1
0
mirror of synced 2025-12-23 03:44:00 -05:00
Files
docs/script/search/build-records.js
James M. Greene 542a459c06 Move script-only dependencies to devDependencies (#19542)
* Move lib/search/sync.js to script/search/sync.js

* Move mdast-util-from-markdown to devDeps

* Move lib/redirects/add-redirect-to-frontmatter.js to script/helpers/

* Move mkdirp to devDeps

* Move linkinator to devDeps

* Move rimraf to devDeps

* Fix script/search/sync.js require paths

* Move lib/search/build-records.js to script/search/

* Move lib/search/find-indexable-pages to script/search/

* Fix require paths for build-records

* Fix require paths for find-indexable-pages

* Move lib/search/algolia-get-remote-index-names.js to script/search/

* Movbe lib/search/algolia-search-index.js to script/search/

* Move lib/search/lunr-search-index.js to script/search/

* Move lib/search/lunr-get-index-names.js to script/search/

* Fix Lunr search index paths

* Move lib/search/validate-records.js to script/search/

* Move is-url to devDeps

* Move lib/search/algolia-client.js to script/search/

* Move lib/search/parse-page-sections-into-records.js to script/search/

* Move lib/search/rank.js to script/search/

* Fix path to cached-index-names.json file

* Normalize require for fs.promises
2021-05-25 20:44:19 +00:00

56 lines
2.2 KiB
JavaScript

const domwaiter = require('domwaiter')
const eventToPromise = require('event-to-promise')
const chalk = require('chalk')
const parsePageSectionsIntoRecords = require('./parse-page-sections-into-records')
const languages = require('../../lib/languages')
const pageMarker = chalk.green('|')
const recordMarker = chalk.grey('.')
const port = 4002
module.exports = async function buildRecords (indexName, indexablePages, pageVersion, languageCode) {
console.log(`\n\nBuilding records for index '${indexName}' (${languages[languageCode].name})`)
const records = []
const pages = indexablePages
// exclude pages that are not in the current language
.filter(page => page.languageCode === languageCode)
// exclude pages that don't have a permalink for the current product version
.filter(page => page.permalinks.some(permalink => permalink.pageVersion === pageVersion))
// Find the approve permalink for the given language and GitHub product variant (dotcom v enterprise)
const permalinks = pages
.map(page => {
return page.permalinks.find(permalink => {
return permalink.languageCode === languageCode && permalink.pageVersion === pageVersion
})
})
.map(permalink => {
permalink.url = `http://localhost:${port}${permalink.href}`
return permalink
})
console.log('indexable pages', indexablePages.length)
console.log('pages in index', pages.length)
console.log('permalinks in index', permalinks.length)
console.log(pageMarker, 'denotes pages')
console.log(recordMarker, 'denotes records derived from sections of pages')
const waiter = domwaiter(permalinks, { maxConcurrent: 200, minTime: 5 })
.on('page', (page) => {
process.stdout.write(pageMarker)
const newRecords = parsePageSectionsIntoRecords(page.href, page.$)
if (!newRecords.length) {
console.log(chalk.red(`\nno records found: ${page.href}`))
}
process.stdout.write(recordMarker.repeat(newRecords.length))
records.push(...newRecords)
})
.on('error', (err) => {
console.error(err)
})
return eventToPromise(waiter, 'done').then(() => {
console.log('\nrecords in index: ', records.length)
return records
})
}