From 9facb914c72bf4d0473b49faa797cdb5c9faf68b Mon Sep 17 00:00:00 2001
From: David Robillard <d@drobilla.net>
Date: Fri, 29 Mar 2024 06:55:05 -0400
Subject: Fix lax NQuads parsing

---
 src/n3.c                                | 14 ++++---
 test/extra/lax/manifest.ttl             | 68 ++++++++++++++++++++++++++-------
 test/extra/lax/test-bad-string-out.nt   |  1 +
 test/extra/lax/test-bad-string.nq       |  3 ++
 test/extra/lax/test-bad-string.nt       |  2 +
 test/extra/lax/test-bad-uri-nq-out.nq   |  4 ++
 test/extra/lax/test-bad-uri-out.nt      |  4 ++
 test/extra/lax/test-bad-uri.nq          |  4 ++
 test/extra/lax/test-bad-uri.nt          |  4 +-
 test/extra/lax/test-bad-uri.ttl         |  9 +++--
 test/extra/lax/test-bad-utf8-nq-out.nq  |  3 ++
 test/extra/lax/test-bad-utf8-nt-out.nt  |  3 ++
 test/extra/lax/test-bad-utf8-ttl-out.nt |  6 +++
 test/extra/lax/test-bad-utf8.nq         |  3 ++
 test/extra/lax/test-bad-utf8.nt         |  9 ++---
 15 files changed, 107 insertions(+), 30 deletions(-)
 create mode 100644 test/extra/lax/test-bad-string-out.nt
 create mode 100644 test/extra/lax/test-bad-string.nq
 create mode 100644 test/extra/lax/test-bad-uri-nq-out.nq
 create mode 100644 test/extra/lax/test-bad-uri-out.nt
 create mode 100644 test/extra/lax/test-bad-uri.nq
 create mode 100644 test/extra/lax/test-bad-utf8-nq-out.nq
 create mode 100644 test/extra/lax/test-bad-utf8-nt-out.nt
 create mode 100644 test/extra/lax/test-bad-utf8-ttl-out.nt
 create mode 100644 test/extra/lax/test-bad-utf8.nq

diff --git a/src/n3.c b/src/n3.c
index da3862db..6684fae8 100644
--- a/src/n3.c
+++ b/src/n3.c
@@ -1783,13 +1783,17 @@ read_nquads_statement(SerdReader* const reader)
 SerdStatus
 read_nquadsDoc(SerdReader* const reader)
 {
-  SerdStatus st = SERD_SUCCESS;
-
-  while (!reader->source.eof && !st) {
-    st = read_nquads_statement(reader);
+  while (!reader->source.eof) {
+    const SerdStatus st = read_nquads_statement(reader);
+    if (st > SERD_FAILURE) {
+      if (reader->strict) {
+        return st;
+      }
+      serd_reader_skip_until_byte(reader, '\n');
+    }
   }
 
-  return st;
+  return SERD_SUCCESS;
 }
 
 #if defined(__clang__) && __clang_major__ >= 10
diff --git a/test/extra/lax/manifest.ttl b/test/extra/lax/manifest.ttl
index a04a74f6..b9890e14 100644
--- a/test/extra/lax/manifest.ttl
+++ b/test/extra/lax/manifest.ttl
@@ -6,32 +6,74 @@
 	a mf:Manifest ;
 	rdfs:comment "Serd lax parsing test suite" ;
 	mf:entries (
-		<#test-bad-string>
-		<#test-bad-uri>
-		<#test-bad-utf8>
+		<#test-bad-string-nq>
+		<#test-bad-string-nt>
+		<#test-bad-string-ttl>
+		<#test-bad-uri-nq>
+		<#test-bad-uri-nt>
+		<#test-bad-uri-ttl>
+		<#test-bad-utf8-nq>
+		<#test-bad-utf8-nt>
+		<#test-bad-utf8-ttl>
 		<#test-lone-list>
 	) .
 
