Treat out of range unicode characters as errors

7 files changed, 12 insertions, 10 deletions

diff --git a/NEWS b/NEWS
index fc35155c..f051f9ee 100644
--- a/NEWS
+++ b/NEWS
@@ -4,8 +4,9 @@ serd (0.32.3) unstable; urgency=medium
   * Fix library current_version on MacOS
   * Fix parsing NQuads lines with no space before the final dot
   * Support reading lone lists in lax mode
+  * Treat out of range unicode characters as errors
 
- -- David Robillard <d@drobilla.net>  Fri, 29 Mar 2024 13:36:36 +0000
+ -- David Robillard <d@drobilla.net>  Sun, 09 Jun 2024 22:15:14 +0000
 
 serd (0.32.2) stable; urgency=medium
 
diff --git a/src/n3.c b/src/n3.c
index 6684fae8..39b00d13 100644
--- a/src/n3.c
+++ b/src/n3.c
@@ -97,7 +97,7 @@ read_UCHAR(SerdReader* const reader, const Ref dest, uint32_t* const char_code)
           code);
     push_bytes(reader, dest, replacement_char, 3);
     *char_code = 0xFFFD;
-    return SERD_SUCCESS;
+    return reader->strict ? SERD_ERR_BAD_SYNTAX : SERD_SUCCESS;
   }
 
   // Build output in buf
@@ -325,7 +325,7 @@ read_string_escape(SerdReader* const    reader,
   uint32_t   code = 0;
   if ((st = read_ECHAR(reader, ref, flags)) &&
       (st = read_UCHAR(reader, ref, &code))) {
-    return r_err(reader, st, "invalid escape '\\%c'\n", peek_byte(reader));
+    return r_err(reader, st, "expected string escape sequence\n");
   }
 
   return st;
diff --git a/test/extra/bad/manifest.ttl b/test/extra/bad/manifest.ttl
index 7e35e5ff..dfb84ce9 100644
--- a/test/extra/bad/manifest.ttl
+++ b/test/extra/bad/manifest.ttl
@@ -13,6 +13,7 @@
 		<#bad-blank-node-label>
 		<#bad-blank-predicate>
 		<#bad-blank-syntax>
+		<#bad-blank-syntax>
 		<#bad-bom>
 		<#bad-char-in-local>
 		<#bad-char-in-prefix>
diff --git a/test/extra/good/manifest.ttl b/test/extra/good/manifest.ttl
index 3b911990..c464d700 100644
--- a/test/extra/good/manifest.ttl
+++ b/test/extra/good/manifest.ttl
@@ -48,7 +48,6 @@
 		<#test-nt-syntax-eol-crlf>
 		<#test-nt-syntax-eol-lf>
 		<#test-nt-syntax-eol-lfcr>
-		<#test-out-of-range-unicode>
 		<#test-prefix>
 		<#test-quote-escapes>
 		<#test-rel>
@@ -295,12 +294,6 @@
 	mf:action <test-nt-syntax-eol-lfcr.nt> ;
 	mf:name "test-nt-syntax-eol-lfcr" .
 
-<#test-out-of-range-unicode>
-	a rdft:TestTurtleEval ;
-	mf:action <test-out-of-range-unicode.ttl> ;
-	mf:name "test-out-of-range-unicode" ;
-	mf:result <test-out-of-range-unicode.nt> .
-
 <#test-prefix>
 	a rdft:TestTurtleEval ;
 	mf:action <test-prefix.ttl> ;
diff --git a/test/extra/lax/manifest.ttl b/test/extra/lax/manifest.ttl
index bf51f931..f79e1984 100644
--- a/test/extra/lax/manifest.ttl
+++ b/test/extra/lax/manifest.ttl
@@ -19,6 +19,7 @@
 		<#test-bad-utf8-ttl>
 		<#test-bad-utf8-trig>
 		<#test-lone-list>
+		<#test-out-of-range-unicode>
 	) .
 
 <#test-bad-string-nq>
@@ -98,3 +99,9 @@
 	mf:action <test-lone-list.ttl> ;
 	mf:name "test-lone-list" ;
 	mf:result <test-lone-list.nt> .
+
+<#test-out-of-range-unicode>
+	a rdft:TestTurtleNegativeSyntax ;
+	mf:action <test-out-of-range-unicode.ttl> ;
+	mf:name "test-out-of-range-unicode" ;
+	mf:result <test-out-of-range-unicode.nt> .
diff --git a/test/extra/good/test-out-of-range-unicode.nt b/test/extra/lax/test-out-of-range-unicode.nt
index 5def9e31..5def9e31 100644
--- a/test/extra/good/test-out-of-range-unicode.nt
+++ b/test/extra/lax/test-out-of-range-unicode.nt
diff --git a/test/extra/good/test-out-of-range-unicode.ttl b/test/extra/lax/test-out-of-range-unicode.ttl
index 7e64785a..7e64785a 100644
--- a/test/extra/good/test-out-of-range-unicode.ttl
+++ b/test/extra/lax/test-out-of-range-unicode.ttl