Merge pull request #2400 from mikiher/bookfinder-improvements

A few BookFinder improvements (including a fix for #2238)

client/components/modals/item/tabs/Match.vue

@@ -332,6 +332,7 @@ export default {
       if (this.isPodcast) return `term=${encodeURIComponent(this.searchTitle)}`
       var searchQuery = `provider=${this.provider}&fallbackTitleOnly=1&title=${encodeURIComponent(this.searchTitle)}`
       if (this.searchAuthor) searchQuery += `&author=${encodeURIComponent(this.searchAuthor)}`
+      if (this.libraryItemId) searchQuery += `&id=${this.libraryItemId}`
       return searchQuery
     },
     submitSearch() {

server/controllers/SearchController.js

@@ -3,15 +3,18 @@ const BookFinder = require('../finders/BookFinder')
 const PodcastFinder = require('../finders/PodcastFinder')
 const AuthorFinder = require('../finders/AuthorFinder')
 const MusicFinder = require('../finders/MusicFinder')
+const Database = require("../Database")
 
 class SearchController {
   constructor() { }
 
   async findBooks(req, res) {
+    const id = req.query.id
+    const libraryItem = await Database.libraryItemModel.getOldById(id)
     const provider = req.query.provider || 'google'
     const title = req.query.title || ''
     const author = req.query.author || ''
-    const results = await BookFinder.search(provider, title, author)
+    const results = await BookFinder.search(libraryItem, provider, title, author)
     res.json(results)
   }
 

server/finders/BookFinder.js

@@ -167,6 +167,7 @@ class BookFinder {
         [/ (2nd|3rd|\d+th)\s+ed(\.|ition)?/g, ''], // Remove edition
         [/(^| |\.)(m4b|m4a|mp3)( |$)/g, ''],       // Remove file-type
         [/ a novel.*$/g, ''],                      // Remove "a novel"
+        [/(^| )(un)?abridged( |$)/g, ' '],         // Remove "unabridged/abridged"
         [/^\d+ | \d+$/g, ''],                      // Remove preceding/trailing numbers
       ]
 
@@ -298,6 +299,7 @@ class BookFinder {
   /**
    * Search for books including fuzzy searches
    * 
+   * @param {Object} libraryItem
    * @param {string} provider 
    * @param {string} title 
    * @param {string} author 
@@ -306,7 +308,7 @@ class BookFinder {
    * @param {{titleDistance:number, authorDistance:number, maxFuzzySearches:number}} options 
    * @returns {Promise<Object[]>}
    */
-  async search(provider, title, author, isbn, asin, options = {}) {
+  async search(libraryItem, provider, title, author, isbn, asin, options = {}) {
     let books = []
     const maxTitleDistance = !isNaN(options.titleDistance) ? Number(options.titleDistance) : 4
     const maxAuthorDistance = !isNaN(options.authorDistance) ? Number(options.authorDistance) : 4
@@ -335,6 +337,7 @@ class BookFinder {
       for (const titlePart of titleParts)
         authorCandidates.add(titlePart)
       authorCandidates = await authorCandidates.getCandidates()
+      loop_author:
       for (const authorCandidate of authorCandidates) {
         let titleCandidates = new BookFinder.TitleCandidates(authorCandidate)
         for (const titlePart of titleParts)
@@ -342,13 +345,27 @@ class BookFinder {
         titleCandidates = titleCandidates.getCandidates()
         for (const titleCandidate of titleCandidates) {
           if (titleCandidate == title && authorCandidate == author) continue // We already tried this
-          if (++numFuzzySearches > maxFuzzySearches) return books
+          if (++numFuzzySearches > maxFuzzySearches) break loop_author
           books = await this.runSearch(titleCandidate, authorCandidate, provider, asin, maxTitleDistance, maxAuthorDistance)
-          if (books.length) return books
+          if (books.length) break loop_author
         }
       }
     }
 
+    if (books.length) {
+      const resultsHaveDuration = provider.startsWith('audible')
+      if (resultsHaveDuration && libraryItem?.media?.duration) {
+        const libraryItemDurationMinutes = libraryItem.media.duration / 60
+        // If provider results have duration, sort by ascendinge duration difference from libraryItem
+        books.sort((a, b) => {
+          const aDuration = a.duration || Number.POSITIVE_INFINITY
+          const bDuration = b.duration || Number.POSITIVE_INFINITY
+          const aDurationDiff = Math.abs(aDuration - libraryItemDurationMinutes)
+          const bDurationDiff = Math.abs(bDuration - libraryItemDurationMinutes)
+          return aDurationDiff - bDurationDiff
+        })
+      }
+    }
     return books
   }
 
@@ -392,12 +409,12 @@ class BookFinder {
 
     if (provider === 'all') {
       for (const providerString of this.providers) {
-        const providerResults = await this.search(providerString, title, author, options)
+        const providerResults = await this.search(null, providerString, title, author, options)
         Logger.debug(`[BookFinder] Found ${providerResults.length} covers from ${providerString}`)
         searchResults.push(...providerResults)
       }
     } else {
-      searchResults = await this.search(provider, title, author, options)
+      searchResults = await this.search(null, provider, title, author, options)
     }
     Logger.debug(`[BookFinder] FindCovers search results: ${searchResults.length}`)
 
