[WIP] Add support for reading and writing variables

[WIP] Command line option, move later?

This adds a reader flag and serdi option for extending a syntax with support
for SPARQL-like variables, for storing things like patterns or simple queries.

30 files changed, 271 insertions, 23 deletions

diff --git a/NEWS b/NEWS
index d7f5b5e9..a1ba8f6a 100644
--- a/NEWS
+++ b/NEWS
@@ -2,6 +2,7 @@ serd (1.1.1) unstable; urgency=medium
 
   * Add SerdBuffer for mutable buffers to keep SerdChunk const-correct
   * Add SerdWorld for shared library state
+  * Add support for parsing variables
   * Add support for writing terse output with minimal newlines
   * Add support for xsd:float and xsd:double literals
   * Bring read/write interface closer to C standard
@@ -19,7 +20,7 @@ serd (1.1.1) unstable; urgency=medium
   * Use a fixed-size reader stack
   * Use char* for strings in public API
 
- -- David Robillard <d@drobilla.net>  Mon, 19 Dec 2022 20:54:56 +0000
+ -- David Robillard <d@drobilla.net>  Mon, 19 Dec 2022 20:55:34 +0000
 
 serd (0.32.0) stable; urgency=medium
 
diff --git a/doc/man/serd-pipe.1 b/doc/man/serd-pipe.1
index 9b700068..f5908c1d 100644
--- a/doc/man/serd-pipe.1
+++ b/doc/man/serd-pipe.1
@@ -8,7 +8,7 @@
 .Nd read and write RDF data
 .Sh SYNOPSIS
 .Nm serd-pipe
-.Op Fl afhlqtv
+.Op Fl afhlqtvx
 .Op Fl B Ar base
 .Op Fl b Ar bytes
 .Op Fl c Ar prefix
@@ -158,6 +158,12 @@ as input.
 Write terser output without newlines.
 .It Fl v
 Display version information and exit.
+.It Fl x
+Support parsing variable nodes.
+Variables can be written in SPARQL style, for example
+.Dq ?var
+or
+.Dq $var .
 .El
 .Sh EXIT STATUS
 .Nm
diff --git a/include/serd/node.h b/include/serd/node.h
index b308a507..1faba160 100644
--- a/include/serd/node.h
+++ b/include/serd/node.h
@@ -38,9 +38,14 @@ typedef struct SerdNodeImpl SerdNode;
 /**
    Type of a node.
 
-   An abstract RDF node can be either a resource or a literal.  This type is
-   more precise to preserve syntactic differences and support additional
-   features.
+   Note that this set of types is both more precise than, and extended from,
+   the possible types of an abstract RDF node.  Not all types can occur in all
+   contexts, for example, a Turtle document can't contain a variable node.
+
+   The string value of a node never contains quoting or other type indicators.
+   For example, the blank node `_:id3` and the plain literal `"id3"` from a
+   Turtle document would both have the same string, "id3", returned by
+   #serd_node_string.
 */
 typedef enum {
   /**
@@ -82,6 +87,19 @@ typedef enum {
      Turtle](http://www.w3.org/TR/turtle/#grammar-production-BLANK_NODE_LABEL)
   */
   SERD_BLANK = 4,
+
+  /**
+     A variable node.
+
+     A variable's identity is, like blank nodes, local to its context.
+     Variables are typically used in interfaces for querying, rather than
+     present in data (there is no concept of a variable in RDF or its standard
+     syntaxes).
+
+     @see [SPARQL 1.1 Query
+     Language](https://www.w3.org/TR/sparql11-query/#rVar)
+  */
+  SERD_VARIABLE = 5,
 } SerdNodeType;
 
 /// Node flags, which ORed together make a #SerdNodeFlags
diff --git a/include/serd/reader.h b/include/serd/reader.h
index 4c669342..040f3398 100644
--- a/include/serd/reader.h
+++ b/include/serd/reader.h
@@ -29,7 +29,8 @@ typedef struct SerdReaderImpl SerdReader;
 
