added tesseract package, trying to recognize text from base64 image

src/utils/classes.ts

@@ -21,6 +21,7 @@
 
 import { isMainThread, parentPort, workerData } from 'worker_threads'
 
+import { recognizeTextFromBase64, tesseractLoaded } from './tesseract'
 import logger from './logger'
 import {
   Question,
@@ -114,7 +115,10 @@ function normalizeSpaces(input: string): string {
 }
 
 function removeUnnecesarySpaces(toremove: string): string {
-  return normalizeSpaces(toremove).replace(/\s+/g, ' ')
+  return normalizeSpaces(toremove)
+    .replace(/\s+/g, ' ')
+    .replace(/(\r\n|\n|\r)/gm, '')
+    .trim()
 }
 
 function compareString(
@@ -278,6 +282,34 @@ function createQuestion(
   }
 }
 
+async function recognizeQuestionImage(question: Question): Promise<Question> {
+  const base64Data = question.data.base64
+  if (Array.isArray(base64Data) && base64Data.length) {
+    try {
+      const res: string[] = []
+      for (let i = 0; i < base64Data.length; i++) {
+        const base64 = base64Data[i]
+        const text = await recognizeTextFromBase64(base64)
+        res.push(text)
+      }
+
+      return {
+        ...question,
+        Q: res.join(' '),
+        data: {
+          ...question.data,
+          type: 'simple',
+        },
+      }
+    } catch (e) {
+      console.error('Error happened in recognizeQuestionImage!')
+      console.error(e)
+    }
+  }
+
+  return question
+}
+
 function compareImage(data: QuestionData, data2: QuestionData): number {
   if (data.hashedImages && data2.hashedImages) {
     return compareString(
@@ -643,6 +675,10 @@ interface WorkData {
 }
 
 if (!isMainThread) {
+  handleWorkerData()
+}
+
+function handleWorkerData() {
   const {
     workerIndex,
     initData,
@@ -653,11 +689,12 @@ if (!isMainThread) {
     `[THREAD #${workerIndex}]: Worker ${workerIndex} reporting for duty`
   )
 
-  parentPort.on('message', (msg /*: TaskObject */) => {
+  parentPort.on('message', async (msg /*: TaskObject */) => {
+    await tesseractLoaded
     if (msg.type === 'work') {
       const {
         subjName,
-        question,
+        question: originalQuestion,
         searchTillMatchPercent,
         searchInAllIfNoResult,
         searchIn,
@@ -667,6 +704,8 @@ if (!isMainThread) {
       let searchResult: SearchResultQuestion[] = []
       let error = false
 
+      const question = await recognizeQuestionImage(originalQuestion)
+
       try {
         qdbs.forEach((qdb) => {
           if (searchIn.includes(qdb.index)) {
@@ -857,8 +896,6 @@ if (!isMainThread) {
       })
     }
   })
-} else {
-  // console.log('[THREAD]: Main thread!')
 }
 
 export function cleanDb(
@@ -917,4 +954,5 @@ export {
   dataToString,
   doSearch,
   setNoPossibleAnswersPenalties,
+  recognizeQuestionImage,
 }