"""
This is used when you want to copy and paste a paper from pdf and translate it into Google.
.Given the path of txt_conv.Since it will be output in html, open it in chrome and right-click to translate.

・ What you are doing
ctrl in pdf+If you paste it with A, line breaks will be wasted and it will not translate well.
Avoid it.
"""

import sys
import os

filepath = sys.argv[1]

# ---Process text for Google Translate
with open(filepath, "r", encoding="utf-8") as f:

    lines = f.readlines()
    newlines = []
    lenbuf = []  #line len buffer
    for line in lines:
        if len(line) >= 2:
            if line[-2] == ".":  #Insert a line break if the end is a period
                line = line[:-1] + "<br>"
            else:  #If not, put a space
                line = line[:-1] + " "
        newlines.append(line)
        lenbuf.append(len(line))
f.close()

# ---If it is better to insert a line break, insert a line break(average line length*0.Set 8 as the threshold)
ave_len = sum(lenbuf) / len(lenbuf)
for n, linelen in enumerate(lenbuf):
    if linelen < ave_len * 0.8:
        newlines[n] = newlines[n] + "<br>"

# ---_Save with conv
savepath = os.path.splitext(filepath)[0] + "_conv.html"

with open(savepath, "w", encoding="utf-8") as f:
    f.writelines(newlines)

Recommended Posts