Verification of LA data mappings in solr

check-dr-collection-in-solr.sh

#!/bin/bash
# Put this, for instance in /usr/loca/bin/ and adapt to your LA node
#
# Used in conjuntion with this other script:
# https://gist.github.com/vjrj/2b077a2bb786227a517ca03f7e30cf4c
# 
# Depends on jq (apt install jq) 
#

USER=XXX
PASS=XXX
INDEX=http://1.2.3.4:8983
ERRORS=0

drs=$1

for dr in $(echo $drs | tr "," "\n")
do
  NUM_REC=$(curl -s -u $USER:$PASS $INDEX/solr/biocache/select\?fl\=collection_uid\&group.field\=collection_uid\&group\=true\&indent\=on\&q\=data_resource_uid:$dr\&wt\=json | jq '.grouped.collection_uid.matches')
  COL_UID=$(curl -s -u $USER:$PASS $INDEX/solr/biocache/select\?fl\=collection_uid\&group.field\=collection_uid\&group\=true\&indent\=on\&q\=data_resource_uid:$dr\&wt\=json | jq '.grouped.collection_uid.groups[0].groupValue' | sed 's/"//g')
  GROUP_NUM=$(curl -s -u $USER:$PASS $INDEX/solr/biocache/select\?fl\=collection_uid\&group.field\=collection_uid\&group\=true\&indent\=on\&q\=data_resource_uid:$dr\&wt\=json | jq '.grouped.collection_uid.groups[0].doclist.numFound')
  NUM_NULL=$(curl -g -s -u $USER:$PASS "$INDEX/solr/biocache/select?indent=on&q=data_resource_uid:$dr%20AND%20-collection_uid:[\%22\%22%20TO%20*]&wt=json" | jq '.response.numFound')

  if [[ $COL_UID == "null" ]]
  then
    echo "ERROR: Incorrect collection_uid '$COL_UID' for $dr in SOLR"
    ERRORS=$[ERRORS + 1]
  else 
    if [[ $NUM_NULL != 0 ]] 
    then
      echo "WARN: some collection_uid are null" 
    fi
    if [[ $NUM_REC != $GROUP_NUM ]]
    then
      echo "WARN: collection_uid $COL_UID for $dr in SOLR has $GROUP_NUM records but the are other collection_uids in this dr. Total: $NUM_REC"
      /usr/local/bin/check-mappings-in-ocurrences.sh $dr
    else
      echo "INFO: Correct collection_uid $COL_UID for $dr in SOLR, $NUM_REC records"
    fi
  fi
done

if [[ $ERRORS -gt 0 ]]
then
  exit 1
else 
  exit 0
fi