Marge the characters of the box file generated by jTessBoxEditor line by line [Tesseract]

Overview

It was very troublesome to merge characters in the box file generated by jTessBoxEditor, so I created it. I wasn't sure if marge would improve the accuracy, but I wrote it as a memorial service.

conditions

Create the following files. Paste this content into jTessBoxEditor to generate tiff files, box files, etc.

text.txt

letter
test
programming
mgs

program

This program regenerates a box file in which the character string defined in "text.txt" is merged line by line in the box file. Replace the generated file with the box file generated together with the tiff file, and specify it when training with jTessBoxEditor.

sample.py

import shutil

def read_words(input_file):
    with open(input_file, "r") as f:    
        data = [ t.replace("\n", "") for t in f.readlines()  ]  

def marge_box(input_file, words):
    with open(input_file, "r") as f:    
        data = [ t.replace("\n", "").split(" ") for t in f.readlines()  ]
        
        res = []
        start = 0
        for word in words:
            length = len(word)
            end = start + length
            tmp = data[start:end]
            print("word", word)
            print("start", start)
            print("end", end)
            print("length", length)
            
            a = [t[1] for t in tmp]
            b = [t[2] for t in tmp]
            c = [t[3] for t in tmp]
            d = [t[4] for t in tmp]
            e = [t[5] for t in tmp]
            a = a[0]
            b = min(b)
            c = max(c)
            d = max(d)
            e = e[0]
            print("abcde", [a, b, c, d, e])
            res.append([word, a, b, c, d, e])
            start += length
        
        shutil.copy(input_file, "copy_" + input_file) #Copy the original file as a backup
    
    with open(input_file, "w", encoding='utf-8') as f:  
        for t in res:
            print(" ".join(t), file=f)

if __name__ == "__main__":
    words = read_words("text.txt")
    marge_box("~~.box", words) #Specify the box file generated by jTessBoxEditor