replace continuous spaces and new lines with space in preview content

This commit is contained in:
Hongbo Wu 2024-06-07 13:29:38 +08:00
parent 76383d8a33
commit 0db1108675
6 changed files with 58 additions and 40 deletions

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

@ -23,12 +23,19 @@ export const generatePreviewContent = async (
return
}
const libraryItem = await findLibraryItemById(libraryItemId, userId)
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}`)
@ -39,7 +46,11 @@ export const generatePreviewContent = async (
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?.slice(0, 600)
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}`
@ -47,6 +58,11 @@ export const generatePreviewContent = async (
return
}
logger.info('Generated preview for library item', {
libraryItemId,
previewContent,
})
await updateLibraryItem(
libraryItemId,
{

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

@ -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

@ -16,10 +16,6 @@ import {
SaveResult,
} from '../generated/graphql'
import { Merge } from '../util'
import {
enqueueGeneratePreviewContentJob,
enqueueThumbnailJob,
} from '../utils/createTask'
import {
cleanUrl,
generateSlug,
@ -166,27 +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)
}
}
const excerpt = parseResult.parsedContent?.excerpt
// generate preview content if excerpt is less than 180 characters
if (!excerpt || excerpt.length < 180) {
try {
await enqueueGeneratePreviewContentJob(clientRequestId, user.id)
} catch (e) {
logger.error('Failed to enqueue generate preview job', e)
}
}
if (parseResult.highlightData) {
const highlight: DeepPartial<Highlight> = {
...parseResult.highlightData,