bash script to remove unwanted files in xnat using the XnatDataClient

spiderdel.sh

#!/bin/bash

# bash script to remove unwanted files in xnat using the XnatDataClient
# iterates through all sessions looking for particular scan types with particular file types and file name patterns

# CLI parameters:
# 1. url - xnat url
# 2. username - xnat username
# 3. project - xnat project
# 4. scan type - e.g. "STRUCTURAL", "FLAIR"
# 5. file type - e.g. "NIFTI", "SNAPSHOT"
# 6. filesuffix - the suffix of the files to be remove, e.g. "_N.nii.gz"

# set this to 0,1,2 depending on required level of feedback
verbose=1
# set this to 0 to actually delete the files
dryrun=0

function log {
  if [[ $verbose -gt 0 ]]; then
    echo "spiderdel: $1"
  fi
}

function debug {
  if [[ $verbose -eq 2 ]]; then
    echo "spiderdel (debug): $1"
  fi
}

# check for correct number of arguments
if [ "$#" -ne 6 ]; then
  echo -e "Incorrect number of arguments - expected 6"
  echo "Usage: spiderdel.sh xnat-url xnat-username xnat-project scan_type file_type filesuffix"
else
  echo -n "XNAT password: "
  read -s password
  echo
  XNAT_URL=$1
  XNAT_USER=$2
  XNAT_PROJECT=$3
  SCAN_TYPE=$4
  FILE_TYPE=$5
  FILE_SUFFIX=$6
  session=$(./XnatDataClient -c -u $XNAT_USER -p $password -r $XNAT_URL/data/JSESSION)
  if [[ $(echo "$session" | tail -n 1) == HTTP\ status\:\ 200 ]]; then
    session_id=$(echo "$session" | head -c 32)
    debug "tomcat session: $session_id"
    # head -n -1 strips bottom line, i.e. "Z bytes"
    sessions=$(./XnatDataClient -s $session_id -r $XNAT_URL/data/archive/projects/$XNAT_PROJECT/experiments\?format=csv | head -n -1 )
    # sessions is csv file with a header:
    # xnat:subjectassessordata/id,ID,project,date,xsiType,label,insert_date,URI
    session_count=$(($(echo "$sessions" | wc -l)-1))
    debug "session_count=$session_count from $sessions"
    if [[ $session_count -gt 0 ]]; then
      echo "$sessions" | while IFS="," read -r col1 col2 col3; do
        if [ "$col2" != "ID" ]; then
          # scans is csv file with a header:
          # xnat_imagescandata_id,ID,type,quality,xsiType,note,series_description,URI
          scans=$(./XnatDataClient -s $session_id -r $XNAT_URL/data/archive/experiments/$col2/scans?format=csv | head -n -1 | tail -n +2 | grep $SCAN_TYPE)
          debug "scans: $scans"
          echo "$scans" | while IFS="," read -r sc1 sc2 sc3; do
            resources=$(./XnatDataClient -s $session_id -r $XNAT_URL/data/archive/experiments/$col2/scans/$sc2/resources?format=csv | head -n -1 | tail -n +2 | grep $FILE_TYPE)
            debug "resources: $resources"
            echo "$resources" | while IFS="," read -r rs1 rs2; do
              files=$(./XnatDataClient -s $session_id -r $XNAT_URL/data/archive/experiments/$col2/scans/$sc2/resources/$rs1/files?format=csv | head -n -1 | tail -n +2)
              debug "files: $files"
              echo "$files" | while IFS="," read -r f1 f2; do
                if [[ "$f1" == *$FILE_SUFFIX ]]; then
                  if [ $dryrun == 1 ]; then
                    echo "TO DELETE:"
                    echo "scan_id: $sc1"
                    echo "resource_id: $rs1"
                    echo "file: $f1"
                  else
                    #TODO: implement delete call.  Can I use XnatDataClient?
                    echo "deleting: $f1"
                    ./XnatDataClient -s $session_id -m DELETE -r $XNAT_URL/data/archive/experiments/$col2/scans/$sc2/resources/$rs1/files/$f1
                    echo "deleted"
                  fi
                fi
              done
            done
          done
        fi
      done
    fi
  else
    log "Failed: authentication failed"
    log "xnat: $url"
    log "user: $1"
  fi
fi