I need to merge text of 2 lines depending on their x1,x2,y1 and y2 values - ocr

Question

I am working on HOCR format of tesseract OCR, i am stuck when i have a text based on multiple lines. HOCR Format contains bounding box for every word and multi-line text issue can be solve if i merge all the texts of words if they have same x1.
I am using java script for this task.
This is sample image i am using, due to privacy issues i have masked the confidential data.
Here is my code, i know its not well written so i am open to any cretic.
//merging 2 lines(currentLine, previousLine)
function merge2lines(ocrLine, ocrPrevLine)
{
var stringOutCome="";
var stringOutComeOuterLine="";
//get all the words in this line
var lineItems = $(ocrLine).children();
var CoordinatesOfItemForFirstLineOfSecondLine='';
var textOfItemForFirstLineOfSecondLine='';
var prevLineItems = $(ocrPrevLine).children();
//loop over all words
var lineItemsCount = lineItems.length;
for (var i=0;i<lineItemsCount;i++)
{
var stringInfo = lineItems[i].textContent;
if (typeof stringInfo === 'string' && stringInfo.trim().length > 0)
{
var str = lineItems[i].title;
var title = (str).split(";");
var firstSpace = title[0].indexOf(" ");
var newStr = title[0].slice(firstSpace);
var coordinates = newStr.split(' ');
if(i==0){
CoordinatesOfItemForFirstWordOfSecondLine = coordinates;
textOfItemForFirstWordOfSecondLine = lineItems[i].textContent;
}
//check and remove text
if(! stringOutComeOuterLine.includes(lineItems[i].textContent))
{
stringOutComeOuterLine = stringOutComeOuterLine+" "+lineItems[i].textContent;
}
// stringOutComeOuterLine = stringOutComeOuterLine+" "+lineItems[i].textContent;
if(i<lineItemsCount-1)
{
var str2 = lineItems[i+1].title;
var title2 = (str2).split(";");
var firstSpace2 = title2[0].indexOf(" ");
var newStr2 = title2[0].slice(firstSpace2);
var coordinates2 = newStr2.split(' ');
var differenceOfCordinates = Math.abs(coordinates[3] - coordinates2[1]);
// console.log(lineItems[i].textContent,+" pakistan <- -> Block B",lineItems[i+1].textContent," DIfference: ",differenceOfCordinates);
if(parseInt(differenceOfCordinates) < 100)
{
stringOutComeOuterLine = stringOutComeOuterLine+" "+lineItems[i+1].textContent;
}
else{
var check = true;
// check previousLineitem x
for (var j=0;j<prevLineItems.length;j++)
{
var prevString = prevLineItems[j].textContent;
if (typeof prevString === 'string' && prevString.trim().length > 0)
{
var previousLineTitle = prevLineItems[j].title;
var splittedPreviousLineTitle = (previousLineTitle).split(";");
var previousLineItemFirstSpace = splittedPreviousLineTitle[0].indexOf(" ");
var previousLineItemCoordinates = splittedPreviousLineTitle[0].slice(previousLineItemFirstSpace);
var previousLineItemCoordinatesList = previousLineItemCoordinates.split(' ');
var stringDifference = Math.abs(previousLineItemCoordinatesList[1] - CoordinatesOfItemForFirstWordOfSecondLine[1]);
// console.log(prevLineItems[j].textContent,"->",textOfItemForFirstWordOfSecondLine,"Diff:", stringDifference);
if(stringDifference < 20 && check == true)
{
check = false;
// stringOutComeOuterLine = prevLineItems[j].textContent +" "+ stringOutComeOuterLine;
stringOutCome = stringOutCome+" "+prevLineItems[j].textContent;
}
else{
if(j<prevLineItems.length-1)
{
console.log(prevLineItems[j+1].textContent);
var previousLineTitleNext = prevLineItems[j+1].title;
var splittedPreviousLineTitleNext = (previousLineTitleNext).split(";");
