From ea4b6e7d109ae3abc7f8ecdf99e3eb33e3484b77 Mon Sep 17 00:00:00 2001
From: David Robillard <d@drobilla.net>
Date: Sun, 5 Feb 2023 12:42:52 -0500
Subject: Strengthen handling of corrupt UTF-8 input

---
 test/extra/lax/test-bad-utf8.ttl | 66 ++++++++++++++++++++++++++++++++++++----
 1 file changed, 60 insertions(+), 6 deletions(-)

(limited to 'test/extra/lax/test-bad-utf8.ttl')

diff --git a/test/extra/lax/test-bad-utf8.ttl b/test/extra/lax/test-bad-utf8.ttl
index 0e177366..e5640078 100644
--- a/test/extra/lax/test-bad-utf8.ttl
+++ b/test/extra/lax/test-bad-utf8.ttl
@@ -1,6 +1,60 @@
-<http://example.org/s> <http://example.org/p> "Impossible bytes: � �" .
-<http://example.org/s> <http://example.org/p> "2 continuation bytes: ��" .
-<http://example.org/s> <http://example.org/p> "Missing continuation: �" .
-<http://example.org/s> <http://example.org/p> """Impossible bytes: � �""" .
-<http://example.org/s> <http://example.org/p> """2 continuation bytes: ��""" .
-<http://example.org/s> <http://example.org/p> """Missing continuation: �""" .
+<http://example.org/s> <http://example.org/p> "The other values of this property should align nicely" .
+<http://example.org/s> <http://example.org/p> "Impossible byte 1:                                   �                               |" .
+<http://example.org/s> <http://example.org/p> "Impossible byte 2:                                   �                               |" .
+<http://example.org/s> <http://example.org/p> "Four impossible bytes:                               ����                            |" .
+<http://example.org/s> <http://example.org/p> "First continuation byte:                             �                               |" .
+<http://example.org/s> <http://example.org/p> "Last continuation byte:                              �                               |" .
+<http://example.org/s> <http://example.org/p> "2 continuation bytes:                                ��                              |" .
+<http://example.org/s> <http://example.org/p> "3 continuation bytes:                                ���                             |" .
+<http://example.org/s> <http://example.org/p> "4 continuation bytes:                                ����                            |" .
+<http://example.org/s> <http://example.org/p> "5 continuation bytes:                                �����                           |" .
+<http://example.org/s> <http://example.org/p> "6 continuation bytes:                                ������                          |" .
+<http://example.org/s> <http://example.org/p> "7 continuation bytes:                                �������                         |" .
+<http://example.org/s> <http://example.org/p> "Continuation bytes 1:                                ����������������                |" .
+<http://example.org/s> <http://example.org/p> "Continuation bytes 2:                                ����������������                |" .
+<http://example.org/s> <http://example.org/p> "Continuation bytes 3:                                ����������������                |" .
+<http://example.org/s> <http://example.org/p> "Continuation bytes 4:                                ����������������                |" .
+<http://example.org/s> <http://example.org/p> "Lonely leading bytes of 2-byte sequences 1:          � � � � � � � � � � � � � � � � |" .
+<http://example.org/s> <http://example.org/p> "Lonely leading bytes of 2-byte sequences 2:          � � � � � � � � � � � � � � � � |" .
+<http://example.org/s> <http://example.org/p> "Lonely leading bytes of 3-byte sequences:            � � � � � � � � � � � � � � � � |" .
+<http://example.org/s> <http://example.org/p> "Lonely leading bytes of 4-byte sequences:            � � � � � � � �                 |" .
+<http://example.org/s> <http://example.org/p> "Lonely leading bytes of 5-byte sequences:            � � � �                         |" .
+<http://example.org/s> <http://example.org/p> "Lonely leading bytes of 6-byte sequences:            � �                             |" .
+<http://example.org/s> <http://example.org/p> "2-byte sequence with last byte missing (U+0000):     �                               |" .
+<http://example.org/s> <http://example.org/p> "3-byte sequence with last byte missing (U+0000):     ��                               |" .
+<http://example.org/s> <http://example.org/p> "4-byte sequence with last byte missing (U+0000):     ���                               |" .
+<http://example.org/s> <http://example.org/p> "2-byte sequence with last byte missing (U-000007FF): �                               |" .
+<http://example.org/s> <http://example.org/p> "3-byte sequence with last byte missing (U-0000FFFF): �                               |" .
+<http://example.org/s> <http://example.org/p> "4-byte sequence with last byte missing (U-001FFFFF): ���                               |" .
+<http://example.org/s> <http://example.org/p> "6 sequences with last byte missing:                  �����������                          |" .
+
+<http://example.org/s> <http://example.org/p> """
+Impossible byte 1:                                   �                               |
+Impossible byte 2:                                   �                               |
+Four impossible bytes:                               ����                            |
+First continuation byte:                             �                               |
+Last continuation byte:                              �                               |
+2 continuation bytes:                                ��                              |
+3 continuation bytes:                                ���                             |
+4 continuation bytes:                                ����                            |
+5 continuation bytes:                                �����                           |
+6 continuation bytes:                                ������                          |
+7 continuation bytes:                                �������                         |
+Continuation bytes 1:                                ����������������                |
+Continuation bytes 2:                                ����������������                |
+Continuation bytes 3:                                ����������������                |
+Continuation bytes 4:                                ����������������                |
+Lonely leading bytes of 2-byte sequences 1:          � � � � � � � � � � � � � � � � |
+Lonely leading bytes of 2-byte sequences 2:          � � � � � � � � � � � � � � � � |
+Lonely leading bytes of 3-byte sequences:            � � � � � � � � � � � � � � � � |
+Lonely leading bytes of 4-byte sequences:            � � � � � � � �                 |
+Lonely leading bytes of 5-byte sequences:            � � � �                         |
+Lonely leading bytes of 6-byte sequences:            � �                             |
+2-byte sequence with last byte missing (U+0000):     �                               |
+3-byte sequence with last byte missing (U+0000):     ��                               |
+4-byte sequence with last byte missing (U+0000):     ���                               |
+2-byte sequence with last byte missing (U-000007FF): �                               |
+3-byte sequence with last byte missing (U-0000FFFF): �                               |
+4-byte sequence with last byte missing (U-001FFFFF): ���                               |
+6 sequences with last byte missing:                  �����������                          |
+""" .
-- 
cgit v1.2.1