This commit is contained in:
Andrew Kozhokaru 2026-05-06 00:22:07 +02:00 committed by GitHub
commit 12600ba04c
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
19 changed files with 2397 additions and 44 deletions

View file

@ -134,6 +134,13 @@ export default {
}
]
if (this.userCanUpdate && this.openAIConfigured) {
items.push({
text: 'Organize Story Order With AI',
action: 'organize-story-order-with-ai'
})
}
if (this.userIsAdminOrUp || this.selectedSeries.rssFeed) {
items.push({
text: this.$strings.LabelOpenRSSFeed,
@ -221,6 +228,9 @@ export default {
userCanDownload() {
return this.$store.getters['user/getUserCanDownload']
},
openAIConfigured() {
return !!this.$store.getters['getServerSetting']('openAIConfigured')
},
currentLibraryId() {
return this.$store.state.libraries.currentLibraryId
},
@ -427,6 +437,12 @@ export default {
seriesContextMenuAction({ action }) {
if (action === 'open-rss-feed') {
this.showOpenSeriesRSSFeed()
} else if (action === 'organize-story-order-with-ai') {
if (this.processingSeries) {
console.warn('Already processing series')
return
}
this.organizeStoryOrderWithAI()
} else if (action === 're-add-to-continue-listening') {
if (this.processingSeries) {
console.warn('Already processing series')
@ -453,6 +469,35 @@ export default {
feed: this.selectedSeries.rssFeed
})
},
organizeStoryOrderWithAI() {
const payload = {
message: `Organize "${this.selectedSeries.name}" into AI-detected story order? This will update the series sequence on every book in the series.`,
callback: (confirmed) => {
if (!confirmed) return
this.processingSeries = true
this.$axios
.$post(`/api/series/${this.seriesId}/organize-story-order`)
.then((data) => {
if (!data.updated) {
this.$toast.info(this.$strings.ToastNoUpdatesNecessary)
} else {
this.$toast.success(`Updated story order for ${data.updated} books`)
}
this.$eventBus.$emit('series-books-updated', { seriesId: this.seriesId })
})
.catch((error) => {
console.error('Failed to organize story order with AI', error)
this.$toast.error(error.response?.data || this.$strings.ToastFailedToUpdate)
})
.finally(() => {
this.processingSeries = false
})
},
type: 'yesNo'
}
this.$store.commit('globals/setConfirmPrompt', payload)
},
reAddSeriesToContinueListening() {
this.processingSeries = true
this.$axios

View file

@ -602,6 +602,11 @@ export default {
this.libraryItemUpdated(ab)
})
},
seriesBooksUpdated(payload) {
if (this.entityName !== 'series-books') return
if (payload?.seriesId !== this.seriesId) return
this.resetEntities(this.currScrollTop)
},
collectionAdded(collection) {
if (this.entityName !== 'collections') return
console.log(`[LazyBookshelf] collectionAdded ${collection.id}`, collection)
@ -791,6 +796,7 @@ export default {
this.$eventBus.$on('bookshelf_clear_selection', this.clearSelectedEntities)
this.$eventBus.$on('user-settings', this.settingsUpdated)
this.$eventBus.$on('series-books-updated', this.seriesBooksUpdated)
if (this.$root.socket) {
this.$root.socket.on('item_updated', this.libraryItemUpdated)
@ -822,6 +828,7 @@ export default {
this.$eventBus.$off('bookshelf_clear_selection', this.clearSelectedEntities)
this.$eventBus.$off('user-settings', this.settingsUpdated)
this.$eventBus.$off('series-books-updated', this.seriesBooksUpdated)
if (this.$root.socket) {
this.$root.socket.off('item_updated', this.libraryItemUpdated)

View file

@ -127,6 +127,7 @@ export default {
autoScanCronExpression: null,
hideSingleBookSeries: false,
onlyShowLaterBooksInContinueSeries: false,
openAIDirectoryGrouping: false,
metadataPrecedence: ['folderStructure', 'audioMetatags', 'nfoFile', 'txtFiles', 'opfFile', 'absMetadata'],
markAsFinishedPercentComplete: null,
markAsFinishedTimeRemaining: 10

View file

@ -1,5 +1,12 @@
<template>
<div class="w-full h-full px-1 md:px-4 py-1 mb-4">
<div class="flex items-center justify-between md:justify-start mb-4">
<div class="flex items-center">
<ui-toggle-switch v-model="openAIDirectoryGrouping" @input="updated" />
<p class="pl-4 text-sm text-gray-300">Use OpenAI to interpret poor directory trees during library scans</p>
</div>
</div>
<div class="flex items-center justify-between mb-2">
<h2 class="text-base md:text-lg text-gray-200">{{ $strings.HeaderMetadataOrderOfPrecedence }}</h2>
<ui-btn small @click="resetToDefault">{{ $strings.ButtonResetToDefault }}</ui-btn>
@ -66,6 +73,11 @@ export default {
name: 'Audio file meta tags OR ebook metadata',
include: true
},
openAIPathMetadata: {
id: 'openAIPathMetadata',
name: 'OpenAI path and filename inference',
include: false
},
nfoFile: {
id: 'nfoFile',
name: 'NFO file',
@ -87,7 +99,8 @@ export default {
include: true
}
},
metadataSourceMapped: []
metadataSourceMapped: [],
openAIDirectoryGrouping: false
}
},
computed: {
@ -126,6 +139,7 @@ export default {
metadataSourceIds.reverse()
return {
settings: {
openAIDirectoryGrouping: !!this.openAIDirectoryGrouping,
metadataPrecedence: metadataSourceIds
}
}
@ -140,6 +154,7 @@ export default {
this.$emit('update', this.getLibraryData())
},
init() {
this.openAIDirectoryGrouping = !!this.librarySettings.openAIDirectoryGrouping
const metadataPrecedence = this.librarySettings.metadataPrecedence || []
this.metadataSourceMapped = metadataPrecedence.map((source) => this.metadataSourceData[source]).filter((s) => s)
@ -157,4 +172,4 @@ export default {
this.init()
}
}
</script>
</script>

View file

@ -1,5 +1,34 @@
<template>
<div class="w-full h-full px-1 md:px-2 py-1 mb-4">
<div v-if="isBookLibrary" class="w-full border border-black-200 p-4 my-8">
<div class="flex flex-wrap items-center">
<div>
<p class="text-lg">Detect Missing Series With AI</p>
<p class="max-w-sm text-sm pt-2 text-gray-300">Analyze books in this library and add missing series names and sequence values using OpenAI. Use the full re-evaluation option after editing book metadata and you want existing series assignments reconsidered.</p>
</div>
<div class="grow" />
<div>
<ui-btn class="mb-3 block" :disabled="processing || !openAIConfigured" @click.stop="detectSeriesWithAI">Detect Missing Series</ui-btn>
<ui-btn :disabled="processing || !openAIConfigured" @click.stop="reEvaluateSeriesWithAI">Re-evaluate All Series</ui-btn>
</div>
</div>
<p v-if="!openAIConfigured" class="text-sm text-yellow-400 mt-3">Configure OpenAI first in server settings.</p>
</div>
<div v-if="isBookLibrary" class="w-full border border-black-200 p-4 my-8">
<div class="flex flex-wrap items-center">
<div>
<p class="text-lg">Dedupe Books With AI</p>
<p class="max-w-sm text-sm pt-2 text-gray-300">Analyze likely duplicate books in this library with OpenAI, keep the best copy, and remove the duplicate items. This deletes duplicate files from disk.</p>
</div>
<div class="grow" />
<div>
<ui-btn :disabled="processing || !openAIConfigured" @click.stop="dedupeBooksWithAI">Dedupe Books</ui-btn>
</div>
</div>
<p v-if="!openAIConfigured" class="text-sm text-yellow-400 mt-3">Configure OpenAI first in server settings.</p>
</div>
<div class="w-full border border-black-200 p-4 my-8">
<div class="flex flex-wrap items-center">
<div>
@ -38,9 +67,86 @@ export default {
},
isBookLibrary() {
return this.mediaType === 'book'
},
openAIConfigured() {
return !!this.$store.getters['getServerSetting']('openAIConfigured')
}
},
methods: {
detectSeriesWithAI() {
const payload = {
message: 'Detect missing series in this library with AI? This only fills books that currently have no series metadata.',
callback: (confirmed) => {
if (confirmed) {
this.runSeriesDetection(true)
}
},
type: 'yesNo'
}
this.$store.commit('globals/setConfirmPrompt', payload)
},
reEvaluateSeriesWithAI() {
const payload = {
message: 'Re-evaluate all books in this library with AI? This can update sequence values for books that already have series metadata.',
callback: (confirmed) => {
if (confirmed) {
this.runSeriesDetection(false)
}
},
type: 'yesNo'
}
this.$store.commit('globals/setConfirmPrompt', payload)
},
dedupeBooksWithAI() {
const payload = {
message: 'Deduplicate books in this library with AI? Duplicate items chosen for removal will be deleted from the database and file system.',
callback: (confirmed) => {
if (confirmed) {
this.runBookDedupe()
}
},
type: 'yesNo'
}
this.$store.commit('globals/setConfirmPrompt', payload)
},
runSeriesDetection(onlyMissing = true) {
this.$emit('update:processing', true)
this.$axios
.$post(`/api/libraries/${this.libraryId}/detect-series-with-ai?onlyMissing=${onlyMissing ? 1 : 0}`)
.then((data) => {
if (!data.updated) {
this.$toast.info(this.$strings.ToastNoUpdatesNecessary)
} else {
this.$toast.success(onlyMissing ? `AI added series data to ${data.updated} books` : `AI re-evaluated series data for ${data.updated} books`)
}
})
.catch((error) => {
console.error('Failed to detect series with AI', error)
this.$toast.error(error.response?.data || this.$strings.ToastFailedToUpdate)
})
.finally(() => {
this.$emit('update:processing', false)
})
},
runBookDedupe() {
this.$emit('update:processing', true)
this.$axios
.$post(`/api/libraries/${this.libraryId}/dedupe-books-with-ai?hard=1`)
.then((data) => {
if (!data.duplicatesRemoved) {
this.$toast.info(this.$strings.ToastNoUpdatesNecessary)
} else {
this.$toast.success(`AI removed ${data.duplicatesRemoved} duplicate books`)
}
})
.catch((error) => {
console.error('Failed to dedupe books with AI', error)
this.$toast.error(error.response?.data || this.$strings.ToastFailedToUpdate)
})
.finally(() => {
this.$emit('update:processing', false)
})
},
removeAllMetadataClick(ext) {
const payload = {
message: this.$getString('MessageConfirmRemoveMetadataFiles', [ext]),

