0.8.18.21:

[sbcl.git] / src / code / target-c-call.lisp

diff --git a/src/code/target-c-call.lisp b/src/code/target-c-call.lisp
index 1d312c3..ccbdc53 100644 (file)
--- a/src/code/target-c-call.lisp
+++ b/src/code/target-c-call.lisp
@@ -57,93 +57,12 @@
   (declare (type system-area-pointer sap))
   (locally
     (declare (optimize (speed 3) (safety 0)))
-    (let ((length (do* ((offset 0)
-                        (byte (sap-ref-8 sap offset) (sap-ref-8 sap offset))
-                        (index 0 (1+ index)))
-                       ((zerop byte) index)
-                    (declare (type fixnum offset index))
-                    (cond
-                      ;; FIXME: Here, and below, we don't defend
-                      ;; against malformed utf-8 with any degree of
-                      ;; rigour.
-                      ((< byte #x80) (incf offset))
-                      ((< byte #xe0) (incf offset 2))
-                      ((< byte #xf0) (incf offset 3))
-                      (t (incf offset 4))))))
-      (let ((result (make-string length :element-type 'character)))
-        (do* ((offset 0)
-              (byte (sap-ref-8 sap offset) (sap-ref-8 sap offset))
-              (index 0 (1+ index)))
-            ((>= index length) result)
-          (declare (type fixnum offset index))
-          (setf (char result index)
-                (cond
-                  ((< byte #x80)
-                   (prog1 (code-char byte) (incf offset)))
-                  ((< byte #xe0)
-                   (prog1 (code-char (dpb byte (byte 5 6)
-                                          (sap-ref-8 sap (1+ offset))))
-                     (incf offset 2)))
-                  ((< byte #xf0)
-                   (prog1 (code-char
-                           (dpb byte (byte 4 12)
-                                (dpb (sap-ref-8 sap (1+ offset)) (byte 6 6)
-                                     (sap-ref-8 sap (+ 2 offset)))))
-                     (incf offset 3)))
-                  (t
-                   (prog1
-                       (code-char
-                        (dpb byte (byte 3 18)
-                             (dpb (sap-ref-8 sap (1+ offset)) (byte 6 12)
-                                  (dpb (sap-ref-8 sap (+ 2 offset)) (byte 6 6)
-                                       (sap-ref-8 sap (+ 3 offset))))))
-                     (incf offset 4))))))))))
+    (let ((byte-length (do* ((offset 0 (1+ offset))
+                            (byte #1=(sap-ref-8 sap offset) #1#))
+                           ((zerop byte) offset))))
+      (handler-bind ((sb!impl::octet-decoding-error #'sb!impl::use-unicode-replacement-char))
+       (sb!impl::utf8->string-sap-ref-8 sap 0 byte-length)))))
 
 (defun %deport-utf8-string (string)
   (declare (type simple-string string))
-  (locally
-    (declare (optimize (speed 3) (safety 0)))
-    (let ((length (1+ (do* ((offset 0)
-                            (length (length string))
-                            (index 0 (1+ index)))
-                           ((= index length) offset)
-                        (declare (type fixnum offset))
-                        (let ((bits (char-code (char string index))))
-                          (cond
-                            ((< bits #x80) (incf offset 1))
-                            ((< bits #x800) (incf offset 2))
-                            ((< bits #x10000) (incf offset 3))
-                            (t (incf offset 4))))))))
-      (let ((vector (make-array length :element-type '(unsigned-byte 8)
-                                :initial-element 0)))
-        (do* ((offset 0)
-              (length (length string))
-              (index 0 (1+ index)))
-             ((= index length) vector)
-          (declare (type fixnum offset))
-          (let ((bits (char-code (char string index))))
-            (cond
-              ((< bits #x80)
-               (setf (aref vector offset) bits)
-               (incf offset))
-              ((< bits #x800)
-               (setf (aref vector offset) (logior #xc0 (ldb (byte 5 6) bits)))
-               (setf (aref vector (1+ offset))
-                     (logior #x80 (ldb (byte 6 0) bits)))
-               (incf offset 2))
-              ((< bits #x10000)
-               (setf (aref vector offset) (logior #xe0 (ldb (byte 4 12) bits)))
-               (setf (aref vector (1+ offset))
-                     (logior #x80 (ldb (byte 6 6) bits)))
-               (setf (aref vector (+ offset 2))
-                     (logior #x80 (ldb (byte 6 0) bits)))
-               (incf offset 3))
-              (t
-               (setf (aref vector offset) (logior #xf0 (ldb (byte 3 18) bits)))
-               (setf (aref vector (1+ offset))
-                     (logior #x80 (ldb (byte 6 12) bits)))
-               (setf (aref vector (+ offset 2))
-                     (logior #x80 (ldb (byte 6 6) bits)))
-               (setf (aref vector (+ offset 3))
-                     (logior #x80 (ldb (byte 6 0) bits)))
-               (incf offset 4)))))))))
+  (sb!impl::string->utf8 string 0 (length string) 1))