leon-ai/leon
1
1
Fork 0
mirror of https://github.com/leon-ai/leon.git synced 2024-11-30 19:07:39 +03:00
Code Issues Projects Releases Wiki Activity

refactor(server): default NLU result

Browse Source
This commit is contained in:
louistiti 2023-03-20 22:35:44 +08:00
parent a8e3fb8b87
commit a215fc8770
2 changed files with 52 additions and 70 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

14
server/src/core/brain/brain.ts
View File

@ -1,17 +1,3 @@
/**
* TODO next:
* 1. [OK] Fix brain.ts TS errors
* 2. [OK] Refactor brain.ts; split "execute" into smaller functions:
* // [OK] handle this scope into its own method:
// - handleLogicActionSkillProcessOutput
// - handleLogicActionSkillProcessError
// - handleLogicActionSkillProcessClose
* 3. Fix nlu.ts TS errors
* 4. Refactor nlu.ts; split into smaller functions
* 5. Restore multi client support on HTTP server / socket server
* 6. Publish to "develop" (or just fix TS errors only and publish first, then refactor)
*/
import fs from 'node:fs'
import path from 'node:path'
import { ChildProcessWithoutNullStreams, spawn } from 'node:child_process'

108
server/src/core/nlp/nlu/nlu.ts
View File

@ -5,6 +5,7 @@ import { spawn } from 'node:child_process'
import axios from 'axios'
import kill from 'tree-kill'
import type { NLUResult } from '@/core/nlp/types'
import { langs } from '@@/core/langs.json'
import { version } from '@@/package.json'
import { HAS_LOGGER, IS_TESTING_ENV, TCP_SERVER_BIN_PATH } from '@/constants'
@ -14,35 +15,30 @@ import { StringHelper } from '@/helpers/string-helper'
import { LangHelper } from '@/helpers/lang-helper'
import Conversation from '@/core/conversation'
const defaultNluResultObj = {
utterance: null,
const DEFAULT_NLU_RESULT = {
utterance: '',
currentEntities: [],
entities: [],
currentResolvers: [],
resolvers: [],
slots: null,
configDataFilePath: null,
slots: {},
configDataFilePath: '',
answers: [], // For dialog action type
classification: {
domain: null,
skill: null,
action: null,
domain: '',
skill: '',
action: '',
confidence: 0
}
}
export default class NLU {
private static instance: NLU
private conversation = new Conversation('conv0')
private nluResult: NLUResult = DEFAULT_NLU_RESULT
constructor() {
if (!NLU.instance) {
this.globalResolversNlp = {}
this.skillsResolversNlp = {}
this.mainNlp = {}
this.ner = {}
this.conv = new Conversation('conv0')
this.nluResult = defaultNluResultObj // TODO
LogHelper.title('NLU')
LogHelper.success('New instance')
@ -118,7 +114,7 @@ export default class NLU {
* Handle in action loop logic before NLU processing
*/
async handleActionLoop(utterance) {
const { domain, intent } = this.conv.activeContext
const { domain, intent } = this.conversation.activeContext
const [skillName, actionName] = intent.split('.')
const configDataFilePath = join(
process.cwd(),
@ -128,8 +124,8 @@ export default class NLU {
`config/${BRAIN.lang}.json`
)
this.nluResult = {
...defaultNluResultObj, // Reset entities, slots, etc.
slots: this.conv.activeContext.slots,
...DEFAULT_NLU_RESULT, // Reset entities, slots, etc.
slots: this.conversation.activeContext.slots,
utterance,
configDataFilePath,
classification: {
@ -209,7 +205,7 @@ export default class NLU {
// Ensure expected items are in the utterance, otherwise clean context and reprocess
if (!hasMatchingEntity && !hasMatchingResolver) {
BRAIN.talk(`${BRAIN.wernicke('random_context_out_of_topic')}.`)
this.conv.cleanActiveContext()
this.conversation.cleanActiveContext()
await this.process(utterance)
return null
}
@ -218,9 +214,9 @@ export default class NLU {
const processedData = await BRAIN.execute(this.nluResult)
// Reprocess with the original utterance that triggered the context at first
if (processedData.core?.restart === true) {
const { originalUtterance } = this.conv.activeContext
