ale/fediblock-instance
ale/fediblock-instance
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

taskdeletedup with streams
All checks were successful
continuous-integration/drone/push Build is passing
Details

Browse Source
This commit is contained in:
ale 2024-10-19 20:37:54 +02:00
parent b8f927be2b
commit 339ab9c192
1 changed files with 56 additions and 39 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

53
lib/taskdeletedup.js
View File

@ -1,12 +1,19 @@
module.exports = client => {
const constant = require('./constant'),
schedule = require('node-schedule'),
{ pick } = require('stream-json/filters/Pick'),
{ parser } = require('stream-json'),
{ streamArray } = require('stream-json/streamers/StreamArray'),
{ chain } = require('stream-chain'),
size = 500,
deleteDup = async () => {
const count = { deleted: 0, total: 0 }
let lastsort = undefined, result = undefined, last = undefined
do {
result = await client.search({
const count = { deleted: 0, total: 0, current: 0 }
let lastsort = undefined, last = undefined
await searchDup(count, lastsort, last)
},
searchDup = async (count, lastsort, last) => {
count.current = 0
const result = await client.search({
index: constant.index,
size: size,
body: {
@ -25,26 +32,36 @@ module.exports = client => {
}],
search_after: lastsort
}
})
for (const instance of result.hits.hits) {
if (last && instance._source && last.instance === instance._source.instance) {
await client.delete({ index: constant.index, id: instance._id })
}, { asStream: true, meta: false }),
pipeline = chain([
parser(),
pick({ filter: 'hits.hits' }),
streamArray(),
data => data.value
])
pipeline.on('data', async data => {
count.current++
if (last && last.instance === data._source.instance) {
await client.delete({ index: constant.index, id: data._id })
count.deleted++
console.log('deleted ' + instance._id + ': ' + instance._source.instance)
console.log('deleted ' + data._id + ': ' + data._source.instance)
}
else {
last = instance._source
last = data._source
}
if (instance._id === result.hits.hits[result.hits.hits.length - 1]._id) {
lastsort = instance.sort
if (count.current === size) {
lastsort = data.sort
}
}
count.total += result.hits.hits.length
if (result.hits.hits.length !== size) {
break
}
} while (result.hits && result.hits.hits && result.hits.hits.length > 0)
})
pipeline.on('end', async () => {
count.total += size
if (count.current === size) {
await searchDup(count, lastsort, last)
} else {
console.log('Index: ' + constant.index + ' - Total: ' + count.total + ' - Deleted: ' + count.deleted)
}
})
result.pipe(pipeline)
},
job = schedule.scheduleJob('0 ' + constant.taskdeletedup + ' * * *', async () => {
await deleteDup()