View file

@ -151,6 +151,38 @@
<div class="py-2">
<ui-multi-select v-model="newServerSettings.allowedOrigins" :items="newServerSettings.allowedOrigins" :label="$strings.LabelCorsAllowed" class="max-w-72" @input="updateCorsOrigins" />
</div>
<div class="pt-4">
<h2 class="font-semibold">AI</h2>
</div>
<div class="border border-white/10 rounded-lg p-4 mt-2">
<div class="flex items-center">
<p class="text-base font-semibold">OpenAI Series Tools</p>
<ui-tooltip text="Environment overrides supported: OPENAI_API_KEY, OPENAI_MODEL, OPENAI_BASE_URL.">
<span class="material-symbols icon-text ml-2 cursor-help">info</span>
</ui-tooltip>
</div>
<p class="text-sm text-white/70 mt-2">Use OpenAI to detect missing series in a book library and organize books inside a series into story order.</p>
<p class="text-sm text-white/70 mt-2">Status: {{ openAIStatusLabel }}</p>
<div class="mt-4">
<ui-text-input-with-label v-model="openAISettings.openAIModel" :disabled="savingOpenAISettings" label="Model" />
</div>
<div class="mt-2">
<ui-text-input-with-label v-model="openAISettings.openAIBaseURL" :disabled="savingOpenAISettings" label="Base URL" />
</div>
<div class="mt-2">
<ui-text-input-with-label v-model="openAISettings.openAIApiKey" type="password" :disabled="savingOpenAISettings" :placeholder="openAIConfigured ? 'Leave blank to keep current key' : 'sk-...'" label="API Key" />
</div>
<div class="flex items-center pt-4">
<ui-btn color="bg-success" small :loading="savingOpenAISettings" @click="saveOpenAISettings">Save AI Settings</ui-btn>
<ui-btn v-if="serverSettings && serverSettings.openAIConfigurationSource === 'settings'" color="bg-bg" small class="ml-2" :loading="savingOpenAISettings" @click="clearOpenAISettingsKey">Clear Saved Key</ui-btn>
</div>
</div>
</div>
</div>
</app-settings-content>
@ -232,7 +264,13 @@ export default {
hasPrefixesChanged: false,
newServerSettings: {},
showConfirmPurgeCache: false,
savingPrefixes: false
savingPrefixes: false,
savingOpenAISettings: false,
openAISettings: {
openAIModel: 'gpt-5.4-mini',
openAIBaseURL: 'https://api.openai.com/v1',
openAIApiKey: ''
}
}
},
watch: {
@ -256,6 +294,14 @@ export default {
timeFormats() {
return this.$store.state.globals.timeFormats
},
openAIConfigured() {
return !!this.serverSettings?.openAIConfigured
},
openAIStatusLabel() {
if (!this.serverSettings?.openAIConfigured) return 'Not configured'
if (this.serverSettings.openAIConfigurationSource === 'environment') return 'Configured from environment variables'
return 'Configured in server settings'
},
dateExample() {
const date = new Date(2014, 2, 25)
return this.$formatJsDate(date, this.newServerSettings.dateFormat)
@ -362,11 +408,68 @@ export default {
}
})
},
saveOpenAISettings() {
const openAIModel = this.openAISettings.openAIModel.trim()
const openAIBaseURL = this.openAISettings.openAIBaseURL.trim().replace(/\/+$/, '')
const payload = {
openAIModel,
openAIBaseURL
}
if (!openAIModel) {
this.$toast.error('OpenAI model is required')
return
}
if (!openAIBaseURL) {
this.$toast.error('OpenAI base URL is required')
return
}
try {
new URL(openAIBaseURL)
} catch {
this.$toast.error('OpenAI base URL must be a valid URL')
return
}
const openAIApiKey = this.openAISettings.openAIApiKey.trim()
if (openAIApiKey) {
payload.openAIApiKey = openAIApiKey
}
this.savingOpenAISettings = true
this.$store.dispatch('updateServerSettings', payload).then((response) => {
if (response.error) {
this.$toast.error(response.error)
} else {
this.$toast.success(this.$strings.ToastServerSettingsUpdateSuccess)
this.initServerSettings()
}
this.savingOpenAISettings = false
})
},
clearOpenAISettingsKey() {
this.savingOpenAISettings = true
this.$store.dispatch('updateServerSettings', { openAIApiKey: null }).then((response) => {
if (response.error) {
this.$toast.error(response.error)
} else {
this.$toast.success(this.$strings.ToastServerSettingsUpdateSuccess)
this.initServerSettings()
}
this.savingOpenAISettings = false
})
},
initServerSettings() {
this.newServerSettings = this.serverSettings ? { ...this.serverSettings } : {}
this.newServerSettings.sortingPrefixes = [...(this.newServerSettings.sortingPrefixes || [])]
this.newServerSettings.allowedOrigins = [...(this.newServerSettings.allowedOrigins || [])]
this.scannerEnableWatcher = !this.newServerSettings.scannerDisableWatcher
this.openAISettings = {
openAIModel: this.serverSettings?.openAIModel || 'gpt-5.4-mini',
openAIBaseURL: this.serverSettings?.openAIBaseURL || 'https://api.openai.com/v1',
openAIApiKey: ''
}
this.homepageUseBookshelfView = this.newServerSettings.homeBookshelfView != this.$constants.BookshelfView.DETAIL
this.useBookshelfView = this.newServerSettings.bookshelfView != this.$constants.BookshelfView.DETAIL

View file

@ -29,6 +29,7 @@ Audiobookshelf is a self-hosted audiobook and podcast server.
- Progressive Web App (PWA)
- Chromecast support on the web app and android app
- Fetch metadata and cover art from several sources
- OpenAI-assisted library tools for series detection, story ordering, scan-time metadata inference, poor directory-tree interpretation, and duplicate-book cleanup
- Chapter editor and chapter lookup (using [Audnexus API](https://audnex.us/))
- Merge your audio files into a single m4b
- Embed metadata and cover image into your audio files
@ -73,6 +74,42 @@ Check out the [API documentation](https://api.audiobookshelf.org/)
See [documentation](https://audiobookshelf.org/docs#book-directory-structure) for supported directory structure, folder naming conventions, and audio file metadata usage.
### OpenAI-assisted library organization
Audiobookshelf can optionally use OpenAI to help organize difficult audiobook libraries when filenames, folders, and metadata are inconsistent.
Available OpenAI features in this branch:
- Detect missing series assignments for books in a library
- Re-evaluate existing series assignments after metadata changes
- Organize books inside a series into story order
- Infer book metadata during scans from messy paths, filenames, and embedded tag data
- Interpret poor directory trees during scans to split ambiguous folders into better logical library items
- Detect likely duplicate books in library tools and remove duplicate copies
Configuration:
- Open the web app and go to `Settings -> AI`
- Enter your OpenAI API key, model, and base URL
- Environment overrides are also supported through `OPENAI_API_KEY`, `OPENAI_MODEL`, and `OPENAI_BASE_URL`
Library usage:
- `Config -> Libraries -> Edit Library -> Scanner`
- Enable `OpenAI path and filename inference` to let scans infer metadata from weak path structure
- Enable `Use OpenAI to interpret poor directory trees during library scans` to let scans group difficult folder layouts more intelligently
Library tools:
- `Config -> Libraries -> Edit Library -> Tools`
- `Detect Missing Series`
- `Re-evaluate All Series`
- `Dedupe Books`
Series page usage:
- Open a series page and use `Organize Story Order With AI`
<br />
# Installation

