Diff versions of a code list

code_list_diff.sh

#!/bin/bash
#
# Compute the diff between SKOS code lists.
# Usage: ./code_list_diff.sh [-h|--help] old_version new_version

set -e
shopt -s extglob

V1=$1
V2=$2

TEST_QUERY="ASK { [] a <http://www.w3.org/2004/02/skos/core#ConceptScheme> . }"

die () {
  echo >&2 "$@"
  exit 1
}

select_query () {
  arq --results CSV --namedGraph $V1 --namedGraph $V2 "${1}" | column -s, -t
}

select_one () {
  select_query "$1" | tail -1
}

ask_query () {
  select_one "$1" | grep -q true && echo "Yes" || echo "No"
}

usage () {
  echo "Usage: ./code_list_diff.sh [-h|--help] old_version new_version"
  echo "Computes differences between old_version and new_version of a SKOS code list."
  echo "Versions of code lists must be provided in an RDF serialization."
  exit 0
}

heading () {
  printf "\n### %s ###\n" "$1" 
}

[[ "-h --help" =~ $1 ]] && usage

# Test if arq is installed.
command -v arq >/dev/null 2>&1 || die "Missing Jena ARQ!"

# Test correct number of arguments.
if [ "$#" -ne 2 ]; then
  die "You must provide two code lists!"
fi

! [ $V1 == $V2 ] || die "Cannot diff the same file."

for i in "$@"
do
  # Test if the diffed files exist.
  if [ ! -f $i ]; then
    die "File $i not found!"
  fi
  # Test if the diffed files contain skos:ConceptSchemes.
  arq --results CSV --data $i "$TEST_QUERY" | tail -1 | grep -q true ||
    die "File $i is missing a skos:ConceptScheme!"
done

heading "Number of concepts in $(basename "$V1")"
COUNT_Q1=$(cat <<-END
PREFIX skos: <http://www.w3.org/2004/02/skos/core#>

SELECT (COUNT(DISTINCT ?concept) AS ?count)
WHERE {
  GRAPH <$V1> {
    ?concept a skos:Concept .
  }
}
END)

select_one "$COUNT_Q1"

heading "Number of concepts in $(basename "$V2")"
COUNT_Q2=$(cat <<-END
PREFIX skos: <http://www.w3.org/2004/02/skos/core#>

SELECT (COUNT(DISTINCT ?concept) AS ?count)
WHERE {
  GRAPH <$V2> {
    ?concept a skos:Concept .
  }
}
END)

select_one "$COUNT_Q2"

heading "Are there concepts sharing labels in $(basename "$V1")?"
DUPLICATE_LABELS_Q1=$(cat <<-END
PREFIX skos: <http://www.w3.org/2004/02/skos/core#>

ASK
WHERE {
  GRAPH <$V1> {
    ?concept1 a skos:Concept ;
      skos:prefLabel ?prefLabel .
    ?concept2 a skos:Concept ;
      skos:prefLabel ?prefLabel .
    FILTER (!sameTerm(?concept1, ?concept2))
  }
}
END)

ask_query "$DUPLICATE_LABELS_Q1"

heading "Are there concepts sharing notations in $(basename "$V1")?"
DUPLICATE_NOTATIONS_Q1=$(cat <<-END
PREFIX skos: <http://www.w3.org/2004/02/skos/core#>

ASK
WHERE {
  GRAPH <$V1> {
    ?concept1 a skos:Concept ;
      skos:notation ?notation .
    ?concept2 a skos:Concept ;
      skos:notation ?notation .
    FILTER (!sameTerm(?concept1, ?concept2))
  }
}
END)

ask_query "$DUPLICATE_NOTATIONS_Q1"

heading "Count the same concepts" 
Q1=$(cat <<-END
PREFIX skos: <http://www.w3.org/2004/02/skos/core#>

SELECT (COUNT(DISTINCT ?concept) AS ?count)
WHERE {
  GRAPH <$V1> {
    ?concept a skos:Concept .
  }
  GRAPH <$V2> {
    ?concept a skos:Concept .
  }
}
END)

select_one "$Q1"

heading "Count concepts sharing the same label and notation"
Q2=$(cat <<-END
PREFIX skos: <http://www.w3.org/2004/02/skos/core#>

