feat(book-audible-api): ♻️ use audible API for series data instead of…

… scraping

src/helpers/books/audibleApi.ts

@@ -2,7 +2,7 @@ import fetch from 'isomorphic-fetch'
 // For merchandising_summary
 import { htmlToText } from 'html-to-text'
 import { AudibleInterface } from '../../interfaces/audible/index'
-import { ApiBookInterface } from '../../interfaces/books/index'
+import { ApiBookInterface, SeriesInterface } from '../../interfaces/books/index'
 import { AuthorInterface, NarratorInterface } from '../../interfaces/people/index'
 import SharedHelper from '.././shared'
 
@@ -14,7 +14,7 @@ class ApiHelper {
         const helper = new SharedHelper()
         const baseDomain: string = 'https://api.audible.com'
         const baseUrl: string = '1.0/catalog/products'
-        const params = '?response_groups=contributors,product_desc,product_extended_attrs,product_attrs,media,rating&image_sizes=500,1024'
+        const params = '?response_groups=contributors,product_desc,product_extended_attrs,product_attrs,media,rating,series&image_sizes=500,1024'
         this.reqUrl = helper.buildUrl(asin, baseDomain, baseUrl, params)
     }
 
@@ -164,10 +164,30 @@ class ApiHelper {
         newKey = 'runtimeLengthMin'
         optionalKeyHandling(key, newKey)
 
-        // SeriesPrimary
-        key = 'publication_name'
-        newKey = 'publicationName'
-        optionalKeyHandling(key, newKey)
+        // Series
+        key = 'series'
+        if (key in inputJson) {
+            inputJson[key].forEach((series: { asin: string | undefined; title: string; sequence: string | undefined }) => {
+                const seriesJson = <SeriesInterface>{}
+                if ('asin' in series) {
+                    seriesJson.asin = series.asin
+                }
+                if ('name' in series) {
+                    seriesJson.name = series.title
+                } else {
+                    console.log(`Series name not available on: ${inputJson.asin}`)
+                    return undefined
+                }
+                if ('sequence' in series) {
+                    seriesJson.position = series.sequence
+                }
+                if (series.title === inputJson.publication_name!) {
+                    finalJson.seriesPrimary = seriesJson
+                } else if (inputJson.series.length > 1 && series.title !== inputJson.publication_name) {
+                    finalJson.seriesSecondary = seriesJson
+                }
+            })
+        }
 
         // Subtitle
         key = 'subtitle'

src/helpers/books/audibleScrape.ts

@@ -1,5 +1,5 @@
 // Import interfaces
-import { HtmlBookInterface, SeriesInterface } from '../../interfaces/books/index'
+import { HtmlBookInterface } from '../../interfaces/books/index'
 import { GenreInterface } from '../../interfaces/audible'
 import fetch from 'isomorphic-fetch'
 // For HTML scraping
@@ -53,45 +53,6 @@ class ScrapeHelper {
         return genreArr
     }
 
-    /**
-     * Checks the presence of series' on html page and formats them into JSON
-     * @param {NodeListOf<Element>} series selected source from seriesLabel
-     * @param {string} seriesRaw innerHTML of the series node
-     * @returns {SeriesInterface[]}
-     */
-    collectSeries (series: cheerio.Cheerio<cheerio.Element>[], seriesRaw: string): SeriesInterface[] | undefined {
-        const bookPos = this.getBookFromHTML(seriesRaw)
-
-        // What is the singular of series? Who knows
-        const seriesArr: SeriesInterface[] | undefined = series.map((serie, index): any => {
-            const thisSeries = {} as SeriesInterface
-            let asin: string
-            let href: string
-            if (serie.attr('href')) {
-                href = serie.attr('href')!
-                asin = this.getAsinFromUrl(href)
-
-                if (serie.text()) {
-                    thisSeries.asin = asin
-                    thisSeries.name = serie.text()
-
-                    if (bookPos && bookPos[0]) {
-                        thisSeries.position = bookPos[0]
-                    }
-
-                    return thisSeries
-                } else {
-                    console.log(`Series ${index} name not available on: ${this.asin}`)
-                }
-            } else {
-                console.log(`Series ${index} asin not available on: ${this.asin}`)
-            }
-            return undefined
-        }) as SeriesInterface[]
-
-        return seriesArr
-    }
-
     /**
      * Fetches the html page and checks it's response
      * @returns {Promise<cheerio.CheerioAPI | undefined>} return text from the html page
@@ -126,17 +87,12 @@ class ScrapeHelper {
         .toArray()
         .map(element => dom(element))
 
