Write statements with invalid URI characters in lax mode

7 files changed, 24 insertions, 10 deletions

diff --git a/NEWS b/NEWS
index 03636d41..ef15d1fb 100644
--- a/NEWS
+++ b/NEWS
@@ -1,8 +1,9 @@
 serd (0.30.11) unstable;
 
   * Fix writing long literals with triple quotes
+  * Write statements with invalid URI characters in lax mode
 
- -- David Robillard <d@drobilla.net>  Mon, 15 Feb 2021 16:34:35 +0000
+ -- David Robillard <d@drobilla.net>  Fri, 09 Apr 2021 19:40:44 +0000
 
 serd (0.30.10) stable;
 
diff --git a/src/n3.c b/src/n3.c
index 9aae8a19..9e9d5327 100644
--- a/src/n3.c
+++ b/src/n3.c
@@ -723,7 +723,7 @@ read_IRIREF(SerdReader* reader, Ref* dest)
 
   SerdStatus st   = SERD_SUCCESS;
   uint32_t   code = 0;
-  while (!st) {
+  while (st <= SERD_FAILURE) {
     const int c = eat_byte_safe(reader, peek_byte(reader));
     switch (c) {
     case '"':
@@ -767,16 +767,16 @@ read_IRIREF(SerdReader* reader, Ref* dest)
 
     default:
       if (c <= 0x20) {
-        r_err(reader,
-              SERD_ERR_BAD_SYNTAX,
-              "invalid IRI character (escape %%%02X)\n",
-              (unsigned)c);
+        st = r_err(reader,
+                   SERD_ERR_BAD_SYNTAX,
+                   "invalid IRI character (escape %%%02X)\n",
+                   (unsigned)c);
         if (reader->strict) {
-          *dest = pop_node(reader, *dest);
-          return SERD_ERR_BAD_SYNTAX;
+          break;
+        } else {
+          st = SERD_FAILURE;
+          push_byte(reader, *dest, c);
         }
-        st = SERD_FAILURE;
-        push_byte(reader, *dest, c);
       } else if (!(c & 0x80)) {
         push_byte(reader, *dest, c);
       } else if (read_utf8_character(reader, *dest, (uint8_t)c)) {
diff --git a/test/lax/manifest.ttl b/test/lax/manifest.ttl
index e3b11ff4..0cfdb776 100644
--- a/test/lax/manifest.ttl
+++ b/test/lax/manifest.ttl
@@ -7,10 +7,17 @@
 	rdf:type mf:Manifest ;
 	rdfs:comment "Serd lax syntax test cases" ;
 	mf:entries (
+		<#test-bad-string>
 		<#test-bad-uri>
 		<#test-bad-utf8>
 	) .
 
+<#test-bad-string>
+	rdf:type rdft:TestTurtleNegativeSyntax ;
+	mf:name "test-bad-string" ;
+	mf:action <test-bad-string.ttl> ;
+	mf:result <test-bad-string.nt> .
+
 <#test-bad-uri>
 	rdf:type rdft:TestTurtleNegativeSyntax ;
 	mf:name "test-bad-uri" ;
diff --git a/test/lax/test-bad-string.nt b/test/lax/test-bad-string.nt
new file mode 100644
index 00000000..24f80a2e
--- /dev/null
+++ b/test/lax/test-bad-string.nt
@@ -0,0 +1 @@
+<http://example.org/s1> <http://example.org/p2> "Good" .
diff --git a/test/lax/test-bad-string.ttl b/test/lax/test-bad-string.ttl
new file mode 100644
index 00000000..72eb9621
--- /dev/null
+++ b/test/lax/test-bad-string.ttl
@@ -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/lax/test-bad-uri.nt b/test/lax/test-bad-uri.nt
index 7458e12a..8cb00ba7 100644
--- a/test/lax/test-bad-uri.nt
+++ b/test/lax/test-bad-uri.nt
@@ -1,3 +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/test_reader_writer.c b/test/test_reader_writer.c
index 573b13de..134c7a8e 100644
--- a/test/test_reader_writer.c
+++ b/test/test_reader_writer.c
@@ -59,6 +59,7 @@ test_sink(void*              handle,
 static size_t
 eof_test_read(void* buf, size_t size, size_t nmemb, void* stream)
 {
+  assert(size == 1);
   assert(nmemb == 1);
 
   static const char* const string = "_:s1 <http://example.org/p> _:o1 .\n"