Duplicate remover: managing possible questions

src/standaloneUtils/rmDuplicates.js

@@ -1,7 +1,13 @@
 const utils = require('../../dist/utils/utils.js').default // eslint-disable-line
 const logger = require('../../dist/utils/logger.js').default // eslint-disable-line
-const { addQuestion, doSearch } = require('../../dist/utils/classes.js') // eslint-disable-line
+const {
+  addQuestion,
+  doSearch,
+  compareQuestionObj,
+  createQuestion,
+} = require('../../dist/utils/classes.js') // eslint-disable-line
 const { loadData, writeData } = require('../../dist/utils/actions.js') // eslint-disable-line
+const fs = require('fs') // eslint-disable-line
 
 // Params [ 'publicDirs/qminingPublic/questionDbs/elektro.json' ]
 // load: 2.767ms
@@ -38,17 +44,33 @@ utils.WriteFile('', globalLog)
 
 const params = process.argv.splice(2)
 
-const fileA = params[0]
-const fileB = params[1]
+const pathA = params[0]
+const pathB = params[1]
 
+const stat = fs.lstatSync(pathA)
+if (stat.isDirectory()) {
+  if (pathB) {
+    log(
+      `Clearing possible questions from ${C(
+        'green'
+      )}${pathA}${C()} based on ${C('green')}${pathB}${C()} db`
+    )
+    const db = pathB ? loadData(pathB) : null
+
+    clearPossibleAnswers(pathA, db)
+  } else {
+    removePossibleAnswersDuplicates(pathA)
+  }
+} else {
   console.time('load')
-const dbA = loadData(fileA)
-const dbB = fileB ? loadData(fileB) : null
+  const dbA = loadData(pathA)
+  const dbB = pathB ? loadData(pathB) : null
   console.timeEnd('load')
 
   console.time('rmduplicates')
+
   if (!dbB) {
-  log(`Removing duplicate questions from ${fileA}`)
+    log(`Removing duplicate questions from ${C('green')}${pathA}${C()}`)
     rmDuplicates(dbA).then((res) => {
       console.timeEnd('rmduplicates')
       writeData(res, resultDbFileName)
@@ -56,9 +78,9 @@ if (!dbB) {
     })
   } else {
     log(
-    `Removing questions found in ${C('green')}${fileB}${C()} from ${C(
+      `Removing questions found in ${C('green')}${pathB}${C()} from ${C(
         'green'
-    )}${fileA}${C()}`
+      )}${pathA}${C()}`
     )
     difference({ dbA: dbA, dbB: dbB }).then((res) => {
       console.timeEnd('rmduplicates')
@@ -66,12 +88,105 @@ if (!dbB) {
       log('File written')
     })
   }
-
-async function rmDuplicates(db) {
-  return await difference({ dbA: db })
 }
 
