parsesynonyms

#!/usr/bin/env sh

# Parameters:
# $1 = input filename
# $2 = server, defaults to localhost:9200
# $3 = output filename, defaults to [input filename]_generated.txt

if [ -z "$1" ]
then
  echo Please apply an input file
  exit 1
fi
if [ -z "$2" ]
then
  server="localhost:9200"
else
  server="$2"
fi
if [ -z "$3" ]
then
  filename=$(basename "$1")
  outputFile="${filename%.*}_generated.txt"
else
  outputFile="$3"
fi

# Create test index for analysis
curl -s -XPUT "$server/parse-synonyms-utility-index" -d "$(jq -n '{settings:{analysis:{analyzer:{synonyms:{type:"custom",tokenizer:"commasep",filter:["trim","asciifolding","lowercase","unique"]}},tokenizer:{commasep:{type:"pattern",pattern:","}}}}}')" > /dev/null

while read line; do
  # sample input: cimodule,ciplusmodule,,,
  # expected response example: {"tokens":[{"token":"cimodule","start_offset":0,"end_offset":8,"type":"word","position":1},{"token":"ciplusmodule","start_offset":9,"end_offset":21,"type":"word","position":2}]}
  response=$(curl -s -XGET "$server/parse-synonyms-utility-index/_analyze?analyzer=synonyms" -d "$line" | jq -r '[.tokens[].token] | join(",")')

  if [ -z "$response" ] || [ "$response" = "null" ]
  then
    echo Elasticsearch returned a malformed response
    exit 2
  fi

  if [ -n "$outputResponse" ]
  then
    #only add newline if outputresponse already exists
    outputResponse=$outputResponse$"\n"
  fi

  #concatenate string
  outputResponse=$outputResponse$response
done < $1

echo $outputResponse > $outputFile