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-uri-nq-out.nq   |  2 +-
 test/extra/lax/test-bad-uri-out.nt      |  2 +-
 test/extra/lax/test-bad-utf8-nq-out.nq  | 32 ++++++++++++++--
 test/extra/lax/test-bad-utf8-nt-out.nt  | 32 ++++++++++++++--
 test/extra/lax/test-bad-utf8-ttl-out.nt | 36 +++++++++++++++---
 test/extra/lax/test-bad-utf8.nq         | 32 ++++++++++++++--
 test/extra/lax/test-bad-utf8.nt         | 32 ++++++++++++++--
 test/extra/lax/test-bad-utf8.ttl        | 66 ++++++++++++++++++++++++++++++---
 8 files changed, 208 insertions(+), 26 deletions(-)

(limited to 'test')

diff --git a/test/extra/lax/test-bad-uri-nq-out.nq b/test/extra/lax/test-bad-uri-nq-out.nq
index 8cb00ba7..66da9b0d 100644
--- a/test/extra/lax/test-bad-uri-nq-out.nq
+++ b/test/extra/lax/test-bad-uri-nq-out.nq
@@ -1,4 +1,4 @@
 <http://example.org/s> <http://example.org/p> <http://example.org/\u0009bado1> .
 <http://example.org/s> <http://example.org/p> <http://example.org/goodo1> .
-<http://example.org/s> <http://example.org/p> <http://example.org/\uFFFDbado2> .
+<http://example.org/s> <http://example.org/p> <http://example.org/\uFFFD\uFFFDbado2> .
 <http://example.org/s> <http://example.org/p> <http://example.org/goodo2> .
diff --git a/test/extra/lax/test-bad-uri-out.nt b/test/extra/lax/test-bad-uri-out.nt
index 8cb00ba7..66da9b0d 100644
--- a/test/extra/lax/test-bad-uri-out.nt
+++ b/test/extra/lax/test-bad-uri-out.nt
@@ -1,4 +1,4 @@
 <http://example.org/s> <http://example.org/p> <http://example.org/\u0009bado1> .
 <http://example.org/s> <http://example.org/p> <http://example.org/goodo1> .
-<http://example.org/s> <http://example.org/p> <http://example.org/\uFFFDbado2> .
+<http://example.org/s> <http://example.org/p> <http://example.org/\uFFFD\uFFFDbado2> .
 <http://example.org/s> <http://example.org/p> <http://example.org/goodo2> .
diff --git a/test/extra/lax/test-bad-utf8-nq-out.nq b/test/extra/lax/test-bad-utf8-nq-out.nq
index 8cefa258..9970c1dd 100644
--- a/test/extra/lax/test-bad-utf8-nq-out.nq
+++ b/test/extra/lax/test-bad-utf8-nq-out.nq
@@ -1,3 +1,29 @@
-<http://example.org/s> <http://example.org/p> "Impossible bytes: \uFFFD \uFFFD" .
-<http://example.org/s> <http://example.org/p> "2 continuation bytes: \uFFFD" .
-<http://example.org/s> <http://example.org/p> "Missing continuation: \uFFFD" .
+<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:                                   \uFFFD                               |" .
+<http://example.org/s> <http://example.org/p> "Impossible byte 2:                                   \uFFFD                               |" .
+<http://example.org/s> <http://example.org/p> "Four impossible bytes:                               \uFFFD\uFFFD\uFFFD\uFFFD                            |" .
+<http://example.org/s> <http://example.org/p> "First continuation byte:                             \uFFFD                               |" .
+<http://example.org/s> <http://example.org/p> "Last continuation byte:                              \uFFFD                               |" .
+<http://example.org/s> <http://example.org/p> "2 continuation bytes:                                \uFFFD\uFFFD                              |" .
+<http://example.org/s> <http://example.org/p> "3 continuation bytes:                                \uFFFD\uFFFD\uFFFD                             |" .
+<http://example.org/s> <http://example.org/p> "4 continuation bytes:                                \uFFFD\uFFFD\uFFFD\uFFFD                            |" .
+<http://example.org/s> <http://example.org/p> "5 continuation bytes:                                \uFFFD\uFFFD\uFFFD\uFFFD\uFFFD                           |" .
+<http://example.org/s> <http://example.org/p> "6 continuation bytes:                                \uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD                          |" .
+<http://example.org/s> <http://example.org/p> "7 continuation bytes:                                \uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD                         |" .
