Move more parsing of strings and html server-side

modules/node_modules/@frogpond/ccc-google-calendar/index.js

@@ -1,21 +1,19 @@
-import {get} from '@frogpond/ccc-lib'
+import {get, parseHtml} from '@frogpond/ccc-lib'
 import moment from 'moment'
 import getUrls from 'get-urls'
-import _jsdom from 'jsdom'
-const {JSDOM} = _jsdom
 
 function convertGoogleEvents(data, now = moment()) {
 	let events = data.map((event) => {
+		const title = parseHtml(event.summary || '')
 		const startTime = moment(event.start.date || event.start.dateTime)
 		const endTime = moment(event.end.date || event.end.dateTime)
-		let description = (event.description || '').replace('<br>', '\n')
-		description = JSDOM.fragment(description).textContent.trim()
+		let description = cleanTextBlock(event.description || '')
 
 		return {
 			dataSource: 'google',
 			startTime,
 			endTime,
-			title: event.summary || '',
+			title,
 			description: description,
 			location: event.location || '',
 			isOngoing: startTime.isBefore(now, 'day'),
@@ -31,6 +29,10 @@ function convertGoogleEvents(data, now = moment()) {
 	return events
 }
 
+function cleanTextBlock(text) {
+	return parseHtml(text).replace(/\t/g, ' ').replace('<br>', '\n').trim()
+}
+
 export async function googleCalendar(calendarId, now = moment()) {
 	let calendarUrl = `https://www.googleapis.com/calendar/v3/calendars/${calendarId}/events`
 

modules/node_modules/@frogpond/ccc-lib/html.js

@@ -0,0 +1,14 @@
+import {toLaxTitleCase} from '@frogpond/titlecase'
+
+import _jsdom from 'jsdom'
+const {JSDOM} = _jsdom
+
+// Html
+
+export function parseHtml(string) {
+	return JSDOM.fragment(string).textContent.trim()
+}
+
+export function innerTextWithSpaces(elem) {
+	return JSDOM.fragment(elem).split(/\s+/u).join(' ').trim()
+}

modules/node_modules/@frogpond/ccc-lib/index.js

@@ -1,3 +1,4 @@
 export {get} from './http'
 export * from './cache'
 export * from './url'
+export * from './html'

modules/node_modules/@frogpond/ccc-presence/index.js

@@ -1,10 +1,8 @@
-import {get, ONE_HOUR} from '@frogpond/ccc-lib'
+import {get, ONE_HOUR, parseHtml} from '@frogpond/ccc-lib'
 import mem from 'mem'
 import lodash from 'lodash'
-import _jsdom from 'jsdom'
 import pMap from 'p-map'
 const {sortBy, startCase} = lodash
-const {JSDOM} = _jsdom
 
 /*
 type ContactPersonType = {
@@ -45,10 +43,13 @@ export function cleanOrg(org) {
 	// )
 
 	let category = org.categories.join(', ')
-	let meetings =
-		(org.regularMeetingLocation || '').trim() +
-		(org.regularMeetingTime || '').trim()
-	let description = JSDOM.fragment(org.description).textContent.trim()
+
+	let meetingTime = org.regularMeetingTime || ''
+	let meetingLocation = parseHtml(org.regularMeetingLocation || '')
+	let meetings = `${meetingTime} ${meetingLocation}`.trim()
+
+	let description = parseHtml(org.description)
+
 	let website = (org.website || '').trim()
 	if (website && !/^https?:\/\//.test(website)) {
 		website = `http://${website}`

modules/node_modules/@frogpond/ccc-reason-calendar/index.js

@@ -1,13 +1,11 @@
 /* eslint-disable camelcase */
 
-import {get} from '@frogpond/ccc-lib'
+import {get, parseHtml} from '@frogpond/ccc-lib'
 import moment from 'moment-timezone'
 import dropWhile from 'lodash/dropWhile'
 import dropRightWhile from 'lodash/dropRightWhile'
 import sortBy from 'lodash/sortBy'
 import getUrls from 'get-urls'
-import _jsdom from 'jsdom'
-const {JSDOM} = _jsdom
 
 const TZ = 'US/Central'
 
@@ -113,16 +111,16 @@ function convertReasonEvent(event, now = moment()) {
 		moment(event.startTime).isBefore(now, 'day') &&
 		moment(event.endTime).isSameOrAfter(now)
 
-	let description = (event.description || '').replace('<br>', '\n')
-	description = JSDOM.fragment(description).textContent.trim()
+	let title = parseHtml(event.name || '')
+	let description = cleanTextBlock(event.description || '')
 
 	let links = description ? [...getUrls(description)] : []
 
 	return {
 		dataSource: 'reason',
 		startTime: event.startTime,
 		endTime: event.endTime,
-		title: event.name || '',
+		title: title,
 		description: description,
 		location: event.location || '',
 		links: links,
@@ -138,6 +136,10 @@ function convertReasonEvent(event, now = moment()) {
 	}
 }
 
