[WIP] Add support for URI hex escape decoding

2 files changed, 25 insertions, 0 deletions

diff --git a/include/serd/reader.h b/include/serd/reader.h
index 78b51d00..57c8b2c3 100644
--- a/include/serd/reader.h
+++ b/include/serd/reader.h
@@ -101,6 +101,15 @@ typedef enum {
      from documents when the statements are sorted, such as in a model.
   */
   SERD_READ_ORDERED = 1U << 5U,
+
+  /**
+     Read URIs with percent-encoded UTF-8 characters decoded.
+
+     Normally, percent-encoded octets in URIs are treated as plain text and
+     preserved.  This flags enables UTF-8 decoding of URIs, so octet escapes
+     like "%7E" in URIs will be decoded to UTF-8 characters like "~".
+  */
+  SERD_READ_DECODED = 1U << 6U,
 } SerdReaderFlag;
 
 /// Bitwise OR of SerdReaderFlag values
diff --git a/include/serd/writer.h b/include/serd/writer.h
index bf54c46c..e3915a86 100644
--- a/include/serd/writer.h
+++ b/include/serd/writer.h
@@ -42,6 +42,10 @@ typedef enum {
      Although all the supported syntaxes are UTF-8 by definition, this can be
      used to escape all non-ASCII characters so that data will survive
      transmission through ASCII-only channels.
+
+     Non-printable-ASCII characters will be written as "U" escapes like
+     "\u007F" in string literals, and as hex-encoded UTF-8 bytes like "%07F" in
+     URIs.
   */
   SERD_WRITE_ASCII = 1U << 0U,
 
@@ -96,6 +100,18 @@ typedef enum {
      implicit context, so it will only be readable in a suitable enviromnent.
   */
   SERD_WRITE_CONTEXTUAL = 1U << 6U,
+
+  /**
+     Escape additional characters in RDF Test Cases format.
+
+     This writes "extended" characters as printable ASCII, using "U" escapes in
+     URIs instead of hex-encoding (escapes like "\u007F" instead of "%7F").
+     This is the format used by the outputs in the Turtle test suite (which
+     predates RDF 1.1 NTriples).  This style makes NTriples output
+     non-canonical, so it generally shouldn't be used except for compatibility
+     purposes.  See <https://www.w3.org/TR/rdf-testcases/>.
+  */
+  SERD_WRITE_ESCAPES = 1U << 7U,
 } SerdWriterFlag;
 
 /// Bitwise OR of #SerdWriterFlag values