+<http://example.org/s> <http://example.org/p> "Continuation bytes 1:                                \uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD                |" .
+<http://example.org/s> <http://example.org/p> "Continuation bytes 2:                                \uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD                |" .
+<http://example.org/s> <http://example.org/p> "Continuation bytes 3:                                \uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD                |" .
+<http://example.org/s> <http://example.org/p> "Continuation bytes 4:                                \uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD                |" .
+<http://example.org/s> <http://example.org/p> "Lonely leading bytes of 2-byte sequences 1:          \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD |" .
+<http://example.org/s> <http://example.org/p> "Lonely leading bytes of 2-byte sequences 2:          \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD |" .
+<http://example.org/s> <http://example.org/p> "Lonely leading bytes of 3-byte sequences:            \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD |" .
+<http://example.org/s> <http://example.org/p> "Lonely leading bytes of 4-byte sequences:            \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD                 |" .
+<http://example.org/s> <http://example.org/p> "Lonely leading bytes of 5-byte sequences:            \uFFFD \uFFFD \uFFFD \uFFFD                         |" .
+<http://example.org/s> <http://example.org/p> "Lonely leading bytes of 6-byte sequences:            \uFFFD \uFFFD                             |" .
+<http://example.org/s> <http://example.org/p> "2-byte sequence with last byte missing (U+0000):     \uFFFD                               |" .
+<http://example.org/s> <http://example.org/p> "3-byte sequence with last byte missing (U+0000):     \uFFFD                               |" .
+<http://example.org/s> <http://example.org/p> "4-byte sequence with last byte missing (U+0000):     \uFFFD                               |" .
+<http://example.org/s> <http://example.org/p> "2-byte sequence with last byte missing (U-000007FF): \uFFFD                               |" .
+<http://example.org/s> <http://example.org/p> "3-byte sequence with last byte missing (U-0000FFFF): \uFFFD                               |" .
+<http://example.org/s> <http://example.org/p> "4-byte sequence with last byte missing (U-001FFFFF): \uFFFD                               |" .
+<http://example.org/s> <http://example.org/p> "6 sequences with last byte missing:                  \uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD                          |" .
diff --git a/test/extra/lax/test-bad-utf8-nt-out.nt b/test/extra/lax/test-bad-utf8-nt-out.nt
index 8cefa258..9970c1dd 100644
--- a/test/extra/lax/test-bad-utf8-nt-out.nt
+++ b/test/extra/lax/test-bad-utf8-nt-out.nt
@@ -1,3 +1,29 @@
-<http://example.org/s> <http://example.org/p> "Impossible bytes: \uFFFD \uFFFD" .
-<http://example.org/s> <http://example.org/p> "2 continuation bytes: \uFFFD" .
-<http://example.org/s> <http://example.org/p> "Missing continuation: \uFFFD" .
+<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:                                   \uFFFD                               |" .
+<http://example.org/s> <http://example.org/p> "Impossible byte 2:                                   \uFFFD                               |" .
+<http://example.org/s> <http://example.org/p> "Four impossible bytes:                               \uFFFD\uFFFD\uFFFD\uFFFD                            |" .
+<http://example.org/s> <http://example.org/p> "First continuation byte:                             \uFFFD                               |" .
+<http://example.org/s> <http://example.org/p> "Last continuation byte:                              \uFFFD                               |" .
+<http://example.org/s> <http://example.org/p> "2 continuation bytes:                                \uFFFD\uFFFD                              |" .
+<http://example.org/s> <http://example.org/p> "3 continuation bytes:                                \uFFFD\uFFFD\uFFFD                             |" .
+<http://example.org/s> <http://example.org/p> "4 continuation bytes:                                \uFFFD\uFFFD\uFFFD\uFFFD                            |" .
+<http://example.org/s> <http://example.org/p> "5 continuation bytes:                                \uFFFD\uFFFD\uFFFD\uFFFD\uFFFD                           |" .
+<http://example.org/s> <http://example.org/p> "6 continuation bytes:                                \uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD                          |" .
+<http://example.org/s> <http://example.org/p> "7 continuation bytes:                                \uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD                         |" .
+<http://example.org/s> <http://example.org/p> "Continuation bytes 1:                                \uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD                |" .
+<http://example.org/s> <http://example.org/p> "Continuation bytes 2:                                \uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD                |" .
+<http://example.org/s> <http://example.org/p> "Continuation bytes 3:                                \uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD                |" .
+<http://example.org/s> <http://example.org/p> "Continuation bytes 4:                                \uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD                |" .