 /// Reader options
 typedef enum {
-  SERD_READ_LAX = 1U << 0U, ///< Tolerate invalid input where possible
+  SERD_READ_LAX       = 1U << 0U, ///< Tolerate invalid input where possible
+  SERD_READ_VARIABLES = 1U << 1U, ///< Support variable nodes
 } SerdReaderFlag;
 
 /// Bitwise OR of SerdReaderFlag values
diff --git a/src/env.c b/src/env.c
index 5159292d..afbf1064 100644
--- a/src/env.c
+++ b/src/env.c
@@ -319,6 +319,7 @@ serd_env_expand(const SerdEnv* env, const SerdNode* node)
   case SERD_CURIE:
     return expand_curie(env, node);
   case SERD_BLANK:
+  case SERD_VARIABLE:
     break;
   }
 
diff --git a/src/read_ntriples.c b/src/read_ntriples.c
index da726658..bec59c13 100644
--- a/src/read_ntriples.c
+++ b/src/read_ntriples.c
@@ -520,6 +520,51 @@ read_HEX(SerdReader* const reader)
   return 0;
 }
 
+/**
+   Read a variable name, starting after the '?' or '$'.
+
+   This is an extension that serd uses in certain contexts to support patterns.
+
+   Restricted version of SPARQL 1.1: [166] VARNAME
+*/
+static SerdStatus
+read_VARNAME(SerdReader* const reader, SerdNode** const dest)
+{
+  // Simplified from SPARQL: VARNAME ::= (PN_CHARS_U | [0-9])+
+  SerdNode*  n  = *dest;
+  SerdStatus st = SERD_SUCCESS;
+  int        c  = 0;
+
+  while ((c = peek_byte(reader))) {
+    if (is_digit(c) || c == '_') {
+      st = push_byte(reader, n, eat_byte_safe(reader, c));
+    } else if ((st = read_PN_CHARS(reader, n))) {
+      st = st > SERD_FAILURE ? st : SERD_SUCCESS;
+      break;
+    }
+  }
+
+  return st;
+}
+
+SerdStatus
+read_Var(SerdReader* const reader, SerdNode** const dest)
+{
+  if (!(reader->flags & SERD_READ_VARIABLES)) {
+    return r_err(reader, SERD_BAD_SYNTAX, "syntax does not support variables");
+  }
+
+  const int c = peek_byte(reader);
+  assert(c == '$' || c == '?');
+  skip_byte(reader, c);
+
+  if (!(*dest = push_node(reader, SERD_VARIABLE, "", 0))) {
+    return SERD_BAD_STACK;
+  }
+
+  return read_VARNAME(reader, dest);
+}
+
 // Nonterminals
 
 // comment ::= '#' ( [^#xA #xD] )*
@@ -576,6 +621,8 @@ read_nt_subject(SerdReader* const reader, SerdNode** const dest)
   switch (peek_byte(reader)) {
   case '<':
     return read_IRI(reader, dest);
+  case '?':
+    return read_Var(reader, dest);
   case '_':
     return read_BLANK_NODE_LABEL(reader, dest, &ate_dot);
   default:
@@ -589,7 +636,8 @@ read_nt_subject(SerdReader* const reader, SerdNode** const dest)
 SerdStatus
 read_nt_predicate(SerdReader* const reader, SerdNode** const dest)
 {
-  return read_IRI(reader, dest);
+  return (peek_byte(reader) == '?') ? read_Var(reader, dest)
+                                    : read_IRI(reader, dest);
 }
 
 /// [4] object
@@ -605,6 +653,8 @@ read_nt_object(SerdReader* const reader,
     return read_literal(reader, dest);
   case '<':
     return read_IRI(reader, dest);
+  case '?':
+    return read_Var(reader, dest);
   case '_':
     return read_BLANK_NODE_LABEL(reader, dest, ate_dot);
   default:
diff --git a/src/read_turtle.c b/src/read_turtle.c
index 828818ba..027f0918 100644
--- a/src/read_turtle.c
+++ b/src/read_turtle.c
@@ -458,10 +458,15 @@ read_literal(SerdReader* const reader,
 }
 
 static SerdStatus
-read_verb(SerdReader* const reader, SerdNode** const dest)
+read_verb(SerdReader* reader, SerdNode** const dest)
 {
   const size_t orig_stack_size = reader->stack.size;
-  if (peek_byte(reader) == '<') {
+
+  switch (peek_byte(reader)) {
+  case '$':
+  case '?':
+    return read_Var(reader, dest);
+  case '<':
     return read_IRIREF(reader, dest);
   }
 