const { originalUtterance } = this.conversation.activeContext
this.conv.cleanActiveContext()
this.conversation.cleanActiveContext()
await this.process(originalUtterance)
return null
}
@ -233,15 +229,15 @@ export default class NLU {
!processedData.action.next_action &&
processedData.core?.isInActionLoop === false
) {
this.conv.cleanActiveContext()
this.conversation.cleanActiveContext()
return null
}
// Break the action loop and prepare for the next action if necessary
if (processedData.core?.isInActionLoop === false) {
this.conv.activeContext.isInActionLoop = !!processedData.action.loop
this.conv.activeContext.actionName = processedData.action.next_action
this.conv.activeContext.intent = `${processedData.classification.skill}.${processedData.action.next_action}`
this.conversation.activeContext.isInActionLoop = !!processedData.action.loop
this.conversation.activeContext.actionName = processedData.action.next_action
this.conversation.activeContext.intent = `${processedData.classification.skill}.${processedData.action.next_action}`
}
return processedData
@ -268,7 +264,7 @@ export default class NLU {
if (processedData && Object.keys(processedData).length > 0) {
// Set new context with the next action if there is one
if (processedData.action.next_action) {
this.conv.activeContext = {
this.conversation.activeContext = {
lang: BRAIN.lang,
slots: processedData.slots,
isInActionLoop: !!processedData.nextAction.loop,
@ -313,14 +309,14 @@ export default class NLU {
await this.mergeSpacyEntities(utterance)
// Pre NLU processing according to the active context if there is one
if (this.conv.hasActiveContext()) {
if (this.conversation.hasActiveContext()) {
// When the active context is in an action loop, then directly trigger the action
if (this.conv.activeContext.isInActionLoop) {
if (this.conversation.activeContext.isInActionLoop) {
return resolve(await this.handleActionLoop(utterance))
}
// When the active context has slots filled
if (Object.keys(this.conv.activeContext.slots).length > 0) {
if (Object.keys(this.conversation.activeContext.slots).length > 0) {
try {
return resolve(await this.handleSlotFilling(utterance))
} catch (e) {
@ -339,13 +335,13 @@ export default class NLU {
* If there are several "delete it" across skills, Leon needs to make use of
* the current context ({domain}.{skill}) to define the most accurate classification
*/
if (this.conv.hasActiveContext()) {
if (this.conversation.hasActiveContext()) {
classifications.forEach(({ intent: newIntent, score: newScore }) => {
if (newScore > 0.6) {
const [skillName] = newIntent.split('.')
const newDomain = MODEL_LOADER.mainNLPContainer.getIntentDomain(locale, newIntent)
const contextName = `${newDomain}.${skillName}`
if (this.conv.activeContext.name === contextName) {
if (this.conversation.activeContext.name === contextName) {
score = newScore
intent = newIntent
domain = newDomain
@ -356,7 +352,7 @@ export default class NLU {
const [skillName, actionName] = intent.split('.')
this.nluResult = {
...defaultNluResultObj, // Reset entities, slots, etc.
...DEFAULT_NLU_RESULT, // Reset entities, slots, etc.
utterance,
answers, // For dialog action type
classification: {
@ -452,8 +448,8 @@ export default class NLU {
// In case all slots have been filled in the first utterance
if (
this.conv.hasActiveContext() &&
Object.keys(this.conv.activeContext.slots).length > 0
this.conversation.hasActiveContext() &&
Object.keys(this.conversation.activeContext.slots).length > 0
) {
try {
return resolve(await this.handleSlotFilling(utterance))
@ -463,10 +459,10 @@ export default class NLU {
}
const newContextName = `${this.nluResult.classification.domain}.${skillName}`
if (this.conv.activeContext.name !== newContextName) {
this.conv.cleanActiveContext()
if (this.conversation.activeContext.name !== newContextName) {
