| Server IP : 153.92.12.235 / Your IP : 216.73.217.6 Web Server : LiteSpeed System : Linux id-dci-web1986.main-hosting.eu 5.14.0-611.26.1.el9_7.x86_64 #1 SMP PREEMPT_DYNAMIC Thu Jan 29 05:24:47 EST 2026 x86_64 User : u686484674 ( 686484674) PHP Version : 8.0.30 Disable Function : system, exec, shell_exec, passthru, mysql_list_dbs, ini_alter, dl, symlink, link, chgrp, leak, popen, apache_child_terminate, virtual, mb_send_mail MySQL : OFF | cURL : ON | WGET : ON | Perl : OFF | Python : OFF | Sudo : OFF | Pkexec : OFF Directory : /proc/self/root/opt/gsutil/third_party/charset_normalizer/tests/ |
Upload File : |
from charset_normalizer.api import from_path
import pytest
from os import path, pardir
DIR_PATH = path.join(
path.dirname(path.realpath(__file__)),
pardir
)
@pytest.mark.parametrize(
"input_data_file, expected_charset, expected_language",
[
('sample-arabic-1.txt', 'cp1256', 'Arabic'),
('sample-french-1.txt', 'cp1252', 'French'),
('sample-arabic.txt', 'utf_8', 'Arabic'),
('sample-russian-3.txt', 'utf_8', 'Russian'),
('sample-french.txt', 'utf_8', 'French'),
('sample-chinese.txt', 'big5', 'Chinese'),
('sample-greek.txt', 'cp1253', 'Greek'),
('sample-greek-2.txt', 'cp1253', 'Greek'),
('sample-hebrew-2.txt', 'cp1255', 'Hebrew'),
('sample-hebrew-3.txt', 'cp1255', 'Hebrew'),
('sample-bulgarian.txt', 'utf_8', 'Bulgarian'),
('sample-english.bom.txt', 'utf_8', 'English'),
('sample-spanish.txt', 'utf_8', 'Spanish'),
('sample-korean.txt', 'cp949', 'Korean'),
('sample-turkish.txt', 'cp1254', 'Turkish'),
('sample-russian-2.txt', 'utf_8', 'Russian'),
('sample-russian.txt', 'mac_cyrillic', 'Russian'),
('sample-polish.txt', 'utf_8', 'Polish'),
]
)
def test_elementary_detection(
input_data_file: str,
expected_charset: str,
expected_language: str,
):
best_guess = from_path(DIR_PATH + "/data/{}".format(input_data_file)).best()
assert best_guess is not None, "Elementary detection has failed upon '{}'".format(input_data_file)
assert best_guess.encoding == expected_charset, "Elementary charset detection has failed upon '{}'".format(input_data_file)
assert best_guess.language == expected_language, "Elementary language detection has failed upon '{}'".format(input_data_file)