-<#test-bad-string>
+<#test-bad-string-nq>
+	a rdft:TestNQuadsNegativeSyntax ;
+	mf:action <test-bad-string.nq> ;
+	mf:name "test-bad-string-nq" ;
+	mf:result <test-bad-string-out.nt> .
+
+<#test-bad-string-nt>
+	a rdft:TestNTriplesNegativeSyntax ;
+	mf:action <test-bad-string.nt> ;
+	mf:name "test-bad-string-nt" ;
+	mf:result <test-bad-string-out.nt> .
+
+<#test-bad-string-ttl>
 	a rdft:TestTurtleNegativeSyntax ;
 	mf:action <test-bad-string.ttl> ;
-	mf:name "test-bad-string" ;
-	mf:result <test-bad-string.nt> .
+	mf:name "test-bad-string-ttl" ;
+	mf:result <test-bad-string-out.nt> .
+
+<#test-bad-uri-nq>
+	a rdft:TestNQuadsNegativeSyntax ;
+	mf:action <test-bad-uri.nq> ;
+	mf:name "test-bad-uri-nq" ;
+	mf:result <test-bad-uri-nq-out.nq> .
+
+<#test-bad-uri-nt>
+	a rdft:TestNTriplesNegativeSyntax ;
+	mf:action <test-bad-uri.nt> ;
+	mf:name "test-bad-uri-nt" ;
+	mf:result <test-bad-uri-out.nt> .
 
-<#test-bad-uri>
+<#test-bad-uri-ttl>
 	a rdft:TestTurtleNegativeSyntax ;
 	mf:action <test-bad-uri.ttl> ;
-	mf:name "test-bad-uri" ;
-	mf:result <test-bad-uri.nt> .
+	mf:name "test-bad-uri-ttl" ;
+	mf:result <test-bad-uri-out.nt> .
 
-<#test-bad-utf8>
+<#test-bad-utf8-nq>
+	a rdft:TestNQuadsNegativeSyntax ;
+	mf:action <test-bad-utf8.nq> ;
+	mf:name "test-bad-utf8-nq" ;
+	mf:result <test-bad-utf8-nq-out.nq> .
+
+<#test-bad-utf8-nt>
+	a rdft:TestNTriplesNegativeSyntax ;
+	mf:action <test-bad-utf8.nt> ;
+	mf:name "test-bad-utf8-nt" ;
+	mf:result <test-bad-utf8-nt-out.nt> .
+
+<#test-bad-utf8-ttl>
 	a rdft:TestTurtleNegativeSyntax ;
 	mf:action <test-bad-utf8.ttl> ;
-	mf:name "test-bad-utf8" ;
-	mf:result <test-bad-utf8.nt> .
+	mf:name "test-bad-utf8-ttl" ;
+	mf:result <test-bad-utf8-ttl-out.nt> .
 