this.conversation.cleanActiveContext()
}
this.conv.activeContext = {
this.conversation.activeContext = {
lang: BRAIN.lang,
slots: {},
isInActionLoop: false,
@ -479,17 +475,17 @@ export default class NLU {
}
// Pass current utterance entities to the NLU result object
this.nluResult.currentEntities =
this.conv.activeContext.currentEntities
this.conversation.activeContext.currentEntities
// Pass context entities to the NLU result object
this.nluResult.entities = this.conv.activeContext.entities
this.nluResult.entities = this.conversation.activeContext.entities
try {
const processedData = await BRAIN.execute(this.nluResult)
// Prepare next action if there is one queuing
if (processedData.nextAction) {
this.conv.cleanActiveContext()
this.conv.activeContext = {
this.conversation.cleanActiveContext()
this.conversation.activeContext = {
lang: BRAIN.lang,
slots: {},
isInActionLoop: !!processedData.nextAction.loop,
@ -528,11 +524,11 @@ export default class NLU {
* and ask for more entities if necessary
*/
async slotFill(utterance) {
if (!this.conv.activeContext.nextAction) {
if (!this.conversation.activeContext.nextAction) {
return null
}
const { domain, intent } = this.conv.activeContext
const { domain, intent } = this.conversation.activeContext
const [skillName, actionName] = intent.split('.')
const configDataFilePath = join(
process.cwd(),
@ -543,7 +539,7 @@ export default class NLU {
)
this.nluResult = {
...defaultNluResultObj, // Reset entities, slots, etc.
...DEFAULT_NLU_RESULT, // Reset entities, slots, etc.
utterance,
classification: {
domain,
@ -559,16 +555,16 @@ export default class NLU {
)
// Continue to loop for questions if a slot has been filled correctly
let notFilledSlot = this.conv.getNotFilledSlot()
let notFilledSlot = this.conversation.getNotFilledSlot()
if (notFilledSlot && entities.length > 0) {
const hasMatch = entities.some(
({ entity }) => entity === notFilledSlot.expectedEntity
)
if (hasMatch) {
this.conv.setSlots(BRAIN.lang, entities)
this.conversation.setSlots(BRAIN.lang, entities)
notFilledSlot = this.conv.getNotFilledSlot()
notFilledSlot = this.conversation.getNotFilledSlot()
if (notFilledSlot) {
BRAIN.talk(notFilledSlot.pickedQuestion)
SOCKET_SERVER.socket.emit('is-typing', false)
@ -578,31 +574,31 @@ export default class NLU {
}
}
if (!this.conv.areSlotsAllFilled()) {
if (!this.conversation.areSlotsAllFilled()) {
BRAIN.talk(`${BRAIN.wernicke('random_context_out_of_topic')}.`)
} else {
this.nluResult = {
...defaultNluResultObj, // Reset entities, slots, etc.
...DEFAULT_NLU_RESULT, // Reset entities, slots, etc.
// Assign slots only if there is a next action
slots: this.conv.activeContext.nextAction
? this.conv.activeContext.slots
slots: this.conversation.activeContext.nextAction
? this.conversation.activeContext.slots
: {},
utterance: this.conv.activeContext.originalUtterance,
utterance: this.conversation.activeContext.originalUtterance,
configDataFilePath,
classification: {
domain,
skill: skillName,
action: this.conv.activeContext.nextAction,
action: this.conversation.activeContext.nextAction,
confidence: 1
}
}
this.conv.cleanActiveContext()
this.conversation.cleanActiveContext()
return BRAIN.execute(this.nluResult)
}
this.conv.cleanActiveContext()
this.conversation.cleanActiveContext()
return null
}
@ -617,7 +613,7 @@ export default class NLU {
const hasMandatorySlots = Object.keys(slots)?.length > 0
if (hasMandatorySlots) {
this.conv.activeContext = {
this.conversation.activeContext = {
lang: BRAIN.lang,
slots,
isInActionLoop: false,
@ -629,7 +625,7 @@ export default class NLU {
entities: this.nluResult.entities
}
const notFilledSlot = this.conv.getNotFilledSlot()
const notFilledSlot = this.conversation.getNotFilledSlot()
// Loop for questions if a slot hasn't been filled
if (notFilledSlot) {
const { actions } = JSON.parse(