a utility script for CCA to check for duplicate syllabi in VAULT

dedupe-syllabi.js

#!/usr/bin/env node
// a utility script for CCA to check for duplicate syllabi in VAULT
let defaults = {
    uuid: '9ec74523-e018-4e01-ab4e-be4dd06cdd68',
    // 50 is max length
    length: 50,
    term: 'Spring 2018',
}
// .equellarc file with credentials for API use
let options = require('rc')('equella', defaults)
let request = require('request')

// construction API URL
let url = `https://vault.cca.edu/api/search/?q=&collections=${options.uuid}&where=%2Fxml%2Fmods%2FtitleInfo%2Ftitle%20LIKE%20%27${encodeURIComponent(options.term)}%25%27&length=${options.length}`

// each syllabus is an object like { name, uuid, ... }
let syllabi = []
let dupes = []
let counts = {
    dupes: 0,
    total: Infinity,
    retrieved: 0
}

request.get(
    url,
    {
        'headers': { 'X-Authorization': 'access_token=' + options.token },
        'json': true
    },
    (err, response, data) => {
        counts.total = data.available
        counts.retrieved = options.length
        syllabi.concat(data.results)

        console.log(`${counts.total} total syllabi`)

        while (counts.retrieved < counts.total) {
            console.log(`Retrieved ${counts.retrieved} of ${counts.total} syllabi`)
            counts.retrieved += options.length

            request.get(
                url + `&start=${counts.retrieved}`,
                {
                    'headers': { 'X-Authorization': 'access_token=' + options.token },
                    'json': true
                },
                (err, response, data) => {
                    syllabi.push(...data.results)
                }
            )
        }
    }
)

// recursive check for dupes fn
function check_for_dupes(arr) {
    let list = arr.slice().sort((a, b) => {
        if ([a.name, b.name].sort()[0] === a.name) {
            return -1
        } else if ([a.name, b.name].sort()[0] === b.name) {
            return 1
        }
    })
    list.forEach((item, index) => {
        let next = list[index + 1]
        if (next && item.name === next.name) {
            dupes.push([item, next])
            counts.dupes++
        }
    })
}

function main() {
    check_for_dupes(syllabi)

    // print results
    console.log(`${counts.dupes} dupes out of ${counts.total} syllabi for ${options.term}`)
    if (counts.dupes) {
        console.log('List of possible dupes:\n')
        for (let dupe of dupes) {
            console.log(`${dupe[0].name}\thttps://vault.cca.edu/items/${dupe[0].uuid}/0/`)
            console.log(`${dupe[1].name}\thttps://vault.cca.edu/items/${dupe[1].uuid}/0/\n`)
        }
    }
}

// lol async coding is hard
setTimeout(main, 4000)