Add serd_new_real_file_uri()

4 files changed, 92 insertions, 1 deletions

diff --git a/include/serd/serd.h b/include/serd/serd.h
index a2e767ff..dd165360 100644
--- a/include/serd/serd.h
+++ b/include/serd/serd.h
@@ -674,6 +674,22 @@ SerdNode* SERD_ALLOCATED
 serd_new_boolean(bool b);
 
 /**
+   Create a new file URI node for a file that exists on this system.
+
+   This is like serd_new_file_uri() except it resolves and canonicalizes the
+   path, so the returned node is always a complete file URI with a scheme and
+   absolute path that does not contain any dot references or links, or NULL if
+   this is impossible (for example, because the path does not exist).
+
+   This should be used wherever the URI for an existent file is required, for
+   example to set the base URI of a document.
+*/
+SERD_API
+SerdNode* SERD_ALLOCATED
+serd_new_real_file_uri(const char* SERD_NULLABLE path,
+                       const char* SERD_NULLABLE hostname);
+
+/**
    Create a new node by serialising `d` into an xsd:decimal string
 
    The resulting node will always contain a `.', start with a digit, and end
diff --git a/src/node.c b/src/node.c
index 566e8839..76c12235 100644
--- a/src/node.c
+++ b/src/node.c
@@ -14,6 +14,8 @@
   OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
 */
 
+#define _XOPEN_SOURCE 600 /* for realpath */
+
 #include "node.h"
 
 #include "namespaces.h"
@@ -24,6 +26,11 @@
 #include "exess/exess.h"
 #include "serd/serd.h"
 
+#ifdef _WIN32
+#  define WIN32_LEAN_AND_MEAN 1
+#  include <windows.h>
+#endif
+
 #include <assert.h>
 #include <float.h>
 #include <math.h>
@@ -539,6 +546,32 @@ is_uri_path_char(const char c)
   }
 }
 
+static char*
+serd_realpath(const char* const path)
+{
+  if (!path) {
+    return NULL;
+  }
+
+#ifdef _WIN32
+  const DWORD size = GetFullPathName(path, 0, NULL, NULL);
+  if (size == 0) {
+    return NULL;
+  }
+
+  char* const out = (char*)calloc(size, 1);
+  const DWORD ret = GetFullPathName(path, MAX_PATH, out, NULL);
+  if (ret == 0 || ret >= size) {
+    free(out);
+    return NULL;
+  }
+
+  return out;
+#else
+  return realpath(path, NULL);
+#endif
+}
+
 SerdNode*
 serd_new_file_uri(const SerdStringView path, const SerdStringView hostname)
 {
@@ -586,6 +619,20 @@ serd_new_file_uri(const SerdStringView path, const SerdStringView hostname)
   return node;
 }
 
+SerdNode*
+serd_new_real_file_uri(const char* const path, const char* const hostname)
+{
+  char* const real_path = serd_realpath(path);
+  if (!real_path) {
+    return NULL;
+  }
+
+  SerdNode* const node = serd_new_file_uri(SERD_MEASURE_STRING(real_path),
+                                           SERD_MEASURE_STRING(hostname));
+  free(real_path);
+  return node;
+}
+
 typedef size_t (*SerdWriteLiteralFunc)(const void* const user_data,
                                        const size_t      buf_size,
                                        char* const       buf);
diff --git a/test/.clang-tidy b/test/.clang-tidy
index de654d8d..b7dd752b 100644
--- a/test/.clang-tidy
+++ b/test/.clang-tidy
@@ -3,6 +3,10 @@ Checks: >
   -*-magic-numbers,
   -*-uppercase-literal-suffix,
   -android-cloexec-fopen,
+  -bugprone-reserved-identifier,
+  -bugprone-suspicious-string-compare,
+  -cert-dcl37-c,
+  -cert-dcl51-cpp,
   -clang-analyzer-nullability.NullabilityBase,
   -clang-analyzer-nullability.NullableDereferenced,
   -clang-analyzer-security.insecureAPI.DeprecatedOrUnsafeBufferHandling,
diff --git a/test/test_uri.c b/test/test_uri.c
index 88aa342c..dff4d999 100644
--- a/test/test_uri.c
+++ b/test/test_uri.c
@@ -20,6 +20,11 @@
 
 #include "serd/serd.h"
 
+#if defined(_WIN32) && !defined(__MINGW32__)
+#  include <io.h>
+#  define mkstemp(pat) _mktemp(pat)
+#endif
+
 #include <assert.h>
 #include <stdio.h>
 #include <stdlib.h>
@@ -80,6 +85,24 @@ test_uri_parsing(void)
 }
 
 static void
+test_real_file_uri(void)
+{
+  char path[16];
+  strncpy(path, "serd_XXXXXX", sizeof(path));
+  assert(mkstemp(path) > 0);
+
+  SerdNode* const good = serd_new_real_file_uri(path, NULL);
+  assert(good);
+  assert(!strncmp(serd_node_string(good), "file://", 7));
+  serd_node_free(good);
+
+  SerdNode* const bad = serd_new_real_file_uri("not_a_file", NULL);
+  assert(!bad);
+
+  assert(!remove(path));
+}
+
+static void
 test_parse_uri(void)
 {
   const SerdStringView base = SERD_STATIC_STRING("http://example.org/a/b/c/");
@@ -204,7 +227,7 @@ test_uri_resolution(void)
   const SerdURIView rel_foo_uri  = serd_relative_uri(abs_foo_uri, base_uri);
   const SerdURIView resolved_uri = serd_resolve_uri(rel_foo_uri, base_uri);
 
-SerdNode* const resolved = serd_new_parsed_uri(resolved_uri);
+  SerdNode* const resolved = serd_new_parsed_uri(resolved_uri);
   assert(!strcmp(serd_node_string(resolved), "http://example.org/a/b/c/foo"));
 
   serd_node_free(resolved);
@@ -215,6 +238,7 @@ main(void)
 {
   test_uri_parsing();
   test_parse_uri();
+  test_real_file_uri();
   test_is_within();
   test_relative_uri();
   test_uri_resolution();