@@ -617,6 +622,10 @@ read_object(SerdReader* const  reader,
   case EOF:
   case ')':
     return r_err(reader, SERD_BAD_SYNTAX, "expected object");
+  case '$':
+  case '?':
+    st = read_Var(reader, &o);
+    break;
   case '[':
     simple = false;
     st     = read_anon(reader, *ctx, false, &o);
@@ -813,6 +822,10 @@ read_turtle_subject(SerdReader* const reader,
   SerdStatus st      = SERD_SUCCESS;
   bool       ate_dot = false;
   switch ((*s_type = peek_byte(reader))) {
+  case '$':
+  case '?':
+    st = read_Var(reader, dest);
+    break;
   case '[':
     st = read_anon(reader, ctx, true, dest);
     break;
diff --git a/src/reader.c b/src/reader.c
index bf6d697f..6f5ccce6 100644
--- a/src/reader.c
+++ b/src/reader.c
@@ -233,6 +233,7 @@ serd_reader_new(SerdWorld* const      world,
   me->sink    = sink;
   me->stack   = serd_stack_new(stack_size, serd_node_align);
   me->syntax  = syntax;
+  me->flags   = flags;
   me->next_id = 1;
   me->strict  = !(flags & SERD_READ_LAX);
 
diff --git a/src/reader.h b/src/reader.h
index 73647307..01ad5890 100644
--- a/src/reader.h
+++ b/src/reader.h
@@ -44,6 +44,7 @@ struct SerdReaderImpl {
   SerdByteSource* source;
   SerdStack       stack;
   SerdSyntax      syntax;
+  SerdReaderFlags flags;
   unsigned        next_id;
   char*           bprefix;
   size_t          bprefix_len;
diff --git a/src/statement.c b/src/statement.c
index 434d41d4..9d2d4229 100644
--- a/src/statement.c
+++ b/src/statement.c
@@ -17,7 +17,8 @@ static bool
 is_resource(const SerdNode* const node)
 {
   const SerdNodeType type = node ? serd_node_type(node) : (SerdNodeType)0;
-  return type == SERD_URI || type == SERD_CURIE || type == SERD_BLANK;
+  return type == SERD_URI || type == SERD_CURIE || type == SERD_BLANK ||
+         type == SERD_VARIABLE;
 }
 
 bool
diff --git a/src/writer.c b/src/writer.c
index 60a387c0..372e4279 100644
--- a/src/writer.c
+++ b/src/writer.c
@@ -943,6 +943,18 @@ write_blank(SerdWriter* const        writer,
 }
 
 SERD_NODISCARD static SerdStatus
+write_variable(SerdWriter* const writer, const SerdNode* const node)
+{
+  SerdStatus st = SERD_SUCCESS;
+
+  TRY(st, esink("?", 1, writer));
+  TRY(st, esink(serd_node_string(node), node->length, writer));
+
+  writer->last_sep = SEP_NONE;
+  return st;
+}
+
+SERD_NODISCARD static SerdStatus
 write_node(SerdWriter* const        writer,
            const SerdNode* const    node,
            const SerdField          field,
@@ -963,6 +975,9 @@ write_node(SerdWriter* const        writer,
   case SERD_BLANK:
     st = write_blank(writer, node, field, flags);
     break;
+  case SERD_VARIABLE:
+    st = write_variable(writer, node);
+    break;
   }
 