View file

@ -19,12 +19,15 @@ const Scanner = require('../scanner/Scanner')
const Database = require('../Database')
const Watcher = require('../Watcher')
const RssFeedManager = require('../managers/RssFeedManager')
const OpenAI = require('../providers/OpenAI')
const libraryFilters = require('../utils/queries/libraryFilters')
const libraryItemsPodcastFilters = require('../utils/queries/libraryItemsPodcastFilters')
const authorFilters = require('../utils/queries/authorFilters')
const zipHelpers = require('../utils/zipHelpers')
const openAI = new OpenAI()
/**
* @typedef RequestUserObject
* @property {import('../models/User')} user
@ -1463,6 +1466,422 @@ class LibraryController {
}
}
async getLibraryBooksForAISeriesDetection(libraryId) {
const books = await Database.bookModel.findAll({
include: [
{
model: Database.libraryItemModel,
required: true,
where: {
libraryId,
isMissing: false,
isInvalid: false
}
},
{
model: Database.bookAuthorModel,
include: {
model: Database.authorModel
},
separate: true,
order: [['createdAt', 'ASC']]
},
{
model: Database.bookSeriesModel,
include: {
model: Database.seriesModel
},
separate: true,
order: [['createdAt', 'ASC']]
}
],
order: [['title', 'ASC']]
})
return books.map((book) => {
const libraryItem = book.libraryItem
delete book.dataValues.libraryItem
book.authors = book.bookAuthors?.map((bookAuthor) => bookAuthor.author) || []
delete book.dataValues.bookAuthors
book.series =
book.bookSeries?.map((bookSeries) => {
const series = bookSeries.series
delete bookSeries.dataValues.series
series.bookSeries = bookSeries
return series
}) || []
delete book.dataValues.bookSeries
libraryItem.media = book
return libraryItem
})
}
groupLibraryBooksByPrimaryAuthor(libraryItems) {
const groups = new Map()
for (const libraryItem of libraryItems) {
const primaryAuthor = libraryItem.media.authors?.[0]?.name?.trim()
if (!primaryAuthor) continue
const key = primaryAuthor.toLowerCase()
if (!groups.has(key)) {
groups.set(key, {
authorName: primaryAuthor,
libraryItems: []
})
}
groups.get(key).libraryItems.push(libraryItem)
}
return [...groups.values()]
}
getLibraryItemFolderKey(libraryItem) {
const basePath = (libraryItem.relPath || libraryItem.path || '').replace(/\\/g, '/')
if (!basePath) return null
const itemPath = libraryItem.isFile ? Path.posix.dirname(basePath) : basePath
const parentPath = Path.posix.dirname(itemPath)
if (!parentPath || parentPath === '.' || parentPath === '/') return null
return parentPath.toLowerCase()
}
groupLibraryBooksByFolder(libraryItems) {
const groups = new Map()
for (const libraryItem of libraryItems) {
const folderKey = LibraryController.prototype.getLibraryItemFolderKey.call(this, libraryItem)
if (!folderKey) continue
if (!groups.has(folderKey)) {
groups.set(folderKey, {
label: folderKey,
libraryItems: []
})
}
groups.get(folderKey).libraryItems.push(libraryItem)
}
return [...groups.values()]
}
normalizeBookTitleForAIDedupe(title) {
if (!title || typeof title !== 'string') return null
return title
.toLowerCase()
.replace(/\([^)]*\)/g, ' ')
.replace(/\[[^\]]*]/g, ' ')
.replace(/\b(unabridged|abridged|audiobook|audio book)\b/g, ' ')
.replace(/[^a-z0-9]+/g, ' ')
.replace(/\s+/g, ' ')
.trim()
}
groupLibraryBooksForAIDedupe(libraryItems) {
const parent = new Map()
const find = (id) => {
if (parent.get(id) !== id) {
parent.set(id, find(parent.get(id)))
}
return parent.get(id)
}
const union = (a, b) => {
const rootA = find(a)
const rootB = find(b)
if (rootA !== rootB) parent.set(rootB, rootA)
}
libraryItems.forEach((libraryItem) => parent.set(libraryItem.id, libraryItem.id))
const candidateMaps = [new Map(), new Map(), new Map()]
libraryItems.forEach((libraryItem) => {
const metadata = libraryItem.media.oldMetadataToJSON()
const primaryAuthor = metadata.authors?.[0]?.name?.trim().toLowerCase() || null
const normalizedTitle = LibraryController.prototype.normalizeBookTitleForAIDedupe.call(this, metadata.title || '')
const isbn = metadata.isbn?.replace(/[-\s]/g, '').toLowerCase() || null
const asin = metadata.asin?.trim().toLowerCase() || null
const candidateKeys = []
if (primaryAuthor && normalizedTitle) candidateKeys.push([candidateMaps[0], `${primaryAuthor}::${normalizedTitle}`])
if (isbn) candidateKeys.push([candidateMaps[1], `isbn::${isbn}`])
if (asin) candidateKeys.push([candidateMaps[2], `asin::${asin}`])
candidateKeys.forEach(([candidateMap, key]) => {
if (!candidateMap.has(key)) candidateMap.set(key, [])
candidateMap.get(key).push(libraryItem)
})
})
candidateMaps.forEach((candidateMap) => {
candidateMap.forEach((groupItems) => {
if (groupItems.length < 2) return
for (let i = 1; i < groupItems.length; i++) {
union(groupItems[0].id, groupItems[i].id)
}
})
})
const grouped = new Map()
libraryItems.forEach((libraryItem) => {
const root = find(libraryItem.id)
if (!grouped.has(root)) grouped.set(root, [])
grouped.get(root).push(libraryItem)
})
return [...grouped.values()]
.filter((groupItems) => groupItems.length > 1)
.map((groupItems) => ({
label: groupItems.map((libraryItem) => libraryItem.media.title).join(' | '),
libraryItems: groupItems.sort((a, b) => a.media.title.localeCompare(b.media.title))
}))
}
getDeleteDependenciesForLibraryItem(libraryItem) {
const mediaItemIds = []
const authorIds = []
const seriesIds = []
mediaItemIds.push(libraryItem.media.id)
if (libraryItem.media.authors?.length) {
authorIds.push(...libraryItem.media.authors.map((author) => author.id))
}
if (libraryItem.media.series?.length) {
seriesIds.push(...libraryItem.media.series.map((series) => series.id))
}
return {
mediaItemIds,
authorIds,
seriesIds
}
}
/**
* POST: /api/libraries/:id/detect-series-with-ai
*
* @param {LibraryControllerRequest} req
* @param {Response} res
*/
async detectSeriesWithAI(req, res) {
if (!req.user.canUpdate) {
Logger.warn(`[LibraryController] User "${req.user.username}" attempted AI series detection without update permissions`)
return res.sendStatus(403)
}
if (req.library.mediaType !== 'book') {
return res.status(400).send('AI series detection is only available for book libraries')
}
if (!openAI.isConfigured) {
return res.status(400).send('OpenAI is not configured')
}
try {
const onlyMissingSeries = req.query.onlyMissing !== '0'
const libraryItems = await LibraryController.prototype.getLibraryBooksForAISeriesDetection.call(this, req.library.id)
const authorGroups = LibraryController.prototype.groupLibraryBooksByPrimaryAuthor.call(this, libraryItems).filter(({ libraryItems }) => {
if (libraryItems.length < 2) return false
if (!onlyMissingSeries) return true
return libraryItems.some((libraryItem) => !libraryItem.media.series.length)
})
const authorCoveredEligibleIds = new Set()
authorGroups.forEach((authorGroup) => {
authorGroup.libraryItems.forEach((libraryItem) => {
if (!onlyMissingSeries || !libraryItem.media.series.length) {
authorCoveredEligibleIds.add(libraryItem.id)
}
})
})
const folderGroups = LibraryController.prototype.groupLibraryBooksByFolder
.call(this, libraryItems)
.filter(({ libraryItems }) => {
if (libraryItems.length < 2) return false
const eligibleItems = onlyMissingSeries ? libraryItems.filter((libraryItem) => !libraryItem.media.series.length) : libraryItems
if (!eligibleItems.length) return false
return eligibleItems.some((libraryItem) => !authorCoveredEligibleIds.has(libraryItem.id))
})
const evaluationGroups = [
...authorGroups.map((group) => ({ type: 'author', label: group.authorName, libraryItems: group.libraryItems })),
...folderGroups.map((group) => ({ type: 'folder', label: group.label, libraryItems: group.libraryItems }))
]
let groupsProcessed = 0
let booksConsidered = 0
let booksUpdated = 0
for (const evaluationGroup of evaluationGroups) {
const eligibleLibraryItems = onlyMissingSeries ? evaluationGroup.libraryItems.filter((libraryItem) => !libraryItem.media.series.length) : evaluationGroup.libraryItems
if (!eligibleLibraryItems.length) continue
if (evaluationGroup.type === 'folder') {
const remainingEligibleItems = eligibleLibraryItems.filter((libraryItem) => !authorCoveredEligibleIds.has(libraryItem.id))
if (!remainingEligibleItems.length) continue
Logger.info(
`[LibraryController] AI series detection evaluating folder group "${evaluationGroup.label}" with ${evaluationGroup.libraryItems.length} books (${remainingEligibleItems.length} eligible for update)`
)
evaluationGroup.eligibleLibraryItems = remainingEligibleItems
} else {
Logger.info(
`[LibraryController] AI series detection evaluating author "${evaluationGroup.label}" with ${evaluationGroup.libraryItems.length} books (${eligibleLibraryItems.length} eligible for update)`
)
evaluationGroup.eligibleLibraryItems = eligibleLibraryItems
}
const assignments = await openAI.detectSeriesAssignments(evaluationGroup.label, evaluationGroup.libraryItems, evaluationGroup.type)
const assignmentsByLibraryItemId = new Map(assignments.map((assignment) => [assignment.id, assignment]))
groupsProcessed++
for (const libraryItem of evaluationGroup.eligibleLibraryItems) {
booksConsidered++
const assignment = assignmentsByLibraryItemId.get(libraryItem.id)
if (!assignment?.seriesName || !assignment.sequence) {
Logger.info(`[LibraryController] AI series detection skipped "${libraryItem.media.title}" (${libraryItem.id})`)
continue
}
Logger.info(
`[LibraryController] AI series detection applying "${libraryItem.media.title}" (${libraryItem.id}) -> series "${assignment.seriesName}" sequence "${assignment.sequence}"`
)
const existingSeries = libraryItem.media.series.find((series) => series.name.toLowerCase() === assignment.seriesName.toLowerCase())
const seriesPayload = libraryItem.media.series.map((series) => ({
id: series.id,
name: series.name,
sequence: series.bookSeries?.sequence || null
}))
if (existingSeries) {
const existingSeriesIndex = seriesPayload.findIndex((series) => series.id === existingSeries.id)
if (existingSeriesIndex >= 0) {
seriesPayload[existingSeriesIndex].sequence = assignment.sequence
}
} else {
seriesPayload.push({
name: assignment.seriesName,
sequence: assignment.sequence
})
}
const seriesUpdate = await libraryItem.media.updateSeriesFromRequest(seriesPayload, libraryItem.libraryId)
if (!seriesUpdate?.hasUpdates) {
Logger.info(`[LibraryController] AI series detection found no metadata changes for "${libraryItem.media.title}" (${libraryItem.id})`)
continue
}
if (seriesUpdate.seriesAdded?.length) {
seriesUpdate.seriesAdded.forEach((series) => {
Database.addSeriesToFilterData(req.library.id, series.name, series.id)
})
}
libraryItem.changed('updatedAt', true)
await libraryItem.save()
await libraryItem.saveMetadataFile()
booksUpdated++
SocketAuthority.libraryItemEmitter('item_updated', libraryItem)
}
}
Logger.info(
`[LibraryController] AI series detection completed for library "${req.library.name}" - groupsProcessed=${groupsProcessed}, booksConsidered=${booksConsidered}, booksUpdated=${booksUpdated}`
)
res.json({
groupsProcessed,
booksConsidered,
updated: booksUpdated
})
} catch (error) {
Logger.error(`[LibraryController] Failed AI series detection for library "${req.library.name}"`, error)
res.status(500).send(error.message || 'Failed to detect series with AI')
}
}
/**
* POST: /api/libraries/:id/dedupe-books-with-ai
*
* @this {import('../routers/ApiRouter')}
*
* @param {LibraryControllerRequest} req
* @param {Response} res
*/
async dedupeBooksWithAI(req, res) {
if (!req.user.canUpdate) {
Logger.warn(`[LibraryController] User "${req.user.username}" attempted AI dedupe without update permissions`)
return res.sendStatus(403)
}
if (req.library.mediaType !== 'book') {
return res.status(400).send('AI book dedupe is only available for book libraries')
}
if (!openAI.isConfigured) {
return res.status(400).send('OpenAI is not configured')
}
const hardDelete = req.query.hard !== '0'
try {
const libraryItems = await LibraryController.prototype.getLibraryBooksForAISeriesDetection.call(this, req.library.id)
const candidateGroups = LibraryController.prototype.groupLibraryBooksForAIDedupe.call(this, libraryItems)
let groupsProcessed = 0
let duplicatesRemoved = 0
const removedIds = new Set()
const authorIdsToCheck = new Set()
const seriesIdsToCheck = new Set()
for (const candidateGroup of candidateGroups) {
const activeLibraryItems = candidateGroup.libraryItems.filter((libraryItem) => !removedIds.has(libraryItem.id))
if (activeLibraryItems.length < 2) continue
Logger.info(`[LibraryController] AI dedupe evaluating candidate group "${candidateGroup.label}" with ${activeLibraryItems.length} books`)
const decisions = await openAI.detectDuplicateBooks(activeLibraryItems)
groupsProcessed++
for (const decision of decisions) {
for (const duplicateId of decision.duplicateIds) {
if (removedIds.has(duplicateId) || duplicateId === decision.keepId) continue
const duplicateItem = activeLibraryItems.find((libraryItem) => libraryItem.id === duplicateId)
if (!duplicateItem) continue
Logger.info(
`[LibraryController] AI dedupe removing duplicate "${duplicateItem.media.title}" (${duplicateItem.id}) keeping "${decision.keepId}" reason="${decision.reason || ''}"`
)
const deleteDependencies = LibraryController.prototype.getDeleteDependenciesForLibraryItem.call(this, duplicateItem)
await this.handleDeleteLibraryItem(duplicateItem.id, deleteDependencies.mediaItemIds, req.library.id)
if (hardDelete) {
await fs.remove(duplicateItem.path).catch((error) => {
Logger.error(`[LibraryController] Failed to hard-delete duplicate item path "${duplicateItem.path}"`, error)
})
}
deleteDependencies.authorIds.forEach((authorId) => authorIdsToCheck.add(authorId))
deleteDependencies.seriesIds.forEach((seriesId) => seriesIdsToCheck.add(seriesId))
removedIds.add(duplicateItem.id)
duplicatesRemoved++
}
}
}
await this.checkRemoveAuthorsWithNoBooks([...authorIdsToCheck])
await this.checkRemoveEmptySeries([...seriesIdsToCheck])
await Database.resetLibraryIssuesFilterData(req.library.id)
Logger.info(
`[LibraryController] AI book dedupe completed for library "${req.library.name}" - groupsProcessed=${groupsProcessed}, duplicatesRemoved=${duplicatesRemoved}, hardDelete=${hardDelete}`
)
res.json({
groupsProcessed,
duplicatesRemoved,
hardDelete
})
} catch (error) {
Logger.error(`[LibraryController] Failed AI dedupe for library "${req.library.name}"`, error)
res.status(500).send(error.message || 'Failed to dedupe books with AI')
}
}
/**
*
* @param {RequestWithUser} req

View file

@ -145,6 +145,36 @@ class MiscController {
if (settingsUpdate.allowedOrigins && !Array.isArray(settingsUpdate.allowedOrigins)) {
return res.status(400).send('allowedOrigins must be an array')
}
if (settingsUpdate.openAIApiKey !== undefined && settingsUpdate.openAIApiKey !== null && typeof settingsUpdate.openAIApiKey !== 'string') {
return res.status(400).send('openAIApiKey must be a string or null')
}
if (settingsUpdate.openAIBaseURL !== undefined && typeof settingsUpdate.openAIBaseURL !== 'string') {
return res.status(400).send('openAIBaseURL must be a string')
}
if (settingsUpdate.openAIModel !== undefined && typeof settingsUpdate.openAIModel !== 'string') {
return res.status(400).send('openAIModel must be a string')
}
if (typeof settingsUpdate.openAIApiKey === 'string') {
settingsUpdate.openAIApiKey = settingsUpdate.openAIApiKey.trim() || null
}
if (typeof settingsUpdate.openAIBaseURL === 'string') {
settingsUpdate.openAIBaseURL = settingsUpdate.openAIBaseURL.trim().replace(/\/+$/, '')
if (!settingsUpdate.openAIBaseURL) {
return res.status(400).send('openAIBaseURL is required')
}
try {
new URL(settingsUpdate.openAIBaseURL)
} catch {
return res.status(400).send('openAIBaseURL must be a valid URL')
}
}
if (typeof settingsUpdate.openAIModel === 'string') {
settingsUpdate.openAIModel = settingsUpdate.openAIModel.trim()
if (!settingsUpdate.openAIModel) {
return res.status(400).send('openAIModel is required')
}
}
const madeUpdates = Database.serverSettings.update(settingsUpdate)
if (madeUpdates) {

View file

@ -2,11 +2,14 @@ const { Request, Response, NextFunction } = require('express')
const Logger = require('../Logger')
const SocketAuthority = require('../SocketAuthority')
const Database = require('../Database')
const OpenAI = require('../providers/OpenAI')
const RssFeedManager = require('../managers/RssFeedManager')
const libraryItemsBookFilters = require('../utils/queries/libraryItemsBookFilters')
const openAI = new OpenAI()
/**
* @typedef RequestUserObject
* @property {import('../models/User')} user
@ -86,6 +89,68 @@ class SeriesController {
res.json(req.series.toOldJSON())
}
/**
* POST: /api/series/:id/organize-story-order
*
* @param {SeriesControllerRequest} req
* @param {Response} res
*/
async organizeStoryOrder(req, res) {
if (!openAI.isConfigured) {
return res.status(400).send('OpenAI is not configured')
}
if (!req.libraryItemsInSeries.length) {
return res.status(400).send('No books found in this series')
}
try {
const seriesOrder = await openAI.getSeriesOrder(req.series, req.libraryItemsInSeries)
const sequenceByLibraryItemId = new Map(seriesOrder.map((book) => [book.id, book.sequence]))
const updatedItems = []
Logger.info(`[SeriesController] AI story-order evaluation returned ${seriesOrder.length} books for series "${req.series.name}"`)
for (const libraryItem of req.libraryItemsInSeries) {
const nextSequence = sequenceByLibraryItemId.get(libraryItem.id)
if (!nextSequence) continue
Logger.info(`[SeriesController] AI story-order applying "${libraryItem.media.title}" (${libraryItem.id}) -> sequence "${nextSequence}" in series "${req.series.name}"`)
const seriesPayload = libraryItem.media.series.map((series) => ({
id: series.id,
name: series.name,
sequence: series.id === req.series.id ? nextSequence : series.bookSeries?.sequence || null
}))
const seriesUpdate = await libraryItem.media.updateSeriesFromRequest(seriesPayload, libraryItem.libraryId)
if (!seriesUpdate?.hasUpdates) {
Logger.info(`[SeriesController] AI story-order found no change for "${libraryItem.media.title}" (${libraryItem.id})`)
continue
}
libraryItem.changed('updatedAt', true)
await libraryItem.save()
await libraryItem.saveMetadataFile()
updatedItems.push(libraryItem)
SocketAuthority.libraryItemEmitter('item_updated', libraryItem)
}
if (updatedItems.length) {
SocketAuthority.emitter('series_updated', req.series.toOldJSON())
}
Logger.info(`[SeriesController] AI story-order completed for series "${req.series.name}" - updated=${updatedItems.length}, total=${req.libraryItemsInSeries.length}`)
res.json({
updated: updatedItems.length,
total: req.libraryItemsInSeries.length
})
} catch (error) {
Logger.error(`[SeriesController] Failed to organize story order for "${req.series.name}"`, error)
res.status(500).send(error.message || 'Failed to organize story order')
}
}
/**
*
* @param {RequestWithUser} req