 <#test-lone-list>
 	a rdft:TestTurtleNegativeSyntax ;
-	mf:name "test-lone-list" ;
 	mf:action <test-lone-list.ttl> ;
+	mf:name "test-lone-list" ;
 	mf:result <test-lone-list.nt> .
diff --git a/test/extra/lax/test-bad-string-out.nt b/test/extra/lax/test-bad-string-out.nt
new file mode 100644
index 00000000..24f80a2e
--- /dev/null
+++ b/test/extra/lax/test-bad-string-out.nt
@@ -0,0 +1 @@
+<http://example.org/s1> <http://example.org/p2> "Good" .
diff --git a/test/extra/lax/test-bad-string.nq b/test/extra/lax/test-bad-string.nq
new file mode 100644
index 00000000..72eb9621
--- /dev/null
+++ b/test/extra/lax/test-bad-string.nq
@@ -0,0 +1,3 @@
+<http://example.org/s1> <http://example.org/p1> "Truncated line
+<http://example.org/s1> <http://example.org/p1> "Bad escape \? " .
+<http://example.org/s1> <http://example.org/p2> "Good" .
diff --git a/test/extra/lax/test-bad-string.nt b/test/extra/lax/test-bad-string.nt
index 24f80a2e..72eb9621 100644
--- a/test/extra/lax/test-bad-string.nt
+++ b/test/extra/lax/test-bad-string.nt
@@ -1 +1,3 @@
+<http://example.org/s1> <http://example.org/p1> "Truncated line
+<http://example.org/s1> <http://example.org/p1> "Bad escape \? " .
 <http://example.org/s1> <http://example.org/p2> "Good" .
diff --git a/test/extra/lax/test-bad-uri-nq-out.nq b/test/extra/lax/test-bad-uri-nq-out.nq
new file mode 100644
index 00000000..e14f3a08
--- /dev/null
+++ b/test/extra/lax/test-bad-uri-nq-out.nq
@@ -0,0 +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/�bado2> .
+<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
new file mode 100644
index 00000000..8cb00ba7
--- /dev/null
+++ b/test/extra/lax/test-bad-uri-out.nt
@@ -0,0 +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/goodo2> .
diff --git a/test/extra/lax/test-bad-uri.nq b/test/extra/lax/test-bad-uri.nq
new file mode 100644
index 00000000..8f11b1d7
--- /dev/null
+++ b/test/extra/lax/test-bad-uri.nq
@@ -0,0 +1,4 @@
+<http://example.org/s> <http://example.org/p> <http://example.org/	bado1> .
+<http://example.org/s> <http://example.org/p> <http://example.org/goodo1> .
+<http://example.org/s> <http://example.org/p> <http://example.org/ÿÿbado2> .
+<http://example.org/s> <http://example.org/p> <http://example.org/goodo2> .
diff --git a/test/extra/lax/test-bad-uri.nt b/test/extra/lax/test-bad-uri.nt
index 8cb00ba7..8f11b1d7 100644
--- a/test/extra/lax/test-bad-uri.nt
+++ b/test/extra/lax/test-bad-uri.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/	bado1> .
 <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/ÿÿbado2> .
 <http://example.org/s> <http://example.org/p> <http://example.org/goodo2> .
diff --git a/test/extra/lax/test-bad-uri.ttl b/test/extra/lax/test-bad-uri.ttl
index 8f11b1d7..1a724fd1 100644
--- a/test/extra/lax/test-bad-uri.ttl
+++ b/test/extra/lax/test-bad-uri.ttl
@@ -1,4 +1,5 @@
-<http://example.org/s> <http://example.org/p> <http://example.org/	bado1> .
-<http://example.org/s> <http://example.org/p> <http://example.org/goodo1> .
-<http://example.org/s> <http://example.org/p> <http://example.org/ÿÿbado2> .
-<http://example.org/s> <http://example.org/p> <http://example.org/goodo2> .
+<http://example.org/s>
+	<http://example.org/p> <http://example.org/	bado1> ,
+		<http://example.org/goodo1> ;
+	<http://example.org/p> <http://example.org/ÿÿbado2> ;
+	<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
new file mode 100644
index 00000000..554cf199
--- /dev/null
+++ b/test/extra/lax/test-bad-utf8-nq-out.nq
@@ -0,0 +1,3 @@
+<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: �" .
diff --git a/test/extra/lax/test-bad-utf8-nt-out.nt b/test/extra/lax/test-bad-utf8-nt-out.nt
new file mode 100644
index 00000000..8cefa258
--- /dev/null
+++ b/test/extra/lax/test-bad-utf8-nt-out.nt
@@ -0,0 +1,3 @@
+<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" .
diff --git a/test/extra/lax/test-bad-utf8-ttl-out.nt b/test/extra/lax/test-bad-utf8-ttl-out.nt
new file mode 100644
index 00000000..58f2c52b
--- /dev/null
+++ b/test/extra/lax/test-bad-utf8-ttl-out.nt
@@ -0,0 +1,6 @@
+<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" .
diff --git a/test/extra/lax/test-bad-utf8.nq b/test/extra/lax/test-bad-utf8.nq
new file mode 100644
index 00000000..b8c04637
--- /dev/null
+++ b/test/extra/lax/test-bad-utf8.nq
@@ -0,0 +1,3 @@
+<http://example.org/s> <http://example.org/p> "Impossible bytes: þ ÿ" .
+<http://example.org/s> <http://example.org/p> "2 continuation bytes: €¿" .
+<http://example.org/s> <http://example.org/p> "Missing continuation: À" .
diff --git a/test/extra/lax/test-bad-utf8.nt b/test/extra/lax/test-bad-utf8.nt
index 58f2c52b..b8c04637 100644
--- a/test/extra/lax/test-bad-utf8.nt
+++ b/test/extra/lax/test-bad-utf8.nt
@@ -1,6 +1,3 @@
-<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> "Impossible bytes: þ ÿ" .
+<http://example.org/s> <http://example.org/p> "2 continuation bytes: €¿" .
+<http://example.org/s> <http://example.org/p> "Missing continuation: À" .
-- 
cgit v1.2.1