From 3fd55b938294d5714e80a813474501b750b8413d Mon Sep 17 00:00:00 2001
From: Sylvain Perron <slvn.perron+github@gmail.com>
Date: Tue, 30 Jul 2019 13:37:08 -0400
Subject: [PATCH] fix(nlu): known slots infinite loop

---
 modules/nlu/src/backend/engine.ts             | 18 ++++------
 .../backend/pipelines/slots/pre-processor.ts  | 36 +++++++------------
 2 files changed, 19 insertions(+), 35 deletions(-)

diff --git a/modules/nlu/src/backend/engine.ts b/modules/nlu/src/backend/engine.ts
index b9ae9e5f70f..a1413c84f9d 100644
--- a/modules/nlu/src/backend/engine.ts
+++ b/modules/nlu/src/backend/engine.ts
@@ -18,8 +18,8 @@ import { sanitize } from './pipelines/language/sanitizer'
 import CRFExtractor from './pipelines/slots/crf_extractor'
 import { generateTrainingSequence } from './pipelines/slots/pre-processor'
 import Storage from './storage'
-import { makeTokens } from './tools/token-utils'
 import { allInRange } from './tools/math'
+import { makeTokens } from './tools/token-utils'
 import { LanguageProvider, NluMlRecommendations, TrainingSequence } from './typings'
 import {
   Engine,
@@ -248,20 +248,14 @@ export default class ScopedEngine implements Engine {
       .value()
 
   private getTrainingSets = async (intentDefs: sdk.NLU.IntentDefinition[], lang: string): Promise<TrainingSequence[]> =>
-    await Promise.all(
-      _.chain(intentDefs)
-        .flatMap(await this.generateTrainingSequenceFromIntent(lang))
-        .value()
-    ).reduce(this.flatMapIdendity, [])
+    _.flatten(await Promise.map(intentDefs, intentDef => this.generateTrainingSequenceFromIntent(lang, intentDef)))
 
-  private generateTrainingSequenceFromIntent = (lang: string) => async (
+  private generateTrainingSequenceFromIntent = async (
+    lang: string,
     intent: sdk.NLU.IntentDefinition
   ): Promise<TrainingSequence[]> =>
-    Promise.all(
-      (intent.utterances[lang] || []).map(
-        async utterance =>
-          await this.scopedGenerateTrainingSequence(utterance, lang, intent.slots, intent.name, intent.contexts)
-      )
+    await Promise.map(intent.utterances[lang] || [], utterance =>
+      this.scopedGenerateTrainingSequence(utterance, lang, intent.slots, intent.name, intent.contexts)
     )
 
   protected async loadModels(intents: sdk.NLU.IntentDefinition[], modelHash: string) {
diff --git a/modules/nlu/src/backend/pipelines/slots/pre-processor.ts b/modules/nlu/src/backend/pipelines/slots/pre-processor.ts
index 47e90e4ef2e..3913d0f19c2 100644
--- a/modules/nlu/src/backend/pipelines/slots/pre-processor.ts
+++ b/modules/nlu/src/backend/pipelines/slots/pre-processor.ts
@@ -1,9 +1,9 @@
 import * as sdk from 'botpress/sdk'
 import _ from 'lodash'
 
-import { makeTokens, mergeSpecialCharactersTokens } from '../../tools/token-utils'
 import { allInRange } from '../../tools/math'
-import { LanguageProvider, KnownSlot, TrainingSequence } from '../../typings'
+import { makeTokens, mergeSpecialCharactersTokens } from '../../tools/token-utils'
+import { KnownSlot, LanguageProvider, TrainingSequence } from '../../typings'
 import { BIO, Sequence, Token } from '../../typings'
 
 const ALL_SLOTS_REGEX = /\[(.+?)\]\(([\w_\.-]+)\)/gi
@@ -19,30 +19,20 @@ export function keepEntityValues(text: string): string {
 
 export function getKnownSlots(text: string, slotDefinitions: sdk.NLU.SlotDefinition[]): KnownSlot[] {
   const slots = [] as KnownSlot[]
+  const localSlotsRegex = /\[(.+?)\]\(([\w_\.-]+)\)/gi // local because it is stateful
 
   let regResult: RegExpExecArray | null
-  let cursor = 0
-  do {
-    const textCpy = text.substring(cursor)
-    regResult = ITTERATIVE_SLOTS_REGEX.exec(textCpy)
-    if (regResult) {
-      const rawMatch = regResult[0]
-      const source = regResult[1] as string
-      const slotName = regResult[2] as string
-
-      const previousSlot = _.last(slots)
-      const start = (previousSlot ? previousSlot.end : 0) + regResult.index
-      const end = start + source.length
-
-      cursor = start + rawMatch.length
-
-      const slotDef = slotDefinitions.find(sd => sd.name === slotName)
-
-      if (slotDef) {
-        slots.push({ ...slotDef, start, end, source })
-      }
+  while ((regResult = localSlotsRegex.exec(text))) {
+    const rawMatch = regResult[0]
+    const source = regResult[1] as string
+    const slotName = regResult[2] as string
+
+    const slotDef = slotDefinitions.find(sd => sd.name === slotName)
+
+    if (slotDef) {
+      slots.push({ ...slotDef, start: regResult.index, end: regResult.index + rawMatch.length, source })
     }
-  } while (regResult)
+  }
 
   return slots
 }