   if (node->type != SERD_BLANK) {
diff --git a/test/extra/bad/bad-var.ttl b/test/extra/bad/bad-var.ttl
new file mode 100644
index 00000000..29b5b008
--- /dev/null
+++ b/test/extra/bad/bad-var.ttl
@@ -0,0 +1,2 @@
+?s
+	<http://example.org/p> <http://example.org/o> .
diff --git a/test/extra/bad/manifest.ttl b/test/extra/bad/manifest.ttl
index 7ab5a427..cd4aee24 100644
--- a/test/extra/bad/manifest.ttl
+++ b/test/extra/bad/manifest.ttl
@@ -87,6 +87,7 @@
 		<#bad-uri-scheme>
 		<#bad-uri-scheme-start>
 		<#bad-uri-truncated>
+		<#bad-var>
 		<#bad-verb>
 	) .
 
@@ -490,6 +491,11 @@
 	mf:action <bad-uri-truncated.nt> ;
 	mf:name "bad-uri-truncated" .
 
+<#bad-var>
+	a rdft:TestTurtleNegativeSyntax ;
+	mf:action <bad-var.ttl> ;
+	mf:name "bad-var" .
+
 <#bad-verb>
 	a rdft:TestTurtleNegativeSyntax ;
 	mf:action <bad-verb.ttl> ;
diff --git a/test/extra/pattern/bad-pattern-graph.nq b/test/extra/pattern/bad-pattern-graph.nq
new file mode 100644
index 00000000..9c77ac9a
--- /dev/null
+++ b/test/extra/pattern/bad-pattern-graph.nq
@@ -0,0 +1 @@
+<http://example.org/s> <http://example.org/p> <http://example.org/o> ?gµ .
diff --git a/test/extra/pattern/bad-pattern-object.nq b/test/extra/pattern/bad-pattern-object.nq
new file mode 100644
index 00000000..7f22b520
--- /dev/null
+++ b/test/extra/pattern/bad-pattern-object.nq
@@ -0,0 +1 @@
+<http://example.org/s> <http://example.org/p> ?oµ <http://example.org/g> .
diff --git a/test/extra/pattern/bad-pattern-object.nt b/test/extra/pattern/bad-pattern-object.nt
new file mode 100644
index 00000000..aef057e4
--- /dev/null
+++ b/test/extra/pattern/bad-pattern-object.nt
@@ -0,0 +1 @@
+<http://example.org/s> <http://example.org/p> ?oµ .
diff --git a/test/extra/pattern/bad-pattern-object.ttl b/test/extra/pattern/bad-pattern-object.ttl
new file mode 100644
index 00000000..cc2eee77
--- /dev/null
+++ b/test/extra/pattern/bad-pattern-object.ttl
@@ -0,0 +1,2 @@
+<http://example.org/s>
+	<http://example.org/o> ?oµ .
diff --git a/test/extra/pattern/bad-pattern-predicate.nq b/test/extra/pattern/bad-pattern-predicate.nq
new file mode 100644
index 00000000..cfe466c9
--- /dev/null
+++ b/test/extra/pattern/bad-pattern-predicate.nq
@@ -0,0 +1 @@
+<http://example.org/s> ?pµ <http://example.org/o> <http://example.org/g> .
diff --git a/test/extra/pattern/bad-pattern-predicate.nt b/test/extra/pattern/bad-pattern-predicate.nt
new file mode 100644
index 00000000..a7e0c994
--- /dev/null
+++ b/test/extra/pattern/bad-pattern-predicate.nt
@@ -0,0 +1 @@
+<http://example.org/s> ?pµ <http://example.org/o> .
diff --git a/test/extra/pattern/bad-pattern-predicate.ttl b/test/extra/pattern/bad-pattern-predicate.ttl
new file mode 100644
index 00000000..cc2bd6b1
--- /dev/null
+++ b/test/extra/pattern/bad-pattern-predicate.ttl
@@ -0,0 +1,2 @@
+<http://example.org/s>
+	?pµ <http://example.org/o> .