-async function difference({ dbA, dbB }) {
+// ---------------------------------------------------------------------------------
+// possible answers duplicate removing
+// ---------------------------------------------------------------------------------
+
+function removePossibleAnswersDuplicates(path) {
+  let count = 0
+  let currIndex = 1
+  let delets = 0
+
+  iterateDir(path, () => {
+    count++
+  })
+
+  iterateDir(path, (currPath) => {
+    currIndex++
+    if (currPath.includes('savedQuestions.json')) {
+      return
+    }
+    if (!utils.FileExists(currPath)) {
+      return
+    }
+    const currData = utils.ReadJSON(currPath)
+    currData.questions.forEach((q1) => {
+      iterateDir(path, (currPath2) => {
+        if (currPath === currPath2) {
+          return
+        }
+        if (currPath2.includes('savedQuestions.json')) {
+          return
+        }
+        if (!utils.FileExists(currPath)) {
+          return
+        }
+        const dataB = utils.ReadJSON(currPath2)
+
+        dataB.questions.forEach((q2) => {
+          const percent = compareQuestionObj(
+            createQuestion(q1),
+            '',
+            createQuestion(q2),
+            ''
+          )
+          if (percent.avg === 100) {
+            utils.deleteFile(currPath2)
+            count--
+            delets++
+          }
+        })
+      })
+    })
+
+    printProgressBar(currIndex, count)
+  })
+  log(`Deleted ${C('green')}${delets}${C()} files`)
+}
+
+function clearPossibleAnswers(path, db) {
+  let count = 0
+  let currIndex = 1
+  let delets = 0
+  iterateDir(path, () => {
+    count++
+  })
+
+  iterateDir(path, (currPath) => {
+    currIndex++
+    if (currPath.includes('savedQuestions.json')) {
+      return
+    }
+    const { subj, questions } = utils.ReadJSON(currPath)
+
+    questions.forEach((question) => {
+      const searchRes = search({
+        qdb: db,
+        subjName: subj,
+        question: question,
+        searchTillMatchPercent: 80,
+      })
+      if (searchRes.length > 0) {
+        utils.deleteFile(currPath)
+        delets++
+      }
+    })
+    printProgressBar(currIndex, count)
+  })
+  log(`Deleted ${C('green')}${delets}${C()} files`)
+}
+
+// ---------------------------------------------------------------------------------
+// difference
+// ---------------------------------------------------------------------------------
+
+function rmDuplicates(db) {
+  return difference({ dbA: db })
+}
+
+function difference({ dbA, dbB }) {
   const doingDifference = !!dbB
   // Stuff only from A
   const resultDb = []
@@ -105,7 +220,7 @@ async function difference({ dbA, dbB }) {
 
     for (let j = 0; j < subj.Questions.length; j++) {
       const question = subj.Questions[j]
-      const searchRes = await search({
+      const searchRes = search({
         qdb: doingDifference ? dbB : resultDb,
         subjName: subj.Name,
         question: question,
@@ -167,11 +282,14 @@ function hasRequiredPercent(result, minpercent) {
 // ---------------------------------------------------------------------------------
 
 function search({ qdb, subjName, question, searchInAllIfNoResult }) {
-  return new Promise((resolve) => {
-    resolve(
-      doSearch(qdb, subjName, question, null, minpercent, searchInAllIfNoResult)
+  return doSearch(
+    qdb,
+    subjName,
+    question,
+    null,
+    minpercent,
+    searchInAllIfNoResult
   )
-  })
 }
 
 function iterateSubjects(db, fn) {
@@ -182,6 +300,26 @@ function iterateSubjects(db, fn) {
   })
 }
 
+// ---------------------------------------------------------------------------------
+// possible answers tools
+// ---------------------------------------------------------------------------------
+
+function iterateDir(path, action) {
+  if (!utils.FileExists(path)) {
+    return
+  }
+
+  const stat = fs.lstatSync(path)
+  if (stat.isDirectory()) {
+    const content = fs.readdirSync(path)
+    content.forEach((currContent) => {
+      iterateDir(`${path}/${currContent}`, action)
+    })
+  } else {
+    action(path)
+  }
+}
+
 // ---------------------------------------------------------------------------------
 // logging and tools
 // ---------------------------------------------------------------------------------

src/utils/classes.ts

@@ -302,7 +302,7 @@ function compareQuestionObj(
   q2: Question,
   q2subjName: string,
   data: QuestionData
-) {
+): any {
   assert(data !== undefined || data !== null)
   assert(q1)
   assert(typeof q1 === 'object')
@@ -678,6 +678,7 @@ if (!isMainThread) {
 // ------------------------------------------------------------------------
 
 export {
+  compareQuestionObj,
   minMatchAmmount,
   getSubjNameWithoutYear,
   createQuestion,

submodules/qmining-data-editor

@@ -1 +1 @@
-Subproject commit 33e8b3a49e7ddbf5c52721c51e655dc28b6ff877
+Subproject commit 1446a57c28072384b471fcad71e8c637c98ff207