From 503e73bcb1dee0065719c01e9cf4e6eaca61155e Mon Sep 17 00:00:00 2001
From: Chris Hanson <org/chris-hanson/cph>
Date: Tue, 22 May 2018 00:46:51 -0700
Subject: [PATCH] Move knowledge of unicode strings into sdata.h and object.h.

---
 src/microcode/object.h | 20 ++++++++++++++++++++
 src/microcode/sdata.h  |  6 +++++-
 src/microcode/string.c |  7 +------
 3 files changed, 26 insertions(+), 7 deletions(-)

diff --git a/src/microcode/object.h b/src/microcode/object.h
index ccaa59f02..5dc0c51df 100644
--- a/src/microcode/object.h
+++ b/src/microcode/object.h
@@ -275,6 +275,26 @@ extern bool string_p (SCHEME_OBJECT);
 #define BYTEVECTOR_REF(s, i) (* (BYTEVECTOR_LOC ((s), (i))))
 #define BYTEVECTOR_SET(s, i, c) ((* (BYTEVECTOR_LOC ((s), (i)))) = (c))
 
+/* Unicode string operations */
+
+#define UNICODE_STRING_CP_LENGTH(u)					\
+  (OBJECT_DATUM (MEMORY_REF ((u), UNICODE_STRING_LENGTH_INDEX)))
+
+#define UNICODE_STRING_FLAGS(u)						\
+  (OBJECT_TYPE (MEMORY_REF ((u), UNICODE_STRING_LENGTH_INDEX)))
+
+/* This must be kept in sync with "runtime/string.scm". */
+#define UNICODE_STRING_BYTES_PER_CP(u)					\
+  ((((UNICODE_STRING_FLAGS (u)) & 0x3) == 0)				\
+   ? 3									\
+   : ((UNICODE_STRING_FLAGS (u)) & 0x3))
+
+#define UNICODE_STRING_BYTE_LENGTH(u)					\
+  ((UNICODE_STRING_CP_LENGTH (u)) * (UNICODE_STRING_BYTES_PER_CP (u)))
+
+#define UNICODE_STRING_POINTER(u)					\
+  ((uint8_t *) (MEMORY_LOC ((u), UNICODE_STRING_DATA)))
+
 /* Legacy string operations */
 
 /* Legacy strings are laid out exactly the same way as bytevectors,
diff --git a/src/microcode/sdata.h b/src/microcode/sdata.h
index 6caf5070d..6396a03bc 100644
--- a/src/microcode/sdata.h
+++ b/src/microcode/sdata.h
@@ -56,7 +56,7 @@ USA.
  * An object that points to one other object (extra indirection).
  * Used by the compiler to share objects.
  */
-#define CELL_CONTENTS 		0
+#define CELL_CONTENTS		0
 
 /* BYTEVECTOR
  * Format consists of the normal non-marked vector header
@@ -67,6 +67,10 @@ USA.
 #define BYTEVECTOR_LENGTH_INDEX	1
 #define BYTEVECTOR_LENGTH_SIZE	1
 #define BYTEVECTOR_DATA		2
+
+#define UNICODE_STRING_HEADER 0
+#define UNICODE_STRING_LENGTH_INDEX 1
+#define UNICODE_STRING_DATA 2
 
 /* COMPILED_PROCEDURE */
 #define COMP_PROCEDURE_ADDRESS	0
diff --git a/src/microcode/string.c b/src/microcode/string.c
index 0f249f4a5..cfef1b36d 100644
--- a/src/microcode/string.c
+++ b/src/microcode/string.c
@@ -36,12 +36,7 @@ string_p (SCHEME_OBJECT object)
   return ((LEGACY_STRING_P (object))
           || (BYTEVECTOR_P (object))
           || ((UNICODE_STRING_P (object))
-              // This tests that the ustring-cp-size is == 1, meaning
-              // one byte per code point.  This must be kept in sync
-              // with "runtime/ustring.scm".
-              && (((OBJECT_TYPE (MEMORY_REF (object, BYTEVECTOR_LENGTH_INDEX)))
-                   & 0x03)
-                  == 0x01)));
+	      && ((UNICODE_STRING_BYTES_PER_CP (object)) == 1)));
 }
 
 SCHEME_OBJECT
-- 
2.25.1