diff --git a/test/extra/pattern/bad-pattern-subject.nq b/test/extra/pattern/bad-pattern-subject.nq
new file mode 100644
index 00000000..341c437f
--- /dev/null
+++ b/test/extra/pattern/bad-pattern-subject.nq
@@ -0,0 +1 @@
+?sµ <http://example.org/p> <http://example.org/o> <http://example.org/g> .
diff --git a/test/extra/pattern/bad-pattern-subject.nt b/test/extra/pattern/bad-pattern-subject.nt
new file mode 100644
index 00000000..5bbb29ff
--- /dev/null
+++ b/test/extra/pattern/bad-pattern-subject.nt
@@ -0,0 +1 @@
+?sµ <http://example.org/p> <http://example.org/o> .
diff --git a/test/extra/pattern/bad-pattern-subject.ttl b/test/extra/pattern/bad-pattern-subject.ttl
new file mode 100644
index 00000000..5f3dbfdd
--- /dev/null
+++ b/test/extra/pattern/bad-pattern-subject.ttl
@@ -0,0 +1,2 @@
+?sµ
+	<http://example.org/p> <http://example.org/o> .
diff --git a/test/extra/pattern/manifest.ttl b/test/extra/pattern/manifest.ttl
new file mode 100644
index 00000000..e5690218
--- /dev/null
+++ b/test/extra/pattern/manifest.ttl
@@ -0,0 +1,88 @@
+@prefix mf: <http://www.w3.org/2001/sw/DataAccess/tests/test-manifest#> .
+@prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#> .
+@prefix rdft: <http://www.w3.org/ns/rdftest#> .
+
+<>
+	a mf:Manifest ;
+	rdfs:comment "Serd pattern syntax test suite" ;
+	mf:entries (
+		<#bad-pattern-graph-nq>
+		<#bad-pattern-object-nq>
+		<#bad-pattern-object-nt>
+		<#bad-pattern-object-ttl>
+		<#bad-pattern-predicate-nq>
+		<#bad-pattern-predicate-nt>
+		<#bad-pattern-predicate-ttl>
+		<#bad-pattern-subject-nq>
+		<#bad-pattern-subject-nt>
+		<#bad-pattern-subject-ttl>
+		<#test-graph-pattern>
+		<#test-pattern-nt>
+		<#test-pattern-ttl>
+	) .
+
+<#bad-pattern-graph-nq>
+	a rdft:TestNQuadsNegativeSyntax ;
+	mf:action <bad-pattern-graph.nq> ;
+	mf:name "bad-pattern-graph-nq" .
+
+<#bad-pattern-object-nq>
+	a rdft:TestNQuadsNegativeSyntax ;
+	mf:action <bad-pattern-object.nq> ;
+	mf:name "bad-pattern-object-nq" .
+
+<#bad-pattern-object-nt>
+	a rdft:TestNTriplesNegativeSyntax ;
+	mf:action <bad-pattern-object.nt> ;
+	mf:name "bad-pattern-object-nt" .
+
+<#bad-pattern-object-ttl>
+	a rdft:TestTurtleNegativeSyntax ;
+	mf:action <bad-pattern-object.ttl> ;
+	mf:name "bad-pattern-object-ttl" .
+
+<#bad-pattern-predicate-nq>
+	a rdft:TestNQuadsNegativeSyntax ;
+	mf:action <bad-pattern-predicate.nq> ;
+	mf:name "bad-pattern-predicate-nq" .
+
+<#bad-pattern-predicate-nt>
+	a rdft:TestNTriplesNegativeSyntax ;
+	mf:action <bad-pattern-predicate.nt> ;
+	mf:name "bad-pattern-predicate-nt" .
+
+<#bad-pattern-predicate-ttl>
+	a rdft:TestTurtleNegativeSyntax ;
+	mf:action <bad-pattern-predicate.ttl> ;
+	mf:name "bad-pattern-predicate-ttl" .
+
+<#bad-pattern-subject-nq>
+	a rdft:TestNQuadsNegativeSyntax ;
+	mf:action <bad-pattern-subject.nq> ;
+	mf:name "bad-pattern-subject-nq" .
+
+<#bad-pattern-subject-nt>