var previousLineItemFirstSpaceNext = splittedPreviousLineTitleNext[0].indexOf(" ");
var previousLineItemCoordinatesNext = splittedPreviousLineTitleNext[0].slice(previousLineItemFirstSpaceNext);
var previousLineItemCoordinatesListNext = previousLineItemCoordinatesNext.split(' ');
var previousLineDiff = Math.abs(previousLineItemCoordinatesList[3] - previousLineItemCoordinatesListNext[1]);
// console.log("Previous Item",prevLineItems[j].textContent,"->Next Item:",prevLineItems[j+1].textContent,"previousLineDiff:", previousLineDiff);
// console.log("<br> left coordinat: ",previousLineItemCoordinatesList[3]," Right Item: ",previousLineItemCoordinatesListNext[1]);
if(previousLineDiff < 120){
stringOutCome = stringOutCome+" "+prevLineItems[j].textContent;//+" "+prevLineItems[j+1].textContent;
// console.log(stringOutCome,"previousLineDiff:", previousLineDiff);
}
else{
// stringOutComeOuterLine = prevLineItems[j].textContent +" "+ stringOutComeOuterLine;
stringOutCome = stringOutCome +" "+stringOutComeOuterLine;
}
}else{
// console.log("Else Part");
stringOutComeOuterLine = stringOutCome +" "+stringOutComeOuterLine;
}
}
// console.log(stringOutComeOuterLine);
}
}
}
}else{
// stringOutComeOuterLine = lineItems[i].textContent+" "+stringOutComeOuterLine;
// check previousLineitem x
for (var j=0;j<prevLineItems.length;j++)
{
var prevString = prevLineItems[j].textContent;
if (typeof prevString === 'string' && prevString.trim().length > 0)
{
var previousLineTitle = prevLineItems[j].title;
var splittedPreviousLineTitle = (previousLineTitle).split(";");
var previousLineItemFirstSpace = splittedPreviousLineTitle[0].indexOf(" ");
var previousLineItemCoordinates = splittedPreviousLineTitle[0].slice(previousLineItemFirstSpace);
var previousLineItemCoordinatesList = previousLineItemCoordinates.split(' ');
// console.log(prevLineItems[j].textContent,"->",textOfItemForFirstWordOfSecondLine,"Diff:", stringDifference);
if(j<prevLineItems.length-1)
{
console.log(prevLineItems[j+1].textContent);
var previousLineTitleNext = prevLineItems[j+1].title;
var splittedPreviousLineTitleNext = (previousLineTitleNext).split(";");
var previousLineItemFirstSpaceNext = splittedPreviousLineTitleNext[0].indexOf(" ");
var previousLineItemCoordinatesNext = splittedPreviousLineTitleNext[0].slice(previousLineItemFirstSpaceNext);
var previousLineItemCoordinatesListNext = previousLineItemCoordinatesNext.split(' ');
var previousLineDiff = Math.abs(previousLineItemCoordinatesList[3] - previousLineItemCoordinatesListNext[1]);
// console.log("Previous Item",prevLineItems[j].textContent,"->Next Item:",prevLineItems[j+1].textContent,"previousLineDiff:", previousLineDiff);
// console.log("<br> left coordinat: ",previousLineItemCoordinatesList[3]," Right Item: ",previousLineItemCoordinatesListNext[1]);
if(previousLineDiff < 120){
stringOutCome = stringOutCome+" "+prevLineItems[j].textContent;//+" "+prevLineItems[j+1].textContent;
// console.log(stringOutCome,"previousLineDiff:", previousLineDiff);
}
else{
// stringOutComeOuterLine = prevLineItems[j].textContent +" "+ stringOutComeOuterLine;
stringOutCome = stringOutCome +" "+stringOutComeOuterLine;
}
}else{
// console.log("Else Part");
stringOutComeOuterLine = stringOutCome +" "+stringOutComeOuterLine;
}
// console.log(stringOutComeOuterLine);
}
}
}
}
}
console.log(stringOutComeOuterLine);
return stringOutComeOuterLine;
}