Improve reader error handling

2 files changed, 153 insertions, 0 deletions

diff --git a/test/meson.build b/test/meson.build
index d9a386c3..ffd65a17 100644
--- a/test/meson.build
+++ b/test/meson.build
@@ -122,6 +122,7 @@ unit_tests = [
   'env',
   'free_null',
   'node',
+  'overflow',
   'reader_writer',
   'sink',
   'string',
diff --git a/test/test_overflow.c b/test/test_overflow.c
new file mode 100644
index 00000000..ac4a490b
--- /dev/null
+++ b/test/test_overflow.c
@@ -0,0 +1,152 @@
+// Copyright 2018 David Robillard <d@drobilla.net>
+// SPDX-License-Identifier: ISC
+
+#undef NDEBUG
+
+#include "serd/serd.h"
+
+#include <assert.h>
+#include <stdio.h>
+
+static const size_t min_stack_size = 4U * sizeof(size_t) + 240U;
+static const size_t max_stack_size = 1024U;
+
+static SerdStatus
+test_size(const char* const str,
+          const SerdSyntax  syntax,
+          const size_t      stack_size)
+{
+  SerdSink*         sink   = serd_sink_new(NULL, NULL);
+  SerdReader* const reader = serd_reader_new(syntax, sink, stack_size);
+  if (!reader) {
+    return SERD_BAD_STACK;
+  }
+
+  serd_reader_start_string(reader, str);
+  const SerdStatus st = serd_reader_read_document(reader);
+  serd_reader_free(reader);
+  serd_sink_free(sink);
+
+  return st;
+}
+
+static void
+test_all_sizes(const char* const str, const SerdSyntax syntax)
+{
+  // Ensure reading with the maximum stack size succeeds
+  SerdStatus st = test_size(str, syntax, 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(str, syntax, size))) {
+      assert(st == SERD_BAD_STACK);
+    }
+  }
+
+  assert(st == SERD_BAD_STACK);
+}
+
+static void
+test_ntriples_overflow(void)
+{
+  static const char* const test_strings[] = {
+    "<http://example.org/s> <http://example.org/p> <http://example.org/o> .",
+    NULL,
+  };
+
+  for (const char* const* t = test_strings; *t; ++t) {
+    test_all_sizes(*t, SERD_NTRIPLES);
+  }
+}
+
+static void
+test_turtle_overflow(void)
+{
+  static const char* const test_strings[] = {
+    "<http://example.org/s> <http://example.org/p> <http://example.org/> .",
+    "<http://example.org/s> <http://example.org/p> "
+    "<thisisanabsurdlylongurischeme://because/testing/> .",
+    "<http://example.org/s> <http://example.org/p> 1234 .",
+    "<http://example.org/s> <http://example.org/p> (1 2 3 4) .",
+    "<http://example.org/s> <http://example.org/p> ((((((((42)))))))) .",
+    "<http://example.org/s> <http://example.org/p> \"literal\" .",
+    "<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 .",
+    "(((((((((42))))))))) <http://example.org/p> <http://example.org/o> .",
+    "@prefix eg: <http://example.org/ns/test> .",
+    "@base <http://example.org/base> .",
+
+    "@prefix eg: <http://example.org/> . \neg:s eg:p eg:o .\n",
+
+    "@prefix ug.dot: <http://example.org/> . \nug.dot:s ug.dot:p ug.dot:o .\n",
+
+    // NOLINTNEXTLINE(bugprone-suspicious-missing-comma)
+    "@prefix øøøøøøøøø: <http://example.org/long> . \n"
+    "<http://example.org/somewhatlongsubjecttooffsetthepredicate> øøøøøøøøø:p "
+    "øøøøøøøøø:o .\n",
+
+    // NOLINTNEXTLINE(bugprone-suspicious-missing-comma)
+    "<http://example.org/subject/with/a/long/path> "
+    "<http://example.org/predicate/with/a/long/path> "
+    "<http://example.org/object/with/a/long/path> .",
+
+    // NOLINTNEXTLINE(bugprone-suspicious-missing-comma)
+    "<http://example.org/s> <http://example.org/p> "
+    "\"typed\"^^<http://example.org/Datatype> .",
+
+    // NOLINTNEXTLINE(bugprone-suspicious-missing-comma)
+    "@prefix eg: <http://example.org/ns/test> .\n"
+    "<http://example.org/s> <http://example.org/p> "
+    "\"typed\"^^eg:Datatype .",
+
+    // NOLINTNEXTLINE(bugprone-suspicious-missing-comma)
+    "@prefix eg: <http://example.org/ns/test> .\n"
+    "<http://example.org/s> <http://example.org/p> eg:foo .",
+
+    // NOLINTNEXTLINE(bugprone-suspicious-missing-comma)
+    "@prefix prefix: <http://example.org/testing/curies> .\n"
+    "prefix:subject prefix:predicate prefix:object .\n",
+
+    // NOLINTNEXTLINE(bugprone-suspicious-missing-comma)
+    "@prefix prefix: <http://example.org/testing/curies> .\n"
+    "prefix:subjectthatwillcomearoundtobeingfinishedanycharacternow "
+    "prefix:predicate prefix:object .\n",
+
+    // NOLINTNEXTLINE(bugprone-suspicious-missing-comma)
+    "@prefix eg: <http://example.org/> .\n"
+    "eg:s eg:p [ eg:p [ eg:p [ eg:p [ eg:p eg:o ] ] ] ] .\n",
+
+    // NOLINTNEXTLINE(bugprone-suspicious-missing-comma)
+    "@prefix eg: <http://example.org/> .\n"
+    "eg:s eg:p ( 1 2 3 ( 4 5 6 ( 7 8 9 ) ) ) .\n",
+
+    // NOLINTNEXTLINE(bugprone-suspicious-missing-comma)
+    "@prefix eg: <http://example.org/ns/test> .\n"
+    "<http://example.org/s> <http://example.org/p> eg:%99 .",
+
+    // NOLINTNEXTLINE(bugprone-suspicious-missing-comma)
+    "@prefix øøøøøøøøø: <http://example.org/long> .\n"
+    "<http://example.org/somewhatlongsubjecttooffsetthepredicate> øøøøøøøøø:p "
+    "øøøøøøøøø:o .\n",
+
+    // NOLINTNEXTLINE(bugprone-suspicious-missing-comma)
+    "@base <http://example.org/ns/test> .\n"
+    "<http://example.org/s> <http://example.org/p> <rel> .",
+
+    NULL,
+  };
+
+  for (const char* const* t = test_strings; *t; ++t) {
+    test_all_sizes(*t, SERD_TURTLE);
+  }
+}
+
+int
+main(void)
+{
+  test_ntriples_overflow();
+  test_turtle_overflow();
+  return 0;
+}