+	a rdft:TestNTriplesNegativeSyntax ;
+	mf:action <bad-pattern-subject.nt> ;
+	mf:name "bad-pattern-subject-nt" .
+
+<#bad-pattern-subject-ttl>
+	a rdft:TestTurtleNegativeSyntax ;
+	mf:action <bad-pattern-subject.ttl> ;
+	mf:name "bad-pattern-subject-ttl" .
+
+<#test-graph-pattern>
+	a rdft:TestNQuadsPositiveSyntax ;
+	mf:action <test-graph-pattern.nq> ;
+	mf:name "test-graph-pattern" .
+
+<#test-pattern-nt>
+	a rdft:TestNTriplesPositiveSyntax ;
+	mf:action <test-pattern.nt> ;
+	mf:name "test-pattern-nt" .
+
+<#test-pattern-ttl>
+	a rdft:TestTurtleEval ;
+	mf:action <test-pattern.ttl> ;
+	mf:name "test-pattern" ;
+	mf:result <test-pattern.nt> .
diff --git a/test/extra/pattern/test-graph-pattern.nq b/test/extra/pattern/test-graph-pattern.nq
new file mode 100644
index 00000000..453e9516
--- /dev/null
+++ b/test/extra/pattern/test-graph-pattern.nq
@@ -0,0 +1 @@
+<http://a.example/s> <http://a.example/p> <http://a.example/o> ?g .
diff --git a/test/extra/pattern/test-pattern.nt b/test/extra/pattern/test-pattern.nt
new file mode 100644
index 00000000..ddfe6d3c
--- /dev/null
+++ b/test/extra/pattern/test-pattern.nt
@@ -0,0 +1,6 @@
+?s <http://example.org/p1> <http://example.org/o1> .
+<http://example.org/s> ?p <http://example.org/o1> .
+<http://example.org/s> <http://example.org/p1> ?o .
+<http://example.org/s> <http://example.org/p2> _:b1 .
+_:b1 ?2p <http://example.org/o2> .
+_:b1 <http://example.org/p3> ?_o .
diff --git a/test/extra/pattern/test-pattern.ttl b/test/extra/pattern/test-pattern.ttl
new file mode 100644
index 00000000..3742e5ed
--- /dev/null
+++ b/test/extra/pattern/test-pattern.ttl
@@ -0,0 +1,10 @@
+?s
+	<http://example.org/p1> <http://example.org/o1> .
+
+<http://example.org/s>
+	?p <http://example.org/o1> ;
+	<http://example.org/p1> ?o ;
+	<http://example.org/p2> [
+		?2p <http://example.org/o2> ;
+		<http://example.org/p3> ?_o
+	] .
diff --git a/test/meson.build b/test/meson.build
index e6b0fe5c..a2436b22 100644
--- a/test/meson.build
+++ b/test/meson.build
@@ -32,6 +32,7 @@ ttl_metadata_file_paths = [
   'extra/full/manifest.ttl',
   'extra/good/manifest.ttl',
   'extra/lax/manifest.ttl',
+  'extra/pattern/manifest.ttl',
   'extra/perfect/manifest.ttl',
   'extra/prefix/manifest.ttl',
   'extra/pretty/manifest.ttl',
@@ -430,6 +431,12 @@ test_suites = {
     files('extra/lax/manifest.ttl'),
     ns_serdtest + 'lax/',
   ],
+  'pattern': [
+    files('extra/pattern/manifest.ttl'),
+    ns_serdtest + 'pattern/',
+    '--',
+    '-x',
+  ],
   'perfect_forward': [
     files('extra/perfect/manifest.ttl'),
     ns_serdtest + 'perfect/',
diff --git a/test/test_overflow.c b/test/test_overflow.c
index 086d353b..b81d4367 100644
--- a/test/test_overflow.c
+++ b/test/test_overflow.c
@@ -12,17 +12,18 @@ static const size_t min_stack_size = 4U * sizeof(size_t) + 240U;
 static const size_t max_stack_size = 1024U;
 
