Merge pull request #4035 from omnivore-app/feature/generate-preview-content

feat: generate preview content if excerpt less than 180 chars
This commit is contained in:
Hongbo Wu 2024-06-07 13:40:17 +08:00 committed by GitHub
commit bafc20c9cd
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
12 changed files with 174 additions and 26 deletions

View file

@ -219,4 +219,7 @@ export class LibraryItem {
@Column('float')
score?: number
@Column('text')
previewContent?: string
}

View file

@ -22,7 +22,7 @@ interface ImageSize {
export const THUMBNAIL_JOB = 'find-thumbnail'
const fetchImage = async (url: string): Promise<AxiosResponse | null> => {
logger.info('fetching image', url)
logger.info('fetching image', { url })
try {
// get image file by url
return await axios.get(url, {

View file

@ -0,0 +1,75 @@
import { parseHTML } from 'linkedom'
import {
findLibraryItemById,
updateLibraryItem,
} from '../services/library_item'
import { findActiveUser } from '../services/user'
import { logger } from '../utils/logger'
export const GENERATE_PREVIEW_CONTENT_JOB = 'generate-preview'
interface GeneratePreviewContentData {
libraryItemId: string
userId: string
}
export const generatePreviewContent = async (
job: GeneratePreviewContentData
) => {
const { libraryItemId, userId } = job
const user = await findActiveUser(userId)
if (!user) {
logger.error(`User not found: ${userId}`)
return
}
const libraryItem = await findLibraryItemById(libraryItemId, userId, {
select: ['id', 'readableContent', 'previewContent'],
})
if (!libraryItem) {
logger.error(`Library item not found: ${libraryItemId}`)
return
}
if (libraryItem.previewContent && libraryItem.previewContent.length >= 180) {
logger.info(`Preview content already exists: ${libraryItemId}`)
return
}
const content = libraryItem.readableContent
if (!content) {
logger.error(`Library item has no content: ${libraryItemId}`)
return
}
// Generate preview content
logger.info(`Generating preview for library item: ${libraryItemId}`)
// the preview content should be within 600 characters
const document = parseHTML(content).document
const previewContent = document.documentElement.textContent
?.trim()
?.replace(/\s+/g, ' ')
?.replace(/\n/g, ' ')
?.slice(0, 600)
if (!previewContent) {
logger.error(
`Failed to generate preview for library item: ${libraryItemId}`
)
return
}
logger.info('Generated preview for library item', {
libraryItemId,
previewContent,
})
await updateLibraryItem(
libraryItemId,
{
previewContent,
},
userId,
undefined,
true
)
}

View file

@ -4,8 +4,10 @@ import { RuleEventType } from './entity/rule'
import { env } from './env'
import { ReportType } from './generated/graphql'
import {
enqueueGeneratePreviewContentJob,
enqueueProcessYouTubeVideo,
enqueueScoreJob,
enqueueThumbnailJob,
enqueueTriggerRuleJob,
} from './utils/createTask'
import { logger } from './utils/logger'
@ -80,6 +82,42 @@ export const createPubSubClient = (): PubsubClient => {
userId,
libraryItemId: data.id,
})
const hasThumbnail = (
data: any
): data is { thumbnail: string | null } => {
return 'thumbnail' in data
}
// we don't want to create thumbnail for imported pages and pages that already have thumbnail
if (!hasThumbnail(data) || !data.thumbnail) {
try {
// create a task to update thumbnail and pre-cache all images
const job = await enqueueThumbnailJob(userId, data.id)
logger.info('Thumbnail job created', { id: job?.id })
} catch (e) {
logger.error('Failed to enqueue thumbnail job', e)
}
}
const hasPreviewContent = (
data: any
): data is { previewContent: string | null } => {
return 'previewContent' in data
}
// generate preview content if it is less than 180 characters
if (
!hasPreviewContent(data) ||
(data.previewContent && data.previewContent.length < 180)
) {
try {
const job = await enqueueGeneratePreviewContentJob(data.id, userId)
logger.info('Generate preview job created', { id: job?.id })
} catch (e) {
logger.error('Failed to enqueue generate preview job', e)
}
}
}
},
entityUpdated: async <T extends EntityEvent>(

View file

@ -31,6 +31,10 @@ import {
} from './jobs/email/inbound_emails'
import { sendEmailJob, SEND_EMAIL_JOB } from './jobs/email/send_email'
import { findThumbnail, THUMBNAIL_JOB } from './jobs/find_thumbnail'
import {
generatePreviewContent,
GENERATE_PREVIEW_CONTENT_JOB,
} from './jobs/generate_preview_content'
import {
exportAllItems,
EXPORT_ALL_ITEMS_JOB_NAME,
@ -194,6 +198,8 @@ export const createWorker = (connection: ConnectionOptions) =>
return updateHome(job.data)
case SCORE_LIBRARY_ITEM_JOB:
return scoreLibraryItem(job.data)
case GENERATE_PREVIEW_CONTENT_JOB:
return generatePreviewContent(job.data)
default:
logger.warning(`[queue-processor] unhandled job: ${job.name}`)
}