View file

@ -11,6 +11,7 @@ const Logger = require('../Logger')
* @property {boolean} audiobooksOnly
* @property {boolean} hideSingleBookSeries Do not show series that only have 1 book
* @property {boolean} onlyShowLaterBooksInContinueSeries Skip showing books that are earlier than the max sequence read
* @property {boolean} openAIDirectoryGrouping Allow OpenAI to infer library-item grouping from poor directory structures
* @property {string[]} metadataPrecedence
* @property {number} markAsFinishedTimeRemaining Time remaining in seconds to mark as finished. (defaults to 10s)
* @property {number} markAsFinishedPercentComplete Percent complete to mark as finished (0-100). If this is set it will be used over markAsFinishedTimeRemaining.
@ -74,6 +75,7 @@ class Library extends Model {
epubsAllowScriptedContent: false,
hideSingleBookSeries: false,
onlyShowLaterBooksInContinueSeries: false,
openAIDirectoryGrouping: false,
metadataPrecedence: this.defaultMetadataPrecedence,
markAsFinishedPercentComplete: null,
markAsFinishedTimeRemaining: 10

View file

@ -84,6 +84,11 @@ class ServerSettings {
this.authOpenIDAdvancedPermsClaim = ''
this.authOpenIDSubfolderForRedirectURLs = undefined
// OpenAI
this.openAIApiKey = null
this.openAIBaseURL = 'https://api.openai.com/v1'
this.openAIModel = 'gpt-5.4-mini'
if (settings) {
this.construct(settings)
}
@ -147,6 +152,9 @@ class ServerSettings {
this.authOpenIDGroupClaim = settings.authOpenIDGroupClaim || ''
this.authOpenIDAdvancedPermsClaim = settings.authOpenIDAdvancedPermsClaim || ''
this.authOpenIDSubfolderForRedirectURLs = settings.authOpenIDSubfolderForRedirectURLs
this.openAIApiKey = settings.openAIApiKey || null
this.openAIBaseURL = settings.openAIBaseURL || 'https://api.openai.com/v1'
this.openAIModel = settings.openAIModel || 'gpt-5.4-mini'
if (!Array.isArray(this.authActiveAuthMethods)) {
this.authActiveAuthMethods = ['local']
@ -256,7 +264,10 @@ class ServerSettings {
authOpenIDMobileRedirectURIs: this.authOpenIDMobileRedirectURIs, // Do not return to client
authOpenIDGroupClaim: this.authOpenIDGroupClaim, // Do not return to client
authOpenIDAdvancedPermsClaim: this.authOpenIDAdvancedPermsClaim, // Do not return to client
authOpenIDSubfolderForRedirectURLs: this.authOpenIDSubfolderForRedirectURLs
authOpenIDSubfolderForRedirectURLs: this.authOpenIDSubfolderForRedirectURLs,
openAIApiKey: this.openAIApiKey, // Do not return to client
openAIBaseURL: this.openAIBaseURL,
openAIModel: this.openAIModel
}
}
@ -268,9 +279,36 @@ class ServerSettings {
delete json.authOpenIDMobileRedirectURIs
delete json.authOpenIDGroupClaim
delete json.authOpenIDAdvancedPermsClaim
delete json.openAIApiKey
json.openAIConfigured = this.openAIConfigured
json.openAIConfigurationSource = this.openAIConfigurationSource
json.openAIBaseURL = this.openAIResolvedBaseURL
json.openAIModel = this.openAIResolvedModel
return json
}
get openAIResolvedApiKey() {
return process.env.OPENAI_API_KEY || this.openAIApiKey || null
}
get openAIResolvedBaseURL() {
return process.env.OPENAI_BASE_URL || this.openAIBaseURL || 'https://api.openai.com/v1'
}
get openAIResolvedModel() {
return process.env.OPENAI_MODEL || this.openAIModel || 'gpt-5.4-mini'
}
get openAIConfigured() {
return !!this.openAIResolvedApiKey
}
get openAIConfigurationSource() {
if (process.env.OPENAI_API_KEY) return 'environment'
if (this.openAIApiKey) return 'settings'
return null
}
get supportedAuthMethods() {
return ['local', 'openid']
}

