| Server IP : 153.92.12.164 / Your IP : 216.73.216.217 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 : /opt/gsutil/third_party/charset_normalizer/tests/ |
Upload File : |
import pytest
from charset_normalizer import from_bytes
from charset_normalizer.constant import TOO_BIG_SEQUENCE
def test_large_payload_u8_sig_basic_entry():
payload = ('0' * TOO_BIG_SEQUENCE).encode("utf_8_sig")
best_guess = from_bytes(payload).best()
assert best_guess is not None, "Large U8 payload case detection completely failed"
assert best_guess.encoding == "utf_8", "Large U8 payload case detection wrongly detected!"
assert best_guess.bom is True, "SIG/BOM property should be True"
assert len(best_guess.raw) == len(payload), "Large payload should remain untouched when accessed through .raw"
assert best_guess._string is not None, "str should be decoded before direct access (sig available)"
def test_large_payload_ascii_basic_entry():
payload = ('0' * TOO_BIG_SEQUENCE).encode("utf_8")
best_guess = from_bytes(payload).best()
assert best_guess is not None, "Large ASCII payload case detection completely failed"
assert best_guess.encoding == "ascii", "Large ASCII payload case detection wrongly detected!"
assert best_guess.bom is False, "SIG/BOM property should be False"
assert len(best_guess.raw) == len(payload), "Large payload should remain untouched when accessed through .raw"
assert best_guess._string is None, "str should not be decoded until direct access"
def test_misleading_large_sequence():
content = (("hello simple ascii " * TOO_BIG_SEQUENCE) + ('我没有埋怨，磋砣的只是一些时间。 磋砣的只是一些时间。')) .encode('utf_8')
guesses = from_bytes(content)
assert len(guesses) > 0
match = guesses.best()
assert match is not None
assert match._string is not None, "str should be cached as only match"
assert match.encoding == 'utf_8'
assert str(match) is not None