+function cleanTextBlock(text) {
+	return parseHtml(text).replace('<br>', '\n').trim()
+}
+
 export async function reasonCalendar(calendarUrl, now = moment()) {
 	let dateParams = {
 		// eslint-disable-next-line camelcase

modules/node_modules/@frogpond/ccc-rss-feed/index.js

@@ -1,4 +1,4 @@
-import {get} from '@frogpond/ccc-lib'
+import {get, parseHtml} from '@frogpond/ccc-lib'
 import _jsdom from 'jsdom'
 const {JSDOM} = _jsdom
 
@@ -25,7 +25,7 @@ export function convertRssItemToStory(item) {
 
 	let title = item.querySelector('title')
 	title = title ? title.textContent : '(no title)'
-	title = JSDOM.fragment(title).textContent.trim()
+	title = parseHtml(title)
 
 	let datePublished = item.querySelector('pubDate')
 	datePublished = datePublished ? datePublished.textContent : null
@@ -35,12 +35,12 @@ export function convertRssItemToStory(item) {
 	let content = item.getAttribute('content:encoded')
 	content = content || (descriptionEl && descriptionEl.textContent)
 	content = content || '(no content)'
-	content = JSDOM.fragment(content).textContent.trim()
+	content = parseHtml(content)
 
 	let excerpt = descriptionEl
 		? descriptionEl.textContent
 		: content.substr(0, 250)
-	excerpt = JSDOM.fragment(excerpt).textContent.trim()
+	excerpt = parseHtml(excerpt)
 
 	let featuredImage = null
 	if (item.querySelector('enclosure')) {

modules/node_modules/@frogpond/ccc-wpjson-feed/index.js

@@ -1,6 +1,4 @@
-import {get} from '@frogpond/ccc-lib'
-import _jsdom from 'jsdom'
-const {JSDOM} = _jsdom
+import {get, parseHtml} from '@frogpond/ccc-lib'
 
 export async function fetchWpJson(url, query = {}) {
 	const feed = await get(url, {query, json: true})
@@ -47,10 +45,10 @@ export function convertWpJsonItemToStory(item) {
 		categories: categories,
 		content: item.content.rendered,
 		datePublished: item.date_gmt,
-		excerpt: JSDOM.fragment(item.excerpt.rendered).textContent.trim(),
+		excerpt: parseHtml(item.excerpt.rendered),
 		featuredImage: featuredImage,
 		link: item.link,
-		title: JSDOM.fragment(item.title.rendered).textContent.trim(),
+		title: parseHtml(item.title.rendered),
 	}
 }
 

modules/node_modules/@frogpond/ccci-carleton-college/v1/convos/index.js

@@ -1,4 +1,4 @@
-import {get, ONE_HOUR, makeAbsoluteUrl} from '@frogpond/ccc-lib'
+import {get, ONE_HOUR, makeAbsoluteUrl, parseHtml} from '@frogpond/ccc-lib'
 import {fromHtml} from '@frogpond/ccc-markdown'
 import mem from 'mem'
 import _jsdom from 'jsdom'
@@ -9,15 +9,15 @@ const archiveBase =
 	'https://apps.carleton.edu/events/convocations/feeds/media_files?page_id=342645'
 
 function processConvo(event) {
-	let title = JSDOM.fragment(
+	let title = parseHtml(
 		event.querySelector('title').textContent,
-	).textContent.trim()
+	)
 
 	let description = event.querySelector('description')
 	description = description
-		? JSDOM.fragment(
+		? parseHtml(
 				event.querySelector('description').textContent,
-		  ).textContent.trim()
+		  )
 		: ''
 
 	let pubDate = moment(event.querySelector('pubDate').textContent)

modules/node_modules/@frogpond/ccci-carleton-college/v1/news/nnb.js

@@ -1,4 +1,4 @@
-import {get} from '@frogpond/ccc-lib'
+import {get, parseHtml} from '@frogpond/ccc-lib'
 import _jsdom from 'jsdom'
 import lodash from 'lodash'
 const {groupBy, toPairs} = lodash
@@ -13,9 +13,9 @@ export async function noonNewsBulletein() {
 	let bulletinEls = [...dom.window.document.querySelectorAll('item')]
 	let bulletins = bulletinEls.map((item) => {
 		let description = item.querySelector('description').textContent
-		description = JSDOM.fragment(description).textContent.trim()
+		description = parseHtml(description)
 		let category = item.querySelector('category').textContent
-		category = JSDOM.fragment(category).textContent.trim()
+		category = parseHtml(category)
 		return {description, category}
 	})
 

modules/node_modules/@frogpond/ccci-stolaf-college/v1/jobs/index.js

@@ -1,22 +1,19 @@
-import {get, ONE_DAY} from '@frogpond/ccc-lib'
+import {get, ONE_DAY, parseHtml} from '@frogpond/ccc-lib'
 import mem from 'mem'
-import _jsdom from 'jsdom'
 import getUrls from 'get-urls'
 