View file

@ -686,7 +686,8 @@ export const functionResolvers = {
canComment: false,
canShare: true,
dir: libraryItem.directionality,
previewContent: libraryItem.description,
previewContent:
libraryItem.previewContent || libraryItem.description,
subscription: libraryItem.subscription,
siteName: libraryItem.siteName,
siteIcon: libraryItem.siteIcon,

View file

@ -150,6 +150,7 @@ export const batchGetLibraryItems = async (ids: readonly string[]) => {
'archivedAt',
'deletedAt',
'slug',
'previewContent',
],
})

View file

@ -3,7 +3,6 @@ import {
LibraryItem,
LibraryItemState,
} from '../entity/library_item'
import { enqueueThumbnailJob } from '../utils/createTask'
import {
cleanUrl,
generateSlug,
@ -12,7 +11,6 @@ import {
validatedDate,
wordsCount,
} from '../utils/helpers'
import { logger } from '../utils/logger'
import {
FAKE_URL_PREFIX,
fetchFavicon,
@ -143,15 +141,5 @@ export const saveEmail = async (
await updateReceivedEmail(input.receivedEmailId, 'article', input.userId)
if (!newLibraryItem.thumbnail) {
// create a task to update thumbnail and pre-cache all images
try {
const job = await enqueueThumbnailJob(input.userId, newLibraryItem.id)
logger.info('Created thumbnail job', { taskId: job })
} catch (e) {
logger.error('Failed to create thumbnail job', e)
}
}
return newLibraryItem
}

View file

@ -16,7 +16,6 @@ import {
SaveResult,
} from '../generated/graphql'
import { Merge } from '../util'
import { enqueueThumbnailJob } from '../utils/createTask'
import {
cleanUrl,
generateSlug,
@ -163,17 +162,6 @@ export const savePage = async (
input.rssFeedUrl
)
// we don't want to create thumbnail for imported pages and pages that already have thumbnail
if (!isImported && !parseResult.parsedContent?.previewImage) {
try {
// create a task to update thumbnail and pre-cache all images
const job = await enqueueThumbnailJob(user.id, clientRequestId)
logger.info('Created thumbnail job', { job })
} catch (e) {
logger.error('Failed to enqueue thumbnail job', e)
}
}
if (parseResult.highlightData) {
const highlight: DeepPartial<Highlight> = {
...parseResult.highlightData,
@ -255,6 +243,7 @@ export const parsedContentToLibraryItem = ({
originalContent: originalHtml,
readableContent: parsedContent?.content || '',
description: parsedContent?.excerpt,
previewContent: parsedContent?.excerpt,
title:
title ||
parsedContent?.title ||

View file

@ -29,6 +29,7 @@ import { BulkActionData, BULK_ACTION_JOB_NAME } from '../jobs/bulk_action'
import { CallWebhookJobData, CALL_WEBHOOK_JOB_NAME } from '../jobs/call_webhook'
import { SendEmailJobData, SEND_EMAIL_JOB } from '../jobs/email/send_email'
import { THUMBNAIL_JOB } from '../jobs/find_thumbnail'
import { GENERATE_PREVIEW_CONTENT_JOB } from '../jobs/generate_preview_content'
import { EXPORT_ALL_ITEMS_JOB_NAME } from '../jobs/integration/export_all_items'
import {
ExportItemJobData,
@ -110,6 +111,7 @@ export const getJobPriority = (jobName: string): number => {
case EXPORT_ALL_ITEMS_JOB_NAME:
case REFRESH_ALL_FEEDS_JOB_NAME:
case THUMBNAIL_JOB:
case GENERATE_PREVIEW_CONTENT_JOB:
return 100
default:
@ -1024,4 +1026,29 @@ export const enqueueScoreJob = async (data: ScoreLibraryItemJobData) => {
})
}
export const enqueueGeneratePreviewContentJob = async (
libraryItemId: string,
userId: string
) => {
const queue = await getBackendQueue()
if (!queue) {
return undefined
}
return queue.add(
GENERATE_PREVIEW_CONTENT_JOB,
{
libraryItemId,
userId,
},
{
jobId: `${GENERATE_PREVIEW_CONTENT_JOB}_${libraryItemId}_${JOB_VERSION}`,
removeOnComplete: true,
removeOnFail: true,
priority: getJobPriority(GENERATE_PREVIEW_CONTENT_JOB),
attempts: 3,
}
)
}
export default createHttpTaskWithToken

View file

@ -0,0 +1,10 @@
-- Type: DO
-- Name: add_preview_content_to_library_item
-- Description: Add preview_content column to library_item table
BEGIN;
ALTER TABLE omnivore.library_item
ADD COLUMN preview_content TEXT;
COMMIT;

View file

@ -0,0 +1,10 @@
-- Type: UNDO
-- Name: add_preview_content_to_library_item
-- Description: Add preview_content column to library_item table
BEGIN;
ALTER TABLE omnivore.library_item
DROP COLUMN preview_content;
COMMIT;