-        const series = dom('li.seriesLabel a')
-        .toArray()
-        .map(element => dom(element))
-
         const tags = dom('div.bc-chip-group a')
         .toArray()
         .map(element => dom(element))
 
         const returnJson = {
-            genres: Array<GenreInterface>(genres.length + tags.length),
-            series: Array<SeriesInterface>(series.length)
+            genres: Array<GenreInterface>(genres.length + tags.length)
         } as HtmlBookInterface
 
         // Combine genres and tags
@@ -153,12 +109,6 @@ class ScrapeHelper {
             returnJson.genres = genreArr as GenreInterface[]
         }
 
-        // Series
-        if (series.length) {
-            const seriesRaw = dom('li.seriesLabel').html()!
-            returnJson.series = this.collectSeries(series, seriesRaw)
-        }
-
         return returnJson
     }
 
@@ -173,17 +123,6 @@ class ScrapeHelper {
         const ASIN = url.match(asinRegex)![0]
         return ASIN
     }
-
-    /**
-     * Regex to return just the book position from HTML input
-     * @param {JSDOM} html block/object to retrieve book number from.
-     * @returns {string} Cleaned book position string, like "Book 3"
-     */
-    getBookFromHTML (html): string {
-        const bookRegex = /(Book ?(\d*\.)?\d+[+-]?[\d]?)/gm
-        const matches = html.match(bookRegex)
-        return matches
-    }
 }
 
 export default ScrapeHelper

src/helpers/books/audibleStitch.ts

@@ -1,4 +1,4 @@
-import { ApiBookInterface, BookInterface, HtmlBookInterface, SeriesInterface } from '../../interfaces/books/index'
+import { ApiBookInterface, BookInterface, HtmlBookInterface } from '../../interfaces/books/index'
 
 class StitchHelper {
     apiRes: ApiBookInterface;
@@ -19,41 +19,12 @@ class StitchHelper {
         }
     }
 
-    /**
-     * Sets series' keys if they exist
-     */
-    async setSeriesOrder () {
-        if (this.apiRes.publicationName) {
-            if (this.htmlRes) {
-                const htmlSeries = this.htmlRes.series
-
-                // If multiple series, set one with seriesPrimary as primary
-                if (htmlSeries) {
-                    if (htmlSeries.length > 1) {
-                        htmlSeries.forEach((item) => {
-                            if (item.name === this.apiRes.publicationName) {
-                                this.tempJson.seriesPrimary = item
-                            } else {
-                                this.tempJson.seriesSecondary = item
-                            }
-                        })
-                    } else {
-                        this.tempJson.seriesPrimary = htmlSeries[0]
-                    }
-                }
-            } else {
-                this.tempJson.seriesPrimary = { name: this.tempJson.publicationName } as SeriesInterface
-            }
-            delete this.tempJson.publicationName
-        }
-    }
-
     /**
      * Call functions in the class to parse final JSON
      * @returns {Promise<BookInterface>}
      */
     async process (): Promise<BookInterface> {
-        Promise.all([this.includeGenres(), this.setSeriesOrder()])
+        Promise.all([this.includeGenres()])
         this.bookJson = this.tempJson
         return this.bookJson
     }

src/interfaces/audible/index.ts

@@ -1,4 +1,5 @@
 /* eslint-disable camelcase */
+import { SeriesInterface } from '../books'
 import { AuthorInterface, NarratorInterface } from '../people/index'
 
 interface Codecs {
@@ -56,6 +57,7 @@ export interface AudibleInterface {
         rating: Ratings;
         release_date: string;
         runtime_length_min: number;
+        series: SeriesInterface[];
         social_media_images: any;
         subtitle?: string;
         thesaurus_subject_keywords: [string]

src/interfaces/books/index.ts

@@ -57,7 +57,4 @@ export interface ApiBookInterface extends CoreBook {
 // What we expect to keep from Audible's HTML pages
 export interface HtmlBookInterface {
     genres?: GenreInterface[];
-    series?: SeriesInterface[];
-    seriesPrimary?: SeriesInterface;
-    seriesSecondary?: SeriesInterface;
 }