887
server/providers/OpenAI.js Normal file
View file

@ -0,0 +1,887 @@
const Path = require('path')
const axios = require('axios')
const Database = require('../Database')
const Logger = require('../Logger')
const htmlSanitizer = require('../utils/htmlSanitizer')
const DEFAULT_BASE_URL = 'https://api.openai.com/v1'
const DEFAULT_MODEL = 'gpt-5.4-mini'
const RESPONSE_TIMEOUT_MS = 60000
const SEQUENCE_REGEX = /^(?:0|[1-9]\d{0,3})(?:\.\d{1,2})?$/
class OpenAI {
summarizeBookForLog(book) {
return JSON.stringify({
id: book.id,
title: book.title,
authors: book.authors,
fullPath: book.fullPath,
relPath: book.relPath,
existingSeries: book.existingSeries,
currentSequence: book.currentSequence
})
}
summarizeAssignmentForLog(assignment) {
return JSON.stringify({
id: assignment.id,
seriesName: assignment.seriesName || null,
sequence: assignment.sequence || null,
reason: assignment.reason || ''
})
}
summarizeScanMetadataForLog(metadata) {
return JSON.stringify({
title: metadata.title || null,
authors: metadata.authors || [],
seriesName: metadata.seriesName || null,
sequence: metadata.sequence || null,
publishedYear: metadata.publishedYear || null,
reason: metadata.reason || ''
})
}
summarizeDirectoryGroupingForLog(grouping) {
return JSON.stringify({
path: grouping.path,
groupId: grouping.groupId,
reason: grouping.reason || ''
})
}
summarizeDuplicateDecisionForLog(decision) {
return JSON.stringify({
keepId: decision.keepId,
duplicateIds: decision.duplicateIds,
reason: decision.reason || ''
})
}
normalizePathForPrompt(filePath) {
if (!filePath || typeof filePath !== 'string') return null
return filePath.replace(/\\/g, '/')
}
getFolderContext(libraryItem) {
const absolutePath = this.normalizePathForPrompt(libraryItem.path)
const relativePath = this.normalizePathForPrompt(libraryItem.relPath)
const basePath = relativePath || absolutePath
if (!basePath && !absolutePath) {
return {
fullPath: null,
relPath: null,
itemFolderName: null,
parentFolderName: null,
folderHierarchy: [],
fullPathHierarchy: []
}
}
const itemPath = libraryItem.isFile ? Path.posix.dirname(basePath) : basePath
const absoluteItemPath = absolutePath ? (libraryItem.isFile ? Path.posix.dirname(absolutePath) : absolutePath) : null
const folderHierarchy = itemPath
.split('/')
.map((segment) => segment.trim())
.filter(Boolean)
const fullPathHierarchy = absoluteItemPath
? absoluteItemPath
.split('/')
.map((segment) => segment.trim())
.filter(Boolean)
: []
const itemFolderName = folderHierarchy.length ? folderHierarchy[folderHierarchy.length - 1] : null
const parentFolderName = folderHierarchy.length > 1 ? folderHierarchy[folderHierarchy.length - 2] : null
return {
fullPath: absolutePath,
relPath: relativePath || basePath,
itemFolderName,
parentFolderName,
folderHierarchy,
fullPathHierarchy
}
}
normalizeBaseURL(url) {
return (url || DEFAULT_BASE_URL).replace(/\/+$/, '')
}
get apiKey() {
return Database.serverSettings?.openAIResolvedApiKey || null
}
get baseURL() {
return this.normalizeBaseURL(Database.serverSettings?.openAIResolvedBaseURL || DEFAULT_BASE_URL)
}
get model() {
return Database.serverSettings?.openAIResolvedModel || DEFAULT_MODEL
}
get isConfigured() {
return !!this.apiKey
}
cleanDescription(description) {
if (!description || typeof description !== 'string') return null
const plain = htmlSanitizer.stripAllTags(description).replace(/\s+/g, ' ').trim()
if (!plain) return null
return plain.slice(0, 600)
}
buildBookPayload(libraryItem) {
const book = libraryItem.media
const metadata = book.oldMetadataToJSON()
const folderContext = this.getFolderContext(libraryItem)
return {
id: libraryItem.id,
title: metadata.title || null,
subtitle: metadata.subtitle || null,
publishedYear: metadata.publishedYear || null,
authors: (metadata.authors || []).map((author) => author.name),
description: this.cleanDescription(metadata.description),
fullPath: folderContext.fullPath,
relPath: folderContext.relPath,
itemFolderName: folderContext.itemFolderName,
parentFolderName: folderContext.parentFolderName,
folderHierarchy: folderContext.folderHierarchy,
fullPathHierarchy: folderContext.fullPathHierarchy,
existingSeries: (metadata.series || []).map((series) => ({
name: series.name,
sequence: series.sequence || null
}))
}
}
extractTextFromResponse(data) {
if (typeof data?.output_text === 'string' && data.output_text.trim()) {
return data.output_text.trim()
}
const texts = []
for (const outputItem of data?.output || []) {
if (typeof outputItem?.text === 'string' && outputItem.text.trim()) {
texts.push(outputItem.text.trim())
}
for (const contentItem of outputItem?.content || []) {
if (typeof contentItem?.text === 'string' && contentItem.text.trim()) {
texts.push(contentItem.text.trim())
} else if (typeof contentItem?.output_text === 'string' && contentItem.output_text.trim()) {
texts.push(contentItem.output_text.trim())
}
}
}
return texts.join('\n').trim()
}
parseJsonResponse(text) {
if (!text || typeof text !== 'string') {
throw new Error('OpenAI returned an empty response')
}
let cleanedText = text.trim()
const fencedMatch = cleanedText.match(/^```(?:json)?\s*([\s\S]*?)\s*```$/i)
if (fencedMatch) {
cleanedText = fencedMatch[1].trim()
}
try {
return JSON.parse(cleanedText)
} catch (error) {
const start = cleanedText.indexOf('{')
const end = cleanedText.lastIndexOf('}')
if (start >= 0 && end > start) {
return JSON.parse(cleanedText.slice(start, end + 1))
}
throw error
}
}
normalizeSeriesName(value) {
if (!value || typeof value !== 'string') return null
const seriesName = value.trim()
return seriesName || null
}
normalizeSequence(value) {
if (value === null || value === undefined) return null
if (typeof value === 'number') value = String(value)
if (typeof value !== 'string') return null
const sequence = value.trim()
if (!SEQUENCE_REGEX.test(sequence)) return null
return sequence
}
normalizeOptionalString(value, maxLength = 300) {
if (value === null || value === undefined) return null
if (typeof value !== 'string') return null
const normalized = value.replace(/\s+/g, ' ').trim()
if (!normalized) return null
return normalized.slice(0, maxLength)
}
normalizeStringArray(value, maxItems = 10, maxLength = 120) {
if (!Array.isArray(value)) return []
const deduped = []
const seen = new Set()
for (const item of value) {
const normalized = this.normalizeOptionalString(item, maxLength)
if (!normalized) continue
const key = normalized.toLowerCase()
if (seen.has(key)) continue
seen.add(key)
deduped.push(normalized)
if (deduped.length >= maxItems) break
}
return deduped
}
normalizePublishedYear(value) {
if (value === null || value === undefined) return null
if (typeof value === 'number' && Number.isInteger(value)) value = String(value)
if (typeof value !== 'string') return null
const normalized = value.trim()
if (!/^\d{4}$/.test(normalized)) return null
return normalized
}
normalizeIsbn(value) {
const normalized = this.normalizeOptionalString(value, 20)
if (!normalized) return null
const compact = normalized.replace(/[-\s]/g, '')
if (!/^(?:\d{10}|\d{13}|[0-9X]{10})$/i.test(compact)) return null
return compact
}
normalizeAsin(value) {
const normalized = this.normalizeOptionalString(value, 10)
if (!normalized) return null
return /^[A-Z0-9]{10}$/i.test(normalized) ? normalized.toUpperCase() : null
}
validateScanMetadataPayload(payload) {
const book = payload?.book && typeof payload.book === 'object' ? payload.book : payload
if (!book || typeof book !== 'object' || Array.isArray(book)) {
throw new Error('OpenAI returned invalid scan metadata payload')
}
const seriesName = this.normalizeSeriesName(book.seriesName)
return {
title: this.normalizeOptionalString(book.title),
subtitle: this.normalizeOptionalString(book.subtitle),
publishedYear: this.normalizePublishedYear(book.publishedYear),
publisher: this.normalizeOptionalString(book.publisher),
isbn: this.normalizeIsbn(book.isbn),
asin: this.normalizeAsin(book.asin),
language: this.normalizeOptionalString(book.language, 40),
authors: this.normalizeStringArray(book.authors),
narrators: this.normalizeStringArray(book.narrators),
seriesName,
sequence: seriesName ? this.normalizeSequence(book.sequence) : null,
reason: this.normalizeOptionalString(book.reason, 600) || ''
}
}
validateDirectoryGroupingPayload(payload, mediaFiles) {
const resultFiles = payload?.files
if (!Array.isArray(resultFiles)) {
throw new Error('OpenAI returned invalid directory-grouping payload')
}
const expectedPaths = new Set(mediaFiles.map((file) => file.path))
const resultByPath = new Map()
resultFiles.forEach((file) => {
if (!expectedPaths.has(file?.path)) {
Logger.warn(`[OpenAI] Ignoring unknown media path "${file?.path}" in directory-grouping response`)
return
}
if (resultByPath.has(file.path)) {
Logger.warn(`[OpenAI] Ignoring duplicate media path "${file.path}" in directory-grouping response`)
return
}
resultByPath.set(file.path, file)
})
return mediaFiles.map((file) => {
const result = resultByPath.get(file.path)
const groupId = this.normalizeOptionalString(result?.groupId, 120) || file.path
const reason = this.normalizeOptionalString(result?.reason, 600) || (result ? '' : 'OpenAI omitted this media file; kept it as its own item')
if (!result) {
Logger.warn(`[OpenAI] Missing directory-grouping result for media path "${file.path}" - keeping it separate`)
}
return {
path: file.path,
groupId,
reason
}
})
}
validateDuplicateBooksPayload(payload, books) {
const resultGroups = Array.isArray(payload?.groups) ? payload.groups : []
const expectedIds = new Set(books.map((book) => book.id))
const consumedIds = new Set()
const validated = []
resultGroups.forEach((group) => {
const keepId = this.normalizeOptionalString(group?.keepId, 120)
if (!keepId || !expectedIds.has(keepId)) {
Logger.warn(`[OpenAI] Ignoring duplicate-books group with invalid keepId "${group?.keepId}"`)
return
}
if (consumedIds.has(keepId)) {
Logger.warn(`[OpenAI] Ignoring duplicate-books group because keepId "${keepId}" was already used`)
return
}
const duplicateIds = Array.isArray(group?.duplicateIds)
? group.duplicateIds
.map((duplicateId) => this.normalizeOptionalString(duplicateId, 120))
.filter((duplicateId) => duplicateId && expectedIds.has(duplicateId) && duplicateId !== keepId)
: []
const dedupedDuplicateIds = []
const seenDuplicateIds = new Set()
duplicateIds.forEach((duplicateId) => {
if (seenDuplicateIds.has(duplicateId) || consumedIds.has(duplicateId)) return
seenDuplicateIds.add(duplicateId)
dedupedDuplicateIds.push(duplicateId)
})
if (!dedupedDuplicateIds.length) return
consumedIds.add(keepId)
dedupedDuplicateIds.forEach((duplicateId) => consumedIds.add(duplicateId))
validated.push({
keepId,
duplicateIds: dedupedDuplicateIds,
reason: this.normalizeOptionalString(group?.reason, 600) || ''
})
})
return validated
}
validateBookIds(resultBooks, books) {
if (!Array.isArray(resultBooks) || resultBooks.length !== books.length) {
throw new Error('OpenAI returned an invalid number of books')
}
const expectedIds = new Set(books.map((book) => book.id))
const seenIds = new Set()
resultBooks.forEach((book) => {
if (!expectedIds.has(book?.id)) {
throw new Error(`OpenAI returned an unknown book id "${book?.id}"`)
}
if (seenIds.has(book.id)) {
throw new Error(`OpenAI returned duplicate book id "${book.id}"`)
}
seenIds.add(book.id)
})
}
normalizeDetectionResultBooks(resultBooks, books) {
if (!Array.isArray(resultBooks)) {
throw new Error('OpenAI returned an invalid books payload')
}
const expectedIds = new Set(books.map((book) => book.id))
const resultBooksById = new Map()
resultBooks.forEach((book) => {
if (!expectedIds.has(book?.id)) {
Logger.warn(`[OpenAI] Ignoring unknown book id "${book?.id}" in series-detection response`)
return
}
if (resultBooksById.has(book.id)) {
Logger.warn(`[OpenAI] Ignoring duplicate book id "${book.id}" in series-detection response`)
return
}
resultBooksById.set(book.id, book)
})
return books.map((book) => {
if (resultBooksById.has(book.id)) {
return resultBooksById.get(book.id)
}
Logger.warn(`[OpenAI] Missing series-detection result for book "${book.id}" - skipping assignment`)
return {
id: book.id,
seriesName: null,
sequence: null,
reason: 'Skipped because OpenAI omitted this book from the response'
}
})
}
validateSeriesOrderPayload(payload, books) {
const resultBooks = payload?.books
this.validateBookIds(resultBooks, books)
const sequences = new Set()
return resultBooks
.map((book) => {
const sequence = this.normalizeSequence(book.sequence)
if (!sequence) {
throw new Error(`OpenAI returned an invalid sequence for "${book.id}"`)
}
if (sequences.has(sequence)) {
throw new Error(`OpenAI returned a duplicate sequence "${sequence}"`)
}
sequences.add(sequence)
return {
id: book.id,
sequence,
reason: typeof book.reason === 'string' ? book.reason.trim() : ''
}
})
.sort((a, b) => Number(a.sequence) - Number(b.sequence))
}
validateSeriesDetectionPayload(payload, books) {
const resultBooks = this.normalizeDetectionResultBooks(payload?.books, books)
const seriesSequences = new Map()
return resultBooks.map((book) => {
const seriesName = this.normalizeSeriesName(book.seriesName)
const sequence = this.normalizeSequence(book.sequence)
const reason = typeof book.reason === 'string' ? book.reason.trim() : ''
if (seriesName && !sequence) {
Logger.warn(`[OpenAI] Series "${seriesName}" for book "${book.id}" did not include a valid sequence - skipping assignment`)
return {
id: book.id,
seriesName: null,
sequence: null,
reason: reason ? `${reason} (skipped due to missing or invalid sequence)` : 'Skipped due to missing or invalid sequence'
}
}
if (!seriesName && sequence) {
Logger.warn(`[OpenAI] Sequence "${sequence}" for book "${book.id}" did not include a series name - skipping assignment`)
return {
id: book.id,
seriesName: null,
sequence: null,
reason: reason ? `${reason} (skipped due to missing series name)` : 'Skipped due to missing series name'
}
}
if (seriesName && sequence) {
const key = seriesName.toLowerCase()
if (!seriesSequences.has(key)) {
seriesSequences.set(key, new Set())
}
if (seriesSequences.get(key).has(sequence)) {
Logger.warn(`[OpenAI] Duplicate inferred sequence "${sequence}" inside "${seriesName}" for book "${book.id}" - skipping assignment`)
return {
id: book.id,
seriesName: null,
sequence: null,
reason: reason ? `${reason} (skipped due to duplicate inferred sequence)` : 'Skipped due to duplicate inferred sequence'
}
}
seriesSequences.get(key).add(sequence)
}
return {
id: book.id,
seriesName,
sequence,
reason
}
})
}
async createResponse(prompt) {
if (!this.isConfigured) {
throw new Error('OpenAI API key is not configured')
}
const url = `${this.baseURL}/responses`
Logger.debug(`[OpenAI] Requesting ${url} with model "${this.model}"`)
const response = await axios
.post(
url,
{
model: this.model,
input: prompt
},
{
timeout: RESPONSE_TIMEOUT_MS,
headers: {
Authorization: `Bearer ${this.apiKey}`,
'Content-Type': 'application/json'
}
}
)
.catch((error) => {
const status = error.response?.status
const message = error.response?.data?.error?.message || error.message
Logger.error(`[OpenAI] Responses API request failed (${status || 'no-status'})`, message)
if (status === 401) {
throw new Error('OpenAI rejected the API key')
} else if (status === 429) {
throw new Error('OpenAI rate limit reached')
} else if (status) {
throw new Error(`OpenAI request failed with status ${status}`)
}
throw new Error(`OpenAI request failed: ${message}`)
})
const text = this.extractTextFromResponse(response.data)
const parsed = this.parseJsonResponse(text)
Logger.debug(`[OpenAI] Parsed response payload: ${JSON.stringify(parsed)}`)
return parsed
}
async getSeriesOrder(series, libraryItems) {
const books = libraryItems.map((libraryItem) => {
const book = this.buildBookPayload(libraryItem)
const currentSeries = book.existingSeries.find((existingSeries) => existingSeries.name.toLowerCase() === series.name.toLowerCase())
return {
...book,
currentSequence: currentSeries?.sequence || null
}
})
Logger.info(`[OpenAI] Evaluating story order for series "${series.name}" with ${books.length} books`)
books.forEach((book) => {
Logger.info(`[OpenAI] Story-order candidate ${this.summarizeBookForLog(book)}`)
})
const prompt = `You organize audiobooks into the correct in-universe story order for a single series.
Return only valid JSON in this shape:
{
"books": [
{
"id": "library-item-id",
"sequence": "1",
"reason": "brief reason"
}
]
}
Rules:
- Include every provided book exactly once.
- Use numeric string sequences only.
- Sequences must be unique and reflect story order, not shelf order.
- Prefer existing sequence values when they already look plausible.
- If evidence is weak, preserve the current sequence when present; otherwise fall back to publishedYear, then title.
- Do not invent books or series.
Series:
${JSON.stringify({ id: series.id, name: series.name, description: this.cleanDescription(series.description) }, null, 2)}
Books:
${JSON.stringify(books, null, 2)}`
const payload = await this.createResponse(prompt)
const validated = this.validateSeriesOrderPayload(payload, books)
validated.forEach((book) => {
Logger.info(`[OpenAI] Story-order result ${JSON.stringify({ id: book.id, sequence: book.sequence, reason: book.reason || '' })}`)
})
return validated
}
async detectSeriesAssignments(contextLabel, libraryItems, contextType = 'author') {
const books = libraryItems.map((libraryItem) => this.buildBookPayload(libraryItem))
Logger.info(`[OpenAI] Detecting series assignments for ${contextType} "${contextLabel}" with ${books.length} books`)
books.forEach((book) => {
Logger.info(`[OpenAI] Series-detection candidate ${this.summarizeBookForLog(book)}`)
})
const contextDescription =
contextType === 'folder'
? 'These books were grouped because they share the same folder context. Folder structure may be more reliable than author metadata for this group.'
: 'These books were grouped by primary author.'
const contextHeading = contextType === 'folder' ? 'Grouping context' : 'Primary author'
const prompt = `You detect audiobook series membership for a group of related books.
Return only valid JSON in this shape:
{
"books": [
{
"id": "library-item-id",
"seriesName": "Series Name or null",
"sequence": "1 or null",
"reason": "brief reason"
}
]
}
Rules:
- Include every provided book exactly once.
- Use "seriesName": null and "sequence": null for standalones or uncertain books.
- When assigning a series, use a numeric string sequence.
- Reuse an existing series name when it already appears in the provided data.
- Full absolute path and relative path are both available and should be used as evidence.
- Books sharing the same parent folder or series-like folder names are strong evidence they belong together.
- Use folder hierarchy as evidence alongside title, subtitle, description, and existing metadata.
- Do not invent series when the evidence is weak.
- Existing series metadata is trusted context and should usually be preserved.
${contextHeading}:
${JSON.stringify(contextLabel)}
Context note:
${contextDescription}
Books:
${JSON.stringify(books, null, 2)}`
const payload = await this.createResponse(prompt)
const validated = this.validateSeriesDetectionPayload(payload, books)
validated.forEach((assignment) => {
Logger.info(`[OpenAI] Series-detection result ${this.summarizeAssignmentForLog(assignment)}`)
})
return validated
}
async inferBookMetadataFromScan(libraryItemData, audioFiles = [], ebookFileScanData = null) {
if (!this.isConfigured) {
throw new Error('OpenAI API key is not configured')
}
const folderContext = this.getFolderContext(libraryItemData)
const audioFileCandidates = (audioFiles || []).slice(0, 25).map((audioFile) => ({
relPath: this.normalizePathForPrompt(audioFile.metadata?.relPath),
filename: audioFile.metadata?.filename || null,
duration: audioFile.duration || null,
trackNumber: audioFile.trackNumFromMeta || audioFile.metaTags?.trackNumber || null,
discNumber: audioFile.discNumFromMeta || audioFile.metaTags?.discNumber || null,
metaTags: {
tagTitle: audioFile.metaTags?.tagTitle || null,
tagAlbum: audioFile.metaTags?.tagAlbum || null,
tagArtist: audioFile.metaTags?.tagArtist || null,
tagAlbumArtist: audioFile.metaTags?.tagAlbumArtist || null,
tagSeries: audioFile.metaTags?.tagSeries || null,
tagSeriesPart: audioFile.metaTags?.tagSeriesPart || null,
tagSubtitle: audioFile.metaTags?.tagSubtitle || null,
tagDate: audioFile.metaTags?.tagDate || null,
tagASIN: audioFile.metaTags?.tagASIN || null
}
}))
const ebookMetadata = ebookFileScanData?.metadata
? {
title: ebookFileScanData.metadata.title || null,
subtitle: ebookFileScanData.metadata.subtitle || null,
authors: ebookFileScanData.metadata.authors || [],
narrators: ebookFileScanData.metadata.narrators || [],
series: ebookFileScanData.metadata.series || [],
publishedYear: ebookFileScanData.metadata.publishedYear || null,
isbn: ebookFileScanData.metadata.isbn || null,
asin: ebookFileScanData.metadata.asin || null
}
: null
const currentPathMetadata = {
title: libraryItemData.mediaMetadata?.title || null,
subtitle: libraryItemData.mediaMetadata?.subtitle || null,
authors: libraryItemData.mediaMetadata?.authors || [],
narrators: libraryItemData.mediaMetadata?.narrators || [],
seriesName: libraryItemData.mediaMetadata?.seriesName || null,
sequence: libraryItemData.mediaMetadata?.seriesSequence || null,
publishedYear: libraryItemData.mediaMetadata?.publishedYear || null
}
Logger.info(`[OpenAI] Inferring scan metadata for "${libraryItemData.relPath}"`)
const prompt = `You infer audiobook metadata from weak or messy directory structures.
Return only valid JSON in this shape:
{
"book": {
"title": "Book title or null",
"subtitle": "Subtitle or null",
"authors": ["Author Name"],
"narrators": ["Narrator Name"],
"seriesName": "Series name or null",
"sequence": "1 or null",
"publishedYear": "2004 or null",
"publisher": "Publisher or null",
"isbn": "ISBN or null",
"asin": "ASIN or null",
"language": "Language or null",
"reason": "brief reason"
}
}
Rules:
- Infer metadata from full path, relative path, folder names, filenames, and any provided tag metadata.
- Prefer title/author/series evidence that is explicit in filenames or tags.
- Use null when uncertain.
- If a series is provided, sequence may be null when it cannot be inferred confidently.
- Do not invent authors or series when there is weak evidence.
- Respond with one book object only.
Current path-derived metadata:
${JSON.stringify(currentPathMetadata, null, 2)}
Folder context:
${JSON.stringify(folderContext, null, 2)}
Audio files:
${JSON.stringify(audioFileCandidates, null, 2)}
Ebook metadata:
${JSON.stringify(ebookMetadata, null, 2)}`
const payload = await this.createResponse(prompt)
const validated = this.validateScanMetadataPayload(payload)
Logger.info(`[OpenAI] Scan-metadata result for "${libraryItemData.relPath}" ${this.summarizeScanMetadataForLog(validated)}`)
return validated
}
async inferDirectoryGroupingFromPaths(containerPath, mediaFiles) {
if (!this.isConfigured) {
throw new Error('OpenAI API key is not configured')
}
if (!Array.isArray(mediaFiles) || !mediaFiles.length) {
return []
}
Logger.info(`[OpenAI] Inferring directory grouping for "${containerPath}" with ${mediaFiles.length} media files`)
mediaFiles.forEach((file) => {
Logger.info(`[OpenAI] Directory-grouping candidate ${JSON.stringify(file)}`)
})
const prompt = `You infer logical audiobook item grouping from messy filesystem paths.
Return only valid JSON in this shape:
{
"files": [
{
"path": "relative/path/to/media-file.m4b",
"groupId": "short-group-label",
"reason": "brief reason"
}
]
}
Rules:
- Include every provided media file exactly once.
- Files that belong to the same logical audiobook item must share the same groupId.
- Files for different books must use different groupIds even if they are in the same series container.
- Use path, filename, parent directories, and current grouping hints as evidence.
- Prefer preserving existing grouping when it already looks reasonable.
- Do not merge different titled books just because they share a series or author folder.
- groupId only needs to be stable within this one response.
Container path:
${JSON.stringify(containerPath)}
Media files:
${JSON.stringify(mediaFiles, null, 2)}`
const payload = await this.createResponse(prompt)
const validated = this.validateDirectoryGroupingPayload(payload, mediaFiles)
validated.forEach((grouping) => {
Logger.info(`[OpenAI] Directory-grouping result ${this.summarizeDirectoryGroupingForLog(grouping)}`)
})
return validated
}
async detectDuplicateBooks(libraryItems) {
if (!this.isConfigured) {
throw new Error('OpenAI API key is not configured')
}
const books = libraryItems.map((libraryItem) => {
const metadata = libraryItem.media.oldMetadataToJSON()
const folderContext = this.getFolderContext(libraryItem)
const metadataCompletenessScore = [
metadata.title,
metadata.subtitle,
metadata.description,
metadata.isbn,
metadata.asin,
metadata.publisher,
metadata.language,
metadata.publishedYear,
metadata.authors?.length ? 'authors' : null,
metadata.series?.length ? 'series' : null,
metadata.narrators?.length ? 'narrators' : null,
libraryItem.media.coverPath ? 'cover' : null
].filter(Boolean).length
return {
id: libraryItem.id,
title: metadata.title || null,
subtitle: metadata.subtitle || null,
authors: (metadata.authors || []).map((author) => author.name),
narrators: metadata.narrators || [],
series: (metadata.series || []).map((series) => ({ name: series.name, sequence: series.sequence || null })),
publishedYear: metadata.publishedYear || null,
description: this.cleanDescription(metadata.description),
language: metadata.language || null,
abridged: !!metadata.abridged,
explicit: !!metadata.explicit,
isbn: metadata.isbn || null,
asin: metadata.asin || null,
duration: libraryItem.media.duration || null,
size: libraryItem.media.size || libraryItem.size || null,
numAudioFiles: libraryItem.media.audioFiles?.length || 0,
numChapters: libraryItem.media.chapters?.length || 0,
hasCover: !!libraryItem.media.coverPath,
ebookFormat: libraryItem.media.ebookFile?.ebookFormat || null,
isFile: !!libraryItem.isFile,
fullPath: folderContext.fullPath,
relPath: folderContext.relPath,
metadataCompletenessScore
}
})
Logger.info(`[OpenAI] Evaluating duplicate books for ${books.length} candidates`)
books.forEach((book) => {
Logger.info(`[OpenAI] Duplicate-books candidate ${JSON.stringify(book)}`)
})
const prompt = `You identify duplicate audiobook library items that represent the same underlying book/work and choose which copy to keep.
Return only valid JSON in this shape:
{
"groups": [
{
"keepId": "library-item-id-to-keep",
"duplicateIds": ["library-item-id-to-remove"],
"reason": "brief reason"
}
]
}
Rules:
- Only mark items as duplicates if they are clearly the same book/work.
- Books in the same series are not duplicates unless they are the same title/work.
- Different abridged vs unabridged editions, different languages, dramatizations, companions, or supplemental books are not duplicates unless the evidence strongly indicates they are just duplicate copies.
- Prefer keeping the copy with richer metadata, cleaner path naming, cover art, more complete file data, and generally better organization.
- Do not include a group if no duplicates should be removed.
- Do not include the same id in more than one group.
Books:
${JSON.stringify(books, null, 2)}`
const payload = await this.createResponse(prompt)
const validated = this.validateDuplicateBooksPayload(payload, books)
validated.forEach((decision) => {
Logger.info(`[OpenAI] Duplicate-books result ${this.summarizeDuplicateDecisionForLog(decision)}`)
})
return validated
}
}
module.exports = OpenAI

