev/null
+++ b/spark_custom_install
@@ -0,0 +1,100 @@
+#!/bin/bash
+
+# This script is shamelessly extended from https://github.com/saalfeldlab/n5-utils, thanks @axtimwalde & co!
+
+USERTHREADS="-1"
+USERMEM="-1"
+
+while [[ $# -gt 0 ]]; do
+  case $1 in
+    -t|--threads)
+      USERTHREADS="$2"
+      shift # past argument
+      shift # past value
+      ;;
+    -m|--mem)
+      USERMEM="$2"
+      shift # past argument
+      shift # past value
+      ;;
+    -*|--*)
+      echo "Unknown option $1"
+      exit 1
+      ;;
+  esac
+done
+
+if [ $USERTHREADS == "-1" ]; then
+  echo "You did not define the number of threads for Java/Spark can use, will be set automatically. You could do it by e.g.: './install -t 8' for 8 threads."
+fi
+
+
+VERSION="0.0.2-SNAPSHOT"
+INSTALL_DIR=$(pwd)
+#INSTALL_DIR=${1:-$(pwd)}
+
+echo ""
+echo "Installing into $INSTALL_DIR (for local execution)"
+
+echo 'Building the code'
+
+sleep 2
+
+#mvn clean install
+#echo 'Building a fatjar, which can also be used for cluster/cloud execution' 
+mvn clean install -P fatjar -Dmaven.repo.local=/opt/BigStitcher-Spark/repo
+mvn -Dmdep.outputFile=cp.txt -Dmdep.includeScope=runtime dependency:build-classpath -Dmaven.repo.local=/opt/BigStitcher-Spark/repo
+
+
+# function that installs one command
+# $1 - command name
+# $2 - java class containing the functionality
+install_command () {
+    echo "Installing '$1' command into" $INSTALL_DIR
+
+    echo '#!/bin/bash' > $1
+    echo '' >> $1
+    echo 'if [ "$#" -lt 2 ]; then echo "Usage: $0 <mem_gb> <threads> <other parameters (enter in mem_gb and threads to see usage)>"; exit 1; fi' >> $1
+    echo 'MEM=$1' >> $1
+    echo 'THREADS=$2' >> $1
+    echo 'shift 2' >> $1
+    echo "JAR=/opt/BigStitcher-Spark/repo/net/preibisch/BigStitcher-Spark/${VERSION}/BigStitcher-Spark-${VERSION}.jar" >> $1
+    echo 'java \' >> $1
+    echo '  -Xmx${MEM}g -Dspark.master=local[${THREADS}] \' >> $1
+    echo -n '  -cp $JAR:' >> $1
+    echo -n $(cat cp.txt) >> $1
+    echo ' \' >> $1
+    echo '  '$2' "$@"' >> $1
+
+    chmod a+x $1
+}
+
+echo 'Installing workflow tools ...'
+
+install_command resave "net.preibisch.bigstitcher.spark.SparkResaveN5"
+install_command detect-interestpoints "net.preibisch.bigstitcher.spark.SparkInterestPointDetection"
+install_command match-interestpoints "net.preibisch.bigstitcher.spark.SparkGeometricDescriptorMatching"
+install_command stitching "net.preibisch.bigstitcher.spark.SparkPairwiseStitching"
+install_command solver "net.preibisch.bigstitcher.spark.Solver"
+install_command affine-fusion "net.preibisch.bigstitcher.spark.SparkAffineFusion"
+install_command nonrigid-fusion "net.preibisch.bigstitcher.spark.SparkNonRigidFusion"
+
+echo 'Installing utils ...'
+
+install_command downsample "net.preibisch.bigstitcher.spark.SparkDownsample"
+install_command clear-interestpoints "net.preibisch.bigstitcher.spark.ClearInterestPoints"
+install_command clear-registrations "net.preibisch.bigstitcher.spark.ClearRegistrations"
+install_command transform-points "net.preibisch.bigstitcher.spark.TransformPoints"
+
+
+if [ $(pwd) == "$INSTALL_DIR" ]; then
+    echo "Installation directory equals current directory, we are done."
+else
+	echo "Creating directory $INSTALL_DIR and moving files..."
+    mkdir -p $INSTALL_DIR
+    mv affine-fusion $INSTALL_DIR/
+fi
+
+rm cp.txt
+
+echo "Installation finished."