SELECT (COUNT(*) AS ?count) 
WHERE {
  {
    SELECT DISTINCT ?prefLabel ?notation
    WHERE {
      GRAPH <$V1> {
        [] a skos:Concept ;
          skos:prefLabel ?prefLabel ;
          skos:notation ?notation .
      }
      GRAPH <$V2> {
        [] a skos:Concept ;
          skos:prefLabel ?prefLabel ;
          skos:notation ?notation .
      }
    }
  }
}
END)

select_one "$Q2"

heading "Concepts sharing the same notation with labels differing in character case"
Q3=$(cat <<-END
PREFIX skos: <http://www.w3.org/2004/02/skos/core#>

SELECT DISTINCT ?notation ?prefLabel1 ?prefLabel2
WHERE {
  GRAPH <$V1> {
    [] a skos:Concept ;
      skos:prefLabel ?prefLabel1 ;
      skos:notation ?notation .
  }
  GRAPH <$V2> {
    [] a skos:Concept ;
      skos:prefLabel ?prefLabel2 ;
      skos:notation ?notation .
  }
  FILTER (!sameTerm(?prefLabel1, ?prefLabel2) && (LCASE(?prefLabel1) = LCASE(?prefLabel2)))
  FILTER NOT EXISTS {
    GRAPH <$V1> {
      [] a skos:Concept ;
        skos:prefLabel ?prefLabel2 ;
        skos:notation ?notation .
    }
  }
}
ORDER BY ?notation
END)

select_query "$Q3"

heading "Concepts sharing the same label that changed notations"
Q4=$(cat <<-END
PREFIX skos: <http://www.w3.org/2004/02/skos/core#>

SELECT DISTINCT ?prefLabel ?notation1 ?notation2
WHERE {
  GRAPH <$V1> {
    [] a skos:Concept ;
      skos:prefLabel ?prefLabel ;
      skos:notation ?notation1 .
  }
  GRAPH <$V2> {
    [] a skos:Concept ;
      skos:prefLabel ?prefLabel ;
      skos:notation ?notation2 .
  }
  FILTER (!sameTerm(?notation1, ?notation2))
  FILTER NOT EXISTS {
    GRAPH <$V1> {
      [] a skos:Concept ;
        skos:prefLabel ?prefLabel ;
        skos:notation ?notation2 .
    }
  }
}
ORDER BY ?prefLabel
END)

select_query "$Q4"

heading "Concepts with prolonged labels"
Q5=$(cat <<-END
PREFIX skos: <http://www.w3.org/2004/02/skos/core#>

SELECT DISTINCT ?notation ?prefLabel1 ?prefLabel2
WHERE {
  GRAPH <$V1> {
    [] a skos:Concept ;
      skos:prefLabel ?prefLabel1 ;
      skos:notation ?notation .
  }
  GRAPH <$V2> {
    [] a skos:Concept ;
      skos:prefLabel ?prefLabel2 ;
      skos:notation ?notation .
  }
  FILTER (!sameTerm(?prefLabel1, ?prefLabel2) && (STRSTARTS(?prefLabel2, ?prefLabel1)))
  FILTER NOT EXISTS {
    GRAPH <$V1> {
      [] a skos:Concept ;
        skos:notation ?notation ;
        skos:prefLabel ?prefLabel2 .
    }
  }
}
END)

select_query "$Q5"

heading "Concepts with shortened labels"
Q6=$(cat <<-END
PREFIX skos: <http://www.w3.org/2004/02/skos/core#>

SELECT DISTINCT ?notation ?prefLabel1 ?prefLabel2
WHERE {
  GRAPH <$V1> {
    [] a skos:Concept ;
      skos:prefLabel ?prefLabel1 ;
      skos:notation ?notation .
  }
  GRAPH <$V2> {
    [] a skos:Concept ;
      skos:prefLabel ?prefLabel2 ;
      skos:notation ?notation .
  }
  FILTER (!sameTerm(?prefLabel1, ?prefLabel2) && (STRSTARTS(?prefLabel1, ?prefLabel2)))
  FILTER NOT EXISTS {
    GRAPH <$V1> {
      [] a skos:Concept ;
        skos:notation ?notation ;
        skos:prefLabel ?prefLabel2 .
    }
  }
}
END)

select_query "$Q6"