@@ -461,6 +478,8 @@ function cleanAuthorForCompares(author) {
   cleanAuthor = cleanAuthor.replace(/([a-z])\.([a-z])/g, '$1. $2')
   // remove middle initials
   cleanAuthor = cleanAuthor.replace(/(?<=\w\w)(\s+[a-z]\.?)+(?=\s+\w\w)/g, '')
+  // remove et al.
+  cleanAuthor = cleanAuthor.replace(/ et al\.?(?= |$)/g, '')
   return cleanAuthor
 }
 

server/scanner/Scanner.js

@@ -37,7 +37,7 @@ class Scanner {
       var searchISBN = options.isbn || libraryItem.media.metadata.isbn
       var searchASIN = options.asin || libraryItem.media.metadata.asin
 
-      var results = await BookFinder.search(provider, searchTitle, searchAuthor, searchISBN, searchASIN, { maxFuzzySearches: 2 })
+      var results = await BookFinder.search(libraryItem, provider, searchTitle, searchAuthor, searchISBN, searchASIN, { maxFuzzySearches: 2 })
       if (!results.length) {
         return {
           warning: `No ${provider} match found`

test/server/finders/BookFinder.test.js

@@ -35,6 +35,8 @@ describe('TitleCandidates', () => {
         ['adds candidate + variant, removing edition 2', 'anna karenina 4th ed.', ['anna karenina', 'anna karenina 4th ed.']],
         ['adds candidate + variant, removing fie type', 'anna karenina.mp3', ['anna karenina', 'anna karenina.mp3']],
         ['adds candidate + variant, removing "a novel"', 'anna karenina a novel', ['anna karenina', 'anna karenina a novel']],
+        ['adds candidate + variant, removing "abridged"', 'abridged anna karenina', ['anna karenina', 'abridged anna karenina']],
+        ['adds candidate + variant, removing "unabridged"', 'anna karenina unabridged', ['anna karenina', 'anna karenina unabridged']],
         ['adds candidate + variant, removing preceding/trailing numbers', '1 anna karenina 2', ['anna karenina', '1 anna karenina 2']],
         ['does not add empty candidate', '', []],
         ['does not add spaces-only candidate', '   ', []],
@@ -109,6 +111,7 @@ describe('AuthorCandidates', () => {
         ['adds recognized author if edit distance from candidate is small', 'nicolai gogol', ['nikolai gogol']],
         ['does not add candidate if edit distance from any recognized author is large', 'nikolai google', []],
         ['adds normalized recognized candidate (contains redundant spaces)', 'nikolai    gogol', ['nikolai gogol']],
+        ['adds normalized recognized candidate (et al removed)', 'nikolai gogol et al.', ['nikolai gogol']],
         ['adds normalized recognized candidate (normalized initials)', 'j.k. rowling', ['j. k. rowling']],
       ].forEach(([name, author, expected]) => it(name, async () => {
         authorCandidates.add(author)
@@ -222,14 +225,14 @@ describe('search', () => {
 
   describe('search title is empty', () => {
     it('returns empty result', async () => {
-      expect(await bookFinder.search('', '', a)).to.deep.equal([])
+      expect(await bookFinder.search(null, '', '', a)).to.deep.equal([])
       sinon.assert.callCount(bookFinder.runSearch, 0)
     })
   })
 
   describe('search title is a recognized title and search author is a recognized author', () => {
     it('returns non-empty result (no fuzzy searches)', async () => {
-      expect(await bookFinder.search('', t, a)).to.deep.equal(r)
+      expect(await bookFinder.search(null, '', t, a)).to.deep.equal(r)
       sinon.assert.callCount(bookFinder.runSearch, 1)
     })
   })
@@ -251,7 +254,7 @@ describe('search', () => {
       [`2022_${t}_HQ`],
     ].forEach(([searchTitle]) => {
       it(`search('${searchTitle}', '${a}') returns non-empty result (with 1 fuzzy search)`, async () => {
-        expect(await bookFinder.search('', searchTitle, a)).to.deep.equal(r)
+        expect(await bookFinder.search(null, '', searchTitle, a)).to.deep.equal(r)
         sinon.assert.callCount(bookFinder.runSearch, 2)
       })
     });
@@ -261,7 +264,7 @@ describe('search', () => {
       [`${a} - series 01 - ${t}`],
     ].forEach(([searchTitle]) => {
       it(`search('${searchTitle}', '${a}') returns non-empty result (with 2 fuzzy searches)`, async () => {
-        expect(await bookFinder.search('', searchTitle, a)).to.deep.equal(r)
+        expect(await bookFinder.search(null, '', searchTitle, a)).to.deep.equal(r)
         sinon.assert.callCount(bookFinder.runSearch, 3)
       })
     });
@@ -271,7 +274,7 @@ describe('search', () => {
       [`${t} junk`],
     ].forEach(([searchTitle]) => {
       it(`search('${searchTitle}', '${a}') returns an empty result`, async () => {
-        expect(await bookFinder.search('', searchTitle, a)).to.deep.equal([])
+        expect(await bookFinder.search(null, '', searchTitle, a)).to.deep.equal([])
       })
     })
 