View file

@ -89,6 +89,8 @@ class ApiRouter {
this.router.delete('/libraries/:id/narrators/:narratorId', LibraryController.middleware.bind(this), LibraryController.removeNarrator.bind(this))
this.router.get('/libraries/:id/matchall', LibraryController.middleware.bind(this), LibraryController.matchAll.bind(this))
this.router.post('/libraries/:id/scan', LibraryController.middleware.bind(this), LibraryController.scan.bind(this))
this.router.post('/libraries/:id/detect-series-with-ai', LibraryController.middleware.bind(this), LibraryController.detectSeriesWithAI.bind(this))
this.router.post('/libraries/:id/dedupe-books-with-ai', LibraryController.middleware.bind(this), LibraryController.dedupeBooksWithAI.bind(this))
this.router.get('/libraries/:id/recent-episodes', LibraryController.middleware.bind(this), LibraryController.getRecentEpisodes.bind(this))
this.router.get('/libraries/:id/opml', LibraryController.middleware.bind(this), LibraryController.getOPMLFile.bind(this))
this.router.post('/libraries/order', LibraryController.reorder.bind(this))
@ -222,6 +224,7 @@ class ApiRouter {
//
this.router.get('/series/:id', SeriesController.middleware.bind(this), SeriesController.findOne.bind(this))
this.router.patch('/series/:id', SeriesController.middleware.bind(this), SeriesController.update.bind(this))
this.router.post('/series/:id/organize-story-order', SeriesController.middleware.bind(this), SeriesController.organizeStoryOrder.bind(this))
//
// Playback Session Routes

View file

@ -20,12 +20,15 @@ const LibraryFile = require('../objects/files/LibraryFile')
const RssFeedManager = require('../managers/RssFeedManager')
const CoverManager = require('../managers/CoverManager')
const OpenAI = require('../providers/OpenAI')
const LibraryScan = require('./LibraryScan')
const OpfFileScanner = require('./OpfFileScanner')
const NfoFileScanner = require('./NfoFileScanner')
const AbsMetadataFileScanner = require('./AbsMetadataFileScanner')
const openAI = new OpenAI()
/**
* Metadata for books pulled from files
* @typedef BookMetadataObject
@ -761,6 +764,43 @@ class BookScanner {
return null
}
async openAIPathMetadata() {
if (!openAI.isConfigured) {
this.libraryScan.addLog(LogLevel.DEBUG, `Skipping OpenAI path metadata for "${this.libraryItemData.relPath}" because OpenAI is not configured`)
return
}
const inferredMetadata = await openAI
.inferBookMetadataFromScan(this.libraryItemData, this.audioFiles, this.ebookFileScanData)
.catch((error) => {
this.libraryScan.addLog(LogLevel.WARN, `OpenAI path metadata failed for "${this.libraryItemData.relPath}": ${error.message}`)
return null
})
if (!inferredMetadata) return
const directKeys = ['title', 'subtitle', 'publishedYear', 'publisher', 'isbn', 'asin', 'language']
directKeys.forEach((key) => {
if (inferredMetadata[key]) {
this.bookMetadata[key] = inferredMetadata[key]
}
})
if (inferredMetadata.authors.length) {
this.bookMetadata.authors = inferredMetadata.authors
}
if (inferredMetadata.narrators.length) {
this.bookMetadata.narrators = inferredMetadata.narrators
}
if (inferredMetadata.seriesName) {
this.bookMetadata.series = [
{
name: inferredMetadata.seriesName,
sequence: inferredMetadata.sequence || null
}
]
}
}
/**
* Metadata from .nfo file
*/