+<http://example.org/s> <http://example.org/p> "Lonely leading bytes of 2-byte sequences 1:          \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD |" .
+<http://example.org/s> <http://example.org/p> "Lonely leading bytes of 2-byte sequences 2:          \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD |" .
+<http://example.org/s> <http://example.org/p> "Lonely leading bytes of 3-byte sequences:            \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD |" .
+<http://example.org/s> <http://example.org/p> "Lonely leading bytes of 4-byte sequences:            \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD                 |" .
+<http://example.org/s> <http://example.org/p> "Lonely leading bytes of 5-byte sequences:            \uFFFD \uFFFD \uFFFD \uFFFD                         |" .
+<http://example.org/s> <http://example.org/p> "Lonely leading bytes of 6-byte sequences:            \uFFFD \uFFFD                             |" .
+<http://example.org/s> <http://example.org/p> "2-byte sequence with last byte missing (U+0000):     \uFFFD                               |" .
+<http://example.org/s> <http://example.org/p> "3-byte sequence with last byte missing (U+0000):     \uFFFD                               |" .
+<http://example.org/s> <http://example.org/p> "4-byte sequence with last byte missing (U+0000):     \uFFFD                               |" .
+<http://example.org/s> <http://example.org/p> "2-byte sequence with last byte missing (U-000007FF): \uFFFD                               |" .
+<http://example.org/s> <http://example.org/p> "3-byte sequence with last byte missing (U-0000FFFF): \uFFFD                               |" .
+<http://example.org/s> <http://example.org/p> "4-byte sequence with last byte missing (U-001FFFFF): \uFFFD                               |" .
+<http://example.org/s> <http://example.org/p> "6 sequences with last byte missing:                  \uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD                          |" .
diff --git a/test/extra/lax/test-bad-utf8-ttl-out.nt b/test/extra/lax/test-bad-utf8-ttl-out.nt
index 58f2c52b..c8364786 100644
--- a/test/extra/lax/test-bad-utf8-ttl-out.nt
+++ b/test/extra/lax/test-bad-utf8-ttl-out.nt
@@ -1,6 +1,30 @@
-<http://example.org/s> <http://example.org/p> "Impossible bytes: \uFFFD \uFFFD" .
-<http://example.org/s> <http://example.org/p> "2 continuation bytes: \uFFFD" .
-<http://example.org/s> <http://example.org/p> "Missing continuation: \uFFFD" .
-<http://example.org/s> <http://example.org/p> "Impossible bytes: \uFFFD \uFFFD" .
-<http://example.org/s> <http://example.org/p> "2 continuation bytes: \uFFFD" .
-<http://example.org/s> <http://example.org/p> "Missing continuation: \uFFFD" .
+<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:                                   \uFFFD                               |" .
+<http://example.org/s> <http://example.org/p> "Impossible byte 2:                                   \uFFFD                               |" .
+<http://example.org/s> <http://example.org/p> "Four impossible bytes:                               \uFFFD\uFFFD\uFFFD\uFFFD                            |" .
+<http://example.org/s> <http://example.org/p> "First continuation byte:                             \uFFFD                               |" .
+<http://example.org/s> <http://example.org/p> "Last continuation byte:                              \uFFFD                               |" .
+<http://example.org/s> <http://example.org/p> "2 continuation bytes:                                \uFFFD\uFFFD                              |" .
+<http://example.org/s> <http://example.org/p> "3 continuation bytes:                                \uFFFD\uFFFD\uFFFD                             |" .
+<http://example.org/s> <http://example.org/p> "4 continuation bytes:                                \uFFFD\uFFFD\uFFFD\uFFFD                            |" .
+<http://example.org/s> <http://example.org/p> "5 continuation bytes:                                \uFFFD\uFFFD\uFFFD\uFFFD\uFFFD                           |" .
+<http://example.org/s> <http://example.org/p> "6 continuation bytes:                                \uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD                          |" .
+<http://example.org/s> <http://example.org/p> "7 continuation bytes:                                \uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD                         |" .
+<http://example.org/s> <http://example.org/p> "Continuation bytes 1:                                \uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD                |" .
+<http://example.org/s> <http://example.org/p> "Continuation bytes 2:                                \uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD                |" .
+<http://example.org/s> <http://example.org/p> "Continuation bytes 3:                                \uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD                |" .
+<http://example.org/s> <http://example.org/p> "Continuation bytes 4:                                \uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD                |" .