@@ -280,7 +283,7 @@ describe('search', () => {
         [`${t} - ${a}`],
       ].forEach(([searchTitle]) => {
         it(`search('${searchTitle}', '${a}') returns an empty result (with no fuzzy searches)`, async () => {
-          expect(await bookFinder.search('', searchTitle, a, null, null, { maxFuzzySearches: 0 })).to.deep.equal([])
+          expect(await bookFinder.search(null, '', searchTitle, a, null, null, { maxFuzzySearches: 0 })).to.deep.equal([])
           sinon.assert.callCount(bookFinder.runSearch, 1)
         })
       })
@@ -292,7 +295,7 @@ describe('search', () => {
         [`${a} - series 01 - ${t}`],
       ].forEach(([searchTitle]) => {
         it(`search('${searchTitle}', '${a}') returns an empty result (1 fuzzy search)`, async () => {
-          expect(await bookFinder.search('', searchTitle, a, null, null, { maxFuzzySearches: 1 })).to.deep.equal([])
+          expect(await bookFinder.search(null, '', searchTitle, a, null, null, { maxFuzzySearches: 1 })).to.deep.equal([])
           sinon.assert.callCount(bookFinder.runSearch, 2)
         })
       })
@@ -305,7 +308,7 @@ describe('search', () => {
       [`${a} - ${t}`],
     ].forEach(([searchTitle]) => {
       it(`search('${searchTitle}', '') returns a non-empty result (1 fuzzy search)`, async () => {
-        expect(await bookFinder.search('', searchTitle, '')).to.deep.equal(r)
+        expect(await bookFinder.search(null, '', searchTitle, '')).to.deep.equal(r)
         sinon.assert.callCount(bookFinder.runSearch, 2)
       })
     });
@@ -316,7 +319,7 @@ describe('search', () => {
       [`${u} - ${t}`]
     ].forEach(([searchTitle]) => {
       it(`search('${searchTitle}', '') returns an empty result`, async () => {
-        expect(await bookFinder.search('', searchTitle, '')).to.deep.equal([])
+        expect(await bookFinder.search(null, '', searchTitle, '')).to.deep.equal([])
       })
     })
   })
@@ -327,7 +330,7 @@ describe('search', () => {
       [`${u} - ${t}`]
     ].forEach(([searchTitle]) => {
       it(`search('${searchTitle}', '${u}') returns a non-empty result (1 fuzzy search)`, async () => {
-        expect(await bookFinder.search('', searchTitle, u)).to.deep.equal(r)
+        expect(await bookFinder.search(null, '', searchTitle, u)).to.deep.equal(r)
         sinon.assert.callCount(bookFinder.runSearch, 2)
       })
     });
@@ -336,9 +339,41 @@ describe('search', () => {
       [`${t}`]
     ].forEach(([searchTitle]) => {
       it(`search('${searchTitle}', '${u}') returns a non-empty result (no fuzzy search)`, async () => {
-        expect(await bookFinder.search('', searchTitle, u)).to.deep.equal(r)
+        expect(await bookFinder.search(null, '', searchTitle, u)).to.deep.equal(r)
         sinon.assert.callCount(bookFinder.runSearch, 1)
       })
     })
   })
+
+  describe('search provider results have duration', () => {
+    const  libraryItem = { media: { duration: 60 * 1000 } }
+    const provider = 'audible'
+    const unsorted = [{ duration: 3000 }, { duration: 2000 }, { duration: 1000 }, { duration: 500 }]
+    const sorted = [{ duration: 1000 }, { duration: 500 }, { duration: 2000 }, { duration: 3000 }]
+    runSearchStub.withArgs(t, a, provider).resolves(unsorted)
+
+    it('returns results sorted by library item duration diff', async () => {
+      expect(await bookFinder.search(libraryItem, provider, t, a)).to.deep.equal(sorted)
+    })
+    
+    it('returns unsorted results if library item is null', async () => {
+      expect(await bookFinder.search(null, provider, t, a)).to.deep.equal(unsorted)
+    })
+
+    it('returns unsorted results if library item duration is undefined', async () => {
+      expect(await bookFinder.search({ media: {} }, provider, t, a)).to.deep.equal(unsorted)
+    })
+
+    it('returns unsorted results if library item media is undefined', async () => {
+      expect(await bookFinder.search({ }, provider, t, a)).to.deep.equal(unsorted)
+    })
+
+    it ('should return a result last if it has no duration', async () => {
+      const unsorted = [{}, { duration: 3000 }, { duration: 2000 }, { duration: 1000 }, { duration: 500 }]
+      const sorted = [{ duration: 1000 }, { duration: 500 }, { duration: 2000 }, { duration: 3000 }, {}]
+      runSearchStub.withArgs(t, a, provider).resolves(unsorted)
+
+      expect(await bookFinder.search(libraryItem, provider, t, a)).to.deep.equal(sorted)
+    })
+  })
 })