| 1 |
nino.borges |
890 |
"""
|
| 2 |
|
|
|
| 3 |
|
|
FileEncodingLib
|
| 4 |
|
|
|
| 5 |
|
|
Created by:
|
| 6 |
|
|
Emanuel Borges
|
| 7 |
|
|
07.12.2019
|
| 8 |
|
|
|
| 9 |
|
|
A set of functions that will help with file encoding detection and issues.
|
| 10 |
|
|
This is version 2 of this library which **requires carset-normalizer be installed**. Previously used chardet.
|
| 11 |
|
|
|
| 12 |
|
|
"""
|
| 13 |
|
|
|
| 14 |
|
|
|
| 15 |
|
|
import os, sys
|
| 16 |
|
|
from charset_normalizer import from_path
|
| 17 |
|
|
|
| 18 |
|
|
|
| 19 |
|
|
def DetectFileEncoding(fileName):
|
| 20 |
|
|
"""Attempts to detect the file encoding. Returns encoding and prints the confidence level. fileName should also include the full path to the file."""
|
| 21 |
|
|
|
| 22 |
|
|
## First verify if charset_normalizer is installed.
|
| 23 |
|
|
if 'charset_normalizer' in sys.modules:
|
| 24 |
|
|
pass
|
| 25 |
|
|
else:
|
| 26 |
|
|
print("charset_normalizer is not intalled. Please install with PIP before using.")
|
| 27 |
|
|
return None
|
| 28 |
|
|
|
| 29 |
|
|
## Attempt to determin encoding.
|
| 30 |
|
|
try:
|
| 31 |
|
|
testResult = from_path(fileName).best()
|
| 32 |
|
|
if testResult:
|
| 33 |
|
|
fileEncoding = testResult.encoding
|
| 34 |
|
|
print(f"File appears to have the encoding {fileEncoding} (Confidence: {testResult.encoding_aliases})")
|
| 35 |
|
|
return fileEncoding
|
| 36 |
|
|
else:
|
| 37 |
|
|
print("FAILED: Unable to detect encoding from this file. Suggest trying common encodings 'utf-8' or 'latin1'.")
|
| 38 |
|
|
return None
|
| 39 |
|
|
except Exception as e:
|
| 40 |
|
|
print(f"Other error occurred: {e}")
|
| 41 |
|
|
return None
|
| 42 |
|
|
|