-const {JSDOM} = _jsdom
-
 export function cleanJob(job) {
+	const title = parseHtml(job.title)
+	const office = parseHtml(job.office)
+	const hoursPerWeek = parseHtml(job.hoursPerWeek)
+	const timeOfHours = parseHtml(job.timeOfHours)
+
 	// these all need to retain their newlines
-	const description = cleanTextBlock(
-		JSDOM.fragment(job.description).textContent,
-	)
-	const comments = cleanTextBlock(JSDOM.fragment(job.comments).textContent)
-	const skills = cleanTextBlock(JSDOM.fragment(job.skills).textContent)
-	const howToApply = cleanTextBlock(JSDOM.fragment(job.howToApply).textContent)
-	const timeline = cleanTextBlock(JSDOM.fragment(job.timeline).textContent)
-	const timeOfHours = cleanTextBlock(
-		JSDOM.fragment(job.timeOfHours).textContent,
-	)
+	const description = cleanTextBlock(job.description)
+	const comments = cleanTextBlock(job.comments)
+	const skills = cleanTextBlock(job.skills)
+	const howToApply = cleanTextBlock(job.howToApply)
+	const timeline = cleanTextBlock(job.timeline)
 
 	const contactEmail = fixupEmailFormat(job.contactEmail)
 	const contactPhone = fixupPhoneFormat(job.contactPhone)
@@ -34,12 +31,12 @@ export function cleanJob(job) {
 		contactPhone: contactPhone,
 		description: description,
 		goodForIncomingStudents: job.goodForIncomingStudents,
-		hoursPerWeek: job.hoursPerWeek,
+		hoursPerWeek: hoursPerWeek,
 		howToApply: howToApply,
 		id: job.id,
 		lastModified: job.lastModified,
 		links: links,
-		office: job.office,
+		office: office,
 		openPositions: job.openPositions,
 		skills: skills,
 		timeline: timeline,
@@ -51,7 +48,7 @@ export function cleanJob(job) {
 }
 
 function cleanTextBlock(text) {
-	return text.replace(/\s+/g, ' ')
+	return parseHtml(text).replace(/\t/g, ' ').trim()
 }
 
 export function getLinksFromJob({description, comments, skills, howToApply}) {

modules/node_modules/@frogpond/ccci-stolaf-college/v1/streams/index.js

@@ -1,4 +1,4 @@
-import {get, ONE_HOUR} from '@frogpond/ccc-lib'
+import {get, ONE_HOUR, parseHtml} from '@frogpond/ccc-lib'
 import mem from 'mem'
 import moment from 'moment-timezone'
 
@@ -19,9 +19,15 @@ export async function getStreams({streamClass, sort, dateFrom, dateTo}) {
 		(resp) => resp.body,
 	)
 	const processed = data.results.map((stream) => {
-		let {starttime} = stream
+		let {starttime, title, subtitle, performer} = stream
+
+		let streamTitle = parseHtml(title)
+		let detail = parseHtml(subtitle || performer || '')
+
 		return {
 			...stream,
+			title: streamTitle,
+			subtitle: detail,
 			starttime: moment
 				.tz(starttime, 'YYYY-MM-DD HH:mm', 'America/Chicago')
 				.toISOString(),

package.json

@@ -22,11 +22,11 @@
     "test": "./scripts/smoke-test.sh"
   },
   "dependencies": {
+    "@frogpond/titlecase": "^1.0.0",
     "dotenv": "10.0.0",
     "esm": "3.2.25",
     "get-urls": "10.0.1",
     "got": "9.6.0",
-    "html-entities": "2.3.2",
     "is-absolute-url": "3.0.3",
     "jsdom": "16.6.0",
     "koa": "2.13.4",

yarn.lock

@@ -109,6 +109,11 @@
     minimatch "^3.0.4"
     strip-json-comments "^3.1.1"
 
+"@frogpond/titlecase@^1.0.0":
+  version "1.0.0"
+  resolved "https://registry.yarnpkg.com/@frogpond/titlecase/-/titlecase-1.0.0.tgz#993e5371c31b58a839e76fdc531053c55b1a2736"
+  integrity sha512-C1qKm/J+B+cXo+7+ZHbRnt2iApx/IrxMEXwOxe+ZkeTvSC1nZ2XRWf0xzFuAanpBNDhFcgrxdnR8FMKNHo1scQ==
+
 "@gar/promisify@^1.0.1":
   version "1.1.2"
   resolved "https://registry.yarnpkg.com/@gar/promisify/-/promisify-1.1.2.tgz#30aa825f11d438671d585bd44e7fd564535fc210"
@@ -1130,11 +1135,6 @@ html-encoding-sniffer@^2.0.1:
   dependencies:
     whatwg-encoding "^1.0.5"
 
-[email protected]:
-  version "2.3.2"
-  resolved "https://registry.yarnpkg.com/html-entities/-/html-entities-2.3.2.tgz#760b404685cb1d794e4f4b744332e3b00dcfe488"
-  integrity sha512-c3Ab/url5ksaT0WyleslpBEthOzWhrjQbg75y7XUsfSzi3Dgzt0l8w5e7DylRn15MTlMMD58dTfzddNS2kcAjQ==
-
 http-assert@^1.3.0:
   version "1.4.1"
   resolved "https://registry.yarnpkg.com/http-assert/-/http-assert-1.4.1.tgz#c5f725d677aa7e873ef736199b89686cceb37878"