 static SerdStatus
-test_size(SerdWorld* const  world,
-          const char* const str,
-          const SerdSyntax  syntax,
-          const size_t      stack_size)
+test_size(SerdWorld* const      world,
+          const char* const     str,
+          const SerdSyntax      syntax,
+          const SerdReaderFlags flags,
+          const size_t          stack_size)
 {
   SerdLimits limits        = serd_world_limits(world);
   limits.reader_stack_size = stack_size;
   serd_world_set_limits(world, limits);
 
   SerdSink*         sink   = serd_sink_new(NULL, NULL, NULL);
-  SerdReader* const reader = serd_reader_new(world, syntax, 0U, sink);
+  SerdReader* const reader = serd_reader_new(world, syntax, flags, sink);
   if (!reader) {
     return SERD_BAD_STACK;
   }
@@ -43,17 +44,18 @@ test_size(SerdWorld* const  world,
 }
 
 static void
-test_all_sizes(SerdWorld* const  world,
-               const char* const str,
-               const SerdSyntax  syntax)
+test_all_sizes(SerdWorld* const      world,
+               const char* const     str,
+               const SerdSyntax      syntax,
+               const SerdReaderFlags flags)
 {
   // Ensure reading with the maximum stack size succeeds
-  SerdStatus st = test_size(world, str, syntax, max_stack_size);
+  SerdStatus st = test_size(world, str, syntax, flags, max_stack_size);
   assert(!st);
 
   // Test with an increasingly smaller stack
   for (size_t size = max_stack_size; size > min_stack_size; --size) {
-    if ((st = test_size(world, str, syntax, size))) {
+    if ((st = test_size(world, str, syntax, flags, size))) {
       assert(st == SERD_BAD_STACK);
     }
   }
@@ -75,7 +77,7 @@ test_ntriples_overflow(void)
   SerdWorld* const world = serd_world_new();
 
   for (const char* const* t = test_strings; *t; ++t) {
-    test_all_sizes(world, *t, SERD_NTRIPLES);
+    test_all_sizes(world, *t, SERD_NTRIPLES, 0U);
   }
 
   serd_world_free(world);
@@ -97,6 +99,7 @@ test_turtle_overflow(void)
     "<http://example.org/s> <http://example.org/p> _:blank .",
     "<http://example.org/s> <http://example.org/p> true .",
     "<http://example.org/s> <http://example.org/p> \"\"@en .",
+    "?subject ?predicate ?object .",
     "(((((((((42))))))))) <http://example.org/p> <http://example.org/o> .",
     "@prefix eg: <http://example.org/ns/test> .",
     "@base <http://example.org/base> .",
@@ -164,7 +167,7 @@ test_turtle_overflow(void)
   SerdWorld* const world = serd_world_new();
 
   for (const char* const* t = test_strings; *t; ++t) {
-    test_all_sizes(world, *t, SERD_TURTLE);
+    test_all_sizes(world, *t, SERD_TURTLE, SERD_READ_VARIABLES);
   }
 
   serd_world_free(world);
diff --git a/tools/serd-pipe.c b/tools/serd-pipe.c
index 480afa6e..b09c12e1 100644
--- a/tools/serd-pipe.c
+++ b/tools/serd-pipe.c
@@ -51,7 +51,8 @@ print_usage(const char* const name, const bool error)
     "  -r ROOT_URI  Keep relative URIs within ROOT_URI.\n"
     "  -s STRING    Parse STRING as input.\n"
     "  -t           Write terser output without newlines.\n"
-    "  -v           Display version information and exit.\n";
+    "  -v           Display version information and exit.\n"
+    "  -x           Support parsing variable nodes like \"?x\".\n";
 
   FILE* const os = error ? stderr : stdout;
   fprintf(os, "%s", error ? "\n" : "");
@@ -161,6 +162,9 @@ main(int argc, char** argv)
         writer_flags |= SERD_WRITE_TERSE;
       } else if (opt == 'v') {
         return serd_print_version(argv[0]);
+      } else if (opt == 'x') {
+        reader_flags |= SERD_READ_VARIABLES;
+        break;
       } else if (argv[a][1] == 'B') {
         if (++a == argc) {
           return missing_arg(prog, 'B');