Check zotero library against crossref.org

checkzoterodb.sh

#!/bin/bash

##
## Requires: sqlite3, jq, wdiff, and curl
##
## It is strongly recommended to use the -m flag with your email address,
## as this will put your crossref reuqests in a priority queue, and help
## the crossref folks keep their service running smoothly.

printhelp() { echo 'Usage: checkzoterodb.sh [-m|--mailto <email>] <database>' ; }

mailto=""
DB="$HOME/Zotero/zotero.sqlite"

while [ "$1" ]; do
    case "$1" in
        '-h'|'--help')
            printhelp
            exit 0
            ;;
        '-m'|'--mailto')
            mailto=$2
            shift
            ;;
        *)
            if [ -f $1 ]; then
                DB="$1"
            else
                printhelp
                exit 1
            fi
            ;;
    esac
    shift
done

mkdir -p /tmp/crossref
doiID=$(sqlite3 "${DB}" 'SELECT fieldID FROM fields WHERE fieldName="DOI";')
dateID=$(sqlite3 "${DB}" 'SELECT fieldID FROM fields WHERE fieldName="date";')
titleID=$(sqlite3 "${DB}" 'SELECT fieldID FROM fields WHERE fieldName="title";')

for itemID in $(sqlite3 "${DB}" 'SELECT itemID FROM items;'); do
    [ "$itemID" == "" ] && continue
    idateID=$(sqlite3 "${DB}"  'SELECT valueID FROM itemData WHERE fieldID="'${dateID}'" and itemID="'${itemID}'"')
    idoiID=$(sqlite3 "${DB}"   'SELECT valueID FROM itemData WHERE fieldID="'${doiID}'" and itemID="'${itemID}'"')
    [ "${idoiID}" == "" ] && continue
    ititleID=$(sqlite3 "${DB}" 'SELECT valueID FROM itemData WHERE fieldID="'${titleID}'" and itemID="'${itemID}'"')
    [ "${ititleID}" == "" ] && continue

    title=$(sqlite3 "${DB}"   'SELECT value FROM itemDataValues WHERE valueID="'${ititleID}'"')
    doi=$(sqlite3 "${DB}"     'SELECT value FROM itemDataValues WHERE valueID="'${idoiID}'"')
    [ "${doi}" == "" ] && echo " - No DOI value on item: $title" && continue
    year=$(sqlite3 "${DB}"    'SELECT value FROM itemDataValues WHERE valueID="'${idateID}'"' | cut -d '-' -f 1)

    ## Get crossref data (first check for cached version)
    refdata=/tmp/crossref/$(echo "$doi" | md5sum | cut -d ' ' -f 1)
    if [ ! -f $refdata ]; then
        if [ "$mailto" ]; then
            curl -s https://api.crossref.org/works/${doi}?mailto=${mailto} > "${refdata}" 2>/dev/null
        else
            curl -s https://api.crossref.org/works/${doi} > "${refdata}" 2>/dev/null
        fi
    fi

    ## Compare year
    crossrefyear="null"
    [ "$crossrefyear" == "null" ] && crossrefyear=$(jq '.message.issued."date-parts"[0][0]' "${refdata}")
    [ "$crossrefyear" == "null" ] && crossrefyear=$(jq '.message."journal-issue"."published-print"."date-parts"[0][0]' "${refdata}")
    [ "$crossrefyear" == "null" ] && crossrefyear=$(jq '.message."published-print"."date-parts"[0][0]' "${refdata}")
    if [ ! "$year" == "$crossrefyear" ]; then
        echo "$crossrefyear != $year :: $title :: $doi"
    fi

    ## Compare title
    crossreftitle=$(jq '.message.title[0]' "${refdata}")
    diff -i -q <(echo "$crossreftitle") <(echo "\"$title\"") >/dev/null
    if [ ! $? -eq 0 ]; then
        echo -n "Title does not match (DOI: $doi): "
        wdiff -i -w "$(tput bold;tput setaf 1)" -x "$(tput sgr0)" \
            -y "$(tput bold;tput setaf 2)" -z "$(tput sgr0)"  \
            <(echo "$crossreftitle") <(echo "\"$title\"")
    fi
done