View file

@ -7,6 +7,7 @@ const Database = require('../Database')
const fs = require('../libs/fsExtra')
const fileUtils = require('../utils/fileUtils')
const scanUtils = require('../utils/scandir')
const globals = require('../utils/globals')
const { LogLevel, ScanResult } = require('../utils/constants')
const libraryFilters = require('../utils/queries/libraryFilters')
const TaskManager = require('../managers/TaskManager')
@ -14,6 +15,10 @@ const LibraryItemScanner = require('./LibraryItemScanner')
const LibraryScan = require('./LibraryScan')
const LibraryItemScanData = require('./LibraryItemScanData')
const Task = require('../objects/Task')
const OpenAI = require('../providers/OpenAI')
const openAI = new OpenAI()
const DISC_DIR_REGEX = /^(cd|dis[ck])\s*\d{1,3}$/i
class LibraryScanner {
constructor() {
@ -309,7 +314,10 @@ class LibraryScanner {
}
const fileItems = await fileUtils.recurseFiles(folderPath)
const libraryItemGrouping = scanUtils.groupFileItemsIntoLibraryItemDirs(library.mediaType, fileItems, library.settings.audiobooksOnly)
let libraryItemGrouping = scanUtils.groupFileItemsIntoLibraryItemDirs(library.mediaType, fileItems, library.settings.audiobooksOnly)
if (library.mediaType === 'book' && library.settings.openAIDirectoryGrouping && openAI.isConfigured) {
libraryItemGrouping = await this.applyOpenAIDirectoryGrouping(folderPath, fileItems, libraryItemGrouping, library.settings.audiobooksOnly)
}
if (!Object.keys(libraryItemGrouping).length) {
Logger.error(`Root path has no media folders: ${folderPath}`)
@ -321,8 +329,12 @@ class LibraryScanner {
let isFile = false // item is not in a folder
let libraryItemData = null
let fileObjs = []
if (libraryItemPath === libraryItemGrouping[libraryItemPath]) {
// Media file in root only get title
const groupedFiles = libraryItemGrouping[libraryItemPath]
const isSingleFileGroup =
libraryItemPath === groupedFiles || (Array.isArray(groupedFiles) && groupedFiles.includes(libraryItemPath))
if (isSingleFileGroup) {
// Media file item may exist in the library root or inside a poorly-structured parent folder.
libraryItemData = {
mediaMetadata: {
title: Path.basename(libraryItemPath, Path.extname(libraryItemPath))
@ -330,11 +342,11 @@ class LibraryScanner {
path: Path.posix.join(folderPath, libraryItemPath),
relPath: libraryItemPath
}
fileObjs = await scanUtils.buildLibraryFile(folderPath, [libraryItemPath])
fileObjs = await scanUtils.buildLibraryFile(folderPath, Array.isArray(groupedFiles) ? groupedFiles : [libraryItemPath])
isFile = true
} else {
libraryItemData = scanUtils.getDataFromMediaDir(library.mediaType, folderPath, libraryItemPath)
fileObjs = await scanUtils.buildLibraryFile(libraryItemData.path, libraryItemGrouping[libraryItemPath])
fileObjs = await scanUtils.buildLibraryFile(libraryItemData.path, groupedFiles)
}
const libraryItemFolderStats = await fileUtils.getFileTimestampsWithIno(libraryItemData.path)
@ -364,6 +376,201 @@ class LibraryScanner {
return items
}
expandGroupingFiles(groupPath, groupedFiles) {
if (groupPath === groupedFiles) return [groupPath]
return groupedFiles.map((file) => {
if (file === groupPath || file.startsWith(groupPath + '/')) return file
return Path.posix.join(groupPath, file)
})
}
getOpenAIDirectoryGroupingCandidates(fileItems, mediaType, audiobooksOnly, libraryItemGrouping) {
if (mediaType !== 'book') return []
const mediaFileItems = fileItems.filter((item) => isMediaFilePath(mediaType, item.path, audiobooksOnly))
const candidatesByContainer = new Map()
mediaFileItems.forEach((item) => {
const topLevelDir = item.path.split('/').filter(Boolean)[0]
if (!topLevelDir || !item.path.includes('/')) return
if (!candidatesByContainer.has(topLevelDir)) {
candidatesByContainer.set(topLevelDir, [])
}
candidatesByContainer.get(topLevelDir).push(item)
})
return [...candidatesByContainer.entries()]
.map(([containerPath, containerMediaFileItems]) => {
const defaultGroupKeys = Object.keys(libraryItemGrouping).filter((groupPath) => groupPath === containerPath || groupPath.startsWith(containerPath + '/'))
const hasDirectMediaFileInContainer = containerMediaFileItems.some((item) => Path.posix.dirname(item.path) === containerPath)
const hasMixedDefaultFileAndDirectoryGroups =
defaultGroupKeys.some((groupPath) => Path.posix.extname(groupPath)) && defaultGroupKeys.some((groupPath) => !Path.posix.extname(groupPath))
const maxRelativeDepth = Math.max(...containerMediaFileItems.map((item) => Path.posix.relative(containerPath, item.path).split('/').filter(Boolean).length))
const suspicious = defaultGroupKeys.length <= 1 || hasDirectMediaFileInContainer || hasMixedDefaultFileAndDirectoryGroups || maxRelativeDepth > 2
if (!suspicious || containerMediaFileItems.length < 2 || containerMediaFileItems.length > 40) {
return null
}
const groupingHints = containerMediaFileItems.map((item) => ({
path: item.path,
filename: item.name,
parentDir: item.reldirpath || '',
folderHierarchy: item.path.split('/').slice(0, -1).filter(Boolean),
currentGroup: defaultGroupKeys.find((groupPath) => this.expandGroupingFiles(groupPath, libraryItemGrouping[groupPath]).includes(item.path)) || null
}))
return {
containerPath,
groupingHints
}
})
.filter(Boolean)
}
getDirectoryGroupingDescriptor(containerPath, mediaPaths) {
const sortedMediaPaths = [...mediaPaths].sort((a, b) => a.localeCompare(b))
if (sortedMediaPaths.length === 1) {
const mediaDir = Path.posix.dirname(sortedMediaPaths[0])
if (mediaDir && mediaDir !== '.' && mediaDir !== containerPath) {
return {
groupPath: mediaDir,
isFile: false
}
}
return {
groupPath: sortedMediaPaths[0],
isFile: true
}
}
const splitPaths = sortedMediaPaths.map((mediaPath) => mediaPath.split('/'))
const commonParts = []
for (let i = 0; i < Math.min(...splitPaths.map((parts) => parts.length - 1)); i++) {
const segment = splitPaths[0][i]
if (splitPaths.every((parts) => parts[i] === segment)) {
commonParts.push(segment)
} else {
break
}
}
const commonDir = commonParts.join('/')
if (commonDir) {
const canUseFolderGroup = sortedMediaPaths.every((mediaPath) => {
const relativeParts = Path.posix.relative(commonDir, mediaPath).split('/').filter(Boolean)
return relativeParts.length === 1 || (relativeParts.length === 2 && DISC_DIR_REGEX.test(relativeParts[0]))
})
if (canUseFolderGroup) {
return {
groupPath: commonDir,
isFile: false
}
}
}
return {
groupPath: sortedMediaPaths[0],
isFile: true
}
}
buildLibraryItemGroupingFromOpenAIAssignments(containerPath, fileItems, assignments, mediaType, audiobooksOnly) {
const mediaPathsByGroupId = new Map()
assignments.forEach((assignment) => {
if (!mediaPathsByGroupId.has(assignment.groupId)) {
mediaPathsByGroupId.set(assignment.groupId, [])
}
mediaPathsByGroupId.get(assignment.groupId).push(assignment.path)
})
const groupRecords = [...mediaPathsByGroupId.entries()].map(([groupId, mediaPaths]) => {
const descriptor = this.getDirectoryGroupingDescriptor(containerPath, mediaPaths)
return {
groupId,
descriptor,
mediaPaths: [...mediaPaths].sort((a, b) => a.localeCompare(b)),
files: []
}
})
groupRecords.forEach((groupRecord) => {
if (groupRecord.descriptor.isFile) {
groupRecord.files.push(...groupRecord.mediaPaths)
} else {
groupRecord.files.push(...groupRecord.mediaPaths.map((mediaPath) => Path.posix.relative(groupRecord.descriptor.groupPath, mediaPath)))
}
})
const nonMediaItems = fileItems.filter((item) => !isMediaFilePath(mediaType, item.path, audiobooksOnly))
nonMediaItems.forEach((item) => {
const itemStem = Path.basename(item.name, item.extension)
let matchingGroup = null
const basenameMatches = groupRecords.filter((groupRecord) =>
groupRecord.mediaPaths.some((mediaPath) => Path.posix.dirname(mediaPath) === item.reldirpath && Path.basename(mediaPath, Path.extname(mediaPath)) === itemStem)
)
if (basenameMatches.length === 1) {
matchingGroup = basenameMatches[0]
}
if (!matchingGroup) {
const directoryMatches = groupRecords.filter((groupRecord) => {
if (!groupRecord.descriptor.isFile) {
return item.path.startsWith(groupRecord.descriptor.groupPath + '/')
}
return Path.posix.dirname(groupRecord.descriptor.groupPath) === item.reldirpath
})
if (directoryMatches.length === 1) {
matchingGroup = directoryMatches[0]
}
}
if (!matchingGroup) return
const fileEntry = matchingGroup.descriptor.isFile ? item.path : Path.posix.relative(matchingGroup.descriptor.groupPath, item.path)
if (!matchingGroup.files.includes(fileEntry)) {
matchingGroup.files.push(fileEntry)
}
})
return groupRecords.reduce((acc, groupRecord) => {
acc[groupRecord.descriptor.groupPath] = [...new Set(groupRecord.files)]
return acc
}, {})
}
async applyOpenAIDirectoryGrouping(folderPath, fileItems, libraryItemGrouping, audiobooksOnly) {
const candidates = this.getOpenAIDirectoryGroupingCandidates(fileItems, 'book', audiobooksOnly, libraryItemGrouping)
if (!candidates.length) return libraryItemGrouping
let updatedGrouping = { ...libraryItemGrouping }
for (const candidate of candidates) {
Logger.info(`[LibraryScanner] Evaluating OpenAI directory grouping for "${candidate.containerPath}" with ${candidate.groupingHints.length} media files`)
const containerFileItems = fileItems.filter((item) => item.path === candidate.containerPath || item.path.startsWith(candidate.containerPath + '/'))
const assignments = await openAI.inferDirectoryGroupingFromPaths(candidate.containerPath, candidate.groupingHints).catch((error) => {
Logger.warn(`[LibraryScanner] OpenAI directory grouping failed for "${candidate.containerPath}": ${error.message}`)
return null
})
if (!assignments?.length) continue
const aiGrouping = this.buildLibraryItemGroupingFromOpenAIAssignments(candidate.containerPath, containerFileItems, assignments, 'book', audiobooksOnly)
if (!Object.keys(aiGrouping).length) continue
updatedGrouping = Object.fromEntries(
Object.entries(updatedGrouping).filter(([groupPath]) => !(groupPath === candidate.containerPath || groupPath.startsWith(candidate.containerPath + '/')))
)
updatedGrouping = {
...updatedGrouping,
...aiGrouping
}
Logger.info(`[LibraryScanner] Applied OpenAI directory grouping for "${candidate.containerPath}" -> ${Object.keys(aiGrouping).length} library items`)
}
return updatedGrouping
}
/**
* Scan files changed from Watcher
* @param {import('../Watcher').PendingFileUpdate[]} fileUpdates
@ -646,16 +853,24 @@ function ItemToFileInoMatch(libraryItem1, libraryItem2) {
return libraryItem1.isFile && libraryItem2.libraryFiles.some((lf) => lf.ino === libraryItem1.ino)
}
function isMediaFilePath(mediaType, filepath, audiobooksOnly = false) {
const ext = Path.extname(filepath).slice(1).toLowerCase()
if (!ext) return false
if (mediaType === 'podcast') return globals.SupportedAudioTypes.includes(ext)
if (audiobooksOnly) return globals.SupportedAudioTypes.includes(ext)
return globals.SupportedAudioTypes.includes(ext) || globals.SupportedEbookTypes.includes(ext)
}
function ItemToItemInoMatch(libraryItem1, libraryItem2) {
return libraryItem1.ino === libraryItem2.ino
}
function hasAudioFiles(fileUpdateGroup, itemDir) {
return isSingleMediaFile(fileUpdateGroup, itemDir) ? scanUtils.checkFilepathIsAudioFile(fileUpdateGroup[itemDir]) : fileUpdateGroup[itemDir].some(scanUtils.checkFilepathIsAudioFile)
return isSingleMediaFile(fileUpdateGroup, itemDir) ? scanUtils.checkFilepathIsAudioFile(itemDir) : fileUpdateGroup[itemDir].some(scanUtils.checkFilepathIsAudioFile)
}
function isSingleMediaFile(fileUpdateGroup, itemDir) {
return itemDir === fileUpdateGroup[itemDir]
return itemDir === fileUpdateGroup[itemDir] || (Array.isArray(fileUpdateGroup[itemDir]) && fileUpdateGroup[itemDir].includes(itemDir))
}
async function findLibraryItemByItemToItemInoMatch(libraryId, fullPath) {

