course-renderer/src/tasks/summary-parser.ts

Manages CA School Courses file system storage and HTML conversion

import * as cheerio from 'cheerio'
import * as MarkdownIt from 'markdown-it'
import * as escapeHTML from 'escape-html'

const through = require('through2')
const replaceExtension = require('replace-ext')

module.exports = (options: any) => (through.obj((file: any, _: any, callback: any) => {
    if (file.isNull()) {
        callback(null, file);
        return;
    }

    if (file.isStream()) {
        callback(new Error('Streaming not supported'));
        return;
    }

    const md = new MarkdownIt()

    const ch = cheerio.load(md.render(file.contents.toString()))

    let output: Array<{}> = new Array()

    traverseSummary(output, ch, ch('ul').first())

    file.contents = new Buffer(JSON.stringify(output, null, 4))
    file.path = replaceExtension(file.path, '.json')
    callback(null, file)
}))

function traverseSummary(output: Array<{}>, ch: CheerioStatic, curEle: Cheerio): Array<{}> {
    curEle.children('li').each( function( index: number, element: CheerioElement ) {
            const childLI       = ch(element)
            const link          = childLI.find('a')
            const label         = link.text()
            const url           = link.attr('href').replace(/\.md$/, '.html')
            const desc          = escapeHTML(childLI.children('blockquote').text().replace(/^\s+|\s+$/g, ''))
            const duration      = childLI.find('strong').first().text()
            
            const summaryToken: any = {
                content     : `[${label}](${url})`,
                text        : label,
                url         : url,
                description : desc,
                duration    : duration,
                children    : []
            };

            if ( childLI.children('ul').length > 0 ) {
                summaryToken.children = traverseSummary( [], ch, childLI.children('ul') )
            }

            output.push( summaryToken )
    })

    return output
}