+<http://example.org/s> <http://example.org/p> "Lonely leading bytes of 2-byte sequences 1:          \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD |" .
+<http://example.org/s> <http://example.org/p> "Lonely leading bytes of 2-byte sequences 2:          \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD |" .
+<http://example.org/s> <http://example.org/p> "Lonely leading bytes of 3-byte sequences:            \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD |" .
+<http://example.org/s> <http://example.org/p> "Lonely leading bytes of 4-byte sequences:            \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD                 |" .
+<http://example.org/s> <http://example.org/p> "Lonely leading bytes of 5-byte sequences:            \uFFFD \uFFFD \uFFFD \uFFFD                         |" .
+<http://example.org/s> <http://example.org/p> "Lonely leading bytes of 6-byte sequences:            \uFFFD \uFFFD                             |" .
+<http://example.org/s> <http://example.org/p> "2-byte sequence with last byte missing (U+0000):     \uFFFD                               |" .
+<http://example.org/s> <http://example.org/p> "3-byte sequence with last byte missing (U+0000):     \uFFFD                               |" .
+<http://example.org/s> <http://example.org/p> "4-byte sequence with last byte missing (U+0000):     \uFFFD                               |" .
+<http://example.org/s> <http://example.org/p> "2-byte sequence with last byte missing (U-000007FF): \uFFFD                               |" .
+<http://example.org/s> <http://example.org/p> "3-byte sequence with last byte missing (U-0000FFFF): \uFFFD                               |" .
+<http://example.org/s> <http://example.org/p> "4-byte sequence with last byte missing (U-001FFFFF): \uFFFD                               |" .
+<http://example.org/s> <http://example.org/p> "6 sequences with last byte missing:                  \uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD                          |" .
+<http://example.org/s> <http://example.org/p> "\nImpossible byte 1:                                   \uFFFD                               |\nImpossible byte 2:                                   \uFFFD                               |\nFour impossible bytes:                               \uFFFD\uFFFD\uFFFD\uFFFD                            |\nFirst continuation byte:                             \uFFFD                               |\nLast continuation byte:                              \uFFFD                               |\n2 continuation bytes:                                \uFFFD\uFFFD                              |\n3 continuation bytes:                                \uFFFD\uFFFD\uFFFD                             |\n4 continuation bytes:                                \uFFFD\uFFFD\uFFFD\uFFFD                            |\n5 continuation bytes:                                \uFFFD\uFFFD\uFFFD\uFFFD\uFFFD                           |\n6 continuation bytes:                                \uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD                          |\n7 continuation bytes:                                \uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD                         |\nContinuation bytes 1:                                \uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD                |\nContinuation bytes 2:                                \uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD                |\nContinuation bytes 3:                                \uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD                |\nContinuation bytes 4:                                \uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD                |\nLonely leading bytes of 2-byte sequences 1:          \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD |\nLonely leading bytes of 2-byte sequences 2:          \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD |\nLonely leading bytes of 3-byte sequences:            \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD |\nLonely leading bytes of 4-byte sequences:            \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD \uFFFD                 |\nLonely leading bytes of 5-byte sequences:            \uFFFD \uFFFD \uFFFD \uFFFD                         |\nLonely leading bytes of 6-byte sequences:            \uFFFD \uFFFD                             |\n2-byte sequence with last byte missing (U+0000):     \uFFFD                               |\n3-byte sequence with last byte missing (U+0000):     \uFFFD                               |\n4-byte sequence with last byte missing (U+0000):     \uFFFD                               |\n2-byte sequence with last byte missing (U-000007FF): \uFFFD                               |\n3-byte sequence with last byte missing (U-0000FFFF): \uFFFD                               |\n4-byte sequence with last byte missing (U-001FFFFF): \uFFFD                               |\n6 sequences with last byte missing:                  \uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD                          |\n" .
diff --git a/test/extra/lax/test-bad-utf8.nq b/test/extra/lax/test-bad-utf8.nq
index b8c04637..de67830b 100644
--- a/test/extra/lax/test-bad-utf8.nq
+++ b/test/extra/lax/test-bad-utf8.nq
@@ -1,3 +1,29 @@
-<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:                  �����������                          |" .
diff --git a/test/extra/lax/test-bad-utf8.nt b/test/extra/lax/test-bad-utf8.nt
index b8c04637..de67830b 100644
--- a/test/extra/lax/test-bad-utf8.nt
+++ b/test/extra/lax/test-bad-utf8.nt
@@ -1,3 +1,29 @@
-<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:                  �����������                          |" .
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