View file

@ -30,6 +30,10 @@ function isScannableNonMediaFile(ext) {
return globals.TextFileTypes.includes(extclean) || globals.MetadataFileTypes.includes(extclean) || globals.SupportedImageTypes.includes(extclean)
}
function isDiscDirectoryName(name) {
return /^(cd|dis[ck])\s*\d{1,3}$/i.test(name || '')
}
function checkFilepathIsAudioFile(filepath) {
const ext = Path.extname(filepath)
if (!ext) return false
@ -67,52 +71,76 @@ function groupFileItemsIntoLibraryItemDirs(mediaType, fileItems, audiobooksOnly,
// Step 3: Group media files (or non-media files if includeNonMediaFiles is true) in library items
const libraryItemGroup = {}
const directMediaFileCountByDir = {}
const hasNestedNonDiscMediaByDir = {}
mediaFileItems.forEach((item) => {
const dirPath = item.reldirpath || ''
directMediaFileCountByDir[dirPath] = (directMediaFileCountByDir[dirPath] || 0) + 1
const dirParts = dirPath.split('/').filter(Boolean)
for (let i = 0; i < dirParts.length - 1; i++) {
const ancestorPath = dirParts.slice(0, i + 1).join('/')
const nextSegment = dirParts[i + 1]
if (!isDiscDirectoryName(nextSegment)) {
hasNestedNonDiscMediaByDir[ancestorPath] = true
}
}
})
mediaFileItems.forEach((item) => {
const dirparts = item.reldirpath.split('/').filter((p) => !!p)
const numparts = dirparts.length
let _path = ''
if (!dirparts.length) {
// Media file in root
libraryItemGroup[item.name] = item.name
libraryItemGroup[item.path] = item.path
} else {
// Iterate over directories in path
for (let i = 0; i < numparts; i++) {
const dirpart = dirparts.shift()
_path = Path.posix.join(_path, dirpart)
const dirPath = dirparts.join('/')
const lastDir = dirparts[dirparts.length - 1]
const shouldUseFileAsLibraryItem = directMediaFileCountByDir[dirPath] === 1 && hasNestedNonDiscMediaByDir[dirPath]
if (libraryItemGroup[_path]) {
// Directory already has files, add file
const relpath = Path.posix.join(dirparts.join('/'), item.name)
libraryItemGroup[_path].push(relpath)
return
} else if (!dirparts.length) {
// This is the last directory, create group
libraryItemGroup[_path] = [item.name]
return
} else if (dirparts.length === 1 && /^(cd|dis[ck])\s*\d{1,3}$/i.test(dirparts[0])) {
// Next directory is the last and is a CD dir, create group
libraryItemGroup[_path] = [Path.posix.join(dirparts[0], item.name)]
return
}
if (shouldUseFileAsLibraryItem) {
libraryItemGroup[item.path] = [item.path]
return
}
const groupPath = isDiscDirectoryName(lastDir) && dirparts.length > 1 ? dirparts.slice(0, -1).join('/') : dirPath
const relpath = Path.posix.relative(groupPath, item.path)
if (!libraryItemGroup[groupPath]) {
libraryItemGroup[groupPath] = []
}
libraryItemGroup[groupPath].push(relpath)
}
})
// Step 4: Add other files into library item groups
otherFileItems.forEach((item) => {
const dirparts = item.reldirpath.split('/')
const numparts = dirparts.length
let _path = ''
const dirPath = item.reldirpath || ''
const itemPath = item.path
const sameDirFileGroups = Object.keys(libraryItemGroup).filter((groupPath) => {
if (!groupPath || !Path.posix.extname(groupPath)) return false
return Path.posix.dirname(groupPath) === dirPath
})
// Iterate over directories in path
for (let i = 0; i < numparts; i++) {
const dirpart = dirparts.shift()
_path = Path.posix.join(_path, dirpart)
if (libraryItemGroup[_path]) {
// Directory is audiobook group
const relpath = Path.posix.join(dirparts.join('/'), item.name)
libraryItemGroup[_path].push(relpath)
if (sameDirFileGroups.length) {
const itemStem = Path.basename(item.name, item.extension)
const matchingFileGroup =
sameDirFileGroups.find((groupPath) => Path.basename(groupPath, Path.extname(groupPath)) === itemStem) ||
(sameDirFileGroups.length === 1 ? sameDirFileGroups[0] : null)
if (matchingFileGroup) {
if (Array.isArray(libraryItemGroup[matchingFileGroup])) {
libraryItemGroup[matchingFileGroup].push(itemPath)
}
return
}
}
const dirparts = dirPath.split('/').filter(Boolean)
for (let i = dirparts.length; i >= 1; i--) {
const groupPath = dirparts.slice(0, i).join('/')
if (Array.isArray(libraryItemGroup[groupPath])) {
const relpath = Path.posix.relative(groupPath, itemPath)
libraryItemGroup[groupPath].push(relpath)
return
}
}

View file

@ -0,0 +1,272 @@
const { expect } = require('chai')
const OpenAI = require('../../../server/providers/OpenAI')
describe('OpenAI', () => {
let openAI
beforeEach(() => {
openAI = new OpenAI()
})
describe('parseJsonResponse', () => {
it('parses fenced JSON', () => {
const payload = openAI.parseJsonResponse('```json\n{"books":[{"id":"1","sequence":"1"}]}\n```')
expect(payload.books).to.have.length(1)
expect(payload.books[0].id).to.equal('1')
})
it('parses JSON wrapped in extra text', () => {
const payload = openAI.parseJsonResponse('Result:\n{"books":[{"id":"1","sequence":"1"}]}')
expect(payload.books[0].sequence).to.equal('1')
})
})
describe('validateSeriesOrderPayload', () => {
it('normalizes valid ordered books', () => {
const result = openAI.validateSeriesOrderPayload(
{
books: [
{ id: 'b', sequence: '2' },
{ id: 'a', sequence: '1' }
]
},
[{ id: 'a' }, { id: 'b' }]
)
expect(result.map((book) => book.id)).to.deep.equal(['a', 'b'])
})
it('rejects duplicate sequences', () => {
expect(() =>
openAI.validateSeriesOrderPayload(
{
books: [
{ id: 'a', sequence: '1' },
{ id: 'b', sequence: '1' }
]
},
[{ id: 'a' }, { id: 'b' }]
)
).to.throw('duplicate sequence')
})
})
describe('validateSeriesDetectionPayload', () => {
it('allows null standalone assignments', () => {
const result = openAI.validateSeriesDetectionPayload(
{
books: [
{ id: 'a', seriesName: null, sequence: null },
{ id: 'b', seriesName: 'Series Name', sequence: '1.5' }
]
},
[{ id: 'a' }, { id: 'b' }]
)
expect(result[0].seriesName).to.equal(null)
expect(result[1].sequence).to.equal('1.5')
})
it('skips a series assignment without sequence', () => {
const result = openAI.validateSeriesDetectionPayload(
{
books: [
{ id: 'a', seriesName: 'Series Name', sequence: null, reason: 'folder match' }
]
},
[{ id: 'a' }]
)
expect(result[0].seriesName).to.equal(null)
expect(result[0].sequence).to.equal(null)
expect(result[0].reason).to.contain('skipped due to missing or invalid sequence')
})
it('skips a sequence without series name', () => {
const result = openAI.validateSeriesDetectionPayload(
{
books: [
{ id: 'a', seriesName: null, sequence: '2', reason: 'sequence found' }
]
},
[{ id: 'a' }]
)
expect(result[0].seriesName).to.equal(null)
expect(result[0].sequence).to.equal(null)
expect(result[0].reason).to.contain('skipped due to missing series name')
})
it('ignores unknown ids and backfills missing ids', () => {
const result = openAI.validateSeriesDetectionPayload(
{
books: [
{ id: 'z', seriesName: 'Wrong Series', sequence: '9' },
{ id: 'a', seriesName: 'Series Name', sequence: '1' }
]
},
[{ id: 'a' }, { id: 'b' }]
)
expect(result).to.have.length(2)
expect(result[0].id).to.equal('a')
expect(result[0].seriesName).to.equal('Series Name')
expect(result[1].id).to.equal('b')
expect(result[1].seriesName).to.equal(null)
expect(result[1].reason).to.contain('omitted this book')
})
it('ignores duplicate ids in detection payload', () => {
const result = openAI.validateSeriesDetectionPayload(
{
books: [
{ id: 'a', seriesName: 'Series Name', sequence: '1' },
{ id: 'a', seriesName: 'Other Series', sequence: '2' }
]
},
[{ id: 'a' }]
)
expect(result).to.have.length(1)
expect(result[0].seriesName).to.equal('Series Name')
expect(result[0].sequence).to.equal('1')
})
})
describe('validateScanMetadataPayload', () => {
it('normalizes valid scan metadata', () => {
const result = openAI.validateScanMetadataPayload({
book: {
title: ' Neuromancer ',
subtitle: '20th Anniversary Edition',
authors: ['William Gibson', 'William Gibson'],
narrators: ['Robertson Dean'],
seriesName: 'Sprawl Trilogy',
sequence: '1',
publishedYear: '1984',
asin: 'B000FC11ZG',
isbn: '978-0441569595',
language: 'English',
reason: 'path and tags match'
}
})
expect(result.title).to.equal('Neuromancer')
expect(result.authors).to.deep.equal(['William Gibson'])
expect(result.seriesName).to.equal('Sprawl Trilogy')
expect(result.sequence).to.equal('1')
expect(result.asin).to.equal('B000FC11ZG')
expect(result.isbn).to.equal('9780441569595')
})
it('returns nulls for malformed scan metadata fields', () => {
const result = openAI.validateScanMetadataPayload({
title: '',
authors: [''],
seriesName: 'Series Name',
sequence: 'not-a-sequence',
publishedYear: '84',
asin: 'bad',
isbn: 'nope'
})
expect(result.title).to.equal(null)
expect(result.authors).to.deep.equal([])
expect(result.seriesName).to.equal('Series Name')
expect(result.sequence).to.equal(null)
expect(result.publishedYear).to.equal(null)
expect(result.asin).to.equal(null)
expect(result.isbn).to.equal(null)
})
})
describe('validateDirectoryGroupingPayload', () => {
it('normalizes valid directory-grouping payload', () => {
const result = openAI.validateDirectoryGroupingPayload(
{
files: [
{ path: 'Series Alpha/Book One.m4b', groupId: ' book-one ', reason: 'same book' },
{ path: 'Series Alpha/Disc 1/Book Two Part 1.mp3', groupId: 'book-two', reason: 'disc set' }
]
},
[{ path: 'Series Alpha/Book One.m4b' }, { path: 'Series Alpha/Disc 1/Book Two Part 1.mp3' }]
)
expect(result[0].groupId).to.equal('book-one')
expect(result[1].groupId).to.equal('book-two')
})
it('backfills missing or invalid grouping rows', () => {
const result = openAI.validateDirectoryGroupingPayload(
{
files: [
{ path: 'unknown/path.m4b', groupId: 'ignored' },
{ path: 'Series Alpha/Book One.m4b', groupId: '' }
]
},
[{ path: 'Series Alpha/Book One.m4b' }, { path: 'Series Alpha/Book Two.m4b' }]
)
expect(result[0].groupId).to.equal('Series Alpha/Book One.m4b')
expect(result[1].groupId).to.equal('Series Alpha/Book Two.m4b')
expect(result[1].reason).to.contain('omitted this media file')
})
})
describe('validateDuplicateBooksPayload', () => {
it('normalizes valid duplicate-book groups', () => {
const result = openAI.validateDuplicateBooksPayload(
{
groups: [
{
keepId: 'a',
duplicateIds: ['b', 'c'],
reason: 'same book'
}
]
},
[{ id: 'a' }, { id: 'b' }, { id: 'c' }]
)
expect(result).to.deep.equal([
{
keepId: 'a',
duplicateIds: ['b', 'c'],
reason: 'same book'
}
])
})
it('ignores invalid and overlapping duplicate-book groups', () => {
const result = openAI.validateDuplicateBooksPayload(
{
groups: [
{
keepId: 'a',
duplicateIds: ['b', 'missing', 'a'],
reason: 'primary match'
},
{
keepId: 'b',
duplicateIds: ['c'],
reason: 'should be skipped because b was consumed'
},
{
keepId: 'z',
duplicateIds: ['c']
}
]
},
[{ id: 'a' }, { id: 'b' }, { id: 'c' }]
)
expect(result).to.deep.equal([
{
keepId: 'a',
duplicateIds: ['b'],
reason: 'primary match'
}
])
})
})
})

View file

@ -33,6 +33,7 @@ describe('scanUtils', async () => {
const dirname = Path.dirname(filePath)
fileItems.push({
name: Path.basename(filePath),
path: filePath,
reldirpath: dirname === '.' ? '' : dirname,
extension: Path.extname(filePath),
deep: filePath.split('/').length - 1
@ -49,4 +50,43 @@ describe('scanUtils', async () => {
'Author/Series2/Book5/deeply/nested': ['cd 01/audiofile.mp3', 'cd 02/audiofile.mp3']
})
})
it('should keep nested book folders separate when a parent folder also contains a direct media file', async () => {
const filePaths = [
'Series Alpha/Standalone Side Story.m4b',
'Series Alpha/Standalone Side Story.nfo',
'Series Alpha/Author Example - Book One Main Saga, Book 1/Book One Main Saga, Book 1.m4b',
'Series Alpha/Author Example - Book One Main Saga, Book 1/Book One Main Saga, Book 1.cue',
'Series Alpha/Author Example - Book Two Main Saga, Book 2/Book Two Main Saga, Book 2.m4b',
'Series Alpha/Author Example - Book Two Main Saga, Book 2/Book Two Main Saga, Book 2.nfo'
]
const fileItems = filePaths.map((filePath) => {
const dirname = Path.dirname(filePath)
return {
name: Path.basename(filePath),
path: filePath,
reldirpath: dirname === '.' ? '' : dirname,
extension: Path.extname(filePath),
deep: filePath.split('/').length - 1
}
})
const libraryItemGrouping = scanUtils.groupFileItemsIntoLibraryItemDirs('book', fileItems, false)
expect(libraryItemGrouping).to.deep.equal({
'Series Alpha/Standalone Side Story.m4b': [
'Series Alpha/Standalone Side Story.m4b',
'Series Alpha/Standalone Side Story.nfo'
],
'Series Alpha/Author Example - Book One Main Saga, Book 1': [
'Book One Main Saga, Book 1.m4b',
'Book One Main Saga, Book 1.cue'
],
'Series Alpha/Author Example - Book Two Main Saga, Book 2': [
'Book Two Main Saga, Book 2.m4b',
'Book Two Main Saga, Book 2.nfo'
]
})
})
})